Merge pull request #3003 from karl-zylinski/patch-1

virtual arena: Actually use DEFAULT_ARENA_STATIC_RESERVE_SIZE as default value on arena_init_static
Merge pull request #3004 from flysand7/strings-doc
2026-07-05 11:11:37 -07:00 · 2023-12-05 16:04:25 +00:00 · 2023-12-05 16:04:07 +00:00 · 2023-12-05 23:17:19 +11:00 · 2023-12-04 21:15:46 +01:00 · 2023-12-04 14:06:19 +00:00
851 changed files with 156053 additions and 67891 deletions
@@ -0,0 +1,10 @@
+root = true
+
+[*]
+indent_style = tab
+end_of_line = lf
+
+[*.y{,a}ml]
+indent_size = 2
+indent_style = space
+tab_width = 2
@@ -38,10 +38,10 @@ jobs:
          cd tests/vendor
          make
        timeout-minutes: 10
-      - name: Odin issues tests
+      - name: Odin internals tests
        run: |
-          cd tests/issues
-          ./run.sh
+          cd tests/internal
+          make
        timeout-minutes: 10
      - name: Odin check examples/all for Linux i386
        run: ./odin check examples/all -vet -strict-style -target:linux_i386
@@ -58,8 +58,8 @@ jobs:
      - uses: actions/checkout@v1
      - name: Download LLVM, botan and setup PATH
        run: |
-          brew install llvm@11 botan
-          echo "/usr/local/opt/llvm@11/bin" >> $GITHUB_PATH
+          brew install llvm@13 botan
+          echo "/usr/local/opt/llvm@13/bin" >> $GITHUB_PATH
          TMP_PATH=$(xcrun --show-sdk-path)/user/include
          echo "CPATH=$TMP_PATH" >> $GITHUB_ENV
      - name: build odin
@@ -87,16 +87,11 @@ jobs:
          cd tests/core
          make
        timeout-minutes: 10
-      - name: Vendor library tests
+      - name: Odin internals tests
        run: |
-          cd tests/vendor
+          cd tests/internal
          make
        timeout-minutes: 10
-      - name: Odin issues tests
-        run: |
-          cd tests/issues
-          ./run.sh
-        timeout-minutes: 10
      - name: Odin check examples/all for Darwin arm64
        run: ./odin check examples/all -vet -strict-style -target:darwin_arm64
        timeout-minutes: 10
@@ -104,13 +99,13 @@ jobs:
        run: ./odin check examples/all -vet -strict-style -target:linux_arm64
        timeout-minutes: 10
  build_windows:
-    runs-on: windows-2019
+    runs-on: windows-2022
    steps:
      - uses: actions/checkout@v1
      - name: build Odin
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          ./build.bat 1
      - name: Odin version
        run: ./odin version
@@ -121,58 +116,65 @@ jobs:
      - name: Odin check
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin check examples/demo -vet
        timeout-minutes: 10
      - name: Odin run
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin run examples/demo
        timeout-minutes: 10
      - name: Odin run -debug
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin run examples/demo -debug
        timeout-minutes: 10
      - name: Odin check examples/all
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin check examples/all -strict-style
        timeout-minutes: 10
      - name: Core library tests
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          cd tests\core
          call build.bat
        timeout-minutes: 10
      - name: Vendor library tests
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          cd tests\vendor
          call build.bat
        timeout-minutes: 10
+      - name: Odin internals tests
+        shell: cmd
+        run: |
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          cd tests\internal
+          call build.bat
+        timeout-minutes: 10
+      - name: Odin documentation tests
+        shell: cmd
+        run: |
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          cd tests\documentation
+          call build.bat
+        timeout-minutes: 10
      - name: core:math/big tests
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          cd tests\core\math\big
          call build.bat
        timeout-minutes: 10
-      - name: Odin issues tests
-        shell: cmd
-        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
-          cd tests\issues
-          call run.bat
-        timeout-minutes: 10
      - name: Odin check examples/all for Windows 32bits
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin check examples/all -strict-style -target:windows_i386
        timeout-minutes: 10
@@ -7,24 +7,26 @@ on:

 jobs:
  build_windows:
-    runs-on: windows-2019
+    if: github.repository == 'odin-lang/Odin'
+    runs-on: windows-2022
    steps:
      - uses: actions/checkout@v1
      - name: build Odin
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          ./build.bat 1 1
      - name: Odin run
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin run examples/demo
      - name: Copy artifacts
        run: |
          rm bin/llvm/windows/LLVM-C.lib
          mkdir dist
          cp odin.exe dist
+          cp LICENSE dist
          cp LLVM-C.dll dist
          cp -r shared dist
          cp -r core dist
@@ -37,6 +39,7 @@ jobs:
          name: windows_artifacts
          path: dist
  build_ubuntu:
+    if: github.repository == 'odin-lang/Odin'
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v1
@@ -50,6 +53,8 @@ jobs:
        run: |
          mkdir dist
          cp odin dist
+          cp LICENSE dist
+          cp libLLVM* dist
          cp -r shared dist
          cp -r core dist
          cp -r vendor dist
@@ -60,13 +65,14 @@ jobs:
          name: ubuntu_artifacts
          path: dist
  build_macos:
+    if: github.repository == 'odin-lang/Odin'
    runs-on: macOS-latest
    steps:
      - uses: actions/checkout@v1
      - name: Download LLVM and setup PATH
        run: |
-          brew install llvm@11
-          echo "/usr/local/opt/llvm@11/bin" >> $GITHUB_PATH
+          brew install llvm@13
+          echo "/usr/local/opt/llvm@13/bin" >> $GITHUB_PATH
          TMP_PATH=$(xcrun --show-sdk-path)/user/include
          echo "CPATH=$TMP_PATH" >> $GITHUB_ENV
      - name: build odin
@@ -77,6 +83,7 @@ jobs:
        run: |
          mkdir dist
          cp odin dist
+          cp LICENSE dist
          cp -r shared dist
          cp -r core dist
          cp -r vendor dist
@@ -13,7 +13,7 @@ jobs:
    runs-on: ubuntu-latest
    steps:
      - name: Close Stale Issues
-        uses: actions/stale@v4.1.0
+        uses: actions/stale@v7.0.0
        with:
 #          stale-issue-message: |
 #            Hello!
@@ -36,7 +36,7 @@ jobs:
 #           The motivation for this automation is to help prioritize issues in the backlog and not ignore, reject, or belittle anyone..

          days-before-stale: 120
-          days-before-close: 30
+          days-before-close: -1
          exempt-draft-pr: true
          ascending: true
          operations-per-run: 1000
@@ -22,6 +22,32 @@ bld/
 [Oo]bj/
 [Ll]og/
 ![Cc]ore/[Ll]og/
+tests/documentation/verify/
+tests/documentation/all.odin-doc
+tests/internal/test_map
+tests/internal/test_pow
+tests/internal/test_rtti
+tests/core/test_core_compress
+tests/core/test_core_filepath
+tests/core/test_core_fmt
+tests/core/test_core_i18n
+tests/core/test_core_image
+tests/core/test_core_libc
+tests/core/test_core_match
+tests/core/test_core_math
+tests/core/test_core_net
+tests/core/test_core_os_exit
+tests/core/test_core_reflect
+tests/core/test_core_strings
+tests/core/test_crypto_hash
+tests/core/test_hash
+tests/core/test_hxa
+tests/core/test_json
+tests/core/test_linalg_glsl_math
+tests/core/test_noise
+tests/core/test_varint
+tests/core/test_xml
+tests/vendor/vendor_botan
 # Visual Studio 2015 cache/options directory
 .vs/
 # Visual Studio Code options directory
@@ -268,9 +294,11 @@ bin/

 # - Linux/MacOS
 odin
+!odin/
 odin.dSYM
 *.bin
 demo.bin
+libLLVM*.so*

 # shared collection
 shared/
@@ -283,3 +311,4 @@ shared/
 *.sublime-workspace
 examples/bug/
 build.sh
+!core/debug/
@@ -11,7 +11,7 @@
        <img src="https://img.shields.io/badge/platforms-Windows%20|%20Linux%20|%20macOS-green.svg">
    </a>
    <br>
-    <a href="https://discord.gg/odinlang">
+    <a href="https://discord.com/invite/sVBPHEv">
        <img src="https://img.shields.io/discord/568138951836172421?logo=discord">
    </a>
    <a href="https://github.com/odin-lang/odin/actions">
@@ -82,7 +82,7 @@ A wiki maintained by the Odin community.

 #### [Odin Discord](https://discord.gg/sVBPHEv)

-Get live support and talk with other odiners on the Odin Discord.
+Get live support and talk with other Odin programmers on the Odin Discord.

 ### Articles

@@ -2,6 +2,23 @@

 setlocal EnableDelayedExpansion

+where /Q cl.exe || (
+	set __VSCMD_ARG_NO_LOGO=1
+	for /f "tokens=*" %%i in ('"C:\Program Files (x86)\Microsoft Visual Studio\Installer\vswhere.exe" -latest -requires Microsoft.VisualStudio.Workload.NativeDesktop -property installationPath') do set VS=%%i
+	if "!VS!" equ "" (
+		echo ERROR: Visual Studio installation not found
+		exit /b 1
+	)
+	call "!VS!\VC\Auxiliary\Build\vcvarsall.bat" amd64 || exit /b 1
+)
+
+if "%VSCMD_ARG_TGT_ARCH%" neq "x64" (
+	if "%ODIN_IGNORE_MSVC_CHECK%" == "" (
+		echo ERROR: please run this from MSVC x64 native tools command prompt, 32-bit target is not supported!
+		exit /b 1
+	)
+)
+
 for /f "usebackq tokens=1,2 delims=,=- " %%i in (`wmic os get LocalDateTime /value`) do @if %%i==LocalDateTime (
 	set CURR_DATE_TIME=%%j
 )
@@ -33,8 +50,14 @@ set odin_version_raw="dev-%curr_year%-%curr_month%"
 set compiler_flags= -nologo -Oi -TP -fp:precise -Gm- -MP -FC -EHsc- -GR- -GF
 set compiler_defines= -DODIN_VERSION_RAW=\"%odin_version_raw%\"

-for /f %%i in ('git rev-parse --short HEAD') do set GIT_SHA=%%i
+if not exist .git\ goto skip_git_hash
+for /f "tokens=1,2" %%i IN ('git show "--pretty=%%cd %%h" "--date=format:%%Y-%%m" --no-patch --no-notes HEAD') do (
+	set odin_version_raw=dev-%%i
+	set GIT_SHA=%%j
+)
 if %ERRORLEVEL% equ 0 set compiler_defines=%compiler_defines% -DGIT_SHA=\"%GIT_SHA%\"
+:skip_git_hash
+
 if %nightly% equ 1 set compiler_defines=%compiler_defines% -DNIGHTLY

 if %release_mode% EQU 0 ( rem Debug
@@ -47,14 +70,26 @@ if %release_mode% EQU 0 ( rem Debug
 set compiler_warnings= ^
 	-W4 -WX ^
 	-wd4100 -wd4101 -wd4127 -wd4146 ^
+	-wd4505 ^
 	-wd4456 -wd4457

 set compiler_includes= ^
 	/Isrc\
 set libs= ^
 	kernel32.lib ^
+	Synchronization.lib ^
 	bin\llvm\windows\LLVM-C.lib

+rem DO NOT TOUCH!
+rem THIS TILDE STUFF IS FOR DEVELOPMENT ONLY!
+set tilde_backend=0
+if %tilde_backend% EQU 1 (
+	set libs=%libs% src\tilde\tb.lib
+	set compiler_defines=%compiler_defines% -DODIN_TILDE_BACKEND
+)
+rem DO NOT TOUCH!
+
+
 set linker_flags= -incremental:no -opt:ref -subsystem:console

 if %release_mode% EQU 0 ( rem Debug
@@ -75,7 +110,8 @@ if %errorlevel% neq 0 goto end_of_build
 call build_vendor.bat
 if %errorlevel% neq 0 goto end_of_build

-if %release_mode% EQU 0 odin run examples/demo
+rem If the demo doesn't run for you and your CPU is more than a decade old, try -microarch:native
+if %release_mode% EQU 0 odin run examples/demo -- Hellope World

 del *.obj > NUL 2> NUL

@@ -1,94 +1,92 @@
-#!/usr/bin/env bash
+#!/usr/bin/env sh
 set -eu

-: ${CXX=clang++}
 : ${CPPFLAGS=}
+: ${CXX=clang++}
 : ${CXXFLAGS=}
 : ${LDFLAGS=}
-: ${ODIN_VERSION=dev-$(date +"%Y-%m")}
+: ${LLVM_CONFIG=}

-CPPFLAGS="$CPPFLAGS -DODIN_VERSION_RAW=\"$ODIN_VERSION\""
+CPPFLAGS="$CPPFLAGS -DODIN_VERSION_RAW=\"dev-$(date +"%Y-%m")\""
 CXXFLAGS="$CXXFLAGS -std=c++14"
-LDFLAGS="$LDFLAGS -pthread -lm -lstdc++"
-
-GIT_SHA=$(git rev-parse --short HEAD || :)
-if [ "$GIT_SHA" ]; then CPPFLAGS="$CPPFLAGS -DGIT_SHA=\"$GIT_SHA\""; fi
-
 DISABLED_WARNINGS="-Wno-switch -Wno-macro-redefined -Wno-unused-value"
-OS=$(uname)
+LDFLAGS="$LDFLAGS -pthread -lm -lstdc++"
+OS_ARCH="$(uname -m)"
+OS_NAME="$(uname -s)"

-panic() {
-	printf "%s\n" "$1"
+if [ -d ".git" ] && [ -n "$(command -v git)" ]; then
+	GIT_SHA=$(git show --pretty='%h' --no-patch --no-notes HEAD)
+	CPPFLAGS="$CPPFLAGS -DGIT_SHA=\"$GIT_SHA\""
+fi
+
+error() {
+	printf "ERROR: %s\n" "$1"
 	exit 1
 }

-version() { echo "$@" | awk -F. '{ printf("%d%03d%03d%03d\n", $1,$2,$3,$4); }'; }
-
-config_darwin() {
-	ARCH=$(uname -m)
-	: ${LLVM_CONFIG=llvm-config}
-
-	# allow for arm only llvm's with version 13
-	if [ ARCH == arm64 ]; then
-		MIN_LLVM_VERSION=("13.0.0")
+if [ -z "$LLVM_CONFIG" ]; then
+	# darwin, linux, openbsd
+	if   [ -n "$(command -v llvm-config-17)" ]; then LLVM_CONFIG="llvm-config-17"
+	elif [ -n "$(command -v llvm-config-14)" ]; then LLVM_CONFIG="llvm-config-14"
+	elif [ -n "$(command -v llvm-config-13)" ]; then LLVM_CONFIG="llvm-config-13"
+	elif [ -n "$(command -v llvm-config-12)" ]; then LLVM_CONFIG="llvm-config-12"
+	elif [ -n "$(command -v llvm-config-11)" ]; then LLVM_CONFIG="llvm-config-11"
+	# freebsd
+	elif [ -n "$(command -v llvm-config17)" ]; then  LLVM_CONFIG="llvm-config-17"
+	elif [ -n "$(command -v llvm-config14)" ]; then  LLVM_CONFIG="llvm-config-14"
+	elif [ -n "$(command -v llvm-config13)" ]; then  LLVM_CONFIG="llvm-config-13"
+	elif [ -n "$(command -v llvm-config12)" ]; then  LLVM_CONFIG="llvm-config-12"
+	elif [ -n "$(command -v llvm-config11)" ]; then  LLVM_CONFIG="llvm-config-11"
+	# fallback
+	elif [ -n "$(command -v llvm-config)" ]; then LLVM_CONFIG="llvm-config"
 	else
-		# allow for x86 / amd64 all llvm versions begining from 11
-		MIN_LLVM_VERSION=("11.1.0")
+		error "No llvm-config command found. Set LLVM_CONFIG to proceed."
 	fi
+fi

-	if [ $(version $($LLVM_CONFIG --version)) -lt $(version $MIN_LLVM_VERSION) ]; then
-		if [ ARCH == arm64 ]; then
-			panic "Requirement: llvm-config must be base version 13 for arm64"
-		else
-			panic "Requirement: llvm-config must be base version greater than 11 for amd64/x86"
+LLVM_VERSION="$($LLVM_CONFIG --version)"
+LLVM_VERSION_MAJOR="$(echo $LLVM_VERSION | awk -F. '{print $1}')"
+LLVM_VERSION_MINOR="$(echo $LLVM_VERSION | awk -F. '{print $2}')"
+LLVM_VERSION_PATCH="$(echo $LLVM_VERSION | awk -F. '{print $3}')"
+
+if [ $LLVM_VERSION_MAJOR -lt 11 ] ||
+	([ $LLVM_VERSION_MAJOR -gt 14 ] && [ $LLVM_VERSION_MAJOR -lt 17 ]); then
+	error "Invalid LLVM version $LLVM_VERSION: must be 11, 12, 13, 14 or 17"
+fi
+
+case "$OS_NAME" in
+Darwin)
+	if [ "$OS_ARCH" == "arm64" ]; then
+		if [ $LLVM_VERSION_MAJOR -lt 13 ] || [ $LLVM_VERSION_MAJOR -gt 17 ]; then
+			error "Darwin Arm64 requires LLVM 13, 14 or 17"
 		fi
 	fi

-	LDFLAGS="$LDFLAGS -liconv -ldl"
 	CXXFLAGS="$CXXFLAGS $($LLVM_CONFIG --cxxflags --ldflags)"
+	LDFLAGS="$LDFLAGS -liconv -ldl -framework System"
 	LDFLAGS="$LDFLAGS -lLLVM-C"
-}
-
-config_freebsd() {
-	: ${LLVM_CONFIG=/usr/local/bin/llvm-config11}
-
+	;;
+FreeBSD)
 	CXXFLAGS="$CXXFLAGS $($LLVM_CONFIG --cxxflags --ldflags)"
 	LDFLAGS="$LDFLAGS $($LLVM_CONFIG --libs core native --system-libs)"
-}
-
-config_openbsd() {
-	: ${LLVM_CONFIG=/usr/local/bin/llvm-config}
-
+	;;
+Linux)
+	CXXFLAGS="$CXXFLAGS $($LLVM_CONFIG --cxxflags --ldflags)"
+	LDFLAGS="$LDFLAGS -ldl $($LLVM_CONFIG --libs core native --system-libs --libfiles)"
+	# Copy libLLVM*.so into current directory for linking
+	# NOTE: This is needed by the Linux release pipeline!
+	cp $(readlink -f $($LLVM_CONFIG --libfiles)) ./
+	LDFLAGS="$LDFLAGS -Wl,-rpath=\$ORIGIN"
+	;;
+OpenBSD)
+	CXXFLAGS="$CXXFLAGS $($LLVM_CONFIG --cxxflags --ldflags)"
 	LDFLAGS="$LDFLAGS -liconv"
-	CXXFLAGS="$CXXFLAGS $($LLVM_CONFIG --cxxflags --ldflags)"
 	LDFLAGS="$LDFLAGS $($LLVM_CONFIG --libs core native --system-libs)"
-}
-
-config_linux() {
-	: ${LLVM_CONFIG=}
-
-	if [ ! "$LLVM_CONFIG" ]; then
-		if which llvm-config > /dev/null 2>&1; then
-			LLVM_CONFIG=llvm-config
-		elif which llvm-config-11 > /dev/null 2>&1; then
-			LLVM_CONFIG=llvm-config-11
-		elif which llvm-config-11-64 > /dev/null 2>&1; then
-			LLVM_CONFIG=llvm-config-11-64
-		else
-			panic "Unable to find LLVM-config"
-		fi
-	fi
-
-	MIN_LLVM_VERSION=("11.0.0")
-	if [ $(version $($LLVM_CONFIG --version)) -lt $(version $MIN_LLVM_VERSION) ]; then
-		echo "Tried to use " $(which $LLVM_CONFIG) "version" $($LLVM_CONFIG --version)
-		panic "Requirement: llvm-config must be base version greater than 11"
-	fi
-
-	LDFLAGS="$LDFLAGS -ldl"
-	CXXFLAGS="$CXXFLAGS $($LLVM_CONFIG --cxxflags --ldflags)"
-	LDFLAGS="$LDFLAGS $($LLVM_CONFIG --libs core native --system-libs)"
-}
+	;;
+*)
+	error "Platform \"$OS_NAME\" unsupported"
+	;;
+esac

 build_odin() {
 	case $1 in
@@ -99,13 +97,20 @@ build_odin() {
 		EXTRAFLAGS="-O3"
 		;;
 	release-native)
-		EXTRAFLAGS="-O3 -march=native"
+		if [ "$OS_ARCH" == "arm64" ]; then
+			# Use preferred flag for Arm (ie arm64 / aarch64 / etc)
+			EXTRAFLAGS="-O3 -mcpu=native"
+		else
+			# Use preferred flag for x86 / amd64
+			EXTRAFLAGS="-O3 -march=native"
+		fi
 		;;
 	nightly)
 		EXTRAFLAGS="-DNIGHTLY -O3"
 		;;
 	*)
-		panic "Build mode unsupported!"
+		error "Build mode \"$1\" unsupported!"
+		;;
 	esac

 	set -x
@@ -114,49 +119,23 @@ build_odin() {
 }

 run_demo() {
-	./odin run examples/demo/demo.odin -file
+	./odin run examples/demo/demo.odin -file -- Hellope World
 }

-case $OS in
-Linux)
-	config_linux
-	;;
-Darwin)
-	config_darwin
-	;;
-OpenBSD)
-	config_openbsd
-	;;
-FreeBSD)
-	config_freebsd
-	;;
-*)
-	panic "Platform unsupported!"
-esac
-
-if [[ $# -eq 0 ]]; then
+if [ $# -eq 0 ]; then
 	build_odin debug
 	run_demo
-	exit 0
-fi
-
-if [[ $# -eq 1 ]]; then
+elif [ $# -eq 1 ]; then
 	case $1 in
 	report)
-		if [[ ! -f "./odin" ]]; then
-			build_odin debug
-		fi
-
+		[ ! -f "./odin" ] && build_odin debug
 		./odin report
-		exit 0
 		;;
 	*)
 		build_odin $1
 		;;
 	esac
-
 	run_demo
-	exit 0
 else
-	panic "Too many arguments!"
+	error "Too many arguments!"
 fi
@@ -15,3 +15,10 @@ if not exist "vendor\miniaudio\lib\*.lib" (
 		call build.bat
 	popd
 )
+
+
+if not exist "vendor\cgltf\lib\*.lib" (
+	pushd vendor\cgltf\src
+		call build.bat
+	popd
+)
@@ -2,25 +2,25 @@ package bufio

 import "core:io"

-// Loadahead_Reader provides io lookahead.
+// Lookahead_Reader provides io lookahead.
 // This is useful for tokenizers/parsers.
-// Loadahead_Reader is similar to bufio.Reader, but unlike bufio.Reader, Loadahead_Reader's buffer size
+// Lookahead_Reader is similar to bufio.Reader, but unlike bufio.Reader, Lookahead_Reader's buffer size
 // will EXACTLY match the specified size, whereas bufio.Reader's buffer size may differ from the specified size.
 // This makes sure that the buffer will not be accidentally read beyond the expected size.
-Loadahead_Reader :: struct {
+Lookahead_Reader :: struct {
 	r:   io.Reader,
 	buf: []byte,
 	n:   int,
 }

-lookahead_reader_init :: proc(lr: ^Loadahead_Reader, r: io.Reader, buf: []byte) -> ^Loadahead_Reader {
+lookahead_reader_init :: proc(lr: ^Lookahead_Reader, r: io.Reader, buf: []byte) -> ^Lookahead_Reader {
 	lr.r = r
 	lr.buf = buf
 	lr.n = 0
 	return lr
 }

-lookahead_reader_buffer :: proc(lr: ^Loadahead_Reader) -> []byte {
+lookahead_reader_buffer :: proc(lr: ^Lookahead_Reader) -> []byte {
 	return lr.buf[:lr.n]
 }

@@ -28,7 +28,7 @@ lookahead_reader_buffer :: proc(lr: ^Loadahead_Reader) -> []byte {
 // lookahead_reader_peek returns a slice of the Lookahead_Reader which holds n bytes
 // If the Lookahead_Reader cannot hold enough bytes, it will read from the underlying reader to populate the rest.
 // NOTE: The returned buffer is not a copy of the underlying buffer
-lookahead_reader_peek :: proc(lr: ^Loadahead_Reader, n: int) -> ([]byte, io.Error) {
+lookahead_reader_peek :: proc(lr: ^Lookahead_Reader, n: int) -> ([]byte, io.Error) {
 	switch {
 	case n < 0:
 		return nil, .Negative_Read
@@ -58,13 +58,13 @@ lookahead_reader_peek :: proc(lr: ^Loadahead_Reader, n: int) -> ([]byte, io.Erro
 // lookahead_reader_peek_all returns a slice of the Lookahead_Reader populating the full buffer
 // If the Lookahead_Reader cannot hold enough bytes, it will read from the underlying reader to populate the rest.
 // NOTE: The returned buffer is not a copy of the underlying buffer
-lookahead_reader_peek_all :: proc(lr: ^Loadahead_Reader) -> ([]byte, io.Error) {
+lookahead_reader_peek_all :: proc(lr: ^Lookahead_Reader) -> ([]byte, io.Error) {
 	return lookahead_reader_peek(lr, len(lr.buf))
 }


 // lookahead_reader_consume drops the first n populated bytes from the Lookahead_Reader.
-lookahead_reader_consume :: proc(lr: ^Loadahead_Reader, n: int) -> io.Error {
+lookahead_reader_consume :: proc(lr: ^Lookahead_Reader, n: int) -> io.Error {
 	switch {
 	case n == 0:
 		return nil
@@ -78,6 +78,6 @@ lookahead_reader_consume :: proc(lr: ^Loadahead_Reader, n: int) -> io.Error {
 	return nil
 }

-lookahead_reader_consume_all :: proc(lr: ^Loadahead_Reader) -> io.Error {
+lookahead_reader_consume_all :: proc(lr: ^Lookahead_Reader) -> io.Error {
 	return lookahead_reader_consume(lr, lr.n)
 }
@@ -14,55 +14,29 @@ read_writer_init :: proc(rw: ^Read_Writer, r: ^Reader, w: ^Writer) {
 }

 read_writer_to_stream :: proc(rw: ^Read_Writer) -> (s: io.Stream) {
-	s.stream_data = rw
-	s.stream_vtable = _read_writer_vtable
+	s.procedure = _read_writer_procedure
+	s.data = rw
 	return
 }

@(private)
-_read_writer_vtable := &io.Stream_VTable{
-	impl_read = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_read(b, p)
-	},
-	impl_read_byte = proc(s: io.Stream) -> (c: byte, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_read_byte(b)
-	},
-	impl_unread_byte = proc(s: io.Stream) -> io.Error {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_unread_byte(b)
-	},
-	impl_read_rune = proc(s: io.Stream) -> (r: rune, size: int, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_read_rune(b)
-	},
-	impl_unread_rune = proc(s: io.Stream) -> io.Error {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_unread_rune(b)
-	},
-	impl_write_to = proc(s: io.Stream, w: io.Writer) -> (n: i64, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_write_to(b, w)
-	},
-	impl_flush = proc(s: io.Stream)  -> io.Error {
-		b := (^Read_Writer)(s.stream_data).w
-		return writer_flush(b)
-	},
-	impl_write = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).w
-		return writer_write(b, p)
-	},
-	impl_write_byte = proc(s: io.Stream, c: byte) -> io.Error {
-		b := (^Read_Writer)(s.stream_data).w
-		return writer_write_byte(b, c)
-	},
-	impl_write_rune = proc(s: io.Stream, r: rune) -> (int, io.Error) {
-		b := (^Read_Writer)(s.stream_data).w
-		return writer_write_rune(b, r)
-	},
-	impl_read_from = proc(s: io.Stream, r: io.Reader) -> (n: i64, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).w
-		return writer_read_from(b, r)
-	},
-}
+_read_writer_procedure := proc(stream_data: rawptr, mode: io.Stream_Mode, p: []byte, offset: i64, whence: io.Seek_From) -> (n: i64, err: io.Error) {
+	rw := (^Read_Writer)(stream_data)
+	n_int: int
+	#partial switch mode {
+	case .Flush:
+		err = writer_flush(rw.w)
+		return
+	case .Read:
+		n_int, err = reader_read(rw.r, p)
+		n = i64(n_int)
+		return
+	case .Write:
+		n_int, err = writer_write(rw.w, p)
+		n = i64(n_int)
+		return
+	case .Query:
+		return io.query_utility({.Flush, .Read, .Write, .Query})
+	}
+	return 0, .Empty
+}
@@ -118,9 +118,7 @@ reader_peek :: proc(b: ^Reader, n: int) -> (data: []byte, err: io.Error) {
 	b.last_rune_size = -1

 	for b.w-b.r < n && b.w-b.r < len(b.buf) && b.err == nil {
-		if fill_err := _reader_read_new_chunk(b); fill_err != nil {
-			return nil, fill_err
-		}
+		_reader_read_new_chunk(b) or_return
 	}

 	if n > len(b.buf) {
@@ -156,9 +154,7 @@ reader_discard :: proc(b: ^Reader, n: int) -> (discarded: int, err: io.Error) {
 	for {
 		skip := reader_buffered(b)
 		if skip == 0 {
-			if fill_err := _reader_read_new_chunk(b); fill_err != nil {
-				return 0, fill_err
-			}
+			_reader_read_new_chunk(b) or_return
 			skip = reader_buffered(b)
 		}
 		skip = min(skip, remaining)
@@ -223,20 +219,18 @@ reader_read :: proc(b: ^Reader, p: []byte) -> (n: int, err: io.Error) {

 // reader_read_byte reads and returns a single byte
 // If no byte is available, it return an error
-reader_read_byte :: proc(b: ^Reader) -> (byte, io.Error) {
+reader_read_byte :: proc(b: ^Reader) -> (c: byte, err: io.Error) {
 	b.last_rune_size = -1
 	for b.r == b.w {
 		if b.err != nil {
 			return 0, _reader_consume_err(b)
 		}
-		if err := _reader_read_new_chunk(b); err != nil {
-			return 0, err
-		}
+		_reader_read_new_chunk(b) or_return
 	}
-	c := b.buf[b.r]
+	c = b.buf[b.r]
 	b.r += 1
 	b.last_byte = int(c)
-	return c, nil
+	return
 }

 // reader_unread_byte unreads the last byte. Only the most recently read byte can be unread
@@ -264,15 +258,12 @@ reader_read_rune :: proc(b: ^Reader) -> (r: rune, size: int, err: io.Error) {
 	    !utf8.full_rune(b.buf[b.r:b.w]) &&
 	    b.err == nil &&
 	    b.w-b.w < len(b.buf) {
-		if err = _reader_read_new_chunk(b); err != nil {
-			return
-		}
+		_reader_read_new_chunk(b) or_return
 	}

 	b.last_rune_size = -1
 	if b.r == b.w {
-		err = _reader_consume_err(b)
-		return
+		return 0, 0, _reader_consume_err(b)
 	}
 	r, size = rune(b.buf[b.r]), 1
 	if r >= utf8.RUNE_SELF {
@@ -305,39 +296,20 @@ reader_write_to :: proc(b: ^Reader, w: io.Writer) -> (n: i64, err: io.Error) {
 		return i64(n), err
 	}

-	n, err = write_buf(b, w)
-	if err != nil {
-		return
-	}
+	n = write_buf(b, w) or_return

 	m: i64
-	if nr, ok := io.to_writer_to(b.rd); ok {
-		m, err = io.write_to(nr, w)
-		n += m
-		return n, err
-	}
-
-	if nw, ok := io.to_reader_from(w); ok {
-		m, err = io.read_from(nw, b.rd)
-		n += m
-		return n, err
-	}
-
 	if b.w-b.r < len(b.buf) {
-		if err = _reader_read_new_chunk(b); err != nil {
-			return
-		}
+		_reader_read_new_chunk(b) or_return
 	}

 	for b.r < b.w {
 		m, err = write_buf(b, w)
-		n += m
+		n += m // this needs to be done before returning
 		if err != nil {
 			return
 		}
-		if err = _reader_read_new_chunk(b); err != nil {
-			return
-		}
+		_reader_read_new_chunk(b) or_return
 	}

 	if b.err == .EOF {
@@ -352,48 +324,28 @@ reader_write_to :: proc(b: ^Reader, w: io.Writer) -> (n: i64, err: io.Error) {

 // reader_to_stream converts a Reader into an io.Stream
 reader_to_stream :: proc(b: ^Reader) -> (s: io.Stream) {
-	s.stream_data = b
-	s.stream_vtable = _reader_vtable
+	s.data = b
+	s.procedure = _reader_proc
 	return
 }



@(private)
-_reader_vtable := &io.Stream_VTable{
-	impl_destroy = proc(s: io.Stream) -> io.Error {
-		b := (^Reader)(s.stream_data)
+_reader_proc :: proc(stream_data: rawptr, mode: io.Stream_Mode, p: []byte, offset: i64, whence: io.Seek_From) -> (n: i64, err: io.Error) {
+	b := (^Reader)(stream_data)
+	#partial switch mode {
+	case .Read:
+		return io._i64_err(reader_read(b, p))
+	case .Destroy:
 		reader_destroy(b)
-		return nil
-	},
-	impl_read = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Reader)(s.stream_data)
-		return reader_read(b, p)
-	},
-	impl_read_byte = proc(s: io.Stream) -> (c: byte, err: io.Error) {
-		b := (^Reader)(s.stream_data)
-		return reader_read_byte(b)
-	},
-	impl_unread_byte = proc(s: io.Stream) -> io.Error {
-		b := (^Reader)(s.stream_data)
-		return reader_unread_byte(b)
-	},
-	impl_read_rune = proc(s: io.Stream) -> (r: rune, size: int, err: io.Error) {
-		b := (^Reader)(s.stream_data)
-		return reader_read_rune(b)
-	},
-	impl_unread_rune = proc(s: io.Stream) -> io.Error {
-		b := (^Reader)(s.stream_data)
-		return reader_unread_rune(b)
-	},
-	impl_write_to = proc(s: io.Stream, w: io.Writer) -> (n: i64, err: io.Error) {
-		b := (^Reader)(s.stream_data)
-		return reader_write_to(b, w)
-	},
+		return
+	case .Query:
+		return io.query_utility({.Read, .Destroy, .Query})
+	}
+	return 0, .Empty
 }

-
-
 //
 // Utility procedures
 //
@@ -435,9 +387,7 @@ reader_read_slice :: proc(b: ^Reader, delim: byte) -> (line: []byte, err: io.Err

 		s = b.w - b.r

-		if err = _reader_read_new_chunk(b); err != nil {
-			break
-		}
+		_reader_read_new_chunk(b) or_break
 	}

 	if i := len(line)-1; i >= 0 {
@@ -66,7 +66,7 @@ scanner_destroy :: proc(s: ^Scanner) {
 }


-// Returns the first non-EOF error that was encounted by the scanner
+// Returns the first non-EOF error that was encountered by the scanner
 scanner_error :: proc(s: ^Scanner) -> Scanner_Error {
 	switch s._err {
 	case .EOF, nil:
@@ -159,7 +159,7 @@ writer_write_rune :: proc(b: ^Writer, r: rune) -> (size: int, err: io.Error) {
 	return
 }

-// writer_write writes a string into the buffer
+// writer_write_string writes a string into the buffer
 // It returns the number of bytes written
 // If n < len(p), it will return an error explaining why the write is short
 writer_write_string :: proc(b: ^Writer, s: string) -> (int, io.Error) {
@@ -173,14 +173,6 @@ writer_read_from :: proc(b: ^Writer, r: io.Reader) -> (n: i64, err: io.Error) {
 	if b.err != nil {
 		return 0, b.err
 	}
-	if writer_buffered(b) == 0 {
-		if w, ok := io.to_reader_from(b.wr); !ok {
-			n, err = io.read_from(w, r)
-			b.err = err
-			return
-		}
-	}
-
 	for {
 		if writer_available(b) == 0 {
 			writer_flush(b) or_return
@@ -222,38 +214,35 @@ writer_read_from :: proc(b: ^Writer, r: io.Reader) -> (n: i64, err: io.Error) {

 // writer_to_stream converts a Writer into an io.Stream
 writer_to_stream :: proc(b: ^Writer) -> (s: io.Stream) {
-	s.stream_data = b
-	s.stream_vtable = _writer_vtable
+	s.data = b
+	s.procedure = _writer_proc
 	return
 }

+// writer_to_stream converts a Writer into an io.Stream
+writer_to_writer :: proc(b: ^Writer) -> (s: io.Writer) {
+	return writer_to_stream(b)
+}
+


@(private)
-_writer_vtable := &io.Stream_VTable{
-	impl_destroy = proc(s: io.Stream) -> io.Error {
-		b := (^Writer)(s.stream_data)
+_writer_proc :: proc(stream_data: rawptr, mode: io.Stream_Mode, p: []byte, offset: i64, whence: io.Seek_From) -> (n: i64, err: io.Error) {
+	b := (^Writer)(stream_data)
+	#partial switch mode {
+	case .Flush:
+		err = writer_flush(b)
+		return
+	case .Write:
+		n_int: int
+		n_int, err = writer_write(b, p)
+		n = i64(n_int)
+		return
+	case .Destroy:
 		writer_destroy(b)
-		return nil
-	},
-	impl_flush = proc(s: io.Stream)  -> io.Error {
-		b := (^Writer)(s.stream_data)
-		return writer_flush(b)
-	},
-	impl_write = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Writer)(s.stream_data)
-		return writer_write(b, p)
-	},
-	impl_write_byte = proc(s: io.Stream, c: byte) -> io.Error {
-		b := (^Writer)(s.stream_data)
-		return writer_write_byte(b, c)
-	},
-	impl_write_rune = proc(s: io.Stream, r: rune) -> (int, io.Error) {
-		b := (^Writer)(s.stream_data)
-		return writer_write_rune(b, r)
-	},
-	impl_read_from = proc(s: io.Stream, r: io.Reader) -> (n: i64, err: io.Error) {
-		b := (^Writer)(s.stream_data)
-		return writer_read_from(b, r)
-	},
+		return
+	case .Query:
+		return io.query_utility({.Flush, .Write, .Destroy, .Query})
+	}
+	return 0, .Empty
 }
@@ -94,7 +94,15 @@ cap :: proc(array: Array_Type) -> int ---

 size_of      :: proc($T: typeid) -> int ---
 align_of     :: proc($T: typeid) -> int ---
-offset_of    :: proc($T: typeid) -> uintptr ---
+
+// e.g. offset_of(t.f), where t is an instance of the type T
+offset_of_selector :: proc(selector: $T) -> uintptr ---
+// e.g. offset_of(T, f), where T can be the type instead of a variable
+offset_of_member   :: proc($T: typeid, member: $M) -> uintptr ---
+offset_of :: proc{offset_of_selector, offset_of_member}
+// e.g. offset_of(T, "f"), where T can be the type instead of a variable
+offset_of_by_string :: proc($T: typeid, member: string) -> uintptr ---
+
 type_of      :: proc(x: expr) -> type ---
 type_info_of :: proc($T: typeid) -> ^runtime.Type_Info ---
 typeid_of    :: proc($T: typeid) -> typeid ---
@@ -109,7 +117,7 @@ jmag       :: proc(value: Quaternion) -> Float ---
 kmag       :: proc(value: Quaternion) -> Float ---
 conj       :: proc(value: Complex_Or_Quaternion) -> Complex_Or_Quaternion ---

-expand_to_tuple :: proc(value: Struct_Or_Array) -> (A, B, C, ...) ---
+expand_values :: proc(value: Struct_Or_Array) -> (A, B, C, ...) ---

 min   :: proc(values: ..T) -> T ---
 max   :: proc(values: ..T) -> T ---
@@ -38,6 +38,11 @@ buffer_init_string :: proc(b: ^Buffer, s: string) {
 }

 buffer_init_allocator :: proc(b: ^Buffer, len, cap: int, allocator := context.allocator) {
+	if b.buf == nil {
+		b.buf = make([dynamic]byte, len, cap, allocator)
+		return
+	}
+
 	b.buf.allocator = allocator
 	reserve(&b.buf, cap)
 	resize(&b.buf, len)
@@ -108,8 +113,11 @@ _buffer_grow :: proc(b: ^Buffer, n: int) -> int {
 	if i, ok := _buffer_try_grow(b, n); ok {
 		return i
 	}
+
 	if b.buf == nil && n <= SMALL_BUFFER_SIZE {
-		b.buf = make([dynamic]byte, n, SMALL_BUFFER_SIZE)
+		// Fixes #2756 by preserving allocator if already set on Buffer via init_buffer_allocator
+		reserve(&b.buf, SMALL_BUFFER_SIZE)
+		resize(&b.buf, n)
 		return 0
 	}

@@ -240,14 +248,11 @@ buffer_read_ptr :: proc(b: ^Buffer, ptr: rawptr, size: int) -> (n: int, err: io.
 buffer_read_at :: proc(b: ^Buffer, p: []byte, offset: int) -> (n: int, err: io.Error) {
 	b.last_read = .Invalid

-	if offset < 0 || offset >= len(b.buf) {
+	if uint(offset) >= len(b.buf) {
 		err = .Invalid_Offset
 		return
 	}
-
-	if 0 <= offset && offset < len(b.buf) {
-		n = copy(p, b.buf[offset:])
-	}
+	n = copy(p, b.buf[offset:])
 	if n > 0 {
 		b.last_read = .Read
 	}
@@ -373,69 +378,31 @@ buffer_read_from :: proc(b: ^Buffer, r: io.Reader) -> (n: i64, err: io.Error) #n


 buffer_to_stream :: proc(b: ^Buffer) -> (s: io.Stream) {
-	s.stream_data = b
-	s.stream_vtable = _buffer_vtable
+	s.data = b
+	s.procedure = _buffer_proc
 	return
 }

@(private)
-_buffer_vtable := &io.Stream_VTable{
-	impl_size = proc(s: io.Stream) -> i64 {
-		b := (^Buffer)(s.stream_data)
-		return i64(buffer_capacity(b))
-	},
-	impl_read = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_read(b, p)
-	},
-	impl_read_at = proc(s: io.Stream, p: []byte, offset: i64) -> (n: int, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_read_at(b, p, int(offset))
-	},
-	impl_read_byte = proc(s: io.Stream) -> (byte, io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_read_byte(b)
-	},
-	impl_read_rune = proc(s: io.Stream) -> (r: rune, size: int, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_read_rune(b)
-	},
-	impl_write = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_write(b, p)
-	},
-	impl_write_at = proc(s: io.Stream, p: []byte, offset: i64) -> (n: int, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_write_at(b, p, int(offset))
-	},
-	impl_write_byte = proc(s: io.Stream, c: byte) -> io.Error {
-		b := (^Buffer)(s.stream_data)
-		return buffer_write_byte(b, c)
-	},
-	impl_write_rune = proc(s: io.Stream, r: rune) -> (int, io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_write_rune(b, r)
-	},
-	impl_unread_byte = proc(s: io.Stream) -> io.Error {
-		b := (^Buffer)(s.stream_data)
-		return buffer_unread_byte(b)
-	},
-	impl_unread_rune = proc(s: io.Stream) -> io.Error {
-		b := (^Buffer)(s.stream_data)
-		return buffer_unread_rune(b)
-	},
-	impl_destroy = proc(s: io.Stream) -> io.Error {
-		b := (^Buffer)(s.stream_data)
+_buffer_proc :: proc(stream_data: rawptr, mode: io.Stream_Mode, p: []byte, offset: i64, whence: io.Seek_From) -> (n: i64, err: io.Error) {
+	b := (^Buffer)(stream_data)
+	#partial switch mode {
+	case .Read:
+		return io._i64_err(buffer_read(b, p))
+	case .Read_At:
+		return io._i64_err(buffer_read_at(b, p, int(offset)))
+	case .Write:
+		return io._i64_err(buffer_write(b, p))
+	case .Write_At:
+		return io._i64_err(buffer_write_at(b, p, int(offset)))
+	case .Size:
+		n = i64(buffer_capacity(b))
+		return
+	case .Destroy:
 		buffer_destroy(b)
-		return nil
-	},
-	impl_write_to = proc(s: io.Stream, w: io.Writer) -> (n: i64, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_write_to(b, w)
-	},
-	impl_read_from = proc(s: io.Stream, r: io.Reader) -> (n: i64, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_read_from(b, r)
-	},
+		return
+	case .Query:
+		return io.query_utility({.Read, .Read_At, .Write, .Write_At, .Size, .Destroy})
+	}
+	return 0, .Empty
 }
-
@@ -638,7 +638,7 @@ trim_left_proc :: proc(s: []byte, p: proc(rune) -> bool) -> []byte {

 index_rune :: proc(s: []byte, r: rune) -> int {
 	switch {
-	case 0 <= r && r < utf8.RUNE_SELF:
+	case u32(r) < utf8.RUNE_SELF:
 		return index_byte(s, byte(r))

 	case r == utf8.RUNE_ERROR:
@@ -16,8 +16,8 @@ reader_init :: proc(r: ^Reader, s: []byte) {
 }

 reader_to_stream :: proc(r: ^Reader) -> (s: io.Stream) {
-	s.stream_data = r
-	s.stream_vtable = _reader_vtable
+	s.data = r
+	s.procedure = _reader_proc
 	return
 }

@@ -137,41 +137,22 @@ reader_write_to :: proc(r: ^Reader, w: io.Writer) -> (n: i64, err: io.Error) {


@(private)
-_reader_vtable := &io.Stream_VTable{
-	impl_size = proc(s: io.Stream) -> i64 {
-		r := (^Reader)(s.stream_data)
-		return reader_size(r)
-	},
-	impl_read = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_read(r, p)
-	},
-	impl_read_at = proc(s: io.Stream, p: []byte, off: i64) -> (n: int, err: io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_read_at(r, p, off)
-	},
-	impl_read_byte = proc(s: io.Stream) -> (byte, io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_read_byte(r)
-	},
-	impl_unread_byte = proc(s: io.Stream) -> io.Error {
-		r := (^Reader)(s.stream_data)
-		return reader_unread_byte(r)
-	},
-	impl_read_rune = proc(s: io.Stream) -> (ch: rune, size: int, err: io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_read_rune(r)
-	},
-	impl_unread_rune = proc(s: io.Stream) -> io.Error {
-		r := (^Reader)(s.stream_data)
-		return reader_unread_rune(r)
-	},
-	impl_seek = proc(s: io.Stream, offset: i64, whence: io.Seek_From) -> (i64, io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_seek(r, offset, whence)
-	},
-	impl_write_to = proc(s: io.Stream, w: io.Writer) -> (n: i64, err: io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_write_to(r, w)
-	},
+_reader_proc :: proc(stream_data: rawptr, mode: io.Stream_Mode, p: []byte, offset: i64, whence: io.Seek_From) -> (n: i64, err: io.Error) {
+	r := (^Reader)(stream_data)
+	#partial switch mode {
+	case .Read:
+		return io._i64_err(reader_read(r, p))
+	case .Read_At:
+		return io._i64_err(reader_read_at(r, p, offset))
+	case .Seek:
+		n, err = reader_seek(r, offset, whence)
+		return
+	case .Size:
+		n = reader_size(r)
+		return
+	case .Query:
+		return io.query_utility({.Read, .Read_At, .Seek, .Size, .Query})
+	}
+	return 0, .Empty
 }
+
@@ -1118,7 +1118,7 @@ expand_macro :: proc(cpp: ^Preprocessor, rest: ^^Token, tok: ^Token) -> bool {

 search_include_next :: proc(cpp: ^Preprocessor, filename: string) -> (path: string, ok: bool) {
 	for ; cpp.include_next_index < len(cpp.include_paths); cpp.include_next_index += 1 {
-		tpath := filepath.join(elems={cpp.include_paths[cpp.include_next_index], filename}, allocator=context.temp_allocator)
+		tpath := filepath.join({cpp.include_paths[cpp.include_next_index], filename}, allocator=context.temp_allocator)
 		if os.exists(tpath) {
 			return strings.clone(tpath), true
 		}
@@ -1136,7 +1136,7 @@ search_include_paths :: proc(cpp: ^Preprocessor, filename: string) -> (path: str
 	}

 	for include_path in cpp.include_paths {
-		tpath := filepath.join(elems={include_path, filename}, allocator=context.temp_allocator)
+		tpath := filepath.join({include_path, filename}, allocator=context.temp_allocator)
 		if os.exists(tpath) {
 			path, ok = strings.clone(tpath), true
 			cpp.filepath_cache[filename] = path
@@ -127,7 +127,7 @@ advance_rune :: proc(t: ^Tokenizer) {
 }

 advance_rune_n :: proc(t: ^Tokenizer, n: int) {
-	for in 0..<n {
+	for _ in 0..<n {
 		advance_rune(t)
 	}
 }
@@ -49,8 +49,8 @@ foreign libc {
 	// 7.3.8 Power and absolute-value functions
 	cabs    :: proc(z: complex_double) -> complex_double ---
 	cabsf   :: proc(z: complex_float) -> complex_float ---
-	cpow    :: proc(z: complex_double) -> complex_double ---
-	cpowf   :: proc(z: complex_float) -> complex_float ---
+	cpow    :: proc(x, y: complex_double) -> complex_double ---
+	cpowf   :: proc(x, y: complex_float) -> complex_float ---
 	csqrt   :: proc(z: complex_double) -> complex_double ---
 	csqrtf  :: proc(z: complex_float) -> complex_float ---

@@ -14,11 +14,24 @@ when ODIN_OS == .Windows {
 //	EDOM,
 //	EILSEQ
 //	ERANGE
-when ODIN_OS == .Linux || ODIN_OS == .FreeBSD {
+when ODIN_OS == .Linux {
 	@(private="file")
 	@(default_calling_convention="c")
 	foreign libc {
-		@(link_name="__libc_errno_location")
+		@(link_name="__errno_location")
+		_get_errno :: proc() -> ^int ---
+	}
+
+	EDOM   :: 33
+	EILSEQ :: 84
+	ERANGE :: 34
+}
+
+when ODIN_OS == .FreeBSD {
+	@(private="file")
+	@(default_calling_convention="c")
+	foreign libc {
+		@(link_name="__error")
 		_get_errno :: proc() -> ^int ---
 	}

@@ -331,7 +331,7 @@ fmin       :: proc{libc_fmin, libc_fminf}
 fma        :: proc{libc_fma, libc_fmaf}

 // But retain the 'f' suffix-variant functions as well so they can be used,
-// a trick is used here where we use explicit procedrual overloading of one
+// a trick is used here where we use explicit procedural overloading of one
 // procedure. This is done because the foreign block is marked @(private) and
 // aliasing functions does not remove privateness from the entity.
 acosf      :: proc{libc_acosf}
@@ -63,4 +63,4 @@ foreign libc {
 // strictly conformant C implementation is 16 on the platforms we care about.
 // The choice of 4096 bytes for storage of this type is more than enough on all
 // relevant platforms.
-jmp_buf :: struct #align 16 { _: [4096]char, }
+jmp_buf :: struct #align(16) { _: [4096]char, }
@@ -18,7 +18,7 @@ foreign _ {
 // strictly conformant C implementation is 16 on the platforms we care about.
 // The choice of 4096 bytes for storage of this type is more than enough on all
 // relevant platforms.
-va_list :: struct #align 16 {
+va_list :: struct #align(16) {
 	_: [4096]u8,
 }

@@ -1,7 +1,10 @@
 package libc

 when ODIN_OS == .Windows {
-	foreign import libc "system:libucrt.lib"
+	foreign import libc {
+		"system:libucrt.lib",
+		"system:legacy_stdio_definitions.lib",
+	}
 } else when ODIN_OS == .Darwin {
 	foreign import libc "system:System.framework"
 } else {
@@ -179,7 +182,7 @@ foreign libc {
 	fscanf    :: proc(stream: ^FILE, format: cstring, #c_vararg args: ..any) -> int ---
 	printf    :: proc(format: cstring, #c_vararg args: ..any) -> int ---
 	scanf     :: proc(format: cstring, #c_vararg args: ..any) -> int ---
-	snprintf  :: proc(s: [^]char, format: cstring, #c_vararg args: ..any) -> int ---
+	snprintf  :: proc(s: [^]char, n: size_t, format: cstring, #c_vararg args: ..any) -> int ---
 	sscanf    :: proc(s, format: cstring, #c_vararg args: ..any) -> int ---
 	vfprintf  :: proc(stream: ^FILE, format: cstring, arg: ^va_list) -> int ---
 	vfscanf   :: proc(stream: ^FILE, format: cstring, arg: ^va_list) -> int ---
@@ -88,7 +88,6 @@ foreign libc {
 	srand         :: proc(seed: uint) ---

 	// 7.22.3 Memory management functions
-	aligned_alloc :: proc(aligment, size: size_t) -> rawptr ---
 	calloc        :: proc(nmemb, size: size_t) -> rawptr ---
 	free          :: proc(ptr: rawptr) ---
 	malloc        :: proc(size: size_t) -> rawptr ---
@@ -125,3 +124,30 @@ foreign libc {
 	mbstowcs      :: proc(pwcs: ^wchar_t, s: cstring, n: size_t) -> size_t ---
 	wcstombs      :: proc(s: [^]char, pwcs: ^wchar_t, n: size_t) -> size_t ---
 }
+
+
+aligned_alloc :: #force_inline proc "c" (alignment, size: size_t) -> rawptr {
+	when ODIN_OS == .Windows {
+		foreign libc {
+			_aligned_malloc :: proc(size, alignment: size_t) -> rawptr ---
+		}
+		return _aligned_malloc(size=size, alignment=alignment)
+	} else {
+		foreign libc {
+			aligned_alloc :: proc(alignment, size: size_t) -> rawptr ---
+		}
+		return aligned_alloc(alignment=alignment, size=size)
+	}
+}
+
+
+aligned_free :: #force_inline proc "c" (ptr: rawptr) {
+	when ODIN_OS == .Windows {
+		foreign libc {
+			_aligned_free :: proc(ptr: rawptr) ---
+		}
+		_aligned_free(ptr)
+	} else {
+		free(ptr)
+	}
+}
@@ -44,7 +44,7 @@ when ODIN_OS == .Windows {
 		@(link_name="_Cnd_destroy")   cnd_destroy   :: proc(cond: ^cnd_t) ---
 		@(link_name="_Cnd_init")      cnd_init      :: proc(cond: ^cnd_t) -> int ---
 		@(link_name="_Cnd_signal")    cnd_signal    :: proc(cond: ^cnd_t) -> int ---
-		@(link_name="_Cnd_timedwait") cnd_timedwait :: proc(cond: ^cnd_t, ts: ^timespec) -> int ---
+		@(link_name="_Cnd_timedwait") cnd_timedwait :: proc(cond: ^cnd_t, mtx: ^mtx_t, ts: ^timespec) -> int ---
 		@(link_name="_Cnd_wait")      cnd_wait      :: proc(cond: ^cnd_t, mtx: ^mtx_t) -> int ---
 		
 		// 7.26.4 Mutex functions
@@ -108,7 +108,7 @@ when ODIN_OS == .Linux {
 		cnd_destroy   :: proc(cond: ^cnd_t) ---
 		cnd_init      :: proc(cond: ^cnd_t) -> int ---
 		cnd_signal    :: proc(cond: ^cnd_t) -> int ---
-		cnd_timedwait :: proc(cond: ^cnd_t, ts: ^timespec) -> int ---
+		cnd_timedwait :: proc(cond: ^cnd_t, mtx: ^mtx_t, ts: ^timespec) -> int ---
 		cnd_wait      :: proc(cond: ^cnd_t, mtx: ^mtx_t) -> int ---
 		
 		// 7.26.4 Mutex functions
@@ -35,12 +35,12 @@ when ODIN_OS == .Windows {
 	clock_t        :: distinct long
 	time_t         :: distinct i64

-	timespec :: struct #align 8 {
+	timespec :: struct #align(8) {
 		tv_sec:  time_t,
 		tv_nsec: long,
 	}

-	tm :: struct #align 8 {
+	tm :: struct #align(8) {
 		tm_sec, tm_min, tm_hour, tm_mday, tm_mon, tm_year, tm_wday, tm_yday, tm_isdst: int,
 	}
 }
@@ -2,6 +2,8 @@ package libc

 import "core:c"

+#assert(!ODIN_NO_CRT, `"core:c/libc" cannot be imported when '-no-crt' is used`)
+
 char           :: c.char // assuming -funsigned-char

 schar          :: c.schar
@@ -98,7 +98,7 @@ foreign libc {
 }

 // Large enough and aligned enough for any wide-spread in-use libc.
-mbstate_t :: struct #align 16 { _: [32]char, }
+mbstate_t :: struct #align(16) { _: [32]char, }

 // Odin does not have default argument promotion so the need for a separate type
 // here isn't necessary, though make it distinct just to be safe.
@@ -188,7 +188,8 @@ input_size_from_memory :: proc(z: ^Context_Memory_Input) -> (res: i64, err: Erro
 }

 input_size_from_stream :: proc(z: ^Context_Stream_Input) -> (res: i64, err: Error) {
-	return io.size(z.input), nil
+	res, _ = io.size(z.input)
+	return
 }

 input_size :: proc{input_size_from_memory, input_size_from_stream}
@@ -212,25 +213,19 @@ read_slice_from_memory :: #force_inline proc(z: ^Context_Memory_Input, size: int

@(optimization_mode="speed")
 read_slice_from_stream :: #force_inline proc(z: ^Context_Stream_Input, size: int) -> (res: []u8, err: io.Error) {
+	// TODO: REMOVE ALL USE OF context.temp_allocator here
+	// the is literally no need for it
 	b := make([]u8, size, context.temp_allocator)
-	_, e := z.input->impl_read(b[:])
-	if e == .None {
-		return b, .None
-	}
-
-	return []u8{}, e
+	_ = io.read(z.input, b[:]) or_return
+	return b, nil
 }

 read_slice :: proc{read_slice_from_memory, read_slice_from_stream}

@(optimization_mode="speed")
 read_data :: #force_inline proc(z: ^$C, $T: typeid) -> (res: T, err: io.Error) {
-	b, e := read_slice(z, size_of(T))
-	if e == .None {
-		return (^T)(&b[0])^, .None
-	}
-
-	return T{}, e
+	b := read_slice(z, size_of(T)) or_return
+	return (^T)(&b[0])^, nil
 }

@(optimization_mode="speed")
@@ -247,12 +242,8 @@ read_u8_from_memory :: #force_inline proc(z: ^Context_Memory_Input) -> (res: u8,

@(optimization_mode="speed")
 read_u8_from_stream :: #force_inline proc(z: ^Context_Stream_Input) -> (res: u8, err: io.Error) {
-	b, e := read_slice_from_stream(z, 1)
-	if e == .None {
-		return b[0], .None
-	}
-
-	return 0, e
+	b := read_slice_from_stream(z, 1) or_return
+	return b[0], nil
 }

 read_u8 :: proc{read_u8_from_memory, read_u8_from_stream}
@@ -294,17 +285,32 @@ peek_data_from_memory :: #force_inline proc(z: ^Context_Memory_Input, $T: typeid
 	}
 }

+@(optimization_mode="speed")
+peek_data_at_offset_from_memory :: #force_inline proc(z: ^Context_Memory_Input, $T: typeid, #any_int offset: int) -> (res: T, err: io.Error) {
+	size :: size_of(T)
+
+	#no_bounds_check {
+		if len(z.input_data) >= size + offset {
+			buf := z.input_data[offset:][:size]
+			return (^T)(&buf[0])^, .None
+		}
+	}
+
+	if len(z.input_data) == 0 {
+		return T{}, .EOF
+	} else {
+		return T{}, .Short_Buffer
+	}
+}
+
@(optimization_mode="speed")
 peek_data_from_stream :: #force_inline proc(z: ^Context_Stream_Input, $T: typeid) -> (res: T, err: io.Error) {
 	size :: size_of(T)

 	// Get current position to read from.
-	curr, e1 := z.input->impl_seek(0, .Current)
-	if e1 != .None {
-		return T{}, e1
-	}
-	r, e2 := io.to_reader_at(z.input)
-	if !e2 {
+	curr := z.input->impl_seek(0, .Current) or_return
+	r, e1 := io.to_reader_at(z.input)
+	if !e1 {
 		return T{}, .Empty
 	}
 	when size <= 128 {
@@ -312,8 +318,8 @@ peek_data_from_stream :: #force_inline proc(z: ^Context_Stream_Input, $T: typeid
 	} else {
 		b := make([]u8, size, context.temp_allocator)
 	}
-	_, e3 := io.read_at(r, b[:], curr)
-	if e3 != .None {
+	_, e2 := io.read_at(r, b[:], curr)
+	if e2 != .None {
 		return T{}, .Empty
 	}

@@ -321,7 +327,37 @@ peek_data_from_stream :: #force_inline proc(z: ^Context_Stream_Input, $T: typeid
 	return res, .None
 }

-peek_data :: proc{peek_data_from_memory, peek_data_from_stream}
+@(optimization_mode="speed")
+peek_data_at_offset_from_stream :: #force_inline proc(z: ^Context_Stream_Input, $T: typeid, #any_int offset: int) -> (res: T, err: io.Error) {
+	size :: size_of(T)
+
+	// Get current position to return to.
+	cur_pos := z.input->impl_seek(0, .Current) or_return
+	// Seek to offset.
+	pos := z.input->impl_seek(offset, .Start) or_return
+
+	r, e3 := io.to_reader_at(z.input)
+	if !e3 {
+		return T{}, .Empty
+	}
+	when size <= 128 {
+		b: [size]u8
+	} else {
+		b := make([]u8, size, context.temp_allocator)
+	}
+	_, e4 := io.read_at(r, b[:], pos)
+	if e4 != .None {
+		return T{}, .Empty
+	}
+
+	// Return read head to original position.
+	z.input->impl_seek(cur_pos, .Start)
+
+	res = (^T)(&b[0])^
+	return res, .None
+}
+
+peek_data :: proc{peek_data_from_memory, peek_data_from_stream, peek_data_at_offset_from_memory, peek_data_at_offset_from_stream}



@@ -407,7 +443,7 @@ peek_bits_lsb_from_memory :: #force_inline proc(z: ^Context_Memory_Input, width:
 	if z.num_bits < u64(width) {
 		refill_lsb(z)
 	}
-	return u32(z.code_buffer & ~(~u64(0) << width))
+	return u32(z.code_buffer &~ (~u64(0) << width))
 }

@(optimization_mode="speed")
@@ -415,7 +451,7 @@ peek_bits_lsb_from_stream :: #force_inline proc(z: ^Context_Stream_Input, width:
 	if z.num_bits < u64(width) {
 		refill_lsb(z)
 	}
-	return u32(z.code_buffer & ~(~u64(0) << width))
+	return u32(z.code_buffer &~ (~u64(0) << width))
 }

 peek_bits_lsb :: proc{peek_bits_lsb_from_memory, peek_bits_lsb_from_stream}
@@ -423,13 +459,13 @@ peek_bits_lsb :: proc{peek_bits_lsb_from_memory, peek_bits_lsb_from_stream}
@(optimization_mode="speed")
 peek_bits_no_refill_lsb_from_memory :: #force_inline proc(z: ^Context_Memory_Input, width: u8) -> u32 {
 	assert(z.num_bits >= u64(width))
-	return u32(z.code_buffer & ~(~u64(0) << width))
+	return u32(z.code_buffer &~ (~u64(0) << width))
 }

@(optimization_mode="speed")
 peek_bits_no_refill_lsb_from_stream :: #force_inline proc(z: ^Context_Stream_Input, width: u8) -> u32 {
 	assert(z.num_bits >= u64(width))
-	return u32(z.code_buffer & ~(~u64(0) << width))
+	return u32(z.code_buffer &~ (~u64(0) << width))
 }

 peek_bits_no_refill_lsb :: proc{peek_bits_no_refill_lsb_from_memory, peek_bits_no_refill_lsb_from_stream}
@@ -1,4 +1,4 @@
-//+ignore
+//+build ignore
 package gzip

 /*
@@ -335,10 +335,8 @@ load_from_context :: proc(z: ^$C, buf: ^bytes.Buffer, known_gzip_size := -1, exp

 	// fmt.printf("GZIP: Expected Payload Size: %v\n", expected_output_size);

-	zlib_error := zlib.inflate_raw(z=z, expected_output_size=expected_output_size)
-	if zlib_error != nil {
-		return zlib_error
-	}
+	zlib.inflate_raw(z, expected_output_size=expected_output_size) or_return
+
 	/*
 		Read CRC32 using the ctx bit reader because zlib may leave bytes in there.
 	*/
@@ -177,12 +177,10 @@ decompress_slice_to_string :: proc(input: []u8, model := DEFAULT_MODEL, allocato
 	max_output_size := decompress_bound(len(input), model)

 	buf: [dynamic]u8
-	if !resize(&buf, max_output_size) {
-		return "", .Out_Of_Memory
-	}
+	resize(&buf, max_output_size) or_return

 	length, result := decompress_slice_to_output_buffer(input, buf[:])
-	resize(&buf, length)
+	resize(&buf, length) or_return
 	return string(buf[:]), result
 }
 decompress :: proc{decompress_slice_to_output_buffer, decompress_slice_to_string}
@@ -307,12 +305,10 @@ compress_string :: proc(input: string, model := DEFAULT_MODEL, allocator := cont
 	max_output_size := compress_bound(len(input))

 	buf: [dynamic]u8
-	if !resize(&buf, max_output_size) {
-		return {}, .Out_Of_Memory
-	}
+	resize(&buf, max_output_size) or_return

 	length, result := compress_string_to_buffer(input, buf[:])
-	resize(&buf, length)
+	resize(&buf, length) or_return
 	return buf[:length], result
 }
 compress :: proc{compress_string_to_buffer, compress_string}
@@ -1,4 +1,4 @@
-//+ignore
+//+build ignore
 package zlib

 /*
@@ -1,3 +1,4 @@
+//+vet !using-param
 package zlib

 /*
@@ -146,15 +147,7 @@ grow_buffer :: proc(buf: ^[dynamic]u8) -> (err: compress.Error) {
 		Double until we reach the maximum allowed.
 	*/
 	new_size := min(len(buf) << 1, compress.COMPRESS_OUTPUT_ALLOCATE_MAX)
-	resize(buf, new_size)
-	if len(buf) != new_size {
-		/*
-			Resize failed.
-		*/
-		return .Resize_Failed
-	}
-
-	return nil
+	return resize(buf, new_size)
 }

 /*
@@ -181,7 +174,7 @@ write_byte :: #force_inline proc(z: ^$C, c: u8) -> (err: io.Error) #no_bounds_ch
 }

@(optimization_mode="speed")
-repl_byte :: proc(z: ^$C, count: u16, c: u8) -> (err: io.Error) 	#no_bounds_check {
+repl_byte :: proc(z: ^$C, count: u16, c: u8) -> (err: io.Error) #no_bounds_check {
 	/*
 		TODO(Jeroen): Once we have a magic ring buffer, we can just peek/write into it
 		without having to worry about wrapping, so no need for a temp allocation to give to
@@ -305,10 +298,10 @@ decode_huffman_slowpath :: proc(z: ^$C, t: ^Huffman_Table) -> (r: u16, err: Erro
 	code := u16(compress.peek_bits_lsb(z,16))

 	k := int(z_bit_reverse(code, 16))
-	s: u8

-	#no_bounds_check for s = HUFFMAN_FAST_BITS+1; ; {
-		if k < t.maxcode[s] {
+	s: u8 = HUFFMAN_FAST_BITS+1
+	for {
+		#no_bounds_check if k < t.maxcode[s] {
 			break
 		}
 		s += 1
@@ -471,7 +464,7 @@ inflate_from_context :: proc(using ctx: ^compress.Context_Memory_Input, raw := f
 	}

 	// Parse ZLIB stream without header.
-	inflate_raw(z=ctx, expected_output_size=expected_output_size) or_return
+	inflate_raw(ctx, expected_output_size=expected_output_size) or_return

 	if !raw {
 		compress.discard_to_next_byte_lsb(ctx)
@@ -509,8 +502,8 @@ inflate_raw :: proc(z: ^$C, expected_output_size := -1, allocator := context.all
 		/*
 			Try to pre-allocate the output buffer.
 		*/
-		reserve(&z.output.buf, expected_output_size)
-		resize (&z.output.buf, expected_output_size)
+		reserve(&z.output.buf, expected_output_size) or_return
+		resize (&z.output.buf, expected_output_size) or_return
 	}

 	if len(z.output.buf) != expected_output_size {
@@ -653,7 +646,7 @@ inflate_raw :: proc(z: ^$C, expected_output_size := -1, allocator := context.all
 	}

 	if int(z.bytes_written) != len(z.output.buf) {
-		resize(&z.output.buf, int(z.bytes_written))
+		resize(&z.output.buf, int(z.bytes_written)) or_return
 	}

 	return nil
@@ -665,7 +658,7 @@ inflate_from_byte_array :: proc(input: []u8, buf: ^bytes.Buffer, raw := false, e
 	ctx.input_data = input
 	ctx.output = buf

-	return inflate_from_context(ctx=&ctx, raw=raw, expected_output_size=expected_output_size)
+	return inflate_from_context(&ctx, raw=raw, expected_output_size=expected_output_size)
 }

 inflate_from_byte_array_raw :: proc(input: []u8, buf: ^bytes.Buffer, raw := false, expected_output_size := -1) -> (err: Error) {
@@ -674,7 +667,7 @@ inflate_from_byte_array_raw :: proc(input: []u8, buf: ^bytes.Buffer, raw := fals
 	ctx.input_data = input
 	ctx.output = buf

-	return inflate_raw(z=&ctx, expected_output_size=expected_output_size)
+	return inflate_raw(&ctx, expected_output_size=expected_output_size)
 }

 inflate :: proc{inflate_from_context, inflate_from_byte_array}
@@ -27,27 +27,28 @@ Bit_Array_Iterator :: struct {
 	word_idx: int,
 	bit_idx:  uint,
 }
-
 /*
-	In:
-		- ba:   ^Bit_Array - the array to iterate over
+Wraps a `Bit_Array` into an Iterator

-	Out:
-		- it:   ^Bit_Array_Iterator - the iterator that holds iteration state
+Inputs:
+- ba: Pointer to the Bit_Array
+
+Returns:
+- it: Iterator struct
 */
 make_iterator :: proc (ba: ^Bit_Array) -> (it: Bit_Array_Iterator) {
 	return Bit_Array_Iterator { array = ba }
 }
-
 /*
-	In:
-		- it:    ^Bit_Array_Iterator - the iterator struct that holds the state.
+Returns the next bit, including its set-state. ok=false once exhausted

-	Out:
-		- set:    bool - the state of the bit at `index`
-		- index:  int - the next bit of the Bit_Array referenced by `it`.
-		- ok:	  bool - `true` if the iterator returned a valid index,
-			  `false` if there were no more bits
+Inputs:
+- it: The iterator that holds the state.
+
+Returns:
+- set: `true` if the bit at `index` is set.
+- index: The next bit of the Bit_Array referenced by `it`.
+- ok: `true` if the iterator can continue, `false` if the iterator is done
 */
 iterate_by_all :: proc (it: ^Bit_Array_Iterator) -> (set: bool, index: int, ok: bool) {
 	index = it.word_idx * NUM_BITS + int(it.bit_idx) + it.array.bias
@@ -64,39 +65,51 @@ iterate_by_all :: proc (it: ^Bit_Array_Iterator) -> (set: bool, index: int, ok:

 	return set, index, true
 }
-
 /*
-	In:
-		- it:     ^Bit_Array_Iterator - the iterator struct that holds the state.
+Returns the next Set Bit, for example if `0b1010`, then the iterator will return index={1, 3} over two calls.

-	Out:
-		- index:  int - the next set bit of the Bit_Array referenced by `it`.
-		- ok:	  bool - `true` if the iterator returned a valid index,
-			  `false` if there were no more bits set
+Inputs:
+- it: The iterator that holds the state.
+
+Returns:
+- index: The next *set* bit of the Bit_Array referenced by `it`.
+- ok: `true` if the iterator can continue, `false` if the iterator is done
 */
 iterate_by_set :: proc (it: ^Bit_Array_Iterator) -> (index: int, ok: bool) {
 	return iterate_internal_(it, true)
 }
-
 /*
-	In:
-		- it:	  ^Bit_Array_Iterator - the iterator struct that holds the state.
+Returns the next Unset Bit, for example if `0b1010`, then the iterator will return index={0, 2} over two calls.

-	Out:
-		- index:  int - the next unset bit of the Bit_Array referenced by `it`.
-		- ok:	  bool - `true` if the iterator returned a valid index,
-			  `false` if there were no more unset bits
+Inputs:
+- it: The iterator that holds the state.
+
+Returns:
+- index: The next *unset* bit of the Bit_Array referenced by `it`.
+- ok: `true` if the iterator can continue, `false` if the iterator is done
 */
 iterate_by_unset:: proc (it: ^Bit_Array_Iterator) -> (index: int, ok: bool) {
 	return iterate_internal_(it, false)
 }
+/*
+Iterates through set/unset bits

+*Private*
+
+Inputs:
+- it: The iterator that holds the state.
+- ITERATE_SET_BITS: `true` for returning only set bits, false for returning only unset bits
+
+Returns:
+- index: The next *unset* bit of the Bit_Array referenced by `it`.
+- ok: `true` if the iterator can continue, `false` if the iterator is done
+*/
@(private="file")
 iterate_internal_ :: proc (it: ^Bit_Array_Iterator, $ITERATE_SET_BITS: bool) -> (index: int, ok: bool) {
 	word := it.array.bits[it.word_idx] if len(it.array.bits) > it.word_idx else 0
 	when ! ITERATE_SET_BITS { word = ~word }

-	// if the word is empty or we have already gone over all the bits in it,
+	// If the word is empty or we have already gone over all the bits in it,
 	// b.bit_idx is greater than the index of any set bit in the word,
 	// meaning that word >> b.bit_idx == 0.
 	for it.word_idx < len(it.array.bits) && word >> it.bit_idx == 0 {
@@ -106,14 +119,14 @@ iterate_internal_ :: proc (it: ^Bit_Array_Iterator, $ITERATE_SET_BITS: bool) ->
 		when ! ITERATE_SET_BITS { word = ~word }
 	}

-	// if we are iterating the set bits, reaching the end of the array means we have no more bits to check
+	// If we are iterating the set bits, reaching the end of the array means we have no more bits to check
 	when ITERATE_SET_BITS {
 		if it.word_idx >= len(it.array.bits) {
 			return 0, false
 		}
 	}

-	// reaching here means that the word has some set bits
+	// Reaching here means that the word has some set bits
 	it.bit_idx += uint(intrinsics.count_trailing_zeros(word >> it.bit_idx))
 	index = it.word_idx * NUM_BITS + int(it.bit_idx) + it.array.bias

@@ -124,24 +137,21 @@ iterate_internal_ :: proc (it: ^Bit_Array_Iterator, $ITERATE_SET_BITS: bool) ->
 	}
 	return index, index <= it.array.max_index
 }
-
-
 /*
-	In:
-		- ba:    ^Bit_Array - a pointer to the Bit Array
-		- index: The bit index. Can be an enum member.
+Gets the state of a bit in the bit-array

-	Out:
-		- res:   The bit you're interested in.
-		- ok:    Whether the index was valid. Returns `false` if the index is smaller than the bias.
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array

-	The `ok` return value may be ignored.
+Returns:
+- res: `true` if the bit at `index` is set.
+- ok: Whether the index was valid. Returns `false` if the index is smaller than the bias.
 */
-get :: proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator) -> (res: bool, ok: bool) {
+get :: proc(ba: ^Bit_Array, #any_int index: uint) -> (res: bool, ok: bool) #optional_ok {
 	idx := int(index) - ba.bias

 	if ba == nil || int(index) < ba.bias { return false, false }
-	context.allocator = allocator

 	leg_index := idx >> INDEX_SHIFT
 	bit_index := idx &  INDEX_MASK
@@ -157,18 +167,36 @@ get :: proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator

 	return res, true
 }
-
 /*
-	In:
-		- ba:    ^Bit_Array - a pointer to the Bit Array
-		- index: The bit index. Can be an enum member.
+Gets the state of a bit in the bit-array

-	Out:
-		- ok:    Whether or not we managed to set requested bit.
+*Bypasses all Checks*

-	`set` automatically resizes the Bit Array to accommodate the requested index if needed.
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array
+
+Returns:
+- `true` if bit is set
 */
-set :: proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator) -> (ok: bool) {
+unsafe_get :: #force_inline proc(ba: ^Bit_Array, #any_int index: uint) -> bool #no_bounds_check {
+	return bool((ba.bits[index >> INDEX_SHIFT] >> uint(index & INDEX_MASK)) & 1)
+}
+/*
+Sets the state of a bit in the bit-array
+
+*Conditionally Allocates (Resizes backing data when `index > len(ba.bits)`)*
+
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array
+- set_to: `true` sets the bit on, `false` to turn it off
+- allocator: (default is context.allocator)
+
+Returns:
+- ok: Whether the set was successful, `false` on allocation failure or bad index
+*/
+set :: proc(ba: ^Bit_Array, #any_int index: uint, set_to: bool = true, allocator := context.allocator) -> (ok: bool) {

 	idx := int(index) - ba.bias

@@ -181,65 +209,97 @@ set :: proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator
 	resize_if_needed(ba, leg_index) or_return

 	ba.max_index = max(idx, ba.max_index)
-	ba.bits[leg_index] |= 1 << uint(bit_index)
+
+	if set_to{ ba.bits[leg_index] |= 1 << uint(bit_index) }
+	else { ba.bits[leg_index] &= ~(1 << uint(bit_index)) }
+
 	return true
 }
-
 /*
-	In:
-		- ba:    ^Bit_Array - a pointer to the Bit Array
-		- index: The bit index. Can be an enum member.
+Sets the state of a bit in the bit-array

-	Out:
-		- ok:    Whether or not we managed to unset requested bit.
+*Bypasses all checks*

-	`unset` automatically resizes the Bit Array to accommodate the requested index if needed.
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array
 */
-unset :: proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator) -> (ok: bool) {
-
-	idx := int(index) - ba.bias
-
-	if ba == nil || int(index) < ba.bias { return false }
-	context.allocator = allocator
-
-	leg_index := idx >> INDEX_SHIFT
-	bit_index := idx &  INDEX_MASK
-
-	resize_if_needed(ba, leg_index) or_return
-
-	ba.max_index = max(idx, ba.max_index)
-	ba.bits[leg_index] &= ~(1 << uint(bit_index))
-	return true
+unsafe_set :: proc(ba: ^Bit_Array, bit: int) #no_bounds_check {
+	ba.bits[bit >> INDEX_SHIFT] |= 1 << uint(bit & INDEX_MASK)
 }
-
 /*
-	A helper function to create a Bit Array with optional bias, in case your smallest index is non-zero (including negative).
+Unsets the state of a bit in the bit-array. (Convienence wrapper for `set`)
+
+*Conditionally Allocates (Resizes backing data when `index > len(ba.bits)`)*
+
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array
+- allocator: (default is context.allocator)
+
+Returns:
+- ok: Whether the unset was successful, `false` on allocation failure or bad index
 */
-create :: proc(max_index: int, min_index := 0, allocator := context.allocator) -> (res: ^Bit_Array, ok: bool) #optional_ok {
+unset :: #force_inline proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator) -> (ok: bool) {
+	return set(ba, index, false, allocator)
+}
+/*
+Unsets the state of a bit in the bit-array
+
+*Bypasses all Checks*
+
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array
+*/
+unsafe_unset :: proc(b: ^Bit_Array, bit: int) #no_bounds_check {
+	b.bits[bit >> INDEX_SHIFT] &= ~(1 << uint(bit & INDEX_MASK))
+}
+/*
+A helper function to create a Bit Array with optional bias, in case your smallest index is non-zero (including negative).
+
+*Allocates (`new(Bit_Array) & make(ba.bits)`)*
+
+Inputs:
+- max_index: maximum starting index
+- min_index: minimum starting index (used as a bias)
+- allocator: (default is context.allocator)
+
+Returns:
+- ba: Allocates a bit_Array, backing data is set to `max-min / 64` indices, rounded up (eg 65 - 0 allocates for [2]u64).
+*/
+create :: proc(max_index: int, min_index: int = 0, allocator := context.allocator) -> (res: ^Bit_Array, ok: bool) #optional_ok {
 	context.allocator = allocator
 	size_in_bits := max_index - min_index

 	if size_in_bits < 1 { return {}, false }

 	legs := size_in_bits >> INDEX_SHIFT
-
+	if size_in_bits & INDEX_MASK > 0 {legs+=1}
+	bits, err := make([dynamic]u64, legs)
+	ok = err == mem.Allocator_Error.None
 	res = new(Bit_Array)
+	res.bits         = bits
 	res.bias         = min_index
 	res.max_index    = max_index
 	res.free_pointer = true
-	return res, resize_if_needed(res, legs)
+	return
 }
-
 /*
-	Sets all bits to `false`.
+Sets all values in the Bit_Array to zero.
+
+Inputs:
+- ba: The target Bit_Array
 */
 clear :: proc(ba: ^Bit_Array) {
 	if ba == nil { return }
 	mem.zero_slice(ba.bits[:])
 }
-
 /*
-	Releases the memory used by the Bit Array.
+Deallocates the Bit_Array and its backing storage
+
+Inputs:
+- ba: The target Bit_Array
 */
 destroy :: proc(ba: ^Bit_Array) {
 	if ba == nil { return }
@@ -248,9 +308,8 @@ destroy :: proc(ba: ^Bit_Array) {
 		free(ba)
 	}
 }
-
 /*
-	Resizes the Bit Array. For internal use.
+	Resizes the Bit Array. For internal use. Provisions needed capacity+1
 	If you want to reserve the memory for a given-sized Bit Array up front, you can use `create`.
 */
@(private="file")
@@ -20,10 +20,10 @@ List :: struct {


 Node :: struct {
-	next, prev: ^Node,
+	prev, next: ^Node,
 }

-push_front :: proc(list: ^List, node: ^Node) {
+push_front :: proc "contextless" (list: ^List, node: ^Node) {
 	if list.head != nil {
 		list.head.prev = node
 		node.prev, node.next = nil, list.head
@@ -34,7 +34,7 @@ push_front :: proc(list: ^List, node: ^Node) {
 	}
 }

-push_back :: proc(list: ^List, node: ^Node) {
+push_back :: proc "contextless" (list: ^List, node: ^Node) {
 	if list.tail != nil {
 		list.tail.next = node
 		node.prev, node.next = list.tail, nil
@@ -45,7 +45,7 @@ push_back :: proc(list: ^List, node: ^Node) {
 	}
 }

-remove :: proc(list: ^List, node: ^Node) {
+remove :: proc "contextless" (list: ^List, node: ^Node) {
 	if node != nil {
 		if node.next != nil {
 			node.next.prev = node.prev
@@ -83,12 +83,34 @@ remove_by_proc :: proc(list: ^List, to_erase: proc(^Node) -> bool) {
 	}
 }

+remove_by_proc_contextless :: proc(list: ^List, to_erase: proc "contextless" (^Node) -> bool) {
+	for node := list.head; node != nil; {
+		next := node.next
+		if to_erase(node) {
+			if node.next != nil {
+				node.next.prev = node.prev
+			}
+			if node.prev != nil {
+				node.prev.next = node.next
+			}
+			if list.head == node {
+				list.head = node.next
+			}
+			if list.tail == node {
+				list.tail = node.prev
+			}
+		}
+		node = next
+	}
+}

-is_empty :: proc(list: ^List) -> bool {
+
+
+is_empty :: proc "contextless" (list: ^List) -> bool {
 	return list.head == nil
 }

-pop_front :: proc(list: ^List) -> ^Node {
+pop_front :: proc "contextless" (list: ^List) -> ^Node {
 	link := list.head
 	if link == nil {
 		return nil
@@ -108,7 +130,7 @@ pop_front :: proc(list: ^List) -> ^Node {
 	return link

 }
-pop_back :: proc(list: ^List) -> ^Node {
+pop_back :: proc "contextless" (list: ^List) -> ^Node {
 	link := list.tail
 	if link == nil {
 		return nil
@@ -134,25 +156,25 @@ Iterator :: struct($T: typeid) {
 	offset: uintptr,
 }

-iterator_head :: proc(list: List, $T: typeid, $field_name: string) -> Iterator(T)
+iterator_head :: proc "contextless" (list: List, $T: typeid, $field_name: string) -> Iterator(T)
 	where intrinsics.type_has_field(T, field_name),
 	      intrinsics.type_field_type(T, field_name) == Node {
 	return {list.head, offset_of_by_string(T, field_name)}
 }

-iterator_tail :: proc(list: List, $T: typeid, $field_name: string) -> Iterator(T)
+iterator_tail :: proc "contextless" (list: List, $T: typeid, $field_name: string) -> Iterator(T)
 	where intrinsics.type_has_field(T, field_name),
 	      intrinsics.type_field_type(T, field_name) == Node {
 	return {list.tail, offset_of_by_string(T, field_name)}
 }

-iterator_from_node :: proc(node: ^Node, $T: typeid, $field_name: string) -> Iterator(T)
+iterator_from_node :: proc "contextless" (node: ^Node, $T: typeid, $field_name: string) -> Iterator(T)
 	where intrinsics.type_has_field(T, field_name),
 	      intrinsics.type_field_type(T, field_name) == Node {
 	return {node, offset_of_by_string(T, field_name)}
 }

-iterate_next :: proc(it: ^Iterator($T)) -> (ptr: ^T, ok: bool) {
+iterate_next :: proc "contextless" (it: ^Iterator($T)) -> (ptr: ^T, ok: bool) {
 	node := it.curr
 	if node == nil {
 		return nil, false
@@ -162,7 +184,7 @@ iterate_next :: proc(it: ^Iterator($T)) -> (ptr: ^T, ok: bool) {
 	return (^T)(uintptr(node) - it.offset), true
 }

-iterate_prev :: proc(it: ^Iterator($T)) -> (ptr: ^T, ok: bool) {
+iterate_prev :: proc "contextless" (it: ^Iterator($T)) -> (ptr: ^T, ok: bool) {
 	node := it.curr
 	if node == nil {
 		return nil, false
@@ -85,7 +85,6 @@ _shift_down :: proc(pq: ^$Q/Priority_Queue($T), i0, n: int) -> bool {
 _shift_up :: proc(pq: ^$Q/Priority_Queue($T), j: int) {
 	j := j
 	queue := pq.queue[:]
-	n := builtin.len(queue)
 	for 0 <= j {
 		i := (j-1)/2
 		if i == j || !pq.less(queue[j], queue[i]) {
@@ -14,7 +14,7 @@ Queue :: struct($T: typeid) {
 DEFAULT_CAPACITY :: 16

 // Procedure to initialize a queue
-init :: proc(q: ^$Q/Queue($T), capacity := DEFAULT_CAPACITY, allocator := context.allocator) -> bool {
+init :: proc(q: ^$Q/Queue($T), capacity := DEFAULT_CAPACITY, allocator := context.allocator) -> runtime.Allocator_Error {
 	if q.data.allocator.procedure == nil {
 		q.data.allocator = allocator
 	}
@@ -22,7 +22,9 @@ init :: proc(q: ^$Q/Queue($T), capacity := DEFAULT_CAPACITY, allocator := contex
 	return reserve(q, capacity)
 }

-// Procedure to initialize a queue from a fixed backing slice
+// Procedure to initialize a queue from a fixed backing slice.
+// The contents of the `backing` will be overwritten as items are pushed onto the `Queue`.
+// Any previous contents are not available.
 init_from_slice :: proc(q: ^$Q/Queue($T), backing: []T) -> bool {
 	clear(q)
 	q.data = transmute([dynamic]T)runtime.Raw_Dynamic_Array{
@@ -34,6 +36,21 @@ init_from_slice :: proc(q: ^$Q/Queue($T), backing: []T) -> bool {
 	return true
 }

+// Procedure to initialize a queue from a fixed backing slice.
+// Existing contents are preserved and available on the queue.
+init_with_contents :: proc(q: ^$Q/Queue($T), backing: []T) -> bool {
+	clear(q)
+	q.data = transmute([dynamic]T)runtime.Raw_Dynamic_Array{
+		data = raw_data(backing),
+		len = builtin.len(backing),
+		cap = builtin.len(backing),
+		allocator = {procedure=runtime.nil_allocator_proc, data=nil},
+	}
+	q.len    = len(backing)
+	q.offset = len(backing)
+	return true
+}
+
 // Procedure to destroy a queue
 destroy :: proc(q: ^$Q/Queue($T)) {
 	delete(q.data)
@@ -55,11 +72,11 @@ space :: proc(q: $Q/Queue($T)) -> int {
 }

 // Reserve enough space for at least the specified capacity
-reserve :: proc(q: ^$Q/Queue($T), capacity: int) -> bool {
-	if uint(capacity) > q.len {
+reserve :: proc(q: ^$Q/Queue($T), capacity: int) -> runtime.Allocator_Error {
+	if capacity > space(q^) {
 		return _grow(q, uint(capacity)) 
 	}
-	return true
+	return nil
 }


@@ -73,11 +90,18 @@ get :: proc(q: ^$Q/Queue($T), #any_int i: int, loc := #caller_location) -> T {
 front :: proc(q: ^$Q/Queue($T)) -> T {
 	return q.data[q.offset]
 }
+front_ptr :: proc(q: ^$Q/Queue($T)) -> ^T {
+	return &q.data[q.offset]
+}

 back :: proc(q: ^$Q/Queue($T)) -> T {
 	idx := (q.offset+uint(q.len))%builtin.len(q.data)
 	return q.data[idx]
 }
+back_ptr :: proc(q: ^$Q/Queue($T)) -> ^T {
+	idx := (q.offset+uint(q.len))%builtin.len(q.data)
+	return &q.data[idx]
+}

 set :: proc(q: ^$Q/Queue($T), #any_int i: int, val: T, loc := #caller_location) {
 	runtime.bounds_check_error_loc(loc, i, builtin.len(q.data))
@@ -92,26 +116,38 @@ get_ptr :: proc(q: ^$Q/Queue($T), #any_int i: int, loc := #caller_location) -> ^
 	return &q.data[idx]
 }

+peek_front :: proc(q: ^$Q/Queue($T), loc := #caller_location) -> ^T {
+	runtime.bounds_check_error_loc(loc, 0, builtin.len(q.data))
+	idx := q.offset%builtin.len(q.data)
+	return &q.data[idx]
+}
+
+peek_back :: proc(q: ^$Q/Queue($T), loc := #caller_location) -> ^T {
+	runtime.bounds_check_error_loc(loc, int(q.len - 1), builtin.len(q.data))
+	idx := (uint(q.len - 1)+q.offset)%builtin.len(q.data)
+	return &q.data[idx]
+}
+
 // Push an element to the back of the queue
-push_back :: proc(q: ^$Q/Queue($T), elem: T) -> bool {
+push_back :: proc(q: ^$Q/Queue($T), elem: T) -> (ok: bool, err: runtime.Allocator_Error) {
 	if space(q^) == 0 {
 		_grow(q) or_return
 	}
 	idx := (q.offset+uint(q.len))%builtin.len(q.data)
 	q.data[idx] = elem
 	q.len += 1
-	return true
+	return true, nil
 }

 // Push an element to the front of the queue
-push_front :: proc(q: ^$Q/Queue($T), elem: T) -> bool {
+push_front :: proc(q: ^$Q/Queue($T), elem: T) -> (ok: bool, err: runtime.Allocator_Error)  {
 	if space(q^) == 0 {
 		_grow(q) or_return
 	}	
 	q.offset = uint(q.offset - 1 + builtin.len(q.data)) % builtin.len(q.data)
 	q.len += 1
 	q.data[q.offset] = elem
-	return true
+	return true, nil
 }


@@ -154,7 +190,7 @@ pop_front_safe :: proc(q: ^$Q/Queue($T)) -> (elem: T, ok: bool) {
 }

 // Push multiple elements to the front of the queue
-push_back_elems :: proc(q: ^$Q/Queue($T), elems: ..T) -> bool {
+push_back_elems :: proc(q: ^$Q/Queue($T), elems: ..T) -> (ok: bool, err: runtime.Allocator_Error)  {
 	n := uint(builtin.len(elems))
 	if space(q^) < int(n) {
 		_grow(q, q.len + n) or_return
@@ -169,7 +205,7 @@ push_back_elems :: proc(q: ^$Q/Queue($T), elems: ..T) -> bool {
 	copy(q.data[insert_from:], elems[:insert_to])
 	copy(q.data[:insert_from], elems[insert_to:])
 	q.len += n
-	return true
+	return true, nil
 }

 // Consume `n` elements from the front of the queue
@@ -206,7 +242,7 @@ clear :: proc(q: ^$Q/Queue($T)) {


 // Internal growinh procedure
-_grow :: proc(q: ^$Q/Queue($T), min_capacity: uint = 0) -> bool {
+_grow :: proc(q: ^$Q/Queue($T), min_capacity: uint = 0) -> runtime.Allocator_Error {
 	new_capacity := max(min_capacity, uint(8), uint(builtin.len(q.data))*2)
 	n := uint(builtin.len(q.data))
 	builtin.resize(&q.data, int(new_capacity)) or_return
@@ -215,5 +251,5 @@ _grow :: proc(q: ^$Q/Queue($T), min_capacity: uint = 0) -> bool {
 		copy(q.data[new_capacity-diff:], q.data[q.offset:][:diff])
 		q.offset += new_capacity - n
 	}
-	return true
+	return nil
 }
@@ -1,6 +1,8 @@
 package container_small_array

 import "core:builtin"
+import "core:runtime"
+_ :: runtime

 Small_Array :: struct($N: int, $T: typeid) where N >= 0 {
 	data: [N]T,
@@ -8,40 +10,54 @@ Small_Array :: struct($N: int, $T: typeid) where N >= 0 {
 }


-len :: proc(a: $A/Small_Array) -> int {
+len :: proc "contextless" (a: $A/Small_Array) -> int {
 	return a.len
 }

-cap :: proc(a: $A/Small_Array) -> int {
+cap :: proc "contextless" (a: $A/Small_Array) -> int {
 	return builtin.len(a.data)
 }

-space :: proc(a: $A/Small_Array) -> int {
+space :: proc "contextless" (a: $A/Small_Array) -> int {
 	return builtin.len(a.data) - a.len
 }

-slice :: proc(a: ^$A/Small_Array($N, $T)) -> []T {
+slice :: proc "contextless" (a: ^$A/Small_Array($N, $T)) -> []T {
 	return a.data[:a.len]
 }


-get :: proc(a: $A/Small_Array($N, $T), index: int) -> T {
+get :: proc "contextless" (a: $A/Small_Array($N, $T), index: int) -> T {
 	return a.data[index]
 }
-get_ptr :: proc(a: ^$A/Small_Array($N, $T), index: int) -> ^T {
+get_ptr :: proc "contextless" (a: ^$A/Small_Array($N, $T), index: int) -> ^T {
 	return &a.data[index]
 }

-set :: proc(a: ^$A/Small_Array($N, $T), index: int, item: T) {
+get_safe :: proc(a: $A/Small_Array($N, $T), index: int) -> (T, bool) #no_bounds_check {
+	if index < 0 || index >= a.len {
+		return {}, false
+	}
+	return a.data[index], true
+}
+
+get_ptr_safe :: proc(a: ^$A/Small_Array($N, $T), index: int) -> (^T, bool) #no_bounds_check {
+	if index < 0 || index >= a.len {
+		return {}, false
+	}
+	return &a.data[index], true
+}
+
+set :: proc "contextless" (a: ^$A/Small_Array($N, $T), index: int, item: T) {
 	a.data[index] = item
 }

-resize :: proc(a: ^$A/Small_Array, length: int) {
+resize :: proc "contextless" (a: ^$A/Small_Array, length: int) {
 	a.len = min(length, builtin.len(a.data))
 }


-push_back :: proc(a: ^$A/Small_Array($N, $T), item: T) -> bool {
+push_back :: proc "contextless" (a: ^$A/Small_Array($N, $T), item: T) -> bool {
 	if a.len < cap(a^) {
 		a.data[a.len] = item
 		a.len += 1
@@ -50,7 +66,7 @@ push_back :: proc(a: ^$A/Small_Array($N, $T), item: T) -> bool {
 	return false
 }

-push_front :: proc(a: ^$A/Small_Array($N, $T), item: T) -> bool {
+push_front :: proc "contextless" (a: ^$A/Small_Array($N, $T), item: T) -> bool {
 	if a.len < cap(a^) {
 		a.len += 1
 		data := slice(a)
@@ -61,14 +77,14 @@ push_front :: proc(a: ^$A/Small_Array($N, $T), item: T) -> bool {
 	return false
 }

-pop_back :: proc(a: ^$A/Small_Array($N, $T), loc := #caller_location) -> T {
+pop_back :: proc "odin" (a: ^$A/Small_Array($N, $T), loc := #caller_location) -> T {
 	assert(condition=(N > 0 && a.len > 0), loc=loc)
 	item := a.data[a.len-1]
 	a.len -= 1
 	return item
 }

-pop_front :: proc(a: ^$A/Small_Array($N, $T), loc := #caller_location) -> T {
+pop_front :: proc "odin" (a: ^$A/Small_Array($N, $T), loc := #caller_location) -> T {
 	assert(condition=(N > 0 && a.len > 0), loc=loc)
 	item := a.data[0]
 	s := slice(a)
@@ -77,7 +93,7 @@ pop_front :: proc(a: ^$A/Small_Array($N, $T), loc := #caller_location) -> T {
 	return item
 }

-pop_back_safe :: proc(a: ^$A/Small_Array($N, $T)) -> (item: T, ok: bool) {
+pop_back_safe :: proc "contextless" (a: ^$A/Small_Array($N, $T)) -> (item: T, ok: bool) {
 	if N > 0 && a.len > 0 {
 		item = a.data[a.len-1]
 		a.len -= 1
@@ -86,31 +102,60 @@ pop_back_safe :: proc(a: ^$A/Small_Array($N, $T)) -> (item: T, ok: bool) {
 	return
 }

-pop_front_safe :: proc(a: ^$A/Small_Array($N, $T)) -> (item: T, ok: bool) {
+pop_front_safe :: proc "contextless" (a: ^$A/Small_Array($N, $T)) -> (item: T, ok: bool) {
 	if N > 0 && a.len > 0 {
 		item = a.data[0]
 		s := slice(a)
 		copy(s[:], s[1:])
 		a.len -= 1
 		ok = true
-	} 
+	}
 	return
 }

-consume :: proc(a: ^$A/Small_Array($N, $T), count: int, loc := #caller_location) {
+consume :: proc "odin" (a: ^$A/Small_Array($N, $T), count: int, loc := #caller_location) {
 	assert(condition=a.len >= count, loc=loc)
 	a.len -= count
 }

-clear :: proc(a: ^$A/Small_Array($N, $T)) {
+ordered_remove :: proc "contextless" (a: ^$A/Small_Array($N, $T), index: int, loc := #caller_location) #no_bounds_check {
+    runtime.bounds_check_error_loc(loc, index, a.len)
+    if index+1 < a.len {
+		copy(a.data[index:], a.data[index+1:])
+	}
+	a.len -= 1
+}
+
+unordered_remove :: proc "contextless" (a: ^$A/Small_Array($N, $T), index: int, loc := #caller_location) #no_bounds_check {
+    runtime.bounds_check_error_loc(loc, index, a.len)
+	n := a.len-1
+	if index != n {
+		a.data[index] = a.data[n]
+	}
+    a.len -= 1
+}
+
+clear :: proc "contextless" (a: ^$A/Small_Array($N, $T)) {
 	resize(a, 0)
 }

-push_back_elems :: proc(a: ^$A/Small_Array($N, $T), items: ..T) {
+push_back_elems :: proc "contextless" (a: ^$A/Small_Array($N, $T), items: ..T) {
 	n := copy(a.data[a.len:], items[:])
 	a.len += n
 }

+inject_at :: proc "contextless" (a: ^$A/Small_Array($N, $T), item: T, index: int) -> bool #no_bounds_check {
+	if a.len < cap(a^) && index >= 0 && index <= len(a^) {
+		a.len += 1
+		for i := a.len - 1; i >= index + 1; i -= 1 {
+			a.data[i] = a.data[i - 1]
+		}
+		a.data[index] = item
+		return true
+	}
+	return false
+}
+
 append_elem  :: push_back
 append_elems :: push_back_elems
 push   :: proc{push_back, push_back_elems}
@@ -32,7 +32,7 @@ init :: proc(sorter: ^$S/Sorter($K)) {
 }

 destroy :: proc(sorter: ^$S/Sorter($K)) {
-	for _, v in &sorter.relations {
+	for _, v in sorter.relations {
 		delete(v.dependents)
 	}
 	delete(sorter.relations)
@@ -80,11 +80,13 @@ sort :: proc(sorter: ^$S/Sorter($K)) -> (sorted, cycled: [dynamic]K) {
 		}
 	}

-	for root in &sorted do for k, _ in relations[root].dependents {
-		relation := &relations[k]
-		relation.dependencies -= 1
-		if relation.dependencies == 0 {
-			append(&sorted, k)
+	for root in sorted {
+		for k, _ in relations[root].dependents {
+			relation := &relations[k]
+			relation.dependencies -= 1
+			if relation.dependencies == 0 {
+				append(&sorted, k)
+			}
 		}
 	}

@@ -1,95 +1,86 @@
 # crypto
-A crypto library for the Odin language
+
+A cryptography library for the Odin language

 ## Supported
+
 This library offers various algorithms implemented in Odin.
-Please see the chart below for the options.  
+Please see the chart below for some of the options.

 ## Hashing algorithms
+
 | Algorithm                                                                                                    |                  |
 |:-------------------------------------------------------------------------------------------------------------|:-----------------|
-| [BLAKE](https://web.archive.org/web/20190915215948/https://131002.net/blake)                                 | &#10004;&#65039; |
 | [BLAKE2B](https://datatracker.ietf.org/doc/html/rfc7693)                                                     | &#10004;&#65039; |
 | [BLAKE2S](https://datatracker.ietf.org/doc/html/rfc7693)                                                     | &#10004;&#65039; |
-| [GOST](https://datatracker.ietf.org/doc/html/rfc5831)                                                        | &#10004;&#65039; |
-| [Grøstl](http://www.groestl.info/Groestl.zip)                                                                | &#10004;&#65039; |
-| [HAVAL](https://web.archive.org/web/20150111210116/http://labs.calyptix.com/haval.php)                       | &#10004;&#65039; |
-| [JH](https://www3.ntu.edu.sg/home/wuhj/research/jh/index.html)                                               | &#10004;&#65039; |
-| [Keccak](https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.202.pdf)                                           | &#10004;&#65039; |
-| [MD2](https://datatracker.ietf.org/doc/html/rfc1319)                                                         | &#10004;&#65039; |
-| [MD4](https://datatracker.ietf.org/doc/html/rfc1320)                                                         | &#10004;&#65039; |
-| [MD5](https://datatracker.ietf.org/doc/html/rfc1321)                                                         | &#10004;&#65039; |
-| [RIPEMD](https://homes.esat.kuleuven.be/~bosselae/ripemd160.html)                                            | &#10004;&#65039; |
-| [SHA-1](https://datatracker.ietf.org/doc/html/rfc3174)                                                       | &#10004;&#65039; |
 | [SHA-2](https://csrc.nist.gov/csrc/media/publications/fips/180/2/archive/2002-08-01/documents/fips180-2.pdf) | &#10004;&#65039; |
 | [SHA-3](https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.202.pdf)                                            | &#10004;&#65039; |
 | [SHAKE](https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.202.pdf)                                            | &#10004;&#65039; |
 | [SM3](https://datatracker.ietf.org/doc/html/draft-sca-cfrg-sm3-02)                                           | &#10004;&#65039; |
-| [Streebog](https://datatracker.ietf.org/doc/html/rfc6986)                                                    | &#10004;&#65039; |
-| [Tiger](https://www.cs.technion.ac.il/~biham/Reports/Tiger/)                                                 | &#10004;&#65039; |
-| [Tiger2](https://www.cs.technion.ac.il/~biham/Reports/Tiger/)                                                | &#10004;&#65039; |
-| [Whirlpool](https://web.archive.org/web/20171129084214/http://www.larc.usp.br/~pbarreto/WhirlpoolPage.html)  | &#10004;&#65039; |
+| legacy/[Keccak](https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.202.pdf)                                    | &#10004;&#65039; |
+| legacy/[MD5](https://datatracker.ietf.org/doc/html/rfc1321)                                                  | &#10004;&#65039; |
+| legacy/[SHA-1](https://datatracker.ietf.org/doc/html/rfc3174)                                                | &#10004;&#65039; |

 #### High level API
-Each hash algorithm contains a procedure group named `hash`, or if the algorithm provides more than one digest size `hash_<size>`\*.  
-Included in these groups are six procedures.
-* `hash_string` - Hash a given string and return the computed hash. Just calls `hash_bytes` internally
-* `hash_bytes` - Hash a given byte slice and return the computed hash
-* `hash_string_to_buffer` - Hash a given string and put the computed hash in the second proc parameter. Just calls `hash_bytes_to_buffer` internally
-* `hash_bytes_to_buffer` - Hash a given string and put the computed hash in the second proc parameter. The destination buffer has to be at least as big as the digest size of the hash
-* `hash_stream` - Takes a stream from io.Stream and returns the computed hash from it
-* `hash_file` - Takes a file handle and returns the computed hash from it. A second optional boolean parameter controls if the file is streamed (this is the default) or read at once (set to true)

-\* On some algorithms there is another part to the name, since they might offer control about additional parameters.  
-For instance, `HAVAL` offers different sizes as well as three different round amounts.  
-Computing a 256-bit hash with 3 rounds is therefore achieved by calling `haval.hash_256_3(...)`.
+Each hash algorithm contains a procedure group named `hash`, or if the algorithm provides more than one digest size `hash_<size>`\*.
+Included in these groups are six procedures.
+- `hash_string` - Hash a given string and return the computed hash. Just calls `hash_bytes` internally
+- `hash_bytes` - Hash a given byte slice and return the computed hash
+- `hash_string_to_buffer` - Hash a given string and put the computed hash in the second proc parameter. Just calls `hash_bytes_to_buffer` internally
+- `hash_bytes_to_buffer` - Hash a given string and put the computed hash in the second proc parameter. The destination buffer has to be at least as big as the digest size of the hash
+- `hash_stream` - Takes a stream from io.Stream and returns the computed hash from it
+- `hash_file` - Takes a file handle and returns the computed hash from it. A second optional boolean parameter controls if the file is streamed (this is the default) or read at once (set to true)
+
+\* On some algorithms there is another part to the name, since they might offer control about additional parameters.
+For instance, `SHA-2` offers different sizes.
+Computing a 512-bit hash is therefore achieved by calling `sha2.hash_512(...)`.

 #### Low level API
+
 The above mentioned procedures internally call three procedures: `init`, `update` and `final`.
 You may also directly call them, if you wish.

 #### Example
+
 ```odin
 package crypto_example

 // Import the desired package
-import "core:crypto/md4"
+import "core:crypto/blake2b"

 main :: proc() {
    input := "foo"

    // Compute the hash, using the high level API
-    computed_hash := md4.hash(input)
+    computed_hash := blake2b.hash(input)

    // Variant that takes a destination buffer, instead of returning the computed hash
-    hash := make([]byte, md4.DIGEST_SIZE) // @note: Destination buffer has to be at least as big as the digest size of the hash
-    md4.hash(input, hash[:])
+    hash := make([]byte, sha2.DIGEST_SIZE) // @note: Destination buffer has to be at least as big as the digest size of the hash
+    blake2b.hash(input, hash[:])

    // Compute the hash, using the low level API
-    ctx: md4.Md4_Context
-    computed_hash_low: [16]byte
-    md4.init(&ctx)
-    md4.update(&ctx, transmute([]byte)input)
-    md4.final(&ctx, computed_hash_low[:])
+    ctx: blake2b.Context
+    computed_hash_low: [blake2b.DIGEST_SIZE]byte
+    blake2b.init(&ctx)
+    blake2b.update(&ctx, transmute([]byte)input)
+    blake2b.final(&ctx, computed_hash_low[:])
 }
 ```
 For example uses of all available algorithms, please see the tests within `tests/core/crypto`.

-#### Thread safety
-The crypto package is not thread-safe at the moment. This may change in the future.
+## Implementation considerations

-### Disclaimer
-The algorithms were ported out of curiosity and due to interest in the field.
-We have not had any of the code verified by a third party or tested/fuzzed by any automatic means.
-Whereever we were able to find official test vectors, those were used to verify the implementation.
-We do not recommend using them in a production environment, without any additional testing and/or verification.
+- The crypto packages are not thread-safe.
+- Best-effort is make to mitigate timing side-channels on reasonable
+  architectures. Architectures that are known to be unreasonable include
+  but are not limited to i386, i486, and WebAssembly.
+- Some but not all of the packages attempt to santize sensitive data,
+  however this is not done consistently through the library at the moment.
+  As Thomas Pornin puts it "In general, such memory cleansing is a fool's
+  quest."
+- All of these packages have not received independent third party review.

-### ToDo
-* Ciphers (Symmetric, Asymmetric)
-* MACs (Message Authentication Code)
-* CSPRNGs (Cryptographically Secure PseudoRandom Number Generator)
-* KDFs (Key Derivation Function)
-* KEAs (Key Exchange Algorithm)
+## License

-### License
 This library is made available under the BSD-3 license.
@@ -10,12 +10,12 @@ package _blake2
    Implementation of the BLAKE2 hashing algorithm, as defined in <https://datatracker.ietf.org/doc/html/rfc7693> and <https://www.blake2.net/>
 */

-import "../util"
+import "core:encoding/endian"

-BLAKE2S_BLOCK_SIZE  :: 64
-BLAKE2S_SIZE        :: 32
-BLAKE2B_BLOCK_SIZE  :: 128
-BLAKE2B_SIZE        :: 64
+BLAKE2S_BLOCK_SIZE :: 64
+BLAKE2S_SIZE :: 32
+BLAKE2B_BLOCK_SIZE :: 128
+BLAKE2B_SIZE :: 64

 Blake2s_Context :: struct {
 	h:            [8]u32,
@@ -28,7 +28,9 @@ Blake2s_Context :: struct {
 	is_keyed:     bool,
 	size:         byte,
 	is_last_node: bool,
-	cfg:		  Blake2_Config,
+	cfg:          Blake2_Config,
+
+	is_initialized: bool,
 }

 Blake2b_Context :: struct {
@@ -42,15 +44,19 @@ Blake2b_Context :: struct {
 	is_keyed:     bool,
 	size:         byte,
 	is_last_node: bool,
-	cfg:		  Blake2_Config,
+	cfg:          Blake2_Config,
+
+	is_initialized: bool,
 }

 Blake2_Config :: struct {
-    size:   byte,
-	key:    []byte, 
-	salt:   []byte, 
+	size:   byte,
+	key:    []byte,
+	salt:   []byte,
 	person: []byte,
-	tree:   union{Blake2_Tree},
+	tree:   union {
+		Blake2_Tree,
+	},
 }

 Blake2_Tree :: struct {
@@ -63,11 +69,13 @@ Blake2_Tree :: struct {
 	is_last_node:    bool,
 }

+@(private)
 BLAKE2S_IV := [8]u32 {
 	0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a,
 	0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19,
 }

+@(private)
 BLAKE2B_IV := [8]u64 {
 	0x6a09e667f3bcc908, 0xbb67ae8584caa73b,
 	0x3c6ef372fe94f82b, 0xa54ff53a5f1d36f1,
@@ -78,8 +86,14 @@ BLAKE2B_IV := [8]u64 {
 init :: proc(ctx: ^$T) {
 	when T == Blake2s_Context {
 		block_size :: BLAKE2S_BLOCK_SIZE
+		max_size :: BLAKE2S_SIZE
 	} else when T == Blake2b_Context {
 		block_size :: BLAKE2B_BLOCK_SIZE
+		max_size :: BLAKE2B_SIZE
+	}
+
+	if ctx.cfg.size > max_size {
+		panic("blake2: requested output size exceeeds algorithm max")
 	}

 	p := make([]byte, block_size)
@@ -106,10 +120,10 @@ init :: proc(ctx: ^$T) {
 	if ctx.cfg.tree != nil {
 		p[2] = ctx.cfg.tree.(Blake2_Tree).fanout
 		p[3] = ctx.cfg.tree.(Blake2_Tree).max_depth
-		util.PUT_U32_LE(p[4:], ctx.cfg.tree.(Blake2_Tree).leaf_size)
+		endian.unchecked_put_u32le(p[4:], ctx.cfg.tree.(Blake2_Tree).leaf_size)
 		when T == Blake2s_Context {
-			p[8]  = byte(ctx.cfg.tree.(Blake2_Tree).node_offset)
-			p[9]  = byte(ctx.cfg.tree.(Blake2_Tree).node_offset >> 8)
+			p[8] = byte(ctx.cfg.tree.(Blake2_Tree).node_offset)
+			p[9] = byte(ctx.cfg.tree.(Blake2_Tree).node_offset >> 8)
 			p[10] = byte(ctx.cfg.tree.(Blake2_Tree).node_offset >> 16)
 			p[11] = byte(ctx.cfg.tree.(Blake2_Tree).node_offset >> 24)
 			p[12] = byte(ctx.cfg.tree.(Blake2_Tree).node_offset >> 32)
@@ -117,7 +131,7 @@ init :: proc(ctx: ^$T) {
 			p[14] = ctx.cfg.tree.(Blake2_Tree).node_depth
 			p[15] = ctx.cfg.tree.(Blake2_Tree).inner_hash_size
 		} else when T == Blake2b_Context {
-			util.PUT_U64_LE(p[8:], ctx.cfg.tree.(Blake2_Tree).node_offset)
+			endian.unchecked_put_u64le(p[8:], ctx.cfg.tree.(Blake2_Tree).node_offset)
 			p[16] = ctx.cfg.tree.(Blake2_Tree).node_depth
 			p[17] = ctx.cfg.tree.(Blake2_Tree).inner_hash_size
 		}
@@ -127,10 +141,10 @@ init :: proc(ctx: ^$T) {
 	ctx.size = ctx.cfg.size
 	for i := 0; i < 8; i += 1 {
 		when T == Blake2s_Context {
-			ctx.h[i] = BLAKE2S_IV[i] ~ util.U32_LE(p[i * 4:])
+			ctx.h[i] = BLAKE2S_IV[i] ~ endian.unchecked_get_u32le(p[i * 4:])
 		}
 		when T == Blake2b_Context {
-			ctx.h[i] = BLAKE2B_IV[i] ~ util.U64_LE(p[i * 8:])
+			ctx.h[i] = BLAKE2B_IV[i] ~ endian.unchecked_get_u64le(p[i * 8:])
 		}
 	}
 	if ctx.cfg.tree != nil && ctx.cfg.tree.(Blake2_Tree).is_last_node {
@@ -142,13 +156,19 @@ init :: proc(ctx: ^$T) {
 		ctx.is_keyed = true
 	}
 	copy(ctx.ih[:], ctx.h[:])
-	copy(ctx.h[:],  ctx.ih[:])
+	copy(ctx.h[:], ctx.ih[:])
 	if ctx.is_keyed {
 		update(ctx, ctx.padded_key[:])
 	}
+
+	ctx.nx = 0
+
+	ctx.is_initialized = true
 }

-update :: proc "contextless" (ctx: ^$T, p: []byte) {
+update :: proc(ctx: ^$T, p: []byte) {
+	assert(ctx.is_initialized)
+
 	p := p
 	when T == Blake2s_Context {
 		block_size :: BLAKE2S_BLOCK_SIZE
@@ -174,15 +194,25 @@ update :: proc "contextless" (ctx: ^$T, p: []byte) {
 	ctx.nx += copy(ctx.x[ctx.nx:], p)
 }

-final :: proc "contextless" (ctx: ^$T, hash: []byte) {
+final :: proc(ctx: ^$T, hash: []byte) {
+	assert(ctx.is_initialized)
+
 	when T == Blake2s_Context {
+		if len(hash) < int(ctx.cfg.size) {
+			panic("crypto/blake2s: invalid destination digest size")
+		}
 		blake2s_final(ctx, hash)
-	}
-	when T == Blake2b_Context {
+	} else when T == Blake2b_Context {
+		if len(hash) < int(ctx.cfg.size) {
+			panic("crypto/blake2b: invalid destination digest size")
+		}
 		blake2b_final(ctx, hash)
 	}
+
+	ctx.is_initialized = false
 }

+@(private)
 blake2s_final :: proc "contextless" (ctx: ^Blake2s_Context, hash: []byte) {
 	if ctx.is_keyed {
 		for i := 0; i < len(ctx.padded_key); i += 1 {
@@ -203,16 +233,14 @@ blake2s_final :: proc "contextless" (ctx: ^Blake2s_Context, hash: []byte) {

 	blocks(ctx, ctx.x[:])

-	j := 0
-	for s, _ in ctx.h[:(ctx.size - 1) / 4 + 1] {
-		hash[j + 0] = byte(s >> 0)
-		hash[j + 1] = byte(s >> 8)
-		hash[j + 2] = byte(s >> 16)
-		hash[j + 3] = byte(s >> 24)
-		j += 4
+	dst: [BLAKE2S_SIZE]byte
+	for i := 0; i < BLAKE2S_SIZE / 4; i += 1 {
+		endian.unchecked_put_u32le(dst[i * 4:], ctx.h[i])
 	}
+	copy(hash, dst[:])
 }

+@(private)
 blake2b_final :: proc "contextless" (ctx: ^Blake2b_Context, hash: []byte) {
 	if ctx.is_keyed {
 		for i := 0; i < len(ctx.padded_key); i += 1 {
@@ -229,56 +257,52 @@ blake2b_final :: proc "contextless" (ctx: ^Blake2b_Context, hash: []byte) {
 	ctx.f[0] = 0xffffffffffffffff
 	if ctx.is_last_node {
 		ctx.f[1] = 0xffffffffffffffff
-	} 
+	}

 	blocks(ctx, ctx.x[:])

-	j := 0
-	for s, _ in ctx.h[:(ctx.size - 1) / 8 + 1] {
-		hash[j + 0] = byte(s >> 0)
-		hash[j + 1] = byte(s >> 8)
-		hash[j + 2] = byte(s >> 16)
-		hash[j + 3] = byte(s >> 24)
-		hash[j + 4] = byte(s >> 32)
-		hash[j + 5] = byte(s >> 40)
-		hash[j + 6] = byte(s >> 48)
-		hash[j + 7] = byte(s >> 56)
-		j += 8
+	dst: [BLAKE2B_SIZE]byte
+	for i := 0; i < BLAKE2B_SIZE / 8; i += 1 {
+		endian.unchecked_put_u64le(dst[i * 8:], ctx.h[i])
 	}
+	copy(hash, dst[:])
 }

+@(private)
 blocks :: proc "contextless" (ctx: ^$T, p: []byte) {
 	when T == Blake2s_Context {
 		blake2s_blocks(ctx, p)
-	}
-	when T == Blake2b_Context {
+	} else when T == Blake2b_Context {
 		blake2b_blocks(ctx, p)
 	}
 }

+@(private)
 blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []byte) {
-	h0, h1, h2, h3, h4, h5, h6, h7 := ctx.h[0], ctx.h[1], ctx.h[2], ctx.h[3], ctx.h[4], ctx.h[5], ctx.h[6], ctx.h[7]
+	h0, h1, h2, h3, h4, h5, h6, h7 :=
+		ctx.h[0], ctx.h[1], ctx.h[2], ctx.h[3], ctx.h[4], ctx.h[5], ctx.h[6], ctx.h[7]
 	p := p
 	for len(p) >= BLAKE2S_BLOCK_SIZE {
 		ctx.t[0] += BLAKE2S_BLOCK_SIZE
 		if ctx.t[0] < BLAKE2S_BLOCK_SIZE {
 			ctx.t[1] += 1
-		} 
+		}
 		v0, v1, v2, v3, v4, v5, v6, v7 := h0, h1, h2, h3, h4, h5, h6, h7
-		v8  := BLAKE2S_IV[0]
-		v9  := BLAKE2S_IV[1]
+		v8 := BLAKE2S_IV[0]
+		v9 := BLAKE2S_IV[1]
 		v10 := BLAKE2S_IV[2]
 		v11 := BLAKE2S_IV[3]
 		v12 := BLAKE2S_IV[4] ~ ctx.t[0]
 		v13 := BLAKE2S_IV[5] ~ ctx.t[1]
 		v14 := BLAKE2S_IV[6] ~ ctx.f[0]
 		v15 := BLAKE2S_IV[7] ~ ctx.f[1]
-		m: [16]u32
-		j := 0
+
+		m: [16]u32 = ---
 		for i := 0; i < 16; i += 1 {
-			m[i] = u32(p[j]) | u32(p[j + 1]) << 8 | u32(p[j + 2]) << 16 | u32(p[j + 3]) << 24
-			j += 4
+			m[i] = endian.unchecked_get_u32le(p[i * 4:])
 		}
+
+		// Round 1
 		v0 += m[0]
 		v0 += v4
 		v12 ~= v0
@@ -391,6 +415,8 @@ blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (32 - 7) | v5 >> 7
+
+		// Round 2
 		v0 += m[14]
 		v0 += v4
 		v12 ~= v0
@@ -503,6 +529,8 @@ blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (32 - 7) | v5 >> 7
+
+		// Round 3
 		v0 += m[11]
 		v0 += v4
 		v12 ~= v0
@@ -615,6 +643,8 @@ blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (32 - 7) | v5 >> 7
+
+		// Round 4
 		v0 += m[7]
 		v0 += v4
 		v12 ~= v0
@@ -727,6 +757,8 @@ blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (32 - 7) | v5 >> 7
+
+		// Round 5
 		v0 += m[9]
 		v0 += v4
 		v12 ~= v0
@@ -839,6 +871,8 @@ blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (32 - 7) | v5 >> 7
+
+		// Round 6
 		v0 += m[2]
 		v0 += v4
 		v12 ~= v0
@@ -951,6 +985,8 @@ blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (32 - 7) | v5 >> 7
+
+		// Round 7
 		v0 += m[12]
 		v0 += v4
 		v12 ~= v0
@@ -1063,6 +1099,8 @@ blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (32 - 7) | v5 >> 7
+
+		// Round 8
 		v0 += m[13]
 		v0 += v4
 		v12 ~= v0
@@ -1175,6 +1213,8 @@ blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (32 - 7) | v5 >> 7
+
+		// Round 9
 		v0 += m[6]
 		v0 += v4
 		v12 ~= v0
@@ -1287,6 +1327,8 @@ blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (32 - 7) | v5 >> 7
+
+		// Round 10
 		v0 += m[10]
 		v0 += v4
 		v12 ~= v0
@@ -1399,6 +1441,7 @@ blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (32 - 7) | v5 >> 7
+
 		h0 ~= v0 ~ v8
 		h1 ~= v1 ~ v9
 		h2 ~= v2 ~ v10
@@ -1407,19 +1450,23 @@ blake2s_blocks :: #force_inline proc "contextless" (ctx: ^Blake2s_Context, p: []
 		h5 ~= v5 ~ v13
 		h6 ~= v6 ~ v14
 		h7 ~= v7 ~ v15
+
 		p = p[BLAKE2S_BLOCK_SIZE:]
 	}
-	ctx.h[0], ctx.h[1], ctx.h[2], ctx.h[3], ctx.h[4], ctx.h[5], ctx.h[6], ctx.h[7] = h0, h1, h2, h3, h4, h5, h6, h7
+	ctx.h[0], ctx.h[1], ctx.h[2], ctx.h[3], ctx.h[4], ctx.h[5], ctx.h[6], ctx.h[7] =
+		h0, h1, h2, h3, h4, h5, h6, h7
 }

+@(private)
 blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []byte) {
-	h0, h1, h2, h3, h4, h5, h6, h7 := ctx.h[0], ctx.h[1], ctx.h[2], ctx.h[3], ctx.h[4], ctx.h[5], ctx.h[6], ctx.h[7]
+	h0, h1, h2, h3, h4, h5, h6, h7 :=
+		ctx.h[0], ctx.h[1], ctx.h[2], ctx.h[3], ctx.h[4], ctx.h[5], ctx.h[6], ctx.h[7]
 	p := p
 	for len(p) >= BLAKE2B_BLOCK_SIZE {
 		ctx.t[0] += BLAKE2B_BLOCK_SIZE
 		if ctx.t[0] < BLAKE2B_BLOCK_SIZE {
-			ctx.t[1]+=1
-		} 
+			ctx.t[1] += 1
+		}
 		v0, v1, v2, v3, v4, v5, v6, v7 := h0, h1, h2, h3, h4, h5, h6, h7
 		v8 := BLAKE2B_IV[0]
 		v9 := BLAKE2B_IV[1]
@@ -1429,13 +1476,13 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v13 := BLAKE2B_IV[5] ~ ctx.t[1]
 		v14 := BLAKE2B_IV[6] ~ ctx.f[0]
 		v15 := BLAKE2B_IV[7] ~ ctx.f[1]
+
 		m: [16]u64 = ---
-		j := 0
-		for i := 0; i < 16; i+=1 {
-			m[i] = u64(p[j]) 		   | u64(p[j + 1]) << 8  | u64(p[j + 2]) << 16 | u64(p[j + 3]) << 24 |
-				   u64(p[j + 4]) << 32 | u64(p[j + 5]) << 40 | u64(p[j + 6]) << 48 | u64(p[j + 7]) << 56
-			j += 8
+		for i := 0; i < 16; i += 1 {
+			m[i] = endian.unchecked_get_u64le(p[i * 8:])
 		}
+
+		// Round 1
 		v0 += m[0]
 		v0 += v4
 		v12 ~= v0
@@ -1548,6 +1595,8 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
+		// Round 2
 		v0 += m[14]
 		v0 += v4
 		v12 ~= v0
@@ -1660,6 +1709,8 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
+		// Round 3
 		v0 += m[11]
 		v0 += v4
 		v12 ~= v0
@@ -1772,6 +1823,8 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
+		// Round 4
 		v0 += m[7]
 		v0 += v4
 		v12 ~= v0
@@ -1884,6 +1937,8 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
+		// Round 5
 		v0 += m[9]
 		v0 += v4
 		v12 ~= v0
@@ -1996,6 +2051,8 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
+		// Round 6
 		v0 += m[2]
 		v0 += v4
 		v12 ~= v0
@@ -2108,6 +2165,8 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
+		// Round 7
 		v0 += m[12]
 		v0 += v4
 		v12 ~= v0
@@ -2220,6 +2279,8 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
+		// Round 8
 		v0 += m[13]
 		v0 += v4
 		v12 ~= v0
@@ -2332,6 +2393,8 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
+		// Round 9
 		v0 += m[6]
 		v0 += v4
 		v12 ~= v0
@@ -2444,6 +2507,8 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
+		// Round 10
 		v0 += m[10]
 		v0 += v4
 		v12 ~= v0
@@ -2556,6 +2621,8 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
+		// Round 11
 		v0 += m[0]
 		v0 += v4
 		v12 ~= v0
@@ -2668,6 +2735,8 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
+		// Round 12
 		v0 += m[14]
 		v0 += v4
 		v12 ~= v0
@@ -2780,6 +2849,7 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		v10 += v15
 		v5 ~= v10
 		v5 = v5 << (64 - 63) | v5 >> 63
+
 		h0 ~= v0 ~ v8
 		h1 ~= v1 ~ v9
 		h2 ~= v2 ~ v10
@@ -2788,7 +2858,9 @@ blake2b_blocks :: #force_inline proc "contextless" (ctx: ^Blake2b_Context, p: []
 		h5 ~= v5 ~ v13
 		h6 ~= v6 ~ v14
 		h7 ~= v7 ~ v15
+
 		p = p[BLAKE2B_BLOCK_SIZE:]
 	}
-	ctx.h[0], ctx.h[1], ctx.h[2], ctx.h[3], ctx.h[4], ctx.h[5], ctx.h[6], ctx.h[7] = h0, h1, h2, h3, h4, h5, h6, h7
-}
+	ctx.h[0], ctx.h[1], ctx.h[2], ctx.h[3], ctx.h[4], ctx.h[5], ctx.h[6], ctx.h[7] =
+		h0, h1, h2, h3, h4, h5, h6, h7
+}
@@ -30,6 +30,6 @@ equivalence.

 For the most part, alterations to the base fiat-crypto generated code was
 kept to a minimum, to aid auditability.  This results in a somewhat
-ideosyncratic style, and in some cases minor performance penalties.
+idiosyncratic style, and in some cases minor performance penalties.

 [1]: https://github.com/mit-plv/fiat-crypto
@@ -9,14 +9,16 @@ package fiat
 u1 :: distinct u8
 i1 :: distinct i8

-cmovznz_u64 :: #force_inline proc "contextless" (arg1: u1, arg2, arg3: u64) -> (out1: u64) {
+@(optimization_mode="none")
+cmovznz_u64 :: proc "contextless" (arg1: u1, arg2, arg3: u64) -> (out1: u64) {
 	x1 := (u64(arg1) * 0xffffffffffffffff)
 	x2 := ((x1 & arg3) | ((~x1) & arg2))
 	out1 = x2
 	return
 }

-cmovznz_u32 :: #force_inline proc "contextless" (arg1: u1, arg2, arg3: u32) -> (out1: u32) {
+@(optimization_mode="none")
+cmovznz_u32 :: proc "contextless" (arg1: u1, arg2, arg3: u32) -> (out1: u32) {
 	x1 := (u32(arg1) * 0xffffffff)
 	x2 := ((x1 & arg3) | ((~x1) & arg2))
 	out1 = x2
@@ -305,7 +305,8 @@ fe_opp :: proc "contextless" (out1: ^Loose_Field_Element, arg1: ^Tight_Field_Ele
 	out1[4] = x5
 }

-fe_cond_assign :: proc "contextless" (out1, arg1: ^Tight_Field_Element, arg2: int) {
+@(optimization_mode="none")
+fe_cond_assign :: #force_no_inline proc "contextless" (out1, arg1: ^Tight_Field_Element, arg2: int) {
 	x1 := fiat.cmovznz_u64(fiat.u1(arg2), out1[0], arg1[0])
 	x2 := fiat.cmovznz_u64(fiat.u1(arg2), out1[1], arg1[1])
 	x3 := fiat.cmovznz_u64(fiat.u1(arg2), out1[2], arg1[2])
@@ -596,7 +597,8 @@ fe_set :: proc "contextless" (out1, arg1: ^Tight_Field_Element) {
 	out1[4] = x5
 }

-fe_cond_swap :: proc "contextless" (out1, out2: ^Tight_Field_Element, arg1: int) {
+@(optimization_mode="none")
+fe_cond_swap :: #force_no_inline proc "contextless" (out1, out2: ^Tight_Field_Element, arg1: int) {
 	mask := -u64(arg1)
 	x := (out1[0] ~ out2[0]) & mask
 	x1, y1 := out1[0] ~ x, out2[0] ~ x
@@ -1,6 +1,6 @@
 package field_poly1305

-import "core:crypto/util"
+import "core:encoding/endian"
 import "core:mem"

 fe_relax_cast :: #force_inline proc "contextless" (arg1: ^Tight_Field_Element) -> ^Loose_Field_Element {
@@ -11,7 +11,7 @@ fe_tighten_cast :: #force_inline proc "contextless" (arg1: ^Loose_Field_Element)
 	return transmute(^Tight_Field_Element)(arg1)
 }

-fe_from_bytes :: #force_inline proc (out1: ^Tight_Field_Element, arg1: []byte, arg2: byte, sanitize: bool = true) {
+fe_from_bytes :: #force_inline proc (out1: ^Tight_Field_Element, arg1: []byte, arg2: byte) {
 	// fiat-crypto's deserialization routine effectively processes a
 	// single byte at a time, and wants 256-bits of input for a value
 	// that will be 128-bits or 129-bits.
@@ -22,42 +22,29 @@ fe_from_bytes :: #force_inline proc (out1: ^Tight_Field_Element, arg1: []byte, a

 	assert(len(arg1) == 16)

-	when ODIN_ARCH == .i386 || ODIN_ARCH == .amd64 {
-		// While it may be unwise to do deserialization here on our
-		// own when fiat-crypto provides equivalent functionality,
-		// doing it this way provides a little under 3x performance
-		// improvement when optimization is enabled.
-		src_p := transmute(^[2]u64)(&arg1[0])
-		lo := src_p[0]
-		hi := src_p[1]
+	// While it may be unwise to do deserialization here on our
+	// own when fiat-crypto provides equivalent functionality,
+	// doing it this way provides a little under 3x performance
+	// improvement when optimization is enabled.
+	lo := endian.unchecked_get_u64le(arg1[0:])
+	hi := endian.unchecked_get_u64le(arg1[8:])

-		// This is inspired by poly1305-donna, though adjustments were
-		// made since a Tight_Field_Element's limbs are 44-bits, 43-bits,
-		// and 43-bits wide.
-		//
-		// Note: This could be transplated into fe_from_u64s, but that
-		// code is called once per MAC, and is non-criticial path.
-		hibit := u64(arg2) << 41 // arg2 << 128
-		out1[0] = lo & 0xfffffffffff
-		out1[1] = ((lo >> 44) | (hi << 20)) & 0x7ffffffffff
-		out1[2] = ((hi >> 23) & 0x7ffffffffff) | hibit
-	} else {
-		tmp: [32]byte
-		copy_slice(tmp[0:16], arg1[:])
-		tmp[16] = arg2
-
-		_fe_from_bytes(out1, &tmp)
-		if sanitize {
-			// This is used to deserialize `s` which is confidential.
-			mem.zero_explicit(&tmp, size_of(tmp))
-		}
-	}
+	// This is inspired by poly1305-donna, though adjustments were
+	// made since a Tight_Field_Element's limbs are 44-bits, 43-bits,
+	// and 43-bits wide.
+	//
+	// Note: This could be transplated into fe_from_u64s, but that
+	// code is called once per MAC, and is non-criticial path.
+	hibit := u64(arg2) << 41 // arg2 << 128
+	out1[0] = lo & 0xfffffffffff
+	out1[1] = ((lo >> 44) | (hi << 20)) & 0x7ffffffffff
+	out1[2] = ((hi >> 23) & 0x7ffffffffff) | hibit
 }

 fe_from_u64s :: proc "contextless" (out1: ^Tight_Field_Element, lo, hi: u64) {
 	tmp: [32]byte
-	util.PUT_U64_LE(tmp[0:8], lo)
-	util.PUT_U64_LE(tmp[8:16], hi)
+	endian.unchecked_put_u64le(tmp[0:], lo)
+	endian.unchecked_put_u64le(tmp[8:], hi)

 	_fe_from_bytes(out1, &tmp)

@@ -201,7 +201,8 @@ fe_opp :: proc "contextless" (out1: ^Loose_Field_Element, arg1: ^Tight_Field_Ele
 	out1[2] = x3
 }

-fe_cond_assign :: proc "contextless" (out1, arg1: ^Tight_Field_Element, arg2: bool) {
+@(optimization_mode="none")
+fe_cond_assign :: #force_no_inline proc "contextless" (out1, arg1: ^Tight_Field_Element, arg2: bool) {
 	x1 := fiat.cmovznz_u64(fiat.u1(arg2), out1[0], arg1[0])
 	x2 := fiat.cmovznz_u64(fiat.u1(arg2), out1[1], arg1[1])
 	x3 := fiat.cmovznz_u64(fiat.u1(arg2), out1[2], arg1[2])
@@ -342,7 +343,8 @@ fe_set :: #force_inline proc "contextless" (out1, arg1: ^Tight_Field_Element) {
 	out1[2] = x3
 }

-fe_cond_swap :: proc "contextless" (out1, out2: ^Tight_Field_Element, arg1: bool) {
+@(optimization_mode="none")
+fe_cond_swap :: #force_no_inline proc "contextless" (out1, out2: ^Tight_Field_Element, arg1: bool) {
 	mask := -u64(arg1)
 	x := (out1[0] ~ out2[0]) & mask
 	x1, y1 := out1[0] ~ x, out2[0] ~ x
@@ -1,169 +0,0 @@
-package _sha3
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation of the Keccak hashing algorithm, standardized as SHA3 in <https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.202.pdf>
-    To use the original Keccak padding, set the is_keccak bool to true, otherwise it will use SHA3 padding.
-*/
-
-import "../util"
-
-ROUNDS :: 24
-
-Sha3_Context :: struct {
-    st: struct #raw_union {
-        b: [200]u8,
-        q: [25]u64,
-    },
-    pt:        int,
-    rsiz:      int,
-    mdlen:     int,
-    is_keccak: bool,
-}
-
-keccakf :: proc "contextless" (st: ^[25]u64) {
-    keccakf_rndc := [?]u64 {
-        0x0000000000000001, 0x0000000000008082, 0x800000000000808a,
-        0x8000000080008000, 0x000000000000808b, 0x0000000080000001,
-        0x8000000080008081, 0x8000000000008009, 0x000000000000008a,
-        0x0000000000000088, 0x0000000080008009, 0x000000008000000a,
-        0x000000008000808b, 0x800000000000008b, 0x8000000000008089,
-        0x8000000000008003, 0x8000000000008002, 0x8000000000000080,
-        0x000000000000800a, 0x800000008000000a, 0x8000000080008081,
-        0x8000000000008080, 0x0000000080000001, 0x8000000080008008,
-    }
-
-    keccakf_rotc := [?]i32 {
-        1,  3,  6,  10, 15, 21, 28, 36, 45, 55, 2,  14,
-        27, 41, 56, 8,  25, 43, 62, 18, 39, 61, 20, 44,
-    }
-
-    keccakf_piln := [?]i32 {
-        10, 7,  11, 17, 18, 3, 5,  16, 8,  21, 24, 4,
-        15, 23, 19, 13, 12, 2, 20, 14, 22, 9,  6,  1,
-    }
-
-    i, j, r: i32 = ---, ---, ---
-    t: u64       = ---
-    bc: [5]u64   = ---
-
-    when ODIN_ENDIAN != .Little {
-        v: uintptr = ---
-        for i = 0; i < 25; i += 1 {
-            v := uintptr(&st[i])
-            st[i] = u64((^u8)(v + 0)^ << 0)  | u64((^u8)(v + 1)^ << 8)  |
-                    u64((^u8)(v + 2)^ << 16) | u64((^u8)(v + 3)^ << 24) |
-                    u64((^u8)(v + 4)^ << 32) | u64((^u8)(v + 5)^ << 40) |
-                    u64((^u8)(v + 6)^ << 48) | u64((^u8)(v + 7)^ << 56)
-        }
-    }
-
-    for r = 0; r < ROUNDS; r += 1 {
-        // theta
-        for i = 0; i < 5; i += 1 {
-            bc[i] = st[i] ~ st[i + 5] ~ st[i + 10] ~ st[i + 15] ~ st[i + 20]
-        }
-
-        for i = 0; i < 5; i += 1 {
-            t = bc[(i + 4) % 5] ~ util.ROTL64(bc[(i + 1) % 5], 1)
-            for j = 0; j < 25; j += 5 {
-                st[j + i] ~= t
-            }
-        }
-
-        // rho pi
-        t = st[1]
-        for i = 0; i < 24; i += 1 {
-            j = keccakf_piln[i]
-            bc[0] = st[j]
-            st[j] = util.ROTL64(t, u64(keccakf_rotc[i]))
-            t = bc[0]
-        }
-
-        // chi
-        for j = 0; j < 25; j += 5 {
-            for i = 0; i < 5; i += 1 {
-                bc[i] = st[j + i]
-            }
-            for i = 0; i < 5; i += 1 {
-                st[j + i] ~= ~bc[(i + 1) % 5] & bc[(i + 2) % 5]
-            }
-        }
-
-        st[0] ~= keccakf_rndc[r]
-    }
-
-    when ODIN_ENDIAN != .Little {
-        for i = 0; i < 25; i += 1 {
-            v = uintptr(&st[i])
-            t = st[i]
-            (^u8)(v + 0)^ = (t >> 0)  & 0xff
-            (^u8)(v + 1)^ = (t >> 8)  & 0xff
-            (^u8)(v + 2)^ = (t >> 16) & 0xff
-            (^u8)(v + 3)^ = (t >> 24) & 0xff
-            (^u8)(v + 4)^ = (t >> 32) & 0xff
-            (^u8)(v + 5)^ = (t >> 40) & 0xff
-            (^u8)(v + 6)^ = (t >> 48) & 0xff
-            (^u8)(v + 7)^ = (t >> 56) & 0xff
-        }
-    }
-}
-
-init :: proc "contextless" (c: ^Sha3_Context) {
-    for i := 0; i < 25; i += 1 {
-        c.st.q[i] = 0
-    }
-    c.rsiz = 200 - 2 * c.mdlen
-}
-
-update :: proc "contextless" (c: ^Sha3_Context, data: []byte) {
-    j := c.pt
-    for i := 0; i < len(data); i += 1 {
-        c.st.b[j] ~= data[i]
-        j += 1
-        if j >= c.rsiz {
-            keccakf(&c.st.q)
-            j = 0
-        }
-    }
-    c.pt = j
-}
-
-final :: proc "contextless" (c: ^Sha3_Context, hash: []byte) {
-    if c.is_keccak {
-        c.st.b[c.pt] ~= 0x01
-    } else {
-        c.st.b[c.pt] ~= 0x06
-    }
-    
-    c.st.b[c.rsiz - 1] ~= 0x80
-    keccakf(&c.st.q)
-    for i := 0; i < c.mdlen; i += 1 {
-        hash[i] = c.st.b[i]
-    }
-}
-
-shake_xof :: proc "contextless" (c: ^Sha3_Context) {
-    c.st.b[c.pt]       ~= 0x1F
-    c.st.b[c.rsiz - 1] ~= 0x80
-    keccakf(&c.st.q)
-    c.pt = 0
-}
-
-shake_out :: proc "contextless" (c: ^Sha3_Context, hash: []byte) {
-    j := c.pt
-    for i := 0; i < len(hash); i += 1 {
-        if j >= c.rsiz {
-            keccakf(&c.st.q)
-            j = 0
-        }
-        hash[i] = c.st.b[j]
-        j += 1
-    }
-    c.pt = j
-}
@@ -0,0 +1,183 @@
+package _sha3
+
+/*
+    Copyright 2021 zhibog
+    Made available under the BSD-3 license.
+
+    List of contributors:
+        zhibog, dotbmp:  Initial implementation.
+
+    Implementation of the Keccak hashing algorithm, standardized as SHA3 in <https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.202.pdf>
+    To use the original Keccak padding, set the is_keccak bool to true, otherwise it will use SHA3 padding.
+*/
+
+import "core:math/bits"
+
+ROUNDS :: 24
+
+Sha3_Context :: struct {
+	st:        struct #raw_union {
+		b: [200]u8,
+		q: [25]u64,
+	},
+	pt:        int,
+	rsiz:      int,
+	mdlen:     int,
+	is_keccak: bool,
+
+	is_initialized: bool,
+	is_finalized:   bool, // For SHAKE (unlimited squeeze is allowed)
+}
+
+keccakf :: proc "contextless" (st: ^[25]u64) {
+	keccakf_rndc := [?]u64 {
+		0x0000000000000001, 0x0000000000008082, 0x800000000000808a,
+		0x8000000080008000, 0x000000000000808b, 0x0000000080000001,
+		0x8000000080008081, 0x8000000000008009, 0x000000000000008a,
+		0x0000000000000088, 0x0000000080008009, 0x000000008000000a,
+		0x000000008000808b, 0x800000000000008b, 0x8000000000008089,
+		0x8000000000008003, 0x8000000000008002, 0x8000000000000080,
+		0x000000000000800a, 0x800000008000000a, 0x8000000080008081,
+		0x8000000000008080, 0x0000000080000001, 0x8000000080008008,
+	}
+
+	keccakf_rotc := [?]int {
+		1, 3, 6, 10, 15, 21, 28, 36, 45, 55, 2, 14,
+		27, 41, 56, 8, 25, 43, 62, 18, 39, 61, 20, 44,
+	}
+
+	keccakf_piln := [?]i32 {
+		10, 7, 11, 17, 18, 3, 5, 16, 8, 21, 24, 4,
+		15, 23, 19, 13, 12, 2, 20, 14, 22, 9, 6, 1,
+	}
+
+	i, j, r: i32 = ---, ---, ---
+	t: u64 = ---
+	bc: [5]u64 = ---
+
+	when ODIN_ENDIAN != .Little {
+		for i = 0; i < 25; i += 1 {
+			st[i] = bits.byte_swap(st[i])
+		}
+	}
+
+	for r = 0; r < ROUNDS; r += 1 {
+		// theta
+		for i = 0; i < 5; i += 1 {
+			bc[i] = st[i] ~ st[i + 5] ~ st[i + 10] ~ st[i + 15] ~ st[i + 20]
+		}
+
+		for i = 0; i < 5; i += 1 {
+			t = bc[(i + 4) % 5] ~ bits.rotate_left64(bc[(i + 1) % 5], 1)
+			for j = 0; j < 25; j += 5 {
+				st[j + i] ~= t
+			}
+		}
+
+		// rho pi
+		t = st[1]
+		for i = 0; i < 24; i += 1 {
+			j = keccakf_piln[i]
+			bc[0] = st[j]
+			st[j] = bits.rotate_left64(t, keccakf_rotc[i])
+			t = bc[0]
+		}
+
+		// chi
+		for j = 0; j < 25; j += 5 {
+			for i = 0; i < 5; i += 1 {
+				bc[i] = st[j + i]
+			}
+			for i = 0; i < 5; i += 1 {
+				st[j + i] ~= ~bc[(i + 1) % 5] & bc[(i + 2) % 5]
+			}
+		}
+
+		st[0] ~= keccakf_rndc[r]
+	}
+
+	when ODIN_ENDIAN != .Little {
+		for i = 0; i < 25; i += 1 {
+			st[i] = bits.byte_swap(st[i])
+		}
+	}
+}
+
+init :: proc(c: ^Sha3_Context) {
+	for i := 0; i < 25; i += 1 {
+		c.st.q[i] = 0
+	}
+	c.rsiz = 200 - 2 * c.mdlen
+	c.pt = 0
+
+	c.is_initialized = true
+	c.is_finalized = false
+}
+
+update :: proc(c: ^Sha3_Context, data: []byte) {
+	assert(c.is_initialized)
+	assert(!c.is_finalized)
+
+	j := c.pt
+	for i := 0; i < len(data); i += 1 {
+		c.st.b[j] ~= data[i]
+		j += 1
+		if j >= c.rsiz {
+			keccakf(&c.st.q)
+			j = 0
+		}
+	}
+	c.pt = j
+}
+
+final :: proc(c: ^Sha3_Context, hash: []byte) {
+	assert(c.is_initialized)
+
+	if len(hash) < c.mdlen {
+		if c.is_keccak {
+			panic("crypto/keccac: invalid destination digest size")
+		}
+		panic("crypto/sha3: invalid destination digest size")
+	}
+	if c.is_keccak {
+		c.st.b[c.pt] ~= 0x01
+	} else {
+		c.st.b[c.pt] ~= 0x06
+	}
+
+	c.st.b[c.rsiz - 1] ~= 0x80
+	keccakf(&c.st.q)
+	for i := 0; i < c.mdlen; i += 1 {
+		hash[i] = c.st.b[i]
+	}
+
+	c.is_initialized = false // No more absorb, no more squeeze.
+}
+
+shake_xof :: proc(c: ^Sha3_Context) {
+	assert(c.is_initialized)
+	assert(!c.is_finalized)
+
+	c.st.b[c.pt] ~= 0x1F
+	c.st.b[c.rsiz - 1] ~= 0x80
+	keccakf(&c.st.q)
+	c.pt = 0
+
+	c.is_finalized = true // No more absorb, unlimited squeeze.
+}
+
+shake_out :: proc(c: ^Sha3_Context, hash: []byte) {
+	assert(c.is_initialized)
+	assert(c.is_finalized)
+
+	j := c.pt
+	for i := 0; i < len(hash); i += 1 {
+		if j >= c.rsiz {
+			keccakf(&c.st.q)
+			j = 0
+		}
+		hash[i] = c.st.b[j]
+		j += 1
+	}
+	c.pt = j
+}
@@ -1,410 +0,0 @@
-package _tiger
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation of the Tiger hashing algorithm, as defined in <https://www.cs.technion.ac.il/~biham/Reports/Tiger/>
-*/
-
-import "../util"
-
-T1 := [?]u64 {
-	0x02aab17cf7e90c5e, 0xac424b03e243a8ec, 0x72cd5be30dd5fcd3, 0x6d019b93f6f97f3a,
-	0xcd9978ffd21f9193, 0x7573a1c9708029e2, 0xb164326b922a83c3, 0x46883eee04915870,
-	0xeaace3057103ece6, 0xc54169b808a3535c, 0x4ce754918ddec47c, 0x0aa2f4dfdc0df40c,
-	0x10b76f18a74dbefa, 0xc6ccb6235ad1ab6a, 0x13726121572fe2ff, 0x1a488c6f199d921e,
-	0x4bc9f9f4da0007ca, 0x26f5e6f6e85241c7, 0x859079dbea5947b6, 0x4f1885c5c99e8c92,
-	0xd78e761ea96f864b, 0x8e36428c52b5c17d, 0x69cf6827373063c1, 0xb607c93d9bb4c56e,
-	0x7d820e760e76b5ea, 0x645c9cc6f07fdc42, 0xbf38a078243342e0, 0x5f6b343c9d2e7d04,
-	0xf2c28aeb600b0ec6, 0x6c0ed85f7254bcac, 0x71592281a4db4fe5, 0x1967fa69ce0fed9f,
-	0xfd5293f8b96545db, 0xc879e9d7f2a7600b, 0x860248920193194e, 0xa4f9533b2d9cc0b3,
-	0x9053836c15957613, 0xdb6dcf8afc357bf1, 0x18beea7a7a370f57, 0x037117ca50b99066,
-	0x6ab30a9774424a35, 0xf4e92f02e325249b, 0x7739db07061ccae1, 0xd8f3b49ceca42a05,
-	0xbd56be3f51382f73, 0x45faed5843b0bb28, 0x1c813d5c11bf1f83, 0x8af0e4b6d75fa169,
-	0x33ee18a487ad9999, 0x3c26e8eab1c94410, 0xb510102bc0a822f9, 0x141eef310ce6123b,
-	0xfc65b90059ddb154, 0xe0158640c5e0e607, 0x884e079826c3a3cf, 0x930d0d9523c535fd,
-	0x35638d754e9a2b00, 0x4085fccf40469dd5, 0xc4b17ad28be23a4c, 0xcab2f0fc6a3e6a2e,
-	0x2860971a6b943fcd, 0x3dde6ee212e30446, 0x6222f32ae01765ae, 0x5d550bb5478308fe,
-	0xa9efa98da0eda22a, 0xc351a71686c40da7, 0x1105586d9c867c84, 0xdcffee85fda22853,
-	0xccfbd0262c5eef76, 0xbaf294cb8990d201, 0xe69464f52afad975, 0x94b013afdf133e14,
-	0x06a7d1a32823c958, 0x6f95fe5130f61119, 0xd92ab34e462c06c0, 0xed7bde33887c71d2,
-	0x79746d6e6518393e, 0x5ba419385d713329, 0x7c1ba6b948a97564, 0x31987c197bfdac67,
-	0xde6c23c44b053d02, 0x581c49fed002d64d, 0xdd474d6338261571, 0xaa4546c3e473d062,
-	0x928fce349455f860, 0x48161bbacaab94d9, 0x63912430770e6f68, 0x6ec8a5e602c6641c,
-	0x87282515337ddd2b, 0x2cda6b42034b701b, 0xb03d37c181cb096d, 0xe108438266c71c6f,
-	0x2b3180c7eb51b255, 0xdf92b82f96c08bbc, 0x5c68c8c0a632f3ba, 0x5504cc861c3d0556,
-	0xabbfa4e55fb26b8f, 0x41848b0ab3baceb4, 0xb334a273aa445d32, 0xbca696f0a85ad881,
-	0x24f6ec65b528d56c, 0x0ce1512e90f4524a, 0x4e9dd79d5506d35a, 0x258905fac6ce9779,
-	0x2019295b3e109b33, 0xf8a9478b73a054cc, 0x2924f2f934417eb0, 0x3993357d536d1bc4,
-	0x38a81ac21db6ff8b, 0x47c4fbf17d6016bf, 0x1e0faadd7667e3f5, 0x7abcff62938beb96,
-	0xa78dad948fc179c9, 0x8f1f98b72911e50d, 0x61e48eae27121a91, 0x4d62f7ad31859808,
-	0xeceba345ef5ceaeb, 0xf5ceb25ebc9684ce, 0xf633e20cb7f76221, 0xa32cdf06ab8293e4,
-	0x985a202ca5ee2ca4, 0xcf0b8447cc8a8fb1, 0x9f765244979859a3, 0xa8d516b1a1240017,
-	0x0bd7ba3ebb5dc726, 0xe54bca55b86adb39, 0x1d7a3afd6c478063, 0x519ec608e7669edd,
-	0x0e5715a2d149aa23, 0x177d4571848ff194, 0xeeb55f3241014c22, 0x0f5e5ca13a6e2ec2,
-	0x8029927b75f5c361, 0xad139fabc3d6e436, 0x0d5df1a94ccf402f, 0x3e8bd948bea5dfc8,
-	0xa5a0d357bd3ff77e, 0xa2d12e251f74f645, 0x66fd9e525e81a082, 0x2e0c90ce7f687a49,
-	0xc2e8bcbeba973bc5, 0x000001bce509745f, 0x423777bbe6dab3d6, 0xd1661c7eaef06eb5,
-	0xa1781f354daacfd8, 0x2d11284a2b16affc, 0xf1fc4f67fa891d1f, 0x73ecc25dcb920ada,
-	0xae610c22c2a12651, 0x96e0a810d356b78a, 0x5a9a381f2fe7870f, 0xd5ad62ede94e5530,
-	0xd225e5e8368d1427, 0x65977b70c7af4631, 0x99f889b2de39d74f, 0x233f30bf54e1d143,
-	0x9a9675d3d9a63c97, 0x5470554ff334f9a8, 0x166acb744a4f5688, 0x70c74caab2e4aead,
-	0xf0d091646f294d12, 0x57b82a89684031d1, 0xefd95a5a61be0b6b, 0x2fbd12e969f2f29a,
-	0x9bd37013feff9fe8, 0x3f9b0404d6085a06, 0x4940c1f3166cfe15, 0x09542c4dcdf3defb,
-	0xb4c5218385cd5ce3, 0xc935b7dc4462a641, 0x3417f8a68ed3b63f, 0xb80959295b215b40,
-	0xf99cdaef3b8c8572, 0x018c0614f8fcb95d, 0x1b14accd1a3acdf3, 0x84d471f200bb732d,
-	0xc1a3110e95e8da16, 0x430a7220bf1a82b8, 0xb77e090d39df210e, 0x5ef4bd9f3cd05e9d,
-	0x9d4ff6da7e57a444, 0xda1d60e183d4a5f8, 0xb287c38417998e47, 0xfe3edc121bb31886,
-	0xc7fe3ccc980ccbef, 0xe46fb590189bfd03, 0x3732fd469a4c57dc, 0x7ef700a07cf1ad65,
-	0x59c64468a31d8859, 0x762fb0b4d45b61f6, 0x155baed099047718, 0x68755e4c3d50baa6,
-	0xe9214e7f22d8b4df, 0x2addbf532eac95f4, 0x32ae3909b4bd0109, 0x834df537b08e3450,
-	0xfa209da84220728d, 0x9e691d9b9efe23f7, 0x0446d288c4ae8d7f, 0x7b4cc524e169785b,
-	0x21d87f0135ca1385, 0xcebb400f137b8aa5, 0x272e2b66580796be, 0x3612264125c2b0de,
-	0x057702bdad1efbb2, 0xd4babb8eacf84be9, 0x91583139641bc67b, 0x8bdc2de08036e024,
-	0x603c8156f49f68ed, 0xf7d236f7dbef5111, 0x9727c4598ad21e80, 0xa08a0896670a5fd7,
-	0xcb4a8f4309eba9cb, 0x81af564b0f7036a1, 0xc0b99aa778199abd, 0x959f1ec83fc8e952,
-	0x8c505077794a81b9, 0x3acaaf8f056338f0, 0x07b43f50627a6778, 0x4a44ab49f5eccc77,
-	0x3bc3d6e4b679ee98, 0x9cc0d4d1cf14108c, 0x4406c00b206bc8a0, 0x82a18854c8d72d89,
-	0x67e366b35c3c432c, 0xb923dd61102b37f2, 0x56ab2779d884271d, 0xbe83e1b0ff1525af,
-	0xfb7c65d4217e49a9, 0x6bdbe0e76d48e7d4, 0x08df828745d9179e, 0x22ea6a9add53bd34,
-	0xe36e141c5622200a, 0x7f805d1b8cb750ee, 0xafe5c7a59f58e837, 0xe27f996a4fb1c23c,
-	0xd3867dfb0775f0d0, 0xd0e673de6e88891a, 0x123aeb9eafb86c25, 0x30f1d5d5c145b895,
-	0xbb434a2dee7269e7, 0x78cb67ecf931fa38, 0xf33b0372323bbf9c, 0x52d66336fb279c74,
-	0x505f33ac0afb4eaa, 0xe8a5cd99a2cce187, 0x534974801e2d30bb, 0x8d2d5711d5876d90,
-	0x1f1a412891bc038e, 0xd6e2e71d82e56648, 0x74036c3a497732b7, 0x89b67ed96361f5ab,
-	0xffed95d8f1ea02a2, 0xe72b3bd61464d43d, 0xa6300f170bdc4820, 0xebc18760ed78a77a,
-}
-
-T2 := [?]u64 {
-	0xe6a6be5a05a12138, 0xb5a122a5b4f87c98, 0x563c6089140b6990, 0x4c46cb2e391f5dd5,
-	0xd932addbc9b79434, 0x08ea70e42015aff5, 0xd765a6673e478cf1, 0xc4fb757eab278d99,
-	0xdf11c6862d6e0692, 0xddeb84f10d7f3b16, 0x6f2ef604a665ea04, 0x4a8e0f0ff0e0dfb3,
-	0xa5edeef83dbcba51, 0xfc4f0a2a0ea4371e, 0xe83e1da85cb38429, 0xdc8ff882ba1b1ce2,
-	0xcd45505e8353e80d, 0x18d19a00d4db0717, 0x34a0cfeda5f38101, 0x0be77e518887caf2,
-	0x1e341438b3c45136, 0xe05797f49089ccf9, 0xffd23f9df2591d14, 0x543dda228595c5cd,
-	0x661f81fd99052a33, 0x8736e641db0f7b76, 0x15227725418e5307, 0xe25f7f46162eb2fa,
-	0x48a8b2126c13d9fe, 0xafdc541792e76eea, 0x03d912bfc6d1898f, 0x31b1aafa1b83f51b,
-	0xf1ac2796e42ab7d9, 0x40a3a7d7fcd2ebac, 0x1056136d0afbbcc5, 0x7889e1dd9a6d0c85,
-	0xd33525782a7974aa, 0xa7e25d09078ac09b, 0xbd4138b3eac6edd0, 0x920abfbe71eb9e70,
-	0xa2a5d0f54fc2625c, 0xc054e36b0b1290a3, 0xf6dd59ff62fe932b, 0x3537354511a8ac7d,
-	0xca845e9172fadcd4, 0x84f82b60329d20dc, 0x79c62ce1cd672f18, 0x8b09a2add124642c,
-	0xd0c1e96a19d9e726, 0x5a786a9b4ba9500c, 0x0e020336634c43f3, 0xc17b474aeb66d822,
-	0x6a731ae3ec9baac2, 0x8226667ae0840258, 0x67d4567691caeca5, 0x1d94155c4875adb5,
-	0x6d00fd985b813fdf, 0x51286efcb774cd06, 0x5e8834471fa744af, 0xf72ca0aee761ae2e,
-	0xbe40e4cdaee8e09a, 0xe9970bbb5118f665, 0x726e4beb33df1964, 0x703b000729199762,
-	0x4631d816f5ef30a7, 0xb880b5b51504a6be, 0x641793c37ed84b6c, 0x7b21ed77f6e97d96,
-	0x776306312ef96b73, 0xae528948e86ff3f4, 0x53dbd7f286a3f8f8, 0x16cadce74cfc1063,
-	0x005c19bdfa52c6dd, 0x68868f5d64d46ad3, 0x3a9d512ccf1e186a, 0x367e62c2385660ae,
-	0xe359e7ea77dcb1d7, 0x526c0773749abe6e, 0x735ae5f9d09f734b, 0x493fc7cc8a558ba8,
-	0xb0b9c1533041ab45, 0x321958ba470a59bd, 0x852db00b5f46c393, 0x91209b2bd336b0e5,
-	0x6e604f7d659ef19f, 0xb99a8ae2782ccb24, 0xccf52ab6c814c4c7, 0x4727d9afbe11727b,
-	0x7e950d0c0121b34d, 0x756f435670ad471f, 0xf5add442615a6849, 0x4e87e09980b9957a,
-	0x2acfa1df50aee355, 0xd898263afd2fd556, 0xc8f4924dd80c8fd6, 0xcf99ca3d754a173a,
-	0xfe477bacaf91bf3c, 0xed5371f6d690c12d, 0x831a5c285e687094, 0xc5d3c90a3708a0a4,
-	0x0f7f903717d06580, 0x19f9bb13b8fdf27f, 0xb1bd6f1b4d502843, 0x1c761ba38fff4012,
-	0x0d1530c4e2e21f3b, 0x8943ce69a7372c8a, 0xe5184e11feb5ce66, 0x618bdb80bd736621,
-	0x7d29bad68b574d0b, 0x81bb613e25e6fe5b, 0x071c9c10bc07913f, 0xc7beeb7909ac2d97,
-	0xc3e58d353bc5d757, 0xeb017892f38f61e8, 0xd4effb9c9b1cc21a, 0x99727d26f494f7ab,
-	0xa3e063a2956b3e03, 0x9d4a8b9a4aa09c30, 0x3f6ab7d500090fb4, 0x9cc0f2a057268ac0,
-	0x3dee9d2dedbf42d1, 0x330f49c87960a972, 0xc6b2720287421b41, 0x0ac59ec07c00369c,
-	0xef4eac49cb353425, 0xf450244eef0129d8, 0x8acc46e5caf4deb6, 0x2ffeab63989263f7,
-	0x8f7cb9fe5d7a4578, 0x5bd8f7644e634635, 0x427a7315bf2dc900, 0x17d0c4aa2125261c,
-	0x3992486c93518e50, 0xb4cbfee0a2d7d4c3, 0x7c75d6202c5ddd8d, 0xdbc295d8e35b6c61,
-	0x60b369d302032b19, 0xce42685fdce44132, 0x06f3ddb9ddf65610, 0x8ea4d21db5e148f0,
-	0x20b0fce62fcd496f, 0x2c1b912358b0ee31, 0xb28317b818f5a308, 0xa89c1e189ca6d2cf,
-	0x0c6b18576aaadbc8, 0xb65deaa91299fae3, 0xfb2b794b7f1027e7, 0x04e4317f443b5beb,
-	0x4b852d325939d0a6, 0xd5ae6beefb207ffc, 0x309682b281c7d374, 0xbae309a194c3b475,
-	0x8cc3f97b13b49f05, 0x98a9422ff8293967, 0x244b16b01076ff7c, 0xf8bf571c663d67ee,
-	0x1f0d6758eee30da1, 0xc9b611d97adeb9b7, 0xb7afd5887b6c57a2, 0x6290ae846b984fe1,
-	0x94df4cdeacc1a5fd, 0x058a5bd1c5483aff, 0x63166cc142ba3c37, 0x8db8526eb2f76f40,
-	0xe10880036f0d6d4e, 0x9e0523c9971d311d, 0x45ec2824cc7cd691, 0x575b8359e62382c9,
-	0xfa9e400dc4889995, 0xd1823ecb45721568, 0xdafd983b8206082f, 0xaa7d29082386a8cb,
-	0x269fcd4403b87588, 0x1b91f5f728bdd1e0, 0xe4669f39040201f6, 0x7a1d7c218cf04ade,
-	0x65623c29d79ce5ce, 0x2368449096c00bb1, 0xab9bf1879da503ba, 0xbc23ecb1a458058e,
-	0x9a58df01bb401ecc, 0xa070e868a85f143d, 0x4ff188307df2239e, 0x14d565b41a641183,
-	0xee13337452701602, 0x950e3dcf3f285e09, 0x59930254b9c80953, 0x3bf299408930da6d,
-	0xa955943f53691387, 0xa15edecaa9cb8784, 0x29142127352be9a0, 0x76f0371fff4e7afb,
-	0x0239f450274f2228, 0xbb073af01d5e868b, 0xbfc80571c10e96c1, 0xd267088568222e23,
-	0x9671a3d48e80b5b0, 0x55b5d38ae193bb81, 0x693ae2d0a18b04b8, 0x5c48b4ecadd5335f,
-	0xfd743b194916a1ca, 0x2577018134be98c4, 0xe77987e83c54a4ad, 0x28e11014da33e1b9,
-	0x270cc59e226aa213, 0x71495f756d1a5f60, 0x9be853fb60afef77, 0xadc786a7f7443dbf,
-	0x0904456173b29a82, 0x58bc7a66c232bd5e, 0xf306558c673ac8b2, 0x41f639c6b6c9772a,
-	0x216defe99fda35da, 0x11640cc71c7be615, 0x93c43694565c5527, 0xea038e6246777839,
-	0xf9abf3ce5a3e2469, 0x741e768d0fd312d2, 0x0144b883ced652c6, 0xc20b5a5ba33f8552,
-	0x1ae69633c3435a9d, 0x97a28ca4088cfdec, 0x8824a43c1e96f420, 0x37612fa66eeea746,
-	0x6b4cb165f9cf0e5a, 0x43aa1c06a0abfb4a, 0x7f4dc26ff162796b, 0x6cbacc8e54ed9b0f,
-	0xa6b7ffefd2bb253e, 0x2e25bc95b0a29d4f, 0x86d6a58bdef1388c, 0xded74ac576b6f054,
-	0x8030bdbc2b45805d, 0x3c81af70e94d9289, 0x3eff6dda9e3100db, 0xb38dc39fdfcc8847,
-	0x123885528d17b87e, 0xf2da0ed240b1b642, 0x44cefadcd54bf9a9, 0x1312200e433c7ee6,
-	0x9ffcc84f3a78c748, 0xf0cd1f72248576bb, 0xec6974053638cfe4, 0x2ba7b67c0cec4e4c,
-	0xac2f4df3e5ce32ed, 0xcb33d14326ea4c11, 0xa4e9044cc77e58bc, 0x5f513293d934fcef,
-	0x5dc9645506e55444, 0x50de418f317de40a, 0x388cb31a69dde259, 0x2db4a83455820a86,
-	0x9010a91e84711ae9, 0x4df7f0b7b1498371, 0xd62a2eabc0977179, 0x22fac097aa8d5c0e,
-}
-
-T3 := [?]u64 {
-	0xf49fcc2ff1daf39b, 0x487fd5c66ff29281, 0xe8a30667fcdca83f, 0x2c9b4be3d2fcce63,
-	0xda3ff74b93fbbbc2, 0x2fa165d2fe70ba66, 0xa103e279970e93d4, 0xbecdec77b0e45e71,
-	0xcfb41e723985e497, 0xb70aaa025ef75017, 0xd42309f03840b8e0, 0x8efc1ad035898579,
-	0x96c6920be2b2abc5, 0x66af4163375a9172, 0x2174abdcca7127fb, 0xb33ccea64a72ff41,
-	0xf04a4933083066a5, 0x8d970acdd7289af5, 0x8f96e8e031c8c25e, 0xf3fec02276875d47,
-	0xec7bf310056190dd, 0xf5adb0aebb0f1491, 0x9b50f8850fd58892, 0x4975488358b74de8,
-	0xa3354ff691531c61, 0x0702bbe481d2c6ee, 0x89fb24057deded98, 0xac3075138596e902,
-	0x1d2d3580172772ed, 0xeb738fc28e6bc30d, 0x5854ef8f63044326, 0x9e5c52325add3bbe,
-	0x90aa53cf325c4623, 0xc1d24d51349dd067, 0x2051cfeea69ea624, 0x13220f0a862e7e4f,
-	0xce39399404e04864, 0xd9c42ca47086fcb7, 0x685ad2238a03e7cc, 0x066484b2ab2ff1db,
-	0xfe9d5d70efbf79ec, 0x5b13b9dd9c481854, 0x15f0d475ed1509ad, 0x0bebcd060ec79851,
-	0xd58c6791183ab7f8, 0xd1187c5052f3eee4, 0xc95d1192e54e82ff, 0x86eea14cb9ac6ca2,
-	0x3485beb153677d5d, 0xdd191d781f8c492a, 0xf60866baa784ebf9, 0x518f643ba2d08c74,
-	0x8852e956e1087c22, 0xa768cb8dc410ae8d, 0x38047726bfec8e1a, 0xa67738b4cd3b45aa,
-	0xad16691cec0dde19, 0xc6d4319380462e07, 0xc5a5876d0ba61938, 0x16b9fa1fa58fd840,
-	0x188ab1173ca74f18, 0xabda2f98c99c021f, 0x3e0580ab134ae816, 0x5f3b05b773645abb,
-	0x2501a2be5575f2f6, 0x1b2f74004e7e8ba9, 0x1cd7580371e8d953, 0x7f6ed89562764e30,
-	0xb15926ff596f003d, 0x9f65293da8c5d6b9, 0x6ecef04dd690f84c, 0x4782275fff33af88,
-	0xe41433083f820801, 0xfd0dfe409a1af9b5, 0x4325a3342cdb396b, 0x8ae77e62b301b252,
-	0xc36f9e9f6655615a, 0x85455a2d92d32c09, 0xf2c7dea949477485, 0x63cfb4c133a39eba,
-	0x83b040cc6ebc5462, 0x3b9454c8fdb326b0, 0x56f56a9e87ffd78c, 0x2dc2940d99f42bc6,
-	0x98f7df096b096e2d, 0x19a6e01e3ad852bf, 0x42a99ccbdbd4b40b, 0xa59998af45e9c559,
-	0x366295e807d93186, 0x6b48181bfaa1f773, 0x1fec57e2157a0a1d, 0x4667446af6201ad5,
-	0xe615ebcacfb0f075, 0xb8f31f4f68290778, 0x22713ed6ce22d11e, 0x3057c1a72ec3c93b,
-	0xcb46acc37c3f1f2f, 0xdbb893fd02aaf50e, 0x331fd92e600b9fcf, 0xa498f96148ea3ad6,
-	0xa8d8426e8b6a83ea, 0xa089b274b7735cdc, 0x87f6b3731e524a11, 0x118808e5cbc96749,
-	0x9906e4c7b19bd394, 0xafed7f7e9b24a20c, 0x6509eadeeb3644a7, 0x6c1ef1d3e8ef0ede,
-	0xb9c97d43e9798fb4, 0xa2f2d784740c28a3, 0x7b8496476197566f, 0x7a5be3e6b65f069d,
-	0xf96330ed78be6f10, 0xeee60de77a076a15, 0x2b4bee4aa08b9bd0, 0x6a56a63ec7b8894e,
-	0x02121359ba34fef4, 0x4cbf99f8283703fc, 0x398071350caf30c8, 0xd0a77a89f017687a,
-	0xf1c1a9eb9e423569, 0x8c7976282dee8199, 0x5d1737a5dd1f7abd, 0x4f53433c09a9fa80,
-	0xfa8b0c53df7ca1d9, 0x3fd9dcbc886ccb77, 0xc040917ca91b4720, 0x7dd00142f9d1dcdf,
-	0x8476fc1d4f387b58, 0x23f8e7c5f3316503, 0x032a2244e7e37339, 0x5c87a5d750f5a74b,
-	0x082b4cc43698992e, 0xdf917becb858f63c, 0x3270b8fc5bf86dda, 0x10ae72bb29b5dd76,
-	0x576ac94e7700362b, 0x1ad112dac61efb8f, 0x691bc30ec5faa427, 0xff246311cc327143,
-	0x3142368e30e53206, 0x71380e31e02ca396, 0x958d5c960aad76f1, 0xf8d6f430c16da536,
-	0xc8ffd13f1be7e1d2, 0x7578ae66004ddbe1, 0x05833f01067be646, 0xbb34b5ad3bfe586d,
-	0x095f34c9a12b97f0, 0x247ab64525d60ca8, 0xdcdbc6f3017477d1, 0x4a2e14d4decad24d,
-	0xbdb5e6d9be0a1eeb, 0x2a7e70f7794301ab, 0xdef42d8a270540fd, 0x01078ec0a34c22c1,
-	0xe5de511af4c16387, 0x7ebb3a52bd9a330a, 0x77697857aa7d6435, 0x004e831603ae4c32,
-	0xe7a21020ad78e312, 0x9d41a70c6ab420f2, 0x28e06c18ea1141e6, 0xd2b28cbd984f6b28,
-	0x26b75f6c446e9d83, 0xba47568c4d418d7f, 0xd80badbfe6183d8e, 0x0e206d7f5f166044,
-	0xe258a43911cbca3e, 0x723a1746b21dc0bc, 0xc7caa854f5d7cdd3, 0x7cac32883d261d9c,
-	0x7690c26423ba942c, 0x17e55524478042b8, 0xe0be477656a2389f, 0x4d289b5e67ab2da0,
-	0x44862b9c8fbbfd31, 0xb47cc8049d141365, 0x822c1b362b91c793, 0x4eb14655fb13dfd8,
-	0x1ecbba0714e2a97b, 0x6143459d5cde5f14, 0x53a8fbf1d5f0ac89, 0x97ea04d81c5e5b00,
-	0x622181a8d4fdb3f3, 0xe9bcd341572a1208, 0x1411258643cce58a, 0x9144c5fea4c6e0a4,
-	0x0d33d06565cf620f, 0x54a48d489f219ca1, 0xc43e5eac6d63c821, 0xa9728b3a72770daf,
-	0xd7934e7b20df87ef, 0xe35503b61a3e86e5, 0xcae321fbc819d504, 0x129a50b3ac60bfa6,
-	0xcd5e68ea7e9fb6c3, 0xb01c90199483b1c7, 0x3de93cd5c295376c, 0xaed52edf2ab9ad13,
-	0x2e60f512c0a07884, 0xbc3d86a3e36210c9, 0x35269d9b163951ce, 0x0c7d6e2ad0cdb5fa,
-	0x59e86297d87f5733, 0x298ef221898db0e7, 0x55000029d1a5aa7e, 0x8bc08ae1b5061b45,
-	0xc2c31c2b6c92703a, 0x94cc596baf25ef42, 0x0a1d73db22540456, 0x04b6a0f9d9c4179a,
-	0xeffdafa2ae3d3c60, 0xf7c8075bb49496c4, 0x9cc5c7141d1cd4e3, 0x78bd1638218e5534,
-	0xb2f11568f850246a, 0xedfabcfa9502bc29, 0x796ce5f2da23051b, 0xaae128b0dc93537c,
-	0x3a493da0ee4b29ae, 0xb5df6b2c416895d7, 0xfcabbd25122d7f37, 0x70810b58105dc4b1,
-	0xe10fdd37f7882a90, 0x524dcab5518a3f5c, 0x3c9e85878451255b, 0x4029828119bd34e2,
-	0x74a05b6f5d3ceccb, 0xb610021542e13eca, 0x0ff979d12f59e2ac, 0x6037da27e4f9cc50,
-	0x5e92975a0df1847d, 0xd66de190d3e623fe, 0x5032d6b87b568048, 0x9a36b7ce8235216e,
-	0x80272a7a24f64b4a, 0x93efed8b8c6916f7, 0x37ddbff44cce1555, 0x4b95db5d4b99bd25,
-	0x92d3fda169812fc0, 0xfb1a4a9a90660bb6, 0x730c196946a4b9b2, 0x81e289aa7f49da68,
-	0x64669a0f83b1a05f, 0x27b3ff7d9644f48b, 0xcc6b615c8db675b3, 0x674f20b9bcebbe95,
-	0x6f31238275655982, 0x5ae488713e45cf05, 0xbf619f9954c21157, 0xeabac46040a8eae9,
-	0x454c6fe9f2c0c1cd, 0x419cf6496412691c, 0xd3dc3bef265b0f70, 0x6d0e60f5c3578a9e,
-}
-
-T4 := [?]u64 {
-	0x5b0e608526323c55, 0x1a46c1a9fa1b59f5, 0xa9e245a17c4c8ffa, 0x65ca5159db2955d7,
-	0x05db0a76ce35afc2, 0x81eac77ea9113d45, 0x528ef88ab6ac0a0d, 0xa09ea253597be3ff,
-	0x430ddfb3ac48cd56, 0xc4b3a67af45ce46f, 0x4ececfd8fbe2d05e, 0x3ef56f10b39935f0,
-	0x0b22d6829cd619c6, 0x17fd460a74df2069, 0x6cf8cc8e8510ed40, 0xd6c824bf3a6ecaa7,
-	0x61243d581a817049, 0x048bacb6bbc163a2, 0xd9a38ac27d44cc32, 0x7fddff5baaf410ab,
-	0xad6d495aa804824b, 0xe1a6a74f2d8c9f94, 0xd4f7851235dee8e3, 0xfd4b7f886540d893,
-	0x247c20042aa4bfda, 0x096ea1c517d1327c, 0xd56966b4361a6685, 0x277da5c31221057d,
-	0x94d59893a43acff7, 0x64f0c51ccdc02281, 0x3d33bcc4ff6189db, 0xe005cb184ce66af1,
-	0xff5ccd1d1db99bea, 0xb0b854a7fe42980f, 0x7bd46a6a718d4b9f, 0xd10fa8cc22a5fd8c,
-	0xd31484952be4bd31, 0xc7fa975fcb243847, 0x4886ed1e5846c407, 0x28cddb791eb70b04,
-	0xc2b00be2f573417f, 0x5c9590452180f877, 0x7a6bddfff370eb00, 0xce509e38d6d9d6a4,
-	0xebeb0f00647fa702, 0x1dcc06cf76606f06, 0xe4d9f28ba286ff0a, 0xd85a305dc918c262,
-	0x475b1d8732225f54, 0x2d4fb51668ccb5fe, 0xa679b9d9d72bba20, 0x53841c0d912d43a5,
-	0x3b7eaa48bf12a4e8, 0x781e0e47f22f1ddf, 0xeff20ce60ab50973, 0x20d261d19dffb742,
-	0x16a12b03062a2e39, 0x1960eb2239650495, 0x251c16fed50eb8b8, 0x9ac0c330f826016e,
-	0xed152665953e7671, 0x02d63194a6369570, 0x5074f08394b1c987, 0x70ba598c90b25ce1,
-	0x794a15810b9742f6, 0x0d5925e9fcaf8c6c, 0x3067716cd868744e, 0x910ab077e8d7731b,
-	0x6a61bbdb5ac42f61, 0x93513efbf0851567, 0xf494724b9e83e9d5, 0xe887e1985c09648d,
-	0x34b1d3c675370cfd, 0xdc35e433bc0d255d, 0xd0aab84234131be0, 0x08042a50b48b7eaf,
-	0x9997c4ee44a3ab35, 0x829a7b49201799d0, 0x263b8307b7c54441, 0x752f95f4fd6a6ca6,
-	0x927217402c08c6e5, 0x2a8ab754a795d9ee, 0xa442f7552f72943d, 0x2c31334e19781208,
-	0x4fa98d7ceaee6291, 0x55c3862f665db309, 0xbd0610175d53b1f3, 0x46fe6cb840413f27,
-	0x3fe03792df0cfa59, 0xcfe700372eb85e8f, 0xa7be29e7adbce118, 0xe544ee5cde8431dd,
-	0x8a781b1b41f1873e, 0xa5c94c78a0d2f0e7, 0x39412e2877b60728, 0xa1265ef3afc9a62c,
-	0xbcc2770c6a2506c5, 0x3ab66dd5dce1ce12, 0xe65499d04a675b37, 0x7d8f523481bfd216,
-	0x0f6f64fcec15f389, 0x74efbe618b5b13c8, 0xacdc82b714273e1d, 0xdd40bfe003199d17,
-	0x37e99257e7e061f8, 0xfa52626904775aaa, 0x8bbbf63a463d56f9, 0xf0013f1543a26e64,
-	0xa8307e9f879ec898, 0xcc4c27a4150177cc, 0x1b432f2cca1d3348, 0xde1d1f8f9f6fa013,
-	0x606602a047a7ddd6, 0xd237ab64cc1cb2c7, 0x9b938e7225fcd1d3, 0xec4e03708e0ff476,
-	0xfeb2fbda3d03c12d, 0xae0bced2ee43889a, 0x22cb8923ebfb4f43, 0x69360d013cf7396d,
-	0x855e3602d2d4e022, 0x073805bad01f784c, 0x33e17a133852f546, 0xdf4874058ac7b638,
-	0xba92b29c678aa14a, 0x0ce89fc76cfaadcd, 0x5f9d4e0908339e34, 0xf1afe9291f5923b9,
-	0x6e3480f60f4a265f, 0xeebf3a2ab29b841c, 0xe21938a88f91b4ad, 0x57dfeff845c6d3c3,
-	0x2f006b0bf62caaf2, 0x62f479ef6f75ee78, 0x11a55ad41c8916a9, 0xf229d29084fed453,
-	0x42f1c27b16b000e6, 0x2b1f76749823c074, 0x4b76eca3c2745360, 0x8c98f463b91691bd,
-	0x14bcc93cf1ade66a, 0x8885213e6d458397, 0x8e177df0274d4711, 0xb49b73b5503f2951,
-	0x10168168c3f96b6b, 0x0e3d963b63cab0ae, 0x8dfc4b5655a1db14, 0xf789f1356e14de5c,
-	0x683e68af4e51dac1, 0xc9a84f9d8d4b0fd9, 0x3691e03f52a0f9d1, 0x5ed86e46e1878e80,
-	0x3c711a0e99d07150, 0x5a0865b20c4e9310, 0x56fbfc1fe4f0682e, 0xea8d5de3105edf9b,
-	0x71abfdb12379187a, 0x2eb99de1bee77b9c, 0x21ecc0ea33cf4523, 0x59a4d7521805c7a1,
-	0x3896f5eb56ae7c72, 0xaa638f3db18f75dc, 0x9f39358dabe9808e, 0xb7defa91c00b72ac,
-	0x6b5541fd62492d92, 0x6dc6dee8f92e4d5b, 0x353f57abc4beea7e, 0x735769d6da5690ce,
-	0x0a234aa642391484, 0xf6f9508028f80d9d, 0xb8e319a27ab3f215, 0x31ad9c1151341a4d,
-	0x773c22a57bef5805, 0x45c7561a07968633, 0xf913da9e249dbe36, 0xda652d9b78a64c68,
-	0x4c27a97f3bc334ef, 0x76621220e66b17f4, 0x967743899acd7d0b, 0xf3ee5bcae0ed6782,
-	0x409f753600c879fc, 0x06d09a39b5926db6, 0x6f83aeb0317ac588, 0x01e6ca4a86381f21,
-	0x66ff3462d19f3025, 0x72207c24ddfd3bfb, 0x4af6b6d3e2ece2eb, 0x9c994dbec7ea08de,
-	0x49ace597b09a8bc4, 0xb38c4766cf0797ba, 0x131b9373c57c2a75, 0xb1822cce61931e58,
-	0x9d7555b909ba1c0c, 0x127fafdd937d11d2, 0x29da3badc66d92e4, 0xa2c1d57154c2ecbc,
-	0x58c5134d82f6fe24, 0x1c3ae3515b62274f, 0xe907c82e01cb8126, 0xf8ed091913e37fcb,
-	0x3249d8f9c80046c9, 0x80cf9bede388fb63, 0x1881539a116cf19e, 0x5103f3f76bd52457,
-	0x15b7e6f5ae47f7a8, 0xdbd7c6ded47e9ccf, 0x44e55c410228bb1a, 0xb647d4255edb4e99,
-	0x5d11882bb8aafc30, 0xf5098bbb29d3212a, 0x8fb5ea14e90296b3, 0x677b942157dd025a,
-	0xfb58e7c0a390acb5, 0x89d3674c83bd4a01, 0x9e2da4df4bf3b93b, 0xfcc41e328cab4829,
-	0x03f38c96ba582c52, 0xcad1bdbd7fd85db2, 0xbbb442c16082ae83, 0xb95fe86ba5da9ab0,
-	0xb22e04673771a93f, 0x845358c9493152d8, 0xbe2a488697b4541e, 0x95a2dc2dd38e6966,
-	0xc02c11ac923c852b, 0x2388b1990df2a87b, 0x7c8008fa1b4f37be, 0x1f70d0c84d54e503,
-	0x5490adec7ece57d4, 0x002b3c27d9063a3a, 0x7eaea3848030a2bf, 0xc602326ded2003c0,
-	0x83a7287d69a94086, 0xc57a5fcb30f57a8a, 0xb56844e479ebe779, 0xa373b40f05dcbce9,
-	0xd71a786e88570ee2, 0x879cbacdbde8f6a0, 0x976ad1bcc164a32f, 0xab21e25e9666d78b,
-	0x901063aae5e5c33c, 0x9818b34448698d90, 0xe36487ae3e1e8abb, 0xafbdf931893bdcb4,
-	0x6345a0dc5fbbd519, 0x8628fe269b9465ca, 0x1e5d01603f9c51ec, 0x4de44006a15049b7,
-	0xbf6c70e5f776cbb1, 0x411218f2ef552bed, 0xcb0c0708705a36a3, 0xe74d14754f986044,
-	0xcd56d9430ea8280e, 0xc12591d7535f5065, 0xc83223f1720aef96, 0xc3a0396f7363a51f,
-}
-
-Tiger_Context :: struct {
-    a:      u64,
-	b:      u64,
-	c:      u64,
-	x:      [64]byte,
-	nx:     int,
-	length: u64,
-	ver:    int,
-}
-
-round :: #force_inline proc "contextless" (a, b, c, x, mul: u64) -> (u64, u64, u64) {
-	a, b, c := a, b, c
-	c ~= x
-	a -= T1[c & 0xff] ~ T2[(c >> 16) & 0xff] ~ T3[(c >> 32) & 0xff] ~ T4[(c >> 48) & 0xff]
-	b += T4[(c >> 8) & 0xff] ~ T3[(c >> 24) & 0xff] ~ T2[(c >> 40) & 0xff] ~ T1[(c >> 56) & 0xff]
-	b *= mul
-	return a, b, c
-}
-
-pass :: #force_inline proc "contextless" (a, b, c: u64, d: []u64, mul: u64) -> (x, y, z: u64) {
-	x, y, z = round(a, b, c, d[0], mul)
-	y, z, x = round(y, z, x, d[1], mul)
-	z, x, y = round(z, x, y, d[2], mul)
-	x, y, z = round(x, y, z, d[3], mul)
-	y, z, x = round(y, z, x, d[4], mul)
-	z, x, y = round(z, x, y, d[5], mul)
-	x, y, z = round(x, y, z, d[6], mul)
-	y, z, x = round(y, z, x, d[7], mul)
-	return
-}
-
-key_schedule :: #force_inline proc "contextless" (x: []u64) {
-	x[0] -= x[7] ~ 0xa5a5a5a5a5a5a5a5
-	x[1] ~= x[0]
-	x[2] += x[1]
-	x[3] -= x[2] ~ ((~x[1]) << 19)
-	x[4] ~= x[3]
-	x[5] += x[4]
-	x[6] -= x[5] ~ ((~x[4]) >> 23)
-	x[7] ~= x[6]
-	x[0] += x[7]
-	x[1] -= x[0] ~ ((~x[7]) << 19)
-	x[2] ~= x[1]
-	x[3] += x[2]
-	x[4] -= x[3] ~ ((~x[2]) >> 23)
-	x[5] ~= x[4]
-	x[6] += x[5]
-	x[7] -= x[6] ~ 0x0123456789abcdef
-}
-
-compress :: #force_inline proc "contextless" (ctx: ^Tiger_Context, data: []byte) {
-	a := ctx.a
-	b := ctx.b
-	c := ctx.c
-	x := util.cast_slice([]u64, data)
-	ctx.a, ctx.b, ctx.c = pass(ctx.a, ctx.b, ctx.c, x, 5)
-	key_schedule(x)
-	ctx.c, ctx.a, ctx.b = pass(ctx.c, ctx.a, ctx.b, x, 7)
-	key_schedule(x)
-	ctx.b, ctx.c, ctx.a = pass(ctx.b, ctx.c, ctx.a, x, 9)
-	ctx.a ~= a
-	ctx.b -= b
-	ctx.c += c
-}
-
-init :: proc "contextless" (ctx: ^Tiger_Context) {
-	ctx.a = 0x0123456789abcdef
-	ctx.b = 0xfedcba9876543210
-	ctx.c = 0xf096a5b4c3b2e187
-}
-
-update :: proc(ctx: ^Tiger_Context, input: []byte) {
-	p := make([]byte, len(input))
-	copy(p, input)
-
-	length     := len(p)
-	ctx.length += u64(length)
-	if ctx.nx > 0 {
-		n := len(p)
-		if n > 64 - ctx.nx {
-			n = 64 - ctx.nx
-		}
-		copy(ctx.x[ctx.nx:ctx.nx + n], p[:n])
-		ctx.nx += n
-		if ctx.nx == 64 {
-			compress(ctx, ctx.x[:64 - 1])
-			ctx.nx = 0
-		}
-		p = p[n:]
-	}
-	for len(p) >= 64 {
-		compress(ctx, p[:64])
-		p = p[64:]
-	}
-	if len(p) > 0 {
-		ctx.nx = copy(ctx.x[:], p)
-	}
-}
-
-final :: proc(ctx: ^Tiger_Context, hash: []byte) {
-	length := ctx.length
-	tmp: [64]byte
-	if ctx.ver == 1 {
-		tmp[0] = 0x01
-	} else {
-		tmp[0] = 0x80
-	}
-
-	size := length & 0x3f
-	if size < 56 {
-		update(ctx, tmp[:56 - size])
-	} else {
-		update(ctx, tmp[:64 + 56 - size])
-	}
-
-	length <<= 3
-	for i := uint(0); i < 8; i += 1 {
-		tmp[i] = byte(length >> (8 * i))
-	}
-	update(ctx, tmp[:8])
-
-	for i := uint(0); i < 8; i += 1 {
-		tmp[i]      = byte(ctx.a >> (8 * i))
-		tmp[i + 8]  = byte(ctx.b >> (8 * i))
-		tmp[i + 16] = byte(ctx.c >> (8 * i))
-	}
-	copy(hash[:], tmp[:len(hash)])
-}
@@ -1,726 +0,0 @@
-package blake
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation of the BLAKE hashing algorithm, as defined in <https://web.archive.org/web/20190915215948/https://131002.net/blake>
-*/
-
-import "core:os"
-import "core:io"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE_224 :: 28
-DIGEST_SIZE_256 :: 32
-DIGEST_SIZE_384 :: 48
-DIGEST_SIZE_512 :: 64
-
-// hash_string_224 will hash the given input and return the
-// computed hash
-hash_string_224 :: proc "contextless" (data: string) -> [DIGEST_SIZE_224]byte {
-    return hash_bytes_224(transmute([]byte)(data))
-}
-
-// hash_bytes_224 will hash the given input and return the
-// computed hash
-hash_bytes_224 :: proc "contextless" (data: []byte) -> [DIGEST_SIZE_224]byte {
-    hash: [DIGEST_SIZE_224]byte
-    ctx: Blake256_Context
-    ctx.is224 = true
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_224 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_224 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_224(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_224 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_224 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_224, "Size of destination buffer is smaller than the digest size")
-    ctx: Blake256_Context
-    ctx.is224 = true
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_224 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
-    hash: [DIGEST_SIZE_224]byte
-    ctx: Blake256_Context
-    ctx.is224 = true
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_224 will read the file provided by the given handle
-// and compute a hash
-hash_file_224 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_224]byte, bool) {
-    if !load_at_once {
-        return hash_stream_224(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_224(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_224]byte{}, false
-}
-
-hash_224 :: proc {
-    hash_stream_224,
-    hash_file_224,
-    hash_bytes_224,
-    hash_string_224,
-    hash_bytes_to_buffer_224,
-    hash_string_to_buffer_224,
-}
-
-// hash_string_256 will hash the given input and return the
-// computed hash
-hash_string_256 :: proc "contextless" (data: string) -> [DIGEST_SIZE_256]byte {
-    return hash_bytes_256(transmute([]byte)(data))
-}
-
-// hash_bytes_256 will hash the given input and return the
-// computed hash
-hash_bytes_256 :: proc "contextless" (data: []byte) -> [DIGEST_SIZE_256]byte {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: Blake256_Context
-    ctx.is224 = false
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_256 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_256 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_256(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_256 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_256 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_256, "Size of destination buffer is smaller than the digest size")
-    ctx: Blake256_Context
-    ctx.is224 = false
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_256 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: Blake256_Context
-    ctx.is224 = false
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_256 will read the file provided by the given handle
-// and compute a hash
-hash_file_256 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_256]byte, bool) {
-    if !load_at_once {
-        return hash_stream_256(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_256(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_256]byte{}, false
-}
-
-hash_256 :: proc {
-    hash_stream_256,
-    hash_file_256,
-    hash_bytes_256,
-    hash_string_256,
-    hash_bytes_to_buffer_256,
-    hash_string_to_buffer_256,
-}
-
-// hash_string_384 will hash the given input and return the
-// computed hash
-hash_string_384 :: proc "contextless" (data: string) -> [DIGEST_SIZE_384]byte {
-    return hash_bytes_384(transmute([]byte)(data))
-}
-
-// hash_bytes_384 will hash the given input and return the
-// computed hash
-hash_bytes_384 :: proc "contextless" (data: []byte) -> [DIGEST_SIZE_384]byte {
-    hash: [DIGEST_SIZE_384]byte
-    ctx: Blake512_Context
-    ctx.is384 = true
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_384 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_384 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_384(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_384 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_384 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_384, "Size of destination buffer is smaller than the digest size")
-    ctx: Blake512_Context
-    ctx.is384 = true
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_384 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
-    hash: [DIGEST_SIZE_384]byte
-    ctx: Blake512_Context
-    ctx.is384 = true
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_384 will read the file provided by the given handle
-// and compute a hash
-hash_file_384 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_384]byte, bool) {
-    if !load_at_once {
-        return hash_stream_384(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_384(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_384]byte{}, false
-}
-
-hash_384 :: proc {
-    hash_stream_384,
-    hash_file_384,
-    hash_bytes_384,
-    hash_string_384,
-    hash_bytes_to_buffer_384,
-    hash_string_to_buffer_384,
-}
-
-// hash_string_512 will hash the given input and return the
-// computed hash
-hash_string_512 :: proc "contextless" (data: string) -> [DIGEST_SIZE_512]byte {
-    return hash_bytes_512(transmute([]byte)(data))
-}
-
-// hash_bytes_512 will hash the given input and return the
-// computed hash
-hash_bytes_512 :: proc "contextless" (data: []byte) -> [DIGEST_SIZE_512]byte {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: Blake512_Context
-    ctx.is384 = false
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_512 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_512 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_512(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_512 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_512 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_512, "Size of destination buffer is smaller than the digest size")
-    ctx: Blake512_Context
-    ctx.is384 = false
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_512 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: Blake512_Context
-    ctx.is384 = false
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_512 will read the file provided by the given handle
-// and compute a hash
-hash_file_512 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_512]byte, bool) {
-    if !load_at_once {
-        return hash_stream_512(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_512(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_512]byte{}, false
-}
-
-hash_512 :: proc {
-    hash_stream_512,
-    hash_file_512,
-    hash_bytes_512,
-    hash_string_512,
-    hash_bytes_to_buffer_512,
-    hash_string_to_buffer_512,
-}
-
-/*
-    Low level API
-*/
-
-init :: proc "contextless" (ctx: ^$T) {
-    when T == Blake256_Context {
-        if ctx.is224 {
-            ctx.h[0] = 0xc1059ed8
-            ctx.h[1] = 0x367cd507
-            ctx.h[2] = 0x3070dd17
-            ctx.h[3] = 0xf70e5939
-            ctx.h[4] = 0xffc00b31
-            ctx.h[5] = 0x68581511
-            ctx.h[6] = 0x64f98fa7
-            ctx.h[7] = 0xbefa4fa4
-        } else {
-            ctx.h[0] = 0x6a09e667
-            ctx.h[1] = 0xbb67ae85
-            ctx.h[2] = 0x3c6ef372
-            ctx.h[3] = 0xa54ff53a
-            ctx.h[4] = 0x510e527f
-            ctx.h[5] = 0x9b05688c
-            ctx.h[6] = 0x1f83d9ab
-            ctx.h[7] = 0x5be0cd19
-        }
-    } else when T == Blake512_Context {
-        if ctx.is384 {
-            ctx.h[0] = 0xcbbb9d5dc1059ed8
-            ctx.h[1] = 0x629a292a367cd507
-            ctx.h[2] = 0x9159015a3070dd17
-            ctx.h[3] = 0x152fecd8f70e5939
-            ctx.h[4] = 0x67332667ffc00b31
-            ctx.h[5] = 0x8eb44a8768581511
-            ctx.h[6] = 0xdb0c2e0d64f98fa7
-            ctx.h[7] = 0x47b5481dbefa4fa4
-        } else {
-            ctx.h[0] = 0x6a09e667f3bcc908
-            ctx.h[1] = 0xbb67ae8584caa73b
-            ctx.h[2] = 0x3c6ef372fe94f82b
-            ctx.h[3] = 0xa54ff53a5f1d36f1
-            ctx.h[4] = 0x510e527fade682d1
-            ctx.h[5] = 0x9b05688c2b3e6c1f
-            ctx.h[6] = 0x1f83d9abfb41bd6b
-            ctx.h[7] = 0x5be0cd19137e2179
-        }
-    }
-}
-
-update :: proc "contextless" (ctx: ^$T, data: []byte) {
-    data := data
-    when T == Blake256_Context {
-        if ctx.nx > 0 {
-            n := copy(ctx.x[ctx.nx:], data)
-            ctx.nx += n
-            if ctx.nx == BLOCKSIZE_256 {
-                block256(ctx, ctx.x[:])
-                ctx.nx = 0
-            }
-            data = data[n:]
-        }
-        if len(data) >= BLOCKSIZE_256 {
-            n := len(data) &~ (BLOCKSIZE_256 - 1)
-            block256(ctx, data[:n])
-            data = data[n:]
-        }
-        if len(data) > 0 {
-            ctx.nx = copy(ctx.x[:], data)
-        }
-    } else when T == Blake512_Context {
-        if ctx.nx > 0 {
-            n := copy(ctx.x[ctx.nx:], data)
-            ctx.nx += n
-            if ctx.nx == BLOCKSIZE_512 {
-                block512(ctx, ctx.x[:])
-                ctx.nx = 0
-            }
-            data = data[n:]
-        }
-        if len(data) >= BLOCKSIZE_512 {
-            n := len(data) &~ (BLOCKSIZE_512 - 1)
-            block512(ctx, data[:n])
-            data = data[n:]
-        }
-        if len(data) > 0 {
-            ctx.nx = copy(ctx.x[:], data)
-        }
-    }
-}
-
-final :: proc "contextless" (ctx: ^$T, hash: []byte) {
-    when T == Blake256_Context {
-        tmp: [65]byte
-    } else when T == Blake512_Context {
-        tmp: [129]byte
-    }
-    nx     := u64(ctx.nx)
-    tmp[0]  = 0x80
-    length := (ctx.t + nx) << 3
-
-    when T == Blake256_Context {
-        if nx == 55 {
-            if ctx.is224 {
-                write_additional(ctx, {0x80})
-            } else {
-                write_additional(ctx, {0x81})
-            }
-        } else {
-            if nx < 55 {
-                if nx == 0 {
-                    ctx.nullt = true
-                }
-                write_additional(ctx, tmp[0 : 55 - nx])
-            } else { 
-                write_additional(ctx, tmp[0 : 64 - nx])
-                write_additional(ctx, tmp[1:56])
-                ctx.nullt = true
-            }
-            if ctx.is224 {
-                write_additional(ctx, {0x00})
-            } else {
-                write_additional(ctx, {0x01})
-            }
-        }
-
-        for i : uint = 0; i < 8; i += 1 {
-            tmp[i] = byte(length >> (56 - 8 * i))
-        }
-        write_additional(ctx, tmp[0:8])
-
-        h := ctx.h[:]
-        if ctx.is224 {
-            h = h[0:7]
-        }
-        for s, i in h {
-            hash[i * 4]     = byte(s >> 24)
-            hash[i * 4 + 1] = byte(s >> 16)
-            hash[i * 4 + 2] = byte(s >> 8)
-            hash[i * 4 + 3] = byte(s)
-        }
-    } else when T == Blake512_Context {
-        if nx == 111 {
-            if ctx.is384 {
-                write_additional(ctx, {0x80})
-            } else {
-                write_additional(ctx, {0x81})
-            }
-        } else {
-            if nx < 111 {
-                if nx == 0 {
-                    ctx.nullt = true
-                }
-                write_additional(ctx, tmp[0 : 111 - nx])
-            } else { 
-                write_additional(ctx, tmp[0 : 128 - nx])
-                write_additional(ctx, tmp[1:112])
-                ctx.nullt = true
-            }
-            if ctx.is384 {
-                write_additional(ctx, {0x00})
-            } else {
-                write_additional(ctx, {0x01})
-            }
-        }
-
-        for i : uint = 0; i < 16; i += 1 {
-            tmp[i] = byte(length >> (120 - 8 * i))
-        }
-        write_additional(ctx, tmp[0:16])
-
-        h := ctx.h[:]
-        if ctx.is384 {
-            h = h[0:6]
-        }
-        for s, i in h {
-            hash[i * 8]     = byte(s >> 56)
-            hash[i * 8 + 1] = byte(s >> 48)
-            hash[i * 8 + 2] = byte(s >> 40)
-            hash[i * 8 + 3] = byte(s >> 32)
-            hash[i * 8 + 4] = byte(s >> 24)
-            hash[i * 8 + 5] = byte(s >> 16)
-            hash[i * 8 + 6] = byte(s >> 8)
-            hash[i * 8 + 7] = byte(s)
-        }
-    }
-}
-
-SIZE_224 :: 28
-SIZE_256 :: 32
-SIZE_384 :: 48
-SIZE_512 :: 64
-BLOCKSIZE_256 :: 64
-BLOCKSIZE_512 :: 128
-
-Blake256_Context :: struct {
-    h:     [8]u32,
-    s:     [4]u32,
-    t:     u64,
-    x:     [64]byte,
-    nx:    int,
-    is224: bool,
-    nullt: bool,
-}
-
-Blake512_Context :: struct {
-    h:     [8]u64,
-    s:     [4]u64,
-    t:     u64,
-    x:     [128]byte,
-    nx:    int,
-    is384: bool,
-    nullt: bool,
-}
-
-SIGMA := [?]int {
-    0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,
-    14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3,
-    11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4,
-    7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8,
-    9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13,
-    2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9,
-    12, 5, 1, 15, 14, 13, 4, 10, 0, 7, 6, 3, 9, 2, 8, 11,
-    13, 11, 7, 14, 12, 1, 3, 9, 5, 0, 15, 4, 8, 6, 2, 10,
-    6, 15, 14, 9, 11, 3, 0, 8, 12, 2, 13, 7, 1, 4, 10, 5,
-    10, 2, 8, 4, 7, 6, 1, 5, 15, 11, 9, 14, 3, 12, 13, 0,
-}
-
-U256 := [16]u32 {
-    0x243f6a88, 0x85a308d3, 0x13198a2e, 0x03707344,
-    0xa4093822, 0x299f31d0, 0x082efa98, 0xec4e6c89,
-    0x452821e6, 0x38d01377, 0xbe5466cf, 0x34e90c6c,
-    0xc0ac29b7, 0xc97c50dd, 0x3f84d5b5, 0xb5470917,
-}
-
-U512 := [16]u64 {
-    0x243f6a8885a308d3, 0x13198a2e03707344, 0xa4093822299f31d0, 0x082efa98ec4e6c89,
-    0x452821e638d01377, 0xbe5466cf34e90c6c, 0xc0ac29b7c97c50dd, 0x3f84d5b5b5470917,
-    0x9216d5d98979fb1b, 0xd1310ba698dfb5ac, 0x2ffd72dbd01adfb7, 0xb8e1afed6a267e96,
-    0xba7c9045f12c7f99, 0x24a19947b3916cf7, 0x0801f2e2858efc16, 0x636920d871574e69,
-}
-
-G256 :: #force_inline proc "contextless" (a, b, c, d: u32, m: [16]u32, i, j: int) -> (u32, u32, u32, u32) {
-    a, b, c, d := a, b, c, d
-    a += m[SIGMA[(i % 10) * 16 + (2 * j)]] ~ U256[SIGMA[(i % 10) * 16 + (2 * j + 1)]]
-    a += b
-    d ~= a
-    d = d << (32 - 16) | d >> 16
-    c += d
-    b ~= c
-    b = b << (32 - 12) | b >> 12
-    a += m[SIGMA[(i % 10) * 16 + (2 * j + 1)]] ~ U256[SIGMA[(i % 10) * 16 + (2 * j)]]
-    a += b
-    d ~= a
-    d = d << (32 - 8) | d >> 8
-    c += d
-    b ~= c
-    b = b << (32 - 7) | b >> 7
-    return a, b, c, d
-}
-
-G512 :: #force_inline proc "contextless" (a, b, c, d: u64, m: [16]u64, i, j: int) -> (u64, u64, u64, u64) {
-    a, b, c, d := a, b, c, d
-    a += m[SIGMA[(i % 10) * 16 + (2 * j)]] ~ U512[SIGMA[(i % 10) * 16 + (2 * j + 1)]]
-    a += b
-    d ~= a
-    d = d << (64 - 32) | d >> 32
-    c += d
-    b ~= c
-    b = b << (64 - 25) | b >> 25
-    a += m[SIGMA[(i % 10) * 16 + (2 * j + 1)]] ~ U512[SIGMA[(i % 10) * 16 + (2 * j)]]
-    a += b
-    d ~= a
-    d = d << (64 - 16) | d >> 16
-    c += d
-    b ~= c
-    b = b << (64 - 11) | b >> 11
-    return a, b, c, d
-}
-
-block256 :: proc "contextless" (ctx: ^Blake256_Context, p: []byte) #no_bounds_check {
-    i, j: int     = ---, ---
-    v, m: [16]u32 = ---, ---
-    p := p
-    for len(p) >= BLOCKSIZE_256 {
-        v[0]  = ctx.h[0]
-        v[1]  = ctx.h[1]
-        v[2]  = ctx.h[2]
-        v[3]  = ctx.h[3]
-        v[4]  = ctx.h[4]
-        v[5]  = ctx.h[5]
-        v[6]  = ctx.h[6]
-        v[7]  = ctx.h[7]
-        v[8]  = ctx.s[0] ~ U256[0]
-        v[9]  = ctx.s[1] ~ U256[1]
-        v[10] = ctx.s[2] ~ U256[2]
-        v[11] = ctx.s[3] ~ U256[3]
-        v[12] = U256[4]
-        v[13] = U256[5]
-        v[14] = U256[6]
-        v[15] = U256[7]
-
-        ctx.t += 512
-        if !ctx.nullt {
-            v[12] ~= u32(ctx.t)
-            v[13] ~= u32(ctx.t)
-            v[14] ~= u32(ctx.t >> 32)
-            v[15] ~= u32(ctx.t >> 32)
-        }
-
-        for i, j = 0, 0; i < 16; i, j = i+1, j+4 {
-            m[i] = u32(p[j]) << 24 | u32(p[j + 1]) << 16 | u32(p[j + 2]) << 8 | u32(p[j + 3])
-        }
-
-        for i = 0; i < 14; i += 1 {
-            v[0], v[4], v[8],  v[12] = G256(v[0], v[4], v[8],  v[12], m, i, 0)
-            v[1], v[5], v[9],  v[13] = G256(v[1], v[5], v[9],  v[13], m, i, 1)
-            v[2], v[6], v[10], v[14] = G256(v[2], v[6], v[10], v[14], m, i, 2)
-            v[3], v[7], v[11], v[15] = G256(v[3], v[7], v[11], v[15], m, i, 3)
-            v[0], v[5], v[10], v[15] = G256(v[0], v[5], v[10], v[15], m, i, 4)
-            v[1], v[6], v[11], v[12] = G256(v[1], v[6], v[11], v[12], m, i, 5)
-            v[2], v[7], v[8],  v[13] = G256(v[2], v[7], v[8],  v[13], m, i, 6)
-            v[3], v[4], v[9],  v[14] = G256(v[3], v[4], v[9],  v[14], m, i, 7)
-        }
-
-        for i = 0; i < 8; i += 1 {
-            ctx.h[i] ~= ctx.s[i % 4] ~ v[i] ~ v[i + 8]
-        }
-        p = p[BLOCKSIZE_256:]
-    }
-}
-
-block512 :: proc "contextless" (ctx: ^Blake512_Context, p: []byte) #no_bounds_check {
-    i, j: int     = ---, ---
-    v, m: [16]u64 = ---, ---
-    p := p
-    for len(p) >= BLOCKSIZE_512 {
-        v[0]  = ctx.h[0]
-        v[1]  = ctx.h[1]
-        v[2]  = ctx.h[2]
-        v[3]  = ctx.h[3]
-        v[4]  = ctx.h[4]
-        v[5]  = ctx.h[5]
-        v[6]  = ctx.h[6]
-        v[7]  = ctx.h[7]
-        v[8]  = ctx.s[0] ~ U512[0]
-        v[9]  = ctx.s[1] ~ U512[1]
-        v[10] = ctx.s[2] ~ U512[2]
-        v[11] = ctx.s[3] ~ U512[3]
-        v[12] = U512[4]
-        v[13] = U512[5]
-        v[14] = U512[6]
-        v[15] = U512[7]
-
-        ctx.t += 1024
-        if !ctx.nullt {
-            v[12] ~= ctx.t
-            v[13] ~= ctx.t
-            v[14] ~= 0
-            v[15] ~= 0
-        }
-
-        for i, j = 0, 0; i < 16; i, j = i + 1, j + 8 {
-            m[i] = u64(p[j]) << 56     | u64(p[j + 1]) << 48 | u64(p[j + 2]) << 40 | u64(p[j + 3]) << 32 | 
-                   u64(p[j + 4]) << 24 | u64(p[j + 5]) << 16 | u64(p[j + 6]) << 8  | u64(p[j + 7])
-        }
-        for i = 0; i < 16; i += 1 {
-            v[0], v[4], v[8],  v[12] = G512(v[0], v[4], v[8],  v[12], m, i, 0)
-            v[1], v[5], v[9],  v[13] = G512(v[1], v[5], v[9],  v[13], m, i, 1)
-            v[2], v[6], v[10], v[14] = G512(v[2], v[6], v[10], v[14], m, i, 2)
-            v[3], v[7], v[11], v[15] = G512(v[3], v[7], v[11], v[15], m, i, 3)
-            v[0], v[5], v[10], v[15] = G512(v[0], v[5], v[10], v[15], m, i, 4)
-            v[1], v[6], v[11], v[12] = G512(v[1], v[6], v[11], v[12], m, i, 5)
-            v[2], v[7], v[8],  v[13] = G512(v[2], v[7], v[8],  v[13], m, i, 6)
-            v[3], v[4], v[9],  v[14] = G512(v[3], v[4], v[9],  v[14], m, i, 7)
-        }
-
-        for i = 0; i < 8; i += 1 {
-            ctx.h[i] ~= ctx.s[i % 4] ~ v[i] ~ v[i + 8]
-        }
-        p = p[BLOCKSIZE_512:]
-    }
-}
-
-write_additional :: proc "contextless" (ctx: ^$T, data: []byte) {
-	ctx.t -= u64(len(data)) << 3
-    update(ctx, data)
-}
@@ -7,12 +7,12 @@ package blake2b
    List of contributors:
        zhibog, dotbmp:  Initial implementation.

-    Interface for the BLAKE2B hashing algorithm.
-    BLAKE2B and BLAKE2B share the implementation in the _blake2 package.
+    Interface for the BLAKE2b hashing algorithm.
+    BLAKE2b and BLAKE2s share the implementation in the _blake2 package.
 */

-import "core:os"
 import "core:io"
+import "core:os"

 import "../_blake2"

@@ -25,103 +25,103 @@ DIGEST_SIZE :: 64
 // hash_string will hash the given input and return the
 // computed hash
 hash_string :: proc(data: string) -> [DIGEST_SIZE]byte {
-    return hash_bytes(transmute([]byte)(data))
+	return hash_bytes(transmute([]byte)(data))
 }

 // hash_bytes will hash the given input and return the
 // computed hash
 hash_bytes :: proc(data: []byte) -> [DIGEST_SIZE]byte {
-    hash: [DIGEST_SIZE]byte
-    ctx: _blake2.Blake2b_Context
-    cfg: _blake2.Blake2_Config
-    cfg.size = _blake2.BLAKE2B_SIZE
-    ctx.cfg  = cfg
-    _blake2.init(&ctx)
-    _blake2.update(&ctx, data)
-    _blake2.final(&ctx, hash[:])
-    return hash
+	hash: [DIGEST_SIZE]byte
+	ctx: Context
+	cfg: _blake2.Blake2_Config
+	cfg.size = _blake2.BLAKE2B_SIZE
+	ctx.cfg = cfg
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
 }

 // hash_string_to_buffer will hash the given input and assign the
 // computed hash to the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_string_to_buffer :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer(transmute([]byte)(data), hash)
+	hash_bytes_to_buffer(transmute([]byte)(data), hash)
 }

 // hash_bytes_to_buffer will hash the given input and write the
 // computed hash into the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_bytes_to_buffer :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE, "Size of destination buffer is smaller than the digest size")
-    ctx: _blake2.Blake2b_Context
-    cfg: _blake2.Blake2_Config
-    cfg.size = _blake2.BLAKE2B_SIZE
-    ctx.cfg  = cfg
-    _blake2.init(&ctx)
-    _blake2.update(&ctx, data)
-    _blake2.final(&ctx, hash)
+	ctx: Context
+	cfg: _blake2.Blake2_Config
+	cfg.size = _blake2.BLAKE2B_SIZE
+	ctx.cfg = cfg
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
 }

-
 // hash_stream will read the stream in chunks and compute a
 // hash from its contents
 hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
-    hash: [DIGEST_SIZE]byte
-    ctx: _blake2.Blake2b_Context
-    cfg: _blake2.Blake2_Config
-    cfg.size = _blake2.BLAKE2B_SIZE
-    ctx.cfg  = cfg
-    _blake2.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _blake2.update(&ctx, buf[:read])
-        } 
-    }
-    _blake2.final(&ctx, hash[:])
-    return hash, true 
+	hash: [DIGEST_SIZE]byte
+	ctx: Context
+	cfg: _blake2.Blake2_Config
+	cfg.size = _blake2.BLAKE2B_SIZE
+	ctx.cfg = cfg
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
 }

 // hash_file will read the file provided by the given handle
 // and compute a hash
 hash_file :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE]byte, bool) {
-    if !load_at_once {
-        return hash_stream(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE]byte{}, false
+	if !load_at_once {
+		return hash_stream(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE]byte{}, false
 }

 hash :: proc {
-    hash_stream,
-    hash_file,
-    hash_bytes,
-    hash_string,
-    hash_bytes_to_buffer,
-    hash_string_to_buffer,
+	hash_stream,
+	hash_file,
+	hash_bytes,
+	hash_string,
+	hash_bytes_to_buffer,
+	hash_string_to_buffer,
 }

 /*
    Low level API
 */

-Blake2b_Context :: _blake2.Blake2b_Context
+Context :: _blake2.Blake2b_Context

-init :: proc(ctx: ^_blake2.Blake2b_Context) {
-    _blake2.init(ctx)
+init :: proc(ctx: ^Context) {
+	_blake2.init(ctx)
 }

-update :: proc "contextless" (ctx: ^_blake2.Blake2b_Context, data: []byte) {
-    _blake2.update(ctx, data)
+update :: proc(ctx: ^Context, data: []byte) {
+	_blake2.update(ctx, data)
 }

-final :: proc "contextless" (ctx: ^_blake2.Blake2b_Context, hash: []byte) {
-    _blake2.final(ctx, hash)
+final :: proc(ctx: ^Context, hash: []byte) {
+	_blake2.final(ctx, hash)
 }
@@ -7,12 +7,12 @@ package blake2s
    List of contributors:
        zhibog, dotbmp:  Initial implementation.

-    Interface for the BLAKE2S hashing algorithm.
-    BLAKE2B and BLAKE2B share the implementation in the _blake2 package.
+    Interface for the BLAKE2s hashing algorithm.
+    BLAKE2s and BLAKE2b share the implementation in the _blake2 package.
 */

-import "core:os"
 import "core:io"
+import "core:os"

 import "../_blake2"

@@ -25,103 +25,103 @@ DIGEST_SIZE :: 32
 // hash_string will hash the given input and return the
 // computed hash
 hash_string :: proc(data: string) -> [DIGEST_SIZE]byte {
-    return hash_bytes(transmute([]byte)(data))
+	return hash_bytes(transmute([]byte)(data))
 }

 // hash_bytes will hash the given input and return the
 // computed hash
 hash_bytes :: proc(data: []byte) -> [DIGEST_SIZE]byte {
-    hash: [DIGEST_SIZE]byte
-    ctx: _blake2.Blake2s_Context
-    cfg: _blake2.Blake2_Config
-    cfg.size = _blake2.BLAKE2S_SIZE
-    ctx.cfg  = cfg
-    _blake2.init(&ctx)
-    _blake2.update(&ctx, data)
-    _blake2.final(&ctx, hash[:])
-    return hash
+	hash: [DIGEST_SIZE]byte
+	ctx: Context
+	cfg: _blake2.Blake2_Config
+	cfg.size = _blake2.BLAKE2S_SIZE
+	ctx.cfg = cfg
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
 }

-
 // hash_string_to_buffer will hash the given input and assign the
 // computed hash to the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_string_to_buffer :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer(transmute([]byte)(data), hash)
+	hash_bytes_to_buffer(transmute([]byte)(data), hash)
 }

 // hash_bytes_to_buffer will hash the given input and write the
 // computed hash into the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_bytes_to_buffer :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE, "Size of destination buffer is smaller than the digest size")
-    ctx: _blake2.Blake2s_Context
-    cfg: _blake2.Blake2_Config
-    cfg.size = _blake2.BLAKE2S_SIZE
-    ctx.cfg  = cfg
-    _blake2.init(&ctx)
-    _blake2.update(&ctx, data)
-    _blake2.final(&ctx, hash)
+	ctx: Context
+	cfg: _blake2.Blake2_Config
+	cfg.size = _blake2.BLAKE2S_SIZE
+	ctx.cfg = cfg
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
 }

 // hash_stream will read the stream in chunks and compute a
 // hash from its contents
 hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
-    hash: [DIGEST_SIZE]byte
-    ctx: _blake2.Blake2s_Context
-    cfg: _blake2.Blake2_Config
-    cfg.size = _blake2.BLAKE2S_SIZE
-    ctx.cfg  = cfg
-    _blake2.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _blake2.update(&ctx, buf[:read])
-        } 
-    }
-    _blake2.final(&ctx, hash[:])
-    return hash, true 
+	hash: [DIGEST_SIZE]byte
+	ctx: Context
+	cfg: _blake2.Blake2_Config
+	cfg.size = _blake2.BLAKE2S_SIZE
+	ctx.cfg = cfg
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
 }

 // hash_file will read the file provided by the given handle
 // and compute a hash
 hash_file :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE]byte, bool) {
-    if !load_at_once {
-        return hash_stream(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE]byte{}, false
+	if !load_at_once {
+		return hash_stream(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE]byte{}, false
 }

 hash :: proc {
-    hash_stream,
-    hash_file,
-    hash_bytes,
-    hash_string,
-    hash_bytes_to_buffer,
-    hash_string_to_buffer,
+	hash_stream,
+	hash_file,
+	hash_bytes,
+	hash_string,
+	hash_bytes_to_buffer,
+	hash_string_to_buffer,
 }

 /*
    Low level API
 */

-Blake2s_Context :: _blake2.Blake2b_Context
+Context :: _blake2.Blake2s_Context

-init :: proc(ctx: ^_blake2.Blake2s_Context) {
-    _blake2.init(ctx)
+init :: proc(ctx: ^Context) {
+	_blake2.init(ctx)
 }

-update :: proc "contextless" (ctx: ^_blake2.Blake2s_Context, data: []byte) {
-    _blake2.update(ctx, data)
+update :: proc(ctx: ^Context, data: []byte) {
+	_blake2.update(ctx, data)
 }

-final :: proc "contextless" (ctx: ^_blake2.Blake2s_Context, hash: []byte) {
-    _blake2.final(ctx, hash)
+final :: proc(ctx: ^Context, hash: []byte) {
+	_blake2.final(ctx, hash)
 }
@@ -1,6 +1,6 @@
 package chacha20

-import "core:crypto/util"
+import "core:encoding/endian"
 import "core:math/bits"
 import "core:mem"

@@ -8,15 +8,23 @@ KEY_SIZE :: 32
 NONCE_SIZE :: 12
 XNONCE_SIZE :: 24

+@(private)
 _MAX_CTR_IETF :: 0xffffffff

+@(private)
 _BLOCK_SIZE :: 64
+@(private)
 _STATE_SIZE_U32 :: 16
+@(private)
 _ROUNDS :: 20

+@(private)
 _SIGMA_0 : u32 : 0x61707865
+@(private)
 _SIGMA_1 : u32 : 0x3320646e
+@(private)
 _SIGMA_2 : u32 : 0x79622d32
+@(private)
 _SIGMA_3 : u32 : 0x6b206574

 Context :: struct {
@@ -52,23 +60,23 @@ init :: proc (ctx: ^Context, key, nonce: []byte) {
 	ctx._s[1] = _SIGMA_1
 	ctx._s[2] = _SIGMA_2
 	ctx._s[3] = _SIGMA_3
-	ctx._s[4] = util.U32_LE(k[0:4])
-	ctx._s[5] = util.U32_LE(k[4:8])
-	ctx._s[6] = util.U32_LE(k[8:12])
-	ctx._s[7] = util.U32_LE(k[12:16])
-	ctx._s[8] = util.U32_LE(k[16:20])
-	ctx._s[9] = util.U32_LE(k[20:24])
-	ctx._s[10] = util.U32_LE(k[24:28])
-	ctx._s[11] = util.U32_LE(k[28:32])
+	ctx._s[4] = endian.unchecked_get_u32le(k[0:4])
+	ctx._s[5] = endian.unchecked_get_u32le(k[4:8])
+	ctx._s[6] = endian.unchecked_get_u32le(k[8:12])
+	ctx._s[7] = endian.unchecked_get_u32le(k[12:16])
+	ctx._s[8] = endian.unchecked_get_u32le(k[16:20])
+	ctx._s[9] = endian.unchecked_get_u32le(k[20:24])
+	ctx._s[10] = endian.unchecked_get_u32le(k[24:28])
+	ctx._s[11] = endian.unchecked_get_u32le(k[28:32])
 	ctx._s[12] = 0
 	if !is_xchacha {
-		ctx._s[13] = util.U32_LE(n[0:4])
-		ctx._s[14] = util.U32_LE(n[4:8])
-		ctx._s[15] = util.U32_LE(n[8:12])
+		ctx._s[13] = endian.unchecked_get_u32le(n[0:4])
+		ctx._s[14] = endian.unchecked_get_u32le(n[4:8])
+		ctx._s[15] = endian.unchecked_get_u32le(n[8:12])
 	} else {
 		ctx._s[13] = 0
-		ctx._s[14] = util.U32_LE(n[0:4])
-		ctx._s[15] = util.U32_LE(n[4:8])
+		ctx._s[14] = endian.unchecked_get_u32le(n[0:4])
+		ctx._s[15] = endian.unchecked_get_u32le(n[4:8])

 		// The sub-key is stored in the keystream buffer.  While
 		// this will be overwritten in most circumstances, explicitly
@@ -179,6 +187,7 @@ reset :: proc (ctx: ^Context) {
 	ctx._is_initialized = false
 }

+@(private)
 _do_blocks :: proc (ctx: ^Context, dst, src: []byte, nr_blocks: int) {
 	// Enforce the maximum consumed keystream per nonce.
 	//
@@ -212,114 +221,114 @@ _do_blocks :: proc (ctx: ^Context, dst, src: []byte, nr_blocks: int) {
 			// quarterround(x, 0, 4, 8, 12)
 			x0 += x4
 			x12 ~= x0
-			x12 = util.ROTL32(x12, 16)
+			x12 = bits.rotate_left32(x12, 16)
 			x8 += x12
 			x4 ~= x8
-			x4 = util.ROTL32(x4, 12)
+			x4 = bits.rotate_left32(x4, 12)
 			x0 += x4
 			x12 ~= x0
-			x12 = util.ROTL32(x12, 8)
+			x12 = bits.rotate_left32(x12, 8)
 			x8 += x12
 			x4 ~= x8
-			x4 = util.ROTL32(x4, 7)
+			x4 = bits.rotate_left32(x4, 7)

 			// quarterround(x, 1, 5, 9, 13)
 			x1 += x5
 			x13 ~= x1
-			x13 = util.ROTL32(x13, 16)
+			x13 = bits.rotate_left32(x13, 16)
 			x9 += x13
 			x5 ~= x9
-			x5 = util.ROTL32(x5, 12)
+			x5 = bits.rotate_left32(x5, 12)
 			x1 += x5
 			x13 ~= x1
-			x13 = util.ROTL32(x13, 8)
+			x13 = bits.rotate_left32(x13, 8)
 			x9 += x13
 			x5 ~= x9
-			x5 = util.ROTL32(x5, 7)
+			x5 = bits.rotate_left32(x5, 7)

 			// quarterround(x, 2, 6, 10, 14)
 			x2 += x6
 			x14 ~= x2
-			x14 = util.ROTL32(x14, 16)
+			x14 = bits.rotate_left32(x14, 16)
 			x10 += x14
 			x6 ~= x10
-			x6 = util.ROTL32(x6, 12)
+			x6 = bits.rotate_left32(x6, 12)
 			x2 += x6
 			x14 ~= x2
-			x14 = util.ROTL32(x14, 8)
+			x14 = bits.rotate_left32(x14, 8)
 			x10 += x14
 			x6 ~= x10
-			x6 = util.ROTL32(x6, 7)
+			x6 = bits.rotate_left32(x6, 7)

 			// quarterround(x, 3, 7, 11, 15)
 			x3 += x7
 			x15 ~= x3
-			x15 = util.ROTL32(x15, 16)
+			x15 = bits.rotate_left32(x15, 16)
 			x11 += x15
 			x7 ~= x11
-			x7 = util.ROTL32(x7, 12)
+			x7 = bits.rotate_left32(x7, 12)
 			x3 += x7
 			x15 ~= x3
-			x15 = util.ROTL32(x15, 8)
+			x15 = bits.rotate_left32(x15, 8)
 			x11 += x15
 			x7 ~= x11
-			x7 = util.ROTL32(x7, 7)
+			x7 = bits.rotate_left32(x7, 7)

 			// quarterround(x, 0, 5, 10, 15)
 			x0 += x5
 			x15 ~= x0
-			x15 = util.ROTL32(x15, 16)
+			x15 = bits.rotate_left32(x15, 16)
 			x10 += x15
 			x5 ~= x10
-			x5 = util.ROTL32(x5, 12)
+			x5 = bits.rotate_left32(x5, 12)
 			x0 += x5
 			x15 ~= x0
-			x15 = util.ROTL32(x15, 8)
+			x15 = bits.rotate_left32(x15, 8)
 			x10 += x15
 			x5 ~= x10
-			x5 = util.ROTL32(x5, 7)
+			x5 = bits.rotate_left32(x5, 7)

 			// quarterround(x, 1, 6, 11, 12)
 			x1 += x6
 			x12 ~= x1
-			x12 = util.ROTL32(x12, 16)
+			x12 = bits.rotate_left32(x12, 16)
 			x11 += x12
 			x6 ~= x11
-			x6 = util.ROTL32(x6, 12)
+			x6 = bits.rotate_left32(x6, 12)
 			x1 += x6
 			x12 ~= x1
-			x12 = util.ROTL32(x12, 8)
+			x12 = bits.rotate_left32(x12, 8)
 			x11 += x12
 			x6 ~= x11
-			x6 = util.ROTL32(x6, 7)
+			x6 = bits.rotate_left32(x6, 7)

 			// quarterround(x, 2, 7, 8, 13)
 			x2 += x7
 			x13 ~= x2
-			x13 = util.ROTL32(x13, 16)
+			x13 = bits.rotate_left32(x13, 16)
 			x8 += x13
 			x7 ~= x8
-			x7 = util.ROTL32(x7, 12)
+			x7 = bits.rotate_left32(x7, 12)
 			x2 += x7
 			x13 ~= x2
-			x13 = util.ROTL32(x13, 8)
+			x13 = bits.rotate_left32(x13, 8)
 			x8 += x13
 			x7 ~= x8
-			x7 = util.ROTL32(x7, 7)
+			x7 = bits.rotate_left32(x7, 7)

 			// quarterround(x, 3, 4, 9, 14)
 			x3 += x4
 			x14 ~= x3
-			x14 = util.ROTL32(x14, 16)
+			x14 = bits.rotate_left32(x14, 16)
 			x9 += x14
 			x4 ~= x9
-			x4 = util.ROTL32(x4, 12)
+			x4 = bits.rotate_left32(x4, 12)
 			x3 += x4
 			x14 ~= x3
-			x14 = util.ROTL32(x14, 8)
+			x14 = bits.rotate_left32(x14, 8)
 			x9 += x14
 			x4 ~= x9
-			x4 = util.ROTL32(x4, 7)
+			x4 = bits.rotate_left32(x4, 7)
 		}

 		x0 += _SIGMA_0
@@ -343,93 +352,48 @@ _do_blocks :: proc (ctx: ^Context, dst, src: []byte, nr_blocks: int) {
 		// this is "use vector operations", support for that is currently
 		// a work in progress/to be designed.
 		//
-		// Until dedicated assembly can be written leverage the fact that
-		// the callers of this routine ensure that src/dst are valid.
+		// In the meantime:
+		// - The caller(s) ensure that src/dst are valid.
+		// - The compiler knows if the target is picky about alignment.

-		when ODIN_ARCH == .i386 || ODIN_ARCH == .amd64 {
-			// util.PUT_U32_LE/util.U32_LE are not required on little-endian
-			// systems that also happen to not be strict about aligned
-			// memory access.
-
-			dst_p := transmute(^[16]u32)(&dst[0])
+		#no_bounds_check {
 			if src != nil {
-				src_p := transmute(^[16]u32)(&src[0])
-				dst_p[0] = src_p[0] ~ x0
-				dst_p[1] = src_p[1] ~ x1
-				dst_p[2] = src_p[2] ~ x2
-				dst_p[3] = src_p[3] ~ x3
-				dst_p[4] = src_p[4] ~ x4
-				dst_p[5] = src_p[5] ~ x5
-				dst_p[6] = src_p[6] ~ x6
-				dst_p[7] = src_p[7] ~ x7
-				dst_p[8] = src_p[8] ~ x8
-				dst_p[9] = src_p[9] ~ x9
-				dst_p[10] = src_p[10] ~ x10
-				dst_p[11] = src_p[11] ~ x11
-				dst_p[12] = src_p[12] ~ x12
-				dst_p[13] = src_p[13] ~ x13
-				dst_p[14] = src_p[14] ~ x14
-				dst_p[15] = src_p[15] ~ x15
+				endian.unchecked_put_u32le(dst[0:4], endian.unchecked_get_u32le(src[0:4]) ~ x0)
+				endian.unchecked_put_u32le(dst[4:8], endian.unchecked_get_u32le(src[4:8]) ~ x1)
+				endian.unchecked_put_u32le(dst[8:12], endian.unchecked_get_u32le(src[8:12]) ~ x2)
+				endian.unchecked_put_u32le(dst[12:16], endian.unchecked_get_u32le(src[12:16]) ~ x3)
+				endian.unchecked_put_u32le(dst[16:20], endian.unchecked_get_u32le(src[16:20]) ~ x4)
+				endian.unchecked_put_u32le(dst[20:24], endian.unchecked_get_u32le(src[20:24]) ~ x5)
+				endian.unchecked_put_u32le(dst[24:28], endian.unchecked_get_u32le(src[24:28]) ~ x6)
+				endian.unchecked_put_u32le(dst[28:32], endian.unchecked_get_u32le(src[28:32]) ~ x7)
+				endian.unchecked_put_u32le(dst[32:36], endian.unchecked_get_u32le(src[32:36]) ~ x8)
+				endian.unchecked_put_u32le(dst[36:40], endian.unchecked_get_u32le(src[36:40]) ~ x9)
+				endian.unchecked_put_u32le(dst[40:44], endian.unchecked_get_u32le(src[40:44]) ~ x10)
+				endian.unchecked_put_u32le(dst[44:48], endian.unchecked_get_u32le(src[44:48]) ~ x11)
+				endian.unchecked_put_u32le(dst[48:52], endian.unchecked_get_u32le(src[48:52]) ~ x12)
+				endian.unchecked_put_u32le(dst[52:56], endian.unchecked_get_u32le(src[52:56]) ~ x13)
+				endian.unchecked_put_u32le(dst[56:60], endian.unchecked_get_u32le(src[56:60]) ~ x14)
+				endian.unchecked_put_u32le(dst[60:64], endian.unchecked_get_u32le(src[60:64]) ~ x15)
 				src = src[_BLOCK_SIZE:]
 			} else {
-				dst_p[0] = x0
-				dst_p[1] = x1
-				dst_p[2] = x2
-				dst_p[3] = x3
-				dst_p[4] = x4
-				dst_p[5] = x5
-				dst_p[6] = x6
-				dst_p[7] = x7
-				dst_p[8] = x8
-				dst_p[9] = x9
-				dst_p[10] = x10
-				dst_p[11] = x11
-				dst_p[12] = x12
-				dst_p[13] = x13
-				dst_p[14] = x14
-				dst_p[15] = x15
+				endian.unchecked_put_u32le(dst[0:4], x0)
+				endian.unchecked_put_u32le(dst[4:8], x1)
+				endian.unchecked_put_u32le(dst[8:12], x2)
+				endian.unchecked_put_u32le(dst[12:16], x3)
+				endian.unchecked_put_u32le(dst[16:20], x4)
+				endian.unchecked_put_u32le(dst[20:24], x5)
+				endian.unchecked_put_u32le(dst[24:28], x6)
+				endian.unchecked_put_u32le(dst[28:32], x7)
+				endian.unchecked_put_u32le(dst[32:36], x8)
+				endian.unchecked_put_u32le(dst[36:40], x9)
+				endian.unchecked_put_u32le(dst[40:44], x10)
+				endian.unchecked_put_u32le(dst[44:48], x11)
+				endian.unchecked_put_u32le(dst[48:52], x12)
+				endian.unchecked_put_u32le(dst[52:56], x13)
+				endian.unchecked_put_u32le(dst[56:60], x14)
+				endian.unchecked_put_u32le(dst[60:64], x15)
 			}
 			dst = dst[_BLOCK_SIZE:]
-		} else {
-			#no_bounds_check {
-				if src != nil {
-					util.PUT_U32_LE(dst[0:4], util.U32_LE(src[0:4]) ~ x0)
-					util.PUT_U32_LE(dst[4:8], util.U32_LE(src[4:8]) ~ x1)
-					util.PUT_U32_LE(dst[8:12], util.U32_LE(src[8:12]) ~ x2)
-					util.PUT_U32_LE(dst[12:16], util.U32_LE(src[12:16]) ~ x3)
-					util.PUT_U32_LE(dst[16:20], util.U32_LE(src[16:20]) ~ x4)
-					util.PUT_U32_LE(dst[20:24], util.U32_LE(src[20:24]) ~ x5)
-					util.PUT_U32_LE(dst[24:28], util.U32_LE(src[24:28]) ~ x6)
-					util.PUT_U32_LE(dst[28:32], util.U32_LE(src[28:32]) ~ x7)
-					util.PUT_U32_LE(dst[32:36], util.U32_LE(src[32:36]) ~ x8)
-					util.PUT_U32_LE(dst[36:40], util.U32_LE(src[36:40]) ~ x9)
-					util.PUT_U32_LE(dst[40:44], util.U32_LE(src[40:44]) ~ x10)
-					util.PUT_U32_LE(dst[44:48], util.U32_LE(src[44:48]) ~ x11)
-					util.PUT_U32_LE(dst[48:52], util.U32_LE(src[48:52]) ~ x12)
-					util.PUT_U32_LE(dst[52:56], util.U32_LE(src[52:56]) ~ x13)
-					util.PUT_U32_LE(dst[56:60], util.U32_LE(src[56:60]) ~ x14)
-					util.PUT_U32_LE(dst[60:64], util.U32_LE(src[60:64]) ~ x15)
-					src = src[_BLOCK_SIZE:]
-				} else {
-					util.PUT_U32_LE(dst[0:4], x0)
-					util.PUT_U32_LE(dst[4:8], x1)
-					util.PUT_U32_LE(dst[8:12], x2)
-					util.PUT_U32_LE(dst[12:16], x3)
-					util.PUT_U32_LE(dst[16:20], x4)
-					util.PUT_U32_LE(dst[20:24], x5)
-					util.PUT_U32_LE(dst[24:28], x6)
-					util.PUT_U32_LE(dst[28:32], x7)
-					util.PUT_U32_LE(dst[32:36], x8)
-					util.PUT_U32_LE(dst[36:40], x9)
-					util.PUT_U32_LE(dst[40:44], x10)
-					util.PUT_U32_LE(dst[44:48], x11)
-					util.PUT_U32_LE(dst[48:52], x12)
-					util.PUT_U32_LE(dst[52:56], x13)
-					util.PUT_U32_LE(dst[56:60], x14)
-					util.PUT_U32_LE(dst[60:64], x15)
-				}
-				dst = dst[_BLOCK_SIZE:]
-			}
 		}

 		// Increment the counter.  Overflow checking is done upon
@@ -441,141 +405,142 @@ _do_blocks :: proc (ctx: ^Context, dst, src: []byte, nr_blocks: int) {
 	}
 }

-_hchacha20 :: proc (dst, key, nonce: []byte) {
+@(private)
+_hchacha20 :: proc "contextless" (dst, key, nonce: []byte) {
 	x0, x1, x2, x3 := _SIGMA_0, _SIGMA_1, _SIGMA_2, _SIGMA_3
-	x4 := util.U32_LE(key[0:4])
-	x5 := util.U32_LE(key[4:8])
-	x6 := util.U32_LE(key[8:12])
-	x7 := util.U32_LE(key[12:16])
-	x8 := util.U32_LE(key[16:20])
-	x9 := util.U32_LE(key[20:24])
-	x10 := util.U32_LE(key[24:28])
-	x11 := util.U32_LE(key[28:32])
-	x12 := util.U32_LE(nonce[0:4])
-	x13 := util.U32_LE(nonce[4:8])
-	x14 := util.U32_LE(nonce[8:12])
-	x15 := util.U32_LE(nonce[12:16])
+	x4 := endian.unchecked_get_u32le(key[0:4])
+	x5 := endian.unchecked_get_u32le(key[4:8])
+	x6 := endian.unchecked_get_u32le(key[8:12])
+	x7 := endian.unchecked_get_u32le(key[12:16])
+	x8 := endian.unchecked_get_u32le(key[16:20])
+	x9 := endian.unchecked_get_u32le(key[20:24])
+	x10 := endian.unchecked_get_u32le(key[24:28])
+	x11 := endian.unchecked_get_u32le(key[28:32])
+	x12 := endian.unchecked_get_u32le(nonce[0:4])
+	x13 := endian.unchecked_get_u32le(nonce[4:8])
+	x14 := endian.unchecked_get_u32le(nonce[8:12])
+	x15 := endian.unchecked_get_u32le(nonce[12:16])

 	for i := _ROUNDS; i > 0; i = i - 2 {
 		// quarterround(x, 0, 4, 8, 12)
 		x0 += x4
 		x12 ~= x0
-		x12 = util.ROTL32(x12, 16)
+		x12 = bits.rotate_left32(x12, 16)
 		x8 += x12
 		x4 ~= x8
-		x4 = util.ROTL32(x4, 12)
+		x4 = bits.rotate_left32(x4, 12)
 		x0 += x4
 		x12 ~= x0
-		x12 = util.ROTL32(x12, 8)
+		x12 = bits.rotate_left32(x12, 8)
 		x8 += x12
 		x4 ~= x8
-		x4 = util.ROTL32(x4, 7)
+		x4 = bits.rotate_left32(x4, 7)

 		// quarterround(x, 1, 5, 9, 13)
 		x1 += x5
 		x13 ~= x1
-		x13 = util.ROTL32(x13, 16)
+		x13 = bits.rotate_left32(x13, 16)
 		x9 += x13
 		x5 ~= x9
-		x5 = util.ROTL32(x5, 12)
+		x5 = bits.rotate_left32(x5, 12)
 		x1 += x5
 		x13 ~= x1
-		x13 = util.ROTL32(x13, 8)
+		x13 = bits.rotate_left32(x13, 8)
 		x9 += x13
 		x5 ~= x9
-		x5 = util.ROTL32(x5, 7)
+		x5 = bits.rotate_left32(x5, 7)

 		// quarterround(x, 2, 6, 10, 14)
 		x2 += x6
 		x14 ~= x2
-		x14 = util.ROTL32(x14, 16)
+		x14 = bits.rotate_left32(x14, 16)
 		x10 += x14
 		x6 ~= x10
-		x6 = util.ROTL32(x6, 12)
+		x6 = bits.rotate_left32(x6, 12)
 		x2 += x6
 		x14 ~= x2
-		x14 = util.ROTL32(x14, 8)
+		x14 = bits.rotate_left32(x14, 8)
 		x10 += x14
 		x6 ~= x10
-		x6 = util.ROTL32(x6, 7)
+		x6 = bits.rotate_left32(x6, 7)

 		// quarterround(x, 3, 7, 11, 15)
 		x3 += x7
 		x15 ~= x3
-		x15 = util.ROTL32(x15, 16)
+		x15 = bits.rotate_left32(x15, 16)
 		x11 += x15
 		x7 ~= x11
-		x7 = util.ROTL32(x7, 12)
+		x7 = bits.rotate_left32(x7, 12)
 		x3 += x7
 		x15 ~= x3
-		x15 = util.ROTL32(x15, 8)
+		x15 = bits.rotate_left32(x15, 8)
 		x11 += x15
 		x7 ~= x11
-		x7 = util.ROTL32(x7, 7)
+		x7 = bits.rotate_left32(x7, 7)

 		// quarterround(x, 0, 5, 10, 15)
 		x0 += x5
 		x15 ~= x0
-		x15 = util.ROTL32(x15, 16)
+		x15 = bits.rotate_left32(x15, 16)
 		x10 += x15
 		x5 ~= x10
-		x5 = util.ROTL32(x5, 12)
+		x5 = bits.rotate_left32(x5, 12)
 		x0 += x5
 		x15 ~= x0
-		x15 = util.ROTL32(x15, 8)
+		x15 = bits.rotate_left32(x15, 8)
 		x10 += x15
 		x5 ~= x10
-		x5 = util.ROTL32(x5, 7)
+		x5 = bits.rotate_left32(x5, 7)

 		// quarterround(x, 1, 6, 11, 12)
 		x1 += x6
 		x12 ~= x1
-		x12 = util.ROTL32(x12, 16)
+		x12 = bits.rotate_left32(x12, 16)
 		x11 += x12
 		x6 ~= x11
-		x6 = util.ROTL32(x6, 12)
+		x6 = bits.rotate_left32(x6, 12)
 		x1 += x6
 		x12 ~= x1
-		x12 = util.ROTL32(x12, 8)
+		x12 = bits.rotate_left32(x12, 8)
 		x11 += x12
 		x6 ~= x11
-		x6 = util.ROTL32(x6, 7)
+		x6 = bits.rotate_left32(x6, 7)

 		// quarterround(x, 2, 7, 8, 13)
 		x2 += x7
 		x13 ~= x2
-		x13 = util.ROTL32(x13, 16)
+		x13 = bits.rotate_left32(x13, 16)
 		x8 += x13
 		x7 ~= x8
-		x7 = util.ROTL32(x7, 12)
+		x7 = bits.rotate_left32(x7, 12)
 		x2 += x7
 		x13 ~= x2
-		x13 = util.ROTL32(x13, 8)
+		x13 = bits.rotate_left32(x13, 8)
 		x8 += x13
 		x7 ~= x8
-		x7 = util.ROTL32(x7, 7)
+		x7 = bits.rotate_left32(x7, 7)

 		// quarterround(x, 3, 4, 9, 14)
 		x3 += x4
 		x14 ~= x3
-		x14 = util.ROTL32(x14, 16)
+		x14 = bits.rotate_left32(x14, 16)
 		x9 += x14
 		x4 ~= x9
-		x4 = util.ROTL32(x4, 12)
+		x4 = bits.rotate_left32(x4, 12)
 		x3 += x4
 		x14 ~= x3
-		x14 = util.ROTL32(x14, 8)
+		x14 = bits.rotate_left32(x14, 8)
 		x9 += x14
 		x4 ~= x9
-		x4 = util.ROTL32(x4, 7)
+		x4 = bits.rotate_left32(x4, 7)
 	}

-	util.PUT_U32_LE(dst[0:4], x0)
-	util.PUT_U32_LE(dst[4:8], x1)
-	util.PUT_U32_LE(dst[8:12], x2)
-	util.PUT_U32_LE(dst[12:16], x3)
-	util.PUT_U32_LE(dst[16:20], x12)
-	util.PUT_U32_LE(dst[20:24], x13)
-	util.PUT_U32_LE(dst[24:28], x14)
-	util.PUT_U32_LE(dst[28:32], x15)
+	endian.unchecked_put_u32le(dst[0:4], x0)
+	endian.unchecked_put_u32le(dst[4:8], x1)
+	endian.unchecked_put_u32le(dst[8:12], x2)
+	endian.unchecked_put_u32le(dst[12:16], x3)
+	endian.unchecked_put_u32le(dst[16:20], x12)
+	endian.unchecked_put_u32le(dst[20:24], x13)
+	endian.unchecked_put_u32le(dst[24:28], x14)
+	endian.unchecked_put_u32le(dst[28:32], x15)
 }
@@ -3,15 +3,17 @@ package chacha20poly1305
 import "core:crypto"
 import "core:crypto/chacha20"
 import "core:crypto/poly1305"
-import "core:crypto/util"
+import "core:encoding/endian"
 import "core:mem"

 KEY_SIZE :: chacha20.KEY_SIZE
 NONCE_SIZE :: chacha20.NONCE_SIZE
 TAG_SIZE :: poly1305.TAG_SIZE

+@(private)
 _P_MAX :: 64 * 0xffffffff // 64 * (2^32-1)

+@(private)
 _validate_common_slice_sizes :: proc (tag, key, nonce, aad, text: []byte) {
 	if len(tag) != TAG_SIZE {
 		panic("crypto/chacha20poly1305: invalid destination tag size")
@@ -37,7 +39,10 @@ _validate_common_slice_sizes :: proc (tag, key, nonce, aad, text: []byte) {
 	}
 }

+@(private)
 _PAD: [16]byte
+
+@(private)
 _update_mac_pad16 :: #force_inline proc (ctx: ^poly1305.Context, x_len: int) {
 	if pad_len := 16 - (x_len & (16-1)); pad_len != 16 {
 		poly1305.update(ctx, _PAD[:pad_len])
@@ -82,8 +87,8 @@ encrypt :: proc (ciphertext, tag, key, nonce, aad, plaintext: []byte) {
 	// mac_data |= num_to_8_le_bytes(aad.length)
 	// mac_data |= num_to_8_le_bytes(ciphertext.length)
 	l_buf := otk[0:16] // Reuse the scratch buffer.
-	util.PUT_U64_LE(l_buf[0:8], u64(aad_len))
-	util.PUT_U64_LE(l_buf[8:16], u64(ciphertext_len))
+	endian.unchecked_put_u64le(l_buf[0:8], u64(aad_len))
+	endian.unchecked_put_u64le(l_buf[8:16], u64(ciphertext_len))
 	poly1305.update(&mac_ctx, l_buf)

 	// tag = poly1305_mac(mac_data, otk)
@@ -123,8 +128,8 @@ decrypt :: proc (plaintext, tag, key, nonce, aad, ciphertext: []byte) -> bool {
 	poly1305.update(&mac_ctx, ciphertext)
 	_update_mac_pad16(&mac_ctx, ciphertext_len)
 	l_buf := otk[0:16] // Reuse the scratch buffer.
-	util.PUT_U64_LE(l_buf[0:8], u64(aad_len))
-	util.PUT_U64_LE(l_buf[8:16], u64(ciphertext_len))
+	endian.unchecked_put_u64le(l_buf[0:8], u64(aad_len))
+	endian.unchecked_put_u64le(l_buf[8:16], u64(ciphertext_len))
 	poly1305.update(&mac_ctx, l_buf)

 	// tag = poly1305_mac(mac_data, otk)
@@ -26,6 +26,7 @@ compare_constant_time :: proc "contextless" (a, b: []byte) -> int {
 //
 // The execution time of this routine is constant regardless of the
 // contents of the memory being compared.
+@(optimization_mode="none")
 compare_byte_ptrs_constant_time :: proc "contextless" (a, b: ^byte, n: int) -> int {
 	x := mem.slice_ptr(a, n)
 	y := mem.slice_ptr(b, n)
@@ -1,382 +0,0 @@
-package gost
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation of the GOST hashing algorithm, as defined in RFC 5831 <https://datatracker.ietf.org/doc/html/rfc5831>
-*/
-
-import "core:mem"
-import "core:os"
-import "core:io"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE :: 32
-
-// hash_string will hash the given input and return the
-// computed hash
-hash_string :: proc(data: string) -> [DIGEST_SIZE]byte {
-    return hash_bytes(transmute([]byte)(data))
-}
-
-// hash_bytes will hash the given input and return the
-// computed hash
-hash_bytes :: proc(data: []byte) -> [DIGEST_SIZE]byte {
-    hash: [DIGEST_SIZE]byte
-    ctx: Gost_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE, "Size of destination buffer is smaller than the digest size")
-    ctx: Gost_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream will read the stream in chunks and compute a
-// hash from its contents
-hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
-    hash: [DIGEST_SIZE]byte
-    ctx: Gost_Context
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file will read the file provided by the given handle
-// and compute a hash
-hash_file :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE]byte, bool) {
-    if !load_at_once {
-        return hash_stream(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE]byte{}, false
-}
-
-hash :: proc {
-    hash_stream,
-    hash_file,
-    hash_bytes,
-    hash_string,
-    hash_bytes_to_buffer,
-    hash_string_to_buffer,
-}
-
-/*
-    Low level API
-*/
-
-init :: proc "contextless" (ctx: ^Gost_Context) {
-    sbox: [8][16]u32 = {
-        { 10, 4,  5,  6,  8,  1,  3,  7,  13, 12, 14, 0,  9,  2,  11, 15 },
-        { 5,  15, 4,  0,  2,  13, 11, 9,  1,  7,  6,  3,  12, 14, 10, 8  },
-        { 7,  15, 12, 14, 9,  4,  1,  0,  3,  11, 5,  2,  6,  10, 8,  13 },
-        { 4,  10, 7,  12, 0,  15, 2,  8,  14, 1,  6,  5,  13, 11, 9,  3  },
-        { 7,  6,  4,  11, 9,  12, 2,  10, 1,  8,  0,  14, 15, 13, 3,  5  },
-        { 7,  6,  2,  4,  13, 9,  15, 0,  10, 1,  5,  11, 8,  14, 12, 3  },
-        { 13, 14, 4,  1,  7,  0,  5,  10, 3,  12, 8,  15, 6,  2,  9,  11 },
-        { 1,  3,  10, 9,  5,  11, 4,  15, 8,  6,  7,  14, 13, 0,  2,  12 },
-    }
-
-    i := 0
-    for a := 0; a < 16; a += 1 {
-        ax := sbox[1][a] << 15
-        bx := sbox[3][a] << 23
-        cx := sbox[5][a]
-        cx = (cx >> 1) | (cx << 31)
-        dx := sbox[7][a] << 7
-        for b := 0; b < 16; b, i = b + 1, i + 1 {
-            SBOX_1[i] = ax | (sbox[0][b] << 11)
-            SBOX_2[i] = bx | (sbox[2][b] << 19)
-            SBOX_3[i] = cx | (sbox[4][b] << 27)
-            SBOX_4[i] = dx | (sbox[6][b] << 3)
-        }
-    }
-}
-
-update :: proc(ctx: ^Gost_Context, data: []byte) {
-    length := byte(len(data))
-    j: byte
-
-    i := ctx.partial_bytes
-    for i < 32 && j < length {
-        ctx.partial[i] = data[j]
-        i, j = i + 1, j + 1
-    }
-
-    if i < 32 {
-        ctx.partial_bytes = i
-        return
-    }
-    bytes(ctx, ctx.partial[:], 256)
-
-    for (j + 32) < length {
-        bytes(ctx, data[j:], 256)
-        j += 32
-    }
-
-    i = 0
-    for j < length {
-        ctx.partial[i] = data[j]
-        i, j = i + 1, j + 1
-    }
-    ctx.partial_bytes = i
-}
-
-final :: proc(ctx: ^Gost_Context, hash: []byte) {
-    if ctx.partial_bytes > 0 {
-        mem.set(&ctx.partial[ctx.partial_bytes], 0, 32 - int(ctx.partial_bytes))
-        bytes(ctx, ctx.partial[:], u32(ctx.partial_bytes) << 3)
-    }
-  
-    compress(ctx.hash[:], ctx.len[:])
-    compress(ctx.hash[:], ctx.sum[:])
-
-    for i, j := 0, 0; i < 8; i, j = i + 1, j + 4 {
-        hash[j]     = byte(ctx.hash[i])
-        hash[j + 1] = byte(ctx.hash[i] >> 8)
-        hash[j + 2] = byte(ctx.hash[i] >> 16)
-        hash[j + 3] = byte(ctx.hash[i] >> 24)
-    }
-}
-
-/*
-    GOST implementation
-*/
-
-Gost_Context :: struct {
-    sum:           [8]u32,
-    hash:          [8]u32,
-    len:           [8]u32,
-    partial:       [32]byte,
-    partial_bytes: byte,
-}
-
-SBOX_1: [256]u32
-SBOX_2: [256]u32
-SBOX_3: [256]u32
-SBOX_4: [256]u32
-
-ENCRYPT_ROUND :: #force_inline proc "contextless" (l, r, t, k1, k2: u32) -> (u32, u32, u32) {
-    l, r, t := l, r, t
-    t  = (k1) + r
-    l ~= SBOX_1[t & 0xff] ~ SBOX_2[(t >> 8) & 0xff] ~ SBOX_3[(t >> 16) & 0xff] ~ SBOX_4[t >> 24]
-    t  = (k2) + l
-    r ~= SBOX_1[t & 0xff] ~ SBOX_2[(t >> 8) & 0xff] ~ SBOX_3[(t >> 16) & 0xff] ~ SBOX_4[t >> 24]
-    return l, r, t
-}
-
-ENCRYPT :: #force_inline proc "contextless" (a, b, c: u32, key: []u32) -> (l, r, t: u32) {
-    l, r, t = ENCRYPT_ROUND(a, b, c, key[0], key[1])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[2], key[3])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[4], key[5])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[6], key[7])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[0], key[1])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[2], key[3])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[4], key[5])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[6], key[7])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[0], key[1])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[2], key[3])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[4], key[5])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[6], key[7])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[7], key[6])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[5], key[4])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[3], key[2])
-    l, r, t = ENCRYPT_ROUND(l, r, t, key[1], key[0])
-    t = r
-    r = l
-    l = t
-    return
-}
-
-bytes :: proc(ctx: ^Gost_Context, buf: []byte, bits: u32) {
-    a, c: u32
-    m: [8]u32
-
-    for i, j := 0, 0; i < 8; i += 1 {
-        a = u32(buf[j]) | u32(buf[j + 1]) << 8 | u32(buf[j + 2]) << 16 | u32(buf[j + 3]) << 24
-        j += 4
-        m[i] = a
-        c = a + c + ctx.sum[i]
-        ctx.sum[i] = c
-        c = c < a ? 1 : 0
-    }
-
-    compress(ctx.hash[:], m[:])
-    ctx.len[0] += bits
-    if ctx.len[0] < bits {
-        ctx.len[1] += 1
-    }
-}
-
-compress :: proc(h, m: []u32) {
-    key, u, v, w, s: [8]u32
-
-    copy(u[:], h)
-    copy(v[:], m)
-
-    for i := 0; i < 8; i += 2 {
-        w[0] = u[0] ~ v[0]
-        w[1] = u[1] ~ v[1]
-        w[2] = u[2] ~ v[2]
-        w[3] = u[3] ~ v[3]
-        w[4] = u[4] ~ v[4]
-        w[5] = u[5] ~ v[5]
-        w[6] = u[6] ~ v[6]
-        w[7] = u[7] ~ v[7]
-
-        key[0] = (w[0] & 0x000000ff)       | (w[2] & 0x000000ff) <<  8 | (w[4] & 0x000000ff) << 16 | (w[6] & 0x000000ff) << 24
-        key[1] = (w[0] & 0x0000ff00) >>  8 | (w[2] & 0x0000ff00)       | (w[4] & 0x0000ff00) <<  8 | (w[6] & 0x0000ff00) << 16
-        key[2] = (w[0] & 0x00ff0000) >> 16 | (w[2] & 0x00ff0000) >>  8 | (w[4] & 0x00ff0000)       | (w[6] & 0x00ff0000) <<  8
-        key[3] = (w[0] & 0xff000000) >> 24 | (w[2] & 0xff000000) >> 16 | (w[4] & 0xff000000) >>  8 | (w[6] & 0xff000000)
-        key[4] = (w[1] & 0x000000ff)       | (w[3] & 0x000000ff) <<  8 | (w[5] & 0x000000ff) << 16 | (w[7] & 0x000000ff) << 24
-        key[5] = (w[1] & 0x0000ff00) >>  8 | (w[3] & 0x0000ff00)       | (w[5] & 0x0000ff00) <<  8 | (w[7] & 0x0000ff00) << 16
-        key[6] = (w[1] & 0x00ff0000) >> 16 | (w[3] & 0x00ff0000) >>  8 | (w[5] & 0x00ff0000)       | (w[7] & 0x00ff0000) <<  8
-        key[7] = (w[1] & 0xff000000) >> 24 | (w[3] & 0xff000000) >> 16 | (w[5] & 0xff000000) >>  8 | (w[7] & 0xff000000)
-
-        r := h[i]
-        l := h[i + 1]
-        t: u32
-        l, r, t = ENCRYPT(l, r, 0, key[:])
-
-        s[i] = r
-        s[i + 1] = l
-
-        if i == 6 {
-            break
-        }
-
-        l    = u[0] ~ u[2]
-        r    = u[1] ~ u[3]
-        u[0] = u[2]
-        u[1] = u[3]
-        u[2] = u[4]
-        u[3] = u[5]
-        u[4] = u[6]
-        u[5] = u[7]
-        u[6] = l
-        u[7] = r
-
-        if i == 2 {
-            u[0] ~= 0xff00ff00
-            u[1] ~= 0xff00ff00
-            u[2] ~= 0x00ff00ff
-            u[3] ~= 0x00ff00ff
-            u[4] ~= 0x00ffff00
-            u[5] ~= 0xff0000ff
-            u[6] ~= 0x000000ff
-            u[7] ~= 0xff00ffff
-        }
-
-        l    = v[0]
-        r    = v[2]
-        v[0] = v[4]
-        v[2] = v[6]
-        v[4] = l ~ r
-        v[6] = v[0] ~ r
-        l    = v[1]
-        r    = v[3]
-        v[1] = v[5]
-        v[3] = v[7]
-        v[5] = l ~ r
-        v[7] = v[1] ~ r
-    }
-
-    u[0] = m[0] ~ s[6]
-    u[1] = m[1] ~ s[7]
-    u[2] = m[2] ~ (s[0] << 16) ~ (s[0] >> 16) ~ (s[0] & 0xffff) ~ 
-        (s[1] & 0xffff) ~ (s[1] >> 16) ~ (s[2] << 16) ~ s[6] ~ (s[6] << 16) ~
-        (s[7] & 0xffff0000) ~ (s[7] >> 16)
-    u[3] = m[3] ~ (s[0] & 0xffff) ~ (s[0] << 16) ~ (s[1] & 0xffff) ~
-        (s[1] << 16) ~ (s[1] >> 16) ~ (s[2] << 16) ~ (s[2] >> 16) ~
-        (s[3] << 16) ~ s[6] ~ (s[6] << 16) ~ (s[6] >> 16) ~ (s[7] & 0xffff) ~
-        (s[7] << 16) ~ (s[7] >> 16)
-    u[4] = m[4] ~
-        (s[0] & 0xffff0000) ~ (s[0] << 16) ~ (s[0] >> 16) ~
-        (s[1] & 0xffff0000) ~ (s[1] >> 16) ~ (s[2] << 16) ~ (s[2] >> 16) ~
-        (s[3] << 16) ~ (s[3] >> 16) ~ (s[4] << 16) ~ (s[6] << 16) ~
-        (s[6] >> 16) ~(s[7] & 0xffff) ~ (s[7] << 16) ~ (s[7] >> 16)
-    u[5] = m[5] ~ (s[0] << 16) ~ (s[0] >> 16) ~ (s[0] & 0xffff0000) ~
-        (s[1] & 0xffff) ~ s[2] ~ (s[2] >> 16) ~ (s[3] << 16) ~ (s[3] >> 16) ~
-        (s[4] << 16) ~ (s[4] >> 16) ~ (s[5] << 16) ~  (s[6] << 16) ~
-        (s[6] >> 16) ~ (s[7] & 0xffff0000) ~ (s[7] << 16) ~ (s[7] >> 16)
-    u[6] = m[6] ~ s[0] ~ (s[1] >> 16) ~ (s[2] << 16) ~ s[3] ~ (s[3] >> 16) ~
-        (s[4] << 16) ~ (s[4] >> 16) ~ (s[5] << 16) ~ (s[5] >> 16) ~ s[6] ~
-        (s[6] << 16) ~ (s[6] >> 16) ~ (s[7] << 16)
-    u[7] = m[7] ~ (s[0] & 0xffff0000) ~ (s[0] << 16) ~ (s[1] & 0xffff) ~
-        (s[1] << 16) ~ (s[2] >> 16) ~ (s[3] << 16) ~ s[4] ~ (s[4] >> 16) ~
-        (s[5] << 16) ~ (s[5] >> 16) ~ (s[6] >> 16) ~ (s[7] & 0xffff) ~
-        (s[7] << 16) ~ (s[7] >> 16)
-
-    v[0] = h[0] ~ (u[1] << 16) ~ (u[0] >> 16)
-    v[1] = h[1] ~ (u[2] << 16) ~ (u[1] >> 16)
-    v[2] = h[2] ~ (u[3] << 16) ~ (u[2] >> 16)
-    v[3] = h[3] ~ (u[4] << 16) ~ (u[3] >> 16)
-    v[4] = h[4] ~ (u[5] << 16) ~ (u[4] >> 16)
-    v[5] = h[5] ~ (u[6] << 16) ~ (u[5] >> 16)
-    v[6] = h[6] ~ (u[7] << 16) ~ (u[6] >> 16)
-    v[7] = h[7] ~ (u[0] & 0xffff0000) ~ (u[0] << 16) ~ (u[7] >> 16) ~ (u[1] & 0xffff0000) ~ (u[1] << 16) ~ (u[6] << 16) ~ (u[7] & 0xffff0000)
-
-    h[0] = (v[0] & 0xffff0000) ~ (v[0] << 16) ~ (v[0] >> 16) ~ (v[1] >> 16) ~
-        (v[1] & 0xffff0000) ~ (v[2] << 16) ~ (v[3] >> 16) ~ (v[4] << 16) ~
-        (v[5] >> 16) ~ v[5] ~ (v[6] >> 16) ~ (v[7] << 16) ~ (v[7] >> 16) ~
-        (v[7] & 0xffff)
-    h[1] = (v[0] << 16) ~ (v[0] >> 16) ~ (v[0] & 0xffff0000) ~ (v[1] & 0xffff) ~
-        v[2] ~ (v[2] >> 16) ~ (v[3] << 16) ~ (v[4] >> 16) ~ (v[5] << 16) ~
-        (v[6] << 16) ~ v[6] ~ (v[7] & 0xffff0000) ~ (v[7] >> 16)
-    h[2] = (v[0] & 0xffff) ~ (v[0] << 16) ~ (v[1] << 16) ~ (v[1] >> 16) ~
-        (v[1] & 0xffff0000) ~ (v[2] << 16) ~ (v[3] >> 16) ~ v[3] ~ (v[4] << 16) ~
-        (v[5] >> 16) ~ v[6] ~ (v[6] >> 16) ~ (v[7] & 0xffff) ~ (v[7] << 16) ~
-        (v[7] >> 16)
-    h[3] = (v[0] << 16) ~ (v[0] >> 16) ~ (v[0] & 0xffff0000) ~
-        (v[1] & 0xffff0000) ~ (v[1] >> 16) ~ (v[2] << 16) ~ (v[2] >> 16) ~ v[2] ~
-        (v[3] << 16) ~ (v[4] >> 16) ~ v[4] ~ (v[5] << 16) ~ (v[6] << 16) ~
-        (v[7] & 0xffff) ~ (v[7] >> 16)
-    h[4] = (v[0] >> 16) ~ (v[1] << 16) ~ v[1] ~ (v[2] >> 16) ~ v[2] ~
-        (v[3] << 16) ~ (v[3] >> 16) ~ v[3] ~ (v[4] << 16) ~ (v[5] >> 16) ~
-        v[5] ~ (v[6] << 16) ~ (v[6] >> 16) ~ (v[7] << 16)
-    h[5] = (v[0] << 16) ~ (v[0] & 0xffff0000) ~ (v[1] << 16) ~ (v[1] >> 16) ~
-        (v[1] & 0xffff0000) ~ (v[2] << 16) ~ v[2] ~ (v[3] >> 16) ~ v[3] ~
-        (v[4] << 16) ~ (v[4] >> 16) ~ v[4] ~ (v[5] << 16) ~ (v[6] << 16) ~
-        (v[6] >> 16) ~ v[6] ~ (v[7] << 16) ~ (v[7] >> 16) ~ (v[7] & 0xffff0000)
-    h[6] = v[0] ~ v[2] ~ (v[2] >> 16) ~ v[3] ~ (v[3] << 16) ~ v[4] ~
-        (v[4] >> 16) ~ (v[5] << 16) ~ (v[5] >> 16) ~ v[5] ~ (v[6] << 16) ~
-        (v[6] >> 16) ~ v[6] ~ (v[7] << 16) ~ v[7]
-    h[7] = v[0] ~ (v[0] >> 16) ~ (v[1] << 16) ~ (v[1] >> 16) ~ (v[2] << 16) ~
-        (v[3] >> 16) ~ v[3] ~ (v[4] << 16) ~ v[4] ~ (v[5] >> 16) ~ v[5] ~
-        (v[6] << 16) ~ (v[6] >> 16) ~ (v[7] << 16) ~ v[7]
-}
@@ -1,653 +0,0 @@
-package groestl
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation of the GROESTL hashing algorithm, as defined in <http://www.groestl.info/Groestl.zip>
-*/
-
-import "core:os"
-import "core:io"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE_224 :: 28
-DIGEST_SIZE_256 :: 32
-DIGEST_SIZE_384 :: 48
-DIGEST_SIZE_512 :: 64
-
-// hash_string_224 will hash the given input and return the
-// computed hash
-hash_string_224 :: proc(data: string) -> [DIGEST_SIZE_224]byte {
-    return hash_bytes_224(transmute([]byte)(data))
-}
-
-// hash_bytes_224 will hash the given input and return the
-// computed hash
-hash_bytes_224 :: proc(data: []byte) -> [DIGEST_SIZE_224]byte {
-    hash: [DIGEST_SIZE_224]byte
-    ctx: Groestl_Context
-    ctx.hashbitlen = 224
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_224 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_224 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_224(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_224 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_224 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_224, "Size of destination buffer is smaller than the digest size")
-    ctx: Groestl_Context
-    ctx.hashbitlen = 224
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_224 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
-    hash: [DIGEST_SIZE_224]byte
-    ctx: Groestl_Context
-    ctx.hashbitlen = 224
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_224 will read the file provided by the given handle
-// and compute a hash
-hash_file_224 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_224]byte, bool) {
-    if !load_at_once {
-        return hash_stream_224(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_224(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_224]byte{}, false
-}
-
-hash_224 :: proc {
-    hash_stream_224,
-    hash_file_224,
-    hash_bytes_224,
-    hash_string_224,
-    hash_bytes_to_buffer_224,
-    hash_string_to_buffer_224,
-}
-
-// hash_string_256 will hash the given input and return the
-// computed hash
-hash_string_256 :: proc(data: string) -> [DIGEST_SIZE_256]byte {
-    return hash_bytes_256(transmute([]byte)(data))
-}
-
-// hash_bytes_256 will hash the given input and return the
-// computed hash
-hash_bytes_256 :: proc(data: []byte) -> [DIGEST_SIZE_256]byte {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: Groestl_Context
-    ctx.hashbitlen = 256
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_256 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_256 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_256(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_256 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_256 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_256, "Size of destination buffer is smaller than the digest size")
-    ctx: Groestl_Context
-    ctx.hashbitlen = 256
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_256 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: Groestl_Context
-    ctx.hashbitlen = 256
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_256 will read the file provided by the given handle
-// and compute a hash
-hash_file_256 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_256]byte, bool) {
-    if !load_at_once {
-        return hash_stream_256(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_256(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_256]byte{}, false
-}
-
-hash_256 :: proc {
-    hash_stream_256,
-    hash_file_256,
-    hash_bytes_256,
-    hash_string_256,
-    hash_bytes_to_buffer_256,
-    hash_string_to_buffer_256,
-}
-
-// hash_string_384 will hash the given input and return the
-// computed hash
-hash_string_384 :: proc(data: string) -> [DIGEST_SIZE_384]byte {
-    return hash_bytes_384(transmute([]byte)(data))
-}
-
-// hash_bytes_384 will hash the given input and return the
-// computed hash
-hash_bytes_384 :: proc(data: []byte) -> [DIGEST_SIZE_384]byte {
-    hash: [DIGEST_SIZE_384]byte
-    ctx: Groestl_Context
-    ctx.hashbitlen = 384
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_384 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_384 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_384(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_384 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_384 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_384, "Size of destination buffer is smaller than the digest size")
-    ctx: Groestl_Context
-    ctx.hashbitlen = 384
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_384 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
-    hash: [DIGEST_SIZE_384]byte
-    ctx: Groestl_Context
-    ctx.hashbitlen = 384
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_384 will read the file provided by the given handle
-// and compute a hash
-hash_file_384 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_384]byte, bool) {
-    if !load_at_once {
-        return hash_stream_384(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_384(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_384]byte{}, false
-}
-
-hash_384 :: proc {
-    hash_stream_384,
-    hash_file_384,
-    hash_bytes_384,
-    hash_string_384,
-    hash_bytes_to_buffer_384,
-    hash_string_to_buffer_384,
-}
-
-// hash_string_512 will hash the given input and return the
-// computed hash
-hash_string_512 :: proc(data: string) -> [DIGEST_SIZE_512]byte {
-    return hash_bytes_512(transmute([]byte)(data))
-}
-
-// hash_bytes_512 will hash the given input and return the
-// computed hash
-hash_bytes_512 :: proc(data: []byte) -> [DIGEST_SIZE_512]byte {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: Groestl_Context
-    ctx.hashbitlen = 512
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_512 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_512 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_512(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_512 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_512 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_512, "Size of destination buffer is smaller than the digest size")
-    ctx: Groestl_Context
-    ctx.hashbitlen = 512
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_512 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: Groestl_Context
-    ctx.hashbitlen = 512
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_512 will read the file provided by the given handle
-// and compute a hash
-hash_file_512 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_512]byte, bool) {
-    if !load_at_once {
-        return hash_stream_512(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_512(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_512]byte{}, false
-}
-
-hash_512 :: proc {
-    hash_stream_512,
-    hash_file_512,
-    hash_bytes_512,
-    hash_string_512,
-    hash_bytes_to_buffer_512,
-    hash_string_to_buffer_512,
-}
-
-/*
-    Low level API
-*/
-
-init :: proc(ctx: ^Groestl_Context) {
-    assert(ctx.hashbitlen == 224 || ctx.hashbitlen == 256 || ctx.hashbitlen == 384 || ctx.hashbitlen == 512, "hashbitlen must be set to 224, 256, 384 or 512")
-    if ctx.hashbitlen <= 256 {
-        ctx.rounds    = 10
-        ctx.columns   = 8
-        ctx.statesize = 64
-    } else {
-        ctx.rounds    = 14
-        ctx.columns   = 16
-        ctx.statesize = 128
-    }
-    for i := 8 - size_of(i32); i < 8; i += 1 {
-        ctx.chaining[i][ctx.columns - 1] = byte(ctx.hashbitlen >> (8 * (7 - uint(i))))
-    }
-}
-
-update :: proc(ctx: ^Groestl_Context, data: []byte) {
-    databitlen := len(data) * 8
-    msglen     := databitlen / 8
-    rem        := databitlen % 8
-
-    i: int
-    assert(ctx.bits_in_last_byte == 0)
-
-    if ctx.buf_ptr != 0 {
-        for i = 0; ctx.buf_ptr < ctx.statesize && i < msglen; i, ctx.buf_ptr =  i + 1, ctx.buf_ptr + 1 {
-            ctx.buffer[ctx.buf_ptr] = data[i]
-        }
-
-        if ctx.buf_ptr < ctx.statesize {
-            if rem != 0 {
-                ctx.bits_in_last_byte    = rem
-                ctx.buffer[ctx.buf_ptr]  = data[i]
-                ctx.buf_ptr             += 1
-            }
-            return
-        }
-
-        ctx.buf_ptr = 0
-        transform(ctx, ctx.buffer[:], u32(ctx.statesize))
-    }
-
-    transform(ctx, data[i:], u32(msglen - i))
-    i += ((msglen - i) / ctx.statesize) * ctx.statesize
-    for i < msglen {
-        ctx.buffer[ctx.buf_ptr] = data[i]
-        i, ctx.buf_ptr          = i + 1, ctx.buf_ptr + 1
-    }
-    
-    if rem != 0 {
-        ctx.bits_in_last_byte    = rem
-        ctx.buffer[ctx.buf_ptr]  = data[i]
-        ctx.buf_ptr             += 1
-    }
-}
-
-final :: proc(ctx: ^Groestl_Context, hash: []byte) {
-    hashbytelen := ctx.hashbitlen / 8
-
-    if ctx.bits_in_last_byte != 0 {
-        ctx.buffer[ctx.buf_ptr - 1] &= ((1 << uint(ctx.bits_in_last_byte)) - 1) << (8 - uint(ctx.bits_in_last_byte))
-        ctx.buffer[ctx.buf_ptr - 1] ~= 0x1 << (7 - uint(ctx.bits_in_last_byte))
-    } else {
-        ctx.buffer[ctx.buf_ptr]  = 0x80
-        ctx.buf_ptr             += 1
-    }
-
-    if ctx.buf_ptr > ctx.statesize - 8 {
-        for ctx.buf_ptr < ctx.statesize {
-            ctx.buffer[ctx.buf_ptr]  = 0
-            ctx.buf_ptr             += 1
-        }
-        transform(ctx, ctx.buffer[:], u32(ctx.statesize))
-        ctx.buf_ptr = 0
-    }
-
-    for ctx.buf_ptr < ctx.statesize - 8 {
-        ctx.buffer[ctx.buf_ptr]  = 0
-        ctx.buf_ptr             += 1
-    }
-
-    ctx.block_counter += 1
-    ctx.buf_ptr        = ctx.statesize
-
-    for ctx.buf_ptr > ctx.statesize - 8 {
-        ctx.buf_ptr              -= 1
-        ctx.buffer[ctx.buf_ptr]   = byte(ctx.block_counter)
-        ctx.block_counter       >>= 8
-    }
-
-    transform(ctx, ctx.buffer[:], u32(ctx.statesize))
-    output_transformation(ctx)
-
-    for i, j := ctx.statesize - hashbytelen , 0; i < ctx.statesize; i, j = i + 1, j + 1 {
-        hash[j] = ctx.chaining[i % 8][i / 8]
-    }
-}
-
-/*
-    GROESTL implementation
-*/
-
-SBOX := [256]byte {
-    0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5,
-    0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76,
-    0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0,
-    0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0,
-    0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc,
-    0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15,
-    0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a,
-    0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75,
-    0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0,
-    0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84,
-    0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b,
-    0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf,
-    0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85,
-    0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8,
-    0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5,
-    0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2,
-    0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17,
-    0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73,
-    0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88,
-    0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb,
-    0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c,
-    0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79,
-    0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9,
-    0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08,
-    0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6,
-    0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a,
-    0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e,
-    0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e,
-    0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94,
-    0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf,
-    0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68,
-    0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16,
-}
-
-SHIFT := [2][2][8]int {
-    {{0, 1, 2, 3, 4, 5, 6, 7},  {1, 3, 5, 7,  0, 2, 4, 6}},
-    {{0, 1, 2, 3, 4, 5, 6, 11}, {1, 3, 5, 11, 0, 2, 4, 6}},
-}
-
-Groestl_Context :: struct {
-    chaining:          [8][16]byte,
-    block_counter:     u64,
-    hashbitlen:        int,
-    buffer:            [128]byte,
-    buf_ptr:           int,
-    bits_in_last_byte: int,
-    columns:           int,
-    rounds:            int,
-    statesize:         int,
-}
-
-Groestl_Variant :: enum {
-    P512  = 0, 
-    Q512  = 1, 
-    P1024 = 2, 
-    Q1024 = 3,
-}
-
-MUL2 :: #force_inline proc "contextless"(b: byte) -> byte {
-    return (b >> 7) != 0 ? (b << 1) ~ 0x1b : (b << 1)
-}
-
-MUL3 :: #force_inline proc "contextless"(b: byte) -> byte {
-    return MUL2(b) ~ b
-}
-
-MUL4 :: #force_inline proc "contextless"(b: byte) -> byte {
-    return MUL2(MUL2(b))
-}
-
-MUL5 :: #force_inline proc "contextless"(b: byte) -> byte {
-    return MUL4(b) ~ b
-}
-
-MUL6 :: #force_inline proc "contextless"(b: byte) -> byte {
-    return MUL4(b) ~ MUL2(b)
-}
-
-MUL7 :: #force_inline proc "contextless"(b: byte) -> byte {
-    return MUL4(b) ~ MUL2(b) ~ b
-}
-
-sub_bytes :: #force_inline proc (x: [][16]byte, columns: int) {
-    for i := 0; i < 8; i += 1 {
-        for j := 0; j < columns; j += 1 {
-            x[i][j] = SBOX[x[i][j]]
-        }
-    }
-}
-
-shift_bytes :: #force_inline proc (x: [][16]byte, columns: int, v: Groestl_Variant) {
-    temp: [16]byte
-    R := &SHIFT[int(v) / 2][int(v) & 1]
-
-    for i := 0; i < 8; i += 1 {
-        for j := 0; j < columns; j += 1 {
-            temp[j] = x[i][(j + R[i]) % columns]
-        }
-        for j := 0; j < columns; j += 1 {
-            x[i][j] = temp[j]
-        }
-    }
-}
-
-mix_bytes :: #force_inline proc (x: [][16]byte, columns: int) {
-    temp: [8]byte
-
-    for i := 0; i < columns; i += 1 {
-        for j := 0; j < 8; j += 1 {
-            temp[j] =  MUL2(x[(j + 0) % 8][i]) ~
-                       MUL2(x[(j + 1) % 8][i]) ~
-                       MUL3(x[(j + 2) % 8][i]) ~
-                       MUL4(x[(j + 3) % 8][i]) ~
-                       MUL5(x[(j + 4) % 8][i]) ~
-                       MUL3(x[(j + 5) % 8][i]) ~
-                       MUL5(x[(j + 6) % 8][i]) ~
-                       MUL7(x[(j + 7) % 8][i])
-        }
-        for j := 0; j < 8; j += 1 {
-            x[j][i] = temp[j]
-        }
-    }
-}
-
-p :: #force_inline proc (ctx: ^Groestl_Context, x: [][16]byte) {
-    v := ctx.columns == 8 ? Groestl_Variant.P512 : Groestl_Variant.P1024
-    for i := 0; i < ctx.rounds; i += 1 {
-        add_roundconstant(x, ctx.columns, byte(i), v)
-        sub_bytes(x, ctx.columns)
-        shift_bytes(x, ctx.columns, v)
-        mix_bytes(x, ctx.columns)
-    }
-}
-
-q :: #force_inline proc (ctx: ^Groestl_Context, x: [][16]byte) {
-    v := ctx.columns == 8 ? Groestl_Variant.Q512 : Groestl_Variant.Q1024
-    for i := 0; i < ctx.rounds; i += 1 {
-        add_roundconstant(x, ctx.columns, byte(i), v)
-        sub_bytes(x, ctx.columns)
-        shift_bytes(x, ctx.columns, v)
-        mix_bytes(x, ctx.columns)
-    }
-}
-
-transform :: proc(ctx: ^Groestl_Context, input: []byte, msglen: u32) {
-    tmp1, tmp2: [8][16]byte
-    input, msglen := input, msglen
-
-    for msglen >= u32(ctx.statesize) {
-        for i := 0; i < 8; i += 1 {
-            for j := 0; j < ctx.columns; j += 1 {
-                tmp1[i][j] = ctx.chaining[i][j] ~ input[j * 8 + i]
-                tmp2[i][j] = input[j * 8 + i]
-            }
-        }
-
-        p(ctx, tmp1[:])
-        q(ctx, tmp2[:])
-
-        for i := 0; i < 8; i += 1 {
-            for j := 0; j < ctx.columns; j += 1 {
-                ctx.chaining[i][j] ~= tmp1[i][j] ~ tmp2[i][j]
-            }
-        }
-
-        ctx.block_counter += 1
-        msglen            -= u32(ctx.statesize)
-        input              = input[ctx.statesize:]
-    }
-}
-
-output_transformation :: proc(ctx: ^Groestl_Context) {
-    temp: [8][16]byte
-
-    for i := 0; i < 8; i += 1 {
-        for j := 0; j < ctx.columns; j += 1 {
-            temp[i][j] = ctx.chaining[i][j]
-        }
-    }
-
-    p(ctx, temp[:])
-
-    for i := 0; i < 8; i += 1 {
-        for j := 0; j < ctx.columns; j += 1 {
-            ctx.chaining[i][j] ~= temp[i][j]
-        }
-    }
-}
-
-add_roundconstant :: proc(x: [][16]byte, columns: int, round: byte, v: Groestl_Variant) {
-    switch (i32(v) & 1) {
-        case 0: 
-            for i := 0; i < columns; i += 1 {
-                x[0][i] ~= byte(i << 4) ~ round
-            }
-        case 1:
-            for i := 0; i < columns; i += 1 {
-                for j := 0; j < 7; j += 1 {
-                    x[j][i] ~= 0xff
-                }
-            }
-            for i := 0; i < columns; i += 1 {
-                x[7][i] ~= byte(i << 4) ~ 0xff ~ round
-            }
-    }
-}
@@ -1,584 +0,0 @@
-package jh
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation of the JH hashing algorithm, as defined in <https://www3.ntu.edu.sg/home/wuhj/research/jh/index.html>
-*/
-
-import "core:os"
-import "core:io"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE_224 :: 28
-DIGEST_SIZE_256 :: 32
-DIGEST_SIZE_384 :: 48
-DIGEST_SIZE_512 :: 64
-
-// hash_string_224 will hash the given input and return the
-// computed hash
-hash_string_224 :: proc(data: string) -> [DIGEST_SIZE_224]byte {
-    return hash_bytes_224(transmute([]byte)(data))
-}
-
-// hash_bytes_224 will hash the given input and return the
-// computed hash
-hash_bytes_224 :: proc(data: []byte) -> [DIGEST_SIZE_224]byte {
-    hash: [DIGEST_SIZE_224]byte
-    ctx: Jh_Context
-    ctx.hashbitlen = 224
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_224 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_224 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_224(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_224 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_224 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_224, "Size of destination buffer is smaller than the digest size")
-    ctx: Jh_Context
-    ctx.hashbitlen = 224
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_224 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
-    hash: [DIGEST_SIZE_224]byte
-    ctx: Jh_Context
-    ctx.hashbitlen = 224
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_224 will read the file provided by the given handle
-// and compute a hash
-hash_file_224 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_224]byte, bool) {
-    if !load_at_once {
-        return hash_stream_224(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_224(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_224]byte{}, false
-}
-
-hash_224 :: proc {
-    hash_stream_224,
-    hash_file_224,
-    hash_bytes_224,
-    hash_string_224,
-    hash_bytes_to_buffer_224,
-    hash_string_to_buffer_224,
-}
-
-// hash_string_256 will hash the given input and return the
-// computed hash
-hash_string_256 :: proc(data: string) -> [DIGEST_SIZE_256]byte {
-    return hash_bytes_256(transmute([]byte)(data))
-}
-
-// hash_bytes_256 will hash the given input and return the
-// computed hash
-hash_bytes_256 :: proc(data: []byte) -> [DIGEST_SIZE_256]byte {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: Jh_Context
-    ctx.hashbitlen = 256
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_256 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_256 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_256(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_256 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_256 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_256, "Size of destination buffer is smaller than the digest size")
-    ctx: Jh_Context
-    ctx.hashbitlen = 256
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_256 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: Jh_Context
-    ctx.hashbitlen = 256
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_256 will read the file provided by the given handle
-// and compute a hash
-hash_file_256 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_256]byte, bool) {
-    if !load_at_once {
-        return hash_stream_256(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_256(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_256]byte{}, false
-}
-
-hash_256 :: proc {
-    hash_stream_256,
-    hash_file_256,
-    hash_bytes_256,
-    hash_string_256,
-    hash_bytes_to_buffer_256,
-    hash_string_to_buffer_256,
-}
-
-// hash_string_384 will hash the given input and return the
-// computed hash
-hash_string_384 :: proc(data: string) -> [DIGEST_SIZE_384]byte {
-    return hash_bytes_384(transmute([]byte)(data))
-}
-
-// hash_bytes_384 will hash the given input and return the
-// computed hash
-hash_bytes_384 :: proc(data: []byte) -> [DIGEST_SIZE_384]byte {
-    hash: [DIGEST_SIZE_384]byte
-    ctx: Jh_Context
-    ctx.hashbitlen = 384
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_384 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_384 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_384(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_384 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_384 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_384, "Size of destination buffer is smaller than the digest size")
-    ctx: Jh_Context
-    ctx.hashbitlen = 384
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_384 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
-    hash: [DIGEST_SIZE_384]byte
-    ctx: Jh_Context
-    ctx.hashbitlen = 384
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_384 will read the file provided by the given handle
-// and compute a hash
-hash_file_384 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_384]byte, bool) {
-    if !load_at_once {
-        return hash_stream_384(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_384(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_384]byte{}, false
-}
-
-hash_384 :: proc {
-    hash_stream_384,
-    hash_file_384,
-    hash_bytes_384,
-    hash_string_384,
-    hash_bytes_to_buffer_384,
-    hash_string_to_buffer_384,
-}
-
-// hash_string_512 will hash the given input and return the
-// computed hash
-hash_string_512 :: proc(data: string) -> [DIGEST_SIZE_512]byte {
-    return hash_bytes_512(transmute([]byte)(data))
-}
-
-// hash_bytes_512 will hash the given input and return the
-// computed hash
-hash_bytes_512 :: proc(data: []byte) -> [DIGEST_SIZE_512]byte {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: Jh_Context
-    ctx.hashbitlen = 512
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_512 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_512 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_512(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_512 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_512 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_512, "Size of destination buffer is smaller than the digest size")
-    ctx: Jh_Context
-    ctx.hashbitlen = 512
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_512 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: Jh_Context
-    ctx.hashbitlen = 512
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_512 will read the file provided by the given handle
-// and compute a hash
-hash_file_512 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_512]byte, bool) {
-    if !load_at_once {
-        return hash_stream_512(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_512(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_512]byte{}, false
-}
-
-hash_512 :: proc {
-    hash_stream_512,
-    hash_file_512,
-    hash_bytes_512,
-    hash_string_512,
-    hash_bytes_to_buffer_512,
-    hash_string_to_buffer_512,
-}
-
-/*
-    Low level API
-*/
-
-init :: proc(ctx: ^Jh_Context) {
-    assert(ctx.hashbitlen == 224 || ctx.hashbitlen == 256 || ctx.hashbitlen == 384 || ctx.hashbitlen == 512, "hashbitlen must be set to 224, 256, 384 or 512")
-    ctx.H[1] = byte(ctx.hashbitlen)      & 0xff
-    ctx.H[0] = byte(ctx.hashbitlen >> 8) & 0xff
-    F8(ctx)
-}
-
-update :: proc(ctx: ^Jh_Context, data: []byte) {
-    databitlen     := u64(len(data)) * 8
-    ctx.databitlen += databitlen
-    i              := u64(0)
-
-    if (ctx.buffer_size > 0) && ((ctx.buffer_size + databitlen) < 512) {
-        if (databitlen & 7) == 0 {
-            copy(ctx.buffer[ctx.buffer_size >> 3:], data[:64 - (ctx.buffer_size >> 3)])
-        } else {
-            copy(ctx.buffer[ctx.buffer_size >> 3:], data[:64 - (ctx.buffer_size >> 3) + 1])
-        } 
-        ctx.buffer_size += databitlen
-        databitlen = 0
-    }
-
-    if (ctx.buffer_size > 0 ) && ((ctx.buffer_size + databitlen) >= 512) {
-        copy(ctx.buffer[ctx.buffer_size >> 3:], data[:64 - (ctx.buffer_size >> 3)])
-        i      = 64 - (ctx.buffer_size >> 3)
-        databitlen = databitlen - (512 - ctx.buffer_size)
-        F8(ctx)
-        ctx.buffer_size = 0
-    }
-
-    for databitlen >= 512 {
-        copy(ctx.buffer[:], data[i:i + 64])
-        F8(ctx)
-        i += 64
-        databitlen -= 512
-    }
-
-    if databitlen > 0 {
-        if (databitlen & 7) == 0 {
-            copy(ctx.buffer[:], data[i:i + ((databitlen & 0x1ff) >> 3)])
-        } else {
-            copy(ctx.buffer[:], data[i:i + ((databitlen & 0x1ff) >> 3) + 1])
-        }
-        ctx.buffer_size = databitlen
-    }
-}
-
-final :: proc(ctx: ^Jh_Context, hash: []byte) {
-    if ctx.databitlen & 0x1ff == 0 {
-        for i := 0; i < 64; i += 1 {
-            ctx.buffer[i] = 0
-        }
-        ctx.buffer[0]  = 0x80
-        ctx.buffer[63] = byte(ctx.databitlen)       & 0xff
-        ctx.buffer[62] = byte(ctx.databitlen >> 8)  & 0xff
-        ctx.buffer[61] = byte(ctx.databitlen >> 16) & 0xff
-        ctx.buffer[60] = byte(ctx.databitlen >> 24) & 0xff
-        ctx.buffer[59] = byte(ctx.databitlen >> 32) & 0xff
-        ctx.buffer[58] = byte(ctx.databitlen >> 40) & 0xff
-        ctx.buffer[57] = byte(ctx.databitlen >> 48) & 0xff
-        ctx.buffer[56] = byte(ctx.databitlen >> 56) & 0xff
-        F8(ctx)
-    } else {
-        if ctx.buffer_size & 7 == 0 {
-            for i := (ctx.databitlen & 0x1ff) >> 3; i < 64; i += 1 {
-                ctx.buffer[i] = 0
-            }
-        } else {
-            for i := ((ctx.databitlen & 0x1ff) >> 3) + 1; i < 64; i += 1 {
-                ctx.buffer[i] = 0
-            }
-        }
-        ctx.buffer[(ctx.databitlen & 0x1ff) >> 3] |= 1 << (7 - (ctx.databitlen & 7))
-        F8(ctx)
-        for i := 0; i < 64; i += 1 {
-            ctx.buffer[i] = 0
-        }
-        ctx.buffer[63] = byte(ctx.databitlen)       & 0xff
-        ctx.buffer[62] = byte(ctx.databitlen >> 8)  & 0xff
-        ctx.buffer[61] = byte(ctx.databitlen >> 16) & 0xff
-        ctx.buffer[60] = byte(ctx.databitlen >> 24) & 0xff
-        ctx.buffer[59] = byte(ctx.databitlen >> 32) & 0xff
-        ctx.buffer[58] = byte(ctx.databitlen >> 40) & 0xff
-        ctx.buffer[57] = byte(ctx.databitlen >> 48) & 0xff
-        ctx.buffer[56] = byte(ctx.databitlen >> 56) & 0xff
-        F8(ctx)
-    }
-    switch ctx.hashbitlen {
-        case 224: copy(hash[:], ctx.H[100:128])
-        case 256: copy(hash[:], ctx.H[96:128])
-        case 384: copy(hash[:], ctx.H[80:128])
-        case 512: copy(hash[:], ctx.H[64:128])
-    }
-}
-
-/*
-    JH implementation
-*/
-
-ROUNDCONSTANT_ZERO := [64]byte {
-    0x6, 0xa, 0x0, 0x9, 0xe, 0x6, 0x6, 0x7,
-    0xf, 0x3, 0xb, 0xc, 0xc, 0x9, 0x0, 0x8,
-    0xb, 0x2, 0xf, 0xb, 0x1, 0x3, 0x6, 0x6,
-    0xe, 0xa, 0x9, 0x5, 0x7, 0xd, 0x3, 0xe,
-    0x3, 0xa, 0xd, 0xe, 0xc, 0x1, 0x7, 0x5,
-    0x1, 0x2, 0x7, 0x7, 0x5, 0x0, 0x9, 0x9,
-    0xd, 0xa, 0x2, 0xf, 0x5, 0x9, 0x0, 0xb,
-    0x0, 0x6, 0x6, 0x7, 0x3, 0x2, 0x2, 0xa,
-}
-
-SBOX := [2][16]byte {
-    {9, 0,  4, 11, 13, 12, 3, 15, 1,  10, 2, 6, 7,  5,  8,  14},
-    {3, 12, 6, 13, 5,  7,  1, 9,  15, 2,  0, 4, 11, 10, 14, 8},
-}
-
-Jh_Context :: struct {
-    hashbitlen:    int,
-    databitlen:    u64,
-    buffer_size:   u64,
-    H:             [128]byte,
-    A:             [256]byte,
-    roundconstant: [64]byte,
-    buffer:        [64]byte,
-}
-
-E8_finaldegroup :: proc(ctx: ^Jh_Context) {
-    t0,t1,t2,t3: byte
-    tem: [256]byte
-    for i := 0; i < 128; i += 1 {
-        tem[i]       = ctx.A[i << 1]
-        tem[i + 128] = ctx.A[(i << 1) + 1]
-    }
-    for i := 0; i < 128; i += 1 {
-        ctx.H[i] = 0
-    }
-    for i := 0; i < 256; i += 1 {
-        t0 = (tem[i] >> 3) & 1
-        t1 = (tem[i] >> 2) & 1
-        t2 = (tem[i] >> 1) & 1
-        t3 = (tem[i] >> 0) & 1
-
-        ctx.H[uint(i) >> 3]         |= t0 << (7 - (uint(i) & 7))
-        ctx.H[(uint(i) + 256) >> 3] |= t1 << (7 - (uint(i) & 7))
-        ctx.H[(uint(i) + 512) >> 3] |= t2 << (7 - (uint(i) & 7))
-        ctx.H[(uint(i) + 768) >> 3] |= t3 << (7 - (uint(i) & 7))
-    }
-}
-
-update_roundconstant :: proc(ctx: ^Jh_Context) {
-    tem: [64]byte
-    t: byte
-    for i := 0; i < 64; i += 1 {
-        tem[i] = SBOX[0][ctx.roundconstant[i]]
-    }
-    for i := 0; i < 64; i += 2 {
-        tem[i + 1] ~= ((tem[i]   << 1)   ~ (tem[i]   >> 3)   ~ ((tem[i]   >> 2) & 2))   & 0xf
-        tem[i]     ~= ((tem[i + 1] << 1) ~ (tem[i + 1] >> 3) ~ ((tem[i + 1] >> 2) & 2)) & 0xf
-    }
-    for i := 0; i < 64; i += 4 {
-        t          = tem[i + 2]
-        tem[i + 2] = tem[i + 3]
-        tem[i + 3] = t
-    }
-    for i := 0; i < 32; i += 1 {
-        ctx.roundconstant[i]      = tem[i << 1]
-        ctx.roundconstant[i + 32] = tem[(i << 1) + 1]
-    }
-    for i := 32; i < 64; i += 2 {
-        t                        = ctx.roundconstant[i]
-        ctx.roundconstant[i]     = ctx.roundconstant[i + 1]
-        ctx.roundconstant[i + 1] = t
-    }
-}
-
-R8 :: proc(ctx: ^Jh_Context) {
-    t: byte
-    tem, roundconstant_expanded: [256]byte
-    for i := u32(0); i < 256; i += 1 {
-        roundconstant_expanded[i] = (ctx.roundconstant[i >> 2] >> (3 - (i & 3)) ) & 1
-    }
-    for i := 0; i < 256; i += 1 {
-        tem[i] = SBOX[roundconstant_expanded[i]][ctx.A[i]]
-    }
-    for i := 0; i < 256; i += 2 {
-        tem[i+1] ~= ((tem[i]   << 1)   ~ (tem[i]   >> 3)   ~ ((tem[i]   >> 2) & 2))   & 0xf
-        tem[i]   ~= ((tem[i + 1] << 1) ~ (tem[i + 1] >> 3) ~ ((tem[i + 1] >> 2) & 2)) & 0xf
-    }
-    for i := 0; i < 256; i += 4 {
-        t        = tem[i + 2]
-        tem[i+2] = tem[i + 3]
-        tem[i+3] = t
-    }
-    for i := 0; i < 128; i += 1 {
-        ctx.A[i]       = tem[i << 1]
-        ctx.A[i + 128] = tem[(i << 1) + 1]
-    }
-    for i := 128; i < 256; i += 2 {
-        t            = ctx.A[i]
-        ctx.A[i]     = ctx.A[i + 1]
-        ctx.A[i + 1] = t
-    }
-}
-
-E8_initialgroup :: proc(ctx: ^Jh_Context) {
-    t0, t1, t2, t3: byte
-    tem:            [256]byte
-    for i := u32(0); i < 256; i += 1 {
-        t0     = (ctx.H[i >> 3]   >> (7 - (i & 7)))       & 1
-        t1     = (ctx.H[(i + 256) >> 3] >> (7 - (i & 7))) & 1
-        t2     = (ctx.H[(i + 512) >> 3] >> (7 - (i & 7))) & 1
-        t3     = (ctx.H[(i + 768) >> 3] >> (7 - (i & 7))) & 1
-        tem[i] = (t0 << 3) | (t1 << 2) | (t2 << 1) | (t3 << 0)
-    }
-    for i := 0; i < 128; i += 1 {
-        ctx.A[i << 1]       = tem[i]
-        ctx.A[(i << 1) + 1] = tem[i + 128]
-    }
-}
-
-E8 :: proc(ctx: ^Jh_Context) {
-    for i := 0; i < 64; i += 1 {
-        ctx.roundconstant[i] = ROUNDCONSTANT_ZERO[i]
-    }
-    E8_initialgroup(ctx)
-    for i := 0; i < 42; i += 1 {
-        R8(ctx)
-        update_roundconstant(ctx)
-    }
-    E8_finaldegroup(ctx)
-}
-
-F8 :: proc(ctx: ^Jh_Context) {
-    for i := 0; i < 64; i += 1 {
-        ctx.H[i] ~= ctx.buffer[i]
-    }
-    E8(ctx)
-    for i := 0; i < 64; i += 1 {
-        ctx.H[i + 64] ~= ctx.buffer[i]
-    }
-}
@@ -1,374 +0,0 @@
-package keccak
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Interface for the Keccak hashing algorithm.
-    This is done because the padding in the SHA3 standard was changed by the NIST, resulting in a different output.
-*/
-
-import "core:os"
-import "core:io"
-
-import "../_sha3"
-
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE_224 :: 28
-DIGEST_SIZE_256 :: 32
-DIGEST_SIZE_384 :: 48
-DIGEST_SIZE_512 :: 64
-
-// hash_string_224 will hash the given input and return the
-// computed hash
-hash_string_224 :: proc(data: string) -> [DIGEST_SIZE_224]byte {
-    return hash_bytes_224(transmute([]byte)(data))
-}
-
-// hash_bytes_224 will hash the given input and return the
-// computed hash
-hash_bytes_224 :: proc(data: []byte) -> [DIGEST_SIZE_224]byte {
-    hash: [DIGEST_SIZE_224]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_224
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_224 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_224 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_224(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_224 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_224 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_224, "Size of destination buffer is smaller than the digest size")
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_224
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash)
-}
-
-// hash_stream_224 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
-    hash: [DIGEST_SIZE_224]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_224
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _sha3.update(&ctx, buf[:read])
-        } 
-    }
-    _sha3.final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_224 will read the file provided by the given handle
-// and compute a hash
-hash_file_224 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_224]byte, bool) {
-    if !load_at_once {
-        return hash_stream_224(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_224(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_224]byte{}, false
-}
-
-hash_224 :: proc {
-    hash_stream_224,
-    hash_file_224,
-    hash_bytes_224,
-    hash_string_224,
-    hash_bytes_to_buffer_224,
-    hash_string_to_buffer_224,
-}
-
-// hash_string_256 will hash the given input and return the
-// computed hash
-hash_string_256 :: proc(data: string) -> [DIGEST_SIZE_256]byte {
-    return hash_bytes_256(transmute([]byte)(data))
-}
-
-// hash_bytes_256 will hash the given input and return the
-// computed hash
-hash_bytes_256 :: proc(data: []byte) -> [DIGEST_SIZE_256]byte {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_256
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_256 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_256 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_256(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_256 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_256 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_256, "Size of destination buffer is smaller than the digest size")
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_256
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash)
-}
-
-// hash_stream_256 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_256
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _sha3.update(&ctx, buf[:read])
-        } 
-    }
-    _sha3.final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_256 will read the file provided by the given handle
-// and compute a hash
-hash_file_256 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_256]byte, bool) {
-    if !load_at_once {
-        return hash_stream_256(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_256(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_256]byte{}, false
-}
-
-hash_256 :: proc {
-    hash_stream_256,
-    hash_file_256,
-    hash_bytes_256,
-    hash_string_256,
-    hash_bytes_to_buffer_256,
-    hash_string_to_buffer_256,
-}
-
-// hash_string_384 will hash the given input and return the
-// computed hash
-hash_string_384 :: proc(data: string) -> [DIGEST_SIZE_384]byte {
-    return hash_bytes_384(transmute([]byte)(data))
-}
-
-// hash_bytes_384 will hash the given input and return the
-// computed hash
-hash_bytes_384 :: proc(data: []byte) -> [DIGEST_SIZE_384]byte {
-    hash: [DIGEST_SIZE_384]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_384
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_384 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_384 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_384(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_384 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_384 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_384, "Size of destination buffer is smaller than the digest size")
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_384
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash)
-}
-
-// hash_stream_384 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
-    hash: [DIGEST_SIZE_384]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_384
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _sha3.update(&ctx, buf[:read])
-        } 
-    }
-    _sha3.final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_384 will read the file provided by the given handle
-// and compute a hash
-hash_file_384 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_384]byte, bool) {
-    if !load_at_once {
-        return hash_stream_384(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_384(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_384]byte{}, false
-}
-
-hash_384 :: proc {
-    hash_stream_384,
-    hash_file_384,
-    hash_bytes_384,
-    hash_string_384,
-    hash_bytes_to_buffer_384,
-    hash_string_to_buffer_384,
-}
-
-// hash_string_512 will hash the given input and return the
-// computed hash
-hash_string_512 :: proc(data: string) -> [DIGEST_SIZE_512]byte {
-    return hash_bytes_512(transmute([]byte)(data))
-}
-
-// hash_bytes_512 will hash the given input and return the
-// computed hash
-hash_bytes_512 :: proc(data: []byte) -> [DIGEST_SIZE_512]byte {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_512
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_512 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_512 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_512(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_512 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_512 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_512, "Size of destination buffer is smaller than the digest size")
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_512
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash)
-}
-
-// hash_stream_512 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_512
-    ctx.is_keccak = true
-    _sha3.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _sha3.update(&ctx, buf[:read])
-        } 
-    }
-    _sha3.final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_512 will read the file provided by the given handle
-// and compute a hash
-hash_file_512 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_512]byte, bool) {
-    if !load_at_once {
-        return hash_stream_512(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_512(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_512]byte{}, false
-}
-
-hash_512 :: proc {
-    hash_stream_512,
-    hash_file_512,
-    hash_bytes_512,
-    hash_string_512,
-    hash_bytes_to_buffer_512,
-    hash_string_to_buffer_512,
-}
-
-/*
-    Low level API
-*/
-
-Keccak_Context :: _sha3.Sha3_Context
-
-init :: proc(ctx: ^_sha3.Sha3_Context) {
-    ctx.is_keccak = true
-    _sha3.init(ctx)
-}
-
-update :: proc "contextless" (ctx: ^_sha3.Sha3_Context, data: []byte) {
-    _sha3.update(ctx, data)
-}
-
-final :: proc "contextless" (ctx: ^_sha3.Sha3_Context, hash: []byte) {
-    _sha3.final(ctx, hash)
-}
@@ -0,0 +1,10 @@
+# crypto/legacy
+
+These are algorithms that are shipped solely for the purpose of
+interoperability with legacy systems.  The use of these packages in
+any other capacity is discouraged, especially those that are known
+to be broken.
+
+- keccak - The draft version of the algorithm that became SHA-3
+- MD5 - Broken (https://eprint.iacr.org/2005/075)
+- SHA-1 - Broken (https://eprint.iacr.org/2017/190)
@@ -0,0 +1,377 @@
+package keccak
+
+/*
+    Copyright 2021 zhibog
+    Made available under the BSD-3 license.
+
+    List of contributors:
+        zhibog, dotbmp:  Initial implementation.
+
+    Interface for the Keccak hashing algorithm.
+    This is done because the padding in the SHA3 standard was changed by the NIST, resulting in a different output.
+*/
+
+import "core:io"
+import "core:os"
+
+import "../../_sha3"
+
+/*
+    High level API
+*/
+
+DIGEST_SIZE_224 :: 28
+DIGEST_SIZE_256 :: 32
+DIGEST_SIZE_384 :: 48
+DIGEST_SIZE_512 :: 64
+
+// hash_string_224 will hash the given input and return the
+// computed hash
+hash_string_224 :: proc(data: string) -> [DIGEST_SIZE_224]byte {
+	return hash_bytes_224(transmute([]byte)(data))
+}
+
+// hash_bytes_224 will hash the given input and return the
+// computed hash
+hash_bytes_224 :: proc(data: []byte) -> [DIGEST_SIZE_224]byte {
+	hash: [DIGEST_SIZE_224]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_224
+	ctx.is_keccak = true
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
+}
+
+// hash_string_to_buffer_224 will hash the given input and assign the
+// computed hash to the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_string_to_buffer_224 :: proc(data: string, hash: []byte) {
+	hash_bytes_to_buffer_224(transmute([]byte)(data), hash)
+}
+
+// hash_bytes_to_buffer_224 will hash the given input and write the
+// computed hash into the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_bytes_to_buffer_224 :: proc(data, hash: []byte) {
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_224
+	ctx.is_keccak = true
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
+}
+
+// hash_stream_224 will read the stream in chunks and compute a
+// hash from its contents
+hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
+	hash: [DIGEST_SIZE_224]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_224
+	ctx.is_keccak = true
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
+}
+
+// hash_file_224 will read the file provided by the given handle
+// and compute a hash
+hash_file_224 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_224]byte, bool) {
+	if !load_at_once {
+		return hash_stream_224(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes_224(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE_224]byte{}, false
+}
+
+hash_224 :: proc {
+	hash_stream_224,
+	hash_file_224,
+	hash_bytes_224,
+	hash_string_224,
+	hash_bytes_to_buffer_224,
+	hash_string_to_buffer_224,
+}
+
+// hash_string_256 will hash the given input and return the
+// computed hash
+hash_string_256 :: proc(data: string) -> [DIGEST_SIZE_256]byte {
+	return hash_bytes_256(transmute([]byte)(data))
+}
+
+// hash_bytes_256 will hash the given input and return the
+// computed hash
+hash_bytes_256 :: proc(data: []byte) -> [DIGEST_SIZE_256]byte {
+	hash: [DIGEST_SIZE_256]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_256
+	ctx.is_keccak = true
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
+}
+
+// hash_string_to_buffer_256 will hash the given input and assign the
+// computed hash to the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_string_to_buffer_256 :: proc(data: string, hash: []byte) {
+	hash_bytes_to_buffer_256(transmute([]byte)(data), hash)
+}
+
+// hash_bytes_to_buffer_256 will hash the given input and write the
+// computed hash into the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_bytes_to_buffer_256 :: proc(data, hash: []byte) {
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_256
+	ctx.is_keccak = true
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
+}
+
+// hash_stream_256 will read the stream in chunks and compute a
+// hash from its contents
+hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
+	hash: [DIGEST_SIZE_256]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_256
+	ctx.is_keccak = true
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
+}
+
+// hash_file_256 will read the file provided by the given handle
+// and compute a hash
+hash_file_256 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_256]byte, bool) {
+	if !load_at_once {
+		return hash_stream_256(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes_256(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE_256]byte{}, false
+}
+
+hash_256 :: proc {
+	hash_stream_256,
+	hash_file_256,
+	hash_bytes_256,
+	hash_string_256,
+	hash_bytes_to_buffer_256,
+	hash_string_to_buffer_256,
+}
+
+// hash_string_384 will hash the given input and return the
+// computed hash
+hash_string_384 :: proc(data: string) -> [DIGEST_SIZE_384]byte {
+	return hash_bytes_384(transmute([]byte)(data))
+}
+
+// hash_bytes_384 will hash the given input and return the
+// computed hash
+hash_bytes_384 :: proc(data: []byte) -> [DIGEST_SIZE_384]byte {
+	hash: [DIGEST_SIZE_384]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_384
+	ctx.is_keccak = true
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
+}
+
+// hash_string_to_buffer_384 will hash the given input and assign the
+// computed hash to the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_string_to_buffer_384 :: proc(data: string, hash: []byte) {
+	hash_bytes_to_buffer_384(transmute([]byte)(data), hash)
+}
+
+// hash_bytes_to_buffer_384 will hash the given input and write the
+// computed hash into the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_bytes_to_buffer_384 :: proc(data, hash: []byte) {
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_384
+	ctx.is_keccak = true
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
+}
+
+// hash_stream_384 will read the stream in chunks and compute a
+// hash from its contents
+hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
+	hash: [DIGEST_SIZE_384]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_384
+	ctx.is_keccak = true
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
+}
+
+// hash_file_384 will read the file provided by the given handle
+// and compute a hash
+hash_file_384 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_384]byte, bool) {
+	if !load_at_once {
+		return hash_stream_384(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes_384(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE_384]byte{}, false
+}
+
+hash_384 :: proc {
+	hash_stream_384,
+	hash_file_384,
+	hash_bytes_384,
+	hash_string_384,
+	hash_bytes_to_buffer_384,
+	hash_string_to_buffer_384,
+}
+
+// hash_string_512 will hash the given input and return the
+// computed hash
+hash_string_512 :: proc(data: string) -> [DIGEST_SIZE_512]byte {
+	return hash_bytes_512(transmute([]byte)(data))
+}
+
+// hash_bytes_512 will hash the given input and return the
+// computed hash
+hash_bytes_512 :: proc(data: []byte) -> [DIGEST_SIZE_512]byte {
+	hash: [DIGEST_SIZE_512]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_512
+	ctx.is_keccak = true
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
+}
+
+// hash_string_to_buffer_512 will hash the given input and assign the
+// computed hash to the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_string_to_buffer_512 :: proc(data: string, hash: []byte) {
+	hash_bytes_to_buffer_512(transmute([]byte)(data), hash)
+}
+
+// hash_bytes_to_buffer_512 will hash the given input and write the
+// computed hash into the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_bytes_to_buffer_512 :: proc(data, hash: []byte) {
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_512
+	ctx.is_keccak = true
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
+}
+
+// hash_stream_512 will read the stream in chunks and compute a
+// hash from its contents
+hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
+	hash: [DIGEST_SIZE_512]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_512
+	ctx.is_keccak = true
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
+}
+
+// hash_file_512 will read the file provided by the given handle
+// and compute a hash
+hash_file_512 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_512]byte, bool) {
+	if !load_at_once {
+		return hash_stream_512(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes_512(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE_512]byte{}, false
+}
+
+hash_512 :: proc {
+	hash_stream_512,
+	hash_file_512,
+	hash_bytes_512,
+	hash_string_512,
+	hash_bytes_to_buffer_512,
+	hash_string_to_buffer_512,
+}
+
+/*
+    Low level API
+*/
+
+Context :: _sha3.Sha3_Context
+
+init :: proc(ctx: ^Context) {
+	ctx.is_keccak = true
+	_sha3.init(ctx)
+}
+
+update :: proc(ctx: ^Context, data: []byte) {
+	_sha3.update(ctx, data)
+}
+
+final :: proc(ctx: ^Context, hash: []byte) {
+	_sha3.final(ctx, hash)
+}
@@ -0,0 +1,295 @@
+package md5
+
+/*
+    Copyright 2021 zhibog
+    Made available under the BSD-3 license.
+
+    List of contributors:
+        zhibog, dotbmp:  Initial implementation.
+
+    Implementation of the MD5 hashing algorithm, as defined in RFC 1321 <https://datatracker.ietf.org/doc/html/rfc1321>
+*/
+
+import "core:encoding/endian"
+import "core:io"
+import "core:math/bits"
+import "core:mem"
+import "core:os"
+
+/*
+    High level API
+*/
+
+DIGEST_SIZE :: 16
+
+// hash_string will hash the given input and return the
+// computed hash
+hash_string :: proc(data: string) -> [DIGEST_SIZE]byte {
+	return hash_bytes(transmute([]byte)(data))
+}
+
+// hash_bytes will hash the given input and return the
+// computed hash
+hash_bytes :: proc(data: []byte) -> [DIGEST_SIZE]byte {
+	hash: [DIGEST_SIZE]byte
+	ctx: Context
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
+}
+
+// hash_string_to_buffer will hash the given input and assign the
+// computed hash to the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_string_to_buffer :: proc(data: string, hash: []byte) {
+	hash_bytes_to_buffer(transmute([]byte)(data), hash)
+}
+
+// hash_bytes_to_buffer will hash the given input and write the
+// computed hash into the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_bytes_to_buffer :: proc(data, hash: []byte) {
+	ctx: Context
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
+}
+
+// hash_stream will read the stream in chunks and compute a
+// hash from its contents
+hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
+	hash: [DIGEST_SIZE]byte
+	ctx: Context
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
+}
+
+// hash_file will read the file provided by the given handle
+// and compute a hash
+hash_file :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE]byte, bool) {
+	if !load_at_once {
+		return hash_stream(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE]byte{}, false
+}
+
+hash :: proc {
+	hash_stream,
+	hash_file,
+	hash_bytes,
+	hash_string,
+	hash_bytes_to_buffer,
+	hash_string_to_buffer,
+}
+
+/*
+    Low level API
+*/
+
+init :: proc(ctx: ^Context) {
+	ctx.state[0] = 0x67452301
+	ctx.state[1] = 0xefcdab89
+	ctx.state[2] = 0x98badcfe
+	ctx.state[3] = 0x10325476
+
+	ctx.bitlen = 0
+	ctx.datalen = 0
+
+	ctx.is_initialized = true
+}
+
+update :: proc(ctx: ^Context, data: []byte) {
+	assert(ctx.is_initialized)
+
+	for i := 0; i < len(data); i += 1 {
+		ctx.data[ctx.datalen] = data[i]
+		ctx.datalen += 1
+		if (ctx.datalen == BLOCK_SIZE) {
+			transform(ctx, ctx.data[:])
+			ctx.bitlen += 512
+			ctx.datalen = 0
+		}
+	}
+}
+
+final :: proc(ctx: ^Context, hash: []byte) {
+	assert(ctx.is_initialized)
+
+	if len(hash) < DIGEST_SIZE {
+		panic("crypto/md5: invalid destination digest size")
+	}
+
+	i := ctx.datalen
+
+	if ctx.datalen < 56 {
+		ctx.data[i] = 0x80
+		i += 1
+		for i < 56 {
+			ctx.data[i] = 0x00
+			i += 1
+		}
+	} else if ctx.datalen >= 56 {
+		ctx.data[i] = 0x80
+		i += 1
+		for i < BLOCK_SIZE {
+			ctx.data[i] = 0x00
+			i += 1
+		}
+		transform(ctx, ctx.data[:])
+		mem.set(&ctx.data, 0, 56)
+	}
+
+	ctx.bitlen += u64(ctx.datalen * 8)
+	endian.unchecked_put_u64le(ctx.data[56:], ctx.bitlen)
+	transform(ctx, ctx.data[:])
+
+	for i = 0; i < DIGEST_SIZE / 4; i += 1 {
+		endian.unchecked_put_u32le(hash[i * 4:], ctx.state[i])
+	}
+
+	ctx.is_initialized = false
+}
+
+/*
+    MD5 implementation
+*/
+
+BLOCK_SIZE :: 64
+
+Context :: struct {
+	data:    [BLOCK_SIZE]byte,
+	state:   [4]u32,
+	bitlen:  u64,
+	datalen: u32,
+
+	is_initialized: bool,
+}
+
+/*
+    @note(zh): F, G, H and I, as mentioned in the RFC, have been inlined into FF, GG, HH
+    and II respectively, instead of declaring them separately.
+*/
+
+@(private)
+FF :: #force_inline proc "contextless" (a, b, c, d, m: u32, s: int, t: u32) -> u32 {
+	return b + bits.rotate_left32(a + ((b & c) | (~b & d)) + m + t, s)
+}
+
+@(private)
+GG :: #force_inline proc "contextless" (a, b, c, d, m: u32, s: int, t: u32) -> u32 {
+	return b + bits.rotate_left32(a + ((b & d) | (c & ~d)) + m + t, s)
+}
+
+@(private)
+HH :: #force_inline proc "contextless" (a, b, c, d, m: u32, s: int, t: u32) -> u32 {
+	return b + bits.rotate_left32(a + (b ~ c ~ d) + m + t, s)
+}
+
+@(private)
+II :: #force_inline proc "contextless" (a, b, c, d, m: u32, s: int, t: u32) -> u32 {
+	return b + bits.rotate_left32(a + (c ~ (b | ~d)) + m + t, s)
+}
+
+@(private)
+transform :: proc "contextless" (ctx: ^Context, data: []byte) {
+	m: [DIGEST_SIZE]u32
+
+	for i := 0; i < DIGEST_SIZE; i += 1 {
+		m[i] = endian.unchecked_get_u32le(data[i * 4:])
+	}
+
+	a := ctx.state[0]
+	b := ctx.state[1]
+	c := ctx.state[2]
+	d := ctx.state[3]
+
+	a = FF(a, b, c, d, m[0], 7, 0xd76aa478)
+	d = FF(d, a, b, c, m[1], 12, 0xe8c7b756)
+	c = FF(c, d, a, b, m[2], 17, 0x242070db)
+	b = FF(b, c, d, a, m[3], 22, 0xc1bdceee)
+	a = FF(a, b, c, d, m[4], 7, 0xf57c0faf)
+	d = FF(d, a, b, c, m[5], 12, 0x4787c62a)
+	c = FF(c, d, a, b, m[6], 17, 0xa8304613)
+	b = FF(b, c, d, a, m[7], 22, 0xfd469501)
+	a = FF(a, b, c, d, m[8], 7, 0x698098d8)
+	d = FF(d, a, b, c, m[9], 12, 0x8b44f7af)
+	c = FF(c, d, a, b, m[10], 17, 0xffff5bb1)
+	b = FF(b, c, d, a, m[11], 22, 0x895cd7be)
+	a = FF(a, b, c, d, m[12], 7, 0x6b901122)
+	d = FF(d, a, b, c, m[13], 12, 0xfd987193)
+	c = FF(c, d, a, b, m[14], 17, 0xa679438e)
+	b = FF(b, c, d, a, m[15], 22, 0x49b40821)
+
+	a = GG(a, b, c, d, m[1], 5, 0xf61e2562)
+	d = GG(d, a, b, c, m[6], 9, 0xc040b340)
+	c = GG(c, d, a, b, m[11], 14, 0x265e5a51)
+	b = GG(b, c, d, a, m[0], 20, 0xe9b6c7aa)
+	a = GG(a, b, c, d, m[5], 5, 0xd62f105d)
+	d = GG(d, a, b, c, m[10], 9, 0x02441453)
+	c = GG(c, d, a, b, m[15], 14, 0xd8a1e681)
+	b = GG(b, c, d, a, m[4], 20, 0xe7d3fbc8)
+	a = GG(a, b, c, d, m[9], 5, 0x21e1cde6)
+	d = GG(d, a, b, c, m[14], 9, 0xc33707d6)
+	c = GG(c, d, a, b, m[3], 14, 0xf4d50d87)
+	b = GG(b, c, d, a, m[8], 20, 0x455a14ed)
+	a = GG(a, b, c, d, m[13], 5, 0xa9e3e905)
+	d = GG(d, a, b, c, m[2], 9, 0xfcefa3f8)
+	c = GG(c, d, a, b, m[7], 14, 0x676f02d9)
+	b = GG(b, c, d, a, m[12], 20, 0x8d2a4c8a)
+
+	a = HH(a, b, c, d, m[5], 4, 0xfffa3942)
+	d = HH(d, a, b, c, m[8], 11, 0x8771f681)
+	c = HH(c, d, a, b, m[11], 16, 0x6d9d6122)
+	b = HH(b, c, d, a, m[14], 23, 0xfde5380c)
+	a = HH(a, b, c, d, m[1], 4, 0xa4beea44)
+	d = HH(d, a, b, c, m[4], 11, 0x4bdecfa9)
+	c = HH(c, d, a, b, m[7], 16, 0xf6bb4b60)
+	b = HH(b, c, d, a, m[10], 23, 0xbebfbc70)
+	a = HH(a, b, c, d, m[13], 4, 0x289b7ec6)
+	d = HH(d, a, b, c, m[0], 11, 0xeaa127fa)
+	c = HH(c, d, a, b, m[3], 16, 0xd4ef3085)
+	b = HH(b, c, d, a, m[6], 23, 0x04881d05)
+	a = HH(a, b, c, d, m[9], 4, 0xd9d4d039)
+	d = HH(d, a, b, c, m[12], 11, 0xe6db99e5)
+	c = HH(c, d, a, b, m[15], 16, 0x1fa27cf8)
+	b = HH(b, c, d, a, m[2], 23, 0xc4ac5665)
+
+	a = II(a, b, c, d, m[0], 6, 0xf4292244)
+	d = II(d, a, b, c, m[7], 10, 0x432aff97)
+	c = II(c, d, a, b, m[14], 15, 0xab9423a7)
+	b = II(b, c, d, a, m[5], 21, 0xfc93a039)
+	a = II(a, b, c, d, m[12], 6, 0x655b59c3)
+	d = II(d, a, b, c, m[3], 10, 0x8f0ccc92)
+	c = II(c, d, a, b, m[10], 15, 0xffeff47d)
+	b = II(b, c, d, a, m[1], 21, 0x85845dd1)
+	a = II(a, b, c, d, m[8], 6, 0x6fa87e4f)
+	d = II(d, a, b, c, m[15], 10, 0xfe2ce6e0)
+	c = II(c, d, a, b, m[6], 15, 0xa3014314)
+	b = II(b, c, d, a, m[13], 21, 0x4e0811a1)
+	a = II(a, b, c, d, m[4], 6, 0xf7537e82)
+	d = II(d, a, b, c, m[11], 10, 0xbd3af235)
+	c = II(c, d, a, b, m[2], 15, 0x2ad7d2bb)
+	b = II(b, c, d, a, m[9], 21, 0xeb86d391)
+
+	ctx.state[0] += a
+	ctx.state[1] += b
+	ctx.state[2] += c
+	ctx.state[3] += d
+}
@@ -0,0 +1,252 @@
+package sha1
+
+/*
+    Copyright 2021 zhibog
+    Made available under the BSD-3 license.
+
+    List of contributors:
+        zhibog, dotbmp:  Initial implementation.
+
+    Implementation of the SHA1 hashing algorithm, as defined in RFC 3174 <https://datatracker.ietf.org/doc/html/rfc3174>
+*/
+
+import "core:encoding/endian"
+import "core:io"
+import "core:math/bits"
+import "core:mem"
+import "core:os"
+
+/*
+    High level API
+*/
+
+DIGEST_SIZE :: 20
+
+// hash_string will hash the given input and return the
+// computed hash
+hash_string :: proc(data: string) -> [DIGEST_SIZE]byte {
+	return hash_bytes(transmute([]byte)(data))
+}
+
+// hash_bytes will hash the given input and return the
+// computed hash
+hash_bytes :: proc(data: []byte) -> [DIGEST_SIZE]byte {
+	hash: [DIGEST_SIZE]byte
+	ctx: Context
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
+}
+
+// hash_string_to_buffer will hash the given input and assign the
+// computed hash to the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_string_to_buffer :: proc(data: string, hash: []byte) {
+	hash_bytes_to_buffer(transmute([]byte)(data), hash)
+}
+
+// hash_bytes_to_buffer will hash the given input and write the
+// computed hash into the second parameter.
+// It requires that the destination buffer is at least as big as the digest size
+hash_bytes_to_buffer :: proc(data, hash: []byte) {
+	ctx: Context
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
+}
+
+// hash_stream will read the stream in chunks and compute a
+// hash from its contents
+hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
+	hash: [DIGEST_SIZE]byte
+	ctx: Context
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
+}
+
+// hash_file will read the file provided by the given handle
+// and compute a hash
+hash_file :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE]byte, bool) {
+	if !load_at_once {
+		return hash_stream(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE]byte{}, false
+}
+
+hash :: proc {
+	hash_stream,
+	hash_file,
+	hash_bytes,
+	hash_string,
+	hash_bytes_to_buffer,
+	hash_string_to_buffer,
+}
+
+/*
+    Low level API
+*/
+
+init :: proc(ctx: ^Context) {
+	ctx.state[0] = 0x67452301
+	ctx.state[1] = 0xefcdab89
+	ctx.state[2] = 0x98badcfe
+	ctx.state[3] = 0x10325476
+	ctx.state[4] = 0xc3d2e1f0
+	ctx.k[0] = 0x5a827999
+	ctx.k[1] = 0x6ed9eba1
+	ctx.k[2] = 0x8f1bbcdc
+	ctx.k[3] = 0xca62c1d6
+
+	ctx.datalen = 0
+	ctx.bitlen = 0
+
+	ctx.is_initialized = true
+}
+
+update :: proc(ctx: ^Context, data: []byte) {
+	assert(ctx.is_initialized)
+
+	for i := 0; i < len(data); i += 1 {
+		ctx.data[ctx.datalen] = data[i]
+		ctx.datalen += 1
+		if (ctx.datalen == BLOCK_SIZE) {
+			transform(ctx, ctx.data[:])
+			ctx.bitlen += 512
+			ctx.datalen = 0
+		}
+	}
+}
+
+final :: proc(ctx: ^Context, hash: []byte) {
+	assert(ctx.is_initialized)
+
+	if len(hash) < DIGEST_SIZE {
+		panic("crypto/sha1: invalid destination digest size")
+	}
+
+	i := ctx.datalen
+
+	if ctx.datalen < 56 {
+		ctx.data[i] = 0x80
+		i += 1
+		for i < 56 {
+			ctx.data[i] = 0x00
+			i += 1
+		}
+	} else {
+		ctx.data[i] = 0x80
+		i += 1
+		for i < BLOCK_SIZE {
+			ctx.data[i] = 0x00
+			i += 1
+		}
+		transform(ctx, ctx.data[:])
+		mem.set(&ctx.data, 0, 56)
+	}
+
+	ctx.bitlen += u64(ctx.datalen * 8)
+	endian.unchecked_put_u64be(ctx.data[56:], ctx.bitlen)
+	transform(ctx, ctx.data[:])
+
+	for i = 0; i < DIGEST_SIZE / 4; i += 1 {
+		endian.unchecked_put_u32be(hash[i * 4:], ctx.state[i])
+	}
+
+	ctx.is_initialized = false
+}
+
+/*
+    SHA1 implementation
+*/
+
+BLOCK_SIZE :: 64
+
+Context :: struct {
+	data:    [BLOCK_SIZE]byte,
+	datalen: u32,
+	bitlen:  u64,
+	state:   [5]u32,
+	k:       [4]u32,
+
+	is_initialized: bool,
+}
+
+@(private)
+transform :: proc "contextless" (ctx: ^Context, data: []byte) {
+	a, b, c, d, e, i, t: u32
+	m: [80]u32
+
+	for i = 0; i < 16; i += 1 {
+		m[i] = endian.unchecked_get_u32be(data[i * 4:])
+	}
+	for i < 80 {
+		m[i] = (m[i - 3] ~ m[i - 8] ~ m[i - 14] ~ m[i - 16])
+		m[i] = (m[i] << 1) | (m[i] >> 31)
+		i += 1
+	}
+
+	a = ctx.state[0]
+	b = ctx.state[1]
+	c = ctx.state[2]
+	d = ctx.state[3]
+	e = ctx.state[4]
+
+	for i = 0; i < 20; i += 1 {
+		t = bits.rotate_left32(a, 5) + ((b & c) ~ (~b & d)) + e + ctx.k[0] + m[i]
+		e = d
+		d = c
+		c = bits.rotate_left32(b, 30)
+		b = a
+		a = t
+	}
+	for i < 40 {
+		t = bits.rotate_left32(a, 5) + (b ~ c ~ d) + e + ctx.k[1] + m[i]
+		e = d
+		d = c
+		c = bits.rotate_left32(b, 30)
+		b = a
+		a = t
+		i += 1
+	}
+	for i < 60 {
+		t = bits.rotate_left32(a, 5) + ((b & c) ~ (b & d) ~ (c & d)) + e + ctx.k[2] + m[i]
+		e = d
+		d = c
+		c = bits.rotate_left32(b, 30)
+		b = a
+		a = t
+		i += 1
+	}
+	for i < 80 {
+		t = bits.rotate_left32(a, 5) + (b ~ c ~ d) + e + ctx.k[3] + m[i]
+		e = d
+		d = c
+		c = bits.rotate_left32(b, 30)
+		b = a
+		a = t
+		i += 1
+	}
+
+	ctx.state[0] += a
+	ctx.state[1] += b
+	ctx.state[2] += c
+	ctx.state[3] += d
+	ctx.state[4] += e
+}
@@ -1,182 +0,0 @@
-package md2
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation of the MD2 hashing algorithm, as defined in RFC 1319 <https://datatracker.ietf.org/doc/html/rfc1319>
-*/
-
-import "core:os"
-import "core:io"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE :: 16
-
-// hash_string will hash the given input and return the
-// computed hash
-hash_string :: proc(data: string) -> [DIGEST_SIZE]byte {
-    return hash_bytes(transmute([]byte)(data))
-}
-
-// hash_bytes will hash the given input and return the
-// computed hash
-hash_bytes :: proc(data: []byte) -> [DIGEST_SIZE]byte {
-	hash: [DIGEST_SIZE]byte
-	ctx: Md2_Context
-    // init(&ctx) No-op
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer :: proc(data: string, hash: []byte) {
-	hash_bytes_to_buffer(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer :: proc(data, hash: []byte) {
-	assert(len(hash) >= DIGEST_SIZE, "Size of destination buffer is smaller than the digest size")
-    ctx: Md2_Context
-    // init(&ctx) No-op
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream will read the stream in chunks and compute a
-// hash from its contents
-hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
-	hash: [DIGEST_SIZE]byte
-	ctx: Md2_Context
-	// init(&ctx) No-op
-	buf := make([]byte, 512)
-	defer delete(buf)
-	read := 1
-	for read > 0 {
-	    read, _ = s->impl_read(buf)
-	    if read > 0 {
-			update(&ctx, buf[:read])
-	    } 
-	}
-	final(&ctx, hash[:])
-	return hash, true
-}
-
-// hash_file will read the file provided by the given handle
-// and compute a hash
-hash_file :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE]byte, bool) {
-	if !load_at_once {
-        return hash_stream(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE]byte{}, false
-}
-
-hash :: proc {
-    hash_stream,
-    hash_file,
-    hash_bytes,
-    hash_string,
-    hash_bytes_to_buffer,
-    hash_string_to_buffer,
-}
-
-/*
-    Low level API
-*/
-
-@(warning="Init is a no-op for MD2")
-init :: proc(ctx: ^Md2_Context) {
-	// No action needed here
-}
-
-update :: proc(ctx: ^Md2_Context, data: []byte) {
-	for i := 0; i < len(data); i += 1 {
-		ctx.data[ctx.datalen] = data[i]
-		ctx.datalen += 1
-		if (ctx.datalen == DIGEST_SIZE) {
-			transform(ctx, ctx.data[:])
-			ctx.datalen = 0
-		}
-	}
-}
-
-final :: proc(ctx: ^Md2_Context, hash: []byte) {
-	to_pad := byte(DIGEST_SIZE - ctx.datalen)
-    for ctx.datalen < DIGEST_SIZE {
-        ctx.data[ctx.datalen] = to_pad
-		ctx.datalen += 1
-    }
-	transform(ctx, ctx.data[:])
-	transform(ctx, ctx.checksum[:])
-    for i := 0; i < DIGEST_SIZE; i += 1 {
-        hash[i] = ctx.state[i]
-    }
-}
-
-/*
-    MD2 implementation
-*/
-
-Md2_Context :: struct {
-    data:     [DIGEST_SIZE]byte,
-    state:    [DIGEST_SIZE * 3]byte,
-    checksum: [DIGEST_SIZE]byte,
-    datalen:  int,
-}
-
-PI_TABLE := [?]byte {
-	41,  46,  67,  201, 162, 216, 124, 1,   61,  54,  84,  161, 236, 240, 6,
-	19,  98,  167, 5,   243, 192, 199, 115, 140, 152, 147, 43,  217, 188, 76,
-	130, 202, 30,  155, 87,  60,  253, 212, 224, 22,  103, 66,  111, 24,  138, 
-	23,  229, 18,  190, 78,  196, 214, 218, 158, 222, 73,  160, 251, 245, 142,
-	187, 47,  238, 122, 169, 104, 121, 145, 21,  178, 7,   63,  148, 194, 16,
-	137, 11,  34,  95,  33,  128, 127, 93,  154, 90,  144, 50,  39,  53,  62, 
-	204, 231, 191, 247, 151, 3,   255, 25,  48,  179, 72,  165, 181, 209, 215,
-	94,  146, 42,  172, 86,  170, 198, 79,  184, 56,  210, 150, 164, 125, 182,
-	118, 252, 107, 226, 156, 116, 4,   241, 69,  157, 112, 89,  100, 113, 135,
-	32,  134, 91,  207, 101, 230, 45,  168, 2,   27,  96,  37,  173, 174, 176,
-	185, 246, 28,  70,  97,  105, 52,  64,  126, 15,  85,  71,  163, 35,  221,
-	81,  175, 58,  195, 92,  249, 206, 186, 197, 234, 38,  44,  83,  13,  110,
-	133, 40,  132, 9,   211, 223, 205, 244, 65,  129, 77,  82,  106, 220, 55,
-	200, 108, 193, 171, 250, 36,  225, 123, 8,   12,  189, 177, 74,  120, 136,
-	149, 139, 227, 99,  232, 109, 233, 203, 213, 254, 59,  0,   29,  57,  242,
-	239, 183, 14,  102, 88,  208, 228, 166, 119, 114, 248, 235, 117, 75,  10,
-	49,  68,  80,  180, 143, 237, 31,  26,  219, 153, 141, 51,  159, 17,  131,
-	20,
-}
-
-transform :: proc(ctx: ^Md2_Context, data: []byte) {
-    j,k,t: byte
-	for j = 0; j < DIGEST_SIZE; j += 1 {
-		ctx.state[j + DIGEST_SIZE] = data[j]
-		ctx.state[j + DIGEST_SIZE * 2] = (ctx.state[j + DIGEST_SIZE] ~ ctx.state[j])
-	}
-	t = 0
-	for j = 0; j < DIGEST_SIZE + 2; j += 1 {
-		for k = 0; k < DIGEST_SIZE * 3; k += 1 {
-			ctx.state[k] ~= PI_TABLE[t]
-			t = ctx.state[k]
-		}
-		t = (t + j) & 0xff
-	}
-	t = ctx.checksum[DIGEST_SIZE - 1]
-	for j = 0; j < DIGEST_SIZE; j += 1 {
-		ctx.checksum[j] ~= PI_TABLE[data[j] ~ t]
-		t = ctx.checksum[j]
-	}
-}
@@ -1,263 +0,0 @@
-package md4
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-        Jeroen van Rijn: Context design to be able to change from Odin implementation to bindings.
-
-    Implementation of the MD4 hashing algorithm, as defined in RFC 1320 <https://datatracker.ietf.org/doc/html/rfc1320>
-*/
-
-import "core:mem"
-import "core:os"
-import "core:io"
-
-import "../util"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE :: 16
-
-// hash_string will hash the given input and return the
-// computed hash
-hash_string :: proc(data: string) -> [DIGEST_SIZE]byte {
-    return hash_bytes(transmute([]byte)(data))
-}
-
-// hash_bytes will hash the given input and return the
-// computed hash
-hash_bytes :: proc(data: []byte) -> [DIGEST_SIZE]byte {
-    hash: [DIGEST_SIZE]byte
-    ctx: Md4_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE, "Size of destination buffer is smaller than the digest size")
-    ctx: Md4_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream will read the stream in chunks and compute a
-// hash from its contents
-hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
-    hash: [DIGEST_SIZE]byte
-    ctx: Md4_Context
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file will read the file provided by the given handle
-// and compute a hash
-hash_file :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE]byte, bool) {
-    if !load_at_once {
-        return hash_stream(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE]byte{}, false
-}
-
-hash :: proc {
-    hash_stream,
-    hash_file,
-    hash_bytes,
-    hash_string,
-    hash_bytes_to_buffer,
-    hash_string_to_buffer,
-}
-
-/*
-    Low level API
-*/
-
-init :: proc(ctx: ^Md4_Context) {
-    ctx.state[0] = 0x67452301
-    ctx.state[1] = 0xefcdab89
-    ctx.state[2] = 0x98badcfe
-    ctx.state[3] = 0x10325476
-}
-
-update :: proc(ctx: ^Md4_Context, data: []byte) {
-    for i := 0; i < len(data); i += 1 {
-        ctx.data[ctx.datalen] = data[i]
-        ctx.datalen += 1
-        if(ctx.datalen == BLOCK_SIZE) {
-            transform(ctx, ctx.data[:])
-            ctx.bitlen += 512
-            ctx.datalen = 0
-        }
-    }
-}
-
-final :: proc(ctx: ^Md4_Context, hash: []byte) {
-    i := ctx.datalen
-    if ctx.datalen < 56 {
-        ctx.data[i] = 0x80
-        i += 1
-        for i < 56 {
-            ctx.data[i] = 0x00
-            i += 1
-        }
-    } else if ctx.datalen >= 56 {
-        ctx.data[i] = 0x80
-        i += 1
-        for i < BLOCK_SIZE {
-            ctx.data[i] = 0x00
-            i += 1
-        }
-        transform(ctx, ctx.data[:])
-        mem.set(&ctx.data, 0, 56)
-    }
-
-    ctx.bitlen  += u64(ctx.datalen * 8)
-    ctx.data[56] = byte(ctx.bitlen)
-    ctx.data[57] = byte(ctx.bitlen >> 8)
-    ctx.data[58] = byte(ctx.bitlen >> 16)
-    ctx.data[59] = byte(ctx.bitlen >> 24)
-    ctx.data[60] = byte(ctx.bitlen >> 32)
-    ctx.data[61] = byte(ctx.bitlen >> 40)
-    ctx.data[62] = byte(ctx.bitlen >> 48)
-    ctx.data[63] = byte(ctx.bitlen >> 56)
-    transform(ctx, ctx.data[:])
-
-    for i = 0; i < 4; i += 1 {
-        hash[i]      = byte(ctx.state[0] >> (i * 8)) & 0x000000ff
-        hash[i + 4]  = byte(ctx.state[1] >> (i * 8)) & 0x000000ff
-        hash[i + 8]  = byte(ctx.state[2] >> (i * 8)) & 0x000000ff
-        hash[i + 12] = byte(ctx.state[3] >> (i * 8)) & 0x000000ff
-    }
-}
-
-/*
-    MD4 implementation
-*/
-
-BLOCK_SIZE  :: 64
-
-Md4_Context :: struct {
-    data:    [64]byte,
-    state:   [4]u32,
-    bitlen:  u64,
-    datalen: u32,
-}
-
-/*
-    @note(zh): F, G and H, as mentioned in the RFC, have been inlined into FF, GG 
-    and HH respectively, instead of declaring them separately.
-*/
-
-FF :: #force_inline proc "contextless"(a, b, c, d, x: u32, s : int) -> u32 {
-    return util.ROTL32(a + ((b & c) | (~b & d)) + x, s)
-}
-
-GG :: #force_inline proc "contextless"(a, b, c, d, x: u32, s : int) -> u32 {
-    return util.ROTL32(a + ((b & c) | (b & d) | (c & d)) + x + 0x5a827999, s)
-}
-
-HH :: #force_inline proc "contextless"(a, b, c, d, x: u32, s : int) -> u32 {
-    return util.ROTL32(a + (b ~ c ~ d) + x + 0x6ed9eba1, s)
-}
-
-transform :: proc(ctx: ^Md4_Context, data: []byte) {
-    a, b, c, d, i, j: u32
-    m: [DIGEST_SIZE]u32
-
-    for i, j = 0, 0; i < DIGEST_SIZE; i += 1 {
-        m[i] = u32(data[j]) | (u32(data[j + 1]) << 8) | (u32(data[j + 2]) << 16) | (u32(data[j + 3]) << 24)
-        j += 4
-    }
-
-    a = ctx.state[0]
-    b = ctx.state[1]
-    c = ctx.state[2]
-    d = ctx.state[3]
-
-    a = FF(a, b, c, d, m[0],  3)
-    d = FF(d, a, b, c, m[1],  7)
-    c = FF(c, d, a, b, m[2],  11)
-    b = FF(b, c, d, a, m[3],  19)
-    a = FF(a, b, c, d, m[4],  3)
-    d = FF(d, a, b, c, m[5],  7)
-    c = FF(c, d, a, b, m[6],  11)
-    b = FF(b, c, d, a, m[7],  19)
-    a = FF(a, b, c, d, m[8],  3)
-    d = FF(d, a, b, c, m[9],  7)
-    c = FF(c, d, a, b, m[10], 11)
-    b = FF(b, c, d, a, m[11], 19)
-    a = FF(a, b, c, d, m[12], 3)
-    d = FF(d, a, b, c, m[13], 7)
-    c = FF(c, d, a, b, m[14], 11)
-    b = FF(b, c, d, a, m[15], 19)
-
-    a = GG(a, b, c, d, m[0],  3)
-    d = GG(d, a, b, c, m[4],  5)
-    c = GG(c, d, a, b, m[8],  9)
-    b = GG(b, c, d, a, m[12], 13)
-    a = GG(a, b, c, d, m[1],  3)
-    d = GG(d, a, b, c, m[5],  5)
-    c = GG(c, d, a, b, m[9],  9)
-    b = GG(b, c, d, a, m[13], 13)
-    a = GG(a, b, c, d, m[2],  3)
-    d = GG(d, a, b, c, m[6],  5)
-    c = GG(c, d, a, b, m[10], 9)
-    b = GG(b, c, d, a, m[14], 13)
-    a = GG(a, b, c, d, m[3],  3)
-    d = GG(d, a, b, c, m[7],  5)
-    c = GG(c, d, a, b, m[11], 9)
-    b = GG(b, c, d, a, m[15], 13)
-
-    a = HH(a, b, c, d, m[0],  3)
-    d = HH(d, a, b, c, m[8],  9)
-    c = HH(c, d, a, b, m[4],  11)
-    b = HH(b, c, d, a, m[12], 15)
-    a = HH(a, b, c, d, m[2],  3)
-    d = HH(d, a, b, c, m[10], 9)
-    c = HH(c, d, a, b, m[6],  11)
-    b = HH(b, c, d, a, m[14], 15)
-    a = HH(a, b, c, d, m[1],  3)
-    d = HH(d, a, b, c, m[9],  9)
-    c = HH(c, d, a, b, m[5],  11)
-    b = HH(b, c, d, a, m[13], 15)
-    a = HH(a, b, c, d, m[3],  3)
-    d = HH(d, a, b, c, m[11], 9)
-    c = HH(c, d, a, b, m[7],  11)
-    b = HH(b, c, d, a, m[15], 15)
-
-    ctx.state[0] += a
-    ctx.state[1] += b
-    ctx.state[2] += c
-    ctx.state[3] += d
-}
@@ -1,285 +0,0 @@
-package md5
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation of the MD5 hashing algorithm, as defined in RFC 1321 <https://datatracker.ietf.org/doc/html/rfc1321>
-*/
-
-import "core:mem"
-import "core:os"
-import "core:io"
-
-import "../util"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE :: 16
-
-// hash_string will hash the given input and return the
-// computed hash
-hash_string :: proc(data: string) -> [DIGEST_SIZE]byte {
-    return hash_bytes(transmute([]byte)(data))
-}
-
-// hash_bytes will hash the given input and return the
-// computed hash
-hash_bytes :: proc(data: []byte) -> [DIGEST_SIZE]byte {
-    hash: [DIGEST_SIZE]byte
-    ctx: Md5_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE, "Size of destination buffer is smaller than the digest size")
-    ctx: Md5_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream will read the stream in chunks and compute a
-// hash from its contents
-hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
-    hash: [DIGEST_SIZE]byte
-    ctx: Md5_Context
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file will read the file provided by the given handle
-// and compute a hash
-hash_file :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE]byte, bool) {
-    if !load_at_once {
-        return hash_stream(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE]byte{}, false
-}
-
-hash :: proc {
-    hash_stream,
-    hash_file,
-    hash_bytes,
-    hash_string,
-    hash_bytes_to_buffer,
-    hash_string_to_buffer,
-}
-
-/*
-    Low level API
-*/
-
-init :: proc(ctx: ^Md5_Context) {
-    ctx.state[0] = 0x67452301
-    ctx.state[1] = 0xefcdab89
-    ctx.state[2] = 0x98badcfe
-    ctx.state[3] = 0x10325476
-}
-
-update :: proc(ctx: ^Md5_Context, data: []byte) {
-    for i := 0; i < len(data); i += 1 {
-        ctx.data[ctx.datalen] = data[i]
-        ctx.datalen += 1
-        if(ctx.datalen == BLOCK_SIZE) {
-            transform(ctx, ctx.data[:])
-            ctx.bitlen += 512
-            ctx.datalen = 0
-        }
-    }
-}
-
-final :: proc(ctx: ^Md5_Context, hash: []byte){
-    i : u32
-    i = ctx.datalen
-
-    if ctx.datalen < 56 {
-        ctx.data[i] = 0x80
-        i += 1
-        for i < 56 {
-            ctx.data[i] = 0x00
-            i += 1
-        }
-    } else if ctx.datalen >= 56 {
-        ctx.data[i] = 0x80
-        i += 1
-        for i < BLOCK_SIZE {
-            ctx.data[i] = 0x00
-            i += 1
-        }
-        transform(ctx, ctx.data[:])
-        mem.set(&ctx.data, 0, 56)
-    }
-
-    ctx.bitlen  += u64(ctx.datalen * 8)
-    ctx.data[56] = byte(ctx.bitlen)
-    ctx.data[57] = byte(ctx.bitlen >> 8)
-    ctx.data[58] = byte(ctx.bitlen >> 16)
-    ctx.data[59] = byte(ctx.bitlen >> 24)
-    ctx.data[60] = byte(ctx.bitlen >> 32)
-    ctx.data[61] = byte(ctx.bitlen >> 40)
-    ctx.data[62] = byte(ctx.bitlen >> 48)
-    ctx.data[63] = byte(ctx.bitlen >> 56)
-    transform(ctx, ctx.data[:])
-
-    for i = 0; i < 4; i += 1 {
-        hash[i]      = byte(ctx.state[0] >> (i * 8)) & 0x000000ff
-        hash[i + 4]  = byte(ctx.state[1] >> (i * 8)) & 0x000000ff
-        hash[i + 8]  = byte(ctx.state[2] >> (i * 8)) & 0x000000ff
-        hash[i + 12] = byte(ctx.state[3] >> (i * 8)) & 0x000000ff
-    }
-}
-
-/*
-    MD4 implementation
-*/
-
-BLOCK_SIZE  :: 64
-
-Md5_Context :: struct {
-    data:    [BLOCK_SIZE]byte,
-    state:   [4]u32,
-    bitlen:  u64,
-    datalen: u32,
-}
-
-/*
-    @note(zh): F, G, H and I, as mentioned in the RFC, have been inlined into FF, GG, HH 
-    and II respectively, instead of declaring them separately.
-*/
-
-FF :: #force_inline proc "contextless" (a, b, c, d, m: u32, s: int, t: u32) -> u32 {
-    return b + util.ROTL32(a + ((b & c) | (~b & d)) + m + t, s)
-}
-
-GG :: #force_inline proc "contextless" (a, b, c, d, m: u32, s: int, t: u32) -> u32 {
-    return b + util.ROTL32(a + ((b & d) | (c & ~d)) + m + t, s)
-}
-
-HH :: #force_inline proc "contextless" (a, b, c, d, m: u32, s: int, t: u32) -> u32 {
-    return b + util.ROTL32(a + (b ~ c ~ d) + m + t, s)
-}
-
-II :: #force_inline proc "contextless" (a, b, c, d, m: u32, s: int, t: u32) -> u32 {
-    return b + util.ROTL32(a + (c ~ (b | ~d)) + m + t, s)
-}
-
-transform :: proc(ctx: ^Md5_Context, data: []byte) {
-    i, j: u32
-    m: [DIGEST_SIZE]u32
-
-    for i, j = 0, 0; i < DIGEST_SIZE; i+=1 {
-        m[i] = u32(data[j]) + u32(data[j + 1]) << 8 + u32(data[j + 2]) << 16 + u32(data[j + 3]) << 24
-        j += 4
-    }
-
-    a := ctx.state[0]
-    b := ctx.state[1]
-    c := ctx.state[2]
-    d := ctx.state[3]
-
-    a = FF(a, b, c, d, m[0],   7, 0xd76aa478)
-    d = FF(d, a, b, c, m[1],  12, 0xe8c7b756)
-    c = FF(c, d, a, b, m[2],  17, 0x242070db)
-    b = FF(b, c, d, a, m[3],  22, 0xc1bdceee)
-    a = FF(a, b, c, d, m[4],   7, 0xf57c0faf)
-    d = FF(d, a, b, c, m[5],  12, 0x4787c62a)
-    c = FF(c, d, a, b, m[6],  17, 0xa8304613)
-    b = FF(b, c, d, a, m[7],  22, 0xfd469501)
-    a = FF(a, b, c, d, m[8],   7, 0x698098d8)
-    d = FF(d, a, b, c, m[9],  12, 0x8b44f7af)
-    c = FF(c, d, a, b, m[10], 17, 0xffff5bb1)
-    b = FF(b, c, d, a, m[11], 22, 0x895cd7be)
-    a = FF(a, b, c, d, m[12],  7, 0x6b901122)
-    d = FF(d, a, b, c, m[13], 12, 0xfd987193)
-    c = FF(c, d, a, b, m[14], 17, 0xa679438e)
-    b = FF(b, c, d, a, m[15], 22, 0x49b40821)
-
-    a = GG(a, b, c, d, m[1],   5, 0xf61e2562)
-    d = GG(d, a, b, c, m[6],   9, 0xc040b340)
-    c = GG(c, d, a, b, m[11], 14, 0x265e5a51)
-    b = GG(b, c, d, a, m[0],  20, 0xe9b6c7aa)
-    a = GG(a, b, c, d, m[5],   5, 0xd62f105d)
-    d = GG(d, a, b, c, m[10],  9, 0x02441453)
-    c = GG(c, d, a, b, m[15], 14, 0xd8a1e681)
-    b = GG(b, c, d, a, m[4],  20, 0xe7d3fbc8)
-    a = GG(a, b, c, d, m[9],   5, 0x21e1cde6)
-    d = GG(d, a, b, c, m[14],  9, 0xc33707d6)
-    c = GG(c, d, a, b, m[3],  14, 0xf4d50d87)
-    b = GG(b, c, d, a, m[8],  20, 0x455a14ed)
-    a = GG(a, b, c, d, m[13],  5, 0xa9e3e905)
-    d = GG(d, a, b, c, m[2],   9, 0xfcefa3f8)
-    c = GG(c, d, a, b, m[7],  14, 0x676f02d9)
-    b = GG(b, c, d, a, m[12], 20, 0x8d2a4c8a)
-
-    a = HH(a, b, c, d, m[5],   4, 0xfffa3942)
-    d = HH(d, a, b, c, m[8],  11, 0x8771f681)
-    c = HH(c, d, a, b, m[11], 16, 0x6d9d6122)
-    b = HH(b, c, d, a, m[14], 23, 0xfde5380c)
-    a = HH(a, b, c, d, m[1],   4, 0xa4beea44)
-    d = HH(d, a, b, c, m[4],  11, 0x4bdecfa9)
-    c = HH(c, d, a, b, m[7],  16, 0xf6bb4b60)
-    b = HH(b, c, d, a, m[10], 23, 0xbebfbc70)
-    a = HH(a, b, c, d, m[13],  4, 0x289b7ec6)
-    d = HH(d, a, b, c, m[0],  11, 0xeaa127fa)
-    c = HH(c, d, a, b, m[3],  16, 0xd4ef3085)
-    b = HH(b, c, d, a, m[6],  23, 0x04881d05)
-    a = HH(a, b, c, d, m[9],   4, 0xd9d4d039)
-    d = HH(d, a, b, c, m[12], 11, 0xe6db99e5)
-    c = HH(c, d, a, b, m[15], 16, 0x1fa27cf8)
-    b = HH(b, c, d, a, m[2],  23, 0xc4ac5665)
-
-    a = II(a, b, c, d, m[0],   6, 0xf4292244)
-    d = II(d, a, b, c, m[7],  10, 0x432aff97)
-    c = II(c, d, a, b, m[14], 15, 0xab9423a7)
-    b = II(b, c, d, a, m[5],  21, 0xfc93a039)
-    a = II(a, b, c, d, m[12],  6, 0x655b59c3)
-    d = II(d, a, b, c, m[3],  10, 0x8f0ccc92)
-    c = II(c, d, a, b, m[10], 15, 0xffeff47d)
-    b = II(b, c, d, a, m[1],  21, 0x85845dd1)
-    a = II(a, b, c, d, m[8],   6, 0x6fa87e4f)
-    d = II(d, a, b, c, m[15], 10, 0xfe2ce6e0)
-    c = II(c, d, a, b, m[6],  15, 0xa3014314)
-    b = II(b, c, d, a, m[13], 21, 0x4e0811a1)
-    a = II(a, b, c, d, m[4],   6, 0xf7537e82)
-    d = II(d, a, b, c, m[11], 10, 0xbd3af235)
-    c = II(c, d, a, b, m[2],  15, 0x2ad7d2bb)
-    b = II(b, c, d, a, m[9],  21, 0xeb86d391)
-
-    ctx.state[0] += a
-    ctx.state[1] += b
-    ctx.state[2] += c
-    ctx.state[3] += d
-}
@@ -1,13 +1,14 @@
 package poly1305

 import "core:crypto"
-import "core:crypto/util"
 import field "core:crypto/_fiat/field_poly1305"
+import "core:encoding/endian"
 import "core:mem"

 KEY_SIZE :: 32
 TAG_SIZE :: 16

+@(private)
 _BLOCK_SIZE :: 16

 sum :: proc (dst, msg, key: []byte) {
@@ -51,8 +52,8 @@ init :: proc (ctx: ^Context, key: []byte) {

 	// r = le_bytes_to_num(key[0..15])
 	// r = clamp(r) (r &= 0xffffffc0ffffffc0ffffffc0fffffff)
-	tmp_lo := util.U64_LE(key[0:8]) & 0x0ffffffc0fffffff
-	tmp_hi := util.U64_LE(key[8:16]) & 0xffffffc0ffffffc
+	tmp_lo := endian.unchecked_get_u64le(key[0:]) & 0x0ffffffc0fffffff
+	tmp_hi := endian.unchecked_get_u64le(key[8:]) & 0xffffffc0ffffffc
 	field.fe_from_u64s(&ctx._r, tmp_lo, tmp_hi)

 	// s = le_bytes_to_num(key[16..31])
@@ -141,6 +142,7 @@ reset :: proc (ctx: ^Context) {
 	ctx._is_initialized = false
 }

+@(private)
 _blocks :: proc (ctx: ^Context, msg: []byte, final := false) {
 	n: field.Tight_Field_Element = ---
 	final_byte := byte(!final)
@@ -149,7 +151,7 @@ _blocks :: proc (ctx: ^Context, msg: []byte, final := false) {
 	data_len := len(data)
 	for data_len >= _BLOCK_SIZE {
 		// n = le_bytes_to_num(msg[((i-1)*16)..*i*16] | [0x01])
-		field.fe_from_bytes(&n, data[:_BLOCK_SIZE], final_byte, false)
+		field.fe_from_bytes(&n, data[:_BLOCK_SIZE], final_byte)

 		// a += n
 		field.fe_add(field.fe_relax_cast(&ctx._a), &ctx._a, &n) // _a unreduced
@@ -1,7 +1,7 @@
 package crypto

-when ODIN_OS != .Linux && ODIN_OS != .OpenBSD && ODIN_OS != .Windows {
-	_rand_bytes :: proc (dst: []byte) {
+when ODIN_OS != .Linux && ODIN_OS != .OpenBSD && ODIN_OS != .Windows && ODIN_OS != .JS {
+	_rand_bytes :: proc(dst: []byte) {
 		unimplemented("crypto: rand_bytes not supported on this OS")
 	}
 }
@@ -0,0 +1,20 @@
+package crypto
+
+foreign import "odin_env"
+foreign odin_env {
+	@(link_name = "rand_bytes")
+	env_rand_bytes :: proc "contextless" (buf: []byte) ---
+}
+
+_MAX_PER_CALL_BYTES :: 65536 // 64kiB
+
+_rand_bytes :: proc(dst: []byte) {
+	dst := dst
+
+	for len(dst) > 0 {
+		to_read := min(len(dst), _MAX_PER_CALL_BYTES)
+		env_rand_bytes(dst[:to_read])
+
+		dst = dst[to_read:]
+	}
+}
@@ -1,8 +1,8 @@
 package crypto

 import "core:fmt"
-import "core:os"
-import "core:sys/unix"
+
+import "core:sys/linux"

 _MAX_PER_CALL_BYTES :: 33554431 // 2^25 - 1

@@ -12,26 +12,25 @@ _rand_bytes :: proc (dst: []byte) {

 	for l > 0 {
 		to_read := min(l, _MAX_PER_CALL_BYTES)
-		ret := unix.sys_getrandom(raw_data(dst), to_read, 0)
-		if ret < 0 {
-			switch os.Errno(-ret) {
-			case os.EINTR:
-				// Call interupted by a signal handler, just retry the
-				// request.
-				continue
-			case os.ENOSYS:
-				// The kernel is apparently prehistoric (< 3.17 circa 2014)
-				// and does not support getrandom.
-				panic("crypto: getrandom not available in kernel")
-			case:
-				// All other failures are things that should NEVER happen
-				// unless the kernel interface changes (ie: the Linux
-				// developers break userland).
-				panic(fmt.tprintf("crypto: getrandom failed: %d", ret))
-			}
+		n_read, errno := linux.getrandom(dst[:to_read], {})
+		#partial switch errno {
+		case .NONE:
+			// Do nothing
+		case .EINTR:
+			// Call interupted by a signal handler, just retry the
+			// request.
+			continue
+		case .ENOSYS:
+			// The kernel is apparently prehistoric (< 3.17 circa 2014)
+			// and does not support getrandom.
+			panic("crypto: getrandom not available in kernel")
+		case:
+			// All other failures are things that should NEVER happen
+			// unless the kernel interface changes (ie: the Linux
+			// developers break userland).
+			panic(fmt.tprintf("crypto: getrandom failed: %v", errno))
 		}
-
-		l -= ret
-		dst = dst[ret:]
+		l -= n_read
+		dst = dst[n_read:]
 	}
 }
@@ -1,919 +0,0 @@
-package ripemd
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation for the RIPEMD hashing algorithm as defined in <https://homes.esat.kuleuven.be/~bosselae/ripemd160.html>
-*/
-
-import "core:os"
-import "core:io"
-
-import "../util"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE_128 :: 16
-DIGEST_SIZE_160 :: 20
-DIGEST_SIZE_256 :: 32
-DIGEST_SIZE_320 :: 40
-
-// hash_string_128 will hash the given input and return the
-// computed hash
-hash_string_128 :: proc(data: string) -> [DIGEST_SIZE_128]byte {
-    return hash_bytes_128(transmute([]byte)(data))
-}
-
-// hash_bytes_128 will hash the given input and return the
-// computed hash
-hash_bytes_128 :: proc(data: []byte) -> [DIGEST_SIZE_128]byte {
-    hash: [DIGEST_SIZE_128]byte
-    ctx: Ripemd128_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_128 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_128 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_128(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_128 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_128 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_128, "Size of destination buffer is smaller than the digest size")
-    ctx: Ripemd128_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_128 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_128 :: proc(s: io.Stream) -> ([DIGEST_SIZE_128]byte, bool) {
-    hash: [DIGEST_SIZE_128]byte
-    ctx: Ripemd128_Context
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_128 will read the file provided by the given handle
-// and compute a hash
-hash_file_128 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_128]byte, bool) {
-    if !load_at_once {
-        return hash_stream_128(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_128(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_128]byte{}, false
-}
-
-hash_128 :: proc {
-    hash_stream_128,
-    hash_file_128,
-    hash_bytes_128,
-    hash_string_128,
-    hash_bytes_to_buffer_128,
-    hash_string_to_buffer_128,
-}
-
-// hash_string_160 will hash the given input and return the
-// computed hash
-hash_string_160 :: proc(data: string) -> [DIGEST_SIZE_160]byte {
-    return hash_bytes_160(transmute([]byte)(data))
-}
-
-// hash_bytes_160 will hash the given input and return the
-// computed hash
-hash_bytes_160 :: proc(data: []byte) -> [DIGEST_SIZE_160]byte {
-    hash: [DIGEST_SIZE_160]byte
-    ctx: Ripemd160_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_160 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_160 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_160(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_160 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_160 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_160, "Size of destination buffer is smaller than the digest size")
-    ctx: Ripemd160_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_160 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_160 :: proc(s: io.Stream) -> ([DIGEST_SIZE_160]byte, bool) {
-    hash: [DIGEST_SIZE_160]byte
-    ctx: Ripemd160_Context
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_160 will read the file provided by the given handle
-// and compute a hash
-hash_file_160 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_160]byte, bool) {
-    if !load_at_once {
-        return hash_stream_160(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_160(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_160]byte{}, false
-}
-
-hash_160 :: proc {
-    hash_stream_160,
-    hash_file_160,
-    hash_bytes_160,
-    hash_string_160,
-    hash_bytes_to_buffer_160,
-    hash_string_to_buffer_160,
-}
-
-// hash_string_256 will hash the given input and return the
-// computed hash
-hash_string_256 :: proc(data: string) -> [DIGEST_SIZE_256]byte {
-    return hash_bytes_256(transmute([]byte)(data))
-}
-
-// hash_bytes_256 will hash the given input and return the
-// computed hash
-hash_bytes_256 :: proc(data: []byte) -> [DIGEST_SIZE_256]byte {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: Ripemd256_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_256 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_256 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_256(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_256 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_256 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_256, "Size of destination buffer is smaller than the digest size")
-    ctx: Ripemd256_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_256 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: Ripemd256_Context
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_256 will read the file provided by the given handle
-// and compute a hash
-hash_file_256 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_256]byte, bool) {
-    if !load_at_once {
-        return hash_stream_256(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_256(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_256]byte{}, false
-}
-
-hash_256 :: proc {
-    hash_stream_256,
-    hash_file_256,
-    hash_bytes_256,
-    hash_string_256,
-    hash_bytes_to_buffer_256,
-    hash_string_to_buffer_256,
-}
-
-// hash_string_320 will hash the given input and return the
-// computed hash
-hash_string_320 :: proc(data: string) -> [DIGEST_SIZE_320]byte {
-    return hash_bytes_320(transmute([]byte)(data))
-}
-
-// hash_bytes_320 will hash the given input and return the
-// computed hash
-hash_bytes_320 :: proc(data: []byte) -> [DIGEST_SIZE_320]byte {
-    hash: [DIGEST_SIZE_320]byte
-    ctx: Ripemd320_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_320 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_320 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_320(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_320 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_320 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_320, "Size of destination buffer is smaller than the digest size")
-    ctx: Ripemd320_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream_320 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_320 :: proc(s: io.Stream) -> ([DIGEST_SIZE_320]byte, bool) {
-    hash: [DIGEST_SIZE_320]byte
-    ctx: Ripemd320_Context
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_320 will read the file provided by the given handle
-// and compute a hash
-hash_file_320 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_320]byte, bool) {
-    if !load_at_once {
-        return hash_stream_320(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_320(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_320]byte{}, false
-}
-
-hash_320 :: proc {
-    hash_stream_320,
-    hash_file_320,
-    hash_bytes_320,
-    hash_string_320,
-    hash_bytes_to_buffer_320,
-    hash_string_to_buffer_320,
-}
-
-/*
-    Low level API
-*/
-
-init :: proc(ctx: ^$T) {
-    when T == Ripemd128_Context {
-        ctx.s[0], ctx.s[1], ctx.s[2], ctx.s[3] = S0, S1, S2, S3
-    } else when T == Ripemd160_Context {
-        ctx.s[0], ctx.s[1], ctx.s[2], ctx.s[3], ctx.s[4] = S0, S1, S2, S3, S4
-    } else when T == Ripemd256_Context {
-        ctx.s[0], ctx.s[1], ctx.s[2], ctx.s[3] = S0, S1, S2, S3
-        ctx.s[4], ctx.s[5], ctx.s[6], ctx.s[7] = S5, S6, S7, S8
-    } else when T == Ripemd320_Context {
-        ctx.s[0], ctx.s[1], ctx.s[2], ctx.s[3], ctx.s[4] = S0, S1, S2, S3, S4
-        ctx.s[5], ctx.s[6], ctx.s[7], ctx.s[8], ctx.s[9] = S5, S6, S7, S8, S9
-    }
-}
-
-update :: proc(ctx: ^$T, data: []byte) {
-    ctx.tc += u64(len(data))
-    data := data
-    if ctx.nx > 0 {
-        n := len(data)
-
-        when T == Ripemd128_Context {
-            if n > RIPEMD_128_BLOCK_SIZE - ctx.nx {
-                n = RIPEMD_128_BLOCK_SIZE - ctx.nx
-            }
-        } else when T == Ripemd160_Context {
-            if n > RIPEMD_160_BLOCK_SIZE - ctx.nx {
-                n = RIPEMD_160_BLOCK_SIZE - ctx.nx
-            }
-        } else when T == Ripemd256_Context{
-            if n > RIPEMD_256_BLOCK_SIZE - ctx.nx {
-                n = RIPEMD_256_BLOCK_SIZE - ctx.nx
-            }
-        } else when T == Ripemd320_Context{
-            if n > RIPEMD_320_BLOCK_SIZE - ctx.nx {
-                n = RIPEMD_320_BLOCK_SIZE - ctx.nx
-            }
-        }
-
-        for i := 0; i < n; i += 1 {
-            ctx.x[ctx.nx + i] = data[i]
-        }
-
-        ctx.nx += n
-        when T == Ripemd128_Context {
-            if ctx.nx == RIPEMD_128_BLOCK_SIZE {
-                block(ctx, ctx.x[0:])
-                ctx.nx = 0
-            }
-        } else when T == Ripemd160_Context {
-            if ctx.nx == RIPEMD_160_BLOCK_SIZE {
-                block(ctx, ctx.x[0:])
-                ctx.nx = 0
-            }
-        } else when T == Ripemd256_Context{
-            if ctx.nx == RIPEMD_256_BLOCK_SIZE {
-                block(ctx, ctx.x[0:])
-                ctx.nx = 0
-            }
-        } else when T == Ripemd320_Context{
-            if ctx.nx == RIPEMD_320_BLOCK_SIZE {
-                block(ctx, ctx.x[0:])
-                ctx.nx = 0
-            }
-        }
-        data = data[n:]
-    }
-    n := block(ctx, data)
-    data = data[n:]
-    if len(data) > 0 {
-        ctx.nx = copy(ctx.x[:], data)
-    }
-}
-
-final :: proc(ctx: ^$T, hash: []byte) {
-    d := ctx
-    tc := d.tc
-    tmp: [64]byte
-    tmp[0] = 0x80
-
-    if tc % 64 < 56 {
-        update(d, tmp[0:56 - tc % 64])
-    } else {
-        update(d, tmp[0:64 + 56 - tc % 64])
-    }
-
-    tc <<= 3
-    for i : u32 = 0; i < 8; i += 1 {
-        tmp[i] = byte(tc >> (8 * i))
-    }
-
-    update(d, tmp[0:8])
-
-    when T == Ripemd128_Context {
-        size :: RIPEMD_128_SIZE
-    } else when T == Ripemd160_Context {
-        size :: RIPEMD_160_SIZE
-    } else when T == Ripemd256_Context{
-        size :: RIPEMD_256_SIZE
-    } else when T == Ripemd320_Context{
-        size :: RIPEMD_320_SIZE
-    }
-
-    digest: [size]byte
-    for s, i in d.s {
-        digest[i * 4]     = byte(s)
-        digest[i * 4 + 1] = byte(s >> 8)
-        digest[i * 4 + 2] = byte(s >> 16)
-        digest[i * 4 + 3] = byte(s >> 24)
-    }
-    copy(hash[:], digest[:])
-}
-
-
-/*
-    RIPEMD implementation
-*/
-
-Ripemd128_Context :: struct {
-	s:  [4]u32,
-	x:  [RIPEMD_128_BLOCK_SIZE]byte,
-	nx: int,
-	tc: u64,
-}
-
-Ripemd160_Context :: struct {
-	s:  [5]u32,
-	x:  [RIPEMD_160_BLOCK_SIZE]byte,
-	nx: int,
-	tc: u64,
-}
-
-Ripemd256_Context :: struct {
-	s:  [8]u32,
-	x:  [RIPEMD_256_BLOCK_SIZE]byte,
-	nx: int,
-	tc: u64,
-}
-
-Ripemd320_Context :: struct {
-	s:  [10]u32,
-	x:  [RIPEMD_320_BLOCK_SIZE]byte,
-	nx: int,
-	tc: u64,
-}
-
-RIPEMD_128_SIZE       :: 16
-RIPEMD_128_BLOCK_SIZE :: 64
-RIPEMD_160_SIZE       :: 20
-RIPEMD_160_BLOCK_SIZE :: 64
-RIPEMD_256_SIZE       :: 32
-RIPEMD_256_BLOCK_SIZE :: 64
-RIPEMD_320_SIZE       :: 40
-RIPEMD_320_BLOCK_SIZE :: 64
-
-S0 :: 0x67452301
-S1 :: 0xefcdab89
-S2 :: 0x98badcfe
-S3 :: 0x10325476
-S4 :: 0xc3d2e1f0
-S5 :: 0x76543210
-S6 :: 0xfedcba98
-S7 :: 0x89abcdef
-S8 :: 0x01234567
-S9 :: 0x3c2d1e0f
-
-RIPEMD_128_N0 := [64]uint {
-	0, 1,  2,  3,  4,  5,  6,  7, 8,  9, 10, 11, 12, 13, 14, 15,
-	7, 4,  13, 1,  10, 6,  15, 3, 12, 0, 9,  5,  2,  14, 11, 8,
-	3, 10, 14, 4,  9,  15, 8,  1, 2,  7, 0,  6,  13, 11, 5,  12,
-	1, 9,  11, 10, 0,  8,  12, 4, 13, 3, 7,  15, 14, 5,  6,  2,
-}
-
-RIPEMD_128_R0 := [64]uint {
-	11, 14, 15, 12, 5,  8,  7,  9,  11, 13, 14, 15, 6,  7,  9,  8,
-	7,  6,  8,  13, 11, 9,  7,  15, 7,  12, 15, 9,  11, 7,  13, 12,
-	11, 13, 6,  7,  14, 9,  13, 15, 14, 8,  13, 6,  5,  12, 7,  5,
-	11, 12, 14, 15, 14, 15, 9,  8,  9,  14, 5,  6,  8,  6,  5,  12,
-}
-
-RIPEMD_128_N1 := [64]uint {
-	5, 14, 7, 0, 9, 2, 11, 4, 13, 6, 15, 8, 1, 10, 3, 12,
-	6, 11, 3, 7, 0, 13, 5, 10, 14, 15, 8, 12, 4, 9, 1, 2,
-	15, 5, 1, 3, 7, 14, 6, 9, 11, 8, 12, 2, 10, 0, 4, 13,
-	8, 6, 4, 1, 3, 11, 15, 0, 5, 12, 2, 13, 9, 7, 10, 14,
-}
-
-RIPEMD_128_R1 := [64]uint {
-	8, 9, 9, 11, 13, 15, 15, 5, 7, 7, 8, 11, 14, 14, 12, 6,
-	9, 13, 15, 7, 12, 8, 9, 11, 7, 7, 12, 7, 6, 15, 13, 11,
-	9, 7, 15, 11, 8, 6, 6, 14, 12, 13, 5, 14, 13, 13, 7, 5,
-	15, 5, 8, 11, 14, 14, 6, 14, 6, 9, 12, 9, 12, 5, 15, 8,
-}
-
-RIPEMD_160_N0 := [80]uint {
-	0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,
-	7, 4, 13, 1, 10, 6, 15, 3, 12, 0, 9, 5, 2, 14, 11, 8,
-	3, 10, 14, 4, 9, 15, 8, 1, 2, 7, 0, 6, 13, 11, 5, 12,
-	1, 9, 11, 10, 0, 8, 12, 4, 13, 3, 7, 15, 14, 5, 6, 2,
-	4, 0, 5, 9, 7, 12, 2, 10, 14, 1, 3, 8, 11, 6, 15, 13,
-}
-
-RIPEMD_160_R0 := [80]uint {
-	11, 14, 15, 12, 5, 8, 7, 9, 11, 13, 14, 15, 6, 7, 9, 8,
-	7, 6, 8, 13, 11, 9, 7, 15, 7, 12, 15, 9, 11, 7, 13, 12,
-	11, 13, 6, 7, 14, 9, 13, 15, 14, 8, 13, 6, 5, 12, 7, 5,
-	11, 12, 14, 15, 14, 15, 9, 8, 9, 14, 5, 6, 8, 6, 5, 12,
-	9, 15, 5, 11, 6, 8, 13, 12, 5, 12, 13, 14, 11, 8, 5, 6,
-}
-
-RIPEMD_160_N1 := [80]uint {
-	5, 14, 7, 0, 9, 2, 11, 4, 13, 6, 15, 8, 1, 10, 3, 12,
-	6, 11, 3, 7, 0, 13, 5, 10, 14, 15, 8, 12, 4, 9, 1, 2,
-	15, 5, 1, 3, 7, 14, 6, 9, 11, 8, 12, 2, 10, 0, 4, 13,
-	8, 6, 4, 1, 3, 11, 15, 0, 5, 12, 2, 13, 9, 7, 10, 14,
-	12, 15, 10, 4, 1, 5, 8, 7, 6, 2, 13, 14, 0, 3, 9, 11,
-}
-
-RIPEMD_160_R1 := [80]uint {
-	8, 9, 9, 11, 13, 15, 15, 5, 7, 7, 8, 11, 14, 14, 12, 6,
-	9, 13, 15, 7, 12, 8, 9, 11, 7, 7, 12, 7, 6, 15, 13, 11,
-	9, 7, 15, 11, 8, 6, 6, 14, 12, 13, 5, 14, 13, 13, 7, 5,
-	15, 5, 8, 11, 14, 14, 6, 14, 6, 9, 12, 9, 12, 5, 15, 8,
-	8, 5, 12, 9, 12, 5, 14, 6, 8, 13, 6, 5, 15, 13, 11, 11,
-}
-
-block :: #force_inline proc (ctx: ^$T, p: []byte) -> int {
-    when T == Ripemd128_Context {
-    	return ripemd_128_block(ctx, p)
-    }
-    else when T == Ripemd160_Context {
-    	return ripemd_160_block(ctx, p)
-    }
-    else when T == Ripemd256_Context {
-    	return ripemd_256_block(ctx, p)
-    }
-    else when T == Ripemd320_Context {
-    	return ripemd_320_block(ctx, p)
-    }
-}
-
-ripemd_128_block :: proc(ctx: ^$T, p: []byte) -> int {
-	n := 0
-	x: [16]u32 = ---
-	alpha: u32 = ---
-	p := p
-	for len(p) >= RIPEMD_128_BLOCK_SIZE {
-		a, b, c, d := ctx.s[0], ctx.s[1], ctx.s[2], ctx.s[3]
-		aa, bb, cc, dd := a, b, c, d
-		for i,j := 0, 0; i < 16; i, j = i+1, j+4 {
-			x[i] = u32(p[j]) | u32(p[j+1])<<8 | u32(p[j+2])<<16 | u32(p[j+3])<<24
-		}
-		i := 0
-		for i < 16 {
-			alpha = a + (b ~ c ~ d) + x[RIPEMD_128_N0[i]]
-			s := int(RIPEMD_128_R0[i])
-			alpha = util.ROTL32(alpha, s)
-			a, b, c, d = d, alpha, b, c
-			alpha = aa + (bb & dd | cc &~ dd) + x[RIPEMD_128_N1[i]] + 0x50a28be6
-			s = int(RIPEMD_128_R1[i])
-			alpha = util.ROTL32(alpha, s)
-			aa, bb, cc, dd= dd, alpha, bb, cc
-			i += 1
-		}
-		for i < 32 {
-			alpha = a + (d ~ (b & (c~d))) + x[RIPEMD_128_N0[i]] + 0x5a827999
-			s := int(RIPEMD_128_R0[i])
-			alpha = util.ROTL32(alpha, s)
-			a, b, c, d = d, alpha, b, c
-			alpha = aa + (dd ~ (bb | ~cc)) + x[RIPEMD_128_N1[i]] + 0x5c4dd124
-			s = int(RIPEMD_128_R1[i])
-			alpha = util.ROTL32(alpha, s)
-			aa, bb, cc, dd = dd, alpha, bb, cc
-			i += 1
-		}
-		for i < 48 {
-			alpha = a + (d ~ (b | ~c)) + x[RIPEMD_128_N0[i]] + 0x6ed9eba1
-			s := int(RIPEMD_128_R0[i])
-			alpha = util.ROTL32(alpha, s)
-			a, b, c, d = d, alpha, b, c
-			alpha = aa + (dd ~ (bb & (cc~dd))) + x[RIPEMD_128_N1[i]] + 0x6d703ef3
-			s = int(RIPEMD_128_R1[i])
-			alpha = util.ROTL32(alpha, s)
-			aa, bb, cc, dd = dd, alpha, bb, cc
-			i += 1
-		}
-		for i < 64 {
-			alpha = a + (c ~ (d & (b~c))) + x[RIPEMD_128_N0[i]] + 0x8f1bbcdc
-			s := int(RIPEMD_128_R0[i])
-			alpha = util.ROTL32(alpha, s)
-			a, b, c, d = d, alpha, b, c
-			alpha = aa + (bb ~ cc ~ dd) + x[RIPEMD_128_N1[i]]
-			s = int(RIPEMD_128_R1[i])
-			alpha = util.ROTL32(alpha, s)
-			aa, bb, cc, dd = dd, alpha, bb, cc
-			i += 1
-		}
-		c = ctx.s[1] + c + dd
-		ctx.s[1] = ctx.s[2] + d + aa
-		ctx.s[2] = ctx.s[3] + a + bb
-		ctx.s[3] = ctx.s[0] + b + cc
-		ctx.s[0] = c
-		p = p[RIPEMD_128_BLOCK_SIZE:]
-		n += RIPEMD_128_BLOCK_SIZE
-	}
-	return n
-}
-
-ripemd_160_block :: proc(ctx: ^$T, p: []byte) -> int {
-    n := 0
-	x: [16]u32 = ---
-	alpha, beta: u32 = ---, ---
-	p := p
-	for len(p) >= RIPEMD_160_BLOCK_SIZE {
-		a, b, c, d, e := ctx.s[0], ctx.s[1], ctx.s[2], ctx.s[3], ctx.s[4]
-		aa, bb, cc, dd, ee := a, b, c, d, e
-		for i,j := 0, 0; i < 16; i, j = i+1, j+4 {
-			x[i] = u32(p[j]) | u32(p[j+1])<<8 | u32(p[j+2])<<16 | u32(p[j+3])<<24
-		}
-		i := 0
-		for i < 16 {
-			alpha = a + (b ~ c ~ d) + x[RIPEMD_160_N0[i]]
-			s := int(RIPEMD_160_R0[i])
-			alpha = util.ROTL32(alpha, s) + e
-			beta = util.ROTL32(c, 10)
-			a, b, c, d, e = e, alpha, b, beta, d
-			alpha = aa + (bb ~ (cc | ~dd)) + x[RIPEMD_160_N1[i]] + 0x50a28be6
-			s = int(RIPEMD_160_R1[i])
-			alpha = util.ROTL32(alpha, s) + ee
-			beta = util.ROTL32(cc, 10)
-			aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
-			i += 1
-		}
-		for i < 32 {
-			alpha = a + (b&c | ~b&d) + x[RIPEMD_160_N0[i]] + 0x5a827999
-			s := int(RIPEMD_160_R0[i])
-			alpha = util.ROTL32(alpha, s) + e
-			beta = util.ROTL32(c, 10)
-			a, b, c, d, e = e, alpha, b, beta, d
-			alpha = aa + (bb&dd | cc&~dd) + x[RIPEMD_160_N1[i]] + 0x5c4dd124
-			s = int(RIPEMD_160_R1[i])
-			alpha = util.ROTL32(alpha, s) + ee
-			beta = util.ROTL32(cc, 10)
-			aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
-			i += 1
-		}
-		for i < 48 {
-			alpha = a + (b | ~c ~ d) + x[RIPEMD_160_N0[i]] + 0x6ed9eba1
-			s := int(RIPEMD_160_R0[i])
-			alpha = util.ROTL32(alpha, s) + e
-			beta = util.ROTL32(c, 10)
-			a, b, c, d, e = e, alpha, b, beta, d
-			alpha = aa + (bb | ~cc ~ dd) + x[RIPEMD_160_N1[i]] + 0x6d703ef3
-			s = int(RIPEMD_160_R1[i])
-			alpha = util.ROTL32(alpha, s) + ee
-			beta = util.ROTL32(cc, 10)
-			aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
-			i += 1
-		}
-		for i < 64 {
-			alpha = a + (b&d | c&~d) + x[RIPEMD_160_N0[i]] + 0x8f1bbcdc
-			s := int(RIPEMD_160_R0[i])
-			alpha = util.ROTL32(alpha, s) + e
-			beta = util.ROTL32(c, 10)
-			a, b, c, d, e = e, alpha, b, beta, d
-			alpha = aa + (bb&cc | ~bb&dd) + x[RIPEMD_160_N1[i]] + 0x7a6d76e9
-			s = int(RIPEMD_160_R1[i])
-			alpha = util.ROTL32(alpha, s) + ee
-			beta = util.ROTL32(cc, 10)
-			aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
-			i += 1
-		}
-		for i < 80 {
-			alpha = a + (b ~ (c | ~d)) + x[RIPEMD_160_N0[i]] + 0xa953fd4e
-			s := int(RIPEMD_160_R0[i])
-			alpha = util.ROTL32(alpha, s) + e
-			beta = util.ROTL32(c, 10)
-			a, b, c, d, e = e, alpha, b, beta, d
-			alpha = aa + (bb ~ cc ~ dd) + x[RIPEMD_160_N1[i]]
-			s = int(RIPEMD_160_R1[i])
-			alpha = util.ROTL32(alpha, s) + ee
-			beta = util.ROTL32(cc, 10)
-			aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
-			i += 1
-		}
-		dd += c + ctx.s[1]
-		ctx.s[1] = ctx.s[2] + d + ee
-		ctx.s[2] = ctx.s[3] + e + aa
-		ctx.s[3] = ctx.s[4] + a + bb
-		ctx.s[4] = ctx.s[0] + b + cc
-		ctx.s[0] = dd
-		p = p[RIPEMD_160_BLOCK_SIZE:]
-		n += RIPEMD_160_BLOCK_SIZE
-	}
-	return n
-}
-
-ripemd_256_block :: proc(ctx: ^$T, p: []byte) -> int {
-	n := 0
-	x: [16]u32 = ---
-	alpha: u32 = ---
-	p := p
-	for len(p) >= RIPEMD_256_BLOCK_SIZE {
-		a, b, c, d := ctx.s[0], ctx.s[1], ctx.s[2], ctx.s[3]
-		aa, bb, cc, dd := ctx.s[4], ctx.s[5], ctx.s[6], ctx.s[7]
-		for i,j := 0, 0; i < 16; i, j = i+1, j+4 {
-			x[i] = u32(p[j]) | u32(p[j+1])<<8 | u32(p[j+2])<<16 | u32(p[j+3])<<24
-		}
-		i := 0
-		for i < 16 {
-			alpha = a + (b ~ c ~ d) + x[RIPEMD_128_N0[i]]
-			s := int(RIPEMD_128_R0[i])
-			alpha = util.ROTL32(alpha, s)
-			a, b, c, d = d, alpha, b, c
-			alpha = aa + (bb & dd | cc &~ dd) + x[RIPEMD_128_N1[i]] + 0x50a28be6
-			s = int(RIPEMD_128_R1[i])
-			alpha = util.ROTL32(alpha, s)
-			aa, bb, cc, dd= dd, alpha, bb, cc
-			i += 1
-		}
-		t := a
-		a = aa
-		aa = t
-		for i < 32 {
-			alpha = a + (d ~ (b & (c~d))) + x[RIPEMD_128_N0[i]] + 0x5a827999
-			s := int(RIPEMD_128_R0[i])
-			alpha = util.ROTL32(alpha, s)
-			a, b, c, d = d, alpha, b, c
-			alpha = aa + (dd ~ (bb | ~cc)) + x[RIPEMD_128_N1[i]] + 0x5c4dd124
-			s = int(RIPEMD_128_R1[i])
-			alpha = util.ROTL32(alpha, s)
-			aa, bb, cc, dd = dd, alpha, bb, cc
-			i += 1
-		}
-		t = b
-		b = bb
-		bb = t
-		for i < 48 {
-			alpha = a + (d ~ (b | ~c)) + x[RIPEMD_128_N0[i]] + 0x6ed9eba1
-			s := int(RIPEMD_128_R0[i])
-			alpha = util.ROTL32(alpha, s)
-			a, b, c, d = d, alpha, b, c
-			alpha = aa + (dd ~ (bb & (cc~dd))) + x[RIPEMD_128_N1[i]] + 0x6d703ef3
-			s = int(RIPEMD_128_R1[i])
-			alpha = util.ROTL32(alpha, s)
-			aa, bb, cc, dd = dd, alpha, bb, cc
-			i += 1
-		}
-		t = c
-		c = cc
-		cc = t
-		for i < 64 {
-			alpha = a + (c ~ (d & (b~c))) + x[RIPEMD_128_N0[i]] + 0x8f1bbcdc
-			s := int(RIPEMD_128_R0[i])
-			alpha = util.ROTL32(alpha, s)
-			a, b, c, d = d, alpha, b, c
-			alpha = aa + (bb ~ cc ~ dd) + x[RIPEMD_128_N1[i]]
-			s = int(RIPEMD_128_R1[i])
-			alpha = util.ROTL32(alpha, s)
-			aa, bb, cc, dd = dd, alpha, bb, cc
-			i += 1
-		}
-		t = d
-		d = dd
-		dd = t
-		ctx.s[0] += a
-		ctx.s[1] += b
-		ctx.s[2] += c
-		ctx.s[3] += d
-		ctx.s[4] += aa
-		ctx.s[5] += bb
-		ctx.s[6] += cc
-		ctx.s[7] += dd
-		p = p[RIPEMD_256_BLOCK_SIZE:]
-		n += RIPEMD_256_BLOCK_SIZE
-	}
-	return n
-}
-
-ripemd_320_block :: proc(ctx: ^$T, p: []byte) -> int {
-    n := 0
-	x: [16]u32 = ---
-	alpha, beta: u32 = ---, ---
-	p := p
-	for len(p) >= RIPEMD_320_BLOCK_SIZE {
-		a, b, c, d, e := ctx.s[0], ctx.s[1], ctx.s[2], ctx.s[3], ctx.s[4]
-		aa, bb, cc, dd, ee := ctx.s[5], ctx.s[6], ctx.s[7], ctx.s[8], ctx.s[9]
-		for i,j := 0, 0; i < 16; i, j = i+1, j+4 {
-			x[i] = u32(p[j]) | u32(p[j+1])<<8 | u32(p[j+2])<<16 | u32(p[j+3])<<24
-		}
-		i := 0
-		for i < 16 {
-			alpha = a + (b ~ c ~ d) + x[RIPEMD_160_N0[i]]
-			s := int(RIPEMD_160_R0[i])
-			alpha = util.ROTL32(alpha, s) + e
-			beta = util.ROTL32(c, 10)
-			a, b, c, d, e = e, alpha, b, beta, d
-			alpha = aa + (bb ~ (cc | ~dd)) + x[RIPEMD_160_N1[i]] + 0x50a28be6
-			s = int(RIPEMD_160_R1[i])
-			alpha = util.ROTL32(alpha, s) + ee
-			beta = util.ROTL32(cc, 10)
-			aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
-			i += 1
-		}
-		t := b
-		b = bb
-		bb = t
-		for i < 32 {
-			alpha = a + (b&c | ~b&d) + x[RIPEMD_160_N0[i]] + 0x5a827999
-			s := int(RIPEMD_160_R0[i])
-			alpha = util.ROTL32(alpha, s) + e
-			beta = util.ROTL32(c, 10)
-			a, b, c, d, e = e, alpha, b, beta, d
-			alpha = aa + (bb&dd | cc&~dd) + x[RIPEMD_160_N1[i]] + 0x5c4dd124
-			s = int(RIPEMD_160_R1[i])
-			alpha = util.ROTL32(alpha, s) + ee
-			beta = util.ROTL32(cc, 10)
-			aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
-			i += 1
-		}
-		t = d
-		d = dd
-		dd = t
-		for i < 48 {
-			alpha = a + (b | ~c ~ d) + x[RIPEMD_160_N0[i]] + 0x6ed9eba1
-			s := int(RIPEMD_160_R0[i])
-			alpha = util.ROTL32(alpha, s) + e
-			beta = util.ROTL32(c, 10)
-			a, b, c, d, e = e, alpha, b, beta, d
-			alpha = aa + (bb | ~cc ~ dd) + x[RIPEMD_160_N1[i]] + 0x6d703ef3
-			s = int(RIPEMD_160_R1[i])
-			alpha = util.ROTL32(alpha, s) + ee
-			beta = util.ROTL32(cc, 10)
-			aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
-			i += 1
-		}
-		t = a
-		a = aa
-		aa = t
-		for i < 64 {
-			alpha = a + (b&d | c&~d) + x[RIPEMD_160_N0[i]] + 0x8f1bbcdc
-			s := int(RIPEMD_160_R0[i])
-			alpha = util.ROTL32(alpha, s) + e
-			beta = util.ROTL32(c, 10)
-			a, b, c, d, e = e, alpha, b, beta, d
-			alpha = aa + (bb&cc | ~bb&dd) + x[RIPEMD_160_N1[i]] + 0x7a6d76e9
-			s = int(RIPEMD_160_R1[i])
-			alpha = util.ROTL32(alpha, s) + ee
-			beta = util.ROTL32(cc, 10)
-			aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
-			i += 1
-		}
-		t = c
-		c = cc
-		cc = t
-		for i < 80 {
-			alpha = a + (b ~ (c | ~d)) + x[RIPEMD_160_N0[i]] + 0xa953fd4e
-			s := int(RIPEMD_160_R0[i])
-			alpha = util.ROTL32(alpha, s) + e
-			beta = util.ROTL32(c, 10)
-			a, b, c, d, e = e, alpha, b, beta, d
-			alpha = aa + (bb ~ cc ~ dd) + x[RIPEMD_160_N1[i]]
-			s = int(RIPEMD_160_R1[i])
-			alpha = util.ROTL32(alpha, s) + ee
-			beta = util.ROTL32(cc, 10)
-			aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
-			i += 1
-		}
-		t = e
-		e = ee
-		ee = t
-		ctx.s[0] += a
-		ctx.s[1] += b
-		ctx.s[2] += c
-		ctx.s[3] += d
-		ctx.s[4] += e
-		ctx.s[5] += aa
-		ctx.s[6] += bb
-		ctx.s[7] += cc
-		ctx.s[8] += dd
-		ctx.s[9] += ee
-		p = p[RIPEMD_320_BLOCK_SIZE:]
-		n += RIPEMD_320_BLOCK_SIZE
-	}
-	return n
-}
@@ -1,246 +0,0 @@
-package sha1
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation of the SHA1 hashing algorithm, as defined in RFC 3174 <https://datatracker.ietf.org/doc/html/rfc3174>
-*/
-
-import "core:mem"
-import "core:os"
-import "core:io"
-
-import "../util"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE :: 20
-
-// hash_string will hash the given input and return the
-// computed hash
-hash_string :: proc(data: string) -> [DIGEST_SIZE]byte {
-    return hash_bytes(transmute([]byte)(data))
-}
-
-// hash_bytes will hash the given input and return the
-// computed hash
-hash_bytes :: proc(data: []byte) -> [DIGEST_SIZE]byte {
-    hash: [DIGEST_SIZE]byte
-    ctx: Sha1_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE, "Size of destination buffer is smaller than the digest size")
-    ctx: Sha1_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream will read the stream in chunks and compute a
-// hash from its contents
-hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
-    hash: [DIGEST_SIZE]byte
-    ctx: Sha1_Context
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file will read the file provided by the given handle
-// and compute a hash
-hash_file :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE]byte, bool) {
-    if !load_at_once {
-        return hash_stream(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE]byte{}, false
-}
-
-hash :: proc {
-    hash_stream,
-    hash_file,
-    hash_bytes,
-    hash_string,
-    hash_bytes_to_buffer,
-    hash_string_to_buffer,
-}
-
-/*
-    Low level API
-*/
-
-init :: proc(ctx: ^Sha1_Context) {
-	ctx.state[0] = 0x67452301
-	ctx.state[1] = 0xefcdab89
-	ctx.state[2] = 0x98badcfe
-	ctx.state[3] = 0x10325476
-	ctx.state[4] = 0xc3d2e1f0
-	ctx.k[0]     = 0x5a827999
-	ctx.k[1]     = 0x6ed9eba1
-	ctx.k[2]     = 0x8f1bbcdc
-	ctx.k[3]     = 0xca62c1d6
-}
-
-update :: proc(ctx: ^Sha1_Context, data: []byte) {
-	for i := 0; i < len(data); i += 1 {
-		ctx.data[ctx.datalen] = data[i]
-		ctx.datalen += 1
-		if (ctx.datalen == BLOCK_SIZE) {
-			transform(ctx, ctx.data[:])
-			ctx.bitlen += 512
-			ctx.datalen = 0
-		}
-	}
-}
-
-final :: proc(ctx: ^Sha1_Context, hash: []byte) {
-	i := ctx.datalen
-
-	if ctx.datalen < 56 {
-		ctx.data[i] = 0x80
-        i += 1
-        for i < 56 {
-            ctx.data[i] = 0x00
-            i += 1
-        }
-	}
-	else {
-		ctx.data[i] = 0x80
-        i += 1
-        for i < BLOCK_SIZE {
-            ctx.data[i] = 0x00
-            i += 1
-        }
-		transform(ctx, ctx.data[:])
-		mem.set(&ctx.data, 0, 56)
-	}
-
-	ctx.bitlen  += u64(ctx.datalen * 8)
-	ctx.data[63] = u8(ctx.bitlen)
-	ctx.data[62] = u8(ctx.bitlen >> 8)
-	ctx.data[61] = u8(ctx.bitlen >> 16)
-	ctx.data[60] = u8(ctx.bitlen >> 24)
-	ctx.data[59] = u8(ctx.bitlen >> 32)
-	ctx.data[58] = u8(ctx.bitlen >> 40)
-	ctx.data[57] = u8(ctx.bitlen >> 48)
-	ctx.data[56] = u8(ctx.bitlen >> 56)
-	transform(ctx, ctx.data[:])
-
-	for j: u32 = 0; j < 4; j += 1 {
-		hash[j]      = u8(ctx.state[0] >> (24 - j * 8)) & 0x000000ff
-		hash[j + 4]  = u8(ctx.state[1] >> (24 - j * 8)) & 0x000000ff
-		hash[j + 8]  = u8(ctx.state[2] >> (24 - j * 8)) & 0x000000ff
-		hash[j + 12] = u8(ctx.state[3] >> (24 - j * 8)) & 0x000000ff
-		hash[j + 16] = u8(ctx.state[4] >> (24 - j * 8)) & 0x000000ff
-	}
-}
-
-/*
-    SHA1 implementation
-*/
-
-BLOCK_SIZE  :: 64
-
-Sha1_Context :: struct {
-    data:    [BLOCK_SIZE]byte,
-    datalen: u32,
-    bitlen:  u64,
-    state:   [5]u32,
-    k:       [4]u32,
-}
-
-transform :: proc(ctx: ^Sha1_Context, data: []byte) {
-    a, b, c, d, e, i, j, t: u32
-    m: [80]u32
-
-	for i, j = 0, 0; i < 16; i += 1 {
-        m[i] = u32(data[j]) << 24 + u32(data[j + 1]) << 16 + u32(data[j + 2]) << 8 + u32(data[j + 3])
-        j += 4
-    }
-	for i < 80 {
-		m[i] = (m[i - 3] ~ m[i - 8] ~ m[i - 14] ~ m[i - 16])
-		m[i] = (m[i] << 1) | (m[i] >> 31)
-        i += 1
-	}
-
-	a = ctx.state[0]
-	b = ctx.state[1]
-	c = ctx.state[2]
-	d = ctx.state[3]
-	e = ctx.state[4]
-
-	for i = 0; i < 20; i += 1 {
-		t = util.ROTL32(a, 5) + ((b & c) ~ (~b & d)) + e + ctx.k[0] + m[i]
-		e = d
-		d = c
-		c = util.ROTL32(b, 30)
-		b = a
-		a = t
-	}
-	for i < 40 {
-		t = util.ROTL32(a, 5) + (b ~ c ~ d) + e + ctx.k[1] + m[i]
-		e = d
-		d = c
-		c = util.ROTL32(b, 30)
-		b = a
-		a = t
-        i += 1
-	}
-	for i < 60 {
-		t = util.ROTL32(a, 5) + ((b & c) ~ (b & d) ~ (c & d)) + e + ctx.k[2] + m[i]
-		e = d
-		d = c
-		c = util.ROTL32(b, 30)
-		b = a
-		a = t
-        i += 1
-	}
-	for i < 80 {
-		t = util.ROTL32(a, 5) + (b ~ c ~ d) + e + ctx.k[3] + m[i]
-		e = d
-		d = c
-		c = util.ROTL32(b, 30)
-		b = a
-		a = t
-        i += 1
-	}
-
-	ctx.state[0] += a
-	ctx.state[1] += b
-	ctx.state[2] += c
-	ctx.state[3] += d
-	ctx.state[4] += e
-}
@@ -11,8 +11,8 @@ package sha3
    If you wish to compute a Keccak hash, you can use the keccak package, it will use the original padding.
 */

-import "core:os"
 import "core:io"
+import "core:os"

 import "../_sha3"

@@ -28,333 +28,337 @@ DIGEST_SIZE_512 :: 64
 // hash_string_224 will hash the given input and return the
 // computed hash
 hash_string_224 :: proc(data: string) -> [DIGEST_SIZE_224]byte {
-    return hash_bytes_224(transmute([]byte)(data))
+	return hash_bytes_224(transmute([]byte)(data))
 }

 // hash_bytes_224 will hash the given input and return the
 // computed hash
 hash_bytes_224 :: proc(data: []byte) -> [DIGEST_SIZE_224]byte {
-    hash: [DIGEST_SIZE_224]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_224
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash[:])
-    return hash
+	hash: [DIGEST_SIZE_224]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_224
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
 }

 // hash_string_to_buffer_224 will hash the given input and assign the
 // computed hash to the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_string_to_buffer_224 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_224(transmute([]byte)(data), hash)
+	hash_bytes_to_buffer_224(transmute([]byte)(data), hash)
 }

 // hash_bytes_to_buffer_224 will hash the given input and write the
 // computed hash into the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_bytes_to_buffer_224 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_224, "Size of destination buffer is smaller than the digest size")
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_224
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash)
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_224
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
 }

 // hash_stream_224 will read the stream in chunks and compute a
 // hash from its contents
 hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
-    hash: [DIGEST_SIZE_224]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_224
-    _sha3.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _sha3.update(&ctx, buf[:read])
-        } 
-    }
-    _sha3.final(&ctx, hash[:])
-    return hash, true
+	hash: [DIGEST_SIZE_224]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_224
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
 }

 // hash_file_224 will read the file provided by the given handle
 // and compute a hash
 hash_file_224 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_224]byte, bool) {
-    if !load_at_once {
-        return hash_stream_224(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_224(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_224]byte{}, false
+	if !load_at_once {
+		return hash_stream_224(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes_224(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE_224]byte{}, false
 }

 hash_224 :: proc {
-    hash_stream_224,
-    hash_file_224,
-    hash_bytes_224,
-    hash_string_224,
-    hash_bytes_to_buffer_224,
-    hash_string_to_buffer_224,
+	hash_stream_224,
+	hash_file_224,
+	hash_bytes_224,
+	hash_string_224,
+	hash_bytes_to_buffer_224,
+	hash_string_to_buffer_224,
 }

 // hash_string_256 will hash the given input and return the
 // computed hash
 hash_string_256 :: proc(data: string) -> [DIGEST_SIZE_256]byte {
-    return hash_bytes_256(transmute([]byte)(data))
+	return hash_bytes_256(transmute([]byte)(data))
 }

 // hash_bytes_256 will hash the given input and return the
 // computed hash
 hash_bytes_256 :: proc(data: []byte) -> [DIGEST_SIZE_256]byte {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_256
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash[:])
-    return hash
+	hash: [DIGEST_SIZE_256]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_256
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
 }

 // hash_string_to_buffer_256 will hash the given input and assign the
 // computed hash to the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_string_to_buffer_256 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_256(transmute([]byte)(data), hash)
+	hash_bytes_to_buffer_256(transmute([]byte)(data), hash)
 }

 // hash_bytes_to_buffer_256 will hash the given input and write the
 // computed hash into the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_bytes_to_buffer_256 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_256, "Size of destination buffer is smaller than the digest size")
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_256
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash)
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_256
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
 }

 // hash_stream_256 will read the stream in chunks and compute a
 // hash from its contents
 hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_256
-    _sha3.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _sha3.update(&ctx, buf[:read])
-        } 
-    }
-    _sha3.final(&ctx, hash[:])
-    return hash, true
+	hash: [DIGEST_SIZE_256]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_256
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
 }

 // hash_file_256 will read the file provided by the given handle
 // and compute a hash
 hash_file_256 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_256]byte, bool) {
-    if !load_at_once {
-        return hash_stream_256(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_256(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_256]byte{}, false
+	if !load_at_once {
+		return hash_stream_256(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes_256(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE_256]byte{}, false
 }

 hash_256 :: proc {
-    hash_stream_256,
-    hash_file_256,
-    hash_bytes_256,
-    hash_string_256,
-    hash_bytes_to_buffer_256,
-    hash_string_to_buffer_256,
+	hash_stream_256,
+	hash_file_256,
+	hash_bytes_256,
+	hash_string_256,
+	hash_bytes_to_buffer_256,
+	hash_string_to_buffer_256,
 }

 // hash_string_384 will hash the given input and return the
 // computed hash
 hash_string_384 :: proc(data: string) -> [DIGEST_SIZE_384]byte {
-    return hash_bytes_384(transmute([]byte)(data))
+	return hash_bytes_384(transmute([]byte)(data))
 }

 // hash_bytes_384 will hash the given input and return the
 // computed hash
 hash_bytes_384 :: proc(data: []byte) -> [DIGEST_SIZE_384]byte {
-    hash: [DIGEST_SIZE_384]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_384
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash[:])
-    return hash
+	hash: [DIGEST_SIZE_384]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_384
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
 }

 // hash_string_to_buffer_384 will hash the given input and assign the
 // computed hash to the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_string_to_buffer_384 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_384(transmute([]byte)(data), hash)
+	hash_bytes_to_buffer_384(transmute([]byte)(data), hash)
 }

 // hash_bytes_to_buffer_384 will hash the given input and write the
 // computed hash into the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_bytes_to_buffer_384 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_384, "Size of destination buffer is smaller than the digest size")
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_384
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash)
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_384
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
 }

 // hash_stream_384 will read the stream in chunks and compute a
 // hash from its contents
 hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
-    hash: [DIGEST_SIZE_384]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_384
-    _sha3.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _sha3.update(&ctx, buf[:read])
-        } 
-    }
-    _sha3.final(&ctx, hash[:])
-    return hash, true
+	hash: [DIGEST_SIZE_384]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_384
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
 }

 // hash_file_384 will read the file provided by the given handle
 // and compute a hash
 hash_file_384 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_384]byte, bool) {
-    if !load_at_once {
-        return hash_stream_384(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_384(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_384]byte{}, false
+	if !load_at_once {
+		return hash_stream_384(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes_384(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE_384]byte{}, false
 }

 hash_384 :: proc {
-    hash_stream_384,
-    hash_file_384,
-    hash_bytes_384,
-    hash_string_384,
-    hash_bytes_to_buffer_384,
-    hash_string_to_buffer_384,
+	hash_stream_384,
+	hash_file_384,
+	hash_bytes_384,
+	hash_string_384,
+	hash_bytes_to_buffer_384,
+	hash_string_to_buffer_384,
 }

 // hash_string_512 will hash the given input and return the
 // computed hash
 hash_string_512 :: proc(data: string) -> [DIGEST_SIZE_512]byte {
-    return hash_bytes_512(transmute([]byte)(data))
+	return hash_bytes_512(transmute([]byte)(data))
 }

 // hash_bytes_512 will hash the given input and return the
 // computed hash
 hash_bytes_512 :: proc(data: []byte) -> [DIGEST_SIZE_512]byte {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_512
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash[:])
-    return hash
+	hash: [DIGEST_SIZE_512]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_512
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
 }

 // hash_string_to_buffer_512 will hash the given input and assign the
 // computed hash to the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_string_to_buffer_512 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_512(transmute([]byte)(data), hash)
+	hash_bytes_to_buffer_512(transmute([]byte)(data), hash)
 }

 // hash_bytes_to_buffer_512 will hash the given input and write the
 // computed hash into the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_bytes_to_buffer_512 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_512, "Size of destination buffer is smaller than the digest size")
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_512
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.final(&ctx, hash)
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_512
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
 }

 // hash_stream_512 will read the stream in chunks and compute a
 // hash from its contents
 hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_512
-    _sha3.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _sha3.update(&ctx, buf[:read])
-        } 
-    }
-    _sha3.final(&ctx, hash[:])
-    return hash, true
+	hash: [DIGEST_SIZE_512]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_512
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
 }

 // hash_file_512 will read the file provided by the given handle
 // and compute a hash
 hash_file_512 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_512]byte, bool) {
-    if !load_at_once {
-        return hash_stream_512(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_512(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_512]byte{}, false
+	if !load_at_once {
+		return hash_stream_512(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes_512(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE_512]byte{}, false
 }

 hash_512 :: proc {
-    hash_stream_512,
-    hash_file_512,
-    hash_bytes_512,
-    hash_string_512,
-    hash_bytes_to_buffer_512,
-    hash_string_to_buffer_512,
+	hash_stream_512,
+	hash_file_512,
+	hash_bytes_512,
+	hash_string_512,
+	hash_bytes_to_buffer_512,
+	hash_string_to_buffer_512,
 }

 /*
    Low level API
 */

-Sha3_Context :: _sha3.Sha3_Context
+Context :: _sha3.Sha3_Context

-init :: proc(ctx: ^_sha3.Sha3_Context) {
-    _sha3.init(ctx)
+init :: proc(ctx: ^Context) {
+	_sha3.init(ctx)
 }

-update :: proc "contextless" (ctx: ^_sha3.Sha3_Context, data: []byte) {
-    _sha3.update(ctx, data)
+update :: proc(ctx: ^Context, data: []byte) {
+	_sha3.update(ctx, data)
 }

-final :: proc "contextless" (ctx: ^_sha3.Sha3_Context, hash: []byte) {
-    _sha3.final(ctx, hash)
+final :: proc(ctx: ^Context, hash: []byte) {
+	_sha3.final(ctx, hash)
 }
@@ -9,10 +9,13 @@ package shake

    Interface for the SHAKE hashing algorithm.
    The SHA3 functionality can be found in package sha3.
+
+    TODO: This should provide an incremental squeeze interface, in addition
+    to the one-shot final call.
 */

-import "core:os"
 import "core:io"
+import "core:os"

 import "../_sha3"

@@ -26,182 +29,178 @@ DIGEST_SIZE_256 :: 32
 // hash_string_128 will hash the given input and return the
 // computed hash
 hash_string_128 :: proc(data: string) -> [DIGEST_SIZE_128]byte {
-    return hash_bytes_128(transmute([]byte)(data))
+	return hash_bytes_128(transmute([]byte)(data))
 }

 // hash_bytes_128 will hash the given input and return the
 // computed hash
 hash_bytes_128 :: proc(data: []byte) -> [DIGEST_SIZE_128]byte {
-    hash: [DIGEST_SIZE_128]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_128
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.shake_xof(&ctx)
-    _sha3.shake_out(&ctx, hash[:])
-    return hash
+	hash: [DIGEST_SIZE_128]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_128
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
 }

 // hash_string_to_buffer_128 will hash the given input and assign the
 // computed hash to the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_string_to_buffer_128 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_128(transmute([]byte)(data), hash)
+	hash_bytes_to_buffer_128(transmute([]byte)(data), hash)
 }

 // hash_bytes_to_buffer_128 will hash the given input and write the
 // computed hash into the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_bytes_to_buffer_128 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_128, "Size of destination buffer is smaller than the digest size")
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_128
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.shake_xof(&ctx)
-    _sha3.shake_out(&ctx, hash)
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_128
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
 }

 // hash_stream_128 will read the stream in chunks and compute a
 // hash from its contents
 hash_stream_128 :: proc(s: io.Stream) -> ([DIGEST_SIZE_128]byte, bool) {
-    hash: [DIGEST_SIZE_128]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_128
-    _sha3.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _sha3.update(&ctx, buf[:read])
-        } 
-    }
-    _sha3.shake_xof(&ctx)
-    _sha3.shake_out(&ctx, hash[:])
-    return hash, true
+	hash: [DIGEST_SIZE_128]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_128
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
 }

 // hash_file_128 will read the file provided by the given handle
 // and compute a hash
 hash_file_128 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_128]byte, bool) {
-    if !load_at_once {
-        return hash_stream_128(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_128(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_128]byte{}, false
+	if !load_at_once {
+		return hash_stream_128(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes_128(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE_128]byte{}, false
 }

 hash_128 :: proc {
-    hash_stream_128,
-    hash_file_128,
-    hash_bytes_128,
-    hash_string_128,
-    hash_bytes_to_buffer_128,
-    hash_string_to_buffer_128,
+	hash_stream_128,
+	hash_file_128,
+	hash_bytes_128,
+	hash_string_128,
+	hash_bytes_to_buffer_128,
+	hash_string_to_buffer_128,
 }

 // hash_string_256 will hash the given input and return the
 // computed hash
 hash_string_256 :: proc(data: string) -> [DIGEST_SIZE_256]byte {
-    return hash_bytes_256(transmute([]byte)(data))
+	return hash_bytes_256(transmute([]byte)(data))
 }

 // hash_bytes_256 will hash the given input and return the
 // computed hash
 hash_bytes_256 :: proc(data: []byte) -> [DIGEST_SIZE_256]byte {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_256
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.shake_xof(&ctx)
-    _sha3.shake_out(&ctx, hash[:])
-    return hash
+	hash: [DIGEST_SIZE_256]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_256
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
 }

 // hash_string_to_buffer_256 will hash the given input and assign the
 // computed hash to the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_string_to_buffer_256 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_256(transmute([]byte)(data), hash)
+	hash_bytes_to_buffer_256(transmute([]byte)(data), hash)
 }

 // hash_bytes_to_buffer_256 will hash the given input and write the
 // computed hash into the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_bytes_to_buffer_256 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_256, "Size of destination buffer is smaller than the digest size")
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_256
-    _sha3.init(&ctx)
-    _sha3.update(&ctx, data)
-    _sha3.shake_xof(&ctx)
-    _sha3.shake_out(&ctx, hash)
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_256
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
 }

 // hash_stream_256 will read the stream in chunks and compute a
 // hash from its contents
 hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: _sha3.Sha3_Context
-    ctx.mdlen = DIGEST_SIZE_256
-    _sha3.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _sha3.update(&ctx, buf[:read])
-        } 
-    }
-    _sha3.shake_xof(&ctx)
-    _sha3.shake_out(&ctx, hash[:])
-    return hash, true
+	hash: [DIGEST_SIZE_256]byte
+	ctx: Context
+	ctx.mdlen = DIGEST_SIZE_256
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
 }

 // hash_file_256 will read the file provided by the given handle
 // and compute a hash
 hash_file_256 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_256]byte, bool) {
-    if !load_at_once {
-        return hash_stream_256(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_256(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_256]byte{}, false
+	if !load_at_once {
+		return hash_stream_256(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes_256(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE_256]byte{}, false
 }

 hash_256 :: proc {
-    hash_stream_256,
-    hash_file_256,
-    hash_bytes_256,
-    hash_string_256,
-    hash_bytes_to_buffer_256,
-    hash_string_to_buffer_256,
+	hash_stream_256,
+	hash_file_256,
+	hash_bytes_256,
+	hash_string_256,
+	hash_bytes_to_buffer_256,
+	hash_string_to_buffer_256,
 }

 /*
    Low level API
 */

-Shake_Context :: _sha3.Sha3_Context
+Context :: _sha3.Sha3_Context

-init :: proc(ctx: ^_sha3.Sha3_Context) {
-    _sha3.init(ctx)
+init :: proc(ctx: ^Context) {
+	_sha3.init(ctx)
 }

-update :: proc "contextless" (ctx: ^_sha3.Sha3_Context, data: []byte) {
-    _sha3.update(ctx, data)
+update :: proc(ctx: ^Context, data: []byte) {
+	_sha3.update(ctx, data)
 }

-final :: proc "contextless" (ctx: ^_sha3.Sha3_Context, hash: []byte) {
-    _sha3.shake_xof(ctx)
-    _sha3.shake_out(ctx, hash[:])
+final :: proc(ctx: ^Context, hash: []byte) {
+	_sha3.shake_xof(ctx)
+	_sha3.shake_out(ctx, hash[:])
 }
@@ -13,202 +13,200 @@ package siphash
 */

 import "core:crypto"
-import "core:crypto/util"
+import "core:encoding/endian"
+import "core:math/bits"

 /*
    High level API
 */

-KEY_SIZE    :: 16
+KEY_SIZE :: 16
 DIGEST_SIZE :: 8

 // sum_string_1_3 will hash the given message with the key and return
 // the computed hash as a u64
 sum_string_1_3 :: proc(msg, key: string) -> u64 {
-    return sum_bytes_1_3(transmute([]byte)(msg), transmute([]byte)(key))
+	return sum_bytes_1_3(transmute([]byte)(msg), transmute([]byte)(key))
 }

 // sum_bytes_1_3 will hash the given message with the key and return
 // the computed hash as a u64
-sum_bytes_1_3 :: proc (msg, key: []byte) -> u64 {
-    ctx: Context
-    hash: u64
-    init(&ctx, key, 1, 3)
-    update(&ctx, msg)
-    final(&ctx, &hash)
-    return hash
+sum_bytes_1_3 :: proc(msg, key: []byte) -> u64 {
+	ctx: Context
+	hash: u64
+	init(&ctx, key, 1, 3)
+	update(&ctx, msg)
+	final(&ctx, &hash)
+	return hash
 }

 // sum_string_to_buffer_1_3 will hash the given message with the key and write
 // the computed hash into the provided destination buffer
 sum_string_to_buffer_1_3 :: proc(msg, key: string, dst: []byte) {
-    sum_bytes_to_buffer_1_3(transmute([]byte)(msg), transmute([]byte)(key), dst)
+	sum_bytes_to_buffer_1_3(transmute([]byte)(msg), transmute([]byte)(key), dst)
 }

 // sum_bytes_to_buffer_1_3 will hash the given message with the key and write
 // the computed hash into the provided destination buffer
 sum_bytes_to_buffer_1_3 :: proc(msg, key, dst: []byte) {
-    assert(len(dst) >= DIGEST_SIZE, "crypto/siphash: Destination buffer needs to be at least of size 8")
-    hash  := sum_bytes_1_3(msg, key)
-    _collect_output(dst[:], hash)
+	hash := sum_bytes_1_3(msg, key)
+	_collect_output(dst[:], hash)
 }

 sum_1_3 :: proc {
-    sum_string_1_3,
-    sum_bytes_1_3,
-    sum_string_to_buffer_1_3,
-    sum_bytes_to_buffer_1_3,
+	sum_string_1_3,
+	sum_bytes_1_3,
+	sum_string_to_buffer_1_3,
+	sum_bytes_to_buffer_1_3,
 }

-// verify_u64_1_3 will check if the supplied tag matches with the output you 
+// verify_u64_1_3 will check if the supplied tag matches with the output you
 // will get from the provided message and key
-verify_u64_1_3 :: proc (tag: u64 msg, key: []byte) -> bool {
-    return sum_bytes_1_3(msg, key) == tag
+verify_u64_1_3 :: proc(tag: u64, msg, key: []byte) -> bool {
+	return sum_bytes_1_3(msg, key) == tag
 }

-// verify_bytes will check if the supplied tag matches with the output you 
+// verify_bytes will check if the supplied tag matches with the output you
 // will get from the provided message and key
-verify_bytes_1_3 :: proc (tag, msg, key: []byte) -> bool {
-    derived_tag: [8]byte
-    sum_bytes_to_buffer_1_3(msg, key, derived_tag[:])
-    return crypto.compare_constant_time(derived_tag[:], tag) == 1
+verify_bytes_1_3 :: proc(tag, msg, key: []byte) -> bool {
+	derived_tag: [8]byte
+	sum_bytes_to_buffer_1_3(msg, key, derived_tag[:])
+	return crypto.compare_constant_time(derived_tag[:], tag) == 1
 }

 verify_1_3 :: proc {
-    verify_bytes_1_3,
-    verify_u64_1_3,
+	verify_bytes_1_3,
+	verify_u64_1_3,
 }

 // sum_string_2_4 will hash the given message with the key and return
 // the computed hash as a u64
 sum_string_2_4 :: proc(msg, key: string) -> u64 {
-    return sum_bytes_2_4(transmute([]byte)(msg), transmute([]byte)(key))
+	return sum_bytes_2_4(transmute([]byte)(msg), transmute([]byte)(key))
 }

 // sum_bytes_2_4 will hash the given message with the key and return
 // the computed hash as a u64
-sum_bytes_2_4 :: proc (msg, key: []byte) -> u64 {
-    ctx: Context
-    hash: u64
-    init(&ctx, key, 2, 4)
-    update(&ctx, msg)
-    final(&ctx, &hash)
-    return hash
+sum_bytes_2_4 :: proc(msg, key: []byte) -> u64 {
+	ctx: Context
+	hash: u64
+	init(&ctx, key, 2, 4)
+	update(&ctx, msg)
+	final(&ctx, &hash)
+	return hash
 }

 // sum_string_to_buffer_2_4 will hash the given message with the key and write
 // the computed hash into the provided destination buffer
 sum_string_to_buffer_2_4 :: proc(msg, key: string, dst: []byte) {
-    sum_bytes_to_buffer_2_4(transmute([]byte)(msg), transmute([]byte)(key), dst)
+	sum_bytes_to_buffer_2_4(transmute([]byte)(msg), transmute([]byte)(key), dst)
 }

 // sum_bytes_to_buffer_2_4 will hash the given message with the key and write
 // the computed hash into the provided destination buffer
 sum_bytes_to_buffer_2_4 :: proc(msg, key, dst: []byte) {
-    assert(len(dst) >= DIGEST_SIZE, "crypto/siphash: Destination buffer needs to be at least of size 8")
-    hash  := sum_bytes_2_4(msg, key)
-    _collect_output(dst[:], hash)
+	hash := sum_bytes_2_4(msg, key)
+	_collect_output(dst[:], hash)
 }

 sum_2_4 :: proc {
-    sum_string_2_4,
-    sum_bytes_2_4,
-    sum_string_to_buffer_2_4,
-    sum_bytes_to_buffer_2_4,
+	sum_string_2_4,
+	sum_bytes_2_4,
+	sum_string_to_buffer_2_4,
+	sum_bytes_to_buffer_2_4,
 }

-sum_string           :: sum_string_2_4
-sum_bytes            :: sum_bytes_2_4
+sum_string :: sum_string_2_4
+sum_bytes :: sum_bytes_2_4
 sum_string_to_buffer :: sum_string_to_buffer_2_4
-sum_bytes_to_buffer  :: sum_bytes_to_buffer_2_4
+sum_bytes_to_buffer :: sum_bytes_to_buffer_2_4
 sum :: proc {
-    sum_string,
-    sum_bytes,
-    sum_string_to_buffer,
-    sum_bytes_to_buffer,
+	sum_string,
+	sum_bytes,
+	sum_string_to_buffer,
+	sum_bytes_to_buffer,
 }

-// verify_u64_2_4 will check if the supplied tag matches with the output you 
+// verify_u64_2_4 will check if the supplied tag matches with the output you
 // will get from the provided message and key
-verify_u64_2_4 :: proc (tag: u64 msg, key: []byte) -> bool {
-    return sum_bytes_2_4(msg, key) == tag
+verify_u64_2_4 :: proc(tag: u64, msg, key: []byte) -> bool {
+	return sum_bytes_2_4(msg, key) == tag
 }

-// verify_bytes will check if the supplied tag matches with the output you 
+// verify_bytes will check if the supplied tag matches with the output you
 // will get from the provided message and key
-verify_bytes_2_4 :: proc (tag, msg, key: []byte) -> bool {
-    derived_tag: [8]byte
-    sum_bytes_to_buffer_2_4(msg, key, derived_tag[:])
-    return crypto.compare_constant_time(derived_tag[:], tag) == 1
+verify_bytes_2_4 :: proc(tag, msg, key: []byte) -> bool {
+	derived_tag: [8]byte
+	sum_bytes_to_buffer_2_4(msg, key, derived_tag[:])
+	return crypto.compare_constant_time(derived_tag[:], tag) == 1
 }

 verify_2_4 :: proc {
-    verify_bytes_2_4,
-    verify_u64_2_4,
+	verify_bytes_2_4,
+	verify_u64_2_4,
 }

 verify_bytes :: verify_bytes_2_4
-verify_u64   :: verify_u64_2_4
+verify_u64 :: verify_u64_2_4
 verify :: proc {
-    verify_bytes,
-    verify_u64,
+	verify_bytes,
+	verify_u64,
 }

 // sum_string_4_8 will hash the given message with the key and return
 // the computed hash as a u64
 sum_string_4_8 :: proc(msg, key: string) -> u64 {
-    return sum_bytes_4_8(transmute([]byte)(msg), transmute([]byte)(key))
+	return sum_bytes_4_8(transmute([]byte)(msg), transmute([]byte)(key))
 }

 // sum_bytes_4_8 will hash the given message with the key and return
 // the computed hash as a u64
-sum_bytes_4_8 :: proc (msg, key: []byte) -> u64 {
-    ctx: Context
-    hash: u64
-    init(&ctx, key, 4, 8)
-    update(&ctx, msg)
-    final(&ctx, &hash)
-    return hash
+sum_bytes_4_8 :: proc(msg, key: []byte) -> u64 {
+	ctx: Context
+	hash: u64
+	init(&ctx, key, 4, 8)
+	update(&ctx, msg)
+	final(&ctx, &hash)
+	return hash
 }

 // sum_string_to_buffer_4_8 will hash the given message with the key and write
 // the computed hash into the provided destination buffer
 sum_string_to_buffer_4_8 :: proc(msg, key: string, dst: []byte) {
-    sum_bytes_to_buffer_4_8(transmute([]byte)(msg), transmute([]byte)(key), dst)
+	sum_bytes_to_buffer_4_8(transmute([]byte)(msg), transmute([]byte)(key), dst)
 }

 // sum_bytes_to_buffer_4_8 will hash the given message with the key and write
 // the computed hash into the provided destination buffer
 sum_bytes_to_buffer_4_8 :: proc(msg, key, dst: []byte) {
-    assert(len(dst) >= DIGEST_SIZE, "crypto/siphash: Destination buffer needs to be at least of size 8")
-    hash  := sum_bytes_4_8(msg, key)
-    _collect_output(dst[:], hash)
+	hash := sum_bytes_4_8(msg, key)
+	_collect_output(dst[:], hash)
 }

 sum_4_8 :: proc {
-    sum_string_4_8,
-    sum_bytes_4_8,
-    sum_string_to_buffer_4_8,
-    sum_bytes_to_buffer_4_8,
+	sum_string_4_8,
+	sum_bytes_4_8,
+	sum_string_to_buffer_4_8,
+	sum_bytes_to_buffer_4_8,
 }

-// verify_u64_4_8 will check if the supplied tag matches with the output you 
+// verify_u64_4_8 will check if the supplied tag matches with the output you
 // will get from the provided message and key
-verify_u64_4_8 :: proc (tag: u64 msg, key: []byte) -> bool {
-    return sum_bytes_4_8(msg, key) == tag
+verify_u64_4_8 :: proc(tag: u64, msg, key: []byte) -> bool {
+	return sum_bytes_4_8(msg, key) == tag
 }

-// verify_bytes will check if the supplied tag matches with the output you 
+// verify_bytes will check if the supplied tag matches with the output you
 // will get from the provided message and key
-verify_bytes_4_8 :: proc (tag, msg, key: []byte) -> bool {
-    derived_tag: [8]byte
-    sum_bytes_to_buffer_4_8(msg, key, derived_tag[:])
-    return crypto.compare_constant_time(derived_tag[:], tag) == 1
+verify_bytes_4_8 :: proc(tag, msg, key: []byte) -> bool {
+	derived_tag: [8]byte
+	sum_bytes_to_buffer_4_8(msg, key, derived_tag[:])
+	return crypto.compare_constant_time(derived_tag[:], tag) == 1
 }

 verify_4_8 :: proc {
-    verify_bytes_4_8,
-    verify_u64_4_8,
+	verify_bytes_4_8,
+	verify_u64_4_8,
 }

 /*
@@ -216,120 +214,150 @@ verify_4_8 :: proc {
 */

 init :: proc(ctx: ^Context, key: []byte, c_rounds, d_rounds: int) {
-    assert(len(key) == KEY_SIZE, "crypto/siphash: Invalid key size, want 16")
-    ctx.c_rounds = c_rounds
-    ctx.d_rounds = d_rounds
-    is_valid_setting := (ctx.c_rounds == 1 && ctx.d_rounds == 3) ||
-                        (ctx.c_rounds == 2 && ctx.d_rounds == 4) ||
-                        (ctx.c_rounds == 4 && ctx.d_rounds == 8)
-    assert(is_valid_setting, "crypto/siphash: Incorrect rounds set up. Valid pairs are (1,3), (2,4) and (4,8)")
-    ctx.k0 = util.U64_LE(key[:8])
-    ctx.k1 = util.U64_LE(key[8:])
-    ctx.v0 = 0x736f6d6570736575 ~ ctx.k0
-    ctx.v1 = 0x646f72616e646f6d ~ ctx.k1
-    ctx.v2 = 0x6c7967656e657261 ~ ctx.k0
-    ctx.v3 = 0x7465646279746573 ~ ctx.k1
-    ctx.is_initialized = true
+	if len(key) != KEY_SIZE {
+		panic("crypto/siphash; invalid key size")
+	}
+	ctx.c_rounds = c_rounds
+	ctx.d_rounds = d_rounds
+	is_valid_setting :=
+		(ctx.c_rounds == 1 && ctx.d_rounds == 3) ||
+		(ctx.c_rounds == 2 && ctx.d_rounds == 4) ||
+		(ctx.c_rounds == 4 && ctx.d_rounds == 8)
+	if !is_valid_setting {
+		panic("crypto/siphash: incorrect rounds set up")
+	}
+	ctx.k0 = endian.unchecked_get_u64le(key[:8])
+	ctx.k1 = endian.unchecked_get_u64le(key[8:])
+	ctx.v0 = 0x736f6d6570736575 ~ ctx.k0
+	ctx.v1 = 0x646f72616e646f6d ~ ctx.k1
+	ctx.v2 = 0x6c7967656e657261 ~ ctx.k0
+	ctx.v3 = 0x7465646279746573 ~ ctx.k1
+
+	ctx.last_block = 0
+	ctx.total_length = 0
+
+	ctx.is_initialized = true
 }

 update :: proc(ctx: ^Context, data: []byte) {
-    assert(ctx.is_initialized, "crypto/siphash: Context is not initalized")
-    ctx.last_block = len(data) / 8 * 8
-    ctx.buf = data
-    i := 0
-    m: u64
-    for i < ctx.last_block {
-        m = u64(ctx.buf[i] & 0xff)
-        i += 1
+	assert(ctx.is_initialized, "crypto/siphash: context is not initialized")

-        for r in u64(1)..<8 {
-            m |= u64(ctx.buf[i] & 0xff) << (r * 8)
-            i += 1
-        }
-
-        ctx.v3 ~= m
-        for _ in 0..<ctx.c_rounds {
-            _compress(ctx)
-        }
-
-        ctx.v0 ~= m
-    }
+	data := data
+	ctx.total_length += len(data)
+	if ctx.last_block > 0 {
+		n := copy(ctx.buf[ctx.last_block:], data)
+		ctx.last_block += n
+		if ctx.last_block == BLOCK_SIZE {
+			block(ctx, ctx.buf[:])
+			ctx.last_block = 0
+		}
+		data = data[n:]
+	}
+	if len(data) >= BLOCK_SIZE {
+		n := len(data) &~ (BLOCK_SIZE - 1)
+		block(ctx, data[:n])
+		data = data[n:]
+	}
+	if len(data) > 0 {
+		ctx.last_block = copy(ctx.buf[:], data)
+	}
 }

 final :: proc(ctx: ^Context, dst: ^u64) {
-    m: u64
-    for i := len(ctx.buf) - 1; i >= ctx.last_block; i -= 1 {
-        m <<= 8
-        m |= u64(ctx.buf[i] & 0xff)
-    }
-    m |= u64(len(ctx.buf) << 56)
+	assert(ctx.is_initialized, "crypto/siphash: context is not initialized")

-    ctx.v3 ~= m
+	tmp: [BLOCK_SIZE]byte
+	copy(tmp[:], ctx.buf[:ctx.last_block])
+	tmp[7] = byte(ctx.total_length & 0xff)
+	block(ctx, tmp[:])

-    for _ in 0..<ctx.c_rounds {
-        _compress(ctx)
-    }
+	ctx.v2 ~= 0xff

-    ctx.v0 ~= m
-    ctx.v2 ~= 0xff
+	for _ in 0 ..< ctx.d_rounds {
+		_compress(ctx)
+	}

-    for _ in 0..<ctx.d_rounds {
-        _compress(ctx)
-    }
+	dst^ = ctx.v0 ~ ctx.v1 ~ ctx.v2 ~ ctx.v3

-    dst^ = ctx.v0 ~ ctx.v1 ~ ctx.v2 ~ ctx.v3
-
-    reset(ctx)
+	reset(ctx)
 }

 reset :: proc(ctx: ^Context) {
-    ctx.k0, ctx.k1 = 0, 0
-    ctx.v0, ctx.v1 = 0, 0
-    ctx.v2, ctx.v3 = 0, 0
-    ctx.last_block = 0
-    ctx.c_rounds = 0
-    ctx.d_rounds = 0
-    ctx.is_initialized = false
+	ctx.k0, ctx.k1 = 0, 0
+	ctx.v0, ctx.v1 = 0, 0
+	ctx.v2, ctx.v3 = 0, 0
+	ctx.last_block = 0
+	ctx.total_length = 0
+	ctx.c_rounds = 0
+	ctx.d_rounds = 0
+	ctx.is_initialized = false
 }

+BLOCK_SIZE :: 8
+
 Context :: struct {
-    v0, v1, v2, v3: u64,    // State values
-    k0, k1:         u64,    // Split key
-    c_rounds:       int,    // Number of message rounds
-    d_rounds:       int,    // Number of finalization rounds
-    buf:            []byte, // Provided data
-    last_block:     int,    // Offset from the last block
-    is_initialized: bool,
+	v0, v1, v2, v3: u64, // State values
+	k0, k1:         u64, // Split key
+	c_rounds:       int, // Number of message rounds
+	d_rounds:       int, // Number of finalization rounds
+	buf:            [BLOCK_SIZE]byte, // Provided data
+	last_block:     int, // Offset from the last block
+	total_length:   int,
+	is_initialized: bool,
 }

+@(private)
+block :: proc "contextless" (ctx: ^Context, buf: []byte) {
+	buf := buf
+
+	for len(buf) >= BLOCK_SIZE {
+		m := endian.unchecked_get_u64le(buf)
+
+		ctx.v3 ~= m
+		for _ in 0 ..< ctx.c_rounds {
+			_compress(ctx)
+		}
+
+		ctx.v0 ~= m
+
+		buf = buf[BLOCK_SIZE:]
+	}
+}
+
+@(private)
 _get_byte :: #force_inline proc "contextless" (byte_num: byte, into: u64) -> byte {
-    return byte(into >> (((~byte_num) & (size_of(u64) - 1)) << 3))
+	return byte(into >> (((~byte_num) & (size_of(u64) - 1)) << 3))
 }

-_collect_output :: #force_inline proc "contextless" (dst: []byte, hash: u64) {
-    dst[0] = _get_byte(7, hash)
-    dst[1] = _get_byte(6, hash)
-    dst[2] = _get_byte(5, hash)
-    dst[3] = _get_byte(4, hash)
-    dst[4] = _get_byte(3, hash)
-    dst[5] = _get_byte(2, hash)
-    dst[6] = _get_byte(1, hash)
-    dst[7] = _get_byte(0, hash)
+@(private)
+_collect_output :: #force_inline proc(dst: []byte, hash: u64) {
+	if len(dst) < DIGEST_SIZE {
+		panic("crypto/siphash: invalid tag size")
+	}
+	dst[0] = _get_byte(7, hash)
+	dst[1] = _get_byte(6, hash)
+	dst[2] = _get_byte(5, hash)
+	dst[3] = _get_byte(4, hash)
+	dst[4] = _get_byte(3, hash)
+	dst[5] = _get_byte(2, hash)
+	dst[6] = _get_byte(1, hash)
+	dst[7] = _get_byte(0, hash)
 }

+@(private)
 _compress :: #force_inline proc "contextless" (ctx: ^Context) {
-    ctx.v0 += ctx.v1
-    ctx.v1  = util.ROTL64(ctx.v1, 13)
-    ctx.v1 ~= ctx.v0
-    ctx.v0  = util.ROTL64(ctx.v0, 32)
-    ctx.v2 += ctx.v3
-    ctx.v3  = util.ROTL64(ctx.v3, 16)
-    ctx.v3 ~= ctx.v2
-    ctx.v0 += ctx.v3
-    ctx.v3  = util.ROTL64(ctx.v3, 21)
-    ctx.v3 ~= ctx.v0
-    ctx.v2 += ctx.v1
-    ctx.v1  = util.ROTL64(ctx.v1, 17)
-    ctx.v1 ~= ctx.v2
-    ctx.v2  = util.ROTL64(ctx.v2, 32)
+	ctx.v0 += ctx.v1
+	ctx.v1 = bits.rotate_left64(ctx.v1, 13)
+	ctx.v1 ~= ctx.v0
+	ctx.v0 = bits.rotate_left64(ctx.v0, 32)
+	ctx.v2 += ctx.v3
+	ctx.v3 = bits.rotate_left64(ctx.v3, 16)
+	ctx.v3 ~= ctx.v2
+	ctx.v0 += ctx.v3
+	ctx.v3 = bits.rotate_left64(ctx.v3, 21)
+	ctx.v3 ~= ctx.v0
+	ctx.v2 += ctx.v1
+	ctx.v1 = bits.rotate_left64(ctx.v1, 17)
+	ctx.v1 ~= ctx.v2
+	ctx.v2 = bits.rotate_left64(ctx.v2, 32)
 }
@@ -10,10 +10,10 @@ package sm3
    Implementation of the SM3 hashing algorithm, as defined in <https://datatracker.ietf.org/doc/html/draft-sca-cfrg-sm3-02>
 */

-import "core:os"
+import "core:encoding/endian"
 import "core:io"
-
-import "../util"
+import "core:math/bits"
+import "core:os"

 /*
    High level API
@@ -24,227 +24,256 @@ DIGEST_SIZE :: 32
 // hash_string will hash the given input and return the
 // computed hash
 hash_string :: proc(data: string) -> [DIGEST_SIZE]byte {
-    return hash_bytes(transmute([]byte)(data))
+	return hash_bytes(transmute([]byte)(data))
 }

 // hash_bytes will hash the given input and return the
 // computed hash
 hash_bytes :: proc(data: []byte) -> [DIGEST_SIZE]byte {
-    hash: [DIGEST_SIZE]byte
-    ctx: Sm3_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
+	hash: [DIGEST_SIZE]byte
+	ctx: Context
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash[:])
+	return hash
 }

 // hash_string_to_buffer will hash the given input and assign the
 // computed hash to the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_string_to_buffer :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer(transmute([]byte)(data), hash)
+	hash_bytes_to_buffer(transmute([]byte)(data), hash)
 }

 // hash_bytes_to_buffer will hash the given input and write the
 // computed hash into the second parameter.
 // It requires that the destination buffer is at least as big as the digest size
 hash_bytes_to_buffer :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE, "Size of destination buffer is smaller than the digest size")
-    ctx: Sm3_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash)
+	ctx: Context
+	init(&ctx)
+	update(&ctx, data)
+	final(&ctx, hash)
 }

 // hash_stream will read the stream in chunks and compute a
 // hash from its contents
 hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
-    hash: [DIGEST_SIZE]byte
-    ctx: Sm3_Context
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true 
+	hash: [DIGEST_SIZE]byte
+	ctx: Context
+	init(&ctx)
+
+	buf := make([]byte, 512)
+	defer delete(buf)
+
+	read := 1
+	for read > 0 {
+		read, _ = io.read(s, buf)
+		if read > 0 {
+			update(&ctx, buf[:read])
+		}
+	}
+	final(&ctx, hash[:])
+	return hash, true
 }

 // hash_file will read the file provided by the given handle
 // and compute a hash
 hash_file :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE]byte, bool) {
-    if !load_at_once {
-        return hash_stream(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE]byte{}, false
+	if !load_at_once {
+		return hash_stream(os.stream_from_handle(hd))
+	} else {
+		if buf, ok := os.read_entire_file(hd); ok {
+			return hash_bytes(buf[:]), ok
+		}
+	}
+	return [DIGEST_SIZE]byte{}, false
 }

 hash :: proc {
-    hash_stream,
-    hash_file,
-    hash_bytes,
-    hash_string,
-    hash_bytes_to_buffer,
-    hash_string_to_buffer,
+	hash_stream,
+	hash_file,
+	hash_bytes,
+	hash_string,
+	hash_bytes_to_buffer,
+	hash_string_to_buffer,
 }

 /*
    Low level API
 */

-init :: proc(ctx: ^Sm3_Context) {
-    ctx.state[0] = IV[0]
-    ctx.state[1] = IV[1]
-    ctx.state[2] = IV[2]
-    ctx.state[3] = IV[3]
-    ctx.state[4] = IV[4]
-    ctx.state[5] = IV[5]
-    ctx.state[6] = IV[6]
-    ctx.state[7] = IV[7]
+init :: proc(ctx: ^Context) {
+	ctx.state[0] = IV[0]
+	ctx.state[1] = IV[1]
+	ctx.state[2] = IV[2]
+	ctx.state[3] = IV[3]
+	ctx.state[4] = IV[4]
+	ctx.state[5] = IV[5]
+	ctx.state[6] = IV[6]
+	ctx.state[7] = IV[7]
+
+	ctx.length = 0
+	ctx.bitlength = 0
+
+	ctx.is_initialized = true
 }

-update :: proc(ctx: ^Sm3_Context, data: []byte) {
-    data := data
-    ctx.length += u64(len(data))
+update :: proc(ctx: ^Context, data: []byte) {
+	assert(ctx.is_initialized)

-    if ctx.bitlength > 0 {
-        n := copy(ctx.x[ctx.bitlength:], data[:])
-        ctx.bitlength += u64(n)
-        if ctx.bitlength == 64 {
-            block(ctx, ctx.x[:])
-            ctx.bitlength = 0
-        }
-        data = data[n:]
-    }
-    if len(data) >= 64 {
-        n := len(data) &~ (64 - 1)
-        block(ctx, data[:n])
-        data = data[n:]
-    }
-    if len(data) > 0 {
-        ctx.bitlength = u64(copy(ctx.x[:], data[:]))
-    }
+	data := data
+	ctx.length += u64(len(data))
+
+	if ctx.bitlength > 0 {
+		n := copy(ctx.x[ctx.bitlength:], data[:])
+		ctx.bitlength += u64(n)
+		if ctx.bitlength == BLOCK_SIZE {
+			block(ctx, ctx.x[:])
+			ctx.bitlength = 0
+		}
+		data = data[n:]
+	}
+	if len(data) >= BLOCK_SIZE {
+		n := len(data) &~ (BLOCK_SIZE - 1)
+		block(ctx, data[:n])
+		data = data[n:]
+	}
+	if len(data) > 0 {
+		ctx.bitlength = u64(copy(ctx.x[:], data[:]))
+	}
 }

-final :: proc(ctx: ^Sm3_Context, hash: []byte) {
-    length := ctx.length
+final :: proc(ctx: ^Context, hash: []byte) {
+	assert(ctx.is_initialized)

-    pad: [64]byte
-    pad[0] = 0x80
-    if length % 64 < 56 {
-        update(ctx, pad[0: 56 - length % 64])
-    } else {
-        update(ctx, pad[0: 64 + 56 - length % 64])
-    }
+	if len(hash) < DIGEST_SIZE {
+		panic("crypto/sm3: invalid destination digest size")
+	}

-    length <<= 3
-    util.PUT_U64_BE(pad[:], length)
-    update(ctx, pad[0: 8])
-    assert(ctx.bitlength == 0)
+	length := ctx.length

-    util.PUT_U32_BE(hash[0:],  ctx.state[0])
-    util.PUT_U32_BE(hash[4:],  ctx.state[1])
-    util.PUT_U32_BE(hash[8:],  ctx.state[2])
-    util.PUT_U32_BE(hash[12:], ctx.state[3])
-    util.PUT_U32_BE(hash[16:], ctx.state[4])
-    util.PUT_U32_BE(hash[20:], ctx.state[5])
-    util.PUT_U32_BE(hash[24:], ctx.state[6])
-    util.PUT_U32_BE(hash[28:], ctx.state[7])
+	pad: [BLOCK_SIZE]byte
+	pad[0] = 0x80
+	if length % BLOCK_SIZE < 56 {
+		update(ctx, pad[0:56 - length % BLOCK_SIZE])
+	} else {
+		update(ctx, pad[0:BLOCK_SIZE + 56 - length % BLOCK_SIZE])
+	}
+
+	length <<= 3
+	endian.unchecked_put_u64be(pad[:], length)
+	update(ctx, pad[0:8])
+	assert(ctx.bitlength == 0)
+
+	for i := 0; i < DIGEST_SIZE / 4; i += 1 {
+		endian.unchecked_put_u32be(hash[i * 4:], ctx.state[i])
+	}
+
+	ctx.is_initialized = false
 }

 /*
    SM3 implementation
 */

-Sm3_Context :: struct {
-    state:     [8]u32,
-    x:         [64]byte,
-    bitlength: u64,
-    length:    u64,
+BLOCK_SIZE :: 64
+
+Context :: struct {
+	state:     [8]u32,
+	x:         [BLOCK_SIZE]byte,
+	bitlength: u64,
+	length:    u64,
+
+	is_initialized: bool,
 }

+@(private)
 IV := [8]u32 {
-    0x7380166f, 0x4914b2b9, 0x172442d7, 0xda8a0600,
-    0xa96f30bc, 0x163138aa, 0xe38dee4d, 0xb0fb0e4e,
+	0x7380166f, 0x4914b2b9, 0x172442d7, 0xda8a0600,
+	0xa96f30bc, 0x163138aa, 0xe38dee4d, 0xb0fb0e4e,
 }

-block :: proc "contextless" (ctx: ^Sm3_Context, buf: []byte) {
-    buf := buf
+@(private)
+block :: proc "contextless" (ctx: ^Context, buf: []byte) {
+	buf := buf

-    w:  [68]u32
-    wp: [64]u32
+	w: [68]u32
+	wp: [64]u32

-    state0, state1, state2, state3 := ctx.state[0], ctx.state[1], ctx.state[2], ctx.state[3]
-    state4, state5, state6, state7 := ctx.state[4], ctx.state[5], ctx.state[6], ctx.state[7]
+	state0, state1, state2, state3 := ctx.state[0], ctx.state[1], ctx.state[2], ctx.state[3]
+	state4, state5, state6, state7 := ctx.state[4], ctx.state[5], ctx.state[6], ctx.state[7]

-    for len(buf) >= 64 {
-        for i := 0; i < 16; i += 1 {
-            j := i * 4
-            w[i] = u32(buf[j]) << 24 | u32(buf[j + 1]) << 16 | u32(buf[j + 2]) << 8 | u32(buf[j + 3])
-        }
-        for i := 16; i < 68; i += 1 {
-            p1v := w[i - 16] ~ w[i - 9] ~ util.ROTL32(w[i - 3], 15)
-            // @note(zh): inlined P1
-            w[i] = p1v ~ util.ROTL32(p1v, 15) ~ util.ROTL32(p1v, 23) ~ util.ROTL32(w[i - 13], 7) ~ w[i - 6]
-        }
-        for i := 0; i < 64; i += 1 {
-            wp[i] = w[i] ~ w[i + 4]
-        }
+	for len(buf) >= BLOCK_SIZE {
+		for i := 0; i < 16; i += 1 {
+			w[i] = endian.unchecked_get_u32be(buf[i * 4:])
+		}
+		for i := 16; i < 68; i += 1 {
+			p1v := w[i - 16] ~ w[i - 9] ~ bits.rotate_left32(w[i - 3], 15)
+			// @note(zh): inlined P1
+			w[i] =
+				p1v ~
+				bits.rotate_left32(p1v, 15) ~
+				bits.rotate_left32(p1v, 23) ~
+				bits.rotate_left32(w[i - 13], 7) ~
+				w[i - 6]
+		}
+		for i := 0; i < 64; i += 1 {
+			wp[i] = w[i] ~ w[i + 4]
+		}

-        a, b, c, d := state0, state1, state2, state3
-        e, f, g, h := state4, state5, state6, state7
+		a, b, c, d := state0, state1, state2, state3
+		e, f, g, h := state4, state5, state6, state7

-        for i := 0; i < 16; i += 1 {
-            v1  := util.ROTL32(u32(a), 12)
-            ss1 := util.ROTL32(v1 + u32(e) + util.ROTL32(0x79cc4519, i), 7)
-            ss2 := ss1 ~ v1
+		for i := 0; i < 16; i += 1 {
+			v1 := bits.rotate_left32(u32(a), 12)
+			ss1 := bits.rotate_left32(v1 + u32(e) + bits.rotate_left32(0x79cc4519, i), 7)
+			ss2 := ss1 ~ v1

-            // @note(zh): inlined FF1
-            tt1 := u32(a ~ b ~ c) + u32(d) + ss2 + wp[i]
-            // @note(zh): inlined GG1
-            tt2 := u32(e ~ f ~ g) + u32(h) + ss1 + w[i]
+			// @note(zh): inlined FF1
+			tt1 := u32(a ~ b ~ c) + u32(d) + ss2 + wp[i]
+			// @note(zh): inlined GG1
+			tt2 := u32(e ~ f ~ g) + u32(h) + ss1 + w[i]

-            a, b, c, d = tt1, a, util.ROTL32(u32(b), 9), c
-            // @note(zh): inlined P0
-            e, f, g, h = (tt2 ~ util.ROTL32(tt2, 9) ~ util.ROTL32(tt2, 17)), e, util.ROTL32(u32(f), 19), g
-        }
+			a, b, c, d = tt1, a, bits.rotate_left32(u32(b), 9), c
+			// @note(zh): inlined P0
+			e, f, g, h =
+				(tt2 ~ bits.rotate_left32(tt2, 9) ~ bits.rotate_left32(tt2, 17)),
+				e,
+				bits.rotate_left32(u32(f), 19),
+				g
+		}

-        for i := 16; i < 64; i += 1 {
-            v   := util.ROTL32(u32(a), 12)
-            ss1 := util.ROTL32(v + u32(e) + util.ROTL32(0x7a879d8a, i % 32), 7)
-            ss2 := ss1 ~ v
+		for i := 16; i < 64; i += 1 {
+			v := bits.rotate_left32(u32(a), 12)
+			ss1 := bits.rotate_left32(v + u32(e) + bits.rotate_left32(0x7a879d8a, i % 32), 7)
+			ss2 := ss1 ~ v

-            // @note(zh): inlined FF2
-            tt1 := u32(((a & b) | (a & c) | (b & c)) + d) + ss2 + wp[i]
-            // @note(zh): inlined GG2
-            tt2 := u32(((e & f) | ((~e) & g)) + h) + ss1 + w[i]
+			// @note(zh): inlined FF2
+			tt1 := u32(((a & b) | (a & c) | (b & c)) + d) + ss2 + wp[i]
+			// @note(zh): inlined GG2
+			tt2 := u32(((e & f) | ((~e) & g)) + h) + ss1 + w[i]

-            a, b, c, d = tt1, a, util.ROTL32(u32(b), 9), c
-            // @note(zh): inlined P0
-            e, f, g, h = (tt2 ~ util.ROTL32(tt2, 9) ~ util.ROTL32(tt2, 17)), e, util.ROTL32(u32(f), 19), g
-        }
+			a, b, c, d = tt1, a, bits.rotate_left32(u32(b), 9), c
+			// @note(zh): inlined P0
+			e, f, g, h =
+				(tt2 ~ bits.rotate_left32(tt2, 9) ~ bits.rotate_left32(tt2, 17)),
+				e,
+				bits.rotate_left32(u32(f), 19),
+				g
+		}

-        state0 ~= a
-        state1 ~= b
-        state2 ~= c
-        state3 ~= d
-        state4 ~= e
-        state5 ~= f
-        state6 ~= g
-        state7 ~= h
+		state0 ~= a
+		state1 ~= b
+		state2 ~= c
+		state3 ~= d
+		state4 ~= e
+		state5 ~= f
+		state6 ~= g
+		state7 ~= h

-        buf = buf[64:]
-    }
+		buf = buf[BLOCK_SIZE:]
+	}

-    ctx.state[0], ctx.state[1], ctx.state[2], ctx.state[3] = state0, state1, state2, state3
-    ctx.state[4], ctx.state[5], ctx.state[6], ctx.state[7] = state4, state5, state6, state7
+	ctx.state[0], ctx.state[1], ctx.state[2], ctx.state[3] = state0, state1, state2, state3
+	ctx.state[4], ctx.state[5], ctx.state[6], ctx.state[7] = state4, state5, state6, state7
 }
@@ -1,517 +0,0 @@
-package streebog
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation of the Streebog hashing algorithm, standardized as GOST R 34.11-2012 in RFC 6986 <https://datatracker.ietf.org/doc/html/rfc6986>
-*/
-
-import "core:os"
-import "core:io"
-
-import "../util"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE_256 :: 32
-DIGEST_SIZE_512 :: 64
-
-// hash_string_256 will hash the given input and return the
-// computed hash
-hash_string_256 :: proc(data: string) -> [DIGEST_SIZE_256]byte {
-    return hash_bytes_256(transmute([]byte)(data))
-}
-
-// hash_bytes_256 will hash the given input and return the
-// computed hash
-hash_bytes_256 :: proc(data: []byte) -> [DIGEST_SIZE_256]byte {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: Streebog_Context
-    ctx.is256 = true
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_256 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_256 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_256(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_256 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_256 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_256, "Size of destination buffer is smaller than the digest size")
-    ctx: Streebog_Context
-    ctx.is256 = true
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-}
-
-// hash_stream_256 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
-    hash: [DIGEST_SIZE_256]byte
-    ctx: Streebog_Context
-    ctx.is256 = true
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_256 will read the file provided by the given handle
-// and compute a hash
-hash_file_256 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_256]byte, bool) {
-    if !load_at_once {
-        return hash_stream_256(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_256(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_256]byte{}, false
-}
-
-hash_256 :: proc {
-    hash_stream_256,
-    hash_file_256,
-    hash_bytes_256,
-    hash_string_256,
-	hash_bytes_to_buffer_256,
-    hash_string_to_buffer_256,
-}
-
-// hash_string_512 will hash the given input and return the
-// computed hash
-hash_string_512 :: proc(data: string) -> [DIGEST_SIZE_512]byte {
-    return hash_bytes_512(transmute([]byte)(data))
-}
-
-// hash_bytes_512 will hash the given input and return the
-// computed hash
-hash_bytes_512 :: proc(data: []byte) -> [DIGEST_SIZE_512]byte {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: Streebog_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_512 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_512 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_512(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_512 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_512 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_512, "Size of destination buffer is smaller than the digest size")
-    ctx: Streebog_Context
-    init(&ctx)
-    update(&ctx, data)
-    final(&ctx, hash[:])
-}
-
-// hash_stream_512 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
-    hash: [DIGEST_SIZE_512]byte
-    ctx: Streebog_Context
-    init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            update(&ctx, buf[:read])
-        } 
-    }
-    final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_512 will read the file provided by the given handle
-// and compute a hash
-hash_file_512 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_512]byte, bool) {
-    if !load_at_once {
-        return hash_stream_512(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_512(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_512]byte{}, false
-}
-
-hash_512 :: proc {
-    hash_stream_512,
-    hash_file_512,
-    hash_bytes_512,
-    hash_string_512,
-    hash_bytes_to_buffer_512,
-    hash_string_to_buffer_512,
-}
-
-/*
-    Low level API
-*/
-
-init :: proc(ctx: ^Streebog_Context) {
-	if ctx.is256 {
-		ctx.hash_size = 256
-		for _, i in ctx.h {
-			ctx.h[i] = 0x01
-		}
-	} else {
-		ctx.hash_size = 512
-	}
-	ctx.v_512[1] = 0x02
-}
-
-update :: proc(ctx: ^Streebog_Context, data: []byte) {
-	length := u64(len(data))
-	chk_size: u64
-	data := data
-	for (length > 63) && (ctx.buf_size == 0) {
-		stage2(ctx, data)
-		data = data[64:]
-		length -= 64
-	}
-
-	for length != 0 {
-		chk_size = 64 - ctx.buf_size
-		if chk_size > length {
-			chk_size = length
-		}
-		copy(ctx.buffer[ctx.buf_size:], data[:chk_size])
-		ctx.buf_size += chk_size
-		length -= chk_size
-		data = data[chk_size:]
-		if ctx.buf_size == 64 {
-			stage2(ctx, ctx.buffer[:])
-			ctx.buf_size = 0
-		}
-	}
-}
-
-final :: proc(ctx: ^Streebog_Context, hash: []byte) {
-	t: [64]byte
-	t[1] = byte((ctx.buf_size * 8) >> 8) & 0xff
-	t[0] = byte((ctx.buf_size) * 8) & 0xff
-
-	padding(ctx)
-
-	G(ctx.h[:], ctx.n[:], ctx.buffer[:])
-
-	add_mod_512(ctx.n[:], t[:], ctx.n[:])
-	add_mod_512(ctx.sigma[:], ctx.buffer[:], ctx.sigma[:])
-
-	G(ctx.h[:], ctx.v_0[:], ctx.n[:])
-	G(ctx.h[:], ctx.v_0[:], ctx.sigma[:])
-
-	if ctx.is256 {
-		copy(hash[:], ctx.h[32:])
-	} else {
-		copy(hash[:], ctx.h[:])
-	}
-}
-
-/*
-    Streebog implementation
-*/
-
-PI := [256]byte {
-	252, 238, 221, 17,  207, 110, 49,  22,  251, 196, 250, 218, 35,  197, 4,   77,
-	233, 119, 240, 219, 147, 46,  153, 186, 23,  54,  241, 187, 20,  205, 95,  193,
-	249, 24,  101, 90,  226, 92,  239, 33,  129, 28,  60,  66,  139, 1,   142, 79,
-	5,   132, 2,   174, 227, 106, 143, 160, 6,   11,  237, 152, 127, 212, 211, 31,
-	235, 52,  44,  81,  234, 200, 72,  171, 242, 42,  104, 162, 253, 58,  206, 204,
-	181, 112, 14,  86,  8,   12,  118, 18,  191, 114, 19,  71,  156, 183, 93,  135,
-	21,  161, 150, 41,  16,  123, 154, 199, 243, 145, 120, 111, 157, 158, 178, 177,
-	50,  117, 25,  61,  255, 53,  138, 126, 109, 84,  198, 128, 195, 189, 13,  87,
-	223, 245, 36,  169, 62,  168, 67,  201, 215, 121, 214, 246, 124, 34,  185, 3,
-	224, 15,  236, 222, 122, 148, 176, 188, 220, 232, 40,  80,  78,  51,  10,  74,
-	167, 151, 96,  115, 30,  0,   98,  68,  26,  184, 56,  130, 100, 159, 38,  65,
-	173, 69,  70,  146, 39,  94,  85,  47,  140, 163, 165, 125, 105, 213, 149, 59,
-	7,   88,  179, 64,  134, 172, 29,  247, 48,  55,  107, 228,	136, 217, 231, 137,
-	225, 27,  131, 73,  76,  63,  248, 254, 141, 83,  170, 144, 202, 216, 133, 97,
-	32,  113, 103, 164, 45,  43,  9,   91,  203, 155, 37,  208, 190, 229, 108, 82,
-	89,  166, 116, 210,	230, 244, 180, 192, 209, 102, 175, 194, 57,  75,  99,  182,
-}
-
-TAU := [64]byte {
-	0,  8, 16, 24, 32, 40, 48, 56,
-	1,  9, 17, 25, 33, 41, 49, 57,
-	2, 10, 18, 26, 34, 42, 50, 58,
-	3, 11, 19, 27, 35, 43, 51, 59,
-	4, 12, 20, 28, 36, 44, 52, 60,
-	5, 13, 21, 29, 37, 45, 53, 61,
-	6, 14, 22, 30, 38, 46, 54, 62,
-	7, 15, 23, 31, 39, 47, 55, 63,
-}
-
-STREEBOG_A := [64]u64 {
-	0x8e20faa72ba0b470, 0x47107ddd9b505a38, 0xad08b0e0c3282d1c, 0xd8045870ef14980e,
-	0x6c022c38f90a4c07, 0x3601161cf205268d, 0x1b8e0b0e798c13c8, 0x83478b07b2468764,
-	0xa011d380818e8f40, 0x5086e740ce47c920, 0x2843fd2067adea10, 0x14aff010bdd87508,
-	0x0ad97808d06cb404, 0x05e23c0468365a02, 0x8c711e02341b2d01, 0x46b60f011a83988e,
-	0x90dab52a387ae76f, 0x486dd4151c3dfdb9, 0x24b86a840e90f0d2, 0x125c354207487869,
-	0x092e94218d243cba, 0x8a174a9ec8121e5d, 0x4585254f64090fa0, 0xaccc9ca9328a8950,
-	0x9d4df05d5f661451, 0xc0a878a0a1330aa6, 0x60543c50de970553, 0x302a1e286fc58ca7,
-	0x18150f14b9ec46dd, 0x0c84890ad27623e0, 0x0642ca05693b9f70, 0x0321658cba93c138,
-	0x86275df09ce8aaa8, 0x439da0784e745554, 0xafc0503c273aa42a, 0xd960281e9d1d5215,
-	0xe230140fc0802984, 0x71180a8960409a42, 0xb60c05ca30204d21, 0x5b068c651810a89e,
-	0x456c34887a3805b9, 0xac361a443d1c8cd2, 0x561b0d22900e4669, 0x2b838811480723ba,
-	0x9bcf4486248d9f5d, 0xc3e9224312c8c1a0, 0xeffa11af0964ee50, 0xf97d86d98a327728,
-	0xe4fa2054a80b329c, 0x727d102a548b194e, 0x39b008152acb8227, 0x9258048415eb419d,
-	0x492c024284fbaec0, 0xaa16012142f35760, 0x550b8e9e21f7a530, 0xa48b474f9ef5dc18,
-	0x70a6a56e2440598e, 0x3853dc371220a247, 0x1ca76e95091051ad, 0x0edd37c48a08a6d8,
-	0x07e095624504536c, 0x8d70c431ac02a736, 0xc83862965601dd1b, 0x641c314b2b8ee083,
-}
-
-STREEBOG_C := [12][64]byte { 
-	{
-		0x07, 0x45, 0xa6, 0xf2, 0x59, 0x65, 0x80, 0xdd,
-		0x23, 0x4d, 0x74, 0xcc, 0x36, 0x74, 0x76, 0x05,
-		0x15, 0xd3, 0x60, 0xa4, 0x08, 0x2a, 0x42, 0xa2,
-		0x01, 0x69, 0x67, 0x92, 0x91, 0xe0, 0x7c, 0x4b,
-		0xfc, 0xc4, 0x85, 0x75, 0x8d, 0xb8, 0x4e, 0x71,
-		0x16, 0xd0, 0x45, 0x2e, 0x43, 0x76, 0x6a, 0x2f,
-		0x1f, 0x7c, 0x65, 0xc0, 0x81, 0x2f, 0xcb, 0xeb,
-		0xe9, 0xda, 0xca, 0x1e, 0xda, 0x5b, 0x08, 0xb1,
-	},
-	{
-		0xb7, 0x9b, 0xb1, 0x21, 0x70, 0x04, 0x79, 0xe6,
-		0x56, 0xcd, 0xcb, 0xd7, 0x1b, 0xa2, 0xdd, 0x55,
-		0xca, 0xa7, 0x0a, 0xdb, 0xc2, 0x61, 0xb5, 0x5c,
-		0x58, 0x99, 0xd6, 0x12, 0x6b, 0x17, 0xb5, 0x9a,
-		0x31, 0x01, 0xb5, 0x16, 0x0f, 0x5e, 0xd5, 0x61,
-		0x98, 0x2b, 0x23, 0x0a, 0x72, 0xea, 0xfe, 0xf3,
-		0xd7, 0xb5, 0x70, 0x0f, 0x46, 0x9d, 0xe3, 0x4f,
-		0x1a, 0x2f, 0x9d, 0xa9, 0x8a, 0xb5, 0xa3, 0x6f,
-	},
-	{
-		0xb2, 0x0a, 0xba, 0x0a, 0xf5, 0x96, 0x1e, 0x99,
-		0x31, 0xdb, 0x7a, 0x86, 0x43, 0xf4, 0xb6, 0xc2,
-		0x09, 0xdb, 0x62, 0x60, 0x37, 0x3a, 0xc9, 0xc1,
-		0xb1, 0x9e, 0x35, 0x90, 0xe4, 0x0f, 0xe2, 0xd3,
-		0x7b, 0x7b, 0x29, 0xb1, 0x14, 0x75, 0xea, 0xf2,
-		0x8b, 0x1f, 0x9c, 0x52, 0x5f, 0x5e, 0xf1, 0x06,
-		0x35, 0x84, 0x3d, 0x6a, 0x28, 0xfc, 0x39, 0x0a,
-		0xc7, 0x2f, 0xce, 0x2b, 0xac, 0xdc, 0x74, 0xf5,
-	},
-	{
-		0x2e, 0xd1, 0xe3, 0x84, 0xbc, 0xbe, 0x0c, 0x22,
-		0xf1, 0x37, 0xe8, 0x93, 0xa1, 0xea, 0x53, 0x34,
-		0xbe, 0x03, 0x52, 0x93, 0x33, 0x13, 0xb7, 0xd8,
-		0x75, 0xd6, 0x03, 0xed, 0x82, 0x2c, 0xd7, 0xa9,
-		0x3f, 0x35, 0x5e, 0x68, 0xad, 0x1c, 0x72, 0x9d,
-		0x7d, 0x3c, 0x5c, 0x33, 0x7e, 0x85, 0x8e, 0x48,
-		0xdd, 0xe4, 0x71, 0x5d, 0xa0, 0xe1, 0x48, 0xf9,
-		0xd2, 0x66, 0x15, 0xe8, 0xb3, 0xdf, 0x1f, 0xef,
-	},
-	{
-		0x57, 0xfe, 0x6c, 0x7c, 0xfd, 0x58, 0x17, 0x60,
-		0xf5, 0x63, 0xea, 0xa9, 0x7e, 0xa2, 0x56, 0x7a,
-		0x16, 0x1a, 0x27, 0x23, 0xb7, 0x00, 0xff, 0xdf,
-		0xa3, 0xf5, 0x3a, 0x25, 0x47, 0x17, 0xcd, 0xbf,
-		0xbd, 0xff, 0x0f, 0x80, 0xd7, 0x35, 0x9e, 0x35,
-		0x4a, 0x10, 0x86, 0x16, 0x1f, 0x1c, 0x15, 0x7f,
-		0x63, 0x23, 0xa9, 0x6c, 0x0c, 0x41, 0x3f, 0x9a,
-		0x99, 0x47, 0x47, 0xad, 0xac, 0x6b, 0xea, 0x4b,
-	},
-	{
-		0x6e, 0x7d, 0x64, 0x46, 0x7a, 0x40, 0x68, 0xfa,
-		0x35, 0x4f, 0x90, 0x36, 0x72, 0xc5, 0x71, 0xbf,
-		0xb6, 0xc6, 0xbe, 0xc2, 0x66, 0x1f, 0xf2, 0x0a,
-		0xb4, 0xb7, 0x9a, 0x1c, 0xb7, 0xa6, 0xfa, 0xcf,
-		0xc6, 0x8e, 0xf0, 0x9a, 0xb4, 0x9a, 0x7f, 0x18,
-		0x6c, 0xa4, 0x42, 0x51, 0xf9, 0xc4, 0x66, 0x2d,
-		0xc0, 0x39, 0x30, 0x7a, 0x3b, 0xc3, 0xa4, 0x6f,
-		0xd9, 0xd3, 0x3a, 0x1d, 0xae, 0xae, 0x4f, 0xae,
-	},
-	{
-		0x93, 0xd4, 0x14, 0x3a, 0x4d, 0x56, 0x86, 0x88,
-		0xf3, 0x4a, 0x3c, 0xa2, 0x4c, 0x45, 0x17, 0x35,
-		0x04, 0x05, 0x4a, 0x28, 0x83, 0x69, 0x47, 0x06,
-		0x37, 0x2c, 0x82, 0x2d, 0xc5, 0xab, 0x92, 0x09,
-		0xc9, 0x93, 0x7a, 0x19, 0x33, 0x3e, 0x47, 0xd3,
-		0xc9, 0x87, 0xbf, 0xe6, 0xc7, 0xc6, 0x9e, 0x39,
-		0x54, 0x09, 0x24, 0xbf, 0xfe, 0x86, 0xac, 0x51,
-		0xec, 0xc5, 0xaa, 0xee, 0x16, 0x0e, 0xc7, 0xf4,
-	},
-	{
-		0x1e, 0xe7, 0x02, 0xbf, 0xd4, 0x0d, 0x7f, 0xa4,
-		0xd9, 0xa8, 0x51, 0x59, 0x35, 0xc2, 0xac, 0x36,
-		0x2f, 0xc4, 0xa5, 0xd1, 0x2b, 0x8d, 0xd1, 0x69,
-		0x90, 0x06, 0x9b, 0x92, 0xcb, 0x2b, 0x89, 0xf4,
-		0x9a, 0xc4, 0xdb, 0x4d, 0x3b, 0x44, 0xb4, 0x89,
-		0x1e, 0xde, 0x36, 0x9c, 0x71, 0xf8, 0xb7, 0x4e,
-		0x41, 0x41, 0x6e, 0x0c, 0x02, 0xaa, 0xe7, 0x03,
-		0xa7, 0xc9, 0x93, 0x4d, 0x42, 0x5b, 0x1f, 0x9b,
-	},
-	{
-		0xdb, 0x5a, 0x23, 0x83, 0x51, 0x44, 0x61, 0x72,
-		0x60, 0x2a, 0x1f, 0xcb, 0x92, 0xdc, 0x38, 0x0e,
-		0x54, 0x9c, 0x07, 0xa6, 0x9a, 0x8a, 0x2b, 0x7b,
-		0xb1, 0xce, 0xb2, 0xdb, 0x0b, 0x44, 0x0a, 0x80,
-		0x84, 0x09, 0x0d, 0xe0, 0xb7, 0x55, 0xd9, 0x3c,
-		0x24, 0x42, 0x89, 0x25, 0x1b, 0x3a, 0x7d, 0x3a,
-		0xde, 0x5f, 0x16, 0xec, 0xd8, 0x9a, 0x4c, 0x94,
-		0x9b, 0x22, 0x31, 0x16, 0x54, 0x5a, 0x8f, 0x37,
-	},
-	{
-		0xed, 0x9c, 0x45, 0x98, 0xfb, 0xc7, 0xb4, 0x74,
-		0xc3, 0xb6, 0x3b, 0x15, 0xd1, 0xfa, 0x98, 0x36,
-		0xf4, 0x52, 0x76, 0x3b, 0x30, 0x6c, 0x1e, 0x7a,
-		0x4b, 0x33, 0x69, 0xaf, 0x02, 0x67, 0xe7, 0x9f,
-		0x03, 0x61, 0x33, 0x1b, 0x8a, 0xe1, 0xff, 0x1f,
-		0xdb, 0x78, 0x8a, 0xff, 0x1c, 0xe7, 0x41, 0x89,
-		0xf3, 0xf3, 0xe4, 0xb2, 0x48, 0xe5, 0x2a, 0x38,
-		0x52, 0x6f, 0x05, 0x80, 0xa6, 0xde, 0xbe, 0xab,
-	},
-	{
-		0x1b, 0x2d, 0xf3, 0x81, 0xcd, 0xa4, 0xca, 0x6b,
-		0x5d, 0xd8, 0x6f, 0xc0, 0x4a, 0x59, 0xa2, 0xde,
-		0x98, 0x6e, 0x47, 0x7d, 0x1d, 0xcd, 0xba, 0xef,
-		0xca, 0xb9, 0x48, 0xea, 0xef, 0x71, 0x1d, 0x8a,
-		0x79, 0x66, 0x84, 0x14, 0x21, 0x80, 0x01, 0x20,
-		0x61, 0x07, 0xab, 0xeb, 0xbb, 0x6b, 0xfa, 0xd8,
-		0x94, 0xfe, 0x5a, 0x63, 0xcd, 0xc6, 0x02, 0x30,
-		0xfb, 0x89, 0xc8, 0xef, 0xd0, 0x9e, 0xcd, 0x7b,
-	},
-	{
-		0x20, 0xd7, 0x1b, 0xf1, 0x4a, 0x92, 0xbc, 0x48,
-		0x99, 0x1b, 0xb2, 0xd9, 0xd5, 0x17, 0xf4, 0xfa,
-		0x52, 0x28, 0xe1, 0x88, 0xaa, 0xa4, 0x1d, 0xe7,
-		0x86, 0xcc, 0x91, 0x18, 0x9d, 0xef, 0x80, 0x5d,
-		0x9b, 0x9f, 0x21, 0x30, 0xd4, 0x12, 0x20, 0xf8,
-		0x77, 0x1d, 0xdf, 0xbc, 0x32, 0x3c, 0xa4, 0xcd,
-		0x7a, 0xb1, 0x49, 0x04, 0xb0, 0x80, 0x13, 0xd2,
-		0xba, 0x31, 0x16, 0xf1, 0x67, 0xe7, 0x8e, 0x37,
-	},
-}
-
-Streebog_Context :: struct {
-	buffer:    [64]byte,
-	h:         [64]byte,
-	n:         [64]byte,
-	sigma:     [64]byte,
-	v_0:       [64]byte,
-	v_512:     [64]byte,
-	buf_size:  u64,
-	hash_size: int,
-	is256:	   bool,
-}
-
-add_mod_512 :: proc(first_vector, second_vector, result_vector: []byte) {
-	t: i32 = 0
-	for i: i32 = 0; i < 64; i += 1 {
-		t = i32(first_vector[i]) + i32(second_vector[i]) + (t >> 8)
-		result_vector[i] = byte(t & 0xff)
-	}
-}
-
-X :: #force_inline proc(a, k, out: []byte) {
-	for i := 0; i < 64; i += 1 {
-		out[i] = a[i] ~ k[i]
-	}
-}
-
-S :: #force_inline proc(state: []byte) {
-	t: [64]byte
-	for i: i32 = 63; i >= 0; i -= 1 {
-		t[i] = PI[state[i]]
-	}
-	copy(state, t[:])
-}
-
-P :: #force_inline proc(state: []byte) {
-	t: [64]byte
-	for i: i32 = 63; i >= 0; i -= 1 {
-		t[i] = state[TAU[i]]
-	}
-	copy(state, t[:])
-}
-
-L :: #force_inline proc(state: []byte) {
-	ins := util.cast_slice([]u64, state)
-	out: [8]u64
-	for i: i32 = 7; i >= 0; i -= 1 {
-		for j: i32 = 63; j >= 0; j -= 1 {
-			if (ins[i] >> u32(j)) & 1 != 0 {
-				out[i] ~= STREEBOG_A[63 - j]
-			}	
-		}
-	}
-	copy(state, util.cast_slice([]byte, out[:]))
-}
-
-E :: #force_inline proc(K, m, state: []byte) {
-	X(m, K, state)
-	for i: i32 = 0; i < 12; i += 1 {
-		S(state)
-		P(state)
-		L(state)
-		get_key(K, i)
-		X(state, K, state)
-	}
-}
-
-get_key :: #force_inline proc(K: []byte, i: i32) {
-	X(K, STREEBOG_C[i][:], K)
-	S(K)
-	P(K)
-	L(K)
-}
-
-G :: #force_inline proc(h, N, m: []byte) {
-	t, K: [64]byte
-	X(N, h, K[:])
-	S(K[:])
-	P(K[:])
-	L(K[:])
-	E(K[:], m, t[:])
-	X(t[:], h, t[:])
-	X(t[:], m, h)
-}
-
-stage2 :: proc(ctx: ^Streebog_Context, m: []byte) {
-	G(ctx.h[:], ctx.n[:], m)
-	add_mod_512(ctx.n[:], ctx.v_512[:], ctx.n[:])
-	add_mod_512(ctx.sigma[:], m, ctx.sigma[:])
-}
-
-padding :: proc(ctx: ^Streebog_Context) {
-	if ctx.buf_size < 64 {
-		t: [64]byte
-		copy(t[:], ctx.buffer[:int(ctx.buf_size)])
-		t[ctx.buf_size] = 0x01
-		copy(ctx.buffer[:], t[:])
-	}
-}
@@ -1,280 +0,0 @@
-package tiger
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Interface for the Tiger1 variant of the Tiger hashing algorithm as defined in <https://www.cs.technion.ac.il/~biham/Reports/Tiger/>
-*/
-
-import "core:os"
-import "core:io"
-
-import "../_tiger"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE_128 :: 16
-DIGEST_SIZE_160 :: 20
-DIGEST_SIZE_192 :: 24
-
-// hash_string_128 will hash the given input and return the
-// computed hash
-hash_string_128 :: proc(data: string) -> [DIGEST_SIZE_128]byte {
-    return hash_bytes_128(transmute([]byte)(data))
-}
-
-// hash_bytes_128 will hash the given input and return the
-// computed hash
-hash_bytes_128 :: proc(data: []byte) -> [DIGEST_SIZE_128]byte {
-    hash: [DIGEST_SIZE_128]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 1
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_128 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_128 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_128(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_128 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_128 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_128, "Size of destination buffer is smaller than the digest size")
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 1
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash)
-}
-
-// hash_stream_128 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_128 :: proc(s: io.Stream) -> ([DIGEST_SIZE_128]byte, bool) {
-    hash: [DIGEST_SIZE_128]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 1
-    _tiger.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _tiger.update(&ctx, buf[:read])
-        } 
-    }
-    _tiger.final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_128 will read the file provided by the given handle
-// and compute a hash
-hash_file_128 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_128]byte, bool) {
-    if !load_at_once {
-        return hash_stream_128(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_128(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_128]byte{}, false
-}
-
-hash_128 :: proc {
-    hash_stream_128,
-    hash_file_128,
-    hash_bytes_128,
-    hash_string_128,
-    hash_bytes_to_buffer_128,
-    hash_string_to_buffer_128,
-}
-
-// hash_string_160 will hash the given input and return the
-// computed hash
-hash_string_160 :: proc(data: string) -> [DIGEST_SIZE_160]byte {
-    return hash_bytes_160(transmute([]byte)(data))
-}
-
-// hash_bytes_160 will hash the given input and return the
-// computed hash
-hash_bytes_160 :: proc(data: []byte) -> [DIGEST_SIZE_160]byte {
-    hash: [DIGEST_SIZE_160]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 1
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_160 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_160 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_160(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_160 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_160 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_160, "Size of destination buffer is smaller than the digest size")
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 1
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash)
-}
-
-// hash_stream_160 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_160 :: proc(s: io.Stream) -> ([DIGEST_SIZE_160]byte, bool) {
-    hash: [DIGEST_SIZE_160]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 1
-    _tiger.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _tiger.update(&ctx, buf[:read])
-        } 
-    }
-    _tiger.final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_160 will read the file provided by the given handle
-// and compute a hash
-hash_file_160 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_160]byte, bool) {
-    if !load_at_once {
-        return hash_stream_160(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_160(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_160]byte{}, false
-}
-
-hash_160 :: proc {
-    hash_stream_160,
-    hash_file_160,
-    hash_bytes_160,
-    hash_string_160,
-    hash_bytes_to_buffer_160,
-    hash_string_to_buffer_160,
-}
-
-// hash_string_192 will hash the given input and return the
-// computed hash
-hash_string_192 :: proc(data: string) -> [DIGEST_SIZE_192]byte {
-    return hash_bytes_192(transmute([]byte)(data))
-}
-
-// hash_bytes_192 will hash the given input and return the
-// computed hash
-hash_bytes_192 :: proc(data: []byte) -> [DIGEST_SIZE_192]byte {
-    hash: [DIGEST_SIZE_192]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 1
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_192 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_192 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_192(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_192 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_192 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_192, "Size of destination buffer is smaller than the digest size")
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 1
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash)
-}
-
-// hash_stream_192 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_192 :: proc(s: io.Stream) -> ([DIGEST_SIZE_192]byte, bool) {
-    hash: [DIGEST_SIZE_192]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 1
-    _tiger.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _tiger.update(&ctx, buf[:read])
-        } 
-    }
-    _tiger.final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_192 will read the file provided by the given handle
-// and compute a hash
-hash_file_192 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_192]byte, bool) {
-    if !load_at_once {
-        return hash_stream_192(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_192(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_192]byte{}, false
-}
-
-hash_192 :: proc {
-    hash_stream_192,
-    hash_file_192,
-    hash_bytes_192,
-    hash_string_192,
-    hash_bytes_to_buffer_192,
-    hash_string_to_buffer_192,
-}
-
-/*
-    Low level API
-*/
-
-Tiger_Context :: _tiger.Tiger_Context
-
-init :: proc(ctx: ^_tiger.Tiger_Context) {
-    ctx.ver = 1
-    _tiger.init(ctx)
-}
-
-update :: proc(ctx: ^_tiger.Tiger_Context, data: []byte) {
-    _tiger.update(ctx, data)
-}
-
-final :: proc(ctx: ^_tiger.Tiger_Context, hash: []byte) {
-    _tiger.final(ctx, hash)
-}
@@ -1,280 +0,0 @@
-package tiger2
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Interface for the Tiger2 variant of the Tiger hashing algorithm as defined in <https://www.cs.technion.ac.il/~biham/Reports/Tiger/>
-*/
-
-import "core:os"
-import "core:io"
-
-import "../_tiger"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE_128 :: 16
-DIGEST_SIZE_160 :: 20
-DIGEST_SIZE_192 :: 24
-
-// hash_string_128 will hash the given input and return the
-// computed hash
-hash_string_128 :: proc(data: string) -> [DIGEST_SIZE_128]byte {
-    return hash_bytes_128(transmute([]byte)(data))
-}
-
-// hash_bytes_128 will hash the given input and return the
-// computed hash
-hash_bytes_128 :: proc(data: []byte) -> [DIGEST_SIZE_128]byte {
-    hash: [DIGEST_SIZE_128]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 2
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_128 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_128 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_128(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_128 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_128 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_128, "Size of destination buffer is smaller than the digest size")
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 2
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash)
-}
-
-// hash_stream_128 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_128 :: proc(s: io.Stream) -> ([DIGEST_SIZE_128]byte, bool) {
-    hash: [DIGEST_SIZE_128]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 2
-    _tiger.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _tiger.update(&ctx, buf[:read])
-        } 
-    }
-    _tiger.final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_128 will read the file provided by the given handle
-// and compute a hash
-hash_file_128 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_128]byte, bool) {
-    if !load_at_once {
-        return hash_stream_128(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_128(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_128]byte{}, false
-}
-
-hash_128 :: proc {
-    hash_stream_128,
-    hash_file_128,
-    hash_bytes_128,
-    hash_string_128,
-    hash_bytes_to_buffer_128,
-    hash_string_to_buffer_128,
-}
-
-// hash_string_160 will hash the given input and return the
-// computed hash
-hash_string_160 :: proc(data: string) -> [DIGEST_SIZE_160]byte {
-    return hash_bytes_160(transmute([]byte)(data))
-}
-
-// hash_bytes_160 will hash the given input and return the
-// computed hash
-hash_bytes_160 :: proc(data: []byte) -> [DIGEST_SIZE_160]byte {
-    hash: [DIGEST_SIZE_160]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 2
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_160 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_160 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_160(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_160 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_160 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_160, "Size of destination buffer is smaller than the digest size")
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 2
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash)
-}
-
-// hash_stream_160 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_160 :: proc(s: io.Stream) -> ([DIGEST_SIZE_160]byte, bool) {
-    hash: [DIGEST_SIZE_160]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 2
-    _tiger.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _tiger.update(&ctx, buf[:read])
-        } 
-    }
-    _tiger.final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_160 will read the file provided by the given handle
-// and compute a hash
-hash_file_160 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_160]byte, bool) {
-    if !load_at_once {
-        return hash_stream_160(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_160(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_160]byte{}, false
-}
-
-hash_160 :: proc {
-    hash_stream_160,
-    hash_file_160,
-    hash_bytes_160,
-    hash_string_160,
-    hash_bytes_to_buffer_160,
-    hash_string_to_buffer_160,
-}
-
-// hash_string_192 will hash the given input and return the
-// computed hash
-hash_string_192 :: proc(data: string) -> [DIGEST_SIZE_192]byte {
-    return hash_bytes_192(transmute([]byte)(data))
-}
-
-// hash_bytes_192 will hash the given input and return the
-// computed hash
-hash_bytes_192 :: proc(data: []byte) -> [DIGEST_SIZE_192]byte {
-    hash: [DIGEST_SIZE_192]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 2
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer_192 will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer_192 :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer_192(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer_192 will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer_192 :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE_192, "Size of destination buffer is smaller than the digest size")
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 2
-    _tiger.init(&ctx)
-    _tiger.update(&ctx, data)
-    _tiger.final(&ctx, hash)
-}
-
-// hash_stream_192 will read the stream in chunks and compute a
-// hash from its contents
-hash_stream_192 :: proc(s: io.Stream) -> ([DIGEST_SIZE_192]byte, bool) {
-    hash: [DIGEST_SIZE_192]byte
-    ctx: _tiger.Tiger_Context
-    ctx.ver = 2
-    _tiger.init(&ctx)
-    buf := make([]byte, 512)
-    defer delete(buf)
-    read := 1
-    for read > 0 {
-        read, _ = s->impl_read(buf)
-        if read > 0 {
-            _tiger.update(&ctx, buf[:read])
-        } 
-    }
-    _tiger.final(&ctx, hash[:])
-    return hash, true
-}
-
-// hash_file_192 will read the file provided by the given handle
-// and compute a hash
-hash_file_192 :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE_192]byte, bool) {
-    if !load_at_once {
-        return hash_stream_192(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes_192(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE_192]byte{}, false
-}
-
-hash_192 :: proc {
-    hash_stream_192,
-    hash_file_192,
-    hash_bytes_192,
-    hash_string_192,
-    hash_bytes_to_buffer_192,
-    hash_string_to_buffer_192,
-}
-
-/*
-    Low level API
-*/
-
-Tiger_Context :: _tiger.Tiger_Context
-
-init :: proc(ctx: ^_tiger.Tiger_Context) {
-    ctx.ver = 2
-    _tiger.init(ctx)
-}
-
-update :: proc(ctx: ^_tiger.Tiger_Context, data: []byte) {
-    _tiger.update(ctx, data)
-}
-
-final :: proc(ctx: ^_tiger.Tiger_Context, hash: []byte) {
-    _tiger.final(ctx, hash)
-}
@@ -1,144 +0,0 @@
-package util
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Various utility procedures
-*/
-
-import "core:mem"
-
-// @note(bp): this can replace the other two
-cast_slice :: #force_inline proc "contextless" ($D: typeid/[]$DE, src: $S/[]$SE) -> D {
-    src := src
-    dst := (^mem.Raw_Slice)(&src)
-
-    when size_of(DE) < size_of(SE) {
-        when size_of(DE) % size_of(SE) == 0 {
-            dst.len /= size_of(SE) / size_of(DE)
-        } else {
-            dst.len *= size_of(SE)
-            dst.len /= size_of(DE)
-        }
-    } else when size_of(DE) > size_of(SE) {
-        when size_of(DE) % size_of(SE) == 0 {
-            dst.len *= size_of(DE) / size_of(SE)
-        } else {
-            dst.len *= size_of(SE)
-            dst.len /= size_of(DE)
-        }
-    } else when size_of(DE) != size_of(SE) {
-        #assert(size_of(DE) % size_of(SE) == 0, "Different size detected")
-        dst.len *= size_of(SE)
-        dst.len /= size_of(DE)
-    }
-
-    return (^D)(dst)^
-}
-
-bytes_to_slice :: #force_inline proc "contextless" ($T: typeid/[]$E, bytes: []byte) -> T {
-    s := transmute(mem.Raw_Slice)bytes
-    s.len /= size_of(E)
-    return transmute(T)s
-}
-
-slice_to_bytes :: #force_inline proc "contextless" (slice: $E/[]$T) -> []byte {
-    s := transmute(mem.Raw_Slice)slice
-    s.len *= size_of(T)
-    return transmute([]byte)s
-}
-
-ROTL16 :: #force_inline proc "contextless" (a, b: u16) -> u16 {
-    return ((a << b) | (a >> (16 - b)))
-}
-
-ROTR16 :: #force_inline proc "contextless" (a, b: u16) -> u16 {
-    return ((a >> b) | (a << (16 - b)))
-}
-
-ROTL32 :: #force_inline proc "contextless"(a: u32, b: int) -> u32 {
-    s := uint(b) & 31
-    return (a << s) | (a >> (32 - s))
-}
-
-ROTR32 :: #force_inline proc "contextless" (a: u32, b: int) -> u32 {
-    s := uint(b) & 31
-    return (a >> s) | (a << (32 - s))
-}
-
-ROTL64 :: #force_inline proc "contextless" (a, b: u64) -> u64 {
-    return ((a << b) | (a >> (64 - b)))
-}
-
-ROTR64 :: #force_inline proc "contextless" (a, b: u64) -> u64 {
-    return ((a >> b) | (a << (64 - b)))
-}
-
-ROTL128 :: #force_inline proc "contextless" (a, b, c, d: ^u32, n: uint) {
-    a, b, c, d := a, b, c, d
-    t := a^ >> (32 - n)
-    a^ = ((a^ << n) | (b^ >> (32 - n)))
-    b^ = ((b^ << n) | (c^ >> (32 - n)))
-    c^ = ((c^ << n) | (d^ >> (32 - n)))
-    d^ = ((d^ << n) | t)
-}
-
-U32_LE :: #force_inline proc "contextless" (b: []byte) -> u32 {
-    return u32(b[0]) | u32(b[1]) << 8 | u32(b[2]) << 16 | u32(b[3]) << 24
-}
-
-U64_LE :: #force_inline proc "contextless" (b: []byte) -> u64 {
-    return u64(b[0])       | u64(b[1]) << 8  | u64(b[2]) << 16 | u64(b[3]) << 24 |
-           u64(b[4]) << 32 | u64(b[5]) << 40 | u64(b[6]) << 48 | u64(b[7]) << 56
-}
-
-U64_BE :: #force_inline proc "contextless" (b: []byte) -> u64 {
-    return u64(b[7])       | u64(b[6]) << 8  | u64(b[5]) << 16 | u64(b[4]) << 24 |
-           u64(b[3]) << 32 | u64(b[2]) << 40 | u64(b[1]) << 48 | u64(b[0]) << 56
-}
-
-PUT_U64_LE :: #force_inline proc "contextless" (b: []byte, v: u64) {
-    b[0] = byte(v)
-    b[1] = byte(v >> 8)
-    b[2] = byte(v >> 16)
-    b[3] = byte(v >> 24)
-    b[4] = byte(v >> 32)
-    b[5] = byte(v >> 40)
-    b[6] = byte(v >> 48)
-    b[7] = byte(v >> 56)
-}
-
-PUT_U32_LE :: #force_inline proc "contextless" (b: []byte, v: u32) {
-    b[0] = byte(v)
-    b[1] = byte(v >> 8)
-    b[2] = byte(v >> 16)
-    b[3] = byte(v >> 24)
-}
-
-PUT_U32_BE :: #force_inline proc "contextless" (b: []byte, v: u32) {
-    b[0] = byte(v >> 24)
-    b[1] = byte(v >> 16)
-    b[2] = byte(v >> 8)
-    b[3] = byte(v)
-}
-
-PUT_U64_BE :: #force_inline proc "contextless" (b: []byte, v: u64) {
-    b[0] = byte(v >> 56)
-    b[1] = byte(v >> 48)
-    b[2] = byte(v >> 40)
-    b[3] = byte(v >> 32)
-    b[4] = byte(v >> 24)
-    b[5] = byte(v >> 16)
-    b[6] = byte(v >> 8)
-    b[7] = byte(v)
-}
-
-XOR_BUF :: #force_inline proc "contextless" (input, output: []byte) {
-    for i := 0; i < len(input); i += 1 {
-        output[i] ~= input[i]
-    }
-}
@@ -1,806 +0,0 @@
-package whirlpool
-
-/*
-    Copyright 2021 zhibog
-    Made available under the BSD-3 license.
-
-    List of contributors:
-        zhibog, dotbmp:  Initial implementation.
-
-    Implementation of the Whirlpool hashing algorithm, as defined in <https://web.archive.org/web/20171129084214/http://www.larc.usp.br/~pbarreto/WhirlpoolPage.html>
-*/
-
-import "core:os"
-import "core:io"
-
-import "../util"
-
-/*
-    High level API
-*/
-
-DIGEST_SIZE :: 64
-
-// hash_string will hash the given input and return the
-// computed hash
-hash_string :: proc(data: string) -> [DIGEST_SIZE]byte {
-    return hash_bytes(transmute([]byte)(data))
-}
-
-// hash_bytes will hash the given input and return the
-// computed hash
-hash_bytes :: proc(data: []byte) -> [DIGEST_SIZE]byte {
-	hash: [DIGEST_SIZE]byte
-	ctx: Whirlpool_Context
-    // init(&ctx) No-op
-    update(&ctx, data)
-    final(&ctx, hash[:])
-    return hash
-}
-
-// hash_string_to_buffer will hash the given input and assign the
-// computed hash to the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_string_to_buffer :: proc(data: string, hash: []byte) {
-    hash_bytes_to_buffer(transmute([]byte)(data), hash)
-}
-
-// hash_bytes_to_buffer will hash the given input and write the
-// computed hash into the second parameter.
-// It requires that the destination buffer is at least as big as the digest size
-hash_bytes_to_buffer :: proc(data, hash: []byte) {
-    assert(len(hash) >= DIGEST_SIZE, "Size of destination buffer is smaller than the digest size")
-    ctx: Whirlpool_Context
-    // init(&ctx) No-op
-    update(&ctx, data)
-    final(&ctx, hash)
-}
-
-// hash_stream will read the stream in chunks and compute a
-// hash from its contents
-hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
-	hash: [DIGEST_SIZE]byte
-	ctx: Whirlpool_Context
-	// init(&ctx) No-op
-	buf := make([]byte, 512)
-	defer delete(buf)
-	read := 1
-	for read > 0 {
-	    read, _ = s->impl_read(buf)
-	    if read > 0 {
-			update(&ctx, buf[:read])
-	    } 
-	}
-	final(&ctx, hash[:])
-	return hash, true
-}
-
-// hash_file will read the file provided by the given handle
-// and compute a hash
-hash_file :: proc(hd: os.Handle, load_at_once := false) -> ([DIGEST_SIZE]byte, bool) {
-	if !load_at_once {
-        return hash_stream(os.stream_from_handle(hd))
-    } else {
-        if buf, ok := os.read_entire_file(hd); ok {
-            return hash_bytes(buf[:]), ok
-        }
-    }
-    return [DIGEST_SIZE]byte{}, false
-}
-
-hash :: proc {
-    hash_stream,
-    hash_file,
-    hash_bytes,
-    hash_string,
-    hash_bytes_to_buffer,
-    hash_string_to_buffer,
-}
-
-/*
-    Low level API
-*/
-
-@(warning="Init is a no-op for Whirlpool")
-init :: proc(ctx: ^Whirlpool_Context) {
-	// No action needed here
-}
-
-update :: proc(ctx: ^Whirlpool_Context, source: []byte) {
-    source_pos: int
-    nn := len(source)
-    source_bits := u64(nn * 8)
-    source_gap := u32((8 - (int(source_bits & 7))) & 7)
-    buffer_rem := uint(ctx.buffer_bits & 7)
-    b: u32
-
-	for i, carry, value := 31, u32(0), u32(source_bits); i >= 0 && (carry != 0 || value != 0); i -= 1 {
-		carry += u32(ctx.bitlength[i]) + (u32(value & 0xff))
-		ctx.bitlength[i] = byte(carry)
-		carry >>= 8
-		value >>= 8
-	}
-
-	for source_bits > 8 {
-		b = u32(u32((source[source_pos] << source_gap) & 0xff) | u32((source[source_pos+1] & 0xff) >> (8 - source_gap)))
-
-		ctx.buffer[ctx.buffer_pos] |= u8(b >> buffer_rem)
-		ctx.buffer_pos += 1
-		ctx.buffer_bits += int(8 - buffer_rem)
-
-		if ctx.buffer_bits == 512 {
-			transform(ctx)
-			ctx.buffer_bits = 0
-			ctx.buffer_pos = 0
-		}
-		ctx.buffer[ctx.buffer_pos] = byte(b << (8 - buffer_rem))
-		ctx.buffer_bits += int(buffer_rem)
-		source_bits -= 8
-		source_pos += 1
-	}
-
-	if source_bits > 0 {
-		b = u32((source[source_pos] << source_gap) & 0xff)
-		ctx.buffer[ctx.buffer_pos] |= byte(b) >> buffer_rem
-	} else {b = 0}
-
-	if u64(buffer_rem) + source_bits < 8 {
-		ctx.buffer_bits += int(source_bits)
-	} else {
-		ctx.buffer_pos += 1
-		ctx.buffer_bits += 8 - int(buffer_rem)
-		source_bits -= u64(8 - buffer_rem)
-
-		if ctx.buffer_bits == 512 {
-			transform(ctx)
-			ctx.buffer_bits = 0
-			ctx.buffer_pos = 0
-		}
-		ctx.buffer[ctx.buffer_pos] = byte(b << (8 - buffer_rem))
-		ctx.buffer_bits += int(source_bits)
-	}
-}
-
-final :: proc(ctx: ^Whirlpool_Context, hash: []byte) {
-	n := ctx
-	n.buffer[n.buffer_pos] |= 0x80 >> (uint(n.buffer_bits) & 7)
-	n.buffer_pos += 1
-
-	if n.buffer_pos > 64 - 32 {
-		if n.buffer_pos < 64 {
-			for i := 0; i < 64 - n.buffer_pos; i += 1 {
-				n.buffer[n.buffer_pos + i] = 0
-			}
-		}
-		transform(ctx)
-		n.buffer_pos = 0
-	}
-
-	if n.buffer_pos < 64 - 32 {
-		for i := 0; i < (64 - 32) - n.buffer_pos; i += 1 {
-			n.buffer[n.buffer_pos + i] = 0
-		}
-	}
-	n.buffer_pos = 64 - 32
-
-	for i := 0; i < 32; i += 1 {
-		n.buffer[n.buffer_pos + i] = n.bitlength[i]
-	}
-	transform(ctx)
-
-	for i := 0; i < 8; i += 1 {
-		hash[i * 8]     = byte(n.hash[i] >> 56)
-		hash[i * 8 + 1] = byte(n.hash[i] >> 48)
-		hash[i * 8 + 2] = byte(n.hash[i] >> 40)
-		hash[i * 8 + 3] = byte(n.hash[i] >> 32)
-		hash[i * 8 + 4] = byte(n.hash[i] >> 24)
-		hash[i * 8 + 5] = byte(n.hash[i] >> 16)
-		hash[i * 8 + 6] = byte(n.hash[i] >> 8)
-		hash[i * 8 + 7] = byte(n.hash[i])
-	}
-}
-
-/*
-	Whirlpool implementation
-*/
-
-ROUNDS :: 10
-
-Whirlpool_Context :: struct {
-    bitlength:   [32]byte,
-    buffer:      [64]byte,
-    buffer_bits: int,
-    buffer_pos:  int,
-    hash:        [8]u64,
-}
-
-C0 := [256]u64 {
-	0x18186018c07830d8, 0x23238c2305af4626, 0xc6c63fc67ef991b8, 0xe8e887e8136fcdfb,
-	0x878726874ca113cb, 0xb8b8dab8a9626d11, 0x0101040108050209, 0x4f4f214f426e9e0d,
-	0x3636d836adee6c9b, 0xa6a6a2a6590451ff, 0xd2d26fd2debdb90c, 0xf5f5f3f5fb06f70e,
-	0x7979f979ef80f296, 0x6f6fa16f5fcede30, 0x91917e91fcef3f6d, 0x52525552aa07a4f8,
-	0x60609d6027fdc047, 0xbcbccabc89766535, 0x9b9b569baccd2b37, 0x8e8e028e048c018a,
-	0xa3a3b6a371155bd2, 0x0c0c300c603c186c, 0x7b7bf17bff8af684, 0x3535d435b5e16a80,
-	0x1d1d741de8693af5, 0xe0e0a7e05347ddb3, 0xd7d77bd7f6acb321, 0xc2c22fc25eed999c,
-	0x2e2eb82e6d965c43, 0x4b4b314b627a9629, 0xfefedffea321e15d, 0x575741578216aed5,
-	0x15155415a8412abd, 0x7777c1779fb6eee8, 0x3737dc37a5eb6e92, 0xe5e5b3e57b56d79e,
-	0x9f9f469f8cd92313, 0xf0f0e7f0d317fd23, 0x4a4a354a6a7f9420, 0xdada4fda9e95a944,
-	0x58587d58fa25b0a2, 0xc9c903c906ca8fcf, 0x2929a429558d527c, 0x0a0a280a5022145a,
-	0xb1b1feb1e14f7f50, 0xa0a0baa0691a5dc9, 0x6b6bb16b7fdad614, 0x85852e855cab17d9,
-	0xbdbdcebd8173673c, 0x5d5d695dd234ba8f, 0x1010401080502090, 0xf4f4f7f4f303f507,
-	0xcbcb0bcb16c08bdd, 0x3e3ef83eedc67cd3, 0x0505140528110a2d, 0x676781671fe6ce78,
-	0xe4e4b7e47353d597, 0x27279c2725bb4e02, 0x4141194132588273, 0x8b8b168b2c9d0ba7,
-	0xa7a7a6a7510153f6, 0x7d7de97dcf94fab2, 0x95956e95dcfb3749, 0xd8d847d88e9fad56,
-	0xfbfbcbfb8b30eb70, 0xeeee9fee2371c1cd, 0x7c7ced7cc791f8bb, 0x6666856617e3cc71,
-	0xdddd53dda68ea77b, 0x17175c17b84b2eaf, 0x4747014702468e45, 0x9e9e429e84dc211a,
-	0xcaca0fca1ec589d4, 0x2d2db42d75995a58, 0xbfbfc6bf9179632e, 0x07071c07381b0e3f,
-	0xadad8ead012347ac, 0x5a5a755aea2fb4b0, 0x838336836cb51bef, 0x3333cc3385ff66b6,
-	0x636391633ff2c65c, 0x02020802100a0412, 0xaaaa92aa39384993, 0x7171d971afa8e2de,
-	0xc8c807c80ecf8dc6, 0x19196419c87d32d1, 0x494939497270923b, 0xd9d943d9869aaf5f,
-	0xf2f2eff2c31df931, 0xe3e3abe34b48dba8, 0x5b5b715be22ab6b9, 0x88881a8834920dbc,
-	0x9a9a529aa4c8293e, 0x262698262dbe4c0b, 0x3232c8328dfa64bf, 0xb0b0fab0e94a7d59,
-	0xe9e983e91b6acff2, 0x0f0f3c0f78331e77, 0xd5d573d5e6a6b733, 0x80803a8074ba1df4,
-	0xbebec2be997c6127, 0xcdcd13cd26de87eb, 0x3434d034bde46889, 0x48483d487a759032,
-	0xffffdbffab24e354, 0x7a7af57af78ff48d, 0x90907a90f4ea3d64, 0x5f5f615fc23ebe9d,
-	0x202080201da0403d, 0x6868bd6867d5d00f, 0x1a1a681ad07234ca, 0xaeae82ae192c41b7,
-	0xb4b4eab4c95e757d, 0x54544d549a19a8ce, 0x93937693ece53b7f, 0x222288220daa442f,
-	0x64648d6407e9c863, 0xf1f1e3f1db12ff2a, 0x7373d173bfa2e6cc, 0x12124812905a2482,
-	0x40401d403a5d807a, 0x0808200840281048, 0xc3c32bc356e89b95, 0xecec97ec337bc5df,
-	0xdbdb4bdb9690ab4d, 0xa1a1bea1611f5fc0, 0x8d8d0e8d1c830791, 0x3d3df43df5c97ac8,
-	0x97976697ccf1335b, 0x0000000000000000, 0xcfcf1bcf36d483f9, 0x2b2bac2b4587566e,
-	0x7676c57697b3ece1, 0x8282328264b019e6, 0xd6d67fd6fea9b128, 0x1b1b6c1bd87736c3,
-	0xb5b5eeb5c15b7774, 0xafaf86af112943be, 0x6a6ab56a77dfd41d, 0x50505d50ba0da0ea,
-	0x45450945124c8a57, 0xf3f3ebf3cb18fb38, 0x3030c0309df060ad, 0xefef9bef2b74c3c4,
-	0x3f3ffc3fe5c37eda, 0x55554955921caac7, 0xa2a2b2a2791059db, 0xeaea8fea0365c9e9,
-	0x656589650fecca6a, 0xbabad2bab9686903, 0x2f2fbc2f65935e4a, 0xc0c027c04ee79d8e,
-	0xdede5fdebe81a160, 0x1c1c701ce06c38fc, 0xfdfdd3fdbb2ee746, 0x4d4d294d52649a1f,
-	0x92927292e4e03976, 0x7575c9758fbceafa, 0x06061806301e0c36, 0x8a8a128a249809ae,
-	0xb2b2f2b2f940794b, 0xe6e6bfe66359d185, 0x0e0e380e70361c7e, 0x1f1f7c1ff8633ee7,
-	0x6262956237f7c455, 0xd4d477d4eea3b53a, 0xa8a89aa829324d81, 0x96966296c4f43152,
-	0xf9f9c3f99b3aef62, 0xc5c533c566f697a3, 0x2525942535b14a10, 0x59597959f220b2ab,
-	0x84842a8454ae15d0, 0x7272d572b7a7e4c5, 0x3939e439d5dd72ec, 0x4c4c2d4c5a619816,
-	0x5e5e655eca3bbc94, 0x7878fd78e785f09f, 0x3838e038ddd870e5, 0x8c8c0a8c14860598,
-	0xd1d163d1c6b2bf17, 0xa5a5aea5410b57e4, 0xe2e2afe2434dd9a1, 0x616199612ff8c24e,
-	0xb3b3f6b3f1457b42, 0x2121842115a54234, 0x9c9c4a9c94d62508, 0x1e1e781ef0663cee,
-	0x4343114322528661, 0xc7c73bc776fc93b1, 0xfcfcd7fcb32be54f, 0x0404100420140824,
-	0x51515951b208a2e3, 0x99995e99bcc72f25, 0x6d6da96d4fc4da22, 0x0d0d340d68391a65,
-	0xfafacffa8335e979, 0xdfdf5bdfb684a369, 0x7e7ee57ed79bfca9, 0x242490243db44819,
-	0x3b3bec3bc5d776fe, 0xabab96ab313d4b9a, 0xcece1fce3ed181f0, 0x1111441188552299,
-	0x8f8f068f0c890383, 0x4e4e254e4a6b9c04, 0xb7b7e6b7d1517366, 0xebeb8beb0b60cbe0,
-	0x3c3cf03cfdcc78c1, 0x81813e817cbf1ffd, 0x94946a94d4fe3540, 0xf7f7fbf7eb0cf31c,
-	0xb9b9deb9a1676f18, 0x13134c13985f268b, 0x2c2cb02c7d9c5851, 0xd3d36bd3d6b8bb05,
-	0xe7e7bbe76b5cd38c, 0x6e6ea56e57cbdc39, 0xc4c437c46ef395aa, 0x03030c03180f061b,
-	0x565645568a13acdc, 0x44440d441a49885e, 0x7f7fe17fdf9efea0, 0xa9a99ea921374f88,
-	0x2a2aa82a4d825467, 0xbbbbd6bbb16d6b0a, 0xc1c123c146e29f87, 0x53535153a202a6f1,
-	0xdcdc57dcae8ba572, 0x0b0b2c0b58271653, 0x9d9d4e9d9cd32701, 0x6c6cad6c47c1d82b,
-	0x3131c43195f562a4, 0x7474cd7487b9e8f3, 0xf6f6fff6e309f115, 0x464605460a438c4c,
-	0xacac8aac092645a5, 0x89891e893c970fb5, 0x14145014a04428b4, 0xe1e1a3e15b42dfba,
-	0x16165816b04e2ca6, 0x3a3ae83acdd274f7, 0x6969b9696fd0d206, 0x09092409482d1241,
-	0x7070dd70a7ade0d7, 0xb6b6e2b6d954716f, 0xd0d067d0ceb7bd1e, 0xeded93ed3b7ec7d6,
-	0xcccc17cc2edb85e2, 0x424215422a578468, 0x98985a98b4c22d2c, 0xa4a4aaa4490e55ed,
-	0x2828a0285d885075, 0x5c5c6d5cda31b886, 0xf8f8c7f8933fed6b, 0x8686228644a411c2,
-}
-
-C1 := [256]u64 {
-	0xd818186018c07830, 0x2623238c2305af46, 0xb8c6c63fc67ef991, 0xfbe8e887e8136fcd,
-	0xcb878726874ca113, 0x11b8b8dab8a9626d, 0x0901010401080502, 0x0d4f4f214f426e9e,
-	0x9b3636d836adee6c, 0xffa6a6a2a6590451, 0x0cd2d26fd2debdb9, 0x0ef5f5f3f5fb06f7,
-	0x967979f979ef80f2, 0x306f6fa16f5fcede, 0x6d91917e91fcef3f, 0xf852525552aa07a4,
-	0x4760609d6027fdc0, 0x35bcbccabc897665, 0x379b9b569baccd2b, 0x8a8e8e028e048c01,
-	0xd2a3a3b6a371155b, 0x6c0c0c300c603c18, 0x847b7bf17bff8af6, 0x803535d435b5e16a,
-	0xf51d1d741de8693a, 0xb3e0e0a7e05347dd, 0x21d7d77bd7f6acb3, 0x9cc2c22fc25eed99,
-	0x432e2eb82e6d965c, 0x294b4b314b627a96, 0x5dfefedffea321e1, 0xd5575741578216ae,
-	0xbd15155415a8412a, 0xe87777c1779fb6ee, 0x923737dc37a5eb6e, 0x9ee5e5b3e57b56d7,
-	0x139f9f469f8cd923, 0x23f0f0e7f0d317fd, 0x204a4a354a6a7f94, 0x44dada4fda9e95a9,
-	0xa258587d58fa25b0, 0xcfc9c903c906ca8f, 0x7c2929a429558d52, 0x5a0a0a280a502214,
-	0x50b1b1feb1e14f7f, 0xc9a0a0baa0691a5d, 0x146b6bb16b7fdad6, 0xd985852e855cab17,
-	0x3cbdbdcebd817367, 0x8f5d5d695dd234ba, 0x9010104010805020, 0x07f4f4f7f4f303f5,
-	0xddcbcb0bcb16c08b, 0xd33e3ef83eedc67c, 0x2d0505140528110a, 0x78676781671fe6ce,
-	0x97e4e4b7e47353d5, 0x0227279c2725bb4e, 0x7341411941325882, 0xa78b8b168b2c9d0b,
-	0xf6a7a7a6a7510153, 0xb27d7de97dcf94fa, 0x4995956e95dcfb37, 0x56d8d847d88e9fad,
-	0x70fbfbcbfb8b30eb, 0xcdeeee9fee2371c1, 0xbb7c7ced7cc791f8, 0x716666856617e3cc,
-	0x7bdddd53dda68ea7, 0xaf17175c17b84b2e, 0x454747014702468e, 0x1a9e9e429e84dc21,
-	0xd4caca0fca1ec589, 0x582d2db42d75995a, 0x2ebfbfc6bf917963, 0x3f07071c07381b0e,
-	0xacadad8ead012347, 0xb05a5a755aea2fb4, 0xef838336836cb51b, 0xb63333cc3385ff66,
-	0x5c636391633ff2c6, 0x1202020802100a04, 0x93aaaa92aa393849, 0xde7171d971afa8e2,
-	0xc6c8c807c80ecf8d, 0xd119196419c87d32, 0x3b49493949727092, 0x5fd9d943d9869aaf,
-	0x31f2f2eff2c31df9, 0xa8e3e3abe34b48db, 0xb95b5b715be22ab6, 0xbc88881a8834920d,
-	0x3e9a9a529aa4c829, 0x0b262698262dbe4c, 0xbf3232c8328dfa64, 0x59b0b0fab0e94a7d,
-	0xf2e9e983e91b6acf, 0x770f0f3c0f78331e, 0x33d5d573d5e6a6b7, 0xf480803a8074ba1d,
-	0x27bebec2be997c61, 0xebcdcd13cd26de87, 0x893434d034bde468, 0x3248483d487a7590,
-	0x54ffffdbffab24e3, 0x8d7a7af57af78ff4, 0x6490907a90f4ea3d, 0x9d5f5f615fc23ebe,
-	0x3d202080201da040, 0x0f6868bd6867d5d0, 0xca1a1a681ad07234, 0xb7aeae82ae192c41,
-	0x7db4b4eab4c95e75, 0xce54544d549a19a8, 0x7f93937693ece53b, 0x2f222288220daa44,
-	0x6364648d6407e9c8, 0x2af1f1e3f1db12ff, 0xcc7373d173bfa2e6, 0x8212124812905a24,
-	0x7a40401d403a5d80, 0x4808082008402810, 0x95c3c32bc356e89b, 0xdfecec97ec337bc5,
-	0x4ddbdb4bdb9690ab, 0xc0a1a1bea1611f5f, 0x918d8d0e8d1c8307, 0xc83d3df43df5c97a,
-	0x5b97976697ccf133, 0x0000000000000000, 0xf9cfcf1bcf36d483, 0x6e2b2bac2b458756,
-	0xe17676c57697b3ec, 0xe68282328264b019, 0x28d6d67fd6fea9b1, 0xc31b1b6c1bd87736,
-	0x74b5b5eeb5c15b77, 0xbeafaf86af112943, 0x1d6a6ab56a77dfd4, 0xea50505d50ba0da0,
-	0x5745450945124c8a, 0x38f3f3ebf3cb18fb, 0xad3030c0309df060, 0xc4efef9bef2b74c3,
-	0xda3f3ffc3fe5c37e, 0xc755554955921caa, 0xdba2a2b2a2791059, 0xe9eaea8fea0365c9,
-	0x6a656589650fecca, 0x03babad2bab96869, 0x4a2f2fbc2f65935e, 0x8ec0c027c04ee79d,
-	0x60dede5fdebe81a1, 0xfc1c1c701ce06c38, 0x46fdfdd3fdbb2ee7, 0x1f4d4d294d52649a,
-	0x7692927292e4e039, 0xfa7575c9758fbcea, 0x3606061806301e0c, 0xae8a8a128a249809,
-	0x4bb2b2f2b2f94079, 0x85e6e6bfe66359d1, 0x7e0e0e380e70361c, 0xe71f1f7c1ff8633e,
-	0x556262956237f7c4, 0x3ad4d477d4eea3b5, 0x81a8a89aa829324d, 0x5296966296c4f431,
-	0x62f9f9c3f99b3aef, 0xa3c5c533c566f697, 0x102525942535b14a, 0xab59597959f220b2,
-	0xd084842a8454ae15, 0xc57272d572b7a7e4, 0xec3939e439d5dd72, 0x164c4c2d4c5a6198,
-	0x945e5e655eca3bbc, 0x9f7878fd78e785f0, 0xe53838e038ddd870, 0x988c8c0a8c148605,
-	0x17d1d163d1c6b2bf, 0xe4a5a5aea5410b57, 0xa1e2e2afe2434dd9, 0x4e616199612ff8c2,
-	0x42b3b3f6b3f1457b, 0x342121842115a542, 0x089c9c4a9c94d625, 0xee1e1e781ef0663c,
-	0x6143431143225286, 0xb1c7c73bc776fc93, 0x4ffcfcd7fcb32be5, 0x2404041004201408,
-	0xe351515951b208a2, 0x2599995e99bcc72f, 0x226d6da96d4fc4da, 0x650d0d340d68391a,
-	0x79fafacffa8335e9, 0x69dfdf5bdfb684a3, 0xa97e7ee57ed79bfc, 0x19242490243db448,
-	0xfe3b3bec3bc5d776, 0x9aabab96ab313d4b, 0xf0cece1fce3ed181, 0x9911114411885522,
-	0x838f8f068f0c8903, 0x044e4e254e4a6b9c, 0x66b7b7e6b7d15173, 0xe0ebeb8beb0b60cb,
-	0xc13c3cf03cfdcc78, 0xfd81813e817cbf1f, 0x4094946a94d4fe35, 0x1cf7f7fbf7eb0cf3,
-	0x18b9b9deb9a1676f, 0x8b13134c13985f26, 0x512c2cb02c7d9c58, 0x05d3d36bd3d6b8bb,
-	0x8ce7e7bbe76b5cd3, 0x396e6ea56e57cbdc, 0xaac4c437c46ef395, 0x1b03030c03180f06,
-	0xdc565645568a13ac, 0x5e44440d441a4988, 0xa07f7fe17fdf9efe, 0x88a9a99ea921374f,
-	0x672a2aa82a4d8254, 0x0abbbbd6bbb16d6b, 0x87c1c123c146e29f, 0xf153535153a202a6,
-	0x72dcdc57dcae8ba5, 0x530b0b2c0b582716, 0x019d9d4e9d9cd327, 0x2b6c6cad6c47c1d8,
-	0xa43131c43195f562, 0xf37474cd7487b9e8, 0x15f6f6fff6e309f1, 0x4c464605460a438c,
-	0xa5acac8aac092645, 0xb589891e893c970f, 0xb414145014a04428, 0xbae1e1a3e15b42df,
-	0xa616165816b04e2c, 0xf73a3ae83acdd274, 0x066969b9696fd0d2, 0x4109092409482d12,
-	0xd77070dd70a7ade0, 0x6fb6b6e2b6d95471, 0x1ed0d067d0ceb7bd, 0xd6eded93ed3b7ec7,
-	0xe2cccc17cc2edb85, 0x68424215422a5784, 0x2c98985a98b4c22d, 0xeda4a4aaa4490e55,
-	0x752828a0285d8850, 0x865c5c6d5cda31b8, 0x6bf8f8c7f8933fed, 0xc28686228644a411,
-}
-
-C2 := [256]u64 {
-	0x30d818186018c078, 0x462623238c2305af, 0x91b8c6c63fc67ef9, 0xcdfbe8e887e8136f,
-	0x13cb878726874ca1, 0x6d11b8b8dab8a962, 0x0209010104010805, 0x9e0d4f4f214f426e,
-	0x6c9b3636d836adee, 0x51ffa6a6a2a65904, 0xb90cd2d26fd2debd, 0xf70ef5f5f3f5fb06,
-	0xf2967979f979ef80, 0xde306f6fa16f5fce, 0x3f6d91917e91fcef, 0xa4f852525552aa07,
-	0xc04760609d6027fd, 0x6535bcbccabc8976, 0x2b379b9b569baccd, 0x018a8e8e028e048c,
-	0x5bd2a3a3b6a37115, 0x186c0c0c300c603c, 0xf6847b7bf17bff8a, 0x6a803535d435b5e1,
-	0x3af51d1d741de869, 0xddb3e0e0a7e05347, 0xb321d7d77bd7f6ac, 0x999cc2c22fc25eed,
-	0x5c432e2eb82e6d96, 0x96294b4b314b627a, 0xe15dfefedffea321, 0xaed5575741578216,
-	0x2abd15155415a841, 0xeee87777c1779fb6, 0x6e923737dc37a5eb, 0xd79ee5e5b3e57b56,
-	0x23139f9f469f8cd9, 0xfd23f0f0e7f0d317, 0x94204a4a354a6a7f, 0xa944dada4fda9e95,
-	0xb0a258587d58fa25, 0x8fcfc9c903c906ca, 0x527c2929a429558d, 0x145a0a0a280a5022,
-	0x7f50b1b1feb1e14f, 0x5dc9a0a0baa0691a, 0xd6146b6bb16b7fda, 0x17d985852e855cab,
-	0x673cbdbdcebd8173, 0xba8f5d5d695dd234, 0x2090101040108050, 0xf507f4f4f7f4f303,
-	0x8bddcbcb0bcb16c0, 0x7cd33e3ef83eedc6, 0x0a2d050514052811, 0xce78676781671fe6,
-	0xd597e4e4b7e47353, 0x4e0227279c2725bb, 0x8273414119413258, 0x0ba78b8b168b2c9d,
-	0x53f6a7a7a6a75101, 0xfab27d7de97dcf94, 0x374995956e95dcfb, 0xad56d8d847d88e9f,
-	0xeb70fbfbcbfb8b30, 0xc1cdeeee9fee2371, 0xf8bb7c7ced7cc791, 0xcc716666856617e3,
-	0xa77bdddd53dda68e, 0x2eaf17175c17b84b, 0x8e45474701470246, 0x211a9e9e429e84dc,
-	0x89d4caca0fca1ec5, 0x5a582d2db42d7599, 0x632ebfbfc6bf9179, 0x0e3f07071c07381b,
-	0x47acadad8ead0123, 0xb4b05a5a755aea2f, 0x1bef838336836cb5, 0x66b63333cc3385ff,
-	0xc65c636391633ff2, 0x041202020802100a, 0x4993aaaa92aa3938, 0xe2de7171d971afa8,
-	0x8dc6c8c807c80ecf, 0x32d119196419c87d, 0x923b494939497270, 0xaf5fd9d943d9869a,
-	0xf931f2f2eff2c31d, 0xdba8e3e3abe34b48, 0xb6b95b5b715be22a, 0x0dbc88881a883492,
-	0x293e9a9a529aa4c8, 0x4c0b262698262dbe, 0x64bf3232c8328dfa, 0x7d59b0b0fab0e94a,
-	0xcff2e9e983e91b6a, 0x1e770f0f3c0f7833, 0xb733d5d573d5e6a6, 0x1df480803a8074ba,
-	0x6127bebec2be997c, 0x87ebcdcd13cd26de, 0x68893434d034bde4, 0x903248483d487a75,
-	0xe354ffffdbffab24, 0xf48d7a7af57af78f, 0x3d6490907a90f4ea, 0xbe9d5f5f615fc23e,
-	0x403d202080201da0, 0xd00f6868bd6867d5, 0x34ca1a1a681ad072, 0x41b7aeae82ae192c,
-	0x757db4b4eab4c95e, 0xa8ce54544d549a19, 0x3b7f93937693ece5, 0x442f222288220daa,
-	0xc86364648d6407e9, 0xff2af1f1e3f1db12, 0xe6cc7373d173bfa2, 0x248212124812905a,
-	0x807a40401d403a5d, 0x1048080820084028, 0x9b95c3c32bc356e8, 0xc5dfecec97ec337b,
-	0xab4ddbdb4bdb9690, 0x5fc0a1a1bea1611f, 0x07918d8d0e8d1c83, 0x7ac83d3df43df5c9,
-	0x335b97976697ccf1, 0x0000000000000000, 0x83f9cfcf1bcf36d4, 0x566e2b2bac2b4587,
-	0xece17676c57697b3, 0x19e68282328264b0, 0xb128d6d67fd6fea9, 0x36c31b1b6c1bd877,
-	0x7774b5b5eeb5c15b, 0x43beafaf86af1129, 0xd41d6a6ab56a77df, 0xa0ea50505d50ba0d,
-	0x8a5745450945124c, 0xfb38f3f3ebf3cb18, 0x60ad3030c0309df0, 0xc3c4efef9bef2b74,
-	0x7eda3f3ffc3fe5c3, 0xaac755554955921c, 0x59dba2a2b2a27910, 0xc9e9eaea8fea0365,
-	0xca6a656589650fec, 0x6903babad2bab968, 0x5e4a2f2fbc2f6593, 0x9d8ec0c027c04ee7,
-	0xa160dede5fdebe81, 0x38fc1c1c701ce06c, 0xe746fdfdd3fdbb2e, 0x9a1f4d4d294d5264,
-	0x397692927292e4e0, 0xeafa7575c9758fbc, 0x0c3606061806301e, 0x09ae8a8a128a2498,
-	0x794bb2b2f2b2f940, 0xd185e6e6bfe66359, 0x1c7e0e0e380e7036, 0x3ee71f1f7c1ff863,
-	0xc4556262956237f7, 0xb53ad4d477d4eea3, 0x4d81a8a89aa82932, 0x315296966296c4f4,
-	0xef62f9f9c3f99b3a, 0x97a3c5c533c566f6, 0x4a102525942535b1, 0xb2ab59597959f220,
-	0x15d084842a8454ae, 0xe4c57272d572b7a7, 0x72ec3939e439d5dd, 0x98164c4c2d4c5a61,
-	0xbc945e5e655eca3b, 0xf09f7878fd78e785, 0x70e53838e038ddd8, 0x05988c8c0a8c1486,
-	0xbf17d1d163d1c6b2, 0x57e4a5a5aea5410b, 0xd9a1e2e2afe2434d, 0xc24e616199612ff8,
-	0x7b42b3b3f6b3f145, 0x42342121842115a5, 0x25089c9c4a9c94d6, 0x3cee1e1e781ef066,
-	0x8661434311432252, 0x93b1c7c73bc776fc, 0xe54ffcfcd7fcb32b, 0x0824040410042014,
-	0xa2e351515951b208, 0x2f2599995e99bcc7, 0xda226d6da96d4fc4, 0x1a650d0d340d6839,
-	0xe979fafacffa8335, 0xa369dfdf5bdfb684, 0xfca97e7ee57ed79b, 0x4819242490243db4,
-	0x76fe3b3bec3bc5d7, 0x4b9aabab96ab313d, 0x81f0cece1fce3ed1, 0x2299111144118855,
-	0x03838f8f068f0c89, 0x9c044e4e254e4a6b, 0x7366b7b7e6b7d151, 0xcbe0ebeb8beb0b60,
-	0x78c13c3cf03cfdcc, 0x1ffd81813e817cbf, 0x354094946a94d4fe, 0xf31cf7f7fbf7eb0c,
-	0x6f18b9b9deb9a167, 0x268b13134c13985f, 0x58512c2cb02c7d9c, 0xbb05d3d36bd3d6b8,
-	0xd38ce7e7bbe76b5c, 0xdc396e6ea56e57cb, 0x95aac4c437c46ef3, 0x061b03030c03180f,
-	0xacdc565645568a13, 0x885e44440d441a49, 0xfea07f7fe17fdf9e, 0x4f88a9a99ea92137,
-	0x54672a2aa82a4d82, 0x6b0abbbbd6bbb16d, 0x9f87c1c123c146e2, 0xa6f153535153a202,
-	0xa572dcdc57dcae8b, 0x16530b0b2c0b5827, 0x27019d9d4e9d9cd3, 0xd82b6c6cad6c47c1,
-	0x62a43131c43195f5, 0xe8f37474cd7487b9, 0xf115f6f6fff6e309, 0x8c4c464605460a43,
-	0x45a5acac8aac0926, 0x0fb589891e893c97, 0x28b414145014a044, 0xdfbae1e1a3e15b42,
-	0x2ca616165816b04e, 0x74f73a3ae83acdd2, 0xd2066969b9696fd0, 0x124109092409482d,
-	0xe0d77070dd70a7ad, 0x716fb6b6e2b6d954, 0xbd1ed0d067d0ceb7, 0xc7d6eded93ed3b7e,
-	0x85e2cccc17cc2edb, 0x8468424215422a57, 0x2d2c98985a98b4c2, 0x55eda4a4aaa4490e,
-	0x50752828a0285d88, 0xb8865c5c6d5cda31, 0xed6bf8f8c7f8933f, 0x11c28686228644a4,
-}
-
-C3 := [256]u64 {
-	0x7830d818186018c0, 0xaf462623238c2305, 0xf991b8c6c63fc67e, 0x6fcdfbe8e887e813,
-	0xa113cb878726874c, 0x626d11b8b8dab8a9, 0x0502090101040108, 0x6e9e0d4f4f214f42,
-	0xee6c9b3636d836ad, 0x0451ffa6a6a2a659, 0xbdb90cd2d26fd2de, 0x06f70ef5f5f3f5fb,
-	0x80f2967979f979ef, 0xcede306f6fa16f5f, 0xef3f6d91917e91fc, 0x07a4f852525552aa,
-	0xfdc04760609d6027, 0x766535bcbccabc89, 0xcd2b379b9b569bac, 0x8c018a8e8e028e04,
-	0x155bd2a3a3b6a371, 0x3c186c0c0c300c60, 0x8af6847b7bf17bff, 0xe16a803535d435b5,
-	0x693af51d1d741de8, 0x47ddb3e0e0a7e053, 0xacb321d7d77bd7f6, 0xed999cc2c22fc25e,
-	0x965c432e2eb82e6d, 0x7a96294b4b314b62, 0x21e15dfefedffea3, 0x16aed55757415782,
-	0x412abd15155415a8, 0xb6eee87777c1779f, 0xeb6e923737dc37a5, 0x56d79ee5e5b3e57b,
-	0xd923139f9f469f8c, 0x17fd23f0f0e7f0d3, 0x7f94204a4a354a6a, 0x95a944dada4fda9e,
-	0x25b0a258587d58fa, 0xca8fcfc9c903c906, 0x8d527c2929a42955, 0x22145a0a0a280a50,
-	0x4f7f50b1b1feb1e1, 0x1a5dc9a0a0baa069, 0xdad6146b6bb16b7f, 0xab17d985852e855c,
-	0x73673cbdbdcebd81, 0x34ba8f5d5d695dd2, 0x5020901010401080, 0x03f507f4f4f7f4f3,
-	0xc08bddcbcb0bcb16, 0xc67cd33e3ef83eed, 0x110a2d0505140528, 0xe6ce78676781671f,
-	0x53d597e4e4b7e473, 0xbb4e0227279c2725, 0x5882734141194132, 0x9d0ba78b8b168b2c,
-	0x0153f6a7a7a6a751, 0x94fab27d7de97dcf, 0xfb374995956e95dc, 0x9fad56d8d847d88e,
-	0x30eb70fbfbcbfb8b, 0x71c1cdeeee9fee23, 0x91f8bb7c7ced7cc7, 0xe3cc716666856617,
-	0x8ea77bdddd53dda6, 0x4b2eaf17175c17b8, 0x468e454747014702, 0xdc211a9e9e429e84,
-	0xc589d4caca0fca1e, 0x995a582d2db42d75, 0x79632ebfbfc6bf91, 0x1b0e3f07071c0738,
-	0x2347acadad8ead01, 0x2fb4b05a5a755aea, 0xb51bef838336836c, 0xff66b63333cc3385,
-	0xf2c65c636391633f, 0x0a04120202080210, 0x384993aaaa92aa39, 0xa8e2de7171d971af,
-	0xcf8dc6c8c807c80e, 0x7d32d119196419c8, 0x70923b4949394972, 0x9aaf5fd9d943d986,
-	0x1df931f2f2eff2c3, 0x48dba8e3e3abe34b, 0x2ab6b95b5b715be2, 0x920dbc88881a8834,
-	0xc8293e9a9a529aa4, 0xbe4c0b262698262d, 0xfa64bf3232c8328d, 0x4a7d59b0b0fab0e9,
-	0x6acff2e9e983e91b, 0x331e770f0f3c0f78, 0xa6b733d5d573d5e6, 0xba1df480803a8074,
-	0x7c6127bebec2be99, 0xde87ebcdcd13cd26, 0xe468893434d034bd, 0x75903248483d487a,
-	0x24e354ffffdbffab, 0x8ff48d7a7af57af7, 0xea3d6490907a90f4, 0x3ebe9d5f5f615fc2,
-	0xa0403d202080201d, 0xd5d00f6868bd6867, 0x7234ca1a1a681ad0, 0x2c41b7aeae82ae19,
-	0x5e757db4b4eab4c9, 0x19a8ce54544d549a, 0xe53b7f93937693ec, 0xaa442f222288220d,
-	0xe9c86364648d6407, 0x12ff2af1f1e3f1db, 0xa2e6cc7373d173bf, 0x5a24821212481290,
-	0x5d807a40401d403a, 0x2810480808200840, 0xe89b95c3c32bc356, 0x7bc5dfecec97ec33,
-	0x90ab4ddbdb4bdb96, 0x1f5fc0a1a1bea161, 0x8307918d8d0e8d1c, 0xc97ac83d3df43df5,
-	0xf1335b97976697cc, 0x0000000000000000, 0xd483f9cfcf1bcf36, 0x87566e2b2bac2b45,
-	0xb3ece17676c57697, 0xb019e68282328264, 0xa9b128d6d67fd6fe, 0x7736c31b1b6c1bd8,
-	0x5b7774b5b5eeb5c1, 0x2943beafaf86af11, 0xdfd41d6a6ab56a77, 0x0da0ea50505d50ba,
-	0x4c8a574545094512, 0x18fb38f3f3ebf3cb, 0xf060ad3030c0309d, 0x74c3c4efef9bef2b,
-	0xc37eda3f3ffc3fe5, 0x1caac75555495592, 0x1059dba2a2b2a279, 0x65c9e9eaea8fea03,
-	0xecca6a656589650f, 0x686903babad2bab9, 0x935e4a2f2fbc2f65, 0xe79d8ec0c027c04e,
-	0x81a160dede5fdebe, 0x6c38fc1c1c701ce0, 0x2ee746fdfdd3fdbb, 0x649a1f4d4d294d52,
-	0xe0397692927292e4, 0xbceafa7575c9758f, 0x1e0c360606180630, 0x9809ae8a8a128a24,
-	0x40794bb2b2f2b2f9, 0x59d185e6e6bfe663, 0x361c7e0e0e380e70, 0x633ee71f1f7c1ff8,
-	0xf7c4556262956237, 0xa3b53ad4d477d4ee, 0x324d81a8a89aa829, 0xf4315296966296c4,
-	0x3aef62f9f9c3f99b, 0xf697a3c5c533c566, 0xb14a102525942535, 0x20b2ab59597959f2,
-	0xae15d084842a8454, 0xa7e4c57272d572b7, 0xdd72ec3939e439d5, 0x6198164c4c2d4c5a,
-	0x3bbc945e5e655eca, 0x85f09f7878fd78e7, 0xd870e53838e038dd, 0x8605988c8c0a8c14,
-	0xb2bf17d1d163d1c6, 0x0b57e4a5a5aea541, 0x4dd9a1e2e2afe243, 0xf8c24e616199612f,
-	0x457b42b3b3f6b3f1, 0xa542342121842115, 0xd625089c9c4a9c94, 0x663cee1e1e781ef0,
-	0x5286614343114322, 0xfc93b1c7c73bc776, 0x2be54ffcfcd7fcb3, 0x1408240404100420,
-	0x08a2e351515951b2, 0xc72f2599995e99bc, 0xc4da226d6da96d4f, 0x391a650d0d340d68,
-	0x35e979fafacffa83, 0x84a369dfdf5bdfb6, 0x9bfca97e7ee57ed7, 0xb44819242490243d,
-	0xd776fe3b3bec3bc5, 0x3d4b9aabab96ab31, 0xd181f0cece1fce3e, 0x5522991111441188,
-	0x8903838f8f068f0c, 0x6b9c044e4e254e4a, 0x517366b7b7e6b7d1, 0x60cbe0ebeb8beb0b,
-	0xcc78c13c3cf03cfd, 0xbf1ffd81813e817c, 0xfe354094946a94d4, 0x0cf31cf7f7fbf7eb,
-	0x676f18b9b9deb9a1, 0x5f268b13134c1398, 0x9c58512c2cb02c7d, 0xb8bb05d3d36bd3d6,
-	0x5cd38ce7e7bbe76b, 0xcbdc396e6ea56e57, 0xf395aac4c437c46e, 0x0f061b03030c0318,
-	0x13acdc565645568a, 0x49885e44440d441a, 0x9efea07f7fe17fdf, 0x374f88a9a99ea921,
-	0x8254672a2aa82a4d, 0x6d6b0abbbbd6bbb1, 0xe29f87c1c123c146, 0x02a6f153535153a2,
-	0x8ba572dcdc57dcae, 0x2716530b0b2c0b58, 0xd327019d9d4e9d9c, 0xc1d82b6c6cad6c47,
-	0xf562a43131c43195, 0xb9e8f37474cd7487, 0x09f115f6f6fff6e3, 0x438c4c464605460a,
-	0x2645a5acac8aac09, 0x970fb589891e893c, 0x4428b414145014a0, 0x42dfbae1e1a3e15b,
-	0x4e2ca616165816b0, 0xd274f73a3ae83acd, 0xd0d2066969b9696f, 0x2d12410909240948,
-	0xade0d77070dd70a7, 0x54716fb6b6e2b6d9, 0xb7bd1ed0d067d0ce, 0x7ec7d6eded93ed3b,
-	0xdb85e2cccc17cc2e, 0x578468424215422a, 0xc22d2c98985a98b4, 0x0e55eda4a4aaa449,
-	0x8850752828a0285d, 0x31b8865c5c6d5cda, 0x3fed6bf8f8c7f893, 0xa411c28686228644,
-}
-
-C4 := [256]u64 {
-	0xc07830d818186018, 0x05af462623238c23, 0x7ef991b8c6c63fc6, 0x136fcdfbe8e887e8,
-	0x4ca113cb87872687, 0xa9626d11b8b8dab8, 0x0805020901010401, 0x426e9e0d4f4f214f,
-	0xadee6c9b3636d836, 0x590451ffa6a6a2a6, 0xdebdb90cd2d26fd2, 0xfb06f70ef5f5f3f5,
-	0xef80f2967979f979, 0x5fcede306f6fa16f, 0xfcef3f6d91917e91, 0xaa07a4f852525552,
-	0x27fdc04760609d60, 0x89766535bcbccabc, 0xaccd2b379b9b569b, 0x048c018a8e8e028e,
-	0x71155bd2a3a3b6a3, 0x603c186c0c0c300c, 0xff8af6847b7bf17b, 0xb5e16a803535d435,
-	0xe8693af51d1d741d, 0x5347ddb3e0e0a7e0, 0xf6acb321d7d77bd7, 0x5eed999cc2c22fc2,
-	0x6d965c432e2eb82e, 0x627a96294b4b314b, 0xa321e15dfefedffe, 0x8216aed557574157,
-	0xa8412abd15155415, 0x9fb6eee87777c177, 0xa5eb6e923737dc37, 0x7b56d79ee5e5b3e5,
-	0x8cd923139f9f469f, 0xd317fd23f0f0e7f0, 0x6a7f94204a4a354a, 0x9e95a944dada4fda,
-	0xfa25b0a258587d58, 0x06ca8fcfc9c903c9, 0x558d527c2929a429, 0x5022145a0a0a280a,
-	0xe14f7f50b1b1feb1, 0x691a5dc9a0a0baa0, 0x7fdad6146b6bb16b, 0x5cab17d985852e85,
-	0x8173673cbdbdcebd, 0xd234ba8f5d5d695d, 0x8050209010104010, 0xf303f507f4f4f7f4,
-	0x16c08bddcbcb0bcb, 0xedc67cd33e3ef83e, 0x28110a2d05051405, 0x1fe6ce7867678167,
-	0x7353d597e4e4b7e4, 0x25bb4e0227279c27, 0x3258827341411941, 0x2c9d0ba78b8b168b,
-	0x510153f6a7a7a6a7, 0xcf94fab27d7de97d, 0xdcfb374995956e95, 0x8e9fad56d8d847d8,
-	0x8b30eb70fbfbcbfb, 0x2371c1cdeeee9fee, 0xc791f8bb7c7ced7c, 0x17e3cc7166668566,
-	0xa68ea77bdddd53dd, 0xb84b2eaf17175c17, 0x02468e4547470147, 0x84dc211a9e9e429e,
-	0x1ec589d4caca0fca, 0x75995a582d2db42d, 0x9179632ebfbfc6bf, 0x381b0e3f07071c07,
-	0x012347acadad8ead, 0xea2fb4b05a5a755a, 0x6cb51bef83833683, 0x85ff66b63333cc33,
-	0x3ff2c65c63639163, 0x100a041202020802, 0x39384993aaaa92aa, 0xafa8e2de7171d971,
-	0x0ecf8dc6c8c807c8, 0xc87d32d119196419, 0x7270923b49493949, 0x869aaf5fd9d943d9,
-	0xc31df931f2f2eff2, 0x4b48dba8e3e3abe3, 0xe22ab6b95b5b715b, 0x34920dbc88881a88,
-	0xa4c8293e9a9a529a, 0x2dbe4c0b26269826, 0x8dfa64bf3232c832, 0xe94a7d59b0b0fab0,
-	0x1b6acff2e9e983e9, 0x78331e770f0f3c0f, 0xe6a6b733d5d573d5, 0x74ba1df480803a80,
-	0x997c6127bebec2be, 0x26de87ebcdcd13cd, 0xbde468893434d034, 0x7a75903248483d48,
-	0xab24e354ffffdbff, 0xf78ff48d7a7af57a, 0xf4ea3d6490907a90, 0xc23ebe9d5f5f615f,
-	0x1da0403d20208020, 0x67d5d00f6868bd68, 0xd07234ca1a1a681a, 0x192c41b7aeae82ae,
-	0xc95e757db4b4eab4, 0x9a19a8ce54544d54, 0xece53b7f93937693, 0x0daa442f22228822,
-	0x07e9c86364648d64, 0xdb12ff2af1f1e3f1, 0xbfa2e6cc7373d173, 0x905a248212124812,
-	0x3a5d807a40401d40, 0x4028104808082008, 0x56e89b95c3c32bc3, 0x337bc5dfecec97ec,
-	0x9690ab4ddbdb4bdb, 0x611f5fc0a1a1bea1, 0x1c8307918d8d0e8d, 0xf5c97ac83d3df43d,
-	0xccf1335b97976697, 0x0000000000000000, 0x36d483f9cfcf1bcf, 0x4587566e2b2bac2b,
-	0x97b3ece17676c576, 0x64b019e682823282, 0xfea9b128d6d67fd6, 0xd87736c31b1b6c1b,
-	0xc15b7774b5b5eeb5, 0x112943beafaf86af, 0x77dfd41d6a6ab56a, 0xba0da0ea50505d50,
-	0x124c8a5745450945, 0xcb18fb38f3f3ebf3, 0x9df060ad3030c030, 0x2b74c3c4efef9bef,
-	0xe5c37eda3f3ffc3f, 0x921caac755554955, 0x791059dba2a2b2a2, 0x0365c9e9eaea8fea,
-	0x0fecca6a65658965, 0xb9686903babad2ba, 0x65935e4a2f2fbc2f, 0x4ee79d8ec0c027c0,
-	0xbe81a160dede5fde, 0xe06c38fc1c1c701c, 0xbb2ee746fdfdd3fd, 0x52649a1f4d4d294d,
-	0xe4e0397692927292, 0x8fbceafa7575c975, 0x301e0c3606061806, 0x249809ae8a8a128a,
-	0xf940794bb2b2f2b2, 0x6359d185e6e6bfe6, 0x70361c7e0e0e380e, 0xf8633ee71f1f7c1f,
-	0x37f7c45562629562, 0xeea3b53ad4d477d4, 0x29324d81a8a89aa8, 0xc4f4315296966296,
-	0x9b3aef62f9f9c3f9, 0x66f697a3c5c533c5, 0x35b14a1025259425, 0xf220b2ab59597959,
-	0x54ae15d084842a84, 0xb7a7e4c57272d572, 0xd5dd72ec3939e439, 0x5a6198164c4c2d4c,
-	0xca3bbc945e5e655e, 0xe785f09f7878fd78, 0xddd870e53838e038, 0x148605988c8c0a8c,
-	0xc6b2bf17d1d163d1, 0x410b57e4a5a5aea5, 0x434dd9a1e2e2afe2, 0x2ff8c24e61619961,
-	0xf1457b42b3b3f6b3, 0x15a5423421218421, 0x94d625089c9c4a9c, 0xf0663cee1e1e781e,
-	0x2252866143431143, 0x76fc93b1c7c73bc7, 0xb32be54ffcfcd7fc, 0x2014082404041004,
-	0xb208a2e351515951, 0xbcc72f2599995e99, 0x4fc4da226d6da96d, 0x68391a650d0d340d,
-	0x8335e979fafacffa, 0xb684a369dfdf5bdf, 0xd79bfca97e7ee57e, 0x3db4481924249024,
-	0xc5d776fe3b3bec3b, 0x313d4b9aabab96ab, 0x3ed181f0cece1fce, 0x8855229911114411,
-	0x0c8903838f8f068f, 0x4a6b9c044e4e254e, 0xd1517366b7b7e6b7, 0x0b60cbe0ebeb8beb,
-	0xfdcc78c13c3cf03c, 0x7cbf1ffd81813e81, 0xd4fe354094946a94, 0xeb0cf31cf7f7fbf7,
-	0xa1676f18b9b9deb9, 0x985f268b13134c13, 0x7d9c58512c2cb02c, 0xd6b8bb05d3d36bd3,
-	0x6b5cd38ce7e7bbe7, 0x57cbdc396e6ea56e, 0x6ef395aac4c437c4, 0x180f061b03030c03,
-	0x8a13acdc56564556, 0x1a49885e44440d44, 0xdf9efea07f7fe17f, 0x21374f88a9a99ea9,
-	0x4d8254672a2aa82a, 0xb16d6b0abbbbd6bb, 0x46e29f87c1c123c1, 0xa202a6f153535153,
-	0xae8ba572dcdc57dc, 0x582716530b0b2c0b, 0x9cd327019d9d4e9d, 0x47c1d82b6c6cad6c,
-	0x95f562a43131c431, 0x87b9e8f37474cd74, 0xe309f115f6f6fff6, 0x0a438c4c46460546,
-	0x092645a5acac8aac, 0x3c970fb589891e89, 0xa04428b414145014, 0x5b42dfbae1e1a3e1,
-	0xb04e2ca616165816, 0xcdd274f73a3ae83a, 0x6fd0d2066969b969, 0x482d124109092409,
-	0xa7ade0d77070dd70, 0xd954716fb6b6e2b6, 0xceb7bd1ed0d067d0, 0x3b7ec7d6eded93ed,
-	0x2edb85e2cccc17cc, 0x2a57846842421542, 0xb4c22d2c98985a98, 0x490e55eda4a4aaa4,
-	0x5d8850752828a028, 0xda31b8865c5c6d5c, 0x933fed6bf8f8c7f8, 0x44a411c286862286,
-}
-
-C5 := [256]u64 {
-	0x18c07830d8181860, 0x2305af462623238c, 0xc67ef991b8c6c63f, 0xe8136fcdfbe8e887,
-	0x874ca113cb878726, 0xb8a9626d11b8b8da, 0x0108050209010104, 0x4f426e9e0d4f4f21,
-	0x36adee6c9b3636d8, 0xa6590451ffa6a6a2, 0xd2debdb90cd2d26f, 0xf5fb06f70ef5f5f3,
-	0x79ef80f2967979f9, 0x6f5fcede306f6fa1, 0x91fcef3f6d91917e, 0x52aa07a4f8525255,
-	0x6027fdc04760609d, 0xbc89766535bcbcca, 0x9baccd2b379b9b56, 0x8e048c018a8e8e02,
-	0xa371155bd2a3a3b6, 0x0c603c186c0c0c30, 0x7bff8af6847b7bf1, 0x35b5e16a803535d4,
-	0x1de8693af51d1d74, 0xe05347ddb3e0e0a7, 0xd7f6acb321d7d77b, 0xc25eed999cc2c22f,
-	0x2e6d965c432e2eb8, 0x4b627a96294b4b31, 0xfea321e15dfefedf, 0x578216aed5575741,
-	0x15a8412abd151554, 0x779fb6eee87777c1, 0x37a5eb6e923737dc, 0xe57b56d79ee5e5b3,
-	0x9f8cd923139f9f46, 0xf0d317fd23f0f0e7, 0x4a6a7f94204a4a35, 0xda9e95a944dada4f,
-	0x58fa25b0a258587d, 0xc906ca8fcfc9c903, 0x29558d527c2929a4, 0x0a5022145a0a0a28,
-	0xb1e14f7f50b1b1fe, 0xa0691a5dc9a0a0ba, 0x6b7fdad6146b6bb1, 0x855cab17d985852e,
-	0xbd8173673cbdbdce, 0x5dd234ba8f5d5d69, 0x1080502090101040, 0xf4f303f507f4f4f7,
-	0xcb16c08bddcbcb0b, 0x3eedc67cd33e3ef8, 0x0528110a2d050514, 0x671fe6ce78676781,
-	0xe47353d597e4e4b7, 0x2725bb4e0227279c, 0x4132588273414119, 0x8b2c9d0ba78b8b16,
-	0xa7510153f6a7a7a6, 0x7dcf94fab27d7de9, 0x95dcfb374995956e, 0xd88e9fad56d8d847,
-	0xfb8b30eb70fbfbcb, 0xee2371c1cdeeee9f, 0x7cc791f8bb7c7ced, 0x6617e3cc71666685,
-	0xdda68ea77bdddd53, 0x17b84b2eaf17175c, 0x4702468e45474701, 0x9e84dc211a9e9e42,
-	0xca1ec589d4caca0f, 0x2d75995a582d2db4, 0xbf9179632ebfbfc6, 0x07381b0e3f07071c,
-	0xad012347acadad8e, 0x5aea2fb4b05a5a75, 0x836cb51bef838336, 0x3385ff66b63333cc,
-	0x633ff2c65c636391, 0x02100a0412020208, 0xaa39384993aaaa92, 0x71afa8e2de7171d9,
-	0xc80ecf8dc6c8c807, 0x19c87d32d1191964, 0x497270923b494939, 0xd9869aaf5fd9d943,
-	0xf2c31df931f2f2ef, 0xe34b48dba8e3e3ab, 0x5be22ab6b95b5b71, 0x8834920dbc88881a,
-	0x9aa4c8293e9a9a52, 0x262dbe4c0b262698, 0x328dfa64bf3232c8, 0xb0e94a7d59b0b0fa,
-	0xe91b6acff2e9e983, 0x0f78331e770f0f3c, 0xd5e6a6b733d5d573, 0x8074ba1df480803a,
-	0xbe997c6127bebec2, 0xcd26de87ebcdcd13, 0x34bde468893434d0, 0x487a75903248483d,
-	0xffab24e354ffffdb, 0x7af78ff48d7a7af5, 0x90f4ea3d6490907a, 0x5fc23ebe9d5f5f61,
-	0x201da0403d202080, 0x6867d5d00f6868bd, 0x1ad07234ca1a1a68, 0xae192c41b7aeae82,
-	0xb4c95e757db4b4ea, 0x549a19a8ce54544d, 0x93ece53b7f939376, 0x220daa442f222288,
-	0x6407e9c86364648d, 0xf1db12ff2af1f1e3, 0x73bfa2e6cc7373d1, 0x12905a2482121248,
-	0x403a5d807a40401d, 0x0840281048080820, 0xc356e89b95c3c32b, 0xec337bc5dfecec97,
-	0xdb9690ab4ddbdb4b, 0xa1611f5fc0a1a1be, 0x8d1c8307918d8d0e, 0x3df5c97ac83d3df4,
-	0x97ccf1335b979766, 0x0000000000000000, 0xcf36d483f9cfcf1b, 0x2b4587566e2b2bac,
-	0x7697b3ece17676c5, 0x8264b019e6828232, 0xd6fea9b128d6d67f, 0x1bd87736c31b1b6c,
-	0xb5c15b7774b5b5ee, 0xaf112943beafaf86, 0x6a77dfd41d6a6ab5, 0x50ba0da0ea50505d,
-	0x45124c8a57454509, 0xf3cb18fb38f3f3eb, 0x309df060ad3030c0, 0xef2b74c3c4efef9b,
-	0x3fe5c37eda3f3ffc, 0x55921caac7555549, 0xa2791059dba2a2b2, 0xea0365c9e9eaea8f,
-	0x650fecca6a656589, 0xbab9686903babad2, 0x2f65935e4a2f2fbc, 0xc04ee79d8ec0c027,
-	0xdebe81a160dede5f, 0x1ce06c38fc1c1c70, 0xfdbb2ee746fdfdd3, 0x4d52649a1f4d4d29,
-	0x92e4e03976929272, 0x758fbceafa7575c9, 0x06301e0c36060618, 0x8a249809ae8a8a12,
-	0xb2f940794bb2b2f2, 0xe66359d185e6e6bf, 0x0e70361c7e0e0e38, 0x1ff8633ee71f1f7c,
-	0x6237f7c455626295, 0xd4eea3b53ad4d477, 0xa829324d81a8a89a, 0x96c4f43152969662,
-	0xf99b3aef62f9f9c3, 0xc566f697a3c5c533, 0x2535b14a10252594, 0x59f220b2ab595979,
-	0x8454ae15d084842a, 0x72b7a7e4c57272d5, 0x39d5dd72ec3939e4, 0x4c5a6198164c4c2d,
-	0x5eca3bbc945e5e65, 0x78e785f09f7878fd, 0x38ddd870e53838e0, 0x8c148605988c8c0a,
-	0xd1c6b2bf17d1d163, 0xa5410b57e4a5a5ae, 0xe2434dd9a1e2e2af, 0x612ff8c24e616199,
-	0xb3f1457b42b3b3f6, 0x2115a54234212184, 0x9c94d625089c9c4a, 0x1ef0663cee1e1e78,
-	0x4322528661434311, 0xc776fc93b1c7c73b, 0xfcb32be54ffcfcd7, 0x0420140824040410,
-	0x51b208a2e3515159, 0x99bcc72f2599995e, 0x6d4fc4da226d6da9, 0x0d68391a650d0d34,
-	0xfa8335e979fafacf, 0xdfb684a369dfdf5b, 0x7ed79bfca97e7ee5, 0x243db44819242490,
-	0x3bc5d776fe3b3bec, 0xab313d4b9aabab96, 0xce3ed181f0cece1f, 0x1188552299111144,
-	0x8f0c8903838f8f06, 0x4e4a6b9c044e4e25, 0xb7d1517366b7b7e6, 0xeb0b60cbe0ebeb8b,
-	0x3cfdcc78c13c3cf0, 0x817cbf1ffd81813e, 0x94d4fe354094946a, 0xf7eb0cf31cf7f7fb,
-	0xb9a1676f18b9b9de, 0x13985f268b13134c, 0x2c7d9c58512c2cb0, 0xd3d6b8bb05d3d36b,
-	0xe76b5cd38ce7e7bb, 0x6e57cbdc396e6ea5, 0xc46ef395aac4c437, 0x03180f061b03030c,
-	0x568a13acdc565645, 0x441a49885e44440d, 0x7fdf9efea07f7fe1, 0xa921374f88a9a99e,
-	0x2a4d8254672a2aa8, 0xbbb16d6b0abbbbd6, 0xc146e29f87c1c123, 0x53a202a6f1535351,
-	0xdcae8ba572dcdc57, 0x0b582716530b0b2c, 0x9d9cd327019d9d4e, 0x6c47c1d82b6c6cad,
-	0x3195f562a43131c4, 0x7487b9e8f37474cd, 0xf6e309f115f6f6ff, 0x460a438c4c464605,
-	0xac092645a5acac8a, 0x893c970fb589891e, 0x14a04428b4141450, 0xe15b42dfbae1e1a3,
-	0x16b04e2ca6161658, 0x3acdd274f73a3ae8, 0x696fd0d2066969b9, 0x09482d1241090924,
-	0x70a7ade0d77070dd, 0xb6d954716fb6b6e2, 0xd0ceb7bd1ed0d067, 0xed3b7ec7d6eded93,
-	0xcc2edb85e2cccc17, 0x422a578468424215, 0x98b4c22d2c98985a, 0xa4490e55eda4a4aa,
-	0x285d8850752828a0, 0x5cda31b8865c5c6d, 0xf8933fed6bf8f8c7, 0x8644a411c2868622,
-}
-
-C6 := [256]u64 {
-	0x6018c07830d81818, 0x8c2305af46262323, 0x3fc67ef991b8c6c6, 0x87e8136fcdfbe8e8,
-	0x26874ca113cb8787, 0xdab8a9626d11b8b8, 0x0401080502090101, 0x214f426e9e0d4f4f,
-	0xd836adee6c9b3636, 0xa2a6590451ffa6a6, 0x6fd2debdb90cd2d2, 0xf3f5fb06f70ef5f5,
-	0xf979ef80f2967979, 0xa16f5fcede306f6f, 0x7e91fcef3f6d9191, 0x5552aa07a4f85252,
-	0x9d6027fdc0476060, 0xcabc89766535bcbc, 0x569baccd2b379b9b, 0x028e048c018a8e8e,
-	0xb6a371155bd2a3a3, 0x300c603c186c0c0c, 0xf17bff8af6847b7b, 0xd435b5e16a803535,
-	0x741de8693af51d1d, 0xa7e05347ddb3e0e0, 0x7bd7f6acb321d7d7, 0x2fc25eed999cc2c2,
-	0xb82e6d965c432e2e, 0x314b627a96294b4b, 0xdffea321e15dfefe, 0x41578216aed55757,
-	0x5415a8412abd1515, 0xc1779fb6eee87777, 0xdc37a5eb6e923737, 0xb3e57b56d79ee5e5,
-	0x469f8cd923139f9f, 0xe7f0d317fd23f0f0, 0x354a6a7f94204a4a, 0x4fda9e95a944dada,
-	0x7d58fa25b0a25858, 0x03c906ca8fcfc9c9, 0xa429558d527c2929, 0x280a5022145a0a0a,
-	0xfeb1e14f7f50b1b1, 0xbaa0691a5dc9a0a0, 0xb16b7fdad6146b6b, 0x2e855cab17d98585,
-	0xcebd8173673cbdbd, 0x695dd234ba8f5d5d, 0x4010805020901010, 0xf7f4f303f507f4f4,
-	0x0bcb16c08bddcbcb, 0xf83eedc67cd33e3e, 0x140528110a2d0505, 0x81671fe6ce786767,
-	0xb7e47353d597e4e4, 0x9c2725bb4e022727, 0x1941325882734141, 0x168b2c9d0ba78b8b,
-	0xa6a7510153f6a7a7, 0xe97dcf94fab27d7d, 0x6e95dcfb37499595, 0x47d88e9fad56d8d8,
-	0xcbfb8b30eb70fbfb, 0x9fee2371c1cdeeee, 0xed7cc791f8bb7c7c, 0x856617e3cc716666,
-	0x53dda68ea77bdddd, 0x5c17b84b2eaf1717, 0x014702468e454747, 0x429e84dc211a9e9e,
-	0x0fca1ec589d4caca, 0xb42d75995a582d2d, 0xc6bf9179632ebfbf, 0x1c07381b0e3f0707,
-	0x8ead012347acadad, 0x755aea2fb4b05a5a, 0x36836cb51bef8383, 0xcc3385ff66b63333,
-	0x91633ff2c65c6363, 0x0802100a04120202, 0x92aa39384993aaaa, 0xd971afa8e2de7171,
-	0x07c80ecf8dc6c8c8, 0x6419c87d32d11919, 0x39497270923b4949, 0x43d9869aaf5fd9d9,
-	0xeff2c31df931f2f2, 0xabe34b48dba8e3e3, 0x715be22ab6b95b5b, 0x1a8834920dbc8888,
-	0x529aa4c8293e9a9a, 0x98262dbe4c0b2626, 0xc8328dfa64bf3232, 0xfab0e94a7d59b0b0,
-	0x83e91b6acff2e9e9, 0x3c0f78331e770f0f, 0x73d5e6a6b733d5d5, 0x3a8074ba1df48080,
-	0xc2be997c6127bebe, 0x13cd26de87ebcdcd, 0xd034bde468893434, 0x3d487a7590324848,
-	0xdbffab24e354ffff, 0xf57af78ff48d7a7a, 0x7a90f4ea3d649090, 0x615fc23ebe9d5f5f,
-	0x80201da0403d2020, 0xbd6867d5d00f6868, 0x681ad07234ca1a1a, 0x82ae192c41b7aeae,
-	0xeab4c95e757db4b4, 0x4d549a19a8ce5454, 0x7693ece53b7f9393, 0x88220daa442f2222,
-	0x8d6407e9c8636464, 0xe3f1db12ff2af1f1, 0xd173bfa2e6cc7373, 0x4812905a24821212,
-	0x1d403a5d807a4040, 0x2008402810480808, 0x2bc356e89b95c3c3, 0x97ec337bc5dfecec,
-	0x4bdb9690ab4ddbdb, 0xbea1611f5fc0a1a1, 0x0e8d1c8307918d8d, 0xf43df5c97ac83d3d,
-	0x6697ccf1335b9797, 0x0000000000000000, 0x1bcf36d483f9cfcf, 0xac2b4587566e2b2b,
-	0xc57697b3ece17676, 0x328264b019e68282, 0x7fd6fea9b128d6d6, 0x6c1bd87736c31b1b,
-	0xeeb5c15b7774b5b5, 0x86af112943beafaf, 0xb56a77dfd41d6a6a, 0x5d50ba0da0ea5050,
-	0x0945124c8a574545, 0xebf3cb18fb38f3f3, 0xc0309df060ad3030, 0x9bef2b74c3c4efef,
-	0xfc3fe5c37eda3f3f, 0x4955921caac75555, 0xb2a2791059dba2a2, 0x8fea0365c9e9eaea,
-	0x89650fecca6a6565, 0xd2bab9686903baba, 0xbc2f65935e4a2f2f, 0x27c04ee79d8ec0c0,
-	0x5fdebe81a160dede, 0x701ce06c38fc1c1c, 0xd3fdbb2ee746fdfd, 0x294d52649a1f4d4d,
-	0x7292e4e039769292, 0xc9758fbceafa7575, 0x1806301e0c360606, 0x128a249809ae8a8a,
-	0xf2b2f940794bb2b2, 0xbfe66359d185e6e6, 0x380e70361c7e0e0e, 0x7c1ff8633ee71f1f,
-	0x956237f7c4556262, 0x77d4eea3b53ad4d4, 0x9aa829324d81a8a8, 0x6296c4f431529696,
-	0xc3f99b3aef62f9f9, 0x33c566f697a3c5c5, 0x942535b14a102525, 0x7959f220b2ab5959,
-	0x2a8454ae15d08484, 0xd572b7a7e4c57272, 0xe439d5dd72ec3939, 0x2d4c5a6198164c4c,
-	0x655eca3bbc945e5e, 0xfd78e785f09f7878, 0xe038ddd870e53838, 0x0a8c148605988c8c,
-	0x63d1c6b2bf17d1d1, 0xaea5410b57e4a5a5, 0xafe2434dd9a1e2e2, 0x99612ff8c24e6161,
-	0xf6b3f1457b42b3b3, 0x842115a542342121, 0x4a9c94d625089c9c, 0x781ef0663cee1e1e,
-	0x1143225286614343, 0x3bc776fc93b1c7c7, 0xd7fcb32be54ffcfc, 0x1004201408240404,
-	0x5951b208a2e35151, 0x5e99bcc72f259999, 0xa96d4fc4da226d6d, 0x340d68391a650d0d,
-	0xcffa8335e979fafa, 0x5bdfb684a369dfdf, 0xe57ed79bfca97e7e, 0x90243db448192424,
-	0xec3bc5d776fe3b3b, 0x96ab313d4b9aabab, 0x1fce3ed181f0cece, 0x4411885522991111,
-	0x068f0c8903838f8f, 0x254e4a6b9c044e4e, 0xe6b7d1517366b7b7, 0x8beb0b60cbe0ebeb,
-	0xf03cfdcc78c13c3c, 0x3e817cbf1ffd8181, 0x6a94d4fe35409494, 0xfbf7eb0cf31cf7f7,
-	0xdeb9a1676f18b9b9, 0x4c13985f268b1313, 0xb02c7d9c58512c2c, 0x6bd3d6b8bb05d3d3,
-	0xbbe76b5cd38ce7e7, 0xa56e57cbdc396e6e, 0x37c46ef395aac4c4, 0x0c03180f061b0303,
-	0x45568a13acdc5656, 0x0d441a49885e4444, 0xe17fdf9efea07f7f, 0x9ea921374f88a9a9,
-	0xa82a4d8254672a2a, 0xd6bbb16d6b0abbbb, 0x23c146e29f87c1c1, 0x5153a202a6f15353,
-	0x57dcae8ba572dcdc, 0x2c0b582716530b0b, 0x4e9d9cd327019d9d, 0xad6c47c1d82b6c6c,
-	0xc43195f562a43131, 0xcd7487b9e8f37474, 0xfff6e309f115f6f6, 0x05460a438c4c4646,
-	0x8aac092645a5acac, 0x1e893c970fb58989, 0x5014a04428b41414, 0xa3e15b42dfbae1e1,
-	0x5816b04e2ca61616, 0xe83acdd274f73a3a, 0xb9696fd0d2066969, 0x2409482d12410909,
-	0xdd70a7ade0d77070, 0xe2b6d954716fb6b6, 0x67d0ceb7bd1ed0d0, 0x93ed3b7ec7d6eded,
-	0x17cc2edb85e2cccc, 0x15422a5784684242, 0x5a98b4c22d2c9898, 0xaaa4490e55eda4a4,
-	0xa0285d8850752828, 0x6d5cda31b8865c5c, 0xc7f8933fed6bf8f8, 0x228644a411c28686,
-}
-
-C7 := [256]u64 {
-	0x186018c07830d818, 0x238c2305af462623, 0xc63fc67ef991b8c6, 0xe887e8136fcdfbe8,
-	0x8726874ca113cb87, 0xb8dab8a9626d11b8, 0x0104010805020901, 0x4f214f426e9e0d4f,
-	0x36d836adee6c9b36, 0xa6a2a6590451ffa6, 0xd26fd2debdb90cd2, 0xf5f3f5fb06f70ef5,
-	0x79f979ef80f29679, 0x6fa16f5fcede306f, 0x917e91fcef3f6d91, 0x525552aa07a4f852,
-	0x609d6027fdc04760, 0xbccabc89766535bc, 0x9b569baccd2b379b, 0x8e028e048c018a8e,
-	0xa3b6a371155bd2a3, 0x0c300c603c186c0c, 0x7bf17bff8af6847b, 0x35d435b5e16a8035,
-	0x1d741de8693af51d, 0xe0a7e05347ddb3e0, 0xd77bd7f6acb321d7, 0xc22fc25eed999cc2,
-	0x2eb82e6d965c432e, 0x4b314b627a96294b, 0xfedffea321e15dfe, 0x5741578216aed557,
-	0x155415a8412abd15, 0x77c1779fb6eee877, 0x37dc37a5eb6e9237, 0xe5b3e57b56d79ee5,
-	0x9f469f8cd923139f, 0xf0e7f0d317fd23f0, 0x4a354a6a7f94204a, 0xda4fda9e95a944da,
-	0x587d58fa25b0a258, 0xc903c906ca8fcfc9, 0x29a429558d527c29, 0x0a280a5022145a0a,
-	0xb1feb1e14f7f50b1, 0xa0baa0691a5dc9a0, 0x6bb16b7fdad6146b, 0x852e855cab17d985,
-	0xbdcebd8173673cbd, 0x5d695dd234ba8f5d, 0x1040108050209010, 0xf4f7f4f303f507f4,
-	0xcb0bcb16c08bddcb, 0x3ef83eedc67cd33e, 0x05140528110a2d05, 0x6781671fe6ce7867,
-	0xe4b7e47353d597e4, 0x279c2725bb4e0227, 0x4119413258827341, 0x8b168b2c9d0ba78b,
-	0xa7a6a7510153f6a7, 0x7de97dcf94fab27d, 0x956e95dcfb374995, 0xd847d88e9fad56d8,
-	0xfbcbfb8b30eb70fb, 0xee9fee2371c1cdee, 0x7ced7cc791f8bb7c, 0x66856617e3cc7166,
-	0xdd53dda68ea77bdd, 0x175c17b84b2eaf17, 0x47014702468e4547, 0x9e429e84dc211a9e,
-	0xca0fca1ec589d4ca, 0x2db42d75995a582d, 0xbfc6bf9179632ebf, 0x071c07381b0e3f07,
-	0xad8ead012347acad, 0x5a755aea2fb4b05a, 0x8336836cb51bef83, 0x33cc3385ff66b633,
-	0x6391633ff2c65c63, 0x020802100a041202, 0xaa92aa39384993aa, 0x71d971afa8e2de71,
-	0xc807c80ecf8dc6c8, 0x196419c87d32d119, 0x4939497270923b49, 0xd943d9869aaf5fd9,
-	0xf2eff2c31df931f2, 0xe3abe34b48dba8e3, 0x5b715be22ab6b95b, 0x881a8834920dbc88,
-	0x9a529aa4c8293e9a, 0x2698262dbe4c0b26, 0x32c8328dfa64bf32, 0xb0fab0e94a7d59b0,
-	0xe983e91b6acff2e9, 0x0f3c0f78331e770f, 0xd573d5e6a6b733d5, 0x803a8074ba1df480,
-	0xbec2be997c6127be, 0xcd13cd26de87ebcd, 0x34d034bde4688934, 0x483d487a75903248,
-	0xffdbffab24e354ff, 0x7af57af78ff48d7a, 0x907a90f4ea3d6490, 0x5f615fc23ebe9d5f,
-	0x2080201da0403d20, 0x68bd6867d5d00f68, 0x1a681ad07234ca1a, 0xae82ae192c41b7ae,
-	0xb4eab4c95e757db4, 0x544d549a19a8ce54, 0x937693ece53b7f93, 0x2288220daa442f22,
-	0x648d6407e9c86364, 0xf1e3f1db12ff2af1, 0x73d173bfa2e6cc73, 0x124812905a248212,
-	0x401d403a5d807a40, 0x0820084028104808, 0xc32bc356e89b95c3, 0xec97ec337bc5dfec,
-	0xdb4bdb9690ab4ddb, 0xa1bea1611f5fc0a1, 0x8d0e8d1c8307918d, 0x3df43df5c97ac83d,
-	0x976697ccf1335b97, 0x0000000000000000, 0xcf1bcf36d483f9cf, 0x2bac2b4587566e2b,
-	0x76c57697b3ece176, 0x82328264b019e682, 0xd67fd6fea9b128d6, 0x1b6c1bd87736c31b,
-	0xb5eeb5c15b7774b5, 0xaf86af112943beaf, 0x6ab56a77dfd41d6a, 0x505d50ba0da0ea50,
-	0x450945124c8a5745, 0xf3ebf3cb18fb38f3, 0x30c0309df060ad30, 0xef9bef2b74c3c4ef,
-	0x3ffc3fe5c37eda3f, 0x554955921caac755, 0xa2b2a2791059dba2, 0xea8fea0365c9e9ea,
-	0x6589650fecca6a65, 0xbad2bab9686903ba, 0x2fbc2f65935e4a2f, 0xc027c04ee79d8ec0,
-	0xde5fdebe81a160de, 0x1c701ce06c38fc1c, 0xfdd3fdbb2ee746fd, 0x4d294d52649a1f4d,
-	0x927292e4e0397692, 0x75c9758fbceafa75, 0x061806301e0c3606, 0x8a128a249809ae8a,
-	0xb2f2b2f940794bb2, 0xe6bfe66359d185e6, 0x0e380e70361c7e0e, 0x1f7c1ff8633ee71f,
-	0x62956237f7c45562, 0xd477d4eea3b53ad4, 0xa89aa829324d81a8, 0x966296c4f4315296,
-	0xf9c3f99b3aef62f9, 0xc533c566f697a3c5, 0x25942535b14a1025, 0x597959f220b2ab59,
-	0x842a8454ae15d084, 0x72d572b7a7e4c572, 0x39e439d5dd72ec39, 0x4c2d4c5a6198164c,
-	0x5e655eca3bbc945e, 0x78fd78e785f09f78, 0x38e038ddd870e538, 0x8c0a8c148605988c,
-	0xd163d1c6b2bf17d1, 0xa5aea5410b57e4a5, 0xe2afe2434dd9a1e2, 0x6199612ff8c24e61,
-	0xb3f6b3f1457b42b3, 0x21842115a5423421, 0x9c4a9c94d625089c, 0x1e781ef0663cee1e,
-	0x4311432252866143, 0xc73bc776fc93b1c7, 0xfcd7fcb32be54ffc, 0x0410042014082404,
-	0x515951b208a2e351, 0x995e99bcc72f2599, 0x6da96d4fc4da226d, 0x0d340d68391a650d,
-	0xfacffa8335e979fa, 0xdf5bdfb684a369df, 0x7ee57ed79bfca97e, 0x2490243db4481924,
-	0x3bec3bc5d776fe3b, 0xab96ab313d4b9aab, 0xce1fce3ed181f0ce, 0x1144118855229911,
-	0x8f068f0c8903838f, 0x4e254e4a6b9c044e, 0xb7e6b7d1517366b7, 0xeb8beb0b60cbe0eb,
-	0x3cf03cfdcc78c13c, 0x813e817cbf1ffd81, 0x946a94d4fe354094, 0xf7fbf7eb0cf31cf7,
-	0xb9deb9a1676f18b9, 0x134c13985f268b13, 0x2cb02c7d9c58512c, 0xd36bd3d6b8bb05d3,
-	0xe7bbe76b5cd38ce7, 0x6ea56e57cbdc396e, 0xc437c46ef395aac4, 0x030c03180f061b03,
-	0x5645568a13acdc56, 0x440d441a49885e44, 0x7fe17fdf9efea07f, 0xa99ea921374f88a9,
-	0x2aa82a4d8254672a, 0xbbd6bbb16d6b0abb, 0xc123c146e29f87c1, 0x535153a202a6f153,
-	0xdc57dcae8ba572dc, 0x0b2c0b582716530b, 0x9d4e9d9cd327019d, 0x6cad6c47c1d82b6c,
-	0x31c43195f562a431, 0x74cd7487b9e8f374, 0xf6fff6e309f115f6, 0x4605460a438c4c46,
-	0xac8aac092645a5ac, 0x891e893c970fb589, 0x145014a04428b414, 0xe1a3e15b42dfbae1,
-	0x165816b04e2ca616, 0x3ae83acdd274f73a, 0x69b9696fd0d20669, 0x092409482d124109,
-	0x70dd70a7ade0d770, 0xb6e2b6d954716fb6, 0xd067d0ceb7bd1ed0, 0xed93ed3b7ec7d6ed,
-	0xcc17cc2edb85e2cc, 0x4215422a57846842, 0x985a98b4c22d2c98, 0xa4aaa4490e55eda4,
-	0x28a0285d88507528, 0x5c6d5cda31b8865c, 0xf8c7f8933fed6bf8, 0x86228644a411c286,
-}
-
-RC := [ROUNDS + 1]u64 {
-	0x0000000000000000,
-	0x1823c6e887b8014f,
-	0x36a6d2f5796f9152,
-	0x60bc9b8ea30c7b35,
-	0x1de0d7c22e4bfe57,
-	0x157737e59ff04ada,
-	0x58c9290ab1a06b85,
-	0xbd5d10f4cb3e0567,
-	0xe427418ba77d95d8,
-	0xfbee7c66dd17479e,
-	0xca2dbf07ad5a8333,
-}
-
-transform :: proc (ctx: ^Whirlpool_Context) {
-	K, block, state, L: [8]u64
-
-	for i := 0; i < 8; i += 1 {block[i] = util.U64_BE(ctx.buffer[8 * i:])}
-
-	for i := 0; i < 8; i += 1 {
-		K[i] = ctx.hash[i]
-		state[i] = block[i] ~ K[i]
-	}
-
-	for r := 1; r <= ROUNDS; r += 1 {
-		for i := 0; i < 8; i += 1 {
-			L[i] = C0[byte(K[i % 8] >> 56)] ~
-				C1[byte(K[(i + 7) % 8] >> 48)] ~
-				C2[byte(K[(i + 6) % 8] >> 40)] ~
-				C3[byte(K[(i + 5) % 8] >> 32)] ~
-				C4[byte(K[(i + 4) % 8] >> 24)] ~
-				C5[byte(K[(i + 3) % 8] >> 16)] ~
-				C6[byte(K[(i + 2) % 8] >> 8)] ~
-				C7[byte(K[(i + 1) % 8])]
-		}
-		L[0] ~= RC[r]
-
-		for i := 0; i < 8; i += 1 {K[i] = L[i]}
-
-		for i := 0; i < 8; i += 1 {
-			L[i] = C0[byte(state[i % 8] >> 56)] ~
-				C1[byte(state[(i + 7) % 8] >> 48)] ~
-				C2[byte(state[(i + 6) % 8] >> 40)] ~
-				C3[byte(state[(i + 5) % 8] >> 32)] ~
-				C4[byte(state[(i + 4) % 8] >> 24)] ~
-				C5[byte(state[(i + 3) % 8] >> 16)] ~
-				C6[byte(state[(i + 2) % 8] >> 8)] ~
-				C7[byte(state[(i + 1) % 8])] ~
-				K[i % 8]
-		}
-		for i := 0; i < 8; i += 1 {state[i] = L[i]}
-	}
-	for i := 0; i < 8; i += 1 {ctx.hash[i] ~= state[i] ~ block[i]}
-}
@@ -6,8 +6,10 @@ import "core:mem"
 SCALAR_SIZE :: 32
 POINT_SIZE :: 32

+@(private)
 _BASE_POINT: [32]byte = {9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}

+@(private)
 _scalar_bit :: #force_inline proc "contextless" (s: ^[32]byte, i: int) -> u8 {
 	if i < 0 {
 		return 0
@@ -15,6 +17,7 @@ _scalar_bit :: #force_inline proc "contextless" (s: ^[32]byte, i: int) -> u8 {
 	return (s[i>>3] >> uint(i&7)) & 1
 }

+@(private)
 _scalarmult :: proc (out, scalar, point: ^[32]byte) {
 	// Montgomery pseduo-multiplication taken from Monocypher.

@@ -0,0 +1,221 @@
+package debug_pe
+
+PE_SIGNATURE_OFFSET_INDEX_POS :: 0x3c
+PE_SIGNATURE :: u32le(0x0000_4550) // "PE\x00\x00"
+PE_SIGNATURE_STRING :: "PE\x00\x00"
+
+OPTIONAL_HEADER_MAGIC :: enum u16le {
+	PE32      = 0x010b,
+	PE32_PLUS = 0x020b,
+}
+
+Optional_Header_Base :: struct #packed {
+	magic:                          OPTIONAL_HEADER_MAGIC,
+	major_linker_version:           u8,
+	minor_linker_version:           u8,
+	size_of_code:                   u32le,
+	size_of_initialized_data:       u32le,
+	size_of_uninitialized_data:     u32le,
+	address_of_entry_point:         u32le,
+	base_of_code:                   u32le,
+}
+
+File_Header :: struct #packed {
+	machine:                 IMAGE_FILE_MACHINE,
+	number_of_sections:      u16le,
+	time_date_stamp:         u32le,
+	pointer_to_symbol_table: u32le,
+	number_of_symbols:       u32le,
+	size_of_optional_header: u16le,
+	characteristics:         IMAGE_FILE_CHARACTERISTICS,
+}
+
+Data_Directory :: struct #packed {
+	virtual_address: u32le,
+	size:            u32le,
+}
+
+Optional_Header32 :: struct #packed {
+	using base: Optional_Header_Base,
+	base_of_data:                   u32le,
+	image_base:                     u32le,
+	section_alignment:              u32le,
+	file_alignment:                 u32le,
+	major_operating_system_version: u16le,
+	minor_operating_system_version: u16le,
+	major_image_version:            u16le,
+	minor_image_version:            u16le,
+	major_subsystem_version:        u16le,
+	minor_subsystem_version:        u16le,
+	win32_version_value:            u32le,
+	size_of_image:                  u32le,
+	size_of_headers:                u32le,
+	check_sum:                      u32le,
+	subsystem:                      IMAGE_SUBSYSTEM,
+	dll_characteristics:            IMAGE_DLLCHARACTERISTICS,
+	size_of_stack_reserve:          u32le,
+	size_of_stack_commit:           u32le,
+	size_of_heap_reserve:           u32le,
+	size_of_heap_commit:            u32le,
+	loader_flags:                   u32le,
+	number_of_rva_and_sizes:        u32le,
+	data_directory:                 [16]Data_Directory,
+}
+
+Optional_Header64 :: struct #packed {
+	using base: Optional_Header_Base,
+	image_base:                     u64le,
+	section_alignment:              u32le,
+	file_alignment:                 u32le,
+	major_operating_system_version: u16le,
+	minor_operating_system_version: u16le,
+	major_image_version:            u16le,
+	minor_image_version:            u16le,
+	major_subsystem_version:        u16le,
+	minor_subsystem_version:        u16le,
+	win32_version_value:            u32le,
+	size_of_image:                  u32le,
+	size_of_headers:                u32le,
+	check_sum:                      u32le,
+	subsystem:                      IMAGE_SUBSYSTEM,
+	dll_characteristics:            IMAGE_DLLCHARACTERISTICS,
+	size_of_stack_reserve:          u64le,
+	size_of_stack_commit:           u64le,
+	size_of_heap_reserve:           u64le,
+	size_of_heap_commit:            u64le,
+	loader_flags:                   u32le,
+	number_of_rva_and_sizes:        u32le,
+	data_directory:                 [16]Data_Directory,
+}
+
+// .debug section
+Debug_Directory_Entry :: struct {
+	characteristics:     u32le,
+	time_date_stamp:     u32le,
+	major_version:       u16le,
+	minor_version:       u16le,
+	type:                IMAGE_DEBUG_TYPE,
+	size_of_data:        u32le,
+	address_of_raw_data: u32le,
+	pointer_to_raw_data: u32le,
+}
+
+
+IMAGE_FILE_MACHINE :: enum u16le {
+	UNKNOWN     = 0x0,
+	AM33        = 0x1d3,
+	AMD64       = 0x8664,
+	ARM         = 0x1c0,
+	ARMNT       = 0x1c4,
+	ARM64       = 0xaa64,
+	EBC         = 0xebc,
+	I386        = 0x14c,
+	IA64        = 0x200,
+	LOONGARCH32 = 0x6232,
+	LOONGARCH64 = 0x6264,
+	M32R        = 0x9041,
+	MIPS16      = 0x266,
+	MIPSFPU     = 0x366,
+	MIPSFPU16   = 0x466,
+	POWERPC     = 0x1f0,
+	POWERPCFP   = 0x1f1,
+	R4000       = 0x166,
+	SH3         = 0x1a2,
+	SH3DSP      = 0x1a3,
+	SH4         = 0x1a6,
+	SH5         = 0x1a8,
+	THUMB       = 0x1c2,
+	WCEMIPSV2   = 0x169,
+}
+
+// IMAGE_DIRECTORY_ENTRY constants
+IMAGE_DIRECTORY_ENTRY :: enum u8 {
+	EXPORT         = 0,
+	IMPORT         = 1,
+	RESOURCE       = 2,
+	EXCEPTION      = 3,
+	SECURITY       = 4,
+	BASERELOC      = 5,
+	DEBUG          = 6,
+	ARCHITECTURE   = 7, // reserved
+	GLOBALPTR      = 8,
+	TLS            = 9,
+	LOAD_CONFIG    = 10,
+	BOUND_IMPORT   = 11,
+	IAT            = 12,
+	DELAY_IMPORT   = 13,
+	COM_DESCRIPTOR = 14, // DLR Runtime headers
+	_RESERVED      = 15,
+}
+#assert(len(IMAGE_DIRECTORY_ENTRY) == 16)
+
+
+IMAGE_FILE_CHARACTERISTICS :: distinct bit_set[IMAGE_FILE_CHARACTERISTIC; u16le]
+IMAGE_FILE_CHARACTERISTIC :: enum u16le {
+	RELOCS_STRIPPED         = 0,
+	EXECUTABLE_IMAGE        = 1,
+	LINE_NUMS_STRIPPED      = 2,
+	LOCAL_SYMS_STRIPPED     = 3,
+	AGGRESIVE_WS_TRIM       = 4,
+	LARGE_ADDRESS_AWARE     = 5,
+
+	BYTES_REVERSED_LO       = 7,
+	MACHINE_32BIT           = 8, // IMAGE_FILE_32BIT_MACHINE  originally
+	DEBUG_STRIPPED          = 9,
+	REMOVABLE_RUN_FROM_SWAP = 10,
+	NET_RUN_FROM_SWAP       = 11,
+	SYSTEM                  = 12,
+	DLL                     = 13,
+	UP_SYSTEM_ONLY          = 14,
+	BYTES_REVERSED_HI       = 15,
+}
+
+IMAGE_SUBSYSTEM :: enum u16le {
+	UNKNOWN                  = 0,
+	NATIVE                   = 1,
+	WINDOWS_GUI              = 2,
+	WINDOWS_CUI              = 3,
+	OS2_CUI                  = 5,
+	POSIX_CUI                = 7,
+	NATIVE_WINDOWS           = 8,
+	WINDOWS_CE_GUI           = 9,
+	EFI_APPLICATION          = 10,
+	EFI_BOOT_SERVICE_DRIVER  = 11,
+	EFI_RUNTIME_DRIVER       = 12,
+	EFI_ROM                  = 13,
+	XBOX                     = 14,
+	WINDOWS_BOOT_APPLICATION = 16,
+}
+
+IMAGE_DLLCHARACTERISTICS :: distinct bit_set[IMAGE_DLLCHARACTERISTIC; u16le]
+IMAGE_DLLCHARACTERISTIC :: enum u16le {
+	HIGH_ENTROPY_VA       = 5,
+	DYNAMIC_BASE          = 6,
+	FORCE_INTEGRITY       = 7,
+	NX_COMPAT             = 8,
+	NO_ISOLATION          = 9,
+	NO_SEH                = 10,
+	NO_BIND               = 11,
+	APPCONTAINER          = 12,
+	WDM_DRIVER            = 13,
+	GUARD_CF              = 14,
+	TERMINAL_SERVER_AWARE = 15,
+}
+
+IMAGE_DEBUG_TYPE :: enum u32le {
+	UNKNOWN               = 0,  // An unknown value that is ignored by all tools.
+	COFF                  = 1,  // The COFF debug information (line numbers, symbol table, and string table). This type of debug information is also pointed to by fields in the file headers.
+	CODEVIEW              = 2,  // The Visual C++ debug information.
+	FPO                   = 3,  // The frame pointer omission (FPO) information. This information tells the debugger how to interpret nonstandard stack frames, which use the EBP register for a purpose other than as a frame pointer.
+	MISC                  = 4,  // The location of DBG file.
+	EXCEPTION             = 5,  // A copy of .pdata section.
+	FIXUP                 = 6,  // Reserved.
+	OMAP_TO_SRC           = 7,  // The mapping from an RVA in image to an RVA in source image.
+	OMAP_FROM_SRC         = 8,  // The mapping from an RVA in source image to an RVA in image.
+	BORLAND               = 9,  // Reserved for Borland.
+	RESERVED10            = 10, // Reserved.
+	CLSID                 = 11, // Reserved.
+	REPRO                 = 16, // PE determinism or reproducibility.
+	EX_DLLCHARACTERISTICS = 20, // Extended DLL characteristics bits.
+}
+
@@ -0,0 +1,128 @@
+package debug_pe
+
+Section_Header32 :: struct {
+	name:                    [8]u8,
+	virtual_size:            u32le,
+	virtual_address:         u32le,
+	size_of_raw_data:        u32le,
+	pointer_to_raw_data:     u32le,
+	pointer_to_relocations:  u32le,
+	pointer_to_line_numbers: u32le,
+	number_of_relocations:   u16le,
+	number_of_line_numbers:  u16le,
+	characteristics:         IMAGE_SCN_CHARACTERISTICS,
+}
+
+Reloc :: struct {
+	virtual_address:    u32le,
+	symbol_table_index: u32le,
+	type:               IMAGE_REL,
+}
+
+IMAGE_SCN_CHARACTERISTICS :: enum u32le {
+	TYPE_NO_PAD            = 0x00000008, // The section should not be padded to the next boundary. This flag is obsolete and is replaced by IMAGE_SCN_ALIGN_1BYTES. This is valid only for object files. = 0x00000010, // Reserved for future use.
+	CNT_CODE               = 0x00000020, // The section contains executable code.
+	CNT_INITIALIZED_DATA   = 0x00000040, // The section contains initialized data.
+	CNT_UNINITIALIZED_DATA = 0x00000080, // The section contains uninitialized data.
+	LNK_OTHER              = 0x00000100, // Reserved for future use.
+	LNK_INFO               = 0x00000200, // The section contains comments or other information. The .drectve section has this type. This is valid for object files only. = 0x00000400, // Reserved for future use.
+	LNK_REMOVE             = 0x00000800, // The section will not become part of the image. This is valid only for object files.
+	LNK_COMDAT             = 0x00001000, // The section contains COMDAT data. For more information, see COMDAT Sections (Object Only). This is valid only for object files.
+	GPREL                  = 0x00008000, // The section contains data referenced through the global pointer (GP).
+	MEM_PURGEABLE          = 0x00020000, // Reserved for future use.
+	MEM_16BIT              = 0x00020000, // Reserved for future use.
+	MEM_LOCKED             = 0x00040000, // Reserved for future use.
+	MEM_PRELOAD            = 0x00080000, // Reserved for future use.
+	ALIGN_1BYTES           = 0x00100000, // Align data on a 1-byte boundary. Valid only for object files.
+	ALIGN_2BYTES           = 0x00200000, // Align data on a 2-byte boundary. Valid only for object files.
+	ALIGN_4BYTES           = 0x00300000, // Align data on a 4-byte boundary. Valid only for object files.
+	ALIGN_8BYTES           = 0x00400000, // Align data on an 8-byte boundary. Valid only for object files.
+	ALIGN_16BYTES          = 0x00500000, // Align data on a 16-byte boundary. Valid only for object files.
+	ALIGN_32BYTES          = 0x00600000, // Align data on a 32-byte boundary. Valid only for object files.
+	ALIGN_64BYTES          = 0x00700000, // Align data on a 64-byte boundary. Valid only for object files.
+	ALIGN_128BYTES         = 0x00800000, // Align data on a 128-byte boundary. Valid only for object files.
+	ALIGN_256BYTES         = 0x00900000, // Align data on a 256-byte boundary. Valid only for object files.
+	ALIGN_512BYTES         = 0x00A00000, // Align data on a 512-byte boundary. Valid only for object files.
+	ALIGN_1024BYTES        = 0x00B00000, // Align data on a 1024-byte boundary. Valid only for object files.
+	ALIGN_2048BYTES        = 0x00C00000, // Align data on a 2048-byte boundary. Valid only for object files.
+	ALIGN_4096BYTES        = 0x00D00000, // Align data on a 4096-byte boundary. Valid only for object files.
+	ALIGN_8192BYTES        = 0x00E00000, // Align data on an 8192-byte boundary. Valid only for object files.
+	LNK_NRELOC_OVFL        = 0x01000000, // The section contains extended relocations.
+	MEM_DISCARDABLE        = 0x02000000, // The section can be discarded as needed.
+	MEM_NOT_CACHED         = 0x04000000, // The section cannot be cached.
+	MEM_NOT_PAGED          = 0x08000000, // The section is not pageable.
+	MEM_SHARED             = 0x10000000, // The section can be shared in memory.
+	MEM_EXECUTE            = 0x20000000, // The section can be executed as code.
+	MEM_READ               = 0x40000000, // The section can be read.
+	MEM_WRITE              = 0x80000000, // The section can be written to.
+}
+
+
+IMAGE_REL :: enum u16le {
+	I386_ABSOLUTE         = 0x0000,
+	I386_DIR16            = 0x0001,
+	I386_REL16            = 0x0002,
+	I386_DIR32            = 0x0006,
+	I386_DIR32NB          = 0x0007,
+	I386_SEG12            = 0x0009,
+	I386_SECTION          = 0x000A,
+	I386_SECREL           = 0x000B,
+	I386_TOKEN            = 0x000C,
+	I386_SECREL7          = 0x000D,
+	I386_REL32            = 0x0014,
+
+	AMD64_ABSOLUTE        = 0x0000,
+	AMD64_ADDR64          = 0x0001,
+	AMD64_ADDR32          = 0x0002,
+	AMD64_ADDR32NB        = 0x0003,
+	AMD64_REL32           = 0x0004,
+	AMD64_REL32_1         = 0x0005,
+	AMD64_REL32_2         = 0x0006,
+	AMD64_REL32_3         = 0x0007,
+	AMD64_REL32_4         = 0x0008,
+	AMD64_REL32_5         = 0x0009,
+	AMD64_SECTION         = 0x000A,
+	AMD64_SECREL          = 0x000B,
+	AMD64_SECREL7         = 0x000C,
+	AMD64_TOKEN           = 0x000D,
+	AMD64_SREL32          = 0x000E,
+	AMD64_PAIR            = 0x000F,
+	AMD64_SSPAN32         = 0x0010,
+
+	ARM_ABSOLUTE          = 0x0000,
+	ARM_ADDR32            = 0x0001,
+	ARM_ADDR32NB          = 0x0002,
+	ARM_BRANCH24          = 0x0003,
+	ARM_BRANCH11          = 0x0004,
+	ARM_SECTION           = 0x000E,
+	ARM_SECREL            = 0x000F,
+	ARM_MOV32             = 0x0010,
+
+	THUMB_MOV32           = 0x0011,
+	THUMB_BRANCH20        = 0x0012,
+	THUMB_BRANCH24        = 0x0014,
+	THUMB_BLX23           = 0x0015,
+
+	ARM_PAIR              = 0x0016,
+
+	ARM64_ABSOLUTE        = 0x0000,
+	ARM64_ADDR32          = 0x0001,
+	ARM64_ADDR32NB        = 0x0002,
+	ARM64_BRANCH26        = 0x0003,
+	ARM64_PAGEBASE_REL21  = 0x0004,
+	ARM64_REL21           = 0x0005,
+	ARM64_PAGEOFFSET_12A  = 0x0006,
+	ARM64_PAGEOFFSET_12L  = 0x0007,
+	ARM64_SECREL          = 0x0008,
+	ARM64_SECREL_LOW12A   = 0x0009,
+	ARM64_SECREL_HIGH12A  = 0x000A,
+	ARM64_SECREL_LOW12L   = 0x000B,
+	ARM64_TOKEN           = 0x000C,
+	ARM64_SECTION         = 0x000D,
+	ARM64_ADDR64          = 0x000E,
+	ARM64_BRANCH19        = 0x000F,
+	ARM64_BRANCH14        = 0x0010,
+	ARM64_REL32           = 0x0011,
+}
+
+PE_CODE_VIEW_SIGNATURE_RSDS :: u32le(0x5344_5352)
@@ -0,0 +1,108 @@
+package debug_pe
+
+COFF_SYMBOL_SIZE :: 18
+
+COFF_Symbol :: struct {
+	name:                  [8]u8,
+	value:                 u32le,
+	section_number:        i16le,
+	type:                  IMAGE_SYM_TYPE,
+	storage_class:         IMAGE_SYM_CLASS,
+	number_of_aux_symbols: u8,
+}
+
+// COFF_Symbol_Aux_Format5 describes the expected form of an aux symbol
+// attached to a section definition symbol. The PE format defines a
+// number of different aux symbol formats: format 1 for function
+// definitions, format 2 for .be and .ef symbols, and so on. Format 5
+// holds extra info associated with a section definition, including
+// number of relocations + line numbers, as well as COMDAT info. See
+// https://docs.microsoft.com/en-us/windows/win32/debug/pe-format#auxiliary-format-5-section-definitions
+// for more on what's going on here.
+COFF_Symbol_Aux_Format5 :: struct {
+	size:             u32le,
+	num_relocs:       u16le,
+	num_line_numbers: u16le,
+	checksum:         u32le,
+	sec_num:          u16le,
+	selection:        IMAGE_COMDAT_SELECT,
+	_:                [3]u8, // padding
+}
+
+IMAGE_COMDAT_SELECT :: enum u8 {
+	NODUPLICATES = 1,
+	ANY          = 2,
+	SAME_SIZE    = 3,
+	EXACT_MATCH  = 4,
+	ASSOCIATIVE  = 5,
+	LARGEST      = 6,
+}
+
+
+// The symbol record is not yet assigned a section. A value of zero indicates
+// that a reference to an external symbol is defined elsewhere. A value of
+// non-zero is a common symbol with a size that is specified by the value.
+IMAGE_SYM_UNDEFINED              :: 0
+// The symbol has an absolute (non-relocatable) value and is not an address.
+IMAGE_SYM_ABSOLUTE               :: -1
+// The symbol provides general type or debugging information but does not
+// correspond to a section. Microsoft tools use this setting along
+// with .file records (storage class FILE).
+IMAGE_SYM_DEBUG                  :: -2
+
+IMAGE_SYM_TYPE :: enum u16le {
+	NULL   = 0,
+	VOID   = 1,
+	CHAR   = 2,
+	SHORT  = 3,
+	INT    = 4,
+	LONG   = 5,
+	FLOAT  = 6,
+	DOUBLE = 7,
+	STRUCT = 8,
+	UNION  = 9,
+	ENUM   = 10,
+	MOE    = 11,
+	BYTE   = 12,
+	WORD   = 13,
+	UINT   = 14,
+	DWORD  = 15,
+	PCODE  = 32768,
+
+	DTYPE_NULL     = 0,
+	DTYPE_POINTER  = 0x10,
+	DTYPE_FUNCTION = 0x20,
+	DTYPE_ARRAY    = 0x30,
+}
+
+IMAGE_SYM_CLASS :: enum u8 {
+	NULL             = 0,
+	AUTOMATIC        = 1,
+	EXTERNAL         = 2,
+	STATIC           = 3,
+	REGISTER         = 4,
+	EXTERNAL_DEF     = 5,
+	LABEL            = 6,
+	UNDEFINED_LABEL  = 7,
+	MEMBER_OF_STRUCT = 8,
+	ARGUMENT         = 9,
+	STRUCT_TAG       = 10,
+	MEMBER_OF_UNION  = 11,
+	UNION_TAG        = 12,
+	TYPE_DEFINITION  = 13,
+	UNDEFINED_STATIC = 14,
+	ENUM_TAG         = 15,
+	MEMBER_OF_ENUM   = 16,
+	REGISTER_PARAM   = 17,
+	BIT_FIELD        = 18,
+	FAR_EXTERNAL     = 68, // Not in PECOFF v8 spec
+	BLOCK            = 100,
+	FUNCTION         = 101,
+	END_OF_STRUCT    = 102,
+	FILE             = 103,
+	SECTION          = 104,
+	WEAK_EXTERNAL    = 105,
+	CLR_TOKEN        = 107,
+
+	END_OF_FUNCTION  = 255,
+}
@@ -0,0 +1,7 @@
+/*
+Package core:dynlib implements loading of shared libraries/DLLs and their symbols.
+
+The behaviour of dynamically loaded libraries is specific to the target platform of the program.
+For in depth detail on the underlying behaviour please refer to your target platform's documentation.
+*/
+package dynlib
@@ -1,15 +1,94 @@
 package dynlib

+/*
+A handle to a dynamically loaded library.
+*/
 Library :: distinct rawptr

-load_library :: proc(path: string, global_symbols := false) -> (Library, bool) {
+/*
+Loads a dynamic library from the filesystem. The paramater `global_symbols` makes the symbols in the loaded
+library available to resolve references in subsequently loaded libraries.
+
+The paramater `global_symbols` is only used for the platforms `linux`, `darwin`, `freebsd` and `openbsd`.
+On `windows` this paramater is ignored.
+
+The underlying behaviour is platform specific.  
+On `linux`, `darwin`, `freebsd` and `openbsd` refer to `dlopen`.  
+On `windows` refer to `LoadLibraryW`.
+
+**Implicit Allocators**  
+`context.temp_allocator`
+
+Example:
+	import "core:dynlib"
+	import "core:fmt"
+
+	load_my_library :: proc() {
+		LIBRARY_PATH :: "my_library.dll"
+		library, ok := dynlib.load_library(LIBRARY_PATH)
+		if ! ok {
+			return
+		}
+		fmt.println("The library %q was successfully loaded", LIBRARY_PATH)
+	}
+*/
+load_library :: proc(path: string, global_symbols := false) -> (library: Library, did_load: bool) {
 	return _load_library(path, global_symbols)
 }

-unload_library :: proc(library: Library) -> bool {
+/*
+Unloads a dynamic library.
+
+The underlying behaviour is platform specific.  
+On `linux`, `darwin`, `freebsd` and `openbsd` refer to `dlclose`.  
+On `windows` refer to `FreeLibrary`.
+
+Example:
+	import "core:dynlib"
+	import "core:fmt"
+
+	load_then_unload_my_library :: proc() {
+		LIBRARY_PATH :: "my_library.dll"
+		library, ok := dynlib.load_library(LIBRARY_PATH)
+		if ! ok {
+			return
+		}
+		did_unload := dynlib.unload_library(library)
+		if ! did_unload {
+			return
+		}
+		fmt.println("The library %q was successfully unloaded", LIBRARY_PATH)
+	}
+*/
+unload_library :: proc(library: Library) -> (did_unload: bool) {
 	return _unload_library(library)
 }

+/*
+Loads the address of a procedure/variable from a dynamic library.
+
+The underlying behaviour is platform specific.  
+On `linux`, `darwin`, `freebsd` and `openbsd` refer to `dlsym`.  
+On `windows` refer to `GetProcAddress`.
+
+**Implicit Allocators**  
+`context.temp_allocator`
+
+Example:
+	import "core:dynlib"
+	import "core:fmt"
+
+	find_a_in_my_library :: proc() {
+		LIBRARY_PATH :: "my_library.dll"
+		library, ok := dynlib.load_library(LIBRARY_PATH)
+		if ! ok {
+			return
+		}
+
+		a, found_a := dynlib.symbol_address(library, "a")
+		if found_a do fmt.printf("The symbol %q was found at the address %v", "a", a)
+	}
+*/
 symbol_address :: proc(library: Library, symbol: string) -> (ptr: rawptr, found: bool) #optional_ok {
 	return _symbol_address(library, symbol)
 }
@@ -0,0 +1,15 @@
+//+build js
+//+private
+package dynlib
+
+_load_library :: proc(path: string, global_symbols := false) -> (Library, bool) {
+	return nil, false
+}
+
+_unload_library :: proc(library: Library) -> bool {
+	return false
+}
+
+_symbol_address :: proc(library: Library, symbol: string) -> (ptr: rawptr, found: bool) {
+	return nil, false
+}
@@ -4,10 +4,12 @@ package dynlib

 import win32 "core:sys/windows"
 import "core:strings"
+import "core:runtime"

 _load_library :: proc(path: string, global_symbols := false) -> (Library, bool) {
 	// NOTE(bill): 'global_symbols' is here only for consistency with POSIX which has RTLD_GLOBAL

+	runtime.DEFAULT_TEMP_ALLOCATOR_TEMP_GUARD()
 	wide_path := win32.utf8_to_wstring(path, context.temp_allocator)
 	handle := cast(Library)win32.LoadLibraryW(wide_path)
 	return handle, handle != nil
@@ -19,6 +21,7 @@ _unload_library :: proc(library: Library) -> bool {
 }

 _symbol_address :: proc(library: Library, symbol: string) -> (ptr: rawptr, found: bool) {
+	runtime.DEFAULT_TEMP_ALLOCATOR_TEMP_GUARD()
 	c_str := strings.clone_to_cstring(symbol, context.temp_allocator)
 	ptr = win32.GetProcAddress(cast(win32.HMODULE)library, c_str)
 	found = ptr != nil
@@ -248,7 +248,7 @@ _read_record :: proc(r: ^Reader, dst: ^[dynamic]string, allocator := context.all
 					field_length += 1

 				case '\n', '\r':
-					if !is_quoted { break read_loop }
+					is_quoted or_break read_loop

 				case r.comma:
 					field_length = 0
@@ -42,7 +42,7 @@ write :: proc(w: ^Writer, record: []string) -> io.Error {
 				}
 			}
 		case:
-			if strings.contains_rune(field, w.comma) >= 0 {
+			if strings.contains_rune(field, w.comma) {
 				return true
 			}
 			if strings.contains_any(field, CHAR_SET) {
--- a/Show More
+++ b/Show More