47ac1796a9
use build string literal from base context
2024-12-06 11:53:16 -08:00
fc429449cf
fix out of bounds check
2024-12-06 11:53:16 -08:00
19fac9b1aa
changed symbol table parser to convert symbols as they are being read
2024-11-21 19:42:39 -08:00
a0dbe72820
moved common codeview code to the main layer
2024-11-21 15:58:32 -08:00
fbf180df2f
assert when merging associative chunks
2024-11-21 13:30:13 -08:00
c037d6ad75
WIP function padding
2024-11-21 01:05:33 -08:00
816d8e88bf
remove warning pragma around radsort
2024-11-18 16:26:42 -08:00
31f536334e
third party include fixups
2024-11-18 16:17:26 -08:00
f0a1deec62
merge third party folders
2024-11-18 16:16:15 -08:00
af57e321de
show full lib path for better debuggability
2024-11-18 15:38:00 -08:00
e44dd3aa07
instead of searching for manifest tool manually punt the job to CreateProcess
2024-11-18 15:38:00 -08:00
b99e10224f
simplified chunk size tracking
2024-11-15 16:14:00 -08:00
435f6c4ef6
append .lib to defaultlibs
2024-11-15 16:14:00 -08:00
b631d431fa
create section layout without copying chunk data and perform chunk data
...
copy in parallel given layout
2024-11-15 16:14:00 -08:00
b16d930d13
natvis for linker types
2024-11-15 16:14:00 -08:00
26be351354
eliminate per-build version numbers, just use base layer to define defaults; bump to 14
2024-11-15 15:36:12 -08:00
a8b1805749
updated prof markup usage sites
2024-11-12 13:18:17 -08:00
2bbf3ff987
switch to data size formatter
2024-11-12 10:36:45 -08:00
c6caec657d
moved cstring and find needle string ops to the main layer
2024-11-12 10:36:45 -08:00
3d7fc6fce5
moved memory and count string formatters to the main layer and
...
replaced str8_from_memory_size with more accurate data size units
2024-11-12 10:36:44 -08:00
0d0951bf08
squash some image-related steps into finalize image
2024-11-12 10:36:44 -08:00
3a190ae2c3
show an early warning for /DEBUG:FASTLINK
2024-11-12 10:36:44 -08:00
680ea2ffdc
added /RAD_IDLE_WORKERS
2024-11-08 16:09:56 -08:00
f7d7f6633f
expose file API to enable IO detouring
2024-11-08 16:09:56 -08:00
8e6a0ae4db
errors and warnings fixes
2024-11-08 11:43:05 -08:00
776c8daa27
updated build.sh to define git hash
2024-11-08 11:43:05 -08:00
69b43c76d4
default to llvm-mt on non-windows platforms
2024-11-08 11:43:05 -08:00
158dd742ad
eliminate entry point 'mode'/type; just gather argc/argv as part of command line parse
2024-11-08 10:33:28 -08:00
15d43d7c99
removed dependency on os_ext
2024-11-07 14:12:41 -08:00
7650a83d1e
further simplify linker/os_ext/os_core.c
2024-11-07 14:08:21 -08:00
daeb08e79f
moved os_folder_exists to the main layer and deleted win32 layer from linker folder
2024-11-07 13:55:15 -08:00
ef5287d2a9
switch to os_full_path_from_path
2024-11-07 13:41:47 -08:00
db04792073
moved getter for process start time w32
2024-11-07 13:33:37 -08:00
4091373517
moved operating system -> string to base
2024-11-07 11:54:25 -08:00
214be8c433
moved Guid struct and related helpers to base
2024-11-07 11:54:25 -08:00
f807dc4785
override entry point
2024-11-07 11:54:25 -08:00
0886963d93
promote section definition number to 32-bit when linking big obj
2024-11-07 11:54:25 -08:00
7bb88f480e
mask compute typo
2024-11-07 11:54:24 -08:00
fad817fb9e
apply communal variable rule for largest comdat only
2024-11-07 11:54:24 -08:00
1d42f1bd66
speed up export table build with hash table
2024-11-07 11:54:24 -08:00
f3a36ece13
rebase fixes
2024-10-30 15:56:55 -07:00
c5d198e3e9
build linker with branch name and full commit hash
2024-10-30 15:56:10 -07:00
eba7500f73
/rad_pdb_hash_type_names:lenient
...
Less aggressive optimization for type names. When lenient enabled, linker
replaces only mangled names and compiler generated lambda types.
2024-10-30 15:56:10 -07:00
339f43919c
remove bogus leaf align
2024-10-30 15:56:09 -07:00
f3fbf10754
replace type names with hashes
2024-10-30 15:56:09 -07:00
4a2e1dd8d0
switch to thread safe hash trie map
...
Initial implementation of the symbol table allowed us to push symbols
from one thread at a time and required a wonky lock-step and spread out
logic for replacing symbols. And because it was a externally chained
hash table we had to guesstimate optimal capacity for small and big targets.
We erred on the side of larger capacity to prevent degradation and
rehashing, however this meant small targets had to spent unreasonable
amount of time memory allocing slots. With hash trie we can dynamically
scale up to 4 ^ 32 items and atomically replace symbols when needed. Neat!
2024-10-30 15:43:53 -07:00
baead67a99
minor fixes
2024-10-30 15:42:39 -07:00
e085dae636
pass over atomic operations
...
- use cast to volatile trick for atomic loads on x64 windows
- removed unnecessary includes on windows
- upgraded clang and gcc code paths to use modern atomics
Removed WIN32_LEAN_AND_MEAN because we were already compiling without it
and there is code path in os_gfx_win32.c that relies on including full
windows.h
2024-10-30 15:42:39 -07:00
8848080729
set explicitly heap and stack reserve and size
2024-10-30 15:42:39 -07:00
e1166abc63
typo in str8 compar name
2024-10-30 15:42:39 -07:00
Nikita Smith
Ryan Fleury