From 166b84cf3004fe8b979e46b0851f2e97c9107e99 Mon Sep 17 00:00:00 2001
From: Ryan Fleury <ryan.j.fleury@gmail.com>
Date: Tue, 7 Oct 2025 16:24:54 -0700
Subject: [PATCH] eliminate content-layer-specific u128 hash

---
 src/content/content.c    | 17 +----------------
 src/content/content.h    |  1 -
 src/ctrl/ctrl_core.c     | 18 ++++++++++++------
 src/tester/tester_main.c |  4 ++--
 4 files changed, 15 insertions(+), 25 deletions(-)

diff --git a/src/content/content.c b/src/content/content.c
index 8f353678..04a3b6bb 100644
--- a/src/content/content.c
+++ b/src/content/content.c
@@ -7,21 +7,6 @@
 ////////////////////////////////
 //~ rjf: Basic Helpers
 
-#if !defined(XXH_IMPLEMENTATION)
-# define XXH_IMPLEMENTATION
-# define XXH_STATIC_LINKING_ONLY
-# include "third_party/xxHash/xxhash.h"
-#endif
-
-internal U128
-c_hash_from_data(String8 data)
-{
-  U128 u128 = {0};
-  XXH128_hash_t hash = XXH3_128bits(data.str, data.size);
-  MemoryCopy(&u128, &hash, sizeof(u128));
-  return u128;
-}
-
 internal C_ID
 c_id_make(U64 u64_0, U64 u64_1)
 {
@@ -181,7 +166,7 @@ c_submit_data(C_Key key, Arena **data_arena, String8 data)
   C_Stripe *key_stripe = &c_shared->key_stripes[key_stripe_idx];
   
   //- rjf: hash data, unpack hash
-  U128 hash = c_hash_from_data(data);
+  U128 hash = u128_hash_from_str8(data);
   U64 slot_idx = hash.u64[1]%c_shared->blob_slots_count;
   U64 stripe_idx = slot_idx%c_shared->blob_stripes_count;
   C_BlobSlot *slot = &c_shared->blob_slots[slot_idx];
diff --git a/src/content/content.h b/src/content/content.h
index e314f0f6..90f44cda 100644
--- a/src/content/content.h
+++ b/src/content/content.h
@@ -197,7 +197,6 @@ global C_Shared *c_shared = 0;
 ////////////////////////////////
 //~ rjf: Basic Helpers
 
-internal U128 c_hash_from_data(String8 data);
 internal C_ID c_id_make(U64 u64_0, U64 u64_1);
 internal B32 c_id_match(C_ID a, C_ID b);
 internal C_Key c_key_make(C_Root root, C_ID id);
diff --git a/src/ctrl/ctrl_core.c b/src/ctrl/ctrl_core.c
index 1b0810a8..5f4cdedf 100644
--- a/src/ctrl/ctrl_core.c
+++ b/src/ctrl/ctrl_core.c
@@ -6009,15 +6009,15 @@ ctrl_memory_artifact_create(String8 key, B32 *cancel_signal, B32 *retry_out)
     B32 pre_run_state = ins_atomic_u64_eval(&ctrl_state->ctrl_thread_run_state);
     if(range_size != 0)
     {
+      // rjf: set up arena
       U64 page_size = os_get_system_info()->page_size; // TODO(rjf): @page_size_from_process
       U64 arena_size = AlignPow2(range_size + ARENA_HEADER_SIZE, page_size);
       range_arena = arena_alloc(.reserve_size = range_size+ARENA_HEADER_SIZE, .commit_size = range_size+ARENA_HEADER_SIZE);
-      if(range_arena == 0)
-      {
-        range_size = 0;
-      }
-      else
+      
+      // rjf: if we got an arena -> push buffer & read
+      if(range_arena != 0)
       {
+        // rjf: read as much as possible
         range_base = push_array_no_zero(range_arena, U8, range_size);
         U64 bytes_read = 0;
         U64 retry_count = 0;
@@ -6042,16 +6042,22 @@ ctrl_memory_artifact_create(String8 key, B32 *cancel_signal, B32 *retry_out)
             break;
           }
         }
+        
+        // rjf: if we read nothing, release arena
         if(bytes_read == 0)
         {
           arena_release(range_arena);
           range_base = 0;
           range_arena = 0;
         }
+        
+        // rjf: if we only got a partial read, zero the rest
         else if(bytes_read < range_size)
         {
           MemoryZero((U8 *)range_base + bytes_read, range_size-bytes_read);
         }
+        
+        // rjf: determine final size; zero terminate if needed; pop any unneeded bytes if zero-terminating
         zero_terminated_size = bytes_read;
         if(zero_terminated && range_base != 0)
         {
@@ -6091,7 +6097,7 @@ ctrl_memory_artifact_create(String8 key, B32 *cancel_signal, B32 *retry_out)
     U128 hash = {0};
     if((zero_terminated_size > 0 || !key_has_history) && range_size != 0 && pre_read_mem_gen == post_read_mem_gen)
     {
-      hash = c_submit_data(content_key, &range_arena, str8((U8*)range_base, zero_terminated_size));
+      hash = c_submit_data(content_key, &range_arena, str8((U8 *)range_base, zero_terminated_size));
     }
     else if(range_arena != 0)
     {
diff --git a/src/tester/tester_main.c b/src/tester/tester_main.c
index 4ee85062..9d34aad4 100644
--- a/src/tester/tester_main.c
+++ b/src/tester/tester_main.c
@@ -148,7 +148,7 @@ for(Test *test = test_##name_identifier; test != 0; test = 0)
           Temp scratch = scratch_begin(0, 0);
           String8 path = n->string;
           String8 data = os_data_from_file_path(scratch.arena, path);
-          rdi_hashes[idx] = c_hash_from_data(data);
+          rdi_hashes[idx] = u128_hash_from_str8(data);
           rdi_paths_array[idx] = path;
           scratch_end(scratch);
         }
@@ -160,7 +160,7 @@ for(Test *test = test_##name_identifier; test != 0; test = 0)
           Temp scratch = scratch_begin(0, 0);
           String8 path = n->string;
           String8 data = os_data_from_file_path(scratch.arena, path);
-          dump_hashes[idx] = c_hash_from_data(data);
+          dump_hashes[idx] = u128_hash_from_str8(data);
           dump_paths_array[idx] = path;
           scratch_end(scratch);
         }