[libc-commits] [libc] [libc] Efficiently implement 'realloc' for AMDGPU devices (PR #145960)

[Joseph Huber via libc-commits]

================
@@ -550,5 +551,26 @@ void deallocate(void *ptr) {
   release_slab(slab);
 }
 
+void *reallocate(void *ptr, uint64_t size) {
----------------
jhuber6 wrote:

It made it clearer to me because I intentionally truncate it to `uint32_t` internally once we enter the chunk-based allocator to save the register.

https://github.com/llvm/llvm-project/pull/145960