fix: implement ELF GOT patching for CUDA hooks on Linux/WSL2#27
Open
WASasquatch wants to merge 1 commit into
Open
fix: implement ELF GOT patching for CUDA hooks on Linux/WSL2#27WASasquatch wants to merge 1 commit into
WASasquatch wants to merge 1 commit into
Conversation
On Linux, libcudart resolves driver API symbols via dlsym at runtime, so the no-op stubs left GOT entries unpatched and VRAM tracking broken. This commit: - Adds src-posix/cuda-hook.c with ELF GOT/PLT patching that hooks both CUDA runtime API (cudaMalloc/cudaFree/cudaMallocAsync/cudaFreeAsync) and driver API (cuMemAlloc_v2/cuMemFree_v2/cuMemAllocAsync/cuMemFreeAsync) - Runtime hooks call through real runtime API preserving CUDA memory pools - Self-contained allocation tracking via thread-safe hash table - Resolves runtime symbols via RTLD_DEFAULT (handles versioned sonames) - Leaves GOT pages R+W after patching for lazy resolution compatibility - Adds -ldl to build scripts for dlopen/dlsym/dl_iterate_phdr - Removes dead symbol-interposition overrides from pyt-cu-plug-alloc-async.c Tested on RTX 4090 under WSL2 (Ubuntu 24.04) with PyTorch 2.10+cu128.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
On Linux, libcudart resolves driver API symbols via dlsym at runtime, so the no-op stubs left GOT entries unpatched and VRAM tracking broken.
This commit:
Tested on RTX 4090 under WSL2 (Ubuntu 24.04) with PyTorch 2.10+cu128.
Contribution Agreement