]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: e14e842)
vulkan: Fix test-thread-safety crashes (#17024)
authorJeff Bolz <redacted>
Sat, 8 Nov 2025 07:39:45 +0000 (01:39 -0600)
committerGitHub <redacted>
Sat, 8 Nov 2025 07:39:45 +0000 (08:39 +0100)
commitd6fe40fa0040bf7207b2380d0f3628fb56d6a86f
tree752034b7177a14737e8383b888760bb37e0f01betree
parente14e842e87103ec2a004770cec95a3f94f861bdacommit | diff
vulkan: Fix test-thread-safety crashes (#17024)

The std::map pipeline_flash_attn_f32_f16 could be searched and inserted at the
same time, which needs to hold the lock. To be safe, hold the lock for all of
ggml_vk_load_shaders.
ggml/src/ggml-vulkan/ggml-vulkan.cpp diff | blob | history
Packaging of ggml-org/llama.cpp