Skip to content

GitLab

"composable_kernel/include/utility/array.hpp" did not exist on "eafdabba771fe3a41843b86c85f6e574b2d176e1"
Unverified Commit 76fcdc60 authored by carlushuang's avatar carlushuang Committed by GitHub
Browse files

fix a bug with non-dword-aligned offset when OOB, in case crash (#616) 


Co-authored-by: default avatarzjing14 <zhangjing14@gmail.com>
parent 0ccecc7c
Hide whitespace changes
Inline Side-by-side
Showing with 4 additions and 4 deletions
include/ck/utility/amd_buffer_addressing.hpp
View file @ 76fcdc60
......@@ -1030,7 +1030,7 @@ amd_buffer_load_invalid_element_return_zero(const T* p_src_wave,
constexpr index_t vector_size = scalar_type<vector_t>::vector_size;
#if CK_EXPERIMENTAL_USE_BUFFER_LOAD_OOB_CHECK_OFFSET_TRICK
uint32_t src_addr_shift = src_thread_element_valid ? 0 : 0x7fffffff;
uint32_t src_addr_shift = src_thread_element_valid ? 0 : 0x80000000;
return amd_buffer_load_impl<scalar_t, vector_size>(
src_wave_buffer_resource, src_addr_shift + src_thread_addr_offset, 0);
......@@ -1091,7 +1091,7 @@ __device__ void amd_buffer_store(const typename vector_type_maker<T, N>::type::t
constexpr index_t vector_size = scalar_type<vector_t>::vector_size;
#if CK_EXPERIMENTAL_USE_BUFFER_STORE_OOB_CHECK_OFFSET_TRICK
uint32_t dst_addr_shift = dst_thread_element_valid ? 0 : 0x7fffffff;
uint32_t dst_addr_shift = dst_thread_element_valid ? 0 : 0x80000000;
amd_buffer_store_impl<scalar_t, vector_size>(
src_thread_data, dst_wave_buffer_resource, dst_addr_shift + dst_thread_addr_offset, 0);
......@@ -1126,7 +1126,7 @@ amd_buffer_atomic_add(const typename vector_type_maker<T, N>::type::type src_thr
constexpr index_t vector_size = scalar_type<vector_t>::vector_size;
#if CK_EXPERIMENTAL_USE_BUFFER_ATOMIC_ADD_OOB_CHECK_OFFSET_TRICK
uint32_t dst_addr_shift = dst_thread_element_valid ? 0 : 0x7fffffff;
uint32_t dst_addr_shift = dst_thread_element_valid ? 0 : 0x80000000;
amd_buffer_atomic_add_impl<scalar_t, vector_size>(
src_thread_data, dst_wave_buffer_resource, dst_addr_shift + dst_thread_addr_offset, 0);
......@@ -1161,7 +1161,7 @@ amd_buffer_atomic_max(const typename vector_type_maker<T, N>::type::type src_thr
constexpr index_t vector_size = scalar_type<vector_t>::vector_size;
#if CK_EXPERIMENTAL_USE_BUFFER_ATOMIC_MAX_OOB_CHECK_OFFSET_TRICK
uint32_t dst_addr_shift = dst_thread_element_valid ? 0 : 0x7fffffff;
uint32_t dst_addr_shift = dst_thread_element_valid ? 0 : 0x80000000;
amd_buffer_atomic_max_impl<scalar_t, vector_size>(
src_thread_data, dst_wave_buffer_resource, dst_addr_shift + dst_thread_addr_offset, 0);
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment