diff options
| author | Pablo Marquez Tello <pablo.tello@arm.com> | 2021-03-24 17:50:19 +0000 |
|---|---|---|
| committer | Pablo Marquez Tello <pablo.tello@arm.com> | 2021-03-25 13:25:54 +0000 |
| commit | f73db971cfc36c82c1aa6409257a11f987aaea92 (patch) | |
| tree | 3d2fd738415e057801281d44a49f4b9e15388aba /src/core/CL/cl_kernels/winograd_output_transform.cl | |
| parent | e7254a09c54f09413ac3b23d4997089ebdbc4387 (diff) | |
| download | ComputeLibrary-f73db971cfc36c82c1aa6409257a11f987aaea92.tar.gz | |
bf16_to_float: Fix strict aliasing violation
* This function was accessing a uint32_t with a pointer to float which
violates the strict aliasing rules. With GCC 11, this leads to a
-Wuninitialized warning, which causes the build to fail.
* This change uses memcpy instead of type punning which fixes the strict
aliasing violation, thereby fixing the build with GCC 11.
* Resolves MLCE-420
Change-Id: I9fda5f1e68ac68490b244bd40380910adae00bf3
Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5318
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Diffstat (limited to 'src/core/CL/cl_kernels/winograd_output_transform.cl')
0 files changed, 0 insertions, 0 deletions