[PATCH] D128344: [AMDGPU] Add the "is_dynamic_callstack" of amd_kernel_code_t to the kernel descriptor
Matt Arsenault via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Wed Jun 22 11:14:08 PDT 2022
arsenm added inline comments.
================
Comment at: llvm/docs/AMDGPUUsage.rst:4000
32 mode.
- 463:459 1 bit Reserved, must be 0.
+ 459 1 bit IS_DYNAMIC_CALLSTACK Indicates if the generated
+ machine code is using a
----------------
Since we are actually re-introducing this and not recycling an existing field, we should take the chance to rename it. The call part of the name is misleading since we can need this in situations that do not have calls. How about just DYNAMIC_STACK_SIZE?
================
Comment at: llvm/include/llvm/Support/AMDHSAKernelDescriptor.h:180-181
uint32_t compute_pgm_rsrc2;
- uint16_t kernel_code_properties;
- uint8_t reserved2[6];
+ uint32_t kernel_code_properties;
+ uint8_t reserved2[4];
};
----------------
arsenm wrote:
> I thought this was already defined here, which was the reason to set it. I'm not sure it makes sense to add the field if it's not already there
We apparently do need to place this here since rocr doesn't want to inspect metadata
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D128344/new/
https://reviews.llvm.org/D128344
More information about the llvm-commits
mailing list