[PATCH] D128525: Test and document stackmap corner case: when the frame size isn't statically known

[Edd Barrett via Phabricator via llvm-commits]

vext01 updated this revision to Diff 439735.
vext01 added a comment.

I think the reason the alloca isn't optimised out is that:

> stackmap and patchpoint intrinsics are considered to potentially read and write all memory

https://llvm.org/docs/StackMaps.html#id12

But to be doubly sure it can't be optimised out, in this revised diff I've made use of the value returned by alloca.

Looks good?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D128525/new/

https://reviews.llvm.org/D128525

Files:
  llvm/docs/StackMaps.rst
  llvm/test/CodeGen/AArch64/stackmap-dynamic-alloca.ll
  llvm/test/CodeGen/SystemZ/stackmap-dynamic-alloca.ll
  llvm/test/CodeGen/X86/stackmap-dynamic-alloca.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D128525.439735.patch
Type: text/x-patch
Size: 3429 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20220624/ca0854ee/attachment.bin>