Skip to content

Fix M1 tail call issue when building #44279

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Feb 23, 2022
Merged

Fix M1 tail call issue when building #44279

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7f43963
Change macro
gbaraldi Feb 20, 2022
7565a52
fix macro
gbaraldi Feb 20, 2022
e102b5c
Fix comment
gbaraldi Feb 20, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 3 additions & 2 deletions src/ccall.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -233,10 +233,11 @@ static GlobalVariable *emit_plt_thunk(
else {
// musttail support is very bad on ARM, PPC, PPC64 (as of LLVM 3.9)
// Known failures includes vararg (not needed here) and sret.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

possible improvement?

Suggested change
// Known failures includes vararg (not needed here) and sret.
// Known failures includes vararg (not needed here) and sret.
ret->setTailCall();

Is this only M1, or all AArch64?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is m1 only. I wonder if it's this https://github.com/llvm/llvm-project/blob/d1cd64ffdd832220dbe1829c2f09b880be67be31/llvm/test/CodeGen/AArch64/GlobalISel/call-translator-musttail.ll, this is the only aarch64-apple exclusive thing related to musttail.

That might be a more robust solution for sure. Does doing the ccall as a tail call have some performance benefit?

#if (defined(_CPU_X86_) || defined(_CPU_X86_64_) || \
defined(_CPU_AARCH64_))

#if (defined(_CPU_X86_) || defined(_CPU_X86_64_) || (defined(_CPU_AARCH64_) && !defined(_OS_DARWIN_)))
// Ref https://bugs.llvm.org/show_bug.cgi?id=47058
// LLVM, as of 10.0.1 emits wrong/worse code when musttail is set
// Apple silicon macs give an LLVM ERROR if musttail is set here #44107.
if (!attrs.hasAttrSomewhere(Attribute::ByVal))
ret->setTailCallKind(CallInst::TCK_MustTail);
#endif
Expand Down