LLVM cannot constant fold memchr with local buffer

[philnik777]

LLVM is currently unable to constant fold calls memchr when a local buffer is passed. I noticed this while looking into libc++'s std::find optimizations.

This is a simple example where Clang doesn't constant fold the function even though it should be able to do that trivially: (Godbolt)

auto test() {
  char buffer[] = "Banane";
  return __builtin_memchr(buffer, 'a', __builtin_strlen(buffer)) == nullptr;
}

I think the problem is that __builtin_memchr can escape the pointer, so the buffer isn't recognized as immutable, resulting in the constant folder failing.

I'm in no way an expert in compiler optimizations, but I think it would be possible to check whether the pointer actually escapes, and if not, mark the pointer as noescape to at least partially solve this problem.

[nikic]

Yes, missing nocapture is the issue. It folds if we manually add it to the IR: https://llvm.godbolt.org/z/Ed553Yvxd

It should be easy to infer nocapture in SimplifyLibCalls for memchr calls that are only compared against nullptr.

The case where the return value is used for more than that would be harder though.

Tracking return-only captures (part of https://discourse.llvm.org/t/rfc-improvements-to-capture-tracking/81420) would probably also handle your case as a side effect.

[llvmbot]

Hi!

This issue may be a good introductory issue for people new to working on LLVM. If you would like to work on this issue, your first steps are:

1. Check that no other contributor has already been assigned to this issue. If you believe that no one is actually working on it despite an assignment, ping the person. After one week without a response, the assignee may be changed.
2. In the comments of this issue, request for it to be assigned to you, or just create a pull request after following the steps below. Mention this issue in the description of the pull request.
3. Fix the issue locally.
4. Run the test suite locally. Remember that the subdirectories under test/ create fine-grained testing targets, so you can e.g. use make check-clang-ast to only run Clang's AST tests.
5. Create a Git commit.
6. Run git clang-format HEAD~1 to format your changes.
7. Open a pull request to the upstream repository on GitHub. Detailed instructions can be found in GitHub's documentation. Mention this issue in the description of the pull request.

If you have any further questions about this issue, don't hesitate to ask via a comment in the thread below.

[llvmbot]

@llvm/issue-subscribers-good-first-issue

Author: Nikolas Klauser (philnik777)

LLVM is currently unable to constant fold calls `memchr` when a local buffer is passed. I noticed this while looking into libc++'s `std::find` optimizations.

This is a simple example where Clang doesn't constant fold the function even though it should be able to do that trivially: (Godbolt)

auto test() {
  char buffer[] = "Banane";
  return __builtin_memchr(buffer, 'a', __builtin_strlen(buffer)) == nullptr;
}

I think the problem is that __builtin_memchr can escape the pointer, so the buffer isn't recognized as immutable, resulting in the constant folder failing.

I'm in no way an expert in compiler optimizations, but I think it would be possible to check whether the pointer actually escapes, and if not, mark the pointer as noescape to at least partially solve this problem.

[sidnarsipur]

Hey, can I be assigned this?

[NeyagapulaSiddhartha]

hey if no one is working on this , I would like to be assigned !!

[nikic]

I haven't tested it, but I'd expect that #132007 fixes this. It's currently blocked on #131168 though.