[ConstantFolding] Add ilogb in isMathLibCallNoop

[kshitijvp]

ilogb libcall was not being constant folded correctly. This patch adds ilogb case in
isMathLibCallNoop with correct error condition.

Fixes #101873

[github-actions]

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

[llvmbot]

@llvm/pr-subscribers-llvm-analysis

@llvm/pr-subscribers-llvm-transforms

Author: Kshitij Paranjape (kshitijvp)

Changes

ilogb libcall was not being constant folded correctly. It was due to LibCallShrinkWrap pass not handling
ilogbx, as a result SimplifyCFG pass did not remove redundant libcall. This patch adds ilogb case in
performCallErrors().

Fixes #101873

---

Full diff: https://github.com/llvm/llvm-project/pull/122582.diff

1 Files Affected:

• (modified) llvm/lib/Transforms/Utils/LibCallsShrinkWrap.cpp (+1)

diff --git a/llvm/lib/Transforms/Utils/LibCallsShrinkWrap.cpp b/llvm/lib/Transforms/Utils/LibCallsShrinkWrap.cpp
index 9fe655e548c22c..1316137ca4f96b 100644
--- a/llvm/lib/Transforms/Utils/LibCallsShrinkWrap.cpp
+++ b/llvm/lib/Transforms/Utils/LibCallsShrinkWrap.cpp
@@ -246,6 +246,7 @@ bool LibCallsShrinkWrap::performCallErrors(CallInst *CI,
   case LibFunc_logb:   // Same as log
   case LibFunc_logbf:  // Same as log
   case LibFunc_logbl:  // Same as log
+  case LibFunc_ilogb:  // Same as log
   {
     ++NumWrappedOneCond;
     Cond = createCond(CI, CmpInst::FCMP_OLE, 0.0f);

[kshitijvp]

@dtcxzyw @c8ef I tried this approach and it seems to work, could you please review my PR. Thanks!

[dtcxzyw]

As @c8ef pointed out, we should fix it in llvm::isMathLibCallNoop. Constant folding is correct. But we have to prove that no errors occur in the call to eliminate the call to ilogb (See also https://en.cppreference.com/w/cpp/numeric/math/ilogb).

BTW, please pre-commit tests to demonstrate that your patch fixes the issue.
See also https://llvm.org/docs/InstCombineContributorGuide.html#precommit-tests.

[c8ef]

And I am suspicious of the original code, which states that the error in calling ilogb is the same as log. However, according to the man pages, this is not true.

ilogb:

ERRORS         [top](https://man7.org/linux/man-pages/man3/ilogb.3.html#top_of_page)
       See [math_error(7)](https://man7.org/linux/man-pages/man7/math_error.7.html) for information on how to determine whether an
       error has occurred when calling these functions.

       The following errors can occur:

       Domain error: x is 0 or a NaN
              An invalid floating-point exception (FE_INVALID) is
              raised, and [errno](https://man7.org/linux/man-pages/man3/errno.3.html) is set to EDOM (but see BUGS).

       Domain error: x is an infinity
              An invalid floating-point exception (FE_INVALID) is
              raised, and [errno](https://man7.org/linux/man-pages/man3/errno.3.html) is set to EDOM (but see BUGS).

log:

See [math_error(7)](https://man7.org/linux/man-pages/man7/math_error.7.html) for information on how to determine whether an
       error has occurred when calling these functions.

       The following errors can occur:

       Domain error: x is negative
              [errno](https://man7.org/linux/man-pages/man3/errno.3.html) is set to EDOM.  An invalid floating-point exception
              (FE_INVALID) is raised.

       Pole error: x is zero
              [errno](https://man7.org/linux/man-pages/man3/errno.3.html) is set to ERANGE.  A divide-by-zero floating-point
              exception (FE_DIVBYZERO) is raised.

[kshitijvp]

@c8ef @dtcxzyw i have added the required changes could you please review it
Thanks!

[kshitijvp]

Made some minor changes. Can you review it and give me a lgtm.

[arsenm]

LGTM. We probably should use an explicit triple in the test though (I have a WIP patch to switch the default handling to assume no libcalls are available without a triple)

[kshitijvp]

Thanks, I will surely make this change in another patch, can you share your WIP implementation if possible? I want to make more contributions and learn more :)

[arsenm]

Thanks, I will surely make this change in another patch, can you share your WIP implementation if possible? I want to make more contributions and learn more :)

It's not really in a postable state (and I kind of forgot about, it's many months old)

[kshitijvp]

Okay, If you restart working on it, I'll be happy to contribute thanks!

It seems @dtcxzyw review is blocking this? I did address his requested change.

[kshitijvp]

Since the patch is approved can someone merge it.
Thanks!

[github-actions]

@kshitijvp Congratulations on having your first Pull Request (PR) merged into the LLVM Project!

Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR.

Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues.

How to do this, and the rest of the post-merge process, is covered in detail here.

If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again.

If you don't get any reports, no action is required from you. Your changes are working as expected, well done!

[VedantParanjape]

Since the patch is approved can someone merge it. Thanks!

Sure, merged it for you! Thanks for your contribution.