SimplifyDemandedBits should handle fneg/fabs/fcopysign

[arsenm]

fneg, fabs and fcopysign are defined as bit operations, and should be subject to bitwise simplification. Fneg is xor with a signbit mask, fabs is and with a sign mask, and fcopysign is and signbit mask, and inverted signbit mask, and or.

This would allow deleting more specific optimizations in visitFCOPYSIGN

This was brought up in #97180 (comment)_

Attaching starter patch to point where this needs to go:
0001-TODO-Support-FABS-FCOPYSIGN-FNEG-in-SimplifyDemanded.patch

[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: Matt Arsenault (arsenm)

fneg, fabs and fcopysign are defined as bit operations, and should be subject to bitwise simplification. Fneg is xor with a signbit mask, fabs is and with a sign mask, and fcopysign is and signbit mask, and inverted signbit mask, and or.

This would allow deleting more specific optimizations in visitFCOPYSIGN

This was brought up in #97180 (comment)_

Attaching starter patch to point where this needs to go:
0001-TODO-Support-FABS-FCOPYSIGN-FNEG-in-SimplifyDemanded.patch

[hmalgewatta]

Hi! I would like to give this a try.

[Mehdibenhadjkhelifa]

hello @hmalgewatta . Are you still working on this?

[hmalgewatta]

@Mehdibenhadjkhelifa yes

[txff99]

Hi @hmalgewatta, are you still working on it? Maybe I can take a look at this one.

[gdhillon2]

Hey @txff99
Are you still working on this? I'd love to help.

[gdhillon2]

I'm looking at the starter patch and I'm still not quite sure where this code needs to go. I believe it should go in the switch statement, specifically after line 3605 where it states in a comment:

  // TODO: There are more binop opcodes that could be handled here - MIN,
  // MAX, saturated math, etc.

since there are a lot of floating point operations handled there.
Can someone let me know if this is correct?

[arsenm]

I'm looking at the starter patch and I'm still not quite sure where this code needs to go. I believe it should go in the switch statement, specifically after line 3605 where it states in a comment:

The starter patch points you where it would go, in a different switch than the one you mentioned. The correct one is in SimplifyDemandedBits. Fneg for example should look similar to

llvm-project/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp

Line 1590 in a72bfc5

case ISD::XOR: {

, except most of the extra optimizations applied here don't really apply. At a minimum this needs the two SimplifyDemandedBits calls on each operand.

// TODO: There are more binop opcodes that could be handled here - MIN,
// MAX, saturated math, etc.
since there are a lot of floating point operations handled there. Can someone let me know if this is correct?

This is in SimplifyDemandedVectorElts, which is not what this ticket is asking for. You could handle these here as well, but it would be a separate issue.