diff options
| author | Paul Kirth <paulkirth@google.com> | 2025-05-05 18:16:33 -0700 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-05-05 18:16:33 -0700 |
| commit | 43eafc0c4aca0b2fd159c09d4b162c1941b4f4ed (patch) | |
| tree | ed8cf3c481182e8dc7510ddc8d9de00579fc5f15 /llvm/lib/Transforms/Utils/Local.cpp | |
| parent | 1c1238d3615a7e1a99570d1e02de3b538d2e0669 (diff) | |
| download | llvm-43eafc0c4aca0b2fd159c09d4b162c1941b4f4ed.zip llvm-43eafc0c4aca0b2fd159c09d4b162c1941b4f4ed.tar.gz llvm-43eafc0c4aca0b2fd159c09d4b162c1941b4f4ed.tar.bz2 | |
[llvm][gvn-sink] Don't try to sink inline asm (#138414)
Fixes #138345. Before this patch, gvn-sink would try to sink inline
assembly statements. Other GVN passes avoid them (see
https://github.com/llvm/llvm-project/blob/b4fac94181c4cf17dbb7ecc2ae975712b0e4a6d1/llvm/lib/Transforms/Scalar/GVN.cpp#L2932
Similarly, gvn-sink should skip these instructions, since they are not
safe to move. To do this, we update the early exit in
canReplaceOperandWithVariable, since it should have caught this case.
It's more efficient to also skip numbering in GVNSink if the instruction
is InlineAsm, but that should be infrequent.
The test added is reduced from a failure when compiling Fuchsia with
gvn-sink.
Diffstat (limited to 'llvm/lib/Transforms/Utils/Local.cpp')
| -rw-r--r-- | llvm/lib/Transforms/Utils/Local.cpp | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/llvm/lib/Transforms/Utils/Local.cpp b/llvm/lib/Transforms/Utils/Local.cpp index 809a0d7..ce03bc0 100644 --- a/llvm/lib/Transforms/Utils/Local.cpp +++ b/llvm/lib/Transforms/Utils/Local.cpp @@ -4225,8 +4225,9 @@ bool llvm::canReplaceOperandWithVariable(const Instruction *I, unsigned OpIdx) { return false; // Early exit. - if (!isa<Constant>(I->getOperand(OpIdx))) + if (!isa<Constant, InlineAsm>(I->getOperand(OpIdx))) { return true; + } switch (I->getOpcode()) { default: |