[RISCV] Added test case for PR119527. NFC

[mikhailramalho]

C2 can be null here, so we need to check it or clang may crash.

[llvmbot]

@llvm/pr-subscribers-backend-risc-v

Author: Mikhail R. Gadelha (mikhailramalho)

Changes

C2 can be null here, so we need to check it or clang may crash.

---

Full diff: https://github.com/llvm/llvm-project/pull/121816.diff

1 Files Affected:

• (modified) llvm/lib/Target/RISCV/RISCVISelLowering.cpp (+1-1)

diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index 7efe3732d8be13..9acd374225505d 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -18388,7 +18388,7 @@ bool RISCVTargetLowering::isDesirableToCommuteWithShift(
     auto *C2 = dyn_cast<ConstantSDNode>(N->getOperand(1));
 
     // Bail if we might break a sh{1,2,3}add pattern.
-    if (Subtarget.hasStdExtZba() && C2->getZExtValue() >= 1 &&
+    if (Subtarget.hasStdExtZba() && C2 && C2->getZExtValue() >= 1 &&
         C2->getZExtValue() <= 3 && N->hasOneUse() &&
         N->user_begin()->getOpcode() == ISD::ADD &&
         !isUsedByLdSt(*N->user_begin(), nullptr) &&

[mikhailramalho]

creduce is running now to reduce the test case I got, I'll add the test case as soon as it's done.

[lukel97]

Thanks, LGTM w/ Philip's comment about moving the test

[lukel97]

Since a bunch of people were running into this crash I've cherry-picked the RISCVISelLowering.cpp changes to main in b0e05a5, but it still needs the test case from this PR

[lukel97]

LGTM again, just need to update the PR title to reflect that it's adding a test and it's NFC

[preames]

LGTM w/minor comment. (I second Luke's comment on the change to description.)