[Verifier] Remove redundant null-check (NFC)

[apach301]

Fixes #157448

[llvmbot]

@llvm/pr-subscribers-debuginfo

@llvm/pr-subscribers-llvm-ir

Author: Daniel Kuts (apach301)

Changes

Fixes #157448

---

Full diff: https://github.com/llvm/llvm-project/pull/157458.diff

1 Files Affected:

• (modified) llvm/lib/IR/Verifier.cpp (+1-1)

diff --git a/llvm/lib/IR/Verifier.cpp b/llvm/lib/IR/Verifier.cpp
index f38871f09f35f..57ea5b53647aa 100644
--- a/llvm/lib/IR/Verifier.cpp
+++ b/llvm/lib/IR/Verifier.cpp
@@ -3188,7 +3188,7 @@ void Verifier::visitFunction(const Function &F) {
     if (SP && ((Scope != SP) && !Seen.insert(SP).second))
       return;
 
-    CheckDI(SP->describes(&F),
+    CheckDI(SP && SP->describes(&F),
             "!dbg attachment points at wrong subprogram for function", N, &F,
             &I, DL, Scope, SP);
   };

[artagnon]

Kindly add a test, showing that this bug can actually be hit.

[apach301]

Kindly add a test, showing that this bug can actually be hit.

It was a static analyzer report, so I don't know how to run this code and have any input data. But the problem in this code is that 'SP' could be NULL - in previous if-statement it is checked on NULL. So it also shoulf be checked in current function

[artagnon]

I don't think DISubprogram::getSubprogram can return null:

DISubprogram *DILocalScope::getSubprogram() const {
  if (auto *Block = dyn_cast<DILexicalBlockBase>(this))
    return Block->getScope()->getSubprogram();
  return const_cast<DISubprogram *>(cast<DISubprogram>(this));
}

So, the check SP && preceding your patch should probably be stripped, as it is dead?

[apach301]

So, the check SP && preceding your patch should probably be stripped, as it is dead?

If it cant return nullptr after cast, then yes, I could remove both checks

[artagnon]

Please update the PR title/description. Suffix the title with "(NFC)" to make it clear that the change is non-functional.

[artagnon]

LGTM, thanks!