Don't use _bzhi_u32 under MSan

MSan knows that x & 0xFF only uses the lower byte from x but it isn't as smart about _bzhi_u32(val, 8). (I'll file an upstream bug.)
2018-12-13 17:20:06 -08:00
parent 136b3ebc31
commit 5913c5f8e4
1 changed files with 1 additions and 1 deletions
--- a/snappy.cc
+++ b/snappy.cc
@@ -695,7 +695,7 @@ static inline void Report(const char *algorithm, size_t compressed_size,
 static inline uint32 ExtractLowBytes(uint32 v, int n) {
  DCHECK_GE(n, 0);
  DCHECK_LE(n, 4);
-#ifdef __BMI2__
+#if defined(__BMI2__) && !defined(MEMORY_SANITIZER)
  return _bzhi_u32(v, 8 * n);
 #else
  // This needs to be wider than uint32 otherwise `mask << 32` will be