[llvm-branch-commits] [HWASan] [LSan] consider allocations with UAF tag poisoned (PR #191920)

[Vitaly Buka via llvm-branch-commits]

================
@@ -30,7 +31,9 @@ uptr TagMemory(uptr p, uptr size, tag_t tag) {
 // --- Implementation of LSan-specific functions --- {{{1
 namespace __lsan {
 bool WordIsPoisoned(uptr addr) {
----------------
vitalybuka wrote:

I understand that this is how it is now.
Why don't we move into that direction, if we load tag from shadow anyway?

https://github.com/llvm/llvm-project/pull/191920