[PATCH] D87971: [MemLoc] Support lllvm.memcpy.inline in MemoryLocation::getForArgument

[Florian Hahn via Phabricator via llvm-commits]

fhahn added inline comments.


================
Comment at: llvm/test/Transforms/DeadStoreElimination/MSSA/memintrinsics.ll:111
+
+  call void @llvm.memcpy.inline.p0i8.p0i8.i8(i8* %A, i8* %B, i8 -1, i1 false)
+
----------------
why pass `-1` as length? That make the function be UB?

 It would probably also be better to just pass in 2 noalias pointers, rather than allocas. That way the memcpy call won't be removed?


================
Comment at: llvm/test/Transforms/DeadStoreElimination/MSSA/memset-and-memcpy.ll:103
   tail call void @llvm.memcpy.inline.p0i8.p0i8.i64(i8* align 1 %P, i8* align 1 %Q, i64 12, i1 false)
- ret void
+  ret void
 }
----------------
nit: tab? Same below.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D87971/new/

https://reviews.llvm.org/D87971