courbet added inline comments.
================
Comment at: llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp:189
+ L->setAlignment(
+ MaybeAlign(CopySrcAlign)); // Check if we can use Align instead.
if (CopyMD)
----------------
FIXME
================
Comment at: llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp:203
+ S->setAlignment(
+ MaybeAlign(CopyDstAlign)); // Check if we can use Align instead.
if (CopyMD)
----------------
ditto
================
Comment at: llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp:277
MI->isVolatile());
- S->setAlignment(Alignment);
+ assert(Alignment && "Alignment must be set");
+ S->setAlignment(Align(Alignment));
----------------
make it const + remove assert ?
================
Comment at: llvm/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp:1367
+ SI.setAlignment(
+ MaybeAlign(KnownAlign)); // Check if we can use Align instead.
else if (StoreAlign == 0)
----------------
`getOrEnforceKnownAlignment()` seems to always be returning >0
================
Comment at: llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp:4027
- ST->setAlignment(Alignment);
+ ST->setAlignment(Align(Alignment));
Value *V = propagateMetadata(ST, E->Scalars);
----------------
I don't think there are any guarantees that `getABITypeAlignment` is non-zero
for now.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D68268/new/
https://reviews.llvm.org/D68268
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
