As noted in D112464, a pre-AVX target can't fold an under-aligned vector load into another op, so we shouldn't report that as a load folding candidate. I only found one caller where this would make a difference -- combineCommutableSHUFP() -- so that's where I added a test to show the (minor) regression.
Details
Details
Diff Detail
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
llvm/lib/Target/X86/X86ISelLowering.cpp | ||
---|---|---|
5050 | What about Subtarget->hasSSEUnalignedMem()? |
llvm/lib/Target/X86/X86ISelLowering.cpp | ||
---|---|---|
5050 | Ah, forgot about that possibility. I'll add the clause along with another RUN line to test it. |
Comment Actions
Patch updated:
Added clause for hasSSEUnalignedMem() and adjusted test file with extra RUN line for that.
What about Subtarget->hasSSEUnalignedMem()?