[PATCH] D141777: [X86] Reenable store merging post-legalization

[Roman Lebedev via Phabricator via llvm-commits]

lebedev.ri created this revision.
lebedev.ri added reviewers: spatel, RKSimon, craig.topper, efriedma, pengfei.
lebedev.ri added a project: LLVM.
Herald added subscribers: StephenFan, hiraditya.
Herald added a project: All.
lebedev.ri requested review of this revision.

D62498 <https://reviews.llvm.org/D62498> disabled store merging post-legalization,
because we do indeed want to split store of concat.

But the current way it's disabled is a way too heavy of a hammer.
We only want to disable the cases that would be problematic,
but e.g. still want to allow store merging of loads.

This requires D141776 <https://reviews.llvm.org/D141776>, otherwise we have conflicting combines.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D141777

Files:
  llvm/lib/Target/X86/X86ISelLowering.cpp
  llvm/lib/Target/X86/X86ISelLowering.h
  llvm/test/CodeGen/X86/avx-arith.ll
  llvm/test/CodeGen/X86/avx-load-store.ll
  llvm/test/CodeGen/X86/avx-varargs-x86_64.ll
  llvm/test/CodeGen/X86/fold-vector-sext-crash2.ll
  llvm/test/CodeGen/X86/fp-load-trunc.ll
  llvm/test/CodeGen/X86/i128-fpconv-win64-strict.ll
  llvm/test/CodeGen/X86/i128-fpconv-win64.ll
  llvm/test/CodeGen/X86/legalize-shl-vec.ll
  llvm/test/CodeGen/X86/masked_store.ll
  llvm/test/CodeGen/X86/memset-zero.ll
  llvm/test/CodeGen/X86/merge-consecutive-loads-128.ll
  llvm/test/CodeGen/X86/merge-consecutive-stores-nt.ll
  llvm/test/CodeGen/X86/nontemporal-loads-2.ll
  llvm/test/CodeGen/X86/recip-fastmath.ll
  llvm/test/CodeGen/X86/recip-fastmath2.ll
  llvm/test/CodeGen/X86/sandybridge-loads.ll
  llvm/test/CodeGen/X86/shift-i128.ll
  llvm/test/CodeGen/X86/single_elt_vector_memory_operation.ll
  llvm/test/CodeGen/X86/sqrt-fastmath-tune.ll
  llvm/test/CodeGen/X86/sse1.ll
  llvm/test/CodeGen/X86/unaligned-32-byte-memops.ll
  llvm/test/CodeGen/X86/unaligned-load.ll
  llvm/test/CodeGen/X86/unfold-masked-merge-vector-variablemask-const.ll
  llvm/test/CodeGen/X86/unfold-masked-merge-vector-variablemask.ll
  llvm/test/CodeGen/X86/vector-gep.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D141777.489315.patch
Type: text/x-patch
Size: 114274 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230114/ee7be971/attachment-0001.bin>