[PATCH] D28037: [PowerPC, DAGCombiner] Change vec_sl to a << (b % (sizeof(a) * 8)), and fold it back to a << b.

[Tim Shen via Phabricator via llvm-commits]

timshen created this revision.
timshen added reviewers: kbarton, hfinkel, iteratee, echristo, bogner.
timshen added subscribers: llvm-commits, cfe-commits.
Herald added subscribers: amehsan, nemanjai, mehdi_amini.

For a << b (as original vec_sl does), if b >= sizeof(a) * 8, the
behavior is undefined. However, Power instructions do define the
behavior, which is equivalent to a << (b % (sizeof(a) * 8)).

This patch changes altivec.h to use a << (b % (sizeof(a) * 8)), to ensure
the consistent semantic of the instructions. Then it combines
the generated multiple instructions back to a single shift.

This patch handles left shift only. Right shift, on the other hand, is more
complicated, considering arithematic/logical right shift.


https://reviews.llvm.org/D28037

Files:
  clang/lib/Headers/altivec.h
  clang/test/CodeGen/builtins-ppc-altivec.c
  llvm/include/llvm/Target/TargetLowering.h
  llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
  llvm/lib/Target/PowerPC/PPCISelLowering.h
  llvm/test/CodeGen/PowerPC/shift_mask.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D28037.82285.patch
Type: text/x-patch
Size: 16917 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20161222/cd122876/attachment.bin>