[PATCH] D93967: [SLP]Need shrink the load vector after reordering.

[Martin Storsjö via Phabricator via llvm-commits]

mstorsjo added a comment.

This caused misoptimizations for armv7, where code that previously worked correctly now produce different results. (The code is clean under ubsan, so it shouldn't be relying on anything undefined.)

The issue appears with https://martin.st/temp/interplayvideo-preproc.c, compiled with `clang -target armv7-linux-gnueabihf -O2`.

The diff in generated code, before/after, looks like this:

          vmov.32 d16[0], lr
          vmov.32 d16[1], r2
   .LBB27_3:                               @ %if.end
                                           @   in Loop: Header=BB27_4 Depth=1
          vmov.32 r2, d16[1]
          add     r3, r3, #1
          vmov.32 r5, d16[0]
          cmp     r3, #8
  +       vdup.32 d16, d16[0]
          vmov.16 d17[1], r2
          vmov.16 d18[0], r5
          vdup.16 d21, d17[1]
          vdup.16 d20, d18[0]
          vst1.16 {d20, d21}, [r1], r12
          beq     .LBB27_9
   .LBB27_4:                               @ %for.body
                                           @ =>This Inner Loop Header: Depth=1
          tst     r3, #3
          bne     .LBB27_3

If it loops back to `.LBB27_3`, the vector element `d16[1]` no longer has the value it was expected to have.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D93967/new/

https://reviews.llvm.org/D93967