[PATCH] D38128: Handle COPYs of physregs better (regalloc hints)

[Nirav Dave via Phabricator via llvm-commits]

niravd added a comment.

I've gone through and marked all the places.

> Is this just about the extra 'movdqa' in vector-shift-ashr-128.ll, or are there other diffs to look at?





================
Comment at: test/CodeGen/X86/vector-shift-lshr-128.ll:232
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movdqa %xmm2, %xmm0
 ; SSE41-NEXT:    psllw $12, %xmm0
----------------
Extra instruction here


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:4539
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
+; SSE4-NEXT:    movdqa %xmm0, %xmm7
 ; SSE4-NEXT:    movdqa %xmm4, %xmm0
----------------
Extra instruction here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:4660
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
+; SSE4-NEXT:    movdqa %xmm0, %xmm7
 ; SSE4-NEXT:    movdqa %xmm4, %xmm0
----------------
Extra instruction here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:4781
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
+; SSE4-NEXT:    movdqa %xmm0, %xmm7
 ; SSE4-NEXT:    pcmpgtq %xmm4, %xmm0
----------------
Extra Instruction here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:4901
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
+; SSE4-NEXT:    movdqa %xmm0, %xmm7
 ; SSE4-NEXT:    pcmpgtq %xmm4, %xmm0
----------------
Extra instruction here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:5019
 ; SSE4-LABEL: test125:
 ; SSE4:       # %bb.0: # %entry
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
----------------
Extra instructions here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:5170
 ; SSE4:       # %bb.0: # %entry
-; SSE4-NEXT:    movdqa %xmm0, %xmm9
-; SSE4-NEXT:    movdqa {{.*#+}} xmm8 = [9223372036854775808,9223372036854775808]
-; SSE4-NEXT:    movdqa %xmm0, %xmm10
-; SSE4-NEXT:    pxor %xmm8, %xmm10
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
+; SSE4-NEXT:    movdqa %xmm6, %xmm9
----------------
Extra instructions here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:5319
 ; SSE4-LABEL: test127:
 ; SSE4:       # %bb.0: # %entry
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
----------------
Extra instructions here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:5468
 ; SSE4-LABEL: test128:
 ; SSE4:       # %bb.0: # %entry
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
----------------
Extra instructions here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:6995
 ; SSE4-LABEL: test153:
 ; SSE4:       # %bb.0: # %entry
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
----------------
Extra instructions here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:7116
 ; SSE4:       # %bb.0: # %entry
-; SSE4-NEXT:    movdqa %xmm0, %xmm8
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
+; SSE4-NEXT:    movdqa %xmm0, %xmm7
----------------
Extra instruction here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:7235
 ; SSE4-LABEL: test155:
 ; SSE4:       # %bb.0: # %entry
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
----------------
Extra instruction here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:7356
 ; SSE4-LABEL: test156:
 ; SSE4:       # %bb.0: # %entry
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
----------------
Extra instructions here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:7505
 ; SSE4-LABEL: test159:
 ; SSE4:       # %bb.0: # %entry
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
----------------
Extra Instructions here.


================
Comment at: test/CodeGen/X86/vselect-minmax.ll:7655
 ; SSE4-LABEL: test160:
 ; SSE4:       # %bb.0: # %entry
+; SSE4-NEXT:    movdqa %xmm7, %xmm8
----------------
Extra instructions here.


Repository:
  rL LLVM

https://reviews.llvm.org/D38128