[PATCH] D52669: [X86] Improve test instruction shrinking when the sign flag is used and the output of the and is truncated

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: lib/Target/X86/X86ISelDAGToDAG.cpp:3408
+          (!(Mask & 0x80) || CmpVT == MVT::i8 ||
+           hasNoSignedComparisonUses(Node))) {
         // For example, convert "testl %eax, $8" to "testb %al, $8"
----------------
RKSimon wrote:
> Possible to pull this repeated hasNoSignedComparisonUses(Node) call out?
It contains a for loop over uses so I'd rather only call it when it necessary.


================
Comment at: test/CodeGen/X86/test-shrink.ll:693
+; CHECK-X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; CHECK-X86-NEXT:    testb $-128, %al
 ; CHECK-X86-NEXT:    jg .LBB15_2
----------------
RKSimon wrote:
> Missed folding opportunity ?
Yes this is the missed folding alluded to in the summary. I don't think its unique to this change. I'll add a FIXME.


https://reviews.llvm.org/D52669