[AArch64][SVE] Ensure redundant PTEST are removed with an 'invalid' PTRUE

[bsmith]

Summary

When a PTRUE of non-element size is encountered, the PTEST optimization
logic bails out since it cannot handle that type of PTRUE. Instead, it
should be treated as a generic predicate to allow later optimizations trigger.

[c-rhodes]

LGTM

llvm/lib/Target/AArch64/AArch64InstrInfo.cpp

bool AArch64InstrInfo::optimizePTestInstr(
   auto NewOp = Pred->getOpcode();
   bool OpChanged = false;
 
   unsigned MaskOpcode = Mask->getOpcode();
   unsigned PredOpcode = Pred->getOpcode();
   bool PredIsPTestLike = isPTestLikeOpcode(PredOpcode);
   bool PredIsWhileLike = isWhileOpcode(PredOpcode);
 
-  if (isPTrueOpcode(MaskOpcode) && (PredIsPTestLike || PredIsWhileLike)) {
-    // For PTEST(PTRUE_ALL, WHILE), if the element size matches the PTEST is
+  if (isPTrueOpcode(MaskOpcode) && (PredIsPTestLike || PredIsWhileLike) &&
+      getElementSizeForOpcode(MaskOpcode) ==
+          getElementSizeForOpcode(PredOpcode) &&
+      Mask->getOperand(1).getImm() == 31) {
+    // For PTEST(PTRUE_ALL, WHILE), if the element size matches, the PTEST is
     // redundant since WHILE performs an implicit PTEST with an all active
-    // mask.
-    uint64_t MaskElementSize = getElementSizeForOpcode(MaskOpcode);
-    uint64_t PredElementSize = getElementSizeForOpcode(PredOpcode);
-
-    // Must be an all active predicate of matching element size.
-    if ((PredElementSize != MaskElementSize) ||
-        (Mask->getOperand(1).getImm() != 31))
-      return false;
+    // mask. Must be an all active predicate of matching element size.
 
     // For PTEST(PTRUE_ALL, PTEST_LIKE), the PTEST is redundant if the
     // PTEST_LIKE instruction uses the same all active mask and the element
     // size matches. If the PTEST has a condition of any then it is always
     // redundant.
     if (PredIsPTestLike) {
       auto PTestLikeMask = MRI->getUniqueVRegDef(Pred->getOperand(1).getReg());
       if (Mask != PTestLikeMask && PTest->getOpcode() != AArch64::PTEST_PP_ANY)

llvm/test/CodeGen/AArch64/sve-ptest-removal-cmple.ll

 ; CHECK-NEXT:    cset w0, ne
 ; CHECK-NEXT:    ret
   %1 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b)
   %2 = tail call i1 @llvm.aarch64.sve.ptest.any.nxv16i1(<vscale x 16 x i1> %pg, <vscale x 16 x i1> %1)
   ret i1 %2
 }
 
 ;
+; Same as above except PG = incorrectly sized PTRUE
+;
+define i1 @cmp8_ptest_any_px_bad_ptrue(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b) {
+; CHECK-LABEL: cmp8_ptest_any_px_bad_ptrue:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    ptrue p0.s
+; CHECK-NEXT:    cmpge p0.b, p0/z, z0.b, z1.b
+; CHECK-NEXT:    cset w0, ne
+; CHECK-NEXT:    ret
+  %1 = tail call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+  %2 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> %1)
+  %3 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> %2, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b)
+  %4 = tail call i1 @llvm.aarch64.sve.ptest.any.nxv16i1(<vscale x 16 x i1> %2, <vscale x 16 x i1> %3)
+  ret i1 %4
+}
+
+;
 ; PTEST_FIRST(PG, CMP32(PG, A, B)). Can't remove PTEST since PTEST.B vs CMP.S.
 ;
 define i1 @cmp32_ptest_first_px(<vscale x 16 x i1> %pg, <vscale x 4 x i32> %a, <vscale x 4 x i32> %b) {
 ; CHECK-LABEL: cmp32_ptest_first_px:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    cmpge p1.s, p0/z, z0.s, z1.s
 ; CHECK-NEXT:    ptest p0, p1.b
 ; CHECK-NEXT:    cset w0, mi

 ; CHECK-NEXT:    ret
   %1 = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg)
   %2 = tail call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> %1, <vscale x 4 x i32> %a, <vscale x 4 x i32> %b)
   %3 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> %2)
   %4 = tail call i1 @llvm.aarch64.sve.ptest.any.nxv16i1(<vscale x 16 x i1> %pg, <vscale x 16 x i1> %3)
   ret i1 %4
 }
 
+;
+; Same as above except PG = incorrectly sized PTRUE
+;
+define i1 @cmp32_ptest_any_px_bad_ptrue(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b) {
+; CHECK-LABEL: cmp32_ptest_any_px_bad_ptrue:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    ptrue p0.b
+; CHECK-NEXT:    cmpge p0.s, p0/z, z0.s, z1.s
+; CHECK-NEXT:    cset w0, ne
+; CHECK-NEXT:    ret
+  %1 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+  %2 = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %1)
+  %3 = tail call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> %2, <vscale x 4 x i32> %a, <vscale x 4 x i32> %b)
+  %4 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> %3)
+  %5 = tail call i1 @llvm.aarch64.sve.ptest.any.nxv16i1(<vscale x 16 x i1> %1, <vscale x 16 x i1> %4)
+  ret i1 %5
+}
+
 ; ==============================================================================
 ; PTEST_OP(X=CMP(PG, ...), X)
 ; ==============================================================================
 
 ;
 ; PTEST_FIRST(X=CMP8(PG, A, B), X). PTEST is redundant if condition is changed
 ; to any.
 ;