[PowerPC] Do not emit record-form rotates when record-form andi/andis suffices

This is a follow-up to the previous patch that eliminated some of the rotates.
With this addition, we will also emit the record-form andis.

This patch increases the number of record-form rotates we eliminate by
more than 70%.

Differential revision: https://reviews.llvm.org/D44897

llvm-svn: 342478

4 files changed, 111 insertions, 28 deletions

diff --git a/llvm/lib/Target/PowerPC/PPCInstrInfo.cpp b/llvm/lib/Target/PowerPC/PPCInstrInfo.cpp
index c3a280ad9ae..f9d173699ec 100644
--- a/llvm/lib/Target/PowerPC/PPCInstrInfo.cpp
+++ b/llvm/lib/Target/PowerPC/PPCInstrInfo.cpp
@@ -1913,14 +1913,36 @@ bool PPCInstrInfo::optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
     // compare).
 
     // Rotates are expensive instructions. If we're emitting a record-form
-    // rotate that can just be an andi, we should just emit the andi.
-    if ((MIOpC == PPC::RLWINM || MIOpC == PPC::RLWINM8) &&
-        MI->getOperand(2).getImm() == 0) {
+    // rotate that can just be an andi/andis, we should just emit that.
+    if (MIOpC == PPC::RLWINM || MIOpC == PPC::RLWINM8) {
+      unsigned GPRRes = MI->getOperand(0).getReg();
+      int64_t SH = MI->getOperand(2).getImm();
       int64_t MB = MI->getOperand(3).getImm();
       int64_t ME = MI->getOperand(4).getImm();
-      if (MB < ME && MB >= 16) {
-        uint64_t Mask = ((1LLU << (32 - MB)) - 1) & ~((1LLU << (31 - ME)) - 1);
-        NewOpC = MIOpC == PPC::RLWINM ? PPC::ANDIo : PPC::ANDIo8;
+      // We can only do this if both the start and end of the mask are in the
+      // same halfword.
+      bool MBInLoHWord = MB >= 16;
+      bool MEInLoHWord = ME >= 16;
+      uint64_t Mask = ~0LLU;
+
+      if (MB <= ME && MBInLoHWord == MEInLoHWord && SH == 0) {
+        Mask = ((1LLU << (32 - MB)) - 1) & ~((1LLU << (31 - ME)) - 1);
+        // The mask value needs to shift right 16 if we're emitting andis.
+        Mask >>= MBInLoHWord ? 0 : 16;
+        NewOpC = MIOpC == PPC::RLWINM ?
+          (MBInLoHWord ? PPC::ANDIo : PPC::ANDISo) :
+          (MBInLoHWord ? PPC::ANDIo8 :PPC::ANDISo8);
+      } else if (MRI->use_empty(GPRRes) && (ME == 31) &&
+                 (ME - MB + 1 == SH) && (MB >= 16)) {
+        // If we are rotating by the exact number of bits as are in the mask
+        // and the mask is in the least significant bits of the register,
+        // that's just an andis. (as long as the GPR result has no uses).
+        Mask = ((1LLU << 32) - 1) & ~((1LLU << (32 - SH)) - 1);
+        Mask >>= 16;
+        NewOpC = MIOpC == PPC::RLWINM ? PPC::ANDISo :PPC::ANDISo8;
+      }
+      // If we've set the mask, we can transform.
+      if (Mask != ~0LLU) {
         MI->RemoveOperand(4);
         MI->RemoveOperand(3);
         MI->getOperand(2).setImm(Mask);
diff --git a/llvm/test/CodeGen/PowerPC/noPermuteFormasking.ll b/llvm/test/CodeGen/PowerPC/noPermuteFormasking.ll
index d4ce1a7ec66..8cd828f0a22 100644
--- a/llvm/test/CodeGen/PowerPC/noPermuteFormasking.ll
+++ b/llvm/test/CodeGen/PowerPC/noPermuteFormasking.ll
@@ -40,3 +40,64 @@ for.body.i.i.i.i.i.i.i:                           ; preds = %for.body.i.i.i.i.i.
 exitBB: ; preds = %for.body.i.i.i.i.i.i.i.prol.loopexit
   ret void
 }
+
+define signext i32 @andis_bot(i32 signext %a, i32 signext %b) {
+; CHECK-LABEL: andis_bot:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    andis. 5, 3, 1
+; CHECK-NEXT:    li 5, 1
+; CHECK-NEXT:    isel 4, 4, 5, 2
+; CHECK-NEXT:    mullw 3, 4, 3
+; CHECK-NEXT:    extsw 3, 3
+; CHECK-NEXT:    blr
+entry:
+  %and = and i32 %a, 65536
+  %tobool = icmp eq i32 %and, 0
+  %mul = select i1 %tobool, i32 %b, i32 1
+  %cond = mul nsw i32 %mul, %a
+  ret i32 %cond
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @andis_mid(i32 signext %a, i32 signext %b) {
+; CHECK-LABEL: andis_mid:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    andis. 5, 3, 252
+; CHECK-NEXT:    li 5, 1
+; CHECK-NEXT:    isel 4, 4, 5, 2
+; CHECK-NEXT:    mullw 3, 4, 3
+; CHECK-NEXT:    extsw 3, 3
+; CHECK-NEXT:    blr
+entry:
+  %and = and i32 %a, 16515072
+  %tobool = icmp eq i32 %and, 0
+  %mul = select i1 %tobool, i32 %b, i32 1
+  %cond = mul nsw i32 %mul, %a
+  ret i32 %cond
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @andis_top(i32 signext %a, i32 signext %b) {
+; CHECK-LABEL: andis_top:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    andis. 5, 3, 64512
+; CHECK-NEXT:    li 5, 1
+; CHECK-NEXT:    isel 4, 4, 5, 2
+; CHECK-NEXT:    mullw 3, 4, 3
+; CHECK-NEXT:    extsw 3, 3
+; CHECK-NEXT:    blr
+entry:
+  %tobool = icmp ugt i32 %a, 67108863
+  %mul = select i1 %tobool, i32 1, i32 %b
+  %cond = mul nsw i32 %mul, %a
+  ret i32 %cond
+}
+
+define i64 @andis_no_cmp(i64 %a, i64 %b) {
+entry:
+  %and = and i64 %a, 65536
+  %tobool = icmp eq i64 %and, 0
+  %mul = select i1 %tobool, i64 %b, i64 1
+  %cond = mul nsw i64 %mul, %a
+  ret i64 %cond
+}
diff --git a/llvm/test/CodeGen/PowerPC/tail-dup-break-cfg.ll b/llvm/test/CodeGen/PowerPC/tail-dup-break-cfg.ll
index f19b11f2ae4..62c1e068b84 100644
--- a/llvm/test/CodeGen/PowerPC/tail-dup-break-cfg.ll
+++ b/llvm/test/CodeGen/PowerPC/tail-dup-break-cfg.ll
@@ -15,12 +15,12 @@ target triple = "powerpc64le-grtev4-linux-gnu"
 ;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 1
 ;CHECK-NEXT: bc 12, 1, [[BODY1LABEL:[._0-9A-Za-z]+]]
 ;CHECK-NEXT: # %test2
-;CHECK-NEXT: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 30, 30
+;CHECK-NEXT: andi. {{[0-9]+}}, [[TAGREG]], 2
 ;CHECK-NEXT: bne 0, [[BODY2LABEL:[._0-9A-Za-z]+]]
 ;CHECK: [[EXITLABEL:[._0-9A-Za-z]+]]: # %exit
 ;CHECK: blr
 ;CHECK-NEXT: [[BODY1LABEL]]
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 30, 30
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 2
 ;CHECK-NEXT: beq 0, [[EXITLABEL]]
 ;CHECK-NEXT: [[BODY2LABEL:[._0-9A-Za-z]+]]:
 ;CHECK: b [[EXITLABEL]]
@@ -58,7 +58,7 @@ exit:
 ;CHECK-NEXT: bc 4, 1, [[TEST2LABEL:[._0-9A-Za-z]+]]
 ;CHECK-NEXT: # %body1
 ;CHECK: [[TEST2LABEL]]: # %test2
-;CHECK-NEXT: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 30, 30
+;CHECK-NEXT: andi. {{[0-9]+}}, [[TAGREG]], 2
 ;CHECK-NEXT: beq 0, [[EXITLABEL:[._0-9A-Za-z]+]]
 ;CHECK-NEXT: # %body2
 ;CHECK: [[EXITLABEL:[._0-9A-Za-z]+]]: # %exit
@@ -106,7 +106,7 @@ declare void @d()
 ; CHECK: # %succ
 ; CHECK: # %c
 ; CHECK: bl c
-; CHECK: rlwinm. {{[0-9]+}}, {{[0-9]+}}, 0, 29, 29
+; CHECK: andi. {{[0-9]+}}, {{[0-9]+}}, 4
 ; CHECK: beq
 ; CHECK: b
 define void @tail_dup_no_succ(i32 %tag) {
diff --git a/llvm/test/CodeGen/PowerPC/tail-dup-layout.ll b/llvm/test/CodeGen/PowerPC/tail-dup-layout.ll
index a4e232941c8..bcae75f43cb 100644
--- a/llvm/test/CodeGen/PowerPC/tail-dup-layout.ll
+++ b/llvm/test/CodeGen/PowerPC/tail-dup-layout.ll
@@ -28,24 +28,24 @@ target triple = "powerpc64le-grtev4-linux-gnu"
 ;CHECK: andi. {{[0-9]+}}, [[TAGREG:[0-9]+]], 1
 ;CHECK-NEXT: bc 12, 1, .[[OPT1LABEL:[_0-9A-Za-z]+]]
 ;CHECK-NEXT: # %test2
-;CHECK-NEXT: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 30, 30
+;CHECK-NEXT: andi. {{[0-9]+}}, [[TAGREG]], 2
 ;CHECK-NEXT: bne 0, .[[OPT2LABEL:[_0-9A-Za-z]+]]
 ;CHECK-NEXT: .[[TEST3LABEL:[_0-9A-Za-z]+]]: # %test3
-;CHECK-NEXT: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 29, 29
+;CHECK-NEXT: andi. {{[0-9]+}}, [[TAGREG]], 4
 ;CHECK-NEXT: bne 0, .[[OPT3LABEL:[_0-9A-Za-z]+]]
 ;CHECK-NEXT: .[[TEST4LABEL:[_0-9A-Za-z]+]]: # %test4
-;CHECK-NEXT: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 28, 28
+;CHECK-NEXT: andi. {{[0-9]+}}, [[TAGREG]], 8
 ;CHECK-NEXT: bne 0, .[[OPT4LABEL:[_0-9A-Za-z]+]]
 ;CHECK-NEXT: .[[EXITLABEL:[_0-9A-Za-z]+]]: # %exit
 ;CHECK: blr
 ;CHECK-NEXT: .[[OPT1LABEL]]:
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 30, 30
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 2
 ;CHECK-NEXT: beq 0, .[[TEST3LABEL]]
 ;CHECK-NEXT: .[[OPT2LABEL]]:
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 29, 29
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 4
 ;CHECK-NEXT: beq 0, .[[TEST4LABEL]]
 ;CHECK-NEXT: .[[OPT3LABEL]]:
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 28, 28
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 8
 ;CHECK-NEXT: beq 0, .[[EXITLABEL]]
 ;CHECK-NEXT: .[[OPT4LABEL]]:
 ;CHECK: b .[[EXITLABEL]]
@@ -119,18 +119,18 @@ exit:
 ;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 1
 ;CHECK-NEXT: bc 12, 1, .[[OPT1LABEL:[_0-9A-Za-z]+]]
 ;CHECK-NEXT: # %test2
-;CHECK-NEXT: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 30, 30
+;CHECK-NEXT: andi. {{[0-9]+}}, [[TAGREG]], 2
 ;CHECK-NEXT: bne 0, .[[OPT2LABEL:[_0-9A-Za-z]+]]
 ;CHECK-NEXT: .[[TEST3LABEL:[_0-9A-Za-z]+]]: # %test3
-;CHECK-NEXT: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 29, 29
+;CHECK-NEXT: andi. {{[0-9]+}}, [[TAGREG]], 4
 ;CHECK-NEXT: bne 0, .[[OPT3LABEL:[_0-9A-Za-z]+]]
 ;CHECK-NEXT: .[[EXITLABEL:[_0-9A-Za-z]+]]: # %exit
 ;CHECK: blr
 ;CHECK-NEXT: .[[OPT1LABEL]]:
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 30, 30
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 2
 ;CHECK-NEXT: beq 0, .[[TEST3LABEL]]
 ;CHECK-NEXT: .[[OPT2LABEL]]:
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 29, 29
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 4
 ;CHECK-NEXT: beq 0, .[[EXITLABEL]]
 ;CHECK-NEXT: .[[OPT3LABEL]]:
 ;CHECK: b .[[EXITLABEL]]
@@ -285,23 +285,23 @@ exit:
 ;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 1
 ;CHECK-NEXT: bc 12, 1, .[[OPT1LABEL:[._0-9A-Za-z]+]]
 ;CHECK-NEXT: # %test2
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 30, 30
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 2
 ;CHECK-NEXT: bne 0, .[[OPT2LABEL:[._0-9A-Za-z]+]]
 ;CHECK-NEXT: .[[TEST3LABEL:[._0-9A-Za-z]+]]: # %test3
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 29, 29
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 4
 ;CHECK-NEXT: bne 0, .[[OPT3LABEL:[._0-9A-Za-z]+]]
 ;CHECK-NEXT: .[[TEST4LABEL:[._0-9A-Za-z]+]]: # %{{(test4|optional3)}}
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 28, 28
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 8
 ;CHECK-NEXT: beq 0, .[[LATCHLABEL]]
 ;CHECK-NEXT: b .[[OPT4LABEL:[._0-9A-Za-z]+]]
 ;CHECK: [[OPT1LABEL]]
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 30, 30
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 2
 ;CHECK-NEXT: beq 0, .[[TEST3LABEL]]
 ;CHECK-NEXT: .[[OPT2LABEL]]
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 29, 29
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 4
 ;CHECK-NEXT: beq 0, .[[TEST4LABEL]]
 ;CHECK-NEXT: .[[OPT3LABEL]]
-;CHECK: rlwinm. {{[0-9]+}}, [[TAGREG]], 0, 28, 28
+;CHECK: andi. {{[0-9]+}}, [[TAGREG]], 8
 ;CHECK-NEXT: beq 0, .[[LATCHLABEL]]
 ;CHECK: [[OPT4LABEL]]:
 ;CHECK: b .[[LATCHLABEL]]
@@ -375,12 +375,12 @@ exit:
 ; Make sure then2 falls through from test2
 ; CHECK-NOT: # %{{[-_a-zA-Z0-9]+}}
 ; CHECK: # %bb.{{[0-9]+}}: # %then2
-; CHECK: rlwinm. {{[0-9]+}}, {{[0-9]+}}, 0, 29, 29
+; CHECK: andi. {{[0-9]+}}, {{[0-9]+}}, 4
 ; CHECK: # %else1
 ; CHECK: bl a
 ; CHECK: bl a
 ; Make sure then2 was copied into else1
-; CHECK: rlwinm. {{[0-9]+}}, {{[0-9]+}}, 0, 29, 29
+; CHECK: andi. {{[0-9]+}}, {{[0-9]+}}, 4
 ; CHECK: # %end1
 ; CHECK: bl d
 ; CHECK: # %else2