summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorSanjay Patel <spatel@rotateright.com>2018-09-11 22:00:15 +0000
committerSanjay Patel <spatel@rotateright.com>2018-09-11 22:00:15 +0000
commit26725bdc508f0c011e39f7ef0125e961d2ad76ed (patch)
treea605d8148f3cf67598cb64ec4f2e65009d6667ef
parentc72a7259bee5779f8627bbf464f53c0c993274e0 (diff)
downloadbcm5719-llvm-26725bdc508f0c011e39f7ef0125e961d2ad76ed.tar.gz
bcm5719-llvm-26725bdc508f0c011e39f7ef0125e961d2ad76ed.zip
[InstCombine] add folds for icmp with xor mask constant
These are the folds in Alive; Name: xor_ult Pre: isPowerOf2(-C1) %xor = xor i8 %x, C1 %r = icmp ult i8 %xor, C1 => %r = icmp ugt i8 %x, ~C1 Name: xor_ugt Pre: isPowerOf2(C1+1) %xor = xor i8 %x, C1 %r = icmp ugt i8 %xor, C1 => %r = icmp ugt i8 %x, C1 https://rise4fun.com/Alive/Vty The ugt case in its simplest form was already handled by DemandedBits, but that's not ideal as shown in the multi-use test. I'm not sure if these are all of the symmetrical folds, but I adjusted the existing code for one of the folds to try to show the similarities. There's no obvious connection, but this is another preliminary step for PR14613... https://bugs.llvm.org/show_bug.cgi?id=14613 llvm-svn: 341997
-rw-r--r--llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp29
-rw-r--r--llvm/test/Transforms/InstCombine/icmp.ll7
2 files changed, 22 insertions, 14 deletions
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index 4b5891461bc..e8e9e86d08b 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -1499,16 +1499,25 @@ Instruction *InstCombiner::foldICmpXorConstant(ICmpInst &Cmp,
}
}
- // (icmp ugt (xor X, C), ~C) -> (icmp ult X, C)
- // iff -C is a power of 2
- if (Pred == ICmpInst::ICMP_UGT && *XorC == ~C && (C + 1).isPowerOf2())
- return new ICmpInst(ICmpInst::ICMP_ULT, X, Y);
-
- // (icmp ult (xor X, C), -C) -> (icmp uge X, C)
- // iff -C is a power of 2
- if (Pred == ICmpInst::ICMP_ULT && *XorC == -C && C.isPowerOf2())
- return new ICmpInst(ICmpInst::ICMP_UGE, X, Y);
-
+ // Mask constant magic can eliminate an 'xor' with unsigned compares.
+ if (Pred == ICmpInst::ICMP_UGT) {
+ // (xor X, ~C) >u C --> X <u ~C (when C+1 is a power of 2)
+ if (*XorC == ~C && (C + 1).isPowerOf2())
+ return new ICmpInst(ICmpInst::ICMP_ULT, X, Y);
+ // (xor X, C) >u C --> X >u C (when C+1 is a power of 2)
+ if (*XorC == C && (C + 1).isPowerOf2())
+ return new ICmpInst(ICmpInst::ICMP_UGT, X, Y);
+ }
+ if (Pred == ICmpInst::ICMP_ULT) {
+ // (xor X, -C) <u C --> X >u ~C (when C is a power of 2)
+ if (*XorC == -C && C.isPowerOf2())
+ return new ICmpInst(ICmpInst::ICMP_UGT, X,
+ ConstantInt::get(X->getType(), ~C));
+ // (xor X, C) <u C --> X >u ~C (when -C is a power of 2)
+ if (*XorC == C && (-C).isPowerOf2())
+ return new ICmpInst(ICmpInst::ICMP_UGT, X,
+ ConstantInt::get(X->getType(), ~C));
+ }
return nullptr;
}
diff --git a/llvm/test/Transforms/InstCombine/icmp.ll b/llvm/test/Transforms/InstCombine/icmp.ll
index 546ed9ba2d0..623f401b718 100644
--- a/llvm/test/Transforms/InstCombine/icmp.ll
+++ b/llvm/test/Transforms/InstCombine/icmp.ll
@@ -2229,8 +2229,7 @@ define <2 x i1> @icmp_xor_neg4_X_uge_4_vec(<2 x i32> %X) {
define <2 x i1> @xor_ult(<2 x i8> %x) {
; CHECK-LABEL: @xor_ult(
-; CHECK-NEXT: [[XOR:%.*]] = xor <2 x i8> [[X:%.*]], <i8 -4, i8 -4>
-; CHECK-NEXT: [[R:%.*]] = icmp ult <2 x i8> [[XOR]], <i8 -4, i8 -4>
+; CHECK-NEXT: [[R:%.*]] = icmp ugt <2 x i8> [[X:%.*]], <i8 3, i8 3>
; CHECK-NEXT: ret <2 x i1> [[R]]
;
%xor = xor <2 x i8> %x, <i8 -4, i8 -4>
@@ -2242,7 +2241,7 @@ define i1 @xor_ult_extra_use(i8 %x, i8* %p) {
; CHECK-LABEL: @xor_ult_extra_use(
; CHECK-NEXT: [[XOR:%.*]] = xor i8 [[X:%.*]], -32
; CHECK-NEXT: store i8 [[XOR]], i8* [[P:%.*]], align 1
-; CHECK-NEXT: [[R:%.*]] = icmp ult i8 [[XOR]], -32
+; CHECK-NEXT: [[R:%.*]] = icmp ugt i8 [[X]], 31
; CHECK-NEXT: ret i1 [[R]]
;
%xor = xor i8 %x, -32
@@ -2265,7 +2264,7 @@ define i1 @xor_ugt_extra_use(i8 %x, i8* %p) {
; CHECK-LABEL: @xor_ugt_extra_use(
; CHECK-NEXT: [[XOR:%.*]] = xor i8 [[X:%.*]], 63
; CHECK-NEXT: store i8 [[XOR]], i8* [[P:%.*]], align 1
-; CHECK-NEXT: [[R:%.*]] = icmp ugt i8 [[XOR]], 63
+; CHECK-NEXT: [[R:%.*]] = icmp ugt i8 [[X]], 63
; CHECK-NEXT: ret i1 [[R]]
;
%xor = xor i8 %x, 63
OpenPOWER on IntegriCloud