mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-23 03:02:36 +01:00
Revert "[SCEV] Add nuw/nsw to mul ops in StrengthenNoWrapFlags"
This reverts r334428. It incorrectly marks some multiplications as nuw. Tim Shen is working on a proper fix. Original commit message: [SCEV] Add nuw/nsw to mul ops in StrengthenNoWrapFlags where safe. Summary: Previously we would add them for adds, but not multiplies. llvm-svn: 335016
This commit is contained in:
parent
868256cb42
commit
49958771f0
@ -2233,32 +2233,19 @@ StrengthenNoWrapFlags(ScalarEvolution *SE, SCEVTypes Type,
|
||||
|
||||
SignOrUnsignWrap = ScalarEvolution::maskFlags(Flags, SignOrUnsignMask);
|
||||
|
||||
if (SignOrUnsignWrap != SignOrUnsignMask &&
|
||||
(Type == scAddExpr || Type == scMulExpr) && Ops.size() == 2 &&
|
||||
isa<SCEVConstant>(Ops[0])) {
|
||||
if (SignOrUnsignWrap != SignOrUnsignMask && Type == scAddExpr &&
|
||||
Ops.size() == 2 && isa<SCEVConstant>(Ops[0])) {
|
||||
|
||||
auto Opcode = [&] {
|
||||
switch (Type) {
|
||||
case scAddExpr:
|
||||
return Instruction::Add;
|
||||
case scMulExpr:
|
||||
return Instruction::Mul;
|
||||
default:
|
||||
llvm_unreachable("Unexpected SCEV op.");
|
||||
}
|
||||
}();
|
||||
// (A + C) --> (A + C)<nsw> if the addition does not sign overflow
|
||||
// (A + C) --> (A + C)<nuw> if the addition does not unsign overflow
|
||||
|
||||
const APInt &C = cast<SCEVConstant>(Ops[0])->getAPInt();
|
||||
|
||||
// (A <opcode> C) --> (A <opcode> C)<nsw> if the op doesn't sign overflow.
|
||||
if (!(SignOrUnsignWrap & SCEV::FlagNSW)) {
|
||||
auto NSWRegion = ConstantRange::makeGuaranteedNoWrapRegion(
|
||||
Opcode, C, OBO::NoSignedWrap);
|
||||
Instruction::Add, C, OBO::NoSignedWrap);
|
||||
if (NSWRegion.contains(SE->getSignedRange(Ops[1])))
|
||||
Flags = ScalarEvolution::setFlags(Flags, SCEV::FlagNSW);
|
||||
}
|
||||
|
||||
// (A <opcode> C) --> (A <opcode> C)<nuw> if the op doesn't unsign overflow.
|
||||
if (!(SignOrUnsignWrap & SCEV::FlagNUW)) {
|
||||
auto NUWRegion = ConstantRange::makeGuaranteedNoWrapRegion(
|
||||
Instruction::Add, C, OBO::NoUnsignedWrap);
|
||||
|
@ -10,7 +10,7 @@
|
||||
; AddRec: {{{(28 + (4 * (-4 + (3 * %m)) * %o) + %A),+,(8 * %m * %o)}<%for.i>,+,(12 * %o)}<%for.j>,+,20}<%for.k>
|
||||
; CHECK: Base offset: %A
|
||||
; CHECK: ArrayDecl[UnknownSize][%m][%o] with elements of 4 bytes.
|
||||
; CHECK: ArrayRef[{3,+,2}<nw><%for.i>][{-4,+,3}<nw><%for.j>][{7,+,5}<nw><%for.k>]
|
||||
; CHECK: ArrayRef[{3,+,2}<%for.i>][{-4,+,3}<nw><%for.j>][{7,+,5}<nw><%for.k>]
|
||||
|
||||
define void @foo(i64 %n, i64 %m, i64 %o, i32* nocapture %A) #0 {
|
||||
entry:
|
||||
|
@ -11,7 +11,7 @@
|
||||
; AddRec: {{((%m * %b * 8) + %A),+,(2 * %m * 8)}<%for.i>,+,(2 * 8)}<%for.j>
|
||||
; CHECK: Base offset: %A
|
||||
; CHECK: ArrayDecl[UnknownSize][%m] with elements of 8 bytes.
|
||||
; CHECK: ArrayRef[{%b,+,2}<nsw><%for.i>][{0,+,2}<nuw><%for.j>]
|
||||
; CHECK: ArrayRef[{%b,+,2}<nsw><%for.i>][{0,+,2}<%for.j>]
|
||||
|
||||
|
||||
define void @foo(i64 %n, i64 %m, i64 %b, double* %A) {
|
||||
|
@ -70,7 +70,7 @@ exit:
|
||||
; sure they aren't marked as post-inc users.
|
||||
;
|
||||
; CHECK-LABEL: IV Users for loop %test2.loop
|
||||
; CHECK-NO-LCSSA: %sub.cond.us = ((-1 * %sub.us)<nuw><nsw> + {0,+,1}<nuw><nsw><%test2.loop>) (post-inc with loop %test2.loop) in %sext.us = mul i32 %mul.us, %sub.cond.us
|
||||
; CHECK-NO-LCSSA: %sub.cond.us = ((-1 * %sub.us)<nsw> + {0,+,1}<nuw><nsw><%test2.loop>) (post-inc with loop %test2.loop) in %sext.us = mul i32 %mul.us, %sub.cond.us
|
||||
define i32 @test2() {
|
||||
entry:
|
||||
br label %test2.loop
|
||||
|
@ -251,10 +251,10 @@ for.end: ; preds = %for.body
|
||||
; CHECK-NEXT: Member: {((2 * %offset) + %a)<nsw>,+,2}<nsw><%for.body>
|
||||
; CHECK-NEXT: Group {{.*}}[[ONE]]:
|
||||
; CHECK-NEXT: (Low: %a High: (10000 + %a))
|
||||
; CHECK-NEXT: Member: {%a,+,2}<nw><%for.body>
|
||||
; CHECK-NEXT: Member: {%a,+,2}<%for.body>
|
||||
; CHECK-NEXT: Group {{.*}}[[TWO]]:
|
||||
; CHECK-NEXT: (Low: (20000 + %a) High: (30000 + %a))
|
||||
; CHECK-NEXT: Member: {(20000 + %a),+,2}<nw><%for.body>
|
||||
; CHECK-NEXT: Member: {(20000 + %a),+,2}<%for.body>
|
||||
|
||||
define void @testi(i16* %a,
|
||||
i64 %offset) {
|
||||
|
@ -33,7 +33,7 @@ target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
|
||||
; i64 {0,+,2}<%for.body>
|
||||
|
||||
; LAA: [PSE] %arrayidxA = getelementptr i16, i16* %a, i64 %mul_ext:
|
||||
; LAA-NEXT: ((2 * (zext i32 {0,+,2}<%for.body> to i64))<nuw> + %a)
|
||||
; LAA-NEXT: ((2 * (zext i32 {0,+,2}<%for.body> to i64)) + %a)
|
||||
; LAA-NEXT: --> {%a,+,4}<%for.body>
|
||||
|
||||
|
||||
@ -130,7 +130,7 @@ for.end: ; preds = %for.body
|
||||
; i64 {zext i32 (2 * (trunc i64 %N to i32)) to i64,+,-2}<%for.body>
|
||||
|
||||
; LAA: [PSE] %arrayidxA = getelementptr i16, i16* %a, i64 %mul_ext:
|
||||
; LAA-NEXT: ((2 * (zext i32 {(2 * (trunc i64 %N to i32)),+,-2}<%for.body> to i64))<nuw> + %a)
|
||||
; LAA-NEXT: ((2 * (zext i32 {(2 * (trunc i64 %N to i32)),+,-2}<%for.body> to i64)) + %a)
|
||||
; LAA-NEXT: --> {((4 * (zext i31 (trunc i64 %N to i31) to i64)) + %a),+,-4}<%for.body>
|
||||
|
||||
; LV-LABEL: f2
|
||||
|
@ -322,7 +322,7 @@ define void @test_05(i32 %N) {
|
||||
; CHECK: %SQ = mul i32 %i.0, %i.0
|
||||
; CHECK-NEXT: --> {4,+,5,+,2}<%bb3>
|
||||
; CHECK: %tmp4 = mul i32 %i.0, 2
|
||||
; CHECK-NEXT: --> {4,+,2}<nuw><%bb3>
|
||||
; CHECK-NEXT: --> {4,+,2}<%bb3>
|
||||
; CHECK: %tmp5 = sub i32 %SQ, %tmp4
|
||||
; CHECK-NEXT: --> {0,+,3,+,2}<%bb3>
|
||||
|
||||
|
@ -1,15 +0,0 @@
|
||||
; RUN: opt < %s -analyze -scalar-evolution | FileCheck %s
|
||||
|
||||
; Check that we add nuw to multiplies by a constant where we can infer that the
|
||||
; multiply does not have unsigned overflow.
|
||||
declare i32 @get_int();
|
||||
|
||||
define void @foo() {
|
||||
%a = call i32 @get_int(), !range !0
|
||||
%b = mul i32 %a, 4
|
||||
; CHECK: %b
|
||||
; CHECK-NEXT: --> (4 * %a)<nuw>
|
||||
ret void
|
||||
}
|
||||
|
||||
!0 = !{i32 0, i32 100}
|
@ -79,5 +79,5 @@ declare void @llvm.assume(i1) nounwind
|
||||
|
||||
; Note: Without the preheader assume, there is an 'smax' in the
|
||||
; backedge-taken count expression:
|
||||
; CHECK: Loop %bb: backedge-taken count is ((-1 + (2 * (%no /u 2))<nuw>) /u 2)
|
||||
; CHECK: Loop %bb: backedge-taken count is ((-1 + (2 * (%no /u 2))) /u 2)
|
||||
; CHECK: Loop %bb: max backedge-taken count is 1073741822
|
||||
|
@ -73,5 +73,5 @@ return: ; preds = %bb1.return_crit_edg
|
||||
ret void
|
||||
}
|
||||
|
||||
; CHECK: Loop %bb: backedge-taken count is ((-1 + (2 * (%no /u 2))<nuw>) /u 2)
|
||||
; CHECK: Loop %bb: backedge-taken count is ((-1 + (2 * (%no /u 2))) /u 2)
|
||||
; CHECK: Loop %bb: max backedge-taken count is 1073741822
|
||||
|
@ -25,7 +25,7 @@ bb: ; preds = %bb1, %bb.nph
|
||||
%tmp6 = sext i32 %i.01 to i64 ; <i64> [#uses=1]
|
||||
%tmp7 = getelementptr double, double* %p, i64 %tmp6 ; <double*> [#uses=1]
|
||||
; CHECK: %tmp7
|
||||
; CHECK-NEXT: --> {%p,+,8}<nw><%bb>
|
||||
; CHECK-NEXT: --> {%p,+,8}<%bb>
|
||||
store double %tmp5, double* %tmp7, align 8
|
||||
%tmp8 = add nsw i32 %i.01, 1 ; <i32> [#uses=2]
|
||||
; CHECK: %tmp8
|
||||
@ -126,7 +126,7 @@ exit:
|
||||
}
|
||||
|
||||
; CHECK-LABEL: PR12375
|
||||
; CHECK: --> {(4 + %arg)<nsw>,+,4}<nuw><%bb1>{{ U: [^ ]+ S: [^ ]+}}{{ *}}Exits: (4 + (4 * ((-1 + (-1 * %arg) + ((4 + %arg)<nsw> umax (8 + %arg)<nsw>)) /u 4))<nuw> + %arg)
|
||||
; CHECK: --> {(4 + %arg)<nsw>,+,4}<nuw><%bb1>{{ U: [^ ]+ S: [^ ]+}}{{ *}}Exits: (4 + (4 * ((-1 + (-1 * %arg) + ((4 + %arg)<nsw> umax (8 + %arg)<nsw>)) /u 4)) + %arg)
|
||||
define i32 @PR12375(i32* readnone %arg) {
|
||||
bb:
|
||||
%tmp = getelementptr inbounds i32, i32* %arg, i64 2
|
||||
@ -145,7 +145,7 @@ bb7: ; preds = %bb1
|
||||
}
|
||||
|
||||
; CHECK-LABEL: PR12376
|
||||
; CHECK: --> {(4 + %arg)<nsw>,+,4}<nuw><%bb2>{{ U: [^ ]+ S: [^ ]+}}{{ *}}Exits: (4 + (4 * ((-1 + (-1 * %arg) + ((4 + %arg)<nsw> umax %arg1)) /u 4))<nuw> + %arg)
|
||||
; CHECK: --> {(4 + %arg)<nsw>,+,4}<nuw><%bb2>{{ U: [^ ]+ S: [^ ]+}}{{ *}}Exits: (4 + (4 * ((-1 + (-1 * %arg) + ((4 + %arg)<nsw> umax %arg1)) /u 4)) + %arg)
|
||||
define void @PR12376(i32* nocapture %arg, i32* nocapture %arg1) {
|
||||
bb:
|
||||
br label %bb2
|
||||
|
@ -1,7 +1,7 @@
|
||||
; RUN: opt < %s -analyze -scalar-evolution | FileCheck %s
|
||||
|
||||
; CHECK: %tmp9 = shl i64 %tmp8, 33
|
||||
; CHECK-NEXT: --> {{.*}} Exits: (-8589934592 + (8589934592 * (zext i32 %arg2 to i64))<nuw>)
|
||||
; CHECK-NEXT: --> {{.*}} Exits: (-8589934592 + (8589934592 * (zext i32 %arg2 to i64)))
|
||||
; CHECK: %tmp10 = ashr exact i64 %tmp9, 32
|
||||
; CHECK-NEXT: --> {{.*}} Exits: (sext i32 (-2 + (2 * %arg2)) to i64)
|
||||
; CHECK: %tmp11 = getelementptr inbounds i32, i32* %arg, i64 %tmp10
|
||||
@ -48,9 +48,9 @@ bb7: ; preds = %bb7, %bb3
|
||||
}
|
||||
|
||||
; CHECK: %t10 = ashr exact i128 %t9, 1
|
||||
; CHECK-NEXT: --> {{.*}} Exits: (sext i127 (-633825300114114700748351602688 + (633825300114114700748351602688 * (zext i32 %arg5 to i127))<nuw>) to i128)
|
||||
; CHECK-NEXT: --> {{.*}} Exits: (sext i127 (-633825300114114700748351602688 + (633825300114114700748351602688 * (zext i32 %arg5 to i127))) to i128)
|
||||
; CHECK: %t14 = or i128 %t10, 1
|
||||
; CHECK-NEXT: --> {{.*}} Exits: (1 + (sext i127 (-633825300114114700748351602688 + (633825300114114700748351602688 * (zext i32 %arg5 to i127))<nuw>) to i128))<nsw>
|
||||
; CHECK-NEXT: --> {{.*}} Exits: (1 + (sext i127 (-633825300114114700748351602688 + (633825300114114700748351602688 * (zext i32 %arg5 to i127))) to i128))<nsw>
|
||||
; CHECK: Loop %bb7: backedge-taken count is (-1 + (zext i32 %arg5 to i128))<nsw>
|
||||
; CHECK-NEXT: Loop %bb7: max backedge-taken count is -1
|
||||
; CHECK-NEXT: Loop %bb7: Predicated backedge-taken count is (-1 + (zext i32 %arg5 to i128))<nsw>
|
||||
|
@ -1,9 +1,9 @@
|
||||
; RUN: opt < %s -analyze -scalar-evolution | FileCheck %s
|
||||
|
||||
; CHECK: %tmp9 = shl i64 %tmp8, 33
|
||||
; CHECK-NEXT: --> {{.*}} Exits: (-8589934592 + (8589934592 * (zext i32 %arg2 to i64))<nuw>)
|
||||
; CHECK-NEXT: --> {{.*}} Exits: (-8589934592 + (8589934592 * (zext i32 %arg2 to i64)))
|
||||
; CHECK-NEXT: %tmp10 = ashr exact i64 %tmp9, 0
|
||||
; CHECK-NEXT: --> {{.*}} Exits: (-8589934592 + (8589934592 * (zext i32 %arg2 to i64))<nuw>)
|
||||
; CHECK-NEXT: --> {{.*}} Exits: (-8589934592 + (8589934592 * (zext i32 %arg2 to i64)))
|
||||
|
||||
define void @foo(i32* nocapture %arg, i32 %arg1, i32 %arg2) {
|
||||
bb:
|
||||
|
@ -31,7 +31,7 @@ exit:
|
||||
ret i32 %i
|
||||
|
||||
; CHECK-LABEL: @test2
|
||||
; CHECK: Loop %loop: backedge-taken count is ((-32 + (32 * (%n /u 32))<nuw>) /u 32)
|
||||
; CHECK: Loop %loop: backedge-taken count is ((-32 + (32 * (%n /u 32))) /u 32)
|
||||
; CHECK: Loop %loop: max backedge-taken count is 134217727
|
||||
}
|
||||
|
||||
|
@ -19,12 +19,15 @@ define void @no_range() {
|
||||
}
|
||||
|
||||
; CHECK-LABEL: @range
|
||||
;
|
||||
; This had to be disabled when r334428 was reverted. We should enable this test
|
||||
; when r334428 is reapplied with a fix.
|
||||
define void @range() {
|
||||
%a = call i32 @get_int(), !range !0
|
||||
%b = mul i32 %a, 4
|
||||
%c = zext i32 %b to i64
|
||||
; CHECK: %c
|
||||
; CHECK-NEXT: --> (4 * (zext i32 %a to i64))<nuw>
|
||||
; CHECK-NEXT: --> (zext i32 (4 * %a) to i64)
|
||||
ret void
|
||||
}
|
||||
|
||||
|
@ -22,7 +22,7 @@ for.body153: ; preds = %for.body153, %for.b
|
||||
; CHECK: add nuw nsw i64 %indvars.iv, 1
|
||||
; CHECK: sub nsw i64 %indvars.iv, 2
|
||||
; CHECK: sub nsw i64 4, %indvars.iv
|
||||
; CHECK: mul nuw nsw i64 %indvars.iv, 8
|
||||
; CHECK: mul nsw i64 %indvars.iv, 8
|
||||
for.body170: ; preds = %for.body170, %for.body153
|
||||
%i2.19 = phi i32 [ %add249, %for.body170 ], [ 0, %for.body153 ]
|
||||
|
||||
|
Loading…
Reference in New Issue
Block a user