Skip to content

[SCEV] Try to prove no-wrap for AddRecs via BTC. #131538

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

[SCEV] Try to prove no-wrap for AddRecs via BTC. #131538

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
37 changes: 37 additions & 0 deletions llvm/lib/Analysis/ScalarEvolution.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5101,6 +5101,37 @@ ScalarEvolution::proveNoWrapViaConstantRanges(const SCEVAddRecExpr *AR) {
return Result;
}

/// Return true if \p AR is known to not wrap via the loop's backedge-taken
/// count.
static SCEV::NoWrapFlags proveNoWrapViaBTC(const SCEVAddRecExpr *AR,
ScalarEvolution &SE) {
SCEV::NoWrapFlags Result = SCEV::FlagAnyWrap;
if (AR->hasNoUnsignedWrap() && AR->hasNoSignedWrap())
return Result;

const Loop *L = AR->getLoop();
const SCEV *BTC = SE.getBackedgeTakenCount(L);
if (isa<SCEVCouldNotCompute>(BTC) ||
!match(AR->getStepRecurrence(SE), m_scev_One()))
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can use isOne here.

return Result;

auto *WTy = SE.getWiderType(AR->getType(), BTC->getType());
// If AR's type is wider than BTC, we can zero extend BTC, otherwise bail out.
if (WTy != AR->getType())
return Result;

// AR has a step of 1, it is NUW/NSW if Start + BTC >= Start.
auto *Add = SE.getAddExpr(AR->getStart(), SE.getNoopOrZeroExtend(BTC, WTy));
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

AR->evaluateAtIteration(BTC, SE)?

if (!AR->hasNoUnsignedWrap() &&
SE.isKnownPredicate(CmpInst::ICMP_UGE, Add, AR->getStart()))
Result = ScalarEvolution::setFlags(Result, SCEV::FlagNUW);
if (!AR->hasNoSignedWrap() &&
SE.isKnownPredicate(CmpInst::ICMP_SGE, Add, AR->getStart()))
Result = ScalarEvolution::setFlags(Result, SCEV::FlagNSW);

return Result;
}

SCEV::NoWrapFlags
ScalarEvolution::proveNoSignedWrapViaInduction(const SCEVAddRecExpr *AR) {
SCEV::NoWrapFlags Result = AR->getNoWrapFlags();
Expand Down Expand Up @@ -5761,6 +5792,9 @@ const SCEV *ScalarEvolution::createSimpleAffineAddRec(PHINode *PN,
setNoWrapFlags(const_cast<SCEVAddRecExpr *>(AR),
(SCEV::NoWrapFlags)(AR->getNoWrapFlags() |
proveNoWrapViaConstantRanges(AR)));
setNoWrapFlags(const_cast<SCEVAddRecExpr *>(AR),
(SCEV::NoWrapFlags)(AR->getNoWrapFlags() |
proveNoWrapViaBTC(AR, *this)));
}

// We can add Flags to the post-inc expression only if we
Expand Down Expand Up @@ -5892,6 +5926,9 @@ const SCEV *ScalarEvolution::createAddRecFromPHI(PHINode *PN) {
setNoWrapFlags(const_cast<SCEVAddRecExpr *>(AR),
(SCEV::NoWrapFlags)(AR->getNoWrapFlags() |
proveNoWrapViaConstantRanges(AR)));
setNoWrapFlags(const_cast<SCEVAddRecExpr *>(AR),
(SCEV::NoWrapFlags)(AR->getNoWrapFlags() |
proveNoWrapViaBTC(AR, *this)));
}

// We can add Flags to the post-inc expression only if we
Expand Down
2 changes: 1 addition & 1 deletion llvm/test/Analysis/ScalarEvolution/addrec-computed-during-addrec-calculation.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,7 @@ define void @test(ptr %p) {
; CHECK-NEXT: %iv2.ext = sext i32 %iv2 to i64
; CHECK-NEXT: --> (sext i32 {%iv,+,1}<%loop2> to i64) U: [-2147483648,2147483648) S: [-2147483648,2147483648) Exits: <<Unknown>> LoopDispositions: { %loop.header: Variant, %loop2: Computable, %loop3: Invariant }
; CHECK-NEXT: %iv3 = phi i64 [ %iv2.ext, %loop2.end ], [ %iv3.next, %loop3 ]
; CHECK-NEXT: --> {(sext i32 {%iv,+,1}<%loop2> to i64),+,1}<nsw><%loop3> U: [-2147483648,2147483648) S: [-2147483648,2147483648) Exits: (sext i32 {%iv,+,1}<%loop2> to i64) LoopDispositions: { %loop3: Computable, %loop.header: Variant }
; CHECK-NEXT: --> {(sext i32 {%iv,+,1}<%loop2> to i64),+,1}<nuw><nsw><%loop3> U: [-2147483648,2147483648) S: [-2147483648,2147483648) Exits: (sext i32 {%iv,+,1}<%loop2> to i64) LoopDispositions: { %loop3: Computable, %loop.header: Variant }
; CHECK-NEXT: %iv3.next = add nsw i64 %iv3, 1
; CHECK-NEXT: --> {(1 + (sext i32 {%iv,+,1}<%loop2> to i64))<nsw>,+,1}<nsw><%loop3> U: [-2147483647,2147483649) S: [-2147483647,2147483649) Exits: (1 + (sext i32 {%iv,+,1}<%loop2> to i64))<nsw> LoopDispositions: { %loop3: Computable, %loop.header: Variant }
; CHECK-NEXT: %iv.next = trunc i64 %iv3 to i32
Expand Down
26 changes: 13 additions & 13 deletions llvm/test/Analysis/ScalarEvolution/different-loops-recs.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -332,21 +332,21 @@ define void @test_03(i32 %a, i32 %b, i32 %c, ptr %p) {
; CHECK-LABEL: 'test_03'
; CHECK-NEXT: Classifying expressions for: @test_03
; CHECK-NEXT: %phi1 = phi i32 [ %a, %entry ], [ %phi1.inc, %loop1 ]
; CHECK-NEXT: --> {%a,+,1}<%loop1> U: full-set S: full-set Exits: (%a umax %c) LoopDispositions: { %loop1: Computable }
; CHECK-NEXT: --> {%a,+,1}<nuw><%loop1> U: full-set S: full-set Exits: (%a umax %c) LoopDispositions: { %loop1: Computable }
; CHECK-NEXT: %phi1.inc = add i32 %phi1, 1
; CHECK-NEXT: --> {(1 + %a),+,1}<%loop1> U: full-set S: full-set Exits: (1 + (%a umax %c)) LoopDispositions: { %loop1: Computable }
; CHECK-NEXT: --> {(1 + %a),+,1}<nw><%loop1> U: full-set S: full-set Exits: (1 + (%a umax %c)) LoopDispositions: { %loop1: Computable }
; CHECK-NEXT: %phi2 = phi i32 [ %a, %loop1 ], [ %phi2.inc, %loop2 ]
; CHECK-NEXT: --> {%a,+,2}<%loop2> U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %loop2: Computable }
; CHECK-NEXT: %phi2.inc = add i32 %phi2, 2
; CHECK-NEXT: --> {(2 + %a),+,2}<%loop2> U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %loop2: Computable }
; CHECK-NEXT: %v1 = load i32, ptr %p, align 4
; CHECK-NEXT: --> %v1 U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %loop2: Variant }
; CHECK-NEXT: %s1 = add i32 %phi1, %v1
; CHECK-NEXT: --> ({%a,+,1}<%loop1> + %v1) U: full-set S: full-set --> ((%a umax %c) + %v1) U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %loop2: Variant }
; CHECK-NEXT: --> ({%a,+,1}<nuw><%loop1> + %v1) U: full-set S: full-set --> ((%a umax %c) + %v1) U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %loop2: Variant }
; CHECK-NEXT: %s2 = add i32 %s1, %b
; CHECK-NEXT: --> ({(%a + %b),+,1}<%loop1> + %v1) U: full-set S: full-set --> ((%a umax %c) + %b + %v1) U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %loop2: Variant }
; CHECK-NEXT: --> ({(%a + %b),+,1}<nw><%loop1> + %v1) U: full-set S: full-set --> ((%a umax %c) + %b + %v1) U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %loop2: Variant }
; CHECK-NEXT: %s3 = add i32 %s2, %phi2
; CHECK-NEXT: --> ({{\{\{}}((2 * %a) + %b),+,1}<%loop1>,+,2}<%loop2> + %v1) U: full-set S: full-set --> ({((%a umax %c) + %a + %b),+,2}<%loop2> + %v1) U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %loop2: Variant }
; CHECK-NEXT: --> ({{\{\{}}((2 * %a) + %b),+,1}<nw><%loop1>,+,2}<%loop2> + %v1) U: full-set S: full-set --> ({((%a umax %c) + %a + %b),+,2}<%loop2> + %v1) U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %loop2: Variant }
; CHECK-NEXT: Determining loop execution counts for: @test_03
; CHECK-NEXT: Loop %loop2: Unpredictable backedge-taken count.
; CHECK-NEXT: Loop %loop2: Unpredictable constant max backedge-taken count.
Expand Down Expand Up @@ -757,13 +757,13 @@ define i64 @test_09(i32 %param) {
; CHECK-NEXT: %iv1.next = add nuw nsw i64 %iv1, 1
; CHECK-NEXT: --> {1,+,1}<nuw><nsw><%loop1> U: [1,4) S: [1,4) Exits: 3 LoopDispositions: { %loop1: Computable }
; CHECK-NEXT: %iv2 = phi i32 [ %iv2.next, %loop2 ], [ %param, %loop2.preheader ]
; CHECK-NEXT: --> {%param,+,1}<%loop2> U: full-set S: full-set Exits: (2 smax %param) LoopDispositions: { %loop2: Computable }
; CHECK-NEXT: --> {%param,+,1}<nsw><%loop2> U: full-set S: full-set Exits: (2 smax %param) LoopDispositions: { %loop2: Computable }
; CHECK-NEXT: %iv2.next = add i32 %iv2, 1
; CHECK-NEXT: --> {(1 + %param),+,1}<%loop2> U: full-set S: full-set Exits: (1 + (2 smax %param))<nuw> LoopDispositions: { %loop2: Computable }
; CHECK-NEXT: --> {(1 + %param),+,1}<nw><%loop2> U: full-set S: full-set Exits: (1 + (2 smax %param))<nuw> LoopDispositions: { %loop2: Computable }
; CHECK-NEXT: %iv2.ext = sext i32 %iv2.next to i64
; CHECK-NEXT: --> (sext i32 {(1 + %param),+,1}<%loop2> to i64) U: [-2147483648,2147483648) S: [-2147483648,2147483648) --> (sext i32 (1 + (2 smax %param))<nuw> to i64) U: [-2147483648,2147483648) S: [-2147483648,2147483648)
; CHECK-NEXT: --> (sext i32 {(1 + %param),+,1}<nw><%loop2> to i64) U: [-2147483648,2147483648) S: [-2147483648,2147483648) --> (sext i32 (1 + (2 smax %param))<nuw> to i64) U: [-2147483648,2147483648) S: [-2147483648,2147483648)
; CHECK-NEXT: %ret = mul i64 %iv1, %iv2.ext
; CHECK-NEXT: --> ((sext i32 {(1 + %param),+,1}<%loop2> to i64) * {0,+,1}<nuw><nsw><%loop1>) U: [-4294967296,4294967295) S: [-4294967296,4294967295) --> (2 * (sext i32 (1 + (2 smax %param))<nuw> to i64))<nsw> U: [0,-1) S: [-4294967296,4294967295)
; CHECK-NEXT: --> ((sext i32 {(1 + %param),+,1}<nw><%loop2> to i64) * {0,+,1}<nuw><nsw><%loop1>) U: [-4294967296,4294967295) S: [-4294967296,4294967295) --> (2 * (sext i32 (1 + (2 smax %param))<nuw> to i64))<nsw> U: [0,-1) S: [-4294967296,4294967295)
; CHECK-NEXT: Determining loop execution counts for: @test_09
; CHECK-NEXT: Loop %loop2: backedge-taken count is ((-1 * %param) + (2 smax %param))
; CHECK-NEXT: Loop %loop2: constant max backedge-taken count is i32 -2147483646
Expand Down Expand Up @@ -828,13 +828,13 @@ define i64 @test_10(i32 %param) {
; CHECK-NEXT: %uncle.outer.next = add i64 %uncle, 1
; CHECK-NEXT: --> {1,+,1}<nuw><nsw><%uncle.loop> U: [1,2) S: [1,2) Exits: <<Unknown>> LoopDispositions: { %uncle.loop: Computable, %loop1: Invariant }
; CHECK-NEXT: %iv2 = phi i32 [ %iv2.next, %loop2 ], [ %param, %loop2.preheader ]
; CHECK-NEXT: --> {%param,+,1}<%loop2> U: full-set S: full-set Exits: (2 smax %param) LoopDispositions: { %loop2: Computable }
; CHECK-NEXT: --> {%param,+,1}<nsw><%loop2> U: full-set S: full-set Exits: (2 smax %param) LoopDispositions: { %loop2: Computable }
; CHECK-NEXT: %iv2.next = add i32 %iv2, 1
; CHECK-NEXT: --> {(1 + %param),+,1}<%loop2> U: full-set S: full-set Exits: (1 + (2 smax %param))<nuw> LoopDispositions: { %loop2: Computable }
; CHECK-NEXT: --> {(1 + %param),+,1}<nw><%loop2> U: full-set S: full-set Exits: (1 + (2 smax %param))<nuw> LoopDispositions: { %loop2: Computable }
; CHECK-NEXT: %iv2.ext = sext i32 %iv2.next to i64
; CHECK-NEXT: --> (sext i32 {(1 + %param),+,1}<%loop2> to i64) U: [-2147483648,2147483648) S: [-2147483648,2147483648) --> (sext i32 (1 + (2 smax %param))<nuw> to i64) U: [-2147483648,2147483648) S: [-2147483648,2147483648)
; CHECK-NEXT: --> (sext i32 {(1 + %param),+,1}<nw><%loop2> to i64) U: [-2147483648,2147483648) S: [-2147483648,2147483648) --> (sext i32 (1 + (2 smax %param))<nuw> to i64) U: [-2147483648,2147483648) S: [-2147483648,2147483648)
; CHECK-NEXT: %ret = mul i64 %iv1, %iv2.ext
; CHECK-NEXT: --> ((sext i32 {(1 + %param),+,1}<%loop2> to i64) * {0,+,1}<nuw><nsw><%loop1>) U: [-4294967296,4294967295) S: [-4294967296,4294967295) --> (2 * (sext i32 (1 + (2 smax %param))<nuw> to i64))<nsw> U: [0,-1) S: [-4294967296,4294967295)
; CHECK-NEXT: --> ((sext i32 {(1 + %param),+,1}<nw><%loop2> to i64) * {0,+,1}<nuw><nsw><%loop1>) U: [-4294967296,4294967295) S: [-4294967296,4294967295) --> (2 * (sext i32 (1 + (2 smax %param))<nuw> to i64))<nsw> U: [0,-1) S: [-4294967296,4294967295)
; CHECK-NEXT: Determining loop execution counts for: @test_10
; CHECK-NEXT: Loop %loop2: backedge-taken count is ((-1 * %param) + (2 smax %param))
; CHECK-NEXT: Loop %loop2: constant max backedge-taken count is i32 -2147483646
Expand Down
Loading
Loading