Skip to content

[BasicAA][ValueTracking] Increase depth for underlying object search #143714

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

Jump to bottom
Merged
merged 1 commit into from
Jun 12, 2025
Merged

[BasicAA][ValueTracking] Increase depth for underlying object search #143714

merged 1 commit into from
Jun 12, 2025

Conversation

@nikic
Copy link
Contributor

@nikic nikic commented Jun 11, 2025

This depth limits a linear search (rather than the usual potentially exponential one) and is not particularly important for compile-time in practice.

The change in #137297 is going to increase the length of GEP chains, so I'd like to increase this limit a bit to reduce the chance of regressions (dtcxzyw/llvm-opt-benchmark#2419 showed a 13% increase in SearchLimitReached). There is no particular significance to the new value of 10.

Compile-time looks neutral: https://llvm-compile-time-tracker.com/compare.php?from=a72bcda1434c72f9db6687565a361479e0dde572&to=f5129a63ce2c543b07de49b0308d2adb2b600582&stat=instructions:u

@nikic
[BasicAA][ValueTracking] Increase depth for underlying object search
366c671 
This depth limits a linear search (rather than the usual potentially
exponential one) and is not particularly important for compile-time
in practice.

The change in llvm#137297 is going to increase the length of GEP chains,
so I'd like to increase this limit a bit to reduce the chance of
regressions.
@nikic nikic requested review from dtcxzyw and fhahn June 11, 2025 14:42
@llvmbot llvmbot added llvm:analysis Includes value tracking, cost tables and constant folding llvm:transforms labels Jun 11, 2025
@llvmbot
Copy link
Member

llvmbot commented Jun 11, 2025

@llvm/pr-subscribers-llvm-analysis

Author: Nikita Popov (nikic)

Changes

This depth limits a linear search (rather than the usual potentially exponential one) and is not particularly important for compile-time in practice.

The change in #137297 is going to increase the length of GEP chains, so I'd like to increase this limit a bit to reduce the chance of regressions (dtcxzyw/llvm-opt-benchmark#2419 showed a 13% increase in SearchLimitReached). There is no particular significance to the new value of 10.

Compile-time looks neutral: https://llvm-compile-time-tracker.com/compare.php?from=a72bcda1434c72f9db6687565a361479e0dde572&to=f5129a63ce2c543b07de49b0308d2adb2b600582&stat=instructions:u

Full diff: https://github.com/llvm/llvm-project/pull/143714.diff

4 Files Affected:

  • (modified) llvm/include/llvm/Analysis/ValueTracking.h (+1-1)
  • (modified) llvm/test/Analysis/BasicAA/gep-decomposition-limit.ll (+23-15)
  • (modified) llvm/test/Analysis/LoopAccessAnalysis/underlying-objects-2.ll (+4-1)
  • (modified) llvm/test/Transforms/Inline/inline-noalias-unidentify-object.ll (+14-8)
diff --git a/llvm/include/llvm/Analysis/ValueTracking.h b/llvm/include/llvm/Analysis/ValueTracking.h
index 32ab9733d13c9..e215c90b5a72a 100644
--- a/llvm/include/llvm/Analysis/ValueTracking.h
+++ b/llvm/include/llvm/Analysis/ValueTracking.h
@@ -47,7 +47,7 @@ constexpr unsigned MaxAnalysisRecursionDepth = 6;
 
 /// The max limit of the search depth in DecomposeGEPExpression() and
 /// getUnderlyingObject().
-constexpr unsigned MaxLookupSearchDepth = 6;
+constexpr unsigned MaxLookupSearchDepth = 10;
 
 /// Determine which bits of V are known to be either zero or one and return
 /// them in the KnownZero/KnownOne bit sets.
diff --git a/llvm/test/Analysis/BasicAA/gep-decomposition-limit.ll b/llvm/test/Analysis/BasicAA/gep-decomposition-limit.ll
index 23a96ebca8485..a256ececbe565 100644
--- a/llvm/test/Analysis/BasicAA/gep-decomposition-limit.ll
+++ b/llvm/test/Analysis/BasicAA/gep-decomposition-limit.ll
@@ -2,22 +2,22 @@
 
 ; CHECK-LABEL: Function: test
 ;; Before limit:
-; CHECK-DAG: MustAlias: i8* %gep.add5, i8* %gep.inc5
-; CHECK-DAG: NoAlias: i8* %gep.inc3, i8* %gep.inc5
-; CHECK-DAG: NoAlias: i8* %gep.inc4, i8* %gep.inc5
+; CHECK-DAG: MustAlias: i8* %gep.add9, i8* %gep.inc9
+; CHECK-DAG: NoAlias: i8* %gep.inc7, i8* %gep.inc9
+; CHECK-DAG: NoAlias: i8* %gep.inc8, i8* %gep.inc9
 ;; At limit:
-; CHECK-DAG: MustAlias: i8* %gep.add6, i8* %gep.inc6
-; CHECK-DAG: NoAlias: i8* %gep.inc4, i8* %gep.inc6
-; CHECK-DAG: NoAlias: i8* %gep.inc5, i8* %gep.inc6
+; CHECK-DAG: MustAlias: i8* %gep.add10, i8* %gep.inc10
+; CHECK-DAG: NoAlias: i8* %gep.inc10, i8* %gep.inc8
+; CHECK-DAG: NoAlias: i8* %gep.inc10, i8* %gep.inc9
 ;; After limit:
-; CHECK-DAG: MayAlias: i8* %gep.add7, i8* %gep.inc7
-; CHECK-DAG: MayAlias: i8* %gep.inc5, i8* %gep.inc7
-; CHECK-DAG: NoAlias: i8* %gep.inc6, i8* %gep.inc7
+; CHECK-DAG: MayAlias: i8* %gep.add11, i8* %gep.inc11
+; CHECK-DAG: MayAlias: i8* %gep.inc11, i8* %gep.inc9
+; CHECK-DAG: NoAlias: i8* %gep.inc10, i8* %gep.inc11
 
 define void @test(ptr %base) {
-  %gep.add5 = getelementptr i8, ptr %base, i64 5
-  %gep.add6 = getelementptr i8, ptr %base, i64 6
-  %gep.add7 = getelementptr i8, ptr %base, i64 7
+  %gep.add9 = getelementptr i8, ptr %base, i64 9
+  %gep.add10 = getelementptr i8, ptr %base, i64 10
+  %gep.add11 = getelementptr i8, ptr %base, i64 11
 
   %gep.inc1 = getelementptr i8, ptr %base, i64 1
   %gep.inc2 = getelementptr i8, ptr %gep.inc1, i64 1
@@ -26,15 +26,23 @@ define void @test(ptr %base) {
   %gep.inc5 = getelementptr i8, ptr %gep.inc4, i64 1
   %gep.inc6 = getelementptr i8, ptr %gep.inc5, i64 1
   %gep.inc7 = getelementptr i8, ptr %gep.inc6, i64 1
+  %gep.inc8 = getelementptr i8, ptr %gep.inc7, i64 1
+  %gep.inc9 = getelementptr i8, ptr %gep.inc8, i64 1
+  %gep.inc10 = getelementptr i8, ptr %gep.inc9, i64 1
+  %gep.inc11 = getelementptr i8, ptr %gep.inc10, i64 1
 
-  load i8, ptr %gep.add5
-  load i8, ptr %gep.add6
-  load i8, ptr %gep.add7
+  load i8, ptr %gep.add9
+  load i8, ptr %gep.add10
+  load i8, ptr %gep.add11
   load i8, ptr %gep.inc3
   load i8, ptr %gep.inc4
   load i8, ptr %gep.inc5
   load i8, ptr %gep.inc6
   load i8, ptr %gep.inc7
+  load i8, ptr %gep.inc8
+  load i8, ptr %gep.inc9
+  load i8, ptr %gep.inc10
+  load i8, ptr %gep.inc11
 
   ret void
 }
diff --git a/llvm/test/Analysis/LoopAccessAnalysis/underlying-objects-2.ll b/llvm/test/Analysis/LoopAccessAnalysis/underlying-objects-2.ll
index abfdff79dc113..1d3512128678e 100644
--- a/llvm/test/Analysis/LoopAccessAnalysis/underlying-objects-2.ll
+++ b/llvm/test/Analysis/LoopAccessAnalysis/underlying-objects-2.ll
@@ -127,9 +127,12 @@ for_j.body:
   %gepB7 = getelementptr inbounds i8, ptr %gepB6, i64 0
   %gepB8 = getelementptr inbounds i8, ptr %gepB7, i64 0
   %gepB9 = getelementptr inbounds i8, ptr %gepB8, i64 0
+  %gepB10 = getelementptr inbounds i8, ptr %gepB9, i64 0
+  %gepB11 = getelementptr inbounds i8, ptr %gepB10, i64 0
+  %gepB12 = getelementptr inbounds i8, ptr %gepB11, i64 0
 
   %loadPrev = load i8, ptr %gepPrev, align 1
-  %loadB = load i8, ptr %gepB9, align 1
+  %loadB = load i8, ptr %gepB12, align 1
 
   %mul = mul i8 %loadPrev, %loadB
 
diff --git a/llvm/test/Transforms/Inline/inline-noalias-unidentify-object.ll b/llvm/test/Transforms/Inline/inline-noalias-unidentify-object.ll
index 54e9ee0918ae8..b7ba1b32238a7 100644
--- a/llvm/test/Transforms/Inline/inline-noalias-unidentify-object.ll
+++ b/llvm/test/Transforms/Inline/inline-noalias-unidentify-object.ll
@@ -3,15 +3,18 @@
 define i32 @caller(ptr %p) {
 ; CHECK-LABEL: define i32 @caller(ptr %p) {
 ; CHECK-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META0:![0-9]+]])
-; CHECK-NEXT:    [[P_8_I:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 8
-; CHECK-NEXT:    [[V_I:%.*]] = load i32, ptr [[P_8_I]], align 4, !alias.scope !0
-; CHECK-NEXT:    [[P_1_I:%.*]] = getelementptr i8, ptr [[P]], i64 1
+; CHECK-NEXT:    [[P_11_I:%.*]] = getelementptr i8, ptr %p, i64 11
+; CHECK-NEXT:    [[V_I:%.*]] = load i32, ptr [[P_11_I]], align 4, !alias.scope !0
+; CHECK-NEXT:    [[P_1_I:%.*]] = getelementptr i8, ptr %p, i64 1
 ; CHECK-NEXT:    [[P_2_I:%.*]] = getelementptr i8, ptr [[P_1_I]], i64 1
 ; CHECK-NEXT:    [[P_3_I:%.*]] = getelementptr i8, ptr [[P_2_I]], i64 1
 ; CHECK-NEXT:    [[P_4_I:%.*]] = getelementptr i8, ptr [[P_3_I]], i64 1
 ; CHECK-NEXT:    [[P_5_I:%.*]] = getelementptr i8, ptr [[P_4_I]], i64 1
 ; CHECK-NEXT:    [[P_6_I:%.*]] = getelementptr i8, ptr [[P_5_I]], i64 1
-; CHECK-NEXT:    [[P_7_I:%.*]] = getelementptr i8, ptr [[P_6_I]], i64 1
+; CHECK-NEXT:    [[P_7_I1:%.*]] = getelementptr i8, ptr [[P_6_I]], i64 1
+; CHECK-NEXT:    [[P_8_I:%.*]] = getelementptr i8, ptr [[P_7_I1]], i64 1
+; CHECK-NEXT:    [[P_9_I:%.*]] = getelementptr i8, ptr [[P_8_I]], i64 1
+; CHECK-NEXT:    [[P_7_I:%.*]] = getelementptr i8, ptr [[P_9_I]], i64 1
 ; CHECK-NEXT:    [[P_8_ALIAS_I:%.*]] = getelementptr i8, ptr [[P_7_I]], i64 1
 ; CHECK-NEXT:    store i32 42, ptr [[P_8_ALIAS_I]], align 4
 ; CHECK-NEXT:    ret i32 [[V_I]]
@@ -21,8 +24,8 @@ define i32 @caller(ptr %p) {
 }
 
 define internal i32 @callee(ptr noalias %p) {
-  %p.8 = getelementptr i8, ptr %p, i64 8
-  %v = load i32, ptr %p.8
+  %p.11 = getelementptr i8, ptr %p, i64 11
+  %v = load i32, ptr %p.11
   %p.1 = getelementptr i8, ptr %p, i64 1
   %p.2 = getelementptr i8, ptr %p.1, i64 1
   %p.3 = getelementptr i8, ptr %p.2, i64 1
@@ -30,7 +33,10 @@ define internal i32 @callee(ptr noalias %p) {
   %p.5 = getelementptr i8, ptr %p.4, i64 1
   %p.6 = getelementptr i8, ptr %p.5, i64 1
   %p.7 = getelementptr i8, ptr %p.6, i64 1
-  %p.8.alias = getelementptr i8, ptr %p.7, i64 1
-  store i32 42, ptr %p.8.alias
+  %p.8 = getelementptr i8, ptr %p.7, i64 1
+  %p.9 = getelementptr i8, ptr %p.8, i64 1
+  %p.10 = getelementptr i8, ptr %p.9, i64 1
+  %p.11.alias = getelementptr i8, ptr %p.10, i64 1
+  store i32 42, ptr %p.11.alias
   ret i32 %v
 }

@llvmbot
Copy link
Member

llvmbot commented Jun 11, 2025

@llvm/pr-subscribers-llvm-transforms

Author: Nikita Popov (nikic)

Changes

This depth limits a linear search (rather than the usual potentially exponential one) and is not particularly important for compile-time in practice.

The change in #137297 is going to increase the length of GEP chains, so I'd like to increase this limit a bit to reduce the chance of regressions (dtcxzyw/llvm-opt-benchmark#2419 showed a 13% increase in SearchLimitReached). There is no particular significance to the new value of 10.

Compile-time looks neutral: https://llvm-compile-time-tracker.com/compare.php?from=a72bcda1434c72f9db6687565a361479e0dde572&to=f5129a63ce2c543b07de49b0308d2adb2b600582&stat=instructions:u

Full diff: https://github.com/llvm/llvm-project/pull/143714.diff

4 Files Affected:

  • (modified) llvm/include/llvm/Analysis/ValueTracking.h (+1-1)
  • (modified) llvm/test/Analysis/BasicAA/gep-decomposition-limit.ll (+23-15)
  • (modified) llvm/test/Analysis/LoopAccessAnalysis/underlying-objects-2.ll (+4-1)
  • (modified) llvm/test/Transforms/Inline/inline-noalias-unidentify-object.ll (+14-8)
diff --git a/llvm/include/llvm/Analysis/ValueTracking.h b/llvm/include/llvm/Analysis/ValueTracking.h
index 32ab9733d13c9..e215c90b5a72a 100644
--- a/llvm/include/llvm/Analysis/ValueTracking.h
+++ b/llvm/include/llvm/Analysis/ValueTracking.h
@@ -47,7 +47,7 @@ constexpr unsigned MaxAnalysisRecursionDepth = 6;
 
 /// The max limit of the search depth in DecomposeGEPExpression() and
 /// getUnderlyingObject().
-constexpr unsigned MaxLookupSearchDepth = 6;
+constexpr unsigned MaxLookupSearchDepth = 10;
 
 /// Determine which bits of V are known to be either zero or one and return
 /// them in the KnownZero/KnownOne bit sets.
diff --git a/llvm/test/Analysis/BasicAA/gep-decomposition-limit.ll b/llvm/test/Analysis/BasicAA/gep-decomposition-limit.ll
index 23a96ebca8485..a256ececbe565 100644
--- a/llvm/test/Analysis/BasicAA/gep-decomposition-limit.ll
+++ b/llvm/test/Analysis/BasicAA/gep-decomposition-limit.ll
@@ -2,22 +2,22 @@
 
 ; CHECK-LABEL: Function: test
 ;; Before limit:
-; CHECK-DAG: MustAlias: i8* %gep.add5, i8* %gep.inc5
-; CHECK-DAG: NoAlias: i8* %gep.inc3, i8* %gep.inc5
-; CHECK-DAG: NoAlias: i8* %gep.inc4, i8* %gep.inc5
+; CHECK-DAG: MustAlias: i8* %gep.add9, i8* %gep.inc9
+; CHECK-DAG: NoAlias: i8* %gep.inc7, i8* %gep.inc9
+; CHECK-DAG: NoAlias: i8* %gep.inc8, i8* %gep.inc9
 ;; At limit:
-; CHECK-DAG: MustAlias: i8* %gep.add6, i8* %gep.inc6
-; CHECK-DAG: NoAlias: i8* %gep.inc4, i8* %gep.inc6
-; CHECK-DAG: NoAlias: i8* %gep.inc5, i8* %gep.inc6
+; CHECK-DAG: MustAlias: i8* %gep.add10, i8* %gep.inc10
+; CHECK-DAG: NoAlias: i8* %gep.inc10, i8* %gep.inc8
+; CHECK-DAG: NoAlias: i8* %gep.inc10, i8* %gep.inc9
 ;; After limit:
-; CHECK-DAG: MayAlias: i8* %gep.add7, i8* %gep.inc7
-; CHECK-DAG: MayAlias: i8* %gep.inc5, i8* %gep.inc7
-; CHECK-DAG: NoAlias: i8* %gep.inc6, i8* %gep.inc7
+; CHECK-DAG: MayAlias: i8* %gep.add11, i8* %gep.inc11
+; CHECK-DAG: MayAlias: i8* %gep.inc11, i8* %gep.inc9
+; CHECK-DAG: NoAlias: i8* %gep.inc10, i8* %gep.inc11
 
 define void @test(ptr %base) {
-  %gep.add5 = getelementptr i8, ptr %base, i64 5
-  %gep.add6 = getelementptr i8, ptr %base, i64 6
-  %gep.add7 = getelementptr i8, ptr %base, i64 7
+  %gep.add9 = getelementptr i8, ptr %base, i64 9
+  %gep.add10 = getelementptr i8, ptr %base, i64 10
+  %gep.add11 = getelementptr i8, ptr %base, i64 11
 
   %gep.inc1 = getelementptr i8, ptr %base, i64 1
   %gep.inc2 = getelementptr i8, ptr %gep.inc1, i64 1
@@ -26,15 +26,23 @@ define void @test(ptr %base) {
   %gep.inc5 = getelementptr i8, ptr %gep.inc4, i64 1
   %gep.inc6 = getelementptr i8, ptr %gep.inc5, i64 1
   %gep.inc7 = getelementptr i8, ptr %gep.inc6, i64 1
+  %gep.inc8 = getelementptr i8, ptr %gep.inc7, i64 1
+  %gep.inc9 = getelementptr i8, ptr %gep.inc8, i64 1
+  %gep.inc10 = getelementptr i8, ptr %gep.inc9, i64 1
+  %gep.inc11 = getelementptr i8, ptr %gep.inc10, i64 1
 
-  load i8, ptr %gep.add5
-  load i8, ptr %gep.add6
-  load i8, ptr %gep.add7
+  load i8, ptr %gep.add9
+  load i8, ptr %gep.add10
+  load i8, ptr %gep.add11
   load i8, ptr %gep.inc3
   load i8, ptr %gep.inc4
   load i8, ptr %gep.inc5
   load i8, ptr %gep.inc6
   load i8, ptr %gep.inc7
+  load i8, ptr %gep.inc8
+  load i8, ptr %gep.inc9
+  load i8, ptr %gep.inc10
+  load i8, ptr %gep.inc11
 
   ret void
 }
diff --git a/llvm/test/Analysis/LoopAccessAnalysis/underlying-objects-2.ll b/llvm/test/Analysis/LoopAccessAnalysis/underlying-objects-2.ll
index abfdff79dc113..1d3512128678e 100644
--- a/llvm/test/Analysis/LoopAccessAnalysis/underlying-objects-2.ll
+++ b/llvm/test/Analysis/LoopAccessAnalysis/underlying-objects-2.ll
@@ -127,9 +127,12 @@ for_j.body:
   %gepB7 = getelementptr inbounds i8, ptr %gepB6, i64 0
   %gepB8 = getelementptr inbounds i8, ptr %gepB7, i64 0
   %gepB9 = getelementptr inbounds i8, ptr %gepB8, i64 0
+  %gepB10 = getelementptr inbounds i8, ptr %gepB9, i64 0
+  %gepB11 = getelementptr inbounds i8, ptr %gepB10, i64 0
+  %gepB12 = getelementptr inbounds i8, ptr %gepB11, i64 0
 
   %loadPrev = load i8, ptr %gepPrev, align 1
-  %loadB = load i8, ptr %gepB9, align 1
+  %loadB = load i8, ptr %gepB12, align 1
 
   %mul = mul i8 %loadPrev, %loadB
 
diff --git a/llvm/test/Transforms/Inline/inline-noalias-unidentify-object.ll b/llvm/test/Transforms/Inline/inline-noalias-unidentify-object.ll
index 54e9ee0918ae8..b7ba1b32238a7 100644
--- a/llvm/test/Transforms/Inline/inline-noalias-unidentify-object.ll
+++ b/llvm/test/Transforms/Inline/inline-noalias-unidentify-object.ll
@@ -3,15 +3,18 @@
 define i32 @caller(ptr %p) {
 ; CHECK-LABEL: define i32 @caller(ptr %p) {
 ; CHECK-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META0:![0-9]+]])
-; CHECK-NEXT:    [[P_8_I:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 8
-; CHECK-NEXT:    [[V_I:%.*]] = load i32, ptr [[P_8_I]], align 4, !alias.scope !0
-; CHECK-NEXT:    [[P_1_I:%.*]] = getelementptr i8, ptr [[P]], i64 1
+; CHECK-NEXT:    [[P_11_I:%.*]] = getelementptr i8, ptr %p, i64 11
+; CHECK-NEXT:    [[V_I:%.*]] = load i32, ptr [[P_11_I]], align 4, !alias.scope !0
+; CHECK-NEXT:    [[P_1_I:%.*]] = getelementptr i8, ptr %p, i64 1
 ; CHECK-NEXT:    [[P_2_I:%.*]] = getelementptr i8, ptr [[P_1_I]], i64 1
 ; CHECK-NEXT:    [[P_3_I:%.*]] = getelementptr i8, ptr [[P_2_I]], i64 1
 ; CHECK-NEXT:    [[P_4_I:%.*]] = getelementptr i8, ptr [[P_3_I]], i64 1
 ; CHECK-NEXT:    [[P_5_I:%.*]] = getelementptr i8, ptr [[P_4_I]], i64 1
 ; CHECK-NEXT:    [[P_6_I:%.*]] = getelementptr i8, ptr [[P_5_I]], i64 1
-; CHECK-NEXT:    [[P_7_I:%.*]] = getelementptr i8, ptr [[P_6_I]], i64 1
+; CHECK-NEXT:    [[P_7_I1:%.*]] = getelementptr i8, ptr [[P_6_I]], i64 1
+; CHECK-NEXT:    [[P_8_I:%.*]] = getelementptr i8, ptr [[P_7_I1]], i64 1
+; CHECK-NEXT:    [[P_9_I:%.*]] = getelementptr i8, ptr [[P_8_I]], i64 1
+; CHECK-NEXT:    [[P_7_I:%.*]] = getelementptr i8, ptr [[P_9_I]], i64 1
 ; CHECK-NEXT:    [[P_8_ALIAS_I:%.*]] = getelementptr i8, ptr [[P_7_I]], i64 1
 ; CHECK-NEXT:    store i32 42, ptr [[P_8_ALIAS_I]], align 4
 ; CHECK-NEXT:    ret i32 [[V_I]]
@@ -21,8 +24,8 @@ define i32 @caller(ptr %p) {
 }
 
 define internal i32 @callee(ptr noalias %p) {
-  %p.8 = getelementptr i8, ptr %p, i64 8
-  %v = load i32, ptr %p.8
+  %p.11 = getelementptr i8, ptr %p, i64 11
+  %v = load i32, ptr %p.11
   %p.1 = getelementptr i8, ptr %p, i64 1
   %p.2 = getelementptr i8, ptr %p.1, i64 1
   %p.3 = getelementptr i8, ptr %p.2, i64 1
@@ -30,7 +33,10 @@ define internal i32 @callee(ptr noalias %p) {
   %p.5 = getelementptr i8, ptr %p.4, i64 1
   %p.6 = getelementptr i8, ptr %p.5, i64 1
   %p.7 = getelementptr i8, ptr %p.6, i64 1
-  %p.8.alias = getelementptr i8, ptr %p.7, i64 1
-  store i32 42, ptr %p.8.alias
+  %p.8 = getelementptr i8, ptr %p.7, i64 1
+  %p.9 = getelementptr i8, ptr %p.8, i64 1
+  %p.10 = getelementptr i8, ptr %p.9, i64 1
+  %p.11.alias = getelementptr i8, ptr %p.10, i64 1
+  store i32 42, ptr %p.11.alias
   ret i32 %v
 }

dtcxzyw
dtcxzyw approved these changes Jun 11, 2025
View reviewed changes
Copy link
Member

@dtcxzyw dtcxzyw left a comment

Choose a reason for hiding this comment

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

LG

fhahn
fhahn approved these changes Jun 11, 2025
View reviewed changes
Copy link
Contributor

@fhahn fhahn left a comment

Choose a reason for hiding this comment

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

LGTM, thanks

@nikic nikic merged commit 6157028 into llvm:main Jun 12, 2025
10 checks passed
@nikic nikic deleted the basicaa-depth branch June 12, 2025 07:19
@dtcxzyw
Copy link
Member

dtcxzyw commented Jun 12, 2025

basicaa.SearchLimitReached 40216 -> 5083 -87.36%

tomtor pushed a commit to tomtor/llvm-project that referenced this pull request Jun 14, 2025
@nikic @tomtor
[BasicAA][ValueTracking] Increase depth for underlying object search (l…
0dd666e 
…lvm#143714)

This depth limits a linear search (rather than the usual potentially
exponential one) and is not particularly important for compile-time in
practice.

The change in llvm#137297 is going to increase the length of GEP chains, so
I'd like to increase this limit a bit to reduce the chance of
regressions (dtcxzyw/llvm-opt-benchmark#2419
showed a 13% increase in SearchLimitReached). There is no particular
significance to the new value of 10.

Compile-time is neutral.
akuhlens pushed a commit to akuhlens/llvm-project that referenced this pull request Jun 24, 2025
@nikic @akuhlens
[BasicAA][ValueTracking] Increase depth for underlying object search (l…
4ee49fb 
…lvm#143714)

This depth limits a linear search (rather than the usual potentially
exponential one) and is not particularly important for compile-time in
practice.

The change in llvm#137297 is going to increase the length of GEP chains, so
I'd like to increase this limit a bit to reduce the chance of
regressions (dtcxzyw/llvm-opt-benchmark#2419
showed a 13% increase in SearchLimitReached). There is no particular
significance to the new value of 10.

Compile-time is neutral.
@nikic nikic mentioned this pull request Oct 24, 2025
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@fhahn fhahn fhahn approved these changes

@dtcxzyw dtcxzyw dtcxzyw approved these changes

Assignees

No one assigned

Labels

llvm:analysis Includes value tracking, cost tables and constant folding llvm:transforms

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@nikic @llvmbot @dtcxzyw @fhahn