From 7bfaa40086359ed7e41c862ab0a65e0bb1be0aeb Mon Sep 17 00:00:00 2001 From: Eric Christopher Date: Thu, 16 Jul 2020 11:40:43 -0700 Subject: [PATCH] Temporarily Revert "[AssumeBundles] Use operand bundles to encode alignment assumptions" due to the performance bugs filed in https://bugs.llvm.org/show_bug.cgi?id=46753. An SROA change soon may obviate some of these problems. This reverts commit 8d09f20798ac180b1749276bff364682ce0196ab. --- clang/lib/CodeGen/CodeGenFunction.cpp | 36 +----- clang/test/CodeGen/align_value.cpp | 30 ++++- clang/test/CodeGen/alloc-align-attr.c | 44 ++++++-- .../assume-aligned-and-alloc-align-attributes.c | 8 +- clang/test/CodeGen/builtin-align-array.c | 32 ++++-- clang/test/CodeGen/builtin-align.c | 24 +++- clang/test/CodeGen/builtin-assume-aligned.c | 32 +++++- ...-assumption-attribute-align_value-on-lvalue.cpp | 8 +- ...ssumption-attribute-align_value-on-paramvar.cpp | 2 +- ...-attribute-alloc_align-on-function-variable.cpp | 10 +- ...ssumption-attribute-alloc_align-on-function.cpp | 2 +- ...ibute-assume_aligned-on-function-two-params.cpp | 10 +- ...mption-attribute-assume_aligned-on-function.cpp | 2 +- ...uiltin_assume_aligned-three-params-variable.cpp | 10 +- ...umption-builtin_assume_aligned-three-params.cpp | 10 +- ...ssumption-builtin_assume_aligned-two-params.cpp | 8 +- .../CodeGen/catch-alignment-assumption-openmp.cpp | 8 +- .../CodeGen/non-power-of-2-alignment-assumptions.c | 13 ++- clang/test/OpenMP/simd_codegen.cpp | 16 +++ clang/test/OpenMP/simd_metadata.c | 117 ++++++++++++-------- ..._teams_distribute_parallel_for_simd_codegen.cpp | 5 +- llvm/include/llvm/IR/IRBuilder.h | 28 +++-- .../Transforms/Scalar/AlignmentFromAssumptions.h | 6 +- llvm/lib/Analysis/AssumeBundleQueries.cpp | 13 +-- llvm/lib/IR/IRBuilder.cpp | 77 ++++++++----- llvm/lib/IR/Verifier.cpp | 23 +--- .../Transforms/InstCombine/InstCombineCalls.cpp | 15 +-- .../Transforms/Scalar/AlignmentFromAssumptions.cpp | 121 ++++++++++++++++----- .../Transforms/AlignmentFromAssumptions/simple.ll | 75 ++++++++----- .../AlignmentFromAssumptions/simple32.ll | 114 +++++++++++++++---- llvm/test/Transforms/Inline/align.ll | 15 ++- llvm/test/Transforms/InstCombine/assume.ll | 1 - .../inlining-alignment-assumptions.ll | 27 ++++- llvm/test/Verifier/assume-bundles.ll | 16 +-- .../unittests/Analysis/AssumeBundleQueriesTest.cpp | 38 ------- 35 files changed, 627 insertions(+), 369 deletions(-) diff --git a/clang/lib/CodeGen/CodeGenFunction.cpp b/clang/lib/CodeGen/CodeGenFunction.cpp index 4a7c845..8ce488f 100644 --- a/clang/lib/CodeGen/CodeGenFunction.cpp +++ b/clang/lib/CodeGen/CodeGenFunction.cpp @@ -2154,39 +2154,13 @@ void CodeGenFunction::emitAlignmentAssumption(llvm::Value *PtrValue, SourceLocation AssumptionLoc, llvm::Value *Alignment, llvm::Value *OffsetValue) { - if (Alignment->getType() != IntPtrTy) - Alignment = - Builder.CreateIntCast(Alignment, IntPtrTy, false, "casted.align"); - if (OffsetValue && OffsetValue->getType() != IntPtrTy) - OffsetValue = - Builder.CreateIntCast(OffsetValue, IntPtrTy, true, "casted.offset"); - llvm::Value *TheCheck = nullptr; + llvm::Value *TheCheck; + llvm::Instruction *Assumption = Builder.CreateAlignmentAssumption( + CGM.getDataLayout(), PtrValue, Alignment, OffsetValue, &TheCheck); if (SanOpts.has(SanitizerKind::Alignment)) { - llvm::Value *PtrIntValue = - Builder.CreatePtrToInt(PtrValue, IntPtrTy, "ptrint"); - - if (OffsetValue) { - bool IsOffsetZero = false; - if (const auto *CI = dyn_cast(OffsetValue)) - IsOffsetZero = CI->isZero(); - - if (!IsOffsetZero) - PtrIntValue = Builder.CreateSub(PtrIntValue, OffsetValue, "offsetptr"); - } - - llvm::Value *Zero = llvm::ConstantInt::get(IntPtrTy, 0); - llvm::Value *Mask = - Builder.CreateSub(Alignment, llvm::ConstantInt::get(IntPtrTy, 1)); - llvm::Value *MaskedPtr = Builder.CreateAnd(PtrIntValue, Mask, "maskedptr"); - TheCheck = Builder.CreateICmpEQ(MaskedPtr, Zero, "maskcond"); + emitAlignmentAssumptionCheck(PtrValue, Ty, Loc, AssumptionLoc, Alignment, + OffsetValue, TheCheck, Assumption); } - llvm::Instruction *Assumption = Builder.CreateAlignmentAssumption( - CGM.getDataLayout(), PtrValue, Alignment, OffsetValue); - - if (!SanOpts.has(SanitizerKind::Alignment)) - return; - emitAlignmentAssumptionCheck(PtrValue, Ty, Loc, AssumptionLoc, Alignment, - OffsetValue, TheCheck, Assumption); } void CodeGenFunction::emitAlignmentAssumption(llvm::Value *PtrValue, diff --git a/clang/test/CodeGen/align_value.cpp b/clang/test/CodeGen/align_value.cpp index a18cb65..acbfbaf 100644 --- a/clang/test/CodeGen/align_value.cpp +++ b/clang/test/CodeGen/align_value.cpp @@ -29,7 +29,10 @@ struct ad_struct { // CHECK-NEXT: [[TMP0:%.*]] = load %struct.ad_struct*, %struct.ad_struct** [[X_ADDR]], align 8 // CHECK-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_AD_STRUCT:%.*]], %struct.ad_struct* [[TMP0]], i32 0, i32 0 // CHECK-NEXT: [[TMP1:%.*]] = load double*, double** [[A]], align 8 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[TMP1]], i64 64) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint double* [[TMP1]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 63 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: ret double* [[TMP1]] // double *foo(ad_struct& x) { @@ -45,7 +48,10 @@ double *foo(ad_struct& x) { // CHECK-NEXT: [[TMP0:%.*]] = load %struct.ad_struct*, %struct.ad_struct** [[X_ADDR]], align 8 // CHECK-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_AD_STRUCT:%.*]], %struct.ad_struct* [[TMP0]], i32 0, i32 0 // CHECK-NEXT: [[TMP1:%.*]] = load double*, double** [[A]], align 8 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[TMP1]], i64 64) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint double* [[TMP1]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 63 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: ret double* [[TMP1]] // double *goo(ad_struct *x) { @@ -60,7 +66,10 @@ double *goo(ad_struct *x) { // CHECK-NEXT: store double** [[X]], double*** [[X_ADDR]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load double**, double*** [[X_ADDR]], align 8 // CHECK-NEXT: [[TMP1:%.*]] = load double*, double** [[TMP0]], align 8 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[TMP1]], i64 64) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint double* [[TMP1]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 63 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: ret double* [[TMP1]] // double *bar(aligned_double *x) { @@ -75,7 +84,10 @@ double *bar(aligned_double *x) { // CHECK-NEXT: store double** [[X]], double*** [[X_ADDR]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load double**, double*** [[X_ADDR]], align 8 // CHECK-NEXT: [[TMP1:%.*]] = load double*, double** [[TMP0]], align 8 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[TMP1]], i64 64) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint double* [[TMP1]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 63 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: ret double* [[TMP1]] // double *car(aligned_double &x) { @@ -91,7 +103,10 @@ double *car(aligned_double &x) { // CHECK-NEXT: [[TMP0:%.*]] = load double**, double*** [[X_ADDR]], align 8 // CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double*, double** [[TMP0]], i64 5 // CHECK-NEXT: [[TMP1:%.*]] = load double*, double** [[ARRAYIDX]], align 8 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[TMP1]], i64 64) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint double* [[TMP1]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 63 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: ret double* [[TMP1]] // double *dar(aligned_double *x) { @@ -103,7 +118,10 @@ aligned_double eep(); // CHECK-LABEL: define {{[^@]+}}@_Z3retv() #0 // CHECK-NEXT: entry: // CHECK-NEXT: [[CALL:%.*]] = call double* @_Z3eepv() -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[CALL]], i64 64) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint double* [[CALL]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 63 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: ret double* [[CALL]] // double *ret() { diff --git a/clang/test/CodeGen/alloc-align-attr.c b/clang/test/CodeGen/alloc-align-attr.c index 44a5729..9517c50 100644 --- a/clang/test/CodeGen/alloc-align-attr.c +++ b/clang/test/CodeGen/alloc-align-attr.c @@ -11,8 +11,12 @@ __INT32_TYPE__*m1(__INT32_TYPE__ i) __attribute__((alloc_align(1))); // CHECK-NEXT: store i32 [[A]], i32* [[A_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK-NEXT: [[CALL:%.*]] = call i32* @m1(i32 [[TMP0]]) -// CHECK-NEXT: [[CASTED_ALIGN:%.*]] = zext i32 [[TMP0]] to i64 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[CALL]], i64 [[CASTED_ALIGN]]) ] +// CHECK-NEXT: [[ALIGNMENTCAST:%.*]] = zext i32 [[TMP0]] to i64 +// CHECK-NEXT: [[MASK:%.*]] = sub i64 [[ALIGNMENTCAST]], 1 +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[CALL]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK]] +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP1:%.*]] = load i32, i32* [[CALL]], align 4 // CHECK-NEXT: ret i32 [[TMP1]] // @@ -28,8 +32,12 @@ __INT32_TYPE__ test1(__INT32_TYPE__ a) { // CHECK-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8 // CHECK-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32 // CHECK-NEXT: [[CALL:%.*]] = call i32* @m1(i32 [[CONV]]) -// CHECK-NEXT: [[CASTED_ALIGN:%.*]] = zext i32 [[CONV]] to i64 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[CALL]], i64 [[CASTED_ALIGN]]) ] +// CHECK-NEXT: [[ALIGNMENTCAST:%.*]] = zext i32 [[CONV]] to i64 +// CHECK-NEXT: [[MASK:%.*]] = sub i64 [[ALIGNMENTCAST]], 1 +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[CALL]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK]] +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP1:%.*]] = load i32, i32* [[CALL]], align 4 // CHECK-NEXT: ret i32 [[TMP1]] // @@ -47,7 +55,11 @@ __INT32_TYPE__ *m2(__SIZE_TYPE__ i) __attribute__((alloc_align(1))); // CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK-NEXT: [[CONV:%.*]] = sext i32 [[TMP0]] to i64 // CHECK-NEXT: [[CALL:%.*]] = call i32* @m2(i64 [[CONV]]) -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[CALL]], i64 [[CONV]]) ] +// CHECK-NEXT: [[MASK:%.*]] = sub i64 [[CONV]], 1 +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[CALL]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK]] +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP1:%.*]] = load i32, i32* [[CALL]], align 4 // CHECK-NEXT: ret i32 [[TMP1]] // @@ -63,7 +75,11 @@ __INT32_TYPE__ test3(__INT32_TYPE__ a) { // CHECK-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8 // CHECK-NEXT: [[CALL:%.*]] = call i32* @m2(i64 [[TMP0]]) -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[CALL]], i64 [[TMP0]]) ] +// CHECK-NEXT: [[MASK:%.*]] = sub i64 [[TMP0]], 1 +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[CALL]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK]] +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP1:%.*]] = load i32, i32* [[CALL]], align 4 // CHECK-NEXT: ret i32 [[TMP1]] // @@ -99,8 +115,12 @@ __INT32_TYPE__ *m3(struct Empty s, __int128_t i) __attribute__((alloc_align(2))) // CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds { i64, i64 }, { i64, i64 }* [[TMP4]], i32 0, i32 1 // CHECK-NEXT: [[TMP8:%.*]] = load i64, i64* [[TMP7]], align 8 // CHECK-NEXT: [[CALL:%.*]] = call i32* @m3(i64 [[TMP6]], i64 [[TMP8]]) -// CHECK-NEXT: [[CASTED_ALIGN:%.*]] = trunc i128 [[TMP3]] to i64 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[CALL]], i64 [[CASTED_ALIGN]]) ] +// CHECK-NEXT: [[ALIGNMENTCAST:%.*]] = trunc i128 [[TMP3]] to i64 +// CHECK-NEXT: [[MASK:%.*]] = sub i64 [[ALIGNMENTCAST]], 1 +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[CALL]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK]] +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP9:%.*]] = load i32, i32* [[CALL]], align 4 // CHECK-NEXT: ret i32 [[TMP9]] // @@ -137,8 +157,12 @@ __INT32_TYPE__ *m4(struct MultiArgs s, __int128_t i) __attribute__((alloc_align( // CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds { i64, i64 }, { i64, i64 }* [[TMP9]], i32 0, i32 1 // CHECK-NEXT: [[TMP13:%.*]] = load i64, i64* [[TMP12]], align 8 // CHECK-NEXT: [[CALL:%.*]] = call i32* @m4(i64 [[TMP6]], i64 [[TMP8]], i64 [[TMP11]], i64 [[TMP13]]) -// CHECK-NEXT: [[CASTED_ALIGN:%.*]] = trunc i128 [[TMP3]] to i64 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[CALL]], i64 [[CASTED_ALIGN]]) ] +// CHECK-NEXT: [[ALIGNMENTCAST:%.*]] = trunc i128 [[TMP3]] to i64 +// CHECK-NEXT: [[MASK:%.*]] = sub i64 [[ALIGNMENTCAST]], 1 +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[CALL]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK]] +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP14:%.*]] = load i32, i32* [[CALL]], align 4 // CHECK-NEXT: ret i32 [[TMP14]] // diff --git a/clang/test/CodeGen/assume-aligned-and-alloc-align-attributes.c b/clang/test/CodeGen/assume-aligned-and-alloc-align-attributes.c index cd8a6f1..fa4ee8d 100644 --- a/clang/test/CodeGen/assume-aligned-and-alloc-align-attributes.c +++ b/clang/test/CodeGen/assume-aligned-and-alloc-align-attributes.c @@ -36,8 +36,12 @@ void *t2_immediate2() { // CHECK-NEXT: store i32 [[ALIGNMENT:%.*]], i32* [[ALIGNMENT_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[ALIGNMENT_ADDR]], align 4 // CHECK-NEXT: [[CALL:%.*]] = call align 32 i8* @my_aligned_alloc(i32 320, i32 [[TMP0]]) -// CHECK-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[CALL]], i64 [[TMP1]]) ] +// CHECK-NEXT: [[ALIGNMENTCAST:%.*]] = zext i32 [[TMP0]] to i64 +// CHECK-NEXT: [[MASK:%.*]] = sub i64 [[ALIGNMENTCAST]], 1 +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[CALL]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK]] +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: ret i8* [[CALL]] // void *t3_variable(int alignment) { diff --git a/clang/test/CodeGen/builtin-align-array.c b/clang/test/CodeGen/builtin-align-array.c index 31f7b42..97235c3 100644 --- a/clang/test/CodeGen/builtin-align-array.c +++ b/clang/test/CodeGen/builtin-align-array.c @@ -4,7 +4,7 @@ extern int func(char *c); -// CHECK-LABEL: @test_array( +// CHECK-LABEL: define {{[^@]+}}@test_array() #0 // CHECK-NEXT: entry: // CHECK-NEXT: [[BUF:%.*]] = alloca [1024 x i8], align 16 // CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [1024 x i8], [1024 x i8]* [[BUF]], i64 0, i64 44 @@ -12,7 +12,10 @@ extern int func(char *c); // CHECK-NEXT: [[ALIGNED_INTPTR:%.*]] = and i64 [[INTPTR]], -16 // CHECK-NEXT: [[DIFF:%.*]] = sub i64 [[ALIGNED_INTPTR]], [[INTPTR]] // CHECK-NEXT: [[ALIGNED_RESULT:%.*]] = getelementptr inbounds i8, i8* [[ARRAYIDX]], i64 [[DIFF]] -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[ALIGNED_RESULT]], i64 16) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[ALIGNED_RESULT]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 15 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[CALL:%.*]] = call i32 @func(i8* [[ALIGNED_RESULT]]) // CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds [1024 x i8], [1024 x i8]* [[BUF]], i64 0, i64 22 // CHECK-NEXT: [[INTPTR2:%.*]] = ptrtoint i8* [[ARRAYIDX1]] to i64 @@ -20,10 +23,13 @@ extern int func(char *c); // CHECK-NEXT: [[ALIGNED_INTPTR4:%.*]] = and i64 [[OVER_BOUNDARY]], -32 // CHECK-NEXT: [[DIFF5:%.*]] = sub i64 [[ALIGNED_INTPTR4]], [[INTPTR2]] // CHECK-NEXT: [[ALIGNED_RESULT6:%.*]] = getelementptr inbounds i8, i8* [[ARRAYIDX1]], i64 [[DIFF5]] -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[ALIGNED_RESULT6]], i64 32) ] -// CHECK-NEXT: [[CALL7:%.*]] = call i32 @func(i8* [[ALIGNED_RESULT6]]) -// CHECK-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [1024 x i8], [1024 x i8]* [[BUF]], i64 0, i64 16 -// CHECK-NEXT: [[SRC_ADDR:%.*]] = ptrtoint i8* [[ARRAYIDX8]] to i64 +// CHECK-NEXT: [[PTRINT7:%.*]] = ptrtoint i8* [[ALIGNED_RESULT6]] to i64 +// CHECK-NEXT: [[MASKEDPTR8:%.*]] = and i64 [[PTRINT7]], 31 +// CHECK-NEXT: [[MASKCOND9:%.*]] = icmp eq i64 [[MASKEDPTR8]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND9]]) +// CHECK-NEXT: [[CALL10:%.*]] = call i32 @func(i8* [[ALIGNED_RESULT6]]) +// CHECK-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [1024 x i8], [1024 x i8]* [[BUF]], i64 0, i64 16 +// CHECK-NEXT: [[SRC_ADDR:%.*]] = ptrtoint i8* [[ARRAYIDX11]] to i64 // CHECK-NEXT: [[SET_BITS:%.*]] = and i64 [[SRC_ADDR]], 63 // CHECK-NEXT: [[IS_ALIGNED:%.*]] = icmp eq i64 [[SET_BITS]], 0 // CHECK-NEXT: [[CONV:%.*]] = zext i1 [[IS_ALIGNED]] to i32 @@ -36,7 +42,7 @@ int test_array(void) { return __builtin_is_aligned(&buf[16], 64); } -// CHECK-LABEL: @test_array_should_not_mask( +// CHECK-LABEL: define {{[^@]+}}@test_array_should_not_mask() #0 // CHECK-NEXT: entry: // CHECK-NEXT: [[BUF:%.*]] = alloca [1024 x i8], align 32 // CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [1024 x i8], [1024 x i8]* [[BUF]], i64 0, i64 64 @@ -44,7 +50,10 @@ int test_array(void) { // CHECK-NEXT: [[ALIGNED_INTPTR:%.*]] = and i64 [[INTPTR]], -16 // CHECK-NEXT: [[DIFF:%.*]] = sub i64 [[ALIGNED_INTPTR]], [[INTPTR]] // CHECK-NEXT: [[ALIGNED_RESULT:%.*]] = getelementptr inbounds i8, i8* [[ARRAYIDX]], i64 [[DIFF]] -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[ALIGNED_RESULT]], i64 16) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[ALIGNED_RESULT]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 15 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[CALL:%.*]] = call i32 @func(i8* [[ALIGNED_RESULT]]) // CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds [1024 x i8], [1024 x i8]* [[BUF]], i64 0, i64 32 // CHECK-NEXT: [[INTPTR2:%.*]] = ptrtoint i8* [[ARRAYIDX1]] to i64 @@ -52,8 +61,11 @@ int test_array(void) { // CHECK-NEXT: [[ALIGNED_INTPTR4:%.*]] = and i64 [[OVER_BOUNDARY]], -32 // CHECK-NEXT: [[DIFF5:%.*]] = sub i64 [[ALIGNED_INTPTR4]], [[INTPTR2]] // CHECK-NEXT: [[ALIGNED_RESULT6:%.*]] = getelementptr inbounds i8, i8* [[ARRAYIDX1]], i64 [[DIFF5]] -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[ALIGNED_RESULT6]], i64 32) ] -// CHECK-NEXT: [[CALL7:%.*]] = call i32 @func(i8* [[ALIGNED_RESULT6]]) +// CHECK-NEXT: [[PTRINT7:%.*]] = ptrtoint i8* [[ALIGNED_RESULT6]] to i64 +// CHECK-NEXT: [[MASKEDPTR8:%.*]] = and i64 [[PTRINT7]], 31 +// CHECK-NEXT: [[MASKCOND9:%.*]] = icmp eq i64 [[MASKEDPTR8]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND9]]) +// CHECK-NEXT: [[CALL10:%.*]] = call i32 @func(i8* [[ALIGNED_RESULT6]]) // CHECK-NEXT: ret i32 1 // int test_array_should_not_mask(void) { diff --git a/clang/test/CodeGen/builtin-align.c b/clang/test/CodeGen/builtin-align.c index 60f7fc9..7e66e2b 100644 --- a/clang/test/CodeGen/builtin-align.c +++ b/clang/test/CodeGen/builtin-align.c @@ -122,7 +122,11 @@ _Bool is_aligned(TYPE ptr, unsigned align) { // CHECK-VOID_PTR-NEXT: [[ALIGNED_INTPTR:%.*]] = and i64 [[OVER_BOUNDARY]], [[INVERTED_MASK]] // CHECK-VOID_PTR-NEXT: [[DIFF:%.*]] = sub i64 [[ALIGNED_INTPTR]], [[INTPTR]] // CHECK-VOID_PTR-NEXT: [[ALIGNED_RESULT:%.*]] = getelementptr inbounds i8, i8* [[PTR]], i64 [[DIFF]] -// CHECK-VOID_PTR-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[ALIGNED_RESULT]], i64 [[ALIGNMENT]]) ] +// CHECK-VOID_PTR-NEXT: [[MASK1:%.*]] = sub i64 [[ALIGNMENT]], 1 +// CHECK-VOID_PTR-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[ALIGNED_RESULT]] to i64 +// CHECK-VOID_PTR-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK1]] +// CHECK-VOID_PTR-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-VOID_PTR-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-VOID_PTR-NEXT: ret i8* [[ALIGNED_RESULT]] // // CHECK-FLOAT_PTR-LABEL: define {{[^@]+}}@align_up @@ -138,7 +142,11 @@ _Bool is_aligned(TYPE ptr, unsigned align) { // CHECK-FLOAT_PTR-NEXT: [[TMP0:%.*]] = bitcast float* [[PTR]] to i8* // CHECK-FLOAT_PTR-NEXT: [[ALIGNED_RESULT:%.*]] = getelementptr inbounds i8, i8* [[TMP0]], i64 [[DIFF]] // CHECK-FLOAT_PTR-NEXT: [[TMP1:%.*]] = bitcast i8* [[ALIGNED_RESULT]] to float* -// CHECK-FLOAT_PTR-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[TMP1]], i64 [[ALIGNMENT]]) ] +// CHECK-FLOAT_PTR-NEXT: [[MASK1:%.*]] = sub i64 [[ALIGNMENT]], 1 +// CHECK-FLOAT_PTR-NEXT: [[PTRINT:%.*]] = ptrtoint float* [[TMP1]] to i64 +// CHECK-FLOAT_PTR-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK1]] +// CHECK-FLOAT_PTR-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-FLOAT_PTR-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-FLOAT_PTR-NEXT: ret float* [[TMP1]] // // CHECK-LONG-LABEL: define {{[^@]+}}@align_up @@ -176,7 +184,11 @@ TYPE align_up(TYPE ptr, unsigned align) { // CHECK-VOID_PTR-NEXT: [[ALIGNED_INTPTR:%.*]] = and i64 [[INTPTR]], [[INVERTED_MASK]] // CHECK-VOID_PTR-NEXT: [[DIFF:%.*]] = sub i64 [[ALIGNED_INTPTR]], [[INTPTR]] // CHECK-VOID_PTR-NEXT: [[ALIGNED_RESULT:%.*]] = getelementptr inbounds i8, i8* [[PTR]], i64 [[DIFF]] -// CHECK-VOID_PTR-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[ALIGNED_RESULT]], i64 [[ALIGNMENT]]) ] +// CHECK-VOID_PTR-NEXT: [[MASK1:%.*]] = sub i64 [[ALIGNMENT]], 1 +// CHECK-VOID_PTR-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[ALIGNED_RESULT]] to i64 +// CHECK-VOID_PTR-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK1]] +// CHECK-VOID_PTR-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-VOID_PTR-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-VOID_PTR-NEXT: ret i8* [[ALIGNED_RESULT]] // // CHECK-FLOAT_PTR-LABEL: define {{[^@]+}}@align_down @@ -191,7 +203,11 @@ TYPE align_up(TYPE ptr, unsigned align) { // CHECK-FLOAT_PTR-NEXT: [[TMP0:%.*]] = bitcast float* [[PTR]] to i8* // CHECK-FLOAT_PTR-NEXT: [[ALIGNED_RESULT:%.*]] = getelementptr inbounds i8, i8* [[TMP0]], i64 [[DIFF]] // CHECK-FLOAT_PTR-NEXT: [[TMP1:%.*]] = bitcast i8* [[ALIGNED_RESULT]] to float* -// CHECK-FLOAT_PTR-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[TMP1]], i64 [[ALIGNMENT]]) ] +// CHECK-FLOAT_PTR-NEXT: [[MASK1:%.*]] = sub i64 [[ALIGNMENT]], 1 +// CHECK-FLOAT_PTR-NEXT: [[PTRINT:%.*]] = ptrtoint float* [[TMP1]] to i64 +// CHECK-FLOAT_PTR-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK1]] +// CHECK-FLOAT_PTR-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-FLOAT_PTR-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-FLOAT_PTR-NEXT: ret float* [[TMP1]] // // CHECK-LONG-LABEL: define {{[^@]+}}@align_down diff --git a/clang/test/CodeGen/builtin-assume-aligned.c b/clang/test/CodeGen/builtin-assume-aligned.c index b9f1ebf..90693cc 100644 --- a/clang/test/CodeGen/builtin-assume-aligned.c +++ b/clang/test/CodeGen/builtin-assume-aligned.c @@ -8,7 +8,10 @@ // CHECK-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP1:%.*]] = bitcast i32* [[TMP0]] to i8* -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[TMP1]], i64 32, i64 0) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[TMP1]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 31 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32* // CHECK-NEXT: store i32* [[TMP2]], i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP3:%.*]] = load i32*, i32** [[A_ADDR]], align 8 @@ -28,7 +31,10 @@ int test1(int *a) { // CHECK-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP1:%.*]] = bitcast i32* [[TMP0]] to i8* -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[TMP1]], i64 32, i64 0) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[TMP1]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 31 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32* // CHECK-NEXT: store i32* [[TMP2]], i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP3:%.*]] = load i32*, i32** [[A_ADDR]], align 8 @@ -48,7 +54,10 @@ int test2(int *a) { // CHECK-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP1:%.*]] = bitcast i32* [[TMP0]] to i8* -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[TMP1]], i64 32) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[TMP1]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 31 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32* // CHECK-NEXT: store i32* [[TMP2]], i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP3:%.*]] = load i32*, i32** [[A_ADDR]], align 8 @@ -72,7 +81,11 @@ int test3(int *a) { // CHECK-NEXT: [[TMP1:%.*]] = bitcast i32* [[TMP0]] to i8* // CHECK-NEXT: [[TMP2:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK-NEXT: [[CONV:%.*]] = sext i32 [[TMP2]] to i64 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[TMP1]], i64 32, i64 [[CONV]]) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[TMP1]] to i64 +// CHECK-NEXT: [[OFFSETPTR:%.*]] = sub i64 [[PTRINT]], [[CONV]] +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[OFFSETPTR]], 31 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP1]] to i32* // CHECK-NEXT: store i32* [[TMP3]], i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP4:%.*]] = load i32*, i32** [[A_ADDR]], align 8 @@ -102,7 +115,11 @@ int *m2() __attribute__((assume_aligned(64, 12))); // CHECK-LABEL: define {{[^@]+}}@test6() #0 // CHECK-NEXT: entry: // CHECK-NEXT: [[CALL:%.*]] = call i32* (...) @m2() -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[CALL]], i64 64, i64 12) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[CALL]] to i64 +// CHECK-NEXT: [[OFFSETPTR:%.*]] = sub i64 [[PTRINT]], 12 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[OFFSETPTR]], 63 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[CALL]], align 4 // CHECK-NEXT: ret i32 [[TMP0]] // @@ -117,7 +134,10 @@ int test6() { // CHECK-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP1:%.*]] = bitcast i32* [[TMP0]] to i8* -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[TMP1]], i64 536870912) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[TMP1]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 536870911 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32* // CHECK-NEXT: store i32* [[TMP2]], i32** [[A_ADDR]], align 8 // CHECK-NEXT: [[TMP3:%.*]] = load i32*, i32** [[A_ADDR]], align 8 diff --git a/clang/test/CodeGen/catch-alignment-assumption-attribute-align_value-on-lvalue.cpp b/clang/test/CodeGen/catch-alignment-assumption-attribute-align_value-on-lvalue.cpp index fb2b1a7..96d2641 100644 --- a/clang/test/CodeGen/catch-alignment-assumption-attribute-align_value-on-lvalue.cpp +++ b/clang/test/CodeGen/catch-alignment-assumption-attribute-align_value-on-lvalue.cpp @@ -21,9 +21,9 @@ char **load_from_ac_struct(struct ac_struct *x) { // CHECK-NEXT: %[[X_RELOADED:.*]] = load %[[STRUCT_AC_STRUCT]]*, %[[STRUCT_AC_STRUCT]]** %[[STRUCT_AC_STRUCT_ADDR]], align 8 // CHECK: %[[A_ADDR:.*]] = getelementptr inbounds %[[STRUCT_AC_STRUCT]], %[[STRUCT_AC_STRUCT]]* %[[X_RELOADED]], i32 0, i32 0 // CHECK: %[[A:.*]] = load i8**, i8*** %[[A_ADDR]], align 8 - // CHECK-SANITIZE-NEXT: %[[PTRINT:.*]] = ptrtoint i8** %[[A]] to i64 - // CHECK-SANITIZE-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[PTRINT]], 2147483647 - // CHECK-SANITIZE-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 + // CHECK-NEXT: %[[PTRINT:.*]] = ptrtoint i8** %[[A]] to i64 + // CHECK-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[PTRINT]], 2147483647 + // CHECK-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 // CHECK-SANITIZE-NEXT: %[[PTRINT_DUP:.*]] = ptrtoint i8** %[[A]] to i64, !nosanitize // CHECK-SANITIZE-NEXT: br i1 %[[MASKCOND]], label %[[CONT:.*]], label %[[HANDLER_ALIGNMENT_ASSUMPTION:[^,]+]],{{.*}} !nosanitize // CHECK-SANITIZE: [[HANDLER_ALIGNMENT_ASSUMPTION]]: @@ -32,7 +32,7 @@ char **load_from_ac_struct(struct ac_struct *x) { // CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize // CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize // CHECK-SANITIZE: [[CONT]]: - // CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8** %[[A]], i64 2147483648) ] + // CHECK-NEXT: call void @llvm.assume(i1 %[[MASKCOND]]) // CHECK-NEXT: ret i8** %[[A]] // CHECK-NEXT: } #line 100 diff --git a/clang/test/CodeGen/catch-alignment-assumption-attribute-align_value-on-paramvar.cpp b/clang/test/CodeGen/catch-alignment-assumption-attribute-align_value-on-paramvar.cpp index 46f7d09..0e3fa75 100644 --- a/clang/test/CodeGen/catch-alignment-assumption-attribute-align_value-on-paramvar.cpp +++ b/clang/test/CodeGen/catch-alignment-assumption-attribute-align_value-on-paramvar.cpp @@ -24,7 +24,7 @@ char **passthrough(__attribute__((align_value(0x80000000))) char **x) { // CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize // CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize // CHECK-SANITIZE: [[CONT]]: - // CHECK-SANITIZE-NEXT: call void @llvm.assume(i1 true) [ "align"(i8** %[[X_RELOADED]], i64 2147483648) ] + // CHECK-SANITIZE-NEXT: call void @llvm.assume(i1 %[[MASKCOND]]) // CHECK-NEXT: ret i8** %[[X_RELOADED]] // CHECK-NEXT: } #line 100 diff --git a/clang/test/CodeGen/catch-alignment-assumption-attribute-alloc_align-on-function-variable.cpp b/clang/test/CodeGen/catch-alignment-assumption-attribute-alloc_align-on-function-variable.cpp index 40abbc3..591eaa0 100644 --- a/clang/test/CodeGen/catch-alignment-assumption-attribute-alloc_align-on-function-variable.cpp +++ b/clang/test/CodeGen/catch-alignment-assumption-attribute-alloc_align-on-function-variable.cpp @@ -30,10 +30,10 @@ char **caller(char **x, unsigned long alignment) { // CHECK-NEXT: %[[X_RELOADED:.*]] = load i8**, i8*** %[[X_ADDR]], align 8 // CHECK-NEXT: %[[ALIGNMENT_RELOADED:.*]] = load i64, i64* %[[ALIGNMENT_ADDR]], align 8 // CHECK-NEXT: %[[X_RETURNED:.*]] = call i8** @[[PASSTHROUGH]](i8** %[[X_RELOADED]], i64 %[[ALIGNMENT_RELOADED]]) - // CHECK-SANITIZE-NEXT: %[[PTRINT:.*]] = ptrtoint i8** %[[X_RETURNED]] to i64 - // CHECK-SANITIZE-NEXT: %[[MASK:.*]] = sub i64 %[[ALIGNMENT_RELOADED]], 1 - // CHECK-SANITIZE-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[PTRINT]], %[[MASK]] - // CHECK-SANITIZE-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 + // CHECK-NEXT: %[[MASK:.*]] = sub i64 %[[ALIGNMENT_RELOADED]], 1 + // CHECK-NEXT: %[[PTRINT:.*]] = ptrtoint i8** %[[X_RETURNED]] to i64 + // CHECK-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[PTRINT]], %[[MASK]] + // CHECK-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 // CHECK-SANITIZE-NEXT: %[[PTRINT_DUP:.*]] = ptrtoint i8** %[[X_RETURNED]] to i64, !nosanitize // CHECK-SANITIZE-NEXT: br i1 %[[MASKCOND]], label %[[CONT:.*]], label %[[HANDLER_ALIGNMENT_ASSUMPTION:[^,]+]],{{.*}} !nosanitize // CHECK-SANITIZE: [[HANDLER_ALIGNMENT_ASSUMPTION]]: @@ -42,7 +42,7 @@ char **caller(char **x, unsigned long alignment) { // CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize // CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize // CHECK-SANITIZE: [[CONT]]: - // CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8** %[[X_RETURNED]], i64 %1) ] + // CHECK-NEXT: call void @llvm.assume(i1 %[[MASKCOND]]) // CHECK-NEXT: ret i8** %[[X_RETURNED]] // CHECK-NEXT: } #line 100 diff --git a/clang/test/CodeGen/catch-alignment-assumption-attribute-alloc_align-on-function.cpp b/clang/test/CodeGen/catch-alignment-assumption-attribute-alloc_align-on-function.cpp index 87d903c..a413579 100644 --- a/clang/test/CodeGen/catch-alignment-assumption-attribute-alloc_align-on-function.cpp +++ b/clang/test/CodeGen/catch-alignment-assumption-attribute-alloc_align-on-function.cpp @@ -39,7 +39,7 @@ char **caller(char **x) { // CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize // CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize // CHECK-SANITIZE: [[CONT]]: - // CHECK-SANITIZE-NEXT: call void @llvm.assume(i1 true) [ "align"(i8** %[[X_RETURNED]], i64 128) ] + // CHECK-SANITIZE-NEXT: call void @llvm.assume(i1 %[[MASKCOND]]) // CHECK-NEXT: ret i8** %[[X_RETURNED]] // CHECK-NEXT: } #line 100 diff --git a/clang/test/CodeGen/catch-alignment-assumption-attribute-assume_aligned-on-function-two-params.cpp b/clang/test/CodeGen/catch-alignment-assumption-attribute-assume_aligned-on-function-two-params.cpp index ecc96bc..e78667c 100644 --- a/clang/test/CodeGen/catch-alignment-assumption-attribute-assume_aligned-on-function-two-params.cpp +++ b/clang/test/CodeGen/catch-alignment-assumption-attribute-assume_aligned-on-function-two-params.cpp @@ -24,10 +24,10 @@ char **caller(char **x) { // CHECK-NEXT: store i8** %[[X]], i8*** %[[X_ADDR]], align 8 // CHECK-NEXT: %[[X_RELOADED:.*]] = load i8**, i8*** %[[X_ADDR]], align 8 // CHECK-NEXT: %[[X_RETURNED:.*]] = call i8** @[[PASSTHROUGH]](i8** %[[X_RELOADED]]) - // CHECK-SANITIZE-NEXT: %[[PTRINT:.*]] = ptrtoint i8** %[[X_RETURNED]] to i64 - // CHECK-SANITIZE-NEXT: %[[OFFSETPTR:.*]] = sub i64 %[[PTRINT]], 42 - // CHECK-SANITIZE-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[OFFSETPTR]], 2147483647 - // CHECK-SANITIZE-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 + // CHECK-NEXT: %[[PTRINT:.*]] = ptrtoint i8** %[[X_RETURNED]] to i64 + // CHECK-NEXT: %[[OFFSETPTR:.*]] = sub i64 %[[PTRINT]], 42 + // CHECK-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[OFFSETPTR]], 2147483647 + // CHECK-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 // CHECK-SANITIZE-NEXT: %[[PTRINT_DUP:.*]] = ptrtoint i8** %[[X_RETURNED]] to i64, !nosanitize // CHECK-SANITIZE-NEXT: br i1 %[[MASKCOND]], label %[[CONT:.*]], label %[[HANDLER_ALIGNMENT_ASSUMPTION:[^,]+]],{{.*}} !nosanitize // CHECK-SANITIZE: [[HANDLER_ALIGNMENT_ASSUMPTION]]: @@ -36,7 +36,7 @@ char **caller(char **x) { // CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize // CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize // CHECK-SANITIZE: [[CONT]]: - // CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8** %[[X_RETURNED]], i64 2147483648, i64 42) ] + // CHECK-NEXT: call void @llvm.assume(i1 %[[MASKCOND]]) // CHECK-NEXT: ret i8** %[[X_RETURNED]] // CHECK-NEXT: } #line 100 diff --git a/clang/test/CodeGen/catch-alignment-assumption-attribute-assume_aligned-on-function.cpp b/clang/test/CodeGen/catch-alignment-assumption-attribute-assume_aligned-on-function.cpp index 5bbc584..f750bbd 100644 --- a/clang/test/CodeGen/catch-alignment-assumption-attribute-assume_aligned-on-function.cpp +++ b/clang/test/CodeGen/catch-alignment-assumption-attribute-assume_aligned-on-function.cpp @@ -36,7 +36,7 @@ char **caller(char **x) { // CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize // CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize // CHECK-SANITIZE: [[CONT]]: - // CHECK-SANITIZE-NEXT: call void @llvm.assume(i1 true) [ "align"(i8** %[[X_RETURNED]], i64 128) ] + // CHECK-SANITIZE-NEXT: call void @llvm.assume(i1 %[[MASKCOND]]) // CHECK-NEXT: ret i8** %[[X_RETURNED]] // CHECK-NEXT: } #line 100 diff --git a/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-three-params-variable.cpp b/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-three-params-variable.cpp index 9c8944b..4306e32 100644 --- a/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-three-params-variable.cpp +++ b/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-three-params-variable.cpp @@ -16,10 +16,10 @@ void *caller(char **x, unsigned long offset) { // CHECK-NEXT: %[[X_RELOADED:.*]] = load i8**, i8*** %[[X_ADDR]], align 8 // CHECK-NEXT: %[[BITCAST:.*]] = bitcast i8** %[[X_RELOADED]] to i8* // CHECK-NEXT: %[[OFFSET_RELOADED:.*]] = load i64, i64* %[[OFFSET_ADDR]], align 8 - // CHECK-SANITIZE-NEXT: %[[PTRINT:.*]] = ptrtoint i8* %[[BITCAST]] to i64 - // CHECK-SANITIZE-NEXT: %[[OFFSETPTR:.*]] = sub i64 %[[PTRINT]], %[[OFFSET_RELOADED]] - // CHECK-SANITIZE-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[OFFSETPTR]], 536870911 - // CHECK-SANITIZE-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 + // CHECK-NEXT: %[[PTRINT:.*]] = ptrtoint i8* %[[BITCAST]] to i64 + // CHECK-NEXT: %[[OFFSETPTR:.*]] = sub i64 %[[PTRINT]], %[[OFFSET_RELOADED]] + // CHECK-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[OFFSETPTR]], 536870911 + // CHECK-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 // CHECK-SANITIZE-NEXT: %[[PTRINT_DUP:.*]] = ptrtoint i8* %[[BITCAST]] to i64, !nosanitize // CHECK-SANITIZE-NEXT: br i1 %[[MASKCOND]], label %[[CONT:.*]], label %[[HANDLER_ALIGNMENT_ASSUMPTION:[^,]+]],{{.*}} !nosanitize // CHECK-SANITIZE: [[HANDLER_ALIGNMENT_ASSUMPTION]]: @@ -28,7 +28,7 @@ void *caller(char **x, unsigned long offset) { // CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize // CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize // CHECK-SANITIZE: [[CONT]]: - // CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* %[[BITCAST]], i64 536870912, i64 %[[OFFSET_RELOADED]]) ] + // CHECK-NEXT: call void @llvm.assume(i1 %[[MASKCOND]]) // CHECK-NEXT: ret i8* %[[BITCAST]] // CHECK-NEXT: } #line 100 diff --git a/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-three-params.cpp b/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-three-params.cpp index 9f61e08..27f53e9 100644 --- a/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-three-params.cpp +++ b/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-three-params.cpp @@ -13,10 +13,10 @@ void *caller(char **x) { // CHECK-NEXT: store i8** %[[X]], i8*** %[[X_ADDR]], align 8 // CHECK-NEXT: %[[X_RELOADED:.*]] = load i8**, i8*** %[[X_ADDR]], align 8 // CHECK-NEXT: %[[BITCAST:.*]] = bitcast i8** %[[X_RELOADED]] to i8* - // CHECK-SANITIZE-NEXT: %[[PTRINT:.*]] = ptrtoint i8* %[[BITCAST]] to i64 - // CHECK-SANITIZE-NEXT: %[[OFFSETPTR:.*]] = sub i64 %[[PTRINT]], 42 - // CHECK-SANITIZE-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[OFFSETPTR]], 536870911 - // CHECK-SANITIZE-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 + // CHECK-NEXT: %[[PTRINT:.*]] = ptrtoint i8* %[[BITCAST]] to i64 + // CHECK-NEXT: %[[OFFSETPTR:.*]] = sub i64 %[[PTRINT]], 42 + // CHECK-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[OFFSETPTR]], 536870911 + // CHECK-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 // CHECK-SANITIZE-NEXT: %[[PTRINT_DUP:.*]] = ptrtoint i8* %[[BITCAST]] to i64, !nosanitize // CHECK-SANITIZE-NEXT: br i1 %[[MASKCOND]], label %[[CONT:.*]], label %[[HANDLER_ALIGNMENT_ASSUMPTION:[^,]+]],{{.*}} !nosanitize // CHECK-SANITIZE: [[HANDLER_ALIGNMENT_ASSUMPTION]]: @@ -25,7 +25,7 @@ void *caller(char **x) { // CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize // CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize // CHECK-SANITIZE: [[CONT]]: - // CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* %[[BITCAST]], i64 536870912, i64 42) ] + // CHECK-NEXT: call void @llvm.assume(i1 %[[MASKCOND]]) // CHECK-NEXT: ret i8* %[[BITCAST]] // CHECK-NEXT: } #line 100 diff --git a/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-two-params.cpp b/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-two-params.cpp index 20bed64..5412270 100644 --- a/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-two-params.cpp +++ b/clang/test/CodeGen/catch-alignment-assumption-builtin_assume_aligned-two-params.cpp @@ -13,9 +13,9 @@ void *caller(char **x) { // CHECK-NEXT: store i8** %[[X]], i8*** %[[X_ADDR]], align 8 // CHECK-NEXT: %[[X_RELOADED:.*]] = load i8**, i8*** %[[X_ADDR]], align 8 // CHECK-NEXT: %[[BITCAST:.*]] = bitcast i8** %[[X_RELOADED]] to i8* - // CHECK-SANITIZE-NEXT: %[[PTRINT:.*]] = ptrtoint i8* %[[BITCAST]] to i64 - // CHECK-SANITIZE-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[PTRINT]], 536870911 - // CHECK-SANITIZE-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 + // CHECK-NEXT: %[[PTRINT:.*]] = ptrtoint i8* %[[BITCAST]] to i64 + // CHECK-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[PTRINT]], 536870911 + // CHECK-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 // CHECK-SANITIZE-NEXT: %[[PTRINT_DUP:.*]] = ptrtoint i8* %[[BITCAST]] to i64, !nosanitize // CHECK-SANITIZE-NEXT: br i1 %[[MASKCOND]], label %[[CONT:.*]], label %[[HANDLER_ALIGNMENT_ASSUMPTION:[^,]+]],{{.*}} !nosanitize // CHECK-SANITIZE: [[HANDLER_ALIGNMENT_ASSUMPTION]]: @@ -24,7 +24,7 @@ void *caller(char **x) { // CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize // CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize // CHECK-SANITIZE: [[CONT]]: - // CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* %[[BITCAST]], i64 536870912) ] + // CHECK-NEXT: call void @llvm.assume(i1 %[[MASKCOND]]) // CHECK-NEXT: ret i8* %[[BITCAST]] // CHECK-NEXT: } #line 100 diff --git a/clang/test/CodeGen/catch-alignment-assumption-openmp.cpp b/clang/test/CodeGen/catch-alignment-assumption-openmp.cpp index 353f2fd..6d75ee0 100644 --- a/clang/test/CodeGen/catch-alignment-assumption-openmp.cpp +++ b/clang/test/CodeGen/catch-alignment-assumption-openmp.cpp @@ -12,9 +12,9 @@ void func(char *data) { // CHECK-NEXT: %[[DATA_ADDR:.*]] = alloca i8*, align 8 // CHECK: store i8* %[[DATA]], i8** %[[DATA_ADDR]], align 8 // CHECK: %[[DATA_RELOADED:.*]] = load i8*, i8** %[[DATA_ADDR]], align 8 - // CHECK-SANITIZE-NEXT: %[[PTRINT:.*]] = ptrtoint i8* %[[DATA_RELOADED]] to i64 - // CHECK-SANITIZE-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[PTRINT]], 1073741823 - // CHECK-SANITIZE-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 + // CHECK-NEXT: %[[PTRINT:.*]] = ptrtoint i8* %[[DATA_RELOADED]] to i64 + // CHECK-NEXT: %[[MASKEDPTR:.*]] = and i64 %[[PTRINT]], 1073741823 + // CHECK-NEXT: %[[MASKCOND:.*]] = icmp eq i64 %[[MASKEDPTR]], 0 // CHECK-SANITIZE-NEXT: %[[PTRINT_DUP:.*]] = ptrtoint i8* %[[DATA_RELOADED]] to i64, !nosanitize // CHECK-SANITIZE-NEXT: br i1 %[[MASKCOND]], label %[[CONT:.*]], label %[[HANDLER_ALIGNMENT_ASSUMPTION:[^,]+]],{{.*}} !nosanitize // CHECK-SANITIZE: [[HANDLER_ALIGNMENT_ASSUMPTION]]: @@ -23,7 +23,7 @@ void func(char *data) { // CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize // CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize // CHECK-SANITIZE: [[CONT]]: - // CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* %[[DATA_RELOADED]], i64 1073741824) ] + // CHECK-NEXT: call void @llvm.assume(i1 %[[MASKCOND]]) #line 100 #pragma omp for simd aligned(data : 0x40000000) diff --git a/clang/test/CodeGen/non-power-of-2-alignment-assumptions.c b/clang/test/CodeGen/non-power-of-2-alignment-assumptions.c index b8ce169..9467f62 100644 --- a/clang/test/CodeGen/non-power-of-2-alignment-assumptions.c +++ b/clang/test/CodeGen/non-power-of-2-alignment-assumptions.c @@ -9,8 +9,12 @@ void *__attribute__((alloc_align(1))) alloc(int align); // CHECK-NEXT: store i32 [[ALIGN:%.*]], i32* [[ALIGN_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[ALIGN_ADDR]], align 4 // CHECK-NEXT: [[CALL:%.*]] = call i8* @alloc(i32 [[TMP0]]) -// CHECK-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[CALL]], i64 [[TMP1]]) ] +// CHECK-NEXT: [[ALIGNMENTCAST:%.*]] = zext i32 [[TMP0]] to i64 +// CHECK-NEXT: [[MASK:%.*]] = sub i64 [[ALIGNMENTCAST]], 1 +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[CALL]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], [[MASK]] +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: ret void // void t0(int align) { @@ -21,7 +25,10 @@ void t0(int align) { // CHECK-NEXT: [[ALIGN_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: store i32 [[ALIGN:%.*]], i32* [[ALIGN_ADDR]], align 4 // CHECK-NEXT: [[CALL:%.*]] = call i8* @alloc(i32 7) -// CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i8* [[CALL]], i64 7) ] +// CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i8* [[CALL]] to i64 +// CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 6 +// CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) // CHECK-NEXT: ret void // void t1(int align) { diff --git a/clang/test/OpenMP/simd_codegen.cpp b/clang/test/OpenMP/simd_codegen.cpp index 335dfd7..8ba87dc 100644 --- a/clang/test/OpenMP/simd_codegen.cpp +++ b/clang/test/OpenMP/simd_codegen.cpp @@ -817,9 +817,25 @@ void parallel_simd(float *a) { // TERM_DEBUG: !{{[0-9]+}} = !DILocation(line: [[@LINE-11]], // CHECK-LABEL: S8 +// CHECK-DAG: ptrtoint [[SS_TY]]* %{{.+}} to i64 +// CHECK-DAG: ptrtoint [[SS_TY]]* %{{.+}} to i64 +// CHECK-DAG: ptrtoint [[SS_TY]]* %{{.+}} to i64 +// CHECK-DAG: ptrtoint [[SS_TY]]* %{{.+}} to i64 + +// CHECK-DAG: and i64 %{{.+}}, 15 +// CHECK-DAG: icmp eq i64 %{{.+}}, 0 // CHECK-DAG: call void @llvm.assume(i1 + +// CHECK-DAG: and i64 %{{.+}}, 7 +// CHECK-DAG: icmp eq i64 %{{.+}}, 0 // CHECK-DAG: call void @llvm.assume(i1 + +// CHECK-DAG: and i64 %{{.+}}, 15 +// CHECK-DAG: icmp eq i64 %{{.+}}, 0 // CHECK-DAG: call void @llvm.assume(i1 + +// CHECK-DAG: and i64 %{{.+}}, 3 +// CHECK-DAG: icmp eq i64 %{{.+}}, 0 // CHECK-DAG: call void @llvm.assume(i1 struct SS { SS(): a(0) {} diff --git a/clang/test/OpenMP/simd_metadata.c b/clang/test/OpenMP/simd_metadata.c index 18133e3..f0ae020 100644 --- a/clang/test/OpenMP/simd_metadata.c +++ b/clang/test/OpenMP/simd_metadata.c @@ -21,21 +21,30 @@ void h1(float *c, float *a, double b[], int size) // CHECK-LABEL: define void @h1 int t = 0; #pragma omp simd safelen(16) linear(t) aligned(c:32) aligned(a,b) - // CHECK: call void @llvm.assume(i1 true) [ "align"(float* [[PTR4:%.*]], {{i64|i32}} 32) ] - // CHECK-NEXT: load - - // X86-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 16) ] - // X86-AVX-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 32) ] - // X86-AVX512-NEXT:call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 64) ] - // PPC-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 16) ] - // PPC-QPX-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 16) ] - // CHECK-NEXT: load - - // X86-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 16) ] - // X86-AVX-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 32) ] - // X86-AVX512-NEXT:call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 64) ] - // PPC-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 16) ] - // PPC-QPX-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 32) ] +// CHECK: [[C_PTRINT:%.+]] = ptrtoint +// CHECK-NEXT: [[C_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[C_PTRINT]], 31 +// CHECK-NEXT: [[C_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[C_MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[C_MASKCOND]]) +// CHECK: [[A_PTRINT:%.+]] = ptrtoint + +// X86-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15 +// X86-AVX-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 31 +// X86-AVX512-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 63 +// PPC-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15 +// PPC-QPX-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15 + +// CHECK-NEXT: [[A_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[A_MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[A_MASKCOND]]) +// CHECK: [[B_PTRINT:%.+]] = ptrtoint + +// X86-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 15 +// X86-AVX-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 31 +// X86-AVX512-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 63 +// PPC-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 15 +// PPC-QPX-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 31 + +// CHECK-NEXT: [[B_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[B_MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[B_MASKCOND]]) for (int i = 0; i < size; ++i) { c[i] = a[i] * a[i] + b[i] * b[t]; ++t; @@ -43,21 +52,30 @@ void h1(float *c, float *a, double b[], int size) // do not emit llvm.access.group metadata due to usage of safelen clause. // CHECK-NOT: store float {{.+}}, float* {{.+}}, align {{.+}}, !llvm.access.group {{![0-9]+}} #pragma omp simd safelen(16) linear(t) aligned(c:32) aligned(a,b) simdlen(8) - // CHECK: call void @llvm.assume(i1 true) [ "align"(float* [[PTR4:%.*]], {{i64|i32}} 32) ] - // CHECK-NEXT: load - - // X86-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 16) ] - // X86-AVX-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 32) ] - // X86-AVX512-NEXT:call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 64) ] - // PPC-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 16) ] - // PPC-QPX-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 16) ] - // CHECK-NEXT: load - - // X86-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 16) ] - // X86-AVX-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 32) ] - // X86-AVX512-NEXT:call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 64) ] - // PPC-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 16) ] - // PPC-QPX-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 32) ] +// CHECK: [[C_PTRINT:%.+]] = ptrtoint +// CHECK-NEXT: [[C_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[C_PTRINT]], 31 +// CHECK-NEXT: [[C_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[C_MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[C_MASKCOND]]) +// CHECK: [[A_PTRINT:%.+]] = ptrtoint + +// X86-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15 +// X86-AVX-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 31 +// X86-AVX512-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 63 +// PPC-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15 +// PPC-QPX-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15 + +// CHECK-NEXT: [[A_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[A_MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[A_MASKCOND]]) +// CHECK: [[B_PTRINT:%.+]] = ptrtoint + +// X86-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 15 +// X86-AVX-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 31 +// X86-AVX512-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 63 +// PPC-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 15 +// PPC-QPX-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 31 + +// CHECK-NEXT: [[B_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[B_MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[B_MASKCOND]]) for (int i = 0; i < size; ++i) { c[i] = a[i] * a[i] + b[i] * b[t]; ++t; @@ -65,21 +83,30 @@ void h1(float *c, float *a, double b[], int size) // do not emit llvm.access.group metadata due to usage of safelen clause. // CHECK-NOT: store float {{.+}}, float* {{.+}}, align {{.+}}, !llvm.access.group {{![0-9]+}} #pragma omp simd linear(t) aligned(c:32) aligned(a,b) simdlen(8) - // CHECK: call void @llvm.assume(i1 true) [ "align"(float* [[PTR4:%.*]], {{i64|i32}} 32) ] - // CHECK-NEXT: load - - // X86-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 16) ] - // X86-AVX-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 32) ] - // X86-AVX512-NEXT:call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 64) ] - // PPC-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 16) ] - // PPC-QPX-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[PTR5:%.*]], {{i64|i32}} 16) ] - // CHECK-NEXT: load - - // X86-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 16) ] - // X86-AVX-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 32) ] - // X86-AVX512-NEXT:call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 64) ] - // PPC-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 16) ] - // PPC-QPX-NEXT: call void @llvm.assume(i1 true) [ "align"(double* [[PTR6:%.*]], {{i64|i32}} 32) ] +// CHECK: [[C_PTRINT:%.+]] = ptrtoint +// CHECK-NEXT: [[C_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[C_PTRINT]], 31 +// CHECK-NEXT: [[C_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[C_MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[C_MASKCOND]]) +// CHECK: [[A_PTRINT:%.+]] = ptrtoint + +// X86-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15 +// X86-AVX-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 31 +// X86-AVX512-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 63 +// PPC-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15 +// PPC-QPX-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15 + +// CHECK-NEXT: [[A_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[A_MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[A_MASKCOND]]) +// CHECK: [[B_PTRINT:%.+]] = ptrtoint + +// X86-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 15 +// X86-AVX-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 31 +// X86-AVX512-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 63 +// PPC-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 15 +// PPC-QPX-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 31 + +// CHECK-NEXT: [[B_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[B_MASKEDPTR]], 0 +// CHECK-NEXT: call void @llvm.assume(i1 [[B_MASKCOND]]) for (int i = 0; i < size; ++i) { c[i] = a[i] * a[i] + b[i] * b[t]; ++t; diff --git a/clang/test/OpenMP/target_teams_distribute_parallel_for_simd_codegen.cpp b/clang/test/OpenMP/target_teams_distribute_parallel_for_simd_codegen.cpp index 7dff119..d2031d6 100644 --- a/clang/test/OpenMP/target_teams_distribute_parallel_for_simd_codegen.cpp +++ b/clang/test/OpenMP/target_teams_distribute_parallel_for_simd_codegen.cpp @@ -101,7 +101,10 @@ int target_teams_fun(int *g){ // CK1: define internal void @[[OUTL1]]({{.+}}) // CK1: [[ARRDECAY:%.+]] = getelementptr inbounds [1000 x i32], [1000 x i32]* %{{.+}}, i{{32|64}} 0, i{{32|64}} 0 - // CK1: call void @llvm.assume(i1 true) [ "align"(i32* [[ARRDECAY]], {{i64|i32}} 8) ] + // CK1: [[ARR_CAST:%.+]] = ptrtoint i32* [[ARRDECAY]] to i{{32|64}} + // CK1: [[MASKED_PTR:%.+]] = and i{{32|64}} [[ARR_CAST]], 7 + // CK1: [[COND:%.+]] = icmp eq i{{32|64}} [[MASKED_PTR]], 0 + // CK1: call void @llvm.assume(i1 [[COND]]) // CK1: call void @__kmpc_for_static_init_4( // CK1: call void {{.+}} @__kmpc_fork_call( // CK1: call void @__kmpc_for_static_fini( diff --git a/llvm/include/llvm/IR/IRBuilder.h b/llvm/include/llvm/IR/IRBuilder.h index 4552ca0..ffec4ff 100644 --- a/llvm/include/llvm/IR/IRBuilder.h +++ b/llvm/include/llvm/IR/IRBuilder.h @@ -782,11 +782,7 @@ public: /// Create an assume intrinsic call that allows the optimizer to /// assume that the provided condition will be true. - /// - /// The optional argument \p OpBundles specifies operand bundles that are - /// added to the call instruction. - CallInst *CreateAssumption(Value *Cond, - ArrayRef OpBundles = llvm::None); + CallInst *CreateAssumption(Value *Cond); /// Create a call to the experimental.gc.statepoint intrinsic to /// start a new statepoint sequence. @@ -2506,11 +2502,13 @@ public: private: /// Helper function that creates an assume intrinsic call that - /// represents an alignment assumption on the provided pointer \p PtrValue - /// with offset \p OffsetValue and alignment value \p AlignValue. + /// represents an alignment assumption on the provided Ptr, Mask, Type + /// and Offset. It may be sometimes useful to do some other logic + /// based on this alignment check, thus it can be stored into 'TheCheck'. CallInst *CreateAlignmentAssumptionHelper(const DataLayout &DL, - Value *PtrValue, Value *AlignValue, - Value *OffsetValue); + Value *PtrValue, Value *Mask, + Type *IntPtrTy, Value *OffsetValue, + Value **TheCheck); public: /// Create an assume intrinsic call that represents an alignment @@ -2519,9 +2517,13 @@ public: /// An optional offset can be provided, and if it is provided, the offset /// must be subtracted from the provided pointer to get the pointer with the /// specified alignment. + /// + /// It may be sometimes useful to do some other logic + /// based on this alignment check, thus it can be stored into 'TheCheck'. CallInst *CreateAlignmentAssumption(const DataLayout &DL, Value *PtrValue, unsigned Alignment, - Value *OffsetValue = nullptr); + Value *OffsetValue = nullptr, + Value **TheCheck = nullptr); /// Create an assume intrinsic call that represents an alignment /// assumption on the provided pointer. @@ -2530,11 +2532,15 @@ public: /// must be subtracted from the provided pointer to get the pointer with the /// specified alignment. /// + /// It may be sometimes useful to do some other logic + /// based on this alignment check, thus it can be stored into 'TheCheck'. + /// /// This overload handles the condition where the Alignment is dependent /// on an existing value rather than a static value. CallInst *CreateAlignmentAssumption(const DataLayout &DL, Value *PtrValue, Value *Alignment, - Value *OffsetValue = nullptr); + Value *OffsetValue = nullptr, + Value **TheCheck = nullptr); }; /// This provides a uniform API for creating instructions and inserting diff --git a/llvm/include/llvm/Transforms/Scalar/AlignmentFromAssumptions.h b/llvm/include/llvm/Transforms/Scalar/AlignmentFromAssumptions.h index 10b6e1c..be119b8 100644 --- a/llvm/include/llvm/Transforms/Scalar/AlignmentFromAssumptions.h +++ b/llvm/include/llvm/Transforms/Scalar/AlignmentFromAssumptions.h @@ -37,9 +37,9 @@ struct AlignmentFromAssumptionsPass ScalarEvolution *SE = nullptr; DominatorTree *DT = nullptr; - bool extractAlignmentInfo(CallInst *I, unsigned Idx, Value *&AAPtr, - const SCEV *&AlignSCEV, const SCEV *&OffSCEV); - bool processAssumption(CallInst *I, unsigned Idx); + bool extractAlignmentInfo(CallInst *I, Value *&AAPtr, const SCEV *&AlignSCEV, + const SCEV *&OffSCEV); + bool processAssumption(CallInst *I); }; } diff --git a/llvm/lib/Analysis/AssumeBundleQueries.cpp b/llvm/lib/Analysis/AssumeBundleQueries.cpp index 05fe05a..972d0d3 100644 --- a/llvm/lib/Analysis/AssumeBundleQueries.cpp +++ b/llvm/lib/Analysis/AssumeBundleQueries.cpp @@ -108,17 +108,10 @@ llvm::getKnowledgeFromBundle(CallInst &Assume, Result.AttrKind = Attribute::getAttrKindFromName(BOI.Tag->getKey()); if (bundleHasArgument(BOI, ABA_WasOn)) Result.WasOn = getValueFromBundleOpInfo(Assume, BOI, ABA_WasOn); - auto GetArgOr1 = [&](unsigned Idx) -> unsigned { - if (auto *ConstInt = dyn_cast( - getValueFromBundleOpInfo(Assume, BOI, ABA_Argument + Idx))) - return ConstInt->getZExtValue(); - return 1; - }; if (BOI.End - BOI.Begin > ABA_Argument) - Result.ArgValue = GetArgOr1(0); - if (Result.AttrKind == Attribute::Alignment) - if (BOI.End - BOI.Begin > ABA_Argument + 1) - Result.ArgValue = MinAlign(Result.ArgValue, GetArgOr1(1)); + Result.ArgValue = + cast(getValueFromBundleOpInfo(Assume, BOI, ABA_Argument)) + ->getZExtValue(); return Result; } diff --git a/llvm/lib/IR/IRBuilder.cpp b/llvm/lib/IR/IRBuilder.cpp index b87dfe1..1fffce0 100644 --- a/llvm/lib/IR/IRBuilder.cpp +++ b/llvm/lib/IR/IRBuilder.cpp @@ -71,9 +71,8 @@ Value *IRBuilderBase::getCastedInt8PtrValue(Value *Ptr) { static CallInst *createCallHelper(Function *Callee, ArrayRef Ops, IRBuilderBase *Builder, const Twine &Name = "", - Instruction *FMFSource = nullptr, - ArrayRef OpBundles = {}) { - CallInst *CI = Builder->CreateCall(Callee, Ops, OpBundles, Name); + Instruction *FMFSource = nullptr) { + CallInst *CI = Builder->CreateCall(Callee, Ops, Name); if (FMFSource) CI->copyFastMathFlags(FMFSource); return CI; @@ -450,16 +449,14 @@ CallInst *IRBuilderBase::CreateInvariantStart(Value *Ptr, ConstantInt *Size) { return createCallHelper(TheFn, Ops, this); } -CallInst * -IRBuilderBase::CreateAssumption(Value *Cond, - ArrayRef OpBundles) { +CallInst *IRBuilderBase::CreateAssumption(Value *Cond) { assert(Cond->getType() == getInt1Ty() && "an assumption condition must be of type i1"); Value *Ops[] = { Cond }; Module *M = BB->getParent()->getParent(); Function *FnAssume = Intrinsic::getDeclaration(M, Intrinsic::assume); - return createCallHelper(FnAssume, Ops, this, "", nullptr, OpBundles); + return createCallHelper(FnAssume, Ops, this); } /// Create a call to a Masked Load intrinsic. @@ -1110,37 +1107,63 @@ Value *IRBuilderBase::CreatePreserveStructAccessIndex( return Fn; } -CallInst *IRBuilderBase::CreateAlignmentAssumptionHelper(const DataLayout &DL, - Value *PtrValue, - Value *AlignValue, - Value *OffsetValue) { - SmallVector Vals({PtrValue, AlignValue}); - if (OffsetValue) - Vals.push_back(OffsetValue); - OperandBundleDefT AlignOpB("align", Vals); - return CreateAssumption(ConstantInt::getTrue(getContext()), {AlignOpB}); +CallInst *IRBuilderBase::CreateAlignmentAssumptionHelper( + const DataLayout &DL, Value *PtrValue, Value *Mask, Type *IntPtrTy, + Value *OffsetValue, Value **TheCheck) { + Value *PtrIntValue = CreatePtrToInt(PtrValue, IntPtrTy, "ptrint"); + + if (OffsetValue) { + bool IsOffsetZero = false; + if (const auto *CI = dyn_cast(OffsetValue)) + IsOffsetZero = CI->isZero(); + + if (!IsOffsetZero) { + if (OffsetValue->getType() != IntPtrTy) + OffsetValue = CreateIntCast(OffsetValue, IntPtrTy, /*isSigned*/ true, + "offsetcast"); + PtrIntValue = CreateSub(PtrIntValue, OffsetValue, "offsetptr"); + } + } + + Value *Zero = ConstantInt::get(IntPtrTy, 0); + Value *MaskedPtr = CreateAnd(PtrIntValue, Mask, "maskedptr"); + Value *InvCond = CreateICmpEQ(MaskedPtr, Zero, "maskcond"); + if (TheCheck) + *TheCheck = InvCond; + + return CreateAssumption(InvCond); } -CallInst *IRBuilderBase::CreateAlignmentAssumption(const DataLayout &DL, - Value *PtrValue, - unsigned Alignment, - Value *OffsetValue) { +CallInst *IRBuilderBase::CreateAlignmentAssumption( + const DataLayout &DL, Value *PtrValue, unsigned Alignment, + Value *OffsetValue, Value **TheCheck) { assert(isa(PtrValue->getType()) && "trying to create an alignment assumption on a non-pointer?"); assert(Alignment != 0 && "Invalid Alignment"); auto *PtrTy = cast(PtrValue->getType()); Type *IntPtrTy = getIntPtrTy(DL, PtrTy->getAddressSpace()); - Value *AlignValue = ConstantInt::get(IntPtrTy, Alignment); - return CreateAlignmentAssumptionHelper(DL, PtrValue, AlignValue, OffsetValue); + + Value *Mask = ConstantInt::get(IntPtrTy, Alignment - 1); + return CreateAlignmentAssumptionHelper(DL, PtrValue, Mask, IntPtrTy, + OffsetValue, TheCheck); } -CallInst *IRBuilderBase::CreateAlignmentAssumption(const DataLayout &DL, - Value *PtrValue, - Value *Alignment, - Value *OffsetValue) { +CallInst *IRBuilderBase::CreateAlignmentAssumption( + const DataLayout &DL, Value *PtrValue, Value *Alignment, + Value *OffsetValue, Value **TheCheck) { assert(isa(PtrValue->getType()) && "trying to create an alignment assumption on a non-pointer?"); - return CreateAlignmentAssumptionHelper(DL, PtrValue, Alignment, OffsetValue); + auto *PtrTy = cast(PtrValue->getType()); + Type *IntPtrTy = getIntPtrTy(DL, PtrTy->getAddressSpace()); + + if (Alignment->getType() != IntPtrTy) + Alignment = CreateIntCast(Alignment, IntPtrTy, /*isSigned*/ false, + "alignmentcast"); + + Value *Mask = CreateSub(Alignment, ConstantInt::get(IntPtrTy, 1), "mask"); + + return CreateAlignmentAssumptionHelper(DL, PtrValue, Mask, IntPtrTy, + OffsetValue, TheCheck); } IRBuilderDefaultInserter::~IRBuilderDefaultInserter() {} diff --git a/llvm/lib/IR/Verifier.cpp b/llvm/lib/IR/Verifier.cpp index 6df1072..c518ae8 100644 --- a/llvm/lib/IR/Verifier.cpp +++ b/llvm/lib/IR/Verifier.cpp @@ -4449,32 +4449,21 @@ void Verifier::visitIntrinsicCall(Intrinsic::ID ID, CallBase &Call) { Assert(Elem.Tag->getKey() == "ignore" || Attribute::isExistingAttribute(Elem.Tag->getKey()), "tags must be valid attribute names"); + Assert(Elem.End - Elem.Begin <= 2, "to many arguments"); Attribute::AttrKind Kind = Attribute::getAttrKindFromName(Elem.Tag->getKey()); - unsigned ArgCount = Elem.End - Elem.Begin; - if (Kind == Attribute::Alignment) { - Assert(ArgCount <= 3 && ArgCount >= 2, - "alignment assumptions should have 2 or 3 arguments"); - Assert(Call.getOperand(Elem.Begin)->getType()->isPointerTy(), - "first argument should be a pointer"); - Assert(Call.getOperand(Elem.Begin + 1)->getType()->isIntegerTy(), - "second argument should be an integer"); - if (ArgCount == 3) - Assert(Call.getOperand(Elem.Begin + 2)->getType()->isIntegerTy(), - "third argument should be an integer if present"); - return; - } - Assert(ArgCount <= 2, "to many arguments"); if (Kind == Attribute::None) break; if (Attribute::doesAttrKindHaveArgument(Kind)) { - Assert(ArgCount == 2, "this attribute should have 2 arguments"); + Assert(Elem.End - Elem.Begin == 2, + "this attribute should have 2 arguments"); Assert(isa(Call.getOperand(Elem.Begin + 1)), "the second argument should be a constant integral value"); } else if (isFuncOnlyAttr(Kind)) { - Assert((ArgCount) == 0, "this attribute has no argument"); + Assert((Elem.End - Elem.Begin) == 0, "this attribute has no argument"); } else if (!isFuncOrArgAttr(Kind)) { - Assert((ArgCount) == 1, "this attribute should have one argument"); + Assert((Elem.End - Elem.Begin) == 1, + "this attribute should have one argument"); } } break; diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp index c734c9a..836af62 100644 --- a/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp +++ b/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp @@ -4220,16 +4220,11 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) { break; case Intrinsic::assume: { Value *IIOperand = II->getArgOperand(0); - SmallVector OpBundles; - II->getOperandBundlesAsDefs(OpBundles); - bool HasOpBundles = !OpBundles.empty(); // Remove an assume if it is followed by an identical assume. // TODO: Do we need this? Unless there are conflicting assumptions, the // computeKnownBits(IIOperand) below here eliminates redundant assumes. Instruction *Next = II->getNextNonDebugInstruction(); - if (HasOpBundles && - match(Next, m_Intrinsic(m_Specific(IIOperand))) && - !cast(Next)->hasOperandBundles()) + if (match(Next, m_Intrinsic(m_Specific(IIOperand)))) return eraseInstFromFunction(CI); // Canonicalize assume(a && b) -> assume(a); assume(b); @@ -4239,15 +4234,14 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) { Value *AssumeIntrinsic = II->getCalledOperand(); Value *A, *B; if (match(IIOperand, m_And(m_Value(A), m_Value(B)))) { - Builder.CreateCall(AssumeIntrinsicTy, AssumeIntrinsic, A, OpBundles, - II->getName()); + Builder.CreateCall(AssumeIntrinsicTy, AssumeIntrinsic, A, II->getName()); Builder.CreateCall(AssumeIntrinsicTy, AssumeIntrinsic, B, II->getName()); return eraseInstFromFunction(*II); } // assume(!(a || b)) -> assume(!a); assume(!b); if (match(IIOperand, m_Not(m_Or(m_Value(A), m_Value(B))))) { Builder.CreateCall(AssumeIntrinsicTy, AssumeIntrinsic, - Builder.CreateNot(A), OpBundles, II->getName()); + Builder.CreateNot(A), II->getName()); Builder.CreateCall(AssumeIntrinsicTy, AssumeIntrinsic, Builder.CreateNot(B), II->getName()); return eraseInstFromFunction(*II); @@ -4263,8 +4257,7 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) { isValidAssumeForContext(II, LHS, &DT)) { MDNode *MD = MDNode::get(II->getContext(), None); LHS->setMetadata(LLVMContext::MD_nonnull, MD); - if (!HasOpBundles) - return eraseInstFromFunction(*II); + return eraseInstFromFunction(*II); // TODO: apply nonnull return attributes to calls and invokes // TODO: apply range metadata for range check patterns? diff --git a/llvm/lib/Transforms/Scalar/AlignmentFromAssumptions.cpp b/llvm/lib/Transforms/Scalar/AlignmentFromAssumptions.cpp index bccf94f..5c00858 100644 --- a/llvm/lib/Transforms/Scalar/AlignmentFromAssumptions.cpp +++ b/llvm/lib/Transforms/Scalar/AlignmentFromAssumptions.cpp @@ -15,7 +15,6 @@ // //===----------------------------------------------------------------------===// -#include "llvm/IR/Instructions.h" #include "llvm/InitializePasses.h" #define AA_NAME "alignment-from-assumptions" #define DEBUG_TYPE AA_NAME @@ -204,33 +203,103 @@ static Align getNewAlignment(const SCEV *AASCEV, const SCEV *AlignSCEV, } bool AlignmentFromAssumptionsPass::extractAlignmentInfo(CallInst *I, - unsigned Idx, Value *&AAPtr, const SCEV *&AlignSCEV, const SCEV *&OffSCEV) { - Type *Int64Ty = Type::getInt64Ty(I->getContext()); - OperandBundleUse AlignOB = I->getOperandBundleAt(Idx); - if (AlignOB.getTagName() != "align") + // An alignment assume must be a statement about the least-significant + // bits of the pointer being zero, possibly with some offset. + ICmpInst *ICI = dyn_cast(I->getArgOperand(0)); + if (!ICI) return false; - assert(AlignOB.Inputs.size() >= 2); - AAPtr = AlignOB.Inputs[0].get(); - // TODO: Consider accumulating the offset to the base. - AAPtr = AAPtr->stripPointerCastsSameRepresentation(); - AlignSCEV = SE->getSCEV(AlignOB.Inputs[1].get()); - AlignSCEV = SE->getTruncateOrZeroExtend(AlignSCEV, Int64Ty); - if (AlignOB.Inputs.size() == 3) - OffSCEV = SE->getSCEV(AlignOB.Inputs[2].get()); - else + + // This must be an expression of the form: x & m == 0. + if (ICI->getPredicate() != ICmpInst::ICMP_EQ) + return false; + + // Swap things around so that the RHS is 0. + Value *CmpLHS = ICI->getOperand(0); + Value *CmpRHS = ICI->getOperand(1); + const SCEV *CmpLHSSCEV = SE->getSCEV(CmpLHS); + const SCEV *CmpRHSSCEV = SE->getSCEV(CmpRHS); + if (CmpLHSSCEV->isZero()) + std::swap(CmpLHS, CmpRHS); + else if (!CmpRHSSCEV->isZero()) + return false; + + BinaryOperator *CmpBO = dyn_cast(CmpLHS); + if (!CmpBO || CmpBO->getOpcode() != Instruction::And) + return false; + + // Swap things around so that the right operand of the and is a constant + // (the mask); we cannot deal with variable masks. + Value *AndLHS = CmpBO->getOperand(0); + Value *AndRHS = CmpBO->getOperand(1); + const SCEV *AndLHSSCEV = SE->getSCEV(AndLHS); + const SCEV *AndRHSSCEV = SE->getSCEV(AndRHS); + if (isa(AndLHSSCEV)) { + std::swap(AndLHS, AndRHS); + std::swap(AndLHSSCEV, AndRHSSCEV); + } + + const SCEVConstant *MaskSCEV = dyn_cast(AndRHSSCEV); + if (!MaskSCEV) + return false; + + // The mask must have some trailing ones (otherwise the condition is + // trivial and tells us nothing about the alignment of the left operand). + unsigned TrailingOnes = MaskSCEV->getAPInt().countTrailingOnes(); + if (!TrailingOnes) + return false; + + // Cap the alignment at the maximum with which LLVM can deal (and make sure + // we don't overflow the shift). + uint64_t Alignment; + TrailingOnes = std::min(TrailingOnes, + unsigned(sizeof(unsigned) * CHAR_BIT - 1)); + Alignment = std::min(1u << TrailingOnes, +Value::MaximumAlignment); + + Type *Int64Ty = Type::getInt64Ty(I->getParent()->getParent()->getContext()); + AlignSCEV = SE->getConstant(Int64Ty, Alignment); + + // The LHS might be a ptrtoint instruction, or it might be the pointer + // with an offset. + AAPtr = nullptr; + OffSCEV = nullptr; + if (PtrToIntInst *PToI = dyn_cast(AndLHS)) { + AAPtr = PToI->getPointerOperand(); OffSCEV = SE->getZero(Int64Ty); - OffSCEV = SE->getTruncateOrZeroExtend(OffSCEV, Int64Ty); + } else if (const SCEVAddExpr* AndLHSAddSCEV = + dyn_cast(AndLHSSCEV)) { + // Try to find the ptrtoint; subtract it and the rest is the offset. + for (SCEVAddExpr::op_iterator J = AndLHSAddSCEV->op_begin(), + JE = AndLHSAddSCEV->op_end(); J != JE; ++J) + if (const SCEVUnknown *OpUnk = dyn_cast(*J)) + if (PtrToIntInst *PToI = dyn_cast(OpUnk->getValue())) { + AAPtr = PToI->getPointerOperand(); + OffSCEV = SE->getMinusSCEV(AndLHSAddSCEV, *J); + break; + } + } + + if (!AAPtr) + return false; + + // Sign extend the offset to 64 bits (so that it is like all of the other + // expressions). + unsigned OffSCEVBits = OffSCEV->getType()->getPrimitiveSizeInBits(); + if (OffSCEVBits < 64) + OffSCEV = SE->getSignExtendExpr(OffSCEV, Int64Ty); + else if (OffSCEVBits > 64) + return false; + + AAPtr = AAPtr->stripPointerCasts(); return true; } -bool AlignmentFromAssumptionsPass::processAssumption(CallInst *ACall, - unsigned Idx) { +bool AlignmentFromAssumptionsPass::processAssumption(CallInst *ACall) { Value *AAPtr; const SCEV *AlignSCEV, *OffSCEV; - if (!extractAlignmentInfo(ACall, Idx, AAPtr, AlignSCEV, OffSCEV)) + if (!extractAlignmentInfo(ACall, AAPtr, AlignSCEV, OffSCEV)) return false; // Skip ConstantPointerNull and UndefValue. Assumptions on these shouldn't @@ -248,14 +317,13 @@ bool AlignmentFromAssumptionsPass::processAssumption(CallInst *ACall, continue; if (Instruction *K = dyn_cast(J)) + if (isValidAssumeForContext(ACall, K, DT)) WorkList.push_back(K); } while (!WorkList.empty()) { Instruction *J = WorkList.pop_back_val(); if (LoadInst *LI = dyn_cast(J)) { - if (!isValidAssumeForContext(ACall, J, DT)) - continue; Align NewAlignment = getNewAlignment(AASCEV, AlignSCEV, OffSCEV, LI->getPointerOperand(), SE); if (NewAlignment > LI->getAlign()) { @@ -263,8 +331,6 @@ bool AlignmentFromAssumptionsPass::processAssumption(CallInst *ACall, ++NumLoadAlignChanged; } } else if (StoreInst *SI = dyn_cast(J)) { - if (!isValidAssumeForContext(ACall, J, DT)) - continue; Align NewAlignment = getNewAlignment(AASCEV, AlignSCEV, OffSCEV, SI->getPointerOperand(), SE); if (NewAlignment > SI->getAlign()) { @@ -272,8 +338,6 @@ bool AlignmentFromAssumptionsPass::processAssumption(CallInst *ACall, ++NumStoreAlignChanged; } } else if (MemIntrinsic *MI = dyn_cast(J)) { - if (!isValidAssumeForContext(ACall, J, DT)) - continue; Align NewDestAlignment = getNewAlignment(AASCEV, AlignSCEV, OffSCEV, MI->getDest(), SE); @@ -305,7 +369,7 @@ bool AlignmentFromAssumptionsPass::processAssumption(CallInst *ACall, Visited.insert(J); for (User *UJ : J->users()) { Instruction *K = cast(UJ); - if (!Visited.count(K)) + if (!Visited.count(K) && isValidAssumeForContext(ACall, K, DT)) WorkList.push_back(K); } } @@ -332,11 +396,8 @@ bool AlignmentFromAssumptionsPass::runImpl(Function &F, AssumptionCache &AC, bool Changed = false; for (auto &AssumeVH : AC.assumptions()) - if (AssumeVH) { - CallInst *Call = cast(AssumeVH); - for (unsigned Idx = 0; Idx < Call->getNumOperandBundles(); Idx++) - Changed |= processAssumption(Call, Idx); - } + if (AssumeVH) + Changed |= processAssumption(cast(AssumeVH)); return Changed; } diff --git a/llvm/test/Transforms/AlignmentFromAssumptions/simple.ll b/llvm/test/Transforms/AlignmentFromAssumptions/simple.ll index 610fd44..14e764f 100644 --- a/llvm/test/Transforms/AlignmentFromAssumptions/simple.ll +++ b/llvm/test/Transforms/AlignmentFromAssumptions/simple.ll @@ -4,7 +4,10 @@ target datalayout = "e-i64:64-f80:128-n8:16:32:64-S128" define i32 @foo(i32* nocapture %a) nounwind uwtable readonly { entry: - tail call void @llvm.assume(i1 true) ["align"(i32* %a, i32 32)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) %0 = load i32, i32* %a, align 4 ret i32 %0 @@ -15,7 +18,11 @@ entry: define i32 @foo2(i32* nocapture %a) nounwind uwtable readonly { entry: - tail call void @llvm.assume(i1 true) ["align"(i32* %a, i32 32, i32 24)] + %ptrint = ptrtoint i32* %a to i64 + %offsetptr = add i64 %ptrint, 24 + %maskedptr = and i64 %offsetptr, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) %arrayidx = getelementptr inbounds i32, i32* %a, i64 2 %0 = load i32, i32* %arrayidx, align 4 ret i32 %0 @@ -27,7 +34,11 @@ entry: define i32 @foo2a(i32* nocapture %a) nounwind uwtable readonly { entry: - tail call void @llvm.assume(i1 true) ["align"(i32* %a, i32 32, i32 28)] + %ptrint = ptrtoint i32* %a to i64 + %offsetptr = add i64 %ptrint, 28 + %maskedptr = and i64 %offsetptr, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) %arrayidx = getelementptr inbounds i32, i32* %a, i64 -1 %0 = load i32, i32* %arrayidx, align 4 ret i32 %0 @@ -39,7 +50,10 @@ entry: define i32 @goo(i32* nocapture %a) nounwind uwtable readonly { entry: - tail call void @llvm.assume(i1 true) ["align"(i32* %a, i32 32, i32 0)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) %0 = load i32, i32* %a, align 4 ret i32 %0 @@ -50,7 +64,10 @@ entry: define i32 @hoo(i32* nocapture %a) nounwind uwtable readonly { entry: - tail call void @llvm.assume(i1 true) ["align"(i32* %a, i64 32, i32 0)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) br label %for.body for.body: ; preds = %entry, %for.body @@ -81,7 +98,10 @@ for.end: ; preds = %for.body ; load(a, i0+i1+i2+32) define void @hoo2(i32* nocapture %a, i64 %id, i64 %num) nounwind uwtable readonly { entry: - tail call void @llvm.assume(i1 true) ["align"(i32* %a, i8 32, i64 0)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) %id.mul = shl nsw i64 %id, 6 %num.mul = shl nsw i64 %num, 6 br label %for0.body @@ -127,7 +147,10 @@ return: define i32 @joo(i32* nocapture %a) nounwind uwtable readonly { entry: - tail call void @llvm.assume(i1 true) ["align"(i32* %a, i8 32, i8 0)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) br label %for.body for.body: ; preds = %entry, %for.body @@ -152,13 +175,16 @@ for.end: ; preds = %for.body define i32 @koo(i32* nocapture %a) nounwind uwtable readonly { entry: + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) br label %for.body for.body: ; preds = %entry, %for.body %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ] %r.06 = phi i32 [ 0, %entry ], [ %add, %for.body ] %arrayidx = getelementptr inbounds i32, i32* %a, i64 %indvars.iv - tail call void @llvm.assume(i1 true) ["align"(i32* %a, i8 32, i8 0)] %0 = load i32, i32* %arrayidx, align 4 %add = add nsw i32 %0, %r.06 %indvars.iv.next = add i64 %indvars.iv, 4 @@ -177,7 +203,10 @@ for.end: ; preds = %for.body define i32 @koo2(i32* nocapture %a) nounwind uwtable readonly { entry: - tail call void @llvm.assume(i1 true) ["align"(i32* %a, i128 32, i128 0)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) br label %for.body for.body: ; preds = %entry, %for.body @@ -202,7 +231,10 @@ for.end: ; preds = %for.body define i32 @moo(i32* nocapture %a) nounwind uwtable { entry: - tail call void @llvm.assume(i1 true) ["align"(i32* %a, i16 32)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) %0 = bitcast i32* %a to i8* tail call void @llvm.memset.p0i8.i64(i8* align 4 %0, i8 0, i64 64, i1 false) ret i32 undef @@ -214,9 +246,15 @@ entry: define i32 @moo2(i32* nocapture %a, i32* nocapture %b) nounwind uwtable { entry: - tail call void @llvm.assume(i1 true) ["align"(i32* %b, i32 128)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) + %ptrint1 = ptrtoint i32* %b to i64 + %maskedptr3 = and i64 %ptrint1, 127 + %maskcond4 = icmp eq i64 %maskedptr3, 0 + tail call void @llvm.assume(i1 %maskcond4) %0 = bitcast i32* %a to i8* - tail call void @llvm.assume(i1 true) ["align"(i8* %0, i16 32)] %1 = bitcast i32* %b to i8* tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %0, i8* align 4 %1, i64 64, i1 false) ret i32 undef @@ -226,19 +264,6 @@ entry: ; CHECK: ret i32 undef } -define i32 @moo3(i32* nocapture %a, i32* nocapture %b) nounwind uwtable { -entry: - %0 = bitcast i32* %a to i8* - tail call void @llvm.assume(i1 true) ["align"(i8* %0, i16 32), "align"(i32* %b, i32 128)] - %1 = bitcast i32* %b to i8* - tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %0, i8* align 4 %1, i64 64, i1 false) - ret i32 undef - -; CHECK-LABEL: @moo3 -; CHECK: @llvm.memcpy.p0i8.p0i8.i64(i8* align 32 %0, i8* align 128 %1, i64 64, i1 false) -; CHECK: ret i32 undef -} - declare void @llvm.assume(i1) nounwind declare void @llvm.memset.p0i8.i64(i8* nocapture, i8, i64, i1) nounwind diff --git a/llvm/test/Transforms/AlignmentFromAssumptions/simple32.ll b/llvm/test/Transforms/AlignmentFromAssumptions/simple32.ll index 453899c..3f0819e 100644 --- a/llvm/test/Transforms/AlignmentFromAssumptions/simple32.ll +++ b/llvm/test/Transforms/AlignmentFromAssumptions/simple32.ll @@ -7,12 +7,18 @@ define i32 @foo(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-LABEL: define {{[^@]+}}@foo ; CHECK-SAME: (i32* nocapture [[A:%.*]]) #0 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[A]], i64 32) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[A]] to i64 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 31 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: tail call void @llvm.assume(i1 [[MASKCOND]]) ; CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[A]], align 32 ; CHECK-NEXT: ret i32 [[TMP0]] ; entry: - call void @llvm.assume(i1 true) ["align"(i32* %a, i64 32)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) %0 = load i32, i32* %a, align 4 ret i32 %0 @@ -22,13 +28,21 @@ define i32 @foo2(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-LABEL: define {{[^@]+}}@foo2 ; CHECK-SAME: (i32* nocapture [[A:%.*]]) #0 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[A]], i64 32, i64 24) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[A]] to i64 +; CHECK-NEXT: [[OFFSETPTR:%.*]] = add i64 [[PTRINT]], 24 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[OFFSETPTR]], 31 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: tail call void @llvm.assume(i1 [[MASKCOND]]) ; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 2 ; CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[ARRAYIDX]], align 16 ; CHECK-NEXT: ret i32 [[TMP0]] ; entry: - call void @llvm.assume(i1 true) ["align"(i32* %a, i64 32, i64 24)] + %ptrint = ptrtoint i32* %a to i64 + %offsetptr = add i64 %ptrint, 24 + %maskedptr = and i64 %offsetptr, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) %arrayidx = getelementptr inbounds i32, i32* %a, i64 2 %0 = load i32, i32* %arrayidx, align 4 ret i32 %0 @@ -39,13 +53,21 @@ define i32 @foo2a(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-LABEL: define {{[^@]+}}@foo2a ; CHECK-SAME: (i32* nocapture [[A:%.*]]) #0 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[A]], i64 32, i64 28) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[A]] to i64 +; CHECK-NEXT: [[OFFSETPTR:%.*]] = add i64 [[PTRINT]], 28 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[OFFSETPTR]], 31 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: tail call void @llvm.assume(i1 [[MASKCOND]]) ; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 -1 ; CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[ARRAYIDX]], align 32 ; CHECK-NEXT: ret i32 [[TMP0]] ; entry: - call void @llvm.assume(i1 true) ["align"(i32* %a, i64 32, i64 28)] + %ptrint = ptrtoint i32* %a to i64 + %offsetptr = add i64 %ptrint, 28 + %maskedptr = and i64 %offsetptr, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) %arrayidx = getelementptr inbounds i32, i32* %a, i64 -1 %0 = load i32, i32* %arrayidx, align 4 ret i32 %0 @@ -56,12 +78,18 @@ define i32 @goo(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-LABEL: define {{[^@]+}}@goo ; CHECK-SAME: (i32* nocapture [[A:%.*]]) #0 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[A]], i64 32) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[A]] to i64 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 31 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: tail call void @llvm.assume(i1 [[MASKCOND]]) ; CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[A]], align 32 ; CHECK-NEXT: ret i32 [[TMP0]] ; entry: - call void @llvm.assume(i1 true) ["align"(i32* %a, i64 32)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) %0 = load i32, i32* %a, align 4 ret i32 %0 @@ -71,7 +99,10 @@ define i32 @hoo(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-LABEL: define {{[^@]+}}@hoo ; CHECK-SAME: (i32* nocapture [[A:%.*]]) #0 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[A]], i64 32) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[A]] to i64 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 31 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: tail call void @llvm.assume(i1 [[MASKCOND]]) ; CHECK-NEXT: br label [[FOR_BODY:%.*]] ; CHECK: for.body: ; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] @@ -88,7 +119,10 @@ define i32 @hoo(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-NEXT: ret i32 [[ADD_LCSSA]] ; entry: - call void @llvm.assume(i1 true) ["align"(i32* %a, i64 32)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) br label %for.body for.body: ; preds = %entry, %for.body @@ -112,7 +146,10 @@ define i32 @joo(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-LABEL: define {{[^@]+}}@joo ; CHECK-SAME: (i32* nocapture [[A:%.*]]) #0 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[A]], i64 32) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[A]] to i64 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 31 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: tail call void @llvm.assume(i1 [[MASKCOND]]) ; CHECK-NEXT: br label [[FOR_BODY:%.*]] ; CHECK: for.body: ; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 4, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] @@ -129,7 +166,10 @@ define i32 @joo(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-NEXT: ret i32 [[ADD_LCSSA]] ; entry: - call void @llvm.assume(i1 true) ["align"(i32* %a, i64 32)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) br label %for.body for.body: ; preds = %entry, %for.body @@ -153,7 +193,10 @@ define i32 @koo(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-LABEL: define {{[^@]+}}@koo ; CHECK-SAME: (i32* nocapture [[A:%.*]]) #0 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[A]], i64 32) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[A]] to i64 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 31 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: tail call void @llvm.assume(i1 [[MASKCOND]]) ; CHECK-NEXT: br label [[FOR_BODY:%.*]] ; CHECK: for.body: ; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] @@ -170,7 +213,10 @@ define i32 @koo(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-NEXT: ret i32 [[ADD_LCSSA]] ; entry: - call void @llvm.assume(i1 true) ["align"(i32* %a, i64 32)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) br label %for.body for.body: ; preds = %entry, %for.body @@ -194,7 +240,10 @@ define i32 @koo2(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-LABEL: define {{[^@]+}}@koo2 ; CHECK-SAME: (i32* nocapture [[A:%.*]]) #0 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[A]], i64 32) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[A]] to i64 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 31 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: tail call void @llvm.assume(i1 [[MASKCOND]]) ; CHECK-NEXT: br label [[FOR_BODY:%.*]] ; CHECK: for.body: ; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ -4, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] @@ -211,7 +260,10 @@ define i32 @koo2(i32* nocapture %a) nounwind uwtable readonly { ; CHECK-NEXT: ret i32 [[ADD_LCSSA]] ; entry: - call void @llvm.assume(i1 true) ["align"(i32* %a, i64 32)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) br label %for.body for.body: ; preds = %entry, %for.body @@ -235,13 +287,19 @@ define i32 @moo(i32* nocapture %a) nounwind uwtable { ; CHECK-LABEL: define {{[^@]+}}@moo ; CHECK-SAME: (i32* nocapture [[A:%.*]]) #1 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[A]], i64 32) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[A]] to i64 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 31 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: tail call void @llvm.assume(i1 [[MASKCOND]]) ; CHECK-NEXT: [[TMP0:%.*]] = bitcast i32* [[A]] to i8* ; CHECK-NEXT: tail call void @llvm.memset.p0i8.i64(i8* align 32 [[TMP0]], i8 0, i64 64, i1 false) ; CHECK-NEXT: ret i32 undef ; entry: - call void @llvm.assume(i1 true) ["align"(i32* %a, i64 32)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) %0 = bitcast i32* %a to i8* tail call void @llvm.memset.p0i8.i64(i8* align 4 %0, i8 0, i64 64, i1 false) ret i32 undef @@ -252,16 +310,28 @@ define i32 @moo2(i32* nocapture %a, i32* nocapture %b) nounwind uwtable { ; CHECK-LABEL: define {{[^@]+}}@moo2 ; CHECK-SAME: (i32* nocapture [[A:%.*]], i32* nocapture [[B:%.*]]) #1 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[A]], i64 32) ] -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(i32* [[B]], i64 128) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint i32* [[A]] to i64 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 31 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: tail call void @llvm.assume(i1 [[MASKCOND]]) +; CHECK-NEXT: [[PTRINT1:%.*]] = ptrtoint i32* [[B]] to i64 +; CHECK-NEXT: [[MASKEDPTR3:%.*]] = and i64 [[PTRINT1]], 127 +; CHECK-NEXT: [[MASKCOND4:%.*]] = icmp eq i64 [[MASKEDPTR3]], 0 +; CHECK-NEXT: tail call void @llvm.assume(i1 [[MASKCOND4]]) ; CHECK-NEXT: [[TMP0:%.*]] = bitcast i32* [[A]] to i8* ; CHECK-NEXT: [[TMP1:%.*]] = bitcast i32* [[B]] to i8* ; CHECK-NEXT: tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 32 [[TMP0]], i8* align 128 [[TMP1]], i64 64, i1 false) ; CHECK-NEXT: ret i32 undef ; entry: - call void @llvm.assume(i1 true) ["align"(i32* %a, i64 32)] - call void @llvm.assume(i1 true) ["align"(i32* %b, i64 128)] + %ptrint = ptrtoint i32* %a to i64 + %maskedptr = and i64 %ptrint, 31 + %maskcond = icmp eq i64 %maskedptr, 0 + tail call void @llvm.assume(i1 %maskcond) + %ptrint1 = ptrtoint i32* %b to i64 + %maskedptr3 = and i64 %ptrint1, 127 + %maskcond4 = icmp eq i64 %maskedptr3, 0 + tail call void @llvm.assume(i1 %maskcond4) %0 = bitcast i32* %a to i8* %1 = bitcast i32* %b to i8* tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %0, i8* align 4 %1, i64 64, i1 false) diff --git a/llvm/test/Transforms/Inline/align.ll b/llvm/test/Transforms/Inline/align.ll index f3a5184..ede6c3f 100644 --- a/llvm/test/Transforms/Inline/align.ll +++ b/llvm/test/Transforms/Inline/align.ll @@ -23,7 +23,10 @@ define void @foo(float* nocapture %a, float* nocapture readonly %c) #0 { ; CHECK-LABEL: define {{[^@]+}}@foo ; CHECK-SAME: (float* nocapture [[A:%.*]], float* nocapture readonly [[C:%.*]]) #0 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[A]], i64 128) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint float* [[A]] to i64 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 127 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) ; CHECK-NEXT: [[TMP0:%.*]] = load float, float* [[C]], align 4 ; CHECK-NEXT: [[ARRAYIDX_I:%.*]] = getelementptr inbounds float, float* [[A]], i64 5 ; CHECK-NEXT: store float [[TMP0]], float* [[ARRAYIDX_I]], align 4 @@ -84,8 +87,14 @@ define void @foo2(float* nocapture %a, float* nocapture %b, float* nocapture rea ; CHECK-LABEL: define {{[^@]+}}@foo2 ; CHECK-SAME: (float* nocapture [[A:%.*]], float* nocapture [[B:%.*]], float* nocapture readonly [[C:%.*]]) #0 ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[A]], i64 128) ] -; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(float* [[B]], i64 128) ] +; CHECK-NEXT: [[PTRINT:%.*]] = ptrtoint float* [[A]] to i64 +; CHECK-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 127 +; CHECK-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) +; CHECK-NEXT: [[PTRINT1:%.*]] = ptrtoint float* [[B]] to i64 +; CHECK-NEXT: [[MASKEDPTR2:%.*]] = and i64 [[PTRINT1]], 127 +; CHECK-NEXT: [[MASKCOND3:%.*]] = icmp eq i64 [[MASKEDPTR2]], 0 +; CHECK-NEXT: call void @llvm.assume(i1 [[MASKCOND3]]) ; CHECK-NEXT: [[TMP0:%.*]] = load float, float* [[C]], align 4 ; CHECK-NEXT: [[ARRAYIDX_I:%.*]] = getelementptr inbounds float, float* [[A]], i64 5 ; CHECK-NEXT: store float [[TMP0]], float* [[ARRAYIDX_I]], align 4 diff --git a/llvm/test/Transforms/InstCombine/assume.ll b/llvm/test/Transforms/InstCombine/assume.ll index b372f52..6f33e83 100644 --- a/llvm/test/Transforms/InstCombine/assume.ll +++ b/llvm/test/Transforms/InstCombine/assume.ll @@ -377,7 +377,6 @@ define i32 @assumption_conflicts_with_known_bits(i32 %a, i32 %b) { define void @debug_interference(i8 %x) { ; CHECK-LABEL: @debug_interference( ; CHECK-NEXT: [[CMP2:%.*]] = icmp ne i8 [[X:%.*]], 0 -; CHECK-NEXT: tail call void @llvm.assume(i1 false) ; CHECK-NEXT: tail call void @llvm.dbg.value(metadata i32 5, metadata !7, metadata !DIExpression()), !dbg !9 ; CHECK-NEXT: tail call void @llvm.assume(i1 false) ; CHECK-NEXT: tail call void @llvm.dbg.value(metadata i32 5, metadata !7, metadata !DIExpression()), !dbg !9 diff --git a/llvm/test/Transforms/PhaseOrdering/inlining-alignment-assumptions.ll b/llvm/test/Transforms/PhaseOrdering/inlining-alignment-assumptions.ll index 2605701..61287e3 100644 --- a/llvm/test/Transforms/PhaseOrdering/inlining-alignment-assumptions.ll +++ b/llvm/test/Transforms/PhaseOrdering/inlining-alignment-assumptions.ll @@ -41,7 +41,10 @@ define void @caller1(i1 %c, i64* align 1 %ptr) { ; ASSUMPTIONS-ON-NEXT: br i1 [[C:%.*]], label [[TRUE2_CRITEDGE:%.*]], label [[FALSE1:%.*]] ; ASSUMPTIONS-ON: false1: ; ASSUMPTIONS-ON-NEXT: store volatile i64 1, i64* [[PTR:%.*]], align 8 -; ASSUMPTIONS-ON-NEXT: call void @llvm.assume(i1 true) [ "align"(i64* [[PTR]], i64 8) ] +; ASSUMPTIONS-ON-NEXT: [[PTRINT:%.*]] = ptrtoint i64* [[PTR]] to i64 +; ASSUMPTIONS-ON-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 7 +; ASSUMPTIONS-ON-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; ASSUMPTIONS-ON-NEXT: tail call void @llvm.assume(i1 [[MASKCOND]]) ; ASSUMPTIONS-ON-NEXT: store volatile i64 0, i64* [[PTR]], align 8 ; ASSUMPTIONS-ON-NEXT: store volatile i64 -1, i64* [[PTR]], align 8 ; ASSUMPTIONS-ON-NEXT: store volatile i64 -1, i64* [[PTR]], align 8 @@ -51,7 +54,10 @@ define void @caller1(i1 %c, i64* align 1 %ptr) { ; ASSUMPTIONS-ON-NEXT: store volatile i64 3, i64* [[PTR]], align 8 ; ASSUMPTIONS-ON-NEXT: ret void ; ASSUMPTIONS-ON: true2.critedge: -; ASSUMPTIONS-ON-NEXT: call void @llvm.assume(i1 true) [ "align"(i64* [[PTR]], i64 8) ] +; ASSUMPTIONS-ON-NEXT: [[PTRINT_C:%.*]] = ptrtoint i64* [[PTR]] to i64 +; ASSUMPTIONS-ON-NEXT: [[MASKEDPTR_C:%.*]] = and i64 [[PTRINT_C]], 7 +; ASSUMPTIONS-ON-NEXT: [[MASKCOND_C:%.*]] = icmp eq i64 [[MASKEDPTR_C]], 0 +; ASSUMPTIONS-ON-NEXT: tail call void @llvm.assume(i1 [[MASKCOND_C]]) ; ASSUMPTIONS-ON-NEXT: store volatile i64 0, i64* [[PTR]], align 8 ; ASSUMPTIONS-ON-NEXT: store volatile i64 -1, i64* [[PTR]], align 8 ; ASSUMPTIONS-ON-NEXT: store volatile i64 -1, i64* [[PTR]], align 8 @@ -88,17 +94,26 @@ false2: ; This test checks that alignment assumptions do not prevent SROA. ; See PR45763. -define internal void @callee2(i64* noalias sret align 32 %arg) { +define internal void @callee2(i64* noalias sret align 8 %arg) { store i64 0, i64* %arg, align 8 ret void } define amdgpu_kernel void @caller2() { -; CHECK-LABEL: @caller2( -; CHECK-NEXT: ret void +; ASSUMPTIONS-OFF-LABEL: @caller2( +; ASSUMPTIONS-OFF-NEXT: ret void +; +; ASSUMPTIONS-ON-LABEL: @caller2( +; ASSUMPTIONS-ON-NEXT: [[ALLOCA:%.*]] = alloca i64, align 8, addrspace(5) +; ASSUMPTIONS-ON-NEXT: [[CAST:%.*]] = addrspacecast i64 addrspace(5)* [[ALLOCA]] to i64* +; ASSUMPTIONS-ON-NEXT: [[PTRINT:%.*]] = ptrtoint i64* [[CAST]] to i64 +; ASSUMPTIONS-ON-NEXT: [[MASKEDPTR:%.*]] = and i64 [[PTRINT]], 7 +; ASSUMPTIONS-ON-NEXT: [[MASKCOND:%.*]] = icmp eq i64 [[MASKEDPTR]], 0 +; ASSUMPTIONS-ON-NEXT: call void @llvm.assume(i1 [[MASKCOND]]) +; ASSUMPTIONS-ON-NEXT: ret void ; %alloca = alloca i64, align 8, addrspace(5) %cast = addrspacecast i64 addrspace(5)* %alloca to i64* - call void @callee2(i64* sret align 32 %cast) + call void @callee2(i64* sret align 8 %cast) ret void } diff --git a/llvm/test/Verifier/assume-bundles.ll b/llvm/test/Verifier/assume-bundles.ll index 6e260f2..3024217 100644 --- a/llvm/test/Verifier/assume-bundles.ll +++ b/llvm/test/Verifier/assume-bundles.ll @@ -1,4 +1,3 @@ -; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; RUN: not opt -verify < %s 2>&1 | FileCheck %s declare void @llvm.assume(i1) @@ -7,21 +6,14 @@ define void @func(i32* %P, i32 %P1, i32* %P2, i32* %P3) { ; CHECK: tags must be valid attribute names call void @llvm.assume(i1 true) ["adazdazd"()] ; CHECK: the second argument should be a constant integral value - call void @llvm.assume(i1 true) ["dereferenceable"(i32* %P, i32 %P1)] + call void @llvm.assume(i1 true) ["align"(i32* %P, i32 %P1)] ; CHECK: to many arguments - call void @llvm.assume(i1 true) ["dereferenceable"(i32* %P, i32 8, i32 8)] + call void @llvm.assume(i1 true) ["align"(i32* %P, i32 8, i32 8)] ; CHECK: this attribute should have 2 arguments - call void @llvm.assume(i1 true) ["dereferenceable"(i32* %P)] + call void @llvm.assume(i1 true) ["align"(i32* %P)] ; CHECK: this attribute has no argument - call void @llvm.assume(i1 true) ["dereferenceable"(i32* %P, i32 4), "cold"(i32* %P)] + call void @llvm.assume(i1 true) ["align"(i32* %P, i32 4), "cold"(i32* %P)] ; CHECK: this attribute should have one argument call void @llvm.assume(i1 true) ["noalias"()] - call void @llvm.assume(i1 true) ["align"(i32* %P, i32 %P1, i32 4)] -; CHECK: alignment assumptions should have 2 or 3 arguments - call void @llvm.assume(i1 true) ["align"(i32* %P, i32 %P1, i32 4, i32 4)] -; CHECK: second argument should be an integer - call void @llvm.assume(i1 true) ["align"(i32* %P, i32* %P2)] -; CHECK: third argument should be an integer if present - call void @llvm.assume(i1 true) ["align"(i32* %P, i32 %P1, i32* %P2)] ret void } diff --git a/llvm/unittests/Analysis/AssumeBundleQueriesTest.cpp b/llvm/unittests/Analysis/AssumeBundleQueriesTest.cpp index 946368e..d35a77f 100644 --- a/llvm/unittests/Analysis/AssumeBundleQueriesTest.cpp +++ b/llvm/unittests/Analysis/AssumeBundleQueriesTest.cpp @@ -546,41 +546,3 @@ TEST(AssumeQueryAPI, AssumptionCache) { ASSERT_EQ(AR[0].Index, 1u); ASSERT_EQ(AR[0].Assume, &*First); } - -TEST(AssumeQueryAPI, Alignment) { - LLVMContext C; - SMDiagnostic Err; - std::unique_ptr Mod = parseAssemblyString( - "declare void @llvm.assume(i1)\n" - "define void @test(i32* %P, i32* %P1, i32* %P2, i32 %I3, i1 %B) {\n" - "call void @llvm.assume(i1 true) [\"align\"(i32* %P, i32 8, i32 %I3)]\n" - "call void @llvm.assume(i1 true) [\"align\"(i32* %P1, i32 %I3, i32 " - "%I3)]\n" - "call void @llvm.assume(i1 true) [\"align\"(i32* %P2, i32 16, i32 8)]\n" - "ret void\n}\n", - Err, C); - if (!Mod) - Err.print("AssumeQueryAPI", errs()); - - Function *F = Mod->getFunction("test"); - BasicBlock::iterator Start = F->begin()->begin(); - IntrinsicInst *II; - RetainedKnowledge RK; - II = cast(&*Start); - RK = getKnowledgeFromBundle(*II, II->bundle_op_info_begin()[0]); - ASSERT_EQ(RK.AttrKind, Attribute::Alignment); - ASSERT_EQ(RK.WasOn, F->getArg(0)); - ASSERT_EQ(RK.ArgValue, 1u); - Start++; - II = cast(&*Start); - RK = getKnowledgeFromBundle(*II, II->bundle_op_info_begin()[0]); - ASSERT_EQ(RK.AttrKind, Attribute::Alignment); - ASSERT_EQ(RK.WasOn, F->getArg(1)); - ASSERT_EQ(RK.ArgValue, 1u); - Start++; - II = cast(&*Start); - RK = getKnowledgeFromBundle(*II, II->bundle_op_info_begin()[0]); - ASSERT_EQ(RK.AttrKind, Attribute::Alignment); - ASSERT_EQ(RK.WasOn, F->getArg(2)); - ASSERT_EQ(RK.ArgValue, 8u); -} -- 2.7.4