diff options
Diffstat (limited to 'llvm/test/Analysis')
10 files changed, 818 insertions, 6 deletions
diff --git a/llvm/test/Analysis/CostModel/AArch64/extract-last-active.ll b/llvm/test/Analysis/CostModel/AArch64/extract-last-active.ll new file mode 100644 index 0000000..9efcf91 --- /dev/null +++ b/llvm/test/Analysis/CostModel/AArch64/extract-last-active.ll @@ -0,0 +1,216 @@ +; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 6 +; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -S -mtriple=aarch64--linux-gnu | FileCheck %s --check-prefix=NEON +; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -S -mtriple=aarch64--linux-gnu -mattr=+sve | FileCheck %s --check-prefix=SVE +; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -S -mtriple=aarch64--linux-gnu -mattr=+sme -force-streaming | FileCheck %s --check-prefix=SME-STREAMING + +define void @extractions() { +; NEON-LABEL: 'extractions' +; NEON-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %v16i8 = call i8 @llvm.experimental.vector.extract.last.active.v16i8(<16 x i8> poison, <16 x i1> poison, i8 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %v8i16 = call i16 @llvm.experimental.vector.extract.last.active.v8i16(<8 x i16> poison, <8 x i1> poison, i16 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %v4i32 = call i32 @llvm.experimental.vector.extract.last.active.v4i32(<4 x i32> poison, <4 x i1> poison, i32 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %v2i64 = call i64 @llvm.experimental.vector.extract.last.active.v2i64(<2 x i64> poison, <2 x i1> poison, i64 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 38 for instruction: %v8f16 = call half @llvm.experimental.vector.extract.last.active.v8f16(<8 x half> poison, <8 x i1> poison, half poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 38 for instruction: %v8bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v8bf16(<8 x bfloat> poison, <8 x i1> poison, bfloat poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %v4f32 = call float @llvm.experimental.vector.extract.last.active.v4f32(<4 x float> poison, <4 x i1> poison, float poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v2f64 = call double @llvm.experimental.vector.extract.last.active.v2f64(<2 x double> poison, <2 x i1> poison, double poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv16i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> poison, i8 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv8i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> poison, i16 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv4i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> poison, i32 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv2i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> poison, i64 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv8f16 = call half @llvm.experimental.vector.extract.last.active.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> poison, half poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv8bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> poison, bfloat poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv4f32 = call float @llvm.experimental.vector.extract.last.active.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> poison, float poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv2f64 = call double @llvm.experimental.vector.extract.last.active.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> poison, double poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 160 for instruction: %v32i8 = call i8 @llvm.experimental.vector.extract.last.active.v32i8(<32 x i8> poison, <32 x i1> poison, i8 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %v16i16 = call i16 @llvm.experimental.vector.extract.last.active.v16i16(<16 x i16> poison, <16 x i1> poison, i16 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %v8i32 = call i32 @llvm.experimental.vector.extract.last.active.v8i32(<8 x i32> poison, <8 x i1> poison, i32 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %v4i64 = call i64 @llvm.experimental.vector.extract.last.active.v4i64(<4 x i64> poison, <4 x i1> poison, i64 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 76 for instruction: %v16f16 = call half @llvm.experimental.vector.extract.last.active.v16f16(<16 x half> poison, <16 x i1> poison, half poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 76 for instruction: %v16bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v16bf16(<16 x bfloat> poison, <16 x i1> poison, bfloat poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %v8f32 = call float @llvm.experimental.vector.extract.last.active.v8f32(<8 x float> poison, <8 x i1> poison, float poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v4f64 = call double @llvm.experimental.vector.extract.last.active.v4f64(<4 x double> poison, <4 x i1> poison, double poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv32i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv32i8(<vscale x 32 x i8> poison, <vscale x 32 x i1> poison, i8 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv16i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv16i16(<vscale x 16 x i16> poison, <vscale x 16 x i1> poison, i16 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv8i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv8i32(<vscale x 8 x i32> poison, <vscale x 8 x i1> poison, i32 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv4i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv4i64(<vscale x 4 x i64> poison, <vscale x 4 x i1> poison, i64 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv16f16 = call half @llvm.experimental.vector.extract.last.active.nxv16f16(<vscale x 16 x half> poison, <vscale x 16 x i1> poison, half poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv16bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv16bf16(<vscale x 16 x bfloat> poison, <vscale x 16 x i1> poison, bfloat poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv8f32 = call float @llvm.experimental.vector.extract.last.active.nxv8f32(<vscale x 8 x float> poison, <vscale x 8 x i1> poison, float poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv4f64 = call double @llvm.experimental.vector.extract.last.active.nxv4f64(<vscale x 4 x double> poison, <vscale x 4 x i1> poison, double poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %v8i8 = call i8 @llvm.experimental.vector.extract.last.active.v8i8(<8 x i8> poison, <8 x i1> poison, i8 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %v4i16 = call i16 @llvm.experimental.vector.extract.last.active.v4i16(<4 x i16> poison, <4 x i1> poison, i16 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %v2i32 = call i32 @llvm.experimental.vector.extract.last.active.v2i32(<2 x i32> poison, <2 x i1> poison, i32 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v1i64 = call i64 @llvm.experimental.vector.extract.last.active.v1i64(<1 x i64> poison, <1 x i1> poison, i64 poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %v4f16 = call half @llvm.experimental.vector.extract.last.active.v4f16(<4 x half> poison, <4 x i1> poison, half poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %v4bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v4bf16(<4 x bfloat> poison, <4 x i1> poison, bfloat poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v2f32 = call float @llvm.experimental.vector.extract.last.active.v2f32(<2 x float> poison, <2 x i1> poison, float poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v1f64 = call double @llvm.experimental.vector.extract.last.active.v1f64(<1 x double> poison, <1 x i1> poison, double poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv8i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv8i8(<vscale x 8 x i8> poison, <vscale x 8 x i1> poison, i8 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv4i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv4i16(<vscale x 4 x i16> poison, <vscale x 4 x i1> poison, i16 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv2i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv2i32(<vscale x 2 x i32> poison, <vscale x 2 x i1> poison, i32 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv1i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv1i64(<vscale x 1 x i64> poison, <vscale x 1 x i1> poison, i64 poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv4f16 = call half @llvm.experimental.vector.extract.last.active.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> poison, half poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv4bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv4bf16(<vscale x 4 x bfloat> poison, <vscale x 4 x i1> poison, bfloat poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv2f32 = call float @llvm.experimental.vector.extract.last.active.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> poison, float poison) +; NEON-NEXT: Cost Model: Invalid cost for instruction: %nxv1f64 = call double @llvm.experimental.vector.extract.last.active.nxv1f64(<vscale x 1 x double> poison, <vscale x 1 x i1> poison, double poison) +; NEON-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void +; +; SVE-LABEL: 'extractions' +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i8 = call i8 @llvm.experimental.vector.extract.last.active.v16i8(<16 x i8> poison, <16 x i1> poison, i8 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i16 = call i16 @llvm.experimental.vector.extract.last.active.v8i16(<8 x i16> poison, <8 x i1> poison, i16 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i32 = call i32 @llvm.experimental.vector.extract.last.active.v4i32(<4 x i32> poison, <4 x i1> poison, i32 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i64 = call i64 @llvm.experimental.vector.extract.last.active.v2i64(<2 x i64> poison, <2 x i1> poison, i64 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8f16 = call half @llvm.experimental.vector.extract.last.active.v8f16(<8 x half> poison, <8 x i1> poison, half poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v8bf16(<8 x bfloat> poison, <8 x i1> poison, bfloat poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4f32 = call float @llvm.experimental.vector.extract.last.active.v4f32(<4 x float> poison, <4 x i1> poison, float poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f64 = call double @llvm.experimental.vector.extract.last.active.v2f64(<2 x double> poison, <2 x i1> poison, double poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> poison, i8 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> poison, i16 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> poison, i32 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> poison, i64 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8f16 = call half @llvm.experimental.vector.extract.last.active.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> poison, half poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> poison, bfloat poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4f32 = call float @llvm.experimental.vector.extract.last.active.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> poison, float poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2f64 = call double @llvm.experimental.vector.extract.last.active.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> poison, double poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v32i8 = call i8 @llvm.experimental.vector.extract.last.active.v32i8(<32 x i8> poison, <32 x i1> poison, i8 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16i16 = call i16 @llvm.experimental.vector.extract.last.active.v16i16(<16 x i16> poison, <16 x i1> poison, i16 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8i32 = call i32 @llvm.experimental.vector.extract.last.active.v8i32(<8 x i32> poison, <8 x i1> poison, i32 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4i64 = call i64 @llvm.experimental.vector.extract.last.active.v4i64(<4 x i64> poison, <4 x i1> poison, i64 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16f16 = call half @llvm.experimental.vector.extract.last.active.v16f16(<16 x half> poison, <16 x i1> poison, half poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v16bf16(<16 x bfloat> poison, <16 x i1> poison, bfloat poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8f32 = call float @llvm.experimental.vector.extract.last.active.v8f32(<8 x float> poison, <8 x i1> poison, float poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f64 = call double @llvm.experimental.vector.extract.last.active.v4f64(<4 x double> poison, <4 x i1> poison, double poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv32i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv32i8(<vscale x 32 x i8> poison, <vscale x 32 x i1> poison, i8 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv16i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv16i16(<vscale x 16 x i16> poison, <vscale x 16 x i1> poison, i16 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv8i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv8i32(<vscale x 8 x i32> poison, <vscale x 8 x i1> poison, i32 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv4i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv4i64(<vscale x 4 x i64> poison, <vscale x 4 x i1> poison, i64 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv16f16 = call half @llvm.experimental.vector.extract.last.active.nxv16f16(<vscale x 16 x half> poison, <vscale x 16 x i1> poison, half poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv16bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv16bf16(<vscale x 16 x bfloat> poison, <vscale x 16 x i1> poison, bfloat poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv8f32 = call float @llvm.experimental.vector.extract.last.active.nxv8f32(<vscale x 8 x float> poison, <vscale x 8 x i1> poison, float poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv4f64 = call double @llvm.experimental.vector.extract.last.active.nxv4f64(<vscale x 4 x double> poison, <vscale x 4 x i1> poison, double poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i8 = call i8 @llvm.experimental.vector.extract.last.active.v8i8(<8 x i8> poison, <8 x i1> poison, i8 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i16 = call i16 @llvm.experimental.vector.extract.last.active.v4i16(<4 x i16> poison, <4 x i1> poison, i16 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i32 = call i32 @llvm.experimental.vector.extract.last.active.v2i32(<2 x i32> poison, <2 x i1> poison, i32 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v1i64 = call i64 @llvm.experimental.vector.extract.last.active.v1i64(<1 x i64> poison, <1 x i1> poison, i64 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4f16 = call half @llvm.experimental.vector.extract.last.active.v4f16(<4 x half> poison, <4 x i1> poison, half poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v4bf16(<4 x bfloat> poison, <4 x i1> poison, bfloat poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f32 = call float @llvm.experimental.vector.extract.last.active.v2f32(<2 x float> poison, <2 x i1> poison, float poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v1f64 = call double @llvm.experimental.vector.extract.last.active.v1f64(<1 x double> poison, <1 x i1> poison, double poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv8i8(<vscale x 8 x i8> poison, <vscale x 8 x i1> poison, i8 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv4i16(<vscale x 4 x i16> poison, <vscale x 4 x i1> poison, i16 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv2i32(<vscale x 2 x i32> poison, <vscale x 2 x i1> poison, i32 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv1i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv1i64(<vscale x 1 x i64> poison, <vscale x 1 x i1> poison, i64 poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4f16 = call half @llvm.experimental.vector.extract.last.active.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> poison, half poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv4bf16(<vscale x 4 x bfloat> poison, <vscale x 4 x i1> poison, bfloat poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2f32 = call float @llvm.experimental.vector.extract.last.active.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> poison, float poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv1f64 = call double @llvm.experimental.vector.extract.last.active.nxv1f64(<vscale x 1 x double> poison, <vscale x 1 x i1> poison, double poison) +; SVE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void +; +; SME-STREAMING-LABEL: 'extractions' +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i8 = call i8 @llvm.experimental.vector.extract.last.active.v16i8(<16 x i8> poison, <16 x i1> poison, i8 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i16 = call i16 @llvm.experimental.vector.extract.last.active.v8i16(<8 x i16> poison, <8 x i1> poison, i16 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i32 = call i32 @llvm.experimental.vector.extract.last.active.v4i32(<4 x i32> poison, <4 x i1> poison, i32 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i64 = call i64 @llvm.experimental.vector.extract.last.active.v2i64(<2 x i64> poison, <2 x i1> poison, i64 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8f16 = call half @llvm.experimental.vector.extract.last.active.v8f16(<8 x half> poison, <8 x i1> poison, half poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v8bf16(<8 x bfloat> poison, <8 x i1> poison, bfloat poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4f32 = call float @llvm.experimental.vector.extract.last.active.v4f32(<4 x float> poison, <4 x i1> poison, float poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f64 = call double @llvm.experimental.vector.extract.last.active.v2f64(<2 x double> poison, <2 x i1> poison, double poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> poison, i8 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> poison, i16 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> poison, i32 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> poison, i64 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8f16 = call half @llvm.experimental.vector.extract.last.active.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> poison, half poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> poison, bfloat poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4f32 = call float @llvm.experimental.vector.extract.last.active.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> poison, float poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2f64 = call double @llvm.experimental.vector.extract.last.active.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> poison, double poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v32i8 = call i8 @llvm.experimental.vector.extract.last.active.v32i8(<32 x i8> poison, <32 x i1> poison, i8 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16i16 = call i16 @llvm.experimental.vector.extract.last.active.v16i16(<16 x i16> poison, <16 x i1> poison, i16 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8i32 = call i32 @llvm.experimental.vector.extract.last.active.v8i32(<8 x i32> poison, <8 x i1> poison, i32 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4i64 = call i64 @llvm.experimental.vector.extract.last.active.v4i64(<4 x i64> poison, <4 x i1> poison, i64 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16f16 = call half @llvm.experimental.vector.extract.last.active.v16f16(<16 x half> poison, <16 x i1> poison, half poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v16bf16(<16 x bfloat> poison, <16 x i1> poison, bfloat poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8f32 = call float @llvm.experimental.vector.extract.last.active.v8f32(<8 x float> poison, <8 x i1> poison, float poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f64 = call double @llvm.experimental.vector.extract.last.active.v4f64(<4 x double> poison, <4 x i1> poison, double poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv32i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv32i8(<vscale x 32 x i8> poison, <vscale x 32 x i1> poison, i8 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv16i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv16i16(<vscale x 16 x i16> poison, <vscale x 16 x i1> poison, i16 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv8i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv8i32(<vscale x 8 x i32> poison, <vscale x 8 x i1> poison, i32 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv4i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv4i64(<vscale x 4 x i64> poison, <vscale x 4 x i1> poison, i64 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv16f16 = call half @llvm.experimental.vector.extract.last.active.nxv16f16(<vscale x 16 x half> poison, <vscale x 16 x i1> poison, half poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv16bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv16bf16(<vscale x 16 x bfloat> poison, <vscale x 16 x i1> poison, bfloat poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv8f32 = call float @llvm.experimental.vector.extract.last.active.nxv8f32(<vscale x 8 x float> poison, <vscale x 8 x i1> poison, float poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv4f64 = call double @llvm.experimental.vector.extract.last.active.nxv4f64(<vscale x 4 x double> poison, <vscale x 4 x i1> poison, double poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i8 = call i8 @llvm.experimental.vector.extract.last.active.v8i8(<8 x i8> poison, <8 x i1> poison, i8 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i16 = call i16 @llvm.experimental.vector.extract.last.active.v4i16(<4 x i16> poison, <4 x i1> poison, i16 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i32 = call i32 @llvm.experimental.vector.extract.last.active.v2i32(<2 x i32> poison, <2 x i1> poison, i32 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v1i64 = call i64 @llvm.experimental.vector.extract.last.active.v1i64(<1 x i64> poison, <1 x i1> poison, i64 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4f16 = call half @llvm.experimental.vector.extract.last.active.v4f16(<4 x half> poison, <4 x i1> poison, half poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v4bf16(<4 x bfloat> poison, <4 x i1> poison, bfloat poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f32 = call float @llvm.experimental.vector.extract.last.active.v2f32(<2 x float> poison, <2 x i1> poison, float poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v1f64 = call double @llvm.experimental.vector.extract.last.active.v1f64(<1 x double> poison, <1 x i1> poison, double poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv8i8(<vscale x 8 x i8> poison, <vscale x 8 x i1> poison, i8 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv4i16(<vscale x 4 x i16> poison, <vscale x 4 x i1> poison, i16 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv2i32(<vscale x 2 x i32> poison, <vscale x 2 x i1> poison, i32 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv1i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv1i64(<vscale x 1 x i64> poison, <vscale x 1 x i1> poison, i64 poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4f16 = call half @llvm.experimental.vector.extract.last.active.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> poison, half poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv4bf16(<vscale x 4 x bfloat> poison, <vscale x 4 x i1> poison, bfloat poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2f32 = call float @llvm.experimental.vector.extract.last.active.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> poison, float poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv1f64 = call double @llvm.experimental.vector.extract.last.active.nxv1f64(<vscale x 1 x double> poison, <vscale x 1 x i1> poison, double poison) +; SME-STREAMING-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void +; + + ;; Legal types + %v16i8 = call i8 @llvm.experimental.vector.extract.last.active.v16i8(<16 x i8> poison, <16 x i1> poison, i8 poison) + %v8i16 = call i16 @llvm.experimental.vector.extract.last.active.v8i16(<8 x i16> poison, <8 x i1> poison, i16 poison) + %v4i32 = call i32 @llvm.experimental.vector.extract.last.active.v4i32(<4 x i32> poison, <4 x i1> poison, i32 poison) + %v2i64 = call i64 @llvm.experimental.vector.extract.last.active.v2i64(<2 x i64> poison, <2 x i1> poison, i64 poison) + %v8f16 = call half @llvm.experimental.vector.extract.last.active.v8f16(<8 x half> poison, <8 x i1> poison, half poison) + %v8bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v8bf16(<8 x bfloat> poison, <8 x i1> poison, bfloat poison) + %v4f32 = call float @llvm.experimental.vector.extract.last.active.v4f32(<4 x float> poison, <4 x i1> poison, float poison) + %v2f64 = call double @llvm.experimental.vector.extract.last.active.v2f64(<2 x double> poison, <2 x i1> poison, double poison) + %nxv16i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> poison, i8 poison) + %nxv8i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> poison, i16 poison) + %nxv4i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> poison, i32 poison) + %nxv2i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> poison, i64 poison) + %nxv8f16 = call half @llvm.experimental.vector.extract.last.active.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> poison, half poison) + %nxv8bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> poison, bfloat poison) + %nxv4f32 = call float @llvm.experimental.vector.extract.last.active.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> poison, float poison) + %nxv2f64 = call double @llvm.experimental.vector.extract.last.active.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> poison, double poison) + + ;; Wider-than-legal + %v32i8 = call i8 @llvm.experimental.vector.extract.last.active.v32i8(<32 x i8> poison, <32 x i1> poison, i8 poison) + %v16i16 = call i16 @llvm.experimental.vector.extract.last.active.v16i16(<16 x i16> poison, <16 x i1> poison, i16 poison) + %v8i32 = call i32 @llvm.experimental.vector.extract.last.active.v8i32(<8 x i32> poison, <8 x i1> poison, i32 poison) + %v4i64 = call i64 @llvm.experimental.vector.extract.last.active.v4i64(<4 x i64> poison, <4 x i1> poison, i64 poison) + %v16f16 = call half @llvm.experimental.vector.extract.last.active.v16f16(<16 x half> poison, <16 x i1> poison, half poison) + %v16bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v16bf16(<16 x bfloat> poison, <16 x i1> poison, bfloat poison) + %v8f32 = call float @llvm.experimental.vector.extract.last.active.v8f32(<8 x float> poison, <8 x i1> poison, float poison) + %v4f64 = call double @llvm.experimental.vector.extract.last.active.v4f64(<4 x double> poison, <4 x i1> poison, double poison) + %nxv32i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv32i8(<vscale x 32 x i8> poison, <vscale x 32 x i1> poison, i8 poison) + %nxv16i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv16i16(<vscale x 16 x i16> poison, <vscale x 16 x i1> poison, i16 poison) + %nxv8i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv8i32(<vscale x 8 x i32> poison, <vscale x 8 x i1> poison, i32 poison) + %nxv4i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv4i64(<vscale x 4 x i64> poison, <vscale x 4 x i1> poison, i64 poison) + %nxv16f16 = call half @llvm.experimental.vector.extract.last.active.nxv16f16(<vscale x 16 x half> poison, <vscale x 16 x i1> poison, half poison) + %nxv16bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv16bf16(<vscale x 16 x bfloat> poison, <vscale x 16 x i1> poison, bfloat poison) + %nxv8f32 = call float @llvm.experimental.vector.extract.last.active.nxv8f32(<vscale x 8 x float> poison, <vscale x 8 x i1> poison, float poison) + %nxv4f64 = call double @llvm.experimental.vector.extract.last.active.nxv4f64(<vscale x 4 x double> poison, <vscale x 4 x i1> poison, double poison) + + ;; Narrower-than-legal + %v8i8 = call i8 @llvm.experimental.vector.extract.last.active.v8i8(<8 x i8> poison, <8 x i1> poison, i8 poison) + %v4i16 = call i16 @llvm.experimental.vector.extract.last.active.v4i16(<4 x i16> poison, <4 x i1> poison, i16 poison) + %v2i32 = call i32 @llvm.experimental.vector.extract.last.active.v2i32(<2 x i32> poison, <2 x i1> poison, i32 poison) + %v1i64 = call i64 @llvm.experimental.vector.extract.last.active.v1i64(<1 x i64> poison, <1 x i1> poison, i64 poison) + %v4f16 = call half @llvm.experimental.vector.extract.last.active.v4f16(<4 x half> poison, <4 x i1> poison, half poison) + %v4bf16 = call bfloat @llvm.experimental.vector.extract.last.active.v4bf16(<4 x bfloat> poison, <4 x i1> poison, bfloat poison) + %v2f32 = call float @llvm.experimental.vector.extract.last.active.v2f32(<2 x float> poison, <2 x i1> poison, float poison) + %v1f64 = call double @llvm.experimental.vector.extract.last.active.v1f64(<1 x double> poison, <1 x i1> poison, double poison) + %nxv8i8 = call i8 @llvm.experimental.vector.extract.last.active.nxv8i8(<vscale x 8 x i8> poison, <vscale x 8 x i1> poison, i8 poison) + %nxv4i16 = call i16 @llvm.experimental.vector.extract.last.active.nxv4i16(<vscale x 4 x i16> poison, <vscale x 4 x i1> poison, i16 poison) + %nxv2i32 = call i32 @llvm.experimental.vector.extract.last.active.nxv2i32(<vscale x 2 x i32> poison, <vscale x 2 x i1> poison, i32 poison) + %nxv1i64 = call i64 @llvm.experimental.vector.extract.last.active.nxv1i64(<vscale x 1 x i64> poison, <vscale x 1 x i1> poison, i64 poison) + %nxv4f16 = call half @llvm.experimental.vector.extract.last.active.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> poison, half poison) + %nxv4bf16 = call bfloat @llvm.experimental.vector.extract.last.active.nxv4bf16(<vscale x 4 x bfloat> poison, <vscale x 4 x i1> poison, bfloat poison) + %nxv2f32 = call float @llvm.experimental.vector.extract.last.active.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> poison, float poison) + %nxv1f64 = call double @llvm.experimental.vector.extract.last.active.nxv1f64(<vscale x 1 x double> poison, <vscale x 1 x i1> poison, double poison) + + ret void +} diff --git a/llvm/test/Analysis/DependenceAnalysis/GCD.ll b/llvm/test/Analysis/DependenceAnalysis/GCD.ll index 03343e7..cb14d18 100644 --- a/llvm/test/Analysis/DependenceAnalysis/GCD.ll +++ b/llvm/test/Analysis/DependenceAnalysis/GCD.ll @@ -254,7 +254,7 @@ define void @gcd4(ptr %A, ptr %B, i64 %M, i64 %N) nounwind uwtable ssp { ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: store i32 %conv, ptr %arrayidx, align 4 ; CHECK-NEXT: da analyze - output [* *]! ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: %0 = load i32, ptr %arrayidx16, align 4 -; CHECK-NEXT: da analyze - none! +; CHECK-NEXT: da analyze - flow [* *|<]! ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: store i32 %0, ptr %B.addr.11, align 4 ; CHECK-NEXT: da analyze - confused! ; CHECK-NEXT: Src: %0 = load i32, ptr %arrayidx16, align 4 --> Dst: %0 = load i32, ptr %arrayidx16, align 4 @@ -322,7 +322,7 @@ define void @gcd5(ptr %A, ptr %B, i64 %M, i64 %N) nounwind uwtable ssp { ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: store i32 %conv, ptr %arrayidx, align 4 ; CHECK-NEXT: da analyze - output [* *]! ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: %0 = load i32, ptr %arrayidx16, align 4 -; CHECK-NEXT: da analyze - flow [<> *]! +; CHECK-NEXT: da analyze - flow [* *|<]! ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: store i32 %0, ptr %B.addr.11, align 4 ; CHECK-NEXT: da analyze - confused! ; CHECK-NEXT: Src: %0 = load i32, ptr %arrayidx16, align 4 --> Dst: %0 = load i32, ptr %arrayidx16, align 4 @@ -390,7 +390,7 @@ define void @gcd6(i64 %n, ptr %A, ptr %B) nounwind uwtable ssp { ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx5, align 4 --> Dst: store i32 %conv, ptr %arrayidx5, align 4 ; CHECK-NEXT: da analyze - output [* *]! ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx5, align 4 --> Dst: %2 = load i32, ptr %arrayidx9, align 4 -; CHECK-NEXT: da analyze - none! +; CHECK-NEXT: da analyze - flow [* *|<]! ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx5, align 4 --> Dst: store i32 %2, ptr %B.addr.12, align 4 ; CHECK-NEXT: da analyze - confused! ; CHECK-NEXT: Src: %2 = load i32, ptr %arrayidx9, align 4 --> Dst: %2 = load i32, ptr %arrayidx9, align 4 diff --git a/llvm/test/Analysis/DependenceAnalysis/SymbolicSIV.ll b/llvm/test/Analysis/DependenceAnalysis/SymbolicSIV.ll index cdfaec7..73a415b 100644 --- a/llvm/test/Analysis/DependenceAnalysis/SymbolicSIV.ll +++ b/llvm/test/Analysis/DependenceAnalysis/SymbolicSIV.ll @@ -384,7 +384,7 @@ define void @symbolicsiv6(ptr %A, ptr %B, i64 %n, i64 %N, i64 %M) nounwind uwtab ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: store i32 %conv, ptr %arrayidx, align 4 ; CHECK-NEXT: da analyze - none! ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: %0 = load i32, ptr %arrayidx7, align 4 -; CHECK-NEXT: da analyze - none! +; CHECK-NEXT: da analyze - flow [*|<]! ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: store i32 %0, ptr %B.addr.02, align 4 ; CHECK-NEXT: da analyze - confused! ; CHECK-NEXT: Src: %0 = load i32, ptr %arrayidx7, align 4 --> Dst: %0 = load i32, ptr %arrayidx7, align 4 @@ -440,7 +440,7 @@ define void @symbolicsiv7(ptr %A, ptr %B, i64 %n, i64 %N, i64 %M) nounwind uwtab ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: store i32 %conv, ptr %arrayidx, align 4 ; CHECK-NEXT: da analyze - none! ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: %1 = load i32, ptr %arrayidx6, align 4 -; CHECK-NEXT: da analyze - flow [<>]! +; CHECK-NEXT: da analyze - flow [*|<]! ; CHECK-NEXT: Src: store i32 %conv, ptr %arrayidx, align 4 --> Dst: store i32 %1, ptr %B.addr.02, align 4 ; CHECK-NEXT: da analyze - confused! ; CHECK-NEXT: Src: %1 = load i32, ptr %arrayidx6, align 4 --> Dst: %1 = load i32, ptr %arrayidx6, align 4 diff --git a/llvm/test/Analysis/DependenceAnalysis/compute-absolute-value.ll b/llvm/test/Analysis/DependenceAnalysis/compute-absolute-value.ll index 64fad37..783150a 100644 --- a/llvm/test/Analysis/DependenceAnalysis/compute-absolute-value.ll +++ b/llvm/test/Analysis/DependenceAnalysis/compute-absolute-value.ll @@ -18,7 +18,7 @@ define void @unknown_sign(ptr %a, i64 %k) { ; CHECK-NEXT: Src: store i8 1, ptr %idx.0, align 1 --> Dst: store i8 1, ptr %idx.0, align 1 ; CHECK-NEXT: da analyze - none! ; CHECK-NEXT: Src: store i8 1, ptr %idx.0, align 1 --> Dst: store i8 2, ptr %idx.1, align 1 -; CHECK-NEXT: da analyze - output [<>]! +; CHECK-NEXT: da analyze - output [*|<]! ; CHECK-NEXT: Src: store i8 2, ptr %idx.1, align 1 --> Dst: store i8 2, ptr %idx.1, align 1 ; CHECK-NEXT: da analyze - none! ; diff --git a/llvm/test/Analysis/DependenceAnalysis/gcd-miv-overflow.ll b/llvm/test/Analysis/DependenceAnalysis/gcd-miv-overflow.ll new file mode 100644 index 0000000..9169ac3 --- /dev/null +++ b/llvm/test/Analysis/DependenceAnalysis/gcd-miv-overflow.ll @@ -0,0 +1,63 @@ +; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 6 +; RUN: opt < %s -disable-output "-passes=print<da>" 2>&1 \ +; RUN: | FileCheck %s --check-prefixes=CHECK,CHECK-ALL +; RUN: opt < %s -disable-output "-passes=print<da>" -da-enable-dependence-test=gcd-miv 2>&1 \ +; RUN: | FileCheck %s --check-prefixes=CHECK,CHECK-GCD-MIV + +; offset0 = 4; +; offset1 = 0; +; for (i = 0; i < 100; i++) { +; A[offset0] = 1; +; A[offset1] = 2; +; offset0 += 3*m; +; offset1 += 3; +; } +; +; Dependency exists between the two stores. E.g., consider `m` is +; 12297829382473034411, which is a modular multiplicative inverse of 3 under +; modulo 2^64. Then `offset0` is effectively `i + 4`, so accesses will be as +; follows: +; +; - A[offset0] : A[4], A[5], A[6], ... +; - A[offset1] : A[0], A[3], A[6], ... +; +define void @gcdmiv_coef_ovfl(ptr %A, i64 %m) { +; CHECK-ALL-LABEL: 'gcdmiv_coef_ovfl' +; CHECK-ALL-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 1, ptr %gep.0, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-ALL-NEXT: da analyze - output [*|<]! +; CHECK-ALL-NEXT: Src: store i8 2, ptr %gep.1, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; +; CHECK-GCD-MIV-LABEL: 'gcdmiv_coef_ovfl' +; CHECK-GCD-MIV-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 1, ptr %gep.0, align 1 +; CHECK-GCD-MIV-NEXT: da analyze - consistent output [*]! +; CHECK-GCD-MIV-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-GCD-MIV-NEXT: da analyze - consistent output [*|<]! +; CHECK-GCD-MIV-NEXT: Src: store i8 2, ptr %gep.1, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-GCD-MIV-NEXT: da analyze - consistent output [*]! +; +entry: + %step = mul i64 3, %m + br label %loop + +loop: + %i = phi i64 [ 0, %entry ], [ %i.inc, %loop ] + %offset.0 = phi i64 [ 4, %entry ] , [ %offset.0.next, %loop ] + %offset.1 = phi i64 [ 0, %entry ] , [ %offset.1.next, %loop ] + %gep.0 = getelementptr inbounds i8, ptr %A, i64 %offset.0 + %gep.1 = getelementptr inbounds i8, ptr %A, i64 %offset.1 + store i8 1, ptr %gep.0 + store i8 2, ptr %gep.1 + %i.inc = add nuw nsw i64 %i, 1 + %offset.0.next = add nsw i64 %offset.0, %step + %offset.1.next = add nsw i64 %offset.1, 3 + %ec = icmp eq i64 %i.inc, 100 + br i1 %ec, label %exit, label %loop + +exit: + ret void +} +;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line: +; CHECK: {{.*}} diff --git a/llvm/test/Analysis/DependenceAnalysis/strong-siv-overflow.ll b/llvm/test/Analysis/DependenceAnalysis/strong-siv-overflow.ll new file mode 100644 index 0000000..bf0fafc --- /dev/null +++ b/llvm/test/Analysis/DependenceAnalysis/strong-siv-overflow.ll @@ -0,0 +1,68 @@ +; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 6 +; RUN: opt < %s -disable-output "-passes=print<da>" 2>&1 \ +; RUN: | FileCheck %s --check-prefixes=CHECK,CHECK-ALL +; RUN: opt < %s -disable-output "-passes=print<da>" -da-enable-dependence-test=strong-siv 2>&1 \ +; RUN: | FileCheck %s --check-prefixes=CHECK,CHECK-STRONG-SIV + +; for (i = 0; i < (1LL << 62); i++) { +; if (0 <= 2*i - 2) +; A[2*i - 2] = 1; +; +; if (0 <= 2*i - 4) +; A[2*i - 4] = 2; +; } +; +; FIXME: DependenceAnalysis currently detects no dependency between the two +; stores, but it does exist. For example, each store will access A[0] when i +; is 1 and 2 respectively. +; The root cause is that the product of the BTC and the coefficient +; ((1LL << 62) - 1 and 2) overflows in a signed sense. +define void @strongsiv_const_ovfl(ptr %A) { +; CHECK-LABEL: 'strongsiv_const_ovfl' +; CHECK-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 1, ptr %gep.0, align 1 +; CHECK-NEXT: da analyze - none! +; CHECK-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-NEXT: da analyze - none! +; CHECK-NEXT: Src: store i8 2, ptr %gep.1, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-NEXT: da analyze - none! +; +entry: + br label %loop.header + +loop.header: + %i = phi i64 [ 0, %entry ], [ %i.inc, %loop.latch ] + %offset.0 = phi i64 [ -2, %entry ], [ %offset.0.next, %loop.latch ] + %offset.1 = phi i64 [ -4, %entry ], [ %offset.1.next, %loop.latch ] + %ec = icmp eq i64 %i, 4611686018427387904 + br i1 %ec, label %exit, label %loop.body + +loop.body: + %cond.0 = icmp sge i64 %offset.0, 0 + %cond.1 = icmp sge i64 %offset.1, 0 + br i1 %cond.0, label %if.then.0, label %loop.middle + +if.then.0: + %gep.0 = getelementptr inbounds i8, ptr %A, i64 %offset.0 + store i8 1, ptr %gep.0 + br label %loop.middle + +loop.middle: + br i1 %cond.1, label %if.then.1, label %loop.latch + +if.then.1: + %gep.1 = getelementptr inbounds i8, ptr %A, i64 %offset.1 + store i8 2, ptr %gep.1 + br label %loop.latch + +loop.latch: + %i.inc = add nuw nsw i64 %i, 1 + %offset.0.next = add nsw i64 %offset.0, 2 + %offset.1.next = add nsw i64 %offset.1, 2 + br label %loop.header + +exit: + ret void +} +;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line: +; CHECK-ALL: {{.*}} +; CHECK-STRONG-SIV: {{.*}} diff --git a/llvm/test/Analysis/DependenceAnalysis/symbolic-rdiv-overflow.ll b/llvm/test/Analysis/DependenceAnalysis/symbolic-rdiv-overflow.ll new file mode 100644 index 0000000..c5ff988 --- /dev/null +++ b/llvm/test/Analysis/DependenceAnalysis/symbolic-rdiv-overflow.ll @@ -0,0 +1,137 @@ +; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 6 +; RUN: opt < %s -disable-output "-passes=print<da>" 2>&1 \ +; RUN: | FileCheck %s --check-prefixes=CHECK,CHECK-ALL +; RUN: opt < %s -disable-output "-passes=print<da>" -da-enable-dependence-test=symbolic-rdiv 2>&1 \ +; RUN: | FileCheck %s --check-prefixes=CHECK,CHECK-SYMBOLIC-RDIV + +; for (i = 0; i < (1LL << 62); i++) { +; if (0 <= 2*i - 2) +; A[2*i - 2] = 1; +; A[i] = 2; +; } +; +; FIXME: DependenceAnalysis currently detects no dependency between the two +; stores, but it does exist. For example, each store will access A[0] when i +; is 1 and 0 respectively. +; The root cause is that the product of the BTC and the coefficient +; ((1LL << 62) - 1 and 2) overflows in a signed sense. +define void @symbolicrdiv_prod_ovfl(ptr %A) { +; CHECK-ALL-LABEL: 'symbolicrdiv_prod_ovfl' +; CHECK-ALL-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 1, ptr %gep.0, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 2, ptr %gep.1, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; +; CHECK-SYMBOLIC-RDIV-LABEL: 'symbolicrdiv_prod_ovfl' +; CHECK-SYMBOLIC-RDIV-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 1, ptr %gep.0, align 1 +; CHECK-SYMBOLIC-RDIV-NEXT: da analyze - none! +; CHECK-SYMBOLIC-RDIV-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-SYMBOLIC-RDIV-NEXT: da analyze - none! +; CHECK-SYMBOLIC-RDIV-NEXT: Src: store i8 2, ptr %gep.1, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-SYMBOLIC-RDIV-NEXT: da analyze - consistent output [*]! +; +entry: + br label %loop.header + +loop.header: + %i = phi i64 [ 0, %entry ], [ %i.inc, %loop.latch ] + %offset = phi i64 [ -2, %entry ], [ %offset.next, %loop.latch ] + %ec = icmp eq i64 %i, 4611686018427387904 + br i1 %ec, label %exit, label %loop.body + +loop.body: + %cond = icmp sge i64 %offset, 0 + br i1 %cond, label %if.then, label %loop.latch + +if.then: + %gep.0 = getelementptr inbounds i8, ptr %A, i64 %offset + store i8 1, ptr %gep.0 + br label %loop.latch + +loop.latch: + %gep.1 = getelementptr inbounds i8, ptr %A, i64 %i + store i8 2, ptr %gep.1 + %i.inc = add nuw nsw i64 %i, 1 + %offset.next = add nsw i64 %offset, 2 + br label %loop.header + +exit: + ret void +} + +; offset0 = -4611686018427387904; // -2^62 +; offset1 = 4611686018427387904; // 2^62 +; for (i = 0; i < (1LL << 62) - 100; i++) { +; if (0 <= offset0) +; A[offset0] = 1; +; if (0 <= offset1) +; A[offset1] = 2; +; offset0 += 2; +; offset1 -= 1; +; } +; +; FIXME: DependenceAnalysis currently detects no dependency between the two +; stores, but it does exist. For example, +; +; memory access | i == 2^61 | i == 2^61 + 2^59 | i == 2^61 + 2^60 +; -------------------------|-----------|------------------|------------------- +; A[2*i - 2^62] (offset0) | | A[2^60] | A[2^61] +; A[-i + 2^62] (offset1) | A[2^61] | | A[2^60] +; +; The root cause is that the calculation of the differenct between the two +; constants (-2^62 and 2^62) overflows in a signed sense. +define void @symbolicrdiv_delta_ovfl(ptr %A) { +; CHECK-ALL-LABEL: 'symbolicrdiv_delta_ovfl' +; CHECK-ALL-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 1, ptr %gep.0, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 2, ptr %gep.1, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; +; CHECK-SYMBOLIC-RDIV-LABEL: 'symbolicrdiv_delta_ovfl' +; CHECK-SYMBOLIC-RDIV-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 1, ptr %gep.0, align 1 +; CHECK-SYMBOLIC-RDIV-NEXT: da analyze - consistent output [*]! +; CHECK-SYMBOLIC-RDIV-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-SYMBOLIC-RDIV-NEXT: da analyze - none! +; CHECK-SYMBOLIC-RDIV-NEXT: Src: store i8 2, ptr %gep.1, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-SYMBOLIC-RDIV-NEXT: da analyze - consistent output [*]! +; +entry: + br label %loop.header + +loop.header: + %i = phi i64 [ 0, %entry ], [ %i.inc, %loop.latch ] + %offset.0 = phi i64 [ -4611686018427387904, %entry ], [ %offset.0.next, %loop.latch ] + %offset.1 = phi i64 [ 4611686018427387904, %entry ], [ %offset.1.next, %loop.latch ] + %cond.0 = icmp sge i64 %offset.0, 0 + %cond.1 = icmp sge i64 %offset.1, 0 + br i1 %cond.0, label %if.then.0, label %loop.middle + +if.then.0: + %gep.0 = getelementptr inbounds i8, ptr %A, i64 %offset.0 + store i8 1, ptr %gep.0 + br label %loop.middle + +loop.middle: + br i1 %cond.1, label %if.then.1, label %loop.latch + +if.then.1: + %gep.1 = getelementptr inbounds i8, ptr %A, i64 %offset.1 + store i8 2, ptr %gep.1 + br label %loop.latch + +loop.latch: + %i.inc = add nuw nsw i64 %i, 1 + %offset.0.next = add nsw i64 %offset.0, 2 + %offset.1.next = sub nsw i64 %offset.1, 1 + %ec = icmp eq i64 %i.inc, 4611686018427387804 ; 2^62 - 100 + br i1 %ec, label %exit, label %loop.header + +exit: + ret void +} +;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line: +; CHECK: {{.*}} diff --git a/llvm/test/Analysis/DependenceAnalysis/weak-crossing-siv-overflow.ll b/llvm/test/Analysis/DependenceAnalysis/weak-crossing-siv-overflow.ll new file mode 100644 index 0000000..ba57c7b --- /dev/null +++ b/llvm/test/Analysis/DependenceAnalysis/weak-crossing-siv-overflow.ll @@ -0,0 +1,125 @@ +; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 6 +; RUN: opt < %s -disable-output "-passes=print<da>" 2>&1 \ +; RUN: | FileCheck %s --check-prefixes=CHECK,CHECK-ALL +; RUN: opt < %s -disable-output "-passes=print<da>" -da-enable-dependence-test=weak-crossing-siv 2>&1 \ +; RUN: | FileCheck %s --check-prefixes=CHECK,CHECK-WEAK-CROSSING-SIV + +; max_i = INT64_MAX/3 // 3074457345618258602 +; for (long long i = 0; i <= max_i; i++) { +; A[-3*i + INT64_MAX] = 0; +; if (i) +; A[3*i - 2] = 1; +; } +; +; FIXME: DependenceAnalysis currently detects no dependency between +; `A[-3*i + INT64_MAX]` and `A[3*i - 2]`, but it does exist. For example, +; +; memory access | i == 1 | i == max_i +; ---------------------|------------------|------------------ +; A[-3*i + INT64_MAX] | A[INT64_MAX - 3] | A[1] +; A[3*i - 2] | A[1] | A[INT64_MAX - 3] +; +; The root cause is that the calculation of the differenct between the two +; constants (INT64_MAX and -2) triggers an overflow. + +define void @weakcorssing_delta_ovfl(ptr %A) { +; CHECK-ALL-LABEL: 'weakcorssing_delta_ovfl' +; CHECK-ALL-NEXT: Src: store i8 0, ptr %idx.0, align 1 --> Dst: store i8 0, ptr %idx.0, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 0, ptr %idx.0, align 1 --> Dst: store i8 1, ptr %idx.1, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 1, ptr %idx.1, align 1 --> Dst: store i8 1, ptr %idx.1, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; +; CHECK-WEAK-CROSSING-SIV-LABEL: 'weakcorssing_delta_ovfl' +; CHECK-WEAK-CROSSING-SIV-NEXT: Src: store i8 0, ptr %idx.0, align 1 --> Dst: store i8 0, ptr %idx.0, align 1 +; CHECK-WEAK-CROSSING-SIV-NEXT: da analyze - consistent output [*]! +; CHECK-WEAK-CROSSING-SIV-NEXT: Src: store i8 0, ptr %idx.0, align 1 --> Dst: store i8 1, ptr %idx.1, align 1 +; CHECK-WEAK-CROSSING-SIV-NEXT: da analyze - none! +; CHECK-WEAK-CROSSING-SIV-NEXT: Src: store i8 1, ptr %idx.1, align 1 --> Dst: store i8 1, ptr %idx.1, align 1 +; CHECK-WEAK-CROSSING-SIV-NEXT: da analyze - consistent output [*]! +; +entry: + br label %loop.header + +loop.header: + %i = phi i64 [ 0, %entry ], [ %i.inc, %loop.latch ] + %subscript.0 = phi i64 [ 9223372036854775807, %entry ], [ %subscript.0.next, %loop.latch ] + %subscript.1 = phi i64 [ -2, %entry ], [ %subscript.1.next, %loop.latch ] + %idx.0 = getelementptr inbounds i8, ptr %A, i64 %subscript.0 + store i8 0, ptr %idx.0 + %cond.store = icmp ne i64 %i, 0 + br i1 %cond.store, label %if.store, label %loop.latch + +if.store: + %idx.1 = getelementptr inbounds i8, ptr %A, i64 %subscript.1 + store i8 1, ptr %idx.1 + br label %loop.latch + +loop.latch: + %i.inc = add nuw nsw i64 %i, 1 + %subscript.0.next = add nsw i64 %subscript.0, -3 + %subscript.1.next = add nsw i64 %subscript.1, 3 + %ec = icmp sgt i64 %i.inc, 3074457345618258602 + br i1 %ec, label %exit, label %loop.header + +exit: + ret void +} + +; max_i = INT64_MAX/3 // 3074457345618258602 +; for (long long i = 0; i <= max_i; i++) { +; A[-3*i + INT64_MAX] = 0; +; A[3*i + 1] = 1; +; } +; +; FIXME: DependenceAnalysis currently detects no dependency between +; `A[-3*i + INT64_MAX]` and `A[3*i - 2]`, but it does exist. For example, +; +; memory access | i == 0 | i == 1 | i == max_i - 1 | i == max_i +; ---------------------|--------|------------------|----------------|------------------ +; A[-3*i + INT64_MAX] | | A[INT64_MAX - 3] | A[1] | +; A[3*i + 1] | A[1] | | | A[INT64_MAX - 3] +; +; The root cause is that the product of the BTC, the coefficient, and 2 +; triggers an overflow. +; +define void @weakcorssing_prod_ovfl(ptr %A) { +; CHECK-ALL-LABEL: 'weakcorssing_prod_ovfl' +; CHECK-ALL-NEXT: Src: store i8 0, ptr %idx.0, align 1 --> Dst: store i8 0, ptr %idx.0, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 0, ptr %idx.0, align 1 --> Dst: store i8 1, ptr %idx.1, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 1, ptr %idx.1, align 1 --> Dst: store i8 1, ptr %idx.1, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; +; CHECK-WEAK-CROSSING-SIV-LABEL: 'weakcorssing_prod_ovfl' +; CHECK-WEAK-CROSSING-SIV-NEXT: Src: store i8 0, ptr %idx.0, align 1 --> Dst: store i8 0, ptr %idx.0, align 1 +; CHECK-WEAK-CROSSING-SIV-NEXT: da analyze - consistent output [*]! +; CHECK-WEAK-CROSSING-SIV-NEXT: Src: store i8 0, ptr %idx.0, align 1 --> Dst: store i8 1, ptr %idx.1, align 1 +; CHECK-WEAK-CROSSING-SIV-NEXT: da analyze - none! +; CHECK-WEAK-CROSSING-SIV-NEXT: Src: store i8 1, ptr %idx.1, align 1 --> Dst: store i8 1, ptr %idx.1, align 1 +; CHECK-WEAK-CROSSING-SIV-NEXT: da analyze - consistent output [*]! +; +entry: + br label %loop + +loop: + %i = phi i64 [ 0, %entry ], [ %i.inc, %loop ] + %subscript.0 = phi i64 [ 9223372036854775807, %entry ], [ %subscript.0.next, %loop ] + %subscript.1 = phi i64 [ 1, %entry ], [ %subscript.1.next, %loop ] + %idx.0 = getelementptr inbounds i8, ptr %A, i64 %subscript.0 + %idx.1 = getelementptr inbounds i8, ptr %A, i64 %subscript.1 + store i8 0, ptr %idx.0 + store i8 1, ptr %idx.1 + %i.inc = add nuw nsw i64 %i, 1 + %subscript.0.next = add nsw i64 %subscript.0, -3 + %subscript.1.next = add nsw i64 %subscript.1, 3 + %ec = icmp sgt i64 %i.inc, 3074457345618258602 + br i1 %ec, label %exit, label %loop + +exit: + ret void +} +;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line: +; CHECK: {{.*}} diff --git a/llvm/test/Analysis/DependenceAnalysis/weak-zero-siv-overflow.ll b/llvm/test/Analysis/DependenceAnalysis/weak-zero-siv-overflow.ll new file mode 100644 index 0000000..6317c38 --- /dev/null +++ b/llvm/test/Analysis/DependenceAnalysis/weak-zero-siv-overflow.ll @@ -0,0 +1,122 @@ +; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 6 +; RUN: opt < %s -disable-output "-passes=print<da>" 2>&1 \ +; RUN: | FileCheck %s --check-prefixes=CHECK,CHECK-ALL +; RUN: opt < %s -disable-output "-passes=print<da>" -da-enable-dependence-test=weak-zero-siv 2>&1 \ +; RUN: | FileCheck %s --check-prefixes=CHECK,CHECK-WEAK-ZERO-SIV + +; for (i = 0; i < (1LL << 62); i++) { +; if (0 <= 2*i - 2) +; A[2*i - 2] = 1; +; A[2] = 2; +; } +; +; FIXME: DependenceAnalysis currently detects no dependency between the two +; stores, but it does exist. The root cause is that the product of the BTC and +; the coefficient ((1LL << 62) - 1 and 2) overflows in a signed sense. +; +define void @weakzero_dst_siv_prod_ovfl(ptr %A) { +; CHECK-ALL-LABEL: 'weakzero_dst_siv_prod_ovfl' +; CHECK-ALL-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 1, ptr %gep.0, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 2, ptr %gep.1, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-ALL-NEXT: da analyze - consistent output [S]! +; +; CHECK-WEAK-ZERO-SIV-LABEL: 'weakzero_dst_siv_prod_ovfl' +; CHECK-WEAK-ZERO-SIV-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 1, ptr %gep.0, align 1 +; CHECK-WEAK-ZERO-SIV-NEXT: da analyze - consistent output [*]! +; CHECK-WEAK-ZERO-SIV-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-WEAK-ZERO-SIV-NEXT: da analyze - none! +; CHECK-WEAK-ZERO-SIV-NEXT: Src: store i8 2, ptr %gep.1, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-WEAK-ZERO-SIV-NEXT: da analyze - consistent output [S]! +; +entry: + br label %loop.header + +loop.header: + %i = phi i64 [ 0, %entry ], [ %i.inc, %loop.latch ] + %offset = phi i64 [ -2, %entry ], [ %offset.next, %loop.latch ] + %ec = icmp eq i64 %i, 4611686018427387904 + br i1 %ec, label %exit, label %loop.body + +loop.body: + %cond = icmp sge i64 %offset, 0 + br i1 %cond, label %if.then, label %loop.latch + +if.then: + %gep.0 = getelementptr inbounds i8, ptr %A, i64 %offset + store i8 1, ptr %gep.0 + br label %loop.latch + +loop.latch: + %gep.1 = getelementptr inbounds i8, ptr %A, i64 2 + store i8 2, ptr %gep.1 + %i.inc = add nuw nsw i64 %i, 1 + %offset.next = add nsw i64 %offset, 2 + br label %loop.header + +exit: + ret void +} + +; for (i = 0; i < n; i++) { +; if (0 <= 2*i - 1) +; A[2*i - 1] = 1; +; A[INT64_MAX] = 2; +; } +; +; FIXME: DependenceAnalysis currently detects no dependency between the two +; stores, but it does exist. When `%n` is 2^62, the value of `%offset` will be +; the same as INT64_MAX at the last iteration. +; The root cause is that the calculation of the difference between the two +; constants (INT64_MAX and -1) overflows in a signed sense. +; +define void @weakzero_dst_siv_delta_ovfl(ptr %A, i64 %n) { +; CHECK-ALL-LABEL: 'weakzero_dst_siv_delta_ovfl' +; CHECK-ALL-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 1, ptr %gep.0, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-ALL-NEXT: da analyze - none! +; CHECK-ALL-NEXT: Src: store i8 2, ptr %gep.1, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-ALL-NEXT: da analyze - consistent output [S]! +; +; CHECK-WEAK-ZERO-SIV-LABEL: 'weakzero_dst_siv_delta_ovfl' +; CHECK-WEAK-ZERO-SIV-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 1, ptr %gep.0, align 1 +; CHECK-WEAK-ZERO-SIV-NEXT: da analyze - consistent output [*]! +; CHECK-WEAK-ZERO-SIV-NEXT: Src: store i8 1, ptr %gep.0, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-WEAK-ZERO-SIV-NEXT: da analyze - none! +; CHECK-WEAK-ZERO-SIV-NEXT: Src: store i8 2, ptr %gep.1, align 1 --> Dst: store i8 2, ptr %gep.1, align 1 +; CHECK-WEAK-ZERO-SIV-NEXT: da analyze - consistent output [S]! +; +entry: + %guard = icmp sgt i64 %n, 0 + br i1 %guard, label %loop.header, label %exit + +loop.header: + %i = phi i64 [ 0, %entry ], [ %i.inc, %loop.latch ] + %offset = phi i64 [ -2, %entry ], [ %offset.next, %loop.latch ] + %ec = icmp eq i64 %i, %n + br i1 %ec, label %exit, label %loop.body + +loop.body: + %cond = icmp sge i64 %offset, 0 + br i1 %cond, label %if.then, label %loop.latch + +if.then: + %gep.0 = getelementptr inbounds i8, ptr %A, i64 %offset + store i8 1, ptr %gep.0 + br label %loop.latch + +loop.latch: + %gep.1 = getelementptr inbounds i8, ptr %A, i64 9223372036854775807 + store i8 2, ptr %gep.1 + %i.inc = add nuw nsw i64 %i, 1 + %offset.next = add nsw i64 %offset, 2 + br label %loop.header + +exit: + ret void +} +;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line: +; CHECK: {{.*}} diff --git a/llvm/test/Analysis/HashRecognize/cyclic-redundancy-check.ll b/llvm/test/Analysis/HashRecognize/cyclic-redundancy-check.ll index 7dec2f8..78b4139 100644 --- a/llvm/test/Analysis/HashRecognize/cyclic-redundancy-check.ll +++ b/llvm/test/Analysis/HashRecognize/cyclic-redundancy-check.ll @@ -1448,4 +1448,85 @@ exit: ; preds = %loop ret i16 %crc.next } +define i16 @not.crc.data.next.outside.user(i16 %crc.init, i16 %data.init) { +; CHECK-LABEL: 'not.crc.data.next.outside.user' +; CHECK-NEXT: Did not find a hash algorithm +; CHECK-NEXT: Reason: Recurrences have stray uses +; +entry: + br label %loop + +loop: + %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop ] + %crc = phi i16 [ %crc.init, %entry ], [ %crc.next, %loop ] + %data = phi i16 [ %data.init, %entry ], [ %data.next, %loop ] + %xor.crc.data = xor i16 %data, %crc + %crc.shl = shl i16 %crc, 1 + %crc.xor = xor i16 %crc.shl, 3 + %check.sb = icmp slt i16 %xor.crc.data, 0 + %crc.next = select i1 %check.sb, i16 %crc.xor, i16 %crc.shl + %data.next = shl i16 %data, 1 + %iv.next = add nuw nsw i32 %iv, 1 + %exit.cond = icmp samesign ult i32 %iv, 7 + br i1 %exit.cond, label %loop, label %exit + +exit: + %ret = xor i16 %data.next, %crc.next + ret i16 %ret +} + +define i16 @not.crc.data.phi.outside.user(i16 %crc.init, i16 %data.init) { +; CHECK-LABEL: 'not.crc.data.phi.outside.user' +; CHECK-NEXT: Did not find a hash algorithm +; CHECK-NEXT: Reason: Recurrences have stray uses +; +entry: + br label %loop + +loop: + %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop ] + %crc = phi i16 [ %crc.init, %entry ], [ %crc.next, %loop ] + %data = phi i16 [ %data.init, %entry ], [ %data.next, %loop ] + %xor.crc.data = xor i16 %data, %crc + %crc.shl = shl i16 %crc, 1 + %crc.xor = xor i16 %crc.shl, 3 + %check.sb = icmp slt i16 %xor.crc.data, 0 + %crc.next = select i1 %check.sb, i16 %crc.xor, i16 %crc.shl + %data.next = shl i16 %data, 1 + %iv.next = add nuw nsw i32 %iv, 1 + %exit.cond = icmp samesign ult i32 %iv, 7 + br i1 %exit.cond, label %loop, label %exit + +exit: + %ret = xor i16 %data, %crc.next + ret i16 %ret +} + +define i16 @not.crc.crc.phi.outside.user(i16 %crc.init, i16 %data.init) { +; CHECK-LABEL: 'not.crc.crc.phi.outside.user' +; CHECK-NEXT: Did not find a hash algorithm +; CHECK-NEXT: Reason: Recurrences have stray uses +; +entry: + br label %loop + +loop: + %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop ] + %crc = phi i16 [ %crc.init, %entry ], [ %crc.next, %loop ] + %data = phi i16 [ %data.init, %entry ], [ %data.next, %loop ] + %xor.crc.data = xor i16 %data, %crc + %crc.shl = shl i16 %crc, 1 + %crc.xor = xor i16 %crc.shl, 3 + %check.sb = icmp slt i16 %xor.crc.data, 0 + %crc.next = select i1 %check.sb, i16 %crc.xor, i16 %crc.shl + %data.next = shl i16 %data, 1 + %iv.next = add nuw nsw i32 %iv, 1 + %exit.cond = icmp samesign ult i32 %iv, 7 + br i1 %exit.cond, label %loop, label %exit + +exit: + %ret = xor i16 %crc, %crc.next + ret i16 %ret +} + declare i16 @side.effect() |