AMDGPU/GlobalISel: Widen small uaddo/usubo

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352294 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp b/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
index d5c176a7e2169ac9153da2eefe127b1125dd1397..336ca883de5f3c3e4502f6f199b51bf9c5af0c03 100644 (file)
--- a/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
+++ b/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
@@ -103,7 +103,8 @@ AMDGPULegalizerInfo::AMDGPULegalizerInfo(const GCNSubtarget &ST,
 
   getActionDefinitionsBuilder({G_UADDO, G_SADDO, G_USUBO, G_SSUBO,
                                G_UADDE, G_SADDE, G_USUBE, G_SSUBE})
-    .legalFor({{S32, S1}});
+    .legalFor({{S32, S1}})
+    .clampScalar(0, S32, S32);
 
   getActionDefinitionsBuilder(G_BITCAST)
     .legalForCartesianProduct({S32, V2S16})

diff --git a/test/CodeGen/AMDGPU/GlobalISel/legalize-uaddo.mir b/test/CodeGen/AMDGPU/GlobalISel/legalize-uaddo.mir
new file mode 100644 (file)
index 0000000..a4159c3
--- /dev/null
+++ b/test/CodeGen/AMDGPU/GlobalISel/legalize-uaddo.mir
@@ -0,0 +1,97 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -O0 -run-pass=legalizer %s -o - | FileCheck %s
+
+---
+name: test_uaddo_s32
+body: |
+  bb.0:
+    liveins: $vgpr0, $vgpr1
+
+    ; CHECK-LABEL: name: test_uaddo_s32
+    ; CHECK: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+    ; CHECK: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+    ; CHECK: [[UADDO:%[0-9]+]]:_(s32), [[UADDO1:%[0-9]+]]:_(s1) = G_UADDO [[COPY]], [[COPY1]]
+    ; CHECK: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[UADDO1]](s1)
+    ; CHECK: $vgpr0 = COPY [[UADDO]](s32)
+    ; CHECK: $vgpr1 = COPY [[ZEXT]](s32)
+    %0:_(s32) = COPY $vgpr0
+    %1:_(s32) = COPY $vgpr1
+    %2:_(s32), %3:_(s1) = G_UADDO %0, %1
+    %4:_(s32) = G_ZEXT %3
+    $vgpr0 = COPY %2
+    $vgpr1 = COPY %4
+...
+
+---
+name: test_uaddo_s7
+body: |
+  bb.0:
+    liveins: $vgpr0, $vgpr1
+
+    ; CHECK-LABEL: name: test_uaddo_s7
+    ; CHECK: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+    ; CHECK: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+    ; CHECK: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 127
+    ; CHECK: [[COPY2:%[0-9]+]]:_(s32) = COPY [[COPY]](s32)
+    ; CHECK: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
+    ; CHECK: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 127
+    ; CHECK: [[COPY3:%[0-9]+]]:_(s32) = COPY [[COPY1]](s32)
+    ; CHECK: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C1]]
+    ; CHECK: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[AND]], [[AND1]]
+    ; CHECK: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 127
+    ; CHECK: [[AND2:%[0-9]+]]:_(s32) = G_AND [[ADD]], [[C2]]
+    ; CHECK: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(ne), [[ADD]](s32), [[AND2]]
+    ; CHECK: [[C3:%[0-9]+]]:_(s32) = G_CONSTANT i32 127
+    ; CHECK: [[COPY4:%[0-9]+]]:_(s32) = COPY [[ADD]](s32)
+    ; CHECK: [[AND3:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C3]]
+    ; CHECK: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[ICMP]](s1)
+    ; CHECK: $vgpr0 = COPY [[AND3]](s32)
+    ; CHECK: $vgpr1 = COPY [[ZEXT]](s32)
+    %0:_(s32) = COPY $vgpr0
+    %1:_(s32) = COPY $vgpr1
+    %2:_(s7) = G_TRUNC %0
+    %3:_(s7) = G_TRUNC %1
+
+    %4:_(s7), %5:_(s1) = G_UADDO %2, %3
+    %6:_(s32) = G_ZEXT %4
+    %7:_(s32) = G_ZEXT %5
+    $vgpr0 = COPY %6
+    $vgpr1 = COPY %7
+...
+
+---
+name: test_uaddo_s16
+body: |
+  bb.0:
+    liveins: $vgpr0, $vgpr1
+
+    ; CHECK-LABEL: name: test_uaddo_s16
+    ; CHECK: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+    ; CHECK: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+    ; CHECK: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
+    ; CHECK: [[COPY2:%[0-9]+]]:_(s32) = COPY [[COPY]](s32)
+    ; CHECK: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
+    ; CHECK: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
+    ; CHECK: [[COPY3:%[0-9]+]]:_(s32) = COPY [[COPY1]](s32)
+    ; CHECK: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C1]]
+    ; CHECK: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[AND]], [[AND1]]
+    ; CHECK: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
+    ; CHECK: [[AND2:%[0-9]+]]:_(s32) = G_AND [[ADD]], [[C2]]
+    ; CHECK: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(ne), [[ADD]](s32), [[AND2]]
+    ; CHECK: [[C3:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
+    ; CHECK: [[COPY4:%[0-9]+]]:_(s32) = COPY [[ADD]](s32)
+    ; CHECK: [[AND3:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C3]]
+    ; CHECK: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[ICMP]](s1)
+    ; CHECK: $vgpr0 = COPY [[AND3]](s32)
+    ; CHECK: $vgpr1 = COPY [[ZEXT]](s32)
+    %0:_(s32) = COPY $vgpr0
+    %1:_(s32) = COPY $vgpr1
+    %2:_(s16) = G_TRUNC %0
+    %3:_(s16) = G_TRUNC %1
+
+    %4:_(s16), %5:_(s1) = G_UADDO %2, %3
+    %6:_(s32) = G_ZEXT %4
+    %7:_(s32) = G_ZEXT %5
+    $vgpr0 = COPY %6
+    $vgpr1 = COPY %7
+...

diff --git a/test/CodeGen/AMDGPU/GlobalISel/legalize-usubo.mir b/test/CodeGen/AMDGPU/GlobalISel/legalize-usubo.mir
new file mode 100644 (file)
index 0000000..9764231
--- /dev/null
+++ b/test/CodeGen/AMDGPU/GlobalISel/legalize-usubo.mir
@@ -0,0 +1,97 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -O0 -run-pass=legalizer %s -o - | FileCheck %s
+
+---
+name: test_usubo_s32
+body: |
+  bb.0:
+    liveins: $vgpr0, $vgpr1
+
+    ; CHECK-LABEL: name: test_usubo_s32
+    ; CHECK: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+    ; CHECK: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+    ; CHECK: [[USUBO:%[0-9]+]]:_(s32), [[USUBO1:%[0-9]+]]:_(s1) = G_USUBO [[COPY]], [[COPY1]]
+    ; CHECK: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[USUBO1]](s1)
+    ; CHECK: $vgpr0 = COPY [[USUBO]](s32)
+    ; CHECK: $vgpr1 = COPY [[ZEXT]](s32)
+    %0:_(s32) = COPY $vgpr0
+    %1:_(s32) = COPY $vgpr1
+    %2:_(s32), %3:_(s1) = G_USUBO %0, %1
+    %4:_(s32) = G_ZEXT %3
+    $vgpr0 = COPY %2
+    $vgpr1 = COPY %4
+...
+
+---
+name: test_usubo_s7
+body: |
+  bb.0:
+    liveins: $vgpr0, $vgpr1
+
+    ; CHECK-LABEL: name: test_usubo_s7
+    ; CHECK: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+    ; CHECK: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+    ; CHECK: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 127
+    ; CHECK: [[COPY2:%[0-9]+]]:_(s32) = COPY [[COPY]](s32)
+    ; CHECK: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
+    ; CHECK: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 127
+    ; CHECK: [[COPY3:%[0-9]+]]:_(s32) = COPY [[COPY1]](s32)
+    ; CHECK: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C1]]
+    ; CHECK: [[SUB:%[0-9]+]]:_(s32) = G_SUB [[AND]], [[AND1]]
+    ; CHECK: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 127
+    ; CHECK: [[AND2:%[0-9]+]]:_(s32) = G_AND [[SUB]], [[C2]]
+    ; CHECK: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(ne), [[SUB]](s32), [[AND2]]
+    ; CHECK: [[C3:%[0-9]+]]:_(s32) = G_CONSTANT i32 127
+    ; CHECK: [[COPY4:%[0-9]+]]:_(s32) = COPY [[SUB]](s32)
+    ; CHECK: [[AND3:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C3]]
+    ; CHECK: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[ICMP]](s1)
+    ; CHECK: $vgpr0 = COPY [[AND3]](s32)
+    ; CHECK: $vgpr1 = COPY [[ZEXT]](s32)
+    %0:_(s32) = COPY $vgpr0
+    %1:_(s32) = COPY $vgpr1
+    %2:_(s7) = G_TRUNC %0
+    %3:_(s7) = G_TRUNC %1
+
+    %4:_(s7), %5:_(s1) = G_USUBO %2, %3
+    %6:_(s32) = G_ZEXT %4
+    %7:_(s32) = G_ZEXT %5
+    $vgpr0 = COPY %6
+    $vgpr1 = COPY %7
+...
+
+---
+name: test_usubo_s16
+body: |
+  bb.0:
+    liveins: $vgpr0, $vgpr1
+
+    ; CHECK-LABEL: name: test_usubo_s16
+    ; CHECK: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+    ; CHECK: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+    ; CHECK: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
+    ; CHECK: [[COPY2:%[0-9]+]]:_(s32) = COPY [[COPY]](s32)
+    ; CHECK: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
+    ; CHECK: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
+    ; CHECK: [[COPY3:%[0-9]+]]:_(s32) = COPY [[COPY1]](s32)
+    ; CHECK: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C1]]
+    ; CHECK: [[SUB:%[0-9]+]]:_(s32) = G_SUB [[AND]], [[AND1]]
+    ; CHECK: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
+    ; CHECK: [[AND2:%[0-9]+]]:_(s32) = G_AND [[SUB]], [[C2]]
+    ; CHECK: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(ne), [[SUB]](s32), [[AND2]]
+    ; CHECK: [[C3:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
+    ; CHECK: [[COPY4:%[0-9]+]]:_(s32) = COPY [[SUB]](s32)
+    ; CHECK: [[AND3:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C3]]
+    ; CHECK: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[ICMP]](s1)
+    ; CHECK: $vgpr0 = COPY [[AND3]](s32)
+    ; CHECK: $vgpr1 = COPY [[ZEXT]](s32)
+    %0:_(s32) = COPY $vgpr0
+    %1:_(s32) = COPY $vgpr1
+    %2:_(s16) = G_TRUNC %0
+    %3:_(s16) = G_TRUNC %1
+
+    %4:_(s16), %5:_(s1) = G_USUBO %2, %3
+    %6:_(s32) = G_ZEXT %4
+    %7:_(s32) = G_ZEXT %5
+    $vgpr0 = COPY %6
+    $vgpr1 = COPY %7
+...