AMDGPU/GlobalISel: Make G_TRUNC legal

This is required to be legal. I'm not sure how we were getting away
without defining any rules for it.

Author: arsenm

llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp

@@ -460,6 +460,9 @@ AMDGPULegalizerInfo::AMDGPULegalizerInfo(const GCNSubtarget &ST_,
   FMad.scalarize(0)
       .lower();
 
+  getActionDefinitionsBuilder(G_TRUNC)
+    .alwaysLegal();
+
   getActionDefinitionsBuilder({G_SEXT, G_ZEXT, G_ANYEXT})
     .legalFor({{S64, S32}, {S32, S16}, {S64, S16},
                {S32, S1}, {S64, S1}, {S16, S1}})

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-trunc.mir

@@ -0,0 +1,124 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -run-pass=legalizer -global-isel-abort=0 -o - %s | FileCheck %s
+
+---
+name: test_trunc_s64_to_s32
+body: |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+
+    ; CHECK-LABEL: name: test_trunc_s64_to_s32
+    ; CHECK: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
+    ; CHECK: [[TRUNC:%[0-9]+]]:_(s32) = G_TRUNC [[COPY]](s64)
+    ; CHECK: $vgpr0 = COPY [[TRUNC]](s32)
+    %0:_(s64) = COPY $vgpr0_vgpr1
+    %1:_(s32) = G_TRUNC %0
+    $vgpr0 = COPY %1
+...
+
+---
+name: test_trunc_s64_to_s16
+body: |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+
+    ; CHECK-LABEL: name: test_trunc_s64_to_s16
+    ; CHECK: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
+    ; CHECK: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY]](s64)
+    ; CHECK: S_ENDPGM 0, implicit [[TRUNC]](s16)
+    %0:_(s64) = COPY $vgpr0_vgpr1
+    %1:_(s16) = G_TRUNC %0
+    S_ENDPGM 0, implicit %1
+...
+
+---
+name: test_trunc_v2s32_to_v2s16
+body: |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+
+    ; CHECK-LABEL: name: test_trunc_v2s32_to_v2s16
+    ; CHECK: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
+    ; CHECK: [[TRUNC:%[0-9]+]]:_(<2 x s16>) = G_TRUNC [[COPY]](<2 x s32>)
+    ; CHECK: $vgpr0 = COPY [[TRUNC]](<2 x s16>)
+    %0:_(<2 x s32>) = COPY $vgpr0_vgpr1
+    %1:_(<2 x s16>) = G_TRUNC %0
+    $vgpr0 = COPY %1
+...
+
+---
+name: test_trunc_v4s32_to_v4s16
+body: |
+  bb.0:
+    liveins: $vgpr0_vgpr1_vgpr2_vgpr3
+
+    ; CHECK-LABEL: name: test_trunc_v4s32_to_v4s16
+    ; CHECK: [[COPY:%[0-9]+]]:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    ; CHECK: [[TRUNC:%[0-9]+]]:_(<4 x s16>) = G_TRUNC [[COPY]](<4 x s32>)
+    ; CHECK: $vgpr0_vgpr1 = COPY [[TRUNC]](<4 x s16>)
+    %0:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    %1:_(<4 x s16>) = G_TRUNC %0
+    $vgpr0_vgpr1 = COPY %1
+...
+
+---
+name: test_trunc_s64_to_s1
+body: |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+
+    ; CHECK-LABEL: name: test_trunc_s64_to_s1
+    ; CHECK: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
+    ; CHECK: [[TRUNC:%[0-9]+]]:_(s1) = G_TRUNC [[COPY]](s64)
+    ; CHECK: S_ENDPGM 0, implicit [[TRUNC]](s1)
+    %0:_(s64) = COPY $vgpr0_vgpr1
+    %1:_(s1) = G_TRUNC %0
+    S_ENDPGM 0, implicit %1
+...
+
+---
+name: test_trunc_s32_to_s1
+body: |
+  bb.0:
+    liveins: $vgpr0
+
+    ; CHECK-LABEL: name: test_trunc_s32_to_s1
+    ; CHECK: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+    ; CHECK: [[TRUNC:%[0-9]+]]:_(s1) = G_TRUNC [[COPY]](s32)
+    ; CHECK: S_ENDPGM 0, implicit [[TRUNC]](s1)
+    %0:_(s32) = COPY $vgpr0
+    %1:_(s1) = G_TRUNC %0
+    S_ENDPGM 0, implicit %1
+...
+
+---
+name: test_trunc_s16_to_s1
+body: |
+  bb.0:
+    liveins: $vgpr0
+
+    ; CHECK-LABEL: name: test_trunc_s16_to_s1
+    ; CHECK: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+    ; CHECK: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY]](s32)
+    ; CHECK: [[TRUNC1:%[0-9]+]]:_(s1) = G_TRUNC [[TRUNC]](s16)
+    ; CHECK: S_ENDPGM 0, implicit [[TRUNC1]](s1)
+    %0:_(s32) = COPY $vgpr0
+    %1:_(s16) = G_TRUNC %0
+    %2:_(s1) = G_TRUNC %1
+    S_ENDPGM 0, implicit %2
+...
+
+---
+name: test_trunc_v4s32_to_v4s1
+body: |
+  bb.0:
+    liveins: $vgpr0_vgpr1_vgpr2_vgpr3
+
+    ; CHECK-LABEL: name: test_trunc_v4s32_to_v4s1
+    ; CHECK: [[COPY:%[0-9]+]]:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    ; CHECK: [[TRUNC:%[0-9]+]]:_(<4 x s1>) = G_TRUNC [[COPY]](<4 x s32>)
+    ; CHECK: S_ENDPGM 0, implicit [[TRUNC]](<4 x s1>)
+    %0:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    %1:_(<4 x s1>) = G_TRUNC %0
+    S_ENDPGM 0, implicit %1
+...