[AArch64][GlobalISel] Add legalization & selection support for G_INTRINSIC_LRINT

[aemerson]

Summary

Once the node equivalence is added selection becomes free.

[arsenm]

llvm/test/CodeGen/AArch64/GlobalISel/legalize-lrint.mir

3

Don't need -global-isel with -run-pass

17

Don't need the IR section

llvm/test/CodeGen/AArch64/GlobalISel/legalizer-info-validation.mir

132

Probably needed this to exist in the patch defining the opcodes

[aemerson]

Update test.

[aemerson]

@paquette ping

[paquette]

LGTM

llvm/include/llvm/Target/GlobalISel/SelectionDAGCompat.td

 def : GINodeEquiv<G_FSIN, fsin>;
 def : GINodeEquiv<G_FABS, fabs>;
 def : GINodeEquiv<G_FSQRT, fsqrt>;
 def : GINodeEquiv<G_FFLOOR, ffloor>;
 def : GINodeEquiv<G_FRINT, frint>;
 def : GINodeEquiv<G_FNEARBYINT, fnearbyint>;
 def : GINodeEquiv<G_INTRINSIC_TRUNC, ftrunc>;
 def : GINodeEquiv<G_INTRINSIC_ROUND, fround>;
+def : GINodeEquiv<G_INTRINSIC_LRINT, lrint>;
 def : GINodeEquiv<G_FCOPYSIGN, fcopysign>;
 def : GINodeEquiv<G_SMIN, smin>;
 def : GINodeEquiv<G_SMAX, smax>;
 def : GINodeEquiv<G_UMIN, umin>;
 def : GINodeEquiv<G_UMAX, umax>;
 def : GINodeEquiv<G_FMINNUM, fminnum>;
 def : GINodeEquiv<G_FMAXNUM, fmaxnum>;
 def : GINodeEquiv<G_FMINNUM_IEEE, fminnum_ieee>;

llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp

AArch64LegalizerInfo::AArch64LegalizerInfo(const AArch64Subtarget &ST)
 
   getActionDefinitionsBuilder({G_FADD, G_FSUB, G_FMUL, G_FDIV, G_FNEG})
     .legalFor({s32, s64, v2s64, v4s32, v2s32});
 
   getActionDefinitionsBuilder(G_FREM).libcallFor({s32, s64});
 
   getActionDefinitionsBuilder({G_FCEIL, G_FABS, G_FSQRT, G_FFLOOR, G_FRINT,
                                G_FMA, G_INTRINSIC_TRUNC, G_INTRINSIC_ROUND,
-                               G_FNEARBYINT})
+                               G_FNEARBYINT, G_INTRINSIC_LRINT})
       // If we don't have full FP16 support, then scalarize the elements of
       // vectors containing fp16 types.
       .fewerElementsIf(
           [=, &ST](const LegalityQuery &Query) {
             const auto &Ty = Query.Types[0];
             return Ty.isVector() && Ty.getElementType() == s16 &&
                    !ST.hasFullFP16();
           },

llvm/test/CodeGen/AArch64/GlobalISel/legalize-lrint.mir

+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -verify-machineinstrs -mtriple aarch64-unknown-unknown -run-pass=legalizer %s -o - | FileCheck %s
+---
+name:            testmsws
+alignment:       4
+tracksRegLiveness: true
+liveins:
+  - { reg: '$s0' }
+body:             |
+  bb.1:
+    liveins: $s0
+
+    ; CHECK-LABEL: name: testmsws
+    ; CHECK: liveins: $s0
+    ; CHECK: [[COPY:%[0-9]+]]:_(s32) = COPY $s0
+    ; CHECK: [[INTRINSIC_LRINT:%[0-9]+]]:_(s64) = G_INTRINSIC_LRINT [[COPY]](s32)
+    ; CHECK: [[TRUNC:%[0-9]+]]:_(s32) = G_TRUNC [[INTRINSIC_LRINT]](s64)
+    ; CHECK: $w0 = COPY [[TRUNC]](s32)
+    ; CHECK: RET_ReallyLR implicit $w0
+    %0:_(s32) = COPY $s0
+    %1:_(s64) = G_INTRINSIC_LRINT %0(s32)
+    %2:_(s32) = G_TRUNC %1(s64)
+    $w0 = COPY %2(s32)
+    RET_ReallyLR implicit $w0
+
+...
+---
+name:            testmsxs
+alignment:       4
+tracksRegLiveness: true
+liveins:
+  - { reg: '$s0' }
+body:             |
+  bb.1:
+    liveins: $s0
+
+    ; CHECK-LABEL: name: testmsxs
+    ; CHECK: liveins: $s0
+    ; CHECK: [[COPY:%[0-9]+]]:_(s32) = COPY $s0
+    ; CHECK: [[INTRINSIC_LRINT:%[0-9]+]]:_(s64) = G_INTRINSIC_LRINT [[COPY]](s32)
+    ; CHECK: $x0 = COPY [[INTRINSIC_LRINT]](s64)
+    ; CHECK: RET_ReallyLR implicit $x0
+    %0:_(s32) = COPY $s0
+    %1:_(s64) = G_INTRINSIC_LRINT %0(s32)
+    $x0 = COPY %1(s64)
+    RET_ReallyLR implicit $x0
+
+...
+---
+name:            testmswd
+alignment:       4
+tracksRegLiveness: true
+liveins:
+  - { reg: '$d0' }
+body:             |
+  bb.1:
+    liveins: $d0
+
+    ; CHECK-LABEL: name: testmswd
+    ; CHECK: liveins: $d0
+    ; CHECK: [[COPY:%[0-9]+]]:_(s64) = COPY $d0
+    ; CHECK: [[INTRINSIC_LRINT:%[0-9]+]]:_(s64) = G_INTRINSIC_LRINT [[COPY]](s64)
+    ; CHECK: [[TRUNC:%[0-9]+]]:_(s32) = G_TRUNC [[INTRINSIC_LRINT]](s64)
+    ; CHECK: $w0 = COPY [[TRUNC]](s32)
+    ; CHECK: RET_ReallyLR implicit $w0
+    %0:_(s64) = COPY $d0
+    %1:_(s64) = G_INTRINSIC_LRINT %0(s64)
+    %2:_(s32) = G_TRUNC %1(s64)
+    $w0 = COPY %2(s32)
+    RET_ReallyLR implicit $w0
+
+...
+---
+name:            testmsxd
+alignment:       4
+tracksRegLiveness: true
+liveins:
+  - { reg: '$d0' }
+body:             |
+  bb.1:
+    liveins: $d0
+
+    ; CHECK-LABEL: name: testmsxd
+    ; CHECK: liveins: $d0
+    ; CHECK: [[COPY:%[0-9]+]]:_(s64) = COPY $d0
+    ; CHECK: [[INTRINSIC_LRINT:%[0-9]+]]:_(s64) = G_INTRINSIC_LRINT [[COPY]](s64)
+    ; CHECK: $x0 = COPY [[INTRINSIC_LRINT]](s64)
+    ; CHECK: RET_ReallyLR implicit $x0
+    %0:_(s64) = COPY $d0
+    %1:_(s64) = G_INTRINSIC_LRINT %0(s64)
+    $x0 = COPY %1(s64)
+    RET_ReallyLR implicit $x0
+
+...

llvm/test/CodeGen/AArch64/GlobalISel/legalizer-info-validation.mir

 # DEBUG-NEXT: G_INTRINSIC_TRUNC (opcode {{[0-9]+}}): 1 type index, 0 imm indices
 # DEBUG-NEXT: .. opcode {{[0-9]+}} is aliased to {{[0-9]+}}
 # DEBUG-NEXT: .. type index coverage check SKIPPED: user-defined predicate detected
 # DEBUG-NEXT: .. imm index coverage check SKIPPED: user-defined predicate detected
 # DEBUG-NEXT: G_INTRINSIC_ROUND (opcode {{[0-9]+}}): 1 type index, 0 imm indices
 # DEBUG-NEXT: .. opcode {{[0-9]+}} is aliased to {{[0-9]+}}
 # DEBUG-NEXT: .. type index coverage check SKIPPED: user-defined predicate detected
 # DEBUG-NEXT: .. imm index coverage check SKIPPED: user-defined predicate detected
 # DEBUG-NEXT: G_INTRINSIC_LRINT (opcode {{[0-9]+}}): 2 type indices, 0 imm indices
-# DEBUG-NEXT: .. type index coverage check SKIPPED: no rules defined
-# DEBUG-NEXT:.. imm index coverage check SKIPPED: no rules defined
+# DEBUG-NEXT: .. opcode {{[0-9]+}} is aliased to {{[0-9]+}}
+# DEBUG-NEXT: .. type index coverage check SKIPPED: user-defined predicate detected
+# DEBUG-NEXT: .. imm index coverage check SKIPPED: user-defined predicate detected
 # DEBUG-NEXT: G_INTRINSIC_ROUNDEVEN (opcode {{[0-9]+}}): 1 type index, 0 imm indices
 # DEBUG-NEXT: .. type index coverage check SKIPPED: no rules defined
 # DEBUG-NEXT: .. imm index coverage check SKIPPED: no rules defined
 
 # DEBUG-NEXT: G_READCYCLECOUNTER (opcode {{[0-9]+}}): 1 type index, 0 imm indices
 # DEBUG-NEXT: .. type index coverage check SKIPPED: no rules defined
 # DEBUG-NEXT: .. imm index coverage check SKIPPED: no rules defined
 

llvm/test/CodeGen/AArch64/lrint-conv.ll

 ; RUN: llc < %s -mtriple=aarch64 -mattr=+neon | FileCheck %s
+; RUN: llc < %s -global-isel -global-isel-abort=2 -pass-remarks-missed=gisel* -mtriple=aarch64 | FileCheck %s --check-prefixes=FALLBACK,CHECK
 
 ; CHECK-LABEL: testmsws:
 ; CHECK:       frintx  [[REG:s[0-9]]], s0
 ; CHECK-NEXT:  fcvtzs  x0, [[REG]]
 ; CHECK:       ret
+; FALLBACK-NOT: remark{{.*}}testmsws
 define i32 @testmsws(float %x) {
 entry:
   %0 = tail call i64 @llvm.lrint.i64.f32(float %x)
   %conv = trunc i64 %0 to i32
   ret i32 %conv
 }
 
 ; CHECK-LABEL: testmsxs:
 ; CHECK:       frintx  [[REG:s[0-9]]], s0
 ; CHECK-NEXT:  fcvtzs  x0, [[REG]]
 ; CHECK-NEXT:  ret
+; FALLBACK-NOT: remark{{.*}}testmsxs
 define i64 @testmsxs(float %x) {
 entry:
   %0 = tail call i64 @llvm.lrint.i64.f32(float %x)
   ret i64 %0
 }
 
 ; CHECK-LABEL: testmswd:
 ; CHECK:       frintx  [[REG:d[0-9]]], d0
 ; CHECK-NEXT:  fcvtzs  x0, [[REG]]
 ; CHECK:       ret
+; FALLBACK-NOT: remark{{.*}}testmswd
 define i32 @testmswd(double %x) {
 entry:
   %0 = tail call i64 @llvm.lrint.i64.f64(double %x)
   %conv = trunc i64 %0 to i32
   ret i32 %conv
 }
 
 ; CHECK-LABEL: testmsxd:
 ; CHECK:       frintx  [[REG:d[0-9]]], d0
 ; CHECK-NEXT:  fcvtzs  x0, [[REG]]
 ; CHECK-NEXT:  ret
+; FALLBACK-NOT: remark{{.*}}testmsxd
 define i64 @testmsxd(double %x) {
 entry:
   %0 = tail call i64 @llvm.lrint.i64.f64(double %x)
   ret i64 %0
 }
 
 ; CHECK-LABEL: testmswl:
 ; CHECK:       bl      lrintl