[AMDGPU] Add llvm.amdgcn.s.setprio intrinsic

Reviewed By: rampitec, arsenm

Differential Revision: https://reviews.llvm.org/D120976

Author: kerbowa

clang/include/clang/Basic/BuiltinsAMDGPU.def

@@ -116,6 +116,7 @@ BUILTIN(__builtin_amdgcn_cubema, "ffff", "nc")
 BUILTIN(__builtin_amdgcn_s_sleep, "vIi", "n")
 BUILTIN(__builtin_amdgcn_s_incperflevel, "vIi", "n")
 BUILTIN(__builtin_amdgcn_s_decperflevel, "vIi", "n")
+BUILTIN(__builtin_amdgcn_s_setprio, "vIs", "n")
 BUILTIN(__builtin_amdgcn_uicmp, "WUiUiUiIi", "nc")
 BUILTIN(__builtin_amdgcn_uicmpl, "WUiWUiWUiIi", "nc")
 BUILTIN(__builtin_amdgcn_sicmp, "WUiiiIi", "nc")

clang/test/CodeGenOpenCL/builtins-amdgcn.cl

@@ -423,6 +423,15 @@ void test_s_decperflevel()
   __builtin_amdgcn_s_decperflevel(15);
 }
 
+// CHECK-LABEL: @test_s_setprio
+// CHECK: call void @llvm.amdgcn.s.setprio(i16 0)
+// CHECK: call void @llvm.amdgcn.s.setprio(i16 3)
+void test_s_setprio()
+{
+  __builtin_amdgcn_s_setprio(0);
+  __builtin_amdgcn_s_setprio(3);
+}
+
 // CHECK-LABEL: @test_cubeid(
 // CHECK: call float @llvm.amdgcn.cubeid(float %a, float %b, float %c)
 void test_cubeid(global float* out, float a, float b, float c) {

clang/test/SemaOpenCL/builtins-amdgcn-error.cl

@@ -54,6 +54,12 @@ void test_s_decperflevel(int x)
   __builtin_amdgcn_s_decperflevel(x); // expected-error {{argument to '__builtin_amdgcn_s_decperflevel' must be a constant integer}}
 }
 
+void test_s_setprio(int x)
+{
+  __builtin_amdgcn_s_setprio(x); // expected-error {{argument to '__builtin_amdgcn_s_setprio' must be a constant integer}}
+  __builtin_amdgcn_s_setprio(65536); // expected-warning {{implicit conversion from 'int' to 'short' changes value from 65536 to 0}}
+}
+
 void test_sicmp_i32(global ulong* out, int a, int b, uint c)
 {
   *out = __builtin_amdgcn_sicmp(a, b, c); // expected-error {{argument to '__builtin_amdgcn_sicmp' must be a constant integer}}

llvm/include/llvm/IR/IntrinsicsAMDGPU.td

@@ -1329,6 +1329,11 @@ def int_amdgcn_s_sethalt :
   Intrinsic<[], [llvm_i32_ty], [ImmArg<ArgIndex<0>>, IntrNoMem,
                                 IntrHasSideEffects, IntrWillReturn]>;
 
+def int_amdgcn_s_setprio :
+  GCCBuiltin<"__builtin_amdgcn_s_setprio">,
+  Intrinsic<[], [llvm_i16_ty], [ImmArg<ArgIndex<0>>, IntrNoMem,
+                                IntrHasSideEffects, IntrWillReturn]>;
+
 def int_amdgcn_s_getreg :
   GCCBuiltin<"__builtin_amdgcn_s_getreg">,
   Intrinsic<[llvm_i32_ty], [llvm_i32_ty],

llvm/lib/Target/AMDGPU/SOPInstructions.td

@@ -1278,7 +1278,10 @@ def S_SLEEP : SOPP_Pseudo <"s_sleep", (ins i32imm:$simm16),
   let hasSideEffects = 1;
 }
 
-def S_SETPRIO : SOPP_Pseudo <"s_setprio" , (ins i16imm:$simm16), "$simm16">;
+def S_SETPRIO : SOPP_Pseudo <"s_setprio", (ins i16imm:$simm16), "$simm16",
+  [(int_amdgcn_s_setprio timm:$simm16)]> {
+  let hasSideEffects = 1;
+}
 
 let Uses = [EXEC, M0] in {
 // FIXME: Should this be mayLoad+mayStore?

llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.setprio.ll

@@ -0,0 +1,49 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -march=amdgcn -mcpu=gfx90a -show-mc-encoding -verify-machineinstrs < %s | FileCheck -check-prefix=GFX9 %s
+; RUN: llc -march=amdgcn -show-mc-encoding -verify-machineinstrs < %s | FileCheck -check-prefix=SI %s
+; RUN: llc -global-isel -march=amdgcn -show-mc-encoding -verify-machineinstrs < %s | FileCheck -check-prefix=SI %s
+; RUN: llc -global-isel -march=amdgcn -mcpu=gfx90a -show-mc-encoding -verify-machineinstrs < %s | FileCheck -check-prefix=GFX9 %s
+
+declare void @llvm.amdgcn.s.setprio(i16) #0
+
+define void @test_llvm_amdgcn_s_setprio() #0 {
+; GFX9-LABEL: test_llvm_amdgcn_s_setprio:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; encoding: [0x00,0x00,0x8c,0xbf]
+; GFX9-NEXT:    s_setprio 0 ; encoding: [0x00,0x00,0x8f,0xbf]
+; GFX9-NEXT:    s_setprio 1 ; encoding: [0x01,0x00,0x8f,0xbf]
+; GFX9-NEXT:    s_setprio 2 ; encoding: [0x02,0x00,0x8f,0xbf]
+; GFX9-NEXT:    s_setprio 3 ; encoding: [0x03,0x00,0x8f,0xbf]
+; GFX9-NEXT:    s_setprio 10 ; encoding: [0x0a,0x00,0x8f,0xbf]
+; GFX9-NEXT:    s_setprio -1 ; encoding: [0xff,0xff,0x8f,0xbf]
+; GFX9-NEXT:    s_setprio 0 ; encoding: [0x00,0x00,0x8f,0xbf]
+; GFX9-NEXT:    s_setprio 1 ; encoding: [0x01,0x00,0x8f,0xbf]
+; GFX9-NEXT:    s_setprio -1 ; encoding: [0xff,0xff,0x8f,0xbf]
+; GFX9-NEXT:    s_setpc_b64 s[30:31] ; encoding: [0x1e,0x1d,0x80,0xbe]
+;
+; SI-LABEL: test_llvm_amdgcn_s_setprio:
+; SI:       ; %bb.0:
+; SI-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; encoding: [0x00,0x00,0x8c,0xbf]
+; SI-NEXT:    s_setprio 0 ; encoding: [0x00,0x00,0x8f,0xbf]
+; SI-NEXT:    s_setprio 1 ; encoding: [0x01,0x00,0x8f,0xbf]
+; SI-NEXT:    s_setprio 2 ; encoding: [0x02,0x00,0x8f,0xbf]
+; SI-NEXT:    s_setprio 3 ; encoding: [0x03,0x00,0x8f,0xbf]
+; SI-NEXT:    s_setprio 10 ; encoding: [0x0a,0x00,0x8f,0xbf]
+; SI-NEXT:    s_setprio -1 ; encoding: [0xff,0xff,0x8f,0xbf]
+; SI-NEXT:    s_setprio 0 ; encoding: [0x00,0x00,0x8f,0xbf]
+; SI-NEXT:    s_setprio 1 ; encoding: [0x01,0x00,0x8f,0xbf]
+; SI-NEXT:    s_setprio -1 ; encoding: [0xff,0xff,0x8f,0xbf]
+; SI-NEXT:    s_setpc_b64 s[30:31] ; encoding: [0x1e,0x20,0x80,0xbe]
+  call void @llvm.amdgcn.s.setprio(i16 0)
+  call void @llvm.amdgcn.s.setprio(i16 1)
+  call void @llvm.amdgcn.s.setprio(i16 2)
+  call void @llvm.amdgcn.s.setprio(i16 3)
+  call void @llvm.amdgcn.s.setprio(i16 10)
+  call void @llvm.amdgcn.s.setprio(i16 65535)
+  call void @llvm.amdgcn.s.setprio(i16 65536)
+  call void @llvm.amdgcn.s.setprio(i16 65537)
+  call void @llvm.amdgcn.s.setprio(i16 -1)
+  ret void
+}
+
+attributes #0 = { nounwind }