[Clang][AArch64] Fix up immediate range of f64f64 mopa/mops intrinsics

[bryanpkc]

clang/include/clang/Basic/arm_sme.td

254

Shouldn't this say ImmCheck0_7 instead?

[sdesmalen]

Replace by ImmCheck0_7.

[sdesmalen]

clang/include/clang/Basic/arm_sme.td

254

I briefly changed it to ImmCheck0_0 to see if any other tests would fail and then committed the wrong version of the patch. Thanks for pointing out!

[bryanpkc]

Thanks for the catch!

clang/include/clang/Basic/arm_sme.td

def NAME # _ZA32_S: SInst<"sv" # n_suffix # "_za32[_{d}]", "viPPdd", "f",
                               [IsStreaming, IsSharedZA],
                               [ImmCheck<0, ImmCheck0_3>]>;
   }
 
   let TargetGuard = "sme-f64f64" in {
     def NAME # _ZA64_D: SInst<"sv" # n_suffix # "_za64[_{d}]", "viPPdd", "d",
                               MergeOp1, "aarch64_sme_" # n_suffix,
                               [IsStreaming, IsSharedZA],
-                              [ImmCheck<0, ImmCheck0_3>]>;
+                              [ImmCheck<0, ImmCheck0_7>]>;
   }
 }
 
 defm SVMOPA : ZAFPOuterProd<"mopa">;
 defm SVMOPS : ZAFPOuterProd<"mops">;

clang/test/CodeGen/aarch64-sme-intrinsics/acle_sme_mopa-za64.c

 #define SME_ACLE_FUNC(A1,A2,A3) A1##A2##A3
 #endif
 
 // CHECK-C-LABEL: @test_svmopa_za64_s16(
 // CHECK-CXX-LABEL: @_Z20test_svmopa_za64_s16u10__SVBool_tu10__SVBool_tu11__SVInt16_tu11__SVInt16_t(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
-// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smopa.wide.nxv8i16(i32 1, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smopa.wide.nxv8i16(i32 7, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
 // CHECK-NEXT:    ret void
 //
 void test_svmopa_za64_s16(svbool_t pn, svbool_t pm, svint16_t zn, svint16_t zm) {
-  SME_ACLE_FUNC(svmopa_za64, _s16, _m)(1, pn, pm, zn, zm);
+  SME_ACLE_FUNC(svmopa_za64, _s16, _m)(7, pn, pm, zn, zm);
 }
 
 // CHECK-C-LABEL: @test_svmopa_za64_u16(
 // CHECK-CXX-LABEL: @_Z20test_svmopa_za64_u16u10__SVBool_tu10__SVBool_tu12__SVUint16_tu12__SVUint16_t(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
 // CHECK-NEXT:    tail call void @llvm.aarch64.sme.umopa.wide.nxv8i16(i32 0, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
 // CHECK-NEXT:    ret void
 //
 void test_svmopa_za64_u16(svbool_t pn, svbool_t pm, svuint16_t zn, svuint16_t zm) {
   SME_ACLE_FUNC(svmopa_za64, _u16, _m)(0, pn, pm, zn, zm);
 }
 
 // CHECK-C-LABEL: @test_svmopa_za64_f64(
 // CHECK-CXX-LABEL: @_Z20test_svmopa_za64_f64u10__SVBool_tu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_t(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PN:%.*]])
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PM:%.*]])
-// CHECK-NEXT:    tail call void @llvm.aarch64.sme.mopa.nxv2f64(i32 1, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i1> [[TMP1]], <vscale x 2 x double> [[ZN:%.*]], <vscale x 2 x double> [[ZM:%.*]])
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.mopa.nxv2f64(i32 7, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i1> [[TMP1]], <vscale x 2 x double> [[ZN:%.*]], <vscale x 2 x double> [[ZM:%.*]])
 // CHECK-NEXT:    ret void
 //
 void test_svmopa_za64_f64(svbool_t pn, svbool_t pm, svfloat64_t zn, svfloat64_t zm) {
-  SME_ACLE_FUNC(svmopa_za64, _f64, _m)(1, pn, pm, zn, zm);
+  SME_ACLE_FUNC(svmopa_za64, _f64, _m)(7, pn, pm, zn, zm);
 }
 
 // CHECK-C-LABEL: @test_svsumopa_za64_s16(
 // CHECK-CXX-LABEL: @_Z22test_svsumopa_za64_s16u10__SVBool_tu10__SVBool_tu11__SVInt16_tu12__SVUint16_t(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
 // CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumopa.wide.nxv8i16(i32 0, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
 // CHECK-NEXT:    ret void
 //
 void test_svsumopa_za64_s16(svbool_t pn, svbool_t pm, svint16_t zn, svuint16_t zm) {
  SME_ACLE_FUNC(svsumopa_za64, _s16, _m)(0, pn, pm, zn, zm);
 }
 
 // CHECK-C-LABEL: @test_svusmopa_za64_u16(
 // CHECK-CXX-LABEL: @_Z22test_svusmopa_za64_u16u10__SVBool_tu10__SVBool_tu12__SVUint16_tu11__SVInt16_t(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
-// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmopa.wide.nxv8i16(i32 2, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmopa.wide.nxv8i16(i32 7, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
 // CHECK-NEXT:    ret void
 //
 void test_svusmopa_za64_u16(svbool_t pn, svbool_t pm, svuint16_t zn, svint16_t zm) {
-  SME_ACLE_FUNC(svusmopa_za64, _u16, _m)(2, pn, pm, zn, zm);
+  SME_ACLE_FUNC(svusmopa_za64, _u16, _m)(7, pn, pm, zn, zm);
 }

clang/test/CodeGen/aarch64-sme-intrinsics/acle_sme_mops-za64.c

 #define SME_ACLE_FUNC(A1,A2,A3) A1##A2##A3
 #endif
 
 // CHECK-C-LABEL: @test_svmops_za64_s16(
 // CHECK-CXX-LABEL: @_Z20test_svmops_za64_s16u10__SVBool_tu10__SVBool_tu11__SVInt16_tu11__SVInt16_t(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
-// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smops.wide.nxv8i16(i32 1, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smops.wide.nxv8i16(i32 7, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
 // CHECK-NEXT:    ret void
 //
 void test_svmops_za64_s16(svbool_t pn, svbool_t pm, svint16_t zn, svint16_t zm) {
-  SME_ACLE_FUNC(svmops_za64, _s16, _m)(1, pn, pm, zn, zm);
+  SME_ACLE_FUNC(svmops_za64, _s16, _m)(7, pn, pm, zn, zm);
 }
 
 // CHECK-C-LABEL: @test_svmops_za64_u16(
 // CHECK-CXX-LABEL: @_Z20test_svmops_za64_u16u10__SVBool_tu10__SVBool_tu12__SVUint16_tu12__SVUint16_t(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
 // CHECK-NEXT:    tail call void @llvm.aarch64.sme.umops.wide.nxv8i16(i32 0, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
 // CHECK-NEXT:    ret void
 //
 void test_svmops_za64_u16(svbool_t pn, svbool_t pm, svuint16_t zn, svuint16_t zm) {
   SME_ACLE_FUNC(svmops_za64, _u16, _m)(0, pn, pm, zn, zm);
 }
 
 // CHECK-C-LABEL: @test_svmops_za64_f64(
 // CHECK-CXX-LABEL: @_Z20test_svmops_za64_f64u10__SVBool_tu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_t(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PN:%.*]])
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PM:%.*]])
-// CHECK-NEXT:    tail call void @llvm.aarch64.sme.mops.nxv2f64(i32 1, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i1> [[TMP1]], <vscale x 2 x double> [[ZN:%.*]], <vscale x 2 x double> [[ZM:%.*]])
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.mops.nxv2f64(i32 7, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i1> [[TMP1]], <vscale x 2 x double> [[ZN:%.*]], <vscale x 2 x double> [[ZM:%.*]])
 // CHECK-NEXT:    ret void
 //
 void test_svmops_za64_f64(svbool_t pn, svbool_t pm, svfloat64_t zn, svfloat64_t zm) {
-  SME_ACLE_FUNC(svmops_za64, _f64, _m)(1, pn, pm, zn, zm);
+  SME_ACLE_FUNC(svmops_za64, _f64, _m)(7, pn, pm, zn, zm);
 }
 
 // CHECK-C-LABEL: @test_svsumops_za64_s16(
 // CHECK-CXX-LABEL: @_Z22test_svsumops_za64_s16u10__SVBool_tu10__SVBool_tu11__SVInt16_tu12__SVUint16_t(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
 // CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumops.wide.nxv8i16(i32 0, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
 // CHECK-NEXT:    ret void
 //
 void test_svsumops_za64_s16(svbool_t pn, svbool_t pm, svint16_t zn, svuint16_t zm) {
  SME_ACLE_FUNC(svsumops_za64, _s16, _m)(0, pn, pm, zn, zm);
 }
 
 // CHECK-C-LABEL: @test_svusmops_za64_u16(
 // CHECK-CXX-LABEL: @_Z22test_svusmops_za64_u16u10__SVBool_tu10__SVBool_tu12__SVUint16_tu11__SVInt16_t(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
-// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmops.wide.nxv8i16(i32 2, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmops.wide.nxv8i16(i32 7, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
 // CHECK-NEXT:    ret void
 //
 void test_svusmops_za64_u16(svbool_t pn, svbool_t pm, svuint16_t zn, svint16_t zm) {
-  SME_ACLE_FUNC(svusmops_za64, _u16, _m)(2, pn, pm, zn, zm);
+  SME_ACLE_FUNC(svusmops_za64, _u16, _m)(7, pn, pm, zn, zm);
 }

clang/test/Sema/aarch64-sme-intrinsics/acle_sme_imm.cpp

void test_range_0_7(svbool_t pg, void *ptr) {
   // expected-error@+1 {{argument value 8 is outside the valid range [0, 7]}}
   SVE_ACLE_FUNC(svsumopa_za64, _s16, _m,)(8, pg, pg, svundef_s16(), svundef_u16());
   // expected-error@+1 {{argument value 18446744073709551615 is outside the valid range [0, 7]}}
   SVE_ACLE_FUNC(svsumops_za64, _s16, _m,)(-1, pg, pg, svundef_s16(), svundef_u16());
   // expected-error@+1 {{argument value 8 is outside the valid range [0, 7]}}
   SVE_ACLE_FUNC(svusmopa_za64, _u16, _m,)(8, pg, pg, svundef_u16(), svundef_s16());
   // expected-error@+1 {{argument value 18446744073709551615 is outside the valid range [0, 7]}}
   SVE_ACLE_FUNC(svusmops_za64, _u16, _m,)(-1, pg, pg, svundef_u16(), svundef_s16());
+
+  // expected-error@+1 {{argument value 8 is outside the valid range [0, 7]}}
+  SVE_ACLE_FUNC(svmopa_za64, _f64, _m,)(8, pg, pg, svundef_f64(), svundef_f64());
+  // expected-error@+1 {{argument value 18446744073709551615 is outside the valid range [0, 7]}}
+  SVE_ACLE_FUNC(svmops_za64, _f64, _m,)(-1, pg, pg, svundef_f64(), svundef_f64());
 }
 
 void test_range_0_15(svbool_t pg, void *ptr) {
   // expected-error@+1 {{argument value 18446744073709551615 is outside the valid range [0, 15]}}
   SVE_ACLE_FUNC(svld1_hor_za128,,,)(-1, -1, 0, pg, ptr);
   // expected-error@+1 {{argument value 16 is outside the valid range [0, 15]}}
   SVE_ACLE_FUNC(svst1_ver_za128,,,)(16, -1, 0, pg, ptr);
   // expected-error@+1 {{argument value 18446744073709551615 is outside the valid range [0, 15]}}