[Clang] Rename and enable boolean get, set, create and undef for sme2 #77338

SamTebbs33 · 2024-01-08T16:35:15Z

This patch renames the get, set, create and undef functions that deal with tuples of booleans to match the ACLE at
https://github.com/ARM-software/acle/pull/257/files . It also enables them for SME2.

This patch renames the get, set, create and undef functions that deal with tuples of booleans to match the ACLE at https://github.com/ARM-software/acle/pull/257/files . It also enables them for SME2.

llvmbot · 2024-01-08T16:35:42Z

@llvm/pr-subscribers-clang

Author: Sam Tebbs (SamTebbs33)

Changes

This patch renames the get, set, create and undef functions that deal with tuples of booleans to match the ACLE at
https://github.com/ARM-software/acle/pull/257/files . It also enables them for SME2.

Patch is 30.48 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/77338.diff

9 Files Affected:

(modified) clang/include/clang/Basic/arm_sve.td (+13-8)
(modified) clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_create2_bool.c (+7-13)
(modified) clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_create4_bool.c (+7-13)
(modified) clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_get2_bool.c (+11-17)
(modified) clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_get4_bool.c (+15-21)
(modified) clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_set2_bool.c (+11-17)
(modified) clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_set4_bool.c (+15-22)
(added) clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_undef_bool.c (+35)
(modified) clang/test/Sema/aarch64-sve2p1-intrinsics/acle_sve2p1_imm.cpp (+14-14)

diff --git a/clang/include/clang/Basic/arm_sve.td b/clang/include/clang/Basic/arm_sve.td
index 7f80fb0386cc77..5dec2eaa9f2536 100644
--- a/clang/include/clang/Basic/arm_sve.td
+++ b/clang/include/clang/Basic/arm_sve.td
@@ -1296,9 +1296,9 @@ def SVCREATE_3_BF16 : SInst<"svcreate3[_{d}]", "3ddd",  "b", MergeNone, "", [IsT
 def SVCREATE_4_BF16 : SInst<"svcreate4[_{d}]", "4dddd", "b", MergeNone, "", [IsTupleCreate]>;
 }
 
-let TargetGuard = "sve2p1" in {
-  def SVCREATE_2_B : SInst<"svcreate2[_{d}]", "2dd",   "Pc", MergeNone, "", [IsTupleCreate]>;
-  def SVCREATE_4_B : SInst<"svcreate4[_{d}]", "4dddd", "Pc", MergeNone, "", [IsTupleCreate]>;
+let TargetGuard = "sve2p1|sme2" in {
+  def SVCREATE_2_B : SInst<"svcreate2_b", "2dd",   "Pc", MergeNone, "", [IsTupleCreate, IsStreamingCompatible]>;
+  def SVCREATE_4_B : SInst<"svcreate4_b", "4dddd", "Pc", MergeNone, "", [IsTupleCreate, IsStreamingCompatible]>;
 }
 
 ////////////////////////////////////////////////////////////////////////////////
@@ -1321,12 +1321,17 @@ def SVSET_3_BF16 : SInst<"svset3[_{d}]", "33id", "b", MergeNone, "", [IsTupleSet
 def SVSET_4_BF16 : SInst<"svset4[_{d}]", "44id", "b", MergeNone, "", [IsTupleSet], [ImmCheck<1, ImmCheck0_3>]>;
 }
 
-let TargetGuard = "sve2p1" in {
-  def SVGET_2_B : SInst<"svget2[_{d}]", "d2i", "Pc", MergeNone, "", [IsTupleGet], [ImmCheck<1, ImmCheck0_1>]>;
-  def SVGET_4_B : SInst<"svget4[_{d}]", "d4i", "Pc", MergeNone, "", [IsTupleGet], [ImmCheck<1, ImmCheck0_3>]>;
+let TargetGuard = "sve2p1|sme2" in {
+  def SVGET_2_B : SInst<"svget2_b", "d2i", "Pc", MergeNone, "", [IsTupleGet], [ImmCheck<1, ImmCheck0_1>]>;
+  def SVGET_4_B : SInst<"svget4_b", "d4i", "Pc", MergeNone, "", [IsTupleGet], [ImmCheck<1, ImmCheck0_3>]>;
 
-  def SVSET_2_B : SInst<"svset2[_{d}]", "22id", "Pc", MergeNone, "", [IsTupleSet], [ImmCheck<1, ImmCheck0_1>]>;
-  def SVSET_4_B : SInst<"svset4[_{d}]", "44id", "Pc", MergeNone, "", [IsTupleSet], [ImmCheck<1, ImmCheck0_3>]>;
+  def SVSET_2_B : SInst<"svset2_b", "22id", "Pc", MergeNone, "", [IsTupleSet], [ImmCheck<1, ImmCheck0_1>]>;
+  def SVSET_4_B : SInst<"svset4_b", "44id", "Pc", MergeNone, "", [IsTupleSet], [ImmCheck<1, ImmCheck0_3>]>;
+}
+
+let TargetGuard = "sve2p1|sme2" in {
+  def SVUNDEF_2_B: Inst<"svundef2_b", "2", "Pc", MergeNone, "", [IsUndef, IsStreamingCompatible], []>;
+  def SVUNDEF_4_B: Inst<"svundef4_b", "4", "Pc", MergeNone, "", [IsUndef, IsStreamingCompatible], []>;
 }
 ////////////////////////////////////////////////////////////////////////////////
 // SVE2 WhileGE/GT
diff --git a/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_create2_bool.c b/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_create2_bool.c
index eb5a19b1d9d326..b251d12b9b6349 100644
--- a/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_create2_bool.c
+++ b/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_create2_bool.c
@@ -3,36 +3,30 @@
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s \
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s\
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
 
 // REQUIRES: aarch64-registered-target
 
 #include <arm_sve.h>
 
-#ifdef SVE_OVERLOADED_FORMS
-// A simple used,unused... macro, long enough to represent any SVE builtin.
-#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
-#else
-#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
-#endif
-
-// CHECK-LABEL: @test_svcreate2_s8(
+// CHECK-LABEL: @test_svcreate2_b(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 32 x i1> @llvm.vector.insert.nxv32i1.nxv16i1(<vscale x 32 x i1> poison, <vscale x 16 x i1> [[X0:%.*]], i64 0)
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 32 x i1> @llvm.vector.insert.nxv32i1.nxv16i1(<vscale x 32 x i1> [[TMP0]], <vscale x 16 x i1> [[X1:%.*]], i64 16)
 // CHECK-NEXT:    ret <vscale x 32 x i1> [[TMP1]]
 //
-// CPP-CHECK-LABEL: @_Z17test_svcreate2_s8u10__SVBool_tS_(
+// CPP-CHECK-LABEL: @_Z16test_svcreate2_bu10__SVBool_tS_(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 32 x i1> @llvm.vector.insert.nxv32i1.nxv16i1(<vscale x 32 x i1> poison, <vscale x 16 x i1> [[X0:%.*]], i64 0)
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 32 x i1> @llvm.vector.insert.nxv32i1.nxv16i1(<vscale x 32 x i1> [[TMP0]], <vscale x 16 x i1> [[X1:%.*]], i64 16)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x i1> [[TMP1]]
 //
-svboolx2_t test_svcreate2_s8(svbool_t x0, svbool_t x1)
+svboolx2_t test_svcreate2_b(svbool_t x0, svbool_t x1)
 {
-  return SVE_ACLE_FUNC(svcreate2,_b8,,)(x0, x1);
+  return svcreate2_b(x0, x1);
 }
diff --git a/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_create4_bool.c b/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_create4_bool.c
index 3f99ee6e79c076..678689f8347283 100644
--- a/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_create4_bool.c
+++ b/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_create4_bool.c
@@ -3,24 +3,18 @@
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s \
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s\
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
 
 // REQUIRES: aarch64-registered-target
 
 #include <arm_sve.h>
 
-#ifdef SVE_OVERLOADED_FORMS
-// A simple used,unused... macro, long enough to represent any SVE builtin.
-#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
-#else
-#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
-#endif
-
-// CHECK-LABEL: @test_svcreate4_b8(
+// CHECK-LABEL: @test_svcreate4_b(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 64 x i1> @llvm.vector.insert.nxv64i1.nxv16i1(<vscale x 64 x i1> poison, <vscale x 16 x i1> [[X0:%.*]], i64 0)
 // CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 64 x i1> @llvm.vector.insert.nxv64i1.nxv16i1(<vscale x 64 x i1> [[TMP0]], <vscale x 16 x i1> [[X1:%.*]], i64 16)
@@ -28,7 +22,7 @@
 // CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 64 x i1> @llvm.vector.insert.nxv64i1.nxv16i1(<vscale x 64 x i1> [[TMP2]], <vscale x 16 x i1> [[X4:%.*]], i64 48)
 // CHECK-NEXT:    ret <vscale x 64 x i1> [[TMP3]]
 //
-// CPP-CHECK-LABEL: @_Z17test_svcreate4_b8u10__SVBool_tS_S_S_(
+// CPP-CHECK-LABEL: @_Z16test_svcreate4_bu10__SVBool_tS_S_S_(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 64 x i1> @llvm.vector.insert.nxv64i1.nxv16i1(<vscale x 64 x i1> poison, <vscale x 16 x i1> [[X0:%.*]], i64 0)
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 64 x i1> @llvm.vector.insert.nxv64i1.nxv16i1(<vscale x 64 x i1> [[TMP0]], <vscale x 16 x i1> [[X1:%.*]], i64 16)
@@ -36,7 +30,7 @@
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 64 x i1> @llvm.vector.insert.nxv64i1.nxv16i1(<vscale x 64 x i1> [[TMP2]], <vscale x 16 x i1> [[X4:%.*]], i64 48)
 // CPP-CHECK-NEXT:    ret <vscale x 64 x i1> [[TMP3]]
 //
-svboolx4_t test_svcreate4_b8(svbool_t x0, svbool_t x1, svbool_t x2, svbool_t x4)
+svboolx4_t test_svcreate4_b(svbool_t x0, svbool_t x1, svbool_t x2, svbool_t x4)
 {
-  return SVE_ACLE_FUNC(svcreate4,_b8,,)(x0, x1, x2, x4);
+  return svcreate4_b(x0, x1, x2, x4);
 }
diff --git a/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_get2_bool.c b/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_get2_bool.c
index 5d38f72b34b164..5adb3685f6252d 100644
--- a/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_get2_bool.c
+++ b/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_get2_bool.c
@@ -3,47 +3,41 @@
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s \
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s\
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
 // REQUIRES: aarch64-registered-target
 #include <arm_sve.h>
 
-#ifdef SVE_OVERLOADED_FORMS
-// A simple used,unused... macro, long enough to represent any SVE builtin.
-#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
-#else
-#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
-#endif
-
-// CHECK-LABEL: @test_svget2_b8_0(
+// CHECK-LABEL: @test_svget2_b_0(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i1> @llvm.vector.extract.nxv16i1.nxv32i1(<vscale x 32 x i1> [[TUPLE:%.*]], i64 0)
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
-// CPP-CHECK-LABEL: @_Z16test_svget2_b8_010svboolx2_t(
+// CPP-CHECK-LABEL: @_Z15test_svget2_b_010svboolx2_t(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i1> @llvm.vector.extract.nxv16i1.nxv32i1(<vscale x 32 x i1> [[TUPLE:%.*]], i64 0)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
-svbool_t test_svget2_b8_0(svboolx2_t tuple)
+svbool_t test_svget2_b_0(svboolx2_t tuple)
 {
-  return SVE_ACLE_FUNC(svget2,_b8,,)(tuple, 0);
+  return svget2_b(tuple, 0);
 }
 
-// CHECK-LABEL: @test_svget2_b8_1(
+// CHECK-LABEL: @test_svget2_b_1(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i1> @llvm.vector.extract.nxv16i1.nxv32i1(<vscale x 32 x i1> [[TUPLE:%.*]], i64 16)
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
-// CPP-CHECK-LABEL: @_Z16test_svget2_b8_110svboolx2_t(
+// CPP-CHECK-LABEL: @_Z15test_svget2_b_110svboolx2_t(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i1> @llvm.vector.extract.nxv16i1.nxv32i1(<vscale x 32 x i1> [[TUPLE:%.*]], i64 16)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
-svbool_t test_svget2_b8_1(svboolx2_t tuple)
+svbool_t test_svget2_b_1(svboolx2_t tuple)
 {
-  return SVE_ACLE_FUNC(svget2,_b8,,)(tuple, 1);
+  return svget2_b(tuple, 1);
 }
diff --git a/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_get4_bool.c b/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_get4_bool.c
index 821a5be3254eba..521b3a62225f10 100644
--- a/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_get4_bool.c
+++ b/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_get4_bool.c
@@ -3,70 +3,64 @@
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s \
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s\
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
 
 // REQUIRES: aarch64-registered-target
 
 #include <arm_sve.h>
 
-#ifdef SVE_OVERLOADED_FORMS
-// A simple used,unused... macro, long enough to represent any SVE builtin.
-#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
-#else
-#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
-#endif
-
 // NOTE: For these tests clang converts the struct parameter into
 // several parameters, one for each member of the original struct.
-// CHECK-LABEL: @test_svget4_b8_0(
+// CHECK-LABEL: @test_svget4_b_0(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i1> @llvm.vector.extract.nxv16i1.nxv64i1(<vscale x 64 x i1> [[TUPLE:%.*]], i64 0)
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
-// CPP-CHECK-LABEL: @_Z16test_svget4_b8_010svboolx4_t(
+// CPP-CHECK-LABEL: @_Z15test_svget4_b_010svboolx4_t(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i1> @llvm.vector.extract.nxv16i1.nxv64i1(<vscale x 64 x i1> [[TUPLE:%.*]], i64 0)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
-svbool_t test_svget4_b8_0(svboolx4_t tuple)
+svbool_t test_svget4_b_0(svboolx4_t tuple)
 {
-  return SVE_ACLE_FUNC(svget4,_b8,,)(tuple, 0);
+  return svget4_b(tuple, 0);
 }
 
 // NOTE: For these tests clang converts the struct parameter into
 // several parameters, one for each member of the original struct.
-// CHECK-LABEL: @test_svget4_b8_1(
+// CHECK-LABEL: @test_svget4_b_1(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i1> @llvm.vector.extract.nxv16i1.nxv64i1(<vscale x 64 x i1> [[TUPLE:%.*]], i64 16)
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
-// CPP-CHECK-LABEL: @_Z16test_svget4_b8_110svboolx4_t(
+// CPP-CHECK-LABEL: @_Z15test_svget4_b_110svboolx4_t(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i1> @llvm.vector.extract.nxv16i1.nxv64i1(<vscale x 64 x i1> [[TUPLE:%.*]], i64 16)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
-svbool_t test_svget4_b8_1(svboolx4_t tuple)
+svbool_t test_svget4_b_1(svboolx4_t tuple)
 {
-  return SVE_ACLE_FUNC(svget4,_b8,,)(tuple, 1);
+  return svget4_b(tuple, 1);
 }
 
 // NOTE: For these tests clang converts the struct parameter into
 // several parameters, one for each member of the original struct.
-// CHECK-LABEL: @test_svget4_b8_3(
+// CHECK-LABEL: @test_svget4_b_3(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i1> @llvm.vector.extract.nxv16i1.nxv64i1(<vscale x 64 x i1> [[TUPLE:%.*]], i64 48)
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
-// CPP-CHECK-LABEL: @_Z16test_svget4_b8_310svboolx4_t(
+// CPP-CHECK-LABEL: @_Z15test_svget4_b_310svboolx4_t(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i1> @llvm.vector.extract.nxv16i1.nxv64i1(<vscale x 64 x i1> [[TUPLE:%.*]], i64 48)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
-svbool_t test_svget4_b8_3(svboolx4_t tuple)
+svbool_t test_svget4_b_3(svboolx4_t tuple)
 {
-  return SVE_ACLE_FUNC(svget4,_b8,,)(tuple, 3);
+  return svget4_b(tuple, 3);
 }
diff --git a/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_set2_bool.c b/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_set2_bool.c
index 298e5b80ceb004..bcf96318b576aa 100644
--- a/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_set2_bool.c
+++ b/clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_set2_bool.c
@@ -3,50 +3,44 @@
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s \
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s\
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s \
 // RUN: | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
 
 // REQUIRES: aarch64-registered-target
 
 #include <arm_sve.h>
 
-#ifdef SVE_OVERLOADED_FORMS
-// A simple used,unused... macro, long enough to represent any SVE builtin.
-#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
-#else
-#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
-#endif
-
-// CHECK-LABEL: @test_svset2_b8_0(
+// CHECK-LABEL: @test_svset2_b_0(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 32 x i1> @llvm.vector.insert.nxv32i1.nxv16i1(<vscale x 32 x i1> [[TUPLE:%.*]], <vscale x 16 x i1> [[X:%.*]], i64 0)
 // CHECK-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
 //
-// CPP-CHECK-LABEL: @_Z16test_svset2_b8_010svboolx2_tu10__SVBool_t(
+// CPP-CHECK-LABEL: @_Z15test_svset2_b_010svboolx2_tu10__SVBool_t(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 32 x i1> @llvm.vector.insert.nxv32i1.nxv16i1(<vscale x 32 x i1> [[TUPLE:%.*]], <vscale x 16 x i1> [[X:%.*]], i64 0)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
 //
-svboolx2_t test_svset2_b8_0(svboolx2_t tuple, svbool_t x)
+svboolx2_t test_svset2_b_0(svboolx2_t tuple, svbool_t x)
 {
-  return SVE_ACLE_FUNC(...
[truncated]

clang/include/clang/Basic/arm_sve.td

kmclaughlin-arm · 2024-01-11T10:48:00Z

clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_undef_bool.c

+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -O2 -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -O2 -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -O2 -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -O2 -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK


Please can you also add a line to test these builtins end to end? For example:
RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s

Ah thanks for spotting that, added now.

…llvm#77338) This patch renames the get, set, create and undef functions that deal with tuples of booleans to match the ACLE at https://github.com/ARM-software/acle/pull/257/files . It also enables them for SME2.

[Clang] Rename and enable boolean get, set, create and undef for sme2

091048a

This patch renames the get, set, create and undef functions that deal with tuples of booleans to match the ACLE at https://github.com/ARM-software/acle/pull/257/files . It also enables them for SME2.

SamTebbs33 requested review from momchil-velikov, dtemirbulatov, sdesmalen-arm, CarolineConcatto, kmclaughlin-arm and MDevereau January 8, 2024 16:35

llvmbot added clang Clang issues not falling into any other category clang:frontend Language frontend issues, e.g. anything involving "Sema" labels Jan 8, 2024

kmclaughlin-arm reviewed Jan 10, 2024

View reviewed changes

clang/include/clang/Basic/arm_sve.td Outdated Show resolved Hide resolved

fixup: make _b optional and add IsStreamingCompatible to get and set

4518b63

kmclaughlin-arm reviewed Jan 11, 2024

View reviewed changes

fixup: add end-to-end RUN line to undef test

e96d9d7

kmclaughlin-arm approved these changes Jan 15, 2024

View reviewed changes

SamTebbs33 merged commit f112caf into llvm:main Jan 15, 2024

SamTebbs33 deleted the getsetcreateundef branch January 15, 2024 15:15

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[Clang] Rename and enable boolean get, set, create and undef for sme2 #77338

[Clang] Rename and enable boolean get, set, create and undef for sme2 #77338

Uh oh!

SamTebbs33 commented Jan 8, 2024

Uh oh!

llvmbot commented Jan 8, 2024

Uh oh!

Uh oh!

kmclaughlin-arm Jan 11, 2024

Uh oh!

SamTebbs33 Jan 11, 2024

Uh oh!

Uh oh!

[Clang] Rename and enable boolean get, set, create and undef for sme2 #77338

[Clang] Rename and enable boolean get, set, create and undef for sme2 #77338

Uh oh!

Conversation

SamTebbs33 commented Jan 8, 2024

Uh oh!

llvmbot commented Jan 8, 2024

Uh oh!

Uh oh!

kmclaughlin-arm Jan 11, 2024

Choose a reason for hiding this comment

Uh oh!

SamTebbs33 Jan 11, 2024

Choose a reason for hiding this comment

Uh oh!

Uh oh!