Skip to content

[AARCH64] Add FEAT_SSVE_FEXPA and fix unsupported features list #134368

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 2 commits into from
Apr 16, 2025
Merged

[AARCH64] Add FEAT_SSVE_FEXPA and fix unsupported features list #134368

merged 2 commits into from
Apr 16, 2025

Conversation

Lukacma
Copy link
Contributor

@Lukacma Lukacma commented Apr 4, 2025

This patch adds new feature introduced in 2025-03 release and changes feature requirements for fexpa instructions and intrinsics.

Additionally it fixes unsupported features list by moving fearures dependent on sme2p1 to correct location.

@llvmbot llvmbot added clang Clang issues not falling into any other category backend:AArch64 clang:driver 'clang' and 'clang++' user-facing binaries. Not 'clang-cl' clang:frontend Language frontend issues, e.g. anything involving "Sema" mc Machine (object) code labels Apr 4, 2025
@llvmbot
Copy link
Member

llvmbot commented Apr 4, 2025 •
edited
Loading

@llvm/pr-subscribers-clang
@llvm/pr-subscribers-mc

@llvm/pr-subscribers-clang-driver

Author: None (Lukacma)

Changes

This patch adds new feature introduced in 2025-03 release and changes feature requirements for fexpa instructions and intrinsics.

Additionally it fixes unsupported features list by moving fearures dependent on sme2p1 to correct location.

Full diff: https://github.com/llvm/llvm-project/pull/134368.diff

9 Files Affected:

  • (modified) clang/include/clang/Basic/arm_sve.td (+4-2)
  • (modified) clang/test/CodeGen/AArch64/sve-intrinsics/acle_sve_expa.c (+11-3)
  • (modified) clang/test/Driver/print-supported-extensions-aarch64.c (+1)
  • (modified) llvm/lib/Target/AArch64/AArch64.td (+2-3)
  • (modified) llvm/lib/Target/AArch64/AArch64Features.td (+3-1)
  • (modified) llvm/lib/Target/AArch64/AArch64InstrInfo.td (+4)
  • (modified) llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td (+1-1)
  • (modified) llvm/test/CodeGen/AArch64/sve-intrinsics-fexpa.ll (+1-1)
  • (modified) llvm/test/MC/AArch64/SVE/fexpa.s (+4-4)
diff --git a/clang/include/clang/Basic/arm_sve.td b/clang/include/clang/Basic/arm_sve.td
index 35263541b67ae..f09f40ce9202e 100644
--- a/clang/include/clang/Basic/arm_sve.td
+++ b/clang/include/clang/Basic/arm_sve.td
@@ -836,9 +836,11 @@ defm SVRINTP : SInstZPZ<"svrintp", "hfd", "aarch64_sve_frintp">;
 defm SVRINTX : SInstZPZ<"svrintx", "hfd", "aarch64_sve_frintx">;
 defm SVRINTZ : SInstZPZ<"svrintz", "hfd", "aarch64_sve_frintz">;
 defm SVSQRT  : SInstZPZ<"svsqrt",  "hfd", "aarch64_sve_fsqrt">;
-
+def SVEXPA  : SInst<"svexpa[_{d}]",  "du",   "hfd", MergeNone, "aarch64_sve_fexpa_x", [VerifyRuntimeMode]>{
+   let SVETargetGuard = "sve";
+   let SMETargetGuard = "sme2,ssve-fexpa";
+}
 let SVETargetGuard = "sve", SMETargetGuard = InvalidMode in {
-def SVEXPA  : SInst<"svexpa[_{d}]",  "du",   "hfd", MergeNone, "aarch64_sve_fexpa_x">;
 def SVTMAD  : SInst<"svtmad[_{d}]",  "dddi", "hfd", MergeNone, "aarch64_sve_ftmad_x", [], [ImmCheck<2, ImmCheck0_7>]>;
 def SVTSMUL : SInst<"svtsmul[_{d}]", "ddu",  "hfd", MergeNone, "aarch64_sve_ftsmul_x">;
 def SVTSSEL : SInst<"svtssel[_{d}]", "ddu",  "hfd", MergeNone, "aarch64_sve_ftssel_x">;
diff --git a/clang/test/CodeGen/AArch64/sve-intrinsics/acle_sve_expa.c b/clang/test/CodeGen/AArch64/sve-intrinsics/acle_sve_expa.c
index 52b6822a833f7..8c34017b7750b 100644
--- a/clang/test/CodeGen/AArch64/sve-intrinsics/acle_sve_expa.c
+++ b/clang/test/CodeGen/AArch64/sve-intrinsics/acle_sve_expa.c
@@ -1,10 +1,12 @@
 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
 // REQUIRES: aarch64-registered-target
 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64 -target-feature +ssve-fexpa -target-feature +sme2 -target-feature +sme -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64 -target-feature +sve -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64 -target-feature +sve -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+// RUN: %clang_cc1 -triple aarch64 -target-feature +ssve-fexpa -target-feature +sme2 -target-feature +sme -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
 #include <arm_sve.h>
 
 #ifdef SVE_OVERLOADED_FORMS
@@ -14,6 +16,12 @@
 #define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
 #endif
 
+#ifdef __ARM_FEATURE_SME
+#define STREAMING __arm_streaming
+#else
+#define STREAMING
+#endif
+
 // CHECK-LABEL: @test_svexpa_f16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.aarch64.sve.fexpa.x.nxv8f16(<vscale x 8 x i16> [[OP:%.*]])
@@ -24,7 +32,7 @@
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.aarch64.sve.fexpa.x.nxv8f16(<vscale x 8 x i16> [[OP:%.*]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP0]]
 //
-svfloat16_t test_svexpa_f16(svuint16_t op)
+svfloat16_t test_svexpa_f16(svuint16_t op) STREAMING
 {
   return SVE_ACLE_FUNC(svexpa,_f16,,)(op);
 }
@@ -39,7 +47,7 @@ svfloat16_t test_svexpa_f16(svuint16_t op)
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.fexpa.x.nxv4f32(<vscale x 4 x i32> [[OP:%.*]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
-svfloat32_t test_svexpa_f32(svuint32_t op)
+svfloat32_t test_svexpa_f32(svuint32_t op) STREAMING
 {
   return SVE_ACLE_FUNC(svexpa,_f32,,)(op);
 }
@@ -54,7 +62,7 @@ svfloat32_t test_svexpa_f32(svuint32_t op)
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.aarch64.sve.fexpa.x.nxv2f64(<vscale x 2 x i64> [[OP:%.*]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
-svfloat64_t test_svexpa_f64(svuint64_t op)
+svfloat64_t test_svexpa_f64(svuint64_t op) STREAMING
 {
   return SVE_ACLE_FUNC(svexpa,_f64,,)(op);
 }
diff --git a/clang/test/Driver/print-supported-extensions-aarch64.c b/clang/test/Driver/print-supported-extensions-aarch64.c
index 38a3f54eb4794..539c1937a9712 100644
--- a/clang/test/Driver/print-supported-extensions-aarch64.c
+++ b/clang/test/Driver/print-supported-extensions-aarch64.c
@@ -81,6 +81,7 @@
 // CHECK-NEXT:     ssbs                FEAT_SSBS, FEAT_SSBS2                                  Enable Speculative Store Bypass Safe bit
 // CHECK-NEXT:     ssve-aes            FEAT_SSVE_AES                                          Enable Armv9.6-A SVE AES support in streaming SVE mode
 // CHECK-NEXT:     ssve-bitperm        FEAT_SSVE_BitPerm                                      Enable Armv9.6-A SVE BitPerm support in streaming SVE mode
+// CHECK-NEXT:     ssve-fexpa          FEAT_SSVE_FEXPA                                        Enable SVE FEXPA instruction in Streaming SVE mode
 // CHECK-NEXT:     ssve-fp8dot2        FEAT_SSVE_FP8DOT2                                      Enable SVE2 FP8 2-way dot product instructions
 // CHECK-NEXT:     ssve-fp8dot4        FEAT_SSVE_FP8DOT4                                      Enable SVE2 FP8 4-way dot product instructions
 // CHECK-NEXT:     ssve-fp8fma         FEAT_SSVE_FP8FMA                                       Enable SVE2 FP8 multiply-add instructions
diff --git a/llvm/lib/Target/AArch64/AArch64.td b/llvm/lib/Target/AArch64/AArch64.td
index 3677f669c3481..b66c88e2d6245 100644
--- a/llvm/lib/Target/AArch64/AArch64.td
+++ b/llvm/lib/Target/AArch64/AArch64.td
@@ -74,12 +74,11 @@ def SVEUnsupported : AArch64Unsupported {
 }
 
 let F = [HasSME2p2, HasSVE2p2_or_SME2p2, HasNonStreamingSVE_or_SME2p2,
-         HasNonStreamingSVE2p2_or_SME2p2, HasNonStreamingSVE2_or_SSVE_BitPerm,
-         HasSME_MOP4, HasSME_TMOP] in
+         HasNonStreamingSVE2p2_or_SME2p2] in
 def SME2p2Unsupported : AArch64Unsupported;
 
 def SME2p1Unsupported : AArch64Unsupported {
-  let F = !listconcat([HasSME2p1, HasSVE2p1_or_SME2p1, HasNonStreamingSVE2p1_or_SSVE_AES],
+  let F = !listconcat([HasSME2p1, HasSVE2p1_or_SME2p1, HasNonStreamingSVE2p1_or_SSVE_AES, HasSME_MOP4, HasSME_TMOP, HasNonStreamingSVE_or_SSVE_FEXPA, HasNonStreamingSVE2_or_SSVE_BitPerm],
                       SME2p2Unsupported.F);
 }
 
diff --git a/llvm/lib/Target/AArch64/AArch64Features.td b/llvm/lib/Target/AArch64/AArch64Features.td
index 357f526d5e308..f4f931a5cdab1 100644
--- a/llvm/lib/Target/AArch64/AArch64Features.td
+++ b/llvm/lib/Target/AArch64/AArch64Features.td
@@ -576,7 +576,9 @@ def FeatureSME_MOP4: ExtensionWithMArch<"sme-mop4", "SME_MOP4", "FEAT_SME_MOP4",
 def FeatureSME_TMOP: ExtensionWithMArch<"sme-tmop", "SME_TMOP", "FEAT_SME_TMOP",
   "Enable SME Structured sparsity outer product instructions.", [FeatureSME2]>;
 
-//===----------------------------------------------------------------------===//
+def FeatureSSVE_FEXPA : ExtensionWithMArch<"ssve-fexpa", "SSVE_FEXPA", "FEAT_SSVE_FEXPA",
+  "Enable SVE FEXPA instruction in Streaming SVE mode", [FeatureSME2]>;
+
 //  Other Features
 //===----------------------------------------------------------------------===//
 
diff --git a/llvm/lib/Target/AArch64/AArch64InstrInfo.td b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
index a3b1ae55df028..7d0c453201dec 100644
--- a/llvm/lib/Target/AArch64/AArch64InstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
@@ -294,6 +294,10 @@ def HasNonStreamingSVE2_or_SSVE_BitPerm
     : Predicate<"(Subtarget->isSVEAvailable() && Subtarget->hasSVE2()) ||"
                 "(Subtarget->isSVEorStreamingSVEAvailable() && Subtarget->hasSSVE_BitPerm())">,
                 AssemblerPredicateWithAll<(any_of FeatureSVE2, FeatureSSVE_BitPerm), "sve2 or ssve-bitperm">;
+def HasNonStreamingSVE_or_SSVE_FEXPA
+    : Predicate<"(Subtarget->isSVEAvailable() && Subtarget->hasSVE()) ||"
+                "(Subtarget->isSVEorStreamingSVEAvailable() && Subtarget->hasSSVE_FEXPA())">,
+                AssemblerPredicateWithAll<(any_of FeatureSVE, FeatureSSVE_FEXPA), "sve or ssve-fexpa">;
 
 // A subset of NEON instructions are legal in Streaming SVE execution mode,
 // so don't need the additional check for 'isNeonAvailable'.
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index a2f326c994c2f..b40c82a25e7ba 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -980,7 +980,7 @@ let Predicates = [HasSVE_or_SME] in {
   def MOVPRFX_ZZ : sve_int_bin_cons_misc_0_c<0b00000001, "movprfx", ZPRAny>;
 } // End HasSVE_or_SME
 
-let Predicates = [HasNonStreamingSVE_or_SME2p2] in {
+let Predicates = [HasNonStreamingSVE_or_SSVE_FEXPA] in {
   defm FEXPA_ZZ : sve_int_bin_cons_misc_0_c_fexpa<"fexpa", int_aarch64_sve_fexpa_x>;
 } // End HasSVE
 
diff --git a/llvm/test/CodeGen/AArch64/sve-intrinsics-fexpa.ll b/llvm/test/CodeGen/AArch64/sve-intrinsics-fexpa.ll
index 00e000f642377..021d4855905e7 100644
--- a/llvm/test/CodeGen/AArch64/sve-intrinsics-fexpa.ll
+++ b/llvm/test/CodeGen/AArch64/sve-intrinsics-fexpa.ll
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s | FileCheck %s
-; RUN: llc -mtriple=aarch64-linux-gnu -force-streaming -mattr=+sme2p2 < %s | FileCheck %s
+; RUN: llc -mtriple=aarch64-linux-gnu -force-streaming -mattr=+ssve-fexpa < %s | FileCheck %s
 
 define <vscale x 8 x half> @fexpa_h(<vscale x 8 x i16> %a) {
 ; CHECK-LABEL: fexpa_h:
diff --git a/llvm/test/MC/AArch64/SVE/fexpa.s b/llvm/test/MC/AArch64/SVE/fexpa.s
index c51b1e2b1d3e5..c6386255b274e 100644
--- a/llvm/test/MC/AArch64/SVE/fexpa.s
+++ b/llvm/test/MC/AArch64/SVE/fexpa.s
@@ -1,6 +1,6 @@
 // RUN: llvm-mc -triple=aarch64 -show-encoding -mattr=+sve < %s \
 // RUN:        | FileCheck %s --check-prefixes=CHECK-ENCODING,CHECK-INST
-// RUN: llvm-mc -triple=aarch64 -show-encoding -mattr=+sme2p2 < %s \
+// RUN: llvm-mc -triple=aarch64 -show-encoding -mattr=+ssve-fexpa < %s \
 // RUN:        | FileCheck %s --check-prefixes=CHECK-ENCODING,CHECK-INST
 // RUN: not llvm-mc -triple=aarch64 -show-encoding < %s 2>&1 \
 // RUN:        | FileCheck %s --check-prefix=CHECK-ERROR
@@ -14,17 +14,17 @@
 fexpa z0.h, z31.h
 // CHECK-INST: fexpa	z0.h, z31.h
 // CHECK-ENCODING: [0xe0,0xbb,0x60,0x04]
-// CHECK-ERROR: instruction requires: sve or sme2p2
+// CHECK-ERROR: instruction requires: sve or ssve-fexpa
 // CHECK-UNKNOWN: 0460bbe0 <unknown>
 
 fexpa z0.s, z31.s
 // CHECK-INST: fexpa	z0.s, z31.s
 // CHECK-ENCODING: [0xe0,0xbb,0xa0,0x04]
-// CHECK-ERROR: instruction requires: sve or sme2p2
+// CHECK-ERROR: instruction requires: sve or ssve-fexpa
 // CHECK-UNKNOWN: 04a0bbe0 <unknown>
 
 fexpa z0.d, z31.d
 // CHECK-INST: fexpa	z0.d, z31.d
 // CHECK-ENCODING: [0xe0,0xbb,0xe0,0x04]
-// CHECK-ERROR: instruction requires: sve or sme2p2
+// CHECK-ERROR: instruction requires: sve or ssve-fexpa
 // CHECK-UNKNOWN: 04e0bbe0 <unknown>

@llvmbot
Copy link
Member

llvmbot commented Apr 4, 2025

@llvm/pr-subscribers-backend-aarch64

Author: None (Lukacma)

Changes

This patch adds new feature introduced in 2025-03 release and changes feature requirements for fexpa instructions and intrinsics.

Additionally it fixes unsupported features list by moving fearures dependent on sme2p1 to correct location.

Full diff: https://github.com/llvm/llvm-project/pull/134368.diff

9 Files Affected:

  • (modified) clang/include/clang/Basic/arm_sve.td (+4-2)
  • (modified) clang/test/CodeGen/AArch64/sve-intrinsics/acle_sve_expa.c (+11-3)
  • (modified) clang/test/Driver/print-supported-extensions-aarch64.c (+1)
  • (modified) llvm/lib/Target/AArch64/AArch64.td (+2-3)
  • (modified) llvm/lib/Target/AArch64/AArch64Features.td (+3-1)
  • (modified) llvm/lib/Target/AArch64/AArch64InstrInfo.td (+4)
  • (modified) llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td (+1-1)
  • (modified) llvm/test/CodeGen/AArch64/sve-intrinsics-fexpa.ll (+1-1)
  • (modified) llvm/test/MC/AArch64/SVE/fexpa.s (+4-4)
diff --git a/clang/include/clang/Basic/arm_sve.td b/clang/include/clang/Basic/arm_sve.td
index 35263541b67ae..f09f40ce9202e 100644
--- a/clang/include/clang/Basic/arm_sve.td
+++ b/clang/include/clang/Basic/arm_sve.td
@@ -836,9 +836,11 @@ defm SVRINTP : SInstZPZ<"svrintp", "hfd", "aarch64_sve_frintp">;
 defm SVRINTX : SInstZPZ<"svrintx", "hfd", "aarch64_sve_frintx">;
 defm SVRINTZ : SInstZPZ<"svrintz", "hfd", "aarch64_sve_frintz">;
 defm SVSQRT  : SInstZPZ<"svsqrt",  "hfd", "aarch64_sve_fsqrt">;
-
+def SVEXPA  : SInst<"svexpa[_{d}]",  "du",   "hfd", MergeNone, "aarch64_sve_fexpa_x", [VerifyRuntimeMode]>{
+   let SVETargetGuard = "sve";
+   let SMETargetGuard = "sme2,ssve-fexpa";
+}
 let SVETargetGuard = "sve", SMETargetGuard = InvalidMode in {
-def SVEXPA  : SInst<"svexpa[_{d}]",  "du",   "hfd", MergeNone, "aarch64_sve_fexpa_x">;
 def SVTMAD  : SInst<"svtmad[_{d}]",  "dddi", "hfd", MergeNone, "aarch64_sve_ftmad_x", [], [ImmCheck<2, ImmCheck0_7>]>;
 def SVTSMUL : SInst<"svtsmul[_{d}]", "ddu",  "hfd", MergeNone, "aarch64_sve_ftsmul_x">;
 def SVTSSEL : SInst<"svtssel[_{d}]", "ddu",  "hfd", MergeNone, "aarch64_sve_ftssel_x">;
diff --git a/clang/test/CodeGen/AArch64/sve-intrinsics/acle_sve_expa.c b/clang/test/CodeGen/AArch64/sve-intrinsics/acle_sve_expa.c
index 52b6822a833f7..8c34017b7750b 100644
--- a/clang/test/CodeGen/AArch64/sve-intrinsics/acle_sve_expa.c
+++ b/clang/test/CodeGen/AArch64/sve-intrinsics/acle_sve_expa.c
@@ -1,10 +1,12 @@
 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
 // REQUIRES: aarch64-registered-target
 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64 -target-feature +ssve-fexpa -target-feature +sme2 -target-feature +sme -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64 -target-feature +sve -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64 -target-feature +sve -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+// RUN: %clang_cc1 -triple aarch64 -target-feature +ssve-fexpa -target-feature +sme2 -target-feature +sme -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
 #include <arm_sve.h>
 
 #ifdef SVE_OVERLOADED_FORMS
@@ -14,6 +16,12 @@
 #define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
 #endif
 
+#ifdef __ARM_FEATURE_SME
+#define STREAMING __arm_streaming
+#else
+#define STREAMING
+#endif
+
 // CHECK-LABEL: @test_svexpa_f16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.aarch64.sve.fexpa.x.nxv8f16(<vscale x 8 x i16> [[OP:%.*]])
@@ -24,7 +32,7 @@
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.aarch64.sve.fexpa.x.nxv8f16(<vscale x 8 x i16> [[OP:%.*]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP0]]
 //
-svfloat16_t test_svexpa_f16(svuint16_t op)
+svfloat16_t test_svexpa_f16(svuint16_t op) STREAMING
 {
   return SVE_ACLE_FUNC(svexpa,_f16,,)(op);
 }
@@ -39,7 +47,7 @@ svfloat16_t test_svexpa_f16(svuint16_t op)
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.fexpa.x.nxv4f32(<vscale x 4 x i32> [[OP:%.*]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
-svfloat32_t test_svexpa_f32(svuint32_t op)
+svfloat32_t test_svexpa_f32(svuint32_t op) STREAMING
 {
   return SVE_ACLE_FUNC(svexpa,_f32,,)(op);
 }
@@ -54,7 +62,7 @@ svfloat32_t test_svexpa_f32(svuint32_t op)
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.aarch64.sve.fexpa.x.nxv2f64(<vscale x 2 x i64> [[OP:%.*]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
-svfloat64_t test_svexpa_f64(svuint64_t op)
+svfloat64_t test_svexpa_f64(svuint64_t op) STREAMING
 {
   return SVE_ACLE_FUNC(svexpa,_f64,,)(op);
 }
diff --git a/clang/test/Driver/print-supported-extensions-aarch64.c b/clang/test/Driver/print-supported-extensions-aarch64.c
index 38a3f54eb4794..539c1937a9712 100644
--- a/clang/test/Driver/print-supported-extensions-aarch64.c
+++ b/clang/test/Driver/print-supported-extensions-aarch64.c
@@ -81,6 +81,7 @@
 // CHECK-NEXT:     ssbs                FEAT_SSBS, FEAT_SSBS2                                  Enable Speculative Store Bypass Safe bit
 // CHECK-NEXT:     ssve-aes            FEAT_SSVE_AES                                          Enable Armv9.6-A SVE AES support in streaming SVE mode
 // CHECK-NEXT:     ssve-bitperm        FEAT_SSVE_BitPerm                                      Enable Armv9.6-A SVE BitPerm support in streaming SVE mode
+// CHECK-NEXT:     ssve-fexpa          FEAT_SSVE_FEXPA                                        Enable SVE FEXPA instruction in Streaming SVE mode
 // CHECK-NEXT:     ssve-fp8dot2        FEAT_SSVE_FP8DOT2                                      Enable SVE2 FP8 2-way dot product instructions
 // CHECK-NEXT:     ssve-fp8dot4        FEAT_SSVE_FP8DOT4                                      Enable SVE2 FP8 4-way dot product instructions
 // CHECK-NEXT:     ssve-fp8fma         FEAT_SSVE_FP8FMA                                       Enable SVE2 FP8 multiply-add instructions
diff --git a/llvm/lib/Target/AArch64/AArch64.td b/llvm/lib/Target/AArch64/AArch64.td
index 3677f669c3481..b66c88e2d6245 100644
--- a/llvm/lib/Target/AArch64/AArch64.td
+++ b/llvm/lib/Target/AArch64/AArch64.td
@@ -74,12 +74,11 @@ def SVEUnsupported : AArch64Unsupported {
 }
 
 let F = [HasSME2p2, HasSVE2p2_or_SME2p2, HasNonStreamingSVE_or_SME2p2,
-         HasNonStreamingSVE2p2_or_SME2p2, HasNonStreamingSVE2_or_SSVE_BitPerm,
-         HasSME_MOP4, HasSME_TMOP] in
+         HasNonStreamingSVE2p2_or_SME2p2] in
 def SME2p2Unsupported : AArch64Unsupported;
 
 def SME2p1Unsupported : AArch64Unsupported {
-  let F = !listconcat([HasSME2p1, HasSVE2p1_or_SME2p1, HasNonStreamingSVE2p1_or_SSVE_AES],
+  let F = !listconcat([HasSME2p1, HasSVE2p1_or_SME2p1, HasNonStreamingSVE2p1_or_SSVE_AES, HasSME_MOP4, HasSME_TMOP, HasNonStreamingSVE_or_SSVE_FEXPA, HasNonStreamingSVE2_or_SSVE_BitPerm],
                       SME2p2Unsupported.F);
 }
 
diff --git a/llvm/lib/Target/AArch64/AArch64Features.td b/llvm/lib/Target/AArch64/AArch64Features.td
index 357f526d5e308..f4f931a5cdab1 100644
--- a/llvm/lib/Target/AArch64/AArch64Features.td
+++ b/llvm/lib/Target/AArch64/AArch64Features.td
@@ -576,7 +576,9 @@ def FeatureSME_MOP4: ExtensionWithMArch<"sme-mop4", "SME_MOP4", "FEAT_SME_MOP4",
 def FeatureSME_TMOP: ExtensionWithMArch<"sme-tmop", "SME_TMOP", "FEAT_SME_TMOP",
   "Enable SME Structured sparsity outer product instructions.", [FeatureSME2]>;
 
-//===----------------------------------------------------------------------===//
+def FeatureSSVE_FEXPA : ExtensionWithMArch<"ssve-fexpa", "SSVE_FEXPA", "FEAT_SSVE_FEXPA",
+  "Enable SVE FEXPA instruction in Streaming SVE mode", [FeatureSME2]>;
+
 //  Other Features
 //===----------------------------------------------------------------------===//
 
diff --git a/llvm/lib/Target/AArch64/AArch64InstrInfo.td b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
index a3b1ae55df028..7d0c453201dec 100644
--- a/llvm/lib/Target/AArch64/AArch64InstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
@@ -294,6 +294,10 @@ def HasNonStreamingSVE2_or_SSVE_BitPerm
     : Predicate<"(Subtarget->isSVEAvailable() && Subtarget->hasSVE2()) ||"
                 "(Subtarget->isSVEorStreamingSVEAvailable() && Subtarget->hasSSVE_BitPerm())">,
                 AssemblerPredicateWithAll<(any_of FeatureSVE2, FeatureSSVE_BitPerm), "sve2 or ssve-bitperm">;
+def HasNonStreamingSVE_or_SSVE_FEXPA
+    : Predicate<"(Subtarget->isSVEAvailable() && Subtarget->hasSVE()) ||"
+                "(Subtarget->isSVEorStreamingSVEAvailable() && Subtarget->hasSSVE_FEXPA())">,
+                AssemblerPredicateWithAll<(any_of FeatureSVE, FeatureSSVE_FEXPA), "sve or ssve-fexpa">;
 
 // A subset of NEON instructions are legal in Streaming SVE execution mode,
 // so don't need the additional check for 'isNeonAvailable'.
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index a2f326c994c2f..b40c82a25e7ba 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -980,7 +980,7 @@ let Predicates = [HasSVE_or_SME] in {
   def MOVPRFX_ZZ : sve_int_bin_cons_misc_0_c<0b00000001, "movprfx", ZPRAny>;
 } // End HasSVE_or_SME
 
-let Predicates = [HasNonStreamingSVE_or_SME2p2] in {
+let Predicates = [HasNonStreamingSVE_or_SSVE_FEXPA] in {
   defm FEXPA_ZZ : sve_int_bin_cons_misc_0_c_fexpa<"fexpa", int_aarch64_sve_fexpa_x>;
 } // End HasSVE
 
diff --git a/llvm/test/CodeGen/AArch64/sve-intrinsics-fexpa.ll b/llvm/test/CodeGen/AArch64/sve-intrinsics-fexpa.ll
index 00e000f642377..021d4855905e7 100644
--- a/llvm/test/CodeGen/AArch64/sve-intrinsics-fexpa.ll
+++ b/llvm/test/CodeGen/AArch64/sve-intrinsics-fexpa.ll
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s | FileCheck %s
-; RUN: llc -mtriple=aarch64-linux-gnu -force-streaming -mattr=+sme2p2 < %s | FileCheck %s
+; RUN: llc -mtriple=aarch64-linux-gnu -force-streaming -mattr=+ssve-fexpa < %s | FileCheck %s
 
 define <vscale x 8 x half> @fexpa_h(<vscale x 8 x i16> %a) {
 ; CHECK-LABEL: fexpa_h:
diff --git a/llvm/test/MC/AArch64/SVE/fexpa.s b/llvm/test/MC/AArch64/SVE/fexpa.s
index c51b1e2b1d3e5..c6386255b274e 100644
--- a/llvm/test/MC/AArch64/SVE/fexpa.s
+++ b/llvm/test/MC/AArch64/SVE/fexpa.s
@@ -1,6 +1,6 @@
 // RUN: llvm-mc -triple=aarch64 -show-encoding -mattr=+sve < %s \
 // RUN:        | FileCheck %s --check-prefixes=CHECK-ENCODING,CHECK-INST
-// RUN: llvm-mc -triple=aarch64 -show-encoding -mattr=+sme2p2 < %s \
+// RUN: llvm-mc -triple=aarch64 -show-encoding -mattr=+ssve-fexpa < %s \
 // RUN:        | FileCheck %s --check-prefixes=CHECK-ENCODING,CHECK-INST
 // RUN: not llvm-mc -triple=aarch64 -show-encoding < %s 2>&1 \
 // RUN:        | FileCheck %s --check-prefix=CHECK-ERROR
@@ -14,17 +14,17 @@
 fexpa z0.h, z31.h
 // CHECK-INST: fexpa	z0.h, z31.h
 // CHECK-ENCODING: [0xe0,0xbb,0x60,0x04]
-// CHECK-ERROR: instruction requires: sve or sme2p2
+// CHECK-ERROR: instruction requires: sve or ssve-fexpa
 // CHECK-UNKNOWN: 0460bbe0 <unknown>
 
 fexpa z0.s, z31.s
 // CHECK-INST: fexpa	z0.s, z31.s
 // CHECK-ENCODING: [0xe0,0xbb,0xa0,0x04]
-// CHECK-ERROR: instruction requires: sve or sme2p2
+// CHECK-ERROR: instruction requires: sve or ssve-fexpa
 // CHECK-UNKNOWN: 04a0bbe0 <unknown>
 
 fexpa z0.d, z31.d
 // CHECK-INST: fexpa	z0.d, z31.d
 // CHECK-ENCODING: [0xe0,0xbb,0xe0,0x04]
-// CHECK-ERROR: instruction requires: sve or sme2p2
+// CHECK-ERROR: instruction requires: sve or ssve-fexpa
 // CHECK-UNKNOWN: 04e0bbe0 <unknown>

jthackray
jthackray approved these changes Apr 4, 2025
View reviewed changes
Copy link
Contributor

@jthackray jthackray left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

CarolineConcatto
CarolineConcatto reviewed Apr 16, 2025
View reviewed changes
llvm/lib/Target/AArch64/AArch64.td Outdated
def SME2p2Unsupported : AArch64Unsupported;

def SME2p1Unsupported : AArch64Unsupported {
let F = !listconcat([HasSME2p1, HasSVE2p1_or_SME2p1, HasNonStreamingSVE2p1_or_SSVE_AES],
let F = !listconcat([HasSME2p1, HasSVE2p1_or_SME2p1, HasNonStreamingSVE2p1_or_SSVE_AES, HasSME_MOP4, HasSME_TMOP, HasNonStreamingSVE_or_SSVE_FEXPA, HasNonStreamingSVE2_or_SSVE_BitPerm],
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: align

CarolineConcatto
CarolineConcatto approved these changes Apr 16, 2025
View reviewed changes
Copy link
Contributor

@CarolineConcatto CarolineConcatto left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Before merging address the nit in: llvm/lib/Target/AArch64/AArch64.td

@Lukacma Lukacma merged commit de90487 into llvm:main Apr 16, 2025
9 of 11 checks passed
var-const pushed a commit to ldionne/llvm-project that referenced this pull request Apr 17, 2025
@Lukacma @var-const
[AARCH64] Add FEAT_SSVE_FEXPA and fix unsupported features list (llvm…
c8766b8 
…#134368)

This patch adds new feature introduced in [2025-03
release](https://developer.arm.com/documentation/ddi0602/2025-03/SVE-Instructions/FEXPA--Floating-point-exponential-accelerator-)
and changes feature requirements for fexpa instructions and intrinsics.

Additionally it fixes unsupported features list by moving fearures
dependent on sme2p1 to correct location.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@jthackray jthackray jthackray approved these changes

@CarolineConcatto CarolineConcatto CarolineConcatto approved these changes

Assignees
No one assigned
Labels
backend:AArch64 clang:driver 'clang' and 'clang++' user-facing binaries. Not 'clang-cl' clang:frontend Language frontend issues, e.g. anything involving "Sema" clang Clang issues not falling into any other category mc Machine (object) code
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@Lukacma @llvmbot @jthackray @CarolineConcatto