[clang] 62744f3 - [AArch64][NEON] NEON intrinsic compilation error with -fno-lax-vector-conversion flag fix (#149329)

[via cfe-commits]

Author: Amina Chabane
Date: 2025-07-30T10:56:14+01:00
New Revision: 62744f368166f223740e5f6105aeb11ea854728d

URL: https://github.com/llvm/llvm-project/commit/62744f368166f223740e5f6105aeb11ea854728d
DIFF: https://github.com/llvm/llvm-project/commit/62744f368166f223740e5f6105aeb11ea854728d.diff

LOG: [AArch64][NEON] NEON intrinsic compilation error with -fno-lax-vector-conversion flag fix (#149329)

Issue originally raised in
https://github.com/llvm/llvm-project/issues/71362#issuecomment-3028515618.
Certain NEON intrinsics that operate on poly types (e.g. poly8x8_t)
failed to compile with the -fno-lax-vector-conversions flag. This patch
updates NeonEmitter.cpp to insert an explicit __builtin_bit_cast from
poly types to the required signed integer vector types when generating
lane-related intrinsics. A test 'neon-bitcast-poly.ll' is included.

Added: 
    

Modified: 
    clang/test/CodeGen/AArch64/neon-scalar-copy.c
    clang/test/CodeGen/AArch64/neon-vget.c
    clang/test/CodeGen/AArch64/poly64.c
    clang/utils/TableGen/NeonEmitter.cpp

Removed: 
    


################################################################################
diff  --git a/clang/test/CodeGen/AArch64/neon-scalar-copy.c b/clang/test/CodeGen/AArch64/neon-scalar-copy.c
index 4ad1ce53a3b39..bd80068e5bf85 100644
--- a/clang/test/CodeGen/AArch64/neon-scalar-copy.c
+++ b/clang/test/CodeGen/AArch64/neon-scalar-copy.c
@@ -1,4 +1,4 @@
-// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon \
+// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon -flax-vector-conversions=none\
 // RUN: -disable-O0-optnone -emit-llvm -o - %s | opt -S -passes=mem2reg | FileCheck %s
 
 // REQUIRES: aarch64-registered-target || arm-registered-target

diff  --git a/clang/test/CodeGen/AArch64/neon-vget.c b/clang/test/CodeGen/AArch64/neon-vget.c
index b17a7ab342817..ebc8c2f9228ed 100644
--- a/clang/test/CodeGen/AArch64/neon-vget.c
+++ b/clang/test/CodeGen/AArch64/neon-vget.c
@@ -1,4 +1,4 @@
-// RUN: %clang_cc1 -triple arm64-apple-darwin -target-feature +neon \
+// RUN: %clang_cc1 -triple arm64-apple-darwin -target-feature +neon -flax-vector-conversions=none \
 // RUN:   -disable-O0-optnone -emit-llvm -o - %s \
 // RUN: | opt -S -passes=mem2reg | FileCheck %s
 

diff  --git a/clang/test/CodeGen/AArch64/poly64.c b/clang/test/CodeGen/AArch64/poly64.c
index 578dd2054dc66..00838b648dfe6 100644
--- a/clang/test/CodeGen/AArch64/poly64.c
+++ b/clang/test/CodeGen/AArch64/poly64.c
@@ -1,5 +1,5 @@
 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
-// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon \
+// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon -flax-vector-conversions=none\
 // RUN:  -ffp-contract=fast -disable-O0-optnone -emit-llvm -o - %s | opt -S -passes=mem2reg,sroa \
 // RUN:  | FileCheck %s
 

diff  --git a/clang/utils/TableGen/NeonEmitter.cpp b/clang/utils/TableGen/NeonEmitter.cpp
index d4fb56e6a39b7..89891fd9310e9 100644
--- a/clang/utils/TableGen/NeonEmitter.cpp
+++ b/clang/utils/TableGen/NeonEmitter.cpp
@@ -1401,14 +1401,12 @@ void Intrinsic::emitBodyAsBuiltinCall() {
       if (LocalCK == ClassB || (T.isHalf() && !T.isScalarForMangling())) {
         CastToType.makeInteger(8, true);
         Arg = "__builtin_bit_cast(" + CastToType.str() + ", " + Arg + ")";
-      } else if (LocalCK == ClassI) {
-        if (CastToType.isInteger()) {
-          CastToType.makeSigned();
-          Arg = "__builtin_bit_cast(" + CastToType.str() + ", " + Arg + ")";
-        }
+      } else if (LocalCK == ClassI &&
+                 (CastToType.isInteger() || CastToType.isPoly())) {
+        CastToType.makeSigned();
+        Arg = "__builtin_bit_cast(" + CastToType.str() + ", " + Arg + ")";
       }
     }
-
     S += Arg + ", ";
   }