Debian's clang carries a patch that makes the default FPU mode 'vfp3-d16' instead of 'neon' for 'armv7-a' to avoid generating NEON instructions on hardware that does not support them:
https://salsa.debian.org/pkg-llvm-team/llvm-toolchain/-/raw/5a61ca6f21b4ad8c... https://bugs.debian.org/841474 https://bugs.debian.org/842142 https://bugs.debian.org/914268
This results in the following build error when clang's integrated assembler is used because the '.arch' directive overrides the '.fpu' directive:
arch/arm/crypto/curve25519-core.S:25:2: error: instruction requires: NEON vmov.i32 q0, #1 ^ arch/arm/crypto/curve25519-core.S:26:2: error: instruction requires: NEON vshr.u64 q1, q0, #7 ^ arch/arm/crypto/curve25519-core.S:27:2: error: instruction requires: NEON vshr.u64 q0, q0, #8 ^ arch/arm/crypto/curve25519-core.S:28:2: error: instruction requires: NEON vmov.i32 d4, #19 ^
Shuffle the order of the '.arch' and '.fpu' directives so that the code builds regardless of the default FPU mode. This has been tested against both clang with and without Debian's patch and GCC.
Cc: stable@vger.kernel.org Fixes: d8f1308a025f ("crypto: arm/curve25519 - wire up NEON implementation") Link: https://github.com/ClangBuiltLinux/continuous-integration2/issues/118 Reported-by: Arnd Bergmann arnd@arndb.de Suggested-by: Arnd Bergmann arnd@arndb.de Suggested-by: Jessica Clarke jrtc27@jrtc27.com Signed-off-by: Nathan Chancellor nathan@kernel.org --- arch/arm/crypto/curve25519-core.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/arm/crypto/curve25519-core.S b/arch/arm/crypto/curve25519-core.S index be18af52e7dc..b697fa5d059a 100644 --- a/arch/arm/crypto/curve25519-core.S +++ b/arch/arm/crypto/curve25519-core.S @@ -10,8 +10,8 @@ #include <linux/linkage.h>
.text -.fpu neon .arch armv7-a +.fpu neon .align 4
ENTRY(curve25519_neon)
base-commit: e49d033bddf5b565044e2abe4241353959bc9120
On Fri, Apr 9, 2021 at 3:12 PM Nathan Chancellor nathan@kernel.org wrote:
Debian's clang carries a patch that makes the default FPU mode 'vfp3-d16' instead of 'neon' for 'armv7-a' to avoid generating NEON instructions on hardware that does not support them:
https://salsa.debian.org/pkg-llvm-team/llvm-toolchain/-/raw/5a61ca6f21b4ad8c... https://bugs.debian.org/841474 https://bugs.debian.org/842142 https://bugs.debian.org/914268
Another good link would be the one from Jessica describing more precisely what the ARM targets for Debian are: https://wiki.debian.org/ArchitectureSpecificsMemo#armel
This results in the following build error when clang's integrated assembler is used because the '.arch' directive overrides the '.fpu' directive:
arch/arm/crypto/curve25519-core.S:25:2: error: instruction requires: NEON vmov.i32 q0, #1 ^ arch/arm/crypto/curve25519-core.S:26:2: error: instruction requires: NEON vshr.u64 q1, q0, #7 ^ arch/arm/crypto/curve25519-core.S:27:2: error: instruction requires: NEON vshr.u64 q0, q0, #8 ^ arch/arm/crypto/curve25519-core.S:28:2: error: instruction requires: NEON vmov.i32 d4, #19 ^
Shuffle the order of the '.arch' and '.fpu' directives so that the code builds regardless of the default FPU mode. This has been tested against both clang with and without Debian's patch and GCC.
Cc: stable@vger.kernel.org Fixes: d8f1308a025f ("crypto: arm/curve25519 - wire up NEON implementation") Link: https://github.com/ClangBuiltLinux/continuous-integration2/issues/118 Reported-by: Arnd Bergmann arnd@arndb.de Suggested-by: Arnd Bergmann arnd@arndb.de Suggested-by: Jessica Clarke jrtc27@jrtc27.com Signed-off-by: Nathan Chancellor nathan@kernel.org
Great work tracking down that Debian was carrying patches! Thank you! I've run this through the same 3 assemblers.
Reviewed-by: Nick Desaulniers ndesaulniers@google.com Tested-by: Nick Desaulniers ndesaulniers@google.com
arch/arm/crypto/curve25519-core.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/arm/crypto/curve25519-core.S b/arch/arm/crypto/curve25519-core.S index be18af52e7dc..b697fa5d059a 100644 --- a/arch/arm/crypto/curve25519-core.S +++ b/arch/arm/crypto/curve25519-core.S @@ -10,8 +10,8 @@ #include <linux/linkage.h>
.text -.fpu neon .arch armv7-a +.fpu neon .align 4
ENTRY(curve25519_neon)
base-commit: e49d033bddf5b565044e2abe4241353959bc9120
2.31.1.189.g2e36527f23
On Fri, Apr 09, 2021 at 03:11:55PM -0700, Nathan Chancellor wrote:
Debian's clang carries a patch that makes the default FPU mode 'vfp3-d16' instead of 'neon' for 'armv7-a' to avoid generating NEON instructions on hardware that does not support them:
https://salsa.debian.org/pkg-llvm-team/llvm-toolchain/-/raw/5a61ca6f21b4ad8c... https://bugs.debian.org/841474 https://bugs.debian.org/842142 https://bugs.debian.org/914268
This results in the following build error when clang's integrated assembler is used because the '.arch' directive overrides the '.fpu' directive:
arch/arm/crypto/curve25519-core.S:25:2: error: instruction requires: NEON vmov.i32 q0, #1 ^ arch/arm/crypto/curve25519-core.S:26:2: error: instruction requires: NEON vshr.u64 q1, q0, #7 ^ arch/arm/crypto/curve25519-core.S:27:2: error: instruction requires: NEON vshr.u64 q0, q0, #8 ^ arch/arm/crypto/curve25519-core.S:28:2: error: instruction requires: NEON vmov.i32 d4, #19 ^
Shuffle the order of the '.arch' and '.fpu' directives so that the code builds regardless of the default FPU mode. This has been tested against both clang with and without Debian's patch and GCC.
Cc: stable@vger.kernel.org Fixes: d8f1308a025f ("crypto: arm/curve25519 - wire up NEON implementation") Link: https://github.com/ClangBuiltLinux/continuous-integration2/issues/118 Reported-by: Arnd Bergmann arnd@arndb.de Suggested-by: Arnd Bergmann arnd@arndb.de Suggested-by: Jessica Clarke jrtc27@jrtc27.com Signed-off-by: Nathan Chancellor nathan@kernel.org
arch/arm/crypto/curve25519-core.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Patch applied. Thanks.
linux-stable-mirror@lists.linaro.org
-
Herbert Xu -
Jason A. Donenfeld -
Nathan Chancellor -
Nick Desaulniers