diff mbox series

x86/crc32: improve crc32c_arch() code generation with clang

Message ID 20250210210741.471725-1-ebiggers@kernel.org (mailing list archive)
State Not Applicable
Delegated to: Herbert Xu
Headers show
Series x86/crc32: improve crc32c_arch() code generation with clang | expand

Commit Message

Eric Biggers Feb. 10, 2025, 9:07 p.m. UTC 
From: Eric Biggers <ebiggers@google.com>

crc32c_arch() is affected by
https://github.com/llvm/llvm-project/issues/20571 where clang
unnecessarily spills the inputs to "rm"-constrained operands to the
stack.  Replace "rm" with ASM_INPUT_RM which partially works around this
by expanding to "r" when the compiler is clang.  This results in better
code generation with clang, though still not optimal.

Signed-off-by: Eric Biggers <ebiggers@google.com>
---

This applies to
https://git.kernel.org/pub/scm/linux/kernel/git/ebiggers/linux.git/log/?h=crc-next

 arch/x86/lib/crc32-glue.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)


base-commit: 4ffd50862d41e5aaf2e749efa354afaa1317c309

Comments

Eric Biggers Feb. 12, 2025, 8:10 p.m. UTC | #1 
On Mon, Feb 10, 2025 at 01:07:41PM -0800, Eric Biggers wrote:
> From: Eric Biggers <ebiggers@google.com>
> 
> crc32c_arch() is affected by
> https://github.com/llvm/llvm-project/issues/20571 where clang
> unnecessarily spills the inputs to "rm"-constrained operands to the
> stack.  Replace "rm" with ASM_INPUT_RM which partially works around this
> by expanding to "r" when the compiler is clang.  This results in better
> code generation with clang, though still not optimal.
> 
> Signed-off-by: Eric Biggers <ebiggers@google.com>
> ---
> 
> This applies to
> https://git.kernel.org/pub/scm/linux/kernel/git/ebiggers/linux.git/log/?h=crc-next

Applied to crc-next.

- Eric
diff mbox series

Patch

diff --git a/arch/x86/lib/crc32-glue.c b/arch/x86/lib/crc32-glue.c
index 9c3f9c1b7bb9..4b4721176799 100644
--- a/arch/x86/lib/crc32-glue.c
+++ b/arch/x86/lib/crc32-glue.c
@@ -53,14 +53,14 @@  u32 crc32c_arch(u32 crc, const u8 *p, size_t len)
 		return crc;
 	}
 
 	for (num_longs = len / sizeof(unsigned long);
 	     num_longs != 0; num_longs--, p += sizeof(unsigned long))
-		asm(CRC32_INST : "+r" (crc) : "rm" (*(unsigned long *)p));
+		asm(CRC32_INST : "+r" (crc) : ASM_INPUT_RM (*(unsigned long *)p));
 
 	for (len %= sizeof(unsigned long); len; len--, p++)
-		asm("crc32b %1, %0" : "+r" (crc) : "rm" (*p));
+		asm("crc32b %1, %0" : "+r" (crc) : ASM_INPUT_RM (*p));
 
 	return crc;
 }
 EXPORT_SYMBOL(crc32c_arch);