2018-10-14 20:25:52

by Uros Bizjak

Subject: [PATCH v2] x86: Use assembly instruction mnemonics instead of .byte streams in arch_hweight.h

Recently the minimum required version of binutils was changed to 2.20,
which supports popcnt instruction mnemonics. The patch removes
corresponding #defines that implement popcnt instructions using target-dependant
.byte streams and uses unified instruction mnemonics instead.

Tested by building x86_64 and i386 version of the kernel and comparing
objdump dumps of the pathced and unpatched vmlinux.o. They were the same.

v2: Update commit message to emphasize removal of #defines.

Signed-off-by: Uros Bizjak <[email protected]>
---
arch/x86/include/asm/arch_hweight.h | 12 ++++--------
1 file changed, 4 insertions(+), 8 deletions(-)

diff --git a/arch/x86/include/asm/arch_hweight.h b/arch/x86/include/asm/arch_hweight.h
index 34a10b2d5b73..d668e411bd32 100644
--- a/arch/x86/include/asm/arch_hweight.h
+++ b/arch/x86/include/asm/arch_hweight.h
@@ -5,15 +5,9 @@
#include <asm/cpufeatures.h>

#ifdef CONFIG_64BIT
-/* popcnt %edi, %eax */
-#define POPCNT32 ".byte 0xf3,0x0f,0xb8,0xc7"
-/* popcnt %rdi, %rax */
-#define POPCNT64 ".byte 0xf3,0x48,0x0f,0xb8,0xc7"
#define REG_IN "D"
#define REG_OUT "a"
#else
-/* popcnt %eax, %eax */
-#define POPCNT32 ".byte 0xf3,0x0f,0xb8,0xc0"
#define REG_IN "a"
#define REG_OUT "a"
#endif
@@ -24,7 +18,8 @@ static __always_inline unsigned int __arch_hweight32(unsigned int w)
{
unsigned int res;

- asm (ALTERNATIVE("call __sw_hweight32", POPCNT32, X86_FEATURE_POPCNT)
+ asm (ALTERNATIVE("call __sw_hweight32",
+ "popcntl %1, %0", X86_FEATURE_POPCNT)
: "="REG_OUT (res)
: REG_IN (w));

@@ -52,7 +47,8 @@ static __always_inline unsigned long __arch_hweight64(__u64 w)
{
unsigned long res;

- asm (ALTERNATIVE("call __sw_hweight64", POPCNT64, X86_FEATURE_POPCNT)
+ asm (ALTERNATIVE("call __sw_hweight64",
+ "popcntq %1, %0", X86_FEATURE_POPCNT)
: "="REG_OUT (res)
: REG_IN (w));

--
2.17.2

2018-10-14 20:31:01

by Borislav Petkov

Subject: Re: [PATCH v2] x86: Use assembly instruction mnemonics instead of .byte streams in arch_hweight.h

On Sun, Oct 14, 2018 at 10:23:54PM +0200, Uros Bizjak wrote:
> Recently the minimum required version of binutils was changed to 2.20,
> which supports popcnt instruction mnemonics. The patch removes
> corresponding #defines that implement popcnt instructions using target-dependant
> .byte streams and uses unified instruction mnemonics instead.
>
> Tested by building x86_64 and i386 version of the kernel and comparing
> objdump dumps of the pathced and unpatched vmlinux.o. They were the same.
>
> v2: Update commit message to emphasize removal of #defines.
>
> Signed-off-by: Uros Bizjak <[email protected]>
> ---
> arch/x86/include/asm/arch_hweight.h | 12 ++++--------
> 1 file changed, 4 insertions(+), 8 deletions(-)

Thanks.

But with one week before the merge window and this being not nearly
urgent, it'll have to wait for after the merge window.

--
Regards/Gruss,
Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.

2018-11-05 09:56:11

by tip-bot for Vasyl Gomonovych

Subject: [tip:x86/asm] x86: Use POPCNT mnemonics in arch_hweight.h

Commit-ID: 566b62a3676caed1f8c626e26066ff3ae3a38918
Gitweb: https://git.kernel.org/tip/566b62a3676caed1f8c626e26066ff3ae3a38918
Author: Uros Bizjak <[email protected]>
AuthorDate: Sun, 14 Oct 2018 22:23:54 +0200
Committer: Borislav Petkov <[email protected]>
CommitDate: Mon, 5 Nov 2018 10:42:32 +0100

x86: Use POPCNT mnemonics in arch_hweight.h

Recently, the minimum required version of binutils was changed to
2.20, which supports POPCNT instruction mnemonics.

Replace the byte-wise specification of POPCNT with those proper
mnemonics.

[ bp: massage commit message and remove line breaks. ]

Signed-off-by: Uros Bizjak <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
CC: "H. Peter Anvin" <[email protected]>
CC: Ingo Molnar <[email protected]>
CC: Thomas Gleixner <[email protected]>
Link: http://lkml.kernel.org/r/[email protected]
---
arch/x86/include/asm/arch_hweight.h | 10 ++--------
1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/arch/x86/include/asm/arch_hweight.h b/arch/x86/include/asm/arch_hweight.h
index 34a10b2d5b73..fc0693569f7a 100644
--- a/arch/x86/include/asm/arch_hweight.h
+++ b/arch/x86/include/asm/arch_hweight.h
@@ -5,15 +5,9 @@
#include <asm/cpufeatures.h>

#ifdef CONFIG_64BIT
-/* popcnt %edi, %eax */
-#define POPCNT32 ".byte 0xf3,0x0f,0xb8,0xc7"
-/* popcnt %rdi, %rax */
-#define POPCNT64 ".byte 0xf3,0x48,0x0f,0xb8,0xc7"
#define REG_IN "D"
#define REG_OUT "a"
#else
-/* popcnt %eax, %eax */
-#define POPCNT32 ".byte 0xf3,0x0f,0xb8,0xc0"
#define REG_IN "a"
#define REG_OUT "a"
#endif
@@ -24,7 +18,7 @@ static __always_inline unsigned int __arch_hweight32(unsigned int w)
{
unsigned int res;

- asm (ALTERNATIVE("call __sw_hweight32", POPCNT32, X86_FEATURE_POPCNT)
+ asm (ALTERNATIVE("call __sw_hweight32", "popcntl %1, %0", X86_FEATURE_POPCNT)
: "="REG_OUT (res)
: REG_IN (w));

@@ -52,7 +46,7 @@ static __always_inline unsigned long __arch_hweight64(__u64 w)
{
unsigned long res;

- asm (ALTERNATIVE("call __sw_hweight64", POPCNT64, X86_FEATURE_POPCNT)
+ asm (ALTERNATIVE("call __sw_hweight64", "popcntq %1, %0", X86_FEATURE_POPCNT)
: "="REG_OUT (res)
: REG_IN (w));