linux-kernel - Re: [PATCH] x86: drop REG_OUT macro from hweight functions

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20190729094329.GW31381@hirez.programming.kicks-ass.net>
Date:   Mon, 29 Jul 2019 11:43:29 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Alexey Dobriyan <adobriyan@...il.com>
Cc:     tglx@...utronix.de, mingo@...hat.com, bp@...en8.de, hpa@...or.com,
        x86@...nel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] x86: drop REG_OUT macro from hweight functions

On Sun, Jul 28, 2019 at 02:51:40PM +0300, Alexey Dobriyan wrote:
> Output register is always RAX/EAX.
> 
> Signed-off-by: Alexey Dobriyan <adobriyan@...il.com>
> ---
> 
>  arch/x86/include/asm/arch_hweight.h |    6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
> 
> --- a/arch/x86/include/asm/arch_hweight.h
> +++ b/arch/x86/include/asm/arch_hweight.h
> @@ -6,10 +6,8 @@
>  
>  #ifdef CONFIG_64BIT
>  #define REG_IN "D"
> -#define REG_OUT "a"
>  #else
>  #define REG_IN "a"
> -#define REG_OUT "a"
>  #endif
>  
>  static __always_inline unsigned int __arch_hweight32(unsigned int w)
> @@ -17,7 +15,7 @@ static __always_inline unsigned int __arch_hweight32(unsigned int w)
>  	unsigned int res;
>  
>  	asm (ALTERNATIVE("call __sw_hweight32", "popcntl %1, %0", X86_FEATURE_POPCNT)
> -			 : "="REG_OUT (res)
> +			 : "=a" (res)
>  			 : REG_IN (w));
>  
>  	return res;
> @@ -45,7 +43,7 @@ static __always_inline unsigned long __arch_hweight64(__u64 w)
>  	unsigned long res;
>  
>  	asm (ALTERNATIVE("call __sw_hweight64", "popcntq %1, %0", X86_FEATURE_POPCNT)
> -			 : "="REG_OUT (res)
> +			 : "=a" (res)
>  			 : REG_IN (w));
>  
>  	return res;

I _think_ something like the below should also work:

(fwiw _ASM_ARG 5 and 6 are broken, as are all the QLWB variants)

---
diff --git a/arch/x86/include/asm/arch_hweight.h b/arch/x86/include/asm/arch_hweight.h
index ba88edd0d58b..88704612b2f7 100644
--- a/arch/x86/include/asm/arch_hweight.h
+++ b/arch/x86/include/asm/arch_hweight.h
@@ -3,22 +3,15 @@
 #define _ASM_X86_HWEIGHT_H
 
 #include <asm/cpufeatures.h>
-
-#ifdef CONFIG_64BIT
-#define REG_IN "D"
-#define REG_OUT "a"
-#else
-#define REG_IN "a"
-#define REG_OUT "a"
-#endif
+#include <asm/asm.h>
 
 static __always_inline unsigned int __arch_hweight32(unsigned int w)
 {
 	unsigned int res;
 
 	asm (ALTERNATIVE("call __sw_hweight32", "popcntl %1, %0", X86_FEATURE_POPCNT)
-			 : "="REG_OUT (res)
-			 : REG_IN (w));
+			 : "=a" (res)
+			 : _ASM_ARG1 (w));
 
 	return res;
 }
@@ -45,8 +38,8 @@ static __always_inline unsigned long __arch_hweight64(__u64 w)
 	unsigned long res;
 
 	asm (ALTERNATIVE("call __sw_hweight64", "popcntq %1, %0", X86_FEATURE_POPCNT)
-			 : "="REG_OUT (res)
-			 : REG_IN (w));
+			 : "=a" (res)
+			 : _ASM_ARG1 (w));
 
 	return res;
 }