From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S965621Ab3DTC3l (ORCPT ); Fri, 19 Apr 2013 22:29:41 -0400 Received: from intranet.asianux.com ([58.214.24.6]:20347 "EHLO intranet.asianux.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S965552Ab3DTC3k (ORCPT ); Fri, 19 Apr 2013 22:29:40 -0400 X-Spam-Score: -100.8 Message-ID: <5171FD67.2080104@asianux.com> Date: Sat, 20 Apr 2013 10:28:55 +0800 From: Chen Gang User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:17.0) Gecko/20130110 Thunderbird/17.0.2 MIME-Version: 1.0 To: Arnd Bergmann CC: Catalin Marinas , Will Deacon , Tony Lindgren , "olof@lixom.net" , Santosh Shilimkar , "linux-arm-kernel@lists.infradead.org" , "linux-kernel@vger.kernel.org" Subject: Re: [Suggestion] ARM64: kernel: compiling issue, need implement cmpxchg64 with assembler language. References: <51712E5E.2020809@asianux.com> <201304191412.54601.arnd@arndb.de> In-Reply-To: <201304191412.54601.arnd@arndb.de> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 2013年04月19日 20:12, Arnd Bergmann wrote: > On Friday 19 April 2013, Chen Gang wrote: >> in arch/arm64/include/asm, not define the function cmpxchg64 >> >> when compiling with allmodconfig, >> drivers/block/blockconsole.c will need this function. >> >> I am not quite familiar with ARM64 (neither ARM64 assembler) >> >> can any member helps to send related patch ? >> if no one have time to send related patch, I should try. >> and I am glad to try, but need additional time resources, >> if I try, I should finish it within this month (2013-4-30). >> >> welcome any suggestions or completions. > > cmpxchg64 is the same as cmpxchg on 64-bit platforms, can't the > driver be changed to use the latter? > > Arnd > > can we be sure that cmpxchg64 is equal to cmpchg on all 64-bit platforms ? (I guess, the driver may need cross multiple platforms) (it seems, under x86, s390, better use cmpxchg64, at least they define it). whether we can be sure or not, I still prefer to define the macro cmpxchg64 just the alias of cmpxchg. (and I also guess ARM64 is always on 64-bit platform) the related patch may be like below. -------------------------------patch begin-------------------------------------- diff --git a/arch/arm64/include/asm/cmpxchg.h b/arch/arm64/include/asm/cmpxchg.h index 968b5cb..b572d2b 100644 --- a/arch/arm64/include/asm/cmpxchg.h +++ b/arch/arm64/include/asm/cmpxchg.h @@ -170,4 +170,6 @@ static inline unsigned long __cmpxchg_mb(volatile void *ptr, unsigned long old, (unsigned long)(n), \ sizeof(*(ptr)))) +#define cmpxchg64(ptr,o,n) cmpxchg((ptr),(o),(n)) + #endif /* __ASM_CMPXCHG_H */ -------------------------------patch end---------------------------------------- I think, we can also reference the implementation of s390: it is in arch/s390/include/asm/cmpxchg.h. since we are ARM64, excluding ARM(32,16...), we can only consider 64-bit. if in the future, ARM64 and ARM are merged together: we can use CONFIG_64BIT to switch the cmpxchg64 definition. if define CONFIG_64BIT, use cmpxchg instead of cmpxchg64. else, use the definition of ARM (arch/arm/include/asm/cmpxchg.h already defines cmpxchg64) -------------------------------reference begin---------------------------------- #ifdef CONFIG_64BIT #define cmpxchg64(ptr, o, n) \ ({ \ cmpxchg((ptr), (o), (n)); \ }) #else /* CONFIG_64BIT */ ... -------------------------------reference end------------------------------------ :-) -- Chen Gang Asianux Corporation