From mboxrd@z Thu Jan  1 00:00:00 1970
From: Nate Lawson <nate-Y6VGUYTwhu0@public.gmane.org>
Subject: RE: ACPI global lock macros
Date: Thu, 11 Dec 2003 09:46:30 -0800 (PST)
Sender: acpi-devel-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
Message-ID: <20031211094510.J50052@root.org>
References: <3ACA40606221794F80A5670F0AF15F8401720C22@PDSMSX403.ccr.corp.intel.com>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Return-path: <acpi-devel-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org>
In-Reply-To: <3ACA40606221794F80A5670F0AF15F8401720C22-SRlDPOYGfgogGBtAFL8yw7fspsVTdybXVpNB7YpNyf8@public.gmane.org>
Errors-To: acpi-devel-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
List-Unsubscribe: <https://lists.sourceforge.net/lists/listinfo/acpi-devel>,
	<mailto:acpi-devel-request-5NWGOfrQmneRv+LV9MX5utDmRdBvX5/5ji9XKYC2TYK4Io4DoRGLOrNAH6kLmebB@public.gmane.org>
List-Post: <mailto:acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org>
List-Help: <mailto:acpi-devel-request-5NWGOfrQmneRv+LV9MX5utDmRdBvX5/5lLLH9aP38KF1pqr+6INuaA@public.gmane.org>
List-Subscribe: <https://lists.sourceforge.net/lists/listinfo/acpi-devel>,
	<mailto:acpi-devel-request-5NWGOfrQmneRv+LV9MX5utDmRdBvX5/5R9m/fhtKvoLKtGCp0j7rLg@public.gmane.org>
List-Archive: <http://sourceforge.net/mailarchive/forum.php?forum=acpi-devel>
To: "Yu, Luming" <luming.yu-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Cc: Paul Menage <menage-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>, agrover-qb8aLOKklSjp4P8CbLYnNQ@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
List-Id: linux-acpi@vger.kernel.org

On Thu, 11 Dec 2003, Yu, Luming wrote:
> -----Original Message-----
> From: acpi-devel-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org [mailto:acpi-devel-admin@lists.sourceforge.net]On Behalf Of Yu, Luming
> Sent: 2003?12?11? 15:06
> To: Paul Menage; agrover-qb8aLOKklSjp4P8CbLYnNQ@public.gmane.org
> Cc: linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org; acpi-devel@lists.sourceforge.net
> Subject: RE: [ACPI] ACPI global lock macros
>
>
> >>#define ACPI_ACQUIRE_GLOBAL_LOCK(GLptr, Acq) \
> >>     do { \
> >>        asm volatile("1:movl   (%1),%%eax;" \
> >>             "movl   %%eax,%%edx;" \
> >>             "andl   %2,%%edx;" \
> >>             "btsl   $0x1,%%edx;" \
> >>             "adcl   $0x0,%%edx;" \
> >>             "lock;  cmpxchgl %%edx,(%1);" \
> >>             "jnz    1b;" \
> >>             "cmpb   $0x3,%%dl;" \
> >>             "sbbl   %0,%0" \
> >>             :"=r"(Acq):"r"(GLptr),"i"(~1L):"dx", "ax"); \
> >>     } while(0)
>
> Above code have a bug! Considering below code:
>
> u8	acquired = FALSE;
>
> ACPI_ACQUIRE_GLOBAL_LOC(acpi_gbl_common_fACS.global_lock, acquired);
> if(acquired) {
> ....
> }
>
> Gcc will complain " ERROR: '%cl' not allowed with sbbl ". And I think any other compiler will
> complain that  too !
>
> How about  below changes to your proposal code.
>
> <             "sbbl   %0,%0" \
> <             :"=r"(Acq):"r"(GLptr),"i"(~1L):"dx","ax"); \
> ---
> >             "sbbl   %%eax,%%eax" \
> >             :"=a"(Acq):"r"(GLptr),"i"(~1L):"dx"); \

FYI, that's what we do in FreeBSD also.  The only difference after your
patch is that we use +m for GLptr.

#define ACPI_ACQUIRE_GLOBAL_LOCK(GLptr, Acq) \
    do { \
        asm("1:     movl %1,%%eax;" \
            "movl   %%eax,%%edx;" \
            "andl   %2,%%edx;" \
            "btsl   $0x1,%%edx;" \
            "adcl   $0x0,%%edx;" \
            "lock;  cmpxchgl %%edx,%1;" \
            "jnz    1b;" \
            "cmpb   $0x3,%%dl;" \
            "sbbl   %%eax,%%eax" \
            : "=a" (Acq), "+m" (GLptr) : "i" (~1L) : "edx"); \
    } while(0)

-Nate


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/