From mboxrd@z Thu Jan  1 00:00:00 1970
From: Stefan Roese <sr@denx.de>
Date: Sat, 10 Feb 2007 09:07:36 +0100
Subject: [U-Boot-Users] [PATCH0/2] Re-do the patch for adding DO_SYNC in
	flash_write_cmd
In-Reply-To: <45CD7AF0.2020209@orkun.us>
References: <20070210010420.CD5CB353CAA@atlas.denx.de>
	<200702100840.46587.sr@denx.de> <45CD7AF0.2020209@orkun.us>
Message-ID: <200702100907.37451.sr@denx.de>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: u-boot@lists.denx.de

Hi Tolunay,

On Saturday 10 February 2007 08:57, Tolunay Orkun wrote:
> > I would not do this. Please let a "sync" instruction _not_ do a "isync"
> > too.
> >
>  > There will be times when you explicitly _don't_ what this.
>
> Could you give a specific example.

The "isync" will reload the TLB's on 440's for example and sometimes you 
really don't want this on a "normal" sync instruction.

> There is also "eieio" and "msync" to consider though these usually map
> to former two (or vice versa).

Yes. But I find the name "sync" more platform generic.

> > Why not use
> >
> > #define sync()  __asm__ __volatile__ ("sync" : : : "memory");
> >
> > from include/asm-ppc/io.h? This seems to be exactly what we need.
>
> I would rather prefer an uppercase SYNC since it is a macro but whatever
> style you guys choose is OK with me.

Could be that on other platforms this sync is not a define but a function. But 
I don't have a strong preference here. Wolfgang, you decide. ;-)

Best regards,
Stefan

=====================================================================
DENX Software Engineering GmbH, HRB 165235 Munich, CEO: Wolfgang Denk
Office:  Kirchenstr. 5,       D-82194 Groebenzell,            Germany
=====================================================================