From mboxrd@z Thu Jan  1 00:00:00 1970
From: Lukasz Majewski <l.majewski@samsung.com>
Date: Mon, 22 Aug 2011 09:29:52 +0200
Subject: [U-Boot] [PATCH] mmc:dcache: Cache line size aligned internal
 MMC buffers
In-Reply-To: <201108191135.52532.vapier@gentoo.org>
References: <1313745913-28672-1-git-send-email-l.majewski@samsung.com>
	<201108190957.14298.vapier@gentoo.org>
	<20110819172818.4a950320@lmajewski.digital.local>
	<201108191135.52532.vapier@gentoo.org>
Message-ID: <20110822092952.66a8e2cb@lmajewski.digital.local>
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 Mike,

On Fri, 19 Aug 2011 11:35:50 -0400
Mike Frysinger <vapier@gentoo.org> wrote:

> On Friday, August 19, 2011 11:28:18 Lukasz Majewski wrote:
> > On Fri, 19 Aug 2011 09:57:10 -0400 Mike Frysinger wrote:
> > > On Friday, August 19, 2011 05:25:13 Lukasz Majewski wrote:
> > > > +	cache_align_buf = memalign(get_dcache_line_size(),
> > > 
> > > nowhere do i see get_dcache_line_size() defined
> > 
> > Please look to the following post:
> > http://patchwork.ozlabs.org/patch/110501/
> > 
> > and another related with this issue:
> > http://patchwork.ozlabs.org/patch/110300/
> 
> if you're posting patches with dependencies, you need to mention them 
> explicitly (below the "---" area), or send proper patch series
> ([PATCH N/M]).
> 
> ignoring that, this patch will break all arches except arm.  that's
> bad mmmkay.  you probably need to move that weak def out of arm's
> cache.c and into like lib/cache.c.

Yes, I will prepare two patch series:
One addressing the get_dcache_line_size function for all u-boot
architectures. 

Another patch series will address changes to the drivers/mmc.c file. 

> > > also, what is the code size increase with your patch ?
> > 
> > Code size overhead (s5p_goni target):
> > Without proposed changes: 167928 B (u-boot.bin)
> > With changes: 168208 B (u-boot.bin)
> > 
> > Delta: 280 B
> 
> np if it gives significant (more than system noise) speedups.  any
> details on that ?
> -mike

No tests performed yet. The goal of those patches is to preserve the
MMC subsystem functionality when dcache is enabled (the ext_csd[512]
corruption is observed with d-cache enabled).


I'm particularly interested if the approach with memalign and
get_dcache_line_size is the preferred way to go.

I will think about some reliable ways to measure the MMC performance
with enabled and disabled MMC subsystem.


-- 
Best regards,

Lukasz Majewski

Samsung Poland R&D Center
Platform Group