AF:
NF:0
PS:10
SRH:1
SFN:
DSR:
MID:<20071011110249.17710a6f@ripper.onstor.net>
CFG:
PT:0
S:andy.sharp@onstor.com
RQ:
SSV:onstor-exch02.onstor.net
NSV:
SSH:
R:<ralf@linux-mips.org>
MAID:1
X-Sylpheed-Privacy-System:
X-Sylpheed-Sign:0
SCF:#mh/Mailbox/sent
RMID:#imap/andys@onstor.net@onstor-exch02.onstor.net/INBOX	0	20071011173042.GA7498@linux-mips.org
X-Sylpheed-End-Special-Headers: 1
Date: Thu, 11 Oct 2007 11:04:10 -0700
From: Andrew Sharp <andy.sharp@onstor.com>
To: Ralf Baechle <ralf@linux-mips.org>
Subject: Re: paging problem with ide-cs driver
Message-ID: <20071011110410.268b85ba@ripper.onstor.net>
In-Reply-To: <20071011173042.GA7498@linux-mips.org>
References: <20071009132657.64ec9158@ripper.onstor.net>
	<20071009220530.0416792b@the-village.bc.nu>
	<20071010112550.GA1780@linux-mips.org>
	<20071010075041.60350e8a@ripper.onstor.net>
	<20071011173042.GA7498@linux-mips.org>
Organization: Onstor
X-Mailer: Sylpheed-Claws 2.6.0 (GTK+ 2.8.20; x86_64-pc-linux-gnu)
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

On Thu, 11 Oct 2007 18:30:42 +0100 Ralf Baechle <ralf@linux-mips.org>
wrote:

> On Wed, Oct 10, 2007 at 07:50:41AM -0700, Andrew Sharp wrote:
> 
> Did you make any progress?  I'm interesting in solving those issues,
> especially if they really turn out to be cache issues.

I'm hacking away on it, but I'm feeling less and less likely that it's
a cache issue, but I could be wrong.  NFS-root works fine, and the
network driver is doing DMA, which theoretically should be more at
risk wrt cache issues.  Unless the paging code does something clever
like load to unchached memory addresses so as not to mess up the
caches or something.  I'm thinking it's more likely something whacky
involving the pcmcia stack perhaps, as replacing the ide/ide-cs stack
with libata/pata-pcmcia still exhibits the problem.

One small thing is that this seems to be an 1125H, pass 3, which the
code base doesn't seem to understand (the code seems to think an 1125H
has to be a pass 2 part), so I hacked tlbx.c to not insert the pass2
workarounds, but that had no effect.

I'll let you know what I find out.

Cheers,

a
