From mboxrd@z Thu Jan  1 00:00:00 1970
From: eric@eukrea.com (=?UTF-8?B?RXJpYyBCw6luYXJk?=)
Date: Mon, 04 Oct 2010 10:08:44 +0200
Subject: I.MX35 GPIO IRQ + Preempt -> Oops
In-Reply-To: <20101004073920.GM28679@pengutronix.de>
References: <4CA739B4.1020506@eukrea.com>
	<20101003114103.GC32736@n2100.arm.linux.org.uk>
	<4CA8A06B.3070103@eukrea.com>
	<20101003162021.GD32736@n2100.arm.linux.org.uk>
	<4CA8BA29.5060600@eukrea.com>
	<20101004073920.GM28679@pengutronix.de>
Message-ID: <4CA98B8C.4010106@eukrea.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

Hi Uwe,

Le 04/10/2010 09:39, Uwe Kleine-K?nig a ?crit :
> On Sun, Oct 03, 2010 at 07:15:21PM +0200, Eric B?nard wrote:
>> Le 03/10/2010 18:20, Russell King - ARM Linux a ?crit :
>> the FIQ is used for the imx-ssi driver.
>>
>> It seems that removing the cpu_do_idle() call in
>> plat-mxc/include/mach/system.h "fix" the problem.
> hmmm
>
> Does deselecting imx-ssi "fixes" the problem, too?
>
not tested yet.

> Do you have traffic on the ssi unit when the problem occurs?
>
no but looking at the errata (and the i.MX31's one which is very 
similar) it's not clearly mentionned that the FIQ has to be used in 
order to trigger the problem (they talk of FIQ/IRQ).

Eric