[Gta04-owner] dbus crashing with SIGBUS after suspend/resume

[NeilBrown]

On Tue, 2 Jul 2013 16:03:08 +0200 Radek Polak <psonek2 at seznam.cz> wrote:

> On Thursday, June 27, 2013 11:05:21 AM NeilBrown wrote:
> 
> > On Thu, 27 Jun 2013 10:28:35 +0200 Hermann Schwärzler
> > 
> > <hermann.schwaerzler at chello.at> wrote:
> > > Hello
> > > 
> > > On 06/26/2013 09:22 PM, Radek Polak wrote:
> > > > if you try to use latest QtMoko v55 you'll soon find it unusable for
> > > > daily phone. We have tracked it down (thanks Christ) to dbus crashing
> > > > with SIGBUS after a few suspend/resumes.
> > > 
> > > [...]
> > > 
> > > > GTA04A4 with QtMoko v55 (has latest Neil's 3.7 kernel compiled with
> > > > standard debian armhf gcc). The phone automatically suspends after
> > > > 40s.
> > > 
> > > [...]
> > > 
> > > > Few calls are ok. But then QtMoko stops working. The reason is that
> > > > dbus crashed after receiving SIGBUS and QtMoko cant work without dbus.
> > > 
> > > I can reproduce this (QtMoko v52 kernel 3.7.0-gta04-qtmoko-v52).
> > > I had my device up and running (well mostly suspended of course) for a
> > > few days and I see a difference in the output of dmesg between resumes
> > > that are caused by a call and those caused by the power-button:
> > > 
> > > At those resumes that are caused by a call (and NOT on those caused by
> > > the button) I see this additional line in dmesg:
> > > 
> > > Unhandled fault: imprecise external abort (0x1c06) at 0xbee30d60
> > 
> > I've seen that too.  I spent a while poking and come up with nothing.
> > 
> > My guess is some sort of hardwarey thing.  Maybe some delay during power up
> > is too short.
> > The Power Manage IC has register for sequencing power up/down with various
> > delays.  Maybe something needs to be tweaked there.
> 
> Hi,
> i tried to reproduce this "imprecise external abort" using rtcwake and it does 
> not happen. To me it seems that it only happens when the phone is woken by GSM 
> call. Maybe it has something to do with USB? E.g. i found this thread:
> 
> 	https://lkml.org/lkml/2011/7/20/357
> 
> Maybe it helps somehow...

Maybe it will.  I at least gives a hint as to what sort of thing might fix
the problem, and what sort of problem might be worth looking for.

I think that rtcwake does trigger it for me eventually, but my memory might
be a bit fuzzy.
I sometimes run my phone with a little program running that wakes it up every
5 minutes and records battery status.  I think this would eventually trigger
it.
I've been running it like that for a while with this:
https://bugs.launchpad.net/ubuntu/+source/linux-linaro/+bug/720055/comments/20
patch applied, and haven' seen the problem.  I'll remove it some time soon
and see if it comes back.

Friday is looks clear at the moment - I hope to try some of this out by then,
and try hard to get 3.10 close to working.

NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 828 bytes
Desc: not available
URL: <http://lists.goldelico.com/pipermail/gta04-owner/attachments/20130703/c2bfd684/attachment-0001.bin>