[Letux-kernel] weird problem with pwm_bl on omap3

[H. Nikolaus Schaller]

Hi Andreas,

> Am 29.05.2018 um 18:27 schrieb Andreas Kemnade <andreas at kemnade.info>:
> 
> On Tue, 29 May 2018 17:41:53 +0200
> "H. Nikolaus Schaller" <hns at goldelico.com> wrote:
> 
>> Hi Tony,
>> 
>>> Am 17.05.2018 um 20:08 schrieb Tony Lindgren <tony at atomide.com>:
>>> 
>>> * H. Nikolaus Schaller <hns at goldelico.com> [180517 12:17]:
>>>> Hi Tony,
>>>> we are using for long time the dmtimer/pwm on
>>>> the GTA04 to drive the display panel backlight.
>>>> 
>>>> Starting a while ago (I am not sure when, but it may
>>>> be 4.17-rc1), the device randomly fails to boot
>>>> with a NULL pointer dereference in strcmp().
>>>> Booting again usually runs fine.
>>> 
>>> Hmm maybe enable CONFIG_DEBUG_SLAB=y and POISON options
>>> and see if that catches something. It's might be some
>>> array out of bounds type issue.
>>> 
>>> Regards,
>>> 
>>> Tony
>> 
>> 
>> I have restarted hunting the issue but it is very
>> ghostly. Every time I try to test by another method
>> it disappears and if I remove my printk things or
>> /etc/modprobe/blacklist.conf it comes back.
>> 
> Have you tried to load every module piece by piece before running udev
> by booting the kernel with a init=/something.sh script?
> So you have the order under control?

Well, if I load modules manually everything is fine.
The problem appears to be the concurrency of loading
and deferred probing of modules.

If I block pwm_bl and omap3dss and some others and load
them all manually, there was never a problem :(

> 
> Another thing: Are our fb patches evil again?

It also occurs with omapdss blacklisted.

> 
> The offmode patch? Maybe it uncovers some other bug.

Hm. I am not sure if I have included it. I am running with
mainline with just a handful of our recent patches (not
everything).

But I can try it as soon as I find time (kernel compiling has
got much slower in recent years so that I have new ideas
faster than a test result :).

> 
> Just to sort out some non-mainline stuff
> 
> I am compiling and will try to boot with my scripts this evening.

Yes, please. At least to check if you can observe the same issue.
It seems to happen more often on a GTA04A5 that a GTA04A4.

And the strcmp(NULL, "backlight_pins_pinmux") occurred
first in 4.17-rc1. But it might just be a more visible
symptom and the bug may be older.

BR,
Nikolaus

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://lists.goldelico.com/pipermail/letux-kernel/attachments/20180529/b4d053b2/attachment.asc>