DSS and HDMI kernel panic on OMAP4

DSS and HDMI kernel panic on OMAP4

[Joachim Eastwood]

Hello Tomi,

I wanted to test my Variscite patches after Tony merged them into his
3.16 dt branch so created a base branch from Linus master and pull in
Tony's 3.16 dt and your dss for-next branch.

I discovered that booting with a HDMI monitor connected or plugging in
one cause a kernel panic. See the log at:
http://slexy.org/raw/s20xo68UPx

Any idea what it could be?

I also noted this in the boot log:
[ 0.984527] omapdss_hdmi 58006000.encoder: can't request region for
resource [mem 0x58006400-0x580073ff]

I don't think I have seen this before.

What I previously have tested have been Linus master + my Variscite
patches and your panel-dpi dt + hdmi hpd patches cherry-picked.

regards
Joachim Eastwood

Re: DSS and HDMI kernel panic on OMAP4

[Joachim Eastwood]

On 15 May 2014 23:14, Joachim  Eastwood <manabian@gmail.com> wrote:
> Hello Tomi,
>
> I wanted to test my Variscite patches after Tony merged them into his
> 3.16 dt branch so created a base branch from Linus master and pull in
> Tony's 3.16 dt and your dss for-next branch.
>
> I discovered that booting with a HDMI monitor connected or plugging in
> one cause a kernel panic. See the log at:
> http://slexy.org/raw/s20xo68UPx
>
> Any idea what it could be?
>
> I also noted this in the boot log:
> [ 0.984527] omapdss_hdmi 58006000.encoder: can't request region for
> resource [mem 0x58006400-0x580073ff]
>
> I don't think I have seen this before.
>
> What I previously have tested have been Linus master + my Variscite
> patches and your panel-dpi dt + hdmi hpd patches cherry-picked.

Reverting "OMAPDSS: HDMI: cleanup ioremaps" 59b3d38a3691396783df108
from your for-next branch fixes the problem.

regards
Joachim Eastwood

Re: DSS and HDMI kernel panic on OMAP4

[Tomi Valkeinen]

[-- Attachment #1: Type: text/plain, Size: 1631 bytes --]

On 16/05/14 00:26, Joachim Eastwood wrote:
> On 15 May 2014 23:14, Joachim  Eastwood <manabian@gmail.com> wrote:
>> Hello Tomi,
>>
>> I wanted to test my Variscite patches after Tony merged them into his
>> 3.16 dt branch so created a base branch from Linus master and pull in
>> Tony's 3.16 dt and your dss for-next branch.
>>
>> I discovered that booting with a HDMI monitor connected or plugging in
>> one cause a kernel panic. See the log at:
>> http://slexy.org/raw/s20xo68UPx
>>
>> Any idea what it could be?
>>
>> I also noted this in the boot log:
>> [ 0.984527] omapdss_hdmi 58006000.encoder: can't request region for
>> resource [mem 0x58006400-0x580073ff]
>>
>> I don't think I have seen this before.
>>
>> What I previously have tested have been Linus master + my Variscite
>> patches and your panel-dpi dt + hdmi hpd patches cherry-picked.
> 
> Reverting "OMAPDSS: HDMI: cleanup ioremaps" 59b3d38a3691396783df108
> from your for-next branch fixes the problem.

I think that's somehow HDMI audio enabled. I bet if you disable HDMI
audio, the error goes away.

If I cat /proc/iomem, I see:

58006000-58006fff : omap-hdmi-audio-dai
  58006200-580062ff : pll
  58006300-580063ff : phy

Without sound support, I don't get the error and I see:

58006200-580062ff : pll
58006300-580063ff : phy
58006400-580073ff : core

Reverting 59b3d38a3691396783df108e6afbba30656edccb helps, because before
that patch the hdmi driver didn't actually reserve the regions.

So, I don't really have any idea yet what's going on there, but the
problem is somewhere around hdmi audio.

 Tomi



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: DSS and HDMI kernel panic on OMAP4

[Joachim Eastwood]

On 16 May 2014 07:24, Tomi Valkeinen <tomi.valkeinen@ti.com> wrote:
> On 16/05/14 00:26, Joachim Eastwood wrote:
>> On 15 May 2014 23:14, Joachim  Eastwood <manabian@gmail.com> wrote:
>>> Hello Tomi,
>>>
>>> I wanted to test my Variscite patches after Tony merged them into his
>>> 3.16 dt branch so created a base branch from Linus master and pull in
>>> Tony's 3.16 dt and your dss for-next branch.
>>>
>>> I discovered that booting with a HDMI monitor connected or plugging in
>>> one cause a kernel panic. See the log at:
>>> http://slexy.org/raw/s20xo68UPx
>>>
>>> Any idea what it could be?
>>>
>>> I also noted this in the boot log:
>>> [ 0.984527] omapdss_hdmi 58006000.encoder: can't request region for
>>> resource [mem 0x58006400-0x580073ff]
>>>
>>> I don't think I have seen this before.
>>>
>>> What I previously have tested have been Linus master + my Variscite
>>> patches and your panel-dpi dt + hdmi hpd patches cherry-picked.
>>
>> Reverting "OMAPDSS: HDMI: cleanup ioremaps" 59b3d38a3691396783df108
>> from your for-next branch fixes the problem.
>
> I think that's somehow HDMI audio enabled. I bet if you disable HDMI
> audio, the error goes away.

Yes, disabling HDMI audio config option makes the panic disappear. But
since nothing prevents you from turning on this option this is not a
solution.

> If I cat /proc/iomem, I see:
>
> 58006000-58006fff : omap-hdmi-audio-dai
>   58006200-580062ff : pll
>   58006300-580063ff : phy
>
> Without sound support, I don't get the error and I see:
>
> 58006200-580062ff : pll
> 58006300-580063ff : phy
> 58006400-580073ff : core

With the patch 59b3d38a36913 reverted I actually have 2
"omap-hdmi-audio-dai" entries in iomem:
48046000-48046fff : omap-hdmi-audio-dai
58006000-58006fff : omap-hdmi-audio-dai

The first one disappear after patch 59b3d38a36913 is applied.

> Reverting 59b3d38a3691396783df108e6afbba30656edccb helps, because before
> that patch the hdmi driver didn't actually reserve the regions.
>
> So, I don't really have any idea yet what's going on there, but the
> problem is somewhere around hdmi audio.

I see. Would be nice have this fixed before the stuff hits mainline.

regards
Joachim Eastwood