[Letux-kernel] OMAP5 Pyra: DSI Panel problems with 4.12-rc1

H. Nikolaus Schaller hns at goldelico.com
Tue May 23 12:13:50 CEST 2017


Hi Tomi,

> Am 23.05.2017 um 11:19 schrieb Tomi Valkeinen <tomi.valkeinen at ti.com>:
> 
> On 22/05/17 18:23, H. Nikolaus Schaller wrote:
>> Hi Tomi,
>> 
>>> Am 22.05.2017 um 08:31 schrieb Tomi Valkeinen <tomi.valkeinen at ti.com>:
>>> 
>>> Hi,
>>> 
>>> On 20/05/17 14:30, H. Nikolaus Schaller wrote:
>>>> Hi Tomi,
>>>> 
>>>> after rebasing our Letux kernel to 4.12-rc1 our DSI panel stopped initializing.
>>>> DRM and X11 are up and running, but the panel fails to initialize and show anything.
>>>> 
>>>> We have not touched the code of our btl507212-w677l DSI driver.
>>>> 
>>>> Is there anything to consider on panel driver side?
>>>> 
>>>> What I can't 100% exclude is that we have additional hacks in the omapdss system
>>>> which may now be in conflict. For example we have your "[PATCH] dispc hackfix"
>>>> or "dss: remove export of venc modes".
>>> 
>>> Can you try reverting 7d267f068a8b4944d52e8b0ae4c8fcc1c1c5c5ba
>> 
>> Yes, a revert makes it working again!
>> 
>> The interesting question is why it makes such a difference...
>> 
>> Are there additional dividers which depend on M and N? For example the
>> DSI LP clock? This could explain why a DPI panel (GTA04) simply works as before
>> and there is only a DSI communication problem during DSI panel initialization.
> 
> I digged up my omap5 uevm, which has the dsi panel you sent me long time
> back. After forward porting the related patches, for some reason, I'm
> getting dispc underflows with it (after I revert the patch above). But I
> still get a semi-valid picture, and can do page flips at ~60fps.
> 
> After applying the patch, fps drops into half. I don't see anything
> wrong with the PLL calcs. I was doing testing with some other dividers,
> and, well... results vary. Changing the fint_min to 0.5MHz in dsi.c
> makes it work, but lower limits show different problems. TRM has
> conflicting limits for the PLL's internal frequencies, so I don't know
> the correct limits.
> 
> I'll look at this a bit more when I find time.

Ok, no problem. Take the time it needs to fix.

For the moment we can live with the revert-hack.

BR and thanks,
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/20170523/10fb550c/attachment.asc>


More information about the Letux-kernel mailing list