rcar-h2 du issue between du0 and du2

rcar-h2 du issue between du0 and du2

[Ben Dooks]

I am using the 3.4 renesas backport (0.5.1) and having issue with the
rcar_du driver with any du0 and du2 having seperate operating modes.

du2 seems to be picking up the data width of du0, so if du0 is at
1920x1080 and du2 at 800x600, du2 will show a display as if du2 data
was at a 1920 stride.

If both du0 and du2 are at 800x600 then both show the right display
data,

-- 
Ben Dooks				http://www.codethink.co.uk/
Senior Engineer				Codethink - Providing Genius

Re: rcar-h2 du issue between du0 and du2

[Laurent Pinchart]

Hi Ben,

On Tuesday 12 November 2013 16:04:38 Ben Dooks wrote:
> I am using the 3.4 renesas backport (0.5.1) and having issue with the
> rcar_du driver with any du0 and du2 having seperate operating modes.
> 
> du2 seems to be picking up the data width of du0, so if du0 is at
> 1920x1080 and du2 at 800x600, du2 will show a display as if du2 data
> was at a 1920 stride.
> 
> If both du0 and du2 are at 800x600 then both show the right display
> data,

Ouch. Thank you for the report. I'll test and debug that. What's the urgency 
level ?

-- 
Regards,

Laurent Pinchart

Re: rcar-h2 du issue between du0 and du2

[Ben Dooks]

On 13/11/13 02:44, Laurent Pinchart wrote:
> Hi Ben,
>
> On Tuesday 12 November 2013 16:04:38 Ben Dooks wrote:
>> I am using the 3.4 renesas backport (0.5.1) and having issue with the
>> rcar_du driver with any du0 and du2 having seperate operating modes.
>>
>> du2 seems to be picking up the data width of du0, so if du0 is at
>> 1920x1080 and du2 at 800x600, du2 will show a display as if du2 data
>> was at a 1920 stride.
>>
>> If both du0 and du2 are at 800x600 then both show the right display
>> data,
>
> Ouch. Thank you for the report. I'll test and debug that. What's the urgency
> level ?

Medium, there's other stuff that I can be getting on with.

I've been looking through the H2 manual but there's nothing specifically
saying the DU0 and DU2 planes should be linked. I know DU0_0 and DU0_1
are basically linked.

We are using the Lager board as following:

We have DU0 outputting LVDS to the ADV7511 to provide DVI
DU2 is connected to the DPAD0 to the VGA

DU0 is configured for 1920x1080-60, 32bpp
DU2 is configured for 800x600-60, 32bpp

DU2 has been verified as outputting 800x600-60 display timings.


-- 
Ben Dooks				http://www.codethink.co.uk/
Senior Engineer				Codethink - Providing Genius

Re: rcar-h2 du issue between du0 and du2

[Ben Dooks]

On 13/11/13 02:44, Laurent Pinchart wrote:
> Hi Ben,
>
> On Tuesday 12 November 2013 16:04:38 Ben Dooks wrote:
>> I am using the 3.4 renesas backport (0.5.1) and having issue with the
>> rcar_du driver with any du0 and du2 having seperate operating modes.
>>
>> du2 seems to be picking up the data width of du0, so if du0 is at
>> 1920x1080 and du2 at 800x600, du2 will show a display as if du2 data
>> was at a 1920 stride.
>>
>> If both du0 and du2 are at 800x600 then both show the right display
>> data,
>
> Ouch. Thank you for the report. I'll test and debug that. What's the urgency
> level ?

I've found the issue, it was with a change we did to the kernel to split
the displays and accidentally sent the same block of memory to both du 
units. Sorry for the inconvenience, it is working now.


-- 
Ben Dooks				http://www.codethink.co.uk/
Senior Engineer				Codethink - Providing Genius

Re: rcar-h2 du issue between du0 and du2

[Laurent Pinchart]

Hi Ben,

On Thursday 14 November 2013 12:52:06 Ben Dooks wrote:
> On 13/11/13 02:44, Laurent Pinchart wrote:
> > On Tuesday 12 November 2013 16:04:38 Ben Dooks wrote:
> >> I am using the 3.4 renesas backport (0.5.1) and having issue with the
> >> rcar_du driver with any du0 and du2 having seperate operating modes.
> >> 
> >> du2 seems to be picking up the data width of du0, so if du0 is at
> >> 1920x1080 and du2 at 800x600, du2 will show a display as if du2 data
> >> was at a 1920 stride.
> >> 
> >> If both du0 and du2 are at 800x600 then both show the right display
> >> data,
> > 
> > Ouch. Thank you for the report. I'll test and debug that. What's the
> > urgency level ?
> 
> I've found the issue, it was with a change we did to the kernel to split
> the displays and accidentally sent the same block of memory to both du
> units.

Good, that means less work for me then :-)

> Sorry for the inconvenience, it is working now.

No worries.

-- 
Regards,

Laurent Pinchart