Re: [Dxr3-devel] 0.17.4 is out

[Christian G. <chr...@gm...>]

2009/10/5 Nicolas Boullis <nbo...@de...>:
> Hi Christian,
>
> On Mon, Oct 05, 2009 at 09:15:06PM +0200, Christian Gmeiner wrote:
>> Hi Nicolas,
>>
>> doing only this:
>>
>>         case EM8300_IOCTL_SPU_SETPALETTE:
>>                 if (copy_from_user(clu, (void *) arg, 16 * 4))
>>                         return -EFAULT;
>>                 //em8300_spu_setpalette(em, clu);
>>                 break;
>>
>> results in the same wrong picture as in previous mail.
>
> That's a very interesting point. It means that the
> EM8300_IOCTL_SPU_SETPALETTE ioctl does nothing... But I still have no
> idea why... Does the dxr3 plugin check the return value of this ioctl?

With commit 1b9e3804075c2dc6f70beb2e9d14ecbb6f23710c I done a minute
ago it does check all ioctl syscalls.

>
>
>> doing only this;
>>
>> int em8300_spu_open(struct em8300_s *em)
>> {
>>         em->sp_ptsfifo_ptr = 0;
>>         em->sp_ptsvalid = 0;
>>         em->sp_mode = 1;
>>         //em8300_spu_setpalette(em, default_palette);
>>         write_ucregister(SP_Command, 0x2);
>>
>>         return 0;
>> }
>>
>> results in a new image:
>> http://picasaweb.google.com/christian.gmeiner/Devel#5389194589710723970
>>
>> and doing both results also in the new image.
>>
>>
>> Ohh and..... 32-bit kernel.
>
> Hmmm... Can you also check with both calls to em8300_spu_setpalette
> disabled, the same version of vdr and its dxr3 plugin, but a kernel <
> 2.6.31? And the same with only the first calls to em8300_spu_setpalette
> disabled?

So here are the results with the same plugin/vdr version and kernel 2.6.29:

If i disable both em8300_spu_setpalette calls I get an osd with the
colors like in
http://picasaweb.google.com/christian.gmeiner/Devel#5389194589710723970

If i disable the em8300_spu_setpalette call in em8300_spu_ioctl
function I get a good
looking OSD with different colors. This makes sense as the default
plaette colors are set.

If i enable all em8300_spu_setpalette calls I get a good looking OSD
with correct color palette
used.


>
> If you get the same results as with linux 2.6.31, I think that would
> confirm that the em8300_spu_setpalette function works fine, and that the
> problem is really related to the ioctl...

I dont know if its ioctl related, as on 2.6.31 the default color
palette gets not set. So there
might be a problem with register writes.

Hope I could help you
-- 
Christian Gmeiner, MSc

PS: Sorry for the long delay... I am very busy with vdr-plugin-dxr3 at
the moment.