Re: Problem with em28xx driver in Gumstix Overo

Re: Problem with em28xx driver in Gumstix Overo

[Linus Torvalds]

Hi Marcos,

[ full email quoted for new cc's ]

you really would be much better off sending a diff rather than trying
to explain what changes you did, but quite frankly, even if you did
that, I'd still want this to go through the actual maintainers of the
em28xx driver.

The USB error sounds like some independent unrelated issue, I have no
ideas on it.

                   Linus

On Mon, Jan 3, 2011 at 5:14 AM, Marcos Alejandro Saldivia Delgado
<marcos.saldivia@gmail.com> wrote:
>
> Hi Linus .
>
> My name is Marcos Saldivia, and I have a problem with the gumstix Overo
> board when trying to capture video with a usb card  that it uses the em28xx
> driver.
> I tried to connect this capture card, it recognized it without problem .
> and /dev/video0 appears ok.
> but it does not capture nothing .
>
> probe with gstreamer, mplayer and and with this:
>
> cat /dev/video0 >> sample.mpg
>
> and always the file size is 0
>
> And Modify em28xx-core.c
> Original:
> /* FIXME: this only function read values from dev */
> int em28xx_resolution_set(struct em28xx *dev)
> {
> int width, height;
> width = norm_maxw(dev);
> height = norm_maxh(dev);
>
> /* Properly setup VBI */
> dev->vbi_width = 720;
> if (dev->norm & V4L2_STD_525_60)
> dev->vbi_height = 12;
> else
> dev->vbi_height = 18;
>
> if (!dev->progressive)
> height >>= norm_maxh(dev);
>
> em28xx_set_outfmt(dev);
>
>
> Make it:
> /* FIXME: this only function read values from dev */
> int em28xx_resolution_set(struct em28xx *dev)
> {
> int width, height;
> width = norm_maxw(dev);
> height = norm_maxh(dev);
>
> /* Properly setup VBI */
> dev->vbi_width = 720;
> if (dev->norm & V4L2_STD_525_60)
> dev->vbi_height = 12;
> else
> dev->vbi_height = 18;
>
> // if (!dev->progressive)
> // height >>= norm_maxh(dev);
>
> em28xx_set_outfmt(dev);
>
>
> Once this change is applied, the video is shown ok, but around the 10
> minutes it sends the following message
>
> em28xx #0: cannot change alternate number to 0 (error=-110)
> ehci-omap ehci-omap.0: force halt; handshake fa064814 00004000 00000000 ->
> -110 submit of urb 1 failed (error=-108)
>
> and it does not work ,it is only necessary to reboot and it returns to
> happen the same
>
> you can help me?
>
> thank you very much for any information.
>
> Regards.
>
> --
> Marcos Saldivia Delgado
>
>

Re: Problem with em28xx driver in Gumstix Overo

[Devin Heitmueller]

On Mon, Jan 3, 2011 at 3:13 PM, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>> // if (!dev->progressive)
>> // height >>= norm_maxh(dev);

This would suggest that the device is providing progressive video and
there is a mismatch between the board profile and the actual hardware,
which is certainly possible but I know absolutely nothing about the
product in question.

It would be helpful if we could get the output of dmesg for starters,
so we can see which board profile is being used.

Cheers,

Devin

-- 
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com

Re: Problem with em28xx driver in Gumstix Overo

[Mauro Carvalho Chehab]

Hi Marcos,

Em 03-01-2011 18:13, Linus Torvalds escreveu:
> Hi Marcos,
> 
> [ full email quoted for new cc's ]
> 
> you really would be much better off sending a diff rather than trying
> to explain what changes you did, but quite frankly, even if you did
> that, I'd still want this to go through the actual maintainers of the
> em28xx driver.

As I explained to the email you sent me in priv, It seems that
there's a bug at the scaling function. Not sure why this but doesn't 
show on x86. 

I intend to do some tests with em28xx and fix the issue, but I can't 
do it this week. I have a few arm boards, and I'll test on them with
some em28xx devices.

> The USB error sounds like some independent unrelated issue, I have no
> ideas on it.

Yes, it seems to be unrelated. Those video drivers generate a lot of ISOC
traffic at the USB bus. I suspect that the hardware have some problems
to keep that high bandwidth rate.  Maybe one of the ARM maintainers could
give us some light around this issue, as this driver doesn't suffer any
known issue of this kind on x86 arch. We even did a large stress test on 
em28xx driver in the past, in order to fix a memory leak there.

Cheers,
Mauro.

Re: Problem with em28xx driver in Gumstix Overo

[Martin Seekatz]

Am Montag 03 Januar 2011 schrieb Devin Heitmueller:
> On Mon, Jan 3, 2011 at 3:13 PM, Linus Torvalds
> 
> <torvalds@linux-foundation.org> wrote:
> >> // if (!dev->progressive)
> >> // height >>= norm_maxh(dev);
> 
> This would suggest that the device is providing progressive video
> and there is a mismatch between the board profile and the actual
> hardware, which is certainly possible but I know absolutely
> nothing about the product in question.
> 
> It would be helpful if we could get the output of dmesg for
> starters, so we can see which board profile is being used.

This main problem seems to be similare to the problem I reportet on 
2011-01-01 to the ML, subject: 
Silver Crest VG2000  "USB 2.0 Video Grabber", USB-Id: eb1a:2863 - does 
not work

In the meantime the figures of the partikular device, including dmesg 
output,  is been included in the linuxtv wiki as
http://www.linuxtv.org/wiki/index.php/SilverCrest_USB_2.0_Video_Grabber_VG_2000

Best regards
Martin

-- 
"Was ist der Unterschied zwischen Franken und Oberbayern?" -
"Die Franken haben weniger Berge, aber dafür mehr Horizont."