On Monday 29 January 2007 07:04, Tim Harvey wrote:
> --- Ricardo Lugo <[EMAIL PROTECTED]> wrote:
> > Hans, Tim:
> >
> > I am happy to report that the trunk's DMA works flawlessly on PPC!
> > There's no need for that #ifdef __powerpc__ any more!
>
> hey thats great... we fixed 2 architectures at once :)
>
> Hans, please remove the #ifdef __powerpc__ when you get a chance
Will do.
>
> > Oddly, PIO still doesn't work. Does PIO work on x86?
>
> No I haven't seen it work in trunk for x86 either - I mentioned this
> back in my ARM thread but it may have got missed. From what I saw
> interrupts don't occur after the start of the capture.
I'm not going to bother with this. DMA is now working for other
platforms than x86, so PIO support can be removed.
Rick, I'll look at the framebuffer problems later when I have a bit more
time.
Hans
>
> Tim
>
> > Since I have a PVR-350, I decided to test out every feature of it;
> > here's a scorecard of what works, what doesn't and what's wierd on
> > PPC:
> >
> > PIO: doesn't work
> > DMA: works perfectly!
> > encoder: works perfectly!
> > decoder: works perfectly!
> > framebuffer: works, but the colors are off: everything is yellow-
> > tinted. see attached picture. (fortunately, when you play MPEGs on
> > the decoder the colors are fine! its just the OSD)
> >
> > All in all, excellent progress and many thanks!
> > - Rick
> >
> > On Jan 25, 2007, at 1:41 PM, Hans Verkuil wrote:
> > > On Thursday 25 January 2007 19:02, Ricardo Lugo wrote:
> > >> On Jan 25, 2007, at 10:53 AM, Tim Harvey wrote:
> > >>> --- Ricardo Lugo <[EMAIL PROTECTED]> wrote:
> > >>>> On Jan 24, 2007, at 11:41 PM, Tim Harvey wrote:
> > >>>>> Rick,
> > >>>>>
> > >>>>> I cant comment on your 'all black' issue other than it looks
> > >>>>> like its encoding
> > >>>>> from input 0 (should be composite) and that I see a
> > >>>>> CX2341X_ENC_MUTE_VIDEO API
> > >>>>> call at the start of your capture (not sure why I don't see
> > >>>>> any equiv 'unmute'
> > >>>>> but I don't understand the API much - perhaps Hans can
> > >>>>> comment).
> > >>>>>
> > >>>>> Your still running in PIO mode. There is an '#ifdef
> > >>>>> __powerpc__' in driver/ivtv-driver.h that you'll need to
> > >>>>> change to enable DMA for powerpc.
> > >>>>> Just comment out the ifdef to see if/how DMA works. Again -
> > >>>>> this is a shot in
> > >>>>> the dark - I don't have a ppc platform I just know that your
> > >>>>> big- endian and
> > >>>>> that the endian patches need to be in place for any hope of
> > >>>>> DMA to work on that
> > >>>>> architecture.
> > >>>>
> > >>>> I tried enabling DMA by only leaving the #define IVTV_USE_PIO
> > >>>> 0 and still got the same results as with PIO.
> > >>>
> > >>> can you post the log with ivtv_debug=79 for DMA ('bs=1024
> > >>> count=4' would be 4K
> > >>> and enough to see that DMA is working properly - looks like it
> > >>> is however).
> > >>
> > >> Tim,
> > >>
> > >> I'll have to wait until Monday or so to try this out, as I will
> > >> be out traveling. I will play around with selecting different
> > >> inputs and such.
> > >>
> > >> In the mean time, I noticed that ivtv-0.9.1 is subject to the
> > >> same "Couldn't find start of within 256 bytes" error that your
> > >> cpu_to_le32 (DMA_MAGIC_COOKIE) fixes. Any chance of putting out
> > >> a patch for that version?
> > >
> > > Hi Rick,
> > >
> > > Please work with the trunk driver! I doubt that a new ivtv-0.9.2
> > > release
> > > will be made. As far as I'm concerned any bugs relating to the
> > > powerpc should be fixed in the trunk. That's going to be the
> > > leading code and also basically what will end up in the kernel
> > > when the driver will be merged.
> > >
> > > Hans
> > >
> > >> - Rick
> > >>
> > >>> I'm not sure why you would only be getting 138kb... your
> > >>> obviously pulling data
> > >>> in both PIO and DMA modes and if you can play it back in a
> > >>> player without
> > >>> errors then its MPEG (even if its black). The 'black' issue I
> > >>> would guess
> > >>> would be something simple dealing with the setting of the input
> > >>> of the card.
> > >>>
> > >>> I do notice that the final read call in your dump is:
> > >>>>>> ivtv0 info: read 512 from encoder MPEG, got 4294967292
> > >>>
> > >>> which would indicate a negative value returned from the read
> > >>> thats indicating
> > >>> some error that perhaps isn't being caught.
> > >>>
> > >>>> [EMAIL PROTECTED]:/usr/src# dd if=/dev/video0 of=test.mpg bs=1024
> > >>>> count=1024 135+0 records in
> > >>>> 135+0 records out
> > >>>> 138240 bytes (138 kB) copied, 7.8193 seconds, 17.7 kB/s
> > >>>>
> > >>>>> As far as the amt of data captured, it looks like you stopped
> > >>>>> the capture and
> > >>>>> everything else looked good until then. If you use something
> > >>>>> like:
> > >>>>>
> > >>>>> dd if=/dev/video0 of=test.mpg bs=1024 count=1024
> > >>>>>
> > >>>>> you should get a 1MB file (adjust count for more than 1024
> > >>>>> 1KB blocks).
> > >>>>>
> > >>>>> The endian fixes definitely fix the DMA_MAGIC_COOKIE issue -
> > >>>>> I bet your just at
> > >>>>> the wrong input as far as the black video goes.
> > >>>>
> > >>>> Hrm. Now this is interesting. Input 0 is my Tuner (which is
> > >>>> hooked up to the comcast cable feed). Is that not correct?
> > >>>> I let the capture go for on for 10 minutes and I will still
> > >>>> get a 138kb file.
> > >>>>
> > >>>> [EMAIL PROTECTED]:/usr/src/ivtv/utils# ./v4l2-ctl -I
> > >>>> Video input : 0 (Tuner 1)
> > >>>> [EMAIL PROTECTED]:/usr/src/ivtv/utils# ./v4l2-ctl -T
> > >>>> Tuner:
> > >>>> Capabilities : 62.5 kHz multi-standard stereo
> > >>>> lang1 lang2
> > >>>> Frequency range : 44.0 MHz - 958.0 MHz
> > >>>> Signal strength : 0%
> > >>>> Current audio mode : stereo
> > >>>> Available subchannels: mono
> > >>>>
> > >>>> It seems as though the driver is not actually returning the
> > >>>> data it is reading?
> > >>>
> > >>> no, its returning valid MPEG if your player plays it - its just
> > >>> not recording
> > >>> the correct input/content for some reason.
> > >>>
> > >>> Tim
> > >>>
> > >>>> - Rick
> > >>>>
> > >>>>> Tim
> > >>>>>
> > >>>>> --- Ricardo Lugo <[EMAIL PROTECTED]> wrote:
> > >>>>>> Hans and Tim,
> > >>>>>>
> > >>>>>> Excellent! I tried out the latest trunk (with the
> > >>>>>> DMA_MAGIC_COOKIE changes) and there is some improvement.
> > >>>>>>
> > >>>>>> Unfortunately, there are still problems.
> > >>>>>>
> > >>>>>> Using dd if=/dev/video0 of=bla.mpg leads only to a ~137kb
> > >>>>>> capture, that looks normal but is all black. Attached are my
> > >>>>>> dmesg and the mpeg.
> > >>>>>>
> > >>>>>> - Rick
> > >>>>>> <snip>
> > >>>>>
> > >>>>> _______________________________________________
> > >>>>> ivtv-devel mailing list
> > >>>>> [email protected]
> > >>>>> http://ivtvdriver.org/mailman/listinfo/ivtv-devel
> > >>>>
> > >>>> _______________________________________________
> > >>>> ivtv-devel mailing list
> > >>>> [email protected]
> > >>>> http://ivtvdriver.org/mailman/listinfo/ivtv-devel
> > >>>
> > >>> _______________________________________________
> > >>> ivtv-devel mailing list
> > >>> [email protected]
> > >>> http://ivtvdriver.org/mailman/listinfo/ivtv-devel
> > >>
> > >> _______________________________________________
> > >> ivtv-devel mailing list
> > >> [email protected]
> > >> http://ivtvdriver.org/mailman/listinfo/ivtv-devel
> > >
> > > _______________________________________________
> > > ivtv-devel mailing list
> > > [email protected]
> > > http://ivtvdriver.org/mailman/listinfo/ivtv-devel
> > >
> > > _______________________________________________
> >
> > ivtv-devel mailing list
> > [email protected]
> > http://ivtvdriver.org/mailman/listinfo/ivtv-devel
>
> _______________________________________________
> ivtv-devel mailing list
> [email protected]
> http://ivtvdriver.org/mailman/listinfo/ivtv-devel
_______________________________________________
ivtv-devel mailing list
[email protected]
http://ivtvdriver.org/mailman/listinfo/ivtv-devel
