Mailing List Archive: PVR-350 decoder firmware problem

PVR-350 decoder firmware problem

May 17, 2009, 4:03 PM

Post #1 of 6 (3633 views)

I've been running the tip of the v4l-dvb repository on my 2.6.23 kernel for
some time now, updating periodically to get fixes (especially fixes related
to VBI). This hasn't caused me any trouble up until recently. Let me say at
the outset that this is not urgent for me; happily, I was careful and backed
up the kernel modules as of the point that it was working. I can go back --
and have, several times -- to the last working update from March 29th.
For the last several weeks, the tip of the repository seems to have trouble
loading the decoder firmware and/or the init code on the PVR-350. At least,
I never see the "green flash" as the card is initialized, and while the
console switches to the PVR-350, nothing ends up displaying from the S-video
output. Numerous reboots and inspection of /var/log/messages has turned up
nothing different with regard to ivtv/ivtvfb. That is, until today:

ivtv1: Loaded v4l-cx2341x-enc.fw firmware (376836 bytes)
ivtv1: Loaded v4l-cx2341x-dec.fw firmware (262144 bytes)
ivtv1: Decoder firmware dead!
ivtv1: Retry loading firmware
ivtv1: Loaded v4l-cx2341x-enc.fw firmware (376836 bytes)
ivtv1: Loaded v4l-cx2341x-dec.fw firmware (262144 bytes)
ivtv1: Encoder revision: 0x02060039
ivtv1: Decoder revision: 0x02020023
ivtv1: Loaded v4l-cx2341x-init.mpg firmware (155648 bytes)
ivtvfb1: Framebuffer at 0xf1510000, mapped to 0xe6390000, size 1665k
Console: switching to colour frame buffer device 80x25
ivtvfb1: Framebuffer registered on ivtv1

I should note that the machine acts like everything is working, even the
time this message came up. I have X11 running, and x11vnc set up, and I can
connect and interact with the machine normally. The only problem is that it
doesn't actually seem to be sending video to my television.

I'm handy with code and a compiler, but not so handy with Mercurial, so I'm
a bit challenged with regard to figuring out what might have changed to
cause this issue. Nevertheless I am really interested in getting Andy's
changes to sort out VBI in the cx18 driver, so any gentle nudges to put me
on the right path would be most appreciated.

Thanks, and regards,

--scott

Re: PVR-350 decoder firmware problem [ In reply to ]

awalls at radix

May 17, 2009, 5:10 PM

Post #2 of 6 (3476 views)

Permalink

On Sun, 2009-05-17 at 18:03 -0500, Scott Reynolds wrote:
> I've been running the tip of the v4l-dvb repository on my 2.6.23
> kernel for some time now, updating periodically to get fixes
> (especially fixes related to VBI). This hasn't caused me any trouble
> up until recently. Let me say at the outset that this is not urgent
> for me; happily, I was careful and backed up the kernel modules as of
> the point that it was working. I can go back -- and have, several
> times -- to the last working update from March 29th.
>
>
> For the last several weeks, the tip of the repository seems to have
> trouble loading the decoder firmware and/or the init code on the
> PVR-350. At least, I never see the "green flash" as the card is
> initialized, and while the console switches to the PVR-350, nothing
> ends up displaying from the S-video output. Numerous reboots and
> inspection of /var/log/messages has turned up nothing different with
> regard to ivtv/ivtvfb. That is, until today:
>
>
> ivtv1: Loaded v4l-cx2341x-enc.fw firmware (376836 bytes)
> ivtv1: Loaded v4l-cx2341x-dec.fw firmware (262144 bytes)
> ivtv1: Decoder firmware dead!
> ivtv1: Retry loading firmware
> ivtv1: Loaded v4l-cx2341x-enc.fw firmware (376836 bytes)
> ivtv1: Loaded v4l-cx2341x-dec.fw firmware (262144 bytes)
> ivtv1: Encoder revision: 0x02060039
> ivtv1: Decoder revision: 0x02020023
> ivtv1: Loaded v4l-cx2341x-init.mpg firmware (155648 bytes)
> ivtvfb1: Framebuffer at 0xf1510000, mapped to 0xe6390000, size 1665k
> Console: switching to colour frame buffer device 80x25
> ivtvfb1: Framebuffer registered on ivtv1
>
>
> I should note that the machine acts like everything is working, even
> the time this message came up. I have X11 running, and x11vnc set up,
> and I can connect and interact with the machine normally. The only
> problem is that it doesn't actually seem to be sending video to my
> television.

Scott,

Have you tried setting something like this in your /etc/modporbe.conf

options ivtvfb debug=3
options ivtv debug=1023

and then looking at what gets logged?

>
> I'm handy with code and a compiler, but not so handy with Mercurial,
> so I'm a bit challenged with regard to figuring out what might have
> changed to cause this issue.

Here are some hints on looking for possible problem changes:

$ cd ~/path/v4l-dvb
$ hg help log
$ hg help date
$ hg log -d ">Mar 28" linux/drivers/media/video/ivtv/* linux/drivers/media/video/saa71[12]*
$ hg log -vp -d ">Mar 28" linux/drivers/media/video/ivtv/* linux/drivers/media/video/saa71[12]* | vim -R -

:set syntax=diff

I'm just guessing it's a change in one of those files. That filespec
limits things to 16 changesets. If you can narrow down the date range,
then you can open up the log serach to all the files.

> Nevertheless I am really interested in getting Andy's changes to sort
> out VBI in the cx18 driver, so any gentle nudges to put me on the
> right path would be most appreciated.

I'm pretty happy with the AFE and VBI changes in the cx18 driver. If
you have and HVR-1600 or Compro H900 or some other cx18 supported board,
you can test them if you like.

However, a hassle will be is translating those cx18 AFE (and maybe VBI
changes) to PVR-150/500 type (CX2584x + CX23416) boards, because the
cx25840 module supports a few slightly different chips and the ivtv
module supports a few different decoders.

The PVR-350 uses a different video decoder from the CX25843, so the
changes in the cx18 driver won't be applicable to making VBI on a
PVR-350 better.

Regards,
Andy

> Thanks, and regards,
>
>
> --scott

_______________________________________________
ivtv-users mailing list
ivtv-users@ivtvdriver.org
http://ivtvdriver.org/mailman/listinfo/ivtv-users

Re: PVR-350 decoder firmware problem [ In reply to ]

martin.dauskardt at gmx

May 18, 2009, 11:28 AM

Post #3 of 6 (3478 views)

Permalink

> For the last several weeks, the tip of the repository seems to have trouble
> loading the decoder firmware and/or the init code on the PVR-350. At least,
> I never see the "green flash" as the card is initialized

This is no error, but a feature :-)
I contributed a patch to the ivtv driver, which deactivates the saa712x until
the MPEG decoder has been initialized:
http://linuxtv.org/hg/v4l-dvb/rev/e95d50985dde

I recently built a PVR350 vdr system with Ubuntu Jaunty. The Jaunty kernel
includes this patch. While using the machine without a VGA card, the console
appeared on the TV out, so this is still working.

I also use a very recent hg snapshot from May for my other PVR350 machine. TV
out works fine, both mpeg decoder (/dev/video16) and mplayer with ivtv cvidix
(using /dev/video48).

If you use a current snapshot, there should be no problem with the S-Video.
(There was one, which was fixed in January with
http://linuxtv.org/hg/v4l-dvb/rev/4c1561cccc22. )

You could try the following:
Stop xorg
unload ivtvfb
undload ivtv
load ivtvfb
cat test.mpg > /dev/video16
You should see the video on the TV out. If this is the case, the decoder is
working.

Greets
Martin

> , and while the
> console switches to the PVR-350, nothing ends up displaying from the S-video
> output. Numerous reboots and inspection of /var/log/messages has turned up
> nothing different with regard to ivtv/ivtvfb. That is, until today:
>
> ivtv1: Loaded v4l-cx2341x-enc.fw firmware (376836 bytes)
> ivtv1: Loaded v4l-cx2341x-dec.fw firmware (262144 bytes)
> ivtv1: Decoder firmware dead!
> ivtv1: Retry loading firmware
> ivtv1: Loaded v4l-cx2341x-enc.fw firmware (376836 bytes)
> ivtv1: Loaded v4l-cx2341x-dec.fw firmware (262144 bytes)
> ivtv1: Encoder revision: 0x02060039
> ivtv1: Decoder revision: 0x02020023
> ivtv1: Loaded v4l-cx2341x-init.mpg firmware (155648 bytes)
> ivtvfb1: Framebuffer at 0xf1510000, mapped to 0xe6390000, size 1665k
> Console: switching to colour frame buffer device 80x25
> ivtvfb1: Framebuffer registered on ivtv1
>
> I should note that the machine acts like everything is working, even the
> time this message came up. I have X11 running, and x11vnc set up, and I can
> connect and interact with the machine normally. The only problem is that it
> doesn't actually seem to be sending video to my television.
>
> I'm handy with code and a compiler, but not so handy with Mercurial, so I'm
> a bit challenged with regard to figuring out what might have changed to
> cause this issue. Nevertheless I am really interested in getting Andy's
> changes to sort out VBI in the cx18 driver, so any gentle nudges to put me
> on the right path would be most appreciated.
>
> Thanks, and regards,
>
> --scott
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
http://ivtvdriver.org/pipermail/ivtv-users/attachments/20090517/adb69949/attachment-0001.html
>
> ------------------------------
>
> Message: 2
> Date: Sun, 17 May 2009 20:10:25 -0400
> From: Andy Walls <awalls@radix.net>
> Subject: Re: [ivtv-users] PVR-350 decoder firmware problem
> To: User discussion about IVTV <ivtv-users@ivtvdriver.org>
> Message-ID: <1242605425.10045.31.camel@palomino.walls.org>
> Content-Type: text/plain
>
> On Sun, 2009-05-17 at 18:03 -0500, Scott Reynolds wrote:
> > I've been running the tip of the v4l-dvb repository on my 2.6.23
> > kernel for some time now, updating periodically to get fixes
> > (especially fixes related to VBI). This hasn't caused me any trouble
> > up until recently. Let me say at the outset that this is not urgent
> > for me; happily, I was careful and backed up the kernel modules as of
> > the point that it was working. I can go back -- and have, several
> > times -- to the last working update from March 29th.
> >
> >
> > For the last several weeks, the tip of the repository seems to have
> > trouble loading the decoder firmware and/or the init code on the
> > PVR-350. At least, I never see the "green flash" as the card is
> > initialized, and while the console switches to the PVR-350, nothing
> > ends up displaying from the S-video output. Numerous reboots and
> > inspection of /var/log/messages has turned up nothing different with
> > regard to ivtv/ivtvfb. That is, until today:
> >
> >
> > ivtv1: Loaded v4l-cx2341x-enc.fw firmware (376836 bytes)
> > ivtv1: Loaded v4l-cx2341x-dec.fw firmware (262144 bytes)
> > ivtv1: Decoder firmware dead!
> > ivtv1: Retry loading firmware
> > ivtv1: Loaded v4l-cx2341x-enc.fw firmware (376836 bytes)
> > ivtv1: Loaded v4l-cx2341x-dec.fw firmware (262144 bytes)
> > ivtv1: Encoder revision: 0x02060039
> > ivtv1: Decoder revision: 0x02020023
> > ivtv1: Loaded v4l-cx2341x-init.mpg firmware (155648 bytes)
> > ivtvfb1: Framebuffer at 0xf1510000, mapped to 0xe6390000, size 1665k
> > Console: switching to colour frame buffer device 80x25
> > ivtvfb1: Framebuffer registered on ivtv1
> >
> >
> > I should note that the machine acts like everything is working, even
> > the time this message came up. I have X11 running, and x11vnc set up,
> > and I can connect and interact with the machine normally. The only
> > problem is that it doesn't actually seem to be sending video to my
> > television.
>
> Scott,
>
> Have you tried setting something like this in your /etc/modporbe.conf
>
> options ivtvfb debug=3
> options ivtv debug=1023
>
> and then looking at what gets logged?
>
>
> >
> > I'm handy with code and a compiler, but not so handy with Mercurial,
> > so I'm a bit challenged with regard to figuring out what might have
> > changed to cause this issue.
>
>
> Here are some hints on looking for possible problem changes:
>
> $ cd ~/path/v4l-dvb
> $ hg help log
> $ hg help date
> $ hg log -d ">Mar 28" linux/drivers/media/video/ivtv/*
linux/drivers/media/video/saa71[12]*
> $ hg log -vp -d ">Mar 28" linux/drivers/media/video/ivtv/*
linux/drivers/media/video/saa71[12]* | vim -R -
>
> :set syntax=diff
>
>
> I'm just guessing it's a change in one of those files. That filespec
> limits things to 16 changesets. If you can narrow down the date range,
> then you can open up the log serach to all the files.
>
>
> > Nevertheless I am really interested in getting Andy's changes to sort
> > out VBI in the cx18 driver, so any gentle nudges to put me on the
> > right path would be most appreciated.
>
> I'm pretty happy with the AFE and VBI changes in the cx18 driver. If
> you have and HVR-1600 or Compro H900 or some other cx18 supported board,
> you can test them if you like.
>
> However, a hassle will be is translating those cx18 AFE (and maybe VBI
> changes) to PVR-150/500 type (CX2584x + CX23416) boards, because the
> cx25840 module supports a few slightly different chips and the ivtv
> module supports a few different decoders.
>
> The PVR-350 uses a different video decoder from the CX25843, so the
> changes in the cx18 driver won't be applicable to making VBI on a
> PVR-350 better.
>
> Regards,
> Andy
>
> > Thanks, and regards,
> >
> >
> > --scott
>
>
>
>
>
> ------------------------------
>
> _______________________________________________
> ivtv-users mailing list
> ivtv-users@ivtvdriver.org
> http://ivtvdriver.org/mailman/listinfo/ivtv-users
>
> End of ivtv-users Digest, Vol 45, Issue 23
> ******************************************
>

_______________________________________________
ivtv-users mailing list
ivtv-users@ivtvdriver.org
http://ivtvdriver.org/mailman/listinfo/ivtv-users

Re: PVR-350 decoder firmware problem [ In reply to ]

scottr at clank

May 18, 2009, 11:57 AM

Post #4 of 6 (3454 views)

Permalink

On Mon, May 18, 2009 at 1:28 PM, Martin Dauskardt
<martin.dauskardt@gmx.de>wrote:

> > For the last several weeks, the tip of the repository seems to have
> trouble
> > loading the decoder firmware and/or the init code on the PVR-350. At
> least,
> > I never see the "green flash" as the card is initialized
>
> This is no error, but a feature :-)
> I contributed a patch to the ivtv driver, which deactivates the saa712x
> until
> the MPEG decoder has been initialized:
> http://linuxtv.org/hg/v4l-dvb/rev/e95d50985dde
>

I believe that this change is present in the v4l-dvb repository as of March
29 (which I've since re-confirmed as one of the last dates that things
worked) -- so I'm pretty sure it's not the culprit.

As a progress report, I've narrowed the issue to a date between March 30 and
April 15. Andy's response was exactly what I needed to make some forward
progress (thanks Andy!).

I'm working on this as my work and recording schedules allow. I've got busy
tuners this evening, so probably nothing more until tomorrow.

--scott

Re: PVR-350 decoder firmware problem [ In reply to ]

srey0123+ivtv at gmail

May 20, 2009, 7:54 AM

Post #5 of 6 (3441 views)

Permalink

On Sun, May 17, 2009 at 7:10 PM, Andy Walls <awalls@radix.net> wrote:
>
> Here are some hints on looking for possible problem changes:
> [...]
>
> I'm just guessing it's a change in one of those files. That filespec
> limits things to 16 changesets. If you can narrow down the date range,
> then you can open up the log serach to all the files.

Thanks, Andy, this was well more than enough to get me going in the
right direction. I quickly narrowed down the date to March 30, for
which there were only two changesets for the files you indicated. I
quickly inferred that it was this change that broke PVR-350 output on
kernel 2.6.23:

http://linuxtv.org/hg/v4l-dvb/rev/e66ace4b4a61
v4l2: use old-style i2c API for kernels < 2.6.26 instead of < 2.6.22

Updating to the tip and reverting this changeset (ignoring one
harmless reject) got PVR-350 output working again for me.

Now, I'm not saying that this is the correct solution. I defer to
those with a much better idea of what the general effects of this
might be. It could be that my kernel doesn't have the old-style i2c
API compiled in, perhaps, which is why it failed in the first place.
I'm just guessing and unfortunately my zero megabits-per-second
Internet upgrade of late prevents me from checking right now. (The
"upgrade" is a joke, in case it's not obvious. I expect the problem to
be resolved by this evening, though.)

> I'm pretty happy with the AFE and VBI changes in the cx18 driver. If
> you have and HVR-1600 or Compro H900 or some other cx18 supported board,
> you can test them if you like.

Let me be clear -- I record analog signals from a pair of DTV tuners
on a PVR-150 and an HVR-1600. Playback goes through a PVR-350
connected via S-video to the television. Now, the PVR-150 has been
giving me stable captions (via sliced VBI) for some time now; and as
of March 29, the cx18 driver was mostly working, with some sync issues
and occasionally dropped captions. My goal, obviously, is to see that
last part working.

> The PVR-350 uses a different video decoder from the CX25843, so the
> changes in the cx18 driver won't be applicable to making VBI on a
> PVR-350 better.

I don't mean to be contrary, but the PVR-350 is working -and- I have
appear to have stable captions on playback for recordings from both
the PVR-150 and the HVR-1600. Do you have any explanation?

Thanks again for your help!

Best regards,

--scott

_______________________________________________
ivtv-users mailing list
ivtv-users@ivtvdriver.org
http://ivtvdriver.org/mailman/listinfo/ivtv-users

Re: PVR-350 decoder firmware problem [ In reply to ]

awalls at radix

May 20, 2009, 6:55 PM

Post #6 of 6 (3421 views)

Permalink

On Wed, 2009-05-20 at 09:54 -0500, Scott Reynolds wrote:
> On Sun, May 17, 2009 at 7:10 PM, Andy Walls <awalls@radix.net> wrote:
> >
> > Here are some hints on looking for possible problem changes:
> > [...]
> >
> > I'm just guessing it's a change in one of those files. That filespec
> > limits things to 16 changesets. If you can narrow down the date range,
> > then you can open up the log serach to all the files.
>
> Thanks, Andy, this was well more than enough to get me going in the
> right direction. I quickly narrowed down the date to March 30, for
> which there were only two changesets for the files you indicated. I
> quickly inferred that it was this change that broke PVR-350 output on
> kernel 2.6.23:
>
> http://linuxtv.org/hg/v4l-dvb/rev/e66ace4b4a61
> v4l2: use old-style i2c API for kernels < 2.6.26 instead of < 2.6.22
>
> Updating to the tip and reverting this changeset (ignoring one
> harmless reject) got PVR-350 output working again for me.
>
> Now, I'm not saying that this is the correct solution.

One could argue that if it works for you with no ill effects, then it is
a correct solution for you.

> I defer to
> those with a much better idea of what the general effects of this
> might be. It could be that my kernel doesn't have the old-style i2c
> API compiled in, perhaps, which is why it failed in the first place.

I don't have any old kernels lying around to troubleshoot this one.
Hans would understand the implications best.

Generally though, the line of thought on the linux-media mailing list is
to provide backward compatability when reasonable, but not to work too
hard on solving backward comptability problems. Volunteer time is
scarce.

> I'm just guessing and unfortunately my zero megabits-per-second
> Internet upgrade of late prevents me from checking right now. (The
> "upgrade" is a joke, in case it's not obvious. I expect the problem to
> be resolved by this evening, though.)
>
> > I'm pretty happy with the AFE and VBI changes in the cx18 driver. If
> > you have and HVR-1600 or Compro H900 or some other cx18 supported board,
> > you can test them if you like.
>
> Let me be clear -- I record analog signals from a pair of DTV tuners
> on a PVR-150 and an HVR-1600. Playback goes through a PVR-350
> connected via S-video to the television. Now, the PVR-150 has been
> giving me stable captions (via sliced VBI) for some time now; and as
> of March 29, the cx18 driver was mostly working, with some sync issues
> and occasionally dropped captions. My goal, obviously, is to see that
> last part working.

It's likely the application side buffer handling performance problem I
fixed with the series of patches around April 15:

http://linuxtv.org/hg/v4l-dvb/shortlog/518bb17e199b

> > The PVR-350 uses a different video decoder from the CX25843, so the
> > changes in the cx18 driver won't be applicable to making VBI on a
> > PVR-350 better.
>
> I don't mean to be contrary, but the PVR-350 is working -and- I have
> appear to have stable captions on playback for recordings from both
> the PVR-150 and the HVR-1600. Do you have any explanation?

Hmm. Maybe I was thinking about something else.

So if everything is working for you, have fun.

> Thanks again for your help!

You're welcome.

Regards,
Andy

> Best regards,
>
> --scott

_______________________________________________
ivtv-users mailing list
ivtv-users@ivtvdriver.org
http://ivtvdriver.org/mailman/listinfo/ivtv-users