Re: WinTV 1400 broken with recent versions?

Re: WinTV 1400 broken with recent versions?

[Jarod Wilson]

On Mar 9, 2011, at 11:52 AM, Jean-Michel Bruenn wrote:

> Hey,
> 
> is this driver going to be fixed anytime soon? It was working fine ago a
> half year/year.
> 
> lspci:
> 06:00.0 Multimedia video controller: Conexant Systems, Inc. CX23885 PCI
> Video and Audio Decoder (rev 02)
> 
> uname -a:
> Linux lyra 2.6.37.1 #1 SMP PREEMPT Tue Feb 22 13:22:59 CET 2011 x86_64
> x86_64 x86_64 GNU/Linux
> 
> dmesg:
> xc2028 1-0064: i2c output error: rc = -6 (should be 64)
> xc2028 1-0064: -6 returned from send
> xc2028 1-0064: Error -22 while loading base firmware
> xc2028 1-0064: Loading firmware for type=BASE F8MHZ (3), id
> 0000000000000000.
> xc2028 1-0064: i2c output error: rc = -6 (should be 64)
> xc2028 1-0064: -6 returned from send
> xc2028 1-0064: Error -22 while loading base firmware
> xc2028 1-0064: Loading firmware for type=BASE F8MHZ (3), id
> 0000000000000000.
> xc2028 1-0064: i2c output error: rc = -6 (should be 64)
> xc2028 1-0064: -6 returned from send
> xc2028 1-0064: Error -22 while loading base firmware
> 
> nothing works - if i do scan it finds nothing and those messages appear on
> dmesg. if i try to watch with the channels.conf from my other pc i can play
> nothing, all i get is those messages above.

This may already be fixed, just not in 2.6.37.x yet. Can you give
2.6.38-rc8 (or later) a try and/or the media_build bits?

http://linuxtv.org/wiki/index.php/How_to_Obtain,_Build_and_Install_V4L-DVB_Device_Drivers

-- 
Jarod Wilson
jarod@wilsonet.com

WinTV 1400 broken with recent versions?

[Jean-Michel Bruenn]

Hey,

is this driver going to be fixed anytime soon? It was working fine ago a
half year/year.

lspci:
06:00.0 Multimedia video controller: Conexant Systems, Inc. CX23885 PCI
Video and Audio Decoder (rev 02)

uname -a:
Linux lyra 2.6.37.1 #1 SMP PREEMPT Tue Feb 22 13:22:59 CET 2011 x86_64
x86_64 x86_64 GNU/Linux

dmesg:
xc2028 1-0064: i2c output error: rc = -6 (should be 64)
xc2028 1-0064: -6 returned from send
xc2028 1-0064: Error -22 while loading base firmware
xc2028 1-0064: Loading firmware for type=BASE F8MHZ (3), id
0000000000000000.
xc2028 1-0064: i2c output error: rc = -6 (should be 64)
xc2028 1-0064: -6 returned from send
xc2028 1-0064: Error -22 while loading base firmware
xc2028 1-0064: Loading firmware for type=BASE F8MHZ (3), id
0000000000000000.
xc2028 1-0064: i2c output error: rc = -6 (should be 64)
xc2028 1-0064: -6 returned from send
xc2028 1-0064: Error -22 while loading base firmware

nothing works - if i do scan it finds nothing and those messages appear
on dmesg. if i try to watch with the channels.conf from my other pc i
can play nothing, all i get is those messages above.

WinTV 1400 broken with recent versions?

[Jean-Michel Bruenn]

Hey,

is this driver going to be fixed anytime soon? It was working fine ago a
half year/year.

lspci:
06:00.0 Multimedia video controller: Conexant Systems, Inc. CX23885 PCI
Video and Audio Decoder (rev 02)

uname -a:
Linux lyra 2.6.37.1 #1 SMP PREEMPT Tue Feb 22 13:22:59 CET 2011 x86_64
x86_64 x86_64 GNU/Linux

dmesg:
xc2028 1-0064: i2c output error: rc = -6 (should be 64)
xc2028 1-0064: -6 returned from send
xc2028 1-0064: Error -22 while loading base firmware
xc2028 1-0064: Loading firmware for type=BASE F8MHZ (3), id
0000000000000000.
xc2028 1-0064: i2c output error: rc = -6 (should be 64)
xc2028 1-0064: -6 returned from send
xc2028 1-0064: Error -22 while loading base firmware
xc2028 1-0064: Loading firmware for type=BASE F8MHZ (3), id
0000000000000000.
xc2028 1-0064: i2c output error: rc = -6 (should be 64)
xc2028 1-0064: -6 returned from send
xc2028 1-0064: Error -22 while loading base firmware

nothing works - if i do scan it finds nothing and those messages appear on
dmesg. if i try to watch with the channels.conf from my other pc i can play
nothing, all i get is those messages above.

Re: WinTV 1400 broken with recent versions?

[jean.bruenn]

> This may already be fixed, just not in 2.6.37.x yet. Can you give
> 2.6.38-rc8 (or later) a try and/or the media_build bits?

Tried - Nope, same behaviour (same error messages in dmesg, no results by
scan)

>
http://linuxtv.org/wiki/index.php/How_to_Obtain,_Build_and_Install_V4L-DVB_Device_Drivers

I was unable to get that working; tried with 2.6.37.2 and 2.6.37.3 always
getting "invalid module format" and yeah, i tried with reboot, i tried
with a
fresh variant.. Also tried ./build.sh and make install and such stuff in
2.6.38-rc8, same.

Re: WinTV 1400 broken with recent versions?

[Jarod Wilson]

On Mar 9, 2011, at 7:00 PM, <jean.bruenn@ip-minds.de> wrote:

> 
>> This may already be fixed, just not in 2.6.37.x yet. Can you give
>> 2.6.38-rc8 (or later) a try and/or the media_build bits?
> 
> Tried - Nope, same behaviour (same error messages in dmesg, no results by
> scan)
> 
>> 
> http://linuxtv.org/wiki/index.php/How_to_Obtain,_Build_and_Install_V4L-DVB_Device_Drivers
> 
> I was unable to get that working; tried with 2.6.37.2 and 2.6.37.3 always
> getting "invalid module format" and yeah, i tried with reboot, i tried
> with a
> fresh variant.. Also tried ./build.sh and make install and such stuff in
> 2.6.38-rc8, same.

As discovered on irc, seems to be a mismatch between the headers that
were being built against and the running kernel.

That aside, given that this is a cx23885-based device, I suspect that
this commit may be relevant to the regression in functionality:

commit 44835f197bf1e3f57464f23dfb239fef06cf89be
Author: Jean Delvare <khali@linux-fr.org>
Date:   Sun Jul 18 16:52:05 2010 -0300

    V4L/DVB: cx23885: Check for slave nack on all transactions
    
    Don't just check for nacks on zero-length transactions. Check on
    other transactions too.
    
    Signed-off-by: Jean Delvare <khali@linux-fr.org>
    Signed-off-by: Andy Walls <awalls@md.metrocast.net>
    Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>


The retval the xc2028 firmware load routine is getting is -ENXIO,
which seems to possibly be new behavior as a result of that patch.
However, it may actually be that the xc2028 driver needs to add some
delays or retries in its firmware load function, as this change *is*
technically correct (Jean is the i2c subsystem maintainer, so we can
be pretty sure he knows how i2c stuff is *supposed* to behave). :)

You could try hard-coding a sleep and/or retries into the inner while
loop near the bottom of load_firmware() in tuner-xc2028.c... That's
definitely where things are falling down, anyway.


-- 
Jarod Wilson
jarod@wilsonet.com

Re: WinTV 1400 broken with recent versions?

[Jarod Wilson]

On Mar 10, 2011, at 9:37 AM, Jarod Wilson wrote:

> On Mar 9, 2011, at 7:00 PM, <jean.bruenn@ip-minds.de> wrote:
> 
>> 
>>> This may already be fixed, just not in 2.6.37.x yet. Can you give
>>> 2.6.38-rc8 (or later) a try and/or the media_build bits?
>> 
>> Tried - Nope, same behaviour (same error messages in dmesg, no results by
>> scan)
>> 
>>> 
>> http://linuxtv.org/wiki/index.php/How_to_Obtain,_Build_and_Install_V4L-DVB_Device_Drivers
>> 
>> I was unable to get that working; tried with 2.6.37.2 and 2.6.37.3 always
>> getting "invalid module format" and yeah, i tried with reboot, i tried
>> with a
>> fresh variant.. Also tried ./build.sh and make install and such stuff in
>> 2.6.38-rc8, same.
> 
> As discovered on irc, seems to be a mismatch between the headers that
> were being built against and the running kernel.
> 
> That aside, given that this is a cx23885-based device, I suspect that
> this commit may be relevant to the regression in functionality:
> 
> commit 44835f197bf1e3f57464f23dfb239fef06cf89be
> Author: Jean Delvare <khali@linux-fr.org>
> Date:   Sun Jul 18 16:52:05 2010 -0300
> 
>    V4L/DVB: cx23885: Check for slave nack on all transactions
> 
>    Don't just check for nacks on zero-length transactions. Check on
>    other transactions too.
> 
>    Signed-off-by: Jean Delvare <khali@linux-fr.org>
>    Signed-off-by: Andy Walls <awalls@md.metrocast.net>
>    Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
> 
> 
> The retval the xc2028 firmware load routine is getting is -ENXIO,
> which seems to possibly be new behavior as a result of that patch.
> However, it may actually be that the xc2028 driver needs to add some
> delays or retries in its firmware load function, as this change *is*
> technically correct (Jean is the i2c subsystem maintainer, so we can
> be pretty sure he knows how i2c stuff is *supposed* to behave). :)
> 
> You could try hard-coding a sleep and/or retries into the inner while
> loop near the bottom of load_firmware() in tuner-xc2028.c... That's
> definitely where things are falling down, anyway.

I knew this all seemed too familiar... :)

http://git.kernel.org/?p=linux/kernel/git/mchehab/linux-2.6.git;a=commit;h=67914b5c400d6c213f9e56d7547a2038ab5c06f4

Its already being reverted for 2.6.38 final (hopefully -- Mauro included
that in the pull req sent to Linus today).

-- 
Jarod Wilson
jarod@wilsonet.com

Re: WinTV 1400 broken with recent versions?

[jean.bruenn]

Hey,

i tried to revert that patch manually (e.g. switching into the directory,
vim cx23885-i2c.c, removing the stuff which was added), then "make clean"
"make distclean" followed by "./build.sh" then make rmmod, then plugged in
the card, dmesg shows it loaded the card correctly, all fine, then i did
./scan Scanlist.txt and i get the same i2c related errors. Did a reboot
just to verify, still getting those, scan gives no results always "tuning
failed". 

Then i reverted another patch (just to make sure..
http://git.kernel.org/?p=linux/kernel/git/mchehab/linux-2.6.git;a=commit;h=f4acb3c4ccca74f5448354308f917e87ce83505a)
- However, this didn't help. So, the problem might be somewhere else. 

I did some more research and it seems i'm not the only one with those
issues, tho nearly none gets answers regarding this trouble, e.g.:

(same card - also expresscard, december 2010)
http://www.spinics.net/lists/linux-media/msg27042.html
(not the same card, tho similar error, hvr 1500, i got hvr 1400 - februar
2009)
http://www.linuxtv.org/pipermail/linux-dvb/2009-February/031839.html

I'm running out of ideas where the problem might be located. i also tried
to switch the firmware by extracting the firmware manually, didn't help.
There's a low power version and another one available, tried to replace,
didn't work neither.

So it seems, this driver is broken at least since december 2010. Totally
weird why there's such a mess, i know for sure that this WAS working.

Anyone, any idea? Maybe something wrong configured in kernel? Might
running native 64bit the cause (no multilib/32bit compat libs here)?

> I knew this all seemed too familiar... :)
> 
>
http://git.kernel.org/?p=linux/kernel/git/mchehab/linux-2.6.git;a=commit;h=67914b5c400d6c213f9e56d7547a2038ab5c06f4
> 
> Its already being reverted for 2.6.38 final (hopefully -- Mauro included
> that in the pull req sent to Linus today).

Re: WinTV 1400 broken with recent versions?

[Devin Heitmueller]

On Fri, Mar 11, 2011 at 6:04 PM,  <jean.bruenn@ip-minds.de> wrote:
>
> Hey,
>
> i tried to revert that patch manually (e.g. switching into the directory,
> vim cx23885-i2c.c, removing the stuff which was added), then "make clean"
> "make distclean" followed by "./build.sh" then make rmmod, then plugged in
> the card, dmesg shows it loaded the card correctly, all fine, then i did
> ./scan Scanlist.txt and i get the same i2c related errors. Did a reboot
> just to verify, still getting those, scan gives no results always "tuning
> failed".
>
> Then i reverted another patch (just to make sure..
> http://git.kernel.org/?p=linux/kernel/git/mchehab/linux-2.6.git;a=commit;h=f4acb3c4ccca74f5448354308f917e87ce83505a)
> - However, this didn't help. So, the problem might be somewhere else.
>
> I did some more research and it seems i'm not the only one with those
> issues, tho nearly none gets answers regarding this trouble, e.g.:
>
> (same card - also expresscard, december 2010)
> http://www.spinics.net/lists/linux-media/msg27042.html
> (not the same card, tho similar error, hvr 1500, i got hvr 1400 - februar
> 2009)
> http://www.linuxtv.org/pipermail/linux-dvb/2009-February/031839.html
>
> I'm running out of ideas where the problem might be located. i also tried
> to switch the firmware by extracting the firmware manually, didn't help.
> There's a low power version and another one available, tried to replace,
> didn't work neither.
>
> So it seems, this driver is broken at least since december 2010. Totally
> weird why there's such a mess, i know for sure that this WAS working.

Doesn't seem weird to me at all.  This is a pretty uncommon card, so
it is entirely possible that many revisions could go by without
someone noticing a regression.  I know for example that the HVR-1500Q
(the US version of that board) was broken for months and nobody
noticed.

Devin

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

Re: WinTV 1400 broken with recent versions?

[jean.bruenn]

> Doesn't seem weird to me at all.  This is a pretty uncommon card, so
> it is entirely possible that many revisions could go by without
> someone noticing a regression.  I know for example that the HVR-1500Q
> (the US version of that board) was broken for months and nobody
> noticed.

Well. How was it solved at the hvr-1500q? :) Any other information i
could provide maybe? 

Re: WinTV 1400 broken with recent versions?

[Devin Heitmueller]

On Fri, Mar 11, 2011 at 6:30 PM,  <jean.bruenn@ip-minds.de> wrote:
>
>> Doesn't seem weird to me at all.  This is a pretty uncommon card, so
>> it is entirely possible that many revisions could go by without
>> someone noticing a regression.  I know for example that the HVR-1500Q
>> (the US version of that board) was broken for months and nobody
>> noticed.
>
> Well. How was it solved at the hvr-1500q? :) Any other information i
> could provide maybe?

It's not clear to me that it ever was resolved for the 1500q.  If the
goal is for it to get fixed, it's usually a matter of getting a sample
unit in the hands of a developer who knows how to debug the issue.  A
common problem is the lack of overlap between people who have the
board versus people who know what to do with it.

Devin

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

Re: WinTV 1400 broken with recent versions?

[Jarod Wilson]

On Mar 11, 2011, at 6:35 PM, Devin Heitmueller wrote:

> On Fri, Mar 11, 2011 at 6:30 PM,  <jean.bruenn@ip-minds.de> wrote:
>> 
>>> Doesn't seem weird to me at all.  This is a pretty uncommon card, so
>>> it is entirely possible that many revisions could go by without
>>> someone noticing a regression.  I know for example that the HVR-1500Q
>>> (the US version of that board) was broken for months and nobody
>>> noticed.
>> 
>> Well. How was it solved at the hvr-1500q? :) Any other information i
>> could provide maybe?
> 
> It's not clear to me that it ever was resolved for the 1500q.  If the
> goal is for it to get fixed, it's usually a matter of getting a sample
> unit in the hands of a developer who knows how to debug the issue.  A
> common problem is the lack of overlap between people who have the
> board versus people who know what to do with it.

The HVR-1500Q I've got works fine with the latest media_tree code, but
admittedly, it sits unused most of the time, so it had been several
months since I last tried it before about a week ago...

(Got it for my thinkpad, planning to just leave it plugged in all the
time, until finding out it stuck out a good inch from the side of the
machine).


-- 
Jarod Wilson
jarod@wilsonet.com

Re: WinTV 1400 broken with recent versions?

[Devin Heitmueller]

On Sat, Mar 12, 2011 at 9:22 PM, Jean-Michel Bruenn
<jean.bruenn@ip-minds.de> wrote:
> So i guess, nobody here can help me to solve that problems?
>
> xc2028 1-0064: i2c output error: rc = -6 (should be 64)
> xc2028 1-0064: -6 returned from send
> xc2028 1-0064: Error -22 while loading base firmware
> xc2028 1-0064: Loading firmware for type=BASE F8MHZ (3), id
> 0000000000000000.
>
> and nobody can tell me, whether thats a firmware problem, or is it an
> i2c problem? At least someone who could explain me, what the above
> errors mean?

It means the i2c bus failed to get an ACK back when talking to the
xc3028.  It could be a number of different things:

* broken cx23885 i2c master implementation
* bug in the xc3028 driver
* screwed up GPIOs causing the xc3028 to be held in reset
* i2c bus wedged

> Also, nobody has any idea what i could try (except for what
> i already did, including reverting patches and downgrading the kernel)?

If you're knowledgeable enough to downgrade the kernel, then your best
bet is to learn how to use git bisect so you can identify exactly
which patch introduced the regression.

> I don't think the card is that uncommon, because i've already seen some
> pages stating problems with that card, at least 3 on this maillinglist
> with exactly the same problem (not to talk about those people, who just
> send the card back and replace it with something else/people who don't
> use maillinglists/bugtracker) however, whether the card is common or
> uncommon is not very helpful/useful, so.. it shouldn't really matter.

Tens of thousands of cards sold.  Three complaints.  That should
indeed give you an idea how few people are using the card under Linux.
 That said, the likelihood of a card getting fixed is largely related
to how popular it is and the probability that some developer who
actually cares has the card.

> heh. I'm about to give up. just a pity because the card wasn't
> cheap when it came out and every page states "supported in linux" which
> seems to be not true anymore. I know that it was, back in 2008 when i
> wrote with S. Toth about it.

Well with Linux and open source in general, you get what you paid for.
 The only reason the card worked at all is because of the thousands of
dollars worth of man-hours that developers like Steve donated to bring
up the board in the first place.  The downside of course is that since
you paid *nothing* for Linux support, you cannot really complain when
it stops working.  If the card is unpopular and therefore there aren't
developers willing to work on it, and you're not knowledgeable enough
to fix it yourself, then you are largely SOL.

> http://git.linuxtv.org/anttip/media_tree.git?a=commit;h=6676237398d0c2e61e5a3a27e0951f60d6ef6fe3
>
> Here's the commit when the card was added. I was using it without any
> trouble around that time. Well if someone knows what i could try (i
> could also hack around in the source if someone tells me what to
> change/where to change) i'd be very happy about ANY information.

Yeah, if you can git bisect to identify which patch broke the support,
Andy can probably offer some idea what is going on.

Devin

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

Re: WinTV 1400 broken with recent versions?

[Jean-Michel Bruenn]

So i guess, nobody here can help me to solve that problems?

xc2028 1-0064: i2c output error: rc = -6 (should be 64)
xc2028 1-0064: -6 returned from send
xc2028 1-0064: Error -22 while loading base firmware
xc2028 1-0064: Loading firmware for type=BASE F8MHZ (3), id
0000000000000000.

and nobody can tell me, whether thats a firmware problem, or is it an
i2c problem? At least someone who could explain me, what the above
errors mean? Also, nobody has any idea what i could try (except for what
i already did, including reverting patches and downgrading the kernel)?
I don't think the card is that uncommon, because i've already seen some
pages stating problems with that card, at least 3 on this maillinglist
with exactly the same problem (not to talk about those people, who just
send the card back and replace it with something else/people who don't
use maillinglists/bugtracker) however, whether the card is common or
uncommon is not very helpful/useful, so.. it shouldn't really matter.

heh. I'm about to give up. just a pity because the card wasn't 
cheap when it came out and every page states "supported in linux" which
seems to be not true anymore. I know that it was, back in 2008 when i
wrote with S. Toth about it.

http://git.linuxtv.org/anttip/media_tree.git?a=commit;h=6676237398d0c2e61e5a3a27e0951f60d6ef6fe3

Here's the commit when the card was added. I was using it without any
trouble around that time. Well if someone knows what i could try (i
could also hack around in the source if someone tells me what to
change/where to change) i'd be very happy about ANY information.

thanks so far.

Re: WinTV 1400 broken with recent versions?

[Jean-Michel Bruenn]

> It means the i2c bus failed to get an ACK back when talking to the
> xc3028.  It could be a number of different things:
> 
> * broken cx23885 i2c master implementation
> * bug in the xc3028 driver
> * screwed up GPIOs causing the xc3028 to be held in reset
> * i2c bus wedged

Ah. Thanks. Now i know what to search for.

> > Also, nobody has any idea what i could try (except for what
> > i already did, including reverting patches and downgrading the kernel)?
> 
> If you're knowledgeable enough to downgrade the kernel, then your best
> bet is to learn how to use git bisect so you can identify exactly
> which patch introduced the regression.

Yup, started to try that yesterday, however, going back from today to
2008 will take some time. I'll let you know if i made any progress.

Thanks so far.

Re: WinTV 1400 broken with recent versions?

[Ryan]

I have a WinTV 1200 card that has this exact issue. Have you made any new
discoveries or progress? Thanks.

Re: WinTV 1400 broken with recent versions?

[Jean-Michel Bruenn]

Hello Ryan,

i'm sorry, no. I was trying to revert patches and downgrading the
kernel, but this was way more work then i thought. 

This problem should happen with a few other cards also, it's not
dedicated to the hvr 1400. 

Jean

On Wed, 20 Apr 2011
06:09:40 +0000 (UTC) Ryan <eekrano@eekrano.com> wrote:

> I have a WinTV 1200 card that has this exact issue. Have you made any new
> discoveries or progress? Thanks.
> 
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: WinTV 1400 broken with recent versions?

[Jean-Michel Bruenn]

Hey,

this one still seems to be broken, with all recent kernels. I wasn't
able to see which patches turned the card into a non-working state. I
know that pretty old kernels (i can't use them otherwise i would have
downgraded) work fine with that card.

For those who are interested in the original issue:
http://www.mail-archive.com/linux-media@vger.kernel.org/msg29147.html

Around 2008-2009 the card made it into linux and was working a few
weeks after these mails:
http://www.linuxtv.org/pipermail/linux-dvb/2008-April/025224.html

I'm curious, anyone else using this card (expresscard) successful? if
yes, with which kernel? Any chance that this is going to be fixed or
should i better buy a new expresscard? Is there some suggestion for a
dvb-t expresscard which will still work in 2-3 years in linux?

Thanks in advance,
Jean