Bug 806558

rpm -q xine-lib xine-lib-extra-freeworld
please.

I don't think the KDE SC update can possibly have anything to do with this problem. It looks like a problem in a component closer to the hardware (kernel, maybe?).

I have dvb-t problems as well. tzap (from dvb-apps) does show signal strength but xine won't connect if I am using kernel-3.3.0-4.fc16.x86_64 . If instead I boot with kernel-3.2.10-3.fc16.x86_64 then I can use xine to watch TV.

Looks very much like a kernel bug then. Juhani Jaakola, are you sure it doesn't work with kernel 3.2.9?

(In reply to comment #1)
> rpm -q xine-lib xine-lib-extra-freeworld
> please.

$ rpm -q xine-lib xine-lib-extra-freeworld
xine-lib-1.1.20.1-1.fc16.i686
package xine-lib-extra-freeworld is not installed

It's "xine-lib-extras-freeworld" with an 's'!

(In reply to comment #4)
> Looks very much like a kernel bug then. Juhani Jaakola, are you sure it doesn't
> work with kernel 3.2.9?

I re-tested Kaffeine with kernels 3.3.0-4.fc16.i686.PAE and 3.2.9-2.fc16.i686.PAE today. They both give error message "No device found".

In fact when I looked in Kaffeine at Television -> Configure Television, it shoud two devices:

Device 1
Name: Afatech AF9013 DVB-T
Device not connected.

Device 2
Name: Afatech AF9013
(other data seems sensible, no error messages)

I removed Device 1, so that only one device is left. Then I try to watch a channel, but I still get error No device found.

By the way, I'm using firmware dvb-usb-af9015.fw version 4.95.0.0

(In reply to comment #6)
> It's "xine-lib-extras-freeworld" with an 's'!

$ rpm -q xine-lib-extras-freeworld
xine-lib-extras-freeworld-1.1.20.1-1.fc16.i686

I tried with kernel 3.2.9-2.fc16.i686.PAE with another DVB USB stick Artec T14BR DVB-T (driver dvb_usb_dib0700). Kaffeine gives error No device found.

When I look at Televison -> Configure Television, it now shows two devices: the old AF9013 and the new Artec T14BR, so the configuration found that device, but playing a channel does not find it!

I tested the following commands with both kernels 3.2.9-2.fc16.i686.PAE and 3.3.0-4.fc16.i686.PAE:

scandvb /usr/share/dvb/dvb-t/fi-Espoo > .xine/channels.conf
xine dvb://MTV3

Both commands worked fine with both kernels!

But Kaffeine complains "No device found" with both kernels.

So I guess it is not a kernel problem? Kaffeine worked fine yesterday, but when I upgraded to latest packages (including KDE 4.8.1) it stopped working. The last time I upgraded this system was two weeks ago, so I guess one of the over 200 packages whose upgrade was pushed public during last two weeks breaks Kaffeine.

That's exactly the problem, there are more than 200 packages in your update, from many different update groups. I don't see how the KDE SC one can possibly cause this, it must be something else.

Can you please attach the list of packages which were updated (the lines of yum.log dated the day of your update)?

Created attachment 572559 [details]
List of updated packages

The machine is question is not physically here now, so I can't provide the yum.log right now. But because I save the output of "yum update" before the y/N prompt to a file so that I can use yumdownloader to save the packages to a USB memory for re-use in other computers, I can provide that list to you. I hope this helps you.

I have a similar problem with kaffeine and an AF9013 DVB-T device which worked fine until today. Kaffeine now tells me "No device found". I am using FC16 with

kaffeine-1.2.2-1.fc16.x86_64
kernel-3.3.0-4.fc16.x86_64
xine-lib-1.1.20.1-1.fc16.x86_64
xine-lib-extras-freeworld-1.1.20.1-1.fc16.x86_64

The updates I installed after the last time using kaffeine for dvb-t without any errors are listed below. As you can see there is no
kernel update included. However, if I boot kernel-3.2.10-3.fc16.x86_64
instead of 3.3.0-4 kaffeine works fine again and I can watch tv.

Mar 24 11:44:31 Updated: audit-libs-2.2-1.fc16.x86_64
Mar 24 11:44:33 Updated: 2:libpng-1.2.48-1.fc16.x86_64
Mar 24 11:44:33 Updated: libjpeg-turbo-1.2.0-1.fc16.x86_64
Mar 24 11:44:35 Updated: 1:cups-libs-1.5.2-6.fc16.x86_64
Mar 24 11:45:09 Updated: selinux-policy-3.10.0-80.fc16.noarch
Mar 24 11:47:23 Updated: selinux-policy-targeted-3.10.0-80.fc16.noarch
Mar 24 11:47:30 Updated: 1:cups-1.5.2-6.fc16.x86_64
Mar 24 11:47:31 Updated: audit-2.2-1.fc16.x86_64
Mar 24 11:47:32 Updated: audit-libs-python-2.2-1.fc16.x86_64
Mar 24 11:47:35 Updated: iproute-2.6.39-5.fc16.x86_64
Mar 24 11:47:37 Updated: lohit-punjabi-fonts-2.5.1-1.fc16.noarch
Mar 24 11:47:38 Updated: dash-0.5.7-1.fc16.x86_64
Mar 24 11:47:40 Updated: libquvi-scripts-0.4.3-1.fc16.noarch
Mar 24 11:47:41 Updated: quvi-0.4.2-1.fc16.x86_64
Mar 24 11:47:44 Updated: lohit-kannada-fonts-2.5.1-1.fc16.noarch
Mar 24 11:47:45 Updated: foomatic-filters-4.0.8-8.fc16.x86_64
Mar 24 11:47:47 Updated: lohit-assamese-fonts-2.5.1-1.fc16.noarch
Mar 24 11:47:49 Updated: lohit-telugu-fonts-2.5.1-1.fc16.noarch
Mar 24 11:47:51 Updated: lohit-bengali-fonts-2.5.1-1.fc16.noarch
Mar 24 11:47:53 Updated: lohit-oriya-fonts-2.5.1-1.fc16.noarch
Mar 24 11:47:55 Updated: lohit-gujarati-fonts-2.5.1-1.fc16.noarch
Mar 24 11:47:56 Updated: audit-libs-2.2-1.fc16.i686
Mar 24 11:47:57 Updated: 2:libpng-1.2.48-1.fc16.i686
Mar 24 11:47:57 Updated: libjpeg-turbo-1.2.0-1.fc16.i686
Mar 24 11:47:58 Updated: 1:cups-libs-1.5.2-6.fc16.i686

Hmmm, selinux-policy… Do you get any AVCs? Can you try with SELinux permissive (su -c "setenforce 0") or disabled?

No,I do not get any AVCs. Also setting SELinux to permissive or disabling it does not have any effect.

Sorry, I made a mistake about the last time I watched TV with kaffeine. This was on March 20, so there are some more updates to be considered. In particular, the new kernel 3.3.0-4 is one of them. The additional updates are

Mar 21 23:00:19 Updated: gd-2.0.35-14.fc16.x86_64
Mar 21 23:00:21 Updated: 1:grub2-1.99-13.fc16.2.x86_64
Mar 21 23:00:25 Updated: GraphicsMagick-1.3.14-1.fc16.x86_64
Mar 21 23:00:26 Updated: gd-2.0.35-14.fc16.i686
Mar 23 00:48:53 Updated: libgcc-4.6.3-2.fc16.x86_64
Mar 23 00:48:55 Updated: bash-4.2.24-1.fc16.x86_64
Mar 23 00:48:55 Updated: xorg-x11-server-common-1.11.4-2.fc16.x86_64
Mar 23 00:48:56 Updated: libquadmath-4.6.3-2.fc16.x86_64
Mar 23 00:48:57 Updated: libstdc++-4.6.3-2.fc16.x86_64
Mar 23 00:48:59 Updated: libuser-0.57.4-1.fc16.x86_64
Mar 23 00:49:00 Updated: libuser-python-0.57.4-1.fc16.x86_64
Mar 23 00:49:01 Updated: exempi-2.2.0-1.fc16.x86_64
Mar 23 00:49:02 Updated: libgfortran-4.6.3-2.fc16.x86_64
Mar 23 00:49:03 Updated: xorg-x11-server-Xephyr-1.11.4-2.fc16.x86_64
Mar 23 00:49:05 Updated: xorg-x11-server-Xorg-1.11.4-2.fc16.x86_64
Mar 23 00:49:06 Updated: kernel-tools-3.3.0-4.fc16.x86_64
Mar 23 00:49:14 Updated: libgcj-4.6.3-2.fc16.x86_64
Mar 23 00:49:45 Updated: google-chrome-stable-17.0.963.83-127885.x86_64
Mar 23 00:49:45 Updated: libgomp-4.6.3-2.fc16.x86_64
Mar 23 00:49:46 Updated: libtool-ltdl-2.4-9.fc16.x86_64
Mar 23 00:49:47 Updated: chkconfig-1.3.59-1.fc16.x86_64
Mar 23 00:49:48 Updated: psmisc-22.16-1.fc16.x86_64
Mar 23 00:49:49 Updated: libgcc-4.6.3-2.fc16.i686
Mar 23 00:49:50 Updated: libstdc++-4.6.3-2.fc16.i686
Mar 23 00:49:51 Updated: libquadmath-4.6.3-2.fc16.i686
Mar 23 00:49:52 Updated: libtool-ltdl-2.4-9.fc16.i686
Mar 23 00:50:03 Installed: kernel-3.3.0-4.fc16.x86_64

Might there be 2 separate bugs? Marco Kuehnel's evidence all points to the kernel as the culprit, but Juhani Jaakola seems to be hitting a different issue (possibly in addition to the kernel bug), maybe in FFmpeg?

In any case, I'm reassigning this to the kernel based on comment #3 and followups.

In comment #3 Michael Young writes that he can't watch DVB-T with Xine with kernel-3.3.0-4.fc16.x86_64

However, as I wrote in comment #10, I can watch DVB-T with Xine without problems with kernel 3.3.0-4.fc16.i686.PAE (and 3.2.9-2.fc16.i686.PAE as well). So xine and scandvb work for me with both kernels. I have 32-bit kernel and Michael has 64-bit kernel - are there other differences, such as weak signal? I have a strong signal.

Also the old program Klear 0.7.1 from fc10 can record programs with kernel 3.3.0-4.fc16.i686.PAE.

And as I wrote in comment #9, Kaffeine detects new DVB-T USB sticks for its configuration, but can not play any programs (gives error No device found). This happens with two different USB DVB-T sticks, with two different drivers (af9013 and dvb_usb_dib0700).

Kaffeine can play programs (.m2t files) which were recorded with Kaffeine when it worked.

I do not have kernel-3.2.10-3 installed now, so I have not tested it.

It's hard for me to think that the problem is in the kernel, because Kaffeine worked with kernel 3.2.9-2.fc16.i686.PAE before the updates but does not work after the updates. If the problem really is in the kernel, there would have to be two changes: one change in 3.2.10-3 which is compatible with the other new updates and another change in 3.3.0-4 which breaks it???

Looks like there are definitely 2 different bugs.

Marco Kuehnel, please file a separate bug against the kernel.

I filed the kernel related bug as Bug 807052.

I have another laptop which has kernel-3.3.0-4.fc16.i686 and Terratec CinergyT2 DVB-T USB stick (driver cinergyT2). On that machine Kaffeine does not give the "No device found" error! Currently I do not have a proper antenna, the signal is too weak to display anything. But Kaffeine behaves correctly without any errors if I try to watch a channel or try to scan channels.

On that machine I have updated only the rpm* packages and kernel 3.3.0-4 lately, so it still has these versions:

kernel-3.3.0-4.fc16.i686
xine-lib-1.1.20.1-1.fc16.i686
xine-lib-extras-freeworld-1.1.20.1-1.fc16.i686
kaffeine-1.2.2-1.fc16.i686
kdelibs-4.7.4-1.fc16.i686
ffmpeg-libs-0.8.9-1.fc16.i686

Command "yum update" lists 31 packages to be installed and 220 packages to be updated, such as ffmpeg-libs-0.8.10-1.fc16. Do you want a list of those packages? Do you want me to try to update only one or some of them?

Well, yes, if you could narrow down the package(s) causing the problem, that'd help a lot. ffmpeg would be one possible suspect.

I have a third laptop as well. I performed the same test as in comment #21 for it as well and I got the same results. After that I updated to ffmpeg-libs-0.8.10-1.fc16 and Kaffeine works with CinergyT2 with it as well. So if CinergyT2 suffers from this problem at all, ffmpeg-libs is not the cause.

Do you have any other suggestions for packages to update? Perhaps kdelibs, because Kaffeine uses libkio?

I updated packages kdelibs and kdelibs-common from 4.7.4-1.fc16.i686 to 4.8.1-2.fc16.i686 on the third laptop and tested with CinergyT2. It still works without any error messages. I'm using kernel-3.3.0-4.fc16.i686. Other parts of KDE are still at level 4.7.4 except kdelibs and kdelibs-common.

Any suggestions which package I should try next?

I guess we now have two possibilities:
- The bug affects only USB DVB-T sticks with drivers af9013 and dvb_usb_dib0700, but not CinergyT2. Unfortunately I do not have other sticks than the CinergyT2 now with me, so I can't compare. OR
- The bug is caused by one of the about 200 packages which I have not yet updated on the second and third laptops. I'll continue updating packages one by one.

There was a large re-work on Kernel 3.3, fixing the DVB API support. Part of the issues can be due to that.

In thesis, they shouldn't be visible to users, but regressions might happen.

Also, there is a special case that can be visible by some:

There are a few devices whose the same frontend can work with more than one different video standard. Those frontends are supported by the drxk and cxd2820 frontend drivers. Both support DVB-T and DVB-C. 

There was a hack on those two specific drivers inside the kernel to expose them as two separate devices, but this were causing troubles with some applications that were expecting them to be independent.

This hack were replaced by a proper support at Kernel v3.3. Applications using DVBv5 API (support added on kernel in 2008) shouldn't have any troubles, but, unfortunately, most applications are still using the legacy DVBv3 API.

With the new support, there's now just one DVB frontend on those devices, and a simple ioctl call is enough to switch it between DVB-T, DVB-T2 and DVB-C.

Applications need to either be ported to fully support the DVBv5 API, or at least, implement the device mode switch.

For applications using the legacy API, there's an userspace tool that allows changing the device mode. It is part of the v4l-utils development branch. Fedora
is using the stable branch. I'll open a BZ with regards to either upgrading it to the development branch or to port the DVB utils there to the stable branch.

AFAIKT, no CinergyT2 or dib0700 devices use those multi-standard frontends.

@Kaffeine maintainers:

A quick fix for Kaffeine would be to do something like this:
    http://linuxtv.org/hg/dvb-apps/rev/0c9932885287

in order to use a multi-standard FE. There is also an enum call that allows
detecting what video standards are supported. The code to get a list of the
standards supported by a given frontend should be done with this code:

struct dtv_properties props;
struct dtv_property dvb_prop[1];
dvb_prop[0].cmd = DTV_ENUM_DELSYS;
props.num = 1;
props.props = dvb_prop;
if (ioctl(fd, FE_GET_PROPERTY, &dtv_prop) == -1)
    return -1;

The number of supported standards will be at:
    dvb_prop[0].u.buffer.len

And the standards, in DVBv5 format, will be at:
    parms->dvb_prop[0].u.buffer.data[i]

If you're willing to add such hack at the Kaffeine code, please remember that
SYS_ISDBT and SYS_CMMB should be handled on legacy code just like DVB-T, as the
Kernel drivers will auto-detect the parameters that are specific to those two
terrestrial standards.

Regards,
Mauro

Thanks Mauro!

I updated my second laptop to the newest packages and CinergyT2 works fine. But my two other USB DVB sticks with drivers af9013 and dvb_usb_dib0700 have problems with kernel 3.3.0-4 - this is understandable based on what Mauro wrote.

But as I wrote earlier (comment #7), I get "No device found" with kernel 3.2.9-2 as well. What explains that? Could using Kaffeine with kernel 3.3.0-4 corrupt Kaffeine television configuration so that it causes error "No device found" even with kernel 3.2.9-2???

According to comment #13 AF9013 works with kernel-3.2.10-3.fc16.x86_64, which is in accordance with what Mauro says. I do not have that kernel.

(In reply to comment #27)
> Thanks Mauro!
> 
> I updated my second laptop to the newest packages and CinergyT2 works fine. But
> my two other USB DVB sticks with drivers af9013 and dvb_usb_dib0700 have
> problems with kernel 3.3.0-4 - this is understandable based on what Mauro
> wrote.

dvb_usb_dib0700 shouldn't be affected. As far as I remember, I tested the new stuff with an ISDB-T dib0700 hardware, back in December when the patches were
sent upstream by me, and I remember I successfully tested the new Kernel+kaffeine on that time.

So, if some Kernel bug is affecting dib0700, it is likely different than what's happening with af9013.

> But as I wrote earlier (comment #7), I get "No device found" with kernel
> 3.2.9-2 as well. What explains that? Could using Kaffeine with kernel 3.3.0-4
> corrupt Kaffeine television configuration so that it causes error "No device
> found" even with kernel 3.2.9-2???

You need to provide us the dmesg with both kernels, in order to see if the devices were properly detected by the Kernel driver or not. It would be great if you could split the issues with dib0700 on a separate BZ, as it doesn't seem to be related with this bz.

Well, this bug is supposed to be about the dib0700 (see the original report), I had Marco Kuehnel file Bug #807052 separately.

(In reply to comment #29)
> Well, this bug is supposed to be about the dib0700 (see the original report), I
> had Marco Kuehnel file Bug #807052 separately.

Well, comment #0 says otherwise:

(In reply to comment #0)
> Additional info:
> I have an af9015 DVB-T tuner.

It was also opened against Kafeine. So, my understanding is that this will eventually result on a kaffeine fix to solve the issue with the af9015 cxd2820 multi-delivery-system frontend.

Am I missing something?

Created attachment 573428 [details]
dmesg after inserting an af9015 with kernel 3.2.9-2.fc16.i686.PAE

Created attachment 573429 [details]
dmesg after inserting an af9015 with kernel 3.3.0-4.fc16.i686.PAE

Created attachment 573434 [details]
dmesg after inserting an Artec stick (dvb0700) with kernel 3.2.9-2.fc16.i686.PAE

Unfortunately I have not saved dmesg for this stick with kernel 3.3.0

(In reply to comment #28)
...
> > But as I wrote earlier (comment #7), I get "No device found" with kernel
> > 3.2.9-2 as well. What explains that? Could using Kaffeine with kernel 3.3.0-4
> > corrupt Kaffeine television configuration so that it causes error "No device
> > found" even with kernel 3.2.9-2???
> 
> You need to provide us the dmesg with both kernels, in order to see if the
> devices were properly detected by the Kernel driver or not. It would be great
> if you could split the issues with dib0700 on a separate BZ, as it doesn't seem
> to be related with this bz.

I have now attached some dmesg's.

Regarding comment #29, I created this bug against Kaffeine. Then I tried to narrow the bug in any way that I could imagine. Originally I used an AF9015 stick but I happened to have a dib0700 stick and a CinergyT2 as well, so I just thought that they would give us further information to help us analyze this bug.

(In reply to comment #34)
> (In reply to comment #28)
> ...
> > > But as I wrote earlier (comment #7), I get "No device found" with kernel
> > > 3.2.9-2 as well. What explains that? Could using Kaffeine with kernel 3.3.0-4
> > > corrupt Kaffeine television configuration so that it causes error "No device
> > > found" even with kernel 3.2.9-2???
> > 
> > You need to provide us the dmesg with both kernels, in order to see if the
> > devices were properly detected by the Kernel driver or not. It would be great
> > if you could split the issues with dib0700 on a separate BZ, as it doesn't seem
> > to be related with this bz.
> 
> I have now attached some dmesg's.

Thanks!

From the logs there, none of those have a multi-delivery system frontend. 

(In reply to comment #32)
> Created attachment 573429 [details]
> dmesg after inserting an af9015 with kernel 3.3.0-4.fc16.i686.PAE

The Kernel drivers were properly loaded.

Could you please check if udev has created the dvb device drivers, under /dev/dvb?

It also helps if you can test it with some other tools, and see if the DVB device is working there.

The better is to use the dvbv5 utils. As they're not yet on Fedora, you'll need to compile them from source. You need to install at least git, gcc, g++ and make, in order to compile it from source:

$ git clone git://linuxtv.org/v4l-utils.git
$ cd v4l-utils
$ autoreconf -vfi && ./configure
$ cd utils/dvb
$ sudo make install

You can now use dvbv5-scan to scan DVB channels, and dvbv5-zap to tune.

dvbv5-scan requires a transponders config file, and generates a channel file used by dvbv5-zap. Those tools will use DVBv5 API.

You can also use dvbscan/dvbzap/gnutv as provided at dvb-utils package, in order to test the device, using the DVBv3 API.

(In reply to comment #35)
>...
> It also helps if you can test it with some other tools, and see if the DVB
> device is working there.
>...

Please see comment #18 (and comment #10). So scandvb, xine and klear work.

I installed Fedora 17 to a PC and tried Kaffeine. It works now! I could scan the channels and watch DVB-T!

My versions are:

kaffeine-1.2.2-1.fc16.i686
kernel-3.3.2-1.fc17.i686

I'm using Artec T14BR DVB-T USB stick, which uses driver dvb_usb_dib0700.

Now I could test this with the original machine. Now Kaffeine works!

My versions are:

kaffeine-1.2.2-1.fc16.i686
kernel-PAE-3.3.2-1.fc16.i686