Bugzilla Bug 245106

Hauppauge Nova-T DVB card works just fine with kernel 2.6.25-r7. However, if I
boot kernel 2.6.26 or 2.6.27 and try to watch TV or scan channels, it is locked
to certain channels.

Reproducible: Always

Steps to Reproduce:
1. Boot 2.6.27 kernel
2. Start mythtv-setup
3. Perform channel-scan


Actual Results:  
Channel scan always finds 2-3 channels instead of about 15. The scan tries to
change the frequency but can't.

Expected Results:  
Channel scan should find the full set of channels, as it does under kernel
2.6.25.

theater ~ # emerge --info
Portage 2.2_rc12 (default/linux/x86/2008.0, gcc-4.3.2, glibc-2.8_p20080602-r0,
2.6.25-gentoo-r7 i686)
=================================================================
System uname:
Linux-2.6.25-gentoo-r7-i686-AMD_Athlon-tm-_64_X2_Dual_Core_Processor_5200+-with-glibc2.0
Timestamp of tree: Fri, 31 Oct 2008 22:45:01 +0000
ccache version 2.4 [enabled]
app-shells/bash:     3.2_p39
dev-lang/python:     2.5.2-r8
dev-util/ccache:     2.4-r8
dev-util/cmake:      2.6.2
sys-apps/baselayout: 2.0.0
sys-apps/openrc:     0.3.0-r1
sys-apps/sandbox:    1.2.18.1-r3
sys-devel/autoconf:  2.13, 2.63
sys-devel/automake:  1.5, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10.1-r1
sys-devel/binutils:  2.18-r4
sys-devel/gcc-config: 1.4.0-r4
sys-devel/libtool:   2.2.6a
virtual/os-headers:  2.6.27-r2
ACCEPT_KEYWORDS="x86 ~x86"
CBUILD="i686-pc-linux-gnu"
CFLAGS="-march=opteron -O3 -pipe"
CHOST="i686-pc-linux-gnu"
CONFIG_PROTECT="/etc"
CONFIG_PROTECT_MASK="/etc/ca-certificates.conf /etc/env.d
/etc/eselect/postgresql /etc/fonts/fonts.conf /etc/gconf /etc/gentoo-release
/etc/php/apache2-php5/ext-active/ /etc/php/cgi-php5/ext-active/
/etc/php/cli-php5/ext-active/ /etc/revdep-rebuild /etc/terminfo
/etc/texmf/web2c /etc/udev/rules.d"
CXXFLAGS="-march=opteron -O3 -pipe"
DISTDIR="/usr/portage/distfiles"
FEATURES="ccache distcc distlocks parallel-fetch preserve-libs protect-owned
sfperms strict unmerge-orphans userfetch"
GENTOO_MIRRORS="http://gentoo.mirrors.easynews.com/linux/gentoo/
http://mirror.datapipe.net/gentoo"
LDFLAGS="-Wl,-O1"
MAKEOPTS="-j4"
PKGDIR="/mnt/usbdisk2/packages"
PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress
--force --whole-file --delete --stats --timeout=180 --exclude=/distfiles
--exclude=/local --exclude=/packages"
PORTAGE_TMPDIR="/var/tmp"
PORTDIR="/usr/portage"
PORTDIR_OVERLAY="/root/portage"
SYNC="rsync://rsync.au.gentoo.org/gentoo-portage"
USE="3dnow 3dnowext 7zip X a52 aac acl acpi alsa amd amr audacious audiofile
automount berkdb bzip2 cairo cdr cdrom cli crypt cups curl custom-optimization
dbus dbx dga dhcp dri dv dvb dvbplayer dvd dvdr dvdread dvi encode fam fat
ffmpeg firefox flac fltk fontconfig foomaticdb ftp gd gdbm gif glitz glsa gnome
gnome-print gpm gstreamer gtk gzip h323 hal hddtemp hfs hpn iconv ieee1394 jpeg
jpeg2k lame linuxkeys lirc mad madwifi matroska midi mime mjpeg mmx mmxext mp3
mpeg mpeg2 mudflap mysql ncurses network nfs nls nptl nptlonly nsplugin nvidia
odbc ogg opengl openmp pam pango pcre perl png postgres python quicktime rar
readline rplay slang smp spell sse sse2 ssl startup-notification subtitles svg
sysfs tcpd tga theora threads threadsafe transcode truetype type1 unicode upnp
usb v4l v4l2 vcd vdr vorbis wifi win32codecs wma wmf x264 x86 xorg xosd
xulrunner xv xvid xvmc zip zlib" ALSA_CARDS="ali5451 als4000 atiixp
atiixp-modem bt87x ca0106 cmipci emu10k1 emu10k1x ens1370 ens1371 es1938 es1968
fm801 hda-intel intel8x0 intel8x0m maestro3 trident usb-audio via82xx
via82xx-modem ymfpci" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop
empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mmap_emul
mulaw multi null plug rate route share shm softvol" APACHE2_MODULES="actions
alias auth_basic authn_alias authn_anon authn_dbm authn_default authn_file
authz_dbm authz_default authz_groupfile authz_host authz_owner authz_user
autoindex cache dav dav_fs dav_lock deflate dir disk_cache env expires
ext_filter file_cache filter headers include info log_config logio mem_cache
mime mime_magic negotiation rewrite setenvif speling status unique_id userdir
usertrack vhost_alias" ELIBC="glibc" INPUT_DEVICES="keyboard mouse"
KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001
mtxorb ncurses text" LIRC_DEVICES="streamzap" USERLAND="GNU"
VIDEO_CARDS="nvidia"
Unset:  CPPFLAGS, CTARGET, EMERGE_DEFAULT_OPTS, FFLAGS, INSTALL_MASK, LANG,
LC_ALL, LINGUAS, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS,
PORTAGE_RSYNC_EXTRA_OPTS

1) It's a bit of a mystery to me why you should have a 64-bit -march with a
32-bit OS, but maybe that doesn't matter:

CFLAGS="-march=opteron -O3 -pipe"
CHOST="i686-pc-linux-gnu"

2) Do I infer correctly that the `emerge --info' reflects the working OS,
because of its kernel version of 2.6.25-r7?

3) Can you reproduce the issue with a different (perhaps less complicated) tv
card viewing application? It would be nice to
a) get mythtv out of the loop and
b) collect more information about what really is going on. dmesg output may
help, but if the problem lies in the interaction between userland and kernel
[driver], then the verbose or debugging output of the userland app should prove
useful.

(In reply to comment #1)

> 2) Do I infer correctly that the `emerge --info' reflects the working OS,
> because of its kernel version of 2.6.25-r7?
Yes.

> 3) Can you reproduce the issue with a different (perhaps less complicated) tv
> card viewing application? It would be nice to
> a) get mythtv out of the loop and
> b) collect more information about what really is going on. dmesg output may
> help, but if the problem lies in the interaction between userland and kernel
> [driver], then the verbose or debugging output of the userland app should prove
> useful.
Can you suggest such application? 

Maybe media-tv/tvtime or another simple tv viewing app. Let's ask.

I couldn't get tvtime to work. It expects /dev/video0 device and all I have is
/dev/dvb/... However, I've tried to use dvbtune with different frequencies:

Kernel 2.6.25:
--------------

theater dvb # dvbtune -c 0 -f 177500000
Using DVB card "DiBcom 3000MC/P"
tuning DVB-T (in United Kingdom) to 177500000 Hz
polling....
Getting frontend event
FE_STATUS:
polling....
Getting frontend event
FE_STATUS: FE_HAS_SIGNAL
polling....
^C
theater dvb # dvbtune -c 0 -f 226500   
Using DVB card "DiBcom 3000MC/P"
tuning DVB-T (in United Kingdom) to 226500000 Hz
polling....
Getting frontend event
FE_STATUS:
polling....
Getting frontend event
FE_STATUS: FE_HAS_SIGNAL
polling....
^C

Kernel 2.6.27:
--------------

theater ~ # dvbtune -c 0 -f 177500000
Using DVB card "DiBcom 3000MC/P"
tuning DVB-T (in United Kingdom) to 177500000 Hz
polling....
Getting frontend event
FE_STATUS:
polling....
polling....
polling....
^C
theater ~ # dvbtune -c 0 -f 226500   
Using DVB card "DiBcom 3000MC/P"
tuning DVB-T (in United Kingdom) to 226500000 Hz
polling....
Getting frontend event
FE_STATUS:
polling....
polling....
polling....

So, it looks like with kernel 2.6.27 the card simply can't tune on the
frequency.

(In reply to comment #3)
> Maybe media-tv/tvtime or another simple tv viewing app. Let's ask.
> 

Please attach dmesg output from a booted kernel (doesn't matter which one, just
want to find out more about your hardware and which driver is behind it)

Created an attachment (id=173384) [details]
dmesg of working 2.6.25 kernel

Thanks. Could you also post the output from a non-working kernel, just incase
there are any obvious differences?

Also, the bug summary indicates that this is a PCI device, but your logs
indicate that it is a USB device driven by the dvb-usb driver. Can you clarify?

Created an attachment (id=173831) [details]
dmesg of gentoo-2.6.27-r4 kernel

(In reply to comment #7)
> Thanks. Could you also post the output from a non-working kernel, just incase
> there are any obvious differences?

I've attached dmesg from 2.6.27 kernel, compiled with the same .config as
working 2.6.25.


> Also, the bug summary indicates that this is a PCI device, but your logs
> indicate that it is a USB device driven by the dvb-usb driver. Can you clarify?
>

The Hauppauge DVB card is very special. It exists in both PCI and USB
incarnations, using the same chip. The kernel recognises it as a USB device.

From that dmesg output, it looks like you do not have the driver for the tuner
built: "mt2060_attach: driver disabled by Kconfig"

After 2.6.25 the way tuners modules are selected in kconfig had changed. Could
you please post your kernel config for 2.6.27?

Created an attachment (id=173857) [details]
Kernel config used for 2.6.27

This config file was copied from working 2.6.25 kernel, then I opened make
menuconfig and compiled 2.6.27 kernel.

Can you confirm that the tuner module is loaded when you attempt to do a scan?

You have built the tuner mentioned in the dmesg output as a module, but do not
have CONFIG_MEDIA_ATTACH set in your config, which would stop auto-loading of
the relevant tuner driver.

Ok, switching on CONFIRM_MEDIA_ATTACH solves it.

(In reply to comment #12)
> Can you confirm that the tuner module is loaded when you attempt to do a scan?
> 
> You have built the tuner mentioned in the dmesg output as a module, but do not
> have CONFIG_MEDIA_ATTACH set in your config, which would stop auto-loading of
> the relevant tuner driver.
> 

Sorry, I reported too soon. Unfortunately, due the mistake in grub.conf, I was
still using the old 2.6.25 kernel. The new 2.6.27 kernel still behaves the
same.

I'm attaching the lsmod output of the 2.6.27 kernel.

Created an attachment (id=174695) [details]
lsmod output for 2.6.27 kernel

Please see comment #12. I assume the configuration in comment #11 is still the
one you are using.

Your lsmod shows that you have not loaded your tuner module (mt20xx) and your
config shows that you have not enabled the tuner module auto-loading
(CONFIG_MEDIA_ATTACH), so the behaviour you are seeing is not surprising.

The latest attempt (above) was made after I enabled CONFIG_MEDIA_ATTACH. I'll
try to load mt20xx manually now.

(In reply to comment #16)
> Please see comment #12. I assume the configuration in comment #11 is still the
> one you are using.
> 
> Your lsmod shows that you have not loaded your tuner module (mt20xx) and your
> config shows that you have not enabled the tuner module auto-loading
> (CONFIG_MEDIA_ATTACH), so the behaviour you are seeing is not surprising.
> 

Created an attachment (id=174787) [details]
Output of lsmod

Created an attachment (id=174788) [details]
Kernel config used for 2.6.27

(In reply to comment #16)
> Please see comment #12. I assume the configuration in comment #11 is still the
> one you are using.
> 
> Your lsmod shows that you have not loaded your tuner module (mt20xx) and your
> config shows that you have not enabled the tuner module auto-loading
> (CONFIG_MEDIA_ATTACH), so the behaviour you are seeing is not surprising.
> 
I've manually loaded mt20xx module - it wasn't loaded automatically.
Unfortunately, nothing has changed.
The latest .config and lsmod output are attached.

Hi,
I have the same problem with my Hauppauge WinTV card (driver: cx8800). I use
tvtime to watch TV and didn't have problems with any kernel older than 2.6.26.
With 2.6.26 (and 2.6.27 too) I have the problem that the tv sticks to a certain
channel which cannot be changed using tvtime. However, this is true for 2.6.26
if the driver is built-in the kernel - if I choose M to compile the driver as a
module, I can change channels and everything works fine with tvtime, but NOT
with mencoder (which, like tvtime, worked fine with kernels older than 2.6.26).
 Attached is my emerge --info, as well as my .config files for 2.6.25 and
2.6.26 kernels. I am ready to test different options and software if this yould
help smash the bug, however, I don't have any more TV cards to test.

Created an attachment (id=174884) [details]
emerge --info

Created an attachment (id=174885) [details]
kernel config for 2.6.25-gentoo-r9

Created an attachment (id=174886) [details]
kernel config for 2.6.26-gentoo-r4

(In reply to comment #21)
Please file a separate bug.
It seems you are running quite different hardware/config.

Also, when migrating from an older to a newer kernel, use
"make oldconfig" after copying an old .config file ...

(In reply to comment #11)
> Created an attachment (id=173857) [edit] [details]
> Kernel config used for 2.6.27
> 
> This config file was copied from working 2.6.25 kernel, then I opened make
> menuconfig and compiled 2.6.27 kernel.

I've just noticed that you did not run "make oldconfig".
I would suggest to create a new .config for 2.6.26 by performing
the following steps.

* copy a working .config from linux-2.6.25-* to linux-2.6.26-* folder
* run "make oldconfig" and accept the defaults
* if required, fine tune your config by using "make menuconfig"
* run "make"
* run make "modules_install"
* install/copy the kernel to /boot

(In reply to comment #26)
> (In reply to comment #11)
Altenatively you can select CONFIG_MEDIA_TUNER_CUSTOMIZE and make sure that
CONFIG_MEDIA_TUNER_MT2060 is set.

Hi Alexey,

(In reply to comment #20)
> I've manually loaded mt20xx module - it wasn't loaded automatically.
> Unfortunately, nothing has changed.
> The latest .config and lsmod output are attached.
> 

Could you please attach another dmesg from the kernel built with that latest
config? 

Also can you please try without a tainted kernel. That means a clean boot,
without loading the nvidia drivers. You will need to confirm without a tainted
kernel if we are going to report this bug upstream.

(In reply to comment #25)
> (In reply to comment #21)
> Please file a separate bug.
> It seems you are running quite different hardware/config.
> 
> Also, when migrating from an older to a newer kernel, use
> "make oldconfig" after copying an old .config file ...
> 

make oldconfig is exactly what I did.

I will open another bug.

Alexey, you have indeed found a bug. You have found a configuration which does
not allow you to build your tuner driver.

The workaround is to enable CONFIG_MEDIA_TUNER_CUSTOMISE and then accept the
defaults for options that fall "under" that. This will then allow you to build
the mt2060 driver which is the correct one for your hardware (not the mt20xx
driver, which I wrongly stated earlier).

I will report this upstream...

Reported upstream at http://bugzilla.kernel.org/show_bug.cgi?id=12204

Please let us know if the above workaround doesn't fix things.

(In reply to comment #31)
> Reported upstream at http://bugzilla.kernel.org/show_bug.cgi?id=12204

Thank you so much! Your workaround worked.
Just FYI, I did try using make oldconfig and nv video driver with no success.
Now everything works fine using mt2060.

The fix was merged by Linus for 2.6.28:

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=8b21c1e90e633aee3363e3d87e1a0b829bdb420e

Well spotted! rediffed for next 2.6.27 release as well

*** Bug 254265 has been marked as a duplicate of this bug. ***

(In reply to comment #35)
> *** Bug 254265 has been marked as a duplicate of this bug. ***

ok, since my report 254265 have been merged here, i clarify that i still
observe the bug with gentoo-sources-2.6.28. enabling
CONFIG_MEDIA_TUNER_CUSTOMISE and selecting everything under it doesn't help
either.

fixed in gentoo-sources-2.6.27-r8

(In reply to comment #36)
> (In reply to comment #35)
> > *** Bug 254265 has been marked as a duplicate of this bug. ***
> 
> ok, since my report 254265 have been merged here, i clarify that i still
> observe the bug with gentoo-sources-2.6.28. enabling
> CONFIG_MEDIA_TUNER_CUSTOMISE and selecting everything under it doesn't help
> either.

Assuming that your problem still isn't solved ...
Can you try with tuner(s) compiled as module which is what
you had in your working 2.6.25 config?

Let's continue working on Nick's problem in #254265

(In reply to comment #39)
> Let's continue working on Nick's problem in #254265
> 

OOPS. With kernel 2.6.29 it's broken again. I'm using the same working config
from 2.6.27. I see mt2060 in lsmod. And, once again, my DVB card doesn't tune
on different channels.

Alexey, any chance you could file a new bug for the 2.6.29 regression? This bug
is already quite messy and chances are it is a different problem.
In the new bug report, please attempt to post some details which you think
might be useful based on your experiences here-- for example, .config and
dmesg. Thanks!

Filed Bug 269350.

(In reply to comment #41)
> Alexey, any chance you could file a new bug for the 2.6.29 regression? This bug
> is already quite messy and chances are it is a different problem.
> In the new bug report, please attempt to post some details which you think
> might be useful based on your experiences here-- for example, .config and
> dmesg. Thanks!
>