Mailing List Archive

On 13/06/2023 23:38, John Pilkington wrote:
> On Friday I updated my 'el7' box from the 'elrepo mainline kernel' 6.3.6
> to 6.3.7.   The backend could be started as normal from the Konsole but
> logging stopped at 'starting as the master server' before the Cardutil
> activation.  The frontend couldn't connect, and 'killall mythbackend'
> did nothing.  Eventually I was able to stop it by closing its window.  I
> reported the problem on the elrepo list.
>
> Today 6.3.7 has reached Fedora 37 and I see the same behaviour.  Both
> boxes are fine on 6.3.6
>
> https://www.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.3.7
>
> http://lists.elrepo.org/pipermail/elrepo/2023-June/006513.html

Today's elrepo has kernel-ml 6.3.8. Same problem.

The mythtv build is the last one that worked, from 31 Jan, but current
master in Fedora 37 does much the same.

mythbackend version: HEAD [v32.0-749472ba33] www.mythtv.org

mythbackend --loglevel debug

hangs at

MythBackend: Starting up as the master server.
CardUtil[1]: Start channel 1 on inputid 1

while under 6.3.6 it goes on like this (timestamps edited out)

{{{

MythBackend: Starting up as the master server.
D CardUtil[1]: Start channel 1 on inputid 1
D DTVChan[1](/dev/dvb/adapter0/frontend0): SetChannelByString(1):
Initialize multiplex options m_tunerType:2 mplexid:3
D PortChecker::checkPort(): host localhost port 3306 timeLimit 30000
linkLocalOnly 1
D CardUtil[2]: Start channel 2 on inputid 2
D DTVChan[2](/dev/dvb/adapter1/frontend0): SetChannelByString(2):
Initialize multiplex options m_tunerType:2 mplexid:3
D PortChecker::checkPort(): host localhost port 3306 timeLimit 30000
linkLocalOnly 1
D CardUtil[3]: Start channel 703 on inputid 3
D DTVChan[3](/dev/dvb/adapter2/frontend0): SetChannelByString(703):
Initialize multiplex options m_tunerType:2 mplexid:3
D PortChecker::checkPort(): host localhost port 3306 timeLimit 30000
linkLocalOnly 1
D CardUtil[4]: Start channel 1 on inputid 4
D CardUtil[5]: Start channel 1 on inputid 5

}}}

Any clues? DTV support missing from kernels 6.3.7+ ?

Thanks

>
> John P

_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

Am Do, Jun 15, 2023 at 10:43:43 +0100 schrieb John Pilkington:
>Today's elrepo has kernel-ml 6.3.8. Same problem.

Yes, I have the same problem with my self-made 6.3.7 and 6.3.8 on
Debian/Testing.

The backend hangs in D state and is unkillable.

I’m using DVB-S/T with a card from Digital Devices.

Stephan

--
| If your life was a horse, you'd have to shoot it. |
_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 15/06/2023 12:57, Stephan Seitz wrote:
> Am Do, Jun 15, 2023 at 10:43:43 +0100 schrieb John Pilkington:
>> Today's elrepo has kernel-ml 6.3.8.  Same problem.
>
> Yes, I have the same problem with my self-made 6.3.7 and 6.3.8 on
> Debian/Testing.
>
> The backend hangs in D state and is unkillable.
>
> I’m using DVB-S/T with a card from Digital Devices.
>
>     Stephan

Yes, I get a lot of Trace stuff like this
{{{

[root@HP_Box john]# journalctl -S -8h | grep -v dracut | grep -A 10
"task mythbackend:"
Jun 15 09:34:44 HP_Box kernel: INFO: task mythbackend:2889 blocked for
more than 122 seconds.
Jun 15 09:34:44 HP_Box kernel: Tainted: G E
6.3.8-1.el7.elrepo.x86_64 #1
Jun 15 09:34:44 HP_Box kernel: "echo 0 >
/proc/sys/kernel/hung_task_timeout_secs" disables this message.
Jun 15 09:34:44 HP_Box kernel: task:mythbackend state:D stack:0
pid:2889 ppid:1 flags:0x00004004
Jun 15 09:34:44 HP_Box kernel: Call Trace:
Jun 15 09:34:44 HP_Box kernel: <TASK>
Jun 15 09:34:44 HP_Box kernel: __schedule+0x357/0x9f0
Jun 15 09:34:44 HP_Box kernel: schedule+0x68/0xf0
Jun 15 09:34:44 HP_Box kernel: schedule_preempt_disabled+0x15/0x30
Jun 15 09:34:44 HP_Box kernel: __mutex_lock.isra.0+0x30a/0x6b0
Jun 15 09:34:44 HP_Box kernel: ? __smp_call_single_queue+0x3c/0x50
--
Jun 15 09:36:46 HP_Box kernel: INFO: task mythbackend:2889 blocked for
more than 245 seconds.
Jun 15 09:36:46 HP_Box kernel: Tainted: G E
6.3.8-1.el7.elrepo.x86_64 #1
Jun 15 09:36:46 HP_Box kernel: "echo 0 >
/proc/sys/kernel/hung_task_timeout_secs" disables this message.
Jun 15 09:36:46 HP_Box kernel: task:mythbackend state:D stack:0
pid:2889 ppid:1 flags:0x00004004
Jun 15 09:36:46 HP_Box kernel: Call Trace:
Jun 15 09:36:46 HP_Box kernel: <TASK>
Jun 15 09:36:46 HP_Box kernel: __schedule+0x357/0x9f0
Jun 15 09:36:46 HP_Box kernel: schedule+0x68/0xf0
Jun 15 09:36:46 HP_Box kernel: schedule_preempt_disabled+0x15/0x30
Jun 15 09:36:46 HP_Box kernel: __mutex_lock.isra.0+0x30a/0x6b0
Jun 15 09:36:46 HP_Box kernel: ? __smp_call_single_queue+0x3c/0x50
}}}
_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Thu, Jun 15, 2023 at 01:57:05PM +0200, Stephan Seitz wrote:
> Am Do, Jun 15, 2023 at 10:43:43 +0100 schrieb John Pilkington:
> > Today's elrepo has kernel-ml 6.3.8. Same problem.
>
> Yes, I have the same problem with my self-made 6.3.7 and 6.3.8 on
> Debian/Testing.
>
> The backend hangs in D state and is unkillable.
>
> I’m using DVB-S/T with a card from Digital Devices.
>
> Stephan

Same here:

FC37.

6.3.7-100.fc37.x86_64 mythbackend hangs and is unkillable.
6.3.6-100.fc37.x86_64 works fine

I have an old saa7164 chip / tuner card.

I do get a bunch of odd wayland errors, I would not think those matter since it's the backend job, but they are not happening when it works.

About 100 or so of each of these:

...
Jun 15 19:08:24 umbrella kwin_wayland[2346]: kwin_screencast: Dropping a screencast cursor update because the compositor is slow
...
Jun 15 19:09:39 umbrella kwin_wayland[2346]: kwin_wayland_drm: Atomic commit failed! Permission denied
Jun 15 19:09:39 umbrella kwin_wayland[2346]: kwin_wayland_drm: Presentation failed! Permission denied
...

-- Patrick
_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 16/06/2023 19:30, Patrick Mansfield via mythtv-users wrote:
> On Thu, Jun 15, 2023 at 01:57:05PM +0200, Stephan Seitz wrote:
>> Am Do, Jun 15, 2023 at 10:43:43 +0100 schrieb John Pilkington:
>>> Today's elrepo has kernel-ml 6.3.8. Same problem.
>>
>> Yes, I have the same problem with my self-made 6.3.7 and 6.3.8 on
>> Debian/Testing.
>>
>> The backend hangs in D state and is unkillable.
>>
>> I’m using DVB-S/T with a card from Digital Devices.
>>
>> Stephan
>
> Same here:
>
> FC37.
>
> 6.3.7-100.fc37.x86_64 mythbackend hangs and is unkillable.
> 6.3.6-100.fc37.x86_64 works fine
>
> I have an old saa7164 chip / tuner card.
>
> I do get a bunch of odd wayland errors, I would not think those matter since it's the backend job, but they are not happening when it works.
>
> About 100 or so of each of these:
>
> ...
> Jun 15 19:08:24 umbrella kwin_wayland[2346]: kwin_screencast: Dropping a screencast cursor update because the compositor is slow
> ...
> Jun 15 19:09:39 umbrella kwin_wayland[2346]: kwin_wayland_drm: Atomic commit failed! Permission denied
> Jun 15 19:09:39 umbrella kwin_wayland[2346]: kwin_wayland_drm: Presentation failed! Permission denied
> ...
>
> -- Patrick

I have opened this as a mythtv 'issue' although it is clearly caused by
a change in the kernel. I haven't run much else with the kernels that
give problems in mythtv, so I don't know whether other software is affected.

https://github.com/MythTV/mythtv/issues/761

John

_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 16/06/2023 20:43, John Pilkington wrote:
>
> I have opened this as a mythtv 'issue' although it is clearly caused
> by a change in the kernel.  I haven't run much else with the kernels
> that give problems in mythtv, so I don't know whether other software
> is affected.
>
> https://github.com/MythTV/mythtv/issues/761

It seems that a fix is on the way

https://bbs.archlinux.org/viewtopic.php?id=286421

https://lore.kernel.org/lkml/da5382ad-09d6-20ac-0d53-611594b30861@lio96.de/

Nige.


_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 17/06/2023 07:58, Nigel Jewell wrote:
>
> On 16/06/2023 20:43, John Pilkington wrote:
>>
>> I have opened this as a mythtv 'issue' although it is clearly caused
>> by a change in the kernel.  I haven't run much else with the kernels
>> that give problems in mythtv, so I don't know whether other software
>> is affected.
>>
>> https://github.com/MythTV/mythtv/issues/761
>
> It seems that a fix is on the way
>
> https://bbs.archlinux.org/viewtopic.php?id=286421
>
> https://lore.kernel.org/lkml/da5382ad-09d6-20ac-0d53-611594b30861@lio96.de/
>
> Nige.

Thanks. Good to know. But I've just had a closer look at the kernel
changelog and find 78 instances of 'dvb'. Fingers crossed.

John

_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Sat, Jun 17, 2023 at 6:59?AM Nigel Jewell <nigel.jewell@grufty.co.uk> wrote:

> It seems that a fix is on the way
>
> https://bbs.archlinux.org/viewtopic.php?id=286421
>
> https://lore.kernel.org/lkml/da5382ad-09d6-20ac-0d53-611594b30861@lio96.de/

Well, the immediate proposed fix reverts
a user-after-free fix, which means that
one will once again be vulnerable to
kernel corruption (and with other changes
in 6.3, that is apparently somewhat more
likely, as the nouveau driver encountered,
resulting in various issues including file
system corruption; the underlying bug
resulting in use-after-free had existing
in nouveau for a long time, but 6.3 made
it more likely to be encountered).

The proper fixes (to avoid potential
kernel corruption without causing
cards using the impacted chipset to
not reliably work) will likely come later,
hopefully in the next few weeks to get
into 6.4 and then may be backported.
_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Tue, Jun 13, 2023 at 11:38:08PM +0100, John Pilkington wrote:
> On Friday I updated my 'el7' box from the 'elrepo mainline kernel' 6.3.6 to
> 6.3.7. The backend could be started as normal from the Konsole but logging
> stopped at 'starting as the master server' before the Cardutil activation.
> The frontend couldn't connect, and 'killall mythbackend' did nothing.
> Eventually I was able to stop it by closing its window. I reported the
> problem on the elrepo list.
>
> Today 6.3.7 has reached Fedora 37 and I see the same behaviour. Both boxes
> are fine on 6.3.6
>
> https://www.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.3.7
>
> http://lists.elrepo.org/pipermail/elrepo/2023-June/006513.html

It also failed for me with kernel 6.3.7.

It's now fixed in 6.3.12 / Fedora vmlinuz-6.3.12-100.fc37.x86_64

_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org