X.Org: invoking xinput --disable [ || --enable] <device

Discussion:

X.Org: invoking xinput --disable [ || --enable] <device_xID> kills gdm user session

Cedric Bhihe

2021-04-11 18:39:54 UTC

Hi folks,

[Background info: host OS is Arch linux 5.11.12 with gdm 40.0.1 on xorg]

Â For the past 4 days, I have this weird mix-up between issuing either
of the following cmds in a `tmux` terminal:

Â Â Â Â $ /usr/bin/xinput --disable <xID>
or
Â Â Â $ /usr/bin/xinput --enable <xID>

where <xID> is my Touchpad xorg device ID (=14) on a Dell XPS15,
obtained with
Â Â Â Â $ /usr/bin/xinput --list --short

Issuing either one of the above cmds instantly kills my gdm user
session, along with all that was going on in it. Next it lands me on the
gdm user login frame, where I can start a new user session as if nothing
had happened. Everything else seems completely normal.

The touchpad's driver is: xf86-input-synaptics 1.9.1-2 (could not find a
reference to a recent update)
The xorg-xinput version is 1.6.3-2, upgraded from 1.6.3-1 on 2020.05.19
(way before the issue appeared)
On the other hand, gdm + the linux kernel and headers were upgraded
recently (from /var/log/pacman.log):
Â Â Â - [2021-04-09T08:29:29+0200] [ALPM] upgraded linux (5.11.11.arch1-1
-> 5.11.12.arch1-1)
Â Â - [2021-04-09T08:29:30+0200] [ALPM] upgraded gdm (3.38.2.1-1 -> 40.0-1)

I have looked up FAQs, misc FAQs and extra FAQs and all I could on get
my eyes on at xorg, as well as the knowledge base on StackExchange, in
addition to having scanned the web, by I found absolutely no chatter on
anything similar to that issue.

I'm stumped in part because I've used the above cli cmds for years to
activate/deactivate my laptop's touchpad on the fly (on the same box)
and I never had an issue.

I'm stumped and would be grateful for any pointer.

PS: I have not cross posted on gdm yet. Will do so in a few days only if
needed or recommended by someone deeper than me on the issue.

Peter Hutterer

2021-04-12 02:35:06 UTC

Permalink

Post by Cedric Bhihe
Hi folks,
[Background info: host OS is Arch linux 5.11.12 with gdm 40.0.1 on xorg]
For the past 4 days, I have this weird mix-up between issuing either of the
     $ /usr/bin/xinput --disable <xID>
or
   $ /usr/bin/xinput --enable <xID>
where <xID> is my Touchpad xorg device ID (=14) on a Dell XPS15, obtained
with
     $ /usr/bin/xinput --list --short

First note: you can use the device name, so `xinput disable "my device
name"` - you do not need to use the ID. This has worked for probably a
decade or more now.

Post by Cedric Bhihe
Issuing either one of the above cmds instantly kills my gdm user session,
along with all that was going on in it. Next it lands me on the gdm user
login frame, where I can start a new user session as if nothing had
happened. Everything else seems completely normal.

most likely a crash in the X server, please check your journal for any
backtraces and file an issue against the X server (cc @whot, i.e. me) and
we can have a look.

Cheers,
Peter

Post by Cedric Bhihe
The touchpad's driver is: xf86-input-synaptics 1.9.1-2 (could not find a
reference to a recent update)
The xorg-xinput version is 1.6.3-2, upgraded from 1.6.3-1 on 2020.05.19 (way
before the issue appeared)
On the other hand, gdm + the linux kernel and headers were upgraded recently
- [2021-04-09T08:29:29+0200] [ALPM] upgraded linux (5.11.11.arch1-1 ->
5.11.12.arch1-1)
- [2021-04-09T08:29:30+0200] [ALPM] upgraded gdm (3.38.2.1-1 -> 40.0-1)
I have looked up FAQs, misc FAQs and extra FAQs and all I could on get my
eyes on at xorg, as well as the knowledge base on StackExchange, in addition
to having scanned the web, by I found absolutely no chatter on anything
similar to that issue.
I'm stumped in part because I've used the above cli cmds for years to
activate/deactivate my laptop's touchpad on the fly (on the same box) and I
never had an issue.
I'm stumped and would be grateful for any pointer.
PS: I have not cross posted on gdm yet. Will do so in a few days only if
needed or recommended by someone deeper than me on the issue.

_______________________________________________
***@lists.x.org: X.Org support
Archives: http://lists.freedesktop.org/archives/xorg
Info: https://lists.x.org/mailman/listinfo/xorg
Your subscription address: %(user_address)s

Cedric Bhihe

2021-04-12 08:06:39 UTC

Permalink

@whot:

Thanks for responding so quickly, Peter.
Yes, issuing:
$ xinput --disable 'DLL07BE:01 06CB:7A13 Touchpad'
had the exact same result of crashing X and killing my user session as
`$ xinput --disable 14`.

A journald log for today (with only one trace of X crashing) is at:
Â Â Â Â Â https://paste.c-net.org/ApplauseFootball
($ journalctl -x -o short-full | sed -ne '/Mon 2021-04-12/p' | curl
--upload-file - 'http://paste.c-net.org/')

I did not apply filters (so journald log entries are all there), because
I just don't know (even after consulting `$ man 7
systemd.journal-fields`) which ones are relevant and which aren't.

Cheers,

-cedric

__________________________________________________________________

Post by Peter Hutterer

Post by Cedric Bhihe
Hi folks,
[Background info: host OS is Arch linux 5.11.12 with gdm 40.0.1 on xorg]
Â For the past 4 days, I have this weird mix-up between issuing either of the
Â Â Â Â $ /usr/bin/xinput --disable <xID>
or
Â Â Â $ /usr/bin/xinput --enable <xID>
where <xID> is my Touchpad xorg device ID (=14) on a Dell XPS15, obtained
with
Â Â Â Â $ /usr/bin/xinput --list --short

First note: you can use the device name, so `xinput disable "my device
name"` - you do not need to use the ID. This has worked for probably a
decade or more now.

most likely a crash in the X server, please check your journal for any
we can have a look.
Cheers,
Peter

(5.11.11.arch1-1 ->

Post by Peter Hutterer

Post by Cedric Bhihe
5.11.12.arch1-1)
Â Â - [2021-04-09T08:29:30+0200] [ALPM] upgraded gdm (3.38.2.1-1 -> 40.0-1)
I have looked up FAQs, misc FAQs and extra FAQs and all I could on get my
eyes on at xorg, as well as the knowledge base on StackExchange, in addition
to having scanned the web, by I found absolutely no chatter on anything
similar to that issue.
I'm stumped in part because I've used the above cli cmds for years to
activate/deactivate my laptop's touchpad on the fly (on the same box) and I
never had an issue.
I'm stumped and would be grateful for any pointer.
PS: I have not cross posted on gdm yet. Will do so in a few days only if
needed or recommended by someone deeper than me on the issue.

Cedric Bhihe

2021-04-14 20:58:11 UTC

Permalink

@whot:

Following up on X crashing when invoking `$ xinput --disable
<device_name>`, I tried to isolate the error with:

`$ grep -i -e "xorg\| X" < <(journalctl -xb) | grep -i -e "fail\|error"`

I include what seems to be the relevant section here:
http://paste.c-net.org/ProducedNorway
Lines
Â Â Â "#3Â 0x00005625301421b1 FatalError (Xorg + 0x14c1b1)" Â Â Â just
before 21:43.22 and
Â Â Â "#3Â 0x000055e2520e61b1 FatalError (Xorg + 0x14c1b1)" between
21:43:34 and 22:02:13
correspond to the two times I reproduced the crash on that boot.
Â Â Â Â FatalError (Xorg + 0x14c1b1)
seems to be the indicator for the crash. It led me to the full coredump
trace in systemd available at
http://paste.c-net.org/EscapedElias.

In the past few days I have seen 2 reports appear on that, essentially
from archers running Xorg like me.
- http://paste.c-net.org/BunksTyped
- https://bbs.archlinux.org/viewtopic.php?id=264928

Cheers,

-cedric

__________________________________________________________________

Post by Peter Hutterer

First note: you can use the device name, so `xinput disable "my device
name"` - you do not need to use the ID. This has worked for probably a
decade or more now.

most likely a crash in the X server, please check your journal for any
we can have a look.
Cheers,
Peter

(5.11.11.arch1-1 ->

Post by Peter Hutterer

Peter Hutterer

2021-04-15 02:34:12 UTC

Permalink

please file a gitlab issue against the X server, mailing lists and email in
general are not useful for debugging with logfiles, etc.

Cheers,
Peter

Following up on X crashing when invoking `$ xinput --disable <device_name>`,
`$ grep -i -e "xorg\| X" < <(journalctl -xb) | grep -i -e "fail\|error"`
http://paste.c-net.org/ProducedNorway
Lines
    "#3 0x00005625301421b1 FatalError (Xorg + 0x14c1b1)"     just before
21:43.22 and
    "#3 0x000055e2520e61b1 FatalError (Xorg + 0x14c1b1)" between 21:43:34
and 22:02:13
correspond to the two times I reproduced the crash on that boot.
     FatalError (Xorg + 0x14c1b1)
seems to be the indicator for the crash. It led me to the full coredump
trace in systemd available at
http://paste.c-net.org/EscapedElias.
In the past few days I have seen 2 reports appear on that, essentially from
archers running Xorg like me.
- http://paste.c-net.org/BunksTyped
- https://bbs.archlinux.org/viewtopic.php?id=264928
Cheers,
-cedric
__________________________________________________________________

Post by Peter Hutterer

First note: you can use the device name, so `xinput disable "my device
name"` - you do not need to use the ID. This has worked for probably a
decade or more now.

most likely a crash in the X server, please check your journal for any
we can have a look.
Cheers,
Peter

Post by Cedric Bhihe
The touchpad's driver is: xf86-input-synaptics 1.9.1-2 (could not
find

Post by Peter Hutterer

Post by Cedric Bhihe
reference to a recent update)
The xorg-xinput version is 1.6.3-2, upgraded from 1.6.3-1 on 2020.05.19 (way
before the issue appeared)
On the other hand, gdm + the linux kernel and headers were upgraded recently
- [2021-04-09T08:29:29+0200] [ALPM] upgraded linux

(5.11.11.arch1-1 ->

Post by Peter Hutterer

Post by Cedric Bhihe
5.11.12.arch1-1)
- [2021-04-09T08:29:30+0200] [ALPM] upgraded gdm (3.38.2.1-1 -> 40.0-1)
I have looked up FAQs, misc FAQs and extra FAQs and all I could on get

Post by Peter Hutterer

Post by Cedric Bhihe
eyes on at xorg, as well as the knowledge base on StackExchange, in addition
to having scanned the web, by I found absolutely no chatter on anything
similar to that issue.
I'm stumped in part because I've used the above cli cmds for years to
activate/deactivate my laptop's touchpad on the fly (on the same box) and I
never had an issue.
I'm stumped and would be grateful for any pointer.
PS: I have not cross posted on gdm yet. Will do so in a few days only if
needed or recommended by someone deeper than me on the issue.

Cedric Bhihe

2021-04-15 08:40:51 UTC

Permalink

Peter and all, apologies for the noise.
I had misunderstood Peter's earlier response and thought reporting on
the list was the right thing to do. I filed the issue at
https://gitlab.freedesktop.org/xorg/xserver/-/issues/1162.
Cheers,
-cedric
___________________________________________________________

Post by Peter Hutterer
please file a gitlab issue against the X server, mailing lists and email in
general are not useful for debugging with logfiles, etc.
Cheers,
Peter

Following up on X crashing when invoking `$ xinput --disable <device_name>`,
`$ grep -i -e "xorg\| X" < <(journalctl -xb) | grep -i -e "fail\|error"`
http://paste.c-net.org/ProducedNorway
Lines
Â Â Â "#3Â 0x00005625301421b1 FatalError (Xorg + 0x14c1b1)" Â Â Â just before
21:43.22 and
Â Â Â "#3Â 0x000055e2520e61b1 FatalError (Xorg + 0x14c1b1)" between 21:43:34
and 22:02:13
correspond to the two times I reproduced the crash on that boot.
Â Â Â Â FatalError (Xorg + 0x14c1b1)
seems to be the indicator for the crash. It led me to the full coredump
trace in systemd available at
http://paste.c-net.org/EscapedElias.
In the past few days I have seen 2 reports appear on that, essentially from
archers running Xorg like me.
- http://paste.c-net.org/BunksTyped
- https://bbs.archlinux.org/viewtopic.php?id=264928
Cheers,
-cedric
__________________________________________________________________

Post by Peter Hutterer

First note: you can use the device name, so `xinput disable "my device
name"` - you do not need to use the ID. This has worked for probably a
decade or more now.

most likely a crash in the X server, please check your journal for any

and

Post by Peter Hutterer

Post by Peter Hutterer
we can have a look.
Cheers,
Peter

Post by Cedric Bhihe
The touchpad's driver is: xf86-input-synaptics 1.9.1-2 (could not
find

Post by Peter Hutterer

(5.11.11.arch1-1 ->

Post by Peter Hutterer

Post by Cedric Bhihe
5.11.12.arch1-1)
Â Â - [2021-04-09T08:29:30+0200] [ALPM] upgraded gdm (3.38.2.1-1 -> 40.0-1)
I have looked up FAQs, misc FAQs and extra FAQs and all I could on get

Post by Peter Hutterer