You are not logged in.

#76 2019-09-11 14:25:32

cx
Member
Registered: 2019-08-17
Posts: 60

Re: [pending] Desktop freezing occasionally (pentium4)

... and now I just got kicked right back to the beginning: Caught another freeze when composing a simple mail in thunderbird; at least I was able to send it ...

This is really close to unusable in X, and it always got to with text input! Either it’s a bug in libinput, or for this old keyboard I’m still missing something (drivers/modules), or both. – What makes me wonder that all seemed to work until midst of August.

For anyone interested, the pentium4 is a ThinkPad R51e.

Now I’m gonna look again for specific official or AUR packages, though while I already found the first being generally too modern, the latter often don’t build anymore due to outdated or missing sources – whereas actually I’d expect libinput yet to be smart enough~

Offline

#77 2019-09-14 07:54:20

andreas_baumann
Administrator
From: Zurich, Switzerland
Registered: 2017-08-10
Posts: 817
Website

Re: [pending] Desktop freezing occasionally (pentium4)

I can now defnitely confirm SSE2 related trouble in rust (librsvg and thunderbird/firefox) when using a CPU without SSE2 and i686 (for instance
my lxpanel just crashes in librsvg2 and restarts when moving the mouse over a menu with an SVG icon). But this is on a Pentium III Copermine
without SSE2. Your laptop should have a CPU with SSE2 (you can check with 'grep flags /proc/cpuinfo'), so my problems are not your problems, it seems. :-)

Offline

#78 2019-09-14 18:22:24

levi
Moderator
From: Yorkshire, UK
Registered: 2018-06-16
Posts: 462

Re: [pending] Desktop freezing occasionally (pentium4)

According to the thinkpad wiki linked by cx above, both possible cpu configurations have SSE2 included.  Still worth checking if lscpu* reckons it's compatible of course.

*Or your grepping of /proc/cpuinfo.  Looks the same except it exposes the flags of both hyper-threaded cores on my system, while lscpu consolidates them into one chip with two cores (kind of).


Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.

Offline

#79 2019-09-18 11:19:09

cx
Member
Registered: 2019-08-17
Posts: 60

Re: [pending] Desktop freezing occasionally (pentium4)

Hi guys – I’ve been just too busy with real life things lately, sorry for the late answer!

Anyways, here you are:

$ grep flags /proc/cpuinfo
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov clflush dts acpi mmx fxsr sse sse2 ss tm pbe bts cpuid

The ad hoc partial desktop freezes persist. And I still think libinput is the problem. Maybe I should send a bug report to freedesktop.org.

Offline

#80 2019-09-18 14:23:58

levi
Moderator
From: Yorkshire, UK
Registered: 2018-06-16
Posts: 462

Re: [pending] Desktop freezing occasionally (pentium4)

I've not heard that anyone's definitively reproduced this problem yet.  The freedesktop guys will probably say this is down to your specific configuration and knock you back with a 'cannot reproduce' or something equivalent, esepecially with no error logs indicating where it's failed.

I wonder if it's possible to kill and restart libinput via the ssh connection.  Then you'll have it running in a terminal so that absolutely everything it prints out will go to the two standard output streams.  If it is an invalid opcode that should at the very least result in a termination message from the kernel I'd have thought.

Only problem is I'm not sure how libinput acutally runs.  The name makes it sound like a library, but there is a binary in /usr/bin (which isn't currently running according to ps -a).  One thing that might be worth trying is to run 'libinput -debug-events' either over ssh or in a spare terminal that you can always get to, which records all valid happenings in libinput at least.


Here are my flags for comparison.  I've not been able to crash or freeze libinput fwiw:

Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx constant_tsc arch_perfmon pebs bts cpuid aperfmperf pni dtes64 monitor ds_cpl est tm2 ssse3 xtpr pdcm movbe lahf_lm dtherm

The ones I have on this Atom system that your P4 doesn't are: nx, constant_tsc, arch_perfmon, pebs, aperfmon, pni, dtest64, monitor, ds_cpi, est, tm2, ssse3, xptr, pdcm, movbe, lahf_im and dtherm.  I don't know what all of those do to be honest apart from ssse3, movbe and nx.


Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.

Offline

#81 2019-09-18 22:28:23

cx
Member
Registered: 2019-08-17
Posts: 60

Re: [pending] Desktop freezing occasionally (pentium4)

Good idea! Well, I used libinput debug-gui so far, which didn’t look quite fruitful (except I wondered the drawn lines by mouse appeared with a notable x/y-offset):

# libinput debug-events --verbose
event2  - Power Button: is tagged by udev as: Keyboard
event2  - Power Button: device is a keyboard
event3  - Video Bus: is tagged by udev as: Keyboard
event3  - Video Bus: device is a keyboard
event0  - Lid Switch: is tagged by udev as: Switch
event0  - Lid Switch: device is a switch device
event1  - Sleep Button: is tagged by udev as: Keyboard
event1  - Sleep Button: device is a keyboard
event5  - SIGMACHIP Usb Mouse: is tagged by udev as: Mouse
event5  - SIGMACHIP Usb Mouse: device is a pointer
event4  - AT Translated Set 2 keyboard: is tagged by udev as: Keyboard
event4  - AT Translated Set 2 keyboard: device is a keyboard
event0  - lid: keyboard paired with Lid Switch<->AT Translated Set 2 keyboard
event7  - TPPS/2 IBM TrackPoint: is tagged by udev as: Mouse Pointingstick
event7  - TPPS/2 IBM TrackPoint: device is a pointer
event6  - PC Speaker: not tagged as supported input device
event6  - not using input device '/dev/input/event6'
event8  - ThinkPad Extra Buttons: is tagged by udev as: Keyboard
event8  - ThinkPad Extra Buttons: device is a keyboard
-event2   DEVICE_ADDED     Power Button                      seat0 default group1  cap:k
-event3   DEVICE_ADDED     Video Bus                         seat0 default group2  cap:k
-event0   DEVICE_ADDED     Lid Switch                        seat0 default group3  cap:S
-event1   DEVICE_ADDED     Sleep Button                      seat0 default group4  cap:k
-event5   DEVICE_ADDED     SIGMACHIP Usb Mouse               seat0 default group5  cap:p left scroll-nat scroll-button
-event4   DEVICE_ADDED     AT Translated Set 2 keyboard      seat0 default group6  cap:k
-event7   DEVICE_ADDED     TPPS/2 IBM TrackPoint             seat0 default group7  cap:p left scroll-nat scroll-button
-event8   DEVICE_ADDED     ThinkPad Extra Buttons            seat0 default group8  cap:k
-event4   KEYBOARD_KEY      +4.96s	*** (-1) pressed
 event4   KEYBOARD_KEY      +5.03s	*** (-1) pressed
 event4   KEYBOARD_KEY      +5.10s	*** (-1) released
 event4   KEYBOARD_KEY      +5.20s	*** (-1) released
 event4   KEYBOARD_KEY     +19.60s	*** (-1) pressed
 event4   KEYBOARD_KEY     +19.70s	*** (-1) pressed
 event4   KEYBOARD_KEY     +19.73s	*** (-1) released
 event4   KEYBOARD_KEY     +19.80s	*** (-1) released
 event4   KEYBOARD_KEY     +20.35s	*** (-1) pressed
 event4   KEYBOARD_KEY     +20.51s	*** (-1) released
-event5   POINTER_MOTION   +35.78s	 -0.95/  0.32 ( -3.00/ +1.00)
 event5   POINTER_MOTION   +35.79s	  5.72/  0.00 ( +6.00/ +0.00)
 event5   POINTER_MOTION   +35.80s	 21.24/ -3.54 (+12.00/ -2.00)
 event5   POINTER_MOTION   +35.81s	 26.00/ -4.00 (+13.00/ -2.00)
.
.
.
 event5   POINTER_MOTION   +36.67s	 -0.55/ -0.55 ( -1.00/ -1.00)
 event5   POINTER_MOTION   +36.77s	 -0.90/  0.45 ( -2.00/ +1.00)
 event5   POINTER_MOTION   +36.79s	 -0.82/  0.82 ( -1.00/ +1.00)
 event5   POINTER_MOTION   +36.80s	 -1.00/  1.00 ( -1.00/ +1.00)
 event5   POINTER_MOTION   +36.81s	 -1.00/  2.00 ( -1.00/ +2.00)
event5  - debounce state: DEBOUNCE_STATE_IS_UP → DEBOUNCE_EVENT_OTHERBUTTON → DEBOUNCE_STATE_IS_UP
event5  - debounce state: DEBOUNCE_STATE_IS_UP → DEBOUNCE_EVENT_PRESS → DEBOUNCE_STATE_IS_DOWN_WAITING
 event5   POINTER_BUTTON   +37.49s	BTN_LEFT (272) pressed, seat count: 1
event5  - debounce state: DEBOUNCE_STATE_IS_DOWN_WAITING → DEBOUNCE_EVENT_TIMEOUT → DEBOUNCE_STATE_IS_DOWN
event5  - debounce state: DEBOUNCE_STATE_IS_DOWN → DEBOUNCE_EVENT_RELEASE → DEBOUNCE_STATE_IS_UP_DETECTING_SPURIOUS
 event5   POINTER_BUTTON   +37.56s	BTN_LEFT (272) released, seat count: 0
event5  - debounce state: DEBOUNCE_STATE_IS_UP_DETECTING_SPURIOUS → DEBOUNCE_EVENT_TIMEOUT_SHORT → DEBOUNCE_STATE_IS_UP_WAITING
event5  - debounce state: DEBOUNCE_STATE_IS_UP_WAITING → DEBOUNCE_EVENT_TIMEOUT → DEBOUNCE_STATE_IS_UP
-event4   KEYBOARD_KEY     +43.52s	*** (-1) pressed
 event4   KEYBOARD_KEY     +43.59s	*** (-1) pressed
 event4   KEYBOARD_KEY     +43.65s	*** (-1) released
 event4   KEYBOARD_KEY     +43.71s	*** (-1) released
 event4   KEYBOARD_KEY     +43.85s	*** (-1) pressed
 event4   KEYBOARD_KEY     +43.92s	*** (-1) pressed
 event4   KEYBOARD_KEY     +43.96s	*** (-1) released
 event4   KEYBOARD_KEY     +44.04s	*** (-1) released
 event4   KEYBOARD_KEY     +47.83s	*** (-1) pressed
 event4   KEYBOARD_KEY     +47.92s	*** (-1) pressed
 event4   KEYBOARD_KEY     +47.98s	*** (-1) released
 event4   KEYBOARD_KEY     +48.07s	*** (-1) released
-event5   POINTER_MOTION   +50.62s	  0.89/  1.77 ( +1.00/ +2.00)
-event4   KEYBOARD_KEY     +60.87s	KEY_WAKEUP (143) pressed
 event4   KEYBOARD_KEY     +60.88s	KEY_WAKEUP (143) released
 event4   KEYBOARD_KEY     +69.41s	*** (-1) pressed
 event4   KEYBOARD_KEY     +69.63s	*** (-1) pressed
 event4   KEYBOARD_KEY     +69.75s	*** (-1) released
 event4   KEYBOARD_KEY     +69.79s	*** (-1) released
 event4   KEYBOARD_KEY     +70.34s	*** (-1) pressed
 event4   KEYBOARD_KEY     +70.46s	*** (-1) pressed
 event4   KEYBOARD_KEY     +70.64s	*** (-1) released
 event4   KEYBOARD_KEY     +70.64s	*** (-1) released
-event5   POINTER_MOTION   +78.95s	 -0.32/  0.00 ( -1.00/ +0.00)
 event5   POINTER_MOTION   +78.96s	 -0.84/  0.00 ( -1.00/ +0.00)
 event5   POINTER_MOTION   +78.97s	 -3.00/  1.00 ( -3.00/ +1.00)
 event5   POINTER_MOTION   +78.98s	 -3.00/  1.00 ( -3.00/ +1.00)
 event5   POINTER_MOTION   +78.98s	 -1.00/  1.00 ( -1.00/ +1.00)
 event5   POINTER_MOTION   +78.99s	 -2.00/  1.00 ( -2.00/ +1.00)
 event5   POINTER_MOTION   +79.00s	 -2.00/  1.00 ( -2.00/ +1.00)
 event5   POINTER_MOTION   +79.01s	 -4.17/  2.09 ( -4.00/ +2.00)
 event5   POINTER_MOTION   +79.02s	 -3.24/  0.00 ( -3.00/ +0.00)
 event5   POINTER_MOTION   +79.02s	 -4.22/  1.05 ( -4.00/ +1.00)
 event5   POINTER_MOTION   +79.03s	 -7.66/  2.55 ( -6.00/ +2.00)
 event5   POINTER_MOTION   +79.04s	-10.36/  1.48 ( -7.00/ +1.00)
 event5   POINTER_MOTION   +79.05s	 -8.79/  1.47 ( -6.00/ +1.00)
 event5   POINTER_MOTION   +79.06s	 -8.37/  1.40 ( -6.00/ +1.00)
 event5   POINTER_MOTION   +79.06s	-12.20/  0.00 ( -8.00/ +0.00)
 event5   POINTER_MOTION   +79.07s	-11.14/  0.00 ( -7.00/ +0.00)
 event5   POINTER_MOTION   +79.08s	-12.73/  0.00 ( -8.00/ +0.00)
 event5   POINTER_MOTION   +79.09s	-11.17/ -1.60 ( -7.00/ -1.00)
 event5   POINTER_MOTION   +79.10s	-15.00/  0.00 ( -9.00/ +0.00)
 event5   POINTER_MOTION   +79.10s	-16.19/  0.00 ( -9.00/ +0.00)
 event5   POINTER_MOTION   +79.11s	-16.19/  0.00 ( -9.00/ +0.00)
 event5   POINTER_MOTION   +79.12s	-16.18/  0.00 ( -9.00/ +0.00)
 event5   POINTER_MOTION   +79.13s	-16.17/  0.00 ( -9.00/ +0.00)
 event5   POINTER_MOTION   +79.14s	-13.83/  0.00 ( -8.00/ +0.00)
 event5   POINTER_MOTION   +79.14s	 -7.27/  0.00 ( -5.00/ +0.00)
 event5   POINTER_MOTION   +79.15s	 -7.90/  0.00 ( -6.00/ +0.00)
 event5   POINTER_MOTION   +79.16s	 -4.99/  0.00 ( -4.00/ +0.00)
 event5   POINTER_MOTION   +79.17s	 -4.47/ -1.12 ( -4.00/ -1.00)
 event5   POINTER_MOTION   +79.18s	 -3.18/ -1.06 ( -3.00/ -1.00)
 event5   POINTER_MOTION   +79.18s	 -2.00/ -1.00 ( -2.00/ -1.00)
 event5   POINTER_MOTION   +79.19s	 -1.00/ -1.00 ( -1.00/ -1.00)
 event5   POINTER_MOTION   +79.32s	 -0.90/ -0.90 ( -1.00/ -1.00)
 event5   POINTER_MOTION   +79.33s	 -0.90/ -0.90 ( -1.00/ -1.00)
 event5   POINTER_MOTION   +79.34s	 -2.00/ -1.00 ( -2.00/ -1.00)
 event5   POINTER_MOTION   +79.35s	 -1.00/ -2.00 ( -1.00/ -2.00)
 event5   POINTER_MOTION   +79.36s	 -1.00/ -1.00 ( -1.00/ -1.00)
 event5   POINTER_MOTION   +79.69s	  0.89/  0.00 ( +1.00/ +0.00)
 event5   POINTER_MOTION   +79.70s	  0.85/  0.00 ( +1.00/ +0.00)
 event5   POINTER_MOTION   +79.70s	  2.00/  0.00 ( +2.00/ +0.00)
 event5   POINTER_MOTION   +79.71s	  5.34/  0.00 ( +5.00/ +0.00)
 event5   POINTER_MOTION   +79.72s	  8.03/  2.68 ( +6.00/ +2.00)
 event5   POINTER_MOTION   +79.73s	 12.40/  1.55 ( +8.00/ +1.00)
 event5   POINTER_MOTION   +79.74s	 11.47/  4.91 ( +7.00/ +3.00)
 event5   POINTER_MOTION   +79.74s	 13.20/  3.30 ( +8.00/ +2.00)
 event5   POINTER_MOTION   +79.75s	 11.28/  1.61 ( +7.00/ +1.00)
 event5   POINTER_MOTION   +79.76s	 17.64/  5.29 (+10.00/ +3.00)
 event5   POINTER_MOTION   +79.77s	 19.68/  1.97 (+10.00/ +1.00)
 event5   POINTER_MOTION   +79.78s	 21.89/  3.98 (+11.00/ +2.00)
 event5   POINTER_MOTION   +79.78s	 19.99/  6.00 (+10.00/ +3.00)
 event5   POINTER_MOTION   +79.79s	 23.98/  4.00 (+12.00/ +2.00)
 event5   POINTER_MOTION   +79.80s	 17.74/  3.94 ( +9.00/ +2.00)
 event5   POINTER_MOTION   +79.81s	  9.77/  3.26 ( +6.00/ +2.00)
 event5   POINTER_MOTION   +79.82s	  5.21/  2.60 ( +4.00/ +2.00)
 event5   POINTER_MOTION   +79.87s	 -0.98/  1.96 ( -1.00/ +2.00)
 event5   POINTER_MOTION   +79.89s	 -1.90/  0.95 ( -2.00/ +1.00)
-event4   KEYBOARD_KEY     +82.92s	*** (-1) pressed
 event4   KEYBOARD_KEY     +83.01s	*** (-1) pressed
 event4   KEYBOARD_KEY     +83.07s	*** (-1) released
 event4   KEYBOARD_KEY     +83.17s	*** (-1) released
.
.
.
^C
event2  - Power Button: device removed
event3  - Video Bus: device removed
event0  - Lid Switch: device removed
event1  - Sleep Button: device removed
event5  - SIGMACHIP Usb Mouse: device removed
event4  - AT Translated Set 2 keyboard: device removed
event7  - TPPS/2 IBM TrackPoint: device removed
event8  - ThinkPad Extra Buttons: device removed

So the the lines at 60.87+s are the events with which I can so far reproduce a freeze – frequently caused by pressing the <Fn> key. Maybe it’s a wrong keyboard mapping here, as this key has been recorded as "KEY_WAKEUP (143)". – But how about the other (random) keys?

And the recording continues with other letter keys I pressed right after the freeze but don’t appear on the screen, while mouse actions seem normal, despite the "debounce states" when clicking.



levi wrote:

Only problem is I'm not sure how libinput acutally runs.  The name makes it sound like a library, but there is a binary in /usr/bin (which isn't currently running according to ps -a).

Good question. And you’re right: (/usr/bin/)libinput can’t be killed, at least not by kill(all) – no such process. pkill libinput shows no error, but probably has no effect either. – Interesting (contemporary) marginal observation: Running a new X session, and executing pkill libinput in a vte, when pressing <Fn>, the desktop does not freeze at once, but later.



Now I just plugged in an external keyboard, a pretty standard one, that doesn’t have an extra <Fn> key. I’d say, if no more freezes appear, that’d be an indicator for a hardware peculiarity of that machine, rather than a libinput problem. On the other side, the problem exists on X desktops (and not on console), and even there were no such problems until midst of August (after an update).

Also, when pressing the machine’s <Fn> key, even the external keyboard input can’t be used anymore.

Since xev doesn’t catch the <Fn> key (and xbindkey --key crashes) and hence can’t be disabled in X, a quick fix could be to physically disconnect that key from the keyboard ...

Last edited by cx (2019-09-19 11:41:55)

Offline

#82 2019-09-19 01:14:40

levi
Moderator
From: Yorkshire, UK
Registered: 2018-06-16
Posts: 462

Re: [pending] Desktop freezing occasionally (pentium4)

On my machine, 'sudo libinput debug-events' doesn't pick up the Fn button press.  Perhaps this keyboard handles Fn internally, or maybe it's handled so close to the metal inside a kernel module or something*.  But in any case, it probably shouldn't be your wakeup key.

FWIW those other random keypresses (the ones assigned to -1) could be presses of the modifier keys - my system reports shift, ctrl and alt as -1, although the home key (left meta), right alt and my menu key (COMPOSE_KEY) all have names.

* I don't think it needs to be that close to the metal actually.  The key labelled caps lock on my keyboard which I redefined as a spare ctrl key, and I suspect caps lock is named, and if so it's inspecting events after that layer.  I forget exactly how I redefined what my caps lock key though.


Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.

Offline

#83 2019-09-19 11:39:24

cx
Member
Registered: 2019-08-17
Posts: 60

Re: [pending] Desktop freezing occasionally (pentium4)

levi wrote:

I've not heard that anyone's definitively reproduced this problem yet.  The freedesktop guys will probably say this is down to your specific configuration and knock you back with a 'cannot reproduce' or something equivalent, esepecially with no error logs indicating where it's failed.

There are similar symptoms with partial desktop/input freezes where libinput is involved, which then later miraculously vanished or were probably touchpad or GPU driver problems – while I only have a trackpoint and no touchpad, and IIRC have no other choice than using radeon.


levi wrote:

On my machine, 'sudo libinput debug-events' doesn't pick up the Fn button press.  Perhaps this keyboard handles Fn internally, or maybe it's handled so close to the metal inside a kernel module or something*.  But in any case, it probably shouldn't be your wakeup key.

What do you mean by "close to the metal inside a kernel module" exactly?


I wonder why, as shown in this post, right 11 seconds after libinput laoding the TP Extra Buttons several devices get removed again, /var/log/Xorg.0.log:

...
[  7175.088] (II) Using input driver 'libinput' for 'ThinkPad Extra Buttons'
[  7175.088] (**) ThinkPad Extra Buttons: always reports core events
[  7175.088] (**) Option "Device" "/dev/input/event7"
[  7175.088] (**) Option "_source" "server/udev"
[  7175.091] (II) event7  - ThinkPad Extra Buttons: is tagged by udev as: Keyboard
[  7175.091] (II) event7  - ThinkPad Extra Buttons: device is a keyboard
[  7175.091] (II) event7  - ThinkPad Extra Buttons: device removed
[  7175.096] (**) Option "config_info" "udev:/sys/devices/platform/thinkpad_acpi/input/input7/event7"
[  7175.096] (II) XINPUT: Adding extended input device "ThinkPad Extra Buttons" (type: KEYBOARD, id 12)
[  7175.096] (**) Option "xkb_model" "pc105"
[  7175.096] (**) Option "xkb_layout" "de"
[  7175.098] (II) event7  - ThinkPad Extra Buttons: is tagged by udev as: Keyboard
[  7175.099] (II) event7  - ThinkPad Extra Buttons: device is a keyboard
[  7186.598] (II) event2  - Power Button: device removed
[  7186.606] (II) event3  - Video Bus: device removed
[  7186.626] (II) event1  - Sleep Button: device removed
[  7186.636] (II) event5  - SIGMACHIP Usb Mouse: device removed
[  7186.653] (II) event4  - AT Translated Set 2 keyboard: device removed
[  7186.663] (II) event8  - TPPS/2 IBM TrackPoint: device removed
[  7186.683] (II) event7  - ThinkPad Extra Buttons: device removed
[  7186.692] (II) AIGLX: Suspending AIGLX clients for VT switch

Regarding missing errors, how about pressing e.g. the <Fn> (addressed w/ "KEY_WAKEUP") key would just make libinput think it should shut down itself, similar as it does when closing X?

[  8719.386] (II) UnloadModule: "libinput"
[  8719.387] (II) UnloadModule: "libinput"
[  8719.387] (II) UnloadModule: "libinput"
[  8719.387] (II) UnloadModule: "libinput"
[  8719.387] (II) UnloadModule: "libinput"
[  8719.387] (II) UnloadModule: "libinput"
[  8719.388] (II) UnloadModule: "libinput"
[  8719.439] (II) Server terminated successfully (0). Closing log file.

Unfortunately, the above "UnloadModule"s does not mean, libinput is a (kernel) module – it’s not listed by lsmod. So manually controlling libinput still seems out of scope. – Or is there an alternative listing/monitoring app for X?



Once I got tpb installed, what didn’t make a difference. And as I thought it possibly cause a driver clash responsible for the freezes, I uninstalled it again.

What I can else do is maybe make a BIOS update, if I can find my external floppy and a working disk. Currently I have v. 1.57, and the latest is v. 1.61. But since even this is as early as 2007, I doubt it's of help here.

Offline

#84 2019-09-19 16:54:53

levi
Moderator
From: Yorkshire, UK
Registered: 2018-06-16
Posts: 462

Re: [pending] Desktop freezing occasionally (pentium4)

cx wrote:
levi wrote:

On my machine, 'sudo libinput debug-events' doesn't pick up the Fn button press.  Perhaps this keyboard handles Fn internally, or maybe it's handled so close to the metal inside a kernel module or something*.  But in any case, it probably shouldn't be your wakeup key.

What do you mean by "close to the metal inside a kernel module" exactly?

Sorry, I'm perhaps exposing my current mental model of the operating system a little to openly there, and risking confusing things.  Basically I see the system starting with BIOS, which loads the bootloader, so in a sense that sits atop.  The loads the kernel, which includes kernel modules so they kind of run in the same space of process no. 1.  The closest to bottom of that mental model (the metal) for us is process 1.

cx wrote:

I wonder why, as shown in this post, right 11 seconds after libinput laoding the TP Extra Buttons several devices get removed again,

Maybe TP extra buttons is handling them now so libinput doesn't need to care about them?  That's my guess anyway.

cx wrote:

Regarding missing errors, how about pressing e.g. the <Fn> (addressed w/ "KEY_WAKEUP") key would just make libinput think it should shut down itself, similar as it does when closing X?

On any machine I've owned it doesn't have what I'd call a wakeup key, only a sleep key.  I assume that's what it's referring to.

cx wrote:

What I can else do is maybe make a BIOS update, if I can find my external floppy and a working disk. Currently I have v. 1.57, and the latest is v. 1.61. But since even this is as early as 2007, I doubt it's of help here.

A bios update probably won't hurt, but I don't see how it would help.  The BIOS to my understanding only provides a few library functions which probably don't get called any more after the kernel has changed mode and has control over the framebuffer.


Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.

Offline

#85 2019-09-19 18:49:09

cx
Member
Registered: 2019-08-17
Posts: 60

Re: [pending] Desktop freezing occasionally (pentium4)

Thanks again for all the input, much appreciated!

So if I understand correctly, it is due to this "process 1" level that one sees several kernel "input" events but it’s no explicite module that can't be e.g. un- or reloaded, right?

$ journalctl -r | grep input
Sep 18 23:32:03 arch32box systemd-logind[424]: Watching system buttons on /dev/input/event10 (SEM HCT Keyboard Consumer Control)
Sep 18 23:32:03 arch32box systemd-logind[424]: Watching system buttons on /dev/input/event11 (SEM HCT Keyboard System Control)
Sep 18 23:32:03 arch32box systemd-logind[424]: Watching system buttons on /dev/input/event9 (SEM HCT Keyboard)
Sep 18 23:32:03 arch32box kernel: hid-generic 0003:1C4F:0002.0003: input,hidraw2: USB HID v1.10 Device [SEM HCT Keyboard] on usb-0000:00:13.1-2/input1
Sep 18 23:32:03 arch32box kernel: input: SEM HCT Keyboard System Control as /devices/pci0000:00/0000:00:13.1/usb3/3-2/3-2:1.1/0003:1C4F:0002.0003/input/input12
Sep 18 23:32:03 arch32box kernel: input: SEM HCT Keyboard Consumer Control as /devices/pci0000:00/0000:00:13.1/usb3/3-2/3-2:1.1/0003:1C4F:0002.0003/input/input11
Sep 18 23:32:03 arch32box kernel: hid-generic 0003:1C4F:0002.0002: input,hidraw1: USB HID v1.10 Keyboard [SEM HCT Keyboard] on usb-0000:00:13.1-2/input0
Sep 18 23:32:02 arch32box kernel: input: SEM HCT Keyboard as /devices/pci0000:00/0000:00:13.1/usb3/3-2/3-2:1.0/0003:1C4F:0002.0002/input/input10
Sep 18 17:56:24 arch32box systemd-logind[424]: Watching system buttons on /dev/input/event8 (ThinkPad Extra Buttons)
Sep 18 17:56:24 arch32box systemd-logind[424]: Watching system buttons on /dev/input/event4 (AT Translated Set 2 keyboard)
Sep 18 17:56:24 arch32box systemd-logind[424]: Watching system buttons on /dev/input/event1 (Sleep Button)
Sep 18 17:56:24 arch32box systemd-logind[424]: Watching system buttons on /dev/input/event0 (Lid Switch)
Sep 18 17:56:24 arch32box systemd-logind[424]: Watching system buttons on /dev/input/event2 (Power Button)
Sep 18 17:56:21 arch32box kernel: input: ThinkPad Extra Buttons as /devices/platform/thinkpad_acpi/input/input9
Sep 18 17:56:20 arch32box kernel: input: TPPS/2 IBM TrackPoint as /devices/platform/i8042/serio1/input/input8
Sep 18 17:56:19 arch32box kernel: input: PC Speaker as /devices/platform/pcspkr/input/input7
Sep 18 17:56:17 arch32box kernel: hid-generic 0003:1C4F:0034.0001: input,hidraw0: USB HID v1.10 Mouse [SIGMACHIP Usb Mouse] on usb-0000:00:13.0-2/input0
Sep 18 17:56:17 arch32box kernel: input: SIGMACHIP Usb Mouse as /devices/pci0000:00/0000:00:13.0/usb2/2-2/2-2:1.0/0003:1C4F:0034.0001/input/input6
Sep 18 17:56:17 arch32box kernel: input: AT Translated Set 2 keyboard as /devices/platform/i8042/serio0/input/input4
Sep 18 17:56:17 arch32box kernel: input: Video Bus as /devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:11/LNXVIDEO:00/input/input3
Sep 18 17:56:17 arch32box kernel: input: Power Button as /devices/LNXSYSTM:00/LNXPWRBN:00/input/input2
Sep 18 17:56:17 arch32box kernel: input: Sleep Button as /devices/LNXSYSTM:00/LNXSYBUS:00/PNP0C0E:00/input/input1
Sep 18 17:56:17 arch32box kernel: input: Lid Switch as /devices/LNXSYSTM:00/LNXSYBUS:00/PNP0C0D:00/input/input0
...

Anyway, no matter how libinput (mis)reads the separate <Fn> key, any more suggestions what to do next?
Staying like this, the X desktop is quite unusable!



I’m now going to install libinput 1.14.1-1.1 from testing. Let's see ...

Edit: Nope, no changes, persistent freezes.

Last edited by cx (2019-09-19 19:09:20)

Offline

#86 2019-09-19 19:45:47

andreas_baumann
Administrator
From: Zurich, Switzerland
Registered: 2017-08-10
Posts: 817
Website

Re: [pending] Desktop freezing occasionally (pentium4)

Did you try to run 'libinput debug-events' (as root in a shell for instance logged in via SSH). It should show you all events
happening on the machine. If it suddently freezes, then at least you know it's libinput.

Offline

#87 2019-09-19 20:53:20

levi
Moderator
From: Yorkshire, UK
Registered: 2018-06-16
Posts: 462

Re: [pending] Desktop freezing occasionally (pentium4)

That's how he found the FN key is being treated as KEY_WAKEUP.  And that does seem to lead almost directly to the system disconnecting the keyboard.


Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.

Offline

#88 2019-09-21 00:20:36

cx
Member
Registered: 2019-08-17
Posts: 60

Re: [pending] Desktop freezing occasionally (pentium4)

I wouldn't say, the keyboard gets "disconnected" – libinput debug-event still also records keyboard activities after a freeze! They're just not handed over to the screen.


I just experienced another incident, not caused by <Fn> but this time by <ALT>(+<Q>) or <CTRL>(+<Q>) when quitting a program and the LED showing hard disk activities for 1–2 seconds or so, just before the desktop froze.

As I had not invoked libinput debug-event unfortunately I can't give any log data here. – But when looking into ~/.local/share/xorg/Xorg.0.log, there're masses of continuous warnings (lines 481–352533 [end of the log]) :

...
[ 51250.793] (WW) RADEON(0): flip queue failed: Device or resource busy
[ 51250.793] (WW) RADEON(0): Page flip failed: Device or resource busy
[ 51250.816] (WW) RADEON(0): flip queue failed: Device or resource busy
[ 51250.816] (WW) RADEON(0): Page flip failed: Device or resource busy
[ 51250.821] (WW) RADEON(0): flip queue failed: Device or resource busy
[ 51250.821] (WW) RADEON(0): Page flip failed: Device or resource busy
...
[ 81117.036] (WW) RADEON(0): flip queue failed: Device or resource busy
[ 81117.036] (WW) RADEON(0): Page flip failed: Device or resource busy
[ 81117.542] (WW) RADEON(0): flip queue failed: Device or resource busy
[ 81117.542] (WW) RADEON(0): Page flip failed: Device or resource busy
[ 81117.544] (WW) RADEON(0): flip queue failed: Device or resource busy
[ 81117.544] (WW) RADEON(0): Page flip failed: Device or resource busy

Strange, right?

Offline

#89 2019-09-21 08:35:57

andreas_baumann
Administrator
From: Zurich, Switzerland
Registered: 2017-08-10
Posts: 817
Website

Re: [pending] Desktop freezing occasionally (pentium4)

This could actually point in a completly different direction: libunput gets keyboard events and something triggers
the graphic driver to stop outputting (which looks like a freeze).

Bugs like:
https://bugs.freedesktop.org/show_bug.cgi?id=97025

look promizing.

I'm really disappointed to say that, but the quality of the Xorg and especially the Xorg drivers got really bad lately,
many of them have stability issues, output a black screen or crash on startup. Basically everything not Intel, AMD
or Nvidia may have problems.

Offline

#90 2019-09-21 21:03:51

levi
Moderator
From: Yorkshire, UK
Registered: 2018-06-16
Posts: 462

Re: [pending] Desktop freezing occasionally (pentium4)

The screen's not freezing; apparently the user can still open submenus and things, just not type anything on the keyboard.

I'm not at all familiar with the AMD radeon driver, but it seems to carry on working even through those device not ready errors, so I don't think it's indicating a serious fault.  Unless by coincidence the keyboard became unbound at 51250 seconds, and the system remained up for another 30000 seconds (about 8h20m) after that fact before the user checked the xorg logs over the ssh connection.

Edit: I've been experimenting today with sending my machine to sleep and waking it up again (usually using the power button, but I've found enter also works).  To my pleasant surprise it all seemed to work for me.  Under libinput debug-events though (and xev fwiw), the sleep key is listed as KEY_SLEEP and not KEY_WAKEUP; I don't seem to have a key defined as the latter.  That little keymap snafy doesn't explain why ctrl+q and alt+q have a very similar effect sometimes though.

Last edited by levi (2019-09-21 21:33:26)


Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.

Offline

#91 2019-09-23 09:15:06

cx
Member
Registered: 2019-08-17
Posts: 60

Re: [pending] Desktop freezing occasionally (pentium4)

Unfortunately since the latest update this weekend the situation seemed to become worse for me. The desktop freezes appear now really fast. And right now I couldn't even evoke the file manager thunar (as initial client): the hotkey <Alt>+<Ctrl>+<Return> resulted in some hard disk activity before freezing; and yesterday I believe it was a mouse click, and in the following X session when intending to close a program with <Ctrl>+<Q>. – But I am still able to end xscreensaver by moving the cursor. Further I recognized when e.g. trying to evoke a terminal by hotkey (i.e. <Alt>+<Return>) after a freeze, the shape of the pointer changes for a moment.

Coming from XFCE/Xmonad, days ago I contemporarily changed to Openbox for testing, but the lagging from selecting menu items by mouse was really annoying!

Well, I don't want believe I can't use this machine in X, only because on tty there are no such problems at all.
And I also hesitate to switch to wayland, as I’m not used to it, and it also relies on libinput and I would probably still need xwayland, say, X dependencies. Or are there any reports about being productive in pure wayland though?

Meanwhile I’m off to work now, see you later.

Offline

#92 2019-09-23 14:21:27

levi
Moderator
From: Yorkshire, UK
Registered: 2018-06-16
Posts: 462

Re: [pending] Desktop freezing occasionally (pentium4)

FWIW Thunar was epicly broken for me when it used to be installed on my system.  I can't remember exactly what it did, but whenever it was spawned from e.g. firefox's show download folder it wasn't happy, so at some point last year I think, I blitzed it from my system.

On my openbox setup, I rarely use the system menu to start programs.  I have shortcuts defined in my hiding panel to start claws-mail and firefox, and everything else I start by typing its name from a terminal.  But I do have the menu there, used very occasionally for finding apps by category that I can't remember the name of.  It opens for me in under half a second though currently.


Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.

Offline

#93 2019-09-23 17:52:19

cx
Member
Registered: 2019-08-17
Posts: 60

Re: [pending] Desktop freezing occasionally (pentium4)

Well, by "lagging" I rather wanted say more precisely that when e.g. hovering the menu with the mouse its items got highlighted 1+ seconds later, which makes cursor handling and navigation arduous. And IMO that's also due to libinput's debouncing behavior.

And so far thunar works flawlessly for me; only I encountered another random freeze by pressing <Ctrl>+<Q> for closing it.

Offline

Board footer

Powered by FluxBB