gnarface | hmm, although now that i look at mine that doesn't seem to be true anymore | 00:01 |
---|---|---|
gnarface | nevermind that | 00:01 |
gnarface | just make sure they're all from the same release | 00:01 |
Nrml | gnarface: here it is https://paste.debian.net/1260299/ | 00:10 |
Nrml | (sorry for the delay, was installing/configuring pastebinit) | 00:10 |
Nrml | gnarface: looks pretty mixed to me | 00:10 |
gnarface | yea that might be normal now, i've got nothing sorry | 00:12 |
Nrml | no problem ;-) | 00:12 |
gnarface | i would try a whole different fresh install of something on another SD card to see if it's the software | 00:12 |
Nrml | I will check for apt-cache policy of each one just in case | 00:12 |
gnarface | i can't imagine how it could not be the software but i'm out of ideas | 00:12 |
Nrml | OK, thanks for the great help anyway. Your idea to plug/unplug them with the X server running and then checking Xorg.0.log actually gave me good clues | 00:14 |
Nrml | I will try and pursue them next | 00:14 |
e3d3 | My custom key bindings (xmodmap & setxkbmap) are reset to to defaults when I connect my new USB mouse & keyboard. How can I prevent this ? (Devuan Chimaera) | 00:51 |
gnarface | the first thing that comes to mind is you could alter the udev rule to call xmodmap and setxkbmap again on hotplug, but i would think there should be a window manager configuration you could use instead that would be smart enough to propagate the config across devices | 00:55 |
e3d3 | I've found it in my XFCE main menu "Removable drives & media", tab "Input devices". I never thought about looking there. Thanks. | 00:59 |
gnarface | no problem | 00:59 |
e3d3 | There I'll call xmodmap & setxkbmap again. Thank you for this simple & quick solution. | 01:00 |
Nrml | gnarface, I just kludged it into working. Got a minute to have a look? | 01:06 |
gnarface | sure | 01:06 |
Nrml | So, pursuing the messages on the Xorg.0.log file which were generated only when the mouse or keyboard were unplugged and replugged with the server running, I focused on this one: | 01:07 |
Nrml | (EE) client bug: Invalid path /dev/input/event3 | 01:08 |
Nrml | What I did was to run `startx` under strace (ie, `strace -f -s 1024 -o /tmp/startx.strace startx`) do the unplug/replug thing, and then look at the file strace produced. | 01:10 |
Nrml | I found the point where the message got write()n, then backtracked from there, looking for a cause. | 01:10 |
Nrml | And here's what I found: | 01:10 |
Nrml | 28636 newfstatat(AT_FDCWD, "/sys/devices/platform/scb/fd500000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0/usb1/1-1/1-1.2/1-1.2.1/1-1.2.1:1.0/0003:093A:2510.0015/input/input27/event3/uevent", | 01:11 |
Nrml | 0x7fefff2a18, 0) = -1 EACCES (Permission denied) | 01:11 |
Nrml | So I checked the path and indeed it was rwx for root only at this exact directory: /sys/devices/platform/scb/fd500000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0/usb1/ | 01:12 |
Nrml | So I did this: `sudo chmod g+rX,o+rX /sys/devices/platform/scb/fd500000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0/usb1/` | 01:12 |
Nrml | With that, it became rX for everyone, as expected | 01:12 |
Nrml | then I ran `startx` again, and IT WORKED! In fact, I'm typing this in X right now. | 01:13 |
Nrml | So, why I think this is a kludge: | 01:13 |
Nrml | 1) /sys is a 'dynamic directory' generated/maintained automatically by the OS. So after my next boot (or sooner) my permission setting will be gone. | 01:14 |
Nrml | 2) I think I just opened a security hole in the system (not sure it's a good idea to let anyone access these directories/files) | 01:15 |
gnarface | it shouldn't be like that, something is wrong | 01:16 |
Nrml | 3) I also think this is NOT the root cause; the root cause is whatever is causing the permission to be wrong (and perhaps the owner/group? both were 'root') | 01:16 |
gnarface | maybe you can fix the udev rule but that doesn't explain why it happened if it wasn't already happening | 01:16 |
Nrml | gnarface: my thoughts exactly ;-) | 01:16 |
Nrml | So it's udev the responsible for generating/maintaining the /sys directory? Or does it just meddles there (with permissions, etc) after the fact? | 01:17 |
rrq | note that /sys typically is a mount of sysfs which essentially is a filesystem API for the kernel | 01:20 |
rrq | udev maps /sys into dev nodes under /dev | 01:20 |
Nrml | rrq: that's more or less what I thought | 01:20 |
Nrml | rrq: so the X server should not be messing with /sys but rather with /dev? | 01:21 |
rrq | right | 01:21 |
gnarface | Nrml: no i think the kernel is making them, but i thought it'd obey udev for the /dev/input nodes, and these should match | 01:21 |
Nrml | crap | 01:21 |
gnarface | Nrml: the primary cause i'd expect for this by the way could only be a kernel change | 01:21 |
Nrml | hummrmrmr | 01:22 |
Nrml | lemme check when I upgraded the kernel | 01:22 |
Nrml | -rwxr-xr-x 1 root root 217854 Sep 6 05:41 config-5.15.65 | 01:22 |
gnarface | and it would seem like a bug but i could imagine a xorg library version mismatch maybe could cause the same thing | 01:22 |
gnarface | it's above my skill level | 01:23 |
Nrml | September 6th and this machine is powered off and then back on every day | 01:23 |
rrq | X11changed (some years ago) in its input handling .. from dev nodes to soemthing mroe dynamic.. I'm ot on top of it all | 01:23 |
rrq | ot = not | 01:23 |
Nrml | the fact is that the X server is indeed trying to access (and accessing!) stuff directly under /sys | 01:24 |
Nrml | which it should not, right? | 01:25 |
Nrml | (I say 'fact' because that's totally unambiguous in the strace output) | 01:25 |
rrq | I remember having to install xserver-xorg-input-kbd for some setup to get a funcitoning keyboard | 01:25 |
rrq | (it still swaps characters around... or maybe that's my fingers) | 01:26 |
Nrml | aha | 01:27 |
Nrml | check this out: | 01:27 |
Nrml | 28636 newfstatat(AT_FDCWD, "/dev/input/event3", {st_mode=S_IFCHR|0660, st_rdev=makedev(0xd, 0x43), ...}, 0) = 0 | 01:27 |
Nrml | 28636 readlinkat(AT_FDCWD, "/sys/dev/char/13:67", "../../devices/platform/scb/fd500000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0/usb1/1-1/1-1.2/1-1.2.1/1-1.2.1:1.0/0003:093A:2510.0015/input/ | 01:27 |
Nrml | input27/event3", 1024) = 153 | 01:28 |
Nrml | 28636 newfstatat(AT_FDCWD, "/sys/devices/platform/scb/fd500000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0/usb1/1-1/1-1.2/1-1.2.1/1-1.2.1:1.0/0003:093A:2510.0015/input/input27/event3/uevent", | 01:28 |
Nrml | 0x7fefff2a18, 0) = -1 EACCES (Permission denied) | 01:28 |
Nrml | this is what the X server did up to and including the error | 01:28 |
Nrml | So it seeems to read the link *content* from /dev/input AND THEN parsing it into a /sys path | 01:28 |
Nrml | this looks totally irregular to me | 01:29 |
gnarface | that is more or less how i was guessing it happened, it's just that the /sys one's permissions should match... | 01:29 |
gnarface | they should be the same thing basically | 01:29 |
gnarface | and i'm suspecting that somehow in your case it's not but i can't imagine why | 01:29 |
Nrml | gnarface: would udev be the responsible for setting the permissions in /sys? | 01:30 |
gnarface | like i said, it's never mismatched for me so i'm not sure i guess | 01:30 |
gnarface | i would have assumed the kernel reads them from udev | 01:30 |
gnarface | or provides udev some sort of hook to push them in | 01:31 |
gnarface | which udev are you using? systemd-udevd or eudev? | 01:31 |
Nrml | gnarface: so in your case, what are the owner/group/perms of /dev/input/event3 and /sys/devices/platform/scb/fd500000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0/usb1/1-1/1-1.2/1-1.2.1/1-1.2.1:1.0/0003:093A:2510.0015/input/input27/event3/uevent, or the equivalent in your system? | 01:31 |
Nrml | gnarface: eudev, of course ;-) | 01:32 |
Nrml | (no fucking systemd here, thank you ;-)) | 01:32 |
Nrml | dpkg -l eudev | 01:33 |
gnarface | i wouldn't know how to find the matching path so i can only assume this | 01:33 |
Nrml | ii eudev 3.2.9-10~chimaera1 arm64 /dev/ and hotplug management daemon | 01:33 |
rrq | that "uevent" is the kernel's event record for that device; it is read by udev, which creates the corresponding /dev node and sets its permissions | 01:33 |
Nrml | gnarface: can you please post a `find /dev/input -ls` from your system with X running? | 01:34 |
gnarface | Nrml: do you have these packages? xserver-xorg-input-evdev xserver-xorg-input-kbd xserver-xorg-input-libinput xserver-xorg-input-mouse xserver-xorg-input-all? | 01:34 |
Nrml | rrq: so udev would not change owner/group/perm in /sys? | 01:34 |
rrq | correct | 01:34 |
Nrml | rrq: ouch | 01:34 |
gnarface | Nrml: no, i will not post that, but i have a lot of them, and they're all 0660 root:input except the ones i've made custom udev rules for | 01:35 |
Nrml | rrq: so my X server (or some library or module or whatever in it) is going rogue when trying to access /sys instead of /dev/input? | 01:35 |
Nrml | gnarface: OK | 01:36 |
gnarface | some are event*, some are mouse*, some are js*, and one called just mice... some of these are legacy and i think some are extraneous ones i just didn't delete from an old udev bug | 01:36 |
Nrml | gnarface: here it's all 0660 root:input too, in /dev/input -- not so in /sys/* | 01:36 |
rrq | the perm on that "uevent" should be 644 | 01:36 |
Nrml | rrq: ls -l /sys/devices/platform/scb/fd500000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0/usb1/1-1/1-1.2/1-1.2.1/1-1.2.1:1.0/0003:093A:2510.0015/input/input27/event3/uevent | 01:37 |
Nrml | -rw-r--r-- 1 root root 4096 Nov 10 20:51 /sys/devices/platform/scb/fd500000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0/usb1/1-1/1-1.2/1-1.2.1/1-1.2.1:1.0/0003:093A:2510.0015/input/input27/event3/uevent | 01:37 |
Nrml | rrq: so it is | 01:37 |
gnarface | Nrml: i have /sys/ accesses in my Xorg.0.log too just no errors about them | 01:37 |
Nrml | but no "rx" permission on the directory needed to get there | 01:37 |
Nrml | in /sys/devices/platform/scb/fd500000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0/usb1/ to be precise, it was 0700 root:root | 01:38 |
gnarface | Nrml: do you have all these packages? xserver-xorg-input-evdev xserver-xorg-input-kbd xserver-xorg-input-libinput xserver-xorg-input-mouse xserver-xorg-input-all? | 01:39 |
Nrml | let me check them packages | 01:39 |
Nrml | gnarface: nope, just xserver-xorg-input-all and xserver-xorg-input-libinput. The other 3 (xserver-xorg-input-evdev, xserver-xorg-input-kbd, xserver-xorg-input-mouse) are not installed | 01:40 |
gnarface | at some point i had to add uhid and uinput to /etc/modules to keep my steam controller and roccat mouse working but this isn't chimaera | 01:40 |
Nrml | I remember installing xserver-xorg-input-evdev, but it didn't help so I uninstalled it | 01:40 |
Nrml | Do you think I should install them all? | 01:41 |
gnarface | they take up so little space it can't hurt | 01:41 |
gnarface | i thought the /dev/input/event* nodes specifically needed the evdev one | 01:41 |
Nrml | I will install, reboot and retest, then come back here | 01:41 |
Nrml | give me a few minutes please | 01:42 |
gnarface | try to modprobe uhid and uinput before you start xorg again too | 01:42 |
Nrml | back | 01:51 |
Nrml | unfortunately no go: installing the missing 3 packages did nothing | 01:51 |
Nrml | will try the modprobe thing now | 01:52 |
Nrml | gnarface: nope, modprobbing the two kmods before running `startx` also didn't change anything | 01:56 |
Nrml | only chgrp input /sys/devices/platform/scb/fd500000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0/usb1/ && chmod g+rX /sys/devices/platform/scb/fd500000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0/usb1/ fixes it | 01:57 |
Nrml | (so far) | 01:57 |
Nrml | folks, gotta go for the night. thanks everyone, specially gnarface and rrq, for the assistance | 01:57 |
Nrml | have a great evening y'a;; | 01:58 |
Nrml | *y'all | 01:58 |
* Nrml waves | 01:58 | |
perrierjouet | hello all | 04:04 |
perrierjouet | testing waters here | 04:04 |
perrierjouet | how is the community ? | 04:04 |
perrierjouet | I am systemd vaccinated :) almost 6 month free of the systemdeases | 04:05 |
debdog | the community is surprisingly silent tonight | 04:08 |
fred | that systemd-torrentclient is rly nice | 04:08 |
rwp | Hello perrierjouet. Welcome to the community! Social chatter is welcome in the #devuan-offtopic channel. | 04:08 |
perrierjouet | I was wondering is the package updates are often ? for example big package like gcc are updated easily ? | 04:10 |
rwp | Big packages like gcc upgrade very easily. One of the easier packages actually. | 04:11 |
rwp | As far as how often Stable only receives security upgrades or otherwise important upgrades. Stable won't get new major versions of gcc. | 04:12 |
rwp | The daily upgrade suite is Unstable which gets upgrades as often as maintainers upload them. | 04:12 |
rwp | However Unstable, the daily upgrade suite, is as the name implies sometimes unstable, sometimes broken. Some assembly required. | 04:13 |
fred | (ceres ftw) | 04:14 |
perrierjouet | thanks | 04:14 |
fred | us someone on here using nala instead of apt? | 04:14 |
bb|hcb | perrierjouet: If you want to check a particular package: https://pkginfo.devuan.org/ | 04:15 |
ham5urg | Is it possible to trick Gnome to hibernate instead suspend the laptop? I hate when users forget about the battery and corrupt the filesystem. | 17:23 |
PoundFoolish | Hello. I am installing Devuan 11 Chimaera from a thumbdrive. I am in the "Software selection" portion and my question is why is the option to choose LXDE not present as it is in the tutorial? | 19:26 |
PoundFoolish | Other desktop environments are present. Perhaps LXDE support been removed from the installer since the tutorial was written? | 19:27 |
PoundFoolish | Strangely the option is present when installing Debian Bullseye, upon which Chimaera is based. | 19:30 |
fsmithred | PoundFoolish, it should be there. It'll be a few hours before the person who makes those isos is awake. | 19:33 |
PoundFoolish | I promise it is not there. ;) https://i.imgur.com/pHVlsNy.png I will wait some hours. No problem. | 19:37 |
fsmithred | I'm sure you're right. I just started tasksel in a terminal. | 19:37 |
golinux | PoundFoolish: From https://www.devuan.org/os/: | 19:39 |
golinux | Desktop Environments | 19:39 |
golinux | Default: Xfce | 19:39 |
golinux | Alternate: Cinnamon, KDE, LXQt, MATE | 19:39 |
golinux | Others: available post-install | 19:39 |
fsmithred | Ithought we added it in beowulf, but I don't have a beowulf live-iso to test. | 19:40 |
golinux | So LXDE is only available post-install. I have no idea why . . . | 19:40 |
fsmithred | and I can't download anything big | 19:40 |
PoundFoolish | My mistake. I thought the process under Devuan would be functionally equivalent to Debian- and as I mentioned the user's screenshots did show LXDE in the menu to be checked. | 19:40 |
fsmithred | because in ascii, lxde didn't work right' | 19:40 |
fsmithred | PoundFoolish, some of the udebs for the installer have been forked. | 19:40 |
PoundFoolish | In the "Install Devuan with LXDE tutorial". Shame that the features I like are always first for the chopping block. | 19:41 |
fsmithred | We left lxde out deliberately, but I thought we added it when it started working right again. | 19:41 |
fsmithred | problem was with polkit stuff | 19:41 |
golinux | Well, when it gets figured out, www needs to be aligned with reality . . . | 19:42 |
fsmithred | install system without desktop, before you install the bootloader you can open a console and install task-lxde-desktop then return to the installer | 19:42 |
PoundFoolish | will try it. worst case i can always nuke the partition and try gentoo or void. If I am to futz about configuring the desktop environment I may as well get elbows deep and really break some stuff. | 19:44 |
PoundFoolish | no clue how to get into a console from the bootable installer "execute a shell" seems like the obvious choice but when I do that, "install task-lxde-desktop" gives "install:not found". | 19:50 |
PoundFoolish | "We left lxde out deliberately" is this decision documented anywhere? | 19:51 |
PoundFoolish | feels like I am getting too far off the path the Devuan installer expects me to take. | 19:52 |
PoundFoolish | Maybe I will try getting an older install media for Devuan and updating after install. | 19:53 |
PoundFoolish | my goal is to find a linux distro based on debian without systemd that makes it possible for me to use the LXDE desktop environment without jumping through too many hoops. | 19:56 |
PoundFoolish | You see how this seems a little sketchy, right? https://sourceforge.net/projects/devuanlxde/ | 19:58 |
PoundFoolish | @fsmithred, maybe you were envisioning a fancier installer than I am using. Neither does ALT+F2 to get to a console recognize the "install" command. | 20:01 |
golinux | PoundFoolish: Are you on a laptop? | 20:11 |
PoundFoolish | I am installing it on a thinkpad, yes. | 20:12 |
golinux | If yes, maybe this thread is relevant? https://lists.dyne.org/lurker/message/20221107.100215.8ba8af49.en.html | 20:12 |
* golinux wanders off to tea | 20:12 | |
PoundFoolish | Doubtful it is relevant. It seems to me that the console where "install" ought to be entered is the console of environment that is being installed, but the only console I can access during the actual install process is a limited one that does include the "install" command or similar package management commands. | 20:14 |
PoundFoolish | *does not include | 20:14 |
PoundFoolish | Anyway since I am just trying to find something that will work I will probably use the equivalent Debian installer and circle back at a later time to see whether Devuan has reenabled LXDE. In principle I like the idea of not using systemd but in practice the choice disables LXDE, which I really like. | 20:16 |
PoundFoolish | I am not in XFCE as the complainant in that exchange is, in any case. I am in the Devuan text-based installer, as seen in this screenshot: https://i.imgur.com/pHVlsNy.png | 20:18 |
PoundFoolish | Reading the forums gives me the impression LXDE may have been cut based on systemd entanglements so maybe it is not as easy as just apt-getting it. Hope Devuan can regain support for LXDE at a future date- imho it is the best desktop environment. Thanks for your help, #devuan. | 20:39 |
djph | golinux: what kinda tea? | 20:44 |
eyalroz | Didn't LXDE transition to Qt at some point? Or was the project forked? | 20:50 |
chomwitt | in chimaera trying to install qalculate-gtk i get you have held broken packages. what does that mean ? | 23:22 |
fsmithred | chomwitt, try 'aptitude -s install qalculate-gtk' and you will get some options. | 23:24 |
fsmithred | libqalculate20 : Depends: libqalculate20-data but it is not going to be installed | 23:25 |
chomwitt | fsmithred, so you get the same message ? | 23:27 |
fsmithred | yeah, with 'apt install' | 23:27 |
fsmithred | aptitude gives more info and some solutions | 23:27 |
fsmithred | well, one of the solutions is not to install the package. Other is to install the -data package with it. | 23:28 |
fsmithred | not sure why that doesn't happen automatically | 23:28 |
chomwitt | ok . there is no need to worry about sth then. i'll try aptitude | 23:28 |
fsmithred | oh, it's a version problem | 23:28 |
fsmithred | there are newer versions in backports | 23:32 |
chomwitt | $ sudo apt-get install qalculate-gtk/chimaera-backports also fails | 23:34 |
chomwitt | should i backport using aptitude ? | 23:34 |
fsmithred | Install the following packages: | 23:34 |
fsmithred | 1) libqalculate20-data [4.2.0-1~bpo11+1 (stable-backports)] | 23:34 |
fsmithred | that's one of the suggestions from aptitude | 23:35 |
chomwitt | ok that was installed | 23:35 |
fsmithred | from backports? | 23:36 |
chomwitt | yes and then also qalculate-gtk was installed | 23:36 |
chomwitt | all from backports | 23:36 |
fsmithred | does it work? | 23:37 |
chomwitt | fsmithred, thanks! although i am curious about the reason of the failed attempt | 23:37 |
chomwitt | yes! it works | 23:37 |
chomwitt | both from backports | 23:38 |
fsmithred | probably someone screwed up the version dependencies | 23:38 |
chomwitt | ok | 23:38 |
chomwitt | should i post an issue somewhere ? | 23:38 |
fsmithred | check debian bug reports. It might already be known. We don't fork any of those packages. | 23:39 |
chomwitt | https://tracker.debian.org/pkg/qalculate-gtk | 23:41 |
chomwitt | i dont see a stable package of qalculate-gtk in debian | 23:41 |
chomwitt | and no bug is reported for that package | 23:43 |
fsmithred | oh, right. | 23:43 |
chomwitt | so there is no stable version ? | 23:44 |
fsmithred | I don't see one | 23:47 |
fsmithred | if you look through bug reports and changelogs, you might find an explanation | 23:48 |
chomwitt | why $ sudo aptitude install qalculate-gtk didnt display a more simple error like 'package not found' ? | 23:51 |
phogg | I tried the daedalus netinst about a month ago and kept hitting problems with the keyring install. Anyone know if that's working now? (I am pulling the current netinst to try it anyway) | 23:59 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!