nmrp3 | I feel I should be making a blood sacrifice before I try to reboot with the new nvidia driver... | 00:04 |
---|---|---|
agris | you should get an amd card instead | 00:05 |
agris | the nvidia drivers including the open source effort are terrible | 00:05 |
agris | i did and I havn't had a graphics related kernel lockup since | 00:06 |
agris | and the drivers are built into the kernel | 00:06 |
fsmithred | I use nouveau with no problems, but my gaming is limited to freecell | 00:07 |
fsmithred | watching video is fine | 00:07 |
nmrp3 | I have never had any joy getting amd to work with things like opencl | 00:07 |
nmrp3 | this is not working :( | 00:32 |
fsmithred | I can't help with nvidia. Do whatever it says in the debian wiki. | 00:33 |
nmrp3 | I have done that | 00:34 |
agris | fsmithred, do you have weird DMA and iommu failures in your dmesg? | 00:34 |
fsmithred | checking... | 00:34 |
fsmithred | nope. dmesg looks ok | 00:36 |
nmrp3 | let's see if a dist-upgrade fixes it | 00:39 |
fsmithred | you added contrib and non-free to sources.list and installed packages from repo, right? | 00:40 |
fsmithred | or did you use nvidia binary? | 00:41 |
nmrp3 | yes, I added the contrib and non-free apt sources | 00:42 |
fsmithred | stay logged in here. You might get an answer hours later. | 00:44 |
nmrp3 | ok, got something - in boot, modprobe: FATAL: Module nvidia-current not found in directory /lib/modules/5.10.0-9-amd64 | 00:46 |
fsmithred | modprobe -v nvidia-current | 00:50 |
fsmithred | nm | 00:51 |
fsmithred | that won't work | 00:51 |
fsmithred | upgrade kernel | 00:51 |
fsmithred | upgrade everything. That iso is from December, I think. | 00:52 |
nmrp3 | I've purged nvidia-driver and am trying nvidia-legacy-... | 00:52 |
nmrp3 | nope - that also fails with a similar error | 00:53 |
fsmithred | 5.10.0-16 is current in chimaera | 00:53 |
nmrp3 | it's looking in that version | 00:54 |
nmrp3 | sorry - i have done a dist-upgrade since | 00:55 |
fsmithred | and reboot? | 00:55 |
nmrp3 | yeah - no dice | 01:06 |
nmrp3 | sudo modprobe nvidia can't find it | 01:06 |
nmrp3 | OK - fixed it possibly - with `apt-get install linux-headers-amd64` | 01:11 |
fsmithred | oh, did that get left out of the nvidia install because of no-recommends? | 01:12 |
nmrp3 | looks like it | 01:12 |
fsmithred | sorry about that | 01:12 |
nmrp3 | great :D I have a real desktop again | 01:14 |
fsmithred | cool | 01:15 |
nmrp3 | ok, so now I need to fix the keyboard layout and default localle, somehow | 01:17 |
fsmithred | sudo dpkg-reconfigure keyboard-configuration | 01:17 |
fsmithred | also dpkg-reconfigure locales | 01:17 |
fsmithred | and possibly tzdata | 01:17 |
fsmithred | if you need to be able to switch between different keyboard layouts, there's a panel plugin you can add. | 01:21 |
nmrp3 | no - I have just the one layout | 01:27 |
nmrp3 | what's the default graphical login devuan uses? | 01:27 |
brocashelm | lightdm IIRC | 01:31 |
nmrp3 | oh - apparently it is slim | 01:33 |
nmrp3 | oh, and apparently abandoned in 2013 ... ok ... backs away slowly | 01:34 |
fsmithred | you can replace it with lightdm | 01:35 |
nmrp3 | ok thanks - it's just a bit of an odd choice to use abandonware as a default | 01:36 |
nmrp3 | ok, that's enough for today - I guess I have everything else to fix tomorrow | 01:55 |
brocashelm | there's also sddm if you find lightdm buggy/bloated | 02:15 |
brocashelm | i guess i was thinking of refracta when i said lightdm, lol | 02:15 |
brocashelm | i just use startx | 02:15 |
debdog | agris: I now have a chimaera VM with the kernel meta package installed but I do not have said symlinks (yet, they prolly get created the first "update" on the kernel image is run) | 04:05 |
debdog | but from the cripts inside the package I gather (without having the entire log) that the command "-hda /pub/qemu/Chimaera_LXQt.qcow2 -cdrom /dev/cdrom -boot c" is failling | 04:06 |
debdog | no, that was not the proper paste here | 04:07 |
debdog | cannot copy'n'paste from VM, ok-.. | 04:08 |
debdog | agris: btw. are the link's (./boot/System.map-5.10.0-11-amd64) mode bits in order? | 04:14 |
agris | should it matter when your root? | 04:15 |
debdog | I hope the leading dot does not cause us to look at the wrong directory | 04:15 |
debdog | root when doing what? | 04:15 |
agris | yes i'm logged in as root when performing the upgrade | 04:16 |
debdog | yes, I do that, too | 04:16 |
agris | there is no 5.10.0-11-and64 map | 04:16 |
debdog | so, the failing command: /usr/bin/linux-update-symlinks upgrade 5.10.0-16-amd64 | 04:17 |
debdog | no, still not the proper command | 04:17 |
debdog | missing option... | 04:17 |
agris | http://dpaste.com/6R6BUJFE7 | 04:18 |
agris | nervermind | 04:19 |
agris | yeah there is | 04:19 |
agris | there's no weird acls on it | 04:19 |
agris | it's just a fat32 filesystem | 04:20 |
agris | debdog, how is your partitioning scheme setup? | 04:20 |
debdog | inside that VM /boot is not a seperate package. | 04:21 |
agris | package? | 04:21 |
agris | are you booting efi or bios? | 04:21 |
agris | this is m y setup https://dpaste.com/56SLVW5PB | 04:22 |
debdog | *partition/bios ; my goal here is, try to find which scripts are ran then reproduce them on the command line and see exactly where it fails. then start looking for a solution on that ground | 04:22 |
agris | I think the issue is that the kernel package started trying to setup symlinks and hardlinks at some point | 04:23 |
agris | but those features are not supported on FAT filesystems | 04:23 |
agris | using fat32 is required by a EFI system | 04:24 |
debdog | yah, the fat32 made me nervouse, too. but it seemed to have worked in the past, from what I understand | 04:24 |
debdog | here | 04:24 |
agris | maybe somebody changed the package without understanding the implications | 04:24 |
agris | anyways, just let me know what you need me to do | 04:24 |
agris | btw, partition 1 isn't a filestation. it's just reserve space for grub | 04:25 |
agris | it's common practice when installing grub to reserve the first 2 mib for grub to store modules | 04:26 |
debdog | what if we cp /boot/* to some temp directory, umount /boot then copy the contents back in. what fs would /boot have then on your system? | 04:26 |
agris | you mean put /boot on my rootfs? | 04:27 |
debdog | temporarily, yes | 04:27 |
agris | I don't think i'll be able to boot my system if I did that | 04:27 |
debdog | no, no booting! | 04:27 |
agris | my root filesystem is on a 'Z' filesystem | 04:27 |
debdog | anyway, if you reboot /boot would be re-mounted and everything should™ be fine | 04:28 |
agris | my bootloader (grub2) can't read modern 'Z' filesystems and as such won't be able to read the kernel and initramfs and microcode if I did that | 04:28 |
debdog | ok | 04:28 |
debdog | hehe | 04:28 |
debdog | maybe your setup is above my paygrade :D | 04:28 |
agris | oh no | 04:29 |
agris | it's just this: | 04:29 |
debdog | you still could go through all the commands of the install scripts manually and see where exactly it fails (there probably are easier solutions, have you had a look at the interweb for someone with the same issue) | 04:30 |
agris | hwinit -> EFI -> /boot/efi/grubx64.efi -> grub looks on the boot filesyetem for a kernel, initramfs, and hardware microcode and boots it | 04:30 |
agris | linux kernel executes loads initramfs, patches hardware ucode, loads Zfs kernel module and attempts to mount the real root filesystem then passes everything off to init | 04:31 |
agris | yeah I saw someone having the same issue on Ubuntu | 04:31 |
agris | unfortunately the Ubuntu solution didn't work for me | 04:32 |
agris | which was add a file in /etc/kernel-img make-symlinks: no | 04:32 |
debdog | (it could also be a problem with the ln command. I think "ln -sf" is somehow broken on my system) | 04:32 |
debdog | (but to find out, going through the commands is the only solution _I_ can come up with) | 04:33 |
debdog | what I haven't found out yet, which script calls... (will take some time) | 04:34 |
debdog | /usr/bin/linux-update-symlinks upgrade 5.10.0-16-amd64 /boot/System.map-5.10.0-16-amd64 | 04:36 |
debdog | found the one for vmlinuz-5.10.0-16-amd64 | 04:36 |
debdog | but not System.map-5.10.0-16-amd64 | 04:36 |
agris | do you want me to upgrade that? | 04:36 |
agris | I mean | 04:36 |
agris | run that command | 04:36 |
debdog | not without backing up /boot! | 04:37 |
debdog | then open linux-update-symlinks and practically copy'n'paste command by command into the terminal to see where it fails | 04:39 |
agris | it didn't fail | 04:39 |
agris | I: /vmlinuz.old is now a symlink to boot/vmlinuz-lts | 04:39 |
agris | I: /vmlinuz is now a symlink to boot/vmlinuz-5.10.0-14-amd64 | 04:39 |
agris | I: /initrd.img is now a symlink to boot/initrd.img-5.10.0-14-amd64 | 04:39 |
debdog | for variables they will have to be exported for that terminal beforehand | 04:39 |
debdog | ok | 04:39 |
debdog | but what about System.map-5.10.0-16-amd64 which is the one's failing, right? | 04:40 |
debdog | still havent figured out who or what causes that error exactly | 04:40 |
agris | it didn't error, had a zero exit status | 04:40 |
agris | but there is no /boot/vmlinuz | 04:41 |
agris | oh wait | 04:41 |
agris | there is in the root filesystem | 04:41 |
agris | not in /boot | 04:41 |
agris | should that be in there? | 04:41 |
debdog | plus there is this (--unpack) in "dpkg: error processing archive /var/cache/apt/archives/linux-image-5.10.0-11-amd64_5.10.92-2_amd64.deb (--unpack):" plus the leading dot in "unable to make backup link of './boot/System.map-5.10.0-11-amd64'" | 04:42 |
debdog | I see | 04:42 |
debdog | I don't have it either and thought it will be created at the next kernel update | 04:43 |
debdog | but this is a totally different situation then. where is ./boot? | 04:43 |
debdog | some tmp folder? | 04:43 |
agris | i don't know | 04:44 |
debdog | maybe the problem is not /boot but a tmp folder where initrd.img is created? | 04:44 |
agris | I think . has a special meaning in bash which mean CWD | 04:44 |
agris | (current working directory) | 04:44 |
agris | so whereever that is inside the .deb script | 04:45 |
debdog | right, but what is the scrip's CWD? | 04:45 |
agris | I don't know | 04:45 |
debdog | right | 04:45 |
agris | I'm not an expert in debian packaging | 04:47 |
debdog | hehe, me neither. but now we have new grounds for searching... | 04:47 |
debdog | so the command dpkg --unpack fails. just in which directory does it do that | 04:51 |
debdog | dpkg: error processing archive /var/cache/apt/archives/linux-image-5.10.0-11-amd64_5.10.92-2_amd64.deb (--unpack): | 04:51 |
debdog | unable to make backup link of './boot/System.map-5.10.0-11-amd64' before installing new version: Operation not permitted | 04:51 |
debdog | however, it's 5am and I still try to catch at least some sleep.... | 04:52 |
agris | sleep is for pussies | 04:53 |
gnarface | looks like the error you get if you try to upgrade a kernel when /boot is mounted read-only | 04:54 |
gnarface | or mabye / itself? | 04:54 |
agris | found this also https://forums.raspberrypi.com/viewtopic.php?t=331011 | 04:55 |
agris | /dev/sdg2 on /boot type vfat (rw,noatime,fmask=0022,dmask=0022,codepage=437,iocharset=ascii,shortname=mixed,utf8,errors=remount-ro) | 04:55 |
agris | tank/ROOT/chimaera on / type zfs (rw,relatime,xattr,noacl) | 04:55 |
agris | goodnight debdog, thanks for your help | 04:56 |
agris | takeaway from that forum post: I don't know where you've got that kernel package from, but I can tell you that it's failing because you can't create symlinks on a FAT filesystem, and you get EPERM if you try. | 04:56 |
agris | also this: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1318951 | 04:58 |
agris | the solution here: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1318951/comments/9 doesn't work | 04:58 |
gnarface | OH, /boot is vfat? yea that would hose it too, didn't expect this to be a raspberry pi | 05:01 |
agris | i'm not on a raspberry pi i'm on amd64-efi | 05:02 |
agris | but my /boot is also fat | 05:02 |
gnarface | you should not | 05:03 |
agris | should not what? | 05:03 |
gnarface | should not boot your linux box from a microsoft filesystem | 05:03 |
agris | what should I do instead for a EFI system? | 05:03 |
gnarface | ... unless there's a specific hardware compatibility reason to do so (like being on a RPI which requires it) | 05:03 |
agris | I thought efi systems required boot to be efi | 05:04 |
gnarface | i would just use the same filesystem you use for all the other partitions, i don't know why EFI would be different for that, i could be wrong | 05:04 |
agris | efi can't read Zfs | 05:04 |
agris | nor can grub | 05:04 |
gnarface | oh | 05:05 |
gnarface | did you try anything else? | 05:05 |
gnarface | like anything grub CAN read other than vfat? | 05:05 |
gnarface | xfs,ext2,ext3,ext4 are all good choices | 05:06 |
agris | gnarface, https://wiki.gentoo.org/wiki/EFI_System_Partition says fat32 is required | 05:06 |
agris | https://wiki.debian.org/UEFI | 05:08 |
gnarface | so no EFI systems have ever worked with packaged kernels? | 05:08 |
gnarface | that doesn't seem to mesh with reality | 05:08 |
agris | it used to work and upgrade just fine | 05:09 |
agris | something changed in the kernel packaging i think | 05:09 |
gnarface | isn't the EFI partition just /boot/efi though or something like that? not all of /boot? | 05:09 |
agris | my partition layout's been the same for several years | 05:09 |
agris | gnarface, maybe | 05:09 |
gnarface | do you have a /EFI directory? | 05:10 |
agris | no I do not | 05:10 |
agris | I wait | 05:10 |
agris | I do | 05:10 |
gnarface | you do | 05:10 |
gnarface | is it vfat? | 05:11 |
agris | yes | 05:11 |
agris | but it's just a directory in /boot | 05:11 |
agris | it's not it's own filesystem | 05:11 |
agris | https://dpaste.com/2DUT7GYJT | 05:11 |
brocashelm | is beowulf the last devuan xfce without forced gtk3? | 05:13 |
brocashelm | or is it ascii? | 05:13 |
agris | brocashelm, beowulf uses xfce4.12 which is gtk2 | 05:13 |
agris | going forward upstream xfce dropped gtk2 and became gtk3 | 05:13 |
agris | in chimaera it's gtk3 | 05:13 |
agris | you might want to look at dwm | 05:13 |
agris | it doesn't depend on third party toolkits | 05:14 |
agris | also lxqt or trinity TDE, which has good gtk2 theming support | 05:14 |
gnarface | can you make sure that EFI directory isn't actually another filesystem? it may be, even though it's not a separate partition | 05:14 |
brocashelm | yeah, i use ceres, so i'm obviously not too happy with xfce's "future" (or lack thereof) | 05:14 |
gnarface | check mount | 05:14 |
brocashelm | yes, i know about lxqt, and tde is actually really good (just not used to kde stuff) | 05:14 |
agris | http://trinitydesktop.org/ | 05:14 |
agris | trinity still provides a traditional desktop | 05:15 |
brocashelm | yeah, i've tried it a few times already | 05:15 |
brocashelm | very lightweight, too | 05:15 |
brocashelm | it's cool they provide devuan repo | 05:15 |
agris | gnarface, I'm sure /boot/EFI isn't on a different filesystem the other other efi related fs is efivarfs on /sys/firmware/efi/efivars type efivarfs (rw,nosuid,nodev,noexec,relatime) | 05:15 |
brocashelm | exe gnu/linux is devuan with tde, i've heard | 05:15 |
brocashelm | from one of the former refracta devs | 05:16 |
agris | brocashelm, what is it? | 05:16 |
brocashelm | exe gnu/linux? that's devuan with tde | 05:16 |
brocashelm | trinity desktop | 05:16 |
agris | what's their homepage? | 05:16 |
brocashelm | https://exegnulinux.net/ | 05:17 |
brocashelm | convenient for trying out in a vm before deciding to put trinity on your actual install or whatever | 05:17 |
agris | brocashelm, one other thing. In chimaera vim-gtk2 is broke | 05:17 |
agris | it was removed | 05:17 |
agris | upstream still supports gtk2 but debian removed it | 05:17 |
brocashelm | i am on ceres, but yeah | 05:17 |
agris | also they broke some gtk2 theming engines that support Bluecurve based themes | 05:18 |
brocashelm | i've just been thinking about dual booting with beowulf so i keep xfce before the bloatware took place | 05:18 |
agris | it's still in the packaging source code they just commented it out so people can't install it | 05:18 |
agris | it's why I haven't upgraded my main laptop yet | 05:18 |
brocashelm | lxqt and mate lack the rich customization of xfce IMO | 05:18 |
brocashelm | but i don't wanna derail this (let's take this to #devuan-offtopic) | 05:18 |
agris | there's an xAthena based version of vim available in the chimaera repos tho | 05:19 |
agris | it doesn't work as well | 05:19 |
agris | gnarface, did you get my last message about /boot/EFI not being a seperate fs? | 05:21 |
gnarface | agris: yea, unfortunately i don't know enough about this to guess what's wrong, i don't trust EFI so i don't use it | 05:22 |
gnarface | requiring that /boot be vfat though doesn't seem right to me | 05:22 |
gnarface | only place i've seen that is with RPi | 05:23 |
agris | gnarface, what don't you trust about EFI? | 05:24 |
gnarface | everything | 05:24 |
gnarface | this does ring a bell though, fsmithred knows what's going on | 05:24 |
gnarface | *maybe fsmithred knows what's going on, i meant | 05:24 |
gnarface | i thought he did last time anyway | 05:25 |
agris | well, you may have to learn it sooner or later. The new archer city saphire rapids platform coming out is all uefi based | 05:25 |
agris | you can't boot those platforms without uefi | 05:25 |
gnarface | i may just switch to ARM to avoid it | 05:25 |
agris | the ampere arm64 cpus are uefi based too | 05:25 |
agris | modern gpus also require uefi to post | 05:27 |
gnarface | how modern are we talking, exactly? | 05:29 |
gnarface | my RX5500 still boots fine without it | 05:30 |
agris | Ellesmere | 05:30 |
agris | GCN4 | 05:31 |
gnarface | https://bbs.archlinux.org/viewtopic.php?id=236633 | 05:31 |
gnarface | this forum thread suggests only /boot/efi needs to be vfat | 05:32 |
onefang | The UEFI files have to be an a FAT partition, but not the kernel and initrd. | 05:33 |
onefang | So GRUB on VFAT, everything else in /boot with a more Linux friendly file system that supports symlinks and stuff. | 05:34 |
onefang | Then you can mount the UEFI VFAT partition at /boot/efi, might even be done for you. | 05:34 |
agris | ugh | 05:35 |
onefang | Though I find rEFInd to be better than GRUB. | 05:35 |
agris | what is rEFInd? | 05:35 |
agris | I'm looking for a replacement to grub because it has terrible zroot support | 05:36 |
onefang | A boot loader like GRUB. Obviously EFI specific. Devuan has it in the repos. I dunno about zroot suppor,t I don't use that. | 05:36 |
agris | the upgrade grub command which gets called on kernel updates fails to parse zfs boot environments and I end up needing to write the grub.cfg manually each time which is a massive pita | 05:36 |
agris | hmm | 05:37 |
agris | what filesystems does refind support? | 05:37 |
gnarface | so, according to the info i'm seeing, the "Ellesmere" gpu family came before Navi 14, which my card uses... so it's significantly less modern | 05:37 |
onefang | That's one of the good thing sabout rEFInd, it finds your operating systems at boot time, no need to configure or update if you don't want to, but you can. | 05:37 |
agris | what about being able to interupt boot and change the kernel cmdline | 05:38 |
agris | and change the initramfs and kernel selected | 05:38 |
agris | i sometimes need to do that in emergencies | 05:38 |
agris | can refind do that? | 05:38 |
onefang | http://www.rodsbooks.com/refind/ | 05:38 |
agris | alright | 05:39 |
agris | it looks like I need to completely rework my boot setup to accomodate the packaging changes | 05:39 |
agris | thanks for the pointers | 05:40 |
onefang | You are welcome. | 05:40 |
agris | gnarface, it's possible that gpu just really doesn't like my system in bios mode for whatever reason | 05:40 |
* onefang goes back to weekend stuff. | 05:40 | |
gnarface | could be a bug in your bios, or maybe whoever made it had an agenda to force a narrative... | 05:41 |
agris | the bios was made by american megatrends so it's probably just a piece of sh*t | 05:42 |
agris | I almost got a job there ounce | 05:42 |
agris | I'm glad i didn't | 05:42 |
agris | i'm going to go with bug | 05:42 |
agris | I already had to RMA it for bad caps | 05:43 |
agris | alright partition tables updated, new filesystems mkfsed, tarball unpacked in place here goes | 05:57 |
agris | i used ext2 | 05:58 |
agris | it worked | 05:58 |
gnarface | in theory the ext4 driver should also be able to support ext2 with the proper kernel build options, saving you a tiny bit of ram | 06:00 |
agris | now i just cross my fingers and hope i'm able to reboot | 06:00 |
agris | I should change out the bootloader first though | 06:01 |
agris | gnarface, are you using refind to boot linux or are you using to to load grub and then boot linux? | 06:01 |
gnarface | grub on most the x86 machines, lilo on a couple of them, u-boot on the ARM ones | 06:02 |
gnarface | the ext2 through ext4 trick i picked up when i realized how much slower the ARM systems boot if /boot is a journaling filesystem | 06:02 |
gnarface | like crippling slowdown | 06:03 |
onefang | I use rEFInd to boot Linux directly. | 06:03 |
onefang | And the Linux kernel + initrd are on my ext4 /boot | 06:04 |
gnarface | i don't understand quite how this works, is there a reason the EFI partition can't be in the initrd? | 06:04 |
gnarface | i think i may have asked that before and someone told me why | 06:05 |
onefang | Coz UEFI expects EFI stuff to be on VFAT. | 06:05 |
agris | the spec for uefi only mandates inclusion of the fat32 filesystem driver | 06:05 |
agris | in theory other filesysems could be supported but you'd have to get a driver from your board manufactuer | 06:05 |
gnarface | any boards in the wild support other filesystems? | 06:05 |
onefang | So rEFInd / GRUB is on the VFAT so the BIOS can find and run them, but then rEFInd / GRUB loads your linux from ext4 or whatever. | 06:06 |
agris | ih no | 06:08 |
agris | oh no | 06:08 |
agris | i think i goofed | 06:08 |
agris | /dev/sdg4 on /boot/EFI type vfat (rw,noatime,fmask=0022,dmask=0022,codepage=437,iocharset=ascii,shortname=mixed,utf8,errors=remount-ro) | 06:08 |
agris | /dev/sdg4 on /boot/efi type vfat (rw,relatime,fmask=0022,dmask=0022,codepage=437,iocharset=ascii,shortname=mixed,utf8,errors=remount-ro) | 06:08 |
agris | how can a system be mounted twice | 06:09 |
onefang | Bind mount? | 06:09 |
agris | is it? | 06:09 |
gnarface | dunno if it even needs to be a bind mount | 06:15 |
gnarface | either way, looks easy to fix | 06:18 |
agris | well i installed refind | 06:22 |
agris | i'm not sure if it need to be configured manually or not but i can fallback to grub.efi | 06:22 |
agris | i'm going to reboot and check it out | 06:22 |
gnarface | good luck | 06:22 |
agris | thanks | 06:22 |
agris | it seems to have worked just fine | 06:56 |
debdog | phew | 06:56 |
onefang | Yay! | 06:57 |
agris | there's a weird thing where one of my daemons thinks it fails to start up but it didn't really | 06:58 |
agris | i have to manually kill it and then restart it and it's fine | 06:58 |
agris | but i don't think that's related | 06:58 |
agris | init related | 06:58 |
agris | zfs 2.1.x from chimaera backports has noticable performance improvements | 06:59 |
agris | thankyou all for your help | 07:02 |
onefang | You are welcome agris. | 07:02 |
Croco | Hi everyone, is anyone alive here? :) | 13:47 |
Croco | I've got a question/suggestion/plea, hmm, whatever you call it | 13:47 |
Croco | Aren't you going to get rid of elogind in the distro? I'm running beowulf, and I'm disappointed by things that damn useless daemon does, mounting unrequested and unwanted virtual filesystems, etc | 13:47 |
ted-ious | Croco: Can you disable what you don't like? | 14:10 |
ted-ious | I don't know but I think elogind is required to avoid the rest of systemd. | 14:11 |
fsmithred | it's also possible to use consolekit instead of elogind | 14:11 |
Croco | ted-ious: well, the system doesn't let me uninstall elogind | 14:12 |
Croco | and I don't think all these "features" can be disabled, as they are what elogind is | 14:13 |
Croco | anyway, well, there are packages that depend on it, like X window, but perhaps they can be build without that dependency. Hmm, *BSD systems use them somehow, no? | 14:13 |
Croco | fsmithred: how do I replace one with the other, then? | 14:14 |
fsmithred | apt install consolekit libpam-ck-connector | 14:15 |
fsmithred | or aptitude -s ... (if you want to see what it will do) | 14:15 |
Croco | fsmithred: thanks, I'm familiar with apt* :) | 14:15 |
fsmithred | for simulated runs, I like aptitude for its better layout | 14:17 |
fsmithred | anyway, it looks like that will only remove libpam-elogind and not elogind. But I think it might disable elogind. | 14:17 |
fsmithred | ^^^ in chimaera | 14:17 |
fsmithred | in beowulf, it removes elogind, too. | 14:18 |
lts | "apt remove elogind" looks promising in chimaera. I only ran with --dry-run though. | 14:19 |
Croco | WOW 8-() let's wait and see if my computer will boot again after all that | 14:20 |
Croco | bye-bye KDE :) I didn't use it anyway | 14:21 |
fsmithred | I just checked my no-dbus build, and it has neither elogind nor consolekit | 14:21 |
fsmithred | in beowulf, xfce and cinnamon worked with consolekit | 14:22 |
fsmithred | I'm not sure about chimaera | 14:22 |
Croco | fsmithred: that's great as dbus is, too, one of the things i hate. Is there a place to get the no-dbus version? | 14:22 |
fsmithred | agris, for future reference if you ever use grub again, you can put custom entries in /etc/grub.d/40_custom and they will be added to the boot menu when you run update-grub. | 14:23 |
fsmithred | yes Croco | 14:23 |
fsmithred | hang on | 14:23 |
Croco | Okay, I'm going to reboot now, hope to get back in minutes. In case of troubles, I'll read the web-based archive for the fsmithred's info :) | 14:24 |
fsmithred | oh, only beowulf nodbus.https://get.refracta.org/files/experimental/?2022 | 14:25 |
fsmithred | I could upload a daedalus nodbus, but there's no chimaera. It's just a proof-of-concept anyway | 14:25 |
fsmithred | when you come back, I'll give you the link for the discussion. You can make your own. | 14:25 |
fsmithred | oops. talking to a ghost | 14:26 |
Croco | Hmmmm... strange feelings | 14:30 |
Croco | there's no elogind in ps ax, at least, so something certainly happen | 14:30 |
Croco | but these damn filesystems are still there: | 14:31 |
Croco | securityfs on /sys/kernel/security type securityfs (rw,relatime) | 14:31 |
Croco | binfmt_misc on /proc/sys/fs/binfmt_misc type binfmt_misc (rw,nosuid,nodev,noexec,relatime) | 14:31 |
Croco | none on /run/user/1000 type tmpfs (rw,relatime,mode=700,uid=1000) | 14:31 |
Croco | pstore on /sys/fs/pstore type pstore (rw,relatime) | 14:31 |
Croco | I also didn't request separate tmpfs'es on /run and /run/lock (what a nonsense) | 14:31 |
fsmithred | you probably still have libelogind0 installed, but that won't do anything | 14:32 |
Croco | fsmithred: well, what do you mean "won't do"? What I want to achieve is to get rid of filesystems I didn't request, they are there, so "it" already did at least this | 14:33 |
fsmithred | I don't think that's where those filesystems are coming from | 14:35 |
Croco | fsmithred: okay... that's possible. The problem is that I've got no idea how to figure out who and why mounts them | 14:37 |
fsmithred | I can't remember where they come from | 14:37 |
fsmithred | checking a nodbus vm now to see what's mounted. | 14:37 |
Croco | fsmithred: thanks anyway! :) you helped me to get rid of 100+ packages I didn't need | 14:38 |
fsmithred | looks like nodbus has them, too | 14:39 |
fsmithred | wait. not all. I'll paste them somewhere. | 14:40 |
fsmithred | https://termbin.com/scoq | 14:41 |
gnarface | Croco: i suspect the binfmt_misc one is coming from binfmt-support, which you don't have to run | 15:11 |
fsmithred | Croco, here's a discussion about running without dbus: https://dev1galaxy.org/viewtopic.php?id=2158 | 15:30 |
Croco | fsmithred, gnarface: thanks folks! | 15:36 |
fsmithred | the link in that thread points to an old nodbus iso. Newer one can be found at https://get.refracta.org/files/experimental/?2022 | 15:37 |
hagbard | libsdl2-mixer-dev is currently unupgradable in devuan testing, because it depends on libfluidsynth-dev, which is also currently uninstallable in devuan testing, because it depends on systemd. | 17:23 |
Kingsy | Hi all, does anyone have the problem with devuan where after you close a laptop lid and it sleeps and you open it again you just get a blank screen? BUT if you press the power button you see X close and it shutting down, so its not crashed, its just like the X session breaks or something. | 20:20 |
Kingsy | I have the same problem on two different laptops using devuan | 20:20 |
fluffywolf | suspend/resume is fragile on linux in general. if the problem is you don't want it trying to suspend but it does, edit /etc/elogind/logind.conf to HandleLidSwitch=ignore or remove elogind. | 20:22 |
Kingsy | hmm no well I would love it to suspend if it could, but I suppose turning it off is a last resort. | 20:22 |
Kingsy | fluffywolf: so its worth just disabling it? or can it be tweaked to work? | 20:26 |
Kingsy | I am surprised its not a common issue, it happens on multiple laptops as I said, different hardware etc. | 20:26 |
fluffywolf | someone who is better at troubleshooting suspend/resume might be able to help, but I've never been good at it. | 20:27 |
xrogaan | Kingsy: this is a conjecture, but I suspected that when you reopen the lid the system doesn't get a "wake" event. So the screen remains shutdown until activity happens. | 20:31 |
xrogaan | Shutting down the lid ought to put your system in standby, should behave exactly the same as a desktop. | 20:33 |
xrogaan | I mean closing the lid, not shutting it down. Sorry. | 20:33 |
xrogaan | The behavior highly depends on what you've installed and what you're running. | 20:34 |
Kingsy | xrogaan: its a default devuan installation, running bspwm launched via xinit | 20:43 |
Kingsy | if thats what you mean. | 20:43 |
Kingsy | xrogaan: any ideas how to proceed? | 21:35 |
xrogaan | you need to understand what happens when your computer wakes from suspend. | 21:38 |
xrogaan | The behavior I had with a previous configuration was that the screensaver would trigger before suspend, then appear at wakeup. | 21:39 |
xrogaan | Energy saving measures where the screen shuts down (DPMS stuff) can also be a cause. | 21:40 |
xrogaan | Forcing a change in virtual console (tty) may force the screen to reactivate. | 21:41 |
xrogaan | I had an issue with something related to lightdm where my keyboard would be non-responsive. Was due to a non-maintained add-on I had installed, or something like that. Can't remember properly. | 21:42 |
Kingsy | xrogaan: yes, but what is the technique to debug this? to understand what is happening? iirc switching to a different tty didnt work | 22:26 |
rwp | Kingsy, It's normal for the screen to be dark in DPMS power save. But then pressing a key should wait it up. That's the part that sounds like the problem. | 22:40 |
rwp | I don't run a normal installation but I think the screen saver will be active. It's possible the screen saver is at fault. | 22:40 |
rwp | I would try disabling the screen saver and seeing if that improves the situation. | 22:41 |
Kingsy | rwp: I don't have a screensaver. I have slock installed but thats just a binary. I cant see ohow that would be getting triggered. | 22:41 |
rwp | If you don't have a screen saver then you don't have a normal installation. Please describe what you do have? | 22:41 |
Kingsy | rwp: devuan, installed without a DE, I installed bspwm, I launch it with xinit, thats it. | 22:42 |
Kingsy | super simple and lightweight | 22:42 |
rwp | Do you have acpid installed? | 22:42 |
rwp | dpkg -l | grep acpi | 22:43 |
Kingsy | yep | 22:43 |
Kingsy | ii acpi 1.7-1.1 amd64 displays information on ACPI devices | 22:43 |
clemz | Are snap packages compatible with devuan? | 22:43 |
fsmithred | clemz, no. Try appimage instead. | 22:44 |
rwp | And acpi-support? (I had to look.) Most of the suspend, resume, reboot actions are triggered through acpid and acpi-support. | 22:44 |
Kingsy | dpkg -l | grep acpi-support <- returns nothing | 22:44 |
rwp | I suggest installing acpi-support and seeing if that includes the support framework that is needed for suspend-resume. | 22:45 |
rwp | That might be the missing parts that are not waking things up again upon resume. | 22:45 |
Kingsy | installed | 22:45 |
rwp | In particular I see it includes /etc/acpi/lid.sh which handles it for me. | 22:45 |
Kingsy | ah this is good to know | 22:45 |
rwp | The /etc/acpi/events/lidbtn file configures to call /etc/acpi/lid.sh which calls other programs to do the work. | 22:46 |
Kingsy | would this require a reboot to take effect? | 22:47 |
rwp | No. Just that acpid is running. I don't think acpid would need a restart. But it might. As root, "service acpid restart" | 22:47 |
Kingsy | cool | 22:48 |
Kingsy | alright need to save some stuff incase it doesnt work.then I'll test. | 22:48 |
rwp | There is also a utility to debug acpi events by printing them as they happen. I don't recall the utility now though... Useful for this type of thing. | 22:48 |
rwp | Ah, it's called "acpi_listen" which lists out acpi events as they happen to debug them. | 22:49 |
rwp | Kingsy, I remember reading of many people with problems where the screen remains dark after suspend. | 22:58 |
rwp | Here is some discussion: https://www.thinkwiki.org/wiki/Problem_with_display_remaining_black_after_resume | 22:58 |
rwp | Noting that that wiki page has rather aged information in general. But perhaps something there will be good background on the problem. | 22:58 |
Kingsy | oh wow nice | 22:59 |
rwp | The suggestion to try Ctrl+Alt+F1 to switch to text console and then Alt+F7 (assuming X on vt7) seems likely to work. Maybe. | 23:01 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!