You are not logged in.
I updated yesterday and after a while my desktop wouldn't connect to websites, I decidee it was a good time to reboot since there was a kernel update. On reboot the process got stuck after loading lvm, doing some searches I found that adding ’modprobe.blacklist=amdgpu,radeon' succesfully gets me to log in (in which point the x server obviously errors).
My gpu is amd r9 270x
Offline
Please post the output of
uname -a
pacman -Qs linux
Can you boot the multi-user.target w/o blacklisting the GPU modules? https://wiki.archlinux.org/index.php/Sy … _boot_into
Offline
✗ ~ uname -a
Linux glad0s 5.3.8-arch1-1 #1 SMP PREEMPT @1572357769 x86_64 GNU/Linux
pacman -Qs linux: https://paste.xinu.at/LXeq1g/
Can you boot the multi-user.target w/o blacklisting the GPU modules? https://wiki.archlinux.org/index.php/Sy … _boot_into
Yes I can! Feels good to be in my normal session again, could you explain why this could've happened? And should I just set the default to multi user or is there a reason not to do that and find another solution?
Thank you for the quick response.
Last edited by indeedwatson (2019-11-02 16:35:57)
Offline
Please post the journal contents from a failed boot.
Offline
Offline
Nov 02 12:57:43 glad0s kernel: amdgpu 0000:01:00.0: Direct firmware load for amdgpu/si58_mc.bin failed with error -2
Nov 02 12:57:43 glad0s kernel: amdgpu 0000:01:00.0: si_mc: Failed to load firmware "amdgpu/si58_mc.bin"
Nov 02 12:57:43 glad0s kernel: amdgpu 0000:01:00.0: Failed to load mc firmware!
Is the linux-firmware package installed?
Offline
Nov 02 12:57:43 glad0s kernel: amdgpu 0000:01:00.0: Direct firmware load for amdgpu/si58_mc.bin failed with error -2 Nov 02 12:57:43 glad0s kernel: amdgpu 0000:01:00.0: si_mc: Failed to load firmware "amdgpu/si58_mc.bin" Nov 02 12:57:43 glad0s kernel: amdgpu 0000:01:00.0: Failed to load mc firmware!
Is the linux-firmware package installed?
It wasn't (and until yesterday it worked fine without it), I spotted that, installed it, and it didn't help.
The error did stop appearing, but it froze in the same way.
Last edited by indeedwatson (2019-11-02 21:36:52)
Offline
From the pacman query:
local/linux-firmware 20191022.2b016af-1
Firmware files for Linux
Might be missing in the inicpio in case of early kms?
Edit:
please post an up-to-date journal and also: what's your graphical target? Can you start it fom the multi-user.target?
Last edited by seth (2019-11-02 21:39:43)
Offline
journalctl -b 0: https://paste.xinu.at/dbgVP/
systemctl list-units --type=target: https://paste.xinu.at/9Tainn/
I'm not sure how to start a target, if you mean
# systemctl isolate graphical.target
that seems to work fine.
Offline
You could try https://wiki.archlinux.org/index.php/Ke … _KMS_start
systemctl list-units --type=target
Try
systemctl list-unit-files --state=enabled
to reveal the graphical.target
Offline
UNIT FILE STATE
autovt@.service enabled
bluetooth.service enabled
dbus-org.bluez.service enabled
dbus-org.freedesktop.thermald.service enabled
dhcpcd.service enabled
getty@.service enabled
netctl.service enabled
thermald.service enabled
remote-fs.target enabled
fstrim.timer enabled
Offline
That doesn't include any graphical.target did you completely disable it in order to enter the multi-user.target?
Please just state what your regular graphical target is, also: did you try enabling early KMS?
Edit: oh, and you've netctl AND dhcpcd enabled, likely throwing off the network config. Please pick one network managing service or ensure to read up the wiki pages and keep the managed interfaces apart.
Last edited by seth (2019-11-03 07:36:34)
Offline
That doesn't include any graphical.target did you completely disable it in order to enter the multi-user.target?
Please just state what your regular graphical target is, also: did you try enabling early KMS?
Nope, I didn't do anything regarding targets, ever, until I set the multi user one yesterday.
I'm not sure what you're asking, which other command should I use to answer your question? Because this isn't something I've ever manually set up or modified.
Early KMS was already enabled.
Edit: oh, and you've netctl AND dhcpcd enabled, likely throwing off the network config. Please pick one network managing service or ensure to read up the wiki pages and keep the managed interfaces apart.
This is my bad after I had a similar problem and solved it, and didn't clean it up well apparently. About a year ago my desktop would suddenly stop connecting to websites, so I switched to netctl.
Offline
You graphical.target would typically be some DM like SDDM, GDM, lightdm, …
If you don't use such, but eg. autolog'd into startx or something, please elaborate what your "regular" boot process (w/o selecting the multi-user.target) looked like. If you've always been booting into a getty and are still, the target selection might have had no impact at all, but the initial delay (if you edit the kernel cmdline by hand) might (ie. the HW is slow to come up and/or a crng issue)
Offline
You graphical.target would typically be some DM like SDDM, GDM, lightdm, …
If you don't use such, but eg. autolog'd into startx or something, please elaborate what your "regular" boot process (w/o selecting the multi-user.target) looked like. If you've always been booting into a getty and are still, the target selection might have had no impact at all, but the initial delay (if you edit the kernel cmdline by hand) might (ie. the HW is slow to come up and/or a crng issue)
Oh right, yeah booting into startx. I'm gonna describe the booting process in vague words because it's the best I can do by memory: grub, select kernel (which I have to manually do because I could never find a way to make grub remember the last selection, apparently something to do with lvm iirc), then white text that goes fast, then different white text with green [ ok ] on the left, then it takes a bit to connect to the network (probably due to what you pointed out about dhcp/netctl) and then x starts right away into bspwm.
When I couldn't boot the point at which it would hang was before the text with the green [ ok ]
Sorry for the crudeness of the explanation, I'm short on time.
Offline
Ok, that makes the question: can you successfully run startx after booting into the multi-user.target?
Offline
Ok, that makes the question: can you successfully run startx after booting into the multi-user.target?
Yes, when I added the multi user line to the kernel it booted perfectly. I can mark this as solved, but I would like to learn why this happened
Offline
This sounds as if the system boots so fast that the GPU isn't fully initialized when attempting to startx - try to add a brief sleep (5 seconds or so) before the automatic startx call.
Offline
This sounds as if the system boots so fast that the GPU isn't fully initialized when attempting to startx - try to add a brief sleep (5 seconds or so) before the automatic startx call.
Why would this start happening after an update?
Offline
*shrug* - something accelerated the boot process or the GPU module/firmware takes longer to initialize it…
Did you try a delay? Does it work?
Offline
What was in the update that triggered the issue?
Does Xorg.log contain:
(EE) systemd-logind: failed to take device /dev/dri/card0: Operation not permitted
Offline
I had dhcpd freeze and use 100% cpu, I couldn't kill the process so I rebooted to get internet again and figure out how to fully disable it.
Upon reboot I ran into the same issue from the start of this thread, except I can't boot into it even with multi user target.
EDIT: Installing linux-firmware again lets me boot, even without multi user target. Still not sure what's going on tho.
And the dhcpcd thing seems to be a bug
Last edited by indeedwatson (2019-11-07 04:44:04)
Offline
*shrug* - something accelerated the boot process or the GPU module/firmware takes longer to initialize it…
Did you try a delay? Does it work?
Does not work, my impression is the issue happens to early to even touch ~/.zprofile.
Last edited by indeedwatson (2019-11-06 14:41:45)
Offline