unofficial linux-ck repo with cpu optimized packages is ONLINE

graysky · 2019-07-24 23:20:56

I saw them but I compiled 5.2.2 prior to the Arch kernel pushing 5.2.2 so they will be waiting for you in 5.2.3.

bkuri · 2019-08-01 17:18:29

.

Last edited by bkuri (2019-08-01 21:26:07)

john29 · 2019-08-01 18:22:56

installing nvidia-utils (430.40-1) breaks dependency 'nvidia-utils=430.34' required by nvidia-ck-zen

graysky · 2019-08-01 19:44:23

@john - Yes, building 5.2.5 now and will then rebuild nvidia-ck against that since 5.2.4 is old news now.

kyak · 2019-08-02 04:31:09

@graysky - Pardon my impudence, but could you please also include bbswitch into the repo? It makes sense, because it often comes hand to hand with nvidia module, which you already build.

Currently I'm using dkms versions of bbswitch, nvidia and virtualbox-host. It is fine, but it forces me to also use dkms for the stock kernel as well.

graysky · 2019-08-02 08:27:20

@kyak - Sorry no. I am already building broadcom + 3 different nvidia modules in addition to the kernel packages all multiplied by 17 architectures. I don't use dkms but it seems like a config file would allow you to customize builds, no?

kyak · 2019-08-03 04:45:39

@graysky - That's fine, thank you! Thing is, <name>-dkms packages conflict with <name> packages. So when I use dkms, I have to use it for all installed kernels. Or did you mean something different?

keepitsimpleengineer · 2019-08-26 19:13:48

Today, after [2019-08-26 08:46] [ALPM] upgraded nvidia-ck-sandybridge (2:430.40-5 -> 2:430.40-6), then startxfce4 gives…

[    16.679] (II) NVIDIA dlloader X Driver  430.40  Sun Jul 21 05:01:54 CDT 2019
[    16.679] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
⋯
[    16.811] (EE) NVIDIA: Failed to initialize the NVIDIA kernel module. Please see the
[    16.811] (EE) NVIDIA:     system's kernel log for additional error messages and
[    16.811] (EE) NVIDIA:     consult the NVIDIA README for details.
[    16.925] (EE) NVIDIA: Failed to initialize the NVIDIA kernel module. Please see the
[    16.925] (EE) NVIDIA:     system's kernel log for additional error messages and
[    16.925] (EE) NVIDIA:     consult the NVIDIA README for details.
[    16.925] (EE) No devices detected.
[    16.925] (EE) 
Fatal server error:
[    16.925] (EE) no screens found(EE) 
[    16.925] (EE)

Boots and loads startxfce4 with Linux 5.2.9-arch1-1-ARCH

Cheers..!

graysky · 2019-08-26 19:24:57

@keepit - Does the corresponding module load after reboot with 430.40-6? Check `dmesg` and `lsmod|grep nvidia`

keepitsimpleengineer · 2019-08-26 23:45:04

graysky wrote:

@keepit - Does the corresponding module load after reboot with 430.40-6? Check `dmesg` and `lsmod|grep nvidia`

dmesg shows this:

[   65.284510] nvidia: version magic '5.2.10-1-ck-sandybridge SMP preempt mod_unload ' should be '5.2.9-1-ck-sandybridge SMP preempt mod_unload '
[   65.397575] nvidia: version magic '5.2.10-1-ck-sandybridge SMP preempt mod_unload ' should be '5.2.9-1-ck-sandybridge SMP preempt mod_unload '

and no mention of 'nvidia' from lsmod

Cheers..!

======

Update today resolved the issue, thanks..!

Cheers..!

Last edited by keepitsimpleengineer (2019-08-27 15:58:54)

willianholtz · 2019-08-27 17:04:21

Man, how many people do I see with problems with Nvidia drivers, including me (nvidia 340xx). Although I am using nouveau. Does anyone know what to do to be able to use the nvidia Driver again? Because even going back to old Kernels doesn't work.

Is there any secret to compiling the kernel with support for it?
OR is something related there are several packages (xorg, wayland, kernel, etc ..)

arun_kpm · 2019-09-12 15:02:59

When running $systemd-detect-virt --container, I get input/output error. So haveged.service doesn't start at boot with this kernel.

loqs · 2019-09-12 16:36:31

@arun_kpm have you tried using strace to determine which call in systemd-detect-virt is producing the IO error?

glitsj16 · 2019-09-12 18:30:48

@arun_kpm The haveged.service file from 1.9.6-3 is also causing issues for me, but IMHO these are not related to linux-ck. I don't use any virtualization, but the haveged service is still being skipped due to 'ConditionVirtualization=!container' not working as expected. I filed a bug report on it. I had things relying on haveged's entropy working again by temporarily copying /usr/lib/systemd/system/haveged.service to /etc/systemd/system and commenting out that line. You might try the same and hopefully it revives haveged for you too.

loqs · 2019-09-12 18:40:08

@glitsj16 see comment #3488 as your bug report was closed as a duplicate.
Edit:
fixed comment number was incorrectly 3489

Last edited by loqs (2019-09-12 18:51:45)

glitsj16 · 2019-09-12 18:50:05

@loqs Yup, just saw that, thanks. Still, it does get haveged started and running as expected on linux-ck. Confusing...

loqs · 2019-09-12 18:52:45

But without knowing why the service fails how can it be fixed?
Edit:
Specifically why systemd-detect-virt fails.

Last edited by loqs (2019-09-12 18:53:44)

arun_kpm · 2019-09-13 02:36:53

loqs wrote:

@arun_kpm have you tried using strace to determine which call in systemd-detect-virt is producing the IO error?

It shows some files are missing.
https://paste.ubuntu.com/p/GcBk7p4WFs/

loqs · 2019-09-13 08:37:38

@arun_kpm What is the contents of /proc/1/sched ? If it exists systemd-detect-virt --container checks if it contains '(' and returns -EIO if not.

https://github.com/systemd/systemd/blob … irt.c#L520

arun_kpm · 2019-09-13 09:19:03

loqs wrote:

@arun_kpm What is the contents of /proc/1/sched ? If it exists systemd-detect-virt --container checks if it contains '(' and returns -EIO if not.
https://github.com/systemd/systemd/blob … irt.c#L520

That file is empty.

Last edited by arun_kpm (2019-09-13 09:30:50)

arun_kpm · 2019-09-13 09:42:49

@loqs , when using arch default kernel

 $cat /proc/1/sched                                                                                                                                    

systemd (1, #threads: 1)
-------------------------------------------------------------------
se.exec_start                                :         73100.979497
se.vruntime                                  :           249.828954
se.sum_exec_runtime                          :           917.149346
se.nr_migrations                             :                  122
nr_switches                                  :                15313
nr_voluntary_switches                        :                 1566
nr_involuntary_switches                      :                13747
se.load.weight                               :              1048576
se.runnable_weight                           :              1048576
se.avg.load_sum                              :                  249
se.avg.runnable_load_sum                     :                  249
se.avg.util_sum                              :               217088
se.avg.load_avg                              :                    0
se.avg.runnable_load_avg                     :                    0
se.avg.util_avg                              :                    0
se.avg.last_update_time                      :          73100979200
se.avg.util_est.ewma                         :                   48
se.avg.util_est.enqueued                     :                   89
policy                                       :                    0
prio                                         :                  120
clock-delta                                  :                   71
mm->numa_scan_seq                            :                    0
numa_pages_migrated                          :                    0
numa_preferred_nid                           :                   -1
total_numa_faults                            :                    0
current_node=0, numa_group_id=0
numa_faults node=0 task_private=0 task_shared=0 group_private=0 group_shared=0

https://paste.ubuntu.com/p/XYKzmfMsqK/

Last edited by arun_kpm (2019-09-13 09:52:25)

loqs · 2019-09-13 10:14:23

So it would appear the issue is the MuQSS scheduler providing an empty /proc/#/sched [1] file while systemd requires it to have the format of the mainline file.

[1] https://github.com/ckolivas/linux/blob/ … SS.c#L7297
Edit:
@graysky would you consider disabling CONFIG_SCHED_DEBUG ?

Last edited by loqs (2019-09-13 15:04:23)

graysky · 2019-09-13 19:08:00

@arun_kpm - can you try building this option disabled? Unless there is a good reason to do so, I do not want to force other users to have this since it deviates from the Arch defaults.

glitsj16 · 2019-09-13 20:12:28

@graysky I'll always respect whatever decision you take regarding linux-ck, we all know and respect the efforts you put in by offering it in your repo. But on this one (arun_kpm's issue and loqs excellent work tracking things down) I'm pressed to ask reconsidering loqs' request. It is not solely a matter of systemd-detect-virt -c failing. Behind the scenes this also creates issues for every systemd unit that uses ConditionVirtualization=!container (and possibly other like it). Systemd treats services that fail to meet such a conditional demand very silently, hence they are rather easily missed by using tools like 'systemctl --failed' and journal grepping. That was actually the gist of what I wanted to put out here in my earlier comment.

graysky · 2019-09-13 20:59:36

I am OK changing a config option if it is as described. I am just asking that someone test the proposed fix and verify that it works before I do it (ie it both fixes the issue and it doesn't introduce more issues). Can someone do this work and report back?

Arch Linux

#3476 2019-07-24 23:20:56

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3477 2019-08-01 17:18:29

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3478 2019-08-01 18:22:56

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3479 2019-08-01 19:44:23

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3480 2019-08-02 04:31:09

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3481 2019-08-02 08:27:20

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3482 2019-08-03 04:45:39

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3483 2019-08-26 19:13:48

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3484 2019-08-26 19:24:57

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3485 2019-08-26 23:45:04

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3486 2019-08-27 17:04:21

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3487 2019-09-12 15:02:59

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3488 2019-09-12 16:36:31

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3489 2019-09-12 18:30:48

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3490 2019-09-12 18:40:08

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3491 2019-09-12 18:50:05

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3492 2019-09-12 18:52:45

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3493 2019-09-13 02:36:53

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3494 2019-09-13 08:37:38

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3495 2019-09-13 09:19:03

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3496 2019-09-13 09:42:49

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3497 2019-09-13 10:14:23

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3498 2019-09-13 19:08:00

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3499 2019-09-13 20:12:28

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3500 2019-09-13 20:59:36

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

Board footer