Package Details: linux-ck 6.11.10-1

Git Clone URL: https://aur.archlinux.org/linux-ck.git (read-only, click to copy)
Package Base: linux-ck
Description: The Linux kernel and modules with ck's hrtimer patches
Upstream URL: https://wiki.archlinux.org/index.php/Linux-ck
Licenses: GPL-2.0-only
Provides: KSMBD-MODULE, VIRTUALBOX-GUEST-MODULES, WIREGUARD-MODULE
Replaces: virtualbox-guest-modules-arch, wireguard-arch
Submitter: graysky
Maintainer: graysky
Last Packager: graysky
Votes: 458
Popularity: 0.123230
First Submitted: 2011-07-22 14:51 (UTC)
Last Updated: 2024-11-26 13:15 (UTC)

Dependencies (14)

Required by (7)

Sources (6)

Latest Comments

« First ‹ Previous 1 .. 30 31 32 33 34 35 36 37 38 39 40 .. 308 Next › Last »

graysky commented on 2019-01-13 12:11 (UTC)

I remember looking at this a while ago and concluding it isn't possible... prove me wrong :)

air-g4p commented on 2019-01-13 09:37 (UTC)

Hi graysky,

I can confirm the build loop issue zoopp mentioned last month is still present in linux-ck 4.20.1-1.

With this PKGBUILD option set:

27. Native optimizations autodetected by GCC (MNATIVE)

_subarch=27

The build fails with hundreds of instances of:

Support for P6_NOPs on Intel chips (X86_P6_NOP) [N/y/?] (NEW) 27

It appears to me that the only way out of this loop is crtl-c.

Of course, the build proceeds normally with any other appropriate subarch assigned.

graysky commented on 2019-01-12 09:38 (UTC)

@artafinde - ck1 disables NUMA

artafinde commented on 2019-01-12 09:32 (UTC)

@vltr: try enabling NUMA on ck.

vltr commented on 2019-01-11 09:53 (UTC)

@marceliq ohhh ... I haven't seen that. And that's actually a real bummer :(

marceliq commented on 2019-01-11 09:46 (UTC) (edited on 2019-01-11 09:47 (UTC) by marceliq)

In patch-4.20-ck1 there is:

@@ -896,6 +913,7 @@ config CGROUP_DEVICE

 config CGROUP_CPUACCT
        bool "Simple CPU accounting controller"
+       depends on !SCHED_MUQSS
        help
          Provides a simple controller for monitoring the
          total CPU consumed by the tasks in a cgroup.

which disable CPUACCT if SCHED_MUQSS is enabled :(

vltr commented on 2019-01-11 09:41 (UTC) (edited on 2019-01-11 10:13 (UTC) by vltr)

Thanks, @graysky! Yeah, I tried to revert some of the changes - docker stopped working after 4.19.4 or 4.19.5 (from repo-ck), but I just got some time to find the source of the error now. I thought it was CONFIG_CGROUP_CPUACCT at first because of error given by docker:

$ docker run --rm hello-world                        
docker: Error response from daemon: unable to find "cpuacct" in controller set: unknown.

That's just an example. Any other docker command (build, run, etc) would eventually ran into this issue (although the docker daemon seem to work normally).

This is something that doesn't happen in the stock kernel (and I make use of testing repositories as well). Anyway, I reverted not only that flag but some others as well (I actually did the diff myself earlier), but docker is still a no go. I'm not that expert in the kernel to go further and analyze what's wrong, unfortunately. I wanted to share my experience here in case anyone bumps into the same problem.

Would you mind providing me CK's blog so I can post there as well?

Thanks a lot!


EDIT

Just found out the issue, pointed out by @marceliq and then CK's blog and comments regarding this issue as well as the forum post related. Thanks!

graysky commented on 2019-01-10 20:25 (UTC) (edited on 2019-01-10 20:28 (UTC) by graysky)

@vltr - It would appear the ck1 patchset does this, see here for a diff of the Arch config before and after simply patching with CK1: https://github.com/graysky2/4.20.1_temp/commit/82d2aee40035766d2e225e4b2a3db36faa187273

Recommend you post to CK's blog referencing that diff and your problem. Might be that he did not mean to do that if I am correct.

vltr commented on 2019-01-10 11:28 (UTC)

Hello, @graysky! Just one question: is there any reason why the CONFIG_CGROUP_CPUACCT flag is not set anymore on linux-ck? Thanks a lot!

graysky commented on 2019-01-01 14:42 (UTC)

@saren - sorry about that... fixed now but did not bump the pkgver.