Package Details: linux-ck 5.18.16-1

Git Clone URL: (read-only, click to copy)
Package Base: linux-ck
Description: The Linux kernel and modules with ck's hrtimer patches
Upstream URL:
Licenses: GPL2
Replaces: virtualbox-guest-modules-arch, wireguard-arch
Submitter: graysky
Maintainer: graysky
Last Packager: graysky
Votes: 449
Popularity: 0.065850
First Submitted: 2011-07-22 14:51 (UTC)
Last Updated: 2022-08-03 18:10 (UTC)

Required by (12)

Sources (10)

Pinned Comments

graysky commented on 2022-04-14 11:45 (UTC)

Is anyone interested in testing the 5.17.y ck kernel? I'd like some broader feedback before I push it out to the AUR and to the repo. Please try it here if interested:

You can download a prebuilt package or build yourself.

Please post feedback good or bad to the repo-ck thread to keep it all in one place:

Latest Comments

air-g4p commented on 2022-07-10 14:34 (UTC) (edited on 2022-07-11 05:08 (UTC) by air-g4p)


Againt kernel 5.18.10-hardened1-1-hardened linux-ck fails to build this way:

LD vmlinux.o OBJTOOL vmlinux.o scripts/ line 107: 190039 Killed tools/objtool/objtool ${objtoolcmd} ${objtoolopt} ${1} make: *** [Makefile:1162: vmlinux] Error 137 ==> ERROR: A failure occurred in build(). Aborting...

real 15m54.769s user 13m3.414s sys 8m38.732s

A second build attempt failed in a very similar way:

LD vmlinux.o OBJTOOL vmlinux.o scripts/ line 107: 283258 Killed tools/objtool/objtool ${objtoolcmd} ${objtoolopt} ${1} make: *** [Makefile:1162: vmlinux] Error 137 ==> ERROR: A failure occurred in build(). Aborting...

real 20m15.477s user 14m19.039s sys 9m15.759s

Any ideas?

snack commented on 2022-04-14 12:37 (UTC)

@zerophase I've been having the (arguably) same problem since very long time (see e.g., and that actually forced me to go back to the stock kernel since I never found a solution.

zerophase commented on 2022-04-14 12:31 (UTC)

Anyone else having issues getting Docker to start up with the CK kernel?

I've noticed issues with mounting archiso too.

graysky commented on 2022-04-14 11:45 (UTC)

Is anyone interested in testing the 5.17.y ck kernel? I'd like some broader feedback before I push it out to the AUR and to the repo. Please try it here if interested:

You can download a prebuilt package or build yourself.

Please post feedback good or bad to the repo-ck thread to keep it all in one place:

graysky commented on 2021-09-14 17:15 (UTC)

It's not MuQSS, but the PKGBUILD now contains CK's hrtimer patch and the recommended 1000 Hz tick rate.

nTia89 commented on 2021-08-31 16:55 (UTC)

sad news

Kr1ss commented on 2021-07-22 09:15 (UTC)

I'm happy to go along with @air-g4p ! Thank you very much @graysky, and all contributors and commenters.

air-g4p commented on 2021-07-22 09:09 (UTC)

@graysky - WOW!! Today marks linux-ck's 10th Birthday!

Thank you very much for all your hard work during the past decade, graysky!

Your efforts are much appreciated...


graysky commented on 2021-06-11 01:28 (UTC) (edited on 2021-06-11 01:28 (UTC) by graysky)

Pretty sure, delete the tar ball and re-download :p

vp1981 commented on 2021-06-11 00:19 (UTC)

@graysky, are you sure about b2sum checksum for 'more-uarches-20210610.tar.gz'? I downloaded it manually and checksum didn't match from PKGBUILD: 30d1df754608bb423cbc99c2097ad521baa091b9a3b39df4bd5c2d50c57eec54d8fa0e4a4a04b847c3d1b87ba682cadc8db45fabeefdc9ad7caaf8e77b96e41a tmp/kernel_compiler_patch-20210610.tar.gz

If the '' git repository is yours would it be feasible to use signed tags to accompany releases? In such cases signed tag would help others to simply change checksum in PKGBUILD. If it is too troublesome then don't bother with it.

snack commented on 2021-06-02 08:05 (UTC)

According to this:

the current version of docker should work with cgroups v2. So I guess it's some interplay between the CK patchset and cgroups v2 that breaks docker. Sadly, I got no reply from CK on his blog...

snack commented on 2021-05-28 06:40 (UTC)

@artafinde your suspect was correct. Setting the kernel parameter you suggested makes docker work again with linux-ck. I'm going to report this on CK's blog. Thanks.

artafinde commented on 2021-05-27 14:06 (UTC)

I suspect your issues has something to do with cgroups v2. Try to replicate the issue after setting (temporarily) systemd.unified_cgroup_hierarchy=0 as a kernel param which forces the cgroup v1 behavior.

GAZ082 commented on 2021-05-27 13:46 (UTC)

@snack had your issue and also a system freeze with the latest release. First time after almost a year of using CK. Went back to Zen.

snack commented on 2021-05-27 06:38 (UTC)

@zerophase I'm not sure. For sure it's something inside the CK patchset (since the stock and many other patched kernels work fine), and it's not MuQSS (see e.g. the comment by Steven Barret on CK's blog stating that Liquorix with MuQSS works fine). I'd need to understand if there is some hope for this to be fixed, otherwise I'll have to ditch linux-ck.

zerophase commented on 2021-05-27 02:30 (UTC)

@snack It has been a bug for awhile. I tried reporting it about a year ago, when trying to author a custom arch install disk, and got some explanation from a guy that this is not a CK issue. (Wasn't CK)

snack commented on 2021-05-26 06:04 (UTC) (edited on 2021-05-26 06:05 (UTC) by snack)

I cannot start docker containers when using the current (5.12) linux-ck-skylake kernel. I get this error:

$ docker run --rm -it -p 80:80 ckulka/baikal:nginx
docker: Error response from daemon: failed to create shim: OCI runtime create failed: container_linux.go:380: starting container process caused: process_linux.go:545: container init caused: process_linux.go:508: setting cgroup config for procHooks process caused: load program: invalid argument: unknown.

With the stock Arch kernel it works, and seemingly also with other patched kernels like Liquorix (although I didn't try by myself).

I know that it is usually suggested to post to CK's blog for this kind of issue, and in fact I did it but from there I got the advice to post to AUR...

vp1981 commented on 2021-05-19 05:45 (UTC)

graysky> GCC 11 reorders things and since GCC 11 is in [core], I updated the PKGBUILD.

That was a culprit! I updated PKGBUILD, rebuild the package (linux-ck-sandybridge) and was able to boot with linux-ck-sandybridge kernel.

elektorronikci commented on 2021-05-18 18:47 (UTC) (edited on 2021-05-18 18:47 (UTC) by elektorronikci)

@graysky, thanks for your suggestion. There is not any problem for me, I had already corrected it when I saw it on nconfig. I mentioned about it just to inform.

graysky commented on 2021-05-18 17:43 (UTC)

@graysky, no it is GCC 10.2

Right, define an updated mirror and update your system :D

elektorronikci commented on 2021-05-18 17:02 (UTC)

@graysky, no it is GCC 10.2

thaewrapt commented on 2021-05-18 16:05 (UTC)

It could also happen on automation level, where the build system is still using old IDs. I'm building for Skylake right now so I can check if it's actually built for Skylake in the end.

thaewrapt commented on 2021-05-18 16:00 (UTC) (edited on 2021-05-18 16:02 (UTC) by thaewrapt)

Yeah, as I hadn't any issues myself I wanted to suggest "a wrong architecture" source of the problem since I've noticed a change of architecture IDs in the package. For example, I've used 25 for Skylake for quite a long time but in the very last release I see it numbered as 26.

graysky commented on 2021-05-18 15:53 (UTC)

@elektorronikci - Is your system up-to-date? GCC 11 reorders things and since GCC 11 is in [core], I updated the PKGBUILD.

elektorronikci commented on 2021-05-18 15:48 (UTC) (edited on 2021-05-18 15:51 (UTC) by elektorronikci)

There is a problem in the PKGBUILD subarch configuration. When I select Haswell, it was actually Broadwell in the .config file. That may cause boot problem because of some unimplemented machine codes.

vp1981 commented on 2021-05-17 08:46 (UTC)

Ok, I posted message on ck's blog.

I suspect an issue with ck patch and GCC11 but I don't understand why newer processors (haswell and skylake specifically) work.

vp1981 commented on 2021-05-17 03:01 (UTC)

@graysky, thanks I'll do that when I finish testing. I compiled linux-ck-sandybridge 5.12.2, 5.12.3 with the same configuration as 5.12.4 (but without 0002-drm* patch applied) but both kernels didn't boot.

graysky commented on 2021-05-17 00:51 (UTC)

Recommend you post to ck's blog and see what he thinks. I have no insights.

vp1981 commented on 2021-05-17 00:19 (UTC)

Similar situation here: my notebook with Intel Core i5 2410M can't boot with linux-ck-sandybridge 5.12.4 but works fine with distro linux 5.12.4. I locally built linux-ck package (though without patch 0002-drm...) with gcc 11 and config the same as for linux package. Seems that something affects "old" processors because on other hosts with Intel processors of haswell and skylake family both linux-ck-haswell and linux-ck-skylake 5.12.4 boot fine. I didn't tried 5.12.3 version but 5.12.2 worked. I'll try to build 5.12.2 and 5.12.3 for sandybridge and report results.

GAZ082 commented on 2021-05-15 20:45 (UTC)

Had to downgrade to 5.12.1-1-ck-nehalem because linux-ck-nehalem-5.12.4-1 freezes the system at boot when "Loading initial ramdisk".

zepar commented on 2021-05-14 01:35 (UTC)


I boot to a black screen with 5.12.3-2 on a Gemini lake CPU. Version 5.12.3-1 worked fine. (i'm using the bin version).

graysky commented on 2021-04-23 14:51 (UTC)

Report/ask on ck's blog:

zerophase commented on 2021-04-23 14:18 (UTC) (edited on 2021-04-23 14:22 (UTC) by zerophase)

@coderkun It's probably something to do with Linux-ck I tried mounting the archiso build system for a custom installer, a year ago, and I can't do it on Linux-ck. Works fine on the default. I told CK about, but got no response. I also received message, from someone else, accusing me of causing the problem.

I think we have the same Arch default settings, other than the CK stuff. I don't see why this would be our fault. It's probably a bug in his code.

vltr commented on 2021-04-23 14:13 (UTC)

@coderkun, I already listed that on the forums:

I'm still looking for a solution for that, btw :-)

coderkun commented on 2021-04-23 06:58 (UTC)

For a week or two I have been getting the following error when trying to start a docker container:

Error response from daemon: OCI runtime create failed: container_linux.go:367: starting container process caused: process_linux.go:495: container init caused: process_linux.go:458: setting cgroup config for procHooks process caused: can't load program: invalid argument: unknown

It work without error on Arch’s default kernel. Is anyone else experience the same and knows what the issue might be?


artafinde commented on 2021-04-22 10:00 (UTC)

@graysky: You can remove the hack for xz since 70261 is fixed.

graysky commented on 2021-04-14 15:06 (UTC)

I sure love updating my kernel 4 times in a week, lol

@powerball253- Comments like this serve absolutely no purpose. In general, this package follows the progress of the official kernel which has bumped several times in the past week.[1] Further, no one is forcing you to rebuild a kernel you opted into using in the first place. Please don't spam the comments page with such statements.


graysky commented on 2021-04-02 13:32 (UTC)

@Kr1ss - Updated in 5.11.11-4, thanks.

Kr1ss commented on 2021-04-02 12:21 (UTC) (edited on 2021-04-02 12:41 (UTC) by Kr1ss)

@graysky - Currently, building fails in environments with a modified SRCDEST folder. (EDIT/ For instance, mine is set to /tmp/makepkg/srcdest)

linux-ck-5.11.11-3-x86_64-prepare.log :

xz: /startdir/patch-5.11-ck1.xz: No such file or directory
==> ERROR: A failure occurred in prepare().
==> Removing installed dependencies...
checking dependencies...

Hence I'd like to suggest a little patch :

diff -U1 old/PKGBUILD new/PKGBUILD
--- old/PKGBUILD    2021-04-02 14:11:24.598889831 +0200
+++ new/PKGBUILD    2021-04-02 14:11:02.978925899 +0200
@@ -107,3 +107,3 @@
     unlink "$srcdir/$_ckpatch.xz"
-    xz -dc "$startdir/$_ckpatch.xz" > "$_ckpatch"
+    xz -dc "$SRCDEST/$_ckpatch.xz" > "$_ckpatch"

Thank you, cheers !

artafinde commented on 2021-04-02 11:01 (UTC) (edited on 2021-04-02 11:01 (UTC) by artafinde)

@graysky: Thanks already using it - noticed the PR from sirlucian :) Thanks!

graysky commented on 2021-04-02 10:05 (UTC)

@Zalgo - 5.11.11-3 should fix this. See the forum thread for discussion:

graysky commented on 2021-04-02 10:05 (UTC)

@artafinde - with some help, the native option now works with Intel and AMD. You will have to select the correct option though:

  33. Intel-Native optimizations autodetected by GCC (MNATIVE_INTEL) (NEW)
  34. AMD-Native optimizations autodetected by GCC (MNATIVE_AMD) (NEW)

Zalgo commented on 2021-04-02 07:20 (UTC)

Thanks for the tip :) If i figure out a way to fix the PKGBUILD that doesn't require manual untarring I'll pose it here

graysky commented on 2021-04-01 20:20 (UTC)

That's odd. Makepkg doesn't seem to be untarring the patch. If you do it manually and add that to the source array, it works as expected.

Zalgo commented on 2021-04-01 17:26 (UTC)

This currently fails on my system with

Setting config...
sed: can't read ../patch-5.11-ck1: No such file or directory
==> ERROR: A failure occurred in prepare().

can anyone else reproduce? I'll test in a clean chroot

graysky commented on 2021-03-07 19:38 (UTC)

@artafinde - I need to dig deeper, but I think it will be difficult to assign a native CPU out of some configs.

For example, X86_INTEL_USERCOPY config. I do that now but no other AMD chips are listed by upstream[1].

Further, I believe AMD chips should not be in the X86_P6_NOP config[2] or X86_CMOV config[3].




artafinde commented on 2021-03-07 18:50 (UTC) (edited on 2021-03-07 18:50 (UTC) by artafinde)

@graysky checksum for the gcc kernel patch is failing - needs an update.

I noticed the changes in the patch repo and it mentions one should not use native on AMD chips. What's the reasoning behind this since the gcc 10 (and clang for what matters) can detect the cpu arch just fine:

$ gcc -march=native -E -v - </dev/null 2>&1 | grep cc1
 /usr/lib/gcc/x86_64-pc-linux-gnu/10.2.0/cc1 -E -quiet -v - -march=znver2 -mmmx -mno-3dnow -msse -msse2 -msse3 -mssse3 -msse4a -mcx16 -msahf -mmovbe -maes -msha -mpclmul -mpopcnt -mabm -mno-lwp -mfma -mno-fma4 -mno-xop -mbmi -mno-sgx -mbmi2 -mno-pconfig -mwbnoinvd -mno-tbm -mavx -mavx2 -msse4.2 -msse4.1 -mlzcnt -mno-rtm -mno-hle -mrdrnd -mf16c -mfsgsbase -mrdseed -mprfchw -madx -mfxsr -mxsave -mxsaveopt -mno-avx512f -mno-avx512er -mno-avx512cd -mno-avx512pf -mno-prefetchwt1 -mclflushopt -mxsavec -mxsaves -mno-avx512dq -mno-avx512bw -mno-avx512vl -mno-avx512ifma -mno-avx512vbmi -mno-avx5124fmaps -mno-avx5124vnniw -mclwb -mmwaitx -mclzero -mno-pku -mrdpid -mno-gfni -mno-shstk -mno-avx512vbmi2 -mno-avx512vnni -mno-vaes -mno-vpclmulqdq -mno-avx512bitalg -mno-avx512vpopcntdq -mno-movdiri -mno-movdir64b -mno-waitpkg -mno-cldemote -mno-ptwrite -mno-avx512bf16 -mno-enqcmd -mno-avx512vp2intersect --param l1-cache-size=32 --param l1-cache-line-size=64 --param l2-cache-size=512 -mtune=znver2

vp1981 commented on 2021-02-23 23:59 (UTC)

@simona, AFAIU, graysky keeps config sync with linux package and these two are in linux-zen but not yet in linux.

simona commented on 2021-02-21 23:19 (UTC)

ashmem_linux and binder_linux missing?

vp1981 commented on 2021-02-08 00:53 (UTC)

@Krlss, same here.

@graysky, why not use Like that 0003-SUNRPC-Fix-NFS-READs-the-start-at-non-page-aligned-o.patch::

P.S. I like the beginning of commit hash ;)

Kr1ss commented on 2021-02-07 19:15 (UTC) (edited on 2021-02-07 19:18 (UTC) by Kr1ss)

That's weird... Github doesn't always serve the exact same patch !

Sometimes when you download it, the object hashes are abbreviated to 13 characters and sometimes to 14.

That's why the b2sum didn't match here, the first time.

Kr1ss commented on 2021-02-07 18:43 (UTC)

@graysky pls disregard my last comment. Seems like I had an issue with an unclean chroot.

graysky commented on 2021-02-07 18:42 (UTC)

@Kr1ss - Matches here. It is pulled as a source/not included with package.

Kr1ss commented on 2021-02-07 18:30 (UTC) (edited on 2021-02-07 18:36 (UTC) by Kr1ss)

@graysky - The checksum of 0003-SUNRPC-Fix-NFS-READs-the-start-at-non-page-aligned-o.patch doesn't match after the latest update. ;)

EDIT the file is missing from the AUR sources.

graysky commented on 2021-02-04 20:27 (UTC)

@artafinde - Yes, it was a typo. It doesn't hurt anything.

artafinde commented on 2021-02-04 20:03 (UTC) (edited on 2021-02-04 20:04 (UTC) by artafinde)

Latest update bumbed version to 5.10.13 but forgot to reset to 1 the pkgrel - not a real issue, but just a reminde for next bump.

bryanjhv commented on 2021-01-09 02:30 (UTC)

@graysky thanks for the update but... I think it's missing a $ ($pkgbase instead of just pkgbase).

graysky commented on 2021-01-09 02:07 (UTC)

@Distorted - It is undone in the PKGBUILD. Maintaining is easier if I take the Arch config as-is. See and else where in the PKGBUILD other Arch specific calls are commented out.

graysky commented on 2021-01-09 02:03 (UTC) (edited on 2021-01-09 02:07 (UTC) by graysky)

@bryanjhv - sure, 5.10.5-2

Distorted commented on 2021-01-09 00:32 (UTC)

Why is CONFIG_DEBUG_INFO=y set? It causes builds to use over 10GB in disk space.

bryanjhv commented on 2021-01-08 22:13 (UTC)

@graysky, is there possibility to change _package-headers' depends array using "$pkgbase" instead of fixed 'linux-ck'? I was building for silvermont (changing $pkgbase to linux-ck-silvermont the way some repositories provide it), and didn't notice it until installing linux-ck-silvermont-headers asked me to also install linux-ck instead of linux-ck-silvermont.

bryanjhv commented on 2021-01-08 22:05 (UTC) (edited on 2021-01-08 22:08 (UTC) by bryanjhv)

Ignoring the error would make things unpredictable. Just remove "-N" option so patch will try to reverse-apply and re-apply.

Kr1ss commented on 2021-01-03 11:14 (UTC)

@LucidComplex That would make it pass though also in situations where the patch itself fails for a different reason. Not sure if that's what you want to achieve.

LucidComplex commented on 2021-01-03 10:47 (UTC)

prepare() fails when patches have already been applied. I suppose you can append a || true to the end of patch commands in prepare() to address this.

Kr1ss commented on 2021-01-02 21:24 (UTC) (edited on 2021-01-02 22:06 (UTC) by Kr1ss)

@graysky The checksum for config doesn't match in the recent release.

EDIT/ Thank you fixing this so quickly ! :)

graysky commented on 2021-01-02 20:36 (UTC)

My bad... git push didn't respond for some reason.

Scimmia commented on 2021-01-02 19:32 (UTC)

I haven't found 5.10.4-2, but as long as linux-headers isn't in the provides array anymore, I'm fine with it.

graysky commented on 2021-01-01 14:33 (UTC)

@scimmia - Please review 5.10.4-2 and comment

Scimmia commented on 2021-01-01 14:24 (UTC)

If packages require that and aren't building against the 'linux' package, they're poorly packaged and should be fixed there, not hacked around here.

graysky commented on 2021-01-01 14:20 (UTC)

At one time there was a need to have the linux-ck-headers package provide linux-headers. Thought it had to do with other packages requiring for actual linux-headers package which is wrong if a user is running linux-ck not linux.

Scimmia commented on 2021-01-01 13:58 (UTC)

See the previous link, this should not be providing linux-headers. Packages saying they provide themselves makes no sense, either.

graysky commented on 2021-01-01 13:53 (UTC)

@Scimmia - Context? "Fixed?"

Scimmia commented on 2021-01-01 13:22 (UTC)

3 updates and the provides has not been fixed. Is there a problem, graysky?

Scimmia commented on 2020-12-11 15:51 (UTC)

Please see the warning here:

graysky commented on 2020-11-28 11:41 (UTC) (edited on 2020-11-28 11:42 (UTC) by graysky)

@nTia89 - fixed in 5.9.11-2

nTia89 commented on 2020-11-28 08:14 (UTC)

If anyone as issues with suspend/poweroff, this is the reason why: Just to share

graysky commented on 2020-10-20 13:10 (UTC)

@nTia89 - My bad. Order of events in the PKGBUILD matter. Fixed in 5.9.1-2

nTia89 commented on 2020-10-20 08:58 (UTC)

Since last update, there is no more the interactive micro-architecture selection before compilation. Does anyone have the same issue?

air-g4p commented on 2020-08-24 08:36 (UTC)

LOLs!!! I cannot fucking keep up!


  1. First we had a fucked up 'unfuck-xx' patch
  2. Then the fucked up unfuck-xx patch got unfucked
  3. Then @graysky accidentally fucked up something else
  4. Finally, @graysky unfucked the most recent fuck up

The bottom line: @graysky, happy to report that linux-ck 5.7.17-1 built and installed without any fuck ups! Cheers!

Kr1ss commented on 2020-08-23 13:33 (UTC)

Working fine here.

Thx a lot @ooo & @graysky, cheers !

graysky commented on 2020-08-23 12:07 (UTC)

@ooo - Thanks for the fix. I do not have time to build/test. Can someone please verify that 5.7.17-1 works as expected?

ooo commented on 2020-08-23 11:02 (UTC) (edited on 2020-08-23 11:15 (UTC) by ooo)

@graysky Apparently the error with 5.7.17 was fixed in zen-kernel already, although I didn't test it myself yet.

See the commits from 2020-08-22 at

EDIT: Applying the fix from zen-sources actually gets a bit confusing, so to make this a little simpler, first apply this to patch-5.7-ck1, to make it apply cleanly against 5.7.17:

Then apply this to fix the build:

graysky commented on 2020-08-21 13:15 (UTC)

5.7.17 introduces some changes that break the ck1 patch. If someone wants to take a look at that and share a solution, that would be good. I am busy at the moment.

enihcam commented on 2020-08-20 23:19 (UTC)

graysky commented on 2020-08-20 17:27 (UTC) (edited on 2020-08-20 17:27 (UTC) by graysky)

Yes, I fucked it up, should be fixed with

You don't need to rebuild unless you have [testing] enabled or have it enabled in a clean build root btw.

Kr1ss commented on 2020-08-20 17:11 (UTC) (edited on 2020-08-20 17:16 (UTC) by Kr1ss)

The recent pkgrel bump broke the validity of the config checksum, @graysky. Could you update that pls ?

Thank you !

air-g4p commented on 2020-08-09 15:18 (UTC) (edited on 2020-08-09 15:25 (UTC) by air-g4p)

@graysky - 5.7.14-1-ck compiled and installed without issue. Good work, done here!

Kr1ss commented on 2020-08-07 13:02 (UTC)

@graysky I'm running 5.7.14-1-ck right now, built w/ localmodconfig.

Looks fine so far, thx.

graysky commented on 2020-08-07 12:39 (UTC)

I cannot reboot now so can someone verify that 5.7.14-ck1 is working properly for me? A small adjustment to ck1 was required.

mrkline commented on 2020-07-31 01:06 (UTC) (edited on 2020-07-31 01:07 (UTC) by mrkline)

Inexplicably, this I ran into the SHA256 issue on one of my machines but not the other. I diffed the two versions of unfuck-ck1.patch - check it out:

---  2020-07-30 18:00:17.955122911 -0700
+++ unfuck-ck1.patch.borked 2020-07-30 18:01:49.000000000 -0700
@@ -8,7 +8,7 @@
  1 file changed, 11 insertions(+)

 diff --git a/kernel/sched/MuQSS.c b/kernel/sched/MuQSS.c
-index 18a9b4a23e44e..c55c158e8b8e4 100644
+index 18a9b4a23e44ea..c55c158e8b8e42 100644
 --- a/kernel/sched/MuQSS.c
 +++ b/kernel/sched/MuQSS.c
 @@ -3246,6 +3246,17 @@ static inline u64 do_task_delta_exec(struct task_struct *p, struct rq *rq)

It seems like is inconsistent in how many digits of the Git SHAs it sends.

vltr commented on 2020-07-27 15:27 (UTC)

I think it's just a wild guess (or just my guts), but I think this might be some caching problem when downloading that particular file. Sometimes you might get it from a different server with a different content (even though this is odd, yes), but might explain why sometimes (and how inconsistently) this happens ...

air-g4p commented on 2020-07-27 05:15 (UTC) (edited on 2020-07-27 05:17 (UTC) by air-g4p)

@graysky and @artafinde: Thanks for trying to reproduce on your respective ends.

Inexplicably, the build completed without any PKGBUILD SHA256 modifications on my other laptop, and again running makepkg -sric. I suppose this success is consistent with @artafinde's '50% of the time' finding when running makepkg -sric.

Cheers guys...

artafinde commented on 2020-07-25 16:38 (UTC) (edited on 2020-07-25 16:40 (UTC) by artafinde)

I was able to reproduce it "consistently" only when using `makepkg -srci` but only 50% of the times. I think it could be related to `` and a combination of makepkg flags. ``` inglor@arch ~/cowerPkg$ auracle clone linux-ck clone complete: /home/inglor/cowerPkg/linux-ck inglor@arch ~/cowerPkg$ cd linux-ck inglor@arch ~/cowerPkg/linux-ck$ makepkg -srci master ==> Making package: linux-ck 5.7.10-1 (Sat 25 Jul 2020 16:35:13 UTC) ==> Checking runtime dependencies... ==> Checking buildtime dependencies... ==> Retrieving sources... -> Downloading linux-5.7.10.tar.xz... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 162 100 162 0 0 2131 0 --:--:-- --:--:-- --:--:-- 2131 100 107M 100 107M 0 0 16.8M 0 0:00:06 0:00:06 --:--:-- 19.2M -> Downloading linux-5.7.10.tar.sign... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 162 100 162 0 0 1653 0 --:--:-- --:--:-- --:--:-- 1653 100 989 100 989 0 0 6181 0 --:--:-- --:--:-- --:--:-- 6181 -> Found config -> Downloading enable_additional_cpu_optimizations-20200615.tar.gz... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 135 100 135 0 0 1928 0 --:--:-- --:--:-- --:--:-- 1928 100 21833 0 21833 0 0 44376 0 --:--:-- --:--:-- --:--:-- 44376 -> Downloading patch-5.7-ck1.xz... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 91244 100 91244 0 0 350k 0 --:--:-- --:--:-- --:--:-- 349k -> Downloading unfuck-ck1.patch... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 909 100 909 0 0 3404 0 --:--:-- --:--:-- --:--:-- 3417 -> Downloading unfuck-ck1-fix-suspend-to-ram.patch... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 1028 100 1028 0 0 3835 0 --:--:-- --:--:-- --:--:-- 3835 -> Found 0000-sphinx-workaround.patch -> Found 0001-ZEN-Add-sysctl-and-CONFIG-to-disallow-unprivileged-C.patch -> Found 0002-PCI-EDR-Log-only-ACPI_NOTIFY_DISCONNECT_RECOVER-even.patch -> Found 0003-iwlwifi-Make-some-Killer-Wireless-AC-1550-cards-work.patch -> Found 0004-virt-vbox-Add-support-for-the-new-VBG_IOCTL_ACQUIRE_.patch ==> Validating source files with sha256sums... linux-5.7.10.tar.xz ... Passed linux-5.7.10.tar.sign ... Skipped config ... Passed enable_additional_cpu_optimizations-20200615.tar.gz ... Passed patch-5.7-ck1.xz ... Passed unfuck-ck1.patch ... Passed unfuck-ck1-fix-suspend-to-ram.patch ... FAILED 0000-sphinx-workaround.patch ... Passed 0001-ZEN-Add-sysctl-and-CONFIG-to-disallow-unprivileged-C.patch ... Passed 0002-PCI-EDR-Log-only-ACPI_NOTIFY_DISCONNECT_RECOVER-even.patch ... Passed 0003-iwlwifi-Make-some-Killer-Wireless-AC-1550-cards-work.patch ... Passed 0004-virt-vbox-Add-support-for-the-new-VBG_IOCTL_ACQUIRE_.patch ... Passed ==> ERROR: One or more files did not pass the validity check! inglor@arch ~/cowerPkg/linux-ck$ sha256sum unfuck-ck1-fix-suspend-to-ram.patch 1 ↵ ✭master 9ad11f720d4dd83ddb8876b453f4526ad35f1ff72465213764fb1d206fea0379 unfuck-ck1-fix-suspend-to-ram.patch ```

graysky commented on 2020-07-25 14:58 (UTC) (edited on 2020-07-25 15:03 (UTC) by graysky)

@air-g4p - I cannot reproduce: ``` % wget % tar xvf linux-ck.tar.gz && cd linux-ck % makepkg -s ==> Making package: linux-ck 5.7.10-1 (Sat 25 Jul 2020 10:57:05 AM EDT) ==> Checking runtime dependencies... ==> Checking buildtime dependencies... ==> Retrieving sources... -> Downloading linux-5.7.10.tar.xz... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 162 100 162 0 0 1408 0 --:--:-- --:--:-- --:--:-- 1421 100 107M 100 107M 0 0 3221k 0 0:00:34 0:00:34 --:--:-- 3656k -> Downloading linux-5.7.10.tar.sign... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 162 100 162 0 0 1421 0 --:--:-- --:--:-- --:--:-- 1421 100 989 100 989 0 0 3129 0 --:--:-- --:--:-- --:--:-- 3129 -> Found config -> Downloading enable_additional_cpu_optimizations-20200615.tar.gz... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 135 100 135 0 0 457 0 --:--:-- --:--:-- --:--:-- 457 100 21833 100 21833 0 0 42394 0 --:--:-- --:--:-- --:--:-- 42394 -> Downloading patch-5.7-ck1.xz... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 91244 100 91244 0 0 97482 0 --:--:-- --:--:-- --:--:-- 97378 -> Downloading unfuck-ck1.patch... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 909 100 909 0 0 3509 0 --:--:-- --:--:-- --:--:-- 3509 -> Downloading unfuck-ck1-fix-suspend-to-ram.patch... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 1026 100 1026 0 0 4007 0 --:--:-- --:--:-- --:--:-- 4007 -> Found 0000-sphinx-workaround.patch -> Found 0001-ZEN-Add-sysctl-and-CONFIG-to-disallow-unprivileged-C.patch -> Found 0002-PCI-EDR-Log-only-ACPI_NOTIFY_DISCONNECT_RECOVER-even.patch -> Found 0003-iwlwifi-Make-some-Killer-Wireless-AC-1550-cards-work.patch -> Found 0004-virt-vbox-Add-support-for-the-new-VBG_IOCTL_ACQUIRE_.patch ==> Validating source files with sha256sums... linux-5.7.10.tar.xz ... Passed linux-5.7.10.tar.sign ... Skipped config ... Passed enable_additional_cpu_optimizations-20200615.tar.gz ... Passed patch-5.7-ck1.xz ... Passed unfuck-ck1.patch ... Passed unfuck-ck1-fix-suspend-to-ram.patch ... Passed 0000-sphinx-workaround.patch ... Passed 0001-ZEN-Add-sysctl-and-CONFIG-to-disallow-unprivileged-C.patch ... Passed 0002-PCI-EDR-Log-only-ACPI_NOTIFY_DISCONNECT_RECOVER-even.patch ... Passed 0003-iwlwifi-Make-some-Killer-Wireless-AC-1550-cards-work.patch ... Passed 0004-virt-vbox-Add-support-for-the-new-VBG_IOCTL_ACQUIRE_.patch ... Passed ==> Verifying source file signatures with gpg... linux-5.7.10.tar ... Passed ==> Extracting sources... -> Extracting linux-5.7.10.tar.xz with bsdtar ```

air-g4p commented on 2020-07-25 07:26 (UTC) (edited on 2020-07-25 08:07 (UTC) by air-g4p)

@graysky - I will be the third person to confirm there is a SHA256 error with the unfuck-ck1-fix-suspend-to-ram.patch in PKGBUILD. `makepkg -sric` fails almost immediately due to this SHA256 validation error. In fact, if I comment out the currently existing value that ends in xxx3558 and use: 9ad11f720d4dd83ddb8876b453f4526ad35f1ff72465213764fb1d206fea0379 instead, as vltr originally noted, the build works fine. BTW, that substituted sha256 value I used matches the sha256sum of the unfuck-ck1-fix-suspend-to-ram.patch downloaded by PKGBUILD, which is why the build succeeds. 9ad11f720d4dd83ddb8876b453f4526ad35f1ff72465213764fb1d206fea0379 unfuck-ck1-fix-suspend-to-ram.patch Cheers

oldsadsongs commented on 2020-07-23 17:24 (UTC)

@graysky I was able to reproduce @vltr's report by typing 'makepkg -si' instead of 'makepkg' on a fresh git clone.

vltr commented on 2020-07-23 14:54 (UTC)

@graysky that's ODD. I always build after cleaning up the git repository (git clean -dfx) ... and the file had that signature I mentioned (hence my build failed). I cleaned (again), ran makepkg and everything went smoothly this time (with unfuck-ck1-fix-suspend-to-ram.patch signature being, indeed, 961ed94b8d905f1e901cacb08d253c4170af0a25828111b7558d9c874e923558). that's really weird ... sorry about the confusion, whatever it happened.

graysky commented on 2020-07-23 13:22 (UTC)

@vltr - You sure?

% sha256sum unfuck-ck1-fix-suspend-to-ram.patch
961ed94b8d905f1e901cacb08d253c4170af0a25828111b7558d9c874e923558  unfuck-ck1-fix-suspend-to-ram.patch

vltr commented on 2020-07-23 12:53 (UTC)

hi there, @graysky! just one problem with 5.7.10: the unfuck-ck1-fix-suspend-to-ram.patch has a sha256sum of value 961ed94b8d905f1e901cacb08d253c4170af0a25828111b7558d9c874e923558 in the PKGBUILD, which would not pass the validity check since the correct sha256sum should be 9ad11f720d4dd83ddb8876b453f4526ad35f1ff72465213764fb1d206fea0379. apart from that, everything works fine. thanks!

nTia89 commented on 2020-06-12 16:05 (UTC)

Above all, thanks @graysky for the change! Anyway, /tmp defaults to tmpfs with size=50% but not everybody has so much RAM. By my own, this is the biggest mistake behind that choice...

dimosd commented on 2020-06-12 15:23 (UTC)

I suggest that you disable CONFIG_DEBUG_INFO, as it was before 5.7.2. That's what I am doing in my own builds anyway to build with a 4GB /tmp.

artafinde commented on 2020-06-12 12:23 (UTC)

graysky: Got my answer:

artafinde commented on 2020-06-12 10:14 (UTC)

Kr1ss: I think devs would probably reply don't build in tmpfs (and require ~15G of space instead of ~3.5G) but that's a discussion for forums possibly :D - Thanks graysky.

Kr1ss commented on 2020-06-12 09:54 (UTC)

Thx @graysky ! I was only able to build 5.7.2-1 after blowing up /tmp to 20GiB, that's kind of ridiculous. If a user needs debug symbols they can enable them; I think most users wouldn't need'em anyways.

Cheers !

graysky commented on 2020-06-12 09:21 (UTC)

@artafinde - Fuck it, 5.7.2-2-ck1. Change my mind.

artafinde commented on 2020-06-12 09:13 (UTC)

@graysky - Agree, especially since later we strip it from the bug symbols. I raised the question in IRC but no answer. I guess forums or a bug might have better luck.

graysky commented on 2020-06-12 07:53 (UTC) (edited on 2020-06-12 08:16 (UTC) by graysky)

@artafinde - Yes, the debug stuff causes a failure on one of my machines with 16G of memory building in tmpfs (allocating 14.5G to the tmpfs partition on which the build root resides!)

This change was introduced with 5.7.2-1-arch1 here:

First question is why do we need the debug options?

"If you say Y here the resulting kernel image will include debugging info resulting in a larger kernel image. This adds debug symbols to the kernel and modules (gcc -g), and is needed if you intend to use kernel crashdump or binary object tools like crash, kgdb, LKCD, gdb, etc on the kernel. Say Y here only if you plan to debug the kernel. If unsure, say N."

msw1 commented on 2020-06-12 06:08 (UTC)

I am getting the same error with /tmp running out of space as of version 5.7.2-1.

artafinde commented on 2020-06-11 21:39 (UTC) (edited on 2020-06-11 21:43 (UTC) by artafinde)

nTia89: The default configuration changed (now we produce some debug info) which we strip on packaging and creation of vmlinuz. See config and change in 5.7.1 ( Config change comes from upstream (arch default config. PS: In my case with 16G /tmp was filled while building without any change in config or PKGBUILD. I suggest to not use tmpfs for now.

nTia89 commented on 2020-06-11 21:14 (UTC) (edited on 2020-06-11 21:17 (UTC) by nTia89)

Staring with 5.7.1, I cannot compile kernel because I run out of space on my /tmp (tmpfs with size of 6GB). Did it change anything recently?

artafinde commented on 2020-06-11 15:59 (UTC)

@as3ii: nvidia-dkms 440.82-2 works for -ck kernel. No idea about bbswitch.

as3ii commented on 2020-06-11 15:53 (UTC)

linux-ck 5.7.2 compiled correctly, but bbswitch and nvidia doesn't work (dkms install error)

Error! Bad return status for module build on kernel: 5.7.2-1-ck

vltr commented on 2020-06-11 15:11 (UTC)

linux-ck 5.7.2 compiled successfully and already up and running. Thanks, @graysky!

duff commented on 2020-06-11 12:29 (UTC)

does not cleanbuild -> 5.7.1-2 tried on two different archlinux hosts (bulldozer und core2)

please fix that

graysky commented on 2020-06-11 10:33 (UTC)

@kwe - Yes, I saw that posted by sir_lucjan. I am reluctant just hack it together without CK's review. See his blog.

kwe commented on 2020-06-11 10:24 (UTC) (edited on 2020-06-11 10:25 (UTC) by kwe)

Getting the same error. It seems to be just that single function that seems to be missing for the compilation to finish, though.

I just added the function that was introduced in to the top part of drivers/thermal/cpufreq_cooling.c and it went through.

Not sure if anything else is missing, though, but please don't consider this a proper fix; it might be dangerous, even. I'm going with this solution, for now, because I have accepted the risk.

henrik_er commented on 2020-06-11 08:51 (UTC)

@graysky yes, exact same error output.

LinguinePenguiny commented on 2020-06-10 16:50 (UTC) (edited on 2020-06-10 16:51 (UTC) by LinguinePenguiny)

As of 5.7.1-2 for me:

LD .tmp_vmlinux.btf ld: drivers/thermal/cpufreq_cooling.o: in function cpufreq_set_cur_state': /opt/makepkg/build/linux-ck/src/linux-5.7.1/drivers/thermal/cpufreq_cooling.c:458: undefined reference toarch_set_thermal_pressure' BTF .btf.vmlinux.bin.o scripts/ line 114: 613105 Segmentation fault (core dumped) LLVM_OBJCOPY=${OBJCOPY} ${PAHOLE} -J ${1} LD .tmp_vmlinux.kallsyms1 .btf.vmlinux.bin.o: file not recognized: file format not recognized make: *** [Makefile:1117: vmlinux] Error 1 ==> ERROR: A failure occurred in build().

sir_lucjan commented on 2020-06-10 13:39 (UTC) (edited on 2020-06-10 13:40 (UTC) by sir_lucjan)


Look at this. I had a similar error using BMQ. Con apparently skipped resync.

vltr commented on 2020-06-10 13:38 (UTC) (edited on 2020-06-10 13:38 (UTC) by vltr)

@graysky yes, first I was going to mention here to add pahole as a dependency but as I was writing it down, you updated the package. Since then, I've been having the error you described.

graysky commented on 2020-06-10 13:34 (UTC)

Is anyone else getting this error:

  GEN     .version
  CHK     include/generated/compile.h
  LD      vmlinux.o
  MODPOST vmlinux.o
  MODINFO modules.builtin.modinfo
  GEN     modules.builtin
  LD      .tmp_vmlinux.btf
ld: drivers/thermal/cpufreq_cooling.o: in function `cpufreq_set_cur_state':
/scratch/linux-ck/src/linux-5.7.1/drivers/thermal/cpufreq_cooling.c:458: undefined reference to `arch_set_thermal_pressure'
  BTF     .btf.vmlinux.bin.o
tag__check_id_drift: subroutine_type id drift, core_id: 1129, btf_type_id: 1127, type_id_off: 0
libbpf: Unsupported BTF_KIND:0
btf_elf__encode: btf__new failed!
free(): double free detected in tcache 2
scripts/ line 114: 3291862 Aborted                 (core dumped) LLVM_OBJCOPY=${OBJCOPY} ${PAHOLE} -J ${1}
  LD      .tmp_vmlinux.kallsyms1
.btf.vmlinux.bin.o: file not recognized: file format not recognized
make: *** [Makefile:1117: vmlinux] Error 1

graysky commented on 2020-04-30 14:25 (UTC)

@stano00 - No, you are correct. Fixed in 5.6.8-2

stano00 commented on 2020-04-30 00:51 (UTC) (edited on 2020-04-30 00:52 (UTC) by stano00)

Is the sphinx-workaround.patch actually being applied by your PKGBUILD? I didn't see it print while the other patches were applied.

nTia89 commented on 2020-03-21 10:06 (UTC)

@graysky @kittenlol I am aware of those two lines, since the change happened years(?) ago. Anyway, using #31 architecture and saying 'y' to P6_NOPs question have never given me any issue. Why?

graysky commented on 2020-03-13 11:14 (UTC)

Gets stuck at asking for X86_P6_NOP with _subarch=31

@kittenlol - See commented line 15 and 16.

kittenlol commented on 2020-03-13 11:09 (UTC)

Gets stuck at asking for X86_P6_NOP with _subarch=31

nTia89 commented on 2020-02-16 21:38 (UTC)


Geeko commented on 2020-02-15 18:40 (UTC)

config checksum don't match

blitz commented on 2020-02-04 01:00 (UTC)

@graysky - `Merge branch '5.5/muqss' into 5.5/master'

graysky commented on 2020-01-30 20:42 (UTC)

@redfang - It's in the output.

  25. Intel Skylake (MSKYLAKE)
  26. Intel Skylake X (MSKYLAKEX)

normalname012 commented on 2020-01-30 09:47 (UTC)

what option should i choose for intel skylate cpu (7th gen) during installation

graysky commented on 2020-01-28 20:07 (UTC)

@dwzg - Not out of date/no ck1. See:

Bambus89 commented on 2020-01-28 06:05 (UTC)

@graysky @superice97 thx

SuperIce97 commented on 2020-01-25 20:20 (UTC)

@Bambus89 MuQSS uses subtick accounting and tickless expiry, and thus is unaffected by the kernel's tick rate. The only thing that would change by increasing the tick rate from 100 to 1000 would be a slight increase in power consumption and a slight decrease in throughput due to increased overhead. This is why the MuQSS patches automatically set 100Hz as the default. This is talked about in the MuQSS design doc (search for "Tickless expiry"):

graysky commented on 2020-01-25 14:06 (UTC)

@bambus89- ck patchset uses this value

Bambus89 commented on 2020-01-25 10:32 (UTC)

Can you set a Option in pkgconfig for set 1000hz? Why is the kernel set to 100hz and are there benefits for 100hz?

graysky commented on 2019-12-31 18:48 (UTC)

Ah, you were asking for a build in [repo-ck]. I can add on with 5.4.7-1-ck.

trthomps commented on 2019-12-31 16:19 (UTC) (edited on 2019-12-31 16:21 (UTC) by trthomps)

@graysky - ah, yes I know I can build my own, was hoping to get a pre-build binary ;-)

Happy to contribute cpu time if that's needed for repo-ck

graysky commented on 2019-12-31 10:28 (UTC)

@trthomps - You can ... it is option 12. See the PKGBUILD.

trthomps commented on 2019-12-31 09:13 (UTC)

Would it be possible to add a zen2 build as well? I'm running zen successfully but it looks like amd added additional optimizations for the latest zen 2 arch.

[travis@travis-pc ~]$ gcc -c -Q -march=native --help=target | grep march
  -march=                           znver2
  Known valid arguments for -march= option:

blazessdd commented on 2019-12-20 12:05 (UTC)

can't add comment a long comment... even if if i code put it in code code

graysky commented on 2019-12-12 17:46 (UTC)

@blazessdd - Can you post a build log and link it here?

blazessdd commented on 2019-12-12 16:01 (UTC)

i have set _localmodcfg=y but whenever i try to compile modprobed wont show up and i get a kernel with all the modules compile what can i do?

graysky commented on 2019-11-28 14:14 (UTC) (edited on 2019-11-28 14:14 (UTC) by graysky)

@Althorion -

# Note - the march=native option is unavailable by this method, use the nconfig
# and manually select it.

Althorion commented on 2019-11-28 13:13 (UTC) (edited on 2019-11-28 13:14 (UTC) by Althorion)

Tired with clean PKGBUILD and -march=native (option 31). Manual patching works just fine, no idea why.

BTW (unrelated problem, I think), I tried manually patching the sources, removing patching from the PKGBUILD, and when I put some value for _subarch=, it fails at make oldconfig—it tries giving it the value I put in _subarch= (that would be 31), thus entering the never-ending loop of wrong arguments.

Finally, I managed to successfully build the package by manual patching and giving it empty _subarch.

graysky commented on 2019-11-28 13:06 (UTC)

Need more to go on in order to reproduce. Any edits to the PKGBUILD? Which number are you selecting on the gcc patch stage?

Althorion commented on 2019-11-28 12:46 (UTC)

No, I manually cloned the package (git clone <>) and run makepkg -i on it. I tried removing the folder and re-cloning and clearing the build folder (rm -rf /tmp/makepkg/linux-ck), but to no avail.

graysky commented on 2019-11-28 12:33 (UTC) (edited on 2019-11-28 12:34 (UTC) by graysky)

How are you building? Works fine here. If you're using an AUR helper, DON'T USE AN AUR HELPER!

patching file kernel/time/Kconfig
patching file kernel/time/clockevents.c
patching file kernel/time/hrtimer.c
patching file kernel/time/posix-cpu-timers.c
patching file kernel/time/timer.c
patching file kernel/trace/trace_selftest.c
patching file mm/vmscan.c
patching file net/core/pktgen.c
patching file sound/pci/maestro3.c
patching file sound/soc/codecs/rt5631.c
patching file sound/soc/codecs/wm8350.c
patching file sound/soc/codecs/wm8900.c
patching file sound/soc/codecs/wm9713.c
patching file sound/soc/soc-dapm.c
patching file sound/usb/line6/pcm.c
  HOSTCC  scripts/basic/fixdep
  HOSTCC  scripts/kconfig/conf.o
  HOSTCC  scripts/kconfig/confdata.o

Althorion commented on 2019-11-28 12:26 (UTC)

Version 5.4-3 fails at patching:

Last few lines:

patching file mm/vmscan.c
patching file net/core/pktgen.c
patching file sound/pci/maestro3.c
patching file sound/soc/codecs/rt5631.c
patching file sound/soc/codecs/wm8350.c
patching file sound/soc/codecs/wm8900.c
patching file sound/soc/codecs/wm9713.c
patching file sound/soc/soc-dapm.c
patching file sound/usb/line6/pcm.c
?[1m    Aborting...?(B?[m

graysky commented on 2019-11-23 22:26 (UTC)

No idea what pikaur is but if you are using an i686 fork of Arch, see man makepkg to see how to ignore the architecture or modify the PKGBUILD itself.

lockheed commented on 2019-11-23 21:11 (UTC)

How can I build this on a 32bit system? I tried it now with pikaur, but I'm getting "linux-ck, linux-ck-headers can't be built on the current arch (pentium4). Supported: x86_64"

I know Arch abandoned 32 bit, but this is a source package so I expected it could be built.

graysky commented on 2019-11-11 18:53 (UTC)

@kyak - Must have missed that when diffing official PKGBUILD vs this. Modifycation will be included in 5.3.11, thanks.

kyak commented on 2019-11-11 18:02 (UTC)

Hi graysky,

Official kernels do now install kernel image to /boot anymore:

Do you plan to follow the same path with linux-ck?

nTia89 commented on 2019-11-08 08:05 (UTC) (edited on 2019-11-08 08:05 (UTC) by nTia89)

Fixed! @sir_lucjan thank you, I tried again and now it compiles flawless. Maybe the reboot fixed the issue, since I upgraded some packages, including mkinitcpio, systemd and gcc...

sir_lucjan commented on 2019-11-07 22:15 (UTC)

mkinitcpio is not involved in the compilation of the kernel at all. It is used only to generate initramfs and it is also possible to do it with dracut.

nTia89 commented on 2019-11-07 18:16 (UTC) (edited on 2019-11-07 18:17 (UTC) by nTia89)

With latest mkinitcpio-27-2, linux-ck compilation (in real it stops before...) stucks just after patching. No errors printed, just "failed in prepare()"

sir_lucjan commented on 2019-11-03 13:11 (UTC)


It's not a linux-ck problem, it's Manjaro. Due to recent updates in Arch Linux (kmod and mkinitcpio) hooks are no longer provided in kernel only in kmod (depmod-hook) and mkinitcpio (mkinitcpio-install and mkinitcpio remove). Unfortunately, at the moment, as a Manjaro user you will not install most of the current AUR kernels and the problem lies solely with your distribution, i.e. Manjaro.

toki1990 commented on 2019-11-03 13:02 (UTC)

I'm manjaro stable repo user. linux-ck 5.3.8-4 not booting. I'm tried linux-slim 5.3.8-2 custom kernel same problem. Xanmod 5.3.8-1 not doing this. Because problem is this:

"Updating linux-xanmod initcpios... ==> Building image from preset: /etc/mkinitcpio.d/linux-xanmod.preset: 'default' -> -k /boot/vmlinuz-linux-xanmod -c /etc/mkinitcpio.conf -g /boot/initramfs-linux-xanmod.img ==> Starting build: 5.3.8-xanmod6-1-xanmod -> Running build hook: [base] -> Running build hook: [udev] -> Running build hook: [autodetect]...."

Xanmod custom kernel doing this last of compilation. Linux-ck 5.3.8-4 not doing this.

graysky commented on 2019-11-01 20:46 (UTC)

@Superice - Yes, I am ignoring that particular change. For example, the substitution will make _srcver=5.3.8-arch1 in this case whereas this package will be 5.3.8-1-ck which is fine for me.

SuperIce97 commented on 2019-11-01 20:39 (UTC) (edited on 2019-11-01 20:40 (UTC) by SuperIce97)

The pkgver and _srcver variables changed in the main linux package:


graysky commented on 2019-11-01 18:48 (UTC)

@sir_lucjan - Good catch, thanks. Fixed in 5.3.8-4.

sir_lucjan commented on 2019-11-01 18:09 (UTC)


As far as I can see, you forgot to change one line.

-  echo "$_kernelname" > localversion.20-pkgname
+  echo "${pkgbase#linux}" > localversion.20-pkgname

air-g4p commented on 2019-10-31 13:36 (UTC) (edited on 2019-10-31 14:34 (UTC) by air-g4p)


It was interesting to see how much 'heat' my wireguard-dkms post generated, and perhaps several of us wireguard-dkms and -ck users have learned some important tips.

Bottom line = tldr version - no action is required provided that: the fou, ip_tunnel and wire modules are found within lsmod and/or modprobed-db list if you run modprobed-db.

Those 3 modules were NOT loaded within MY modprobed-db list, which was quite surprising, given that I have run wireguard-dkms and modprobed-db for a VERY LONG time!

However, once I re-ran: modprobed-db store, it found the three new modules.

Of course, during a subsequent, clean, rebuild of linux-ck, wireguard-dkms was successfully installed.

Thanks for your input graysky, and to the others who offered their comments.

laenco commented on 2019-10-30 22:38 (UTC) (edited on 2019-10-30 22:39 (UTC) by laenco)

@kwe: OK. I'll try to explain.

  1. localmodconfig disables every module, that not exactly listed in provided file

  2. the file modprobed.db filling automatically only by modules that are in memory when modprobed-db service runs.

  3. fou module did not get into the modprobed.db for some reason

  4. But it is needed as indirect dependency for wireguard-dkms

  5. when you build kernel with localmodconfig - fou module become disabled

  6. And consequent build of wireguard-dkms fails.

As I can see - the most simple way to fix such a problem - is to add fou/fou6 modules into modprobed.db manually.

Or - load them and run modprobed-db.service for the same thing.

Blame me, if I wrong, cause there is more coffee than blood in veins)

kwe commented on 2019-10-30 22:17 (UTC)

@laenco: Either I'm misunderstanding you or I have to disagree.
When digging though lsmod AFTER HAVING USED WIREGUARD (so I assume all the required modules have been loaded), I can't seem to find any specific modules that are NOT in modprobed.db. It seems to me like the part where .config is configured for the kernel compilation process is assembled in a wrong way, missing out on the loaded (!) modules. This doesn't seem to have anything to do with modprobed-db as far as I can tell (?).

laenco commented on 2019-10-30 22:13 (UTC)

@kwe: so it's look like wireguard needs fou module for dkms build/install, but does not load it during routine operations, so modprobed-db did not detect it.

You could try to add fou and fou6 modules to modprobed.db manually and rebuild.

kwe commented on 2019-10-30 22:06 (UTC)

@laenco: Here is my current system state:

$ grep -i fou ~/.config/modprobed.db
$ lsmod | grep -i fou
$ lsmod | grep -i wire
wireguard             233472  0
ip6_udp_tunnel         16384  1 wireguard
udp_tunnel             16384  1 wireguard
$ grep -i tun ~/.config/modprobed.db

When I run makepkg to extract the linux-ck src directory, I end up with this .config diff against the current running kernel config:

$ zdiff /proc/config.gz config.last 
> # CONFIG_NET_FOU is not set

laenco commented on 2019-10-30 21:51 (UTC) (edited on 2019-10-30 21:53 (UTC) by laenco)

@kwe could you check ~/.config/modprobed.db for presence of "fou"/"fou6" modules?

The question is "If localmodconfig is unable to detect fou or is it a modprobed-db detection problem"?.

kwe commented on 2019-10-30 21:28 (UTC)

@graysky I am building linux-ck with modprobed-db, and for some reason, I always end up with those configuration options being disabled. I have to re-enable them manually every single time.

graysky commented on 2019-10-30 20:11 (UTC)

% uname -r

% zgrep CONFIG_NET_FOU /proc/config.gz

Should be there as a module. Building wireguard-dkms on CK works for me:

# pacman -S wireguard-dkms
resolving dependencies...
looking for conflicting packages...

Packages (2) dkms-2.7.1-1  wireguard-dkms-0.0.20191012-1

Total Download Size:   0.24 MiB
Total Installed Size:  1.75 MiB

:: Proceed with installation? [Y/n] 
:: Retrieving packages...
 wireguard-dkms-0.0.20191012-1-x86_64            245.7 KiB  3.43 MiB/s 00:00 [############################################] 100%
(2/2) checking keys in keyring                                               [############################################] 100%
(2/2) checking package integrity                                             [############################################] 100%
(2/2) loading package files                                                  [############################################] 100%
(2/2) checking for file conflicts                                            [############################################] 100%
:: Processing package changes...
(1/2) installing dkms                                                        [############################################] 100%
Optional dependencies for dkms
    linux-headers: build modules against the Arch kernel [installed]
    linux-lts-headers: build modules against the LTS kernel
    linux-zen-headers: build modules against the ZEN kernel
    linux-hardened-headers: build modules against the HARDENED kernel
(2/2) installing wireguard-dkms                                              [############################################] 100%
:: Running post-transaction hooks...
(1/2) Arming ConditionNeedsUpdate...
(2/2) Install DKMS modules
==> dkms install wireguard/0.0.20191012 -k 5.3.7-arch1-2-ARCH
==> dkms install wireguard/0.0.20191012 -k 5.3.8-1-ck

Kr1ss commented on 2019-10-30 19:41 (UTC)

FWIW, I'm also setting that option manually, hence I'd support @air-g4p's request. Thx again for maintaining this btw !

Cheers !

air-g4p commented on 2019-10-30 19:36 (UTC)

@graysky - wireguard-dkms failed to build on CK with this output:

I asked the devs at #wireguard about this failure, and was told I needed to enable: CONFIG_NET_FOU

(FOU = foo over udp)

After enabling that and rebuilding linux-ck --> Success.

Per Jason's (zx2c4), the lead wireguard dev, suggestion, can you please enable CONFIG_NET_FOU by default on linux-ck?

Thanks graysky

graysky commented on 2019-10-27 14:46 (UTC) (edited on 2019-10-29 20:42 (UTC) by graysky)

@timo - OK... that is likely because your .config got tweaked due to your CC= line. I opened a flyspray shown below. See it for a solution:

timo_capa commented on 2019-10-27 14:03 (UTC)

@graysky What I did was add CC=clang before make in the PKGBUILD to use Clang as compiler.

What worked was using makepkg for compiling till it gets stuck, and then end it with make modules_install and make install from the src folder. Doesn't give a package for pacman, but works.

I've tried linux-pf the same way, though that got stuck at the same place :)

graysky commented on 2019-10-27 12:34 (UTC)

@simona - the recent failure of dkms is due to the fact that there are both a newer version of the toolchain and of the dkms packages in [testing] currently. The advice I have been given is that it always a good idea to build the kernel against [testing] to prevent version mismatch errors when toolchain packages come out of [testing] and into [core].

I need to figure out what the best option is now.

@timo_capa - I am not sure how you go about building in Clang... I am using the official Arch devtools for clean chroot building. Are you patching the kernel source with ck1?

simona commented on 2019-10-27 11:52 (UTC)

==> dkms install vboxhost/6.0.14_OSE -k 5.3.7-1-ck-skylake Error! Bad return status for module build on kernel: 5.3.7-1-ck-skylake (x86_64)

timo_capa commented on 2019-10-12 12:21 (UTC) (edited on 2019-10-12 14:43 (UTC) by timo_capa)

I'm not sure if this is something you're interested in fixing, but compiling the kernel with Clang gets stuck at Installing boot image.... I'm assuming it's spamming something related to Kconfigs in the background related to GCC, as cancelling the build shows

"install: cannot stat '*'$'\n''* Restart config...'$'\n''*'$'\n''*'$'\n''* GCC plugins'$'\n''*'$'\n''Generate some entropy during boot and runtime (GCC_PLUGIN_LATENT_ENTROPY) [N/y/?] (NEW) ': No such file or directory"

E: Manually compiling a kernel from the tarball from works and boots.

Klotz commented on 2019-10-12 12:21 (UTC)

For version 5.2.21, I think applying the sed expression sed -i -re 's/(CFLAGS[[:space:]]+)\+=/\1:=/' to the ck stuff prior to patching will avoid a patch failure in tools/objtool/Makefile.

Pillgar commented on 2019-10-06 01:01 (UTC) (edited on 2019-10-06 12:34 (UTC) by Pillgar)

I'm getting DKMS errors when upgrading to linux-ck-ivybridge-5.2.19-1. On step (3/4) Install DKMS modules: Error! Bad return status for module build on kernel: 5.2.19-1-ck-ivybridge (x86_64).

It seems initcpio image can't build with any DKMS modules over ck-ivybridge-5.2.19-1

I'm using nvidia 435.21 dkms and vboxhost 6.0.12 dkms.

In make.logs I see error: incompatible gcc/plugin versions

I'm using gcc (9.1.0-2 updated: 2019-06-25)

Edit: There literally was just an update for gcc a few hours after I posted this. All solved.

Lessaj commented on 2019-09-15 19:14 (UTC)

It is building successfully now.

graysky commented on 2019-09-15 18:36 (UTC)

Please try 5.2.14-4

Lessaj commented on 2019-09-15 18:20 (UTC) (edited on 2019-09-15 18:20 (UTC) by Lessaj)

The addition of fix_patch-5.2-ck1_for_systemd-detect-virt_containers.patch seems to cause these last 3 lines to be dropped from kernel/sched/MuQSS.c even though they are in patch-5.2-ck1 which causes a build failure.



air-g4p commented on 2019-08-30 09:39 (UTC) (edited on 2019-08-30 15:38 (UTC) by air-g4p)

@graysky - It now appears the issue is exclusively related to modprobed-db. My latest build worked well in a clean chroot with ccache enabled, but I used an empty _localmodcfg= statement. Of course, the build time was quite lengthy, but it worked.

Any thoughts on a workaround, or a permanent fix?

Thanks for your help graysky....

graysky commented on 2019-08-28 18:41 (UTC)

@air-g4p - I do not use ccache so I cannot comment. I will say that builds using distcc on 5.2 fail where they worked fine with 5.1. Related? The change in the chroot build might be due to the modprobed-db change I made wherein the old behavior was to call it in recall mode via sudo which was always hacky in a PKGBUILD. The current method reads the database directly which the build scripts might not allow since your homedir is outside of the build root. Try building in the chroot without using the modprobed-db enabled; that will probably work.

air-g4p commented on 2019-08-28 13:08 (UTC)

@graysky - built without issue with ccache disabled, while running makepkg -sri so what is driving that fail?

Also, why did -ck previously build (many times) without issue in a clean chroot, but now it fails?


graysky commented on 2019-08-27 19:09 (UTC)

@air-g4p - Disable ccache and try again

air-g4p commented on 2019-08-27 08:25 (UTC) (edited on 2019-08-27 12:34 (UTC) by air-g4p)


In a clean folder, and running makepkg -sri the build failed, but in a different way. I got as far as:

-> Applying patch 0004-drm-amdgpu-pin-the-csb-buffer-on-hw-init-for-gfx-v8.patch... patching file drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c -> Setting config... -> Applying enable_additional_cpu_optimizations_for_gcc_v9.1+_kernel_v4.13+.patch ... patching file arch/x86/include/asm/module.h patching file arch/x86/Kconfig.cpu patching file arch/x86/Makefile patching file arch/x86/Makefile_32.cpu HOSTCC scripts/basic/fixdep /usr/bin/ccache: invalid option -- 'W'

then I also saw:

make[1]: [scripts/ scripts/basic/fixdep] Error 1 make: [Makefile:508: scripts_basic] Error 2 ==> ERROR: A failure occurred in prepare(). Aborting...

Any ideas? Thanks

graysky commented on 2019-08-26 21:05 (UTC)

@air-g4p - Does it build outside of the chroot? I'm thinking the chroot build script doesn't look outside of that environment and will therefore fail.

air-g4p commented on 2019-08-26 16:51 (UTC)


I'm seeing several inexplicable errors with the latest build:

When trying to build ck with time extra-x86_64-build in a clean chroot, I get as far as:

Kernel Live Patching (LIVEPATCH) [N/y/?] n

configuration written to .config

-> No modprobed.db data found /startdir/PKGBUILD: line 157: exit1: command not found ==> ERROR: A failure occurred in prepare(). Aborting... ==> ERROR: Build failed, check /var/lib/archbuild/extra-x86_64/user/build

this DESPITE: modprobed-db store showing:

180 modules currently loaded per /proc/modules 180 modules are in /home/user/.config/modprobed.db

AND the correct path in PKGBUILD:


Also, please have a look at Line 157 of PKGBUILD.

Did you intend to type exit as opposed to your current exit1?

I have never seen this modprobed error previously.

Ideas are welcome...

graysky commented on 2019-08-20 19:31 (UTC)

I haven't made an ISO/no idea. You could ask on CK's blog why it fails with that mount command. I do not edit beyond the changes his patchset make.

zerophase commented on 2019-08-20 00:46 (UTC)

@graysky archiso failing to build appears to be an issue with the CK kernel. No, idea what's different from Arch, as the only thing I changed was the compression algorithm used for initramfs.

Have any idea what might be causing the mount failure with CK?

zerophase commented on 2019-08-17 18:41 (UTC)

@graysky This is a forum post I made on the issue. It looks like folders under the work directory are not able to be identified as mount points. If you install archiso, and just run sudo cp -r /usr/share/archiso/configs/releng/ ~/archlive; cd ~/archlive; sudo ./ -v The issue should reproduce itself.

I just happened to decide to try the Arch kernel to see if that resolved the build failure I was receiving, and everything just built successfully. Looks like others have identified this issue in the past, but did not report it. I have a feeling it has something to do with mkarchiso.

graysky commented on 2019-08-17 11:10 (UTC)

You have provided nothing to go on...

zerophase commented on 2019-08-17 00:42 (UTC) (edited on 2019-08-17 00:43 (UTC) by zerophase)

I'm having issues getting archiso to successfully build an image when running the CK kernel. Everything builds fine with the default kernel. Is anyone else having this issue? Could it be from disabling initramfs support, other than LZ4? Is this a CK issue or archiso issue?

rzax666 commented on 2019-08-09 09:57 (UTC) (edited on 2019-08-09 10:29 (UTC) by rzax666)

@charveey thx but fix pkgbuild please ;D

Deadstar commented on 2019-08-09 05:19 (UTC)

Will the AMD Llano processors family be added to the kernel?

charveey commented on 2019-08-07 00:55 (UTC) (edited on 2019-08-07 00:58 (UTC) by charveey)

@rzax666 if you're interested you can just install my package, linux-bootsplash which provides the kernel with framebuffer bootsplash support. I personally run the 5.2.5 ck kernel with the patch you want. I got a hardware issue on my computer so I can't test the latest version (5.2.6)

rzax666 commented on 2019-08-07 00:31 (UTC) (edited on 2019-08-07 00:33 (UTC) by rzax666)

please add this patch -

graysky commented on 2019-08-03 19:22 (UTC) (edited on 2019-08-03 19:23 (UTC) by graysky)

Yeah, I think you shouldn't use native in non-interactive mode. See the comments:

Note - the march=native option is unavailable by this method, use the and manually select it.

kwe commented on 2019-08-03 15:01 (UTC) (edited on 2019-08-03 15:03 (UTC) by kwe)

Since switching from _subarch=20 (MIVYBRIDGE) to _subarch=28 (MNATIVE) in the PKGBUILD, it won't build:

  24. Intel Skylake X (MSKYLAKEX) (NEW)  
  25. Intel Cannon Lake (MCANNONLAKE) (NEW)  
  26. Intel Ice Lake (MICELAKE) (NEW)  
> 27. Generic-x86-64 (GENERIC_CPU)  
  28. Native optimizations autodetected by GCC (MNATIVE) (NEW)  
choice[1-28?]: 28  
Support for P6_NOPs on Intel chips (X86_P6_NOP) [N/y/?] (NEW) 28  
Support for P6_NOPs on Intel chips (X86_P6_NOP) [N/y/?] (NEW) 28  
Support for P6_NOPs on Intel chips (X86_P6_NOP) [N/y/?] (NEW) 28  
Support for P6_NOPs on Intel chips (X86_P6_NOP) [N/y/?] (NEW) 28  
Support for P6_NOPs on Intel chips (X86_P6_NOP) [N/y/?] (NEW) 28  
Support for P6_NOPs on Intel chips (X86_P6_NOP) [N/y/?] (NEW) 28  

The last message keeps looping forever.

When I switch back to the default _subarch= with no value, the process/interaction seems suboptimal:
- it asks me choice[1-28?]:
- I choose 28
- it asks me Support for P6_NOPs on Intel chips (X86_P6_NOP) [N/y/?] (NEW)
- I choose y
- it does the modprobed-db recall (as configured)
- it asks me BT8XX GPIO abuser (GPIO_BT8XX) [N/m/y/?] (NEW)
- I choose n (why does this ask me every time?)
- it asks me Studio Evolution SE6X (SND_SE6X) [N/m/?] (NEW)
- I choose n (again, why does this ask me every time?)
- ==> Sources are ready.

air-g4p commented on 2019-07-30 06:28 (UTC)

@graysky - Thanks, and understood.

@zerophase - Thanks, what you said made sense, and got me thinking about my chroot. As I had been using makechrootpkg of late to build linux-ck, I completely forgot that command neither cleans, nor updates a chroot.

As soon as I built linux-ck with my typical extra-x86_64-build my chroot got cleaned and updated, which resolved all the DKMS incompatible gcc versions installation issues.

zerophase commented on 2019-07-29 19:11 (UTC)

@air-g4p if you can rollback to the last good kernel you should be able to install. The last time I had dkms issues was from updating the kernel multiple times without rebooting. On the second kernel update I usually receive a bunch of errors about the version of the kernel dkms is looking for does not exist. Pretty sure that's what is happening to you too.

graysky commented on 2019-07-29 19:06 (UTC)

I do not use dkms so I wouldn't know where to start with it.

air-g4p commented on 2019-07-29 13:34 (UTC) (edited on 2019-07-29 13:35 (UTC) by air-g4p)

@graysky - Saw these DKMS installation errors today after building -ck on my fully updated system:

(2/6) Install DKMS modules ==> dkms install bbswitch/0.8 -k 5.2.4-1-ck Error! Bad return status for module build on kernel: 5.2.4-1-ck (x86_64) Consult /var/lib/dkms/bbswitch/0.8/build/make.log for more information. ==> dkms install vboxhost/6.0.10_OSE -k 5.2.4-1-ck Error! Bad return status for module build on kernel: 5.2.4-1-ck (x86_64) Consult /var/lib/dkms/vboxhost/6.0.10_OSE/build/make.log for more information. ==> dkms install nvidia/430.34 -k 5.2.4-1-ck Error! Bad return status for module build on kernel: 5.2.4-1-ck (x86_64) Consult /var/lib/dkms/nvidia/430.34/build/make.log for more information. ==> dkms install wireguard/0.0.20190702 -k 5.2.4-1-ck Error! Bad return status for module build on kernel: 5.2.4-1-ck (x86_64) Consult /var/lib/dkms/wireguard/0.0.20190702/build/make.log for more information.

My /build/make.logs show multiple instances of:

cc1: error: incompatible gcc/plugin versions cc1: error: fail to initialize plugin ./scripts/gcc-plugins/

Is there a workaround, or a solution?

graysky commented on 2019-07-27 22:15 (UTC)

@SuperIce - Right you are thanks, fixed.

SuperIce97 commented on 2019-07-27 22:13 (UTC) (edited on 2019-07-27 22:13 (UTC) by SuperIce97)

@graysky The 26 should be changed to 27 in this comment in the PKGBUILD:

nTia89 commented on 2019-07-27 09:25 (UTC)


graysky commented on 2019-07-26 20:22 (UTC)

@nTia89 - Yes, but was waiting for 5.2.3 to incorporate that and the other 2 patches. See:

nTia89 commented on 2019-07-26 13:02 (UTC)

@graysky Don't you follow upstream changes about CONFIG_GCC_PLUGIN_STACKLEAK[1]?


vinibali commented on 2019-07-21 06:54 (UTC)

5.2 patches are out now:

timo_capa commented on 2019-07-14 07:36 (UTC) (edited on 2019-07-14 07:39 (UTC) by timo_capa)

@nivekuil It shouldn't. I checked the config, and "CONFIG_USER_NS_UNPRIVILEGED" is enabled.

Also see:

nivekuil commented on 2019-07-14 07:18 (UTC)

Does this linux-zen bug also apply here?

graysky commented on 2019-06-17 19:05 (UTC)

@freenes - if the unpatched kernel doesn't work there is little likelihood that the MuQSS scheduler will fix it. You need to find out what's breaking it.

freenestor commented on 2019-06-17 02:06 (UTC)

@graysky @zerophase Thank you for advises. I tried 5.1.8-arch, and updated bios(my laptop model is ThinkPad T430u, bios is updated to Version: H6ETA0WW 2.18 Release Date: 06/01/2018), neither working.

zerophase commented on 2019-06-15 21:31 (UTC)

@graysky The version from the aur seems to have that feature disabled by default.

graysky commented on 2019-06-15 10:28 (UTC)

@gee - I try to keep linux-ck as close to the official ARCH kernel as I can. I do not know what negative effect changing that parameter might have for other users. Recommend you build from the AUR yourself modifying it as you see fit.

gee commented on 2019-06-15 10:12 (UTC)

Hello again,

I'll retry my last request in an easier way if you don't mind graysky.

Would you mind using 'N' for Linux kconfig's RAID6_PQ_BENCHMARK? I have not been able to find a way to do so without altering the kconfig file yet, and since I grab the kernels from your repos, it wouldn't work that way.

Setting it to 'N' save a few seconds on boot time, which I think is quite nice.

zerophase commented on 2019-06-14 17:38 (UTC)

@freenestor if that does not work try updating your bios / uefi. I would also post more system specs. As a last, resort you could try doing a clean install just to rule out configs breaking from waiting awhile to upgrade.

I think the issue probably has something to do with the motherboard, though. If nothing works try to use the warranty for a replacement. A new version of the same mobo might just work.

graysky commented on 2019-06-14 17:14 (UTC)

@freenestor - can you boot 5.1.x-arch without the parameter? Rule out ck1 patchset as the cause in other words.

freenestor commented on 2019-06-14 05:56 (UTC)

I cannot boot linux-ck 5.1.x without nolapic kernel parameter, and with nolapic parameter the performance is worse. I can boot linux-ck 4.19.x normally. Can anyone give some advises?

graysky commented on 2019-06-13 20:27 (UTC)

@ismet - Something happened with the rsync mirror. Fixing now, please try in 5 min or so.

ismet commented on 2019-06-13 16:21 (UTC)

Hey graysky, pre-compiled linux-ck-zen package cannot be downloaded from repo-ck repositories.

error: failed retrieving file 'linux-ck-zen-5.1.9-2-x86_64.pkg.tar.zst' from : The requested URL returned error: 404

Althorion commented on 2019-06-11 08:07 (UTC)

Shame. Seems like a useful option.

graysky commented on 2019-06-10 18:46 (UTC)

@Althorin - Purely archival.

Althorion commented on 2019-06-10 11:18 (UTC)

The PKGBUILD saves config.last for later, but never does anything with it (i.e., giving the option to use one as a basis for compilation). Is there something missing?

graysky commented on 2019-05-06 18:09 (UTC)

@galaxy0419 - Not out of date,

gatlinnewhouse commented on 2019-04-27 21:00 (UTC)


I fixed the issue, it seems there was a problem with my makepkg.cfg

gatlinnewhouse commented on 2019-04-18 10:32 (UTC)

trying to compile this now causes my computer to freeze, drop out of the x server and go back to the vt login screen (and my configs automatically log me in) and just sit there, fans spinning inconsistently

sir_lucjan commented on 2019-04-04 16:07 (UTC)


You could rebase against 5.0.6

graysky commented on 2019-04-03 23:18 (UTC) (edited on 2019-04-03 23:18 (UTC) by graysky)

CK1 needs to be tweaked to patch into 5.0.6 it seems...

patching file tools/objtool/Makefile
Hunk #1 FAILED at 31.
1 out of 1 hunk FAILED -- saving rejects to file tools/objtool/Makefile.rej

Can someone post to CK's blog?

RaskAr commented on 2019-03-29 12:22 (UTC) (edited on 2019-03-29 13:02 (UTC) by RaskAr)

Hello graysky, I use to patch the local copy of the PKGBUILD to allow the currently running config (/proc/config) to be used as base config so maybe it can be useful to merge it ?

BS86 commented on 2019-03-10 08:26 (UTC)

The patchset for 5.0 is live:

He is just too busy to officially announce:

Kr1ss commented on 2019-02-16 19:05 (UTC) (edited on 2019-02-16 19:06 (UTC) by Kr1ss)


Thx for researching and fixing so quickly ! PKGREL 2 built fine. (EDIT without changing any configs)

graysky commented on 2019-02-16 18:30 (UTC) (edited on 2019-02-16 18:40 (UTC) by graysky)

Seems like the ck patchset needs a tweak:

Testing now and will push 4.20.10-2 if successful.

Kr1ss commented on 2019-02-16 18:27 (UTC)

Looks like I could build after setting CONFIG_KVM and CONFIG_KVM_INTEL from <m> to [y]. I'm building for Haswell architecture btw.

graysky commented on 2019-02-16 18:19 (UTC)

Yea it tricked me into thinking it built find until I found 0 packages built from the batch... currently reviewing build log to see wtf.

Kr1ss commented on 2019-02-16 17:59 (UTC)

Yep, me too :

ERROR: "sched_smt_present" [arch/x86/kvm/kvm-intel.ko] undefined!                                                                                                                                      
make[1]: *** [scripts/Makefile.modpost:92: __modpost] Error 1                                                                                                                                          
make: *** [Makefile:1275: modules] Error 2                                                                                                                                                             
==> ERROR: A failure occurred in build().                                                                                                                                                              
==> Removing installed dependencies...                                                                                                                                                                 
checking dependencies...                                                                                                                                                                               

Package (2)  Old Version  Net Change                                                                                                                                                                   

bc           1.07.1-3      -0.17 MiB                                                                                                                                                                   
inetutils    1.9.4-7       -1.12 MiB                                                                                                                                                                   

Total Removed Size:  1.29 MiB                                                                                                                                                                          

:: Do you want to remove these packages? [Y/n]                                                                                                                                                         
:: Running pre-transaction hooks...                                                                                                                                                                    
(1/1) Removing old entries from the info directory file...                                                                                                                                             
:: Processing package changes...                                                                                                                                                                       
removing inetutils...                                                                                                                                                                                  
removing bc...                                                                                                                                                                                         
:: Running post-transaction hooks...                                                                                                                                                                   
(1/2) Reloading system manager configuration...                                                                                                                                                        
  Skipped: Current root is not booted.                                                                                                                                                                 
(2/2) Arming ConditionNeedsUpdate...                                                                                                                                                                   
==> ERROR: Build failed, check /var/lib/buildroot/x86_64/kriss/build

zerophase commented on 2019-02-16 17:54 (UTC)

Is 4.20.10 giving anyone else compile errors?

gee commented on 2019-02-15 09:29 (UTC)

It probably is out of focus, but it would be really nice to add this patchset:

My boot takes 2 sec to test various raid6 algorithms, when I don't even use raid... Being able to specify the algorithm to use would shave that time off and be awesome.

air-g4p commented on 2019-01-17 07:23 (UTC)


Thanks for pointing out my non-Arch standard settings and I forgot there was a new /etc/makepkg.conf.pacnew hanging about. I'm back to standard now.

I see you removed 'subarch 27' from PKGBUILD, and now as expected, linux-ck-4.20.3-1-x86_64.pkg.tar.xz and its headers build without issue.

graysky commented on 2019-01-17 00:50 (UTC)

@air-g4p - As well, your CFLAGS are not the Arch standard... drop -fstack-protector-strong per the new makepkg.conf in pacman-5.1.2-2.

graysky commented on 2019-01-16 21:07 (UTC) (edited on 2019-01-17 00:51 (UTC) by graysky)

@air-g4p - See the comments in the PKGBUILD and the discussion here in the AUR.

<h1>Note - the march=native option is unavailable by this method, use the nconfig and manually select it.</h1>

In short, I do not know how to allow the native option to get set via the method used for the other discretely defined options.

air-g4p commented on 2019-01-16 20:35 (UTC)


I'm running an i7 8550U.

Under both /etc/makepkg.conf and ~/chroot/root/etc/makepkg.conf, I have these identical settings:


CARCH="x86_64" CHOST="x86_64-pc-linux-gnu"

-- Compiler and Linker Flags

CPPFLAGS="-D_FORTIFY_SOURCE=2" CFLAGS="-march=native -O2 -pipe -fstack-protector-strong -fno-plt" CXXFLAGS="-march=native -O2 -pipe -fstack-protector-strong -fno-plt" LDFLAGS="-Wl,-O1,--sort-common,--as-needed,-z,relro,-z,now"

-- Make Flags: change this for DistCC/SMP systems


-- Debugging flags

DEBUG_CFLAGS="-g -fvar-tracking-assignments" DEBUG_CXXFLAGS="-g -fvar-tracking-assignments"

Is there anything you see in my setup that would drive:

'The build fails with hundreds of instances of:

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

that I mentioned when '_subarch=27' is set?

I normally build linux-ck in a clean chroot without issue, but this infinite loop issue disturbs me.

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)


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

        bool "Simple CPU accounting controller"
+       depends on !SCHED_MUQSS
          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!


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:

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.

Saren commented on 2019-01-01 06:13 (UTC)

Hi, since 4.20-ck NUMA is disabled regardless of _NUMAdisable. $ numactl -s shows No NUMA support available on this system.

zoopp commented on 2018-12-27 12:46 (UTC)

I usually build with native optimisations detected by GCC however there's a problem if I assign anything to the _subarch variable. Early during the build this question pops up:

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

.. which is answered with the value of _subarch variable (27 in my case).

Since this isn't an accepted answer, the question loops endlessly.

Not really a big issue but I thought I should mention it.

air-g4p commented on 2018-12-12 09:51 (UTC)


Same here: Built linux-ck 4.19.8-1 without issue yesterday, and all seems to be running well and as expected since then.

Ropid commented on 2018-12-09 03:07 (UTC) (edited on 2018-12-09 03:11 (UTC) by Ropid)


On first look, it seems to work fine for me. I built and booted into this 4.19.8 version a few hours ago.

$ uname -r

$ uptime
 04:05:41 up  5:57,  1 user,  load average: 0.32, 1.34, 1.90

I don't see anything unusual happening in the journal. Programs, games and the desktop don't seem to misbehave.

graysky commented on 2018-12-08 17:22 (UTC)

I heard from CK that it is unlikely he will be able to unfuck 4.19-ck1 before the release of 4.20. Fortunately, a user on the blog posted two patches which seem to do just that. I have adapted them to the PKGBUILD for 4.19.7 and for 4.19.8. I have been testing 4.19.7-ck1 and it has been running without issues. I am building 4.19.8-ck1 now. I pushed 4.19.8 to the AUR just now and am requesting some user feedback on stability and performance. I will build the repo set once we some confirmation that all is well. Thanks.

graysky commented on 2018-12-05 22:53 (UTC)

@lukpod - Not out of date, see pinned comment:

Even though 4.19.7 has been released, several upstream changes therein break ck1. Please do not flag out-of-date until Con releases an update.

graysky commented on 2018-12-05 20:32 (UTC)

Even though 4.19.7 has been released, several upstream changes therein break ck1. Please do not flag out-of-date until Con releases an update.

graysky commented on 2018-12-01 12:25 (UTC)

@FarStar - False alarm... I made a mistake in my config when building. No issues with 4.19.6.

FarStar commented on 2018-12-01 12:08 (UTC)

Hello graysky,

What kind of problems? I have slowdowns at times but I don't know if it's related.

graysky commented on 2018-12-01 12:04 (UTC)

Anyone else experiencing problems relating to networking under 4.19.6?

graysky commented on 2018-11-15 11:09 (UTC)

I need to clear yay's cache after every major kernel update.

Highly discourage users from AUR helpers. 9 times out of 10, problems with builds are the apfsult of the helper.

zerophase commented on 2018-11-15 02:32 (UTC)

@graysky Found the issue. I need to clear yay's cache after every major kernel update.

graysky commented on 2018-11-14 23:41 (UTC)

I don't think this is something with linux-ck but with how you configured/compiled/whatever it. Try the control experiment: build it straight-up and see if it works.

zerophase commented on 2018-11-14 22:57 (UTC)

@graysky I think it might be since I've modified my initramfs to the minimal size.

I could have sworn some of the usb subsystems were added directly to the kernel this release. I just don't want to reboot and find I have an unbootable kernel.

Any idea what to do about modules not being found?

graysky commented on 2018-11-14 22:05 (UTC)

I get no such errors

zerophase commented on 2018-11-14 20:49 (UTC) (edited on 2018-11-14 20:50 (UTC) by zerophase)

When running mkinitcpio, on 4.19.2, I'm getting errors over ehci_pci, xhci_pci, usbhid modules not being found. Have those modules been renamed?

artafinde commented on 2018-11-05 19:51 (UTC)

@graysky: wiki for modprobed-db mention to call the localmodconfig as below:

make LSMOD=$HOME/.config/modprobed.db localmodconfig 

What's the difference of the two methods?

Ranguvar commented on 2018-11-01 14:31 (UTC)

Anyone tried unofficial patch submittied in ck-hack comments?

Could be tampered, I haven't tested yet (and not advocating this package use it).

gustawho commented on 2018-10-29 09:13 (UTC)

Here's patch-4-18.ck1.xz

Ranguvar commented on 2018-10-29 00:38 (UTC)

Anyone have a mirror of patch-4.18-ck1.xz?

artafinde commented on 2018-10-28 22:45 (UTC)

Meh, seems down - can't download patch

somethingtohide commented on 2018-10-18 15:56 (UTC)

I'm getting an error while building this, although it's unclear what it is. All the previously logged patching steps are either skipped or success. Here's just the last bit:

patching file sound/soc/soc-dapm.c patching file sound/usb/line6/pcm.c patching file tools/objtool/Makefile ==> ERROR: A failure occurred in prepare(). Aborting...

I simply clone this AUR link, modify PKGBUILD to "_makeenconfig=y" and "_NUMAdisable=", then "makepkg -sri". Changing "_subarch=" has no effect.

I checked that last Makefile that's patched and it's indeed patched. I think the problem is the next line in the PKGBUILD, "local src"

francoism90 commented on 2018-10-09 13:19 (UTC)

@graysky - seems to be a known issue, thanks for looking into it. :)

graysky commented on 2018-10-05 16:25 (UTC)

@ELMastro - Do not build with an AUR helper. Use makepkg.

ElMastro commented on 2018-10-05 02:49 (UTC)

After [code] INSTALL sound/usb/line6/snd-usb-line6.ko INSTALL sound/usb/line6/snd-usb-pod.ko INSTALL sound/usb/line6/snd-usb-podhd.ko INSTALL sound/usb/line6/snd-usb-toneport.ko INSTALL sound/usb/line6/snd-usb-variax.ko INSTALL sound/usb/misc/snd-ua101.ko INSTALL sound/usb/snd-usb-audio.ko INSTALL sound/usb/snd-usbmidi-lib.ko INSTALL sound/usb/usx2y/snd-usb-us122l.ko INSTALL sound/usb/usx2y/snd-usb-usx2y.ko INSTALL virt/lib/irqbypass.ko DEPMOD 4.18.12-1-ck-ck rm: [/code] mine complains it can't remove '/tmp/packerbuild-1001/linux-ck/linux-ck/pkg/linux-ck/usr/lib/modules/4.18.12-1-ck/source' Can someone please confirm I'm the only one with this problem?

graysky commented on 2018-10-04 21:40 (UTC)

@fran - I mirror the Arch official config just allowing CK1 to set the options it wants to. Does your setup work with the official Arch kernel?

francoism90 commented on 2018-10-01 09:16 (UTC) (edited on 2018-10-01 09:17 (UTC) by francoism90)

@graysky Something goes wrong after the latest kernel update. I use something like this cryptkey=/dev/disk/by-uuid/..:ext4:/keys/mykey as kernel parameter, but it seems the USB is no longer init at boot. When using linux 4.10 it works correctly and the USB contents will be read.

Seems changes have been made in the config ( - don't know why this happens, do you have an idea? :)

air-g4p commented on 2018-09-25 09:44 (UTC)

@graysky - Thank you for sending me down the issues/29 path. I learned quite a bit, and now realize this 'problem' is upstream in nature. I'm sure this will not come as a surprise to you, but I'm running gcc version 8.2.1 20180831 (GCC) and the output of: gcc -c -Q -march=native --help=target | grep march is: -march= skylake Now I understand why, which I do appreciate.

graysky commented on 2018-09-23 17:56 (UTC)

@air-g4p - Your question is really for the gcc patchset I maintain. Issue 29 linked below applies but to my knowledge, there hasn't been any gcc changes to support your CPU. If I am wrong, plz post to the issue against the patch:

air-g4p commented on 2018-09-23 08:03 (UTC) (edited on 2018-09-23 08:35 (UTC) by air-g4p)

o/ graysky:

I'm running an i7-8550U which Intel describes as belonging to the Kaby Lake R family. This family was launched in Q3 2017.

That family of CPUs lives here:

Are you planning to add Kaby Lake support within your PKGBUILD or do you suggest I use another one of the other PKGBUILD 'Lakes'?


graysky commented on 2018-09-16 13:01 (UTC)

@ntia89 -

nTia89 commented on 2018-09-16 12:56 (UTC)

oh, thanks... I was missing the change upstream (kernel dev) :-(

graysky commented on 2018-09-16 12:47 (UTC)

This change mirrors that of the upstream Arch kernel and is due to the change of upstream (kernel source) not signing patches any more. You can modify the PKGBUILD to be like the old behavior if you want, but I will maintain as closely as I can how Arch upstream does it.

nTia89 commented on 2018-09-16 12:15 (UTC)

some releases ago, I used to download 4.X tar (90MB) and a patch-4.X.Y tar (~KB), while now I download 4.X.Y tar, for every new kernel sub-version. Is there a way to revert to a patch-system in order to build the kernel? I have an very very slow internet connection but I bet it will be appreciate for metered-connection users.

graysky commented on 2018-09-16 11:54 (UTC)


nTia89 commented on 2018-09-16 11:49 (UTC) (edited on 2018-09-16 11:54 (UTC) by nTia89)

Is there an easy way to get back with the kernel patch system?

francoism90 commented on 2018-09-06 07:56 (UTC)

@graysky Sorry for the noise, the commit has been reverted. :|

sir_lucjan commented on 2018-09-05 10:54 (UTC)


francoism90 commented on 2018-09-05 08:48 (UTC)

@graysky Thanks for the update! :)

graysky commented on 2018-09-03 11:44 (UTC)

@francoism90 - Was waiting for 4.18.6 but I don't even see an rc for it. I will push 4.18.5-4 with these incorporated shortly.

francoism90 commented on 2018-09-03 07:14 (UTC) (edited on 2018-09-03 07:15 (UTC) by francoism90)

@graysky Could you please implement the following changes:

Should this also be changed?


vp1981 commented on 2018-08-31 08:29 (UTC)

@keren_sky, did you search for 'apm_bigs' in the kernel source? I did and found nothing, but searching for 'apm_bios' gives many results, for example

$ find -name 'apm_bios*'

./include/linux/apm_bios.h ./include/uapi/linux/apm_bios.h

I would assume that the error has nothing to do with 'apm_bios*' files.

P.S. If somehow your kernel source doesn't have 'apm_bios*' files then something wrong on your side.

keren_sky commented on 2018-08-29 23:00 (UTC)

I see the following: CC [M] drivers/gpu/drm/nouveau/nvkm/engine/disp/dmacgv100.o fixdep: error opening file: include/linux/apm_bigs.h: No such file or directory make[4]: *** [scripts/ drivers/gpu/drm/nouveau/nvkm/engine/disp/dmacgv100.o] Error 2

Which pkg provides that file? thanks.

simona commented on 2018-08-29 15:10 (UTC)

repository will be updated soon?

vp1981 commented on 2018-08-29 14:18 (UTC)

@biosin, did you read your log and other comments? Read the log again and these comments: freenestor commented on 2018-08-28 04:17 (edited on 2018-08-28 04:21 by freenestor)

ltsdw commented on 2018-08-28 05:24 (edited on 2018-08-28 05:24 by ltsdw)

skydrome commented on 2018-08-28 09:09 (edited on 2018-08-28 09:09 by skydrome)

P.S. Sorry, no links.

commented on 2018-08-29 09:25 (UTC)

Edit: My fault, somehow the AUR helper didn't delete everything before rebuilding it. 4.18.5-3 does compile.

Somehow, linux-ck 4.18.5-3 and 4.18.5-1 failed building for me at the same seemingly random spot:

LDS     arch/x86/kernel/
  CC [M]  arch/x86/kernel/msr.o
  CC [M]  arch/x86/kernel/cpuid.o
  AR      arch/x86/kernel/built-in.a
  AR      arch/x86/built-in.a
==> ERROR: A failure occurred in build().

Full log:

tomascco commented on 2018-08-29 00:36 (UTC)

Thanks for your hard work @graysky! I can also confirm that it's building fine.

Lessaj commented on 2018-08-29 00:11 (UTC)

Great! Confirming 4.18.5-3 builds fine for me now, you did the same thing I tried by removing the DEPMOD=/doesnt/exist. :) I was just testing it out and for my particular setup there are some regressions (AMDGPU related, I have 4 monitors, tends to be finicky) so I'll hold off for now and stick with 4.17.11-6. Thank you for the quick support!

graysky commented on 2018-08-28 22:42 (UTC)

@zerophase - I don't think I can automate that since there is are 2 prompts...

graysky commented on 2018-08-28 22:42 (UTC)

4.18.5-3 builds, sorry for the fuck-up.

SilverMight commented on 2018-08-28 22:30 (UTC)

Also received the same issue as @Lessaj

ltsdw commented on 2018-08-28 22:29 (UTC)

I made changes in pkgbuild to apply muqss to another package kernel, and it build successfully here.

zerophase commented on 2018-08-28 21:20 (UTC) (edited on 2018-08-28 21:24 (UTC) by zerophase)

setting _subarch=27 for GCC detecting native cpu features causes the setup script to loop infinitely on Support for P6_NOPs on Intel chips (X86_P6_NOP) [N/y/?] Would it be possible to allow user input here for whether to turn this feature on or not?

dunconio commented on 2018-08-28 21:02 (UTC) (edited on 2018-08-28 21:29 (UTC) by dunconio)

I can confirm the updated build 4.18.5-2-ck fails as per @Lessaj. (I am using _subarch=19)

  INSTALL sound/usb/usx2y/snd-usb-usx2y.ko
  INSTALL virt/lib/irqbypass.ko
  DEPMOD  4.18.5-2-ck
'make modules_install' requires /doesnt/exist. Please install it.
This is probably in the kmod package.
make: *** [Makefile:1248: _modinst_post] Error 1
==> ERROR: A failure occurred in package_linux-ck().
==> ERROR: Makepkg was unable to build linux-ck.

graysky commented on 2018-08-28 20:07 (UTC)

I updated with ck's fix in 4.18.5-2 ... please confirm.

Lessaj commented on 2018-08-28 19:19 (UTC) (edited on 2018-08-28 19:27 (UTC) by Lessaj)

Able to complete the build with the patch mentioned by @skydrone, but fails right at the end when installing modules as part of package(). Comparing with previous PKGBUILD this part is unchanged, not sure why I'm experiencing this error now. Confirmed I have the kmod package.

  INSTALL sound/xen/snd_xen_front.ko
  INSTALL virt/lib/irqbypass.ko
  DEPMOD  4.18.5-1-ck
'make modules_install' requires /doesnt/exist. Please install it.
This is probably in the kmod package.
make: *** [Makefile:1248: _modinst_post] Error 1
==> ERROR: A failure occurred in package_linux-ck().

ltsdw commented on 2018-08-28 18:29 (UTC)

@skydrome how can I apply this? Should I edit something in pkgbuild or before/pre pkgbuild?

LChris314 commented on 2018-08-28 12:29 (UTC)

Hello. When I installed my current 4.17.11 kernel, I made some changes to the config with menuconfig (mostly removed unneeded modules). What is the preferred way to integrate the changes I made into the new version? Do I just load the current config from the menu, or should I modify the PKGBUILD in some way?

skydrome commented on 2018-08-28 09:09 (UTC) (edited on 2018-08-28 09:09 (UTC) by skydrome)

ltsdw commented on 2018-08-28 05:24 (UTC) (edited on 2018-08-28 05:24 (UTC) by ltsdw)

I'm getting this:

kernel/sched/MuQSS.c: In function ‘sched_tick_remote’: kernel/sched/MuQSS.c:3321:34: error: ‘struct task_struct’ has no member named ‘se’ delta = rq_clock_task(rq) - curr->se.exec_start;

coldreaver commented on 2018-08-28 04:41 (UTC)

4.18.5 is failing yea :(

freenestor commented on 2018-08-28 04:17 (UTC) (edited on 2018-08-28 04:21 (UTC) by freenestor)

I failed buid it on my computer with linux-4.17-ck installed

CC kernel/rcu/rcu_segcblist.o

AR kernel/rcu/built-in.a

CC kernel/sched/MuQSS.o

kernel/sched/MuQSS.c: In function ‘sched_tick_remote’:

kernel/sched/MuQSS.c:3321:34: error: ‘struct task_struct’ has no member named ‘se’

delta = rq_clock_task(rq) - curr->se.exec_start; ^~

make[2]: *** [scripts/ kernel/sched/MuQSS.o] Error 1

make[1]: *** [scripts/ kernel/sched] Error 2

make: *** [Makefile:1033: kernel] Error 2

==> ERROR: A failure occurred in build(). Aborting...

zerophase commented on 2018-08-28 01:46 (UTC)

Is the 4.18.5 failing to build for anyone else?

francoism90 commented on 2018-08-27 13:22 (UTC)

@graysky - could you also enable this in linux-ck? :)

harryharryharry commented on 2018-08-27 10:06 (UTC)

Just a heads up, a 4.18 ck-patchset has just been released:

sir_lucjan commented on 2018-08-18 15:46 (UTC)


Do you see patchset for 4.18?

simona commented on 2018-08-18 08:53 (UTC)

I hope 4.18.2 will arrive soon...

graysky commented on 2018-08-17 22:54 (UTC)

@Jeffrey4l - I don't see why not. Did you create a ticket for the Arch kernel to do the same? @francoism90 - No idea, email ck.

francoism90 commented on 2018-08-17 20:02 (UTC)

@graysky Does this mean no patch for 4.18 yet?

Jeffrey4l commented on 2018-08-16 04:36 (UTC)

Could you enable


in config in default?

This is need for docker stats and systemd-cgtop

sir_lucjan commented on 2018-08-14 20:47 (UTC)


simona commented on 2018-08-14 20:40 (UTC)

4.17.11-6 ? what change?

keepitsimpleengr commented on 2018-08-13 16:17 (UTC)

:: installing nvidia-utils (396.51-1) breaks dependency 'nvidia-utils=396.45' required by nvidia-ck-sandybridge

simona commented on 2018-08-11 17:53 (UTC)

ok :-))) thx.

graysky commented on 2018-08-11 10:11 (UTC)

@simona - See the pinned post from Aug 4th.

simona commented on 2018-08-11 08:13 (UTC)

stopped to 4.17.11?

gee commented on 2018-08-11 03:35 (UTC)

Oh it's pkgbase, got it!

Yeah I'm running it, but the changes were made in bluetooth that I have not tested much, and schedutils that I don't use. If you look at the patch you'll see it is just a 3-way manual diff, nothing fancy.

graysky commented on 2018-08-10 18:56 (UTC)

@gee - Thanks, any feedback (assuming you're running that patched code)? I passed CK your proposal for review. To your question, just rename the pkgbase from linux-ck to linux-whatever-you-want

sispus commented on 2018-08-10 10:28 (UTC) (edited on 2018-08-10 11:03 (UTC) by sispus)

@ sir_lucjan

thank you, but I think I don't understand what to do. I did:

sudo pacman-key --recv-key 38DBBDC86092693E

it replied:

gpg: bad data signature from key 8396F1D05506E82D: Wrong key usage (0x19, 0x2)

gpg: bad data signature from key 20E8A9C77716EB4F: Wrong key usage (0x19, 0x2)

I did also that without realizing "wrong key usage" warning:

sudo pacman-key --lsign-key 38DBBDC86092693E

Now when I repeat above command (--recv-key) it says: Remote key not fetched correctly from keyserver.

I don't know if these info is helpful. If not, I'm sorry...

edit: I found this from a forum and solved: gpg --recv-keys 38DBBDC86092693E

sir_lucjan commented on 2018-08-10 09:53 (UTC)

@ sispus

sispus commented on 2018-08-10 09:49 (UTC)

I couldn't install it to manjaro xfce. linux-4.17.11.tar ... FAILED (unknown public key 38DBBDC86092693E) ==> ERROR: One or more PGP signatures could not be verified!

gee commented on 2018-08-10 08:29 (UTC) (edited on 2018-08-10 09:27 (UTC) by gee)

@graysky, I've patched ck's patch to build with .14, it was fairly easy, hopefully I made no mistake:

Unrelated, but how do I easily get this package to be named similarly to the one I'd get from your repo, in my case -haswell?

melcor333 commented on 2018-08-09 23:20 (UTC) (edited on 2018-08-09 23:28 (UTC) by melcor333)

@graysky - I'll try thanks

EDIT: Perfect! Working as intended. Thanks a lot! :)

melcor333 commented on 2018-08-09 23:18 (UTC)

Hi, since last 2 updates I'm unable to boot to GUI, because of nvidia isn't found in kernel modules. I get this error during install:

==> ERROR: module not found: nvidia' ==> ERROR: module not found:nvidia_modeset' ==> ERROR: module not found: nvidia_uvm' ==> ERROR: module not found:nvidia_drm'

I have nvidia-ck and linux-ck-headers installed. Is there anything I'm doing wrong?

GPU is Nvidia GTX960. Before this everything worked flawlessly.


graysky commented on 2018-08-09 21:16 (UTC)

@melcor333 - broadcom and nvidia-* have been updated. Please try and reported back. broadcom and nvidia-340xx work for me on my test box.

graysky commented on 2018-08-09 21:06 (UTC)

Yes, the naming schema has changed. I will push some updates shortly.

graysky commented on 2018-08-09 19:29 (UTC)

All - I think I have all the kinks ironed out of the new PKGBUILD structure. We are still unfortunately in the position of ck1 not patching into 4.17.12-current so we seem stuck on 4.17.11. Good news is that 4.18.0 should be released this weekend and CK is likely to work on that rather than doing so to an EOL kernel (4.17.x series).

I just pushed 4.17.11-4 with a few minor tweaks. Builds and runs fine for me. Please d/l and test.

@Saren - Fixed the NUMA issue in -4. Please try.

Saren commented on 2018-08-07 17:42 (UTC)

Hi, my system needs NUMA support, and changing _NUMAdisable=y to _NUMAdisable= still disables NUMA support on 4.17.11-3. I have already compiled the kernel twice and I am sure I have changed that line. It was totally fine before the major PKGBUILD format change.

artafinde commented on 2018-08-07 11:29 (UTC)

Yeah if only they provide a checksum and/or signed packages :)

graysky commented on 2018-08-07 11:03 (UTC)

QuartzDragon commented on 2018-08-07 09:47 (UTC)


That sounds better than my suggestion. :)

artafinde commented on 2018-08-07 08:53 (UTC) (edited on 2018-08-07 09:00 (UTC) by artafinde)

Following the new way for the linux package fetches the whole linux repo (bare) with cloning which downloads 2Gb just to checkout HEAD of it (~150Mb). The ideal would beto download from release i.e.

QuartzDragon commented on 2018-08-07 07:42 (UTC) (edited on 2018-08-07 07:43 (UTC) by QuartzDragon)

Hey graysky,

Is there any reason you're not using It contains all of the relevant Arch-specific changes:

zerophase commented on 2018-08-06 16:09 (UTC) (edited on 2018-08-06 16:11 (UTC) by zerophase)

@graysky should the kernel be installing modules as ck-ck now?

graysky commented on 2018-08-05 19:17 (UTC)

@id942701 - Opps, debug helper. Removed no in 4.17.11-3.

id942701 commented on 2018-08-05 18:18 (UTC)

PKGBUILD problem: exit in prepare(). Build successful.

graysky commented on 2018-08-05 16:02 (UTC) (edited on 2018-08-05 16:02 (UTC) by graysky)

I changed my mind... I just commited 4.17.11-2 which just changes to the PKGBUILD to mirror that of Arch upstream. Builds for me. Can some others please test and report back? CK has been contacted about ck1 not patching in to 4.17.12.

graysky commented on 2018-08-05 11:32 (UTC)

@QuartzDragon - I have already merged it too but am don't see much value in using 4.17.11-2 with just the PKGBUILD changes... as I posted on 08-04, ck1 does not patch against 4.17.12.

QuartzDragon commented on 2018-08-05 09:44 (UTC) (edited on 2018-08-05 09:49 (UTC) by QuartzDragon)

Hey, graysky,

You may want to check out because the kernel PKGBUILD has changed significantly.

I've already merged in the changes locally, and it builds fine.

graysky commented on 2018-08-04 11:26 (UTC) (edited on 2018-08-11 10:11 (UTC) by graysky)

I am unable to patch into 4.17.12+ due to changes upstream... need to fix before updating the PKGBUILD:

Hunk #1 FAILED at 185.
Hunk #2 succeeded at 473 (offset -1 lines).
1 out of 2 hunks FAILED -- saving rejects to file

id942701 commented on 2018-08-04 08:47 (UTC)

Small typo "Note that the generic (default) option is 24" but "26. Generic-x86-64 (GENERIC_CPU)"

Tjuh commented on 2018-08-01 17:58 (UTC)

I'm trying to build 4.14.20 because I have not been able to build a functional version since 4.16.15. Well 4.16 builds fine but it just hangs at boot same with 4.17-ck and vanilla. Only working kernel I got now is latest linux-lts.

graysky commented on 2018-08-01 17:48 (UTC)

You're building 4.16?

Tjuh commented on 2018-08-01 17:05 (UTC)

I am unable to build any version prior to 4.16 due to the following error message:

pager.c: In function ‘pager_preexec’: pager.c:36:12: error: passing argument 2 to restrict-qualified parameter aliases with argument 4 [-Werror=restrict] select(1, &in, NULL, &in, NULL); ^~~ ~~~ cc1: all warnings being treated as errors mv: cannot stat '/tmp/makepkg/linux-ck/src/linux-4.14/tools/objtool/.pager.o.tmp': No such file or directory make[4]: [/tmp/makepkg/linux-ck/src/linux-4.14/tools/build/ /tmp/makepkg/linux-ck/src/linux-4.14/tools/objtool/pager.o] Error 1 make[3]: [Makefile:52: /tmp/makepkg/linux-ck/src/linux-4.14/tools/objtool/libsubcmd-in.o] Error 2 make[2]: [Makefile:54: /tmp/makepkg/linux-ck/src/linux-4.14/tools/objtool/libsubcmd.a] Error 2 make[1]: [Makefile:62: objtool] Error 2 make: *** [Makefile:1637: tools/objtool] Error 2 ==> ERROR: A failure occurred in prepare(). Aborting...

Anyone know what this is about?

artafinde commented on 2018-08-01 15:47 (UTC)

I'm the same person with the one on the forums :)

SilverMight commented on 2018-08-01 15:43 (UTC)

@artafinde Yup, just saw that in the thread. Thanks

artafinde commented on 2018-08-01 06:48 (UTC)

@SilverMight: Enable NUMA, recompile and CUDA now works.

SilverMight commented on 2018-07-31 21:44 (UTC)

Have a few people with the same issue in the thread:

graysky commented on 2018-07-29 21:49 (UTC)

No idea about CUDA/don't have the hardware. You might try CK's blog or someone might reply to what I think is your bbs thread.

SilverMight commented on 2018-07-29 21:39 (UTC)

Now it works, thanks. Do you know if there's an issue with CUDA? Doesn't seem to initialize on this kernel, but every other kernel works fine for me

graysky commented on 2018-07-29 21:04 (UTC)

Did you pacman -Syyu? Fine for me % pacman -Ss core2-h repo-ck/linux-ck-core2-headers 4.17.11-1 (ck-core2) Header files and scripts for building modules for Linux-ck-core2 kernel Intel Core2/Newer Xeon optimized.

SilverMight commented on 2018-07-29 19:19 (UTC) (edited on 2018-07-29 20:10 (UTC) by SilverMight)

Some reason my pacman is trying to fetch 4.10-1 instead of 4.11 from repo-ck, even though I have refreshed the lists.. anyone experiencing this?

Also - is anyone else having an issue with CUDA with this kernel?

sir_lucjan commented on 2018-07-11 23:01 (UTC)

@ graysky


graysky commented on 2018-07-11 20:35 (UTC)

Can anyone else confirm that 4.17.6 causes errors relating to mounting disks.

bezirg commented on 2018-06-28 08:50 (UTC)

Proposed change in PKGBUILD:

Note that the
# generic (default) option is 24.

to ====>

Note that the
# generic (default) option is 26.

bezirg commented on 2018-06-28 08:45 (UTC)

On linux-ck 4.17.3-1 makepkg error:

==> Verifying source file signatures with gpg...
    linux-4.17.tar ... FAILED (unknown public key 79BE3E4300411886)
    patch-4.17.3 ... FAILED (unknown public key 38DBBDC86092693E)
==> ERROR: One or more PGP signatures could not be verified!

graysky commented on 2018-06-25 13:14 (UTC)

I asked for a timeline for 4.17-ck1. I will update the [repo-ck] thread when I hear back. Doesn't make too much sense to go though heroic measures to troubleshoot if 4.17-ck1 is just around the corner. I believe 4.16.x will be EOL'ed shortly as well.

CK expects 4.17-ck1 in a few days so it's not worth trying to unfuck 4.16-ck1 to fit into 4.16.17 or 4.16.18 IMO.

graysky commented on 2018-06-24 21:07 (UTC)

Yea, it's confirmed that 4.16.17-1 doesn't play well with ck's patchset. I asked for a timeline for 4.17-ck1. I will update the [repo-ck] thread when I hear back. Doesn't make too much sense to go though heroic measures to troubleshoot if 4.17-ck1 is just around the corner. I believe 4.16.x will be EOL'ed shortly as well.

Tjuh commented on 2018-06-24 18:36 (UTC)

Same problem as the others, just a black screen during boot, ditto with the pre built pkg from the repo ck.

nTia89 commented on 2018-06-24 10:24 (UTC)

me too. 4.16.17-2 stucks just after boot selection. I am on broadwell, compiled with native and P6_NOP. Downgraded

graysky commented on 2018-06-23 05:37 (UTC)

@zero - Yes, I am having this problem as well.

zerophase commented on 2018-06-22 23:45 (UTC) (edited on 2018-06-22 23:54 (UTC) by zerophase)

Anyone else getting stuck after the boot loader with the last update? Don't believe I have any logs since the system stalls early in boot.

graysky commented on 2018-06-22 19:18 (UTC)

@Saren - Thanks... I ran the test in another terminal but never checked the result. Fixed now.

Saren commented on 2018-06-22 19:05 (UTC)

Fail at prepare() for 4.16.17-1

patching file drivers/acpi/acpi_watchdog.c
Reversed (or previously applied) patch detected!  Skipping patch.
1 out of 1 hunk ignored -- saving rejects to file drivers/acpi/acpi_watchdog.c.rej

graysky commented on 2018-05-24 22:06 (UTC)

@banzay - I like to mirror the official Arch kernel patch-wise @Morgan - I believe that was needed for some reason ... can't remember why but it's harmless as-is.

banzay commented on 2018-05-24 21:22 (UTC) these patches will not be included?

Morganamilo commented on 2018-05-22 04:32 (UTC)

Is there a reason for the linux-ck=4.16.10 provide? All packages automatically provide their own name.

zebulon commented on 2018-05-02 06:24 (UTC)

@graysky: thanks. Do you mean 4.16.6-2? There is no -3 in the repository.

graysky commented on 2018-05-01 23:31 (UTC)

@sir_l - CK1 includes the nvidia patch (0003) and I bumped linux-rc to 4.16.7 which includes 0004 so I accidentally omitted it.... fixing in 4.16.6-3, thanks.

sir_lucjan commented on 2018-05-01 21:03 (UTC) (edited on 2018-05-01 21:04 (UTC) by sir_lucjan)


Probably you forgot about other patches:


  2. (added: 4.16.7-rc1).

graysky commented on 2018-05-01 20:56 (UTC)

@sir_lucjan - Yes, I thought I diffed that out >:( ... fixed now but since I have been building the repo packages, I don't want to bump the pkgver and start all over.

@zebulon - Thanks for reporting. Please try now.

sir_lucjan commented on 2018-05-01 20:51 (UTC) (edited on 2018-05-01 20:52 (UTC) by sir_lucjan)


You should remove

  install -Dt "${_builddir}/drivers/media/dvb-core" -m644 drivers/media/dvb-core/*.h

from headers()



zebulon commented on 2018-05-01 20:47 (UTC) (edited on 2018-05-01 20:48 (UTC) by zebulon)

Hi, packaging fails with this error:

==> Starting package_linux-ck-headers()... install: cannot stat 'drivers/media/dvb-core/*.h': No such file or directory ==> ERROR: A failure occurred in package_linux-ck-headers(). Aborting...

Or am I doing something wrong? The only option I used is Skylake optimisations, but I do not think this is relevant.

sir_lucjan commented on 2018-04-11 14:47 (UTC)


You should ask Con.

simona commented on 2018-04-10 08:13 (UTC)

4.16? soon?

Dennis commented on 2018-03-24 07:14 (UTC)

Today receiving the key just worked and the packages updated as well without issues. Must have been a network issue elsewhere.

@vp1981: The missing sudo on the second command was just an oversight on my part when I prepared a cleaned up, condensed message for posting here. In reality I ran all commands in a root shell and only added sudo in front for the log here to make it clear that I was running those commands as root.

vp1981 commented on 2018-03-23 23:55 (UTC)


I already installed that packages using 'pacman -U'. I had to say that these are the only packages that not upgraded smoothly, on other two platforms (haswell and sandybridge) all was fine. At first I suspected that it is the 'archlinux-keyring' that cause the problem, I tried to update all other packages but without '-ck-skylake*' and only after that tried to update these two packages using usual 'pacman -Su' but 'pacman' reported something about either broken package or a key, I couldn't remember the details, sorry.


I tried to reinstall the key and went fine (actually pacman-key returned that the key is already installed). Besides, according to the your message the problem with 'gpg': it cannot retrieve the key. You may try again but I'm not sure what causes the failure (may be network problem or gpg misconfiguration, you may want to [re]read Also I noticed that you issue command like

sudo pacman-key ... && pacman-key ...

without 'sudo' before the second pacman-key. Try to append 'sudo' to both 'pacman-key'.

P.S. I usually do that everyday:

(reflector stuff) && pacman -Scc && pacman -Syy && pacman -Su && pacman -Scc

Dennis commented on 2018-03-23 19:22 (UTC) (edited on 2018-03-23 19:22 (UTC) by Dennis)

I'm also having problems updated linux-ck-skylake from repo-ck:

$ sudo pacman -Syyu

:: Starting full system upgrade... resolving dependencies... looking for conflicting packages...

Packages (2) linux-ck-skylake-4.15.12-1 linux-ck-skylake-headers-4.15.12-1

Total Installed Size: 152.68 MiB Net Upgrade Size: 0.05 MiB

:: Proceed with installation? [Y/n] Y (2/2) checking keys in keyring [#######################################################################################] 100% downloading required keys... error: key "88A032865EE46C4C" could not be looked up remotely error: required key missing from keyring error: failed to commit transaction (unexpected error) Errors occurred, no packages were upgraded.

$ sudo pacman-key -r 5EE46C4C && pacman-key --lsign-key 5EE46C4C gpg: keyserver receive failed: Address family not supported by protocol ==> ERROR: Remote key not fetched correctly from keyserver.

Also tried updating archlinux-keyring alone first and reinitialized and repopulated the keyring but I am still getting above errors.

graysky commented on 2018-03-23 18:54 (UTC)

@vp1981 and @Silvermight - Please pacman -Syyu and it should work.

vp1981 commented on 2018-03-23 14:35 (UTC)

@SilverMight: same here, I checked downloaded files (-ck-skylake and -ck-skylake-headers) and they are correct and 'pacman -U' installs them fine.

SilverMight commented on 2018-03-23 10:37 (UTC)

Getting error: linux-ck-skylake: missing required signature when trying to update from I have the pacman key added and I tried it again, however it still fails to update

sir_lucjan commented on 2018-03-22 10:30 (UTC)



Works fine.

shanti commented on 2018-03-22 09:12 (UTC)


thanks I added the lines you provided however I am still unable to apply the bfq-sq-mq patch successfully,

Makefile.rej gives:

[code]--- Makefile +++ Makefile @@ -2,7 +2,7 @@ VERSION = 4 PATCHLEVEL = 15 SUBLEVEL = 11 -EXTRAVERSION = +EXTRAVERSION = NAME = Fearless Coyote

DOCUMENTATION --- Makefile +++ Makefile @@ -2,7 +2,7 @@ VERSION = 4 PATCHLEVEL = 15 SUBLEVEL = 11 -EXTRAVERSION = +EXTRAVERSION =-mq NAME = Fearless Coyote

shanti commented on 2018-03-22 09:07 (UTC)


thanks I added the lines you provided however I am still unable to apply the bfq-sq-mq patch successfully,

Makefile.rej gives:

[--- Makefile +++ Makefile @@ -2,7 +2,7 @@ VERSION = 4 PATCHLEVEL = 15 SUBLEVEL = 11 -EXTRAVERSION = +EXTRAVERSION = NAME = Fearless Coyote

# DOCUMENTATION --- Makefile +++ Makefile @@ -2,7 +2,7 @@ VERSION = 4 PATCHLEVEL = 15 SUBLEVEL = 11 -EXTRAVERSION = +EXTRAVERSION =-mq NAME = Fearless Coyote


vp1981 commented on 2018-03-22 01:16 (UTC) (edited on 2018-03-22 01:17 (UTC) by vp1981)


you do not have the option to enable bfq-mq anymore

you mean the kernel cmdline option?

it should appear as bfq-mq in the list of schedulers

I have felling that that's wrong (may be because now BFQ in mainline). Some time ago I tried to follow closely BFQ development and had impression that new 'BFQ-MQ' is now 'BFQ' in mainline. This line

mq-deadline kyber [bfq] none

tells me that bfq in my case is BFQ-MQ (because all these schedulers are MQ).

sir_lucjan commented on 2018-03-21 17:23 (UTC)


Nope, I have my own kernel with bfq patchset.

Terence commented on 2018-03-21 17:20 (UTC)

@sir_lucjan yes, you have bfq-mq and bfq I was just saying vp1981 only had bfq. I see you have bfq-mq. Is this from latest linux-ck? Did you have to do somethign to enable it?

sir_lucjan commented on 2018-03-21 16:05 (UTC)


bfq != bfq-mq

[lucjan@archlinux ~]$ cat /sys/block/sda/queue/scheduler
bfq [bfq-mq] none

Terence commented on 2018-03-21 16:00 (UTC)

@vp1981, I think so because you do not have the option to enable bfq-mq anymore and it should appear as bfq-mq in the list of schedulers.

vp1981 commented on 2018-03-21 14:23 (UTC) (edited on 2018-03-21 14:24 (UTC) by vp1981)

$ uname -a

Linux 4.15.11-1-ck-haswell #1 SMP PREEMPT Tue Mar 20 11:27:59 EDT 2018 x86_64 GNU/Linux

$ cat /proc/cmdline

BOOT_IMAGE=/vmlinuz-linux-ck-haswell root=UUID=162d2744-158f-46b5-b3e8-fdc25739c96a rw systemd.unified_cgroup_hierarchy=1 scsi_mod.use_blk_mq=1 quiet

$ cat /sys/block/sd*/queue/scheduler

mq-deadline kyber [bfq] none mq-deadline kyber [bfq] none mq-deadline kyber [bfq] none mq-deadline kyber [bfq] none mq-deadline kyber [bfq] none

@Terence, is that mean that I don't have BFQ-MQ?

Terence commented on 2018-03-21 13:13 (UTC)

bfq-mq is also missing.

francoism90 commented on 2018-03-21 11:12 (UTC)

Why I'm I only seeing the following options: $ cat /sys/block/sd*/queue/scheduler noop deadline [cfq]

$ uname -a Linux 4.15.11-1-ck-skylake #1 SMP PREEMPT Tue Mar 20 11:44:14 EDT 2018 x86_64 GNU/Linux


Morganamilo commented on 2018-03-21 00:53 (UTC)

@LinX as one of the devs for it I'm glad it does ;)

Kerriganx commented on 2018-03-21 00:51 (UTC)

@Morganamilo thanks, yay works.

Morganamilo commented on 2018-03-21 00:31 (UTC)

@LinX works with me using yay. AUR helpers are not officially supported though. your best bet is makepkg.

Kerriganx commented on 2018-03-21 00:27 (UTC) (edited on 2018-03-21 00:30 (UTC) by Kerriganx)

I tried pacaur instead of yaourt but it freezes at:

==> Validating source files with sha256sums... linux-4.15.tar.xz ... Passed linux-4.15.tar.sign ... Skipped patch-4.15.11.xz ... Passed patch-4.15.11.sign ... Skipped config ... Passed 60-linux.hook ... Passed 90-linux.hook ... Passed linux.preset ... Passed enable_additional_cpu_optimizations-20180310.tar.gz ... Passed patch-4.15-ck1.xz ... Passed 0001-add-sysctl-to-disallow-unprivileged-CLONE_NEWUSER-by.patch ... Passed 0002-drm-i915-edp-Only-use-the-alternate-fixed-mode-if-it.patch ... Passed ==> Verifying source file signatures with gpg... linux-4.15.tar ... Passed patch-4.15.11 ... Passed

Kerriganx commented on 2018-03-20 23:35 (UTC) (edited on 2018-03-20 23:36 (UTC) by Kerriganx)

@Morganamilo I run through yaourt -Syua. Month ago there was no such problem

Morganamilo commented on 2018-03-20 23:13 (UTC)

@LinX It wants user input, are you running makepkg through a script that doesn't pass on stdin or something?

Kerriganx commented on 2018-03-20 23:05 (UTC)

I'm stuck on "Support for P6_NOPs". This line appears endlessly:

sir_lucjan commented on 2018-03-20 15:38 (UTC)


### Patch source with BFQ-SQ-MQ
        msg "Fix naming schema in BFQ-SQ-MQ patch"
        sed -i -e "s|SUBLEVEL = 0|SUBLEVEL = 11|g" \
            -i -e "s|EXTRAVERSION = -bfq|EXTRAVERSION =|g" \
            -i -e "s|EXTRAVERSION =-mq|EXTRAVERSION =|g" ../${_bfq_sq_mq_patch}
        msg "Patching source with BFQ-SQ-MQ patches"
        patch -Np1 -i ../${_bfq_sq_mq_patch}

Works fine.

shanti commented on 2018-03-20 15:34 (UTC)


Thank you very much, unfortunately despite my trials, the patch fails to apply. I reverted to 4.14.12-2-ck which seem to be the last linux-ck including bfq-sq scheduler.

So to my understanding, bfq-sq has been removed from ck patchset itself but I couldn't find relevant information on Paolo Valente's website.

Can somebody elaborate why? or have any information whether it may come back or not.

RaskAr commented on 2018-03-16 10:47 (UTC)

@CarterCox See

CarterCox commented on 2018-03-16 03:02 (UTC)

==> ERROR: One or more PGP signatures could not be verified!

This is what I'm getting. Anyone knows how to solve it?

sir_lucjan commented on 2018-03-13 14:07 (UTC)


You can add bfq-sq/bfq-mq patchset:


shanti commented on 2018-03-13 13:55 (UTC)

As mentioned by someone earlier, bfq-sq scheduler is gone in recent releases. Can someone please explain why such an important feature of this kernel is being omitted or dropped? Thanks

j1simon commented on 2018-03-11 13:31 (UTC)

I see several warnings in my log "NOHZ: local_softirq_pending xxx" 4.15.8-1-ck-skylake

QuartzDragon commented on 2018-03-08 07:42 (UTC)

The problem is referenced here:

Problem should be fixed in systemd 238.0-2.

jambuu commented on 2018-03-07 11:22 (UTC)

The system is crashing in the latest version wrt systemd and glibc, I am having the same issue as @QuartzDragon

QuartzDragon commented on 2018-03-06 23:07 (UTC)

Anyone else get a systemd PID1-related coredump after the 238 systemd update currently in testing?

It's weird that only the stock Arch kernel worked, while systemd wasn't happy with linux-ck...

RaskAr commented on 2018-03-01 23:11 (UTC) (edited on 2018-03-01 23:28 (UTC) by RaskAr)

Yep, removing this file from the cache and fetching again, worked.

For reference and as a follow up on the touchPad + trackPoint of my pc not being fully enabled by this config, enabling _makenconfig=y option in the PKGBUILD permitted adding this two modules to the build:

Device Drivers -> input device support -> (M) RMI4 SMB Support

Device Drivers -> HID support -> Special HID drivers -> (M) Synaptics RMI4 device support

After a reboot everything was working as before, TITS (that is to say :D) the trackPoint was hardly moving in one direction only.

But adding psmouse.synaptics_intertouch=1 as a boot parameter (advertised in boot log) made the two freshly built modules to be loaded at boot, allowing the trackPoint to be usable without shortcomings !

As a proof, using xinput on X11 the touchPad is recognized in a different way : "SynPS/2 Synaptics TouchPad" now is "Synaptics TM2758-001". In Wayland the name is different but both devices work well too.

Terence commented on 2018-03-01 20:23 (UTC)

@RaskAr Remove the one you have and fetch it again.

graysky commented on 2018-03-01 20:12 (UTC)

Hello graysky, linux-ck 4.15.7-1 fails to build here : enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v4.13+.patch ... ÉCHEC

@RaskAr - Not sure what that means... builds fine for me.

RaskAr commented on 2018-03-01 20:01 (UTC)

Hello graysky, linux-ck 4.15.7-1 fails to build here : enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v4.13+.patch ... ÉCHEC

microsoftenator commented on 2018-02-27 06:38 (UTC) (edited on 2018-02-27 06:38 (UTC) by microsoftenator)

I'm getting this output in my dmesg whenever a VM is running:

BUG: using smp_processor_id() in preemptible [00000000] code: CPU 0/KVM/690
BUG: using smp_processor_id() in preemptible [00000000] code: CPU 1/KVM/691

It seems to be related to CONFIG_PREEMPT, since changing to CONFIG_PREEMPT_NONE=y or CONFIG_PREEMPT_VOLUNTARY=y eliminates the message.

This does not occur on the regular Arch kernel

graysky commented on 2018-02-23 20:41 (UTC)

@RaskAr - I use the ARCH .config as the base for linux-ck making only the changes the ck patchset requires. If this is something shipped with the kernel but silent in the config itself, you can enable it and build with it. If this is an out-of-tree addition, you are free to patch the source and build it.

RaskAr commented on 2018-02-23 19:46 (UTC)

Hello graysky, the config seems to be missing a builtin config for laptop users :

From log : « kernel: psmouse serio3: synaptics: The touchpad can support a better bus than the too old PS/2 protocol. Make sure MOUSE_PS2_SYNAPTICS_SMBUS and RMI4_SMB are enabled to get a better touchpad experience. »

After checking the config it only misses RMI4, which is the successor of the PS2 PROTOCOL, see Would be great to have it enabled !

presianbg commented on 2018-02-19 05:03 (UTC) (edited on 2018-02-20 16:54 (UTC) by presianbg)

bfq-sq is gone again.


PS: Putting "scsi_mod.use_blk_mq=1 " in kernel options is enabling the BFQ MQ scheduler. The selecting it with TLP, works just fine.

PS2: elevator option in the kernel parameters is not recognized.

graysky commented on 2018-02-18 20:35 (UTC)

which processor family should i chose for apollo lake (goldmont) processor Celeron N3450?

Did you try:

garwol commented on 2018-02-17 17:34 (UTC)

which processor family should i chose for apollo lake (goldmont) processor Celeron N3450?

Terence commented on 2018-02-17 16:50 (UTC)

@nTia89 It had to be from my side, it's working fine now with 4.14.20.

nTia89 commented on 2018-02-15 22:02 (UTC)

no problem here with .19 and DKMS

Terence commented on 2018-02-13 20:34 (UTC)

Something with 4.14.19 is making DKMS not working for any driver.

kata198 commented on 2018-02-08 05:03 (UTC)

@QuartzDragon - I applied the 4.14 patch with -F5 (fuzz-level 5, dangerous; requires manual review of every patch but "shoves" things in when at all possible. It even managed to shove some code right into the middle of a comment!), fixed the rejections, added one or two new compat functions as required due to API/ABI changes between .14 and .15, did some basic research into where things moved around and such and applied those changes.

I've done some kernel work in the past, so I like to think I'm somewhat familiar with the layout of things. It shouldn't take too long for you to pick up where things are, thanks to you're friendly General Regular Expression Parser (grep) .

Basically, checked the diff between 4.14 and 4.15 on the core scheduler to see any abi/api changes, and others as they arose ( I remember there being a change in some of the "compat" files which required updates/additions/rewrites beyond what the original patch provided, there may have been others ).

Then just built, and fixed any issues that arise during compile (these can be due to changing where #ifdef wrappers are in the baseline, where the patch may force things into the wrong section, or a linker issue from a missing function, etc.)

I'd say just dive right in! Try to merge 4.14 to 4.15 yourself, basically make it till you break it, or succeed! Probably a good idea to use a virtualbox image for testing rather than your primary system, or what I do on this laptop I am using to type to you right now, which is running arch in a virtualbox on top of windows 10 (work / warrenty reasons) is this: I take a snapshot prior to development, so any damage that gets done I can just roll back and alls good.

Hope this helps, I'm not too sure on what you are looking for specifically, so I tried to keep it general!

I would say just dig in! Try you m

kata198 commented on 2018-02-08 04:46 (UTC)

I mentioned specifically that I didn't know if they were true or not -- I just found a couple articles when searching to see if anyone else had ported it as it was time for me to build a new kernel. I mean no disrespect nor harm toward anybody. @beojan is probably right -- they may have been dated 2007 which I mistook for 2017.

If you've spoken with him and he says he's not -- all the better. But please don't try to make me out to be some sort of demon spreading nonsense -- I specifically said "I don't know if these reports are legitimate or not."

You're free to enjoy the early merge and update I did, or you're free to wait for him to release an official version. I'm still running strong without any issues on my merge, so it seems pretty stable to me.

Again, I apoligize if I've upset anybody, this was not my intent at all.

beojan commented on 2018-02-06 11:48 (UTC)

@kata198 I wonder if the "rumours" you're hearing are really old news from when he stopped maintaining the old ck patchset in 2007.

graysky commented on 2018-02-04 11:20 (UTC)

@kata198- While I applause your efforts to help port the ck patchset, announcing uncited and unconfirmed rumors as pretext is not cool.

CK himself replied to me in a private email:

Yeah bullshit. It'll probably be a week or so though.


On 4 February 2018 at 22:06, John <graysky AT archlinux DOT us> wrote:
> Specifically: "Hey guys, so I've read some rumors that Con Kolivas is taking
> a break / no longer maintaining the ck patchset; that 4.14 was the last
> version for a while. "

QuartzDragon commented on 2018-02-04 04:35 (UTC) (edited on 2018-02-04 04:36 (UTC) by QuartzDragon)


Where did you hear about these rumours? I'm curious as to why he'd stop maintaining the patchset without notifying everyone, and passing the reins onto someone capable.

Also, out of curiousity, how did you rebase the patches? I'm curious about learning how to do rebasing, myself.

kata198 commented on 2018-02-04 02:23 (UTC) (edited on 2018-02-04 02:29 (UTC) by kata198)

Linux-4.15 (Unofficial)

Hey guys, I've heard rumros that Con Kolivas may no longer be maintaining the ck patchset? I don't know if these reports are legitimate or not, but just in case they are, I went ahead and manually updated the ck patchset to cleanly apply with 4.15. It's been a little bit of time since I've done any sort of active kernel development, but I believe everything is correct here, and I've been running this ported patchset (ported patch-4.14-ck1.xz to work with linux-4.15) for almost 2 full days now without even the slightestnd ev indication of issue, I'm running on an intel i7 in SMP/SMT mode, through virtualbox as my primary laptop/desktop.

I'm posting to share my unofficial merge of the patchset, in the hopes that it may be of benefit to others.

You can find the patch and even a drop-in replacement of the PKGBUILD on stock arch-4.15 package, HERE:


Plexcon commented on 2018-02-02 20:56 (UTC)

drivers/gpu/drm/amd/amdgpu/.tmp_uvd_v7_0.o: warning: objtool: uvd_v7_0_ring_emit_fence()+0x45a: return with modified stack frame CC [M] drivers/gpu/drm/amd/amdgpu/amdgpu_vce.o /bin/sh: línea 1: 27516 Violación de segmento (`core' generado) ./tools/objtool/objtool orc generate --no-fp "drivers/gpu/drm/amd/amdgpu/.tmp_amdgpu_vce.o" make[4]: [scripts/ drivers/gpu/drm/amd/amdgpu/amdgpu_vce.o] Error 139 make[3]: [scripts/ drivers/gpu/drm/amd/amdgpu] Error 2 make[2]: [scripts/ drivers/gpu/drm] Error 2 make[1]: [scripts/ drivers/gpu] Error 2 make: *** [Makefile:1032: drivers] Error 2 ==> ERROR: Se produjo un fallo en build(). Cancelando...

Plexcon commented on 2018-02-02 19:33 (UTC)

/bin/sh: línea 1: 5200 Violación de segmento (`core' generado) ./tools/objtool/objtool orc generate --no-fp "drivers/infiniband/hw/mlx4/.tmp_mcg.o" make[4]: [scripts/ drivers/infiniband/hw/mlx4/mcg.o] Error 139 make[3]: [scripts/ drivers/infiniband/hw/mlx4] Error 2 make[2]: [scripts/ drivers/infiniband/hw] Error 2 make[1]: [scripts/ drivers/infiniband] Error 2 make: *** [Makefile:1032: drivers] Error 2 ==> ERROR: Se produjo un fallo en build(). Cancelando...

Plexcon commented on 2018-02-02 18:17 (UTC) (edited on 2018-02-05 19:49 (UTC) by Plexcon)

==> Verificando las firmas de las fuentes con gpg... linux-4.14.tar ... HA FALLADO (clave pública desconocida 79BE3E4300411886) patch-4.14.16 ... HA FALLADO (clave pública desconocida 38DBBDC86092693E) ==> ERROR: ¡No se ha podido verificar alguna de las firmas PGP!

sudo pacman-key --recv 38DBBDC86092693E sudo pacman-key --recv 79BE3E4300411886

niklaszantner commented on 2018-01-31 22:22 (UTC) (edited on 2018-01-31 22:23 (UTC) by niklaszantner)


==> Verifying source file signatures with gpg... linux-4.14.tar ... Passed patch-4.14.16 ... Passed

Works for me.

I assume that you have to add the keys to your system (looks like you are missing Linus' and Greg's key)

Following commands should help:

gpg --recv-key 79BE3E4300411886<br> gpg --recv-key 38DBBDC86092693E

AriseEVE commented on 2018-01-31 22:06 (UTC)

==> Verifying source file signatures with gpg... linux-4.14.tar ... FAILED (unknown public key 79BE3E4300411886) patch-4.14.16 ... FAILED (unknown public key 38DBBDC86092693E) ==> ERROR: One or more PGP signatures could not be verified! <br> ==> ERROR: Makepkg was unable to build linux-ck.

Unable to build, key check failed.

graysky commented on 2018-01-29 21:49 (UTC)

@Mthw - Not out-of-date for the same reason as graysky commented on 2018-01-29 09:52

Morganamilo commented on 2018-01-29 11:02 (UTC)

Ah my bad. Wasn't thinking about the patches.

graysky commented on 2018-01-29 09:52 (UTC)

@morganamilo-not out of date until ck1 for 4.15.x is released.

AstroProfundis commented on 2018-01-26 08:26 (UTC)

4.14.15-2 works for me as well, thanks

bratpilz commented on 2018-01-26 01:25 (UTC)

linux-ck-4.14.15-2 builds successfully for me. Thanks for the patch!

graysky commented on 2018-01-26 00:03 (UTC)

I included a patch from loqs that should fix the build error in 4.14.15-2. Please test and give some feedback.

AstroProfundis commented on 2018-01-25 05:53 (UTC)

I tried to disable CONFIG_SCHED_MUQSS in defconfig (and regenerated the config file thus enabled FAIR_GROUP_SCHED and CFS_BANDWIDTH automatically), the build was success without error. Hope this info helps.

vp1981 commented on 2018-01-25 01:09 (UTC)

Hello, just for the record: and corresponding commit is in 4.14.15.

I'm not familiar with MuQSS source but hope that the fix is simple enough and CK make new ck2.

sir_lucjan commented on 2018-01-24 19:55 (UTC)

I tried to compile linux-ck but I had the same error.

sir_lucjan commented on 2018-01-24 19:51 (UTC)

Look, I think there's been a misunderstanding.

Error seems related to ck-patchset/muqss. I don't use muqss / ck patchset.

Sources 4.14.15 build fine without errors.

sir_lucjan commented on 2018-01-24 19:50 (UTC)

Look, I think there's been a misunderstanding.

Error seems related to ck-patchset/muqss. I don't use muqss / ck patchset.

Sources 4.14.15 build fine without errors.

graysky commented on 2018-01-24 19:44 (UTC)

Yes but without MuQSS. Works fine. Seems related to ck-patchset.

...without MuQSS? So if you apply all the patches in the broken out ck1 except for 0001-MuQSS-version-0.162-CPU-scheduler.patch you are able to build without error?

sir_lucjan commented on 2018-01-24 15:48 (UTC)


Yes but without MuQSS. Works fine.

Seems related to ck-patchset.

graysky commented on 2018-01-24 15:35 (UTC)

@sir_l - you compiled 4.14.15-1 without errors? Looks like you're building a differently named kernel. What are the differences vs linux-ck?

sir_lucjan commented on 2018-01-23 22:21 (UTC) (edited on 2018-01-23 22:23 (UTC) by sir_lucjan)


Yes, seems related to MuQSS. I've compiled without errors:

Linux version 4.14.15-0.4-bfq-sq-mq-haswell-git (lucjan@archlinux) (gcc version 7.2.1 20171224 (GCC)) #1 SMP PREEMPT Mon Jan 22 17:03:42 CET 2018
Linux archlinux 4.14.15-0.4-bfq-sq-mq-haswell-git #1 SMP PREEMPT Mon Jan 22 17:03:42 CET 2018 x86_64 GNU/Linux

(stable-rc version)

graysky commented on 2018-01-23 22:19 (UTC)

@sir_l - Seems related to MuQSS. CK has been emailed.

Full log: Clipped log:

kernel/sched/MuQSS.c: In function ‘try_to_wake_up’:
kernel/sched/MuQSS.c:1962:3: error: too few arguments to function ‘delayacct_blkio_end’
In file included from kernel/sched/MuQSS.c:41:0:
./include/linux/delayacct.h:125:20: note: declared here
 static inline void delayacct_blkio_end(struct task_struct *p)
kernel/sched/MuQSS.c: In function ‘try_to_wake_up_local’:
kernel/sched/MuQSS.c:2025:4: error: too few arguments to function ‘delayacct_blkio_end’
In file included from kernel/sched/MuQSS.c:41:0:
./include/linux/delayacct.h:125:20: note: declared here
 static inline void delayacct_blkio_end(struct task_struct *p)
  CC [M]  sound/drivers/vx/vx_pcm.o
  CC [M]  net/8021q/vlanproc.o
make[2]: *** [scripts/ kernel/sched/MuQSS.o] Error 1
make[1]: *** [scripts/ kernel/sched] Error 2
make: *** [Makefile:1032: kernel] Error 2

sir_lucjan commented on 2018-01-23 21:59 (UTC) (edited on 2018-01-23 21:59 (UTC) by sir_lucjan)


Could you paste a log?

graysky commented on 2018-01-23 21:57 (UTC) (edited on 2018-01-23 21:58 (UTC) by graysky)

All - It seems that 4.14.15-1 is erroring out on the build for me. Need to dig into it further to see why.

graysky commented on 2018-01-23 21:39 (UTC)

@sir_lucjan - Thanks. I removed it but since it's harmless, I won't push a -2 release... it can be effective on the next version.

sir_lucjan commented on 2018-01-23 21:28 (UTC) (edited on 2018-01-23 21:31 (UTC) by sir_lucjan)

"chmod +x tools/objtool/" is not needed anymore.

graysky commented on 2018-01-21 12:56 (UTC)

could you please append 'make clean' and 'rm -fr Documentation' after 'make mrproper'?

'rm -fr Documentation' saves >2Gbytes on tmpfs and can avoid 'No space left on device' error which happens in low-ram PC.

@enihcam - The make clean would be redundant and I show linux-4.14/Documentation as being only 38M, not 2G.

zerophase commented on 2018-01-18 00:21 (UTC) (edited on 2018-01-18 00:21 (UTC) by zerophase)

@QuartzDragon Thanks got it working. I was zcatting config over too early in prepare earlier. (might have been missing switching Retpoline on) For some reason running zcat too early in prepare turns every module on.

QuartzDragon commented on 2018-01-17 23:55 (UTC)

@graysky @zerophase

The Retpoline patches for Spectre got backported to the current 4.14.14 release, not before. The Memory Page Isolation patches for Meltdown were earlier, though, yeah.

graysky commented on 2018-01-17 22:57 (UTC)

@zerophase - Yes, since version 4.14.11-1 I believe.

zerophase commented on 2018-01-16 21:29 (UTC)

I tried the whole zcat /proc/config.gz > ./.config. It seems to work fine, but it looks like more modules are turned on, and the kernel ballooned up by 51.66 MiBs for me. Did anyone else notice the kernel get larger?

Are the Spectre and Meltdown protections turned on by default?

cokomoko commented on 2018-01-12 20:28 (UTC)

@graysky No, I do not change anything. But I use the system in Turkish. Could it be a problem?

graysky commented on 2018-01-12 20:18 (UTC) (edited on 2018-01-12 20:18 (UTC) by graysky)

@cokomoko - Builds find for me. Are you modifying anything when building?

@gustawho - Yes.

gustawho commented on 2018-01-12 19:34 (UTC)

Does/will this build include any patches to address Meltdown and/or Spectre?

cokomoko commented on 2018-01-12 19:28 (UTC) (edited on 2018-01-12 19:34 (UTC) by cokomoko)

CC [M] arch/x86/crypto/sha512-mb/sha512_mb.o AS [M] arch/x86/crypto/sha512-mb/sha512_mb_mgr_flush_avx2.o CC [M] arch/x86/crypto/sha512-mb/sha512_mb_mgr_init_avx2.o AS [M] arch/x86/crypto/sha512-mb/sha512_mb_mgr_submit_avx2.o AS [M] arch/x86/crypto/sha512-mb/sha512_x4_avx2.o LD [M] arch/x86/crypto/sha512-mb/sha512-mb.o arch/x86/crypto/sha512-mb/sha512_mb.o: file not recognized: Dosya biçemi tanınmıyor make[3]: [scripts/ arch/x86/crypto/sha512-mb/sha512-mb.o] Error 1 make[2]: [scripts/ arch/x86/crypto/sha512-mb] Error 2 make[1]: [scripts/ arch/x86/crypto] Error 2 make: [Makefile:1032: arch/x86] Error 2

Where's the problem?

enihcam commented on 2018-01-11 04:47 (UTC)

Thank you @QuartzDragon.

Maybe @graysky could simple add a step to zcat /proc/config.gz to somewhere, for example /tmp/.config? I thought this should be harmless. In this way, people who want to reuse current config could save the manual step, i.e. just open the menu and load /tmp/.config.

zerophase commented on 2018-01-11 04:00 (UTC)

I wouldn't mind a solution for config_current either. I can't always remember what I turned on in the kernel every time.

QuartzDragon commented on 2018-01-11 03:21 (UTC) (edited on 2018-01-11 03:22 (UTC) by QuartzDragon)


You can always just keep config.last aside, or just zcat /proc/config.gz manually if you really need to.

enihcam commented on 2018-01-11 01:51 (UTC)

@graysky, is there any alternative solutions to using '_use_current'? I have many specialized devices running linux-ck, removal of _use_current causes a lot of troubles in maintaining the kernels.

Lessaj commented on 2018-01-09 01:26 (UTC) (edited on 2018-01-09 01:26 (UTC) by Lessaj)

As long as it stops to ask you for any CONFIG that it's missing it's good to keep. It used to stop to ask, I think what's preventing it is:

yes "" | make config >/dev/null

graysky commented on 2018-01-07 20:12 (UTC)

...I'm beginning to wonder if the _use_current variable is more trouble than it's worth

Lessaj commented on 2018-01-07 19:16 (UTC) (edited on 2018-01-07 19:17 (UTC) by Lessaj)

@eugen-b @Tjuh The parameter is not in the current config unless you've booted with a kernel that contains it so I added it, I believe it's actually getting stuck further down in prepare but commenting that out fixes it because it doesn't overwrite the current config.

  zcat /proc/config.gz > ./.config
  echo "CONFIG_PAGE_TABLE_ISOLATION=y" >> ./.config

Tjuh commented on 2018-01-07 16:48 (UTC)

Same issue as eugen-b when enabling config_audit=y in the config file. Haven't been able to compile since version 4.14.6-2.

eugen-b commented on 2018-01-07 16:21 (UTC) (edited on 2018-01-07 16:48 (UTC) by eugen-b)

I got endless lines related to CONFIG_PAGE_TABLE_ISOLATION=y after makeconfig -sri

I edited the PKGBUILD with _use_current=n, but it didn't help.

What helped was commenting the following lines in PKGBUILD

  ### Optionally use running kernel's config
  # code originally by nous;
#  if [ -n "$_use_current" ]; then
#    if [[ -s /proc/config.gz ]]; then
#      msg "Extracting config from /proc/config.gz..."
#      # modprobe configs
#      zcat /proc/config.gz > ./.config
#    else
#      warning "Your kernel was not compiled with IKCONFIG_PROC!"
#      warning "You cannot read the current config!"
#      warning "Aborting!"
#      exit
#    fi
#  fi

Saren commented on 2018-01-07 04:36 (UTC)

@graysky Thanks for making the toggle back.

graysky commented on 2018-01-06 12:24 (UTC) (edited on 2018-01-06 12:31 (UTC) by graysky)

@presianbg - My bad. Fixed in 4.14.12-3.

@saren - Yes, some recent benchmarking I did on a dual quad showed a pretty big hit so I reasoned that 99+% of users want it disabled. I reenabled the variable toggle.

presianbg commented on 2018-01-06 09:47 (UTC)

Hi, happy 2018! Wish you all the best! bfq-sq is missing again... what is going on :?

Saren commented on 2018-01-06 03:45 (UTC)

Hi there. Since does that mean I will have to comment out the NUMA disabled code everytime? Thanks.

tosto commented on 2018-01-05 18:01 (UTC)

@sl13kp Disabilita le notifiche, dovresti trovare un link in alto a destra nel riquadro "Package Actions" (e per favore evita di spammare nei commenti).

sl13kp commented on 2017-12-31 00:12 (UTC)

avete scocciato con le notifiche non mi interessa questo programma

QuartzDragon commented on 2017-12-28 09:49 (UTC) (edited on 2017-12-28 09:50 (UTC) by QuartzDragon)


Try deleting your old build files, because the patch applies perfectly fine for me. If you're not using TMPFS for your builds, then your old build files will stick around, causing you issues if you don't delete them every now and then.

kurwall commented on 2017-12-28 01:46 (UTC)

The GCC patch needs to be redone, as currently it is not applying properly

air-g4p commented on 2017-12-19 09:58 (UTC)

Hi grayslake,

SUCCESS! Linux-ck can now be built without issue in a clean chroot.

To make it absolutely clear to anyone who wants to build linux-ck in a clean chroot, grayslake referred to specifying the correct _subarch variable for your system on line 38 of PKGBUILD prior to building.

You, ROCK, grayslake!

Thank you for your assistance.

All the best - and Merry Christmas,


graysky commented on 2017-12-18 20:10 (UTC)

@air-g4p - You need to assign a value to the _subarch variable (line 38) before you try building.

air-g4p commented on 2017-12-18 15:55 (UTC)

oops: my missing (below) linux-ck 4.14.7-1 build failure link is here:

air-g4p commented on 2017-12-18 15:52 (UTC)

Hi grayslake,

I am sorry to have to inform you that using: extra-x86_64-build

fails with either linux-ck 4.14.6-3 or with linux-ck 4.14.7-1.

I have pasted the complete failed build runs at:

linux-ck 4.14.6-3 (per your suggestion) lives here:

and also with: linux-ck 4.14.7-1, here:

Despite the 'advice' generated just prior to build failure:

Console input/output is redirected. Run 'make oldconfig' to update configuration.

$ make oldconfig make: *** No rule to make target 'oldconfig'. Stop.

and ofc, this also:

$ sudo make oldconfig [sudo] password for <userid>: make: *** No rule to make target 'oldconfig'. Stop.</userid>

Those 'make oldconfig' failures occur irrespective of whether a user is in their correct build directory, or in their chroot directory = /var/lib/archbuild/extra-x86_64/<userid>/build</userid>

Also, carefully note, the build process never stopped, prior to failure, to ask me, which CPU I wanted to build support for. The current linux-ck AURs I have tested above both only auto-select '> 23. Generic-x86-64 (GENERIC_CPU)', then immediately fail.

grayslake - I do appreciate your ongoing support, but I would like to get to the bottom of this issue, sometime 'soon'. If you have specific commands you want me to test, let me know.

All the best,


graysky commented on 2017-12-18 12:22 (UTC)

@artafinde - good idea.

sir_lucjan commented on 2017-12-18 11:29 (UTC) (edited on 2017-12-18 11:30 (UTC) by sir_lucjan)



artafinde commented on 2017-12-18 07:58 (UTC)

@graysky I suggest for now to delete the 24 option (line #38) on the PKGBUILD. If a user wants native detection let him go through makepkg.

foi commented on 2017-12-18 06:16 (UTC)


Failed in install (distro: manjaro gnome):

==> Verifying source file signatures with gpg... linux-4.14.tar ... FAILED (unknown public key 79BE3E4300411886) patch-4.14.7 ... FAILED (unknown public key 38DBBDC86092693E) ==> ERROR: One or more PGP signatures could not be verified! ==> ERROR: Makepkg was unable to build linux-ck.

air-g4p commented on 2017-12-18 05:41 (UTC)

Hi graysky,

Thank you very much for attention and extremely quick correction of this issue. Highly impressive, indeed!

When I get a chance, I'll try a re-build with the current release, and let you how things went.

All the best,


graysky commented on 2017-12-17 23:33 (UTC) (edited on 2017-12-17 23:34 (UTC) by graysky)

@artafinde - Short of applying this patch and shipping the corresponding config (which makes building for repo-ck more difficult), I don't know how to cover that use-case.

It might be a bit of a hassle for you, but if you just run makepkg -s outside of the chroot, setup the config to your liking, you can replace the shipped config with config.last and update the checksums before you build in the chroot.

artafinde commented on 2017-12-17 23:23 (UTC)

@graysky: if you choose 24 (native) then you are asked again for P6_NOPS which fails to auto-configure with "yes" Support for P6_NOPs on Intel chips (X86_P6_NOP) [N/y/?] (NEW)

graysky commented on 2017-12-17 09:56 (UTC)

@air-g4p - Yes, the failure is due to user intervention being required to select a sub-arch. Please try 4.14.16-3 in which the _subarch variable should take care of this use case.

air-g4p commented on 2017-12-17 06:21 (UTC)

Hi graysky,

In the event you are not aware, your linux-ck 4.14.6-2 AUR fails to build in a clean chroot, using: extra-x86_64-build.

Your immediately prior release also failed to build using either extra-x86_64-build or makechrootpkg -c -r $CHROOT.

We know for a fact your AUR is broken because when discussing this linux-ck build failure on #archlinux, another experienced AUR user also replicated the same extra-x86_64-build failure on his system.

I have pasted my entire build run for your review at: That paste URL will remain valid for one month.

Additionally, for those users willing to employ this approach, I can confirm your AUR can be successfully built and installed with: makepkg -si.

I hope you are able to determine, and correct, whichever part(s) of your code is/are causing the build failure, as many of us strongly prefer to build AURs exclusively within a clean chroot environment.

Thank you for providing an otherwise excellent alternative kernel. Your effort is immensely appreciated.



eduardoeae commented on 2017-12-16 12:48 (UTC)

Works OK now. Thanks!

graysky commented on 2017-12-16 11:36 (UTC)

@edu and sha - Please try now.

Shaikh commented on 2017-12-16 08:03 (UTC)

I encountered the same problem as @eduardoeae.

eduardoeae commented on 2017-12-16 03:17 (UTC)

Your packages (from are failing with "missing required signature". I have been using them for months without problem. Also tried re-adding your key (pacman-key -r 5EE46C4C && pacman-key --lsign-key 5EE46C4C) as the site suggests, but still not working.

pedrogabriel commented on 2017-11-26 22:29 (UTC)

You plan to include the new runqueue patch?

johnnybegood commented on 2017-11-25 17:44 (UTC)

Thanks a lot for the ck patches for bfq-sq!!! :)

Osleg commented on 2017-11-25 16:21 (UTC)

Kernel panic solved, was unrelated to kernel itself

Osleg commented on 2017-11-25 13:44 (UTC)

Also this since upgrade to >4.14 Kernel panic with "No working init found" Any idea what could be the cause?

Osleg commented on 2017-11-25 13:38 (UTC)

Yeah It downloads the sources twice and it builds it twice. Just finished compilation using plain old and reliable makepkg, worked like a charm w/o duplication. :) Also, @graysky, was looking for you at IRC but since can't find you I'll ask you here. I have a server that standing idling for no reasons for quite long time and it will keep standing there. Would you like to host the$arch there? Cuz afaik lots of people have problems downloading the binary -ck images from your servers. But it's quite anoying to re-build it every time on my machine :) I can give you access to the machine if you'd like?

graysky commented on 2017-11-25 13:05 (UTC)

No idea how yaourt works but if it is running the build function twice, the author of it should be notified as that is wasteful. Are you sure that it isn't just building once, and packaging the kernel then headers? That is how makepkg does it.

Osleg commented on 2017-11-25 12:36 (UTC)

@sir_lucjan Thanks for responding, already found my question in documentation, this PKGBUILD is multi package for both headers and the kernel itself. The question tho is that possible to build both headers and kernel w/o re-downloading and re-building the entire kernel? Right now I'm installing using yaourt. So it download sources twice and building twice. How could I prevent that?

sir_lucjan commented on 2017-11-23 19:40 (UTC)

@Osleg Could you tell us what do you mean?

Osleg commented on 2017-11-23 19:37 (UTC)

I wonder what is the difference between this pkg and linux-ck as their PKGBUILDs are identical

Terence commented on 2017-11-23 15:05 (UTC)

@artafinde @graysky After some more investigations, I figured out those were just warnings with the purpose to warn you they will not be included by default in the kernel, and you can safely ignore them as dkms take care of building them. Thank you very much for your help, I hope this could eventually help someone confused like me.

QuartzDragon commented on 2017-11-23 10:03 (UTC) (edited on 2017-11-23 10:05 (UTC) by QuartzDragon)

Attention @graysky and @everyone: Urgently needed update! FS#56404 - [linux] Using bcache will destroy filesystems (4.14.X) ~ and: 4.14.1-2 patch ~ I've already updated manually for this.

Terence commented on 2017-11-22 23:11 (UTC)

@artafinde: Thanks for your complete answer. I rigorously followed your advices, booting back to the ARCH kernel and making sure once more dkms modules are blacklisted and not stored, but whatever I do, I still get the same error with localmodconfig when either compiling on ARCH or -ck. I don't understand what's wrong... Note that I also use dkms modules on the regular kernel.

artafinde commented on 2017-11-22 18:05 (UTC)

@Terence: modprobed-db is snapshotting your loaded modules and adds them into a database. Provided you run the system long enough and use it you should have all the modules you are using. localmodconfig is a tool to configure the kernel to use the modules which are currently used from the kernel dkms is using the kernel-headers to create new modules for nvidia/virtualbox etc. During linux-ck compilation and if you have set the localmodconfig parameter in PKGBUILD we load all modules which are referenced in modprobed-db prior to calling localmodconfig. Then kernel is configured according to the loaded modules and compiles etc. If you want to make sure you have a clean modprobed-db follow these steps: 1. clean up modprobed-db from modules which don't belong on kernel (nvidia, vbox etc) 2. add those to IGNORE list (optionally test by calling modprobed-db store nd check they are not added) 3. reboot and load -ARCH kernel (to make sure all the modules are available) 4. edit PKGBUILD and set localmodconfig - compile and install 5. dkms should build new modules (nvidia, vbox, exfat etc) 6. reboot to your new kernel The above process is what I follow and works fine with nvidia, vbox and exfat -dkms packages. As a fallback I have -ARCH kernel always installed.

Terence commented on 2017-11-22 17:42 (UTC) (edited on 2017-11-22 23:07 (UTC) by Terence)

@artafinde sorry I didn't make it clear enough but that's what I did :) To me it's like modprobe-db just loads all the modules it detected has been loaded at some point and then localmodconfig just lsmod, which includes dkms modules... There has to be a way to exclude them.

artafinde commented on 2017-11-22 17:38 (UTC)

Open the modprobed-db database (it's a file) and remove them from the list.

Terence commented on 2017-11-22 17:30 (UTC)

@artafinde @graysky Thanks but I knew that, that's why they are built using dkms and I took care of blacklisting all of them, making sure they are not appearing in the modprobe-db file, but problem persists.

artafinde commented on 2017-11-22 17:25 (UTC)

@Terence: The nvidia is external same as virtualbox modules. If you have dkms packages it should trigger a build after you install (provided you have headers). modprobed-db should probably exclude them with IGNORE list - see wiki.

graysky commented on 2017-11-22 17:19 (UTC)

@Terence - These are either provided by some other package or not included in the kernel anymore. See man modprobed-db.

Terence commented on 2017-11-22 17:10 (UTC)

I've never been able to find anywhere a solution to this problem with localmodconfig, modprobed-db and dkms modules: ==> If you have modprobed-db installed, running it in recall mode now Modprobed-db v2.37 164 modules currently loaded per /proc/modules 168 modules are in /home/terence/.config/modprobed.db Attempting to modprobe the contents of /home/terence/.config/modprobed.db 164 modules are now loaded per /proc/modules ==> Running Steven Rostedt's make localmodconfig now using config: '.config' vboxpci config not found!! nvidia config not found!! nvidia_uvm config not found!! vboxnetflt config not found!! vboxnetadp config not found!! vboxdrv config not found!! 8812au config not found!! nvidia_drm config not found!! nvidia_modeset config not found!! Any insight?

sir_lucjan commented on 2017-11-21 21:04 (UTC)

@johnnybegood I also recommended blk-mq patches.

johnnybegood commented on 2017-11-21 20:55 (UTC)

Works perfect for me as well. However, I still miss the old single queue bfq, the now called "bfq-sq" that was never merged into the official kernel, so I compiled ck with sirlucjan latest patch for 4.14 ( I find the combination of -ck and bfq-sq unbeatable. Maybe it's just me but it could be an addition to this excellent kernel. Graysky, thanks for your work and CK's!

zebulon commented on 2017-11-21 12:07 (UTC)

Works for me. Need to use scsi_mod.use_blk_mq=1 init option for blq as expected. @silvik: I do not need to use elevator=bfq, scsi_mod.use_blk_mq=1 is sufficient. I need however to set udev rules to set bfq.

graysky commented on 2017-11-20 21:37 (UTC)

CK pushed MuQSS v0.162 into ck1 for 4.14 so it's official. I merged with ARCH changes and here we go. Please test/provide feedback. I will hold off on pushing to [repo-ck] for at least 24 h to get some feedback.

silvik commented on 2017-11-20 20:12 (UTC)

BFQ works for me with scsi_mod.use_blk_mq=1 elevator=bfq kernel options.

francoism90 commented on 2017-11-04 21:28 (UTC)

@kwe: Do you have a source? We should try to keep the Wiki up-to-date, contributes are welcome. :)

Terence commented on 2017-11-04 20:22 (UTC)

@kwe Thanks and sorry for the off-topic.

kwe commented on 2017-11-04 20:20 (UTC)

@Terence 1. this is not really related to this AUR package anymore 2. don't take the wiki for the golden words from god 3. what the note means is that all sq schedulers are unavailable and (probably from pre-4.12 Kernels) that there were no mq-schedulers. 4.12 added Kyber and BFQ

Terence commented on 2017-11-04 20:14 (UTC) (edited on 2017-11-04 20:17 (UTC) by Terence)

@kwe Ok my bad, I forgot it's not deadline anymore but mk-deadline. Why is the wiki saying that I can't do that though?

kwe commented on 2017-11-04 20:10 (UTC)

@Terence Deadline should be an option: $ cat /sys/block/sda/queue/scheduler mq-deadline kyber [bfq] none

Terence commented on 2017-11-04 20:09 (UTC)

@kwe Ok, I'm indeed using grub and forgot about that, thanks. One more question: I have my root partition on my SSD and my home/var partition on my HDD. I was before using deadline on the SSD and BFQ on the HDD. It seems I can't do that anymore now and it's confirmed by the arch wiki. Will it's still perform ok with the SSD compared to deadline or is there another alternative for my system configuration?

kwe commented on 2017-11-04 20:02 (UTC)

@Terence: From what I can tell, the "elevator" parameter is being slowly phased out and might only set the sq scheduler. Using elevator=[mq-scheduler] seems to crash. There is a patch for the Linux kernel that fixes this crash by introducing a relevant check, but it won't be introduced until 4.15 IIRC. You don't need to boot a live system if you're using GRUB. You can edit the start parameters via the E key and then fix your GRUB config from within your normal system.

Terence commented on 2017-11-04 19:58 (UTC)

@kwe Thanks for your relevant comment, I tried to use the 'elevator' kernel parameter but it crashed the kernel at boot and I needed to boot to the live usb to be able to fix it. I then have 2 questions: 1) Why are there still patches for BFQ included? 2) Can the package be updated with "CONFIG_SCSI_MQ_DEFAULT = y"?

kwe commented on 2017-11-04 18:59 (UTC)

@everybody who is not involved, @Terence I've had some troubles figuring out what's up, so here is my tl;dr of the issue: We used to have BFQ as a part of the ck kernel patches. With a recent kernel version, you have BFQ merged into the vanilla kernel tree. Hence, BFQ was removed from the ck patchset. However, the BFQ in the kernel is for blk-mq, whereas previously we had bfq for non-mq / sq. Because this kernel defaults to sq (CONFIG_SCSI_MQ_DEFAULT = n), the mq BFQ from the Linux kernel is not available. The solution seems to be to either build with that option enabled or alternatively boot the kernel with scsi_mod.use_blk_mq=1

Terence commented on 2017-11-04 18:52 (UTC)

I can't get BFQ to be enabled with the latest version.

keepitsimpleengr commented on 2017-10-31 20:37 (UTC)

@graysky got it

graysky commented on 2017-10-31 19:29 (UTC)

@keepitsimp - It's due to the fact that nvidia-utils-387.22-1 is out but I know and thus didn't update nvidia-ck. It has been done now. In the future please flag the nvidia-ck package out-of-date.

mrkline commented on 2017-10-31 18:04 (UTC)

Apparently I'm blind and/or need more coffee. My apologies.

sir_lucjan commented on 2017-10-31 17:49 (UTC)

@mrkline # Patch source with ck patchset patch -Np1 -i "../${_ckpatchname}" patch -Np1 -R -i "../0015-Enable-and-make-BFQ-default-IO-scheduler-overriding-.patch" patch -Np1 -i "../unfuck_bfq_v2.patch" Do you see "patch -R" or not?

mrkline commented on 2017-10-31 17:36 (UTC)

@sir_lucjan - What does `patch -R` have to do with this? I'm posting because 4.13.10-1 allowed me to select BFQ as a scheduler, while 4.13.10-2 and 4.13.10-3 do not seem to. @artafinde - I've been following along for the most part. Previously BFQ wasn't the default, but I was able to switch to it on boot using a udev rule. I'm confused why this changed.

keepitsimpleengr commented on 2017-10-31 16:51 (UTC) (edited on 2017-10-31 16:52 (UTC) by keepitsimpleengr)

:: nvidia-ck-k10: installing nvidia-utils (387.22-1) breaks dependency 'nvidia-utils=387.12' :: nvidia-ck-sandybridge: installing nvidia-utils (387.22-1) breaks dependency 'nvidia-utils=387.12'

artafinde commented on 2017-10-31 11:04 (UTC)

@mrkline: This has been in discussion in forums official post: (read last two pages at least)

sir_lucjan commented on 2017-10-31 10:59 (UTC)

@mrkline: You should read 'patch --help.' -R --reverse Assume patches were created with old and new files swapped.

mrkline commented on 2017-10-31 00:45 (UTC)

@graysky: The latest two builds (4.13.10-2 and 4.13.10-3) don't seem to be building in BFQ. I see that you added CK's Enable-and-make-BFQ-default-IO-scheduler-overriding-.patch, but after building and installing, I see: $ cat /sys/block/sda/queue/scheduler noop deadline [cfq] Perhaps the patches are being applied in the wrong order?

timofonic commented on 2017-10-30 19:09 (UTC)

Hello. I would love to build ck patches, but need to use drm-tip. I might get one interesting patch you have and maybe scrap for some more, but I doubt patches could apply to this too new version. If it does, I'll happily do a variant of it. Anyway, could you help me to improve my PKGBUILD? I'll do an -anbox variant with the binder and the other subsystem enabled (I don't remember the name right now). I got busy IRL and a little lost about what to improve or continue changing. Thanks in advance.

j1simon commented on 2017-10-24 06:59 (UTC)

I have a problem with 4.13.x. I have opened a bug because this occurs with all kernels 4.13.x that I've tested: stock, zen, ck.

artafinde commented on 2017-10-23 09:09 (UTC)

@kyak: as mentioned [here], build yourself from aur [here]:

kyak commented on 2017-10-23 08:44 (UTC)

Hi graysky, The virtualbox-ck-host-modules-ivybridge doesn't seem to exist anymore in repo-ck. Is it as intended, and if so, what is the suggested replacement?

graysky commented on 2017-10-22 17:30 (UTC) (edited on 2017-10-22 17:36 (UTC) by graysky)

All - again, the old behavior causes boot freezes thus requiring: 1) users to tweak with udev or the like, 2) something I do with the help of loqs, 3) the best option is for CK to do it and release ck2 with the fix. I know 4.13.x has been delayed and I am no happier about that than you guys are. I emailed ck earlier today asking about his decision re: this fix in ck2 (recall the discussion on his blog last week), but haven't heard back yet. As well, if you're following the repo-ck thread on the bbs, I have asked loqs for his best solution as I neither have the hardware nor to the time to dig into this further right now. Stay tuned and thanks for your time to test and provide feedback.

nTia89 commented on 2017-10-22 17:19 (UTC)

@graysky, since changelog says: -CONFIG_DEFAULT_CFQ=y +CONFIG_DEFAULT_BFQ=y -CONFIG_DEFAULT_IOSCHED="cfq" +CONFIG_DEFAULT_IOSCHED="bfq" I expect "bfq" enabled without ANY trick (both kernel param., udev rule or any other...). Correct me if I am wrong.

Tjuh commented on 2017-10-22 17:17 (UTC)

Nice, thanks for that, most appreciated.

FarStar commented on 2017-10-22 17:09 (UTC) (edited on 2017-10-22 17:15 (UTC) by FarStar)

Hi, Personally, I use an udev rule: /etc/udev/rules.d/60-scheduler.rules ACTION=="add|change", KERNEL=="sd[a-z]", ATTR{queue/scheduler}="bfq" And I get: $ cat /sys/block/sda/queue/scheduler mq-deadline kyber [bfq] none

Tjuh commented on 2017-10-22 17:05 (UTC)

A bit confused, so how does one enable BFQ nowadays? Wiki says setting elevator=bfq as a kernel parameter does not work.

graysky commented on 2017-10-22 16:28 (UTC)

@nTia89 and Tjuh - I believe that is the expected output, no?

Tjuh commented on 2017-10-22 15:29 (UTC)

Same as nTia89.

nTia89 commented on 2017-10-22 15:25 (UTC)

I get this, no kernel parameters or anything else (e.g udev rules): $ cat /sys/block/sda/queue/scheduler [mq-deadline] kyber bfq none

graysky commented on 2017-10-22 13:46 (UTC)

Please try 4.13.9-1 which has a workaround incorporated. No special kernel parameters should be needed. Works for me without them. Feedback please!

QuartzDragon commented on 2017-10-20 08:56 (UTC)

Hey @graysky and everyone, The reason for all of the BFQ woes came from SCSI_MQ_DEFAULT defaulting to disabled upstream. I have BFQ automatically set via udev rules, and enabling SCSI_MQ_DEFAULT allowed BFQ to work as usual. This was a problem previously, and I'm surprised no-one seems to have remembered this... unless it was presumed that SCSI_MQ_DEFAULT was still enabled by default.

sir_lucjan commented on 2017-10-17 13:20 (UTC)

@zoidberg Could you read previous comments?

zoidberg commented on 2017-10-17 13:17 (UTC) (edited on 2017-10-17 13:37 (UTC) by zoidberg)

As of version 4.13.7-2, I am not able to select bfq as a scheduler. It is not available as a scheduler option. Is this intentional? $ cat /sys/block/sda/queue/scheduler noop deadline [cfq] Edit: Sorry, did not read previous comments re. the problem with bfq.

artafinde commented on 2017-10-16 21:14 (UTC)

Works fine, good job both. (also nvidia-dkms works fine I blame my debug stripped package)

sir_lucjan commented on 2017-10-16 21:02 (UTC)

No problem! I'm always at your service, John :D

graysky commented on 2017-10-16 21:00 (UTC)

Thanks to sir_lucjan's suggestion, 4.13.7-2 boots and works as expected. Please build and test for yourself.

artafinde commented on 2017-10-16 20:57 (UTC)

Hey lazy people, share the patch :)

sir_lucjan commented on 2017-10-16 20:56 (UTC)

@graysky - I'm lazier than you.... :D

graysky commented on 2017-10-16 20:51 (UTC)

@sir_lucjan - Yes, but I'm too lazy to use that lame captha on the blog :D

sir_lucjan commented on 2017-10-16 20:47 (UTC)

@graysky: As I wrote - of course you can quote me.

graysky commented on 2017-10-16 20:39 (UTC)

@sir_lucjan - Building what you proposed in the email now and will push 4.13.7-2 if it works. Would you mind pasting your email to ck's blog? Let him also consider it.

artafinde commented on 2017-10-16 20:21 (UTC) (edited on 2017-10-16 20:21 (UTC) by artafinde)

@graysky: confirmed bfq to blame. (side note: nvidia-dkms fails also).

sir_lucjan commented on 2017-10-16 20:17 (UTC)

@graysky Please check your mail.

graysky commented on 2017-10-16 20:03 (UTC)

It seems as though bfq is to blame. For me, booting elevator=cfq in the boot loader works as expected. Can others confirm?

artafinde commented on 2017-10-16 18:33 (UTC) (edited on 2017-10-16 19:56 (UTC) by artafinde)

How does the debug symbols impact performance? It would be handy to have backtraces information when this happens but it means we need to compile with debug symbols which probably defeats the purpose of optimizations (MNATIVE etc). It would be handy for CK though.. My kernel panic on i7-6700 here:

nTia89 commented on 2017-10-16 17:52 (UTC)

Same here (not too old hardware, i5-5200U). Compiled with options: MNATIVE and X86_P6_NOP I attach a photo, maybe it can be useful: I have to say that respond to command: CTRL+ALT+CANC reboot the pc

Saren commented on 2017-10-16 15:56 (UTC)

Can confirm it's not just you. I got kernel panic on boot

graysky commented on 2017-10-16 13:40 (UTC)

I just pushed 4.13.7-1 but I am unable to boot after compiling on an admittedly old system. Requesting feedback from other users.

zerophase commented on 2017-10-15 04:35 (UTC)

@ozmartian in /usr/lib/modules-load.d just comment out crypto_user in bluez.conf. I wasn't having issues with booting; the module just wouldn't load for me.

ozmartian commented on 2017-10-14 22:14 (UTC)

@zerophase could you provide more detail re crypto_user and what you edited and for which bluez package, pretty please? my linux-ck partitions are stalling on boot and crypto_user was the only error I could find in journal/logs so would save me having to figure it out myself until 4.13 is out... am temporarily using standard arch 4.13 kernel as a result and it feels like i am cheating on CK :p thanks!

zebulon commented on 2017-10-11 21:07 (UTC)

@zerophase: thanks, always interesting to hear experience from others. Hopefully the wait will not be too long.

zerophase commented on 2017-10-09 18:57 (UTC)

For supporting journaled quotas is another switch needed to be turned on other than QFMT_V2? Switched it on, but still cannot mount my disk with usrjquota and grpjquota turned on.

zerophase commented on 2017-10-06 18:56 (UTC)

@zebulon I just needed to edit one of the bluez packages to not load crypto_user, since that doesn't exist until 4.13. I understand ck has a busy life, and in the past has spent too much time hacking the kernel. I'm just trying to see what's up.

graysky commented on 2017-10-06 18:53 (UTC)

I emailed CK asking about the timelines for MuQSS for 4.13.x and he said 1-2 weeks. I also asked him to post a blog entry since others probably want to know too.

zebulon commented on 2017-10-06 11:29 (UTC)

@zerophase: Con Klivas' blog and repository mention the latest patch for 4.12. We need to wait. What kind of issues do you have?

zerophase commented on 2017-10-06 09:37 (UTC)

Has anyone heard from ck on 4.13 being released? Being on 4.12 still is starting to cause bugs with other arch packages.

keepitsimpleengr commented on 2017-10-04 14:46 (UTC)

:: nvidia-ck-k10: installing nvidia-utils (387.12-1) breaks dependency 'nvidia-utils=384.90'

keepitsimpleengr commented on 2017-09-28 15:12 (UTC)

nvidia-ck-sandybridge: installing nvidia-utils (384.90-1) breaks dependency 'nvidia-utils=384.69'

willianholtz commented on 2017-09-23 22:17 (UTC)

Hello, I have the same SD card problem as the following link. If the SD is plugged in when starting the kernel, it simply hangs, if it is not plugged in, when entering the DE does not mount!

graysky commented on 2017-09-23 17:49 (UTC)

That is to be expected with nvme.

zerophase commented on 2017-09-23 17:20 (UTC)

Should bfq be the default scheduler without setting anything, now? Just noticed my nvme drive is running with none by default.

eschwartz commented on 2017-09-15 13:11 (UTC)

Apparently around 2012 linux-ck-headers started providing linux-headers as a truly ugly workaround for some community package that had a hard dependency on linux-headers. AFAICT no package still requires linux-headers when they work for -ck as well, since we have generally moved to dkms. Can you please remove this provides?

graysky commented on 2017-09-06 19:44 (UTC) (edited on 2017-09-06 19:45 (UTC) by graysky)

@archzz - please do not flag out of date until ck releases a patchset suitable for the 4.13.0 series.

graysky commented on 2017-09-05 16:21 (UTC)

Due to x.69 being pushed to [extra]. Updated now.

keepitsimpleengr commented on 2017-09-05 16:09 (UTC)

nvidia-ck-k10: installing nvidia-utils (384.69-1) breaks dependency 'nvidia-utils=384.59'

hotty commented on 2017-08-26 16:05 (UTC)

Since it is weekend and I finally had some time I found the culprit in the form of my internal sd-card reader. If I have a sd-card inserted udev times out resulting in a blinking cursor. Without a card linux-ck boots successful. Also inserting a card after booting does not detect the sd-card but instead crashes systemd-udev with the kernel blaming the error on block/elevator.c Using an external card-reader works. To report the bug and send in the full error message, is it an bug in udev or in ck?

Tharbad commented on 2017-08-22 19:00 (UTC)

It's appears that there's a conflict between scsi_mod.use_blk_mq=0 and any of the multi queue I/O schedulers. So either set scsi_mod.use_blk_mq to 1 or use the "none" scheduler. Choosing option n2 makes this kernel worthless. I haven't tried it yet. Found it here:

vp1981 commented on 2017-08-22 12:21 (UTC)

@hotty, @Tharbad, guys would you mind to provide more details? For example, your system configuration (HDD/SSD, FS), kernel cmdline, what queue scheduler do you use and etc. And don't forget that graysky only build the kernel with CK patches, so report any problems to CK,

Tharbad commented on 2017-08-22 12:08 (UTC)

System freezes sometime during the graphical login. CPU is at 100%. ctrl+alt+F# and magic keys don't work.

hotty commented on 2017-08-20 21:23 (UTC)

Still got a non-bootable kernel, both from here or from repo-ck freezes at boot since version 4.12.7. Normal arch kernel boots without any issues. Just a blinking cursor and after 2 minutes it shows "Triggering uevents" but thats it.

graysky commented on 2017-08-17 19:13 (UTC)

I bumped to 4.12.8-1 but will hold on building for [repo-ck] until CK verifies the recent merges (github) will not be released to ck3 any time soon.

mrkline commented on 2017-08-15 21:38 (UTC) (edited on 2017-08-15 21:44 (UTC) by mrkline)

@graysky: See!msg/bfq-iosched/j4QUK4IPOhw/CY1LmmhcAwAJ, as previously mentioned - apparently other kernel devs have removed compile-time scheduler defaults. I've been able to "fix" it for now with a udev rule, slightly modified from what @vp1981 posted: ACTION=="add|change", KERNEL=="sd[a-z]", ATTR{queue/scheduler}="bfq" (The wildcard was causing several false positives and errors to show up in the system log while booting.)

graysky commented on 2017-08-15 21:20 (UTC) (edited on 2017-08-15 21:25 (UTC) by graysky)

@mrkline - Interesting... I get the same. EDIT:

mrkline commented on 2017-08-15 20:20 (UTC)

Scratch that, 4.12.7-2 seems to have the same issue.

mrkline commented on 2017-08-15 18:40 (UTC)

Apparently ck-2 contains changes to make BFQ the default scheduler: So we should be good to go as soon as @graysky bumps this.

vp1981 commented on 2017-08-15 12:32 (UTC) (edited on 2017-08-15 12:36 (UTC) by vp1981)

@mrkline, you may ask post-factum to translate new on, see, but this is receipt from the news: [russian] Планировщики ввода-вывода при использовании blk-mq нельзя указывать через командную строку ядра в загрузчике. Вместо этого используйте правило udev, например: ACTION=="add|change", KERNEL=="sd[a-z]*", ATTR{queue/scheduler}="bfq" Чтобы включить подсистему blk-mq, ядро нужно собрать с опцией CONFIG_SCSI_MQ_DEFAULT=y или передать параметр scsi_mod.use_blk_mq=1 из загрузчика. [/russian] (my translation) ... use udev rule, for example: ACTION=="add|change", KERNEL=="sd[a-z]*", ATTR{queue/scheduler}="bfq" In order to enable blk-mq subsystem the kernel has to be configured with option CONFIG_SCSI_MQ_DEFAULT=y or pass parameter scsi_mod.use_blk_mq=1 in boot loader. P.S. Gee, I found the news on post-factum site:, see "v4.12-pf2 is out" (

mrkline commented on 2017-08-15 07:27 (UTC) (edited on 2017-08-15 07:36 (UTC) by mrkline)

@graysky 4.12.7-1 boots and runs fine on several of my boxen. The only curious tidbit is that even though BFQ is set as the default: $ zcat /proc/config.gz | grep -i BFQ CONFIG_DEFAULT_BFQ=y CONFIG_DEFAULT_IOSCHED="bfq" CONFIG_IOSCHED_BFQ=y CONFIG_BFQ_GROUP_IOSCHED=y The kernel is defaulting to mq-deadline: $ cat /sys/block/sda/queue/scheduler [mq-deadline] kyber bfq none Is this what @kwe was discussing below? EDIT: See What's the simplest way to switch all disks to BFQ on boot in the meantime?

artafinde commented on 2017-08-15 06:49 (UTC) (edited on 2017-08-15 07:29 (UTC) by artafinde)

Building 4.12.7-1. Apart for the changes in config so you have to configure lots more than before we are missing the CPU optimizations. Update: Got a non-bootable kernel. Trying a second build, config seems better now as expected (I'll blame ccache which I disabled now). #2 Update: Non bootable again, stuck at "Triggering udevents"

pedrogabriel commented on 2017-08-15 02:03 (UTC)

4.12.7-1 works like a charm.

kwe commented on 2017-08-14 19:58 (UTC)

@graysky That config option will disappear in the future because the behavior will be enabled by default: Apparently this is planned for 4.13: linux/drivers/scsi$ git tag --contains 5c279bd9e40624f4ab6e688671026d6005b066fa v4.13-rc1 v4.13-rc2 v4.13-rc3 v4.13-rc4 v4.13-rc5 So, when it comes around, you might want to remove it again. :)

beest commented on 2017-08-14 19:25 (UTC)

4.12.7-1 boots, everything seems kosher so far.

graysky commented on 2017-08-14 18:31 (UTC)

@zebulon - Please try 4.12.7-1 which contains a suggestion by CK to fix. Does it boot and run? I cannot build and test for some time.

graysky commented on 2017-08-13 19:07 (UTC)

@amoka - There is not yet a stable ck1 patchset for 4.12.

pedrogabriel commented on 2017-08-13 05:11 (UTC)

Got a non-bootable system too.

zebulon commented on 2017-08-12 19:25 (UTC)

For your info, when you wait long enough, you get error messages: "systemd-udevd blocked for more than 120 seconds".

post-factum commented on 2017-08-12 19:16 (UTC)

-ck is still buggy, unfortunately. I'd recommend downgrade for the time being.

graysky commented on 2017-08-12 16:27 (UTC)

@zerophase - Yes, I too get a non-bootable. I emailed CK. @kwe - Agreed, thanks for pointing that out. Will be edited in the -3 release.

kwe commented on 2017-08-12 16:09 (UTC)

@graysky How much sense does the _BFQ_enable_ option have with 4.12? $ grep -i bfq config.x86_64 CONFIG_DEFAULT_BFQ=y CONFIG_DEFAULT_IOSCHED="bfq" CONFIG_IOSCHED_BFQ=y CONFIG_BFQ_GROUP_IOSCHED=y

zerophase commented on 2017-08-12 16:01 (UTC)

I just get a blinking cursor.

zebulon commented on 2017-08-12 15:43 (UTC) (edited on 2017-08-12 15:52 (UTC) by zebulon)

@graysky: unfortunately I cannot boot. It compiled fine (with Skylake optimisation) and seemed to install. But for the first time with the ck kernel, it hangs at the very beginning of the boot sequence. I can only see systemd version, then nothing happens.

zebulon commented on 2017-08-12 15:27 (UTC)

@graysky: thanks. I was wondering why MuQSS had disappeared from config.i686.

graysky commented on 2017-08-12 15:05 (UTC)

@zebulon - Fuck, you're correct. Fixed in 4.12.6-2 just now.

zebulon commented on 2017-08-12 14:53 (UTC)

@graysky: thanks for maintaining this package. One thing I noticed: up to 4.11 it asked via a numbered menu for the kind of CPU optimisation the user wanted. This time it did not, is this normal?

puithove commented on 2017-07-29 11:37 (UTC)

Looks like CK was busy for a while playing with his new Tesla :) Good for him.

zerophase commented on 2017-07-28 19:31 (UTC)

4.12 is up on core. Are we waiting on the ck patch?

JunioCalu commented on 2017-07-16 11:37 (UTC)

About P6_NOPS, should I enable it?

graysky commented on 2017-07-13 22:34 (UTC)

@gnu - done, will be in the next update, thks

beest commented on 2017-07-13 15:02 (UTC)

Just for the sake of pedantry, next release you may want to change the URL preceding _BFQ_enable_ to as the fragment in the original comment is dead and the linux-ck page links there now anyway.

r08 commented on 2017-07-11 20:35 (UTC)

@QuartzDragon @artafinde Thanks for the feedback. I guess I'm not the only one... Will test this using CONFIG_HZ_PERIODIC=y to see if I get the same latency issues..

artafinde commented on 2017-07-09 07:17 (UTC) (edited on 2017-07-09 07:30 (UTC) by artafinde)

@QuartzDragon: I have similar behaviour with my laptop here (Intel(R) Core(TM) i7-4510U CPU @ 2.00GHz). I was setting the tick to 1000Hz before MuQSS but on the first iterations of MuQSS I think it was better. Do you get same freezes with linux from core? Is it fixed if you compile your own -ARCH and set tick to 1000Hz?

QuartzDragon commented on 2017-07-08 22:38 (UTC)

When I change CONFIG_HZ to 1000, CONFIG_HZ_100_MUQSS seems to be automatically unset.

artafinde commented on 2017-07-08 08:57 (UTC) (edited on 2017-07-08 08:57 (UTC) by artafinde)

MuQSS 0.150 reverted the Hz back to 100Hz and I think that's the recommended value from ck. @QuartzDragon, r08 do you change both the below: CONFIG_HZ_100_MUQSS=y CONFIG_HZ=100

graysky commented on 2017-07-08 07:37 (UTC)

@Quartz and r08 - Post to ck's blog?

QuartzDragon commented on 2017-07-08 06:54 (UTC) (edited on 2017-07-08 06:55 (UTC) by QuartzDragon)

In the past, I too had noticed weird stuttering using 100Hz, so now I always set it to 1000Hz. System is smooth. :)

r08 commented on 2017-07-07 10:51 (UTC)

Interesting how I noticed slow downs using the default MuQSS@100Hz while using Firefox and running Steam games like Team Fortress 2. Setting it to 1000Hz fixed the latency issues I was having. Should 1000Hz tick be the recommended setting for a "normal desktop user?" Running old Intel Core 2 T9500@2.60Ghz

zerophase commented on 2017-07-05 22:23 (UTC)

Anyone else getting mce errors at boot, occasionally, on Haswell? I only get them when I hard reboot if the computer stalls after displaying time till next fsck.

nTia89 commented on 2017-07-05 17:58 (UTC)

Have you read it?

vp1981 commented on 2017-07-05 02:43 (UTC)

@j1simon: I think SuperIce97 meant that if you compare output for 'native' for both Skylake and Kaby Lake then you don't see any difference. I have two hosts one with Skylake and other with Kaby Lake. I compared they 'native' output and the difference: $ diff -y --suppress-common-lines native-sl.out native-kl.out -march= skylake | -march= broadwell -mtune= skylake | -mtune= generic For example, this is difference between Sandy Bridge and Skylake: $ diff -y --suppress-common-lines native-sb.out native-sl.out -mabm [disabled] | -mabm [enabled] -madx [disabled] | -madx [enabled] -march= sandybridge | -march= skylake -mavx2 [disabled] | -mavx2 [enabled] -mbmi [disabled] | -mbmi [enabled] -mbmi2 [disabled] | -mbmi2 [enabled] -mclflushopt [disabled] | -mclflushopt [enabled] -mf16c [disabled] | -mf16c [enabled] -mfma [disabled] | -mfma [enabled] -mfsgsbase [disabled] | -mfsgsbase [enabled] -mlzcnt [disabled] | -mlzcnt [enabled] -mmovbe [disabled] | -mmovbe [enabled] -mprfchw [disabled] | -mprfchw [enabled] -mrdrnd [disabled] | -mrdrnd [enabled] -mrdseed [disabled] | -mrdseed [enabled] -msgx [disabled] | -msgx [enabled] -mtune= sandybridge | -mtune= skylake -mxsavec [disabled] | -mxsavec [enabled] -mxsaves [disabled] | -mxsaves [enabled] Sandy Bridge: Intel(R) Core(TM) i5-2410M CPU @ 2.30GHz Skylake: Intel(R) Core(TM) i5-6200U CPU @ 2.30GHz Kaby Lake: Intel(R) Core(TM) i5-7400 CPU @ 3.00GHz @SuperIce97: at first I was confused why 'native' reports that arch is 'broadwell' and even thought that GCC don't support this new processor. But if you are right about behavior of GCC when one passes 'native' vs 'NAME' then it is related only to 'generic' build for Kaby Lake and that processor is fully supported by GCC.

SuperIce97 commented on 2017-07-04 21:57 (UTC) (edited on 2017-07-04 21:59 (UTC) by SuperIce97)

@j1simon: I believe this is because gcc optimizes in a way to make sure that all CPUs from that architecture will run correctly, meaning that if 1 CPU is missing an instruction, it will not be included in the optimization. This leads to differences as some CPUs in the same architecture supports newer instructions. Here is the diff between the broadwell arch optimizations and native for an Intel 3205u Celeron Broadwell CPU: diff -y --suppress-common-lines broadwell.out native.out -mabm [disabled] | -mabm [enabled] -mcx16 [disabled] | -mcx16 [enabled] -mfsgsbase [disabled] | -mfsgsbase [enabled] -mfxsr [disabled] | -mfxsr [enabled] -mlzcnt [disabled] | -mlzcnt [enabled] -mmmx [disabled] | -mmmx [enabled] -mmovbe [disabled] | -mmovbe [enabled] -mno-sse4 [enabled] | -mno-sse4 [disabled] -mpclmul [disabled] | -mpclmul [enabled] -mpopcnt [disabled] | -mpopcnt [enabled] -mprfchw [disabled] | -mprfchw [enabled] -mrdrnd [disabled] | -mrdrnd [enabled] -mrdseed [disabled] | -mrdseed [enabled] -msahf [disabled] | -msahf [enabled] -msse [disabled] | -msse [enabled] -msse2 [disabled] | -msse2 [enabled] -msse3 [disabled] | -msse3 [enabled] -msse4 [disabled] | -msse4 [enabled] -msse4.1 [disabled] | -msse4.1 [enabled] -msse4.2 [disabled] | -msse4.2 [enabled] -mssse3 [disabled] | -mssse3 [enabled] As you can see, on my Broadwell CPU, broadwell optimization != native optimizations. Also notice how "native" only enables optimizations instead of disabling. This is why I usually recommend -march=native, as you will be able to optimize your CPU as much as possible. For Kaby Lake vs Skylake, here is the Wikipedia page listing the differences (which are all clock speed/process, socket/chipset, and GPU changes, not microarchitecture changes):

j1simon commented on 2017-07-04 12:23 (UTC) (edited on 2017-07-04 12:24 (UTC) by j1simon)

@SuperIce97 I don't sure of that. My CPU: Intel(R) Core(TM) i7-7700 CPU @ 3.60GHz I execute: $ gcc -march=skylake -mtune=skylake -Q --help=target > skylake.out and: $ gcc -march=native -mtune=native -Q --help=target > native.out and these are the differences: $ diff -y --suppress-common-lines skylake.out native.out -mabm [disabled] | -mabm [enabled] -madx [disabled] | -madx [enabled] -maes [disabled] | -maes [enabled] -march= skylake | -march= broadwell -mavx [disabled] | -mavx [enabled] -mavx2 [disabled] | -mavx2 [enabled] -mbmi [disabled] | -mbmi [enabled] -mbmi2 [disabled] | -mbmi2 [enabled] -mclflushopt [disabled] | -mclflushopt [enabled] -mcx16 [disabled] | -mcx16 [enabled] -mf16c [disabled] | -mf16c [enabled] -mfma [disabled] | -mfma [enabled] -mfsgsbase [disabled] | -mfsgsbase [enabled] -mfxsr [disabled] | -mfxsr [enabled] -mhle [disabled] | -mhle [enabled] -mlzcnt [disabled] | -mlzcnt [enabled] -mmmx [disabled] | -mmmx [enabled] -mmovbe [disabled] | -mmovbe [enabled] -mno-sse4 [enabled] | -mno-sse4 [disabled] -mpclmul [disabled] | -mpclmul [enabled] -mpopcnt [disabled] | -mpopcnt [enabled] -mprfchw [disabled] | -mprfchw [enabled] -mrdrnd [disabled] | -mrdrnd [enabled] -mrdseed [disabled] | -mrdseed [enabled] -mrtm [disabled] | -mrtm [enabled] -msahf [disabled] | -msahf [enabled] -msgx [disabled] | -msgx [enabled] -msse [disabled] | -msse [enabled] -msse2 [disabled] | -msse2 [enabled] -msse3 [disabled] | -msse3 [enabled] -msse4 [disabled] | -msse4 [enabled] -msse4.1 [disabled] | -msse4.1 [enabled] -msse4.2 [disabled] | -msse4.2 [enabled] -mssse3 [disabled] | -mssse3 [enabled] -mtune= skylake | -mtune= generic -mxsave [disabled] | -mxsave [enabled] -mxsavec [disabled] | -mxsavec [enabled] -mxsaveopt [disabled] | -mxsaveopt [enabled] -mxsaves [disabled] | -mxsaves [enabled] I think it isn't the same.

sir_lucjan commented on 2017-06-27 10:14 (UTC)

@zhenyu: 1. gpg --receive-keys 79BE3E4300411886 38DBBDC86092693E 2., not

zhenyu commented on 2017-06-27 03:30 (UTC) (edited on 2017-06-27 05:28 (UTC) by zhenyu)

two errors: first,two unknown keys 79BE3E4300411886 38DBBDC86092693E,fix it plz. second,the can not update,exec 'pacman -Syy' return me 'can not find ck-repo.db'. plz tell me how to fix it.thx!

graysky commented on 2017-06-23 21:30 (UTC) wrote: "Please do not leave a comment containing the version number every time you update the package. This keeps the comment section usable for valuable content mentioned above." I will therefore stop commenting as I have :)

francoism90 commented on 2017-06-16 17:23 (UTC) (edited on 2017-06-16 17:24 (UTC) by francoism90)

@graysky: sorry for delay, output of Kaby Lake system: $ gcc -c -Q -march=native --help=target | grep 'march\|mtune' -march= broadwell -mtune-ctrl= -mtune= generic $ cat /proc/cpuinfo | grep 'model name' | uniq model name : Intel(R) Core(TM) i3-7100 CPU @ 3.90GHz @graysky @vp1981: weird it shows broadwell :S

vp1981 commented on 2017-06-12 11:26 (UTC) (edited on 2017-06-12 11:27 (UTC) by vp1981)

I have host with $ cat /proc/cpuinfo | grep 'model name' | uniq model name : Intel(R) Core(TM) i5-7400 CPU @ 3.00GHz and $ gcc -c -Q -march=native --help=target | grep 'march\|mtune' gives me -march= broadwell -mtune-ctrl= -mtune= generic While on host with $ cat /proc/cpuinfo | grep 'model name' | uniq model name : Intel(R) Core(TM) i5-6200U CPU @ 2.30GHz the same 'gcc...' gives me -march= skylake -mtune-ctrl= -mtune= skylake I faced with this mismatch for Kaby Lake when was installing Archlinux on that new host. After some investigation I decided that -march|-mtune=skylake is appropriate for Kaby Lake processor. P.S. It is curious enough that both processor on Intel site are categorized as 'Products formely...', not sure what that mean. See and

graysky commented on 2017-06-11 19:30 (UTC)

Are you sure? On your kaby lake what is the output of: gcc -c -Q -march=native --help=target | grep 'march\|mtune'

francoism90 commented on 2017-06-11 18:53 (UTC)

@SuperIce97: Ah thanks, I'll give it a try. :)

SuperIce97 commented on 2017-06-11 18:38 (UTC)

@francoism90: Kaby Lake has an identical architecture to Skylake (they just improved the manufacturing process which simply allows for higher clock speeds), so you could just use use the Skylake optimized version. You could compile it with native optimizations.

francoism90 commented on 2017-06-11 16:27 (UTC)

@graysky: thanks, will install generic instead. :)

graysky commented on 2017-06-11 16:25 (UTC)

There is no KL support in gcc v7.1.1.

francoism90 commented on 2017-06-11 16:21 (UTC)

Hi graysky, Thanks for providing the kernel and packages. :) Is it possible to add support for Kaby Lake?

graysky commented on 2017-05-26 21:52 (UTC)

@mareexx and Davikar - Best to post to ck's blog as well since that is more or less the bug tracker for the patchset.

mareex commented on 2017-05-26 21:26 (UTC)

I get really high cpu load with 4.11. These processes have huge cpu spikes: rcu_preempt kworker/u8:<some_number> irq/279-s-iwlwi ksoftirqd/1 My fans are screaming. Stock arch kernel is calm. Can anybody else confirm?

Davikar commented on 2017-05-26 17:04 (UTC)

Ever since I started using 4.11 I keep getting these kernel panics: ------------[ cut here ]------------ WARNING: CPU: 1 PID: 449 at net/ipv4/tcp_input.c:2819 tcp_fastretrans_alert+0x8e7/0xad0 Modules linked in: ip6table_nat nf_nat_ipv6 ip6t_REJECT nf_reject_ipv6 ip6table_mangle ip6table_raw nf_conntrack_ipv6 nf_defrag_ipv6 xt_recent ipt_REJECT nf_reject_ipv4 xt_comment xt_multiport xt_conntrack xt_hashlimit xt_addrtype xt_mark xt_nat xt_tcpudp xt_CT iptable_raw nf_log_ipv6 xt_NFLOG nfnetlink_log xt_LOG nf_log_ipv4 nf_log_common nf_nat_tftp nf_nat_snmp_basic nf_conntrack_snmp nf_nat_sip nf_nat_pptp nf_nat_proto_gre nf_nat_irc nf_nat_h323 nf_nat_ftp nf_nat_amanda ts_kmp nf_conntrack_amanda nf_conntrack_sane nf_conntrack_tftp nf_conntrack_sip nf_conntrack_pptp nf_conntrack_proto_gre nf_conntrack_netlink nfnetlink nf_conntrack_netbios_ns nf_conntrack_broadcast nf_conntrack_irc nf_conntrack_h323 nf_conntrack_ftp ip6table_filter ip6_tables iptable_filter iptable_mangle ipt_MASQUERADE nf_nat_masquerade_ipv4 iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat nf_conntrack libcrc32c crc32c_generic btrfs xor adt7475 hwmon_vid iTCO_wdt gpio_ich iTCO_vendor_support evdev mac_hid raid6_pq nouveau led_class mxm_wmi wmi video psmouse ttm i2c_i801 drm_kms_helper lpc_ich skge sky2 drm syscopyarea sysfillrect asus_atk0110 sysimgblt fb_sys_fops i2c_algo_bit button shpchp intel_agp intel_gtt acpi_cpufreq tpm_tis tpm_tis_core tpm sch_fq_codel coretemp msr nfsd auth_rpcgss oid_registry nfs_acl lockd grace sunrpc ip_tables x_tables ext4 crc16 jbd2 fscrypto mbcache sd_mod ata_generic pata_acpi serio_raw atkbd libps2 uhci_hcd ehci_pci ehci_hcd ahci libahci usbcore usb_common pata_jmicron mpt3sas raid_class libata scsi_transport_sas scsi_mod i8042 serio CPU: 1 PID: 449 Comm: irq/19-enp5s4 Tainted: G W 4.11.3-1-ck-core2 #1 Hardware name: System manufacturer System Product Name/P5B-Deluxe, BIOS 1238 09/30/2008 Call Trace: <IRQ> dump_stack+0x63/0x83 __warn+0xcb/0xf0 warn_slowpath_null+0x1d/0x20 tcp_fastretrans_alert+0x8e7/0xad0 tcp_ack+0xe57/0x14f0 tcp_rcv_established+0x11f/0x6f0 ? sk_filter_trim_cap+0xb7/0x270 tcp_v4_do_rcv+0x130/0x210 tcp_v4_rcv+0xb39/0xcc0 ip_local_deliver_finish+0xa1/0x200 ip_local_deliver+0x5d/0x100 ? inet_del_offload+0x40/0x40 ip_rcv_finish+0x1eb/0x3f0 ip_rcv+0x2b3/0x3c0 ? ip_local_deliver_finish+0x200/0x200 __netif_receive_skb_core+0x507/0xa70 ? tcp4_gro_receive+0x11a/0x1c0 ? try_preempt+0x160/0x190 __netif_receive_skb+0x18/0x60 netif_receive_skb_internal+0x81/0xd0 napi_gro_receive+0xdb/0x150 skge_poll+0x397/0x880 [skge] net_rx_action+0x242/0x3d0 __do_softirq+0x104/0x2e1 ? irq_finalize_oneshot.part.2+0xe0/0xe0 do_softirq_own_stack+0x1c/0x30 </IRQ> do_softirq.part.4+0x41/0x50 __local_bh_enable_ip+0x88/0xa0 irq_forced_thread_fn+0x59/0x70 irq_thread+0x12f/0x1c0 ? wake_threads_waitq+0x30/0x30 kthread+0x108/0x140 ? irq_thread_dtor+0xc0/0xc0 ? kthread_create_on_node+0x70/0x70 ret_from_fork+0x2c/0x40 ---[ end trace a181bdf0ee69c250 ]---

graysky commented on 2017-05-25 19:03 (UTC)

Bump to v4.11.3-1 Changelog:

mrturcot commented on 2017-05-21 03:03 (UTC)

Thanks guys, I was able to upgrade with @grayskys instructions. Also duly noted @mrkline, cheers.

mrkline commented on 2017-05-20 21:41 (UTC)

@mrturcot nvidia-dkms is also a viable option if you don't mind building it building the driver locally with each kernel update.

graysky commented on 2017-05-20 13:54 (UTC) (edited on 2017-05-20 13:59 (UTC) by graysky)

Bump to v4.11.2-1 Changelog:

graysky commented on 2017-05-20 10:24 (UTC)

I can't read any content from that hatebin provider. Try using a gist on github or another provider. In any case, you will need to build both packages first on a machine that isn't running linux-ck with nvidia-ck installed due to the versioning dependencies. Or, uninstall all of your ck packages first and boot into the standard arch kernel, then build them stepwise. I hope you understand.

mrturcot commented on 2017-05-20 10:16 (UTC) (edited on 2017-05-20 10:17 (UTC) by mrturcot)

@graysky not quite srry if I was not clear. My pending updates are Linux ck and Nvidia ck. It does not matter now I initiate the upgrade, all ways lead to this error Tried to upgrade just Linux-ck first by itself and it fails, tried nvidia-ck & it installs linux-ck as a dependency first and fails with that error as well. Thanks

graysky commented on 2017-05-20 09:51 (UTC)

@mrt - You built and installed linux-ck 4.11.1-2 and then were unable to build nvidia-ck?

mrturcot commented on 2017-05-20 05:30 (UTC) (edited on 2017-05-20 05:31 (UTC) by mrturcot)

Hey getting error when updating with nvidia-ck installed... :: nvidia-ck: installing linux-ck (4.11.1-2) breaks dependency 'linux-ck<4.11' More info - Please help... i'm stuck thanks.

graysky commented on 2017-05-20 01:37 (UTC)

Bump to v4.11.1-2 Changelog: fix pre-made config for cpu type Commit:

graysky commented on 2017-05-20 01:26 (UTC)

@mrkline - It's not a dumb question at all. Mistake on my part. Fixed in 4.11.1-2. Thank you for the report.

mrkline commented on 2017-05-20 00:18 (UTC) (edited on 2017-05-20 00:22 (UTC) by mrkline)

@graysky: Perhaps this is a dumb question, but is there a reason the build process no longer prompts for your CPU type? I now have to go set CONFIG_MNATIVE and CONFIG_X86_P6_NOP with makenconfig again - the menu you introduced in December has been much more convenient.

graysky commented on 2017-05-19 22:14 (UTC) (edited on 2017-05-19 22:15 (UTC) by graysky)

Bump to v4.11.1-1 Changelog: Commit: Linux changes: Notes: I need some feedback on the 4.11 release, specifically the extra packages such as nvidia-304xx-ck, nvidia-340xx-ck, broadcom-wl-ck, and virtualbox-ck modules. I have successfully build all of these but can only test nvidia-ck and virtualbox-ck since I do not have older nvidia hardware nor broadcom-wl hardware. Once I get some positive feedback, I will build up the 4.11.x series in [repo-ck] but for now, will hold. Also 4.11.2 is due out tomorrow.

QuartzDragon commented on 2017-05-15 05:38 (UTC)

Do you have the PKGBUILD for 4.11 sitting around? Would love to test it. :)

graysky commented on 2017-05-14 15:46 (UTC)

The major show stopper are the nvidia drivers at the moment. The kernel builds fine but I don't want mess up users of nvidia-xxx.

hekel commented on 2017-05-14 14:49 (UTC)

@graysky I see, thanks. I've only recently started using your repo/pkgbuild again. I guess I spoiled myself by building my own asap for so long.

graysky commented on 2017-05-14 14:19 (UTC) (edited on 2017-05-14 14:23 (UTC) by graysky)

@hekel - and see (threads titled Kernel 4.11 status).

hekel commented on 2017-05-14 13:25 (UTC)

BFQ has been out for 4.11 since 05/01 4.11-ck1 has been out since 05/11 4.11.1 just dropped today and 4.11.0 has been out since 4/30 …Is there a particular reason this was updated for 4.10.16 instead of 4.11.1?

graysky commented on 2017-05-14 12:20 (UTC)

Bump to v4.10.16-1 Changelog: Commit:

monotykamary commented on 2017-05-13 10:49 (UTC) There is one for 340xx in the nvidia forums, but haven't found one for 304xx.

graysky commented on 2017-05-12 21:59 (UTC)

Link to a patch for 304xx and 340xx?

monotykamary commented on 2017-05-12 21:31 (UTC) (edited on 2017-05-12 21:32 (UTC) by monotykamary)

@graysky Gotcha. For dkms, virtualbox-host-dkms builds correctly here. nvidia-dkms requires a MODULE_LICENSE edit for nvidia-drm/nvidia-drm-linux.c. I have not tested broadcom.

graysky commented on 2017-05-12 19:09 (UTC)

@monotykamary - Need to verify that all broadcom, nvidia-xxxx, and vbox modules build against 4.11.x before I bump it.

keepitsimpleengr commented on 2017-05-11 14:32 (UTC)

nvidia-ck-sandybridge: installing nvidia-utils (381.22-1) breaks dependency 'nvidia-utils=378.13'

graysky commented on 2017-05-08 18:31 (UTC)

Bump to v4.10.15-1 Changelog: Commit:

graysky commented on 2017-05-03 18:03 (UTC)

Bump to v4.10.14-1 Changelog: Commit:

graysky commented on 2017-04-27 19:12 (UTC)

Bump to v4.10.13-1 Changelog: Commit:

graysky commented on 2017-04-21 22:17 (UTC)

Bump to v4.10.12-1 Changelog: Commit:

artafinde commented on 2017-04-21 14:11 (UTC) (edited on 2017-04-21 14:44 (UTC) by artafinde)

BFQ scheduled for mainline - yoohoo!!topic/bfq-iosched/D-7wDP47iBI

graysky commented on 2017-04-18 07:26 (UTC)

Bump to v4.10.11-1 Changelog: Commit:

kyak commented on 2017-04-14 05:29 (UTC)

Having some problems with Docker and Linux-ck, posted here:

graysky commented on 2017-04-12 19:02 (UTC)

Bump to v4.10.10-1 Changelog: Commit:

bacondropped commented on 2017-04-08 18:10 (UTC)

Hm, I'm getting an occasional "kernel bug" crash with BFQ enabled on a per-device basis (4.10.8 and 4.10.9 with NUMA enabled). I've disabled BFQ, let's see if it persists.

graysky commented on 2017-04-08 11:21 (UTC)

Bump to v4.10.9-1 Changelog: Commit:

sidro commented on 2017-04-07 18:06 (UTC)

Other errors: drivers/media/dvb-frontends/tda10071.c:1267:1: fatal error: opening dependency file drivers/media/dvb-frontends/.tda10071.o.d: No such file or directory MODULE_FIRMWARE(TDA10071_FIRMWARE); ^~~~~~~~~~~~~~~ compilation terminated. make[3]: *** [scripts/ drivers/media/dvb-frontends/tda10071.o] Error 1 make[2]: *** [scripts/ drivers/media/dvb-frontends] Error 2 make[1]: *** [scripts/ drivers/media] Error 2 make: *** [Makefile:992: drivers] Error 2 ==> ERROR: A failure occurred in build(). Aborting... ==> ERROR: Makepkg was unable to build linux-ck.

SuperIce97 commented on 2017-04-07 18:03 (UTC)

@sidro: This is just how AUR/Arch packages work. There is no way to "fix" the package. It is expected that you import the gpg keys before you build the package.

sir_lucjan commented on 2017-04-07 18:03 (UTC)

gpg --recv-keys 79BE3E4300411886 gpg --recv-keys 38DBBDC86092693E Code works fine. PEBKAC.

sidro commented on 2017-04-07 18:00 (UTC) (edited on 2017-04-07 18:02 (UTC) by sidro)

Because original Build Script return errors. ==> Verifying source file signatures with gpg... linux-4.10.tar ... FAILED (unknown public key 79BE3E4300411886) patch-4.10.8 ... FAILED (unknown public key 38DBBDC86092693E) ==> ERROR: One or more PGP signatures could not be verified! ==> ERROR: Makepkg was unable to build linux-ck. Fix your code. Other errors: drivers/media/dvb-frontends/tda10071.c:1267:1: fatal error: opening dependency file drivers/media/dvb-frontends/.tda10071.o.d: No such file or directory MODULE_FIRMWARE(TDA10071_FIRMWARE); ^~~~~~~~~~~~~~~ compilation terminated. make[3]: *** [scripts/ drivers/media/dvb-frontends/tda10071.o] Error 1 make[2]: *** [scripts/ drivers/media/dvb-frontends] Error 2 make[1]: *** [scripts/ drivers/media] Error 2 make: *** [Makefile:992: drivers] Error 2 ==> ERROR: A failure occurred in build(). Aborting... ==> ERROR: Makepkg was unable to build linux-ck.

sir_lucjan commented on 2017-04-07 17:04 (UTC)

@sidro: You're wrong.

SuperIce97 commented on 2017-04-07 17:02 (UTC)

@sidro: How is it wrong?

sidro commented on 2017-04-07 16:59 (UTC)

Greg Kroah-Hartman sign is wrong. Fix it.

sir_lucjan commented on 2017-04-07 16:25 (UTC)

gpg --recv-keys 79BE3E4300411886 38DBBDC86092693

sidro commented on 2017-04-07 16:13 (UTC)

==> Verifying source file signatures with gpg... linux-4.10.tar ... FAILED (unknown public key 79BE3E4300411886) patch-4.10.8 ... FAILED (unknown public key 38DBBDC86092693E) ==> ERROR: One or more PGP signatures could not be verified! ==> ERROR: Makepkg was unable to build linux-ck.

graysky commented on 2017-03-31 17:58 (UTC)

Bump to v4.10.8-1 Changelog: Commit:

graysky commented on 2017-03-30 11:14 (UTC)

Bump to v4.10.7-1 Changelog: Commit:

graysky commented on 2017-03-26 14:59 (UTC)

Bump to v4.10.6-1 Changelog: Commit:

graysky commented on 2017-03-22 18:33 (UTC)

Bump to v4.10.5-1 Changelog: Commit:

graysky commented on 2017-03-18 12:48 (UTC)

Bump to v4.10.4-1 Changelog: Commit:

graysky commented on 2017-03-15 19:20 (UTC)

Bump to v4.10.3-1 Changelog: Commit:

graysky commented on 2017-03-13 18:15 (UTC) (edited on 2017-03-13 18:18 (UTC) by graysky)

Bump to v4.10.2-2 Changelog: apply fix for CVE-2017-2636 Commit:

graysky commented on 2017-03-12 18:33 (UTC)

@bugs - Manjaro != Arch. It's not surprising that Arch packages might not work on it.

bugscze commented on 2017-03-12 17:05 (UTC)

Manjaro won't boot with this kernel. In default, Kernel panic shows up. Adding this line into /boot/grub/grub.cfg: initrd /boot/initramfs-linux-ck-ivybridge.img kernel boots but cannot startx (with and without nvidia-ck-ivybridge) Anyone knows how to solve this issue?

graysky commented on 2017-03-12 11:18 (UTC)

Bump to v4.10.2-1 Changelog: Commit:

graysky commented on 2017-03-11 10:16 (UTC)

@ArchNemo: Arch != Antergos.

QuartzDragon commented on 2017-03-11 00:28 (UTC) (edited on 2017-03-11 00:29 (UTC) by QuartzDragon)

Have you tried this command? cat /sys/block/<drive>/queue/scheduler

m8D2 commented on 2017-03-11 00:08 (UTC) (edited on 2017-03-11 01:17 (UTC) by m8D2)

How to persistently enable bfq scheduler? I tried adding "elevator=bfq" in kernel parameter as suggested by Arch's wiki, but it didn't do anything. Thanks! Updates I just found out Antergos by default has a script trying to set scheduler in: /etc/udev/rules.d/60-schedulers.rules So I commented everything in that file and it's good now, nothing is wrong with linux-ck. I guess Antergos isn't real Arch, eh? :D

mradermaxlol commented on 2017-03-03 21:05 (UTC)

@graysky nope, it's *probably* not gonna be there because of AMD devs' habits, I think, but that's a good idea to try to push a backport to 4.10.X. Will do that ASAP.

graysky commented on 2017-03-03 20:22 (UTC)

@mrader - I tend to not put out-of-tree patches in the PKGBUILD and [repo-ck]... Do you know if this is planned to be back-ported in the 4.10 stable queue?[1] I currently see nothing there... 1.

mradermaxlol commented on 2017-03-03 20:13 (UTC)

@graysky is it possible to include this - - patch in the next PKGREL? It works for me and people on bugtracker, so it should be just fine to include it. It's a cherry-pick from the in-development 4.11 kernel, and thus would be useless and safe to remove after you update the package to that version.

graysky commented on 2017-03-01 19:42 (UTC)

@cooljay - again, recommend that you open a FS against the ARCH kernel

cooljay032 commented on 2017-03-01 13:28 (UTC)

commit tells us otherwise -CONFIG_LIRC_SERIAL=m -CONFIG_LIRC_SERIAL_TRANSMITTER=y not sure if there are other (new) options to run serial devices!?! +CONFIG_IR_SERIAL=m +CONFIG_IR_SERIAL_TRANSMITTER=y think i have to dig in here...

graysky commented on 2017-02-28 21:16 (UTC)

@cooljay - It's not in the official ARCH kernel either. I simply mirror ARCH's config (with the ck changes). The module has either been dropped upstream or we aren't configured to build it... you might wanna open up a flyspray against the kernel. When I look at the 4.9 -> 4.10 commit, I don't see that we dropped the config option containing 'serial' though:

cooljay032 commented on 2017-02-28 20:19 (UTC)

seems the lirc_serial driver missing!?! -- Unit systemd-modules-load.service has begun starting up. Feb 28 21:07:41 vdr systemd-modules-load[801]: Inserted module 'acpi_cpufreq' Feb 28 21:07:41 vdr systemd-modules-load[801]: modprobe: ERROR: could not find module by name='lirc_serial' Feb 28 21:07:41 vdr systemd-modules-load[801]: modprobe: ERROR: could not insert 'lirc_serial': Unknown symbol in module, or unknown parameter (see dmesg) Feb 28 21:07:41 vdr systemd[1]: systemd-modules-load.service: Main process exited, code=exited, status=1/FAILURE Feb 28 21:07:41 vdr systemd[1]: Failed to start Load Kernel Modules.

SuperIce97 commented on 2017-02-28 00:29 (UTC)

@mrkline That's a bit odd, yeah. nvidia-libgl is now on version 378.13-3 but nvidia-dkms is 378.13-2. You need 378.13-3 for Linux 4.10 support, so I wonder why they haven't pushed that dkms update yet.

mrkline commented on 2017-02-28 00:26 (UTC)

Ah - turns out the patch needed for 4.10 is in nvidia-dkms 378.13-3, which is currently only available in the testing repo. Installing it by hand from did the trick.

mrkline commented on 2017-02-28 00:11 (UTC) (edited on 2017-02-28 00:14 (UTC) by mrkline)

nvidia-dkms fails to build against 4.10.1-1. Is there a known workaround? Or do we now have to use nvidia-ck? (I've been using nvidia-dkms with linux-ck for quite a while without any trouble.) From the build log: CC [M] /var/lib/dkms/nvidia/378.13/build/nvidia/nv-vtophys.o /var/lib/dkms/nvidia/378.13/build/nvidia/nv-pat.c: In function ‘nvidia_cpu_callback’: /var/lib/dkms/nvidia/378.13/build/nvidia/nv-pat.c:213:14: error: ‘CPU_DOWN_FAILED’ undeclared (first use in this function) case CPU_DOWN_FAILED: ^~~~~~~~~~~~~~~ /var/lib/dkms/nvidia/378.13/build/nvidia/nv-pat.c:213:14: note: each undeclared identifier is reported only once for each function it appears in /var/lib/dkms/nvidia/378.13/build/nvidia/nv-pat.c:220:14: error: ‘CPU_DOWN_PREPARE’ undeclared (first use in this function) case CPU_DOWN_PREPARE: ^~~~~~~~~~~~~~~~ In file included from /var/lib/dkms/nvidia/378.13/build/nvidia/nv-pat.c:15:0: /var/lib/dkms/nvidia/378.13/build/nvidia/nv-pat.c: In function ‘nv_init_pat_support’: /var/lib/dkms/nvidia/378.13/build/common/inc/nv-linux.h:391:34: error: implicit declaration of function ‘register_cpu_notifier’ [-Werror=implicit-function-declaration] #define register_hotcpu_notifier register_cpu_notifier ^ /var/lib/dkms/nvidia/378.13/build/nvidia/nv-pat.c:258:17: note: in expansion of macro ‘register_hotcpu_notifier’ if (register_hotcpu_notifier(&nv_hotcpu_nfb) != 0) ^~~~~~~~~~~~~~~~~~~~~~~~ /var/lib/dkms/nvidia/378.13/build/nvidia/nv-pat.c: In function ‘nv_teardown_pat_support’: /var/lib/dkms/nvidia/378.13/build/common/inc/nv-linux.h:388:36: error: implicit declaration of function ‘unregister_cpu_notifier’ [-Werror=implicit-function-declaration] #define unregister_hotcpu_notifier unregister_cpu_notifier ^ /var/lib/dkms/nvidia/378.13/build/nvidia/nv-pat.c:283:9: note: in expansion of macro ‘unregister_hotcpu_notifier’ unregister_hotcpu_notifier(&nv_hotcpu_nfb); ^~~~~~~~~~~~~~~~~~~~~~~~~~ CC [M] /var/lib/dkms/nvidia/378.13/build/nvidia/os-interface.o cc1: some warnings being treated as errors make[2]: *** [scripts/ /var/lib/dkms/nvidia/378.13/build/nvidia/nv-pat.o] Error 1 make[2]: *** Waiting for unfinished jobs.... make[1]: *** [Makefile:1494: _module_/var/lib/dkms/nvidia/378.13/build] Error 2 make[1]: Leaving directory '/usr/lib/modules/4.10.1-1-ck/build' make: *** [Makefile:81: modules] Error 2

graysky commented on 2017-02-27 22:43 (UTC)

Bump to v4.10.1-1 Changelog: Commit:

graysky commented on 2017-02-26 11:31 (UTC)

Bump to v4.9.13-1 Changelog: Commit:

QuartzDragon commented on 2017-02-26 00:24 (UTC) (edited on 2017-02-26 00:25 (UTC) by QuartzDragon)

Cannot reproduce here, even after many hours of system activity. I also have an Ivy Bridge CPU.

simona commented on 2017-02-25 19:42 (UTC)

Anyone has freeze problem with all CK (CK standard, CK-ivybridge) ?

graysky commented on 2017-02-23 20:32 (UTC)

Bump to v4.9.12-1 Changelog: Commit:

graysky commented on 2017-02-23 20:32 (UTC)

@kyak - Rebuilt a few days ago. Thanks for reporting.

kyak commented on 2017-02-21 18:14 (UTC)

It seems that virtualbox-ck-host-modules-ivybridge requires rebuild: Failed to insert 'vboxnetadp': Invalid argument Failed to insert 'vboxnetflt': Invalid argument Failed to insert 'vboxnetadp': Invalid argument Failed to insert 'vboxnetflt': Invalid argument Kernel 4.9.11-1-ck-ivybridge from repo-ck.

Gelmo commented on 2017-02-21 04:58 (UTC)

@dkaylor Thanks for the heads up. I didn't realize this was following the Core release exclusively, and I've removed the flag. Regardless, I'm leaving the PKGBUILD; it builds 4.10 just fine :D

dkaylor commented on 2017-02-20 09:16 (UTC)

@gelmo Please read graysky's wiki for linux-ck before flagging out-of-date. He's been doing a very good job with this package for years now, your updated PKGBUILD is not needed...

Gelmo commented on 2017-02-20 07:11 (UTC) (edited on 2017-02-20 08:58 (UTC) by Gelmo)

Here's a working PKGBUILD for 4.10: Updated versioning, commented out the upstream patches, commented out the console loglevel patch (implemented now; I got this option presented to me pre-nconfig). Edit: The loglevel option is now available in nconfig under Kernel Hacking Edit 2: I also had to comment out the fpu patch on my setup. If you do this, don't forget to comment it out in the sources and prepare sections. Currently writing this from 4.10.0-1-ck, no issues.

graysky commented on 2017-02-18 17:20 (UTC)

Bump to v4.9.11-1 Changelog: Commit:

graysky commented on 2017-02-15 22:45 (UTC)

Bump to v4.9.10-1 Changelog: Commit:

graysky commented on 2017-02-09 22:34 (UTC)

Bump to v4.9.9-1 Changelog: Commit:

cdown commented on 2017-02-08 16:31 (UTC)

@VerruckteFuchs @fekle This is a bug in pacaur. I've just put up a patch to fix it at

VerruckteFuchs commented on 2017-02-06 20:58 (UTC)

@graysky @artafinde A simple makepkg -sri after clearing out the src directory sorted it out. It looks like something broke after pacaur got updated.

artafinde commented on 2017-02-06 20:07 (UTC)

Just follow the wiki:

fekle commented on 2017-02-06 20:00 (UTC)

@graysky to wait and see if other people experience this problem ;) you're right though, just tried installing manually and it worked just fine - oops.

graysky commented on 2017-02-06 18:54 (UTC)

Why would trying in a few days make a difference? Don't build with an AUR helper. Those things are way more trouble than they're worth. Just build with makepkg and profit.

VerruckteFuchs commented on 2017-02-06 18:40 (UTC)

@artafinde I've run cd ~/.cache/pacaur/linux-ck/src/linux-4.9 make clean #I also tried make mrproper (used in the PKGBUILD) and make distclean I also tried pacaur -Sc to see if that did anything to help. As far as I can tell skimming over the PKGBUILD, in the prepare() step make mrproper comes after the initial patching and should clean the build environment. Oddly enough I had no issues upgrading linux-ck on my laptop, but on my desktop it's having trouble with the prepare() step for some reason. It might be because I upgraded pacaur before upgrading linux-ck and there is a change in PKGBUILD syntax/formatting or something else. On my laptop I'm pretty sure I upgraded linux-ck and pacaur at the same time, which means the previous version of pacaur handled the installation. That's all I can think of that's different.

fekle commented on 2017-02-06 16:57 (UTC)

@VerruckteFuch I've got the same problem: Tried deleting .cache/pacaur, choosing both yes and no when asked (see paste), to no avail. Anyway, I'll try again in a few days, and as always, thanks for the great work @graysky!

artafinde commented on 2017-02-06 16:01 (UTC)

@VerruckteFuchs clean your build environment.

VerruckteFuchs commented on 2017-02-06 15:14 (UTC) (edited on 2017-02-06 15:15 (UTC) by VerruckteFuchs)

I'm getting a couple of errors in the prepare() step: The next patch would create the file arch/x86/include/asm/asm-prototypes.h, which already exists! Assume -R? [n] The next patch would create the file include/asm-generic/asm-prototypes.h, which already exists! Assume -R? [n] I keep getting a failure occurring in prepare() regardless if I answer 'y' or 'n' to the question and to the "Apply anyway? [n]" question. I've started with a fresh download of everything by deleting the ~/.cache/pacaur/linux-ck/ folder. I've had pacaur redownload everything using this method a few times and the failure persists.

graysky commented on 2017-02-04 23:39 (UTC)

Bump to v4.9.8-1 Changelog: Commit:

nicegamer7 commented on 2017-02-04 21:23 (UTC)

@graysky Sorry to bother then, only wanted to make sure :)

walkingrobot commented on 2017-02-03 03:32 (UTC)

@zerophase no I do not. I've never tried it.

zerophase commented on 2017-02-03 02:16 (UTC)

@walkingrobot thanks. I turned the SCSI one on, and I noticed I was able to boot successfully once in a bluemoon without udev compiled into my initramfs. Do you know if there is currently a way to boot without udev, and an nvme drive in the pcie slot?

Smasher816 commented on 2017-02-03 00:44 (UTC)

I am getting this error. If I manually extract the .xz file, then I end up receiving conflicts during the patching. patch: **** Can't open patch file XXXXXXX/linux-ck/src/patch-4.9.7 : No such file or directory

graysky commented on 2017-02-02 21:18 (UTC)

@niceg - It's correct as-is.

nicegamer7 commented on 2017-02-02 14:45 (UTC)

On the PKGBUILD...

nicegamer7 commented on 2017-02-02 14:45 (UTC)

I believe there is a typo on line 168. It is "s'/..." instead of "'s/...".

xsmile commented on 2017-02-02 14:04 (UTC)

@graysky: I would like to request a patch addition to fix low CPU freuqencies after standby, most probably limited to Dell notebooks. Please see for the patch and for more details on the bug tracker. Thanks for the consideration.

walkingrobot commented on 2017-02-02 10:05 (UTC)


graysky commented on 2017-02-02 00:12 (UTC)

Sorry about that... fixed but didn't bump pkgver.

SuperIce97 commented on 2017-02-01 22:07 (UTC)

The config files are failing to pass the sha256sums

zerophase commented on 2017-02-01 21:41 (UTC)

@graysky I'm just running into issues with booting the drive outside of fallback initramfs, but this isn't unique to CK.

graysky commented on 2017-02-01 21:39 (UTC)

Bump to v4.9.7-1 Changelog: Commit:

graysky commented on 2017-02-01 21:38 (UTC)

@zerophase - you shouldn't need anything special

zerophase commented on 2017-01-31 23:40 (UTC)

Does anyone know which NVME switches I should turn on in the kernel? I turned on CONFIG_BLK_DEV_NVME, and CONFIG_NVME_TARGET. I just left NVME_TARGET_LOOP, and CONFIG_BLK_DEV_NVME_SCSI off. Are there any other NVME settings I should turn on?

QuartzDragon commented on 2017-01-27 04:01 (UTC)

Finally updated to 4.9 after watching the bugs get fixed. Nice and stable so far! Thanks graysky! :)

graysky commented on 2017-01-27 00:39 (UTC)

Bump to v4.9.6-1 Changelog: Commit:

graysky commented on 2017-01-20 20:59 (UTC)

Bump to v4.9.5-1 Changelog: Commit:

graysky commented on 2017-01-15 13:28 (UTC)

Bump to v4.9.4-1 Changelog: Commit:

enihcam commented on 2017-01-15 02:18 (UTC)

Since the issue was worked around by enabling suspend-to-RAM, I did not want to spend time in repro it with makepkg. That was for your information only, not a request.

graysky commented on 2017-01-15 01:58 (UTC)

Does that mean it compiles without modification? Please use makepkg and report back.

enihcam commented on 2017-01-15 01:24 (UTC)

@graysky the variable is declared if and only if suspend-to-RAM is enabled, otherwise compiler cannot find the variable therefore complains. I manually compiled the kernel sources with patches you provided.

snack commented on 2017-01-14 18:50 (UTC)

With linux-ck-core2 4.9.3-1 and nvidia-340xx-ck-core2 340.101-3 I cannot start the X server. I get an error "Error: driver Nvidia is already registered. Aborting". No problem for me to stay with linux 4.8, this is just a warning for those wishing to upgrade.

graysky commented on 2017-01-14 15:14 (UTC)

Disabling suspend to ram and affect compilation? Did you try building with makepkg?

enihcam commented on 2017-01-14 14:06 (UTC)

@graysky Disabling suspend-to-RAM can repro the issue.

graysky commented on 2017-01-14 13:24 (UTC)

How does that pm freezing lik you referenced apply to your situation? Did you try building as-is with makepkg?

enihcam commented on 2017-01-14 13:17 (UTC)

@graysky The issue is discussed here: Seems like one patch is missing.

graysky commented on 2017-01-14 12:42 (UTC)

Stop using an AUR helper which I believe you're using and just build with makepkg.

enihcam commented on 2017-01-14 06:41 (UTC) (edited on 2017-01-14 12:05 (UTC) by enihcam)

Both of my two machines failed in compiling Linux-ck 4.9.3 with GCC 6.3.1. Anyone got the same issue? <-Log Start--------------> LD fs/built-in.o ==> ERROR: A failure occurred in build(). Aborting... ==> ERROR: Makepkg was unable to build linux-ck. ==> Restart building linux-ck ? [y/N] ==> --------------------------------- ==> <-Log End------------->

SanskritFritz commented on 2017-01-13 21:55 (UTC)

Users of nvidia-304xx have held back the last version anyway since current version doesn't work at all :D

graysky commented on 2017-01-13 21:47 (UTC)

Bump to v4.9.3-1 Changelog: Commit: Linux changes: Notes: Users of nvidia-304xx will have to pull the corresponding utils package from [testing] until 4.9 is pushed into [core]. Sorry about the inconvenience but way more users of linux-ck want the 4.9 series kernel and I didn't want to hold it up any longer.

graysky commented on 2017-01-12 21:19 (UTC)

@artafinde - READ 2 POSTS DOWN

artafinde commented on 2017-01-12 21:05 (UTC)

Is there a reason the linux 4.9.x is not pushed to repos yet? Someone mentioned in ML that's it's unstable - and seen reports of crashes on BTRFS ML.

graysky commented on 2017-01-12 19:35 (UTC)

For those wanting the 4.9.3 release -

graysky commented on 2017-01-12 07:24 (UTC)

@enihcam - Not out-of-date, plz read comments. graysky commented on 2017-01-07 19:25 graysky commented on 2016-12-22 19:02

zerophase commented on 2017-01-11 05:22 (UTC)

@graysky sounds like an efi issue.

graysky commented on 2017-01-10 09:10 (UTC)

@pedrogabriel - @zero - Haswell here and no problems.

zerophase commented on 2017-01-09 23:29 (UTC)

@pedrogabriel I believe 4.9 currently has issues with some old and new intel chips. That's why it hasn't had a release for arch yet.

pedrogabriel commented on 2017-01-09 23:25 (UTC)

Could you release a pastebin of 4.9?

graysky commented on 2017-01-09 09:14 (UTC)

Bump to v4.8.17-1 Changelog: Commit:

graysky commented on 2017-01-07 19:25 (UTC)

@nTia89 - Fixed thanks. @Saren - Whenever the Arch maintainer of nvidia-304xx updates it which is needed to build against the 4.9 code base. See my comments from 2016-12-22 19:02

Saren commented on 2017-01-07 18:45 (UTC)

Hi there, when will be 4.9 arriving? Thanks.

nTia89 commented on 2017-01-07 18:03 (UTC) (edited on 2017-01-07 18:03 (UTC) by nTia89)

missing a word in the enable* patch: ... It is not recommended to compile on Atom-CPUs with the 'native' option.[4] The recommendation is to *** the 'atom' option instead. ...

graysky commented on 2017-01-06 19:57 (UTC)

Bump to v4.8.16-1 Changelog: Commit:

kogone commented on 2016-12-28 22:13 (UTC)

this is a free for all

graysky commented on 2016-12-25 15:04 (UTC)

The PKGBUILDs in the AUR are free for all to modify.

neTpK commented on 2016-12-25 14:44 (UTC)

I have created my own package from this (forked basically) adding support for Reiser4 and with a config optimized solely for the Dell XPS 9550 laptops (in its various configurations). If its OK for you I'd like to name this Linux-CK-Reiser4 but with clear text that this is not made by you (in case shit hits the fan) but you will ofcourse gain the credit for the inital package the fork is based on. So Im basically using your PKGBUILD but with other options, the same patches but with a couple ones added, and grouped as ck-skylake to support your pre-compiled modules at repo-ck. Is this OK?

graysky commented on 2016-12-22 19:02 (UTC)

I updated linux-ck and all its daughter packages (broadcom-ck nvidia-x-ck, and virtualbox-ck) but currently nvidia 304xx doens't build without an update to 304.134 and currently nvidia-304xx-utils in [extra] is stuck on 304.132. This means I can't push the 4.9 update until our devs put 304.134 in the repos.

graysky commented on 2016-12-22 08:38 (UTC)

@BS - Yes, I saw it. Just need to verify that all modules build ok. Stay tuned.

BS86 commented on 2016-12-22 07:29 (UTC)

finally, 4.9 is out:

graysky commented on 2016-12-21 17:00 (UTC)

@Saren - I see... sorry, you'll need to edit the PKGBUILD. You can do it with a call to sed if you want.

Saren commented on 2016-12-21 16:07 (UTC)

@graysky - lol no, I want to enable NUMA.

graysky commented on 2016-12-21 13:58 (UTC)

@Saren - You shouldn't need to modify the PKGBUILD to disable NUMA... that has been defaulted for some time now.

Saren commented on 2016-12-21 13:30 (UTC)

Hi, since it's now possible to configure cpu march now, is it possible to add whether to disable numa as well so that multi cpu system users don't have to edit PKGBUILD every time linux-ck upgrades? Thanks.

mrkline commented on 2016-12-19 18:23 (UTC)

@Rainmaker: I've seen that if pacman is upgrading bash in another process. I would assume it's just a temporary problem - you'd have some really serious issues if /bin/sh went missing. :)

QuartzDragon commented on 2016-12-19 14:59 (UTC)

@Rainmaker Cannot reproduce here. That's a weird error I've never seen before... What else is happening on your machine while trying to compile?

Rainmaker commented on 2016-12-19 10:50 (UTC)

4.8.15-2 fails because it's trying to call /bin/sh during the build: Cyclomatic Complexity 2 sound/core/hwdep.c:snd_hwdep_proc_read Cyclomatic Complexity 1 sound/core/hwdep.c:alsa_hwdep_init Cyclomatic Complexity 1 sound/core/hwdep.c:alsa_hwdep_exit make[2]: /bin/sh: Command not found make[2]: *** [scripts/ sound/core/snd-hwdep.o] Error 127 make[1]: *** [scripts/ sound/core] Error 2 make: *** [Makefile:974: sound] Error 2 ==> ERROR: A failure occurred in build(). Aborting...

blitz commented on 2016-12-18 23:07 (UTC)

@graysky - Kernel config allows for menu driven cpu arch selection with minimal interaction and pre-selected MNATIVE. (taken from own PKGBUILD) _CPU_NATIVE=y ... ### Optionally set processor type for kernel if [ -n "${_CPU_NATIVE}" ]; then msg "Setting cpu arch to gcc native..." if [ "${CARCH}" = "x86_64" ]; then sed -i -e 's/^CONFIG_GENERIC_CPU=y/# CONFIG_GENERIC_CPU is not set/' \ -i -e 's/^# CONFIG_MNATIVE is not set/CONFIG_MNATIVE=y/' ./.config else sed -i -e 's/^CONFIG_M686=y/# CONFIG_M686 is not set/' \ -i -e 's/^# CONFIG_MNATIVE is not set/CONFIG_MNATIVE=y/' ./.config fi fi

graysky commented on 2016-12-18 19:38 (UTC)

@mrkline - Update the checksum (again without bumping the pkgver). No real easy way to automate the selection of native.

mrkline commented on 2016-12-18 19:29 (UTC)

@graysky - The menu is quite nice! Hopefully I can stop messing around in nconfig soon. On that note, is it possible to automate the selection? I tried adding a `CONFIG_MNATIVE=y` line to config.x86_64, but I was still prompted when I ran makepkg. Also, enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz needs its SHA updated.

graysky commented on 2016-12-18 17:36 (UTC)

OK, just updated without bumping the pkgver with this change. Feedback is welcomed.

graysky commented on 2016-12-18 13:42 (UTC)

@artafinde - see my comment from 2016-12-18 10:44 for the example.

artafinde commented on 2016-12-18 12:37 (UTC)

@graysky Can you provide a PKGBUILD (maybe through gist) to see how much interaction will be needed? I personally have to edit PKGBUILD anyway since I'm building with modprobed and BFS enabled by default (and using "-march=native")

nTia89 commented on 2016-12-18 11:43 (UTC) (edited on 2016-12-18 11:45 (UTC) by nTia89)

yes, missing of optimizations without -march=native is my worry! (I know, this seems false but it is true... hence, any modification is welcome, even if this mean to press some key my doubt is another: who builds this package from AUR, I imagine, wants the best optimizations overall, so, why "-march=native" is not the default choice?

zerophase commented on 2016-12-18 10:59 (UTC)

@graysky depends on how it's implemented. But, in general if a config file has new additions in a package bump it interrupts kernel compilation till I resolve the issue. I imagine whatever you implement will work with yaourt as well.

graysky commented on 2016-12-18 10:44 (UTC)

A modification I am considering is to have the AUR package provide unmodified config files for x86_64 and for i686 with respect to the gcc patch. What this means is that whenever you build linux-ck yourself, makepkg (or am I assuming your AUR helper) will pause and wait for you to select from the list of CPU optimizations (with default being a simple enter key). This will circumvent the need to modify the PKGBUILD enabled the nconfig option, and subsequent poking around in the nconfig itself. Potentially way faster and easier for the user. 1) How do people feel about this? 2) Will the added interactivity break AUR helpers (I am a pure makepkg person myself)? Example output: % makepkg -src ==> Making package: linux-ck 4.8.15-2 (Sun Dec 18 05:42:02 EST 2016) ... Processor family 1. AMD Opteron/Athlon64/Hammer/K8 (MK8) 2. AMD Opteron/Athlon64/Hammer/K8 with SSE3 (MK8SSE3) (NEW) 3. AMD 61xx/7x50/PhenomX3/X4/II/K10 (MK10) (NEW) 4. AMD Barcelona (MBARCELONA) (NEW) 5. AMD Bobcat (MBOBCAT) (NEW) 6. AMD Bulldozer (MBULLDOZER) (NEW) 7. AMD Piledriver (MPILEDRIVER) (NEW) 8. AMD Steamroller (MSTEAMROLLER) (NEW) 9. AMD Jaguar (MJAGUAR) (NEW) 10. Intel P4 / older Netburst based Xeon (MPSC) 11. Intel Atom (MATOM) 12. Intel Core 2 (MCORE2) 13. Intel Nehalem (MNEHALEM) (NEW) 14. Intel Westmere (MWESTMERE) (NEW) 15. Intel Silvermont (MSILVERMONT) (NEW) 16. Intel Sandy Bridge (MSANDYBRIDGE) (NEW) 17. Intel Ivy Bridge (MIVYBRIDGE) (NEW) 18. Intel Haswell (MHASWELL) (NEW) 19. Intel Broadwell (MBROADWELL) (NEW) 20. Intel Skylake (MSKYLAKE) (NEW) > 21. Generic-x86-64 (GENERIC_CPU) 22. Native optimizations autodetected by GCC (MNATIVE) (NEW) choice[1-22?]:

graysky commented on 2016-12-18 10:07 (UTC)

@SuperIce97 @mrkline - Good points about usability and about niche population with aged hardware. I reverted the commit removing native updating 4.8.15-2 without bumping the pkgver since I don't want to rebuild for [repo-ck] since this change has 0 impact to the repo packages.

QuartzDragon commented on 2016-12-18 06:20 (UTC)

@mrkline @SuperIce97 Also agreed! :)

mrkline commented on 2016-12-18 05:29 (UTC) (edited on 2016-12-18 05:35 (UTC) by mrkline)

+1 for bringing back `-march=native`. If you look at what flags are actually passed to the actual compiler (cc1), it provides much more machine-specific information (e.g., cache size, etc.) than `-march=<CPU family>`. As other users have suggested, it seems like there are workarounds for CPUs where it caused trouble. Even if workarounds don't exist, why remove the option from *all* users for the sake of a minority? A simple warning to avoid the flag for the few setups where it causes trouble should suffice.

SuperIce97 commented on 2016-12-17 21:02 (UTC)

Is there anything wrong with the version of your patch before the last commit that removed the native option? If you choose native, it automatically gives you the option for P6 NOPS but it's disabled by default. I think that's a pretty safe way to do it. The help message also has the pretty straightforward message of "Say Y if you have Intel CPU newer than Pentium Pro, N otherwise." at the end, so I believe that anyone who enables Native optimizations would know what to do there. For the issue with Atom CPUs having issues when set to native, you could just add "[causes issues on some Atom CPUs]" to the name of the native option. I don't think this needs to be too complex.

graysky commented on 2016-12-17 19:57 (UTC)

@SuperIce97 - I am not skilled enough to implement the logic for the X86_P6_NOP needed. While true that the native option may pass more tokens to make, if you look at the unpatched Makefile for a selection of core2, they are merely passing march=core2 option[1]. 1.

SuperIce97 commented on 2016-12-17 19:53 (UTC)

Also, I noticed that on my SandyBridge laptop, march=native enables a ton more instructions than march=sandybrdige (my CPU is a 2630QM which is a SandyBridge). That means that we are not getting the full potential by specifying an arch vs native arch

SuperIce97 commented on 2016-12-17 19:40 (UTC)

Looking at issue #17 on the gcc patch's github page though, I don't think it would be that difficult to enable native optimizations and just have a X86_P6_NOP as an additional option like tpruzina suggested. The reason I prefer using native instead of the specific arch is because some lower end CPUs (at least with Intel) don't seem to actually support every instruction of their arch for some reason. I used to have a Chromebook C720 with the Haswell Celeron 2955u (I now have a C740 with a Broadwell Celeron 3205u) and I was using the Haswell kernel until it decided that it wouldn't boot (not sure if a change in the kernel or gcc caused it; it was a long time ago and I was too lazy to hunt down the issue) if I compiled it with Haswell. With native on the other hand, the kernel worked just fine. Would it be possible to reimplement march=native but have the extra configurable as suggested by tpruzina?

graysky commented on 2016-12-17 18:57 (UTC) (edited on 2016-12-17 18:58 (UTC) by graysky)

@nTia89 - I removed it as handling dependencies for newer/older hardware would be too difficult[1]... just select the hardware option that corresponds to your CPU[2] and functionally, there is no difference. 1. 2.

nTia89 commented on 2016-12-17 17:24 (UTC) (edited on 2016-12-17 17:55 (UTC) by nTia89)

@graysky, how does new GCC patch affect "-march=native" users ?

graysky commented on 2016-12-17 16:20 (UTC)

Bump to v4.8.15-2 Changelog: Updated version of gcc patch Commit:

graysky commented on 2016-12-15 23:51 (UTC) (edited on 2016-12-15 23:55 (UTC) by graysky)

@Rainmaker - I like to use the official PKGBUILD as a template for this one since it minimizes both the chance of errors and my time to keep it in sync. Plus, I think many might not want to have the extra bulk of the headers package since it's only really needed to build modules. Not sure what you mean by split out the config. @BSB6 - It's the holidays people travel... seems to get this this pretty consistently this time of year.

QuartzDragon commented on 2016-12-15 22:51 (UTC)

@BS86 So what? Perhaps there's a good reason for it?

QuartzDragon commented on 2016-12-15 22:50 (UTC)

@Rainmaker I believe that the linux-ck PKGBUILD already does this?

BS86 commented on 2016-12-15 22:40 (UTC) (edited on 2016-12-15 22:40 (UTC) by BS86)

lol - arch still has only 4.8.13 in their official repositories, and no 4.9 - even Manjaro already has 4.8.14 and 4.9.0 in the stable repos -.-

Rainmaker commented on 2016-12-15 22:07 (UTC)

First of all, thank you for being very on top of all updates. If I may suggest 2 improvements: - You can build multiple packages in a single PKGBUILD. Because linux-ck and linux-ck-headers are likely to be installed together, may I suggest merging both packages to a single package. Building multiple packages in the same PKGBUILD is something that, for example, clion does. Effectively, it is setting pkgname= to a space separated list. - Could you split out the config? Each time I do a git pull, I need to open up PKGBUILD, and edit the parameters. By having a separate configuration file, I can setup git to ignore that file (with assume-unchanged), which saves me from having to set the parameters over and over.

graysky commented on 2016-12-15 18:57 (UTC)

Bump to v4.8.15-1 Changelog: Commit:

graysky commented on 2016-12-12 11:42 (UTC)

@metaphorex0 - Not out-of-date until 4.9.0 is written.[1] 1.

PotHix commented on 2016-12-11 17:49 (UTC)

@graysky thanks for your reply. Just saw the solution on @nlopez's post: gpg --recv-keys 79BE3E4300411886 38DBBDC86092693E

graysky commented on 2016-12-11 11:59 (UTC)

Bump to v4.8.14-2 Changelog: Bug patch fix following Arch upstream. Commit:

graysky commented on 2016-12-11 11:55 (UTC)

@PotFix - This is by default, see the official PKGBUILD. @mareex - That was due to a diff in the md5sums of the local/remote packages. I didn't catch the error. Fixed now, thanks.

mareex commented on 2016-12-11 10:04 (UTC)

@graysky What has happened with linux-ck-skylake from your repos? Maybe I am blind, but I cannot find it anymore (

PotHix commented on 2016-12-11 02:28 (UTC)

There is a problem with GPG signature verification ==> Validating source files with sha256sums... linux-4.8.tar.xz ... Passed linux-4.8.tar.sign ... Skipped patch-4.8.14.xz ... Passed patch-4.8.14.sign ... Skipped patch-4.8-ck8.xz ... Passed enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz ... Passed config.x86_64 ... Passed config ... Passed 99-linux.hook ... Passed linux.preset ... Passed change-default-console-loglevel.patch ... Passed net_handle_no_dst_on_skb_in_icmp6_send.patch ... Passed ==> Verifying source file signatures with gpg... linux-4.8.tar ... FAILED (error during signature verification) patch-4.8.14 ... FAILED (error during signature verification) ==> ERROR: One or more PGP signatures could not be verified!

graysky commented on 2016-12-10 18:51 (UTC)

Bump to v4.8.14-1 Changelog: Commit:

graysky commented on 2016-12-08 20:43 (UTC)

Bump to v4.8.13-2 Changelog: Add net_handle_no_dst_on_skb_in_icmp6_send.patch from Arch upstream Commit:

graysky commented on 2016-12-08 20:29 (UTC) (edited on 2016-12-08 20:30 (UTC) by graysky)

Bump to v4.8.13-1 Changelog: Commit:

SuperIce97 commented on 2016-12-08 20:06 (UTC)

@niceman The mainline kernel does include the msr module, as does basically every kernel in the main and AUR repos (including this one). If you enable nconfig for this kernel, you can see under "Processor Type and Features -> /dev/cpu/*/msr - Model-specific register support" that msr is indeed enabled as kernel module. Perhaps your device does not support msr? (I have an old Intel Atom machine that I use as a micro server that does not support msr and thus does not support i7z, which I would have liked to have been able to use)

niceman commented on 2016-12-08 19:17 (UTC)

hello, where can I find the msr module ? powertop seems to need it. The mainline kernel doesn't seem to include it too, is the module built in the kernel ?

graysky commented on 2016-12-06 23:02 (UTC)

Bump to v4.8.12-2 Changelog: Security patch fix following Arch upstream. Commit:

zerophase commented on 2016-12-02 23:53 (UTC)

@mlc thanks. I just ended up rm -fing the misc file, and reinstalling.

graysky commented on 2016-12-02 19:41 (UTC)

Bump to v4.8.12-1 Changelog: Commit:

commented on 2016-11-30 12:40 (UTC)

@zerophase: you can firstly remove linux-ck and linux-ck-headers $ sudo pacman -Rdd linux-ck linux-ck-headers Then delete /lib/modules/4.8.11-1-ck/kernel/drivers/usb/misc if this file still exists and reinstall theses two packages: $ sudo pacman -S linux-ck linux-ck-headers It should be alright.

zerophase commented on 2016-11-30 10:10 (UTC)

@mlc think it's safe to delete that file on a running kernel?

commented on 2016-11-30 05:46 (UTC)

@zerophase: no, it's a directory here.

zerophase commented on 2016-11-30 05:43 (UTC)

Is /lib/modules/4.8.11-1-ck/kernel/drivers/usb/misc installing as a file, rather than a directory for anyone else? I tried reinstalling ck, and it still looks corrupted.

graysky commented on 2016-11-26 09:44 (UTC)

Bump to v4.8.11-1 Changelog: Commit:

graysky commented on 2016-11-22 08:36 (UTC)

Bump to v4.8.10-2 Changelog: Bump to ck8 with MuQSS 0.144 Commit:

graysky commented on 2016-11-21 20:17 (UTC)

Bump to v4.8.10-1 Changelog: Commit: Note: CK will be releasing ck8 in several hours so I am holding off on building 4.8.10-1 for [repo-ck]. AUR users are of course free to do it.

graysky commented on 2016-11-20 14:34 (UTC)

Bump to v4.8.9-3 Changelog: Added a warning about initramfs rename for [repo-ck] users. Commit:

graysky commented on 2016-11-20 12:01 (UTC)

@snack - Yes, this naming scheme is the default now with upstream changes (hook and PKGBUILD). Since linux-ck-core2 from [repo-ck] has a pkgbase=linux-ck-core2, all users of [repo-ck] or of the AUR building a non-standard pkgbase will need to update their boot loader. See: If I was more savvy about this, I would have added a pacman warning to the post install, but even then, you only reach a subset of users.

snack commented on 2016-11-20 09:42 (UTC)

I fixed by removing all the old linux-ck files from /boot (keeping the linux-ck-core2 ones) and substituting "linux-ck" with "linux-ck-core2" everywhere in /boot/grub/grub.cfg

snack commented on 2016-11-20 09:31 (UTC) (edited on 2016-11-20 09:34 (UTC) by snack)

Update to linux-ck-core2 4.8.9-1/2 from repo broke my system. Boot fails because my grub configuration (which used to work up to 4.8.6) searches for /boot/vmlinuz-linux-ck, while with 4.8.9 I got /boot/vmlinuz-linux-ck-core2. Fixing it in the grub command line the boot fails again complaining that /lib/modules/4.8.9-2-ck-core2/modules.devname is missing, while the file itself is present. The boot media is an ext4 ssd. Should I update something in my system configuration or is the linux-ck-core2 package broken?

QuartzDragon commented on 2016-11-20 04:18 (UTC)

@graysky Now I understand better; I was just confused by the changes and how they worked. Thanks, graysky! :)

graysky commented on 2016-11-19 19:57 (UTC)

Bump to v4.8.9-2 Changelog: Fix checksums and bump for repo rebuild. Commit:

sidneycrestani commented on 2016-11-19 19:21 (UTC)

@marceliq yes, patch-4.8.9.xz sha256sum fails

marceliq commented on 2016-11-19 14:53 (UTC)

Hi, is it just me or sha256sums of patch-4.8.9.xz doesn't match for you either?

sir_lucjan commented on 2016-11-19 14:13 (UTC)

@graysky - you have absolutely right.

graysky commented on 2016-11-19 12:32 (UTC)

@fafryd - I don't think it's a problem since the PKGBULD renames the hook on the filesystem. See my comment in FS#51818 for more.

graysky commented on 2016-11-19 12:27 (UTC)

@fafryd - Yes, linux-ck is shipping 99-linux.hook since 4.8.8-2. I posted in the bug report you referenced asking if the hook needs a unique name. Right now linux-ck shares the name with linux which might be a problem. We can make the change if needed. Thank you for bring this to my attention.

fafryd commented on 2016-11-19 12:03 (UTC) (edited on 2016-11-19 12:24 (UTC) by fafryd)

>Every linux package (linux, linux-zen, linux-lts, linux-grsec, ... AUR packages >like linux-git, ...) has to ship its own hook. edit:

graysky commented on 2016-11-19 10:55 (UTC)

Bump to v4.8.9-1 Changelog: Commit:

graysky commented on 2016-11-19 10:55 (UTC)

@QD - No sure what you mean. You mention errors but provide no output and you mention references but provide no line numbers. Finally, you mention larger scope that PKGBUILD but provide no examples... I can't really act on anything you're reporting. If the references to the vanilla kernel are some variables in the PKGBUILD or linux.{preset,install} make sure you inspect them in a shell as many are substitutions that simply don't get substituted. For example, PKGBUILD line 262.

QuartzDragon commented on 2016-11-19 05:48 (UTC)

Hey graysky, I was looking at the PKGBUILD, due to build errors, and there are many references that only apply to the vanilla Arch Linux kernel. And it's not just the PKGBUILD that needs fixing... justing letting you know. Cheers, QD

graysky commented on 2016-11-17 23:46 (UTC)

Bump to v4.8.8-2 Changelog: Updates to PKGBUILD per Arch upstream. Commit:

graysky commented on 2016-11-17 23:36 (UTC) (edited on 2016-11-17 23:57 (UTC) by graysky)

Yep, already updated but was waiting for 4.8.9 to bump[1]. EDIT: here you go... better to hear about any bugs/errors before I do the repo rebuild so maybe this 24-36 h period will unearth some. 1.

SuperIce97 commented on 2016-11-17 22:15 (UTC)

The mainline kernel has some interesting changes with 4.8.8-2 and above. I'm not sure if these should be integrated into the ck kernel, but I would say it's worth a look (mainly some sed cleanups and using a new trigger to generate the initramfs as a post-transaction hook):

graysky commented on 2016-11-15 20:07 (UTC)

Bump to v4.8.8-1 Changelog: Commit:

rob77 commented on 2016-11-15 09:53 (UTC)

@QuartzDragon didn't think of that as I have git cloned this packages URL, I just git pull and re build the package. I did build the package so my comment was an FYI for anyone else facing same issue. But when next update comes will delete the file I have locally and re download.

QuartzDragon commented on 2016-11-15 07:59 (UTC) (edited on 2016-11-15 07:59 (UTC) by QuartzDragon)

@rob77 Try deleting the file so it can be redownloaded? I did that just in case.

rob77 commented on 2016-11-15 00:25 (UTC)

sha256 check failed on enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz had to amend line 74 of PKGBUILD to pass the error to cf0f984ebfbb8ca8ffee1a12fd791437064b9ebe0712d6f813fd5681d4840791. Alternatively double check the sha256sum by running sha256sum enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz

graysky commented on 2016-11-12 10:36 (UTC)

Bump to v4.8.7-2 Changelog: Bump to ck7 with MuQSS 0.140 Commit:

QuartzDragon commented on 2016-11-12 01:28 (UTC) :)

graysky commented on 2016-11-11 20:52 (UTC) (edited on 2016-11-11 20:53 (UTC) by graysky)

Bump to v4.8.7-1 Changelog: Commit:

commented on 2016-11-10 10:44 (UTC)

@QuartzDragon & artafinde: => "In addition to the most up to date version of MuQSS replacing BFS, this is the first release with BFQ included. It is configurable and is set by default in -ck though it is entirely optional".

QuartzDragon commented on 2016-11-10 09:27 (UTC) (edited on 2016-11-10 09:29 (UTC) by QuartzDragon)

@artafinde I suspect so: I applaud Con for adding BFQ to the CK patchset! :D Though, I wonder why he did it silently? Or was there an announcement somewhere that we missed?

artafinde commented on 2016-11-09 23:50 (UTC)

@graysky: after the BFQ patches are excluded. But the scheduler is still available. Is this now included with -ck patch or merged on main kernel?

graysky commented on 2016-11-07 20:08 (UTC)

@Tahu - Please report to CK's blog if you haven't already.

Tahu363 commented on 2016-11-07 14:43 (UTC)

Latest changes break wireless from working after suspend. Full reboot required to gain wireless functionality. ifconfig up/down has no effect.

graysky commented on 2016-11-05 19:10 (UTC)

Thanks updated without bumping the pkgver since purely cosmetic.

glitsj16 commented on 2016-11-05 19:05 (UTC)

Suggestions for your most recent PKGBUILD: --- a/PKGBUILD +++ b/PKGBUILD @@ -118,7 +118,7 @@ # this as the default choice for it leads to more throughput and power # savings as well. # - # + # sed -i -e 's/^CONFIG_HZ_300=y/# CONFIG_HZ_300 is not set/' \ -i -e 's/^# CONFIG_HZ_100 is not set/CONFIG_HZ_100=y/' \ -i -e 's/^CONFIG_HZ=300/CONFIG_HZ=100/' .config @@ -149,7 +149,7 @@ # modprobe configs zcat /proc/config.gz > ./.config else - warning "You kernel was not compiled with IKCONFIG_PROC!" + warning "Your kernel was not compiled with IKCONFIG_PROC!" warning "You cannot read the current config!" warning "Aborting!" exit

WFV commented on 2016-11-05 00:55 (UTC) (edited on 2016-11-05 18:24 (UTC) by WFV)

4.8.5-3 and 4.8.6-1 ck-piledriver, Virtualbox guests sluggish to the point unable to use them (and WinXP guest blue-screens @ launch). Problem isn't present in stock kernel 4.8.6-1, all guests function normal. (Asus M5A88-M, FX8350 AM3+). No other host problems noticed in either ck rev. EDIT: problem persists in 4.8.6-2 ck-piledriver. Does it have to do with MuQSS and CGROUPS shared real time (for virtualbox)?

graysky commented on 2016-11-04 23:04 (UTC)

Bump to v4.8.6-2 Changelog: Bump to ck6 with MuQSS 0.135 Commit:

zerophase commented on 2016-11-04 10:34 (UTC)

Definitely noticed MuQSS does a much better job of utilizing cpu resources than BFS. The only side effect I've noticed is if compiling large projects (Unreal Engine) The desktop has a tendency to become unresponsive.

graysky commented on 2016-11-01 17:12 (UTC)

@nTia - This sort of report is not very useful without a comparison back to the ARCH kernel of the same version at a minimum. As always, make sure that whatever you're experiencing is reproducible and not present on the non-ck kernel of the same version and report to CK's blog.

nTia89 commented on 2016-11-01 16:54 (UTC)

just a warning, on my system (dell xps 13, broadwell) latest version (4.8.6-1) is unstable boot and shutdown slowed down, a bunch of errors about audio ASoC, etc...

graysky commented on 2016-10-31 18:44 (UTC) (edited on 2016-11-01 11:15 (UTC) by graysky)

Bump to v4.8.6-1 Changelog: Notes: This release also includes two of CK's pending patches per his recommendation as a ck6 is not on the nearterm horizon. Commit:

BS86 commented on 2016-10-30 14:28 (UTC)

no issues with disabled NUMA here on my Piledriver FX 8350 AM3+ CPU

SuperIce97 commented on 2016-10-30 04:11 (UTC)

@zerophase No, full dynticks is not recommended. I miscommented earlier. Also, I was referring to CPU accounting, not CPU ticks.

zerophase commented on 2016-10-30 04:04 (UTC)

@SuperIce97 Is Full dynticks still recommended?

graysky commented on 2016-10-29 15:53 (UTC)

@zosodk - I can mirror if needed but it's up for me at this point in time...

zosodk69 commented on 2016-10-29 14:05 (UTC)

Is there an alternative download source for patch-4.8-ck5.xz? There appear to be issues with right now.

graysky commented on 2016-10-29 11:58 (UTC)

@Tjuh - Please post to CK's blog

Tjuh commented on 2016-10-29 11:51 (UTC)

I'm getting this msg in journallog with the latest version: BFQ WARNING:last 4611686022722355494 budget 13835058059577131310 jiffies 4294967598 diff 4611686018427387904

Saren commented on 2016-10-29 11:15 (UTC) (edited on 2016-10-29 11:15 (UTC) by Saren) is down for everyone ATM

graysky commented on 2016-10-29 09:02 (UTC)

Bump to v4.8.5-3 Changelog: 100Hz is default by CK now so hardset this in PKGBUILD Commit:

graysky commented on 2016-10-29 09:02 (UTC)

@Saren - Correct: dual socket users will need to build from the AUR since NUMA is disabled by default (recommended by CK). @QuartzD - Yes, it would be... I currently build 20 different package sets for [repo-ck] and toggling this on/off would double that to 40 and only serve to confuse users. After all, what percentage of [repo-ck] users have a dual socket motherboard? @SuperIce and @QuartzD - Yes, corrected in 4.8.5-3.

QuartzDragon commented on 2016-10-29 08:13 (UTC)

Would it be a hassle to maintain NUMA and non-NUMA versions in Repo-CK?

Saren commented on 2016-10-29 06:08 (UTC)

With NUMA disabled by default, as a dual socket workstation user, does it mean I have to recompile kernel every time when upgrading it?

QuartzDragon commented on 2016-10-29 05:04 (UTC) (edited on 2016-10-29 05:05 (UTC) by QuartzDragon)

It's being set to 100 Hz, it's just that the "_1K_HZ_ticks" variable is confusing. Should be renamed to "_100_HZ_ticks", now, to reflect the changes. Just guessing you missed it somehow, graysky?

SuperIce97 commented on 2016-10-29 04:46 (UTC) (edited on 2016-10-29 04:52 (UTC) by SuperIce97)

Can you set the ticks to 100Hz as recommended by ck now?

graysky commented on 2016-10-29 01:19 (UTC)

@vishwin -

vishwin commented on 2016-10-29 00:28 (UTC)

ck in his -ck4 announcement post said that he's cutting -ck5 and MuQSS 120 today. They have just been tagged on his github and I bet he's prepping the tarballs as I type…

mrkline commented on 2016-10-29 00:18 (UTC)

I'm late to the party, but I've run with NUMA disabled since I started using this package (4+ months?). Zero issues on all my machines (Skylake, Haswell, and Ivy Bridge).

graysky commented on 2016-10-28 18:08 (UTC)

Bump to v4.8.5-1 Changelog: Commit:

graysky commented on 2016-10-27 19:15 (UTC)

Thanks for the comments, all.

Saren commented on 2016-10-27 13:58 (UTC)

Dual haswell xeon cpu. 4.8.4-4, NUMA and MuQSS 116 enabled. 0 issue so far and great performance.

artafinde commented on 2016-10-27 07:04 (UTC) (edited on 2016-10-27 07:04 (UTC) by artafinde)

I was one of the persons who had issues with NUMA on an AMD FX 8100. About 6 months ago I changed to i7 6700 and disabled NUMA. No issues so far. I did had btrfs corruption with linux-ck early MuQSS versions (v110 I think) so I was a late adopter but now running pretty stable for days. I think the NUMA issue were related to AM3+ socket.

commented on 2016-10-27 04:28 (UTC)

I'm running linux-ck with NUMA disabled since about two weeks. I haven't yet experienced any problems with my Skylake CPU.

graysky commented on 2016-10-26 21:16 (UTC) (edited on 2016-10-27 19:14 (UTC) by graysky)

@QD - It's a small increase using a make endpoint in my experience (see the flyspray I note in the PKGBUILD comments). It could be that other endpoints reveal more substantial gains as well. In any case, NUMA is really for servers with multiple sockets; it has no point on a single socket motherboard is my understanding. Linux-ck had it disabled for a long time until some combination of upstream/BFS + NUMA disabled was believed to be responsible for problems which is when I disabled the code to disable it. Now that MuQSS has replaced BFS, this may not be the case any more. Several users have posted to the AUR reporting stability with it disabled as is my experience as well.

QuartzDragon commented on 2016-10-26 20:50 (UTC)

How much of a speed increase does disabling NUMA actually give, anyway? I've never really felt the difference.

graysky commented on 2016-10-26 19:17 (UTC)

Greg just tagged 4.8.5-rc1 and has it scheduled to release on Friday/noon UTC[1]. We can give disabling NUMA a whirl in 4.8.5-1 if there are no reports linking this setting to bad behavior with MuQSS (I am not experiencing anything bad on my Haswell). 1.

metaphorex0 commented on 2016-10-26 10:45 (UTC)

I've been running linux-ck with NUMA disabled since you added that option back to PKGBUILD. I've had 0 issues so far.

graysky commented on 2016-10-25 15:49 (UTC)

I can but I would like to get some additional feedback from others who build with NUMA disabled since it caused problems in the past. I too have it disabled on my workstation without ill effects. Can other AUR users please comment. If you have NUMA disabled, are you running a stable 4.8.4-4?

mareex commented on 2016-10-25 15:45 (UTC)

hi graysky, would you mind to disable NUMA in your repo builds? Never had any problems on four machines with NUMA disabled.

graysky commented on 2016-10-25 10:28 (UTC)

@mono - Please report upstream:

monotykamary commented on 2016-10-25 06:42 (UTC)

No more complete freezes, but it now slows to a crawl on the same application.

graysky commented on 2016-10-24 19:33 (UTC)

Bump to v4.8.4-4 Changelog: Update with ck4/MuQSS v0.116. Updates to [repo-ck] are building now. Commit:

eduardoeae commented on 2016-10-24 10:11 (UTC)

linux-4.8-ck4, MuQSS CPU scheduler v0.116

graysky commented on 2016-10-23 16:53 (UTC)

@mono - the difference between AUR and [repo-ck] right now is that AUR uses 4.8.4-3 which has ck3 (MuQSS) whereas [repo-ck] uses 4.8.4-1 with ck1 (bfs). I have a dual core system that black screened on booting with ck3 last I checked but there are two patches in 'pending' that I need to try first. I think if they fix that problem, I will go ahead and build 4.8.4-3 for [repo-ck] but I won't be able to test until tomorrow.

monotykamary commented on 2016-10-23 16:12 (UTC) (edited on 2016-10-23 16:14 (UTC) by monotykamary)

There's a bug that completely freezes my computer when opening or playing OpenGL games for a period of time (with NVIDIA), from the linux-ck-ivybridge package group. The AUR linux-ck package doesn't have that bug, but now specifically crashes osu! on wine everytime upon load (or a few seconds after load). I haven't seen it crash on any other native or wine application so far. --- WINEARCH=win32 WINEPREFIX=~/.wineprefixes/.osu winetricks -q dotnet46 cjkfonts gdiplus wininet winhttp osu! folder size: 3.9G Installed Packages: linux-ck, linux-ck-headers, nvidia-ck, broadcom-wl-ck BFQ and MuQSS enabled; bug also occurs with I/O scheduler CFQ. CPU: Intel Core i5-3317U CPU @ 1.7GHz GPU: GeForce GT 640M LE

kogone commented on 2016-10-23 00:34 (UTC)

everything is been stable on ivybridge. have built every update so far! hard to keep up =P

graysky commented on 2016-10-22 23:59 (UTC) (edited on 2016-10-23 00:00 (UTC) by graysky)

Bump to v4.8.4-3 Changelog: Added two key pending patches from CK. Commit: Notes: I suspect CK will ready a ck4 patchset and continue to hold on pushing MuQSS to [repo-ck] a little longer. AUR users are encouraged to test and communicate via CK's blog results as MuQSS matures.

graysky commented on 2016-10-22 19:51 (UTC)

Bump to v4.8.4-2 Changelog: Fix EXTRAVERSION variable. Commit: Notes: I would like to get some feedback from AUR users around the stability with the MuQSS before I go building [repo-ck] packages with 4.8-ck3. Please test and report back good or bad.

graysky commented on 2016-10-22 19:48 (UTC) (edited on 2016-10-22 19:50 (UTC) by graysky)

@blitz - Thanks for the reminder... accidentally deleted that code.

blitz commented on 2016-10-22 19:11 (UTC)

linux-4.8 ck patchset introduced new versioning schema: patch-4.8-ck3 diff --git a/Makefile b/Makefile +CKVERSION = -ck3 +CKNAME = MuQSS Powered +EXTRAVERSION := $(EXTRAVERSION)$(CKVERSION) From kernel version '4.8' patch '4' extraversion '1' this schema translates into 4.8.4-1-ck3-ck To keep both ck3 and ck is extraneous and not nesessary, imho. # set extraversion to pkgrel # remove ckX version msg "Setting kernel extra version" sed -ri -e "s|^(EXTRAVERSION =).*|\1 -${pkgrel}|" \ -e "s|^(EXTRAVERSION :=).*|\1 -${pkgrel}|" Makefile

Tjuh commented on 2016-10-22 18:02 (UTC)

I dunno yet, keep having mini system freezes when playing certain games on steam with the 4.8 series. No issues whatsoever with 4.7.8.

FarStar commented on 2016-10-22 13:39 (UTC)

Everything is behaving fine with linux-ck 4.8.4 and MuQSS v0.115 (like v0.114). My CPU is an i7-6700K.

nTia89 commented on 2016-10-22 12:40 (UTC)

if with the first release of the 4.8 series, now (4.8.3-3) my system (Dell XPS 13, Broadwell i5-5200U) boots fine and works as in the past days

BS86 commented on 2016-10-22 12:39 (UTC)

No issues here with MuQSS on my piledriver. on 4.8.4-1 I am trying out some advanced amdgpu package, which checks for a specific config. Would it be much effort to include it, so that this check does not fail? Kernel 4.8.4-1-ck3-ck not supported CONFIG_DRM_AMDGPU_CIK is missing

graysky commented on 2016-10-22 11:24 (UTC)

Bump to v4.8.4-1 Changelog: Commit: Notes: I would like to get some feedback from AUR users around the stability with the MuQSS before I go building [repo-ck] packages with 4.8-ck3. Please test and report back good or bad.

commented on 2016-10-22 08:56 (UTC)

Hello graysky, I use linux-ck with BFQ and MuQSS enabled since v106. It works very fine with my Skylake CPU. Performances and responsiveness are great with v114/115 :) Thank you for all!

graysky commented on 2016-10-22 07:01 (UTC)

Bump to v4.8.3-3 Changelog: Included 4.8-ck3 which uses MuQSS v0.115. Commit: Notes: I would like to get some feedback from AUR users around the stability with the MuQSS before I go building [repo-ck] packages with 4.8-ck3. Please test and report back good or bad.

vishwin commented on 2016-10-22 03:21 (UTC) (edited on 2016-10-22 03:21 (UTC) by vishwin)

-ck3 just released, includes MuQSS v0.115

graysky commented on 2016-10-21 19:22 (UTC)

Bump to v4.8.3-2 Changelog: Included 4.8-ck2 which uses MuQSS v0.114. Commit: Notes: I would like to get some feedback from AUR users around the stability with the MuQSS before I go building [repo-ck] packages with 4.8-ck2. Please test and report back good or bad.

graysky commented on 2016-10-21 13:21 (UTC)

Please post these details to ck's blog so he can assist.

xevrem commented on 2016-10-21 12:17 (UTC)

I'm experiencing a kernel panic shortly after boot. Currently I am using the linux-ck-piledriver kernel. from journalctl -k -b -1: Oct 21 07:03:49 vanir kernel: AMD-Vi: Completion-Wait loop timed out Oct 21 07:03:49 vanir kernel: AMD-Vi: Event logged [IOTLB_INV_TIMEOUT device=07:00.0 address=0x000000040cfa2380] Oct 21 07:03:49 vanir kernel: AMD-Vi: Completion-Wait loop timed out in this log its complaining about device 07:00.0, but its complained about other devices as well that work perfectly fine in other kernels... I would do a dump of those dmesg logs if i could, but it appears journalctl didnt capture anything from those boot cycles... only that one and it never catchs the kernel stack trace dump >_< here is info about my CPU from /proc/cpuinfo if that helps: processor : 0 vendor_id : AuthenticAMD cpu family : 21 model : 101 model name : AMD FX-9800P RADEON R7, 12 COMPUTE CORES 4C+8G stepping : 1 microcode : 0x6006113 cpu MHz : 1400.000 cache size : 1024 KB physical id : 0 siblings : 4 core id : 0 cpu cores : 2 apicid : 16 initial apicid : 0 fpu : yes fpu_exception : yes cpuid level : 13 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good acc_power nopl nonstop_tsc extd_apicid aperfmperf eagerfpu pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs xop skinit wdt lwp fma4 tce nodeid_msr tbm topoext perfctr_core perfctr_nb bpext ptsc mwaitx cpb hw_pstate vmmcall fsgsbase bmi1 avx2 smep bmi2 xsaveopt arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic overflow_recov bugs : fxsave_leak sysret_ss_attrs null_seg bogomips : 5392.44 TLB size : 1536 4K pages clflush size : 64 cache_alignment : 64 address sizes : 48 bits physical, 48 bits virtual power management: ts ttp tm 100mhzsteps hwpstate cpb eff_freq_ro acc_power [13] It should be said, i do not experience these issues with stock kernel nor the zen kernel. I've also tried the generic linux-ck kernel and experience similar issues... Any suggestions or additional info i can grab for you that would help, just let me know :)

Taijian commented on 2016-10-21 11:46 (UTC) (edited on 2016-10-21 14:26 (UTC) by Taijian)

I have a question regarding some modules I am missing when I try to compile the kernel locally. These modules also fail to load with the repo-ck 4.8.3 kernel. The 4.7.6 kernel from repo-ck still had them. Specifically, I have a Dell Inspiron laptop, so I am using i8kutils to get my temp-sensors and my fan to work. According to modprobed-db, this is pulling in the following modules: - dell_laptop - dell_smbios - dell_smm_hwmon - dell_wmi However, when trying to build the kernel, these modules are missing/nowhere to be found. Can anybody help point me in the right direction, where I can pull these in from?

ooo commented on 2016-10-21 10:56 (UTC)

@graysky, Fair enough. Altough The pending/ patches for bfs512 aren't really 'development' patches, but fixes to bugs that were discovered after bfs512/ck1 release. My understanding is that ck simply didn't want to make a new release, but still recommends adding the fixes. Since some people seem to be having issues with bfs512, it would make sense to me to apply those. I have no problems myself, but I guess anyone having issues with linux-ck-4.8* should try if adding the pending bfs512 patches helps.

nTia89 commented on 2016-10-21 10:09 (UTC)

@SuperIce97 I posted here just to advice other users...

graysky commented on 2016-10-21 09:41 (UTC)

@ooo - I don't want to include pending patches without CK specifically asking since this is not a development package. You are free to modify the PKGBUILD on your own of course. @eduardoeae - I saw the release but am a bit reluctant to bump and push to [repo-ck] since a few reports to the blog you linked indicate issues still. Perhaps the AUR can have it but I will hold off on publishing to the repo initially.

eduardoeae commented on 2016-10-21 01:33 (UTC)

-ck2 is out. I'm not flagging as out of date because MuQSS is now the default. Probably would want to wait a bit before jumping into it.

ooo commented on 2016-10-20 22:55 (UTC)

How about including the pending patches for bfs512? They contain bugfixes, and are recommended by ck.

rob77 commented on 2016-10-20 21:48 (UTC)

linux-ck 4.8.3-1 working fine here, with MuQSS enabled, BFQ enabled and compiled with intel haswell optimizations.

graysky commented on 2016-10-20 20:26 (UTC)

Please report issues with the patchset to CK via his blog. I am just the packager :p

initnull commented on 2016-10-20 19:34 (UTC) (edited on 2016-10-20 19:51 (UTC) by initnull)

I get a Black screen on boot with repo-ck/linux-ck-skylake 4.8.2-1. Setup is Kaby Lake processor with /boot and /home crypted stock 4.8.2-1-ARCH works fine.

graysky commented on 2016-10-20 18:50 (UTC)

Bump to v4.8.3-1 Changelog: Commit:

SuperIce97 commented on 2016-10-20 17:10 (UTC)

Everybody who is having problems: Have you tried running the stock 4.8.2 kernel from the main Arch repos? If that has problems as well, there's nothing that we can do. Also, if you are running a 4.7 kernel, please update immediately to 4.7.9. A large security vulnerability was found and fixed. The exploit using this vulnerability has been found in the wild (and the bug has existed for 11 years). For people running 4.8.2, I expect @graysky to bump the version soon. Please update immediately when he does.

Saren commented on 2016-10-20 15:33 (UTC) (edited on 2016-10-21 07:56 (UTC) by Saren)

Getting boot freeze on my haswell-ep system as well since 4.8.2-1 with muqss on, no problem if no muqss.

nTia89 commented on 2016-10-20 14:01 (UTC)

I've just updated to ck-4.8.2 and my system (Dell XPS 13, broadwell 9343) stucks on boot (during the systemd boot process...) the error is related to "i2c designware..." going back to latest 4.7 NB: no extra options enabled from the AUR package, just this boot parameters: elevator=bfq and nmi_watchdog=0

bacondropped commented on 2016-10-20 13:02 (UTC)

Jesus, @graysky, you're rolling this package out faster than I update my system. Immense props for that.

rob77 commented on 2016-10-20 09:59 (UTC)

@Tjuh It's an error message introduced in Linux 2.6.35 and up. Basically it happens when your system doesn't support APIC. It's a harmless error, but doesn't look all that nice. You can make it go away by adding noapic to the end of your grub boot line. Adding it from Tazpanel: 1. Boot--> Boot Loader 2. Click on View or edit menu.lst 3. Click Edit 4. Add noapic to the end of the kernel line so it looks something like this: kernel /boot/vmlinuz-2.6.37-slitaz root=/dev/hda1 quiet noapic 5. Click Save and you're done. The error should be gone the next time you boot Not upgraded kernel yet but will check to see if that fixes it this evening

Tjuh commented on 2016-10-20 09:27 (UTC)

Getting this weird error -ERROR: Unable to locate IOAPIC for GSI 37- in my journallogs with MuqSS, 1k_HZ_ticks, NUMAdisable, localmodcfg and BFQ_enable enabled. No such error or any other issues with the same settings enabled on the previous version.

zerophase commented on 2016-10-20 08:56 (UTC)

@VerruckteFuchs I'm on Haswell as well. I've been up for close to 8 hours without issue, compiling projects and running services that stress the cpu. I don't run mpd, though. Have you noticed any issues in terms of cpu load and the crash?

snack commented on 2016-10-20 07:38 (UTC)

It seems that virtualbox-ck-host-modules-core2 (and possibly other virtualbox-ck-host-modules packages) in repo have been built with a wrong dependency from linux-ck-core2 version since pacman says: $ sudo pacman -Su :: Starting full system upgrade... resolving dependencies... looking for conflicting packages... error: failed to prepare transaction (could not satisfy dependencies) :: virtualbox-ck-host-modules-core2: installing linux-ck-core2 (4.8.2-1) breaks dependency 'linux-ck-core2<4.8'

VerruckteFuchs commented on 2016-10-20 01:37 (UTC)

Trying to run mpd results in an instant freeze/lockup of my Haswell machine. I do have a previous playlist queue from when it was last open, if that info helps at all.

graysky commented on 2016-10-19 20:30 (UTC) (edited on 2016-10-19 20:30 (UTC) by graysky)

Bump to v4.8.2-1 Changelog: bfs512/ck1, upstream release 4.8.2, and optional (AUR only) MuQSS 0.112 Commit: Linux changes:

graysky commented on 2016-10-16 16:53 (UTC)

Bump to v4.7.8-1 Changelog: Commit:

graysky commented on 2016-10-15 14:32 (UTC)

@Saren - Once CK incorporates MuQSS into the ck1 patchset, yes. Right now it is considered in development, so AUR users can build it if they wish, but I don't want to server-up experimental code in [repo-ck].

Saren commented on 2016-10-15 14:23 (UTC)

@graysky Will packages on repo-ck adopt MuQSS updates as well?

graysky commented on 2016-10-13 22:07 (UTC)

4.7.7-1 has been updated with MuQSS 111.

graysky commented on 2016-10-11 18:56 (UTC) (edited on 2016-10-13 22:07 (UTC) by graysky)

4.7.7-1 has been updated with MuQSS 110.

graysky commented on 2016-10-09 16:09 (UTC)

@agm28011997 - CK has stuff in his pending queue beyond MuQSS108 but I don't want to add continually add these since the rate of development of MuQSS is as rapid as it has been. Once MuQSS finalized and replaces BFS, this package will be here applying it.

Saren commented on 2016-10-08 07:23 (UTC)

Noticeable performance loss... compared to 4.7.1-ck since BFS 8...

pedrogabriel commented on 2016-10-08 02:14 (UTC)

It's working nicely with or withouth the MuQSS, I'm actually not using MuQSS because it someway brokes the fan driver of my notebook (with or withouth NUMA), but exceptly that, it's working nicely.

graysky commented on 2016-10-07 17:35 (UTC)

Bump to v4.7.7-1 Changelog: Commit:

graysky commented on 2016-10-07 17:34 (UTC)

@Roberth - Not out of date until BFQ is ready for 4.8

graysky commented on 2016-10-07 15:31 (UTC)

Good to hear, pedro... this is with MuQSS disabled, correct? I would like to hear from more users before I enable it globally again (and in the repo).

pedrogabriel commented on 2016-10-07 15:27 (UTC)

I'm using it with NUMA disabled, no problems so far. (Intel Haswell chip)

zerophase commented on 2016-10-07 05:07 (UTC)

@graysky those nvme patches are listed on the archwiki for fixing an issue with powersaving with nvme disks. I'm just trying to figure out if I should add those patches to the pkgbuild when I pick up an nvme drive in a couple weeks. I'll at least test to see if there are any incompatibility issues. If they work without issue, should they get added into the ck repo till reaching mainline?

graysky commented on 2016-10-07 00:07 (UTC)

@jwhickman - Dunno about NUMA. Have to ask some users to test and verify that it is or is not a problem. Anyone willing to test? Also, I don't think MuQSS is stable at this point. We'll probably have BFS until the 4.9 release if I read CK's blog correctly. In either case, the linux-ck brand is here to stay; MuQSS will still have the ck1 brand as I understand it.

jwhickman commented on 2016-10-06 12:52 (UTC) (edited on 2016-10-06 12:58 (UTC) by jwhickman)

Thanks @graysky! I modified this to build MQS for k4.8; no problems besides having to comment out the BFQ patches due to a compile failure, not unsurprisingly. I was hoping the k4.7 BFQ patches would work on k4.8, since the patches applied without issue. :) Question for you; is the 'NUMA disable issue' still relevant, as it's been around a long time, plus now this is MQS. Also, I noticed in my k4.8 build, the config showed a new tick rate of 250 Hz, so was also curious about that. /======================================================= # Running with a 1000 HZ tick rate (vs the ARCH 300) is reported to solve the issues with the bfs/linux 3.1x and suspend. [...] _1k_HZ_ticks=y ### Do not disable NUMA until CK figures out why doing so causes panics for some users! [...] #_NUMAdisable= =======================================================/ In commented original discussion on the tick rate, ck has said, "(16 September 2013 at 18:55) If it still needs 1000Hz to work correctly, then there is still a bug." Anyway, thanks again!

graysky commented on 2016-10-06 09:05 (UTC)

OK... unofficial patches. I was wondering if something was in the merge queue on the topic that I didn't see.

zerophase commented on 2016-10-06 09:03 (UTC)

@graysky These patches

graysky commented on 2016-10-06 08:58 (UTC)

What patches are you referring to?

zerophase commented on 2016-10-06 03:15 (UTC)

Anyone know if the Linux-nvme kernel patches have any issues with ck? Going to be picking up an nvme drive in a couple week and just want to check if anyone knows of any issues.

graysky commented on 2016-10-05 18:09 (UTC)

@Pedrogabriel - Yes, if you enable the build option. I just updated 4.7.6-1 to pull in MuQSS v106 (disabled by default).

pedrogabriel commented on 2016-10-05 15:45 (UTC)

Does MuQSS 106 apply to 4.7?

graysky commented on 2016-10-04 22:12 (UTC)

105 is out but it's jacked up and does not cleanly apply to either 4.7 or 4.8. See CK's latest blog. I have unflagged; please flag again when the 105 is updated.

pedrogabriel commented on 2016-10-04 22:00 (UTC)

MuQSS 105 is out now.

pedrogabriel commented on 2016-10-04 13:27 (UTC)

The kernel don't recognize my fans after compile it with MuQSS.

graysky commented on 2016-10-03 18:36 (UTC) (edited on 2016-10-03 20:08 (UTC) by graysky)

@vishwin - Sorry, but not out of date until BFQ for 4.8 is released. I do not see MuQSS104 that patches against the 4.7 code-base (although CK is working on 105 as we speak)... plus, 4.8-ARCH is probably still in the works as it hasn't been build nor hit [testing] just yet.

artafinde commented on 2016-10-03 13:45 (UTC) (edited on 2016-10-03 13:48 (UTC) by artafinde)

I actually won't be enabling this unless it turns out to be stable enough and give something to the <= 8 CPUs. I've enabled it on my laptop and screwed up the BTRFS ctree. Seems like I'm reformatting today. I can't really blame MuQSS since there's a big WARNING on top, but it's a warning to users here. Could be something else but all started after I compiled and booted the MuQSS enabled kernel. The combination was: CPU Haswell, BTRFS single metadata SSD 1 device, BFQ enabled by default, localmodules enabled and MuQSS.

kogone commented on 2016-10-03 13:40 (UTC)

maybe we should create a rubric of tests to run to see the difference of muqss running on cpus with <= 8 cores?

artafinde commented on 2016-10-02 09:26 (UTC)

According to the blog post CK mentions only CPU not cores. I'm not entirely sure how the runqueues are defined (per CPU or per Core) but my understanding from the post is to target servers and make the BFS more comparable to mainline CFS.

zerophase commented on 2016-10-02 09:05 (UTC)

@artafinde Does MuQSS benefit more than 1 cpu die, or multiple cores? Say 8+?

artafinde commented on 2016-10-02 08:58 (UTC)

MuQSS is really targeting multi CPUs. As per the comment from CK "it will make a difference in 16+ CPUs and with high loads". I suppose we can still use it for testing purposes but unlikely we will produce any lockups to help surface any bugs with the desktop/laptop PCs. 103 is out btw.

graysky commented on 2016-10-01 20:05 (UTC)

@SuperIce - Thanks updated to 102 still in 4.7.6-1.

SuperIce97 commented on 2016-10-01 17:14 (UTC)

@graysky Version 102 is out now as well. According to ck: "I've uploaded version 102 which has a a lot of bugfixes and makes it boot properly again for me."

graysky commented on 2016-10-01 15:39 (UTC)

@SuperIce97 - Thanks for the heads-up about version 101 of muqss. Updated without bumping the pkgver since so few people are actually building and since the repo packages do not use this and I don't want to trigger a rebuild.

graysky commented on 2016-10-01 09:39 (UTC)

Bump to v4.7.6-1 Changelog: Commit: Note: This release allows AUR users to optionally build CK's MuQSS (Multiple Queue Skiplist Scheduler)[1]. See the commented lines in the PKGBUILD. 1.

Saren commented on 2016-10-01 04:38 (UTC)

No freeze observed so far since v4.7.5-1.

graysky commented on 2016-09-24 09:45 (UTC) (edited on 2016-09-24 09:50 (UTC) by graysky)

Bump to v4.7.5-1 Changelog: Commit:

Rainmaker commented on 2016-09-23 22:40 (UTC)

@graysky. You're right. makepkg only builds once. makepkg -C DOES build twice. But that behavior has nothing to do with this package specifically. Thanks.

graysky commented on 2016-09-23 08:28 (UTC)

Bump to v4.7.4-2 Changelog: Update to new BFS[1] Commit: 1.

zerophase commented on 2016-09-19 23:01 (UTC)

@graysky Is the extra patch in the pending directory getting applied? I guess that's supposed to fix the problem with Haswell.

Saren commented on 2016-09-19 16:27 (UTC)

@zerophase That's why I am staying on 4.7.1-1.

zerophase commented on 2016-09-18 23:56 (UTC)

Definitely, getting those crashes on Haswell E as well. The only thing I noticed in the logs was: ENERGY_PERF_BIAS: Set to 'normal', was 'performance' Could cpupower not be setting the cpu governor correctly? I also get that same message while running the stock kernel.

graysky commented on 2016-09-18 13:12 (UTC)

Makepkg does not build twice, it packages linux-ck then linux-ck-headers. Are you sure you are not mischaracterizing?

Rainmaker commented on 2016-09-18 12:51 (UTC)

Well, I'm using makepkg. I have a local directory on which I do a git pull every time there is an update. But, I am doing makepkg -C. Maybe I should try cleaning up the previous source manually before building. Currently, it DOES do two builds. It finishes building linux-ck, and then proceeds with unpacking linux-ck-headers and compiling that. It does warn about the PKGBUILD being a split package. When using yaourt, it does built twice, but I guess that is to be expected.

graysky commented on 2016-09-18 12:45 (UTC)

@Saren - Thanks for clarify. @Rainmaker - As suggested by others, don't use an aur helper. Just build with makepkg and you'll be fine (ie one build, multiple packages). Which one are you using by the way? The fact that is build twice is just wrong.

francoism90 commented on 2016-09-18 10:29 (UTC)

@Saran: How do you build this package? By an AUR wrapper (e.g. pacaur/yaourt)? If so, please just download/unpack the tar and run makepkg instead. I don't have any issues with multiple PKG builds.

Saren commented on 2016-09-18 07:33 (UTC) (edited on 2016-09-18 07:35 (UTC) by Saren)

@graysky I think you got it wrong. What @Rainmaker said is when we build linux-ck and linux-ck-headers packages, the kernel will be compiled twice, which is unnecessary and waste of time and power. When we build linux-ck-headers, linux-ck will be also compiled and provided. We can just accept installing both packages instead of linux-ck-headers only when pacman prompt us.

graysky commented on 2016-09-16 18:44 (UTC)

The headers package has nothing to do with it; edit /etc/mkinitcpio.d/linux-ck.preset and disable the fallback image.

Saren commented on 2016-09-16 13:42 (UTC)

@Rainmaker Install linux-ck-headers only.

Rainmaker commented on 2016-09-16 12:01 (UTC)

When you install linux-ck and linux-ck-headers, the kernel gets built twice. Is there a way to only build it once?

graysky commented on 2016-09-15 22:16 (UTC)

Disable ccache....

artafinde commented on 2016-09-15 22:07 (UTC)

@rko same thing to me with ccache had to clean. Seems like it's a pattern now (build failed due to object fixdep-in.o again) :(

jschorr commented on 2016-09-15 20:12 (UTC)

4.7.3-5 (linux-ck-headers-4.7.3-5 w/ linux-ck-piledriver-4.7.3-5) fail to boot for me- it just freezes partway through so I had to go back to 4.7.3-4. I have an AMD 9590 CPU.

graysky commented on 2016-09-15 19:05 (UTC)

Bump to v4.7.4-1 Changelog: Commit:

boteium commented on 2016-09-15 17:48 (UTC)

Both my baytrail z3775 (silvermont) and E3 1265l v3 (haswell) freeze on 4.7.3.

artafinde commented on 2016-09-15 12:54 (UTC)

I just want to point out that on my Skylake i7 6700 I had no panics for some time now and none with 4.7.x Could be only related to older hardware?

usuariopolivalen commented on 2016-09-15 12:21 (UTC)

please post problems here

kogone commented on 2016-09-15 11:36 (UTC)

freeze happening with my Silvermont on 4.7.3-5 but no problems with my Ivybridge.

Saren commented on 2016-09-15 11:26 (UTC) (edited on 2016-09-15 14:27 (UTC) by Saren)

I compiled linux-ck 4.7.3-5 from AUR last night with native optimizations as set in nconfig. I still got kernel panic when I was playing games inside VM. I am using 4.7.1-1 for now and see if crashes still occur. EDIT: Crashes seem to no longer occur at least for now after 3 hours consecutive VM gaming.

zerophase commented on 2016-09-15 08:49 (UTC)

@agm28011997 Odd, I'm not getting any freezes with Haswell-e that I believe are caused by the kernel. I just freeze when I have CLion, and a Windows vm up. (I give that up to changing the CLion vm options for parsing a large CMake file) I'm pretty sure I didn't use up all of my ram or swap. Maybe, it is the kernel.

evilpot commented on 2016-09-15 08:06 (UTC)

its crashing all over the place...

usuariopolivalen commented on 2016-09-14 17:04 (UTC)

incredible man, I don't know what is it the difference or what have the new code but in my i5 4690 haswell with intel hd graphics from git all is perfect with one hour of gaming now and any freeze, it is something weird because it is supposed to be here yet the freezes

rko commented on 2016-09-14 16:55 (UTC)

Random Comment: I was getting build failures with this, and it turns out that I needed to clear my ccache to get the kernel to build.

graysky commented on 2016-09-14 13:14 (UTC)

Even though 4.7.4 is due out today, I have built and published 4.7.3-5 and all related packages to [repo-ck]. CONFIG_CGROUP_SCHED is enabled per CK's suggestion.

usuariopolivalen commented on 2016-09-13 22:55 (UTC)

@gravysky thanks for your work, both of you, I wish I could help more than testing packages, when you put the new version in the repo I will test for report the issues, but I think that the freezes will continue due to the other comment in the page.. @saren what is BTW? I have seen the journalctl with the hour of freezes and nothing.. the freeze make my computer to do nothing and repeat 2 last seconds of music more or less and the mosue to not move and the keyboard the same

graysky commented on 2016-09-13 19:40 (UTC)

Bump to v4.7.3-5 Changelog: update to ck4/bfs497 and reenable CONFIG_CGROUP_SCHED. Commit: Note: I am waiting to hear back from CK before I build the repo packages since he may suggest disabling CONFIG_CGROUP_SCHED. I like to think that the Arch community plays a large role in helping CK to refine bfs so AUR users please continue provide feedback to ck's blog as usual:

Saren commented on 2016-09-13 19:38 (UTC) (edited on 2016-09-13 19:40 (UTC) by Saren)

@agm28011997 Haswell-EP is the xeon version of Haswell cpus, linux-ck-haswell should be 100% compatible with it. I am downgrading once again from 4.7.2-2 to 4.7.2-1 because of your comment. :/ BTW, can confirm the lockup is kernel panic.

usuariopolivalen commented on 2016-09-13 19:30 (UTC)

@saren what is haswell-EP? my freezes starts at 4.7.2-2 with the bfs 472 patch, and other thing, post your problem in the ck blogspot of con kolivas @th3voic3 other like you it is me, if you want you can downgrade the version to 4.7.2-1 with patch bfs472 that is very stable for me @gravysky try to use cgroups enable if you want because for haswell desktop users there are too loockups for use this kernel, we'll wait until a solution

Saren commented on 2016-09-13 19:24 (UTC)

Haswell-EP user here, using linux-ck-haswell. Getting 3 random lockups already since 4.7.3. 2 of the lockups in the GPU Passthrough gaming, and the remaining one happened while watching youtube videos. I am downgrading to 4.7.2 as suggested by the comments and see if the issue will be going away.

th3voic3 commented on 2016-09-13 19:09 (UTC)

I'm using a Haswell cpu and a qemu vm with PCI passthrough. With the recent kernels my entire host would lockup as soon as I shutdown my VM. I just compiled the kernel with the new patch version and enabled CONFIG_CGROUP_SCHED. Just had a lockup again. Although I had lockups with or without CONFIG_CGROUP_SCHED. Haven't tested this patch yet with the option disabled. For now I'm using the mainline kernel. No issues there.

graysky commented on 2016-09-13 19:04 (UTC)

Just waiting to hear back from CK regarding his recommendation to or not to disable CONFIG_CGROUP_SCHED with ck4/bfs497. Will update once this happens.

puithove commented on 2016-09-13 18:50 (UTC)

I'm on a SandyBridge machine and using the SandyBridge package from the repo. I ended up having the same complete system freezes as well as KP on shutdown/reboot even after updating to 4.7.3-4 As of yesterday evening I rolled back to 4.7.2-1 and I seem to be stable again.

usuariopolivalen commented on 2016-09-12 12:44 (UTC)

CONFIG_CGROUP_PIDS=y CONFIG_CGROUP_FREEZER=y what means this lines with cgroups disabled??

usuariopolivalen commented on 2016-09-12 11:52 (UTC)

@gravysky after a few tests I have yet freeze problems while gaming browsing etc.. something wird because is a problem that affects haswell desktops only .. based on read forums and ck blog, I tried to use generic package of linux ck in the repotoo, but it has freezes too

pnehem commented on 2016-09-11 22:52 (UTC)

Thanks @graysky, running great so far!

usuariopolivalen commented on 2016-09-11 19:43 (UTC)

thanks @gravysky

graysky commented on 2016-09-11 14:47 (UTC)

Bump to v4.7.3-4 Changelog: Disable CONFIG_CGROUP_SCHED per CK's suggestion. Commit:

bacondropped commented on 2016-09-10 08:02 (UTC)

@QuartzDragon @pnehem @graysky I've recompiled the kernel without CONFIG_CGROUP_SCHED as well (GCC 6.2.1), and for me, it DOES seem to solve the persistent kernel oops. I'll try to test it more than just poking it with a stick, meanwhile, you can use this config patch (I know it's like sixty lines with context, shut up):

QuartzDragon commented on 2016-09-10 00:43 (UTC)

I've disabled CONFIG_CGROUP_SCHED and am rebuilding to see if the problem persists.

bacondropped commented on 2016-09-09 23:28 (UTC) (edited on 2016-09-10 00:22 (UTC) by bacondropped)

@pnehem It seems to happen around the same time to me as well, specifically during Xorg launch sequence (I'm running KDE on Manjaro, so to me, it happens during/after the Plymouth splash). Another person reported an oops with a stack trace that looks almost like mine (sans maybe two entries in the middle): Does your stack trace look like the one in that gist? I'm currently rebuilding the kernel with GCC 5.4.0 instead of 6.2.1 to eliminate the toolchain, I'll let you know how that goes. UPDATE: @graysky Do you think disabling cgroups might be worth considering? CK writes they're unstable and are known to cause problems/panics: UPDATE: rebuilding with GCC 5.4.0 does NOT fix the problem (reminder, here's the problem I'm trying to fix, bug report not mine, but very similar stack trace:

pnehem commented on 2016-09-09 20:34 (UTC) (edited on 2016-09-09 23:25 (UTC) by pnehem)

Hello, I'm getting a kernel panic as well, but I was trying to figure out if it where it was happening at because I can get to my login screen, just fine. But when I go to log in then it kernel panics and locks up the computer. I was just getting ready to start reading logs. --update: I updated my system but still getting a seg fault/kernel panic. I get to the login/sddm screen just fine, But the I try to go farther the screen goes black and I can see the cursor and nothing responds locks up the whole system sort of -I can do ctl-alt-F2 and login from the "command line" but I can't make heads or tails of the logs.

usuariopolivalen commented on 2016-09-09 17:51 (UTC)

@gravisky con kollivas has commented that you use the patch of cgroups but he told too that it is unstable yet, I don't think the problem is this but can someone that have my problem try to recompile the kernel without this patch for seeing if the problem is fixed? other thing the people that have my problem has c states in bios enabled? it is an idea..

usuariopolivalen commented on 2016-09-09 14:19 (UTC) (edited on 2016-09-09 14:19 (UTC) by usuariopolivalen)

@gravysky what changes have you put in the kernel ck betwin the 4.7.2-1 and 4.7.2-2 with bfs 480? I revert to older kernel and I have no problems with freezes with cpufreq and pstate, but I am the only person that appear to have this problem, cgroups patch is enabled in the repo?

usuariopolivalen commented on 2016-09-09 14:11 (UTC)

@QuartzDragon kernel opps? What do you mean?

usuariopolivalen commented on 2016-09-09 12:57 (UTC)

I don't know why but I changed the p_state to cpufreq and the freezes continues here.., I used the pstate powersave and the powersave cpufreq and nothint both of them and I think the rest too give me random freezes, for ram? for memory? for cache? I don't know..

QuartzDragon commented on 2016-09-09 10:38 (UTC)

I consistently get kernel oops with BFS 490 ck3 whenever I start my X server. I know it's not BFQ, because it also happens with CFQ. I'll post on his blog, anyways.

bacondropped commented on 2016-09-09 03:21 (UTC) (edited on 2016-09-09 12:16 (UTC) by bacondropped)

@agm28011997 @graysky Can confirm a kernel oops on v4.7.3-3. Update 2016-09-09 12:14+000: @QuartzDragon, yep, happens on the X server start to me as well.

usuariopolivalen commented on 2016-09-09 02:16 (UTC)

I continues having problems with that kernel, I dont know if it is a problem with p_state driver, someone have this problems too???? it is supposed to be fixed.. but for me .. not

graysky commented on 2016-09-08 20:09 (UTC)

Bump to v4.7.3-3 Changelog: Fix config.x86_64 (my mistake), otherwise unchanged from 4.7.3-2. Commit:

graysky commented on 2016-09-08 20:08 (UTC)

@rko - Fuck! You are correct. Fixed in 4.7.3-3. Thank you for the observation.

rko commented on 2016-09-08 20:01 (UTC)

I think the config.x86_64 got copied over with config (which is i386)? Looking at the diffs, this does not seem correct and is specifying 32 bit version? -CONFIG_64BIT=y -CONFIG_X86_64=y +# CONFIG_64BIT is not set +CONFIG_X86_32=y

graysky commented on 2016-09-08 19:59 (UTC)

Is anyone else experiencing a kernel panic upon booting into 4.7.3-2 or is it just me?

graysky commented on 2016-09-08 19:02 (UTC) (edited on 2016-09-08 19:03 (UTC) by graysky)

Bump to v4.7.3-2 Changelog: update bfq v8r2-->v8r3 and incorporate Arch upstream changes to kernel config Commit:

sflor commented on 2016-09-08 08:09 (UTC) (edited on 2016-09-08 16:12 (UTC) by sflor)

My laptop's keyboard doesn't work with 4.7.3, but did with 4.7.2. I can't even enter my password.

usuariopolivalen commented on 2016-09-08 00:53 (UTC)

@zerophase I test the latest kernel and for the moment I only had one freeze since I update.. I am not sure if the problem is the memory, but I don't think so with 8 GB and swap of 6GB the only thing is that a I have the tmp files mount in ram but.. not sure.. before, the pc not freezes..

zerophase commented on 2016-09-08 00:04 (UTC)

@agm28011997 I just run out of memory if I'm doing things with Unreal. It's because I don't have enough swap space. (going to fix that once I get an NVME ssd, and just use my old ssd for swap) I set my cpu to performance mode, so I don't encounter issues with frequencies changing.

kogone commented on 2016-09-07 19:46 (UTC)

with this bump from 4.7.2 to 4.7.3, make oldconfig asked whether to enable CGROUP_SCHED. should I do it?

usuariopolivalen commented on 2016-09-07 19:44 (UTC)

thabks very much graysky, I'll wait until the repo update:)

graysky commented on 2016-09-07 19:31 (UTC)

Bump to v4.7.3-1 Changelog: bfs490/ck3[1] and upstream release 4.7.3[2] Commit: 1. 2.

usuariopolivalen commented on 2016-09-07 18:00 (UTC)

i dont know zerophase, if you use intel new cpu you use p-state driver and for me causes freezes.. and for out of memory the kernel not freezes too..

eduardoeae commented on 2016-09-07 12:22 (UTC)

BFS 490 (-ck3) is also out.

graysky commented on 2016-09-04 18:54 (UTC)

You guys really want to be reporting this on ck's blog if you aren't doing it already:

zerophase commented on 2016-09-04 18:27 (UTC)

Which driver is p state? I'm getting freezes, but thought it was from running out of memory when running virtual machines.

usuariopolivalen commented on 2016-09-04 18:07 (UTC)

someone with haswell desktop p-state driver and the lastest kernel of ck with freeze problems? i am getting freezes all time with that kernel, the older for me don't make me this and in other laptop with amd cpu it didn't do this too

graysky commented on 2016-09-02 17:18 (UTC)

Bump to v4.7.2-2 Changelog: bfs480 and ck2, see: Commit:

graysky commented on 2016-08-29 18:55 (UTC) (edited on 2016-08-29 19:00 (UTC) by graysky)

@blitz - Thanks for that info, see: According the wikipedia[1], these aren't really out there yet for testing purposes. As well, the AMD stuff isn't out until 2017 but still good to keep watch. 1.

blitz commented on 2016-08-29 08:18 (UTC)

Regarding enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz. GCC 6 Series target specific improvements for Intel Skylake -march=skylake-avx512 and AMD Zen (family 17h) processors -march=znver1 and -mtune=znver1 is now available. :: New Targets and Target Specific Improvements

kyak commented on 2016-08-21 06:24 (UTC)

@Distorted That would be great, unless i'm the only one using bbswitch-ck and it is not worth effort.

Distorted commented on 2016-08-20 19:46 (UTC)

@kyak I removed bbswitch-ck from the AUR because of dkms, do you want me to put it back?

kyak commented on 2016-08-20 17:42 (UTC)

It looks like bbswitch-ck is gone from AUR. I don't want to use bbswitch-dkms, are there any other options?

graysky commented on 2016-08-20 16:57 (UTC)

Bump to v4.7.2-1 LinuxChanges: Commit:

graysky commented on 2016-08-19 07:51 (UTC)

@asuh - CK gave the go-ahead so I will include it in 4.7.2-1 once it is available.

asuh commented on 2016-08-17 21:27 (UTC)


graysky commented on 2016-08-17 19:10 (UTC)

If that's true (the lack of support for SMT_NICE) you will need to patch the source prior to building or email CK to ask him to include it in the ck1 patchset as I think your hardware is in the minority of users. I don't want to just apply a patch that without CK's blessing particularly when the code doesn't help 99% of the users :)

enihcam commented on 2016-08-17 01:25 (UTC)

@graysky thanks for pointing it out. Are you suggesting the panic can only be workarounded by setting CONFIG_SMT_NICE=y given that AMD processors (not zen) do not support SMT actually?

graysky commented on 2016-08-16 23:48 (UTC)

Well, they are certainly in the sources: % grep CONFIG_SMT config.x86_64 CONFIG_SMT_NICE=y % grep CONFIG_SMT config CONFIG_SMT_NICE=y And in the [repo-ck] packages but I can't control what you do to the PKGBUILD on your own machine :p % uname -r 4.7.1-1-ck % zgrep CONFIG_SMT /proc/config.gz CONFIG_SMT_NICE=y

enihcam commented on 2016-08-16 23:48 (UTC)

By the way, why would I need to set SMT on an AMD processor?

enihcam commented on 2016-08-16 23:45 (UTC)

@graysky, I didn't find CONFIG_SMT_NICE in /proc/config.gz. Are there any dependencies I need to turn on first?

graysky commented on 2016-08-16 23:31 (UTC)

@enihcam - Regarding configs without CONFIG_SMT_NICE set? Linux-ck and the Arch kernel both have this set. I don't want to add a conditional in the PKGBUILD that inspects $srcdir/linux-$pkgver/.config and patches if so for example.

enihcam commented on 2016-08-16 23:08 (UTC)

@graysky Please refer to

graysky commented on 2016-08-16 19:36 (UTC) (edited on 2016-08-16 19:36 (UTC) by graysky)

Bump to v4.7.0-1 LinuxChanges: Commit:

graysky commented on 2016-08-16 13:31 (UTC)

Is this in CK's pending set? Can you get him to articulate a timeline to implement into the next bfs? If the code is not tested enough for example, and if he feels more time and testing is needed, I do not want to push unstable code to the aur and by analogy to the 4000+ users of [repo-ck].

enihcam commented on 2016-08-16 06:19 (UTC)

@SuperIce97 @graysky The issue exists in bfs code. The following patch fixes it. --- linux-4.7.orig/kernel/sched/bfs.c +++ linux-4.7/kernel/sched/bfs.c @@ -1418,6 +1418,9 @@ static void try_preempt(struct task_stru * a different CPU set. This means waking tasks are * treated differently to rescheduling tasks. */ +#ifndef CONFIG_SMT_NICE + cpu = cpu_of(highest_prio_rq); +#endif set_task_cpu(p, cpu); resched_curr(highest_prio_rq); }

SuperIce97 commented on 2016-08-15 23:50 (UTC)

@enihcam I haven't had any issues with it, and 4.7 is what the mainline Arch kernel is at right now. Do you have issues with the default arch kernel now? If you do, report it to the arch bug tracker so they can try to get it fixed

Distorted commented on 2016-08-15 11:03 (UTC)

@enihcam the aur is git based, just checkout an older version

enihcam commented on 2016-08-15 11:00 (UTC)

Hi Graysky, could you please fallback this package to v4.6.6? I think 4.7 is not stable for now, because two of my boxes failed to boot. One freezes at logon screen, another panics. Thanks a lot!

enihcam commented on 2016-08-15 09:37 (UTC)

"In addition to the resync, a few minor changes have made their way into this release with respect to the way tasks preempt other tasks. See bfs470-updates.patch for details." --<<BFS 472, linux-4.7-ck1>>

ryant0000 commented on 2016-08-12 16:50 (UTC)

Made it here as a warning. Anyways, recent update resolved issue.

graysky commented on 2016-08-11 21:26 (UTC)

Bump to v4.7.0-1 LinuxChanges: Commit:

graysky commented on 2016-08-11 09:58 (UTC)

The AUR is probably not the best place for your report... Try communicating with Palolo, the lead developer.

ryant0000 commented on 2016-08-10 22:14 (UTC)

My machine kernel panicked after updating to 4.6.6-1. I do use BFQ for my ext4 root drive. I did not kernel panic on previous versions.

graysky commented on 2016-08-10 17:39 (UTC)

Bump to v4.6.6-1 Changelog: Commit:

Distorted commented on 2016-08-10 14:48 (UTC)

Does anyone have issues with the new BFQ scheduler? The first v8 did some damage so i want to be careful

graysky commented on 2016-08-08 22:07 (UTC)

Bump to v4.6.5-4 Changelog: v8r2 of BFQ Commit:

graysky commented on 2016-08-05 22:02 (UTC)

Bump to v4.6.5-3 Changelog: v8r1 of BFQ Commit:

Distorted commented on 2016-08-05 20:39 (UTC)

Paolo updated the patch to bfq 4.6.0-v8r1, which causes the current url to be incorrect as wel.

graysky commented on 2016-08-02 17:36 (UTC)

@fafik1234 - Not out of date yet, see[1]. 1.

graysky commented on 2016-07-31 08:10 (UTC)

OK... I recommend using an alternative IO scheduler until Paolo resolves the issue.

glavin commented on 2016-07-31 08:06 (UTC)

yes getting kernel panic with bfq enabled on ext4 filesystem, in my case.

graysky commented on 2016-07-31 07:59 (UTC)

For those getting panics, please verify that it is because you are using BFQ on a device.

Saren commented on 2016-07-31 07:43 (UTC)

Getting kernel panic on Haswell-E as well, I thought it was btrfs's problem, done btrfs check and fixed nothing, then I found the kernel panic message is about bfq and got here.

glavin commented on 2016-07-31 06:59 (UTC)

Getting kernel panic on sandybridge as well, freeze on moderate I/O and sometimes get panic messages on shutdown.

zosodk69 commented on 2016-07-30 23:01 (UTC)

The panics are almost certainly BFQv8 related. Reverting to BFQv7r11 makes them go away. I was able to get good panic output by logging to the serial port. I'm not a linux-ck user, rather a standalone BFQ user. I've started a thread on paolo's newsgroup to work through this issue:!topic/bfq-iosched/80V4U6ak57w Because of these stability issues, I'd recommend linux-ck revert to BFQv7r11.

zerophase commented on 2016-07-30 22:49 (UTC) (edited on 2016-07-30 22:59 (UTC) by zerophase)

@graysky I'm definitely getting kernel panics after the system has been up for a few hours. the numlock key doesn't even respond, once the system locks up. Running the standard kernel for now to see if it's still an issue.

graysky commented on 2016-07-29 19:06 (UTC)

@rucci - That is the default for the PKGBUILD except for _BFQ_enable_=y (and NUMA code has been commented for ages now). The only diff you have is enabling BFQ by default but that shouldn't trigger a bug as you activated it through other means before.

mssdvd commented on 2016-07-29 16:28 (UTC) (edited on 2016-07-29 16:29 (UTC) by mssdvd)

I recompiled again with "_1k_HZ_ticks=y", "_BFQ_enable_=y", "_NUMAdisable=y", gcc patch and now it works without kernel panic...

zerophase commented on 2016-07-29 15:17 (UTC)

@graysky I haven't had any panics yet. Running the kernel compiled for Haswell, with '_1k_hz_ticks' and bfq enabled.

Cake commented on 2016-07-29 14:12 (UTC)

@graysky Yes, I only installed one disk (/dev/sda) and and enabled BFQ on it. Several hours ago, I checked it with cake-pc# cat /sys/block/sda/queue/scheduler noop deadline cfq [bfq] I am sorry that I did not try recompiling `linux-ck=4.6.5` with `_1k_HZ_ticks=`. I just changed `grub.cfg` and rebooted with the stock kernel ... Everything seems to be working fine ... I am wondering what is going on when the system freezes. So if you require any other information, please let me know.

graysky commented on 2016-07-29 13:19 (UTC)

So two reports of the new bfq causing panics? Can you both verify that you have a drive set to use bfq? I don't see how tick rate plays into the equation.

mssdvd commented on 2016-07-29 09:20 (UTC)

It works if I disable "_1k_HZ_ticks" option

Cake commented on 2016-07-29 08:39 (UTC)

@rucci I have the same problem (Gnome BFQ Ivybridge) after updating to 4.6.5-2. I have to restart by pressing the reset button on my machine when the system freezes...

mssdvd commented on 2016-07-29 08:28 (UTC)

Kernel panic after update 4.6.5-1 -> 4.6.5-2 (KDE BFQ IvyBridge)

mradermaxlol commented on 2016-07-29 05:35 (UTC)

@graysky if you wanna know, everything's just fine after the latest update - stuff is buildable, at least for me. I can't understand why -ck patch was ruining the build, though; maybe I'm just dumb.

graysky commented on 2016-07-28 18:58 (UTC)

3.6.5-2 has been updated the corrected name of the 4th patch; the checksums remain the same since upstream did not change the content, only the name.

mradermaxlol commented on 2016-07-28 15:27 (UTC) (edited on 2016-07-28 15:48 (UTC) by mradermaxlol)

I'm getting patching errors (likely) during prepare() execution which result in makepkg build failure. It's happening at -ck patchset applying, and I have completely no idea what's causing it - maybe these patches are outdated, maybe it's just 4.6.5-incompatible (although I don't think so as errors happen in bfq.c/bfq.h files, for example). Right now my best guess is that the fourth BFQ patch is causing that, because that makes sense - it might be okay a while ago, but now its name in PKGBUILD is useless as it's wrong; I've changed it as @herraiz suggested and, well, everything's okay except patching X) Does anybody else experience this issue as well? EDIT: After a little research I've found out (I hope so) that the thing responsible for that problem is... Yup, the -ck patchset itself. Ta-daa!

herraiz commented on 2016-07-28 14:23 (UTC) (edited on 2016-07-28 14:23 (UTC) by herraiz)

@nTia89: the third patch has changed, you have probably an old copy of the third patch in your sources dir.

nTia89 commented on 2016-07-28 13:16 (UTC)

other than corrected forth name's patch, in order to have a working pkgbuild, we need to correct the checksum for the third patch

graysky commented on 2016-07-28 09:25 (UTC)

I will update the source array for the renamed patch but this is technically not out of date since the content of that patch did not change. See my post a few down (we knew this change was coming).

herraiz commented on 2016-07-28 08:25 (UTC)

I think there is a typo in one of the URLs. I am getting the following error: curl: (22) The requested URL returned error: 404 Not Found ==> ERROR: Failure while downloading The correct URL is (blkck -> block)

FadeMind commented on 2016-07-28 02:54 (UTC)

Hi @graysky!topic/bfq-iosched/wAKntB66OHs BFQ changelog. Regards FadeMind

cryzed commented on 2016-07-28 01:03 (UTC) (edited on 2016-07-28 01:44 (UTC) by cryzed)

@grasky, yes: pacaur. Should I run makepkg manually? I want to note that I have been using pacaur previously to install and update linux-ck without any problems. EDIT: Running it manually right now using "makepkg -fsic". EDIT 2: Well that worked. That's the first time an AUR helper bit me, especially pacaur which is supposed to be very reliable ( Sorry about this.

graysky commented on 2016-07-28 00:56 (UTC)

Are you using an aur helper?

cryzed commented on 2016-07-28 00:36 (UTC) (edited on 2016-07-28 00:41 (UTC) by cryzed)

When updating to the latest version, I get the following error after the seemingly successful compilation process: > ==> Finished making: linux-ck 4.6.5-2 (Do 28. Jul 02:33:40 CEST 2016) > ==> Cleaning up... > :: Installing linux-ck,linux-ck-headers package(s)... > :: linux-ck,linux-ck-headers package(s) failed to install. Check .SRCINFO for mismatching data with PKGBUILD. The .SRCINFO looks like this: I did the following modifications to the PKGBUILD: > _1k_HZ_ticks= > _BFQ_enable_=y I don't use the sleep/suspend feature, hence the first change. My /etc/makepkg.conf has the following modifications: > CFLAGS="-march=native -O2 -pipe -fstack-protector-strong" > CXXFLAGS="-march=native -O2 -pipe -fstack-protector-strong" > MAKEFLAGS="-j$(nproc)"

graysky commented on 2016-07-27 21:16 (UTC)

@blitz - Yes, I assumed the filename will change, and I'll have to update it :p If you catch it before someone else posts here, please email me privately so I can push the change and minimize impact of users building from the AUR. Thanks.

blitz commented on 2016-07-27 21:04 (UTC)

@graysky 2016-07-27 21:01 That was fast, while /me wrote mail message to Paolo Valente with regard to broken filenaming sequence, -blkck- vs -block-

graysky commented on 2016-07-27 21:01 (UTC) (edited on 2016-07-27 21:01 (UTC) by graysky)

Bump to v4.6.5-2 Changelog: 4.6.5-1 neglected to apply one of the BFQ patches; 4.6.5-2 fixes it (thanks to Lucjan Lucjanov for pointing this out). Commit:

graysky commented on 2016-07-27 18:21 (UTC)

Bump to v4.6.5-1 Changelog: Commit:

graysky commented on 2016-07-11 21:57 (UTC)

Bump to v4.6.4-1 Changelog: Commit:

graysky commented on 2016-06-25 15:33 (UTC)

Bump to v4.6.3-1 Changelog: Commit: LinuxChanges: Paolo has assured me that BFQ v7 is fine for the 4.6.x series of kernels; the bug that is affecting it applies equally to 4.5.x kernel users so there is no harm in including it against the 4.6.x series. He is testing prev8 of BFQ now but would not predict a release date. Enjoy the 4.6 series.

artafinde commented on 2016-06-20 19:52 (UTC)

That's fine and I agree - just mentioned my findings (though if I knew about the g-groups I'd already know that :D ). Probably the wait will be worth it with the new BFQ v8 [!topic/bfq-iosched/ReUuq4jtZ84]

graysky commented on 2016-06-20 19:32 (UTC)

Paolo has a testing version of BFQ for 4.6 but it and the 4.5 series have problems with hierarchical support[1]. We have 4,000+ [repo-ck] users so I would rather wait fo the ink to dry on the BFQ 4.6-final version before including old or testing patches. 1.!topic/bfq-iosched/7gJMK1DttYQ

artafinde commented on 2016-06-20 07:55 (UTC)

Interestingly the patches of BFQ 4.5 apply successfully on 4.6.2. Although I haven't test those as I don't use BFQ.

graysky commented on 2016-06-08 21:03 (UTC)

@dlh - Not of of date... see the comments from 2016-06-08

graysky commented on 2016-06-08 14:50 (UTC)

@PieGuy128 - Not out of date... see the comments from 2016-06-08

graysky commented on 2016-06-08 11:17 (UTC)

Bump to v4.5.7-1 Changelog: Commit:

graysky commented on 2016-06-08 11:05 (UTC)

@Fusion - Yes, but we still need to wait for Paolo to release a BFQ for 4.6:

fusion809 commented on 2016-06-08 03:13 (UTC)

4.6 is out

graysky commented on 2016-06-01 21:18 (UTC)

Bump to v4.5.6-1 Changelog: Commit:

graysky commented on 2016-05-26 19:06 (UTC)

@Tjuh - Yes, that is likely your problem. I highly recommend using modprobed-db if you want to make localmodconfig for this purpose. Just know that it's only as good as your database and if you want to have a new module that you haven't used before, you will have to boot into a kernel that actually has this module available for modprobed-db to learn that you have an interest. Hope that makes sense. See the wiki for more.

Distorted commented on 2016-05-26 12:46 (UTC)

@Tjuh Disable localmodcfg, it causes the kernel to only compile the modules you are currently using.

Tjuh commented on 2016-05-26 11:07 (UTC)

Ye I also find it weird, cause I checked the config.x86_64 and saw that it is enabled as a module. Hmm maybe I should disable the localmodprobe option in the PKGBUILD. And yes rebooted and recompiled many times.

graysky commented on 2016-05-25 20:56 (UTC)

@Tjuh - Dunno why, it's in the config and it loads on my system running 4.5.5-ck-1 % zgrep -i zram /proc/config.gz CONFIG_ZRAM=m CONFIG_ZRAM_LZ4_COMPRESS=y Have you perhaps not rebooted since updating to 4.5.5?

Tjuh commented on 2016-05-25 20:21 (UTC)

I'm trying to get zram working follwing the wiki, but the module seems to be missing; tjuh ~ $ modprobe zram modprobe: FATAL: Module zram not found in directory /lib/modules/4.5.5-1-ck Any suggestions?

graysky commented on 2016-05-19 06:52 (UTC)

Bump to v4.5.5-1 Changelog: Commit:

francoism90 commented on 2016-05-17 14:47 (UTC)

@sir_lucjan: OK, thanks.. hope it will be available soon. :)

sir_lucjan commented on 2016-05-17 13:08 (UTC)

@francoism90: Nope.

francoism90 commented on 2016-05-17 13:07 (UTC)

Hi graysky, any news about Linux 4.6? Is a patch available? Thanks :)

graysky commented on 2016-05-11 22:54 (UTC)

Bump to v4.5.4-1 Changelog: Commit:

graysky commented on 2016-05-04 22:46 (UTC)

Bump to v4.5.3-1 Changelog: Commit:

herraiz commented on 2016-04-29 17:16 (UTC)

@graysky you are right, I had an old copy of that patch in my sources dir. Thanks!

ScarecrowDM commented on 2016-04-29 14:56 (UTC)

@Tjuh, same problem here. Disabling BUILDDIR=/tmp/makepkg fixed it for me. :)

graysky commented on 2016-04-28 18:44 (UTC)

@herraiz - It's on your end :) Delete the offending file to allow makepkg to redownload it.

herraiz commented on 2016-04-28 17:19 (UTC)

One of the patches fails to pass the checksum: 0003-block-bfq-add-Early-Queue-Merge-EQM-to-BFQ-v7r11-for.patch ... FAILED

graysky commented on 2016-04-27 22:12 (UTC)

@Tjuh - You have something wrong with your setup... compiles for me just fine. I recommend that you try it in a clean chroot. Try clean-chroot-manager here in the AUR.

Tjuh commented on 2016-04-27 21:53 (UTC) (edited on 2016-04-27 21:56 (UTC) by Tjuh)

I did lols, everytime it fails, I unpack the linux-ck.tar.gz and start a fresh makepkg. And furthermore those dirs src and pkg aren't even present when that error occurs.

graysky commented on 2016-04-27 21:33 (UTC)

@francoism - @Tjuh - Clear your src dir before you build (ie rm -rf pkg src) assuming you're using makepkg and not an AUR helper.

Tjuh commented on 2016-04-27 21:14 (UTC)

So I've tried to compile this atleast a dozen times, but it fails everytime at applying a patch; ... ==> Patching source with ck1 including BFS v0.469 patching file arch/powerpc/platforms/cell/spufs/sched.c The next patch would create the file Documentation/scheduler/sched-BFS.txt, which already exists! Skipping patch. 1 out of 1 hunk ignored patching file Documentation/sysctl/kernel.txt patching file fs/proc/base.c patching file include/linux/init_task.h patching file include/linux/ioprio.h patching file include/linux/sched.h patching file init/Kconfig patching file init/main.c patching file kernel/delayacct.c patching file kernel/exit.c patching file kernel/sysctl.c patching file lib/Kconfig.debug patching file include/linux/jiffies.h patching file drivers/cpufreq/cpufreq.c patching file drivers/cpufreq/cpufreq_ondemand.c The next patch would create the file kernel/sched/bfs.c, which already exists! Skipping patch. 1 out of 1 hunk ignored patching file include/uapi/linux/sched.h patching file drivers/cpufreq/cpufreq_conservative.c patching file kernel/time/Kconfig patching file kernel/sched/Makefile The next patch would create the file kernel/sched/bfs_sched.h, which already exists! Skipping patch. 1 out of 1 hunk ignored patching file kernel/sched/stats.c patching file arch/x86/Kconfig Hunk #2 succeeded at 1294 (offset 1 line). Hunk #3 succeeded at 1314 (offset 1 line). Hunk #4 succeeded at 2012 (offset 1 line). Hunk #5 succeeded at 2041 (offset 1 line). patching file include/linux/sched/prio.h patching file drivers/cpufreq/intel_pstate.c patching file kernel/sched/idle.c patching file kernel/time/posix-cpu-timers.c patching file kernel/trace/trace_selftest.c patching file kernel/Kconfig.preempt patching file kernel/Kconfig.hz patching file Makefile ==> ERROR: A failure occurred in prepare(). Aborting... ...

francoism90 commented on 2016-04-27 19:59 (UTC) (edited on 2016-04-27 20:00 (UTC) by francoism90)

@graysky: I mean patches that aren't in the added to the Linux kernel yet, like the BFS patch. :) But that's maybe a bit off-topic, think I'll better ask on the forums.

graysky commented on 2016-04-27 19:40 (UTC)

@francoism - Not sure what you're asking?

francoism90 commented on 2016-04-27 19:37 (UTC)

@graysky: Thanks going to upgrade tomorrow. :D Off-topic: is there a repo somewhere with recommended patches and working with this kernel? :)

graysky commented on 2016-04-27 19:17 (UTC)

Bump to v4.5.2-1 Changelog: Commit: LinuxChanges:

sir_lucjan commented on 2016-04-27 18:44 (UTC)

@FadeMind - a little bit to late, BFQ for 4.5 is out :D

FadeMind commented on 2016-04-27 15:20 (UTC) (edited on 2016-04-28 06:33 (UTC) by FadeMind)

Who is interesting Linux 4.5 with CK patches can download ready PKGBUILD and patches from my repo. BFQ for 4.5 kernel included. git clone Kernel working for me like a charm. See file for details. Regards. FadeMind

francoism90 commented on 2016-04-22 20:43 (UTC)

@graysky: thanks for the skylake update! :)

graysky commented on 2016-04-20 21:21 (UTC)

Bump to v4.4.8-1 Changelog: Commit:

graysky commented on 2016-04-19 19:55 (UTC)

@QuartzDragon - Yes, Paolo told me as much in a private email but I don't want to force all the repo users to be guinea pigs since the 4.5 series of BFQ should be released very shortly per Paolo. @kyak - No problem, thanks for pointing that out. @francoism - No testing is needed, it's just that adding support required tweaks to the entire set of packages (ie adding conflicts arrays) the skylake packages will be included in the next build (either 4.4.8 or 4.5.2 or whenever Paolo releases BFQ for 4.5.x - whichever comes first).

francoism90 commented on 2016-04-19 18:01 (UTC)

Hi graysky, Thanks for providing Linux-ck, really like the BFQ scheduler and also like the easy tweaking/customizing. :) Sorry if this is already ask before, but if testing is needed for the v4.5.x series, please let me know. I'm running on Skylake platform, maybe this platform needs more testing? :P Thanks again!

kyak commented on 2016-04-19 14:54 (UTC)

@graysky - thank you, all works fine now! As for rename, i'm not pushing at all; this is just something to keep in mind. Thanks again!

QuartzDragon commented on 2016-04-19 12:59 (UTC) (edited on 2016-04-19 15:37 (UTC) by QuartzDragon)

Greetings graysky, Some users on the BFQ mailing list have tested the v7r11 4.4.0 patches on 4.5.0 kernel and have found they work without any issues. I've also tested for myself with a modified PKGBUILD and found they work stably and without any issues or quirks compared to CFQ, no matter my system load or uptime. I guess your plans are to wait for the official patches?

graysky commented on 2016-04-18 20:48 (UTC)

@kyak - Thanks, try refreshing and downloading now. As to the renaming of the package, that would require a rewrite of my build script which isn't in the cards for the short term.

kyak commented on 2016-04-18 17:47 (UTC) (edited on 2016-04-18 17:48 (UTC) by kyak)

@graysky, thanks! But virtualbox-ck-host-modules-ivybridge and virtualbox-host-modules-arch are conflicting now. How should we handle this? By the way, does it make sense to change the naming convention in accordance to upstream - i.e. virtualbox-host-modules-ck-ivybridge ?

graysky commented on 2016-04-18 16:51 (UTC)

@kyak - Done.

kyak commented on 2016-04-18 15:57 (UTC)

hi graysky, can we get the virtualbox-host-modules-ck-* back now that the upstream has returned the binary modules?

zipeldiablo commented on 2016-04-17 18:40 (UTC)

Hi guys, question might be weird but i run windows 10 in virtual machine with vga passthrough (qemu-kvm). Should i expect any performance gain on my benchmark or will i just have more perf on my host system?

graysky commented on 2016-04-12 21:08 (UTC)

Bump to v4.4.7-1 Changelog: Commit:

graysky commented on 2016-04-06 19:43 (UTC)

NOTE -- Even though CK has released BFS for linux v4.5.x, there is not a corresponding BFQ I/O scheduler for this series yet so please don't flag out-of-date until Paolo has done so.

graysky commented on 2016-04-06 11:37 (UTC)

@tech4david - See the pinned comment :p

zerophase commented on 2016-04-02 15:34 (UTC)

@pedrogabriel, I tried the steamos-xpad driver. Error! There are no instances of module: steamos-xpad 20160103 located in the DKMS tree. error: command failed to execute correctly I tried deleting the xpad.ko.gz file from 4.4.6-1-ck kernel with no luck.

pedrogabriel commented on 2016-04-02 02:52 (UTC)

@zerophase, use the steamos driver.

zerophase commented on 2016-04-02 01:28 (UTC)

I managed to get my 360 controller working by enabling xpad, but can't figure out how to get the led ring to stop blinking with the ck kernel? Anyone able to help.

graysky commented on 2016-03-26 11:01 (UTC)

@kyak -

kyak commented on 2016-03-26 05:49 (UTC)

Hi graysky, Where has virtualbox-ck-host-modules-ivybridge gone? I only see generic virtualbox-ck-host-modules now. Should i use that?

Nordlicht commented on 2016-03-25 20:33 (UTC)

Runs without problems here with nvidia-ck (AUR).

graysky commented on 2016-03-25 19:04 (UTC)

OK all, please build and test 4.4.6-1 which is running fine on my machine. I will hold off a bit on building/pushing to the repo until I get some feedback from you guys. Thanks.

graysky commented on 2016-03-25 18:27 (UTC) (edited on 2016-03-26 11:02 (UTC) by graysky)

Bump to v4.4.6-1 Changelog: Commit: LinuxChanges: NOTE -- Even though CK has released BFS for linux v4.5.x, there is not a corresponding BFQ I/O scheduler for this series yet so please don't flag out-of-date until Paolo has done so.

Distorted commented on 2016-03-25 12:07 (UTC)

Running it now without issues

graysky commented on 2016-03-25 09:02 (UTC)

Dammit. You will have to rename the file to simply linux-ck-4.4.6-1.src.tar after you download it as it wasn't run through gzip my mistake. Also, you will need to update the checksums as config will fail as-is.

graysky commented on 2016-03-25 08:07 (UTC)

Guys - I don't have time to verify this builds or test so I am holding off pushing up to the AUR until later today or over the weekend depending on my schedule . Please test 4.4.6 and report back. Source tarball:

inti commented on 2016-03-19 17:56 (UTC)

@Peter1992 I'm not sure if this is what you meant, but graysky doesn't create the ck patchset; Con Kolivas does. See: graysky puts together the AUR linux-ck PKGBUILD once Con Kolivas releases updated versions of the patchset.

Peter1992 commented on 2016-03-14 21:28 (UTC) (edited on 2016-03-14 21:28 (UTC) by Peter1992)

@graysky Right,graysky didn't know about that.Well just for curiosity how long would you think it would take for you to put the ck1 patchsets for the 4.4.x kernels?

graysky commented on 2016-03-14 19:31 (UTC)

@Peter1992 - Not out of date... there is no ck1 for 4.4.x.

FadeMind commented on 2016-03-09 08:59 (UTC)

@ thevictoryiswon gpg --recv-keys ABAF11C65A2970B130ABE3C479BE3E4300411886 647F28654894E3BD457199BE38DBBDC86092693E

Distorted commented on 2016-03-09 08:59 (UTC)

@thevictoryiswon Look at some of the older comments, your issue has come up numerous times.

thevictoryiswon commented on 2016-03-09 04:00 (UTC)

It will not build for me. Has it been corrupted? ==> Verifying source file signatures with gpg... linux-4.3.tar ... FAILED (unknown public key 79BE3E4300411886) patch-4.3.6 ... FAILED (unknown public key 38DBBDC86092693E) ==> ERROR: One or more PGP signatures could not be verified!

WFV commented on 2016-03-06 23:13 (UTC)

Thanks graysky and FadeMind, rebuild worked.

FadeMind commented on 2016-03-06 21:50 (UTC) (edited on 2016-03-06 21:53 (UTC) by FadeMind)

@WFV kernel images have stored systemd version from time creating of image kernel. For bump this run just sudo mkinitcpio -P and reboot. Alternative: ignore this or add to grub conf file /etc/default/grub in line GRUB_CMDLINE_LINUX rd.udev.log-priority=3 for prevent print systemd version during boot. Update grub after made changes in a conf file OFC.

graysky commented on 2016-03-06 21:49 (UTC)

@WFV - Rebuild your initramfs image... nothing to do with the packages.

WFV commented on 2016-03-06 21:27 (UTC)

Running 4.3.6-1-ck (piledriver) and current -Syu as of 3/3/16. I notice during boot the message still shows "starting version 228" instead of 229. Doesn't appear to affect anything, but was thinking the version refers to systemd?

Distorted commented on 2016-02-21 21:45 (UTC)

@farwaver i have the same issue, but im still on 4.3.5, so i think its unrelated.

graysky commented on 2016-02-21 21:44 (UTC)

@fawyayer - You need to compare 4.3.6-1-ck to 4.3.6-1-ARCH (which you need to build yourself) to provide a comparison. Comparing to 4.4.1-1-ARCH is the wrong experiment.

farwayer commented on 2016-02-21 21:43 (UTC) (edited on 2016-02-21 21:43 (UTC) by farwayer)

Look like there is some problem with working intel i965 driver with 4.3.6-1-ck kernel. Xorg cursor does not render. However there are no errors in xorg log. Changing acceleration method to old uxa fixs the problem. All ok with arch vanila kernel (4.4.1).

graysky commented on 2016-02-20 00:12 (UTC)

Bump to v4.3.6-1 Changelog: Commit:

nlopez commented on 2016-02-15 17:40 (UTC)

gpg --recv-keys 79BE3E4300411886 38DBBDC86092693E # to properly validate the signatures on this package. As the page @Distored mentions, makepkg uses your user keyring:

sinatosk commented on 2016-02-07 15:33 (UTC) (edited on 2016-02-07 15:51 (UTC) by sinatosk)

nevermind... I see that in order for CGROUP_CPUACCT it to show... BFS needs to be disabled... just noticed

commented on 2016-02-04 02:28 (UTC)

Sent you an e-mail with a small change to the PKGBUILD. Hope it helps.

graysky commented on 2016-02-02 22:30 (UTC)

@hydraz - Not out-of-date until CK releases the BFS for 4.4 :p

commented on 2016-02-01 17:26 (UTC) It seems like linux-ck-headers was installed twice and linux-ck wasn't installed?

graysky commented on 2016-02-01 00:05 (UTC)

Bump to v4.3.5-1 Changelog: Commit:

Distorted commented on 2016-01-29 01:35 (UTC) (edited on 2016-01-29 01:36 (UTC) by Distorted)

@sa1t TL;DR version: use 'makepkg --skippgpcheck'

sa1t commented on 2016-01-29 01:19 (UTC)

==> Verifying source file signatures with gpg... linux-4.3.tar ... FAILED (unknown public key 79BE3E4300411886) patch-4.3.4 ... FAILED (unknown public key 38DBBDC86092693E) ==> ERROR: One or more PGP signatures could not be verified! ==> ERROR: Makepkg was unable to build linux-ck. ==> Restart building linux-ck ? [y/N]

FadeMind commented on 2016-01-24 20:11 (UTC) ;) but for 4.4.0, ck1 is still pending...

eduardoeae commented on 2016-01-24 20:02 (UTC)

BFQ has been updated to v7r11 fixing some bugs.

graysky commented on 2016-01-23 12:42 (UTC)

Bump to v4.3.4-1 Changelog: Commit:

Distorted commented on 2016-01-23 10:24 (UTC)

Would it be possible to update to 4.3.4?

graysky commented on 2016-01-20 15:36 (UTC)

Bump to v4.3.3-3 Changelog: fix #47820 CVE-2016-0728 Commit:

graysky commented on 2016-01-18 20:53 (UTC)

I saw it, FM, thanks. Just need ck1...

FadeMind commented on 2016-01-18 18:26 (UTC)


graysky commented on 2016-01-13 08:43 (UTC)

@gutigen - Not out of date until: -Arch updates to 4.4 -CK releases a BFS for 4.4 -Paolo releases a BFQ for 4.4

graysky commented on 2015-12-24 15:35 (UTC)

Bump to v4.3.3-2 Changelog: fix #46968 Commit:

graysky commented on 2015-12-24 15:35 (UTC)

Sorry about that. Fixed now (didn't bump pkgver since I just rebuilt the repo).

whatot commented on 2015-12-24 14:53 (UTC)

missing the sha256sum of 0001-disabling-primary-plane-in-the-noatomic-case.patch after the eighth.

pedrogabriel commented on 2015-12-24 14:50 (UTC)

==> ERROR: Integrity checks (sha256) differ in size from the source array. ==> ERROR: Makepkg was unable to build linux-ck. It's not just you.

Dijuna commented on 2015-12-24 14:19 (UTC)

==> ERROR: Integrity checks (sha256) differ in size from the source array. ==> ERROR: Makepkg was unable to build linux-ck. Is it just me?

graysky commented on 2015-12-15 08:37 (UTC)

Bump to v4.3.3-1 Changelog: Commit: Note from CK:

graysky commented on 2015-12-14 12:02 (UTC)

@XenGi - I try to be as true to the Arch config as I can be for linux-ck. You are of course free to customize/build your own. My suggestion is for you to open a flyspray again the official Arch linux package asking for this. Might want to include your reasoning why it would be important.

XenGi commented on 2015-12-14 10:51 (UTC)

hi, Could you set CONFIG_CHECKPOINT_RESTORE=y for all ck kernels? I'm using the broadwell build from repo-ck. This would enable fancy process names for LXC. The setting is not set atm. ``` $ zgrep 'CONFIG_CHECKPOINT_RESTORE' /proc/config.gz # CONFIG_CHECKPOINT_RESTORE is not set ``` Here's a thread explaining the issue:

graysky commented on 2015-12-14 07:29 (UTC)

I did find this suggesting it is not related to linux-ck:

graysky commented on 2015-12-14 07:27 (UTC)

Recommend that you post to CK's blog.

enihcam commented on 2015-12-14 07:14 (UTC)

@renodesper I got the exactly same issue.

renodesper commented on 2015-12-14 01:58 (UTC)

I got this weird message when entering boot process: [ 47.772679] r8169 0000:03:00.0 enp3s0: rtl_counters_cond == 1 (loop: 1000, delay: 10). [ 52.807884] r8169 0000:03:00.0 enp3s0: rtl_counters_cond == 1 (loop: 1000, delay: 10). [ 57.842769] r8169 0000:03:00.0 enp3s0: rtl_counters_cond == 1 (loop: 1000, delay: 10). [ 62.877924] r8169 0000:03:00.0 enp3s0: rtl_counters_cond == 1 (loop: 1000, delay: 10). [ 67.912998] r8169 0000:03:00.0 enp3s0: rtl_counters_cond == 1 (loop: 1000, delay: 10). [ 72.947899] r8169 0000:03:00.0 enp3s0: rtl_counters_cond == 1 (loop: 1000, delay: 10). [ 77.982966] r8169 0000:03:00.0 enp3s0: rtl_counters_cond == 1 (loop: 1000, delay: 10). [ 83.017963] r8169 0000:03:00.0 enp3s0: rtl_counters_cond == 1 (loop: 1000, delay: 10). [ 87.661070] r8169 0000:03:00.0 enp3s0: rtl_counters_cond == 1 (loop: 1000, delay: 10). [ 88.053001] r8169 0000:03:00.0 enp3s0: rtl_counters_cond == 1 (loop: 1000, delay: 10). It still showing up when I check from dmesg. The message only appear when I use ck kernel. Is there something wrong with ck kernel?

graysky commented on 2015-12-11 13:21 (UTC)

Bump to v4.3.2-1 Changelog: Commit:

graysky commented on 2015-12-11 13:01 (UTC)

Bump to v4.3.1-3 Changelog: Incorporate BFS 466 in ck2[1] Commit: 1.

graysky commented on 2015-12-10 18:27 (UTC)

Bump to v4.3.1-2 Changelog: Mirroring changes to official package's config. Commit:

graysky commented on 2015-12-09 21:08 (UTC)

Bump to v4.3.1-1 Changelog: Commit:

crobe2 commented on 2015-12-07 21:41 (UTC)

The page seems down :(

neTpK commented on 2015-11-24 22:04 (UTC)

For all those with this problem in dmesg/bootup: [drm:intel_pipe_config_compare [i915]] *ERROR* mismatch in has_drrs (expected 1, found 0) Theres also a warning, I dont know if it is related: WARNING: CPU: 1 PID: 6 at drivers/gpu/drm/i915/intel_display.c:12700 intel_atomic_commit+0xdea/0x1390 [i915]() Bug submittet here:

kyak commented on 2015-11-13 18:18 (UTC)

@Distorted - of course, and thanks again!

Distorted commented on 2015-11-13 18:15 (UTC)

@kyak I dont use the repo-ck, so i'm just guessing here, if linux-ck-ivybridge is not version 4.3 you're absolutely right, but until then you could use an older commit of bbswitch-ck.

FadeMind commented on 2015-11-13 18:13 (UTC)

kyak linux-ck is main package for GENERIC (ALL) CPU's and OFC can be adjust BEFORE compile. Customized by CPU names are designed for powerfull CPU's AMD and Intel See: are available on repo site. IF You have installed linux-ck-ivybridge You have two options: 1. Download sources of linux-ck and custom config file for enable valid CPU family for GCC and compile it. 2. Purge whole -ck packages (linux-ck-ivybridge, bbswitch-ck, and nvidia-ck-ivybridge) 3. Purge and install again bbswitch-ck package (major linux version bumped from 4.1 to 4.3 and package NEED to be recompilled against 4.3) 4. Install new version of nvidia-ck version ready for 4.3 OR JUST WAIT for upload proper packages in to repo-ck repository. SUBFORUM of AUR packages IS designed for report ISSUES with PKGBUILD/package build issues. FOR help with Your misunderstanding of the issues please write directly on BBS Arch Forums: Reference And one more: stop spamming and confuse users of other packages (bbswitch-ck).

kyak commented on 2015-11-13 18:11 (UTC) (edited on 2015-11-13 18:12 (UTC) by kyak)

@Distorted i think i get it now. I should wait for the latest version of linux-ck-ivybridge to hit the repo-ck. After that i'd have something that provides linux-ck=4.3 and be able to install bbswitch-ck. I didn't realize i was still using linux-ck-ivybridge 4.1.3

Distorted commented on 2015-11-13 18:01 (UTC)

@kyak also, did you rebuild the bbswitch-ck package? i updated it for the new version.

Distorted commented on 2015-11-13 18:00 (UTC)

@kyak You could probably edit the bbswitch-ck PKGBUILD file and change depends=linux-ck to depends=linux-ck-ivybridge Unless they changed the actual name of the kernel, not just the package. If this is not the case, the linux-ck-ivybridge pkgbuild should say provide='linux-ck=4.3' to solve these kinds of problems.

kyak commented on 2015-11-13 17:54 (UTC)

Well, wait a second.. bbswitch-ck requires linux-ck, which conflicts with linux-ck-ivybridge that i have installed. What to do? It used to work before

kyak commented on 2015-11-13 17:51 (UTC)

ok, finally.. had to uninstall bbswitch-ck. It doesn't work woth 4.3 kernel. Thanks for the hint, Distorted.

kyak commented on 2015-11-13 17:47 (UTC)

but.. i didn't build anything from aur. I'm using repo-ck

graysky commented on 2015-11-13 17:25 (UTC)

Just remove all repo-ck packages before you try installing one you built yourself from the AUR: pacman -Rs ck-ivybridge

kyak commented on 2015-11-13 17:01 (UTC)

$ LANG=C pacaur -R linux-ck error: target not found: linux-ck Yes, duh. I don't have linux-ck installed, only linux-ck-ivybridge. I suspect a pacaur bug.

sir_lucjan commented on 2015-11-13 16:42 (UTC)

Read carefully: conflicts=('kernel26-ck' 'linux-ck-corex' 'linux-ck-p4' 'linux-ck-pentm' 'linux-ck-atom' 'linux-ck-core2' 'linux-ck-nehalem' 'linux-ck-sandybridge' 'linux-ck-ivybridge' 'linux-ck-broadwell' 'linux-ck-haswell' 'linux-ck-kx' 'linux-ck-k10' 'linux-ck-barcelona' 'linux-ck-bulldozer' 'linux-ck-piledriver' 'linux-ck-silvermont') Duh? :)

Distorted commented on 2015-11-13 15:28 (UTC)

@kyak There are probably some packages that depend on linux-ck-ivybridge that are the cause this error.

kyak commented on 2015-11-13 15:16 (UTC) (edited on 2015-11-13 15:16 (UTC) by kyak)

Running pacaur -Syu: :: linux-ck and linux-ck-ivybridge are in conflict (linux-ck). Remove linux-ck-ivybridge? [y/N] :: unresolvable package conflicts detected :: failed to prepare transaction (conflicting dependencies) :: linux-ck and linux-ck-ivybridge are in conflict Duh?

FadeMind commented on 2015-11-12 21:12 (UTC)

Self compilled with custom modules list (modprobed_db). Linux version 4.3.0-1-ck (tomasz@arch) (gcc version 5.2.0 (GCC) ) #1 SMP PREEMPT Thu Nov 12 21:46:41 CET 2015 custom modules list for my device: kernel tree: enabled blk-mq via grub flag (scsi_mod.use_blk_mq=1) (working like a charm - better than noop) dmesg: config: Good work CK and Graysky!

graysky commented on 2015-11-12 19:41 (UTC)

Bump to v4.3-1 Changelog: Commit: LinuxChanges: Note: I am working on the related packages now (broadcom, nvidia, vbox) and I will upload to the repo once I get some feedback from AUR users as to stability so please use the AUR page to report in if something is bad (or good)!

yannick commented on 2015-11-12 08:25 (UTC)

It's building now, thx

vishwin commented on 2015-11-11 07:53 (UTC)

I just 'SKIP'ped that file for now, will wait for the next release haha.

graysky commented on 2015-11-10 08:45 (UTC)

@yannick and vishwin - Try now.

yannick commented on 2015-11-10 08:41 (UTC) (edited on 2015-11-10 08:42 (UTC) by yannick)

This part of the patch is not good: - sha256sums = 819961379909c028e321f37e27a8b1b08f1f1e3dd58680e07b541921282da532 + sha256sums = cf0f984ebfbb8ca8ffee1a12fd791437064b9ebe0712d6f813fd5681d4840791

vishwin commented on 2015-11-10 00:24 (UTC)

"enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz ... FAILED" …yet the file itself appears to never have been modified and has passed before 4.1.13-1?

graysky commented on 2015-11-09 23:03 (UTC)

Bump to v4.1.13-1 Changelog: Commit:

artafinde commented on 2015-10-28 20:48 (UTC) (edited on 2015-10-28 20:54 (UTC) by artafinde)

Hmm seems like the BFQ patches host [1] is down. Mirror from Oleksandr Natalenko at same SHA256 sum [1]:

graysky commented on 2015-10-27 21:12 (UTC)

Bump to v4.1.12-1 Changelog: Commit:

graysky commented on 2015-10-27 11:58 (UTC)

Thanks for the comments, FM, and SL. I have been following Con's blog and am waiting to hear back from him that the proposed solution is kosher before I apply it to the AUR and repo. Stay tuned.

sir_lucjan commented on 2015-10-27 07:22 (UTC)

@FadeMind: "Just a heads up that 4.1.12 will break compilation of BFS because of a change in the common scheduler API. I already sent a patch to Con to fix it, so please don't panic. :)"

FadeMind commented on 2015-10-27 06:58 (UTC) (edited on 2015-10-27 07:00 (UTC) by FadeMind) Main URL to sources was changed to Sample:

graysky commented on 2015-10-22 23:07 (UTC)

Bump to v4.1.11-1 Changelog: Commit:

glitsj16 commented on 2015-10-14 00:13 (UTC)

Hi graysky, would you consider modifying the links to inside the PKGBUILD cfr. to improve download speeds for folks located outside North America?

graysky commented on 2015-10-07 18:43 (UTC)

Bump to v4.1.10-3 Changelog: Fix potential deadlock in reqsk_queue_unlink[1] Commit: 1.

graysky commented on 2015-10-05 19:22 (UTC)

Bump to v4.1.10-1 Changelog: Commit:

kyak commented on 2015-10-02 11:38 (UTC)

Hi graysky, The repo download speed is doing pretty badly. Even with several Server lines in pacman.conf. Even worse, the download doesn't resume from the same place. At maximum i was able to download around 1000 Kb, and the download starts over and over again from the beginning of the file. It's been like that for a couple of days now (can't update to v4.1.9-1). It would be great if you could have a look. Thank you!

graysky commented on 2015-09-29 19:33 (UTC)

Bump to v4.1.9-1 Changelog: Commit:

graysky commented on 2015-09-22 00:49 (UTC)

Bump to v4.1.8-1 Changelog: Commit:

graysky commented on 2015-08-31 05:29 (UTC)

@ali - Why out of date? No BFQ for 4.2 and no ck1 for 4.2.

NullDivision commented on 2015-08-30 18:30 (UTC)

Yeah. Sorry. My mistake. Forgot to add the key ID to the chain. ^.^

sir_lucjan commented on 2015-08-30 17:19 (UTC)

@NullDivision No.

Tjuh commented on 2015-08-24 14:35 (UTC)

Yes, I do. That's interesting, I'm gonna recompile it with localmodconf disabled. Thanks.

marceliq commented on 2015-08-24 08:01 (UTC)

@Tjuh: Do you use localmodconf with modprobed-db? Because I found out that if I don't use it then it works without panics.

Tjuh commented on 2015-08-23 15:26 (UTC)

I am regularly getting a kernel panic, with and without NUMA disabled :S

graysky commented on 2015-08-17 21:58 (UTC)

Bump to v4.1.6-1 Changelog: Commit:

graysky commented on 2015-08-11 18:38 (UTC)

Bump to v4.1.5-1 Changelog: Commit:

artafinde commented on 2015-08-11 08:20 (UTC)

This should work (it builds but didn't boot it yet)

FadeMind commented on 2015-08-11 06:00 (UTC)

4.1.5 released

graysky commented on 2015-08-09 07:23 (UTC)

Bump to v4.1.4-2 Changelog: Bump to ck2 patchset Commit:

nemesys commented on 2015-08-08 03:16 (UTC)

I consistently disable NUMA on my Lenovo T420 and have not had any recognizable issues. I also customize my kernel pretty heavily to remove support for things I don't use or need.

artafinde commented on 2015-08-06 22:53 (UTC)

I'm running with NUMA disabled and with ck patch for reverting unplugged [1]. I'm fairly stable with BFQ enabled and on BTRFS. I'll post if I got issues. Test: Normal work day, scrubs on BTRFS, kernel compilation. [1]

dkaylor commented on 2015-08-05 10:53 (UTC)

I may have spoken too soon. System had been up and running for about 30 hours with no trouble. Then did a pacman -Syu (with linux-4.1.4 and systemd-224) and got a panic right in the middle of the kernel image generation. Reboot, cold boot, nothing helped. Panic then hang. And generic kernel image generation didn't finish, so system was toast. Will try a repair with ISO before, but wiki instructions seem dodgy. Probably should have saved pacman work for the generic kernel, lesson learned.

marceliq commented on 2015-08-05 08:50 (UTC)

I have NUMA disabled and today kernel panic appears during boot process. After hard reset it booted without problems.

artafinde commented on 2015-08-05 08:28 (UTC)

I've recompiled with disabled NUMA. After half a dozen reboots no panic so far.

pedrogabriel commented on 2015-08-05 04:24 (UTC)

@graysky - The kernel is working fine in my system with NUMA disabled.

dkaylor commented on 2015-08-04 23:52 (UTC)

Additional to my previous post: SMT is turned off in my config, since processor is 4 cores == 4 threads. Doubt it's relevant, but just in case.

graysky commented on 2015-08-04 23:03 (UTC)

@dk - That's good to hear. I guess we should solicit others to try disabling NUMA and checking for stability. Any takers?

dkaylor commented on 2015-08-04 22:13 (UTC)

Regarding NUMA - I've been running a custom linux-ck 4.1.4-1 with NUMA disabled and haven't had any hangs. Everyday workloads, several suspend/resume cycles. On the same hardware, I had to turn NUMA off on 4.0.x to get it to run. Machine is Core2 Quad.

artafinde commented on 2015-08-04 14:00 (UTC)

Maybe it's better to follow up at forums [1] [1]:

marceliq commented on 2015-08-04 12:29 (UTC)

Thanks FadeMind. Sorry, I didnt notice that. Somehow it worked in previous versions. Now I have same problem as artafinde, but without nouveau :)

FadeMind commented on 2015-08-04 09:57 (UTC)

marceliq modprobed_db use _ instead of - for naming modules. You mixed up this. Corrected modules list: dm_bio_prison dm_bufio dm_cache dm_cache_mq dm_crypt dm_log dm_mirror dm_mod dm_persistent_data dm_region_hash dm_snapshot

marceliq commented on 2015-08-04 09:31 (UTC)

Hi, I run into some trouble. Im using localmodconf. In my modprobed.db file I can see modules: dm_bio_prison dm_bufio dm-cache dm-cache-mq dm_crypt dm_log dm_mirror dm_mod dm_persistent_data dm_region_hash dm-snapshot But after compilation, while installing new package when building new init image I get this: >>> Updating module dependencies. Please wait ... >>> Generating initial ramdisk, using mkinitcpio. Please wait... ==> Building image from preset: /etc/mkinitcpio.d/linux-ck.preset: 'default' -> -k /boot/vmlinuz-linux-ck -c /etc/mkinitcpio.conf -g /boot/initramfs-linux-ck.img ==> Starting build: 4.1.4-1-ck -> Running build hook: [base] -> Running build hook: [udev] -> Running build hook: [autodetect] -> Running build hook: [modconf] -> Running build hook: [block] -> Running build hook: [encrypt] -> Running build hook: [lvm2] ==> ERROR: module not found: `dm-snapshot' ==> ERROR: module not found: `dm-mirror' ==> ERROR: module not found: `dm-cache' ==> ERROR: module not found: `dm-cache-mq' -> Running build hook: [filesystems] -> Running build hook: [keyboard] -> Running build hook: [fsck] ==> Generating module dependencies ==> Creating lzop-compressed initcpio image: /boot/initramfs-linux-ck.img ==> WARNING: errors were encountered during the build. The image may not be complete. Any suggestions? Thanks.

graysky commented on 2015-08-03 19:51 (UTC)

Bump to v4.1.4-1 Changelog: Commit:

graysky commented on 2015-08-03 08:42 (UTC)

Post to ck's blog.

artafinde commented on 2015-08-03 07:41 (UTC)

I had a kernel panic while shutdown with latest build [1]. This is build from AUR with BFQ, nouveau and localmodconfig. [1]:

graysky commented on 2015-08-02 22:49 (UTC)

Just enable the _makenconfig in the PKGBUILD and build it. You'll need to select the CPU from the nconfig therein.

artafinde commented on 2015-08-02 21:48 (UTC)

@ummakybes: If you have configured makepkg accordingly yes, see [1]. [1]:

ummakynes commented on 2015-08-02 21:27 (UTC)

Does this PKGBUILD automatically apply a CPU architecture optimization?

graysky commented on 2015-08-02 17:20 (UTC)

Bump to v4.1.3-1 LinuxChanges: Commit:

graysky commented on 2015-08-02 17:19 (UTC)

No idea...

graysky commented on 2015-07-24 22:50 (UTC)

Bump to v4.0.9-2 Changelog: Rebuild against gcc 5.2.0-1. Commit:

graysky commented on 2015-07-22 08:02 (UTC)

Bump to v4.0.9-1 Changelog: Commit:

graysky commented on 2015-07-17 19:12 (UTC)

@spotfire - how ood? No ck patch set means not ood.

Gacha commented on 2015-07-07 19:55 (UTC)

When you see "ERROR: One or more PGP signatures could not be verified!" then import the required keys with: gpg --keyserver --recv-keys ABAF11C65A2970B130ABE3C479BE3E4300411886 647F28654894E3BD457199BE38DBBDC86092693e

graysky commented on 2015-06-30 19:30 (UTC)

Bump to v4.0.7-1 Changelog: Commit:

zoopp commented on 2015-06-25 17:28 (UTC)

I might have found something regarding the Kernel Panic. I usually compile (with the NUMA patch enabled) using native optimizations detected by gcc and I disable kernel debugging. Today I updated and forgot to disable kernel debugging and for the first time since this issue arose I got the kernel panic. I then recompiled with kernel debugging disabled and the kernel panic was gone. Not sure if this is going to be of any use but I'll leave it here just in case.

graysky commented on 2015-06-23 02:05 (UTC)

Bump to v4.0.6-1 Changelog: Commit:

graysky commented on 2015-06-08 22:04 (UTC)

Bump to v4.0.5-2 Changelog: update BFQ to v7r8. Commit:

Det commented on 2015-06-08 05:39 (UTC)

graysky commented on 2015-06-06 16:23 (UTC)

Bump to v4.0.5-1 Changelog: Commit:

FadeMind commented on 2015-06-06 15:53 (UTC)


graysky commented on 2015-05-22 17:37 (UTC)

Bump to v4.0.4-3 Changelog: sync up with ARCH (ext4 data corruption patch) Commit:

graysky commented on 2015-05-20 20:11 (UTC)

Bump to v4.0.4-2 Changelog: sync up config with ARCH trunk. Commit:

graysky commented on 2015-05-17 19:51 (UTC)

Bump to v4.0.4-1 Changelog: Commit:

FadeMind commented on 2015-05-17 19:34 (UTC)

Bump 4.0.4

sir_lucjan commented on 2015-05-14 07:25 (UTC)


ishitatsuyuki commented on 2015-05-14 07:22 (UTC)

Your version bot f*ck out on changelog link

graysky commented on 2015-05-13 19:32 (UTC)

Bump to v4.0.3-1 Changelog: Commit:

graysky commented on 2015-05-12 20:05 (UTC)

Bump to v4.0.2-2 Comment: 4.0.2-2 has been reported by 6 users now to be stable (running for 3-4 days under various conditions). Please post to the repo-ck discussion thread if you experience a problem. Changelog: Commit:

graysky commented on 2015-05-07 00:31 (UTC)

Bump to v3.19.7-1 Changelog: Commit:

graysky commented on 2015-05-06 19:02 (UTC)

Yes, it's in the description as you know so it's a cosmetic mistake. I have corrected it and will update when 3.19.7 goes live. Thank you.

qqqqqqqqq9 commented on 2015-05-06 13:00 (UTC)

actually it should read modprobed-db

qqqqqqqqq9 commented on 2015-05-06 12:58 (UTC)

Hi there's a d missing in line 201: modprobe_db should read modprobed_db

graysky commented on 2015-05-05 19:36 (UTC)

See the official repo-ck thread for a discussion and summary of debugging with CK's help.

XenGi commented on 2015-05-05 11:55 (UTC)

@gravsky Sorry, clicked before thinking. Waiting for 4.0-ck to arrive. Take your time. Maybe look at linux-pf. Seems they have fixed some bfq problems in 4.0.

graysky commented on 2015-05-05 09:08 (UTC)

@XenGi - why out of date?

graysky commented on 2015-04-29 19:34 (UTC)

Bump to v3.19.6-1 Changelog: Commit:

graysky commented on 2015-04-29 19:33 (UTC)

OK. Given all the reports of problems with ck1 and 4.0.x, I have decided to push version 3.19.6 to the AUR (and all associated packages now depend on 3.19.x). If you experienced a kernel panic using linux-ck v4.0.x, please see this post and consider if you can help Con debug: For those interested in the 4.0.x version of PKGBUILDs, I have made them available on repo-ck at this address: PLEASE DO NOT FLAG OUT-OF-DATE UNTIL CK FIXES THE PANICS - I WILL NOT PUSH THE AUR OR REPO UNTIL IT IS STABLE!

graysky commented on 2015-04-26 09:15 (UTC)

Previous versions are available from: May I ask that those of you experiencing problems like this please follow and post in the repo-ck thread in the bbs? Here is where the recent discussion about 4.0 starts:

rat commented on 2015-04-26 08:31 (UTC)

I have the same situation, panic of kernel during the boot, Where I can find the previous version of linux-ck 3.19.5 ??

tomaszc commented on 2015-04-26 07:25 (UTC)

After update linux-ck-nehalem x86_64 from 3.19.5-1 to 4.0-1 there is a kernel panic at boot or later. linux 4.0.0-2-ARCH from testing repo working properly, Jak zdobędę więcej informacji to umieszczę je tutaj,

graysky commented on 2015-04-24 19:48 (UTC)

Bump to v4.0-1 Commit: LinuxChanges:

graysky commented on 2015-04-23 00:02 (UTC)

BFQ for 4.0 is now released. For those wishing to play with version 4.0 now, see the source here: I plan to upload to the AUR shortly (24 h or so) due to updates to other packages.

graysky commented on 2015-04-19 14:36 (UTC)

Bump to v3.19.5-1 Changelog: Commit:

FadeMind commented on 2015-04-19 09:08 (UTC)


graysky commented on 2015-04-18 19:27 (UTC)

@Buddlespit - Why out-of-date? See my comment from 2015-04-16 21:11.

graysky commented on 2015-04-17 19:14 (UTC)

@fademind - dunno what to tell you; running fine on my systems (haswell and ivy with onboard video). @all - heard back from Paolo about BFQ for 4.0. Arianna and he are testing a new patch that is specifically for 4.0 so I don't plan to bump linux-ck until that happens. You can continue using the version I have linked below if you wish until then.

FadeMind commented on 2015-04-17 19:11 (UTC)

@graysky After disabing zramswap. Boot, sddm splash, loading desktop, no wallpaper - black background, freeze mouse and keyboard, capslock key led flash in pulse. I can't do anything except hard poweroff. Keyboard and mouse not respond. Kernel panic is before SDDM startup... IMO it is Intel DRM regression... weird.

graysky commented on 2015-04-17 18:53 (UTC)

@fademind - I dunno about your setup but I see somr zram lines in your dmesg. Disable zram and try again? Also, if you boot to the shell and not to sddm, does it panic? If not, can you start your xsession without sddm by running `xinit` and see if it happens. Trying to isolate which part is causing the hang.

FadeMind commented on 2015-04-17 09:59 (UTC)

@graysky Here it is: Linux 3.19.4-ck boot fine. Here is dmesg from it: Lsmod: inxi -Fxz I can't boot fine on 4.0-ck anyway Regards

graysky commented on 2015-04-17 09:16 (UTC)

@fademind - what video are you using and please report your findings on CK's blog

FadeMind commented on 2015-04-17 05:25 (UTC)

@graysky I'm used Your SRC. After compilling, reboot and boot from linux-ck I have freeze under X11. I saw mouse and SDDM login splash begin animation but hangs. Only hard poweroff (push 5 sec. power button) was usefull for poweroff. After power on and boot from stock kernel I had Recovery Jornal again. It is something wrong with CK for 4.0. I patched -mainline kernel with GCC patch and working fine - without any issues. For SSD I used NOOP. For HDD I used CFQ (in stock) or BFQ (in -ck, -mainline) and I don't had Issues. Dynamic changing scheduler udev rules script. Note: I booted from 3.19.4-ck kernel and PASS fine.

graysky commented on 2015-04-16 21:11 (UTC)

I generated a PKGBUILD+related files for linux-ck-4.0 using ck1-4.0 and using the 3.19.x series of BFQ since there is one report on the BFQ forum that this version works fine in the 4.0 code[1]. You may download it and build/test from this file: I also emailed Paolo to see if this is the official word from his group. I don't want to put linux-ck-4.0 live in the AUR/repo until we hear back from him. Please test this out if you wish and report back good or bad. 1.!topic/bfq-iosched/lUAAQQG6P4A

graysky commented on 2015-04-16 20:57 (UTC)

@FadeMind - So the only difference is ck1 patchset? Are you using BFQ at all?

FadeMind commented on 2015-04-16 14:53 (UTC)

I built own kernel 4.0 with lastest ck patch and I had random kernel panic during boot with hangs. After reboot I had recovery journal of disk. Sometimes is fine, sometimes just hangs and blow up. Safe is waiting for 4.0.1 release update 4.0 kernel series. ON Stock 4.0 sersies ARCH kernel this never happened.

PerfectGentleman commented on 2015-04-16 13:33 (UTC)

yep, BFQ for 3.19 works fine on 4.0

AnAkkk commented on 2015-04-16 10:03 (UTC)

Apparently BFQ for 3.19 works just fine on 4.0 according to their mailing list.

graysky commented on 2015-04-16 10:02 (UTC)

I saw it, butBFQ has not yet been released for v4.0. Please flag again once that happens if I don't beat you to it.

ishitatsuyuki commented on 2015-04-16 07:45 (UTC) It's coming this time.

graysky commented on 2015-04-14 19:54 (UTC)

Bump to v3.19.4-1 Changelog: Commit:

graysky commented on 2015-04-13 08:48 (UTC)

the point is that this package is not outdated even with the release of linux 4.0 or 4.0.1 etc. since there has not yet been a release of ck1 for the 4.0 codebase.

ishitatsuyuki commented on 2015-04-13 08:30 (UTC)

Sorry, flagged with my mistouch.

FadeMind commented on 2015-04-13 07:46 (UTC)

4.0 will be really STABLE in 4.0.1 release after fixing missed regressions. It has always been. Don't rush.

graysky commented on 2015-04-13 07:00 (UTC)

@cryptoluks - Why out of date? 3.19.4 is not released yet, and there is not ck patchset yet for 4.0.

Scimmia commented on 2015-04-09 23:53 (UTC)

There are other ways, too. First off, you don't have to trust the keys. Second, you can automate downloading them in gpg.conf --skippgpcheck is a joke. If you're too lazy to figure out how to use gpg, you shouldn't be building your kernel, to say the least.

Twilight_Genesis commented on 2015-04-09 23:47 (UTC)

2 Ways To fix PGP key errors: (Recommended): Import the PGP keys in the validpgpkeys array in the PKGBUILD and trust them. gpg --recv-keys ABAF11C65A2970B130ABE3C479BE3E4300411886 647F28654894E3BD457199BE38DBBDC86092693E For each key run gpg --edit-key <KEY>, where <KEY> is the PGP key then type trust and hit enter, then 5 and hit enter to trust it ultimately, and finally type quit to quit (Ignore PGP signatures)(Not recommended) Have makepkg skip PGP signature checking just pass the --skippgpcheck argument to makepkg makepkg --skippgpcheck -r -s

Twilight_Genesis commented on 2015-04-09 23:19 (UTC)

@radialhat: You have to either import the PGP keys in the validpgpkeys array in the PKGBUILD and trust them or if you don't care about the PGP signatures then you can pass the --skippgpcheck argument to makepkg and have it ignore the PGP signatures.

graysky commented on 2015-04-06 08:49 (UTC)

@alonhar -

alonhar commented on 2015-04-06 00:43 (UTC)

How can I uninstall linux-ck.. return to the normal kernel and not break arch?

graysky commented on 2015-04-05 12:26 (UTC)

Please take makepkg/gnupg discussions to the forums; I doubt others in the AUR like receiving notifications on these topics which have nothing to do with the package. Thank you.

radialhat commented on 2015-04-05 11:50 (UTC)

What kind of fix do you refer to? I've done nothing to my gnupg setup... It works with other packages

Scimmia commented on 2015-04-05 03:41 (UTC)

@radialhat, if you're having problems, don't work around it. Fix your GnuPG setup. Works fine here.

radialhat commented on 2015-04-05 01:35 (UTC)

Seriously, what is wrong with the PGP keys? Do we have to workaround everytime now?

graysky commented on 2015-03-30 19:25 (UTC)

@sp1d3rmxn - No idea what you're talking about; PKGBUILD is solid as-is.

dkaylor commented on 2015-03-30 12:25 (UTC)

@sp1d3rmxn Uh.... What?

sp1d3rmxn commented on 2015-03-30 11:45 (UTC)

Funny you all are having problems with the pgp key. I brought this up some time ago and was given a bunch of BS about why it "should" work as if I already didn't know it "should". Anyways... I've been seeing this wonderful additional problem with the "enable patch" that for some reason is also being disregarded so here's how to fix it. When you compile the package it's nice to see someone has left or did something stupid (yet again) where the file "enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz" is actually not ending in ".gz". Awesome huh? So what you have to do is this. Rename the file removing the ".gz" part and copy it to the "src" directory where you're building (you can't do this through the yaourt build because you need to intervene beyond editing the PKGBUILD). Recap: - run makepkg -rs (if this is how you do it) - ignore errors since this is becoming commonplace with this package anyway - edit PKGBUILD and remove .sig and .sign PGP crap since no one wants to implement this right anyway - mv enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch - update sha256sums => (4th line down) "e5b0f1882861cb1d0070e49c41a3e7390cf379f183be0bddb54144ecc68fb8f9" - cp enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch to ./src (you should be in the directory you're compiling) - Sit back and wait for fresh kernel and enjoy

Tjuh commented on 2015-03-28 14:01 (UTC)

Unfortunately none of the suggestions below worked, I had to run makepkg with the --skippgpcheck flag to get it to compile.

sir_lucjan commented on 2015-03-27 14:07 (UTC)

1. nano /home/tjuh/.gnupg/gpg.conf 2. Add 'keyserver hkp://' 3. Try again

Tjuh commented on 2015-03-27 13:57 (UTC)

tjuh ~ $ gpg --keyserver hkp:// --recv-keys 00411886 6092693E gpg: directory '/home/tjuh/.gnupg' created gpg: new configuration file '/home/tjuh/.gnupg/gpg.conf' created gpg: WARNING: options in '/home/tjuh/.gnupg/gpg.conf' are not yet active during this run gpg: keybox '/home/tjuh/.gnupg/pubring.kbx' created gpg: keyserver receive failed: No keyserver available Same warning :S

FadeMind commented on 2015-03-27 13:12 (UTC)

Execute this command: gpg --keyserver hkp:// --recv-keys 00411886 6092693E Scimmia thanks for tip.

Scimmia commented on 2015-03-27 12:36 (UTC)

You shouldn't need to sign them.

Tjuh commented on 2015-03-27 11:45 (UTC)

==> Verifying source file signatures with gpg... linux-3.19.tar ... FAILED (unknown public key 79BE3E4300411886) patch-3.19.3 ... FAILED (unknown public key 38DBBDC86092693E) ==> ERROR: One or more PGP signatures could not be verified! ==> Removing installed dependencies... checking dependencies... Packages (1) bc-1.06.95-1 Total Removed Size: 0.18 MiB :: Do you want to remove these packages? [Y/n] y (1/1) removing bc [######################] 100% tjuh ~/Downloads/linux-ck $ gpg --recv-keys 79BE3E4300411886 gpg: keyserver receive failed: No keyserver available tjuh ~/Downloads/linux-ck $ gpg --recv-keys 38DBBDC86092693E gpg: keyserver receive failed: No keyserver available Anyway to fix this? I tried running dirmngr </dev/null as root and then those 2 gpg commands again, but the keyserver is still not available.

commented on 2015-03-26 21:21 (UTC)

Thanks Sir_Lucjan, Graysky and Sir.Tiddlesworth -- removed my comments for errors, it was my mistakes. Fixed download interruption using tip on repo-ck wiki and had to add keys as root (not sure why using sudo wasn't working). Finally I am on ck-kx :)

commented on 2015-03-26 21:18 (UTC)

#dirmngr </dev/null #gpg --recv-keys ABAF11C65A2970B130ABE3C479BE3E4300411886 #gpg --recv-keys 647F28654894E3BD457199BE38DBBDC86092693E gpg: keyserver receive failed: No keyserver available

graysky commented on 2015-03-26 21:17 (UTC)

Bump to v3.19.3-1 Changelog: Commit:

commented on 2015-03-26 20:47 (UTC)

Thanks SirTiddle. It worked but repo-ck download speed is pain in the eyes. When doing it via AUR, its again: ==> Verifying source file signatures with gpg... linux-3.19.tar ... FAILED (unknown public key 79BE3E4300411886) patch-3.19.3 ... FAILED (unknown public key 38DBBDC86092693E) :S

Sir.Tiddlesworth commented on 2015-03-26 11:39 (UTC)

@jackpot Run the following as root, then try again. # dirmngr </dev/null

commented on 2015-03-20 10:59 (UTC)

@graysky: Thanks for reply. Actually I posted after every option I could find failed. When I try to add keys as mentioned in I get following: Quote: gpg: connecting dirmngr at '/root/.gnupg/S.dirmngr' failed: IPC connect call failed gpg: keyserver receive failed: No dirmngr ==> ERROR: Remote key not fetched correctly from keyserver. Unquote This is happening since 2 weeks (i.e. the time I have been trying different things)

graysky commented on 2015-03-19 22:58 (UTC)

@jackpot - See the instructions on which should fix you up.

commented on 2015-03-19 22:17 (UTC)

I am unable to add key for repo-ck (server error) and also having similar problem using yaourt. Followed comments of Fademind but again having remote server error. Please suggest solution. Thanks!

graysky commented on 2015-03-18 22:17 (UTC)

Bump to v3.19.2-1 Changelog: Commit:

graysky commented on 2015-03-12 08:42 (UTC)

If I remember correctly, I did this back when I bought a RPi. No real use for it now... In a drawer somewhere :p

ryanvade commented on 2015-03-12 01:34 (UTC)

Can the CK patchset be used on Arm devices?

FadeMind commented on 2015-03-08 12:21 (UTC)

milan385 add these PGP keys: gpg --recv-keys ABAF11C65A2970B130ABE3C479BE3E4300411886 gpg --recv-keys 647F28654894E3BD457199BE38DBBDC86092693E gpg --lsign ABAF11C65A2970B130ABE3C479BE3E4300411886 gpg --lsign 647F28654894E3BD457199BE38DBBDC86092693E

sir_lucjan commented on 2015-03-08 12:08 (UTC)

==> Verifying source file signatures with gpg... linux-3.19.tar ... Passed patch-3.19.1 ... Passed ==> Entering fakeroot environment... ==> Creating source package... -> Adding PKGBUILD... -> Generating .SRCINFO file... -> Adding config.x86_64... -> Adding config... -> Adding linux-ck.preset... -> Adding change-default-console-loglevel.patch... -> Adding install file (linux-ck.install)... -> Compressing source package... ==> Leaving fakeroot environment. ==> Source package created: linux-ck (Sun Mar 8 13:07:38 CET 2015)

sir_lucjan commented on 2015-03-08 11:52 (UTC)

@milan385 You're wrong. and

graysky commented on 2015-03-08 11:29 (UTC)

Grrr... sorry about that. I refreshed the PKGBUILD for the correct checksum. The issue was that it was in fact correct, but at that time I didn't push the local copy to the repo. I then manually did so when pedrogabriel posted but the sum in the PKGBUILD was reindex against the old version. Anyway, it's right now.

clfarron4 commented on 2015-03-08 11:01 (UTC)

@DaMoo @milan385: Have you downloaded the GCC patch fresh? It was updated to reflect the addition of the Silvermont family of processors.

DaMoo commented on 2015-03-08 01:29 (UTC)

I'm having an issue where a patch is failing to pass a sha256 integrity check. Here's the output from my AUR helper: ==> Validating source files with sha256sums... linux-3.19.tar.xz ... Passed linux-3.19.tar.sign ... Skipped patch-3.19.1.xz ... Passed patch-3.19.1.sign ... Skipped patch-3.19-ck1.xz ... Passed enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz ... FAILED config.x86_64 ... Passed config ... Passed linux-ck.preset ... Passed change-default-console-loglevel.patch ... Passed 0001-block-cgroups-kconfig-build-bits-for-BFQ-v7r7-3.19.patch ... Passed 0002-block-introduce-the-BFQ-v7r7-I-O-sched-for-3.19.patch ... Passed 0003-block-bfq-add-Early-Queue-Merge-EQM-to-BFQ-v7r7-for-3.19.0.patch ... Passed ==> ERROR: One or more files did not pass the validity check! :: failed to verify linux-ck integrity

graysky commented on 2015-03-07 16:25 (UTC)

Bump to v3.19.1-1 Comments: This release includes an updated gcc patch which allows for the optimization Silvermont processors. Changelog:

pedrogabriel commented on 2015-03-07 16:20 (UTC)

Thank you.

graysky commented on 2015-03-07 16:16 (UTC)

Fixed the checksum. Sorry about that.

pedrogabriel commented on 2015-03-07 15:55 (UTC)

linux-3.19.tar.xz ... Passed linux-3.19.tar.sign ... Skipped patch-3.19.1.xz ... Passed patch-3.19.1.sign ... Skipped patch-3.19-ck1.xz ... Passed enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz ... FAILED config.x86_64 ... Passed config ... Passed linux-ck.preset ... Passed change-default-console-loglevel.patch ... Passed 0001-block-cgroups-kconfig-build-bits-for-BFQ-v7r7-3.19.patch ... Passed 0002-block-introduce-the-BFQ-v7r7-I-O-sched-for-3.19.patch ... Passed 0003-block-bfq-add-Early-Queue-Merge-EQM-to-BFQ-v7r7-for-3.19.0.patch ... Passed ERROR: One or more files did not pass the validity check!

graysky commented on 2015-03-07 15:52 (UTC)

Bump to v3.19.1-1 Commit: LinuxChanges:

sir_lucjan commented on 2015-03-01 14:25 (UTC)


dkaylor commented on 2015-03-01 00:01 (UTC)

Downloading from has gone flaky again, can't grab patch-3.18-ck1.bz2

Lompik commented on 2015-02-28 21:34 (UTC)

gpg --recv-keys ABAF11C65A2970B130ABE3C479BE3E4300411886 647F28654894E3BD457199BE38DBBDC86092693E Solved.

sir_lucjan commented on 2015-02-28 21:27 (UTC)

@Lompik and

Lompik commented on 2015-02-28 21:24 (UTC)

Hello, got this error today: ==> Verifying source file signatures with gpg... linux-3.19.tar ... FAILED (unknown public key 79BE3E4300411886) ==> ERROR: One or more PGP signatures could not be verified! ==> ERROR: Makepkg was unable to build linux-ck. Rgds

graysky commented on 2015-02-28 00:10 (UTC)

Bump to v3.19-1 Commit: LinuxChanges:

graysky commented on 2015-02-27 20:54 (UTC)

Bump to v3.18.8-1 Changelog: Commit:

graysky commented on 2015-02-27 20:45 (UTC)

@protake - I see it. Just need to get 3.18.8 squared away and update all package for the 3.19 kernel. I will bump shortly.

sflor commented on 2015-02-27 17:02 (UTC)

BFS 461 is now available:

FadeMind commented on 2015-02-27 15:22 (UTC)

hepha. In PKGBUILD of linux-ck all is fine. [tomasz@arch ~]$ export LC_ALL=C;pacman -Qo modprobed-db /usr/bin/modprobed-db is owned by modprobed-db 2.26-1 [tomasz@arch ~]$ export LC_ALL=C;pacman -Ql modprobed-db modprobed-db /usr/ modprobed-db /usr/bin/ modprobed-db /usr/bin/modprobed-db modprobed-db /usr/bin/modprobed_db modprobed-db /usr/share/ modprobed-db /usr/share/licenses/ modprobed-db /usr/share/licenses/modprobed-db/ modprobed-db /usr/share/licenses/modprobed-db/LICENSE modprobed-db /usr/share/man/ modprobed-db /usr/share/man/man8/ modprobed-db /usr/share/man/man8/modprobed-db.8.gz modprobed-db /usr/share/modprobed-db/ modprobed-db /usr/share/modprobed-db/modprobed-db.skel modprobed-db /usr/share/zsh/ modprobed-db /usr/share/zsh/site-functions/ modprobed-db /usr/share/zsh/site-functions/_modprobed-db

hepha commented on 2015-02-27 15:16 (UTC)

hello /usr/bin/modprobed_db not /usr/bin/modprobed-db if [ -e /usr/bin/modprobed-db ]; then [[ ! -x /usr/bin/sudo ]] && echo "Cannot call modprobe with sudo. Install via pacman -S sudo and configure to work with this user." && exit 1 sudo /usr/bin/modprobed-db recall fi

graysky commented on 2015-02-17 20:58 (UTC)

Been following the discussion on the ck blog but am unwilling to update the package with non-ck patches.

AnAkkk commented on 2015-02-17 20:57 (UTC)

Apparently there is a patch to make BFS work on 3.19: Unofficial but CK already used patches from this guy (Alfred Chen) to make it work on new kernels before.

kyak commented on 2015-02-15 12:02 (UTC)

Thanks, graysky!

graysky commented on 2015-02-15 11:27 (UTC)

@kyak - See the repo-ck wiki for advice on dealing with godaddy's shitty internet services. I haven't considered a donate button; this is my way of giving back to the community.

kyak commented on 2015-02-15 06:57 (UTC)

Hi graysky, When using your repo, download if often aborted with "transfer closed with XXX bytes remaining to read" - due to slow speed. Previsously it was possible to just "pacman -Syu" again and the download would resume where it had been stopped. But now the download is always started from the beginning. I'm struggling now to download 10 Mb of linux-ck-ivybridge-3.18.7-2 - even for such small size, i had to "pacman -Syu" several times, untill it succeded (the download is always started from the beginning). Anyway. Did you think about "Donate" button that would probably help you improve the infrastructure a little bit?

graysky commented on 2015-02-14 22:06 (UTC)

Bump to v3.18.7-2 Changelog: Included ck's patch to correct locked plugged IO[1]. Commit: 1.

graysky commented on 2015-02-11 20:17 (UTC)

Bump to v3.18.7-1 Changelog: Commit:

sir_lucjan commented on 2015-02-11 09:11 (UTC)

3.18.7 is out

clfarron4 commented on 2015-02-09 14:30 (UTC)

@khron: Of course, it goes without saying that if you've done this that you should make note that you've done this. Good thing, as @sir_lucjan points out, is that hosts all the (current) versions of BFQ, so there's that as well.

khron commented on 2015-02-08 23:07 (UTC)

Easy way, just host your own mirror and use your favourite AUR tool: $ echo "" >> /etc/hosts $ mkdir ~/patches $ cd ~/patches $ mkdir -p people/paolo/disk_sched/patches/3.18.0-v7r7/ (download patches from there) $ python2 -m SimpleHTTPServer 80 $ pacaur -S linux-ck :)

graysky commented on 2015-02-07 12:30 (UTC)

I am hosting the BFQ patches until their site is back up. Access them like so: 1) Download the linux-ck source tarball, untar it, enter the linux-ck directory 2) sed -i 's|^_bfqpath=.*$|_bfqpath=\"\"|' PKGBUILD

Tjuh commented on 2015-02-07 11:34 (UTC)

The 0001-block-cgroups-kconfig-build-bits-for-BFQ-v7r7-3.18.patch fails to download during makepkg-ing, I've tried it numerous times.

graysky commented on 2015-02-06 20:12 (UTC)

Bump to v3.18.6-1 Changelog: Commit:

sir_lucjan commented on 2015-02-06 15:33 (UTC)

3.18.6 is out

sir_lucjan commented on 2015-02-02 19:36 (UTC)


clfarron4 commented on 2015-02-02 19:34 (UTC)

I think is down again...

graysky commented on 2015-01-30 20:29 (UTC)

Bump to v3.18.5-1 Changelog: Commit:

clfarron4 commented on 2015-01-28 17:33 (UTC)

@scabdates: Does that happen with the [core] kernel in the repositories as well, or just ck?

scabdates commented on 2015-01-28 15:52 (UTC)

The 3.18.4-1 update caused my init to stop working, just FYI.

graysky commented on 2015-01-27 19:15 (UTC)

Bump to v3.18.4-1 Changelog: Commit:

Xaero252 commented on 2015-01-18 02:53 (UTC)

@graysky: Yup, seems to be resolved here as well. Must have been some momentary downtime somewhere in the pipes preventing it. ¯\(°_o)/¯

graysky commented on 2015-01-17 19:49 (UTC)

@Xaero - No problems downloading sources here.

Xaero252 commented on 2015-01-17 18:23 (UTC)

Correction, all files appear to be failing from the mirror in the PKGBUILD, falling back on repo-ck until things are resolved.

graysky commented on 2015-01-17 14:06 (UTC)

prakharsingh95 commented on 2015-01-17 13:38 (UTC)


Tjuh commented on 2015-01-17 13:15 (UTC)

Trying to build this, but it fails at verifying some files: ==> Verifying source file signatures with gpg... linux-3.18.tar ... FAILED (unknown public key 79BE3E4300411886) patch-3.18.3 ... FAILED (unknown public key 38DBBDC86092693E) ==> ERROR: One or more PGP signatures could not be verified! ==> Removing installed dependencies...

graysky commented on 2015-01-17 12:13 (UTC)

Bump to v3.18.3-1 Changelog: Commit:

graysky commented on 2015-01-15 21:18 (UTC)

There are known issues with users of 3.18.x and the broadcom-wl-* driver.

neTpK commented on 2015-01-15 19:54 (UTC)

Stock kernel (3.18.2) causes major kernel lockups on my Asus laptop. tried with this kernel, same issues. Switched to linux-lts.

melanopsis commented on 2015-01-09 22:08 (UTC)

Just to let you know that my desktop no longer locks up with 3.18.2-1-ck

graysky commented on 2015-01-09 21:32 (UTC)

prakharsingh95 commented on 2015-01-09 21:32 (UTC)

@sir_lucjan Thanks! I came across this article but I was trying the keys present in PKGBUILD. Had to go through comments as well. It worked out.

sir_lucjan commented on 2015-01-09 21:09 (UTC)


prakharsingh95 commented on 2015-01-09 21:05 (UTC)

Hi, The installation broke with this new tarball. ==> Verifying source file signatures with gpg... linux-3.18.tar ... FAILED (unknown public key 79BE3E4300411886) patch-3.18.2 ... FAILED (unknown public key 38DBBDC86092693E) ==> ERROR: One or more PGP signatures could not be verified! I built the previous tarball for this version, and I have installed that, but not rebooted. I deleted all my previous builds. That should boot right? I mean sig files are just for verification.

graysky commented on 2015-01-09 19:56 (UTC)

Accidentally left the sig files for upstream source and patch commented in the PKGBUILD. I have just remedied this now without bumping the pkgrel.

graysky commented on 2015-01-08 19:50 (UTC)

Bump to v3.18.2-1 Changelog: Commit: Note: This PKGBUILD now conforms to the pacman-4.2 standards of being able to verify pgp signatures of upstream files. See the following for info on how to add the needed keys or how to disable this feature entirely,[1] 1.

graysky commented on 2015-01-04 14:21 (UTC)

I am now hosting the BFQ patches until their site is back up. Access them like so: 1) Download the linux-ck source tarball, untar it, enter the linux-ck directory 2) sed -i 's|^_bfqpath=.*$|_bfqpath=\"\"|' PKGBUILD

sir_lucjan commented on 2015-01-04 13:59 (UTC)

sp1d3rmxn commented on 2015-01-04 13:54 (UTC)

I'm lovin it: --------------------- -> Downloading 0001-block-cgroups-kconfig-build-bits-for-BFQ-v7r7-3.18.patch... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0Warning: Transient problem: HTTP error Will retry in 3 seconds. 3 retries Warning: left. 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0Warning: Transient problem: HTTP error Will retry in 3 seconds. 2 retries Warning: left. 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0Warning: Transient problem: HTTP error Will retry in 3 seconds. 1 retries Warning: left. 0 0 0 0 0 0 0 0 --:--:-- 0:00:59 --:--:-- 0curl: (22) The requested URL returned error: 503 Service Unavailable ==> ERROR: Failure while downloading 0001-block-cgroups-kconfig-build-bits-for-BFQ-v7r7-3.18.patch Aborting...

prakharsingh95 commented on 2015-01-03 14:32 (UTC)

Hi, I have been using 3.18.1-3-ck since Dec 26, and I have (thank God) not experienced any lockups, freezes, or wierd stuff. Can anyone elaborate on this issue, or steps to replicate as I would degrade my kernel version. I am using broadcom-wl-ck. I have an i7-3610QM.

fackamato commented on 2015-01-03 14:05 (UTC)

Black screen on sandy bridge even with nomodeset on i915. Single mode works. Any ideas?

graysky commented on 2014-12-23 15:59 (UTC)

As with every report of instability: does the corresponding ARCH kernel of the same version cause issues? In this case you will need to install 3.18.1-1-ARCH from [testing] to properly answer this question.

neTpK commented on 2014-12-23 14:20 (UTC)

Running linux-ck-haswell with the update to 3.18.1 locks up the cpu bad. First thought it was the update to overlayfs in PSD that was causing this, but it seems its a kernel issue. Changed to stock linux solves the issue.

graysky commented on 2014-12-22 18:49 (UTC)

Bump to v3.18.1-3 Changelog: Build against new BFQ patchset (v7r7). Commit: Note: I am not yet implimenting the linux-ck equivalent of 'verify source signatures[1]' due to a bug affecting mkaurball[2]. 1. 2.

sir_lucjan commented on 2014-12-22 15:09 (UTC)

melanopsis commented on 2014-12-19 05:43 (UTC)

@graysky - No, I am not using the broadcom-wl. It is most likely this issue

eduardoeae commented on 2014-12-19 01:50 (UTC)

Mine is an older Intel and also panics. I am using broadcom-wl

graysky commented on 2014-12-19 01:12 (UTC)

@melanopsis - Are you using the broadcom-wl driver by chance? 3.18.1-2-ck is rock solid for me...

melanopsis commented on 2014-12-19 01:11 (UTC)

Just a heads up, my ivy bridge desktop locks up pretty badly with this kernel. I had to downgrade to 3.17.7-1-ck...

graysky commented on 2014-12-18 09:57 (UTC)

Bump to v3.18.1-2 Changelog: Fix FS#43143 Commit:

graysky commented on 2014-12-17 18:41 (UTC)

Bump to v3.18.1-1 Changelog: Commit: LinuxChanges:

graysky commented on 2014-12-17 17:28 (UTC)

@mareex - I saw it on my mobile. Thanks. Will make the switch shortly.

mareex commented on 2014-12-17 16:51 (UTC)

linux 3.18.1-1 is in testing now.

graysky commented on 2014-12-16 20:00 (UTC)

Bump to v3.17.7-1 Changelog: Commit:

graysky commented on 2014-12-16 19:50 (UTC)

Update3: 3.18.1 has just been released and I have updated the source tarball for those of you wishing to compile it up on your own. A reminder that all of the other *-ck packages I maintain have now been updated and are also available on repo-ck BOTH as source tarballs and (for generic x86_64 only) as compiled packages - note that the compiled package of 3.18.1-ck-1 will be updated shortly. *Source tarballs: *Pre-build 3.18-ck packages (x86_64 only):

graysky commented on 2014-12-15 19:12 (UTC)

Update2: I now have all packages updated to build against 3.18-ck-1 including broadcom-wl-ck. Find the source package for it in the same place as I indicated below. Note - I do not have a broadcom wireless chipset so I cannot test the module. I can modprobe it and it does insert. Can someone with this hardware try it and report back?

graysky commented on 2014-12-14 17:51 (UTC)

Update: All packages except for broadcom-wl-ck have been updated to work with 3.18.0-1-ck. Again, I will wait to diff the linux-ck PKGBUILD and associated files against the official ARCH version once it publishes, but those wishing to upgrade to 3.18.0-1-ck now can either build your own from the src tarballs or use pre-build the packages you can download directly with `pacman -U` if you wish. Otherwise, just keep waiting for the official release. *Source tarballs: *Pre-build 3.18-ck packages (x86_64 only):

graysky commented on 2014-12-14 03:05 (UTC)

While we wait the ARCH devs to create a proper PKGBUILD for 3.18.0, I have pieced together one for those wishing to test out the package in the following directory of repo-ck: Note that I have simply done a `make oldconfig` against 3.17.6-1's config files and selected the modules for new features unless that wasn't an option in which case I went with the upstream defaults for new features. Also know that I am noting planning to do an early update to the accompanying packages (broadcom, nvidia*, and vbox*) so if you have a system that depends on them, you will have to modify them yourself until 3.18.x-1-ARCH goes live. Enjoy!

graysky commented on 2014-12-11 18:19 (UTC)

Got em. Just need to wait for the ARCH dev team to push their config and PKGBUILD etc. into svn. I unflagged it; please flag it out of date again once 3.18-1 is disclosed[1]. 1.

theaifam5 commented on 2014-12-11 17:16 (UTC)

sir_lucjan commented on 2014-12-11 09:52 (UTC)

graysky commented on 2014-12-11 00:20 (UTC)

Samsagax - why out of date?

clfarron4 commented on 2014-12-10 12:28 (UTC)

@AnAkkk: There's also the issue about which extra kernels should be provided in the main ArchLinux repositories (which is always a fun one).

graysky commented on 2014-12-09 21:08 (UTC)

Did that a while back and was not voted in; limited time for it now... what's wrong with [repo-ck]? I have been maintaining it since 2011. See the wiki page for more if you haven't already.

AnAkkk commented on 2014-12-09 21:03 (UTC)

Can't you ask to become a TU and do that yourself ? :)

graysky commented on 2014-12-09 20:54 (UTC)

Oh... that's not up to me. I believe a TU would need to adopt the package and bring it in toe [community]. No way would he or she build all the CPU versions that I supply in [repo-ck] though...

AnAkkk commented on 2014-12-09 20:53 (UTC)

I think what he is saying is that you could have your linux-ck package in community maybe, like linux-grsec.

graysky commented on 2014-12-09 20:51 (UTC)

The dev team will not incorporate an out-of-tree scheduler into the official distro kernel package. That's why we have the AUR.

dradec commented on 2014-12-09 19:46 (UTC)

Why is this not in main?

graysky commented on 2014-12-08 10:34 (UTC)

Bump to v3.17.6-1 Changelog: Commit:

graysky commented on 2014-12-07 12:46 (UTC)

Bump to v3.17.5-1 Changelog: Commit:

graysky commented on 2014-11-28 11:59 (UTC)

@Det - You're welcome. Enjoy.

Det commented on 2014-11-28 11:54 (UTC)

Just came to express my appreciation for your beautiful prebuilt packages. I was already tired of constantly rebuilding my own local kernel, and why should I, when we have such a wonderful man like you. That's all. Remember you're a force for good.

graysky commented on 2014-11-21 21:40 (UTC)

Bump to v3.17.4-1 Changelog: Commit:

graysky commented on 2014-11-18 20:20 (UTC)

Bump to v3.17.3-3 Changelog: Bump to latest ck patchset: ck2 Commit:

graysky commented on 2014-11-18 10:34 (UTC)

All - I know ck2 has been released. I will update this package later this afternoon.

graysky commented on 2014-11-14 23:54 (UTC)

Bump to v3.17.3-2 Changelog: Added fix FS#42689 Commit:

graysky commented on 2014-11-14 21:36 (UTC)

Bump to v3.17.3-1 Changelog: Commit:

graysky commented on 2014-11-11 22:48 (UTC)

Bump to v3.17.2-1 Changelog: Commit: LinuxChanges:

jbanks commented on 2014-11-11 12:45 (UTC)

linux-3.17-ck1 is out.

graysky commented on 2014-10-30 18:53 (UTC)

Bump to v3.16.7-1 Changelog: Commit:

graysky commented on 2014-10-23 19:35 (UTC)

Dunno how long as CK does this in his spare time. For reference: BFS for v3.16 = 13 days BFS for v3.15 = 26 days BFS for v3.14 = 36 days BFS for v3.13 = 18 days BFS for v3.12 = 15 days BFS for v3.11 = 7 days BFS for v3.10 = 10 days BFS for v3.9 = 8 days BFS for v3.8 = 13 days BFS for v3.7 = 4 days BFS for v3.6 = 12 days BFS for v3.5 = 10 days BFS for v3.4 = 12 days

AnAkkk commented on 2014-10-23 19:28 (UTC)

I guess we'll have to wait a long time then :/

graysky commented on 2014-10-23 19:27 (UTC)

I'll update to the 3.17 series once ck officially releases the patch. I don't think many users would want a user-created bfs.

AnAkkk commented on 2014-10-23 09:35 (UTC)

There's a patch to make BFS work on 3.17: Any chance you can update to 3.17? :)

graysky commented on 2014-10-23 00:13 (UTC)

Bump to v3.16.6-3 Changelog: enable early microcode updates Commit:

graysky commented on 2014-10-19 23:45 (UTC)

Bump to v3.16.6-2 Changelog: new version of BFQ Commit:

sir_lucjan commented on 2014-10-19 22:56 (UTC)


artafinde commented on 2014-10-17 16:38 (UTC)

Agree, thanks.

graysky commented on 2014-10-17 16:03 (UTC)

Not a linux-ck issue. Please take discussion to the bbs so as not to clutter up the AUR comments.

artafinde commented on 2014-10-17 15:37 (UTC)

I believe it's an issue with ccache and btrfs on SSD. When I disable the ccache it works. The problem is I'm getting a kernel panic so compilation stops half-way. After hard-reset I clean out the src and pkg directory and restart the compilation. Then I am assuming the objects are half created and I hit again a compilation error (not kernel-panic). It's an obscure case so comments in AUR might not be the proper place to solve this.

clfarron4 commented on 2014-10-17 08:32 (UTC)

@artafinde: What is the exact error? Might help us work out what the problem actually is.

artafinde commented on 2014-10-17 06:30 (UTC)

I get a compilation error while using 8 threads and ccache. Anyone else can verify?

graysky commented on 2014-10-16 19:39 (UTC)

@mareex - Of course, but not until the corresponding intel-ucode package (currently 20140913-1) makes it into [extra] as there are key changes therein.

mareex commented on 2014-10-16 19:30 (UTC)

Will linux-ck enable early microcode updates like the Arch stock kernel?

graysky commented on 2014-10-15 22:07 (UTC)

Bump to v3.16.6-1 Changelog: Commit:

Tjuh commented on 2014-10-10 18:28 (UTC)

Fixed now, thanks. I ran a new modprobed_db with the external hdd on and recompiled.

prakharsingh95 commented on 2014-10-10 18:00 (UTC)

@Tjuh. I faced a similar issue. Just boot into arch default, plug in all your peripherals and get them working, and then build linux-ck with localmodconfig=y. Everything will work out of the box them. @graysky. Is it possible for you to enable SCSI, USB, etc by default (ie add an option to PKGBUILD like minimal_support=y)?

graysky commented on 2014-10-10 16:25 (UTC)

Odds are that using that option, you did not compile in USB support. I suggest using modprobe-db if you haven't already done so. Boot into a kernel with all modules, load those you will need, ie insert the external USB drive, populate the db and try rebuilding.

Tjuh commented on 2014-10-10 13:34 (UTC)

Yes, localmodcfg=y in the PKGBUILD. The latest linux-ck-haswell kernel works just fine from the repo-ck. The previous kernel v3.16.4-1 works just fine with the same PKGBUILD settings.

graysky commented on 2014-10-10 13:24 (UTC)

Did you modify the PKGBUILD or build with localmodconfig or the like? What happens if you use one of the kernel packages from repo-ck?

Tjuh commented on 2014-10-10 10:20 (UTC)

After compiling the latest linux-ck my external usb hdd is not being recognized anymore. I had no problem with the previous kernel.

graysky commented on 2014-10-09 20:30 (UTC)

Bump to v3.16.5-1 Changelog: Commit:

synergy commented on 2014-10-08 06:26 (UTC)

@AnAkkk What you're most likely witnessing isn't a bug. By design, BFQ sacrifices throughput for interactivity.

prakharsingh95 commented on 2014-10-08 05:18 (UTC)

@nemesys and @clfarron4 I had to manually enable USB SCSI option. I was just surprised that localmodcfg skipped this.

nemesys commented on 2014-10-07 18:46 (UTC)

@prakharsingh95: Please read the kernel documentation for what different build options do before attempting to install a custom kernel. The fact that you are compiling the main working software component on your system should be warning enough to RTFM.

clfarron4 commented on 2014-10-07 18:39 (UTC)

@prakharsingh95: A warning for what? I know there is a clear warning for what localmodcfg does and how it will only compile probed modules in the PKGBUILD.

graysky commented on 2014-10-05 22:18 (UTC)

Bump to v3.16.4-1 Changelog: Commit:

graysky commented on 2014-09-28 13:57 (UTC)

If you feel strongly enough about it, report your findings upstream:

AnAkkk commented on 2014-09-28 12:40 (UTC)

I haven't timed, but it's quite noticeable. I've tried with and without elevator=bfq, my KDE session opens pretty much instantly without, and there's a delay with BFQ.

graysky commented on 2014-09-28 12:02 (UTC)

No slower here. Have you actually timed between boots?

AnAkkk commented on 2014-09-28 09:35 (UTC)

Has anyone else noticed slower boot times when BFQ is enabled (on a SSD)? My KDE session takes a longer time to load with BFQ.

coderkun commented on 2014-09-19 15:04 (UTC)

@graysky: Thanks, works fine now.

graysky commented on 2014-09-18 18:54 (UTC)

Sorry gang, my bad. Fixed now, but did not bump the pkgver to avoid rebuilding the repo (there is no need to do so).

coderkun commented on 2014-09-18 12:30 (UTC)

@clfarron4: I get the same sha256sum as you but the PKGBUILD contains a different one (4c44ad820fae8afaf3ad55fa7b191cc15e61829293a074126be780a35275b7c6).

clfarron4 commented on 2014-09-18 11:13 (UTC)

@coderkun: What do you get as the sum for that file? I get this: % sha256sum enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz a9ca0e6fc01a3d34058b6cc9cdb560f62233f2eca5917db65dc0b29772bb236e enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz

coderkun commented on 2014-09-18 07:35 (UTC)

Checksum of “enable_additional_cpu_optimizations_for_gcc_v4.9+_kernel_v3.15+.patch.gz” does not match.

graysky commented on 2014-09-17 19:37 (UTC)

Bump to v3.16.3-1 Changelog: Commit:

graysky commented on 2014-09-06 11:14 (UTC)

Bump to v3.16.2-1 Changelog: Commit:

graysky commented on 2014-08-25 00:24 (UTC)

Bump to v3.16.1-2 Changelog: Update to ck2 with bfs v456 and fix FS#41683 (intel ringbuffer). Commit:

AnAkkk commented on 2014-08-20 15:51 (UTC)

According to the kernel config, 1000 Hz is best for desktop systems anyway, so it might be better to just leave it like that.

graysky commented on 2014-08-18 12:18 (UTC)

I haven't checked running at 300 Hz. If you would like to try and report back that would be great. AFAIK ck recommends 1000 Hz.

AnAkkk commented on 2014-08-18 10:48 (UTC)

Haven't the suspend issues been solved since a long time now? 1000 Hz should no longer be required? Aren't there some downsides from using 1000Hz instead of 300Hz?

dkaylor commented on 2014-08-16 18:22 (UTC)

OK. As always, thanks for the work you do on linux-ck, been using it for years now.

graysky commented on 2014-08-16 18:14 (UTC)

No idea about the answer to your SMT question. I just use modprobed-db on my own system and disable a few things that I don't use, NUMA, virtualization, and hibernation.

dkaylor commented on 2014-08-16 18:12 (UTC)

Thanks graysky. Which leads me to a question - SMT nice aside, if I have a non-SMT processor (4 cores = 4 threads, nice and simple) does disabling SMT in the kernel config accomplish anything, other than saving a little overhead? What's your opinion on heavily customized kernel configs in general?

graysky commented on 2014-08-16 17:55 (UTC)

No, I thought about this... it's currently on (ie hardcoded in the .config files). You will need to manually disable it. I could add this to the PKGBUILD as an option, but have no idea how many users would even use it.

dkaylor commented on 2014-08-16 17:53 (UTC)

so as bfs 450 now includes the smt nice work, will I be able to turn it off in the kernel config? I have a non-hyperthreaded core2 quad, so I always turn off SMT scheduling, although I'm not sure it matters much. ck mentions a bfs450-nosmt-buildfix.patch to make it impossible to enable SMT nice on a non-SMT kernel, but not sure it's included in the PKGBUILD. I'm fine with just disabling it in the config.

graysky commented on 2014-08-16 10:09 (UTC)

Bump to v3.16.1-1 Changelog: Commit:

sir_lucjan commented on 2014-08-16 10:03 (UTC)

patch-3.16-ck1 is out!

graysky commented on 2014-08-15 19:16 (UTC)

@sir_lucjan - I see it. Updated but did not bump pkgver since rebuilding the repo packages is silly with this defaulted to off.

sir_lucjan commented on 2014-08-15 17:50 (UTC)

@graysky: smtnice 6 is out

sir_lucjan commented on 2014-08-15 08:58 (UTC)


graysky commented on 2014-08-15 06:46 (UTC)

@ishitatsuyuki - Why out-of-date?

graysky commented on 2014-08-15 06:46 (UTC)

Bump to v3.15.10-1 Changelog: Commit:

graysky commented on 2014-08-08 06:58 (UTC)

Bump to v3.15.9-2 Changelog: Fixed _kernver mistake in 3.15.9-1 Commit:

graysky commented on 2014-08-08 00:50 (UTC)

Bump to v3.15.9-1 Changelog: Commit:

clfarron4 commented on 2014-08-05 21:46 (UTC)

Looks like it's back up now.

ilkyest commented on 2014-08-04 03:09 (UTC)

yeah it's down... have you seen something about? another link to put on "pkgbuild"

ilkyest commented on 2014-08-04 01:51 (UTC)

Graysky... patch from con kolivas, resulting in timeout -> patch-3.15-ck1.bz2...

ilkyest commented on 2014-08-04 01:40 (UTC)

Thanks so much by the kernel Graysky. I'm a folding at home user (, that try improves F@H performance, every time trying new kernels. Sometime ago, about 3.10.1 your kernel was not good to folding. At this time, only in one test, it were the worst, I'm trying NUMA=ON and NUMA=OFF differences, and, withing your changes, and "kernel's studying" in a while, only ck will fold on my pc. But... when you will bump the pkgver? Any date?

graysky commented on 2014-08-01 13:42 (UTC)

Note that CK just posted an experimental patch that in his words, "Works for me flawlessly on all my stuff, but it's still pretty new." You can read about it at the link below. I don't want to enable it by default in the repo but do want to allow users a chance to try it out in the PKGBUILD. See the first set of comments to enable it. As always, post your experiences both good and bad to CK's blog so he can make ck1 and BFS more and more powerful and robust for everyone to enjoy. Note-I am not going to bump the pkgver to 2 since I just rebuild the repo overnight and since this is really no chance unless you enable it.

graysky commented on 2014-08-01 02:36 (UTC)

Bump to v3.15.8-1 Changelog: Commit:

graysky commented on 2014-07-30 19:22 (UTC)

Yeah, I saw them and in the past CK has considered the patch experimental and as noted in his blog post, "this patch by itself does nothing unless other code uses the locks." I don't want to add an experimental patch to the package must less to repo packages people depend on for stable systems.

clfarron4 commented on 2014-07-30 19:10 (UTC)

I assume the flagger saw a new post on the CK blog and flagged out-of-date without reading. That said, the new post on the CK blog refers to these patches: Just after I build 3.15.7-1-ck-pax as well (with no time left today or tomorrow to muse over these new patches).

graysky commented on 2014-07-28 20:07 (UTC)

Bump to v3.15.7-1 Changelog: Commit:

clfarron4 commented on 2014-07-25 11:45 (UTC)

And is back up again.

graysky commented on 2014-07-24 18:59 (UTC)

I am now hosting the BFQ patches until their site is back up. Access them like so: 1) Download the linux-ck source tarball, untar it, enter the linux-ck directory 2) sed -i 's|^_bfqpath=.*$|_bfqpath=\"\"|' PKGBUILD

fx101 commented on 2014-07-24 15:23 (UTC)

For whatever reason, is offline. Any mirrors?

graysky commented on 2014-07-19 02:24 (UTC)

Bump to v3.15.6-1 Changelog: Commit:

mareex commented on 2014-07-18 09:41 (UTC)

Out of date Linux 3.15.6 is out on

graysky commented on 2014-07-11 20:46 (UTC)

Bump to v3.15.5-2 Changelog: add patch for #33745 efistub breakage Commit:

graysky commented on 2014-07-10 20:04 (UTC)

Bump to v3.15.5-1 Changelog: Commit:

mareex commented on 2014-07-09 19:30 (UTC)

Out of date: 3.15.5 is out on

graysky commented on 2014-07-08 10:43 (UTC)

Bump to v3.15.4-1 Changelog: Changelog: bfs v0.449 update Commit:

clfarron4 commented on 2014-07-07 14:57 (UTC)

I see it now: It would be nice for people to post here to say what's out of date. Also, 3.15.4 is out now.

dkaylor commented on 2014-07-07 03:15 (UTC)

I'm not the one flagging it out of date, but it's probably because ck released BFS 449. graysky stated earlier that he was busy with family stuff this weekend, so he just hasn't had a chance to do 3.15.3-2 yet.

fx101 commented on 2014-07-06 20:48 (UTC)

Who keeps flagging this out of date? lists 3.15.3 as the latest release.

clfarron4 commented on 2014-07-05 22:45 (UTC)

How is this out-of-date? Seriously, we need functionality like the main package website so that we understand how it is out-of-date.

graysky commented on 2014-07-02 23:53 (UTC)

Bump to v3.15.3-1 Changelog: Commit: Note - I just pushed the 3.15.3-1 release series but will not have time for any updates until after the holiday due to family being in town.

graysky commented on 2014-07-01 19:08 (UTC)

Bump to v3.14.10-1 Changelog: Commit: Note - CK should be releasing 3.15-ck1 shortly. We are having family come out to visit for the 4th of July holiday which may prevent me from updating all the linux-ck packages (nvidia, broadcom, vbox, etc.) to the 3.15 release until after they leave.

sir_lucjan commented on 2014-07-01 09:30 (UTC)

3.14.10 is out!

graysky commented on 2014-06-26 21:45 (UTC)

Bump to v3.14.9-1 Changelog: Commit:

graysky commented on 2014-06-20 19:24 (UTC)

Bump to v3.14.8-2 Changelog: BFQ updated to v7r5 (thanks for the note clfarron4) Commit:

clfarron4 commented on 2014-06-20 18:25 (UTC)

Oop, it looks like BFQ v7r5 is out!

illis commented on 2014-06-19 02:19 (UTC)

@clfarron4: Sorry this could actually be something on my end. Tried downloading the file another few times and was still getting the mismatched hash. Looked at it a bit further and it turns out that the file was in plain text (not gz) - somewhere along the line it was being uncompressed automatically. Have a feeling it may be something to do with my ISP as I teathered up my mobile issues the same curl -O ... command, and got the correct response.

clfarron4 commented on 2014-06-18 12:12 (UTC)

@illis: I just downloaded it fresh, and I get this: sha256sum enable_additional_cpu_optimizations_for_gcc_v4.9+.patch.gz c6c4a9f77683b95c37636b20c4bc8a1f8214c87feef7fc469e58534fcc32fb4a The sha256sum in the PKGBUILD is correct.

illis commented on 2014-06-17 23:03 (UTC)

sha256 match seems to be failing for the gcc patch. sha256sum enable_additional_cpu_optimizations_for_gcc_v4.9+.patch.gz 4883a4d45fdcb02600af976be8b9b89f459775ac8916712dfdefd29419c3eacf enable_additional_cpu_optimizations_for_gcc_v4.9+.patch.gz

turtil commented on 2014-06-17 21:31 (UTC)

@graysky, @clfarron4, thanks, i removed the src/, and it worked without a charm, must have been during my couple of goes trying to get it up and running, and it patching itself a couple of times or something. Thanks, all working now!

clfarron4 commented on 2014-06-17 17:35 (UTC)

@turtil: Either the below comment or you're building in a directory that needs to the /src and /pkg folders to be removed.

graysky commented on 2014-06-17 08:05 (UTC)

@turtil - Perhaps you have a cached version of the source or something that needs to be cleared? Check your /etc/makepkg.conf

Schmeidenbacher commented on 2014-06-17 07:33 (UTC)

Just compiled this kernel, went through without an issue.

turtil commented on 2014-06-16 23:50 (UTC)

patching file include/uapi/linux/sched.h patching file include/linux/sched/rt.h patching file kernel/stop_machine.c patching file drivers/cpufreq/cpufreq_conservative.c patching file kernel/time/Kconfig patching file kernel/sched/Makefile The next patch would create the file kernel/sched/bfs_sched.h, which already exists! Skipping patch. 1 out of 1 hunk ignored patching file kernel/sched/stats.c patching file arch/ia64/kvm/kvm-ia64.c patching file arch/powerpc/kvm/book3s_hv.c Hunk #1 succeeded at 1509 (offset -2 lines). patching file arch/x86/Kconfig patching file arch/x86/kvm/x86.c Hunk #1 succeeded at 6114 (offset -1 lines). patching file kernel/Kconfig.preempt patching file kernel/Kconfig.hz patching file Makefile ==> ERROR: A failure occurred in prepare(). Aborting... Generally speaking, sure this is an issue with the Makefile being patched, has anyone else had this issue? If its just me, I dont mind trying to patch the patch, and provide back.

graysky commented on 2014-06-16 22:03 (UTC)

Bump to v3.14.8-1 Changelog: Commit:

clfarron4 commented on 2014-06-16 21:56 (UTC)

mariojuniorjp commented on 2014-06-14 00:15 (UTC)

@Schemeidenbacher I changed the folder for the compilation. Now I have to wait to do it again, and see if will work this time. Many thanks for all the help! =D

Schmeidenbacher commented on 2014-06-14 00:02 (UTC)

Or you uncomment the BUILDDIR stanza in /etc/makepkg.conf and set it to a place with more space.

mariojuniorjp commented on 2014-06-14 00:00 (UTC)

@Schemeidenbacher 13.9 MB of 908.7 MB (98% used) That was it. Now I have to figure out how to increase the size of / tmp folder, because what is taking up all the space is precisely the kernel compilation. xD

Schmeidenbacher commented on 2014-06-13 23:54 (UTC)

It's building in /tmp/makepkg then (the default), so yeah, depends on what's mounted there.

mariojuniorjp commented on 2014-06-13 23:52 (UTC)

@graysky @Scimmia I'm building using Yaourt. :P I did not know it would cause an error in compiling the kernel that way. I've only set up the PKGBUILD, other things too, and then sent compile the kernel. @Schemeidenbacher #-- Specify a directory for package building. #BUILDDIR=/tmp/makepkg Is this?

Schmeidenbacher commented on 2014-06-13 23:36 (UTC)

@mariojuniorjp: Also check the file /etc/makepkg.conf for the BUILDDIR stanza. That will tell you where makepkg is building it. Check for space there.

Scimmia commented on 2014-06-13 23:35 (UTC)

The fact that it's asking you if you want to restart means you're using a helper of some sort. From the looks of it, I would say yaourt. By default, yaourt builds in /tmp, which is on tmpfs. That's your problem.

graysky commented on 2014-06-13 23:32 (UTC)

Seems like something thinks otherwise... where are you building (type pwd in the build dir)... also, are you using an aur helper? Don't do that. Just try makepkg -srci and see if that gives problems.

mariojuniorjp commented on 2014-06-13 23:31 (UTC)

I'm getting an error on the part of the kernel build, and do not know how to solve. The error is this one: patch: **** write error : No space left on device ==> ERROR: A failure occurred in prepare(). Aborting... ==> ERROR: Makepkg was unable to build linux-ck. ==> Restart building linux-ck ? [y/N] Have plenty of space on the partition, then do not know what the logic of this error. Oo

graysky commented on 2014-06-11 20:40 (UTC)

Bump to v3.14.7-1 Changelog: Commit:

graysky commented on 2014-06-11 20:40 (UTC)

I think it's because repo-ck moves lots of data and they [godaddy] want me to get sick of them and drop my unlimited service they have me in at an old rate. They have offered to upgrade me to a dedicated server for higher reliability several times now for a large increase in my monthly fee. Each time I remind them that my plan is unlimited and that they can upgrade me without a charge. Bastards. I am working slowly with barrikad to setup a mirror on his service. I have been very busy with work/family shit and traveling though. I am booked this weekend as well but hope to do some testing and setup with him shortly.

dlh commented on 2014-06-09 10:41 (UTC)

snack commented on 2014-06-09 10:12 (UTC)

@clfarron4: I don't understand your comment about "godaddy" (english is not my native speaking, sorry). I am indeed able to resume the download, but it's a bit annoying when you have to launch pacman 4-5 times to update the system...

clfarron4 commented on 2014-06-09 10:11 (UTC)

@snack: That's godaddy being godaddy. You should be able to recommence the download with pacman and just repeat until it finishes.

snack commented on 2014-06-09 07:21 (UTC)

Since some days I get this kind of error while retrieveing packages from repo-ck: error: failed retrieving file 'linux-ck-core2-3.14.6-2-x86_64.pkg.tar.xz' from : transfer closed with 55761293 bytes remaining to read Am I the only one?

graysky commented on 2014-06-08 14:25 (UTC)

Bump to v3.14.6-2 Changelog: Commit:

sir_lucjan commented on 2014-06-08 11:21 (UTC)

dlh commented on 2014-06-08 10:53 (UTC)

Thanks for your work, I'll check that patch in the meantime.

graysky commented on 2014-06-08 10:42 (UTC)

OK. I see that 3.14.6-1 is now in svn. I have further adjusted the gcc patch and am building 3.14.6-2 now. I will upload the source tarball to the AUR in several hours once I see that the various new arches using the patch build just fine. For those wanting a tarball early, find it here[1] and please be sure to post to the AUR your experiences building (good or bad) and for which arch! 1.

graysky commented on 2014-06-08 00:10 (UTC)

OK... I am building with the new patch but will not index the repo until I get some feedback that everything is ok from you guys. So: 1) If you want to build 3.14.6-1-ck yourself, download the source tarball here.1 2) If you want to try a repo-ck package, just browse manually to and navigate to either x86_64 or i686 and manually download the package for your processor. Remember that I am building now and will not have a complete set for another 6 hours or so. 1.

graysky commented on 2014-06-07 23:25 (UTC)

Ivy works for me... no access to older stuff that I can mess with at this point in time.

graysky commented on 2014-06-07 23:06 (UTC)

I need some savvy users of newer intel cpus (any core ix like nehalem, sandy, ivy, or haswell) to try out the new patch I have placed in my 'unstable' branch. Just substitute the line in this PKGBUILD for it. Please build the package selecting the nconfig option for your hardware, boot into it and let me know that my stuff works.

graysky commented on 2014-06-07 21:27 (UTC)

I know that 3.14.6 was just released but I want to wait to make the linux-ck release to see what tpowa does with the official ARCH package as it seems a few new config options are introduced. Please do not flag linux-ck out-of-date until the ARCH 3.14.6-1 is pushed to svn or repo. Thanks!

graysky commented on 2014-06-07 19:37 (UTC)

I see now that this is confirmed: I will need to adjust the patch accordingly.

graysky commented on 2014-06-07 19:29 (UTC)

Interesting that with the new version of gcc, these "older" options for march and mtune seem to have been made more generic. For example, on my ivy: % gcc -c -Q -march=native --help=target | grep march -march= ivybridge Older gcc version returned -march= core-avx-i

graysky commented on 2014-06-07 19:23 (UTC)

@dlh - For haswell: CONFIG_MCOREAVX2 gets set which uses -march=core-avx2 -mtune=core-avx2. See:

dlh commented on 2014-06-07 12:35 (UTC)

@elektorronikci: I have G3220 and with native I have -mno-avx -mno-avx2 But you are re saying that linux-ck-haswell is compiled with those options, right? That is why I am unable to run it, correct me if I am wrong?

elektorronikci commented on 2014-06-06 17:07 (UTC)

@sash-arch, your CPU doesn't have AVX2 set:

elektorronikci commented on 2014-06-06 17:01 (UTC)

I installed your ck-haswell compilation, now it is running. Not much time running, so not tested enough I think, but it really seems fine. If any problem happens, I will acknowledge.

graysky commented on 2014-06-06 09:59 (UTC)

As a test can either of you guys [tweeserhead or elektorronikci] try the ck-haswell packages on your haswell systems to confirm the problem is not with them + the hardware?

elektorronikci commented on 2014-06-06 08:16 (UTC)

I have i7-4770. I don't use ck-haswell repo but compile with -march=native. I use bfs&bfq. (bfqv7r3 had some issues before) There is no problem.

walkingrobot commented on 2014-06-06 00:04 (UTC)

Sorry, I forgot to mention it's a haswell processor

walkingrobot commented on 2014-06-06 00:02 (UTC)

Well this kinda misses the point but I compile with -march=native and CONFIG_MNATIVE

graysky commented on 2014-06-03 19:03 (UTC)

Not sure what to tell you guys... are other haswell users out there using the ck-haswell repo packages without issue?

dlh commented on 2014-06-03 14:03 (UTC)

@sash-arch I have the same problem as you, therefore I switched to generic linux-ck package.

sash-arch commented on 2014-06-02 22:36 (UTC)

Haswell... sascha@server ~ % gcc -c -Q -march=native --help=target | grep march -march= haswell

graysky commented on 2014-06-02 22:35 (UTC)

Not sure what to make of that, sash. Ivy gets CONFIG_MCOREAVXI which haswell gets CONFIG_MCOREAVX2 which are the correct flags. What does the following return for you? gcc -c -Q -march=native --help=target | grep march

sash-arch commented on 2014-06-02 22:30 (UTC)

I have an Celeron G1840 CPU. It's an "Haswell" CPU but the Linux-ck-haswell Package doesn't work with CPU. It stops working while loading the initial ramdisk. If i switch to linux-ck-ivybridge, everything works fine.

graysky commented on 2014-05-31 22:42 (UTC)

Bump to v3.14.5-1 Changelog: Commit:

janck commented on 2014-05-30 20:21 (UTC)

Thank you very much:)

graysky commented on 2014-05-30 18:32 (UTC)

Just refreshed 3.14.4-2 which is a proper split package now that the AUR supports them. @janck - Your suggestion is live now.

koshak commented on 2014-05-30 07:11 (UTC)

@graysky Sorry, seems like local proxy server gives me unzipped version of this patch for some reason, so sha256sum is different and its contents is plain-text... On-the-fly PKGBUILD editing fixed this particular issue.

graysky commented on 2014-05-29 19:45 (UTC)

Let me take them in order: clfarron4 - Yes, that patch is needed for building with SMP but this is the first time I can remember it coming up. In the interest of simplicity when I merge this PKGBUILD with the official ARCH PKGBUILD on major version bumps, I want to minimize the potential for errors on my part and thus am rejecting the request to add the SMP patch. CK may roll it into his patchset which is fine, but there aren't many people building this package on a single core system. @janck - I see, you're saying that the modification is needed to keep the build from FAILING in the case of you stripping out the extraneous config options. OK. I will implement your fix and for my sanity comment the PKGBUILD as such so I don't nuke it on an major version merge. @koshak - The sums match for me, not sure why your sums are different. I also don't understand your comment about the gz being txt. Are you editing with something smart like vim? For example, vim will decompress the gz when you open the gz transparently. I hope that addresses everyone's concerns.

docwild commented on 2014-05-29 13:20 (UTC)

as clfarron4 mentioned below I suggested a CONFIG_SMP patch on the forum repo topic by mistake. That post is now gone but I'll put it here: Adding [code] if grep --quiet "# CONFIG_SMP is not set" "${startdir}/config.last" 2&>0 /dev/null || grep --quiet "# CONFIG_SMP is not set" "${startdir}/config.x86_64.last" 2&>0 /dev/null then msg "Patch NOSMP patch" patch -Np1 -i "${srcdir}/ck-nosmp.patch" fi [/code] at the very end of the prepare() function, after populating the config.old file allows users with uniprocessors to disable symmetric multiprocessing support in kernel config (it has some overhead). The "ck-nosmp.patch" file here: Mentioned on cks blog here:

koshak commented on 2014-05-29 09:40 (UTC)

wrong sum for prepare() is looking for enable_additional_cpu_optimizations_for_gcc.patch, which is absent. instead enable_additional_cpu_optimizations_for_gcc.patch.gz is present but it is not a gzip file. It is pure text. WTF?

janck commented on 2014-05-29 09:39 (UTC)

After removing all dvb shit from kernel config, include/config/dvb does not exist in the source tree anymore, and consequently the build fails at line 432. This fixes my particular problem (have been testing that it works this time): if [ -d include/config/dvb/ ]; then mkdir -p "${pkgdir}/usr/lib/modules/${_kernver}/build/include/config/dvb/" cp include/config/dvb/*.h "${pkgdir}/usr/lib/modules/${_kernver}/build/include/config/dvb/" fi

clfarron4 commented on 2014-05-29 09:29 (UTC)

Some has suggested this in the BBS, to fix something to do with SMP:

graysky commented on 2014-05-28 22:22 (UTC)

@janck - But again, the PKGBUILD as-is should accomplish this: mkdir -p "${pkgdir}/usr/lib/modules/${_kernver}/build/include/config/dvb/" cp include/config/dvb/*.h "${pkgdir}/usr/lib/modules/${_kernver}/build/include/config/dvb/"

janck commented on 2014-05-28 22:13 (UTC)

Found a fix from a previous build. # this line will allow the package to build if a user disables the dvb shit find include/config/dvb -name '*.h' -exec cp {} "${pkgdir}/usr/src/linux-${_kernver}/include/config/dvb/" \;

graysky commented on 2014-05-28 18:55 (UTC)

@janck - Guess I'm confused... your statement says if include/config/dvb exists copy include/config/dvb/*.h to "${pkgdir}/usr/lib/modules/${_kernver}/build/include/config/dvb/" The current PKGBUILD just does that, first making the destination on line 432 and doing the same copy on line 433. What am I missing?

janck commented on 2014-05-28 04:58 (UTC)

sorry to confuse. What needs to be done is to repalce line 432 with this one: [[ -e include/config/dvb/ ]] && cp include/config/dvb/*.h "${pkgdir}/usr/lib/modules/${_kernver}/build/include/config/dvb/"

graysky commented on 2014-05-27 21:40 (UTC)

Yes, those lines are in the PKGBUILD. I am confused.

janck commented on 2014-05-27 20:52 (UTC)

yes... did you check?

graysky commented on 2014-05-27 20:03 (UTC)

@janck - That is contained in this PKGBUILD, no?

janck commented on 2014-05-27 19:25 (UTC)

line 432: # and... # mkdir -p "${pkgdir}/usr/lib/modules/${_kernver}/build/include/config/dvb/" cp include/config/dvb/*.h "${pkgdir}/usr/lib/modules/${_kernver}/build/include/config/dvb/"

graysky commented on 2014-05-26 18:23 (UTC)

Shit... did I merge out that line again? I need to review some past PKGBUILDs. I have a nasty habit of comparing linux-ck to linux the ARCH package to make sure I capture all the new changes. I think I may have merged out our change. I will look into it; faster if you can and point me to it given how busy I am these days.

janck commented on 2014-05-26 17:36 (UTC)

All dvb config has been diabled, and I get this error when compiling: ==> Starting package_linux-ck-headers()... cp: cannot stat ‘include/config/dvb/*.h’: No such file or directory ==> ERROR: A failure occurred in package_linux-ck-headers().

kyak commented on 2014-05-20 16:37 (UTC)

Hi graysky, It seems that nvidia-ck needs to be updated: error: failed to prepare transaction (could not satisfy dependencies) :: nvidia-ck-ivybridge: requires nvidia-utils=337.12 Thank you!

graysky commented on 2014-05-18 13:05 (UTC)

Bump to v3.14.4-2 Changelog: Upstream bump of BFQ to v7r4. Commit:

clfarron4 commented on 2014-05-18 12:43 (UTC)

BFQ v7r4 is out!

sash-arch commented on 2014-05-14 06:49 (UTC)

Since i've updatet to 3.14.3-1 i have the same issue, like it's discribed in: Is it possible to compile nls_iso8859-1 into the Kernel like it's been since 3.13? I don't know how to fix this issue without an Kernel with nls_iso8859- builtin support.

test0 commented on 2014-05-13 20:49 (UTC)

Thanks a lot, graysky! :)

graysky commented on 2014-05-13 18:55 (UTC)

Bump to v3.14.4-1 Changelog: Commit:

sir_lucjan commented on 2014-05-13 17:32 (UTC)

3.14.4 is out.

test0 commented on 2014-05-13 09:25 (UTC)

Please apply patch to fix CVE-2014-0196 (

dlh commented on 2014-05-07 18:42 (UTC)

@graysky: yes, the generic package works fine, but linux-ck-haswell not.

graysky commented on 2014-05-07 18:33 (UTC)

@dlh - If you install the generic packages from repo-ck does that render you with a bootable system? Post #1805 in this thread reports no problem with haswell.

dlh commented on 2014-05-07 15:48 (UTC)

[] linux-ck-haswell doesn't boot at all, the issue not apply for the linux-ck

graysky commented on 2014-05-06 20:15 (UTC)

Bump to v3.14.3-1 Changelog: Bump to new version. Commit:

sir_lucjan commented on 2014-05-06 12:32 (UTC)

graysky commented on 2014-04-26 00:34 (UTC)

Bump to v3.13.11-2 Changelog: Bump for new BFQ. Commit:

graysky commented on 2014-04-24 09:04 (UTC)

So it is... I will update later today. Wish I knew this last night before I built 3.13.11-1 but what can you do...

felixonmars commented on 2014-04-24 08:15 (UTC)

BFQ v7r3 is out!

clfarron4 commented on 2014-04-23 22:53 (UTC)

@brando56894: Nice catch indeed. I've updated those parts in my packages as well!

graysky commented on 2014-04-23 19:53 (UTC)

Bump to v3.13.11-1 Changelog: Commit:

graysky commented on 2014-04-23 19:52 (UTC)

Good catch, updated.

brando56894 commented on 2014-04-23 19:48 (UTC)

This section of the package build's comments needs to be updated: "To keep track of which modules are needed for your specific system/hardware, # give module_db script a try: # This PKGBUILD will call it directly to probe all the modules you have logged!" needs to be changed to To keep track of which modules are needed for your specific system/hardware, # give modprobed_db script a try: # This PKGBUILD will call it directly to probe all the modules you have logged!

graysky commented on 2014-04-12 22:19 (UTC)

I am now hosting the BFQ patches until their site is back up. Access them like so: 1) Download the linux-ck source tarball, untar it, enter the linux-ck directory 2) sed -i 's|^_bfqpath=.*$|_bfqpath=\"\"|' PKGBUILD Now you can build it.

chosig commented on 2014-04-12 19:24 (UTC) redirects to -- which seems to have wrecked havoc with the subdomains, I guess the move isn't completed yet.

commented on 2014-04-12 16:07 (UTC) seems to be offline; is there an alternative source of the patches hosted by it?

dlh commented on 2014-04-09 19:05 (UTC)

Here I described my issue, it's actually not related with kvm, I was confused cause I'm using this feature too

coderkun commented on 2014-04-08 15:23 (UTC)

@adam777, @graysky: On my laptop KVM (Qemu) works flawless with linux-ck 3.13.9-1. @dlh: Maybe it is also related to something else?

adam777 commented on 2014-04-06 18:51 (UTC)

@coderkun, @dlh - In here, 3.13.9-1 from the repo works fine with KVM. I'm using the following: qemu-system-x86_64 -machine type=pc,accel=kvm -cpu host -m 1G -net nic,model=virtio -net user -vga qxl -drive file=*******/XP.qcow2,if=virtio -rtc base=localtime -balloon virtio -spice port=5900,disable-ticketing,image-compression=off,jpeg-wan-compression=never,playback-compression=off -device virtio-serial-pci -device virtserialport,chardev=spicechannel0,name=com.redhat.spice.0 -chardev spicevmc,id=spicechannel0,name=vdagent &

dlh commented on 2014-04-06 15:52 (UTC)

@coderkun: I will try tomorrow, since I am currently on machine with different processor.

coderkun commented on 2014-04-06 15:47 (UTC)

@graysky: I disabled the KVM-path in the PKGBUILD and recompiled linux-ck but KVM is still not working for me. So I’m not sure if the path makes it better or worse. What about you @dlh?

graysky commented on 2014-04-06 15:33 (UTC)

coderkun and dlh - adam777 requested that patch in the bbs and replied that all is well with kvm.[1] I asked him to comment here. If the patch is causing breakage for users, I will disable it. I just need to know since I do not use KVM. 1.

dlh commented on 2014-04-06 15:31 (UTC)

Maybe it was fixed upstream

coderkun commented on 2014-04-06 15:12 (UTC)

@graysky: I have no evidence that the patch really causes this but KVM stopped working for me at around the same time.

graysky commented on 2014-04-06 14:22 (UTC)

@dlh - Yes, since v3.13.6-1 that patch has been included per a user request. Are you saying that your KVM functionality has been broken since that release, in other words, that patch has caused the regression?

dlh commented on 2014-04-06 12:40 (UTC)

graysky do you build your kernel in repo-ck with ? From version 3.13.6 it's stopped working.

graysky commented on 2014-04-04 18:58 (UTC)

Bump to v3.13.9-1 Changelog: Commit:

sir_lucjan commented on 2014-04-04 14:47 (UTC)

graysky - sorry, my mistake.

graysky commented on 2014-04-04 00:28 (UTC)

Can update until after work tomorrow but thanks for the heads up. Odd that Greg pushed it on a Thurs night, must be something critical... Will check the readme once lkml is back up.

clfarron4 commented on 2014-04-03 22:39 (UTC)

Dear lord, 3.13.9 is out all ready...

graysky commented on 2014-03-31 23:58 (UTC)

Bump to v3.13.8-1 Changelog: Commit:

clfarron4 commented on 2014-03-31 17:34 (UTC)

3.13.8 is out:

graysky commented on 2014-03-24 19:46 (UTC)

Bump to v3.13.7-1 Changelog: Commit:

snack commented on 2014-03-24 09:20 (UTC)

I'm experiencing random hangs of touchpad, mouse and keyboards at boot after kdm start. I'm unable to do anything but a hard reboot. This is what I find in the boot log: mar 24 09:58:53 elric kernel: BUG: unable to handle kernel NULL pointer dereference at (null) mar 24 09:58:53 elric kernel: IP: [<ffffffffa0c9b317>] mousedev_open_device+0x77/0x100 [mousedev] mar 24 09:58:53 elric kernel: PGD b646b067 PUD b6476067 PMD 0 mar 24 09:58:53 elric kernel: Oops: 0000 [#1] PREEMPT SMP mar 24 09:58:53 elric kernel: Modules linked in: mousedev(+) psmouse serio_raw atkbd libps2 uvcvideo videobuf2_vmalloc videobuf2_memops videobuf2_core videodev media coretemp arc4 microcode(+) iwldvm mac80211 pcspkr i2c_i801 lpc_ich snd_hda_codec_hdmi r592 memstick iwlwi mar 24 09:58:53 elric kernel: CPU: 0 PID: 367 Comm: acpid Tainted: P O 3.13.6-1-ck #1 mar 24 09:58:53 elric kernel: Hardware name: Sony Corporation VGN-SR21M_S/VAIO, BIOS R1110Y1 08/14/2008 mar 24 09:58:53 elric kernel: task: ffff8800b77dda30 ti: ffff8800b64c0000 task.ti: ffff8800b64c0000 mar 24 09:58:53 elric kernel: RIP: 0010:[<ffffffffa0c9b317>] [<ffffffffa0c9b317>] mousedev_open_device+0x77/0x100 [mousedev] mar 24 09:58:53 elric kernel: RSP: 0018:ffff8800b64c1c10 EFLAGS: 00010202 mar 24 09:58:53 elric kernel: RAX: 0000000000000000 RBX: ffff8800b655a000 RCX: ffff8800b655a068 mar 24 09:58:53 elric kernel: RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000246 mar 24 09:58:53 elric kernel: RBP: ffff8800b64c1c28 R08: 0000000000000000 R09: ffff88013b001600 mar 24 09:58:53 elric kernel: R10: 0000000000000000 R11: 0000000000000004 R12: 0000000000000000 mar 24 09:58:53 elric kernel: R13: ffff8800b655a080 R14: ffff8800b65761a8 R15: ffff8801395cf300 mar 24 09:58:53 elric kernel: FS: 00007fae54adb700(0000) GS:ffff88013fc00000(0000) knlGS:0000000000000000 mar 24 09:58:53 elric kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 mar 24 09:58:53 elric kernel: CR2: 0000000000000000 CR3: 00000000b7726000 CR4: 00000000000007f0 mar 24 09:58:53 elric kernel: Stack: mar 24 09:58:53 elric kernel: ffff880138dda000 ffff8800b655a000 ffff8800b655a078 ffff8800b64c1c60 mar 24 09:58:53 elric kernel: ffffffffa0c9c0cc ffff8800b655a340 ffff8800b65761a8 ffff8801395cf300 mar 24 09:58:53 elric kernel: ffffffffa0c9ce80 ffff8801395cf310 ffff8800b64c1c98 ffffffff8118555f mar 24 09:58:53 elric kernel: Call Trace: mar 24 09:58:53 elric kernel: [<ffffffffa0c9c0cc>] mousedev_open+0xcc/0x150 [mousedev] mar 24 09:58:53 elric kernel: [<ffffffff8118555f>] chrdev_open+0x9f/0x1d0 mar 24 09:58:53 elric kernel: [<ffffffff8117ec97>] do_dentry_open+0x1b7/0x2c0 mar 24 09:58:53 elric kernel: [<ffffffff8118bf41>] ? __