Package Details: linux-clear 5.18.8-1

Git Clone URL: (read-only, click to copy)
Package Base: linux-clear
Description: The Clear Linux kernel and modules
Upstream URL:
Keywords: clear clearlinux intel kernel
Licenses: GPL2
Submitter: metak
Maintainer: metak
Last Packager: metak
Votes: 59
Popularity: 1.88
First Submitted: 2018-01-18 21:47 (UTC)
Last Updated: 2022-06-29 13:53 (UTC)

Pinned Comments

metak commented on 2022-03-27 13:44 (UTC) (edited on 2022-04-02 13:20 (UTC) by metak)

The PKGBUILD file supports a few different configurable build-time options:
  • _makenconfig Tweak kernel options prior to a build via nconfig. Pseudo-graphical menu based on ncurses. 1 2
  • _localmodcfg Only compile active modules to VASTLY reduce the number of modules built and the build time.
  • Please read arch wiki page for detailed explanation on how to use it.
  • _subarch Enable additional optimization/tuning for kernel builds by adding more micro-architectures options. Default is Generic-x86-64 number 36 if nothing else is selected.
  • PKGBUILD file with the list of architectures
  • graysky kernel_compiler_patch
  • _use_current Use the current kernel's .config file. Enabling this option will use the .config of the RUNNING kernel rather than the ARCH defaults. Useful when the package gets updated and you already went through the trouble of customizing your config options. NOT recommended when a new kernel is released, but again, convenient for package bumps.
  • _use_llvm_lto Compile the kernel with LLVM/Clang.
  • _debug Enable y some additional debug features present in arch kernel, but not in Clear upstream. n to force disable or leave empty to ignore.
Lastly, build the kernel by setting your environment variable env _subarch=40 _localmodcfg=y.

metak commented on 2018-01-22 01:49 (UTC) (edited on 2021-12-01 12:47 (UTC) by metak)

1. Binaries available in my repo:
SigLevel = Never
Server =$arch
2. After install adjust your boot cmd line. ClearLinux uses clr-boot-manager which takes care of that. This is upstream default:
quiet console=tty0 console=ttyS0,115200n8 cryptomgr.notests initcall_debug intel_iommu=igfx_off kvm-intel.nested=1 no_timer_check noreplace-smp page_alloc.shuffle=1 rcupdate.rcu_expedited=1 rootfstype=ext4,btrfs,xfs,f2fs tsc=reliable rw 

Latest Comments

metak commented on 2022-05-08 16:02 (UTC)

@Aleksana Try initcall_blacklist=kvm_intel to blacklist built-in module?

Aleksana commented on 2022-05-08 14:46 (UTC)

I got the kvm_intel issue again.When I compile on an intel x86_64 targeted at amd zen3 subarch, the kernel loads both kvm_intel and kvm amd,throwing an error

kvm: no hardware support for 'kvm_intel'

Have tried to append kernel parameter modprobe.blacklist=kvm_intel but it didn't help. It looks a bit annoying as I'm using plymouth.

xuanruiqi commented on 2022-04-19 10:16 (UTC)

OK, I figured out the problem is LLVM LTO. It's broken on this version. If you're using it, maybe it's wise to disable for the time being.

metak commented on 2022-04-19 08:54 (UTC)

@xuanruiqi Have you tried compiling with SYSFB_SIMPLEFB disabled?

xuanruiqi commented on 2022-04-18 22:00 (UTC)

After update to 5.17.3-clear, I have the weird problem that I can't get my screen to initialize anymore. I don't have this problem with the stock Arch kernel. Tried a lot of reconfigurations and recompilations, but still I get stuck on:

fb0 switching to i915 from efi vga

every boot. I wonder if anyone is able to reproduce this at all...

metak commented on 2022-04-17 08:58 (UTC)

@FiestaLake #Patch0110 is already commented in spec file so it's skipped, but I had to manually disable raid6.patch because spec file only commented rpm macro part #%patch0129 and forgot patch source line.

FiestaLake commented on 2022-04-17 02:24 (UTC) (edited on 2022-04-17 02:30 (UTC) by FiestaLake)

@metak Shouldn't you also remove patch0110? It's commented on linux5.17 update now.

EDIT: I guess it's not getting applied now.

FiestaLake commented on 2022-03-31 13:21 (UTC)

@metak You'd better just leave the option for some people that requested a option of the entire debug flags enabled other than him.

@rayzorben tho @metak won't add the option for it, you can easily edit PKGBUILD to add an extra option for it. Just look through the PKGBUILD to see how the system works.

metak commented on 2022-03-31 10:32 (UTC)

@FiestaLake I get it now. Thanks. But, as you've already pointed out using _use_current on stock arch kernel will get you way more extra options. Pretty much you're on your own then. Not supported.

My main question is still should I just remove the _debug option completely from PKGBUILD file? Does anyone find it useful? And, with next update I'll just apply change mentioned bellow for resolve_btfids.
If anyone has DEBUG_INFO_BTF_MODULES enabled and resolve file is generated, it will be installed.

FiestaLake commented on 2022-03-31 08:26 (UTC) (edited on 2022-03-31 08:27 (UTC) by FiestaLake)

@rayzorben But you have to notice that _use_current is intended to be passed in the environment booted with this clear kernel, not the arch kernel.

_use_current: Use the current kernel's .config file. Enabling this option will use the .config of the RUNNING kernel rather than enabling some arch kernel's configs on the top of the clear's default. Useful when the package gets updated and you already went through the trouble of customizing your config options. NOT recommended when a new kernel is released, but again, convenient for package bumps.

FiestaLake commented on 2022-03-31 08:21 (UTC) (edited on 2022-03-31 08:23 (UTC) by FiestaLake)

@metak I think you can just add a _no_debug option that disables debug configs.

What he wanted to say is that when we build the kernel using the arch kernel's config (by passing _use_current in the env booted with the arch kernel), debug configs are enabled whether _debug is passed or not (I guess).

rayzorben commented on 2022-03-30 21:27 (UTC)

@metak i'm honestly not sure. I was just thinking I want those debug flags off even if arch has them on, so I would like to just do _debug=n and have it turn those kernel flags off.

metak commented on 2022-03-30 18:18 (UTC) (edited on 2022-03-30 18:18 (UTC) by metak)

@rayzorben If I'm getting this right then I should just remove the debug option and just check in PKGBUILD file for resolve_btfids if the file exists & install it? (not native English speaker)

    # required when DEBUG_INFO_BTF_MODULES is enabled
    if [ -f tools/bpf/resolve_btfids/resolve_btfids ]; then
        install -Dt "$builddir/tools/bpf/resolve_btfids" tools/bpf/resolve_btfids/resolve_btfids

rayzorben commented on 2022-03-30 16:53 (UTC)

@metak thanks for your work, having the options listed is very helpful. For somebody new coming to this project, it will be very beneficial.

Regarding _debug it looks like you check if it is set, and then enable the proper debug kernel flags. As far as I understand, the issue with btfids is because the default arch kernel has these on, and when we use _use_current it is pulling in those options even if they aren't enabled in clear. Because those are set, you need to run resolve_btfids.

I am thinking that this is one of those 'tradeoffs' of debug vs performance. I might want to disable these. If I pass _debug=n do you think you could disable these kernel flags instead of enabling them if it is set?

I know I could use _makenconfig to disable these, but then if I ever have to boot standard arch kernel and compile linux-clear I will likely forget and miss them again.

metak commented on 2022-03-27 13:59 (UTC)

@rayzorben resolve_btfids should be resolved with the latest update where I've added _debug option which enables some basic debug options.

Those are the default Clear linux boot parameters. I pasted it here for refference only, the kernel should boot just fine without any modifications. SERIAL_8250

metak commented on 2022-03-27 13:44 (UTC) (edited on 2022-04-02 13:20 (UTC) by metak)

The PKGBUILD file supports a few different configurable build-time options:
  • _makenconfig Tweak kernel options prior to a build via nconfig. Pseudo-graphical menu based on ncurses. 1 2
  • _localmodcfg Only compile active modules to VASTLY reduce the number of modules built and the build time.
  • Please read arch wiki page for detailed explanation on how to use it.
  • _subarch Enable additional optimization/tuning for kernel builds by adding more micro-architectures options. Default is Generic-x86-64 number 36 if nothing else is selected.
  • PKGBUILD file with the list of architectures
  • graysky kernel_compiler_patch
  • _use_current Use the current kernel's .config file. Enabling this option will use the .config of the RUNNING kernel rather than the ARCH defaults. Useful when the package gets updated and you already went through the trouble of customizing your config options. NOT recommended when a new kernel is released, but again, convenient for package bumps.
  • _use_llvm_lto Compile the kernel with LLVM/Clang.
  • _debug Enable y some additional debug features present in arch kernel, but not in Clear upstream. n to force disable or leave empty to ignore.
Lastly, build the kernel by setting your environment variable env _subarch=40 _localmodcfg=y.

rayzorben commented on 2022-03-27 04:57 (UTC) (edited on 2022-03-27 04:59 (UTC) by rayzorben)

I also had the resolve_btfids issue. I had it working, but when upgrading the kernel I forgot to modify the PKGBUILD and next thing you know, no dkms modules were installing.

Would be nice to have a _use_resolve_btfids option, or check for the correct kernel option if it is enabled and call it.

Also to note, what is the console=tty0 console=ttyS0,115200n8 line about? I had an issue with fstab, and nothing was being output during boot even with quiet removed. When I removed those I was able to see what it was outputting. This may make boot faster because its not dumping what it is doing to the screen but it made it a pain to debug.

metak commented on 2022-03-15 19:55 (UTC)

@rayzorben You should boot the default arch kernel, plug in all your devices so it loads appropriate modules and then create modprobe.db. It seems like something might be missing in your build and that's why bluetooth is not working? As for the options, I'll add some description in the future.

rayzorben commented on 2022-03-13 16:37 (UTC)

Would love to see some basic instructions pinned with what the options are. I discovered them through the PKGBUILD but still wasn't sure what to do with them.

That being said, I had a few trial and errors regardless. I felt like env _subarch=35 _localmodcfg=y _use_current=y would be the best combo while running my fully functioning system. However it seems that _localmodcfg for some reason is causing issues with bluetooth. Bluetooth was there, it could see devices, it said my mouse was connected, but it would not work. Rebuilding without _localmodcfg works though.

xuanruiqi commented on 2022-01-22 14:57 (UTC) (edited on 2022-01-22 15:11 (UTC) by xuanruiqi)

OK, turning debug info off works.

(Maybe because I have CONFIG_DEBUG_INFO_BTF turned on in my kernel config which causes resolve_btfids to be called during DKMS build. Let me try turning most debug options off.)

metak commented on 2022-01-22 12:41 (UTC)

@xuanruiqi After testing nvidia-dkms I couldn't reproduce it. It builds fine (gcc & llvm). It's definitely something on your system that's causing this. Try building in clean environment?

xuanruiqi commented on 2022-01-22 08:03 (UTC)

I was building nvidia-dkms. Some modules like nvidia need btf info to build properly, so maybe you want to put it under an option or something like that. AFAIK a lot of people use nvidia-dkms.

metak commented on 2022-01-21 23:14 (UTC)

@xuanruiqi I had to comment out that line in PKGBUILD file when I synced it with stock arch because it would fail to build. (not enabled on Clear) It needs some other debugging things enabled like CONFIG_DEBUG_INFO, CONFIG_DEBUG_INFO_BTF, CONFIG_PAHOLE_HAS_SPLIT_BTF which will increase the size and compilation time. I can enable the debug info if others want it, but I would rather not.

As for dkms error, I just tried building random module (rtl88x2bu) and it builds fine. (gcc & llvm) Please specify which module/versions, etc. (whole build log)

xuanruiqi commented on 2022-01-21 17:02 (UTC) (edited on 2022-01-21 20:26 (UTC) by xuanruiqi)

SOLUTION: please fix the PKGBUILD, add this line, otherwise external DKMS modules can't be built:

install -Dt "$builddir/tools/bpf/resolve_btfids" tools/bpf/resolve_btfids/resolve_btfids

See diff in official linux package commit history.

Hey all! With LLVM compiled kernel (5.16.2), I get this error when I try to compile DKMS modules:

/bin/sh: line 1: ./tools/bpf/resolve_btfids/resolve_btfids: No such file or directory

This error is absent in linux-clear <=5.16.1.

GCC doesn't work either. I suspect this is a really silly mistake, but I can't fix it...

k0r0ng commented on 2022-01-12 22:55 (UTC)

@metak Thank you!

metak commented on 2022-01-11 17:37 (UTC)

@k0r0ng It's not enabled by default in CLR config file, I've enabled it in ˙5.15.14-1˙.

k0r0ng commented on 2022-01-11 15:06 (UTC)

Why CONFIG_KPROBES not included by default?

d00d00 commented on 2021-12-11 05:13 (UTC)

@metak @Moviuro You're both legends. Compiling on Skylake for Atom with single short script now, no interaction required. Thank you so much!

FiestaLake commented on 2021-12-10 12:31 (UTC)

@metak I can confirm that 5.15.7-3 and 5.15.7-4 changes work fine here! Happy to use clang kernel with new dkms support :)

metak commented on 2021-12-10 11:58 (UTC)

@d00d00 Added in 5.15.7-2.

@xuanruiqi Added in 5.15.7-3. Haven't tried it too much so test it and we'll see.

@Moviuro Added in 5.15.7-4. I also changed it for the other variables as well.

Moviuro commented on 2021-12-10 11:13 (UTC)

Could we get the following diff in the PKGBUILD? rationale: I can now do _subarch=26 makepkg -sriCfc; and it correctly compiles for skylake, with no interaction from me. That way, I don't need to edit the PKGBUILD, or wait until I'm asked for the subarch in an interactive way.

The : "${var:="value"}" is a POSIX-compliant way to set a variable to a default value (in our case: the empty string), but only if it wasn't already set. This change has no impact for people who simply type makepkg, and it will still ask them (interactively) which subarch to compile for.

diff --git a/PKGBUILD b/PKGBUILD
index 13880d4..ff19d30 100644
@@ -62,7 +62,7 @@ _localmodcfg=
 #  39. Generic-x86-64-v4 (GENERIC_CPU4)
 #  40. Intel-Native optimizations autodetected by GCC (MNATIVE_INTEL)
 #  41. AMD-Native optimizations autodetected by GCC (MNATIVE_AMD)
+: "${_subarch:=""}"

 # Use the current kernel's .config file
 # Enabling this option will use the .config of the RUNNING kernel rather than

xuanruiqi commented on 2021-12-09 06:50 (UTC)

Could we have an option to enable Clang builds for the kernel? I would like to use a Clang/LLVM build with LTO.

d00d00 commented on 2021-12-09 00:21 (UTC) (edited on 2021-12-09 00:30 (UTC) by d00d00)

Hello! Would it be possible for you to add the KSMBD module to the default build for this kernel? It's a kernel-level SMB daemon that works really well on lighter-weight NAS machines and whatnot. It requires these configuration changes to be added to the PKGBUILD (assuming SMB Direct is desired):

scripts/config --module SMB_SERVER \

And it requires KSMBD-MODULE to be added to provides(). Sorry if this is too niche a request, and thank you for your work maintaining this package.

metak commented on 2021-12-01 12:43 (UTC) (edited on 2021-12-01 12:43 (UTC) by metak)

@mizzunet .cmdline file is meant for systemd-boot loader, but kernel should boot just fine without any modifications to cmd boot line.
If you're using GRUB then after installing this package you should only have to regerate grub.cfg file. See: 1 2

mizzunet commented on 2021-11-27 12:41 (UTC)

I'm having this error message for boot entry Arch Linux, with Linux linux-clear-bin.cmdline I do have another boot entry regarding clear, Arch Linux, with Linux linux-clear-bin which can be booted into.

Loading Linux linux
Error: premature end of file /vmlinuz-linux-clear-bin.cmdline
Loading initial ramdisk
Error: you need to load the kernel first
Press a key to continue

I have adjusted cmd boot line as pinned comment, that is

GRUB_CMDLINE_LINUX="quiet console=tty0 console=ttyS0,115200n8 cryptomgr.notests initcall_debug intel_iommu=igfx_off kvm-intel.nested=1 no_timer_check noreplace-smp page_alloc.shuffle=1 rcupdate.rcu_expedited=1 rootfstype=ext4,btrfs,xfs,f2fs tsc=reliable rw"

then had executed sudo grub-mkconfig -o /boot/grub/grub.cfg

FiestaLake commented on 2021-11-10 23:44 (UTC)

@metak It worked. Thanks^^

tam1m commented on 2021-11-10 18:34 (UTC)

@metak Awesome, thanks!

metak commented on 2021-11-10 18:07 (UTC)

@tam1m NTFS3 module enabled in 5.15.1-2. The default config comes from clear team and it's not enabled in there by default, don't know why exactly.

@FiestaLake Can you try the latest update?

FiestaLake commented on 2021-11-10 16:09 (UTC) (edited on 2021-11-10 16:30 (UTC) by FiestaLake)

tam1m commented on 2021-11-10 14:48 (UTC)

@metak Would you mind setting


by default? Is there a specific reason NTFS3 is not enabled?

xuanruiqi commented on 2021-10-18 12:26 (UTC)

@metak Much thanks!

metak commented on 2021-10-17 21:33 (UTC)

@xuanruiqi Thanks for pointing it out. I did applied those chnages in 5.7.2-1, but for some reason I removed them in subsequent commit? I'll add them back again and see.

xuanruiqi commented on 2021-10-17 20:28 (UTC)

There are some discrepancies with the upstream PKGBUILD that I see could use a fix.

For example,

echo "Stripping vmlinux..."
  strip -v $STRIP_STATIC "$builddir/vmlinux"

is missing (so vmlinux is not stripped, it seems).

Also, the upstream uses make all to build the kernel instead of make bzImage . The other difference I notice is the command used to install modules:

make INSTALL_MOD_PATH="$pkgdir/usr" INSTALL_MOD_STRIP=1 modules_install

in the upstream.

Thank you a lot for your work maintaining this package! It just seemed that the upstream PKGBUILD changed a bit in the mean time without getting noticed.

metak commented on 2021-10-14 15:17 (UTC)

@xuanruiqi The package installation size is 64,83 MiB and for headers it's 127,95 MiB for binaries in my repo. Both are smaller in size than default kernel/headers packages. I try to keep my PKGBUILD files as close to upstream linux package as possible.

xuanruiqi commented on 2021-10-14 12:45 (UTC)

I just found that this package, when installed, is very large. On my system, for example, it is 666M in size, as opposed to 78M for the regular kernel. It seems that It seems that the blame is on:


The reason seems to be that this file is not stripped, as is done in the official kernel. In addition, there are some more discrepancies between this package's PKGBUILD and the upstream linux PKGBUILD. I would like to recommend that the PKGBUILD be revised to match the upstream linux PKGBUILD.

metak commented on 2021-10-01 13:18 (UTC)

@Gonzo2028 Add fsync/futex patches to sources in PKGBUILD file and build the package. You can get them at these two locations:

Gonzo2028 commented on 2021-09-30 22:53 (UTC)

Can anyone tell me how to add the fsync-patch for/from steam?

guiodic commented on 2021-09-17 14:59 (UTC)

@metak: I installed linux-clear-current (5.14.rc7). I reported here, not on linux-clear-current, sorry.

metak commented on 2021-09-17 13:08 (UTC) (edited on 2021-09-17 13:08 (UTC) by metak)

@guiodic Well arch kernel is already on 5.14 while this one is still on 5.13. If you can please try linux-clear-bin from my repo just so we can rule out any possible changes introduced in this pkgbuild?

guiodic commented on 2021-09-17 12:46 (UTC)

I was wrong, i915/tgl_dmc_ver2_08.bin is not GUC but for power management, nevertheless also GUC seems not work.

With the default kernel all works well:

17/09/21 13:18 i915 0000 0:02.0: [drm] Finished loading DMC firmware i915/tgl_dmc_ver2_08.bin (v2.8) 17/09/21 13:18 i915 0000 0:02.0: [drm] GuC firmware i915/tgl_guc_49.0.1.bin version 49.0 submission:disabled 17/09/21 13:18 i915 0000 0:02.0: [drm] HuC firmware i915/tgl_huc_7.5.0.bin version 7.5 authenticated:yes

guiodic commented on 2021-09-17 12:43 (UTC)

@metak: yes, it works well with the standard kernel but not for linux-clear

metak commented on 2021-09-17 11:45 (UTC)

@guiodic Have you checked/tried what's in wiki?

guiodic commented on 2021-09-17 10:43 (UTC)

17/09/21 12:40 kernel i915 0000:00:02.0: Direct firmware load for i915/tgl_dmc_ver2_08.bin failed with error -2 17/09/21 12:40 kernel i915 0000:00:02.0: [drm] Failed to load DMC firmware i915/tgl_dmc_ver2_08.bin. Disabling runtime power management. 17/09/21 12:40 kernel i915 0000:00:02.0: [drm] DMC firmware homepage:

guiodic commented on 2021-09-17 09:09 (UTC)

I receive an error on Intel GPU firmware loading (GUC)

Direct firmware load for i915/tgl_dmc_ver2_08.bin failed with error -2

this make me unable to VP9 encoding via LibVa

metak commented on 2021-09-09 17:45 (UTC)

@Tjuh It's optimized for intel hardware, but it should also work for AMD. Haven't tried it on amd.

Tjuh commented on 2021-09-04 09:27 (UTC)

Is this kernel mainly suited for Intel cpus or does it provide the same benefits for an AMD Ryzen 7 laptop?

xstefen commented on 2021-08-28 02:01 (UTC) (edited on 2021-08-28 02:04 (UTC) by xstefen)

For those unfamiliar with CFLAGs/-march optimizations


Info on what option to choose, the differences, etc..

tl;dr -march=native is usually ideal

lucenera commented on 2021-08-19 07:39 (UTC)

==> ERROR: Failure while downloading How to solve?

metak commented on 2021-07-07 10:54 (UTC)

@xawiw Previously I removed that build parameter from PKGBUILD file because the proper way is to set it with MAKEFLAGS in /etc/makepkg.conf.

@Dobbie03 If you don't want native optimizations then choose skylake.

Dobbie03 commented on 2021-07-07 08:08 (UTC)

What option should I choose for a Coffee Lake Core i9 9900?

xawlw commented on 2021-07-06 20:29 (UTC)

@metak Can you please add multi-threading to the build parameters?

make bzImage modules -j$(nproc --all)

metak commented on 2021-06-11 08:28 (UTC)

@PedroDePacas Version 20210606 was removed because of the broken patch for kernel versions 4.19-5.4. Fixed now.

PedroDePacas commented on 2021-06-11 07:00 (UTC) (edited on 2021-06-11 08:02 (UTC) by PedroDePacas)

@metak Hi. I'm getting this error:

Failure while downloading`

How can I fix it?

Khaithang commented on 2021-06-08 16:58 (UTC)

@metak, thanks, that was the case. Apparently 3G proved a bit low!. though i did check the usage and it stop while 250-300M was still available. and why the heck was i even using ram, my nvme SSD seems to be faster then that. thanks again for the tip.

metak commented on 2021-06-07 14:33 (UTC)

@Khaithang Have you tried compiling on SSD/HD instread of in RAM to check you're not running out of space? BUILDDIR=/tmp/makepkg

Khaithang commented on 2021-06-06 20:29 (UTC) (edited on 2021-06-08 17:00 (UTC) by Khaithang)

i am getting a build error! few excerpt from the build log:

ZSTD22  arch/x86/boot/compressed/vmlinux.bin.zst   
make[2]: *** [arch/x86/boot/compressed/Makefile:135: 
arch/x86/boot/compressed/vmlinux.bin.zst] Error 137
make[2]: *** Deleting file 'arch/x86/boot/compressed/vmlinux.bin.zst'
make[2]: *** Waiting for unfinished jobs....
GEN     Module.symvers
make[1]: *** [arch/x86/boot/Makefile:115: arch/x86/boot/compressed/vmlinux] Error 2
make: *** [arch/x86/Makefile:302: bzImage] Error 2
make: *** Waiting for unfinished jobs....

then after almost finishing the build, it fails with this:

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

Here's the full log, most seems normal apart what i already pointed out above. Sorry, i don't which parts are important so adding the whole log, the log file is pretty long. log :

some guys pointed that the issue is with makepkg/file not able to extract simple xz compressed file, this is in regards to the linux-ck package, and downgrading file to 5.39 seems to fix it for them. I downgraded it but the issue persist! Do let me know if i need to attach makepkg.conf as well.

EDIT: anyway i am attaching my makepkg.conf as well ->

oddpuck commented on 2021-06-04 19:58 (UTC)


metak commented on 2021-06-04 19:49 (UTC)

@oddpuck There's no separate option for Comet Lake, you should use skylake.

oddpuck commented on 2021-06-04 18:20 (UTC)

First, thanks for making this package. Why is Intel Comet Lake not an option when it asks you to select processor family during build?

metak commented on 2021-05-02 10:05 (UTC)

@amoka That part in prepare() needs some cleaning up...

amoka commented on 2021-05-02 01:43 (UTC)

Can you please explain a little more about the following (line 227)?

### do not run `make olddefconfig` as it sets default options
    yes "" | make config >/dev/null

Since we run make olddefconfig before that too (line 179)

Thanks in advance!

Ahmedtas commented on 2021-04-29 15:44 (UTC)

@metak thank you

metak commented on 2021-04-29 10:46 (UTC)

@Ahmedtas I don't really know... There are opened issues asking this same question on their github page, but without answer. Their other kernel packages are definitely lagging behind, -current, -preempt-rt,... Maybe they have other more important things to do / test? 5.10 is LTS anyway.

Ahmedtas commented on 2021-04-28 22:25 (UTC)

Just a question: Why there isn't version 5.11 or 5.12 ?

metak commented on 2021-04-07 10:51 (UTC)

@olace & @xuanruiqi It's a problem with new version of file.

xuanruiqi commented on 2021-04-07 07:18 (UTC)

For some reason it seems that makepkg doesn't extract non-tarball compressed files automatically anymore....

xuanruiqi commented on 2021-04-07 07:11 (UTC)

I'm unfortunately having the same problem as @olace.

olace commented on 2021-04-02 16:31 (UTC)

Hi, could help with these?

-> Creating working copy of clearlinux git repo... Reset branch 'makepkg' -> Extracting enable_additional_cpu_optimizations-20210309.tar.gz with bsdtar ==> Starting prepare()... Add upstream patches patch: **** Can't open patch file ../patch-5.10.27 : No such file or directory ==> ERROR: A failure occurred in prepare(). Aborting...

[linux-clear]$ ls 0001-ZEN-Add-sysctl-and-CONFIG-to-disallow-unprivileged-C.patch clearlinux enable_additional_cpu_optimizations-20210309.tar.gz linux-5.10.tar.sign linux-5.10.tar.xz linux.install patch-5.10.27 patch-5.10.27.xz pci-enable-overrides-for-missing-acs-capabilities.patch PKGBUILD src

metak commented on 2021-03-26 16:42 (UTC)

@amoka Workaround is what it is right now where I just removed it from boot cmd line because it's completely optional. init-rdahead clear srpm package provides initra-desktop and it reads ahead list of libraries/files on boot I believe.

amoka commented on 2021-03-26 02:22 (UTC)

It seems that we don't have this file in Arch initra-desktop and Clear Linux cmdline adds: init=/usr/bin/initra-desktop. Any workarounds for this in Arch?

Gonzo2028 commented on 2021-03-09 14:11 (UTC)

enabling CONFIG_NET_VENDOR_AQUANTIA worked for me.

metak commented on 2021-03-09 14:02 (UTC)

@Gonzo2028 Hi. I've enabled CONFIG_NET_VENDOR_AQUANTIA, does that work for you or do you need me to also build the atlantic module?

Gonzo2028 commented on 2021-03-08 08:55 (UTC)

Found CONFIG_NET_VENDOR_AQUANTIA is disabled in config. Compiling and then testing...

Gonzo2028 commented on 2021-03-08 08:47 (UTC)

My new 10GBE Ethernet card seems to not work with that kernel. It hangs on boot. Any ideas? Stock kernel works fine.

lspci -v 05:00.0 Ethernet controller: Aquantia Corp. AQC100 10G Ethernet MAC controller [AQtion] (rev 02)


8021q ac97_bus acpi_cpufreq aesni_intel af_alg agpgart ahci algif_hash algif_skcipher asus_wmi atlantic battery bluetooth bnep btbcm btintel btrtl btusb cbc ccp cec cfg80211 cifs cmac crc16 crc32c_generic crc32c_intel crc32_pclmul crct10dif_pclmul cryptd crypto_simd crypto_user dca dm_crypt dm_integrity dm_mod dns_resolver drm drm_kms_helper ecc ecdh_generic edac_mce_amd eeepc_wmi efivars encrypted_keys evdev ext2 ext3 ext4 f2fs fat fb_sys_fops ff_memless fixed_phy fscache fuse garp ghash_clmulni_intel glue_helper hid hid_fanatec hid_generic hid_logitech_hidpp hid_sony hwmon_vid i2c_algo_bit i2c_nvidia_gpu i2c_piix4 igb input_leds intel_rapl_common intel_rapl_msr ip6table_filter ip6_tables iptable_filter ip_tables irqbypass isofs jbd2 joydev k10temp kvm kvm_amd ledtrig_audio libahci libarc4 libata libdes libphy llc loop mac_hid macsec mbcache mc md4 mdio_devres mousedev mrp msr mxm_wmi nct6775 nls_cp437 nls_iso8859_1 nls_utf8 of_mdio pcspkr pinctrl_amd r8169 rapl rc_core realtek rfcomm rfkill rng_core sch_fq scsi_mod sd_mod sg snd snd_compress snd_hda_codec snd_hda_codec_generic snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_core snd_hda_intel snd_hwdep snd_intel_dspcfg snd_intel_nhlt snd_mixer_oss snd_pcm snd_pcm_dmaengine snd_pcm_oss snd_rawmidi snd_seq_device snd_soc_core snd_timer snd_usb_audio snd_usbmidi_lib soundcore soundwire_bus soundwire_cadence soundwire_generic_allocation soundwire_intel sp5100_tco sparse_keymap stp syscopyarea sysfillrect sysimgblt tcp_bbr tpm tpm_crb tpm_tis tpm_tis_core trusted ttm typec typec_displayport typec_ucsi uas ucsi_ccg uhid uinput usbhid

metak commented on 2021-01-19 20:37 (UTC)

@Gonzo2028 You can check in the kernel config file in /proc/config.gz at # Processor type and features section.

zcat /proc/config.gz > /tmp/config-$(uname -r)

Gonzo2028 commented on 2021-01-19 19:34 (UTC)

Thanks! I will give it a try. Is there a method to read the used subarch on compilation for the current running kernel?

metak commented on 2021-01-19 18:03 (UTC)

@Gonzo2028 You just have to make a small change to pkgbuild file, set _subarch=13 for zen3 and point the source for gcc optimizations patch to latest commit to get the required patch.

Gonzo2028 commented on 2021-01-19 17:18 (UTC)

or just tell me how to implement znver3 then ;). Thanks.

Gonzo2028 commented on 2021-01-19 17:09 (UTC)

i have gcc 11 installed, i just need the option. or will native detect znver3 if available?

metak commented on 2021-01-10 20:08 (UTC)

@Gonzo2028 Hi, znver3 requires gcc from trunk so I can't add that.
Add -march=znver3/requires gcc trunk at this time

Gonzo2028 commented on 2021-01-10 14:51 (UTC)

please add support for -march=znver3

gebau00a commented on 2021-01-07 02:31 (UTC)

I also do not have any xf86-video installed, only nvidia-dkms drivers. I assume your main problem is the missing drm module. Which graphics driver do you want to use?

btd1337 commented on 2020-12-26 23:39 (UTC) (edited on 2020-12-26 23:43 (UTC) by btd1337)

Is anyone able to help?

When library xf86-video-intel isn't installed, Kernel Clear with Gnome won't starts GDM.

gnome-shell[524]: Failed to create backend: No drm devices found
gnome-session-binary[515]: WARNING: App 'org.gnome.Shell.desktop' exited with code 1
gnome-session-binary[515]: Unrecoverable failure in required component org.gnome.Shell.desktop
gdm[465]: Gdm: GdmDisplay: Session never registered, failing
gdm[465]: Gdm: Child process -509 was already dead.
/usr/lib/gdm-x-session[535]: (--) Log file renamed from "/var/log/" to "/var/log/Xorg.0.log"
/usr/lib/gdm-x-session[535]: (WW) Failed to open protocol names file lib/xorg/protocol.txt
gnome-session-binary[569]: WARNING: Failed to reset failed state of units: GDBus.Error:org.freedesktop.DBus.Error.NameHasNoOwner: Name "org.freedesktop.systemd1" does not exist
gnome-session-binary[569]: WARNING: Falling back to non-systemd  startup procedure due to error: GDBus.Error:org.freedesktop.DBus.Error.NameHasNoOwner: Name "org.freedesktop.systemd1" does not exist

In the kernel LTS the system works normally without xf86-video-intel

PS: xf86-video-intel causes screen sharing and recording errors so I shouldn't be using it.

metak commented on 2020-12-08 12:09 (UTC)

@adsoncicilioti Does this help?

@Ryaniskira Not really. If something depends on 'linux' that means it depends on default arch kernel package linux'. If you're building module then you should use dkms. What's the package you're installing?

RyanHakurei commented on 2020-12-06 00:26 (UTC)

Can you update the PKGBUILD so that this package provides Linux? As it stands if you want to just use this kernel and you have a package that depends on Linux, you have to also have another kernel installed as a dependency.

adsoncicilioti commented on 2020-11-25 02:55 (UTC) (edited on 2020-11-25 22:39 (UTC) by adsoncicilioti)

@metak, Hi.

After latests updates the GDM are in loop after try login.

This behavior begin after 5.9.10 update and nvidia-455.45.

I can login after restarts GDM service from an TTY#.

Any idea ?

See this:

journalctl -e | grep error
nov 25 19:38:03 arch gnome-shell[1496]: gnome-shell: Fatal IO error 11 (Recurso temporariamente indisponível) on X server :0.
nov 25 19:38:03 arch gsd-color[1554]: gsd-color: Fatal IO error 4 (Chamada de sistema interrompida) on X server :0.
nov 25 19:38:03 arch gsd-wacom[1553]: gsd-wacom: Fatal IO error 11 (Recurso temporariamente indisponível) on X server :0.
nov 25 19:38:03 arch gsd-media-keys[1566]: gsd-media-keys: Fatal IO error 11 (Recurso temporariamente indisponível) on X server :0.
nov 25 19:38:03 arch gsd-power[1582]: gsd-power: Fatal IO error 11 (Recurso temporariamente indisponível) on X server :0.
nov 25 19:38:03 arch gsd-keyboard[1555]: gsd-keyboard: Fatal IO error 11 (Recurso temporariamente indisponível) on X server :0.

metak commented on 2020-11-22 12:48 (UTC)

@Khaithang Hi. It's just a git repository that you can clone manually or use any of the aur helpers to install & update afterwards.

Khaithang commented on 2020-11-21 20:35 (UTC)

can i track this pkgbuild via asp/abs or some other tools?

metak commented on 2020-11-10 20:28 (UTC)

@adsoncicilioti As @fidasx said by default this PKGBUILD is set for gcc to detect and build native optimization, but for your CPU you can set 25 skylake if you don't want native.

fidasx commented on 2020-11-10 19:36 (UTC)

@adsoncicilioti just choose 33 Native optimizations autodetected by GCC. it will build for your specific CPU

adsoncicilioti commented on 2020-11-10 19:22 (UTC)

Yep @metak! Thanks for maintain this PKG!

I have one question... What number i can apply on _subarch for Comet lake CPU (Intel Core i5 10300H)?

fidasx commented on 2020-11-10 19:03 (UTC)

@metak nice! thx for your work on maintaining this PKG.

metak commented on 2020-11-10 17:27 (UTC)

@fidasx Thank you very much for your input. I've included (5.9.7-1) your nvidia uvm patch and I've enabled the AGP modules @lgomesf mentioned to fix nvidia 340xx driver build.

lgomesf commented on 2020-11-10 14:24 (UTC)

Solve , change , config to :

lgomesf commented on 2020-11-10 11:48 (UTC)

I am trying to compile Nvidis-340xx-DKMS and am having the following error. It compiles normal on CK, can you help me?
ERROR: modpost: "agp_backend_release" [/home/luiz/git-Download/nvidia-340xx/src/NVIDIA-Linux-x86_64-340.108-no-compat32/kernel/nvidia.ko] undefined!
ERROR: modpost: "agp_copy_info" [/home/luiz/git-Download/nvidia-340xx/src/NVIDIA-Linux-x86_64-340.108-no-compat32/kernel/nvidia.ko] undefined!
ERROR: modpost: "agp_unbind_memory" [/home/luiz/git-Download/nvidia-340xx/src/NVIDIA-Linux-x86_64-340.108-no-compat32/kernel/nvidia.ko] undefined!
ERROR: modpost: "agp_free_memory" [/home/luiz/git-Download/nvidia-340xx/src/NVIDIA-Linux-x86_64-340.108-no-compat32/kernel/nvidia.ko] undefined!
ERROR: modpost: "agp_backend_acquire" [/home/luiz/git-Download/nvidia-340xx/src/NVIDIA-Linux-x86_64-340.108-no-compat32/kernel/nvidia.ko] undefined!
ERROR: modpost: "agp_find_bridge" [/home/luiz/git-Download/nvidia-340xx/src/NVIDIA-Linux-x86_64-340.108-no-compat32/kernel/nvidia.ko] undefined!
make[2]: *** [scripts/Makefile.modpost:111: /home/luiz/git-Download/nvidia-340xx/src/NVIDIA-Linux-x86_64-340.108-no-compat32/kernel/Module.symvers]

fidasx commented on 2020-11-08 22:17 (UTC)


glad i helped you!

adsoncicilioti commented on 2020-11-08 22:15 (UTC)

GREAT @fidasx!!

That fixes! nvidia_uvm loaded and apps that uses CUDA working! Thanks!

fidasx commented on 2020-11-08 21:11 (UTC)


download the patch and save it inside the linux-clear with name nvidia-5.9-uvm-fix.patch then open the PKGBUILD and add it under the other patches


Add nvidia-uvm.patch
echo "Applying nvidia-uvm.patch ..."
patch -Np1 -i ../nvidia-5.9-uvm-fix.patch

adsoncicilioti commented on 2020-11-08 20:55 (UTC)

Thanks @fidasx! But, i'm so sorry.. I don't know how make this patch.

Like i says on commentary below, i've use the compiled package from @metak repo. When i try compiles with yay, using modprobed-db to made local .conf modules, my touchpad don't work.

Do you can give me an clarification about how do this patch?

fidasx commented on 2020-11-08 10:59 (UTC)

@adsoncicilioti you need to patch the kernel for nvidia_uvm to work

use this if you want

adsoncicilioti commented on 2020-11-08 05:50 (UTC) (edited on 2020-11-08 05:51 (UTC) by adsoncicilioti)

Hi @metak! nice work with Clear Kernel. I'm using the 5.9.6 from your repo, and all boot parms requireds by Clear, including module.sig_unenforce.

With standard and Zen Kernel, all nvidia modules are loaded correctly - nvidia, nvidia_drm, nvidia_modset, nvidia_uvm. On clear nvidia_uvm module not is loaded, and apps like blender, and other that uses CUDA cores, not recognize the nvidia (cuda device).

I've try with # modprobe nvidia_uvm command but the error is returned: modprobe: ERROR: could not insert 'nvidia_uvm': Unknown symbol in module, or unknown parameter (see dmesg).

Running dmesg i got:

  184.986139] nvidia_uvm: module uses symbols from proprietary module nvidia, inheriting taint.
[  184.986268] nvidia_uvm: Unknown symbol radix_tree_preloads (err -2)

I've don't understand why the standard and zen Kernels loads and Clear not. Any idea?

metak commented on 2020-10-27 16:10 (UTC)

@krathalan I can change that with next update.

krathalan commented on 2020-10-27 02:58 (UTC) (edited on 2020-10-27 02:59 (UTC) by krathalan)

@metak some users (like myself) may have their ~/.config/modprobed.db file actually be a symlink to somewhere else that is shared with other systems. You should change the -f on line 215 from if [ -f $HOME/.config/modprobed.db ]; then to -e, so it reads if [ -e $HOME/.config/modprobed.db ]; then.

From man test, -f checks that: FILE exists and is a regular file (emphasis mine). -e only checks that: FILE exists. -f returns false if the file exists but is a symlink.

NateR100 commented on 2020-09-20 21:12 (UTC)

So once I've built the kernel with my customized config options from nconfig, I see the config file is copied into the main directory. How do I use this when I want to recompile again, or when the kernel updates and there are more config options (like make oldconfig would show you)?

enihcam commented on 2020-08-29 14:55 (UTC)

Thanks, metak!

metak commented on 2020-08-29 13:43 (UTC)

@enihcam Change urls, remove patches and configs added in this package that you don't need. Comment out the Patch0129 otherwise the build will fail.

enihcam commented on 2020-08-29 13:28 (UTC)

@metak, I need patches from instead. Can I just grab your PKGBUILD and change few URLs? what else should I change?

metak commented on 2020-08-19 15:46 (UTC)

@Gonzo2028 Works fine on my end. Please try building with makepkg instead of aur helpers and clean your build directory first.

@Tjuh Well that's up to clear team to update first. The -current & -preempt-rt are both updated to latest clear versions 5.8.0.rc7 & 5.6.19.rt11 - upstream.
5.9rc1 was released and clear team will probably update their -current soon to it.

Tjuh commented on 2020-08-19 14:56 (UTC)

Plese update current and preempt-rt versions.

Tjuh commented on 2020-08-19 14:56 (UTC)

Plese update current and preempt-rt versions.

Gonzo2028 commented on 2020-08-16 18:31 (UTC)

Does not build for me, using localmodconfig=y.

patching file virt/kvm/kvm_main.c ==> FEHLER: Ein Fehler geschah in prepare(). Breche ab...

andsmir commented on 2020-06-26 08:17 (UTC)

@metak Oh, thank you! Perfectly works with makepkg...

metak commented on 2020-06-25 14:47 (UTC)

@andsmir Looks like _localmodcfg is not applying in your build... I just tried it again on my pc and it works as it should. Modified _localmodcfg to "y" in PKGBUILD file and built with makepkg. If you're using any aur helper, please try building with makepkg.

andsmir commented on 2020-06-25 06:54 (UTC)

@metak I tried to set _localmodcfg=y and nothing changed. Here is my log:

metak commented on 2020-06-23 09:38 (UTC)

@andsmir You'll have to provide more information than that, specially what happens in the prepare() section. Also, it's enough to just set _localmodcfg=y.

andsmir commented on 2020-06-22 11:12 (UTC) (edited on 2020-06-22 11:46 (UTC) by andsmir)

Seems that usage of modprobedb database is currently broken. It worked at 5.6, as I remember.
188 modules are in /home/user/.config/modprobed.db

find /lib/modules/5.7.5-1-clear/ -type f | wc -l

includao commented on 2020-05-30 12:35 (UTC)

If you are having problems installing nvidia 440.84 dkms with CL kernel, read this:

lucasrizzini commented on 2020-05-22 05:29 (UTC)

What are the odds you add the add-sysctl-to-disallow-unprivileged-CLONE_NEWUSER-by.patch to this baby?

xuanruiqi commented on 2020-05-17 12:10 (UTC)

@metak: also, _gcc_more_v can be bumped to 20200428.

xuanruiqi commented on 2020-05-17 12:09 (UTC)

If you use nvidia-dkms, then you must configure SECTION_MISMATCH_WARN_ONLY (Kernel hacking -> Compile-time checks and compiler options -> Make section mismatch errors non-fatal) because somehow there's a mismatch problem in the NVIDIA driver, somehow.

Also, GCC 10 is missing GCC plugins support for some reason; I guess it would be fixed by upstream soon.

metak commented on 2020-05-15 15:46 (UTC)

@wshuman3 & @tam1m Thank you for chiming in.

@fidasx nvidia issues are nothing new. Here's the latest reported on clear issue tracker.

tam1m commented on 2020-05-15 15:11 (UTC) (edited on 2020-05-15 15:12 (UTC) by tam1m)

@fidasx That's because gcc10 is now default in arch.

I wasn't able to get nvidia-dkms working with linux-clear and gcc10. As a quick workaround, dowgrading to gcc9 and rebuilding linux-clear AND nvidia-dkms works. This however makes nvidia-dkms incompatible with every linux package compiled with gcc10 (default arch kernel for example). So keep that in mind

fidasx commented on 2020-05-15 15:04 (UTC)


getting errors on nvidia-dkms driver for 5.6.13-1

WARNING: /var/lib/dkms/nvidia/440.82/build/nvidia.o(.text.unlikely+0x317): 
Section mismatch in reference from the function nv_module_init() to the 
function .init.text:nvlink_core_init()
The function nv_module_init() references
the function __init nvlink_core_init().
This is often because nv_module_init lacks a __init 
annotation or the annotation of nvlink_core_init is wrong.

FATAL: modpost: Section mismatches detected.
make[2]: *** [scripts/Makefile.modpost:93: __modpost] Error 1
make[1]: *** [Makefile:1603: modules] Error 2
make[1]: Leaving directory '/usr/lib/modules/5.6.13-1-clear/build'
make: *** [Makefile:81: modules] Error 2

any idea? have been using your kernel for a long time never had this problem

Tjuh commented on 2020-05-15 13:22 (UTC)

Great! Much obliged.

wshuman3 commented on 2020-05-15 13:12 (UTC)

@Tjuh the documentation for kernel params can be found here:

Tjuh commented on 2020-05-15 13:09 (UTC)

I think it would be helpful to get some descriptions on those kernel parameters. What do they do and why are they usefull to have. I know some of them like - quiet, rootfstype and rw - but I'm unfamiliar with the rest.

Tjuh commented on 2020-05-11 13:14 (UTC)

I understand. Thank you.

metak commented on 2020-05-11 13:08 (UTC)

@Tjuh That's because of gcc10 in testing repo that you're using? You can check in arch linux's repo all the commits related to gcc10 warnings.
Clear Linux still uses gcc9 so we'll see when they'll update that.

Tjuh commented on 2020-05-11 12:59 (UTC)

==> Starting build()...
scripts/kallsyms.c: In function ‘read_symbol’:
scripts/kallsyms.c:222:2: warning: ‘strcpy’ writing between 1 and 128 bytes into a region of size 0 [-Wstringop-overflow=]
  222 |  strcpy(sym_name(sym), name);
      |  ^~~~~~~~~~~~~~~~~~~~~~~~~~~
arch/x86/kvm/../../../virt/kvm/kvm_main.c: In function ‘__kvm_gfn_to_hva_cache_init’:
arch/x86/kvm/../../../virt/kvm/kvm_main.c:2236:42: error: ‘nr_pages_avail’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
 2236 |  for ( ; start_gfn <= end_gfn; start_gfn += nr_pages_avail) {
      |                                ~~~~~~~~~~^~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors
make[2]: *** [scripts/ arch/x86/kvm/../../../virt/kvm/kvm_main.o] Error 1
make[1]: *** [scripts/ arch/x86/kvm] Error 2
make[1]: *** Waiting for unfinished jobs....
make: *** [Makefile:1683: arch/x86] Error 2
make: *** Waiting for unfinished jobs....
kernel/bpf/core.c: In function ‘bpf_patch_insn_single’:
cc1: warning: writing 8 bytes into a region of size 0 [-Wstringop-overflow=]
In file included from kernel/bpf/core.c:21:
./include/linux/filter.h:550:20: note: at offset 0 to object ‘insnsi’ with size 0 declared here
  550 |   struct bpf_insn  insnsi[0];
      |                    ^~~~~~
==> ERROR: A failure occurred in build().

No problem with previous iteration, but this one fails.

rvlobato commented on 2020-05-08 21:21 (UTC)

@metak I see, thanks. @lucasrizzini thanks.

metak commented on 2020-05-08 10:35 (UTC) (edited on 2020-05-08 10:36 (UTC) by metak)

@rvlobato As @lucasrizzini has said, you can easily add that patch and build it for your self. I removed fsync patch because clear team don't feel comfortable enough to include it.

lucasrizzini commented on 2020-05-07 23:30 (UTC)

No. You have to patch the source manually. Just add "patch -Np1 -i /path/to/patch/file" just below "### Add Clearlinux patches" around line #108.

Patch ->

rvlobato commented on 2020-05-07 23:21 (UTC)

Hello, is fsync still working on linux-clear?

alitajelsir commented on 2020-05-01 03:47 (UTC)

@metak Thanks.

metak commented on 2020-04-30 13:52 (UTC)

@AliTajelsir Hi. You can ignore that warning.$srcdir
objtool file contains some path references to where the package was built. Regular arch kernel should have that too.

alitajelsir commented on 2020-04-30 07:08 (UTC)

I am getting the following warning while building:

==> WARNING: Package contains reference to $srcdir

bwbuhse commented on 2020-04-29 21:58 (UTC)

@metak Wow thanks! For some reason I had most of base-devel but not flex or bison. I must have messed something up when I was installing Arch and just not needed flex until now. That fixed it though, thank you :)

metak commented on 2020-04-29 21:51 (UTC)

@bwbuhse Hi. If you want to build packages from AUR you have to have base-devel group installed. flex is in there.

bwbuhse commented on 2020-04-29 21:41 (UTC)

I'm having an error while trying to install this (using yay).

Hunk #2 succeeded at 4867 (offset 104 lines).
Setting config...
Enable extra stuff from arch kernel...
  HOSTCC  scripts/basic/fixdep
  HOSTCC  scripts/kconfig/conf.o
  HOSTCC  scripts/kconfig/confdata.o
  HOSTCC  scripts/kconfig/expr.o
  LEX     scripts/kconfig/lexer.lex.c
/bin/sh: flex: command not found
make[1]: *** [scripts/ scripts/kconfig/lexer.lex.c] Error 127
make: *** [Makefile:568: olddefconfig] Error 2
==> ERROR: A failure occurred in prepare().

Any ideas?

dedguy21 commented on 2020-04-27 00:22 (UTC) (edited on 2020-04-27 00:23 (UTC) by dedguy21)

Nvidia-dkms 1) make sure to create a blacklist.conf in you're /etc/modprobe.d/ folder

2) blacklist nouveau blacklist nvidia.fb (this is actually a nouvea driver misabled so it has to be blacklisted)

3) beyond that make sure You include the command lines for Linux clear in your boot script. And For love of God if you're stil using grab switch to systed-boot

idimitro commented on 2020-04-27 00:14 (UTC)

Hi, is anybody using the proprietary Nvidia drivers with this kernel? I guess I will need the dkms version. Any problems, tips/tricks, hints for using the kernel it NVIDIA blob? Thanks.

metak commented on 2020-04-18 09:51 (UTC)

@vivaeltopo It's an issue in mkinitcpio which I already reported.

vivaeltopo commented on 2020-04-18 09:26 (UTC) (edited on 2020-04-18 09:26 (UTC) by vivaeltopo)

I am getting this warning while creating the fallback image.

>> mkinitcpio -v -p linux-clear
    adding file: /usr/lib/
    adding file: /usr/bin/fsck.vfat
    adding file: /usr/bin/fsck.xfs
    adding file: /usr/sbin/fsck
cp: cannot stat '(builtin)': No such file or directory
==> Generating module dependencies
==> Creating gzip-compressed initcpio image: /boot/initramfs-linux-clear-fallback.img
==> WARNING: errors were encountered during the build. The image may not be complete.

metak commented on 2020-04-10 19:48 (UTC)

@MikeDee Aur helpers are not supported in arch. You're supposed to import the pgp key yourself.
Aur hosts pkgbuild files (recipes) for bulding arch packages and my pinned comment point to my repository where I provide built binary packages.
These are built from source on arch for arch, while linux-clear-bin as the name says, takes the built binary code from clear and repackages it. There's a difference.

MikeDee commented on 2020-04-07 00:15 (UTC) (edited on 2020-04-08 21:12 (UTC) by MikeDee)

I got the following error in Octopi when trying to install linux-clear 5.5.15-2:

Installing selected packages...

resolving dependencies... looking for conflicting packages... Packages (1) linux-clear-5.5.15-2 Total Installed Size: 52.97 MiB checking keys in keyring checking package integrity error: linux-clear: missing required signature error: failed to commit transaction (package missing required signature) Errors occurred, no packages were upgraded.

Command finished with errors!

Also, the pinned comment above says the repository contains binaries. This package is source code, not a binary. I went the AUR linux-clear-bin to get the kernel binary.

xuanruiqi commented on 2020-03-18 09:42 (UTC)

@metak: anyways, I fixed the problem. I think the problem is on my own side w/ wrong kernel config.

metak commented on 2020-03-16 20:30 (UTC)

@MikeDee Sorry, that's beyond the scope of this package. You can ask on Clear Linux's github issue page though?

@xuanruiqi Post the error and maybe someone can help you. (I'm not using nvidia)

xuanruiqi commented on 2020-03-16 09:48 (UTC) (edited on 2020-03-16 09:48 (UTC) by xuanruiqi)

I'm having trouble with compiling nvidia-dkms on the current version. Is everyone else doing fine?

MikeDee commented on 2020-03-14 16:08 (UTC)

Can the MuQSS CPU scheduler be added to this kernel? (I tried by adding a patch, but it didn't work).

lybin commented on 2020-02-27 09:45 (UTC)

INFO: if you will get error: modprobe: ERROR: could not insert 'nvidia': No such device Blacklist it: cat /etc/modprobe.d/nvidia.conf blacklist nouveau blacklist nvidiafb

bjoe2k4 commented on 2020-02-27 07:31 (UTC)

Currently the repo domain seems to have some dnssec issues:

metak commented on 2020-02-24 20:15 (UTC)

@maxpshaw It depends which boot loader you're using. I'm using grub so I have a custom file /etc/grub.d/40_custom with all the different clear linux entries.

maxpshaw commented on 2020-02-24 19:14 (UTC)

@metak Hi, Where can I adjust the boot cmd line?

metak commented on 2020-02-24 10:31 (UTC)

@maxpshaw Hi. That's the default boot cmd line for this kernel in clear linux and if you want to change it however you want you're free to do so, the kernel will still boot even without any adjustment.

maxpshaw commented on 2020-02-24 05:47 (UTC)

@metak Hi, is it necessary to adjust boot cmd line to use clr-boot-manager? I was able to boot without adjust the boot cmd line.

eca commented on 2020-01-17 17:08 (UTC)

@metak - Oh I see, adding SigLevel = Optional for now. Thanks!

metak commented on 2020-01-16 20:18 (UTC) (edited on 2020-01-16 20:18 (UTC) by metak)

@eca Hi, yes OBS (open build service) just recently added support for zstd packages in arch and the part that generates the signatures wasn't yet updated. ISSUE REPORT

But, on the other hand the signatures for pacman in OBS are broken since a long time ago so there's that... if you look a page or two back in the comments you'll find my explanation.

eca commented on 2020-01-16 19:49 (UTC)

@metak - I think your repo is missing pgpsigs. error: linux-clear: missing required signature.

idimitro commented on 2020-01-09 11:27 (UTC)

@metak - the config and compile optimizations will be great addition. Please introduce them. Thanks.

metak commented on 2020-01-09 10:46 (UTC)

@idimitro Well, I can add -j$(nproc) to make command, but _subarch backup is beyond the scope of this package. That's up to user.

idimitro commented on 2020-01-09 10:29 (UTC)

@metak - if possible please extract the config. The kernel config is one part of the customizations. Other things that can be extracted are: arch selection "_subarch=" and "make -j8 bzImage modules" (allow multithread compilation as it currently compiles in single core). The end target is to have the package build automatically + customizations with aurto :)

metak commented on 2020-01-09 09:40 (UTC) (edited on 2020-01-09 09:50 (UTC) by metak)

@abdulocracy For i915 module, that's builtin with this kernel.

cat /lib/modules/$(uname -r)/modules.builtin | grep i915


Edit: let me check some more about that error...

metak commented on 2020-01-09 09:37 (UTC)

@idimitro For now the pkgbuild just copies the .config file for archival purposes and you can't easily reuse it later. If it would be useful I might add support for reusing the old file just like it is in linux-lqx package? Check Here.

abdulocracy commented on 2020-01-09 08:07 (UTC) (edited on 2020-01-09 09:05 (UTC) by abdulocracy)

I'm getting these errors on a i7-8550U machine, upon sleeping and resuming:

Jan 09 09:05:05 abdulkadir-xps kernel: [drm:intel_power_domains_verify_state] *ERROR* power well DC off state mismatch (refcount 0/enabled 1)
Jan 09 09:05:05 abdulkadir-xps kernel: DMAR: DRHD: handling fault status reg 2
Jan 09 09:05:05 abdulkadir-xps kernel: DMAR: [INTR-REMAP] Request device [f0:1f.0] fault index 0 [fault reason 37] Blocked a compatibility format interrupt request

Also, the second console option in the upstream cmdline causes my machine to boot to a black screen.

Edit: Another peculiarity, modinfo -p i915 states that the module was not found.

idimitro commented on 2020-01-09 07:36 (UTC)

Hi, the PKGBUILD file has options to optimize the kernel for the specific machine it will run. But how are those changes supposed to be preserved after upgrades? I couldn't find an option for the configs to be read from the outside the build directory. Any suggestions?

tam1m commented on 2019-12-13 14:50 (UTC)

@metak Works great! Thanks

metak commented on 2019-12-13 13:51 (UTC)

@tam1m It's also reported on archlinux bug tracker. I've applied the patch from the bug repport in 5.4.3-2 if you can test it?

tam1m commented on 2019-12-13 10:29 (UTC)

I'm experiencing this bug bug with linux-clear 5.4.2-1. The e1000e driver constantly crashes.

metak commented on 2019-12-08 12:14 (UTC) (edited on 2019-12-08 12:15 (UTC) by metak)

@danhyal I've added it, but if there's any complaints/regressions/etc. I'll have to remove it. So I hope you and others can test it and report back.

amelia commented on 2019-12-07 22:44 (UTC)

consider adding the fsync patch?

metak commented on 2019-12-01 13:40 (UTC) (edited on 2019-12-01 13:41 (UTC) by metak)

@bjoe2k4 You can manually re-add the key (download the key with wget, import and sign it).

pub   rsa2048 2013-05-19 [SC] [expires: 2022-01-24]
uid           [  full  ] home:metakcahura OBS Project <>

But it's more or less useless since signatures are broken in open-build-service for pacman. Might as well set SigLevel = Never.

bjoe2k4 commented on 2019-12-01 12:27 (UTC)

The signature key expired yesterday :-(

pub rsa2048 2013-05-19 [SC] [expired: 2019-11-30] A6BD5EECD7B96FC10806355C50AAD02AEB1F61C9 uid [ expired] home:metakcahura OBS Project

Morisson commented on 2019-11-17 12:52 (UTC)

Thank you vary much

metak commented on 2019-11-17 12:32 (UTC)

@Morisson 5.3.11-3 (clr 5.3.11-868) enable ACPI_TABLE_UPGRADE

Morisson commented on 2019-11-17 06:43 (UTC) (edited on 2019-11-17 06:44 (UTC) by Morisson)

Hi. Can you change kernel config from your reposotory? ACPI_TABLE_UPGRADE=y I need add patched dsdt for my laptop. Please)

metak commented on 2019-11-08 16:48 (UTC)

@danhyal I'm afraid that's not possible and it kinda defeats the purpose of this package IMO. I already added two patches (cpu optimizations and acs overrides) because they're small and don't change the kernel in radical way. On the other hand -ck patches are much bigger, touch many more files, change a bunch of existing config options, etc. There could also be clashes between -ck and -clear patches... Besides, this kernel is already optimized for desktop/laptop users so I don't really see a point?

amelia commented on 2019-11-08 15:40 (UTC)

Could you make a seperate package with the ck patchset applied? Since most people using this kernel are desktop/laptop users, it would be useful.

metak commented on 2019-11-05 12:12 (UTC) (edited on 2019-11-05 12:14 (UTC) by metak)

@danhyal 5.3.8-4 (clr 5.3.7-853) enable CONFIG_NETFILTER_INGRESS

amelia commented on 2019-11-05 09:01 (UTC)

Could you add the config option CONFIG_NETFILTER_INGRESS

dedguy21 commented on 2019-11-04 21:14 (UTC) (edited on 2019-11-04 21:45 (UTC) by dedguy21)

Ah I am running mkinicpio 26 not 27 I will update. Thanks.

_----------- Yes everything is working now.

Thanks again for your time.

metak commented on 2019-11-04 19:10 (UTC)

@dedguy21 I'm a bit confused as to what you want to achieve? You don't need to do 1) and 2). New version of mkinitcpio=27-1 includes hook and preset files so that not every kernel package has to provide them. Just build this package as any other or use the binaries provided in my repo.

dedguy21 commented on 2019-11-04 18:51 (UTC)

@ metak Apologies for the possible noobie questions....

I have the linux-clear kernel in the /usr/lib/module folder.

1) How can I write the proper compressed image to the /boot folder 2) and the .preset files to the mkinitcpio.d folder

I used yay for installation, should I just git files and run make install?

metak commented on 2019-11-04 11:48 (UTC) (edited on 2019-11-04 14:35 (UTC) by metak)

@dedguy21 Hello. linux-clear package is the "main" kernel, it's the default kernel in clearlinux and clear team focuses the most on it. With that said, linux-clear-current is used mostly for testing purposes when kernel nears towards the latest release candidates (and first .0 release and .1 sometimes) then clear team rebase their patches and test this new kernel. These patches are then moved to linux-clear when everything seems ok. After that they don't update the linux-clear-current much, it's stuck on 5.3.1 and I expect it will get updated with 5.4rc7? I don't know. The kernel .config file is pretty much the same as they try to sync changes between their kernels, the exception being patch for exfat patch and some crypto modules that were introduced in linux-clear.

Yes, I'm maintaining both of them and as you can see they're two separate packages and neither is being abandoned. Use linux-clear and later in kernel development cycle if you want to test the latest rc then try linux-clear-current.

I follow stock arch's kernel way of packaging and those changes were brought by Changes for new kmod and mkinitcpio hooks.

dedguy21 commented on 2019-11-04 00:00 (UTC) (edited on 2019-11-04 01:18 (UTC) by dedguy21)

First of all I wanted to say thanks for the Clear Linux packages. I always thought if you could marry their kernel with Arch you would have the perfect distro.

So wondering why does linux-clear package has a kernel number higher than linux-clear-current package?

And are you maintaining both packages separately? Or is one being abandoned for the other?

Also, when I install this package, I notice that there isn't an image for this created to the /boot or any .preset in mkinitcpio.d

I located the file in /usr/lib/module

...just wondering if this was done on purpose on your end.

Thanks again

tam1m commented on 2019-10-30 14:09 (UTC) (edited on 2019-10-30 14:09 (UTC) by tam1m)

@metak Ok I didn't know that /usr/lib/modules/<kernel>/build/.config existed. Copying it to another location seems unnecessary, especially since /usr/lib/kernel/.config would nowhere be mentioned in the "Arch World". CONFIG_IKCONFIG is at least enabled by default in arch's linux package. That's why I thought it would be nice to have it in linux-clear, too.

metak commented on 2019-10-30 13:55 (UTC)

@tam1m That's just a packaging difference. In arch kernel there's a copy of kernel config file in -headers package. /usr/lib/modules/<kernel>/build/.config
I can add a line that copies config file like it's in clearlinux package along with enabling ikconfig.

tam1m commented on 2019-10-30 13:41 (UTC) (edited on 2019-10-30 13:42 (UTC) by tam1m)

@metak Sounds good to me. Having it as a module should be enough. Also regarding this comment from the issue tracker, we don't have the /usr/lib/kernel file. At least I don't..

metak commented on 2019-10-30 12:27 (UTC)

@tam1m With next update I can enable CONFIG_IKCONFIG=m (as a module) and CONFIG_IKCONFIG_PROC=y? This was already brought up upstream on ClearLinux issue tracker and they decided not to enable it because not many people use this and it wastes more memory.

niklaszantner commented on 2019-10-30 12:10 (UTC)


You were right, thanks a lot!

tam1m commented on 2019-10-30 12:07 (UTC) (edited on 2019-10-30 12:07 (UTC) by tam1m)

I think it would be good to enable CONFIG_IKCONFIG_PROC/CONFIG_IKCONFIG by default. Is that something you would consider?

metak commented on 2019-10-30 11:24 (UTC)

@niklaszantner Do you have sha256 and sha512 modules listed in MODULES=() in /etc/mkinitcpio.conf? If you do, try removing them as they're built into the kernel.

niklaszantner commented on 2019-10-30 11:01 (UTC) (edited on 2019-10-30 11:02 (UTC) by niklaszantner)

Am I the only one who gets the following errors past build on install:

  -> Running build hook: [keymap]
  -> Running build hook: [encrypt]
  -> Running build hook: [resume]
  -> Running build hook: [filesystems]
  -> Running build hook: [fsck]
==> ERROR: module not found: `sha256'
==> ERROR: module not found: `sha512'
==> Generating module dependencies
==> Creating gzip-compressed initcpio image: /boot/initramfs-linux-clear-fallback.img
==> WARNING: errors were encountered during the build. The image may not be complete.
error: command failed to execute correctly

wan109020 commented on 2019-10-26 15:57 (UTC)

@metak Big thanks for adding them to the package. Can now set kernel.yama.ptrace_scope=1 and apparmor is working properly. No issues encountered so far.

metak commented on 2019-10-26 14:12 (UTC)

@wan109020 In 5.3.7-3, I've enabled Yama, Apparmor, SELinux and TOMOYO security options just like in stock arch kernel. Please test and report.

wan109020 commented on 2019-10-26 09:47 (UTC)

Package works great, Just that is it possible to add Yama support? Would like to set the kernel.yama.ptrace_scope=1 as it's the Arch default.

Also, chrome sandbox seems to be more powerful with Yama (chrome://sandbox)

Would also be nice to have Apparmor/SELinux support too


rvlobato commented on 2019-10-18 22:55 (UTC) (edited on 2019-10-18 22:55 (UTC) by rvlobato)

@metak It's working for me, as I said, I just blacklisted the nvidiafb module and the nvidia started to work. Thanks.

metak commented on 2019-10-18 19:24 (UTC)

@rvlobato @astroty Sorry I can't be more helpful with the nvidia because I don't have any nvidia cards.

@astroty The extramodules-clear folder was removed in sync with the stock arch kernel package.

astroty commented on 2019-10-18 18:57 (UTC) (edited on 2019-10-18 19:01 (UTC) by astroty)

After some searching, it seems that my Nvidia driver was failing to load because of this flag in my /etc/default/grub: pci=nommconf I didn't need to do that in another kernel I have though, so I don't know what might've caused this

Also, is it just me or are these new packages not generating a new extramodules-clear folder?

rvlobato commented on 2019-10-15 22:58 (UTC)

@calin @metak I was having the same issue, seems that the conflict was between nvidiafb and nvidia, after blacklisting the nvidiafb (the nvidia-dkms blacklist nouveau) the nvidia modules were loaded.

metak commented on 2019-10-11 18:29 (UTC)

@calin I've already disabled CONFIG_MODULE_SIG_FORCE by default to make it easier and you should not need module.sig_unenforce. Check logs for more clues?

calin commented on 2019-10-10 08:54 (UTC)

The nvidia drivers fail to load even when using module.sig_unenforce, they used to work, not anymore unfortunately, is anyone else having this issue?

metak commented on 2019-10-01 20:03 (UTC)

@Elanzer Sorry about that. The pkgbuild file in obs wasn't synced fully and dell smbios should've been enabled long ago (as it is here in aur). The repo binary is already building.

Elanzer commented on 2019-10-01 19:31 (UTC)

@metak can you compile lts-2018 with CONFIG_DELL_SMBIOS_SMM=y or CONFIG_DELL_SMBIOS_WMI=y?? I dont know which one works for mine. My touchpad is weird with 5.3 so I rolledback to LTS. And with the older kernel i'm getting SMBIOS messages again. My laptop is old and it takes forever to compile a kernel, your binaries are a life-saver. Thanks.

metak commented on 2019-09-20 18:45 (UTC)

@toki1990 I believe it should work on amd cpus, but I don't have amd cpus to try. If you can post some logs maybe someone can help.

toki1990 commented on 2019-09-19 16:57 (UTC)

kernel build complete without error. but kernel not work over amd ryzen processor. is there a way run over amd ?

xuanruiqi commented on 2019-09-08 04:30 (UTC)

It seems that I need the cpio package to correctly build linux-clear. Maybe this is due to some kernel options I enabled, but not sure which ones.

metak commented on 2019-08-26 18:12 (UTC)

@btd1337 That's for archlinux Trusted Users and Developers, not some random aur maintainers. ;-)

btd1337 commented on 2019-08-25 21:24 (UTC) (edited on 2019-08-25 21:25 (UTC) by btd1337)

@metak Why you don't use the Arch Servers? Like here:

metak commented on 2019-08-25 12:35 (UTC)

@btd1337 Hey. OpenSUSE Build Service support for archlinux is bad. I reported this issue years ago.
Basically, the build service triggered the package rebuild (after my last commit), but since there's no support for arch to increment pkrel +1 we end up with (invalid or corrupt package (checksum)).
The package itself is fine though. You can install it with pacman -U, but I do realize this is inconvenient. Maybe I should just remove the instructions for binaries if it's causing confusion/problems?

btd1337 commented on 2019-08-25 02:19 (UTC) (edited on 2019-08-25 02:20 (UTC) by btd1337)

:: File /var/cache/pacman/pkg/linux-clear-5.2.9-6-x86_64.pkg.tar.xz is corrupt (invalid or corrupt package (checksum)).
Do you want to delete it? [Y / n] y
:: File /var/cache/pacman/pkg/linux-clear-headers-5.2.9-6-x86_64.pkg.tar.xz is corrupt (invalid or corrupt package (checksum)).
Do you want to delete it? [Y / n] y
error: failed to submit transaction (invalid or corrupted package)
Errors occurred and therefore no packages were updated.
Error installing repo packages

metak commented on 2019-08-18 22:36 (UTC)

@btd1337 That's because OpenSUSE Build Service "support" for pgp-signed repositories is broken.

btd1337 commented on 2019-08-18 22:28 (UTC)

@metak Just adding the key, this error persists:

error: linux-clear: unsupported signature format
error: linux-clear-headers: unsupported signature format

It's necessary to add SigLevel = Never in order to error message disapear, but this is not a good practice.

metak commented on 2019-08-18 20:01 (UTC)

@btd1337 You're supposed to download the key and then import it. It's all in the link in the pinned comment where it says 'Then, import the key.'

wget ""
pacman-key --add home_metakcahura_kernel_Arch_Extra_standard.key

If you're still unable to import the key then just add SigLevel = Never before the Server = line for this repo in /etc/pacman.conf.

btd1337 commented on 2019-08-18 19:04 (UTC) (edited on 2019-08-18 19:04 (UTC) by btd1337)

@Metak What's wrong?

sudo pacman-key --add
gpg: can't open '': No file or directory
==> ERROR: A specified key file could not be added to the keychain.
sudo pacman-key --add  <\>
zsh: arquivo ou diretório inexistente:>
sudo pacman-key --lsign-key 50AAD02AEB1F61C9                                       1 ↵
  -> Locally signing the key 50AAD02AEB1F61C9...
==> Updating trust database...
gpg: marginals needed: 3  completes needed: 1  trust model: pgp
gpg: depth: 0  valid:   1  signed:   7  trust: 0-, 0q, 0n, 0m, 0f, 1u
gpg: depth: 1  valid:   7  signed:  89  trust: 1-, 0q, 0n, 6m, 0f, 0u
gpg: depth: 2  valid:  86  signed:  12  trust: 86-, 0q, 0n, 0m, 0f, 0u
gpg: next verification of the trusted database to 2019-10-12
error: linux-clear: unsupported signature format
error: linux-clear-headers: unsupported signature format

metak commented on 2019-08-18 12:43 (UTC)

@btd1337 I've disabled the CONFIG_MODULE_SIG_FORCE, because many users build modules and to just make it easier. Try 5.2.9-4.

btd1337 commented on 2019-08-18 00:29 (UTC) (edited on 2019-08-18 07:08 (UTC) by btd1337)

@metak I have this error during boot:

Failed to start Load Kernel Modules - systemd-modules-load.service
systemd-modules-load[2218]: Failed to find module 'speakup_soft'
systemd-modules-load[2218]: Failed to insert module 'vboxdrv': Required key not available
systemd-modules-load[2218]: Failed to insert module 'vboxpci': Required key not available
systemd-modules-load[2218]: Failed to insert module 'vboxnetadp': Required key not available
systemd-modules-load[2218]: Failed to insert module 'vboxnetflt': Required key not available
systemd-modules-load[2218]: Failed to insert module 'vboxdrv': Required key not available
systemd-modules-load[2218]: Failed to insert module 'vboxnetadp': Required key not available
systemd-modules-load[2218]: Failed to insert module 'vboxnetflt': Required key not available
systemd-modules-load[2218]: Failed to insert module 'vboxpci': Required key not available
systemd[1]: systemd-modules-load.service: Failed with result 'exit-code'.
systemd[1]: Failed to start Load Kernel Modules.

My /boot/default/grub file contains:

GRUB_CMDLINE_LINUX_DEFAULT="console=tty0 console=ttyS0,115200n8 initcall_debug tsc=reliable no_timer_check noreplace-smp kvm-intel.nested=1 rootfstype=ext4,btrfs,xfs intel_iommu=igfx_off cryptomgr.notests rcupdate.rcu_expedited=1 rcu_nocbs=0-64 module.sig_unenforce rw"

module.sig_unenforce is as a second to last parameter. What am I doing wrong that step 3 doesn't work?

Thanks in advance!


sudo vboxreload                                                                                                                                            ✔
Unloading modules: 
Loading modules: modprobe: ERROR: could not insert 'vboxnetadp': Required key not available
modprobe: ERROR: could not insert 'vboxnetflt': Required key not available
modprobe: ERROR: could not insert 'vboxpci': Required key not available
modprobe: ERROR: could not insert 'vboxdrv': Required key not available

Elanzer commented on 2019-08-17 19:44 (UTC)

SMBIOS messages are gone, thanks.

metak commented on 2019-08-17 19:31 (UTC) (edited on 2019-08-17 19:34 (UTC) by metak)

@btd1337 If you've already created $XDG_CONFIG_HOME/modprobed.db as per WIKI then you only have to set _localmodcfg=y. That's all.

btd1337 commented on 2019-08-17 19:20 (UTC) (edited on 2019-08-17 19:21 (UTC) by btd1337)

What do this variable Pkgbuild, Line 57: _localmodcfg= needs?

Is this?


metak commented on 2019-08-17 17:40 (UTC)

@btd1337 I've updated the repo instructions in the pinned comment.

@Elanzer I've enabled CONFIG_DELL_SMBIOS_SMM=y, can you try the latest version?

Elanzer commented on 2019-08-17 15:37 (UTC)

I'm getting this message on boot:

dell_smbios: No SMBIOS backends available (wmi: -19, smm: -19) dell_smbios: No dell-smbios drivers are loaded

System is running fine, but still worries me. It's was fixed upstream, according

I think dell_smbios is set to user 'smm' but only 'wmi' is available with clear-linux

btd1337 commented on 2019-08-17 15:34 (UTC)


error: home_metakcahura_kernel_Arch_Extra_standard: Signature of "home: metakcahura OBS Project <home:>" has unknown trust
error: failed to update home_metakcahura_kernel_Arch_Extra_standard (invalid or corrupt database (PGP signature))
error: failed to sync all databases
Error installing repo packages

metak commented on 2019-08-16 23:46 (UTC)

@btd1337 Hi. That first line should be [home_metakcahura_kernel_Arch_Extra_standard].

btd1337 commented on 2019-08-16 22:48 (UTC) (edited on 2019-08-16 22:51 (UTC) by btd1337)

@metak how to use your repository in Arch Linux?

I tried this:

# binaries for Clear Linux kernel
Server =

but this error occurs:

error: failed to get "Arch_Extra_standard.db" file from The requested URL returned error: 404
Error: Failed to update Arch_Extra_standard (unexpected error)
error: failed to sync all databases
Error installing repo packages

Thanks in advance

yw662 commented on 2019-08-04 22:52 (UTC)

I have uploaded some packages to download the kernel from clear linux repo, instead of compiling it from source. linux-clear-bin for the kernel, linux-clear-headers-bin for the headers, and intel-ucode-clear for intel cpu and i915 ucode. Hope it helps.

P.S. It is currently at 5.2.5-813, as in

metak commented on 2019-08-01 16:29 (UTC)

No, I only updated to the latest upstream kernel bugfix release 5.2.5.

zheiling commented on 2019-08-01 11:40 (UTC) (edited on 2019-08-01 11:40 (UTC) by zheiling)

After upgrading to linux-clear 5.2.5-1, proprietary nvidia driver becomes to work. Have you applied a new fix?

TillySutton commented on 2019-07-31 08:32 (UTC)

Thanks for this It helped me a lot

zheiling commented on 2019-07-30 13:53 (UTC)

@metak Updated kernel today. No changes are seen. I even can't find Nvidia messages in dmesg, Xorg.0.log also silent about Nvidia. Very strange behavior...

metak commented on 2019-07-29 22:17 (UTC) (edited on 2019-07-29 22:21 (UTC) by metak)

@zheiling On second thought, your issue could be related to this? Latest native kernel issues with custom drivers #1048 Can you try the latest version?

zheiling commented on 2019-07-29 15:04 (UTC)

@danhyal other kernels work normally with switching. In CLI version found this message: Config parsing : unknown option kill_logind in section optimus. Ignoring. Config parsing : unknown option kill_x11 in section optimus. Ignoring. Config parsing : unknown option login_manager_control in section optimus. Ignoring. Config parsing : unknown option pci_reset in section optimus. Ignoring. Config parsing : unknown option terminate_sessions in section optimus. Ignoring. Config parsing : unknown option dri in section nvidia. Ignoring. BTW: how to format text in aurweb? Can't find instructions.

amelia commented on 2019-07-28 21:40 (UTC)

@zheiling Do you still have the problem on the default arch kernel?

zheiling commented on 2019-07-28 21:04 (UTC)

@metak Fully removed nouveau and compiled kernel without nouveau support. The message is disappeared, but the problem with switching still exist. I suppose that an issue can be in patches/flags applied by linux-clear team.

metak commented on 2019-07-27 18:17 (UTC)

@zheiling Can you try unloading nouveau rmmod nouveau and load nvidia modprobe nvidia module?

zheiling commented on 2019-07-27 18:03 (UTC) (edited on 2019-07-27 18:15 (UTC) by zheiling)

Addition to my previous post about nVidia: [ 150.341496] nvidia: module license 'NVIDIA' taints kernel. [ 150.341498] Disabling lock debugging due to kernel taint [ 150.355115] calling nvidia_frontend_init_module+0x0/0x93 [nvidia] @ 1580 [ 150.355150] nvidia-nvlink: Nvlink Core is being initialized, major device number 236 [ 150.355540] NVRM: The NVIDIA probe routine was not called for 1 device(s). [ 150.355541] NVRM: This can occur when a driver such as: NVRM: nouveau, rivafb, nvidiafb or rivatv NVRM: was loaded and obtained ownership of the NVIDIA device(s). [ 150.355542] NVRM: Try unloading the conflicting kernel module (and/or NVRM: reconfigure your kernel without the conflicting NVRM: driver(s)), then try loading the NVIDIA kernel module NVRM: again. [ 150.355543] NVRM: No NVIDIA devices probed. Using NVIDIA Corporation GM108M [GeForce 840M] (rev a2) and optimus-manager-qt

metak commented on 2019-07-23 15:54 (UTC) (edited on 2019-07-23 15:56 (UTC) by metak)

@berturion Don't expect any massive performance boost just from this kernel. Most of the clear linux optimizations are in userland. However, your desktop should be more responsive as it has CONFIG_HZ_1000=y & CONFIG_HZ=1000.

@everyone Version 5.2.2 should be out in a day or two.

berturion commented on 2019-07-20 19:23 (UTC)

Did some people find evidences of a performance boost by using this kernel instead of using arch linux kernel ? Can you share the commands or tools you used that shows significant performance changes please (good or bad) ?

I tried:

sysbench --test=cpu --num-threads=4 --cpu-max-prime=9999 run


time $(i=0; while (( i < 9999999 )); do (( i ++ )); done)

Results between the kernels are similar.

metak commented on 2019-07-19 21:08 (UTC)

@zheiling You'll have to provide more info about your setup and logs if someone is to help you. I don't have nvidia card so there's also that.

zheiling commented on 2019-07-18 22:04 (UTC)

With this kernel optimus-manager switch from Intel to NVIDIA does not work. It just switches to Intel again and again, when I choose NVIDIA.

berturion commented on 2019-07-18 17:54 (UTC)

Thank you @metak, I should have seen this, sorry. It works now. I needed to run this command:

pacman-key --lsign-key 50AAD02AEB1F61C9

metak commented on 2019-07-18 13:32 (UTC)

@berturion Have you imported the key as it says in the wiki? Key ID: 50AAD02AEB1F61C9

berturion commented on 2019-07-18 10:31 (UTC)

Hello, I wanted to include your repo in my pacman configuration so I added this in /etc/pacman.conf :

Include = /etc/pacman.d/metak-kernel-mirrorlist

And this in /etc/pacman.d/metak-kernel-mirrorlist :

Server =

When I update pacman repos, I have this :

erreur : home_metakcahura_kernel_Arch_Extra_standard : la signature de « home:metakcahura OBS Project <> » est de confiance inconnue

Is that can work? How can I trust this gpg?

metak commented on 2019-07-18 00:16 (UTC)

@fidasx It takes some time to rebase all the patches and test everything with major versions. I don't know when it's coming, but my guess is in couple of days since the upstream is already at 5.2.1.

fidasx commented on 2019-07-17 23:06 (UTC)

hey @metak ty amazing pkg any idea when 5.2+ its coming

metak commented on 2019-07-11 15:37 (UTC)

@zheiling That's what clr team set as default, but I can disable it as it is in arch kernel. 5.1.17-2 (clr 5.1.16-795)

zheiling commented on 2019-07-11 10:37 (UTC)

@metak /usr/share/doc/systemd/README suggests to set CONFIG_RT_GROUP_SCHED=n. As default, it turned on.

metak commented on 2019-07-06 21:00 (UTC) (edited on 2019-07-06 21:00 (UTC) by metak)

@zheiling Since it's also enabled in the default arch kernel it's not a problem. Done in 5.1.16-4 (clr 5.1.16-794).

zheiling commented on 2019-07-06 16:27 (UTC) (edited on 2019-07-06 16:27 (UTC) by zheiling)


metak commented on 2019-07-04 22:26 (UTC)

@danhyal Patch is pulled from linux-ck and it's broken over there right now (can't set native this way), but it seems like it works just fine in this package. I'll enable it in next update and if there are some issues please report them.

amelia commented on 2019-07-04 21:04 (UTC)

Why don't you set the subarch to 27? since gcc will detect all the available instructions sets, it shouldn't cause any problems.

amelia commented on 2019-06-30 15:54 (UTC)

@metak I have tested it and the patch seems to be working as expected, thanks for including it.

metak commented on 2019-06-27 20:38 (UTC)

@danhyal Since this patch needs to be enabled with kernel command line option it shouldn't be a problem for those who have no use for it. Would you mind running some tests to see if it's working as it should?

amelia commented on 2019-06-27 16:39 (UTC)

@metak That would be useful for people that are doing hardware passthrough.

metak commented on 2019-06-26 19:29 (UTC)

@danhyal Not sure about adding more extra patches... if others would consider it useful I could add it as an option that can be enabled for testing or something?

amelia commented on 2019-06-26 18:22 (UTC)

Could you include the acs-overrides patch?

metak commented on 2019-06-24 18:58 (UTC)

Hey everyone. I've enabled hibernation and it seems that clr team plans to enable it soon.
Enable CONFIG_HIBERNATION to support 'suspend to disk' #458

berturion commented on 2019-06-24 16:53 (UTC) (edited on 2019-06-24 16:55 (UTC) by berturion)

As all other kernels available in Arch Linux repo and AUR have hibernation, maybe the maintainer could change the config to enable it in this kernel too? What do you think @metak?

As for the ACPI bloc in the PKGBUILD, we could add this one:

    ### Set CONFIG_HIBERNATION to make hibernation possible
        msg2 "Make hibernation possible..."
        sed -i "s|# CONFIG_HIBERNATION is not set|CONFIG_HIBERNATION=y|g" ./.config

Or maybe there is a good reason for disabling hibernation but I don't get it.

EDIT: bad copy/paste

berturion commented on 2019-06-24 16:35 (UTC)

Ok thank you @ltsdw. I will think about a custom script to do that.

ltsdw commented on 2019-06-24 15:19 (UTC)

@berturion sadly I do not know how make this process automatically, so my suggestion is edit PKGBUILD (maybe download and edit everytime that this package is updated) to use your .config previously edit with the changes. Or even using sed to edit. Something like sed -i "s|# CONFIG_HIBERNATION is not set|CONFIG_HIBERNATION=y|g" PATH\ /TO\ /.CONFIG/.config. I think you could possibly write a script to change the PKGBUILD before building it when running yay, but I still don't now how write it. Sorry!

berturion commented on 2019-06-24 14:28 (UTC)

Thanks @ltsdw. Do you know if I can durably set this when an update is available for this AUR package. Like positioning a hook with yay to automatically modify the kernel config before building it ?

ltsdw commented on 2019-06-24 12:51 (UTC)

@berturion try compiling the kernel with CONFIG_HIBERNATION set to Y.

berturion commented on 2019-06-24 11:45 (UTC)

Hi, there. Thank you very much for providing this package. It seems that hibernation is disabled by default on clear linux OS. And, using this AUR package, in fact, I can't get it to work. I ran mkinitcpio -p linux-clear command (resume is in th HOOKS). But it doesn't solve it. Is anyone managed to get hibernation work ? Is it even possible ?

metak commented on 2019-06-08 13:33 (UTC)

@simona Hi. That is not in any way related to this package, but your third-party repo you're using. (chaotic-aur?)

simona commented on 2019-06-08 11:39 (UTC)

help: errore: linux-clear-headers: signature from "Pedro Henrique Lara Campos" is invalid

metak commented on 2019-04-17 21:50 (UTC)

@xuanruiqi Should be fixed now.

xuanruiqi commented on 2019-04-17 20:41 (UTC) (edited on 2019-04-17 20:42 (UTC) by xuanruiqi)

There's a problem with this version. If _subarch is set, then the line:

yes "$_subarch" | make oldconfig

will echo "_subarch" to the newly added interactive item:

Automatically load TTY Line Disciplines (LDISC_AUTOLOAD) [Y/n/?] (NEW)

and will cause infinite looping. The solution is of course unsetting _subarch temporarily.

metak commented on 2019-04-16 15:17 (UTC)

@longspear Never experienced it myself.

longspear commented on 2019-04-15 07:15 (UTC)

Hi there, is anyone else experiencing a recurring "ping: no buffer space available" error with this particular Kernel?

metak commented on 2019-04-09 15:28 (UTC)

@aorth If you're experiencing this issue only on linux-clear kernel it may be due to something clear specific? Dunno. Please link that lvm2 issue in your clearlinux-distribution github comment so that clear team can check it out if they haven't already.

aorth commented on 2019-04-09 13:28 (UTC) (edited on 2019-04-09 13:29 (UTC) by aorth)

@metak it looks like this issue will be fixed in LVM2. I'm not sure why it only affects me on linux-clear, though...

metak commented on 2019-04-04 16:58 (UTC) (edited on 2019-04-04 17:02 (UTC) by metak)

@aorth Seems like this issue is already reported upstream. LVM Clearlinux takes a long time to shut down

aorth commented on 2019-04-04 15:55 (UTC) (edited on 2019-04-06 11:37 (UTC) by aorth)

@metak yeah I saw that thread. They seem to conclude that it's something related to LVM, and I am indeed using LVM on LUKS, though for me the issue only happens on linux-clear. The screen is black with a blinking cursor and eventually (maybe 90 seconds, I should actually time it) it says something about failing to unmount oldroot and then it reboots.

Edit: thanks @metak. I won't spam here anymore. Nice to know I'm not alone and that someone has even reported it upstream. For reference, here is the message that is spammed after the cursor blinks for ~90 seconds and it finally reboots:

[  116.123456] sd-umoun[1040]: Failed to unmount /oldroot: Device or resource busy
[  116.123456] sd-umoun[1041]: Failed to unmount /oldroot/dev: Device or resource busy
[  117.123457] reboot: Restarting system

metak commented on 2019-04-01 15:33 (UTC) (edited on 2019-04-01 15:33 (UTC) by metak)

@aorth Have you checked this thread?

aorth commented on 2019-04-01 15:08 (UTC) (edited on 2019-04-01 15:18 (UTC) by aorth)

Does anyone get a blinking cursor for about one minute on shutdown? It's the one annoying thing that causes me to stop running linux-clear. Not sure if it's a systemd job waiting or something...

metak commented on 2019-03-13 18:56 (UTC)

@ltsdw Thanx. Seems like it's already fixed by clear team in 5.0.1-716.

ltsdw commented on 2019-03-12 17:38 (UTC)

@metak done!

metak commented on 2019-03-12 11:28 (UTC)

@ltsdw Can you please report this issue upstream?

ltsdw commented on 2019-03-11 23:56 (UTC)

This was enabled on the config by the clear's devs, but this is causing a error message on boot:

kvm: already loaded the other module

According to gentoo's documentation this is due to both intel and amd kvm module beeing enabled on the kernel compiling CONFIG_KVM_AMD and CONFIG_KVM_INTEL:

Recompiling the kernel only with respective module for my processor resolved problem. In my case CONFIG_KVM_INTEL

metak commented on 2019-02-23 22:08 (UTC)

@phantomas 1. The upstream patches failed to apply because you tried to re-build the package without cleaning the src dir.
2. That part of the pkgbuild file is taken from linux-ck package. Every custom kernel in aur pretty much has it the same. I guess if you wanted you could change the line to yes "" | make localmodconfig for defaults.

phantomas commented on 2019-02-23 21:49 (UTC)

Weird, now the patch suceeds. The kconfig is not fully automated since it is asking for: Studio Evolution SE6X (SND_SE6X) [N/m/?] (NEW)

phantomas commented on 2019-02-23 21:43 (UTC) (edited on 2019-02-23 21:44 (UTC) by phantomas)

Has anybody else successfully built the 4.20.12 kernel?

For me it fails somewhere in "Add upstream patches":

tools/arch/riscv/include/uapi/asm/bitsperlong.h already exists!

[...] patching file sound/usb/card.c patching file sound/usb/mixer.c patching file sound/usb/pcm.c patching file sound/usb/quirks-table.h patching file sound/usb/quirks.c patching file sound/usb/stream.c The next patch would create the file tools/arch/riscv/include/uapi/asm/bitsperlong.h, which already exists! Skipping patch. 1 out of 1 hunk ignored patching file tools/bpf/bpftool/common.c patching file tools/bpf/bpftool/json_writer.c patching file tools/bpf/bpftool/map.c patching file tools/bpf/bpftool/prog.c patching file tools/bpf/bpftool/xlated_dumper.c patching file tools/build/Makefile.feature patching file tools/build/feature/Makefile patching file tools/hv/hv_kvp_daemon.c patching file tools/iio/iio_generic_buffer.c patching file tools/include/uapi/asm/bitsperlong.h patching file tools/lib/bpf/bpf.c patching file tools/lib/subcmd/Makefile patching file tools/lib/traceevent/event-parse.c patching file tools/perf/Makefile.config patching file tools/perf/arch/common.c patching file tools/perf/arch/common.h patching file tools/perf/arch/x86/util/intel-pt.c patching file tools/perf/arch/x86/util/kvm-stat.c patching file tools/perf/builtin-script.c patching file tools/perf/builtin-stat.c patching file tools/perf/builtin-timechart.c patching file tools/perf/pmu-events/arch/x86/skylake/skl-metrics.json patching file tools/perf/pmu-events/arch/x86/skylakex/skx-metrics.json patching file tools/perf/tests/ patching file tools/perf/tests/bp_signal.c patching file tools/perf/tests/evsel-tp-sched.c patching file tools/perf/tests/shell/lib/ patching file tools/perf/util/callchain.c patching file tools/perf/util/callchain.h patching file tools/perf/util/cs-etm.c patching file tools/perf/util/dso.c patching file tools/perf/util/env.c patching file tools/perf/util/event.c patching file tools/perf/util/evlist.c patching file tools/perf/util/header.c patching file tools/perf/util/machine.c patching file tools/perf/util/machine.h patching file tools/perf/util/parse-events.c patching file tools/perf/util/pmu.c patching file tools/perf/util/probe-file.c patching file tools/perf/util/python.c patching file tools/perf/util/s390-cpumsf.c patching file tools/perf/util/scripting-engines/trace-event-python.c patching file tools/perf/util/session.c patching file tools/perf/util/stat-display.c patching file tools/perf/util/svghelper.c patching file tools/perf/util/thread.h patching file tools/power/x86/intel_pstate_tracer/ patching file tools/testing/nvdimm/test/iomap.c patching file tools/testing/selftests/Makefile patching file tools/testing/selftests/android/Makefile patching file tools/testing/selftests/bpf/Makefile patching file tools/testing/selftests/bpf/ patching file tools/testing/selftests/bpf/test_progs.c patching file tools/testing/selftests/bpf/test_verifier.c patching file tools/testing/selftests/drivers/dma-buf/Makefile patching file tools/testing/selftests/drivers/dma-buf/udmabuf.c patching file tools/testing/selftests/futex/functional/Makefile patching file tools/testing/selftests/gpio/Makefile patching file tools/testing/selftests/kselftest_harness.h patching file tools/testing/selftests/kvm/Makefile patching file tools/testing/selftests/kvm/x86_64/evmcs_test.c patching file tools/testing/selftests/kvm/x86_64/state_test.c patching file tools/testing/selftests/ patching file tools/testing/selftests/networking/timestamping/Makefile patching file tools/testing/selftests/seccomp/seccomp_bpf.c patching file tools/testing/selftests/tc-testing/bpf/Makefile patching file tools/testing/selftests/vm/Makefile patching file tools/testing/selftests/x86/protection_keys.c patching file virt/kvm/arm/arm.c patching file virt/kvm/arm/mmio.c patching file virt/kvm/arm/vgic/vgic-mmio.c patching file virt/kvm/arm/vgic/vgic.c patching file virt/kvm/kvm_main.c

metak commented on 2019-02-23 21:37 (UTC)

Load all the modules that you want eg. modprobe loop, and then run modprobed-db store to create a list of loaded modules.
If you're building this package on top of default arch kernel you have to load loop module first since it's not built into the kernel like with this clear kernel.

phantomas commented on 2019-02-23 20:55 (UTC)

@metak Thanks and sorry for my ignorance. I would like to manually add the "loop" module since I would have to reboot to enforce it.

metak commented on 2019-02-23 20:30 (UTC)

@phantomas Just set it to _localmodcfg=y and that should be it.

phantomas commented on 2019-02-23 20:09 (UTC)

What should I enter at localmodcfg?

I would like to use modprobed-db, so

_localmodcfg=~/.config/modprobed-db.conf ??

xuanruiqi commented on 2019-02-22 20:12 (UTC)

I tried and seems that it doesn't break anything.

metak commented on 2019-02-20 13:56 (UTC)

@xuanruiqi You can try it if you want. If you followed the instructions in my pinned comment and it still for some reason doesn't work you can still just reboot to default kernel. It shouldn't break anything.

xuanruiqi commented on 2019-02-20 07:34 (UTC)

Thinking about installing this, but wanted to confirm that the extra boot parameters won't break other installed kernels.

bruceutut commented on 2019-02-16 05:09 (UTC)

@metak works like a charm on both local and building machine. thanks m8.

metak commented on 2019-02-15 12:40 (UTC) (edited on 2019-02-15 21:26 (UTC) by metak)

@bruceutut Edit: Hey brucetut. I rearranged some things in pkgbuild file and now it appears to build fine. ;-) Can you try it and report back. (I won't bump pkrel for this)

bruceutut commented on 2019-02-14 06:13 (UTC)


Here is the full build log on building machine built with extra-x86_64-build

And built package with issue:

And same package built with makepkg on my local machine:

And my friend another aur packager try to modify pkgbuild to solve this issue refering ths

Thanks again.

metak commented on 2019-02-13 22:37 (UTC)

@bruceutut I'll try replicating this issue when I get some free time, but in the meantime, you could upload full build logs and the built 'bad' package somewhere for me to take a look.

bruceutut commented on 2019-02-13 12:42 (UTC) (edited on 2019-02-13 12:43 (UTC) by bruceutut)


Actually, I can build correct package using makepkg command. Or yay/pikaur something.

But the bad package using extra-x86_64-build command, on local machine or my building machine.

The bad package does not have crypto_user and xz compression module is not enabled as well.

Thanks for your patience :)

metak commented on 2019-02-03 15:09 (UTC)

@bruceutut crypto_user should be built since this pkgbuild enables it. It's in /usr/lib/modules/4.20.6-1-clear/kernel/crypto/crypto_user.ko.xz, both on my local build and in clean chroot for repo.
Have you checked your build log? You haven't provided any info.

bruceutut commented on 2019-02-03 08:05 (UTC)

crypto_user module not found in package after successful build.
Anything am i doing wrong ?

enihcam commented on 2018-11-24 01:29 (UTC)

@metak Yes, after turbo3 patch is removed the package can be built again. :) I have also left this comment in the github.

metak commented on 2018-11-23 15:23 (UTC)

@enihcam That's probably due to the patch turbo3-scheduler.patch in latest commit. No problems on my machine or in repo build. You can try building with that patch disabled to check?

enihcam commented on 2018-11-23 00:27 (UTC)

ld: kernel/sched/fair.o: in function select_task_rq_fair': fair.c:(.text+0x61a): undefined reference tobest_core' ld: fair.c:(.text+0x628): undefined reference to `second_best_core' make[1]: [Makefile:1032: vmlinux] Error 1 make: [Makefile:1001: autoksyms_recursive] Error 2 ==> ERROR: A failure occurred in build(). Aborting... ==> ERROR: Makepkg was unable to build linux-clear. ==> Restart building linux-clear ? [y/N] ==> ------------------------------------

4.19.3-1 has no such problem.

ltsdw commented on 2018-11-21 18:25 (UTC)

this is not related with the package, but I like to update my problem with module that was resolved after recompile the kernel habilitating preemption. Again, thanks.

ltsdw commented on 2018-11-21 02:28 (UTC) (edited on 2018-11-21 02:28 (UTC) by ltsdw)

@metak Yeah, I have module.sig_unenforce as boot parameter... anyway much obliged.

metak commented on 2018-11-21 01:41 (UTC) (edited on 2018-11-21 01:48 (UTC) by metak)

@ltsdw What you described sounds more like an issue with yay than this package. The package is properly built and you can manually install the binary.

Edit: as for your edit, have you checked the third point in the sticky comment for dkms modules?

ltsdw commented on 2018-11-20 21:16 (UTC) (edited on 2018-11-21 01:27 (UTC) by ltsdw)

@metak so, everything seems be 'ok', nothing appears when I was compiling, none error, the only thing that differs the linux-headers builded from aur (my cached headers from build 19917744) is the size from that one on your repo (19404108). The log build says that installing headers, but when I tried install something for example, whit dkms, says "Your kernel headers for kernel 4.19.2-2-clear cannot be found" so I have to install manually or from my cache or from your binary. (I don't know if it matters but I'm using yay)

Edit: I think this is not related to the package itself, but look what @iwfinger said on this issue that I opened on github:

"Your kernel headers and the compiler version MUST match exactly. "SMP mod_unload" is not the same as "SMP preempt mod_unload". I have never seen this problem as most of my kernels are self-generated, which sidesteps this problem. I'm not sure how to fix it."

You can check on this link:

Do you have idea what can I do?

metak commented on 2018-11-20 19:28 (UTC)

@ltsdw Sorry for late reply. I couldn't reproduce your issue. The package builds fine on my machine and in clean chroot. You'll have to provide more details. Build logs etc.
Have you compared the headers package from my repo with your built binary for differences to make sure your built package is okay?

ltsdw commented on 2018-11-20 19:06 (UTC)

And every module that I tried install I get something like this for example:

[ 2725.823458] 8188eu: version magic '4.19.2-2-clear SMP mod_unload ' should be '4.19.2-2-clear SMP preempt mod_unload '

ltsdw commented on 2018-11-20 18:20 (UTC) (edited on 2018-11-20 18:22 (UTC) by ltsdw)

Getting this every time I compile:

Error! echo
Your kernel headers for kernel 4.19.2-2-clear cannot be found at
/usr/lib/modules/4.19.2-2-clear/build or /usr/lib/modules/4.19.2-2-clear/source.

Always have to install the linux-clear-headers binary from your repo.

ltsdw commented on 2018-11-14 20:58 (UTC)

I tried recompile again and still doesn't installing headers for me, so I installed the binary one from your repo.

metak commented on 2018-11-03 13:49 (UTC)

@enihcam I've added the provides wireguard module and I've also enabled the XZ compressed modules to further reduce the size. From roughly ~77 MiB installed size down to ~26 Mib.

enihcam commented on 2018-11-03 12:18 (UTC) (edited on 2018-11-03 12:30 (UTC) by enihcam)

@metak, you need to add the following line into PKGBUILD: provides=('WIREGUARD-MODULE')

Otherwise, when you install wireguard-tools, wireguard-dkms would be installed as its dependency.

I have written a note at

metak commented on 2018-10-17 10:14 (UTC)

@enihcam No need. Bison is in base-devel group and should already be installed. Official linux package also doesn't have it makedepends.

enihcam commented on 2018-10-17 06:22 (UTC)

not a big deal. bison is needed to compile the kernel, but it is not in makedepends.

metak commented on 2018-09-12 21:16 (UTC)

NOTICE! I've set CONFIG_MNATIVE as default like @Enverex asked previously. If there are any issues please let me know. (of course, this doesn't affect the binaries in my repo)

metak commented on 2018-09-10 11:34 (UTC) (edited on 2018-09-10 11:35 (UTC) by metak)

@jwm1611 Slight improvement is what you can expect, but no miracles. As far as clearlinux distribution optimizations go, they're mostly in userland and not the kernel. I don't use Manjaro so I can't comment on that one. If you're having issues with keys try building with --skippgpcheck.

jwm1611 commented on 2018-09-07 19:46 (UTC) (edited on 2018-09-07 20:21 (UTC) by jwm1611)

Not sure if it should be expected that this package will not work on Manjaro, but I decided to test this out. Just attempted install of linux-clear 4.18.6-1 on Manjaro (with kernel 4.18.5-1-MANJARO) with all updates applied as of today, and installation ended prematurely here:

configuration written to .config

==> Sources are ready. :: failed to verify integrity or prepare linux-clear package

After checking the scroll-back, I saw that I needed to import Linus Torvalds' key. I was able to import his key with sudo pacman-key -r 79BE3E4300411886, but was still unable to get the kernel source tarball to verify during the linux-clear package install with pacaur, even after following that up with sudo pacman-key --lsign-key 79BE3E4300411886.

I ended up having to follow these instructions to get the correct signature to import, so I could then get the tarball verification to succeed:

"This isn't a solution to the error message you're getting, but it's a way to import the PGP keys. I'll run through importing Linus' key for Kernel Packaging. Greg's will be the same process.

1) Go to a keyserver, say; 2) Search for the key you're looking for with matching fingerprints, in this case, Linus Torvalds or 00411886 (the last 8 digits of the long fingerprint 79BE3E4300411886 in the error message); 3) Click on the keyID (so 00411886 for Linus) 4) Copy all the text on the page into a text editor and save it as a linus.asc file; 5) Navigate to the folder you saved linus.asc and run:

gpg --import linus.asc

6) Done (provided there are no error messages)."

However, my cached package data for linux-clear still needed to be cleared out before successfully reinstalling the package, because for some reason things still came to a stop with the same error, even though no integrity checks had failed:

==> ERROR: A failure occurred in prepare(). Aborting... :: failed to verify integrity or prepare linux-clear package

The linux-clear kernel successfully built for me, and here are my before and after Geekbench scores:

4.18.5-1-MANJARO: single: 5079 multi: 17426

linux-clear 4.18.6-1: single: 5108 multi: 17502

Looks like modest gains which are likely beyond margin of error, but I haven't done any scientific testing.

As far as my perception goes, my system already felt insanely responsive running on this Samsung 970 Pro 512gb SSD with delidded 3770K @ 4.8ghz on all cores (highest I could get before voltage scaling curve hit the cliff). Will test further, thanks for your work on this package!

metak commented on 2018-09-07 15:07 (UTC) (edited on 2018-09-07 15:08 (UTC) by metak)

@Enverex The only thing clearlinux team changed was to enable pre-GCN 1.2 AMDGPU driver support in 4.18.5-2 (clr 4.18.5-628). Have you checked the wiki or troubleshoot in any way to check if that commit caused this?

Enverex commented on 2018-09-07 12:12 (UTC)

Just a heads up, HDMI audio (on an RX580) stopped working for me a version or two ago with this. Had to switch back to the normal kernel.

metak commented on 2018-09-04 11:02 (UTC)

@aorth I've added the info in the pinned comment about unsigned modules.

aorth commented on 2018-09-04 07:19 (UTC) (edited on 2018-09-06 05:55 (UTC) by aorth)

Does anyone know how to get DKMS modules like VirtualBox to work with this kernel? When I try to modprobe the modules I get the error "required key not available." For what it's worth, I do not have secure boot enabled.

Edit: @metak, thanks for the tip about the module.sig_unenforce boot parameter!

enihcam commented on 2018-08-23 23:37 (UTC)

Yes, of course. Thanks a lot!

metak commented on 2018-08-23 15:46 (UTC) (edited on 2018-08-23 15:48 (UTC) by metak)

@enihcam If others don't disagree then I'll drop linux-clear-docs package in aur (next update), but I'll leave it in my repo build so if anyone needs it they can still get it there. Is this okay?

enihcam commented on 2018-08-23 10:04 (UTC) (edited on 2018-08-23 13:43 (UTC) by enihcam)

@metak, can you set python-sphinx and graphviz as optional dependencies? They are only needed by linux-clear-docs.

Or can you split linux-clear-docs out? like linux-ck. Very few people read local docs.

linux-clear is running in my router and NAS. python-sphinx and graphviz would bring too many unnecessary dependencies.


metak commented on 2018-08-06 18:31 (UTC)

@airend Thanx. That's the best way really. If the clear team are willing to enable something then it's easier for me since I don't have to change/patch anything. Your requested changes are now available with latest update.

airend commented on 2018-08-04 03:13 (UTC)

@metak, done in

metak commented on 2018-08-03 14:38 (UTC) (edited on 2018-08-03 14:41 (UTC) by metak)

Update 2018-08-03 Update your boot parameter line. modprobe.blacklist and i915.fastboot were removed. (The pinned comment is updated version.)

@airend Could your please ask first over at if they would be willing to enable that by default in clr kernel?

airend commented on 2018-07-31 10:23 (UTC)

Is it possible to enable Zswap support (CONFIG_ZSWAP, CONFIG_ZPOOL, CONFIG_ZBUD, CONFIG_Z3FOLD, CONFIG_FRONTSWAP) and CONFIG_CLEANCACHE, like Arch kernels?

metak commented on 2018-07-30 11:51 (UTC)

@Enverex I think it's best to leave it like it is now (taken from linux-ck package and graysky2 patch). By default the _subarch is empty and it will detect/enable optimizations for your cpu family.
The binary in my repo is built with generic_cpu optimizations. You could create a script or something?


git checkout PKGBUILD
git pull
sed -i 's|^_subarch.*|_subarch=27|' PKGBUILD
makepkg -Ccrs

Enverex commented on 2018-07-30 11:04 (UTC)

Any chance of making CONFIG_MNATIVE the default (rather than CONFIG_MCORE2) or having some way of enabling it without having to manually edit the PKGBUILD? (e.g. so that the package can be updated and keep this setting without manual intervention).

metak commented on 2018-07-19 15:43 (UTC)

@enihcam I don't mind at all, but I hope it won't be flagged out-of-date again if it's not following releases. ;-)

enihcam commented on 2018-07-19 14:25 (UTC)

@metak again, this package is great. people using this package should vote for it. :)

this is only a suggestion. Personally i think it might be too frequent for compiling the kernel because the rapidly increasing version. i think it might be good to slow down a little bit.

Rather than watching, maybe you want to release a version only when this RSS is updated:

what do you think?

metak commented on 2018-07-08 17:52 (UTC) (edited on 2018-07-08 20:35 (UTC) by metak)

@big_bun Sorry for late reply, I've been busy. I really don't want to deviate too much from default upstream. (only when needed to comply with archlinux packaging for example)

big_bum commented on 2018-07-05 20:37 (UTC)

@metak could you also add l2tp/pptp module in the next build? I think this should enable them:

metak commented on 2018-06-30 17:58 (UTC) (edited on 2018-06-30 17:58 (UTC) by metak)

@A.J If you have SecureBoot disabled then you simply use this boot parameter module.sig_unenforce to allow loading unsigned kernel modules. Otherwise you'll have to build your kernel with CONFIG_MODULE_SIG_FORCE disabled.

A.J commented on 2018-06-30 17:50 (UTC)

@metak could you elaborate , is it boot option or build option ??

metak commented on 2018-06-30 12:04 (UTC)

@A.J Look below in comments for my comment to @Osleg about option to allow unsigned modules.

A.J commented on 2018-06-30 04:17 (UTC)

sorry for the noop question , but i couldn't get nvidia-dkms working and when i modprobe it i get :

modprobe: ERROR: could not insert 'nvidia': Required key not available

metak commented on 2018-06-25 23:11 (UTC) (edited on 2018-06-25 23:12 (UTC) by metak)

@ltsdw Thank you for reporting back. I don't use bluetooth at all so I never paid too much attention to it, but it seems that quite a few people do use it so I'll enable the crypto_user module in next update.

ltsdw commented on 2018-06-25 22:56 (UTC)

@metak yeah, exactly, bluetooth working like a charm after enabling CONFIG_CRYPTO_USER. I also enabled CONFIG_BT_ATH3K for precaution, but I'm actually using btusb. It's only needed set CONFIG_CRYPTO_USER to get bluetooth working.

metak commented on 2018-06-25 22:10 (UTC) (edited on 2018-06-25 22:30 (UTC) by metak)

@ltsdw Sorry, I'm a bit busy right now. After checking again I'll have to enable user_crypto with next update to comply with arch packaging.

bluez package in arch /usr/lib/modules-load.d/bluez.conf needs this module crypto_user at boot time.

Edit: can you build this kernel with CONFIG_CRYPTO_USER option enabled and check if bluetooth is working properly?

ltsdw commented on 2018-06-25 21:23 (UTC) (edited on 2018-06-25 21:24 (UTC) by ltsdw)


ltsdw commented on 2018-06-25 20:02 (UTC)

@metak really is in the comments, thanks. One more thing, it's directly linked with the fact if bluetooth works or not?

metak commented on 2018-06-25 19:34 (UTC) (edited on 2018-06-25 19:35 (UTC) by metak)

@ltsdw If you need user_crypto module then you'll have to enable it CONFIG_CRYPTO_USER since this .config doesn't. I mentioned this somewhere in comments bellow.

This module is built in stock arch kernel though.

ltsdw commented on 2018-06-25 18:05 (UTC)

Having problem with bluez, if I uncomment user_crypto line in /usr/lib/modules-load.d/bluez.conf will fail to load systemd-modules-load at boot, basicly journalctl acuses not finding user_crypto module. Works well in arch's kernel, some idea?

metak commented on 2018-06-18 04:36 (UTC) (edited on 2018-06-18 04:41 (UTC) by metak)

@enihcam I will probably enable it by default with next update.

@E5ten No, but the kernel has to be packaged differently if you want to use clr-boot-manager.

E5ten commented on 2018-06-18 03:54 (UTC)

Is it required to use clr-boot-manager with the clear kernel?

enihcam commented on 2018-06-18 02:01 (UTC)

linux-clear is going to be the most popular kernel package in AUR. Keep it up!

Also, I think it is safe to unlock additional gcc CPU optimizatons by default. It does not change any kernel logics. It just provides some more compiler options.

simona commented on 2018-06-13 21:45 (UTC)

Now ok... Thanks :-)

metak commented on 2018-06-13 16:02 (UTC)

@Athemis Sorry about that, fixed now. I basically have three slightly different pkgbuild files to update (local, aur, and obs for binaries)... ;-)

Athemis commented on 2018-06-13 15:54 (UTC)

Thanks for keeping this up to date! However, 4.17.1 fails to compile for me, unless I change $_clearver to 27b762cf009a56fa47e00660d602f9db4b672ffb.

metak commented on 2018-06-08 17:45 (UTC) (edited on 2018-06-08 17:46 (UTC) by metak)

@gaougalos That's expected if you're building this kernel on top of the default arch kernel since all those config options are not present in default clear .config but are in stock arch.

gaougalos commented on 2018-06-08 17:24 (UTC)

When i enable localmodconfig it returns

==> Running Steven Rostedt's make localmodconfig now
using config: '.config'
acpi_call config not found!!
WARNING: LZ4_COMPRESS is required, but nothing in the
  current config selects it.
WARNING: LZ4_COMPRESS is required, but nothing in the
  current config selects it.
module mei did not have configs CONFIG_INTEL_MEI
module crct10dif_pclmul did not have configs CONFIG_CRYPTO_CRCT10DIF_PCLMUL
module input_leds did not have configs CONFIG_INPUT_LEDS
module mei_me did not have configs CONFIG_INTEL_MEI_ME
module joydev did not have configs CONFIG_INPUT_JOYDEV
module agpgart did not have configs CONFIG_AGP
module lz4 did not have configs CONFIG_CRYPTO_LZ4
module uas did not have configs CONFIG_USB_UAS
module algif_aead did not have configs CONFIG_CRYPTO_USER_API_AEAD
module iTCO_vendor_support did not have configs CONFIG_ITCO_WDT
module ghash_clmulni_intel did not have configs CONFIG_CRYPTO_GHASH_CLMUL_NI_INTEL
module lz4_compress did not have configs CONFIG_LZ4_COMPRESS
module intel_agp did not have configs CONFIG_AGP_INTEL
module iTCO_wdt did not have configs CONFIG_ITCO_WDT
module f2fs did not have configs CONFIG_F2FS_FS
module mac_hid did not have configs CONFIG_MAC_EMUMOUSEBTN
module crc32_pclmul did not have configs CONFIG_CRYPTO_CRC32_PCLMUL
module pcspkr did not have configs CONFIG_INPUT_PCSPKR
module crypto_user did not have configs CONFIG_CRYPTO_USER
module ax25 did not have configs CONFIG_AX25
module ath3k did not have configs CONFIG_BT_ATH3K
module mousedev did not have configs CONFIG_INPUT_MOUSEDEV

metak commented on 2018-06-05 13:38 (UTC)

As of 4.16.14-1 I've added kernel_gcc_optimizations patch. Disabled by default. Try and report any issues.

simona commented on 2018-05-08 16:56 (UTC)

make[1]: *** Nessuna regola per generare l'obiettivo "firmware/intel-ucode/06-4f-01", necessario per "firmware/intel-ucode/06-4f-01.gen.o". Arresto.

metak commented on 2018-05-07 19:46 (UTC) (edited on 2018-06-05 00:32 (UTC) by metak)

@Osleg Moved acpi_rev_override as a choice. As for those additional patches, I don't know, this kernel is meant for newer intel skylake processors...

Osleg commented on 2018-05-02 02:16 (UTC)

@metak I'm not sure it should be enabled by default as this is rarely used. Tho on the other hand vanilla kernel has it enabled so I don't know which approach would be better. imho tho providing prompts to enable/disable features like it happens in linux-ck, including the -mnative builds requested below would be the best option to allow every user to chose the options he needs the most.

metak commented on 2018-05-01 07:26 (UTC)

@enihcam I'll take a look and maybe I'll add it in as an option if it's not a problem.

enihcam commented on 2018-04-27 23:56 (UTC) (edited on 2018-04-28 00:05 (UTC) by enihcam)

This is a great kernel.

Suggestions: 1. Users should have a change to nconfig the kernel before build. 2. To allow -mnative build, please apply the patch enable_additional_cpu_optimizations_for_gcc.tar.gz

Please refer to

metak commented on 2018-04-24 16:17 (UTC)

@stephencox Sorry, somehow I missed the part where you said you can't boot with clear kernel. (facepalm)

@Osleg I've enabled CONFIG_ACPI_REV_OVERRIDE_POSSIBLE by default if it'll make life a bit easier?

Osleg commented on 2018-04-24 10:36 (UTC)

@stephencox I'm not sure that this is indeed the case but could you try to add nvidia to modprobe.blacklist and see if you can boot your microserver?

Osleg commented on 2018-04-22 21:42 (UTC)

I found what was causing problem in my setup. My laptop have nvidia discreet card and I'm using bbswitch to turn it off. For bbswitch to work I have to set acpi_rev_override=1 to kernel parameters but linux-clear compiled without CONFIG_ACPI_REV_OVERRIDE_POSSIBLE. I have updated PKGBUILD to have the key to optionally enable it

# Maintainer: Josip Ponjavic <josipponjavic at gmail dot com>
# Contributor:

makedepends=('bc' 'git' 'inetutils' 'kmod' 'libelf' 'linux-firmware' 'xmlto')

# If you have laptop with optimus and it hangs on boot one solution might be 
# to set acpi_rev_override. Yet for this to happen kernel should be compiled
# with `CONFIG_ACPI_REV_OVERRIDE_POSSIBLE`. Set next variable to `y` to enable.

  '60-linux.hook'  # pacman hook for depmod
  '90-linux.hook'  # pacman hook for initramfs regeneration
  '99-linux.hook'  # pacman hook for remove initramfs
  'linux.preset'   # standard config files for mkinitcpio ramdisk
  'ABAF11C65A2970B130ABE3C479BE3E4300411886'  # Linus Torvalds
  '647F28654894E3BD457199BE38DBBDC86092693E'  # Greg Kroah-Hartman


prepare() {
  cd linux-${__basekernel}

  # add upstream patch
  patch -p1 -i ../patch-${pkgver}

  cp -Tf $srcdir/clearlinux/config .config
  cp -a /usr/lib/firmware/i915 firmware/
  cp -a ${srcdir}/intel-ucode firmware/
  rm -f firmware/intel-ucode/0f*

  # Apply clearlinux patches
  for i in $(grep '^Patch' ${srcdir}/clearlinux/linux.spec | grep -Ev '^Patch0115|^Patch0410|^Patch0500|^Patch200' | sed -n 's/.*: //p'); do
    msg "Applying ${i}"
    patch -p1 -i "$srcdir/clearlinux/${i}"

  if [ "${_kernelname}" != "" ]; then
    sed -i "s|CONFIG_LOCALVERSION=.*|CONFIG_LOCALVERSION=\"${_kernelname}\"|g" ./.config

  # set extraversion to pkgrel and empty localversion
  sed -e "/^EXTRAVERSION =/s/=.*/= -${pkgrel}/" \
      -i Makefile

  # set ACPI_REV_OVERRIDE_POSSIBLE to prevent optimus lockup
  if [ "${_rev_override}" = "y" ]; then
  # don't run depmod on 'make install'. We'll do this ourselves in packaging
  sed -i '2iexit 0' scripts/

  make olddefconfig

build() {
  cd linux-${__basekernel}

  make bzImage modules

_package() {
  pkgdesc="The ${pkgbase/linux/Linux} kernel and modules"
  depends=('coreutils' 'linux-firmware' 'kmod' 'mkinitcpio>=0.7')
  optdepends=('crda: to set the correct wireless channels of your country')

  cd linux-${__basekernel}

  # get kernel version
  _kernver="$(make kernelrelease)"

  mkdir -p "${pkgdir}"/{boot,usr/lib/modules}
  make INSTALL_MOD_PATH="${pkgdir}/usr" modules_install
  cp arch/x86/boot/bzImage "${pkgdir}/boot/vmlinuz-${pkgbase}"

  # make room for external modules
  local _extramodules="extramodules-${_basekernel}${_kernelname:--ARCH}"
  ln -s "../${_extramodules}" "${pkgdir}/usr/lib/modules/${_kernver}/extramodules"

  # add real version for building modules and running depmod from hook
  echo "${_kernver}" |
    install -Dm644 /dev/stdin "${pkgdir}/usr/lib/modules/${_extramodules}/version"

  # remove build and source links
  rm "${pkgdir}"/usr/lib/modules/${_kernver}/{source,build}

  # now we call depmod...
  depmod -b "${pkgdir}/usr" -F "${_kernver}"

  # add vmlinux
  install -Dt "${pkgdir}/usr/lib/modules/${_kernver}/build" -m644 vmlinux

  # sed expression for following substitutions
  local _subst="

  # hack to allow specifying an initially nonexisting install file
  sed "${_subst}" "${startdir}/${install}" > "${startdir}/${install}.pkg"
  true && install=${install}.pkg

  # install mkinitcpio preset file
  sed "${_subst}" ../linux.preset |
    install -Dm644 /dev/stdin "${pkgdir}/etc/mkinitcpio.d/${pkgbase}.preset"

  # install pacman hooks
  sed "${_subst}" ../60-linux.hook |
    install -Dm644 /dev/stdin "${pkgdir}/usr/share/libalpm/hooks/60-${pkgbase}.hook"
  sed "${_subst}" ../90-linux.hook |
    install -Dm644 /dev/stdin "${pkgdir}/usr/share/libalpm/hooks/90-${pkgbase}.hook"
  sed "${_subst}" ../99-linux.hook |
    install -Dm644 /dev/stdin "${pkgdir}/usr/share/libalpm/hooks/99-${pkgbase}.hook"

_package-headers() {
  pkgdesc="Header files and scripts for building modules for ${pkgbase/linux/Linux} kernel"

  cd linux-${__basekernel}
  local _builddir="${pkgdir}/usr/lib/modules/${_kernver}/build"

  install -Dt "${_builddir}" -m644 Makefile .config Module.symvers
  install -Dt "${_builddir}/kernel" -m644 kernel/Makefile

  mkdir "${_builddir}/.tmp_versions"

  cp -t "${_builddir}" -a include scripts

  install -Dt "${_builddir}/arch/x86" -m644 arch/x86/Makefile
  install -Dt "${_builddir}/arch/x86/kernel" -m644 arch/x86/kernel/asm-offsets.s

  cp -t "${_builddir}/arch/x86" -a arch/x86/include

  install -Dt "${_builddir}/drivers/md" -m644 drivers/md/*.h
  install -Dt "${_builddir}/net/mac80211" -m644 net/mac80211/*.h

  install -Dt "${_builddir}/drivers/media/i2c" -m644 drivers/media/i2c/msp3400-driver.h

  install -Dt "${_builddir}/drivers/media/usb/dvb-usb" -m644 drivers/media/usb/dvb-usb/*.h
  install -Dt "${_builddir}/drivers/media/dvb-frontends" -m644 drivers/media/dvb-frontends/*.h
  install -Dt "${_builddir}/drivers/media/tuners" -m644 drivers/media/tuners/*.h

  # add xfs and shmem for aufs building
  mkdir -p "${_builddir}"/{fs/xfs,mm}

  # copy in Kconfig files
  find . -name Kconfig\* -exec install -Dm644 {} "${_builddir}/{}" \;

  # add objtool for external module building and enabled VALIDATION_STACK option
  install -Dt "${_builddir}/tools/objtool" tools/objtool/objtool

  # remove unneeded architectures
  local _arch
  for _arch in "${_builddir}"/arch/*/; do
    [[ ${_arch} == */x86/ ]] && continue
    rm -r "${_arch}"

  # remove files already in linux-docs package
  rm -r "${_builddir}/Documentation"

  # remove now broken symlinks
  find -L "${_builddir}" -type l -printf 'Removing %P\n' -delete

  # Fix permissions
  chmod -R u=rwX,go=rX "${_builddir}"

  # strip scripts directory
  local _binary _strip
  while read -rd '' _binary; do
    case "$(file -bi "${_binary}")" in
      *application/x-sharedlib*)  _strip="${STRIP_SHARED}"   ;; # Libraries (.so)
      *application/x-archive*)    _strip="${STRIP_STATIC}"   ;; # Libraries (.a)
      *application/x-executable*) _strip="${STRIP_BINARIES}" ;; # Binaries
      *) continue ;;
    /usr/bin/strip ${_strip} "${_binary}"
  done < <(find "${_builddir}/scripts" -type f -perm -u+w -print0 2>/dev/null)

_package-docs() {
  pkgdesc="Kernel hackers manual - HTML documentation that comes with the ${pkgbase/linux/Linux} kernel"

  cd linux-${__basekernel}
  local _builddir="${pkgdir}/usr/lib/modules/${_kernver}/build"

  mkdir -p "${_builddir}"
  cp -t "${_builddir}" -a Documentation

  # Fix permissions
  chmod -R u=rwX,go=rX "${_builddir}"

pkgname=("${pkgbase}" "${pkgbase}-headers" "${pkgbase}-docs")
for _p in ${pkgname[@]}; do
  eval "package_${_p}() {
    $(declare -f "_package${_p#${pkgbase}}")

# vim:set ts=8 sts=2 sw=2 et:

stephencox commented on 2018-04-22 18:39 (UTC)

@metak, I can't boot linux-clear at all on my hp microserver. The lsmod is from linux-mainline which boots. What is the likelyhood of a combination of the default kernel and linux-clear?

metak commented on 2018-04-22 15:32 (UTC) (edited on 2018-04-22 15:32 (UTC) by metak)

@stephencox I'm just packaging this as it is upstream and they are not making this to be a generic kernel, hence many things are not enabled. Like your wifi module.

Have you tried booting in default arch kernel and running lsmod and then compare that to clr kernel? Maybe you need something else that's not enabled in this kernel by default?

stephencox commented on 2018-04-22 14:24 (UTC)

On my HP laptop the WiFi (RTL8723BE) is not working as it's not included in the modules.

My HP microserver doesn't boot at all, stuck at Loading initial ramdisk. lsmod:

metak commented on 2018-04-20 21:44 (UTC)

Osleg, sorry couldn't be more helpful. If you want you can try this linux-clear-lts-4.14.35 pkgbuild?

Osleg commented on 2018-04-20 21:17 (UTC)

Well something really weird, now I can't boot even to rescue with linux-clear. Logs are clean but laptop seems to get stuck at last stage.

metak commented on 2018-04-19 23:31 (UTC)

@Osleg Yes, the easier choice is to not use initrd /intel-ucode.img (clr default), otherwise the build would fail and require some adjustments/patching.

Osleg commented on 2018-04-19 19:40 (UTC)

Hm, I just realized that linux-clear also has ucode built in so we either should not initrd /intel-ucode.img or should not compile it with clear... I think first option is better to follow intel's kernel as close as possible.

Osleg commented on 2018-04-19 19:32 (UTC)

@metak unfortunately I have no idea. The boot process goes well until the end and it stuck when starting to load I could boot into single user mode but that doesn't help in any way :/

And yay for the module.sig_unenforce, will try that... in recovery mode tho :(

metak commented on 2018-04-12 18:21 (UTC) (edited on 2018-04-12 18:22 (UTC) by metak)

@Osleg Do you have any idea what might be the problem? Btw, clr-team introduced new patch in their latest commit called Add boot option to allow unsigned modules so that's why I removed that sed line from pkgbuild.

+Add module.sig_unenforce boot parameter to allow loading unsigned kernel
 +modules. Parameter is only effective if CONFIG_MODULE_SIG_FORCE is
 +enabled and system is *not* SecureBooted.

Osleg commented on 2018-04-12 07:21 (UTC)

Just to be exact, it stopped working even before update back when it was 4.15.15

Osleg commented on 2018-04-12 07:20 (UTC)

Have no idea what happened but the kernel stopped working for me. It reaches target graphical and then laptop hangs. :/

metak commented on 2018-04-07 14:21 (UTC)

@Osleg I disabled CONFIG_MODULE_SIG_FORCE by default and if anyone wants to turn it on just comment that line in pkgbuild.

Osleg commented on 2018-04-07 05:43 (UTC)

The main issue with modules not loading is that clear has CONFIG_MODULE_SIG and CONFIG_MODULE_SIG_FORCE enabled. Meaning most of the "custom" modules won't load. The workaround that I'm using for that is to sign those modules manually as described here

Tho I think it might be a good idea to disable CONFIG_MODULE_SIG_FORCE or at least provide a config for PKGBUILD to disable it for those who want. as this setting is good for hardened kernel for servers and such. For desktop machines, especially for developers, this setting provides nothing but pain :)

volfyd commented on 2018-02-08 20:07 (UTC)

But I don't see the microcode for linux on Intel's website yet.

metak commented on 2018-02-06 11:26 (UTC)

@PythonNut Yes, "CONFIG_CRYPTO_USER" is not set by default.

PythonNut commented on 2018-02-02 07:08 (UTC)

On many arch systems, this will cause systemd-modules-load to fail because the crypto_user module is not available.

metak commented on 2018-01-27 07:34 (UTC) (edited on 2018-01-27 07:35 (UTC) by metak)

@volfyd The srpm thing was a temporary solution. I filled the issue on github and they've updated the git repo which was out of date for two weeks or so.

Now, I've switched to the official kernel tarball/patch and fetching clearlinux patches via git.


This should save a bit of bandwidth and cause less problems. ;-)

volfyd commented on 2018-01-26 17:51 (UTC)

==> Retrieving sources... -> Downloading linux-4.14.15-515.src.rpm... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0 curl: (22) The requested URL returned error: 404 Not Found

Looks like 516 is already out.

metak commented on 2018-01-26 04:46 (UTC)

@ahjolinna nvidia? Sorry, but no. Never had/used any nvidia cards so I wouldn't be able to test anything.

"dm_cache & dm_cache_smq"? I answered this one in my first comment below. It's not enabled by default. It's marked as experimental in kernel.

Here's a patch you can try that enables it. Add it to sources in pkgbuild file and rebuild the package.

ahjolinna commented on 2018-01-26 01:52 (UTC)

why isn't the "dm_cache & dm_cache_smq" enabled in the kernel conf?, as I get module not found error

ahjolinna commented on 2018-01-25 16:34 (UTC)

@metak: btw. is it possible you could add the nvidia driver to repo so there wouldn't be any need to compile least an aur version

ahjolinna commented on 2018-01-25 16:25 (UTC) (edited on 2018-01-25 16:34 (UTC) by ahjolinna)

it seems I have to use a workaround for this problem; by adding "SigLevel = Never" ...not the best solution but it works so I'm happy

about the "intel-ucode" thing, I didnt know/notice as I have skylake cpu so thankfully it hasn't affected me

metak commented on 2018-01-25 16:02 (UTC)

@ahjolinna I can't help with the repo since that is the limitation with archlinux support in opensuse build service. I reported that issue years ago. ;-) Basically, build service rebuilds packages even when that's not actually necessary and thus pkgver-pkgrel don't change but the package is now 'corrupt'. It's safe to just install it with pacman -U. Although, this kernel builds fast even on my old machine.

As for "intel-ucode", there's a bug opened, known issues, debian and redhat also reverted to that older version, and intel removed the 20180108 tarball from their site.

ahjolinna commented on 2018-01-25 15:22 (UTC) (edited on 2018-01-25 15:25 (UTC) by ahjolinna)

I think your repo is "broken" least repo key (Remote key not fetched correctly from keyserver.) and same goes with the pkgs, I get "invalid or corrupted package (PGP signature)" when I tried to install them directly

PS. is there a reason why you are using the older "intel-ucode" pkgs compare to whats is in arch repos?

metak commented on 2018-01-22 01:49 (UTC) (edited on 2021-12-01 12:47 (UTC) by metak)

1. Binaries available in my repo:
SigLevel = Never
Server =$arch
2. After install adjust your boot cmd line. ClearLinux uses clr-boot-manager which takes care of that. This is upstream default:
quiet console=tty0 console=ttyS0,115200n8 cryptomgr.notests initcall_debug intel_iommu=igfx_off kvm-intel.nested=1 no_timer_check noreplace-smp page_alloc.shuffle=1 rcupdate.rcu_expedited=1 rootfstype=ext4,btrfs,xfs,f2fs tsc=reliable rw 

metak commented on 2018-01-22 01:20 (UTC)

@OneObsession Sorry about that, copy/pasta wrong pkgbuild file. Fixed, but I didn't bumped the pkgrel.

OneObsession commented on 2018-01-20 11:42 (UTC)


is there a reason why you changed "make ${MAKEFLAGS} LOCALVERSION= bzImage modules" to "make ARCH=x86_64 LOCALVERSION= CONFIG_DEBUG_SECTION_MISMATCH=y ${MAKEFLAGS}"? I think bzImage and modules should be kept.

metak commented on 2018-01-19 17:57 (UTC)

It's not enabled in this kernel.

CONFIG_DM_CACHE is not set

rku commented on 2018-01-19 09:34 (UTC) (edited on 2018-01-19 09:35 (UTC) by rku)

Using llvm2 hook with this kernel reports following errors:

==> ERROR: module not found: `dm_cache'

==> ERROR: module not found: `dm_cache_smq'