Arch Linux User Repository

Important update
I have updated and renamed some settings envvars. Please check your scripts, aliases, brain memory or whatever you use to store your build command and update them accordingly. Please read the description in the PKGBUILD file instead of the short summary in the pinned comment on the AUR or in the README in the repository as some important details important to this update are only found inside the PKGBUILD.

Information comment
This comment provides useful information about this package. Make sure to read it through, you won't regret it.

Repository URL
This package is hosted at sos!git. If you want to contribute to the package, you can do it here: https://git.staropensource.de/JeremyStarTM/aur-linux-clear

Issue handling
Please only submit issues in the package's repository if they aren't bug reports about the package or help requests. Please keep them to the AUR only. This prevents unnecessary signups.

Updates
I usually publish updates 0-3 days after a kernel release, depending on how busy I am.
If I however get wind of security-relevant updates or any vulnerabilities I will try to update the kernel as soon as possible and write a comment about the vulnerability/update. So make sure to subscribe to this package's comment mailing list.

PKGBUILD environment variable customization
The PKGBUILD file supports customization via environment variables. Here's a list of all of them:
- _makemenuconfig: Invokes make menuconfig (ncurses-based default configuration menu) before compilation starts
- _makenconfig: Invokes make nconfig (ncurses-based nicer configuration menu) before compilation starts
- _makexconfig: Invokes make xconfig (X11-based GUI configuration menu) before compilation starts
- _reuse_current: Will use the configuration of the running kernel, if the running kernel is compiled with IKCONFIG_PROC
- _reuse_file: Will use the configuration file named kconfig contained in the PKGBUILD file's directory
- _update_kconfig_on_reuse: Specifies whether the kernel configuration shall be updated when using any of the 'reuse*' flags
- _copyfinalconfig: Copies the final kernel configuration into the repository root as kconfig-new before compilation starts
- _localmodcfg: Only compiles modules found in modprobed-db's database (which decreases compilation time and kernel size)
- _use_llvm_lto: Compiles the kernel with LLVM instead of GCC. Should work, if not open an issue
- _subarch: Specifies the subarchitecture to compile for (see the PKGBUILD file for a list of all subarches). Default is 41 (Generic x86-64), which is compatible with all amd64 processors. Must be a number
- _use_llvm_lto: Enables compilation with LLVM. Untested but should work. If not write a comment on the AUR.
- _subarch_microarch: Specifies the microarchitecture to compile for. Only applies to and is required by the GENERIC_CPU subarch. Must be a number between 1 and 4 (click for more information)
- _debug_kernel: Force enables kernel debugging options when set to y, force disables debug options when set to n or relies on the kconfig when unset
- _debug_script: Enables debugging for the PKGBUILD file

All of these variables just need to be set for them to apply, except for _subarch, _subarch_microarch and _debug_kernel. For more information please check their documentation inside the PKGBUILD file

GPG errors
If makepkg complains about invalid PGP keys, try running this command first: gpg --locate-keys torvalds@kernel.org gregkh@kernel.org sashal@kernel.org benh@debian.org. This command only needs to be executed once.

Build tool
If you want a dead simple way to compile this package, you can try out my build tool. It includes a few no nonsense changes to the kernel configuration and allows you to configure the PKGBUILD file in a simple manner and decrease build time. And if you don't want that, using makepkg as-is works fine too.

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.

@wshuman3 & @tam1m Thank you for chiming in.

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

@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

@metak

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.
Set CONFIG_SECTION_MISMATCH_WARN_ONLY=y to allow them.
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

Great! Much obliged.

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

https://www.kernel.org/doc/html/v5.6/admin-guide/kernel-parameters.html

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.

I understand. Thank you.

@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.

==> 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/Makefile.build:267: arch/x86/kvm/../../../virt/kvm/kvm_main.o] Error 1
make[1]: *** [scripts/Makefile.build:505: 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().
    Aborting...

No problem with previous iteration, but this one fails.