You are not logged in.
I have been using the linux-ck patched kernel from the AUR for the last few weeks https://aur.archlinux.org/packages/Linux-ck/
I decided I would try to compile it with the gcc optimised patch as well. https://github.com/graysky2/kernel_gcc_patch
I have a laptop using the intel i7-4712MQ CPU which I understand is part of the Haswell family.
I downloaded the tarball from the AUR, extracted it, I edited the PKGBUILD, to enabe the nconfig and also the BFQ scheduler, ran makepkg -s enabled the patch for Haswell chip family and saved it as .config and carried on building the kernel as normal, then installed the resulting packages with pacman -U kernel and the header and then rebooted.
on boot I received a failed to execute /init error 8 and no working init found, I fell back to using the default Arch Kernel and can also rebild the linux-ck patched kernel as normal with the GCC optimisation patch with no problems, but I want to understand what I did wrong when applying the GCC patch and fix it, if there is another log I can get please let me know and I will post it.
I know it is not the new version of the Kernel as I had built version 3.19-1 it without the gcc optimisation patch successfully on Friday
Thanks
Rob
Last edited by robt77 (2015-03-01 12:28:58)
Offline
I downloaded the tarball from the AUR, extracted it, I edited the PKGBUILD, to enabe the nconfig and also the BFQ scheduler, ran makepkg -s enabled the patch for Haswell chip family and saved it as .config and carried on building the kernel
This is not very clear. What do you mean you carried on building the kernel after running makepkg -s? Makepkg -s builds the packages, any configuration you make after that command is not applied to the built packages.
on boot I received a failed to execute /init error 8 and no working init found
Then the kernel loaded just fine - init failed. Did you (re)build an initramfs for the new kernel? What is the bootloader command used for this kernel?
"UNIX is simple and coherent" - Dennis Ritchie; "GNU's Not Unix" - Richard Stallman
Offline
If you built the kernel with makepkg and installed it with pacman, you won't need to rebuild initramfs, it's suppose to be done automatically. The only manual intervention is adding a boot entry. I guess you should post some confs as Trilby pointed out for your bootloader command.
Last edited by smirky (2015-03-01 13:01:16)
Personal spot :: https://www.smirky.net/ :: Try not to get lost!
Offline
robt77 wrote:I downloaded the tarball from the AUR, extracted it, I edited the PKGBUILD, to enabe the nconfig and also the BFQ scheduler, ran makepkg -s enabled the patch for Haswell chip family and saved it as .config and carried on building the kernel
This is not very clear. What do you mean you carried on building the kernel after running makepkg -s? Makepkg -s builds the packages, any configuration you make after that command is not applied to the built packages.
after running makepkg -S it came up with the menu to select different options for the kernel, so I arrowed to Processor type and features ---> Processor family ---> and found Intel Haswell, selected that and saved it by pressing f5 it saved to a file called .config and it started building the kernel
robt77 wrote:on boot I received a failed to execute /init error 8 and no working init found
Then the kernel loaded just fine - init failed. Did you (re)build an initramfs for the new kernel? What is the bootloader command used for this kernel?
44
I did not buikd ab initramfs, I have always just had to download the tar ball from AUR, extract it , then run makepkg -s and then install the packages with pacman -U and reboot.
Here is a link to the bootloader command from grub http://hastebin.com/roweqocuqi.vala
Offline
The problem could be that your chip does not have the full set of extensions that -march=haswell expects... that came up before with a user who had a pentium not full blown i7. I am not familiar enough with the full line of processors to know if yours does indeed have these... May or may not be at work in your case.
Last edited by graysky (2015-03-01 13:16:45)
Offline
The problem could be that your chip does not have the full set of extensions that -march=haswell expects... that came up before with a user who had a pentium not full blown i7. I am not familiar enough with the full line of processors to know if yours does indeed have these...
ok, I will just re build without the additional optimisations, the kernel works great without it, thought I would see what extra I could get out of it
Offline
ok, I will just re build without the additional optimisations, the kernel works great without it, thought I would see what extra I could get out of it
You can also use precomplied packages I build to save time. See my sig.
Offline