unofficial linux-ck repo with cpu optimized packages is ONLINE

graysky · 2018-05-07 18:52:22

@Ropid - I believe that when Arch packages are built, they do so in an environment that has [testing] enabled, whereas I currently build [repo-ck] with [testing] disabled. I can change this behavior moving forward, but probably not until I get some free time (this week maybe but more than likely over the weekend).

loqs · 2018-05-07 19:00:44

Arch builds packages targeting the repo they are initially intended for. As core packages must pass through testing that means they are built against testing.
Packages going straight to extra/community would not be built against testing.
Edit:
You can also extract the .BUILDINFO from a package and see which packages were present in the build environment.

Last edited by loqs (2018-05-07 19:07:01)

graysky · 2018-05-07 19:14:49

@loqs - Thanks for the clarification. Again, when I build [repo-ck] in the future, I will be mirroring this behavior. I just don't have a chunk of time to devote to changing the build process now.

EDIT: Looks like gcc-8.1.0 just hit [core]. I will rebuild 4.16.7 using it now and freshen the repo. Watch for 4.16.7-2 to hit [repo-ck] in a few hours.

Last edited by graysky (2018-05-07 19:16:48)

Ropid · 2018-05-07 19:37:03

Thinking about this some more, I think there's no need to change the build environment. It might even be a mistake? If things are built like for a 'testing' package, the same problem might show up in the opposite direction: linux-ck might be using newer versions of stuff, while a user at home then can't compile a dkms package because of using older versions. There's perhaps no good solution for this and best would be to simply keep things exactly like they are right now.

Last edited by Ropid (2018-05-07 19:37:29)

graysky · 2018-05-07 19:48:49

Ah the status quo!

eschwartz · 2018-05-07 19:49:02

Ropid wrote:

Thinking about this some more, I think there's no need to change the build environment. It might even be a mistake? If things are built like for a 'testing' package, the same problem might show up in the opposite direction: linux-ck might be using newer versions of stuff, while a user at home then can't compile a dkms package because of using older versions. There's perhaps no good solution for this and best would be to simply keep things exactly like they are right now.

Uh, I guess you could have [repo-ck] and [repo-ck-testing].

vuudochile · 2018-05-08 06:16:31

Thnx, bauerbill too I tried
The new -ck-16.7.2 came down in a jiffy
I wonder whether compiling for the next day or two beats it or whether I should get a 2nd gentoo pc for such things

dobie2564 · 2018-05-10 17:05:30

I've been using repo-ck without issue for years. But last time I updated with pacman -Syu, I receive the following eror:

"error: failed retrieving file 'nvidia-ck-haswell-1:390.48-3-x86_64.pkg.tar.xz' from repo-ck.com : The requested URL returned error: 404"

I'm at a loss to know where to start looking. Any help will be greatly appreciated

sknd · 2018-05-10 18:37:33

dobie2564 wrote:

I've been using repo-ck without issue for years. But last time I updated with pacman -Syu, I receive the following eror:
"error: failed retrieving file 'nvidia-ck-haswell-1:390.48-3-x86_64.pkg.tar.xz' from repo-ck.com : The requested URL returned error: 404"
I'm at a loss to know where to start looking. Any help will be greatly appreciated

the same problem here
I also tried mirror I found here - http://repo-ck.segfault.gq/$arch but: could not resolve host

any ideas? i have to use other kernel because i can't start GUI with kernel-ck now...

graysky · 2018-05-10 19:22:18

The missing nv packages are my fault... I bumped to 396.24 but the corresponding util package hasn't yet hit [extra]. Fixed now with nvidia-ck 2:390.48-1. Please refresh and report back.

dobie2564 · 2018-05-10 20:45:25

graysky wrote:

The missing nv packages are my fault... I bumped to 396.24 but the corresponding util package hasn't yet hit [extra]. Fixed now with nvidia-ck 2:390.48-1. Please refresh and report back.

Works great now. Thanks for the quick service.

pr0dukter · 2018-05-11 18:43:40

@graysky, i know this is above and beyond, but could you either teach me, or connect me with instructions that teach me how to pull all the updated modules from drm-next and/or amd-staging-drm-next and include those with the linux-ck bulldozer.....

the tricky part is there is also libdrm which is the other half of whats needed as i understand? could you just show me what part of your package build pulls from there or something so i can kinda modularly inject and remove drm as needed?

i would even push a binary and/or pkgbuild w/ git repo to whichever repo afterwards in order to contribute.

if im not understanding this right, please enlighten me but i keep getting:

0] RIP: 0010:generic_reg_update_ex+0x12b/0x158 [amdgpu]
[25985.267921] RSP: 0018:ffffa6ee00b47790 EFLAGS: 00010246
[25985.267922] RAX: ffffa6ee00b477b0 RBX: ffff9535a7765280 RCX: 0000000000000000
[25985.267923] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff9535a89ccb00
[25985.267923] RBP: ffffa6ee00b477f8 R08: 0000000000000000 R09: 0000000000000000
[25985.267924] R10: 0000000000000001 R11: 00000000fffffffb R12: 0000000000000000
[25985.267925] R13: 0000000000000000 R14: ffff9535a7778000 R15: ffff95359edc9000
[25985.267926] FS:  00007fd270a7f940(0000) GS:ffff9535bed00000(0000) knlGS:0000000000000000
[25985.267926] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[25985.267927] CR2: 0000563f2f92efd0 CR3: 000000041edd8000 CR4: 00000000000406e0
[25985.267928] Call Trace:
[25985.267970]  dce110_stream_encoder_update_hdmi_info_packets+0x205/0x398 [amdgpu]
[25985.268009]  dc_stream_adjust_vmin_vmax+0xb9/0xe8 [amdgpu]
[25985.268048]  set_freesync_on_streams.part.4+0x43/0x210 [amdgpu]
[25985.268087]  mod_freesync_notify_mode_change+0x11b/0x140 [amdgpu]
[25985.268127]  amdgpu_dm_atomic_commit_tail+0x52c/0xd80 [amdgpu]
[25985.268129]  ? preempt_count_add+0x68/0x98
[25985.268131]  ? _raw_spin_unlock_irq+0x1d/0x38
[25985.268132]  ? wait_for_common+0x112/0x190
[25985.268133]  ? _raw_spin_unlock_irq+0x1d/0x38
[25985.268134]  ? wait_for_common+0x112/0x190
[25985.268140]  commit_tail+0x3d/0x68 [drm_kms_helper]
[25985.268146]  drm_atomic_helper_commit+0x103/0x110 [drm_kms_helper]
[25985.268152]  restore_fbdev_mode_atomic+0x1a7/0x210 [drm_kms_helper]
[25985.268158]  drm_fb_helper_restore_fbdev_mode_unlocked+0x45/0x90 [drm_kms_helper]
[25985.268163]  drm_fb_helper_set_par+0x29/0x58 [drm_kms_helper]
[25985.268165]  fb_set_var+0x212/0x3e0
[25985.268167]  ? memcg_check_events.isra.21+0x39/0x1f0
[25985.268170]  fbcon_blank+0x26e/0x310
[25985.268173]  do_unblank_screen+0x9b/0x150
[25985.268175]  ? vt_ioctl+0x363/0x10f8
[25985.268176]  ? filename_parentat.isra.17.part.18+0xd3/0x140
[25985.268178]  ? __offline_pages+0x440/0x7f0
[25985.268179]  ? tty_ioctl+0x215/0x870
[25985.268181]  ? __inode_wait_for_writeback+0x7f/0xe8
[25985.268182]  ? current_time+0x16/0x60
[25985.268184]  ? do_vfs_ioctl+0xa4/0x610
[25985.268186]  ? __fput+0x121/0x1e0
[25985.268187]  ? preempt_count_add+0x68/0x98
[25985.268189]  ? SyS_ioctl+0x61/0x80
[25985.268190]  ? do_syscall_64+0x74/0x190
[25985.268192]  ? entry_SYSCALL_64_after_hwframe+0x3d/0xa2
[25985.268193] Code: 89 da 48 8b 07 48 8b 40 38 e8 42 e8 6b c4 48 8b 74 24 18 65 48 33 34 25 28 00 00 00 89 d8 75 20 48 83 c4 50 5b 41 5c 41 5d 5d c3 <0f> 0b e9 1a ff ff ff 44 89 e0 4d 89 cd 41 bc 01 00 00 00 eb 99 
[25985.268217] ---[ end trace 4534303478d331e5 ]---

which the amd-staging-drm-next kernels fix, but then i dont have all the ck-bulldozer optimizations that make an enormous difference for me.

loqs · 2018-05-11 19:54:54

git checkout amd-staging-drm-next
git remote add ck https://github.com/ckolivas/linux.git
git fetch ck
git cherry-pick caae7ce4055208164b84678a509e8dedc2ba43b7 #repeat for the rest of the commits from https://github.com/ckolivas/linux/commits/4.16-ck

Note this will break once amd-staging-drm-next rebases from 4.16 to 4.17
Then apply the rest of the linux-ck PKGBUILD (the stable patch will probably fail to apply due to conflicting changes relating to AMDGPU)

Last edited by loqs (2018-05-11 20:13:37)

pr0dukter · 2018-05-11 23:31:30

thank you, i really appreciate you for taking the time to help me, between this and mesa bugs i dont know if im doing wrong or is buggy, because eitherway potential to fail :D

thanks so much you really made my night and next several days :D

graysky · 2018-05-12 19:40:52

Now shipping nvidia-390xx-ck on [repo-ck]. Please use this thread to report any bugs as usual.

keepitsimpleengineer · 2018-05-12 21:24:08

nvidia-390xx-ck ~ Very cool... .. !

Last edited by keepitsimpleengineer (2018-05-12 21:25:05)

keepitsimpleengineer · 2018-05-13 00:03:05

graysky wrote:

Now shipping nvidia-390xx-ck on [repo-ck]. Please use this thread to report any bugs as usual.

==> Validating source files with sha256sums...
    NVIDIA-Linux-x86_64-390.48-no-compat32.run ... Passed
    kernel-4.16.patch ... Passed
==> Extracting sources...
==> Starting prepare()...
Creating directory NVIDIA-Linux-x86_64-390.48-no-compat32
Verifying archive integrity... OK
Uncompressing NVIDIA Accelerated Graphics Driver for Linux-x86_64 390.48....
patching file kernel/common/inc/nv-linux.h
patching file kernel/conftest.sh
==> Starting build()...
cat: /usr/lib/modules/extramodules-4.16-ck/version: No such file or directory
==> ERROR: A failure occurred in build().

...and after checking to see if linux-ck was installed (it wasn't)

[ljohnson@KISE-066 nvidia-390xx-ck]$ makepkg -srci
==> Making package: nvidia-390xx-ck 390.48-1 (Sun May 13 11:16:00 PDT 2018)
==> Checking runtime dependencies...
==> Checking buildtime dependencies...
==> Retrieving sources...
  -> Found NVIDIA-Linux-x86_64-390.48-no-compat32.run
  -> Found kernel-4.16.patch
==> Validating source files with sha256sums...
    NVIDIA-Linux-x86_64-390.48-no-compat32.run ... Passed
    kernel-4.16.patch ... Passed
==> Extracting sources...
==> Starting prepare()...
Creating directory NVIDIA-Linux-x86_64-390.48-no-compat32
The directory 'NVIDIA-Linux-x86_64-390.48-no-compat32' already exists.  Please either
move the existing directory out of the way, or specify a
different directory with the '--target' option.
==> ERROR: A failure occurred in prepare().
    Aborting...

UPDATE: Turns out booting with linux-ck-k10 fails due to failing to load nic, still sorting that out
~precompiled nvidia-390xx-ck installs as expected

Possible compile error above due the fact that extra/nvidia-390xx is also installed..?

Last edited by keepitsimpleengineer (2018-05-13 20:17:57)

Trilby · 2018-05-13 01:03:14

keepitsimpleengineer, please edit your post to add code tags. While you're at it, be sure to include the full command you used. The first error is very clearly your failing as linux-ck is properly listed as a dependency. The second most likely is as well depending on the flags you passed to makepkg. It looks like this has nothing in particular to do with the linux-ck packages (and definitely not the precompiled binaries in [repo-ck]) but is just a failure to use makepkg properly.

loqs · 2018-05-13 19:37:13

@keepitsimpleengineer what is the output if you change the flags to -Crsi ?

klingt.net · 2018-05-16 12:29:36

Since one of the last linux-ck kernel releases (maybe since 4.16.8) I am not able to enter the password for my encrypted root partition correctly because some keystrokes are registered more than once.
This does not happen when I boot the default kernel.
I am building the kernel from source with native optimizations.
Any ideas what change could have introduced this issue?

vuudochile · 2018-05-16 15:00:38

I have had the same exact behavior with mine, both compiled and binary download of cpu specific kernels. I was using -ck directly from my distro's repository but in recent months it stopped. I had the same problem then and thought it was hw peculiarity.
This only happens on console though, once X goes up everything is normal. I just have to enter user and pw with really quick sharp strokes to get it right the first time.

klingt.net · 2018-05-16 15:48:36

I can confirm that this goes away when X is loaded.

vuudochile · 2018-05-17 12:01:41

I installed the update 4..16-9 and it still the same. It seems as the setting of how long it lags from key-stroke before it replicates the character is set to be too sort. But this setting seems to change or be reset by X and be slightly longer. It might be as simple as 200ms to 400ms difference.
Small problem but noticeable and irritating when working on a hard to see console.

john29 · 2018-05-26 11:59:06

Will there be any ryzen linux-ck kernel ,I have a 2600x now alway use the piledriver kernel and before that k10

graysky · 2018-05-26 12:59:43

@john29 - I've supported -march=znver1 for a while now...
https://wiki.archlinux.org/index.php/Un … d_packages
https://en.wikipedia.org/wiki/List_of_A … hitectures

Arch Linux

#3301 2018-05-07 18:52:22

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3302 2018-05-07 19:00:44

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3303 2018-05-07 19:14:49

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3304 2018-05-07 19:37:03

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3305 2018-05-07 19:48:49

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3306 2018-05-07 19:49:02

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3307 2018-05-08 06:16:31

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3308 2018-05-10 17:05:30

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3309 2018-05-10 18:37:33

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3310 2018-05-10 19:22:18

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3311 2018-05-10 20:45:25

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3312 2018-05-11 18:43:40

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3313 2018-05-11 19:54:54

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3314 2018-05-11 23:31:30

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3315 2018-05-12 19:40:52

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3316 2018-05-12 21:24:08

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3317 2018-05-13 00:03:05

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3318 2018-05-13 01:03:14

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3319 2018-05-13 19:37:13

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3320 2018-05-16 12:29:36

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3321 2018-05-16 15:00:38

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3322 2018-05-16 15:48:36

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3323 2018-05-17 12:01:41

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3324 2018-05-26 11:59:06

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#3325 2018-05-26 12:59:43

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

Board footer