Welcome!
Important information
-- Spectre and Meltdown vulnerabilities
-- Change in MX sources

News
-- MX Linux on social media: here
-- Mepis support still here

Current releases
-- MX-17.1 Final release info here
-- antiX-17 release info here

New users
-- Please read this first, and don't forget to add system and hardware information to posts!
-- Here are the Forum Rules

kernel options

Help for Current Versions of MX
Message
Author
User avatar
entropyfoe
Forum Regular
Forum Regular
Posts: 801
Joined: Thu Apr 19, 2007 11:42 am

Re: kernel options

#101 Post by entropyfoe » Tue Feb 20, 2018 3:32 pm

Stevo

I see source code here

https://github.com/xanmod/linux/releases
Asus Prime 370X-Pro
AMD Ryzen 1600X (6 cores @ 3.6 GHz)
16 Gig DDR4 3200 (G Skill)
Nvidia -MSI GeForce GT 710 fanless
Samsung 960 NVMe SSD nvme0n1 P1,P2, and P3=MX-17.1, P4=antiX-17
1TB SSD sda1= Data, 2TB WD =backups
on-board ethernet & sound

User avatar
Stevo
Forum Veteran
Forum Veteran
Posts: 16305
Joined: Fri Dec 15, 2006 8:07 pm

Re: kernel options

#102 Post by Stevo » Tue Feb 20, 2018 5:44 pm

Yes, it looks like they add their changes to the vanilla kernel source, then do a manual config and deb package. I don't see their config file or any debianized source anywhere like we get with the Liquorix kernel.

User avatar
entropyfoe
Forum Regular
Forum Regular
Posts: 801
Joined: Thu Apr 19, 2007 11:42 am

Re: kernel options

#103 Post by entropyfoe » Tue Feb 20, 2018 9:22 pm

OK so far. Still running. Almost one day uptime with 4.15.3-antix.1 and nvidia 384.111, and the grub parameter passed to the kernel
rcu_nocbs=0-11
caprea, those links are spot on, and this bug is being shown to amd, and to kernel people, so eventually it will get fixed.
Still not proof, but very encouraging.
Based on last week, I would say I would have had 2 crashes by now. But I have none !

I have a good feeling about this.
A few months ago while researching these Ryzen stability issues, I saw many mentions of the C6 state and to disable it in the BIOS. It is some thing in power management. I looked once in the BIOS for such a setting, but did not see one. I will recheck after the stability testing, but for now, I don't think something with C6 is available to me.
Last edited by entropyfoe on Wed Feb 21, 2018 10:46 am, edited 1 time in total.
Asus Prime 370X-Pro
AMD Ryzen 1600X (6 cores @ 3.6 GHz)
16 Gig DDR4 3200 (G Skill)
Nvidia -MSI GeForce GT 710 fanless
Samsung 960 NVMe SSD nvme0n1 P1,P2, and P3=MX-17.1, P4=antiX-17
1TB SSD sda1= Data, 2TB WD =backups
on-board ethernet & sound

caprea
Forum Regular
Forum Regular
Posts: 113
Joined: Sat Aug 23, 2014 7:01 am

Re: kernel options

#104 Post by caprea » Tue Feb 20, 2018 11:11 pm

Sounds good.
There's a good chance, that the 4.15 with the boot-parameter will be as stable as the 4.10, so why changing any BIOS-settings if it's not necessary.
Image

User avatar
stsoh
Forum Regular
Forum Regular
Posts: 393
Joined: Sun Aug 20, 2017 10:11 am

Re: kernel options

#105 Post by stsoh » Wed Feb 21, 2018 3:24 am

Code: Select all

For 16-threads Ryzen (1800X, 1700X, 1700):
rcu_nocbs=0-15

For 12-threads Ryzen (1600X, 1600):
rcu_nocbs=0-11

For 8-threads Ryzen (1500X, 1500, 1400):
rcu_nocbs=0-7

For 4-threads Ryzen (1300X, 1200):
rcu_nocbs=0-3 
MX-17.1_x64 Horizon, G41M-P33 Combo(MS-7592), Pentium E5400 (min/max: 1203/2700 MHz), 8Gb RAM (800 MT/s),
Intel 4 Series Integrated Graphics, Realtek PCIe Fast RTL8101/2/6E, PCI Gigabit RTL8169 Ethernets.

User avatar
entropyfoe
Forum Regular
Forum Regular
Posts: 801
Joined: Thu Apr 19, 2007 11:42 am

Re: kernel options

#106 Post by entropyfoe » Thu Feb 22, 2018 11:44 am

Two days 11 hours with no crash on kernel 4.15.3-antix.1. I think this problem is resolved. Many in caprea's links reported that the the rcu_nocbs=0-11 is the only thing need to fix. The BIOS changes are not needed, and so far I concur.

This is looking solved by the rcu_nocbs=0-11 line added to grub to pass to the kernel.
I think the rcu was configured correctly in 4.10, and as one of caprea's links showed, this was changed aftewards in later kernels.

So, now it requires the kernel be compiled with the correct rcu config, and then manually add the rcu_nocbs=0-11.
Eventually, this regression will be fixed.

The kernels made available by anti and the technical expertise here in this forum are impressive.
Thanks to all who helped with these kernel issues, anti, D-O, and especially caprea.

I think I have a stable system, latest kernel, latest nvidia, recent BIOS (1201 from November 2017).
I tried a day at very low load, which is the usual cause for the crash, not even firefox, just a terminal open, still no crash.

Now I will try and get the temp sensors to work and see if I can get to the >41 days uptime I had on 4.10 and 4.9.77.
Back to normal use !
Asus Prime 370X-Pro
AMD Ryzen 1600X (6 cores @ 3.6 GHz)
16 Gig DDR4 3200 (G Skill)
Nvidia -MSI GeForce GT 710 fanless
Samsung 960 NVMe SSD nvme0n1 P1,P2, and P3=MX-17.1, P4=antiX-17
1TB SSD sda1= Data, 2TB WD =backups
on-board ethernet & sound

caprea
Forum Regular
Forum Regular
Posts: 113
Joined: Sat Aug 23, 2014 7:01 am

Re: kernel options

#107 Post by caprea » Thu Feb 22, 2018 1:26 pm

That is really good news !

Did you also see this links
https://docs.google.com/forms/d/e/1FAIp ... w/viewform
https://docs.google.com/spreadsheets/d/ ... =221910414

Are you sure your sensors are not working, bc in the inxi 2 sides back there is cpu: 24.0C mobo: N/A gpu: 40C ?

User avatar
Stevo
Forum Veteran
Forum Veteran
Posts: 16305
Joined: Fri Dec 15, 2006 8:07 pm

Re: kernel options

#108 Post by Stevo » Thu Feb 22, 2018 4:30 pm

How many cores do you have on your Ryzen? How long does it take to build a kernel if you use all of them?

User avatar
entropyfoe
Forum Regular
Forum Regular
Posts: 801
Joined: Thu Apr 19, 2007 11:42 am

Re: kernel options

#109 Post by entropyfoe » Thu Feb 22, 2018 4:47 pm

Stevo,

I have 6 cores, 12 threads (logical cores).
This is the sweet spot for value in my opinion.
You get the same top speed of the Ryzen 1800 (it has 8 cores/16 threads 3.6 GHz with 4 GHz turbo), but minus two cores. So 75% of the multi-thread performance, at about half the cost, a good value in my book.

I have never compiled a kernel in my life, so I don't know. But I do the snapshot, and it cranks 1100% CPU (11 threads at 100%) for about 5 minutes when doing the squasfs
I can time one if you want to know the seconds. I am impressed.

With plenty of RAM and the fast NVMe sda, most operations happen instantly. It returns from hibernate in 4(music playing) to 7(full desktop) seconds.
The problem with new hardware is you need the recent kernel, and have to do some troubleshooting to get all working.

Thanks to the all the experts here, I got the kernel and help to get it running.
Last edited by entropyfoe on Thu Feb 22, 2018 5:04 pm, edited 1 time in total.
Asus Prime 370X-Pro
AMD Ryzen 1600X (6 cores @ 3.6 GHz)
16 Gig DDR4 3200 (G Skill)
Nvidia -MSI GeForce GT 710 fanless
Samsung 960 NVMe SSD nvme0n1 P1,P2, and P3=MX-17.1, P4=antiX-17
1TB SSD sda1= Data, 2TB WD =backups
on-board ethernet & sound

User avatar
Stevo
Forum Veteran
Forum Veteran
Posts: 16305
Joined: Fri Dec 15, 2006 8:07 pm

Re: kernel options

#110 Post by Stevo » Thu Feb 22, 2018 4:57 pm

Thanks for the info. It turns out not to be very difficult to set up a pbuilder environment and build something like the Liquorix kernel from the debianized sources with the number of threads that you specify, and even cross-compile (i386 packages on amd64). That's how I'm building Liquorix for our repo, among with most other packages now.

Post Reply

Return to “MX Help”