unofficial linux-ck repo with cpu optimized packages is ONLINE

Lockheed · 2015-02-28 11:10:06

nbvcxz wrote:

Lockheed wrote:
Hi, graysky. What ck kernel, if any, should I install on btver2 Jagura AMD CPU?
According to this wiki page the ck-piledriver is for btver2 . Am I wrong?

Look closely. It says bdver2, not btver2.

graysky · 2015-02-28 11:24:34

nbvcxz wrote:

Lockheed wrote:
Hi, graysky. What ck kernel, if any, should I install on btver2 Jagura AMD CPU?
According to this wiki page the ck-piledriver is for btver2 . Am I wrong?

That is not the same as your processor I think. From my GCC patch:

...
+        cflags-$(CONFIG_MPILEDRIVER) += $(call cc-option,-march=bdver2)
+        cflags-$(CONFIG_MJAGUAR) += $(call cc-option,-march=btver2)
...

EDIT:

Lockheed wrote:

Look closely. It says bdver2, not btver2.

Ha! You beat me to it!

Last edited by graysky (2015-02-28 11:25:21)

macpd · 2015-03-01 21:52:14

Dearest graysky,

I'm having an odd issue with linux-ck precompiled kernel (Specifically version 3.19.0-1-ck), dhcp, and iptables NAT'ing.

I am unable to boot my system using the linux-ck kernel (generic, and sandybridge) due to a kernel panic during dhcp systemd unit initialization for a specific interface. I have narrowed down the issue to a the following single iptables NAT rule (forwards packets to this host on one port to another so that a non-privileged process is reachable on a "privileged" port):
"""
-A PREROUTING -i <iface> -p tcp -m tcp --destination <host IP> --dport <privileged port> -j REDIRECT --to-ports <non-privileged port>
"""

The system boots successfully in the following conditions:
a) the above rule is enabled, and booting with the default arch kernel (3.18.6-1)
b) rule disabled, and booting with linux-ck kernel
c) rule enabled, Ethernet cable removed from the NIC, and booting with the linux-ck kernel

In situation (c) the kernel panic still happens if I plug in the cable and use dhcpd as a systemd unit to assign an IP to the NIC. In situation (b) if I add the rule (via iptables command) and run restart the dhcpd service for this NIC it works.

I have fixed this by configuring the dhcpd unit to start before iptables.

I am unsure why the kernel panic only occurs when using the linux-ck kernel, and wonder if you have any insight as to why or hints for investigating further.

thanks

Lockheed · 2015-03-02 08:56:29

graysky wrote:

Ha! You beat me to it!

So, going back to my original question - what is the answer to it, because I still don't really know?

What ck kernel, if any, should I install on btver2 Jagura AMD CPU? If it is not supported, are there any plans to add its support in the foreseeable future?

Last edited by Lockheed (2015-03-02 08:57:36)

graysky · 2015-03-02 09:50:45

@macpd - sorry, no idea. You did run the control experiment by booting into 3.19-1-ARCH, yes?

@lock - no, I currently don't build a package set for your processor. You'll have to build from the AUR.

Lockheed · 2015-03-02 09:52:20

graysky wrote:

@lock - no, I currently don't build a package set for your processor. You'll have to build from the AUR.

Thanks. So it will automatically compile with support for Jaguar, right?

graysky · 2015-03-02 11:01:37

No, you need to enable the nconfig option and select that processor if you don't want generic x84-64.

Lockheed · 2015-03-02 16:29:15

graysky wrote:

@lock - no, I currently don't build a package set for your processor. You'll have to build from the AUR.

I forgot to ask: do you plan to add its support in the foreseeable future?

graysky · 2015-03-02 20:12:24

Lockheed wrote:

I forgot to ask: do you plan to add its support in the foreseeable future?

No plans to add it. Sorry.

guiniol · 2015-03-03 09:55:58

@graysky I have a server with unused bandwidth, and whenever the connection drops too much from my laptop, I download the packages there and then to my laptop, and on my server I have never seen the connection drop. Would you be interested in my setting up a repo-ck mirror?
I haven't checked the whole thread to see if someone offered already and you didn't want to for some reason, so if that's the case, just disregard my proposal.
Otherwise, I have never made a mirror, but it seems a simple web server and a way to update the mirror should be enough, and not that much of a hassle to setup.

graysky · 2015-03-03 10:16:28

Thanks for the offer, but if we mirror the packages, I can't analyze the package stats to help me understand package usage so I can remove unused groups. See repo-ck.com/stats.pdf for an example.

guiniol · 2015-03-03 10:21:54

I could give you the logs each month if you want.
I saw that and I was wondering if it would be a problem, but then I saw that godaddy wasn't giving you the logs so I thought you had stopped doing them.

graysky · 2015-03-03 19:53:49

Thanks for the generous offer but merging the logs would be nightmarish... you should see the way I parse through them now

guiniol · 2015-03-03 20:09:56

Alright, I can imagine, and no, I don't want to know. ^^
Seriously, though, if you ever change your mind, you know where to find me.

Nordic89 · 2015-03-04 09:16:24

Hey Graysky,

have you considered only releasing linux-ck versions of kernels that are already in the official repos? I've been having some major instabilities and crashes with the latest linux-ck kernels, and I think it might be smarter to not be so many kernel versions ahead of the "official" linux releases from the arch team.

The current linux kernel in arch is 3.18.6-1 and it'd be great to have a linux-ck version of that kernel instead of 3.19 already. I really value all the work you put into this, so I don't want this to sound like I wouldn't appreciate it, but since I've even had a crash that corrupted gnutls and libgcrypt and forced me to reinstall linux instead of linux-ck, I haven't even been using linux-ck anymore because I'm afraid of the instability.

Anyway, I'd be happy if you could put some thought into this (or maybe have both? A linux-ck-fresh and linux-ck-stable?) and good luck with however you decide.

graysky · 2015-03-04 09:47:36

You can find the last version on repo-ck.com. Just browse to the archives section. I have been running 3.19 since the last rc without issue. It's not clear to me why 3.19-1-ARCH remains in [testing] for nearly a month now.

Soukyuu · 2015-03-04 12:49:41

So far when I was having instabilities running the -ck kernel, it turned out to not be the -ck kernel but something else.
For example, for some reason using the BFQ scheduler for my SSD (system only) causes a random kernel panic when I wrote to the machine over SMB, even though I was writing to another HDD which used CFQ...

fhr · 2015-03-05 06:13:42

Hi,

I have a Celeron 2995u (inside a Gigabyte Brix). It's a Haswell processor and "march" matches that:

gcc -c -Q -march=native --help=target | grep march
  -march=                     		haswell

The generic linux-ck kernel works just fine, however linux-ck-haswell doesn't boot, I'm stuck at "Booting the kernel". Any idea? This is the first time I use the CK kernels on this system so I have no idea if this is related to the latest version or if it is an older problem.

Thanks!

graysky · 2015-03-05 07:57:47

@fhr - Celeron != full blown i7 I think. The celeron is missing some key extensions. You can check that on the intel processor finder or by `cat /proc/cpuinfo` and compare it to the output from my i7-4790k below. Just use the generic ck packages.

% cat /proc/cpuinfo 
processor	: 0
vendor_id	: GenuineIntel
cpu family	: 6
model		: 60
model name	: Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz
stepping	: 3
microcode	: 0x1c
cpu MHz		: 4399.843
cache size	: 8192 KB
physical id	: 0
siblings	: 8
core id		: 0
cpu cores	: 4
apicid		: 0
initial apicid	: 0
fpu		: yes
fpu_exception	: yes
cpuid level	: 13
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 fma cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm ida arat pln pts dtherm tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid xsaveopt
bugs		:
bogomips	: 8000.78
clflush size	: 64
cache_alignment	: 64
address sizes	: 39 bits physical, 48 bits virtual
power management:
...

Last edited by graysky (2015-03-05 07:59:00)

fhr · 2015-03-05 17:52:44

Thanks for the input (and for the hard work with repo-ck!).

Comparing your cpuinfo and one for a 2955u (found online, I don't have access to my machine right now), I see that the following extensions are in the i7 but not the celeron:

aes avx avx2 bmi1 bmi2 f16c fma ida smep x2apic

The celeron has 1 extension that the i7 doesn't have, "epb".

Just out of curiousity, do you know what the key one(s) is(are)?

Šaran · 2015-03-05 19:32:50

Hello graysky,

Would you consider building for 'silvermont' architecture? It's code name for Intel's Bay Trail platform, which is
as far as I can tell, hugely popular nowdays. Most of sub-$300 laptops are based around Bay Trail, together with most of
the Chromebooks. CK kernel has done wonders for my Nehalem i7 under heavy load, so I was hoping that there might be
enough people (run poll maybe?) who would benefit from architecture specific optimizations on our underpowered laptops.
I'd be happy to be beta tester or provide any kind of assistance you might need.

graysky · 2015-03-05 20:01:53

@saran - Not sure there is a gcc option specifically for this arch is there? DO you have access to this hardware? Post the output of the gcc command shown on the repo-ck wiki page.

Šaran · 2015-03-05 22:10:59

Here is gcc output:

 % gcc -c -Q -march=native --help=target | grep march
  -march=                     		silvermont

And yes, laptop I use most of the time is based on Bay Trail.

graysky · 2015-03-05 22:51:20

Guess step one would be to add silvermont to the kernel_gcc_patch.

Can you also post the output of

gcc -c -Q -march=native --help=target | grep mtune

and

cat /proc/cpuinfo

Last edited by graysky (2015-03-05 22:52:01)

Šaran · 2015-03-06 10:13:30

There you go.

 % gcc -c -Q -march=native --help=target | grep mtune
  -mtune-ctrl=                		
  -mtune=                     		silvermont

 % cat /proc/cpuinfo
processor	: 0
vendor_id	: GenuineIntel
cpu family	: 6
model		: 55
model name	: Intel(R) Celeron(R) CPU  N2840  @ 2.16GHz
stepping	: 8
microcode	: 0x829
cpu MHz		: 499.741
cache size	: 1024 KB
physical id	: 0
siblings	: 2
core id		: 0
cpu cores	: 2
apicid		: 0
initial apicid	: 0
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 movbe popcnt tsc_deadline_timer rdrand lahf_lm 3dnowprefetch ida arat epb dtherm tpr_shadow vnmi flexpriority ept vpid tsc_adjust smep erms
bugs		:
bogomips	: 4326.40
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 1
vendor_id	: GenuineIntel
cpu family	: 6
model		: 55
model name	: Intel(R) Celeron(R) CPU  N2840  @ 2.16GHz
stepping	: 8
microcode	: 0x829
cpu MHz		: 499.741
cache size	: 1024 KB
physical id	: 0
siblings	: 2
core id		: 1
cpu cores	: 2
apicid		: 2
initial apicid	: 2
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 movbe popcnt tsc_deadline_timer rdrand lahf_lm 3dnowprefetch ida arat epb dtherm tpr_shadow vnmi flexpriority ept vpid tsc_adjust smep erms
bugs		:
bogomips	: 4326.40
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

Arch Linux

#2176 2015-02-28 11:10:06

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

#2177 2015-02-28 11:24:34

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

#2178 2015-03-01 21:52:14

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

#2179 2015-03-02 08:56:29

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

#2180 2015-03-02 09:50:45

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

#2181 2015-03-02 09:52:20

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

#2182 2015-03-02 11:01:37

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

#2183 2015-03-02 16:29:15

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

#2184 2015-03-02 20:12:24

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

#2185 2015-03-03 09:55:58

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

#2186 2015-03-03 10:16:28

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

#2187 2015-03-03 10:21:54

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

#2188 2015-03-03 19:53:49

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

#2189 2015-03-03 20:09:56

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

#2190 2015-03-04 09:16:24

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

#2191 2015-03-04 09:47:36

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

#2192 2015-03-04 12:49:41

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

#2193 2015-03-05 06:13:42

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

#2194 2015-03-05 07:57:47

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

#2195 2015-03-05 17:52:44

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

#2196 2015-03-05 19:32:50

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

#2197 2015-03-05 20:01:53

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

#2198 2015-03-05 22:10:59

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

#2199 2015-03-05 22:51:20

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

#2200 2015-03-06 10:13:30

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

Board footer