Page 1 of 3

nrj >>> nrjQL, the Next Kernel Generation, a step forward...

Posted: 4 September 2012, 21:25
by NicCo
Image

ROSALAB, after some labs tests, choosed NRJ as the default predefined Kernel for ROSA 2012 MARATHON LTS, and I'm developing and maintaining it in a conservative and secure way, but so I couldn't experiment new interesting but sometimes dangerous stuff, we needed a development branch where we could add all new interesting stuff without the risk of harming some user and so we and our brave geeks will be able to test how the latest software technologies are working in the real world...

Some weeks ago, I quietly started working on NRJ 2 project (next generation NRJ Kernels)


The goal, the name, the contents / features?

The main goals are preparing some new and more performant Kernels for Mandriva and ROSA linux distros, but concurrently maintaining the complete previous NRJ patcheset, so if we need we can grab some things from the development branch to copy in the mainline kernel (NRJ) but being almost sure there should be no problems.

The kernel complete name should be something like:
kernel-nrj2-bfs-bfq-toi-realtime-adaptivetickless-lto
I dont think to adopt such a long and difficult name...
it's not yet decided, while temporary codename is > nrjX (meaning as nrj Xtreme)

These kernels will be more experimental and will represent the development branch:
including many not official kernel developments, when these stuff promise to reduce latence and improve responsiveness in many different fields, like task scheduler, disk I/O, multimedia playing of complex Multimedia and MIDI Professional, DAW (Digital Audio Station), Realtime applications, ecc.


Everything is still a big Work in progress...
but among the following items (see below), the points 1> to 10> have been already implemented and are already inside the kernel srpm contents:
for example AUFS3 is ready to be configured, all patches are already in srpms, but now are NOT applied/enabled to avoid possible conflict with OverlayFS


1> Mandriva ex mnb with more than 100 patches mainly improving HW compatibility
It should be possible to build the same kernels also for the new free Foundation born from Mandriva (MoonDrake?), but when ABF will be available...


2> MIB NRJ enabled flavours
You can read something about
viewtopic.php?f=38&t=3244
Image Image


3> CK full patcheset, to enhance Desktop behaviour and BFS (cpu task scheduler)
http://ck.kolivas.org/patches/bfs/bfs-faq.txt

http://users.on.net/~ckolivas/kernel/
"patches designed to improve system responsiveness and interactivity with specific emphasis on the desktop, but suitable to any workload."



4> BFQ - I/O disk scheduler
http://lwn.net/Articles/275978/
http://algo.ing.unimo.it/people/paolo/disk_sched/
here an interesting video



5> TuxOnIce
http://tuxonice.net/



6 > UKSM
For Data Deduplication Of The Linux Kernel
http://www.phoronix.com/scan.php?page=n ... px=MTEzMTI
english translation
http://translate.google.com/translate?s ... chmarks%2F



7> OverlayFS
http://sourceforge.net/projects/olfs/


8> AUFS3
http://aufs.sourceforge.net/


Amongst the others best kernel patchsets, we are considering mainly these

9>liquorix
http://liquorix.net/
kernel replacement built using the best configuration and kernel sources for desktop, multimedia, and gaming workloads.
10>zen
http://zen-kernel.org/
Zen has also taken its own leap to the future by taking a stab at modern release philosophies; the art of rolling code.
11> 40 patches for Adaptive Tickless
http://lwn.net/Articles/495422/
we have no more fixed HZ, for server now we have HZ=100, for desktop HZ=1000, while with Adaptative Tickless we have HZ=no, so the hires timers can change their frequence realtime when it need, so there is no overhead to kernel due to not used interrupts


12> LTO (link time optimization)
http://lwn.net/Articles/512548/
that's possible with gcc >= 4.7
this produces better internal code in kernel using the most efficient cpu instructions
despite this efficiency, the build time will raise by a factor 2X to 4X...


13> many others things to discover by a surprise... :)



First successfully rpms built with nrjX Kernel 3.0.42
the second kernel prepared and in testing is now Kernel nrjX 3.5.3

These files are the results of first implemented items (1> to 4> already in...)
as soon as possible I'll fix the last issues in the spec/scripts/configs so also You, dear brave hearts, could start testing

We hope TuningUp everything and release publicly for the end of September
As usual these Kernels will be prepared for MDV 2010 & 2011 to RM2012 & ROSA 2012.1

Code: Select all

Elaborazione file in corso: kernel-nrjX-source-latest-3.0.42-1mib2010.2                                                                 
Controllo per file scompattato/i in corso: /usr/lib/rpm/check-files /home/utente/rpmbuild/BUILDROOT/kernel-nrjX-3.0.42-1mib2010.2.x86_64
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-desktop-3.0.42-1mib-1-1mib2010.2.x86_64.rpm                                      
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-desktop-devel-3.0.42-1mib-1-1mib2010.2.x86_64.rpm                                
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-desktop-latest-3.0.42-1mib2010.2.x86_64.rpm                                      
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-desktop-devel-latest-3.0.42-1mib2010.2.x86_64.rpm                                
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-laptop-3.0.42-1mib-1-1mib2010.2.x86_64.rpm                                       
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-laptop-devel-3.0.42-1mib-1-1mib2010.2.x86_64.rpm                                 
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-laptop-latest-3.0.42-1mib2010.2.x86_64.rpm                                       
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-laptop-devel-latest-3.0.42-1mib2010.2.x86_64.rpm                                 
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-server-3.0.42-1mib-1-1mib2010.2.x86_64.rpm                                       
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-server-devel-3.0.42-1mib-1-1mib2010.2.x86_64.rpm                                 
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-server-latest-3.0.42-1mib2010.2.x86_64.rpm                                       
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-server-devel-latest-3.0.42-1mib2010.2.x86_64.rpm                                 
Scritto: /home/utente/rpmbuild/RPMS/noarch/kernel-nrjX-source-3.0.42-1mib-1-1mib2010.2.noarch.rpm                                       
Scritto: /home/utente/rpmbuild/RPMS/noarch/kernel-nrjX-source-latest-3.0.42-1mib2010.2.noarch.rpm                                       
Esecuzione(%clean) in corso: /bin/sh -e /var/tmp/rpm-tmp.wU3hH4                                                                         
+ umask 022                                                                                                                             
+ cd /home/utente/rpmbuild/BUILD                                                                                                        
+ cd kernel-nrjX-x86_64                                                                                                                 
+ rm -rf /home/utente/rpmbuild/BUILDROOT/kernel-nrjX-3.0.42-1mib2010.2.x86_64                                                           
+ exit 0   
[/size]
I installed and using since some days

Code: Select all

localhost proc # urpmi /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-desktop-devel-3.0.42-1mib-1-1mib2010.2.x86_64.rpm

installazione di kernel-nrjX-desktop-devel-3.0.42-1mib-1-1mib2010.2.x86_64.rpm da /home/utente/rpmbuild/RPMS/x86_64
In preparazione...               ######################################################################################################
      1/1: kernel-nrjX-desktop-devel-3.0.42-1mib
                                 ######################################################################################################
localhost proc # urpmi /home/utente/rpmbuild/RPMS/x86_64/kernel-nrjX-desktop-3.0.42-1mib-1-1mib2010.2.x86_64.rpm

installazione di kernel-nrjX-desktop-3.0.42-1mib-1-1mib2010.2.x86_64.rpm da /home/utente/rpmbuild/RPMS/x86_64
In preparazione...               ######################################################################################################
      1/1: kernel-nrjX-desktop-3.0.42-1mib
                                 ######################################################################################################
Dovresti riavviare il computer per kernel-nrjX-desktop-3.0.42-1mib
localhost proc #
[/size]

I and "oiram73" (our great professional tester) are using this featured kernels since some days, and first results are really impressive...
in the meantime, we are also testing the stability and reliability while on the PC for days to make several stressful work and some large data transfers

now we are waiting for the results of other tests from our MIB guys like Astragalo, Dago68, GvM...
https://wiki.archlinux.org/index.php/Ke ... taliano%29

bye, NicCo

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 4 September 2012, 23:35
by oiram73
First tests and feedback about kernel 3.0.42 nrjX
Image Let's give a surge of pure energy to our brand new Mandriva / ROSA Linux!

Dear friends, the 3.0.42 kernel-nrjX is the best Kernel I have ever used!
1) Amarok running with a playlist consisting of 40 days of music
2) Rosegarden performing the V Beethoven Symphony = 11 midi channels
3) Kmid running my music piece with 5 midi channels
4) ROMP playing a video in HD
5) 2010.2_x64 and berni-lomax doing her updates on virtualbox
6) Lilypond running which is usually hungry for memory and cpu breaks much less
7) I have 6 Gb of ram and only after I launched Lilypond exceeded 50%
8) 16 active windows, and you log in as if I were slipping on the oil not to mention the cube that starts immediately when the mouse is in the upper right (nrj-old hesitated a moment) -> running kwin effects is outstanding!
9) xruns: here appeared the first in the 20th minute: jackd reported it but honestly I have not heard and that's what counts as with other kernel were not only reported but also felt so at this point the elimination of the report is more a matter of formal elegance and practical.
Conclusions: With this kernel I do not feel inferior to Apple users who always made ​​her master in the audio field.
The only negative note is my touchpad scrolling that this series of kernel (3.0.x) does not work unlike the upper series (> = 3.2.x)!


Image Dai una sprizzata di pura Energia al tuo fiammante Mandriva / ROSA Linux!

Cari amici il 3.0.42-nrjX è il miglior kernel che abbia mai usato!
1) Aamrok in esecuzione con i suoi 40 giorni di musica
2) Rosegarden che esegue la V sinfonia di Beethoven = 11 canali midi
3) Kmid che esegue un mio pezzo con 5 canali midi
4) ROMP che proietta un film HD
5) 2010.2_x64 e berni-lomax che fa i suoi aggiornamenti su virtualbox
6) Lilypond in esecuzione che di solito è avido di memoria e di cpu rompe molto meno
7) Io ho 6 Gb di ram e solo dopo Lilypond ho superato il 50%
8) 16 finestre attive e vi accedo come se io stessi scivolando sull'olio per non parlare del cubo che parte immediatamente quando il mouse va sull'angolo in alto a destra (con nrj-old esitava un attimo) --> esecuzione effetti kwin eccezionale!
9) xruns : anche qui è comparso il primo al 20° minuto: jackd lo ha segnalato ma sinceramente non l'ho sentito e questo è quello che conta visto che con gli altri kernel non solo venivano segnalati ma si sentivano anche perciò a questo punto l'eliminazione della segnalazione è più un fatto di eleganza formale che pratica.
Conclusioni: con questo kernel non mi sento più inferiore agli utenti Apple che da sempre l'ha fatta da padrone in campo audio.
Unica nota negativa lo scrolling del tappetino che anche su questa serie di kernel (3.0.x) non funziona a differenza delle serie superiori (>= 3.2.x)!

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 5 September 2012, 16:00
by rolf
Hi,
I enjoyed watching the video demo and reading ( a little :) ) about the developer's innovation. This looks very interesting to me. Image

Thanks,
Rolf

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 6 September 2012, 17:01
by GvMariani
GvM builded and tested kernel 3.5.3 nrjX - his feedback
Phenom II X4 810 (@ 2600 MHz)
4 Gb RAM
BFQ + CK patches + BFS > Desktop mode with Tick ​​Frequency: HZ=1.000 Hz
Mdv 2010.3-MIB x86_64

Image Let's give a surge of pure energy to our brand new Mandriva / ROSA Linux!

My experience after about 24 hours of use of a local build the new kernel experimentally:
a little mixed.

At a first glance the system is more agile (eg when loading the DE at the start) and holds up well even in stressing situations
(Eg: when playing a movie smoothly from my HDD, maybe also jumping forward, while I'm extracting in parallel in the background a couple of archives from 600-700 Mb each, or while I'm building rpms in a konsole).

The only thing that seems to be worse is the use of VMware 9.0 : if there is strong activity in the VM's HDD, as, eg in a compilation, of rpm you notice major slowdowns in the rest of the system with other windows that are a bit jerky...
With the 3.5.3-nrj standard seems to me that it happens much less visibly, perhaps VMware Wks is optimized for use with the CFQ.


Image Dai una sprizzata di pura Energia al tuo fiammante Mandriva / ROSA Linux!

La mia esperienza dopo circa 24 ore di uso di un build locale del nuovo kernel sperimentale:
un po mista.

A occhio il sistema è più scattante (p.es. nel caricamento del DE alla partenza) e regge bene anche in situazioni di stress
(p.es.: vedere fluidamente un filmato presente sull'HDD, magari saltando in avanti, mentre in background estraggo in parallelo un paio di archivi da 600-700 Mb l'uno, oppure mentre sto buildando rpms in una konsole).

L'unica cosa che sembra andare peggio è l'utilizzo di VmWare: se nella VM c'è forte attività con l'HDD come, p.es., in una compilazione di un rpm si notano forti rallentamenti nel resto del sistema con le altre finestre che vanno un po 'a scatti... Con il 3.5.3-nrj standard mi sembra che la cosa succeda molto meno visibilmente: forse VmWare Wks è ottimizzata per l'uso con il CFQ.

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 6 September 2012, 21:43
by oiram73
TEST DI 3.5.3 Desktop flavour with Preempt and 10.000 Hz

Hardware:
1) pentium4 3Ghz
2) 2Gb di ram
3) video nvidia con 256 Mb fino a 512 in condivisione con quella del PC
4) audio intel integrato
5) HD esterno USB (quello interno è scassato)
Software:
1) ROSA-RP1
2) jackd
RISULTATI TEST
1) amarok in esecuzione con i suoi 3,5 giorni di musica
2) rosegarden che esegue la V sinfonia di Beethoven = 11 canali midi con fluidsynth
3) kmid che esegue in successione 5 files midi con diversi canali con timidity
4) romp e dragonplayer che proiettano 2 film differenti
5) lilypond in esecuzione
6)12 finestre attive cui accedo con una facilità che su questa configurazione HW non avevo mai visto : in passato ho anzi assistito a numerosi system-freeze con qualunque distribuzione, da Mandriva a Fedora e anche Debian.
7) xruns: il primo è stato segnalato al 34° minuto e solo quano ho dato dal terminale dolphin :"urpmi *.rpm" però non l'ho sentito e questo è quello che conta!
CONCLUSIONI: finalmente posso utilizzare questo HW come si deve!

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 8 September 2012, 10:45
by NicCo
Update ^ to 08 September 2012
Go To the previous msg!


It's only to remind You that today we have available all the updated kernel versions from kernel 3.0 to 3.5

Precisely we are maintaining: 6 versions for 4 different distros: mdv2010, mdv2011, RM2012, ROSA 2012.1
We must thank a lot also to the clever and available ABF developers that make all that possible!


and after this,
I'll focus all my attention in completing the first public nrjX, first release will be version 3.5.3

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 10 September 2012, 0:48
by NicCo
Finally, after a lot of difficulties... :D
new MIB kernel flavours with BFS/BFQ/TOI, testing time is really so close now...

Code: Select all

Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-workstation-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-workstation-devel-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-workstation-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-workstation-devel-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-multimedia-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-multimedia-devel-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-multimedia-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-multimedia-devel-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-studio-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-studio-devel-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-studio-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-studio-devel-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-realtime-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-realtime-devel-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-realtime-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-realtime-devel-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-laptop-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-laptop-devel-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-laptop-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-laptop-devel-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-server-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-server-devel-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-server-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-server-devel-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-gameserver-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-gameserver-devel-3.5.3-69mib-1-1mib2010.2.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-gameserver-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/x86_64/kernel-nrj-gameserver-devel-latest-3.5.3-69.x86_64.rpm
Scritto: /home/utente/rpmbuild/RPMS/noarch/kernel-source-3.5.3-69mib-1-1mib2010.2.noarch.rpm
Scritto: /home/utente/rpmbuild/RPMS/noarch/kernel-source-latest-3.5.3-69.noarch.rpm                                              
Scritto: /home/utente/rpmbuild/RPMS/noarch/kernel-doc-3.5.3-69.noarch.rpm                                                        
Esecuzione(%clean) in corso: /bin/sh -e /var/tmp/rpm-tmp.v8tvs5                                                                  
+ umask 022                                                                                                                      
+ cd /home/utente/rpmbuild/BUILD                                                                                                 
+ cd kernel-x86_64                                                                                                               
+ rm -rf /home/utente/rpmbuild/BUILDROOT/kernel-3.5.3-69.x86_64                                                                  
+ exit 0
After having worked very hard for all this week-end, also being close to give up,
in less time than I previewed I'll need your help in testing new kernel behaviour,
on each flavours have been developed points named 1-5 (CK1/BFS/BFQ/TOI)

actually Iinserted the TuxOnIce feature on all flavours, only to test it better,
at the end this feature should be active probably only in -laptop flavours

I did not change the codename for kernel name, I preferred to maintain nrj
so to distinguish different flavours I simply changed the flavour names

Temporary flavour names list

kernel-nrj-workstation = mainline kernel-nrj-desktop plus the add of +CK1 +BFS +BFQ +TOI

kernel-nrj-multimedia = mainline kernel-nrj-desktop plus the add of +CK1 +BFS +BFQ +TOI
this has a higher HZ frequency, so It should be more reactive for all multimedia apps

kernel-nrj-studio = mainline kernel-nrj-desktop plus the add of +CK1 +BFS +BFQ +TOI
this has the hightest HZ frequency possible, it's the one est tuned for Pro Studio Audio (DAW)
This is the one we are testing (oiram73) with Jackd, Ardour, Rosegarden

kernel-nrj-realtime > mainline kernel-nrj-desktop plus the add of +CK1 +BFS +BFQ +TOI
we are experimenting some timers configs for realtime, it will be maintained only if works

kernel-nrj-laptop > mainline kernel-nrj-desktop plus the add of +CK1 +BFS +BFQ +TOI
with laptop configs and lower internal frequency, ONDEMAND profile to improve battery lasting

kernel-nrj-server > mainline kernel-server plus the add of +CK1 +BFS +BFQ +TOI
to be used in a generic server environment (like LAMP) and other server apps

kernel-nrj-gameserver > mainline kernel-server plus the add of +CK1 +BFS +BFQ +TOI
to be used in a game server PC for 3D FPS games, for example CounterStrike


Then we have all PAE flavours, and for now only one CPU oriented (new Intel) + PAE
kernel-nrj-workstation-pae

these have been optimized for all Intel cpus greater than Pentium4 (starting from Centrino upto I3/I5/I7)
kernel-nrj-workstation-core2
kernel-nrj-work-core2-pae


bye, NicCo

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 11 September 2012, 9:06
by NicCo
Yesterday evening I sent the new Kernel development SRPM to ABF
ABF liked the new SRPM, and so, at the first attempt, we have rpms

New kernel flavours are available for MDV 2011, RM 2012 and ROSA 2012.1
about 2010, we'll start soon the local build, we hope to have builded rpms
in the next two days...


the folder with devel experimental kernels to test is:
http://abf.rosalinux.ru/downloads/kerne ... _personal/

you may need to install cpu firmware before
Microcode firmwares for cpu AMD / INTEL updates



here, you can find the description of different flavours
viewtopic.php?p=19434&f=2#p19434


After kernel 3.5.3, will be refined, I wish adding other things inside
(aufs, overlayfs, and many others, I'll go on with previous kernels,
I'll prepare the same for other kernel versions: 3.4, 3.3, 3.2, 3.1, 3.0


now, I'm waiting for Your precious testing and feedback


bye, NicCo

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 11 September 2012, 9:48
by rolf
Good news about ABF :)

There was some problem with, at least, the method I follow:

Code: Select all

[..]
getting exclusive lock on rpm


installing kernel-nrj-realtime-devel-3.5.3-69mib-1-1-mib2011.0.x86_64.rpm kernel-nrj-realtime-3.5.3-69mib-1-1-mib2011.0.x86_64.rpm from /var/cache/urpmi/partial
starting installing packages
created transaction for installing on / (remove=0, install=2, upgrade=0)
Preparing...                     ###########################################################################################
      1/2: kernel-nrj-realtime-3.5.3-69mib
                                 ###########################################################################################
INTERNAL ERROR: hd5 has no mapping in device.map (when translating (hd5))
MDK::Common::Various::internal_error() called from /usr/lib/libDrakX/bootloader.pm:1623
bootloader::grub2dev_and_file() called from /usr/lib/libDrakX/bootloader.pm:1632
bootloader::grub2dev() called from /usr/lib/libDrakX/bootloader.pm:224
bootloader::read_grub() called from /usr/lib/libDrakX/bootloader.pm:174
bootloader::read() called from /usr/sbin/bootloader-config:64
      2/2: kernel-nrj-realtime-devel-3.5.3-69mib
                                 ###########################################################################################

nvidia-current (295.40-2): Installing module.
.........(bad exit status: 10)
  Build failed.  Installation skipped.
vboxhost (4.1.22): Installing module.
...............
......
You should restart your computer for kernel-nrj-realtime-3.5.3-69mib
unlocking urpmi database
unlocking rpm database
[root@localhost rolf]#
initrd and grub entry for this kernel were not produced.
There are only the five hard drives attached, atm.

Code: Select all

# cat /boot/grub/device.map
(fd0) /dev/fd0
(hd0) /dev/sda
(hd1) /dev/sdb
(hd2) /dev/sdc
(hd3) /dev/sdd
(hd4) /dev/sde
Thanks for the work!

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 11 September 2012, 15:54
by NicCo
rolf wrote: There was some problem with, at least, the method I follow:

..........................................
Thanks for the work!
A question:
In which Distro did you try? Mandriva 2011 (I read that from your signature) or RM 2012?
We know about some problems often happens when you install a kernel on a mdv 2011
(I think that the "dracut" version in mdv2011 is rather crap...)
nicco wrote:kernel-nrj-realtime > mainline kernel-nrj-desktop plus the add of +CK1 +BFS +BFQ +TOI
we are experimenting some timers configs for realtime, it will be maintained only if works
About flavour:
Please test also other flavours, each has very different internal configs
The flavour closer to "kernel-nrj-desktop" is new "kernel-nrj-workstation", then follows "kernel-nrj-multimedia" with a higher scheduler freq (5000, instead than 1000)

bye, NicCo

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 11 September 2012, 18:18
by rolf
Hi,
NicCo wrote:A question:
In which Distro did you try? Mandriva 2011 (I read that from your signature) or RM 2012?
We know about some problems often happens when you install a kernel on a mdv 2011
(I think that the "dracut" version in mdv2011 is rather crap...)

Yes, it's on 2011 x86_64 that I am working. I have had problems with initrd not generated for each new kernel installation since, probably, the switch to dracut. Accordingly,

Code: Select all

[root@localhost boot]# ll /etc/alternatives/mkinitrd
lrwxrwxrwx 1 root root 23 Jul 21 02:00 /etc/alternatives/mkinitrd -> /sbin/mkinitrd-mkinitrd*
Probably, the installkernel scripts call dracut explicitly and, so far, I can build the initrd 'by hand' using my mkinitrd-mkintrd alternatives configuration.

The dkms failure with nvidia is another matter and I've seen various reports on cooker and mageia ml about problems with nvidia and newer kernels. There is a build log with some clues:

Code: Select all

[root@localhost linux-3.5.3-nrj-realtime-69mib]# cat /var/lib/dkms/nvidia-current/295.40-2/build/make.log 
DKMS make.log for nvidia-current-295.40-2 for kernel 3.5.3-nrj-workstation-69mib (x86_64)
Tue Sep 11 07:58:50 PDT 2012
NVIDIA: calling KBUILD...
/usr/src/linux-3.5.3-nrj-workstation-69mib/arch/x86/Makefile:96: CONFIG_X86_X32 enabled but no binutils support
test -e include/generated/autoconf.h -a -e include/config/auto.conf || (                \
echo;                                                           \
echo "  ERROR: Kernel configuration is invalid.";               \
echo "         include/generated/autoconf.h or include/config/auto.conf are missing.";\
echo "         Run 'make oldconfig && make prepare' on kernel src to fix it.";  \
echo;                                                           \
/bin/false)
mkdir -p /var/lib/dkms/nvidia-current/295.40-2/build/.tmp_versions ; rm -f /var/lib/dkms/nvidia-current/295.40-2/build/.tmp_versions/*
make -f scripts/Makefile.build obj=/var/lib/dkms/nvidia-current/295.40-2/build
  cc -Wp,-MD,/var/lib/dkms/nvidia-current/295.40-2/build/.nv.o.d  -nostdinc -isystem /usr/lib64/gcc/x86_64-mandriva-linux-gnu/4.6.1/include -I/usr/src/linux-3.5.3-nrj-workstation-69mib/arch/x86/include -Iarch/x86/include/generated -Iinclude  -include /usr/src/linux-3.5.3-nrj-workstation-69mib/include/linux/kconfig.h -D__KERNEL__ -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -Werror-implicit-function-declaration -Wno-format-security -fno-delete-null-pointer-checks -O2 -m64 -mtune=generic -mno-red-zone -mcmodel=kernel -funit-at-a-time -maccumulate-outgoing-args -fstack-protector -DCONFIG_AS_CFI=1 -DCONFIG_AS_CFI_SIGNAL_FRAME=1 -DCONFIG_AS_CFI_SECTIONS=1 -DCONFIG_AS_FXSAVEQ=1 -DCONFIG_AS_AVX=1 -pipe -Wno-sign-compare -fno-asynchronous-unwind-tables -mno-sse -mno-mmx -mno-sse2 -mno-3dnow -mno-avx -Wframe-larger-than=2048 -Wno-unused-but-set-variable -fno-omit-frame-pointer -fno-optimize-sibling-calls -pg -Wdeclaration-after-statement -Wno-pointer-sign -fno-strict-overflow -fconserve-stack -DCC_HAVE_ASM_GOTO -I/var/lib/dkms/nvidia-current/295.40-2/build -Wall -MD -Wsign-compare -Wno-cast-qual -Wno-error -D__KERNEL__ -DMODULE -DNVRM -DNV_VERSION_STRING=\"295.40\" -Wno-unused-function -Wuninitialized -mno-red-zone -mcmodel=kernel -UDEBUG -U_DEBUG -DNDEBUG  -DMODULE  -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(nv)"  -D"KBUILD_MODNAME=KBUILD_STR(nvidia)" -c -o /var/lib/dkms/nvidia-current/295.40-2/build/.tmp_nv.o /var/lib/dkms/nvidia-current/295.40-2/build/nv.c
In file included from include/linux/kernel.h:19:0,
                 from include/linux/sched.h:62,
                 from include/linux/utsname.h:35,
                 from /var/lib/dkms/nvidia-current/295.40-2/build/nv-linux.h:38,
                 from /var/lib/dkms/nvidia-current/295.40-2/build/nv.c:13:
include/linux/bitops.h: In function ‘hweight_long’:
include/linux/bitops.h:66:41: warning: signed and unsigned type in conditional expression [-Wsign-compare]
In file included from /usr/src/linux-3.5.3-nrj-workstation-69mib/arch/x86/include/asm/uaccess.h:586:0,
                 from include/linux/poll.h:14,
                 from /var/lib/dkms/nvidia-current/295.40-2/build/nv-linux.h:97,
                 from /var/lib/dkms/nvidia-current/295.40-2/build/nv.c:13:
/usr/src/linux-3.5.3-nrj-workstation-69mib/arch/x86/include/asm/uaccess_64.h: In function ‘copy_from_user’:
/usr/src/linux-3.5.3-nrj-workstation-69mib/arch/x86/include/asm/uaccess_64.h:53:6: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
In file included from /var/lib/dkms/nvidia-current/295.40-2/build/nv.c:13:0:
/var/lib/dkms/nvidia-current/295.40-2/build/nv-linux.h: At top level:
/var/lib/dkms/nvidia-current/295.40-2/build/nv-linux.h:114:75: fatal error: asm/system.h: No such file or directory
compilation terminated.
make[3]: *** [/var/lib/dkms/nvidia-current/295.40-2/build/nv.o] Error 1
make[2]: *** [_module_/var/lib/dkms/nvidia-current/295.40-2/build] Error 2
NVIDIA: left KBUILD.
nvidia.ko failed to build!
make[1]: *** [module] Error 1
make: *** [module] Error 2
Looking at the error about the kernel config file, I see in the dkms manual:

Code: Select all

--config <kernel-.config-location>
              During a build this option is used to specify an alternate location for  the  kernel  .config  file
              which  was used to compile that kernel.  Normally, dkms uses the Red Hat standard location and con‐
              fig filenames located in /usr/src/linux-<kernel>/configs/.  If the config for the kernel  that  you
              are  building a module for is not located here or does not have the expected name in this location,
              you will need to tell dkms where the necessary .config can be found so  that  your  kernel  can  be
              properly prepared for the module build.
But there is no /usr/src/linux-<kernel>/configs/ directory for this or any other kernel-devel on my machine. The .config is in /usr/src/linux-<kernel>/ as, I think, for all the time I've used this distro (> 12 yrs).

Code: Select all

[root@localhost linux-3.5.3-nrj-realtime-69mib]# ll .config 
-rw-r--r-- 1 root root 132716 Sep 10 14:49 .config
I did "Run 'make oldconfig && make prepare' on kernel src to fix it."

Code: Select all

[root@localhost linux-3.5.3-nrj-realtime-69mib]# make oldconfig && make prepare
  HOSTCC  scripts/basic/fixdep
  HOSTCC  scripts/kconfig/conf.o
  HOSTCC  scripts/kconfig/zconf.tab.o
  HOSTLD  scripts/kconfig/conf
/usr/src/linux-3.5.3-nrj-realtime-69mib/arch/x86/Makefile:96: CONFIG_X86_X32 enabled but no binutils support
make: Nothing to be done for `prepare'.
I gave --config /usr/src/linux-3.5.3-nrj-realtime-69mib/ to dkms but it errors, still. The log:

Code: Select all

[root@localhost linux-3.5.3-nrj-realtime-69mib]# cat /var/lib/dkms/nvidia-current/295.40-2/build/make.log DKMS make.log for nvidia-current-295.40-2 for kernel 3.5.3-nrj-realtime-69mib (x86_64)
Tue Sep 11 08:57:07 PDT 2012
NVIDIA: calling KBUILD...
/usr/src/linux-3.5.3-nrj-realtime-69mib/arch/x86/Makefile:96: CONFIG_X86_X32 enabled but no binutils support
test -e include/generated/autoconf.h -a -e include/config/auto.conf || (                \
echo;                                                           \
echo "  ERROR: Kernel configuration is invalid.";               \
echo "         include/generated/autoconf.h or include/config/auto.conf are missing.";\
echo "         Run 'make oldconfig && make prepare' on kernel src to fix it.";  \
echo;                                                           \
/bin/false)
mkdir -p /var/lib/dkms/nvidia-current/295.40-2/build/.tmp_versions ; rm -f /var/lib/dkms/nvidia-current/295.40-2/build/.tmp_versions/*
make -f scripts/Makefile.build obj=/var/lib/dkms/nvidia-current/295.40-2/build
  cc -Wp,-MD,/var/lib/dkms/nvidia-current/295.40-2/build/.nv.o.d  -nostdinc -isystem /usr/lib64/gcc/x86_64-mandriva-linux-gnu/4.6.1/include -I/usr/src/linux-3.5.3-nrj-realtime-69mib/arch/x86/include -Iarch/x86/include/generated -Iinclude  -include /usr/src/linux-3.5.3-nrj-realtime-69mib/include/linux/kconfig.h -D__KERNEL__ -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -Werror-implicit-function-declaration -Wno-format-security -fno-delete-null-pointer-checks -O2 -m64 -mtune=generic -mno-red-zone -mcmodel=kernel -funit-at-a-time -maccumulate-outgoing-args -fstack-protector -DCONFIG_AS_CFI=1 -DCONFIG_AS_CFI_SIGNAL_FRAME=1 -DCONFIG_AS_CFI_SECTIONS=1 -DCONFIG_AS_FXSAVEQ=1 -DCONFIG_AS_AVX=1 -pipe -Wno-sign-compare -fno-asynchronous-unwind-tables -mno-sse -mno-mmx -mno-sse2 -mno-3dnow -mno-avx -Wframe-larger-than=2048 -Wno-unused-but-set-variable -fno-omit-frame-pointer -fno-optimize-sibling-calls -pg -Wdeclaration-after-statement -Wno-pointer-sign -fno-strict-overflow -fconserve-stack -DCC_HAVE_ASM_GOTO -I/var/lib/dkms/nvidia-current/295.40-2/build -Wall -MD -Wsign-compare -Wno-cast-qual -Wno-error -D__KERNEL__ -DMODULE -DNVRM -DNV_VERSION_STRING=\"295.40\" -Wno-unused-function -Wuninitialized -mno-red-zone -mcmodel=kernel -UDEBUG -U_DEBUG -DNDEBUG  -DMODULE  -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(nv)"  -D"KBUILD_MODNAME=KBUILD_STR(nvidia)" -c -o /var/lib/dkms/nvidia-current/295.40-2/build/.tmp_nv.o /var/lib/dkms/nvidia-current/295.40-2/build/nv.c
In file included from include/linux/kernel.h:19:0,
                 from include/linux/sched.h:62,
                 from include/linux/utsname.h:35,
                 from /var/lib/dkms/nvidia-current/295.40-2/build/nv-linux.h:38,
                 from /var/lib/dkms/nvidia-current/295.40-2/build/nv.c:13:
include/linux/bitops.h: In function ‘hweight_long’:
include/linux/bitops.h:66:41: warning: signed and unsigned type in conditional expression [-Wsign-compare]
In file included from /usr/src/linux-3.5.3-nrj-realtime-69mib/arch/x86/include/asm/uaccess.h:586:0,
                 from include/linux/poll.h:14,
                 from /var/lib/dkms/nvidia-current/295.40-2/build/nv-linux.h:97,
                 from /var/lib/dkms/nvidia-current/295.40-2/build/nv.c:13:
/usr/src/linux-3.5.3-nrj-realtime-69mib/arch/x86/include/asm/uaccess_64.h: In function ‘copy_from_user’:
/usr/src/linux-3.5.3-nrj-realtime-69mib/arch/x86/include/asm/uaccess_64.h:53:6: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
In file included from /var/lib/dkms/nvidia-current/295.40-2/build/nv.c:13:0:
/var/lib/dkms/nvidia-current/295.40-2/build/nv-linux.h: At top level:
/var/lib/dkms/nvidia-current/295.40-2/build/nv-linux.h:114:75: fatal error: asm/system.h: No such file or directory
compilation terminated.
make[3]: *** [/var/lib/dkms/nvidia-current/295.40-2/build/nv.o] Error 1
make[2]: *** [_module_/var/lib/dkms/nvidia-current/295.40-2/build] Error 2
NVIDIA: left KBUILD.
nvidia.ko failed to build!
make[1]: *** [module] Error 1
make: *** [module] Error 2
nicco wrote:kernel-nrj-realtime > mainline kernel-nrj-desktop plus the add of +CK1 +BFS +BFQ +TOI
we are experimenting some timers configs for realtime, it will be maintained only if works
nicco wrote:About flavour:
Please test also other flavours, each has very different internal configs
The flavour closer to "kernel-nrj-desktop" is new "kernel-nrj-workstation", then follows "kernel-nrj-multimedia" with a higher scheduler freq (5000, instead than 1000)

bye, NicCo
I did try with kernel-nrj-workstation and the results seem the same: dkms fails, no initrd, no grub entry.

Thanks.
Rolf

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 11 September 2012, 18:30
by NicCo
rolf wrote:
Yes, it's on 2011 x86_64 that I am working. I have had problems with initrd not generated for each new kernel installation since, probably, the switch to dracut. Accordingly,
Installing a new kernel on mdv 2011 is like a bet,
the most times you loose, rarely you can win

Some months ago, In one Pc of a friend where installing a new kernel seemed impossible,
I replaced the 2011 dracut version with the RM2012 dracut, then the problem was solved,
because at the next attempt we succeeded to update the kernel with a working boot

but sincerey I dont know if the problem was really fixed properly, but you could try...


I abandoned my PWP 2011 because I had so many problems with it, and I came back to MDV 2010
updating it with 2010.3 repo to new KDE 4.8

Now I use daily MDV2010 and RM2012, while in my opinion, MDV2011 is the worst distro ever used

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 11 September 2012, 18:34
by rolf
NicCo wrote: Installing a new kernel on mdv 2011 is like a bet,
the most times you loose, rarely you can win

Some months ago, In one Pc of a friend where installing a new kernel seemed impossible,
I replaced the 2011 dracut version with the RM2012 dracut, then the problem was solved,
because at the next attempt we succeeded to update the kernel with a working boot

but sincerlly I dont know if the problem was really fixed properly, but you could try...
I will try that, when I get caught up with my to-do list. ;)

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 11 September 2012, 19:00
by sylvainsjc
Hi,

i wanted to try this kernel from ABF repository but I get this error :

Code: Select all

Sorry, the following packages cannot be selected:

- kernel-nrj-workstation-3.5.3-69rosa.lts-1-1-rosa.lts2012.0.x86_64 (car microcode[>= 0.20120313-1] est non satisfait)

Re: Kernel nrjX(treme?), the Next Generation... a step forwa

Posted: 11 September 2012, 22:18
by NicCo
sylvainsjc wrote:Hi,

i wanted to try this kernel from ABF repository but I get this error :

Code: Select all

Sorry, the following packages cannot be selected:

- kernel-nrj-workstation-3.5.3-69rosa.lts-1-1-rosa.lts2012.0.x86_64 (car microcode[>= 0.20120313-1] est non satisfait)
You could use the Sarch function

You could find the page also using Active Topics, now is in the first page

BTW, the topic is here:
Microcode firmwares for cpu AMD / INTEL updates


bye. NicCo