___DEPRECATED___ 2.6.14-nitro2

[Tiger683]

nitro-sources-2.6.14-r2 "Drinkin in LA"

Disclaimer
This patchset is in no way affiliated with the Gentoo Foundation and so it's
use does not imply any more or less official support from the gentoo developers.
Do NOT, EVER, post bugs regarding a kernel patched to nitro-sources to
_bugs.gentoo.org_. Post your problems in this thread instead.


What is it?
Custom kernel patchset, based upon ck-patchset from Con Kolivas and
many additional features usually requested by users.

For me, best release since 2.6.12-nitro5

What's in it?

Code: Select all

#
# 2.6.14-CK5
#
patch-2.6.14-ck4
hz-extra_values.patch
patch-2.6.14.1-2
ipw2x00-hz-fix.diff
scsi-low_hz-fix.patch

#
# -mm incl. git stuff
#
mm-patches.patch
05-linux-2.6.13-e100-badeeprom.diff
git-agpgart.patch
git-alsa.patch
git-libata-all-1.patch
git-netdev-all-b44-build-fix.patch
git-netdev-all-ieee80211_get_payload-warning-fix.patch
git-netdev-all-ieee80211_tx-fix.patch
git-netdev-all.patch
git-ntfs.patch
git-scsi-misc.patch

#
# Standard stuff
#
100-suspend2-2.2-rc9-for-2.6.14.patch
vesafb-tng-1.0-rc1-r2-2.6.14.patch
fbsplash-0.9.2-r5-2.6.14.patch
06-mppe_mppc-iptables_layer7-2.6.14.diff
10-svgalib_helper-1.9.21.diff

#
# Filesystems
#
07-squashfs_2.2-r2.diff
08-unionfs-20051014-1448.diff
09-shfs-0.35_for_2.6.14-rc4.diff
reiser4-for-2.6.14-1.patch

#
# Misc - performance
#
acpi_dsdt_initrd_initramfs
ide-probe-delay
increase-MAX_MP_BUSSES
increase-firmware-loading-timeout
kmalloc-max
parport-mutex
posix-locks-match
readahead-tune
stack-limit
unmap_vmas-lat

#
# Rest
#
04-daconfig-2.3.0.patch
11-usbhid-readd-kconfig.diff
12-shutup-unneeded-warnings.diff
13-config-nr-tty-devices.diff
14-Nitro-FB-logo.diff
15-fixes
nitro-menu.patch
nitro.diff

IMPORTANT UPDATES AND EDITS:
*** I am strongly convinced that this will be a stable release, although we never know about drivers, vesafb-tng functional again
*** Sparsemem memory model selectable for x86 and up, as it has (hopefully) been fixed for these archs with a patch from -mm
** To be continued...

Howto install

GENTOO:
1)create directory <OVERLAY_DIR>/sys-kernel/nitro-sources (substitute <OVERLAY_DIR> according to overlay path)
2) PROMPT>cd <OVERLAY_DIR>/sys-kernel/nitro-sources
3) PROMPT>wget http://www.stud.uni-karlsruhe.de/~uyavl ... -r2.ebuild
4) PROMPT>ebuild nitro-sources-2.6.14-r2.ebuild digest
5) PROMPT>USE="symlink" emerge nitro-sources
6) PROMPT>cd /usr/src/linux
7) configure, compile and install your kernel
if in doubt about how to do it, consult the documentation in the kernel source directory

ALL OTHER DISTROS:
1) PROMPT>cd /usr/src
2) PROMPT>wget http://www.kernel.org/pub/linux/kernel/ ... 14.tar.bz2
3) PROMPT>wget http://www.stud.uni-karlsruhe.de/~uyavl ... nitro2.bz2
4) PROMPT>tar -xvjf linux-2.6.14.tar.bz2 && rm linux-2.6.14.tar.bz2
5) PROMPT>mv linux-2.6.14 linux-2.6.14-nitro2 && rm linux && ln -s linux-2.6.14-nitro2 linux
6) PROMPT>bunzip2 patch-2.6.14-nitro2.bz2 && cd linux
7) PROMPT>patch -p1 -i ../patch-2.6.14-nitro2
configure, compile and install your kernel
if in doubt about how to do it, consult the documentation in the kernel source directory

NOTE:
for rpm based distros, try doing "make rpm" after menuconfig, although this is untested

And now, even more important part:

LINKS
MANIFEST FILE
DIGEST FILE
EBUILD
PATCH
BROKEN OUT TARBALL
SERIES (DETAILED PATCH LISTING)

SHOWTIME !!!!

cheers

T

[Thetargos]

I'll be testing this, as I've been having a bit of trouble compiling anything but vanilla kenrels on my x86_64 system.

[SchAmane]

Lets give a try this best kernel ever

[Tiger683]

Ok, ladies and gents, i'll get some sleep now, happy testing,
i hope you will be as positively surprised as i am

cheers

T

[Thetargos]

I'm having trouble compiling on an x86_64 machine, here's the error code:

Code: Select all

kernel/sysctl.c:641: error: ‘KERN_INTERACTIVE’ no se declaró aquí (no en una función)
kernel/sysctl.c:649: error: ‘KERN_COMPUTE’ no se declaró aquí (no en una función)
kernel/sysctl.c:657: error: ‘KERN_ISO_CPU’ no se declaró aquí (no en una función)
kernel/sysctl.c:780: error: ‘VM_MAPPED’ no se declaró aquí (no en una función)
kernel/sysctl.c:791: error: ‘VM_HARDMAPLIMIT’ no se declaró aquí (no en una función)
make[1]: *** [kernel/sysctl.o] Error 1
make: *** [kernel] Error 2

NOTE: no se declaró aquí (no en una función) means not declared here (not in a function)

Edit
This was while compiling sysctl.o

[cheater1034]

I will definitely compile this after testing my livecd I just made the iso for with .14-nitro1

*edit*
Nitro gave me a kernel panic on the livecd Something about the CPU, not syncing

I had this with .14-nitro1 on my system too, but didn't bother to investigate, I recompiled to archck and it worked.
I'm sure this nitro2 is impressive, but I'm already building archck for it, because its more stable, less optimized or whatever - If i get the spare time, i will try .14-nitro2 on my system

[gimpel]

hummm, getting tons of framedrops playing video in xine while emerging something...
using latest nvidia driver + kernel agpgart --> maybe git-agpgart is guilty

514 frames in 5.0 seconds = 102.183 FPS
holy mama

on archck4 it's ~1000FPS with that crappy gfx5200 here

[Cinder6]

Compiling now

[Cinder6]

running it now. seems good so far! No problems with glxgears; still getting 10k+ fps. nice work, tiger.

[Thetargos]

Anyone else running on x86_64 64-bit Linux?? I seem unable to build the kernel, I've tried vanilla CK patch with the same result, so I think I'll write to Con and see what does he think about this whole issue... Very consistent, can't get past sysctl.o compilation.

[John5788]

everything works here as usual. glad to have vesafb-tng back. whats the difference between vesafb and vesafb-tng anyways?

[Thetargos]

everything works here as usual. glad to have vesafb-tng back. whats the difference between vesafb and vesafb-tng anyways?

vesa-tng (AFAIK) results in easier configuration than usual vesafb.

[sirdilznik]

everything works here as usual. glad to have vesafb-tng back. whats the difference between vesafb and vesafb-tng anyways?

vesafb-tng uses different "friendlier" notation for the resolution and color depth. Also vesafb-tng supports higher resolutions (AFAIK) and doesn't work on amd64

[Cinder6]

tng supports different (higher) refresh rates as well.

oh, and I don't think that plain vesafb and tng differ on resolutions, but I could be wrong there...

all in all, though, I wish it were supported on amd64! oh well, it doesn't make any difference, really.

[Tiger683]

Thetargos:
would you please post contents of your toolchain?

In this case i suspect it also to b0rked somehow, as all the includes
are there, and NOT DEFINED HERE is not what i would expect from a sane compiler
in that place .....

EDIT: Everybody:
Don't use the weirdo HZ values (other than 100 or 1000), as for everyone with
non-standard hardware config, (x86, IDE hdd, no extra peripherals) stuff _will_ break at some point....

cheers

T

[Tiger683]

I had this with .14-nitro1 on my system too, but didn't bother to investigate, I recompiled to archck and it worked.
I'm sure this nitro2 is impressive, but I'm already building archck for it, because its more stable, less optimized or whatever

Your bad well, certainly not mine

[Thetargos]

Thetargos:
would you please post contents of your toolchain?

In this case i suspect it also to b0rked somehow, as all the includes
are there, and NOT DEFINED HERE is not what i would expect from a sane compiler
in that place .....

For some unexplicable reason, apparently the problem was due to the mm-swap-prefetch patch in CK-4 that in 32-bit i386 was just fine, but prevented compiilation with x86_64 for some reason. Now CK-5 is out, and I'm doing kind of a mix'n'match with the broken out nitro2 and the broken out ck-5... wish me luck!

Edit: Tiger, by the way, it would be helpful if you included a file with patch precedence in the broken-out tarball Would make this work of mix'n'match a lot easier

In any case, thanks, mate! Yours is one hell of a kernel!!!

[Tiger683]

Thetargos:

listen up closely: this is actually ck5 based, as only change from ck4 to ck5 was sync with
2.6.14.2, which if you look closely at my patch list, is already in nitro2, so save your time and enjoy....

[Thetargos]

For some strange reason, when I applied the patches (both nitro2 and CK4) as whole (upon a fresh 2.6.14.2 tree, did not have plain 2.6.14), compilation would always stop at sysctl.o, and the only patch that touched sysctl.c was mm-swap-prefetch. I can't say why, but when I applied the broken-out CK-5 patches, in the order stated in the series file of the tarball, the problem did not show up. I'm in the process of testing applying the rest of the nitro patches and see what the outcome is... I don't know if the VM patches have some precedence in the order or what... However this happened to me for the last 6 hours or so... And have been testing and ruling out as much as I can in these past hours.

[Tiger683]

hmm, ok, so i bet you posted without reading my answer....

apply nitro to a vanilla 2.6.14
nitro is ck5 based, so save your time with these experiments,
also applying it to 2.6.14.2 probably caused your all problems.

there, i guess this should help..

EDIT: k, changed font back to normal

[Thetargos]

I'm at the process of building.

[sobers_2002]

Thetargos:
EDIT: Everybody:
Don't use the weirdo HZ values (other than 100 or 1000), as for everyone with
non-standard hardware config, (x86, IDE hdd, no extra peripherals) stuff _will_ break at some point....
T

what does this exactly do?? i chose 250 and i don't see any visible differences so to say

[poOoch]

Hehe very nice Patchset.
My Software Raid0 performance increased about 20 MB/sec.
With gentoo-sources I have about 71 MB/sec and with this kernel 94.84 MB/sec measured with hdparm.
Wow

Cheers!

[Thetargos]

Thetargos:
EDIT: Everybody:
Don't use the weirdo HZ values (other than 100 or 1000), as for everyone with
non-standard hardware config, (x86, IDE hdd, no extra peripherals) stuff _will_ break at some point....
T

what does this exactly do?? i chose 250 and i don't see any visible differences so to say

That's the default value for a lot of sources. I find it to be quite nice for servers, as it provides some sort of interactivity (if you ran, for instance, a remote X app on the server from your desktop). However for desktop use, 1000 is WAY better... It also depends a bit on your processor speed. I'm still getting the hang of the whole x86_64 thing, but for instnace in my previous system, a Sempron 2400+, if I had set the internal Hz to 1000 some apps (most notably games running through either Wine or Cedega) would stutter a LOT (especially sound and input), but with a value of say 512 the system was at its fullest. When a linux vendor does not know exactly what type of hardware will run their kenrels, they'd better bet for a sure value, and as such Red Hat, SuSE, Mandriva and Fedora all have a value of about 250 or 100 internal Hz, good for stability and compatibility with older systems, lousy for low-lat desktops.

[Tiger683]

Thetargos:
how did it work out?