View previous topic :: View next topic |
Author |
Message |
martin.k Guru
Joined: 28 Nov 2004 Posts: 493 Location: Wylatowo, Polska
|
Posted: Tue Jul 26, 2005 7:57 am Post subject: 2.6.13-rc3-mm1-fx4 "Dangerous Mutation" |
|
|
2.6.13-rc3-mm1-fx4 codename: "Dangerous Mutation"
This patch is based on linux-2.6.13-rc3-mm1
Short info about main features
1) Linux kernel 2.6.13-rc3-mm1
2) Dynamic Tick Timer for Linux - currently removed (until Tony updates it)
3) Different CPU Schedulers
FX4 is based on Peter Williams' PlugSched 5.2.3-for-2.6.13-rc3
Now you can choose a CPU Scheduler at booting via kernel BOOTPARAMS line
and you can choose the default CPU Scheduler in menu.
Adding "cpusched=staircase" makes the kernel boot with staircase scheduler form Con Kolivas (v11.3).
There are available options:
Quote: | "cpusched=nicksched" - nickscheduler from Nick Piggin
"cpusched=staircase" - staircase cpu scheduler from Con Kolivas
"cpusched=ingosched" - ingosched - Ingo Molnar CPU scheduler from vanilla
"cpusched=zaphod" - P.Williams Zaphod CPU scheduler
"cpusched=spa_no_frills" - P.Williams Single Priority Array CPU Scheduler
|
4) Different IO Schedulers + genetics
FX4 uses genetic libs and patches from Jake Moilanen(2.6.11-gl1 - Jake is going to port it to 2.6.12 soon)
So the available IO Schedulers are :
- deadline
- no-op
- genetic anticipatory ( menuconfig selection )
- CFQ
You can choose default IO Scheduler from menu or via kernel BOOTPARAMS
If You add "elevator=anticipatory" the kernel will bot with genetic anticipatory io scheduler
I'm not sure if 2.6.11-gl1 genetic stuff gives some speedup to as in 2.6.13-rc3-mm1 - needs testing.
Some intensive I/O testing
EDIT: genetic-lib for 2.6.12 has just been released... http://jakem.net/blog
5) IDE boot probe delay patch
This patch switches the default value of boot ide probe delay to 10ms.
If it doesn't work for you switch the value via kernel BOOTPARAMS:
ide-delay=xx
where xx stands for the desired ide probe delay (the default value
for vanilla kernels is set to 50ms).
6) Mapped watermark III from Con Kolivas from 2.6.13-rc1-ck1
Allow "watermarking" of the memory to make it less likely to swap.
Quote: |
The mapped watermark is configurable so a server for example might be happy to have a lower mapped percentage.
The default is 66 for a desktop and 0 for a server. If you have user logins then the default of 66 is still the most suitable setting.
A hybrid setup that mostly serves files and only rarely has user logins might be better with a mixed value of 33.
echo 33 > /proc/sys/vm/mapped
|
6) Reiser4
Reiser4 now in -mm tree.
Since Reiser4 IS NOT stable and can be very buggy use it at your own risk
My advice: use reiserfs not reiser4 or some other fs (ext3).
7) vesafb-tng from Micha³ Januszewski
If you like nice console
But Vesafb-tng works only on 32-bit x86 due to the technology it uses (vm86)
more at: http://dev.gentoo.org/~spock/
fbsplash (gensplash) from Micha³ Januszewski
Some cute bootsplash - try it, livecd-2005.1 theme is nice
more at: http://dev.gentoo.org/~spock/
You can enable gensplash support via USE="gensplash" flag during emerge process
9) Modded some Kconfig, so that you can use devfs. (I know we should use udev
but I like devfs way It's still in 2.6.13 tree but it's just not in the Kconfig
Patch-list goes here:
Code: | patch-2.6.13-rc3
2.6.13-rc3-mm1
1g_lowmem1_i386.diff
plugsched-5.2.3-for-2.6.13-rc1.patch
genetic-as-sched-2.6.11-gl1.patch
genetic-io-sched-2.6.11-gl1.patch
genetic-lib-2.6.11-gl1.patch
mapped_watermark3.diff
fbsplash-0.9.2-r4-2.6.13-rc1.patch
vesafb-tng-0.9-rc7-r1-2.6.13-rc1.patch
pdflush-tweaks.patch
and the last - ugly FX-sources boot logo hack :) |
This is a experimental kernel version, I'm not responsible for breaking your system!
!!! ALWAYS keep a second stable and bootable kernel apart in your boot manager !!!
Do NOT report any bugs to gentoo developers if the problems you'd eventually encounter do not happen
with a kernel supported by gentoo!
I'm not a developer!
I've just put together some patches - for me it works fine (with nvidia-7667 and reiser4)
Ebuild and patches
INFO: http://xthost.info/fxsources/fx/fx4/FX4-NOTES.TXT
Ebuild: http://xthost.info/fxsources/fx/fx4/fx-sources-2.6.13-r1.ebuild
Patch1: http://xthost.info/fxsources/fx/fx4/patch-2.6.13-rc3-fx4-no-gensplash.patch.bz2
Patch2: http://xthost.info/fxsources/fx/fx4/patch-2.6.13-rc3-fx4-gensplash.patch.bz2
If it doesn't work for you try previous rel. 2.6.12-fx3
You can find it on Polish thread https://forums.gentoo.org/viewtopic-t-362078.html _________________ linux-2.6.17 +ck +R4 +lockless +genetic-as +... więcej nie pamiętam
LRU #299256
Last edited by martin.k on Thu Aug 04, 2005 7:36 pm; edited 1 time in total |
|
Back to top |
|
|
m0p Apprentice
Joined: 20 Jun 2005 Posts: 205 Location: en_GB
|
Posted: Tue Jul 26, 2005 1:10 pm Post subject: |
|
|
I'll give it a go later. I will try out those different CPU scheds too. I wonder if any of them outperform staircase. BTW, where do you get your patches from? I've been looking to do some patch experiments for a while. |
|
Back to top |
|
|
martin.k Guru
Joined: 28 Nov 2004 Posts: 493 Location: Wylatowo, Polska
|
Posted: Tue Jul 26, 2005 1:14 pm Post subject: |
|
|
m0p wrote: | I'll give it a go later. I will try out those different CPU scheds too. I wonder if any of them outperform staircase. BTW, where do you get your patches from? I've been looking to do some patch experiments for a while. |
PlugSched - from P.Williams http://cpuse.sf.net
genetic stuff - from J.Moilanen http://jakem.net/blog
-mm stuff - www.kernel.org
-dyn-tic stuff - www.muru.com
-ck stuff - from Con Kolivas http://members.optusnet.com.au/ckolivas/kernel/
-vesafb-tng & gensplash - from Spock - http://dev.gentoo.org/~spock/
-www.google.com - for all other stuff and questions
And some (a lot of) spare time to rediff rejects
It's not that hard as you would think, but I'm not a developer so I can't do some
real codding stuff. Just simple collecting patches. _________________ linux-2.6.17 +ck +R4 +lockless +genetic-as +... więcej nie pamiętam
LRU #299256 |
|
Back to top |
|
|
Tiger683 Veteran
Joined: 08 Jan 2005 Posts: 1347 Location: Heffner's House
|
Posted: Tue Jul 26, 2005 2:19 pm Post subject: |
|
|
actually, jake's kernel page might be more interesting than
his blog....
http://kernel.jakem.net _________________ Retired gentoo user |
|
Back to top |
|
|
Stephonovich n00b
Joined: 20 May 2004 Posts: 72 Location: United States
|
Posted: Tue Aug 02, 2005 3:08 am Post subject: |
|
|
Just what I've been looking for. I'm trying the CVS build of Beagle, which breaks inotify in 2.6.12. At the same time, I need reiser4 and vesafb-tng support; so I was looking at manually patching sources. I've been running 2.6.12-morph7 for awhile, but I'll give this a whirl...
EDIT: The ebuild failed for me. bzip2 was complaining that patch-2.6.13-rc3.bz2 was corrupt. Even tried re-downloading it from kernel.org, checked md5sum; nothing. bzip2 -tvv said it was fine. So I manually bunzip'd it, commented out the appropiate unpack line in the ebuild, and made the necessary changes in Manifest. No idea what's going on. If it matters, I'm using gensplash.
Finally, it failed to compile. First, in include/linux/mmzone.h, line 430, an #endif without an #if. Commenting it out fixed that. On to include/linux/pagemap.h, lines 207-208.
In function `wait_on_page_fs_misc':
error: `PG_fs_misc' undeclared (first use in this function)
I'm not much of a programmer, but I looked around some, and couldn't find anything glaringly obvious.
Then, it's 0043. Perhaps time to give this a rest. _________________ Who needs reincarnation when you've got parallel universes? |
|
Back to top |
|
|
martin.k Guru
Joined: 28 Nov 2004 Posts: 493 Location: Wylatowo, Polska
|
Posted: Thu Aug 04, 2005 7:35 pm Post subject: FX-sources - > Isotope-sources |
|
|
Due to possible trademark infringement (etc.) the FX-sources name is changed to isotope-sources.
All the ebuilds and patches are temporally killed until upcoming 2.6.13-rc4-mm1-isotope1 release _________________ linux-2.6.17 +ck +R4 +lockless +genetic-as +... więcej nie pamiętam
LRU #299256 |
|
Back to top |
|
|
|