View previous topic :: View next topic |
Author |
Message |
Waninkoko Guru
Joined: 13 May 2005 Posts: 549
|
Posted: Thu May 17, 2007 1:23 pm Post subject: 2.6.21-kamikaze3: "Hyperspace" |
|
|
· KAMIKAZE-SOURCES ::.
· 2.6.21-kamikaze3: "Hyperspace"
New version of "kamikaze-sources" !!
I made many interesting changes in this version.
Hotfixes:
Links:
What's new:
Quote: | "Completely Fair Scheduler" updated (works better with 3D applications).
New drivers and many drivers updated. |
Changes in this version:
Quote: | "CFS" updated to v13.
Fixed a bug in "CFS" tunables patch.
"suspend2" updated to 2.2.9.17 (bug when compiling kernel without suspend2 is gone).
"linux-phc" updated to 0.2.10.
"adaptive-readahead" patch has been replaced by the new "adaptive+ondemand-readahead" patch.
"gspca" driver for webcams added.
"mac80211" updated to 7.1.2-ht.
"ipw3945" updated to 1.2.1.
|
Patches:
Quote: | patch-2.6.21.1
fix-sis900-oops.patch
# Completely fair scheduler
sched-cfs-v2.6.21-v13.patch
sched-cfs-v2.6.21-v13-boost.patch
sched-cfs-v2.6.21-v13-boost-tunables.patch
# Con Kolivas patchset
ckpatches-2.6.21-5.patch
# gentoo-sources patchset
genpatches-2.6.21.patch
hz-864-kconfig-option.patch
# cpuidle
cpuidle-2.6.21.patch
# Suspend2
suspend2-2.2.9.17-for-2.6.21.patch
suspend2-2.2.9.17-for-2.6.21-add-debug-mode.patch
# Linux-PHC
linux-phc-0.2.10-2.6.21.patch
realtime-lsm-2.6.21.patch
# ACPI patches
acpi-dsdt-initrd-v0.8.4-2.6.21.patch
sony-laptop-rollup-2.6.21.patch
thinkpad-2.6.21.patch
# Filesystems patches
fuse-2.6.5-for-2.6.21.patch
reiser4-2.6.21.patch
ext4-updates-2.6.21-2.patch
udf-2.50-2.6.21.patch
unionfs-2.6.21.patch
# Adaptive readahead
adaptive+ondemand-readahead-2.6.21.patch
# Execshield
execshield-2.6.21.patch
execshield-kconfig-2.6.21.patch
cpu-support-rollup-2.6.21.patch
# Mactel
mactel-patches-2.6.21.patch
# Video4Linux
gspca-20070508-2.6.21.patch
# Network/Wireless
mac80211-7.1.2-ht-2.6.21.patch
ipw3945-1.2.1-2.6.21.patch
# Input devices
xpad-360-support-2.6.21.patch
# Misc patches
enable-4k-stacks-default-2.6.21.patch
kamikaze-version.patch |
Greetings.
Last edited by Waninkoko on Wed May 23, 2007 6:00 pm; edited 5 times in total |
|
Back to top |
|
|
DocterD Tux's lil' helper
Joined: 15 May 2004 Posts: 129
|
Posted: Thu May 17, 2007 2:07 pm Post subject: |
|
|
can't compile.
Quote: | CC fs/compat_ioctl.o
CC fs/nfsctl.o
CC fs/binfmt_script.o
CC fs/binfmt_elf.o
fs/binfmt_elf.c: In function 'load_elf_binary':
fs/binfmt_elf.c:1048: error: too many arguments to function 'arch_setup_additional_pages'
make[1]: *** [fs/binfmt_elf.o] Error 1
make: *** [fs] Error 2 |
|
|
Back to top |
|
|
Waninkoko Guru
Joined: 13 May 2005 Posts: 549
|
Posted: Thu May 17, 2007 2:18 pm Post subject: |
|
|
DocterD wrote: | can't compile.
Quote: | CC fs/compat_ioctl.o
CC fs/nfsctl.o
CC fs/binfmt_script.o
CC fs/binfmt_elf.o
fs/binfmt_elf.c: In function 'load_elf_binary':
fs/binfmt_elf.c:1048: error: too many arguments to function 'arch_setup_additional_pages'
make[1]: *** [fs/binfmt_elf.o] Error 1
make: *** [fs] Error 2 |
|
x86 or x86_64? |
|
Back to top |
|
|
DocterD Tux's lil' helper
Joined: 15 May 2004 Posts: 129
|
Posted: Thu May 17, 2007 2:21 pm Post subject: |
|
|
x86_64 with Dyntick turned on. |
|
Back to top |
|
|
Waninkoko Guru
Joined: 13 May 2005 Posts: 549
|
Posted: Thu May 17, 2007 2:24 pm Post subject: |
|
|
DocterD wrote: | x86_64 with Dyntick turned on. |
Edit: Patch updated. Now kernel should compile. |
|
Back to top |
|
|
lucke n00b
Joined: 11 May 2005 Posts: 35
|
|
Back to top |
|
|
mrcs Tux's lil' helper
Joined: 10 Oct 2003 Posts: 137
|
Posted: Thu May 17, 2007 5:15 pm Post subject: |
|
|
Great stuff! Thanks |
|
Back to top |
|
|
br0ken n00b
Joined: 15 May 2007 Posts: 18 Location: Kiev, Ukraine
|
Posted: Thu May 17, 2007 5:46 pm Post subject: |
|
|
x86_64 (Turion) - system locks up during kernel boot |
|
Back to top |
|
|
Waninkoko Guru
Joined: 13 May 2005 Posts: 549
|
Posted: Thu May 17, 2007 5:59 pm Post subject: |
|
|
br0ken wrote: | x86_64 (Turion) - system locks up during kernel boot |
Looks like dynticks is not working properly. Did you enabled dynticks? Could you try kernel without dynticks?
If dynticks causes many problems I will drop it and update the patch (again). |
|
Back to top |
|
|
vipernicus Veteran
Joined: 17 Jan 2005 Posts: 1462 Location: Your College IT Dept.
|
Posted: Thu May 17, 2007 6:56 pm Post subject: Re: 2.6.21-kamikaze3: "Hyperspace" |
|
|
Waninkoko wrote: | # Completely fair scheduler
sched-cfs-v2.6.21-v12.patch
sched-cfs-v2.6.21-v12-boost.patch
sched-cfs-v2.6.21-v12-boost-tunables.patch |
Old news! Where is v13! _________________ Viper-Sources Maintainer || nesl247 Projects || vipernicus.org blog |
|
Back to top |
|
|
Waninkoko Guru
Joined: 13 May 2005 Posts: 549
|
Posted: Thu May 17, 2007 7:32 pm Post subject: |
|
|
Patch updated. Now using CFS v13 and dynticks patch dropped. |
|
Back to top |
|
|
kernelOfTruth Watchman
Joined: 20 Dec 2005 Posts: 6111 Location: Vienna, Austria; Germany; hello world :)
|
|
Back to top |
|
|
Waninkoko Guru
Joined: 13 May 2005 Posts: 549
|
Posted: Thu May 17, 2007 7:41 pm Post subject: |
|
|
Quote: | Changes since -v12:
- small tweak: made the "fork flow" of reniced tasks zero-sum
- debugging update: /proc/<PID>/sched is now seqfile based and echoing
0 to it clears the maximum-tracking counters.
- more debugging counters
- small rounding fix to make the statistical average of rounding errors
zero
- scale both the runtime limit and the granularity on SMP too, and make
it dependent on HZ
- misc cleanups |
|
|
Back to top |
|
|
vipernicus Veteran
Joined: 17 Jan 2005 Posts: 1462 Location: Your College IT Dept.
|
|
Back to top |
|
|
Vegan Tux's lil' helper
Joined: 29 Mar 2007 Posts: 133
|
Posted: Fri May 18, 2007 2:31 am Post subject: |
|
|
I've decided to switch form SD to CFS , which value for sched_granularity_ns is the best for desktop ? _________________ Go Vegan! For The Sake of Animals
//Gentoo 2007.0/desktop, portage- 2.1.3_rc8, kernel @ 2.6.22-kamikaze1 (CFSv19) i686, "/" REISER4, ~x86, gcc-4.1.2, glibc-2.5-r3
//P4 2.4Ghz, 768 RAM , ATI Radeon X1600pro AGP, P4P800, SB Live ! 5.1, WDC 80GB |
|
Back to top |
|
|
br0ken n00b
Joined: 15 May 2007 Posts: 18 Location: Kiev, Ukraine
|
Posted: Fri May 18, 2007 6:06 am Post subject: |
|
|
report: after dropping dynticks patch things seems like work on x86_64 Finally, after few updates |
|
Back to top |
|
|
tranquilcool Veteran
Joined: 25 Mar 2005 Posts: 1179
|
Posted: Fri May 18, 2007 8:16 am Post subject: |
|
|
on my box SD performs better than CFS.
under load in CFS emerge takes ages.
only TV (tvtime) runs smoother in CFS under load. _________________ this is a strange strange world. |
|
Back to top |
|
|
Waninkoko Guru
Joined: 13 May 2005 Posts: 549
|
Posted: Fri May 18, 2007 11:34 am Post subject: |
|
|
vipernicus wrote: | Added to custom-kernels overlay. |
Thanks vipernicus |
|
Back to top |
|
|
ekerazha n00b
Joined: 19 May 2007 Posts: 4
|
Posted: Sat May 19, 2007 12:42 pm Post subject: |
|
|
Hello
Is there a good reason because you use ExecShield instead of PaX? |
|
Back to top |
|
|
Waninkoko Guru
Joined: 13 May 2005 Posts: 549
|
Posted: Sat May 19, 2007 1:00 pm Post subject: |
|
|
ekerazha wrote: | Hello
Is there a good reason because you use ExecShield instead of PaX? |
No
Is PaX better than ExecShield? |
|
Back to top |
|
|
ekerazha n00b
Joined: 19 May 2007 Posts: 4
|
Posted: Sat May 19, 2007 1:15 pm Post subject: |
|
|
Waninkoko wrote: | ekerazha wrote: | Hello
Is there a good reason because you use ExecShield instead of PaX? |
No
Is PaX better than ExecShield? |
Good question I think there are pros and cons... maybe PaX is a bit more "secure", but I think it is harder to manage because it seems like it is "enabled by default" for everything and you can disable it for single packages using "paxctl", while ExecShield seems more flexible:
Ingo Molnar wrote: |
There is a new boot-time kernel command line option called exec-shield=,
which has 4 values. Each value represents a different level of security:
exec-shield=0 - always-disabled
exec-shield=1 - default disabled, except binaries that enable it
exec-shield=2 - default enabled, except binaries that disable it
exec-shield=3 - always-enabled
|
PaX also has some limits about memory usage.
So... as I've already said, maybe Pax is a bit more secure and ExeShield is a bit more flexible and easier to manage... but I'm very undecided. |
|
Back to top |
|
|
Waninkoko Guru
Joined: 13 May 2005 Posts: 549
|
Posted: Sat May 19, 2007 1:28 pm Post subject: |
|
|
suspend2 2.2.9.17 to 2.2.10 patch uploaded. |
|
Back to top |
|
|
kernelOfTruth Watchman
Joined: 20 Dec 2005 Posts: 6111 Location: Vienna, Austria; Germany; hello world :)
|
Posted: Sat May 19, 2007 7:11 pm Post subject: |
|
|
ekerazha wrote: |
Good question I think there are pros and cons... maybe PaX is a bit more "secure", but I think it is harder to manage because it seems like it is "enabled by default" for everything and you can disable it for single packages using "paxctl", while ExecShield seems more flexible: |
that's not true, pax also supports "softmode", in which it's disabled by default, it's more secure yes, but this security has a price:
it can't be used in every environment:
ever tried it with SEGMEXEC, PAGEEXEC + mprotect in conjunction with ati-drivers or nvidia-drivers, then you know that it's a no-go, and if you disable those parts, you work less secure than exec-shield
(you can check this with paxtest )
ekerazha wrote: |
PaX also has some limits about memory usage.
So... as I've already said, maybe Pax is a bit more secure and ExeShield is a bit more flexible and easier to manage... but I'm very undecided. |
yeah, what a pity that it can't be used with binary graphics drivers, this would make some linux desktop even more secure
furthermore wine doesn't work with SEGMEXEC as far as I know ... _________________ https://github.com/kernelOfTruth/ZFS-for-SystemRescueCD/tree/ZFS-for-SysRescCD-4.9.0
https://github.com/kernelOfTruth/pulseaudio-equalizer-ladspa
Hardcore Gentoo Linux user since 2004 |
|
Back to top |
|
|
thubble Tux's lil' helper
Joined: 16 May 2003 Posts: 119 Location: Winnipeg, Canada
|
Posted: Sun May 20, 2007 3:08 am Post subject: |
|
|
Doesn't work on my machine. I get to this point:
Code: | VFS: Mounted root (xfs filesystem) readonly. |
and it just sits there. I can plug/unplug my ipod and the kernel hotplug works (I get console messages saying it recognizes a new device) so it's not hung or crashed. It's just not starting the INIT process.
I'm not using Dynticks or anything else that i wasn't using in 2.6.19. |
|
Back to top |
|
|
ekerazha n00b
Joined: 19 May 2007 Posts: 4
|
Posted: Sun May 20, 2007 9:44 am Post subject: |
|
|
kernelOfTruth wrote: |
that's not true, pax also supports "softmode", in which it's disabled by default, it's more secure yes, but this security has a price:
|
Glad to hear this.
kernelOfTruth wrote: |
ever tried it with SEGMEXEC, PAGEEXEC + mprotect in conjunction with ati-drivers or nvidia-drivers, then you know that it's a no-go, and if you disable those parts, you work less secure than exec-shield
(you can check this with paxtest )
yeah, what a pity that it can't be used with binary graphics drivers, this would make some linux desktop even more secure
furthermore wine doesn't work with SEGMEXEC as far as I know ... |
This is the reason because there's "paxctl". |
|
Back to top |
|
|
|