2.6.22-kamikaze4: "Not very experimental!"

Message

dmead · Post by **dmead** » Sat Jul 28, 2007 11:03 pm

any gamers try this out?

eiffel · Post by **eiffel** » Sun Jul 29, 2007 9:14 am

ati-drivers works with this kernel??

mbar · Post by **mbar** » Sun Jul 29, 2007 10:22 am

yes eiffel, but you have to copy ioctl32.h file from vanilla or .21 kernel

p4ddY · Post by **p4ddY** » Sun Jul 29, 2007 7:02 pm

Nice

JDGBOLT · Post by **JDGBOLT** » Sun Jul 29, 2007 9:47 pm

Very good patchset, using it on my laptop, only had a slight issue with the generic ide driver giving me 2M/s throughput, but once I disabled it back to 35M/s. Using it on what I kind of call a livecd running on the hard drive, using squashfs images for everything layered together through aufs, and then booted. Works pretty well, one thing that I didn't expect is hibernate fully works on this monstrosity. Definitely this has been one of the best gentoo reinstalls that I have done. Good job on the kernel.

JDGBOLT

Cadynum · Post by **Cadynum** » Mon Jul 30, 2007 10:30 pm

Very nice, works great for me this far.

kernelOfTruth · Post by **kernelOfTruth** » Mon Jul 30, 2007 11:02 pm

p4ddY wrote:Today someone asked my how to get ndiswrapper running with this awesome kernel.. But it doesn't work because 4k stacks are enabled by default and you can't disable it.
Both the menu option and the option in kernel config are simply missing.
I think enable-4k-stacks-default-2.6.22.patch is responsible for that behaviour and it would be cool if you could remove that patch in kamikaze5.

/edit: This patch does the same as yours but doesn't remove the menu option
Code: Select all
--- linux-2.6.20.orig/arch/i386/Kconfig.debug
+++ linux-2.6.20/arch/i386/Kconfig.debug
@@ -59,12 +59,14 @@
 config 4KSTACKS
        bool "Use 4Kb for kernel stacks instead of 8Kb"
        depends on DEBUG_KERNEL
+       default y
        help
          If you say Y here the kernel will use a 4Kb stacksize for the
          kernel stack attached to each process/thread. This facilitates
          running more threads on a system and also reduces the pressure
          on the VM subsystem for higher order allocations. This option
-         will also use IRQ stacks to compensate for the reduced stackspace.
+         will also use separate 4Kb IRQ stacks to compensate for the
+         reduced stackspace.
 
 config X86_FIND_SMP_CONFIG
        bool
Thanks,
Patrick

so just reverse waninkoko's patch:

Code: Select all

patch -p1 -R < ../4k-stacks_no-ption

(?)

then replace with yours

Code: Select all

patch -p1 < ../4k-stacks_with-option

LoSeR_5150 · Post by **LoSeR_5150** » Mon Jul 30, 2007 11:23 pm

p4ddy, I am using ndiswrapper on my x86 box with this kernel and i have no trouble at all. Ndiswrapper just prints warnings, they are not errors and I have not had any connection troubles thus far.

star.dancer · Post by **star.dancer** » Tue Jul 31, 2007 5:27 am

Hey, thanks so much for this! This is my first Kamikaze kernel and I am very impressed.

I do have a question though, is there a setting to enable CFS in the kernel config? I only have no-op, CFQ, and deadline that come up as scheduler options, although I do get the CFS tunable configs. I have searched all over and heard nothing about needing to config CFS so I assume the patches make the kernel always use CFS. Is this the case? For example, what happens if I choose "deadline" as the default scheduler in the Kamikaze sources?

Ati-drivers: there is a problem with ati-drivers and the 2.6.22 kernel which has nothing to do with the Kamikaze patches. Go here for the patch.

shaumux · Post by **shaumux** » Tue Jul 31, 2007 6:09 am

Any chance of getting this in the patchset
http://linuxcompressed.sourceforge.net/

darklegion · Post by **darklegion** » Tue Jul 31, 2007 8:02 am

star.dancer wrote:Hey, thanks so much for this! This is my first Kamikaze kernel and I am very impressed.

I do have a question though, is there a setting to enable CFS in the kernel config? I only have no-op, CFQ, and deadline that come up as scheduler options, although I do get the CFS tunable configs. I have searched all over and heard nothing about needing to config CFS so I assume the patches make the kernel always use CFS. Is this the case? For example, what happens if I choose "deadline" as the default scheduler in the Kamikaze sources?

Ati-drivers: there is a problem with ati-drivers and the 2.6.22 kernel which has nothing to do with the Kamikaze patches. Go here for the patch.

No-op, CFQ and deadline are "i/o" schedulers.CFS is a "cpu" scheduler and it is always enabled with this patchset.

Waninkoko · Post by **Waninkoko** » Tue Jul 31, 2007 10:52 am

shaumux wrote:Any chance of getting this in the patchset
http://linuxcompressed.sourceforge.net/

It caused some problems so I dropped it.

star.dancer · Post by **star.dancer** » Tue Jul 31, 2007 2:54 pm

darklegion wrote: No-op, CFQ and deadline are "i/o" schedulers.CFS is a "cpu" scheduler and it is always enabled with this patchset.

Oh, oops. Thanks, that makes sense. :)

2.6.22-kamikaze4: "Not very experimental!"

Enabling CFS

Re: Enabling CFS

Re: Enabling CFS