Official thread: "zen-sources" - Part III

Message

soundcheck · Post by **soundcheck** » Tue Mar 18, 2008 1:11 pm

Waninkoko wrote:I'll check that piece of code later. Maybe we removed some changes by mistake.

Hi Waninkoko.

Did you have a chance to look at the scheduler parameter issue?

THX KLS

creidiki · Post by **creidiki** » Tue Mar 18, 2008 1:37 pm

unK wrote:Are you using reiser4?

Yup - /var/tmp/paludis is a reiser4 partition.

gimpel · Post by **gimpel** » Tue Mar 18, 2008 1:45 pm

creidiki wrote:
unK wrote:Are you using reiser4?
Yup - /var/tmp/paludis is a reiser4 partition.

http://bugs.gentoo.org/show_bug.cgi?id=200556

dodo1122 · Post by **dodo1122** » Tue Mar 18, 2008 2:19 pm

SlashBeast wrote:Witam,

Aufs is broken in latest master-devel. But i can't now paste what error I got, sorry.

Yes, that's because it's incompatible with .25. When i get back home i'll check if there is a newer version which is compatible, and merge it. Also, squashfs does not build since .25-rcX. I'll fix that today.

dodo

kernelOfTruth · Post by **kernelOfTruth** » Tue Mar 18, 2008 2:49 pm

dodo1122 wrote:
SlashBeast wrote:Witam,

Aufs is broken in latest master-devel. But i can't now paste what error I got, sorry.
Yes, that's because it's incompatible with .25. When i get back home i'll check if there is a newer version which is compatible, and merge it. Also, squashfs does not build since .25-rcX. I'll fix that today.

dodo

Hi Witam,

is lockless pagecache still included in zen-sources ?

thanks

gimpel · Post by **gimpel** » Tue Mar 18, 2008 3:28 pm

kernelOfTruth wrote:Hi Witam,

is lockless pagecache still included in zen-sources ?

thanks

Looking at "git log /include/linux/radix-tree.h" both lockless and concurrent pagecache is in there.

dodo1122 · Post by **dodo1122** » Tue Mar 18, 2008 7:26 pm

v2.6.25-rc6-zen0 tagged in master-devel. squashfs fixed there as well. aufs is still broken upstream.

dodo

Phlogiston · Post by **Phlogiston** » Tue Mar 18, 2008 8:29 pm

dodo1122 wrote:v2.6.25-rc6-zen0 tagged in master-devel. squashfs fixed there as well. aufs is still broken upstream.

dodo

Thanks what about unionfs?

kernelOfTruth · Post by **kernelOfTruth** » Thu Mar 20, 2008 12:23 pm

Kernel: arch/x86/boot/bzImage is ready (#1)
Building modules, stage 2.
MODPOST 465 modules
ERROR: "btrfs_xattr_acl_access_handler" [fs/btrfs/btrfs.ko] undefined!
ERROR: "btrfs_xattr_acl_default_handler" [fs/btrfs/btrfs.ko] undefined!
make[1]: *** [__modpost] Error 1
make: *** [modules] Error 2

guruvan · Post by **guruvan** » Thu Mar 20, 2008 4:59 pm

I'm having some trouble with the hda-intel driver on one of my machines.

I have an HPxw6200 (2xXeon,hda-intel) and a Toshiba A135-4656 laptop (CeleronM with hda-intel)

I don't know what is different about the two intel sound chipsets, but on the HP the master/master_mono/headphone-sense all work correctly

On the Toshiba, the headphone jack doesn't work at all. (nor does the exterior volume wheel, but that's not important).

Any thoughts on this? (I'm on the zen3 sources, I'll upgrade in a few days & see if that helps)

(either way - this is the bestest-fastest-SMOOTHEST my boxen have run!!! Thanks Guys!!!)
Zen and the Art of Module Maintenance

p.s. just checked - and my gentooVM (vmware) has not lost any appreciable amount of time (1 or 2 seconds at most) in a few days of runtime...

Nsane457 · Post by **Nsane457** » Sun Mar 23, 2008 6:55 am

Is there a way to generate a set of broken-out patches from the git tree?

If not, is there at least a way to get an up-to-date list of patches applied to the kernel tree?

Nsane457 · Post by **Nsane457** » Sun Mar 23, 2008 7:45 am

When attempting to compile cmake-2.4.8 using the latest git zen-sources and gcc-4.3.0, I got:

Code: Select all

In file included from /usr/lib/gcc/i686-pc-linux-gnu/4.3.0/include/g++-v4/bits/basic_ios.h:44,
                 from /usr/lib/gcc/i686-pc-linux-gnu/4.3.0/include/g++-v4/ios:50,
                 from /usr/lib/gcc/i686-pc-linux-gnu/4.3.0/include/g++-v4/istream:45,
                 from /usr/lib/gcc/i686-pc-linux-gnu/4.3.0/include/g++-v4/fstream:45,
                 from /var/tmp/portage/dev-util/cmake-2.4.8/work/cmake-2.4.8/Source/cmStandardIncludes.h:91,
                 from /var/tmp/portage/dev-util/cmake-2.4.8/work/cmake-2.4.8/Source/cmInstallGenerator.h:20,
                 from /var/tmp/portage/dev-util/cmake-2.4.8/work/cmake-2.4.8/Source/cmInstallTargetGenerator.h:20,
                 from /var/tmp/portage/dev-util/cmake-2.4.8/work/cmake-2.4.8/Source/cmInstallTargetGenerator.cxx:17:
/usr/lib/gcc/i686-pc-linux-gnu/4.3.0/include/g++-v4/bits/locale_facets.h:353: internal compiler error: Segmentation fault

Even using -O1 & -march=nocona flags without LDFLAGS or CXXFLAGS same problem.

But when using SystemRescueCD & chrooting, it compiles fine.

I never knew that the kernel environment could lead to a compiler error.

MageSlayer · Post by **MageSlayer** » Sun Mar 23, 2008 5:19 pm

Maybe offtopic, but anyone knows how to apply rt-patchset?
I tried to patch zen-sources-2.6.24-zen2 using http://www.kernel.org/pub/linux/kernel/ ... .2-rt2.bz2
but too many hunk failed. Is there any version coincidence between official and zen sources?

Thanks.

Thargor · Post by **Thargor** » Sun Mar 23, 2008 6:06 pm

Applying -rt on your own is very likely to break if you don't know about kernel internals as it breaks nearly every other patch.
But there is a -rt version of zen-sources.
If you manage you kernel manually and not through portage, simply do this in your zen-sources dir:

Code: Select all

git checkout -b master-devel-rt origin/master-devel-rt

If you are currently using portage, you first have to clone zen-sources manually using this:

Code: Select all

cd /usr/src
git clone git://repo.or.cz/linux-2.6/zen-sources.git

For further questions ask b33fc0d3, it's mainly his work, but the others in #zen-sources on irc.freenode.net can help you too, for sure.

kfiaciarka · Post by **kfiaciarka** » Sun Mar 23, 2008 7:14 pm

kernelOfTruth wrote:
Kernel: arch/x86/boot/bzImage is ready (#1)
Building modules, stage 2.
MODPOST 465 modules
ERROR: "btrfs_xattr_acl_access_handler" [fs/btrfs/btrfs.ko] undefined!
ERROR: "btrfs_xattr_acl_default_handler" [fs/btrfs/btrfs.ko] undefined!
make[1]: *** [__modpost] Error 1
make: *** [modules] Error 2

hi, maybe try not select b-tree fs support ?

BTW - I can't run make clean or make mrproper inside zen-sources directory :/

Code: Select all

make mrproper 
  CLEAN   arch/x86/boot/compressed
  CLEAN   arch/x86/boot
  CLEAN   /usr/src/linux-2.6.24-zen4
  CLEAN   arch/x86/kernel
  CLEAN   drivers/char
  CLEAN   drivers/net/wireless/madwifi/ath_hal
  CLEAN   drivers/video/logo
/usr/src/linux-2.6.24-zen4/fs/hostfs/Makefile:11: arch/um/scripts/Makefile.rules: Nie ma takiego pliku ani katalogu
make[2]: *** Brak reguł do wykonania obiektu `arch/um/scripts/Makefile.rules'. Stop.
make[1]: *** [fs/hostfs] Błąd 2
make: *** [_clean_fs] Błąd 2

MageSlayer · Post by **MageSlayer** » Sun Mar 23, 2008 7:50 pm

Thargor

Not without some git manual reading, but it seems that I managed to check out master-devel-rt!
Thanks

mbar · Post by **mbar** » Sun Mar 23, 2008 8:00 pm

@kfiaciarka:
You have to add "um" USE flag and reemerge zen-sources.

dodo1122 · Post by **dodo1122** » Sun Mar 23, 2008 8:03 pm

kernelOfTruth wrote:
Kernel: arch/x86/boot/bzImage is ready (#1)
Building modules, stage 2.
MODPOST 465 modules
ERROR: "btrfs_xattr_acl_access_handler" [fs/btrfs/btrfs.ko] undefined!
ERROR: "btrfs_xattr_acl_default_handler" [fs/btrfs/btrfs.ko] undefined!
make[1]: *** [__modpost] Error 1
make: *** [modules] Error 2

unselect btrfs posix acl support. it's broken atm.

dodo

kfiaciarka · Post by **kfiaciarka** » Mon Mar 24, 2008 2:08 pm

mbar wrote:@kfiaciarka:
You have to add "um" USE flag and reemerge zen-sources.

thx

mbar · Post by **mbar** » Tue Mar 25, 2008 9:11 am

Is there any chance for inclusion of Ramback patch?
http://lwn.net/Articles/272534/

gimpel · Post by **gimpel** » Tue Mar 25, 2008 9:27 pm

Please apply following Reiser4 patches:

ccreg40 diskfull conditions:
http://marc.info/?l=reiserfs-devel&m=12 ... 32124&q=p3

NFS fixes:
http://marc.info/?l=reiserfs-devel&m=12 ... 29461&q=p3

mroconnor · Post by **mroconnor** » Tue Mar 25, 2008 9:29 pm

As anyone had any success with any version of ati-drivers and 2.6.25 master-devel?? I tried it when .25 first came out and couldnt get ati-drivers to compile. I know someone has made it happen....right?

kernelOfTruth · Post by **kernelOfTruth** » Tue Mar 25, 2008 9:48 pm

gimpel wrote:
kernelOfTruth wrote:Hi Witam,

is lockless pagecache still included in zen-sources ?

thanks
Looking at "git log /include/linux/radix-tree.h" both lockless and concurrent pagecache is in there.

I was just worried whether it's still in there, but it seems to play fine with all loads it seems

@kfiaciarka, dodo1122:

thanks, I'll give it a try within the next days ...

gimpel wrote:Please apply following Reiser4 patches:

ccreg40 diskfull conditions:
http://marc.info/?l=reiserfs-devel&m=12 ... 32124&q=p3

NFS fixes:
http://marc.info/?l=reiserfs-devel&m=12 ... 29461&q=p3

++

both are very important, please include

I'm forced to fill several backup HDDs in the following days &

it would be nice if this was fixed before I run into a null pointer dereference or data loss situation

hirakendu · Post by **hirakendu** » Fri Mar 28, 2008 5:40 am

Is the squashfs-lzma updated ? The latest squashfs-lzma updates upstream fix the data corruption issue. Btw, last seen, the fixes to squashfs lzma made in zen-sources breaks it instead. And the code to make lzma optional makes it pretty complicated

.

soundcheck · Post by **soundcheck** » Fri Mar 28, 2008 1:05 pm

Hi.

Interesting read about CFS: http://www.ibm.com/developerworks/linux/library/l-cfs/

Just to bring it up again:
I am still facing problems with the sched parameters which I can not set.

Am I the only one here using the custom ZEN settings? Respectively having this problem?
I mean, these are IMO key features within ZEN.

I just tried e.g.
$ sysctl -w kernel.sched_min_granularity_ns=40000

giving me
error: "Invalid argument" setting key "kernel.sched_min_granularity_ns"

Any ideas what this could mean?

Cheers
KLS