Zen-Kernel Discussion/Support Thread - Part 4

Message

lkraav · Post by **lkraav** » Tue May 24, 2011 1:33 pm

ponciarello wrote:which bugs are you referring to?

i'm also at a loss with his comment..

dE_logics · Post by **dE_logics** » Tue May 24, 2011 2:33 pm

ponciarello wrote:which bugs are you referring to?

Kernel will not compile with squashfs + XZ, suspend and hibernate not working, Gentoo will not shutdown (hangs after mount / ro), corruption with reiser4 (actually it got worst with vanilla 2.6.38)

ponciarello · Post by **ponciarello** » Tue May 24, 2011 3:01 pm

dE_logics wrote:Kernel will not compile with squashfs + XZ

it builds fine here, also on master-2.6.39.

suspend and hibernate not working

this is an upstream kernel problem

Gentoo will not shutdown (hangs after mount / ro)

can't help much with this, sorry, I use slackware

corruption with reiser4 (actually it got worst with vanilla 2.6.38)

so this is not a zen problem too.

mbar · Post by **mbar** » Tue May 24, 2011 7:26 pm

ponciarello wrote:which bugs are you referring to?

All of them. Like inability to even boot zen based livecd under x64 KVM machine.
I also have had with zen and I'm back to gen2-sources. Just too much trouble with zen.

ponciarello · Post by **ponciarello** » Tue May 24, 2011 7:45 pm

mbar wrote:All of them. Like inability to even boot zen based livecd under x64 KVM machine.

I boot very often slax-remix (i486 and x86_64, zen-kernel based) and the derivate porteus on a x86_64 kvm, really don't know which livecd you're referring to, sorre.
but I think you have to blame only who provides the livecd

pappy_mcfae · Post by **pappy_mcfae** » Tue May 24, 2011 8:53 pm

The only problems I've ever had with zen-sources are their lackluster release timing and format. From only getting them through the git, to having them in portage, to changing the naming, and so on, keeping up with zen sources to be able to make seeds with them has been interesting, to say the least. To say the most, it's been a PITA!

As far as bugs, before you condemn the source, be absolutely SURE you've configured things properly. One bad setting can destroy a kernel's ability to work properly. Something I also recently discovered is if you push the timer frequency beyond 1000Hz, you will make an unstable kernel. This effect is amplified with multi-core processors, and can lead to all manner of trouble...file system corruption, inability to boot, drivers that don't load properly, and so on.

Before you give up on zen, try one of my seeds. The only caveat is that you set the timer less than 1500Hz. With multi-core processors, 300Hz will work just fine. I actually recommend that, as multi-core processors have more than enough horsepower without having to "hold the pedal to the floor" so to speak. Since the timer frequency multiplies itself by the number of processor cores, a four core processor would wind up creating 1200Hz internally.

Cheers,
Pappy

Anon-E-moose · Post by **Anon-E-moose** » Tue May 24, 2011 9:01 pm

I've been running zen through several iterations and have always kept the hertz at 1000 without any problem at all.

pappy_mcfae · Post by **pappy_mcfae** » Tue May 24, 2011 10:09 pm

Be that as it may, please note that I have experience with others who have had issues with that particular setting. I am sending out that reality for those who are having issues, not to compare with others who are not having issues. My laptop works fine at 1500, but none of the multi-core processors I have can handle it. At 300Hz, none of my machines have issues.

Cheers,
Pappy

lkraav · Post by **lkraav** » Wed May 25, 2011 3:01 am

pappy_mcfae wrote:Be that as it may, please note that I have experience with others who have had issues with that particular setting. I am sending out that reality for those who are having issues, not to compare with others who are not having issues. My laptop works fine at 1500, but none of the multi-core processors I have can handle it. At 300Hz, none of my machines have issues.

thanks for the tips. it's stuff like this that makes it worthwhile hanging out in the forums.

i've had a relatively stable experience with 2.6.38.5-zen+, aside from my acer-wmi problems and the occasional btrfs BUG. i now re-checked and indeed i had been running this i3-330UM laptop at 1000Hz. this info was good enough reason to drop the timer down and enable some pm debug stuff as well for next testing leg.

mbar · Post by **mbar** » Wed May 25, 2011 4:12 am

ponciarello wrote:
mbar wrote:All of them. Like inability to even boot zen based livecd under x64 KVM machine.
I boot very often slax-remix (i486 and x86_64, zen-kernel based) and the derivate porteus on a x86_64 kvm, really don't know which livecd you're referring to, sorre.
but I think you have to blame only who provides the livecd

That livecd I my own

It is unable to boot when zen-kernel is both on host and on livecd (I have a laptop with i7) and when livecd kernel has CK BFS enabled. I get only blank screen just after grub.
gentoo-sources do not experience that problem.

dE_logics · Post by **dE_logics** » Wed May 25, 2011 5:10 am

ponciarello wrote:
dE_logics wrote:Kernel will not compile with squashfs + XZ
it builds fine here, also on master-2.6.39.

suspend and hibernate not working
this is an upstream kernel problem

Gentoo will not shutdown (hangs after mount / ro)
can't help much with this, sorry, I use slackware

corruption with reiser4 (actually it got worst with vanilla 2.6.38)
so this is not a zen problem too.

It does not build with 2.6.37 here.

Suspend and hibernate works with gentoo and vanila sources 2.6.37

reiser4 corruption does not persist with systemrescuecd.

ponciarello wrote:I boot very often slax-remix (i486 and x86_64, zen-kernel based) and the derivate porteus on a x86_64 kvm, really don't know which livecd you're referring to, sorre.
but I think you have to blame only who provides the livecd

There are lot of problems that I don't experience but a lot of other users and PCs administered by myself do. That doesn't mean there are no bugs. No wonder software testing is big business.

Lastly, I tend to report a lot of bugs, and the first thing they ask me to do is use vanilla kernel.

dE_logics · Post by **dE_logics** » Wed May 25, 2011 5:13 am

pappy_mcfae wrote:Something I also recently discovered is if you push the timer frequency beyond 1000Hz, you will make an unstable kernel. This effect is amplified with multi-core processors, and can lead to all manner of trouble...file system corruption, inability to boot, drivers that don't load properly, and so on.

One more issue is that the corefrequency tends to jump more often without any reasons.

pappy_mcfae · Post by **pappy_mcfae** » Wed May 25, 2011 6:55 am

This might be the root of the issue. All I know is that I've had at least three seed users who had problems with the timer settings...and all were multi-core. One was an i7, which was brought to its knees at 1000Hz. That was a bit of a humbling experience, but one I am glad I had nonetheless.

Cheers,
Pappy

ponciarello · Post by **ponciarello** » Thu May 26, 2011 6:01 am

bad news for 2.6.39?

pappy_mcfae · Post by **pappy_mcfae** » Thu May 26, 2011 6:04 am

No kidding.

Cheers,
Pappy

albright · Post by **albright** » Thu May 26, 2011 12:48 pm

bad news for 2.6.39?

yes - I had to go back to 2.6.38 because of this exact problem

but it took a while since the symptoms were very weird
(e.g. the "ps" command hanging (happened once),
strange browser errors (honest) and the give away failure
to be able to halt or reboot ...)

Ant P. · Post by **Ant P.** » Thu May 26, 2011 6:59 pm

Just out of curiosity, can the people getting crashes on 2.6.39 post their lspci -n and lsmod?

kernelOfTruth · Post by **kernelOfTruth** » Fri May 27, 2011 4:22 pm

ponciarello wrote:bad news for 2.6.39?

eww

upstream broke it once more

so 2.6.39 without ck-patches (or specifically: BFS) should be fine ?

how about the suspend issue (for me suspend-to-ram is important hibernate kinda doesn't work - I do not use):

what cards are affected ? AMD ? evergreen ?

thanks !

ponciarello · Post by **ponciarello** » Mon May 30, 2011 7:00 pm

good news?

kernelOfTruth · Post by **kernelOfTruth** » Sat Jun 04, 2011 1:37 am

ok I got this really weird issue with screen output

and master-2.6.39

after it says: "booting kernel" there's no change, scrolling of text or anything else

it just sits there and waits for some serious mode change

graphics driver / card: Radeon 5850, opensource radeon-driver

it would be easy if I could compile in the radeon module and the additional dependencies but with that it would hardlock at boot (from the start - no improvements until now - so it has to be compiled as a module)

so i'm forced to fly this baby blind (enter password for the luks partitions) until I get to the login prompt, login as root

and do

Code: Select all

modprobe radeon modeset=1

or

Code: Select all

/etc/init.d/xdm start

after that I can get to X

with xdm start only (without loading the kernel-module) the output of the VTs is still the same like right after launching the kernel

the last line displayed is "booting kernel" - switching through them alt+f1 - f6 ...

shows no difference

but when inserting / modprobeing the module output of the VTs is displayed again and also for all the other VTs

does this ring a bell for anyone ? how can this be fixed ? did I miss to compile a missing dependency into the kernel or is this a known regression ?

Many thanks in advance !

besides that 2.6.39-zen is running great so far - it feels really fast ^^

ponciarello · Post by **ponciarello** » Sun Jun 05, 2011 5:18 am

same here since the first versions of master-2.6.39 (nvidia): I got around it installing the proprietary video driver via ssh and boot after with a graphical login manager (but still no output on the VTs).

kernelOfTruth · Post by **kernelOfTruth** » Mon Jun 06, 2011 3:46 pm

ponciarello wrote:same here since the first versions of master-2.6.39 (nvidia): I got around it installing the proprietary video driver via ssh and boot after with a graphical login manager (but still no output on the VTs).

well, according to the log of 2.6.39.1 it doesn't seem if that update would fix it :\

are any bug reports or reports to lkml known ?

slightly related: http://forums.gentoo.org/viewtopic-p-66 ... 2d2593228b

but I don't have uvesa or vesafb enabled afaik

so it's strange that I'm also affected ...

ponciarello · Post by **ponciarello** » Tue Jun 07, 2011 6:30 am

Steven (thanks!) seems to have fixed it in latest git.

kernelOfTruth · Post by **kernelOfTruth** » Wed Jun 08, 2011 5:28 pm

confirmed !

thanks Steven

where's the Donate / paypal stuff that was used to be on the zen-kernel.org homepage ?

lkraav · Post by **lkraav** » Wed Jun 08, 2011 7:24 pm

how about flattr button instead.