Official thread: "zenmm-sources" - Latest: 2.6.27-rc1-zenmm0

[kernelOfTruth]

if I'm the only one using klive it's useless

@rmh3093:

if you want more feedback klive-wise - drop a note on the zen-repositories, the homepage and the zen-threads, that way the situation (hopefully) will improve

[mroconnor]

grrrrrrrrr.....What the hell am I doing wrong then?? I have hard locks and a ton of those rcu_boost bugs/warnings.

I am using klive now as well.

[kernelOfTruth]

grrrrrrrrr.....What the hell am I doing wrong then?? I have hard locks and a ton of those rcu_boost bugs/warnings.

good luck, buddy - here comes the kernel-config

http://paste.ubuntu.com/27819/

[mroconnor]

woohoo!!! cheat sheet!!

Thanks KoT!

[kernelOfTruth]

woohoo!!! cheat sheet!!

Thanks KoT!

just uploaded to a paste-service (added link),

the forum cuts out several lines

you're welcome

[kernelOfTruth]

woohoo!!! cheat sheet!!

Thanks KoT!

just uploaded to a paste-service (added link),

the forum cuts out several lines

you're welcome

update:

this is a configuration somewhat "specific" to an dell m1330 + p5w dh deluxe asus box, but should work on your box, too (amd64 / 64bit)

good luck;


this kernel (2.6.26-rc8-mm3) is simply amazing ! it has an average load of 29.15 - 29.27 (currently compiling a kernel + testing stability) and has no lags ,

internet streaming of audio (shoutcast) works perfectly, nearly no lagging / delay in input via mouse/keyboard + output via display, working & launching of apps has also nearly no delay


from me you guys get an A+

thanks a thousands times

I hope 2.6.26-mm1 will have this kind of quality, too

[rmh3093]

grrrrrrrrr.....What the hell am I doing wrong then?? I have hard locks and a ton of those rcu_boost bugs/warnings.

I am using klive now as well.

do u still have *_trace enabled? if so get rid of it, u dont need it

[rmh3093]

woohoo!!! cheat sheet!!

Thanks KoT!

just uploaded to a paste-service (added link),

the forum cuts out several lines

you're welcome

update:

this is a configuration somewhat "specific" to an dell m1330 + p5w dh deluxe asus box, but should work on your box, too (amd64 / 64bit)

good luck;


this kernel (2.6.26-rc8-mm3) is simply amazing ! it has an average load of 29.15 - 29.27 (currently compiling a kernel + testing stability) and has no lags ,

internet streaming of audio (shoutcast) works perfectly, nearly no lagging / delay in input via mouse/keyboard + output via display, working & launching of apps has also nearly no delay


from me you guys get an A+

thanks a thousands times

I hope 2.6.26-mm1 will have this kind of quality, too

im glad its working well for you, it works great for me too... i really wish i knew what was blowing up for mroconnor and cheater

[mroconnor]

@rmh3093 & @KoT:
I'll be back at it tonight testing out things. I think a lot of the bugs that *_trace is throwning might be ignored so I will get rid of it. I'm going to check my .config against KoT's and see whats up. But WOW KoT you have everything in your .config


Anyway I think I can get it sorted out and back running smoothly soon.

@rmh3093: how hard is it to go from baselayout2 to einit?

[rmh3093]

@rmh3093 & @KoT:
I'll be back at it tonight testing out things. I think a lot of the bugs that *_trace is throwning might be ignored so I will get rid of it. I'm going to check my .config against KoT's and see whats up. But WOW KoT you have everything in your .config


Anyway I think I can get it sorted out and back running smoothly soon.

@rmh3093: how hard is it to go from baselayout2 to einit?

well they dont mess with each other so u can switch back and forth....

einit takes a little patients editing some xml config files and not every program has an init script so for something obscure you might need to make your own script

[kicior]

I still have hardlocks sometimes with zenmm3 and very often my network is dead (reloading module doesn't help, even ethtool claims there is no link detected - after reboot network is back (module e1000)). rc9-zen1 is working ok, but hdaps is not working:

Code: Select all

hdaps: IBM ThinkPad R61 detected.
hdaps: driver init failed (ret=-6)! 

I have also a lot of garbage in dmesg.

[gimpel]

Runs smooth like oiled ferret again with all fancy stuff enabled: rcu preempt, IRQ threading, LLD-Preempt & tunables, Reiser4, -march=native - all fine.

All I can do to help is maybe also provide my .config: http://phpfi.com/332998

[mroconnor]

I still have hardlocks sometimes with zenmm3 and very often my network is dead (reloading module doesn't help, even ethtool claims there is no link detected - after reboot network is back (module e1000)). rc9-zen1 is working ok, but hdaps is not working:

Code: Select all

hdaps: IBM ThinkPad R61 detected.
hdaps: driver init failed (ret=-6)! 

I have also a lot of garbage in dmesg.

EDIT: posting to forums on a blackberry while driving is both dangerous and a sign of -zen addiction.

I can confirm hdaps works on my z61m w/ -zenmm3. But bec_se I have a lot of other issues right now I can't confirm if it works well w/ the fancy stuff(irq etc etc).

[rmh3093]

git clone git://zen-sources.org/zen/kernel-rt.git zenrt-sources

EDIT: not all the branches are merged but it works on my box, dmesg is a little dirty but i think its non fatal as the kernel is still alive

[mroconnor]

git clone git://zen-sources.org/zen/kernel-rt.git zenrt-sources

EDIT: not all the branches are merged but it works on my box, dmesg is a little dirty but i think its non fatal as the kernel is still alive

What would be the differences between -zenrt & -zenmm?

[rmh3093]

git clone git://zen-sources.org/zen/kernel-rt.git zenrt-sources

EDIT: not all the branches are merged but it works on my box, dmesg is a little dirty but i think its non fatal as the kernel is still alive

What would be the differences between -zenrt & -zenmm?

well i found a git repo that has up-to-date -rt in it so -zenrt will be based on that with the normal zen branches merged into it, also I will try and merge as much of -mm brokenout at some point

[mroconnor]

I am all but giving up on -zenmm3, it just doesnt like me(or the laptop). Going to try ne last thing and get rid of HDAPS and tp_smapi and see what happens.
EDIT:
I removed tp_smapi & HDAPS but still have this in dmesg:

Code: Select all

------------[ cut here ]------------
WARNING: at kernel/lockdep.c:2085 handle_IRQ_event+0xd0/0xee()
Modules linked in: snd_pcm_oss snd_mixer_oss snd_seq_dummy snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device radeon snd_hda_intel snd_pcm snd_timer snd intel_agp iwl3945 soundcore snd_page_alloc
Pid: 3442, comm: IRQ-216 Not tainted 2.6.26-rc8-zenmm3 #26

Call Trace:
 [<ffffffff802304d1>] warn_on_slowpath+0x51/0x77
 [<ffffffff80250dbd>] __lock_acquire+0xf44/0xf53
 [<ffffffff802619f2>] __rcu_read_unlock+0x94/0xa2
 [<ffffffff80210fa4>] native_sched_clock+0x4a/0x67
 [<ffffffff8024df28>] get_lock_stats+0x24/0x50
 [<ffffffff8025e82e>] do_irqd+0x182/0x2f7
 [<ffffffff8024f825>] trace_hardirqs_on_caller+0xb4/0x123
 [<ffffffff8025e349>] handle_IRQ_event+0xd0/0xee
 [<ffffffff8025e839>] do_irqd+0x18d/0x2f7
 [<ffffffff8024f86a>] trace_hardirqs_on_caller+0xf9/0x123
 [<ffffffff8025e6ac>] do_irqd+0x0/0x2f7
 [<ffffffff8024295f>] kthread+0x47/0x73
 [<ffffffff8053337e>] trace_hardirqs_on_thunk+0x3a/0x3f
 [<ffffffff8020c1d8>] child_rip+0xa/0x12
 [<ffffffff8020b8c3>] restore_args+0x0/0x30
 [<ffffffff80242918>] kthread+0x0/0x73
 [<ffffffff8020c1ce>] child_rip+0x0/0x12

INFO: lockdep is turned off.

Now the interesting thing is without HDAPS/tp_smapi I no longer have cpu0 running at 93-96%. cpu0 now operates at about 14% while cpu1 is at 4%, this is while I am typing this and nothing else. I still have a lot of activity on a process call IRQ-0 and I see a lot of softirq process creeping into the top 3.

I am not sure what is going on....more testing i suppose.

[mroconnor]

I just did this:

Code: Select all

tail -f /var/log/messages

Two things are apperant, one my cpufreq is now working kind of correctly, no errors like this:

Code: Select all

Jul 17 14:52:12 cosmo cpufreqd: cpufreqd_loop            : Cannot set policy, Rule unchanged ("AC Rule").
Jul 17 14:52:14 cosmo cpufreqd: cpufreqd_set_profile     : Couldn't set profile "Performance Low" set for cpu0 (1667000-1667000-performance)

And no wacky bug/warning messages like before. AND i haven't hardlocked yet.

Who wants to bet? HDAPs or tp_smapi?

[mroconnor]

HDAPS and something arent playing nicely together. When HDAPS in a module cpu0 goes crazy with utilization, with no HDAPS the laptop is happy with the exception of the lockdep error when booting.

I am going to research more, there was some talk about lockdep and IRQs in 2006 with -mm.

[rmh3093]

HDAPS and something arent playing nicely together. When HDAPS in a module cpu0 goes crazy with utilization, with no HDAPS the laptop is happy with the exception of the lockdep error when booting.

I am going to research more, there was some talk about lockdep and IRQs in 2006 with -mm.

i had a feeling i was hdaps

[mroconnor]

I still get this, even with out HDAPs, grrr.

Code: Select all

Jul 17 16:12:38 cosmo WARNING: at kernel/lockdep.c:2085 handle_IRQ_event+0xd0/0xee()
Jul 17 16:12:38 cosmo Modules linked in: snd_pcm_oss snd_mixer_oss snd_seq_dummy snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device tp_smapi hdaps thinkpad_ec radeon snd_hda_intel snd_pcm snd_timer snd iwl3945 intel_agp soundcore snd_page_alloc
Jul 17 16:12:38 cosmo Pid: 3482, comm: IRQ-216 Not tainted 2.6.26-rc8-zenmm3 #28
.........

But it does seem to run nicely.

[rmh3093]

I still get this, even with out HDAPs, grrr.

Code: Select all

Jul 17 16:12:38 cosmo WARNING: at kernel/lockdep.c:2085 handle_IRQ_event+0xd0/0xee()
Jul 17 16:12:38 cosmo Modules linked in: snd_pcm_oss snd_mixer_oss snd_seq_dummy snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device tp_smapi hdaps thinkpad_ec radeon snd_hda_intel snd_pcm snd_timer snd iwl3945 intel_agp soundcore snd_page_alloc
Jul 17 16:12:38 cosmo Pid: 3482, comm: IRQ-216 Not tainted 2.6.26-rc8-zenmm3 #28
.........

But it does seem to run nicely.

well u say with out hdaps but that message still referrs to tp_smapi and hdaps as if they were loaded

[mroconnor]

I didnt build in or make modules of HDAPS and I forgot to remove it from /etc/conf.d/modules. I think that is the reason.

[tranquilcool]

git clone git://zen-sources.org/zen/kernel-rt.git zenrt-sources

EDIT: not all the branches are merged but it works on my box, dmesg is a little dirty but i think its non fatal as the kernel is still alive

What would be the differences between -zenrt & -zenmm?

well i found a git repo that has up-to-date -rt in it so -zenrt will be based on that with the normal zen branches merged into it, also I will try and merge as much of -mm brokenout at some point

not possible to get nvidia-drivers to install.

[rmh3093]

git clone git://zen-sources.org/zen/kernel-rt.git zenrt-sources

EDIT: not all the branches are merged but it works on my box, dmesg is a little dirty but i think its non fatal as the kernel is still alive

What would be the differences between -zenrt & -zenmm?

well i found a git repo that has up-to-date -rt in it so -zenrt will be based on that with the normal zen branches merged into it, also I will try and merge as much of -mm brokenout at some point

not possible to get nvidia-drivers to install.

thats a shame, radeonhd works like a charm