Acpi slows down my new Asus portable quite a lot

Acpi slows down my new Asus portable quite a lot

[Carlo E. Prelz]

Dear readers, 

I up to now managed to not need to tackle ACPI world. But I have
recently purchased a fresh new Asus portable.

It is a Centrino-based N3700N. Using a kernel with only APM, the audio
chipset would not be recognized (was assigned IRQ 0 - suggestions to
solve this problem by disabling the 'PnP OS' bios option were useless,
since the bios of this machine lacks this (and many other) options). 

Having read that ACPI can manage IRQ's, I compiled it, and lo! audio
chip detected and working! But the machine slows down quite a
lot. This happens both on 2.4.22-pre5 and on 2.6.0-test1, this last
one also with the small ACPI patch applied. A typical top output looks
as follows:

    3 root      15 -10     0    0    0 R 99.9  0.0  59:11.49 events/0
  644 root      15   0  1696  932 1576 R  0.3  0.2   0:00.01 top
    1 root      15   0  1212  468 1180 S  0.0  0.1   0:01.27 init
    2 root      34  19     0    0    0 S  0.0  0.0   0:00.00 ksoftirqd/0
    4 root       5 -10     0    0    0 S  0.0  0.0   0:00.00 kblockd/0
    5 root      16   0     0    0    0 S  0.0  0.0   0:00.05 khubd
    6 root      25   0     0    0    0 S  0.0  0.0   0:00.00 pdflush
    7 root      15   0     0    0    0 S  0.0  0.0   0:00.00 pdflush
    8 root      25   0     0    0    0 S  0.0  0.0   0:00.00 kswapd0
    9 root      10 -10     0    0    0 S  0.0  0.0   0:00.00 aio/0
   10 root      22   0     0    0    0 S  0.0  0.0   0:00.00 pccardd
   11 root      17   0     0    0    0 S  0.0  0.0   0:00.00 kseriod
   12 root      15   0     0    0    0 S  0.0  0.0   0:00.00 kjournald
   70 root      25   0     0    0    0 S  0.0  0.0   0:00.00 kjournald
  111 root      15   0  2052  824 1936 S  0.0  0.2   0:00.00 dhclient
...
...
...

As you can see, 99.9 of the cpu time is sucked by events/0. In 2.4
kernels, the fat process is called keventd, but the result is the
same.

I have compiled ACPI first without support for any sub-module, and
then with fan and asus-specific support. No change happens. What I
noticed is that, soon after boot, the PC's fan starts blowing, and
never stops while the machine is on. This does not happen without
ACPI. 

I decided to write after finding a bug registration on your site ([Bug
843] New: Getting very high CPU load on Laptop with ACPI). The
portable of the submitter of the bug is a Fujitsu-Siemens C6155, but
his situation is otherwise quite similar to mine. I found no reference
of this on the acpi-devel list, so I thought I'd drop a line.

Here is an extract of the boot messages:

--8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<--
...
...
...
ACPI: Subsystem revision 20030714
ACPI: Interpreter enabled
ACPI: Using PIC for interrupt routing
ACPI: PCI Root Bridge [PCI0] (00:00)
PCI: Probing PCI hardware (bus 00)
PCI: Ignoring BAR0-3 of IDE controller 0000:00:1f.1
Transparent bridge - Intel Corp. 82801BAM/CAM PCI Bri
ACPI: PCI Interrupt Routing Table [\_SB_.PCI0._PRT]
ACPI: PCI Interrupt Routing Table [\_SB_.PCI0.P0P1._PRT]
ACPI: Embedded Controller [EC0] (gpe 28)
ACPI: PCI Interrupt Link [LNKA] (IRQs 3 4 5 6 7 *11 12)
ACPI: PCI Interrupt Link [LNKB] (IRQs 3 4 5 6 7 11 12, disabled)
ACPI: PCI Interrupt Link [LNKC] (IRQs *4 12)
ACPI: PCI Interrupt Link [LNKD] (IRQs *5 6)
ACPI: PCI Interrupt Link [LNKE] (IRQs 6 *11)
ACPI: PCI Interrupt Link [LNKF] (IRQs 3 7, disabled)
ACPI: PCI Interrupt Link [LNKG] (IRQs 4 7, disabled)
ACPI: PCI Interrupt Link [LNKH] (IRQs *4 6 12)
ACPI: Power Resource [GFAN] (off)
...
...
...
ACPI: PCI Interrupt Link [LNKB] enabled at IRQ 5
ACPI: PCI Interrupt Link [LNKF] enabled at IRQ 3
ACPI: PCI Interrupt Link [LNKG] enabled at IRQ 7
PCI: Using ACPI for IRQ routing
PCI: if you experience problems, try using option 'pci=noacpi' or even 'acpi=off'
...
...
...
ACPI: Fan [FN00] (off)
Asus Laptop ACPI Extras version 0.24a
  M3N model detected, unsupported, trying default values, contact the
developers
  Notify Handler installed successfully
  Error changing brightness
  Error changing brightness
...
...
...
--8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<--

I am available if more info are needed, or if you need me to test any
patch. Thanks in advance!

Carlo

-- 
  *         Se la Strada e la sua Virtu' non fossero state messe da parte,
* K * Carlo E. Prelz - fluido-1XNsQiTun9D1P9xLtpHBDw@public.gmane.org             che bisogno ci sarebbe
  *               di parlare tanto di amore e di rettitudine? (Chuang-Tzu)


-------------------------------------------------------
This SF.Net email sponsored by: Parasoft
Error proof Web apps, automate testing & more.
Download & eval WebKing and get a free book.
www.parasoft.com/bulletproofapps1

Re: Acpi slows down my new Asus portable quite a lot

[Jan Rychter]

[-- Attachment #1: Type: text/plain, Size: 739 bytes --]

>>>>> "Carlo" == Carlo E Prelz <fluido-1XNsQiTun9D1P9xLtpHBDw@public.gmane.org> writes:
[...]
 Carlo> As you can see, 99.9 of the cpu time is sucked by events/0. In
 Carlo> 2.4 kernels, the fat process is called keventd, but the result
 Carlo> is the same.
[...]

I see the same thing on 2.4.22-pre5, keventd sometimes decides to run a
lot and consume CPU time. When it happens, keventd doesn't run all the
time, but runs for a while, then stops, then runs again.

I've noticed this doesn't happen if I don't use ALSA. But once ALSA has
been used, this happens even after ALSA has been unloaded.

I don't know if this is ACPI's fault or ALSA's fault. In fact, I don't
even know where and how to report it properly.

--J.

[-- Attachment #2: Type: application/pgp-signature, Size: 188 bytes --]

Re: Re: Acpi slows down my new Asus portable quite a lot

[Carlo E. Prelz]

	Subject: [ACPI] Re: Acpi slows down my new Asus portable quite a lot
	Date: Tue, Jul 15, 2003 at 01:22:57PM -0700

Quoting Jan Rychter (jan-JAsPCFd0eodBDgjK7y7TUQ@public.gmane.org):

> I see the same thing on 2.4.22-pre5, keventd sometimes decides to run a
> lot and consume CPU time. When it happens, keventd doesn't run all the
> time, but runs for a while, then stops, then runs again.
> 
> I've noticed this doesn't happen if I don't use ALSA. But once ALSA has
> been used, this happens even after ALSA has been unloaded.
> 
> I don't know if this is ACPI's fault or ALSA's fault. In fact, I don't
> even know where and how to report it properly.

No, here it happened even when ALSA was out of the picture. On 2.4 I
was running the OSS module for Intel ICH (i8xx) chips. I had this IRQ
problem and tried to see if this would be solved with 2.6. The problem
was still present with 2.6 and alpha, but it is solved by letting ACPI
handle the interrupts. Only, then the slowdown comes over. I see
non-stop load at 99.9%, never resting.

I reach an useable status (let's say, being able to playback a DVD
with MPlayer with acceptable performance) by re-nicing the offending
process. As of default it runs at -10; if I set it to 1 (just higher
than normal processes) I manage to do things. 

I wonder if it is possible to find out where these events that clog
the process come from. 

Carlo

-- 
  *         Se la Strada e la sua Virtu' non fossero state messe da parte,
* K * Carlo E. Prelz - fluido-1XNsQiTun9D1P9xLtpHBDw@public.gmane.org             che bisogno ci sarebbe
  *               di parlare tanto di amore e di rettitudine? (Chuang-Tzu)


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines at the
same time. Free trial click here: http://www.vmware.com/wl/offer/345/0

Re: Acpi slows down my new Asus portable quite a lot

[Karol Kozimor]

Thus wrote Carlo E. Prelz:
> It is a Centrino-based N3700N. Using a kernel with only APM, the audio

Though I can't unfortunately help much, I must say...

Wow, there's some really interesting stuff in your DSDT. It seems that the
nice people from ASUS decided to make their users' (and my) life easier, by
implementing frequency control methods for their driver. Since I'm leaving
in a couple of days, this won't probably be implemented for some time yet
(unless Julien gets to work on it), but look forward to getting basic
support for your laptop in the tomorrow's acpi4asus CVS.

BTW: Have you tried recompiling your DSDT? The MXN DSDT's oddly contain
some trivial-to-fix errors (like the If(SSn) statements in your case),
there's even an updated DSDT at http://acpi.sf.net.

Best regards,

-- 
Karol 'sziwan' Kozimor
sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines at the
same time. Free trial click here: http://www.vmware.com/wl/offer/345/0

Re: Acpi slows down my new Asus portable quite a lot

[Carlo E. Prelz]

	Subject: Re: [ACPI] Acpi slows down my new Asus portable quite a lot
	Date: Wed, Jul 16, 2003 at 12:10:27AM +0200

Quoting Karol Kozimor (sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org):

> Thus wrote Carlo E. Prelz:
> > It is a Centrino-based N3700N. Using a kernel with only APM, the audio
> 
> Though I can't unfortunately help much, I must say...
> 
> Wow, there's some really interesting stuff in your DSDT. It seems that the
> nice people from ASUS decided to make their users' (and my) life easier, by
> implementing frequency control methods for their driver. Since I'm leaving
> in a couple of days, this won't probably be implemented for some time yet
> (unless Julien gets to work on it), but look forward to getting basic
> support for your laptop in the tomorrow's acpi4asus CVS.

I'll keep track of it. I have some updates. After turning the portable
off last night, I turned it on this morning, and the problem seemed to
have disappeared... the process was not eating cpu anymore. I started
X, and the fan started working briefly, then stopped when I left X. I
did a reboot, and things were still OK. Then, when I started a 'make
menuconfig,' the fan started working again, and exactly then the
process began again to eat 99.9% of the CPU timæ.

The problem is really connected with fan operation: if I go look at
/proc/acpi/fan/FN00/status, I find

status:       off

while the fan is instead very much in operation. If I do

echo 'on' > state

the content changes to 

status:       on

but if I then try to write 'off', mothing changes. Can this all be
useful for tracking down the problem? 

> BTW: Have you tried recompiling your DSDT? The MXN DSDT's oddly contain
> some trivial-to-fix errors (like the If(SSn) statements in your case),
> there's even an updated DSDT at http://acpi.sf.net.

I read something (and I am under deadline for my work... grgrgr) about
all this, but I could not learn enough to be confident with the
process. 

Maybe later...

Thanks in all cases

Carlo

-- 
  *         Se la Strada e la sua Virtu' non fossero state messe da parte,
* K * Carlo E. Prelz - fluido-1XNsQiTun9D1P9xLtpHBDw@public.gmane.org             che bisogno ci sarebbe
  *               di parlare tanto di amore e di rettitudine? (Chuang-Tzu)


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines at the
same time. Free trial click here: http://www.vmware.com/wl/offer/345/0

Re: Acpi slows down my new Asus portable quite a lot

[Sebastian Henschel]

[-- Attachment #1: Type: text/plain, Size: 1118 bytes --]

hello karol and carlo..

* Karol Kozimor <sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org> [2003-07-16 09:15 +0200]:
> Thus wrote Carlo E. Prelz:
> > It is a Centrino-based N3700N. Using a kernel with only APM, the audio
> 
> Though I can't unfortunately help much, I must say...

> BTW: Have you tried recompiling your DSDT? The MXN DSDT's oddly contain
> some trivial-to-fix errors (like the If(SSn) statements in your case),
> there's even an updated DSDT at http://acpi.sf.net.

http://acpi.sf.net/dsdt/view.php?id=84

does that really fix the keventd-problem as it says in the comment?
i diffed the original and the custom .asl and the only difference seems
the removal of "If (SSn)"-stuff. and only that does not seem to fix it,
because (for the record): the ASUS S1300N is also affected by the very same
problem which persists after syntactically debugging the DSDT.

btw, the hotkeys from asus_acpi work for S1300N, too. but more on that
in a pm, karol.

cheers,
 sebastian
-- 
::: sebastian henschel
::: kodeaffe
::: lynx -source http://www.kodeaffe.de/shensche.pub | gpg --import

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Acpi slows down my new Asus portable quite a lot

[Sebastian Henschel]

[-- Attachment #1: Type: text/plain, Size: 1869 bytes --]

hi again..

* Sebastian Henschel <acpi-1rjuZeEg9oEb1SvskN2V4Q@public.gmane.org> [2003-07-16 14:02 +0200]:
> 
> * Karol Kozimor <sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org> [2003-07-16 09:15 +0200]:
> > Thus wrote Carlo E. Prelz:
> > > It is a Centrino-based N3700N. Using a kernel with only APM, the audio
> > 
> > Though I can't unfortunately help much, I must say...
> 
> > BTW: Have you tried recompiling your DSDT? The MXN DSDT's oddly contain
> > some trivial-to-fix errors (like the If(SSn) statements in your case),
> > there's even an updated DSDT at http://acpi.sf.net.
> 
> http://acpi.sf.net/dsdt/view.php?id=84
> 
> does that really fix the keventd-problem as it says in the comment?
> i diffed the original and the custom .asl and the only difference seems
> the removal of "If (SSn)"-stuff. and only that does not seem to fix it,
> because (for the record): the ASUS S1300N is also affected by the very same
> problem which persists after syntactically debugging the DSDT.

ahh.. archives are up again.. browsing them i found:

http://sourceforge.net/mailarchive/forum.php?thread_id=2693584&forum_id=7803

(check this, carlos) and looked at the aforementioned diff again. i had
overlooked the "comments" patched into the DSDT, which was all that had to
be done for disabling some thermal handling.

and after i commented this one function out in the DSDT for the
S1300N... *tadaaa* ... keventd is using ~0% cpu and the harddisk speeds
up to 29MB/s (hdparm -t). now also the kernel module "thermal" does not
oops when unloading it.

ok, now i only have to find out why speedstep does not work with with
cpufreq and why i cannot unblank the console except when switching to
XFree86.

bye,
 sebastian
-- 
::: sebastian henschel
::: kodeaffe
::: lynx -source http://www.kodeaffe.de/shensche.pub | gpg --import

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Acpi slows down my new Asus portable quite a lot

[Carlo E. Prelz]

	Subject: Re: [ACPI] Acpi slows down my new Asus portable quite a lot
	Date: Wed, Jul 16, 2003 at 02:36:20PM +0200

Quoting Sebastian Henschel (acpi-1rjuZeEg9oEb1SvskN2V4Q@public.gmane.org):

> http://sourceforge.net/mailarchive/forum.php?thread_id=2693584&forum_id=7803

I must publically beg for pardon: I had not caught mention of this
very recent thread in acpi_support. (I could not find a place where I
could do keyword search on sourceforge mailing lists...)

After spending the last 2 hours making the acquaintance with iasl &
friends, I was able to patch the kernel as suggested, and indeed my
M3700S is quite spiffy now. Impressive.

I will have to find some time to understand acpi in depth. It seems to
be a treasure chest of tweakitude... ;-)

Not now. I must go back to my original deadline...

Thanks again. 

Carlo

-- 
  *         Se la Strada e la sua Virtu' non fossero state messe da parte,
* K * Carlo E. Prelz - fluido-1XNsQiTun9D1P9xLtpHBDw@public.gmane.org             che bisogno ci sarebbe
  *               di parlare tanto di amore e di rettitudine? (Chuang-Tzu)


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines at the
same time. Free trial click here: http://www.vmware.com/wl/offer/345/0