C1VN, ACPI and ESD/ASD = const CPU

C1VN, ACPI and ESD/ASD = const CPU

[Jamin W. Collins]

Has anyone else tried using ACPI with the C1VN or similiar laptops?  I've
recently recompiled my kernel, removing APM and enabling ACPI.  Most
things seem to work fine (battery status and events).  However, when I
start esd or asd to mix audio streams my processor utilization immediately
jumps to ~50% and stays there.  The utilization drops to 15-30% if I set
longrun to maximum rather than variable.  I've verified the utilization
with both "top" and "gkrellm".  However, while top shows that my CPU is
not completely idle, it does not list what is using the CPU.  I've tried
sorting by CPU utilization, but there are still large gaps. As shown by
the following:

CPU states:   0.8% user,  25.7% system,   0.0% nice,  73.5% idle
Mem:    110876K total,   107672K used,     3204K free,     9944K buffers
Swap:   996020K total,        0K used,   996020K free,    48224K cached

  PID USER     PRI  NI  SIZE  RSS SHARE STAT %CPU %MEM   TIME COMMAND
  700 jcollins  15   0   960  960   748 R     1.1  0.8   0:16 top
  591 root      10 -10 39296  30M  2052 S <   0.7 27.8   2:14 XFree86
    1 root       8   0   488  488   424 S     0.0  0.4   0:01 init
    2 root       9   0     0    0     0 SW    0.0  0.0   0:00 keventd
    3 root      19  19     0    0     0 SWN   0.0  0.0   0:00
ksoftirqd_CPU0
    4 root       9   0     0    0     0 SW    0.0  0.0   0:00 kswapd

I'm able to run either ESD or ASD with a non-ACPI kernel without these
symptoms and likewise, running the ACPI kernel without ESD or ASD results
in normal CPU utilization.

I've tried both the vanilla ACPI support provided with the 2.4.18 kernel
and the latest patches provided by the ACPI project for the 2.4.18 kernel,
both exhibit the same behaviour.

Any ideas on how to use both ACPI and either ASD or ESD without this high
CPU utilization cropping up?

-- 
Jamin W. Collins


-------------------------------------------------------
This sf.net email is sponsored by: OSDN - Tired of that same old
cell phone?  Get a new here for FREE!
https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390

Re: C1VN, ACPI and ESD/ASD = const CPU

[Faye Pearson]

Jamin W. Collins [jcollins-u0TlgVUp4POoiH3BShWgcg@public.gmane.org] wrote:
> Has anyone else tried using ACPI with the C1VN or similiar laptops?  I've
> recently recompiled my kernel, removing APM and enabling ACPI.  Most
> things seem to work fine (battery status and events).  However, when I
> start esd or asd to mix audio streams my processor utilization immediately
> jumps to ~50% and stays there.  The utilization drops to 15-30% if I set
> longrun to maximum rather than variable.  I've verified the utilization
> with both "top" and "gkrellm".  However, while top shows that my CPU is
> not completely idle, it does not list what is using the CPU.  I've tried
> sorting by CPU utilization, but there are still large gaps. As shown by
> the following:

If you stop acpid (if you run it) and cat /proc/acpi/events, do you get
a constant stream of events during this high system load?

I get a constant stream of Processor 0x80 events if I pass the Passive
cooling point on my laptop (Compaq 2701EA) running 2.4.19-020815.
This causes the processor to run hotter and therefore continue to stream
these events.

I need to find what it is which is causing the problem.  I had a look at
processor and thermal, and when the passive point is reached it does
send an acpi event, but if I comment that out, all it does is stop it
going to acpi/events, it doesn't stop whatever is spinning and causing
the high system load and noone on the list has been able to give me
any tips on where to look.

I don't remember the problem occurring when I was using 2.4.18-020709
kernel.  Perhaps Passive cooling "handling" is new since then?


Faye

-- 
Faye Pearson,
Software Development Manager,
ClaraNET Ltd.       Tel 020 7903 3000

Friends may come and go, but enemies accumulate.
		-- Thomas Jones


-------------------------------------------------------
This sf.net email is sponsored by: OSDN - Tired of that same old
cell phone?  Get a new here for FREE!
https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390

Re: C1VN, ACPI and ESD/ASD = const CPU

[Jamin W. Collins]

On Tue, 27 Aug 2002 15:48:26 +0100
Faye Pearson <faye-6JSjyQ0Qj1ReoWH0uzbU5w@public.gmane.org> wrote:

> If you stop acpid (if you run it) and cat /proc/acpi/events, do you get
> a constant stream of events during this high system load?

Nope a "cat /proc/acpi/event" reports nothing.  The laptop doesn't seem to
suffer any ill effects from the reported CPU utilization.  In fact, I
haven't noticed any slow down to speak of.  Below is the output of "acpi
-V" while asd is running:

     Battery 1: charging, 100%
     Thermal 1: ok, 48.0 degrees C
  AC Adapter 1: on-line

After several minutes of more or less idling this looks like:

     Battery 1: charging, 100%
     Thermal 1: ok, 42.0 degrees C
  AC Adapter 1: on-line

So, the temperature isn't really fluxating that much.

-- 
Jamin W. Collins


-------------------------------------------------------
This sf.net email is sponsored by: OSDN - Tired of that same old
cell phone?  Get a new here for FREE!
https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390

Re: C1VN, ACPI and ESD/ASD = const CPU

[Stephen White]

---- Original Message ----
> From Jamin W. Collins <jcollins-u0TlgVUp4POoiH3BShWgcg@public.gmane.org>
> Date: Tuesday, 27 Aug 2002, 15:18
> 
> CPU states:   0.8% user,  25.7% system,   0.0% nice,  73.5% idle
                            ^^^^^^^^^^^^
The marked bit means that 25.7% of the CPU time is being spent inside 
the kernel, this means that it doesn't show up on the process list as 
it is not a normal process causing the load ..  hence you weren't able 
to spot where it was being used even when sorting by CPU usage.

A traditional way to create a high system CPU load would be lots of
network traffic over an encrypted interface (e.g. cipe) .. so the kernel
itself is spending a lot of time doing cryptography.

The fact that ACPI is causing this is a problem.  One thing to try is 
check that you are not getting lots of ACPI events or that you are not 
getting lots of interrupts occuring.  The later can be found with 'cat 
/proc/interrupts' .. run it a few times and compare the results.

I'm not an ACPI expert so I can't suggest much else to look for.

-- 
Stephen White <stephen-acpi-devel-4QvXXjU8Dv4@public.gmane.org>


-------------------------------------------------------
This sf.net email is sponsored by: OSDN - Tired of that same old
cell phone?  Get a new here for FREE!
https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390

Re: C1VN, ACPI and ESD/ASD = const CPU

[Jamin W. Collins]

On Tue, 27 Aug 2002 15:58:43 +0100
Stephen White <stephen-acpi-devel-4QvXXjU8Dv4@public.gmane.org> wrote:

> ---- Original Message ----
> > From Jamin W. Collins <jcollins-u0TlgVUp4POoiH3BShWgcg@public.gmane.org>
> > Date: Tuesday, 27 Aug 2002, 15:18
> > 
> > CPU states:   0.8% user,  25.7% system,   0.0% nice,  73.5% idle
>                             ^^^^^^^^^^^^
> The marked bit means that 25.7% of the CPU time is being spent inside 
> the kernel, this means that it doesn't show up on the process list as 
> it is not a normal process causing the load ..  hence you weren't able 
> to spot where it was being used even when sorting by CPU usage.

Cool, that makes sense.

> A traditional way to create a high system CPU load would be lots of
> network traffic over an encrypted interface (e.g. cipe) .. so the kernel
> itself is spending a lot of time doing cryptography.

Well, as this only occurs when ASD or ESD is running in conjunction with
ACPI, I doubt it's that.

> The fact that ACPI is causing this is a problem.  One thing to try is 
> check that you are not getting lots of ACPI events or that you are not 
> getting lots of interrupts occuring.  The later can be found with 'cat 
> /proc/interrupts' .. run it a few times and compare the results.

I assume I'm supposed to be looking at the CPU0 column for changes?  If
so, I definitely see a drastic increase in the values for interrupts 0 &
9.  The following were grabbed by simply repeatedly running the command in
rapid succession:

**** ASD running ****

jamin-mini:/# cat /proc/interrupts 
           CPU0       
  0:    1189010          XT-PIC  timer
  1:       1219          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  3:       5329          XT-PIC  orinoco_cs
  8:          3          XT-PIC  rtc
  9:    1924005          XT-PIC  acpi, usb-uhci, Ricoh Co Ltd RL5c475,
YMFPCI
 11:         55          XT-PIC  sonypi
 12:       9081          XT-PIC  PS/2 Mouse
 14:      89778          XT-PIC  ide0
NMI:          0 
LOC:          0 
ERR:          0
MIS:          0
jamin-mini:/# cat /proc/interrupts 
           CPU0       
  0:    1189228          XT-PIC  timer
  1:       1219          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  3:       5343          XT-PIC  orinoco_cs
  8:          3          XT-PIC  rtc
  9:    1924413          XT-PIC  acpi, usb-uhci, Ricoh Co Ltd RL5c475,
YMFPCI
 11:         55          XT-PIC  sonypi
 12:       9081          XT-PIC  PS/2 Mouse
 14:      89778          XT-PIC  ide0
NMI:          0 
LOC:          0 
ERR:          0
MIS:          0
jamin-mini:/# cat /proc/interrupts 
           CPU0       
  0:    1189701          XT-PIC  timer
  1:       1219          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  3:       5357          XT-PIC  orinoco_cs
  8:          3          XT-PIC  rtc
  9:    1925301          XT-PIC  acpi, usb-uhci, Ricoh Co Ltd RL5c475,
YMFPCI
 11:         55          XT-PIC  sonypi
 12:       9081          XT-PIC  PS/2 Mouse
 14:      89780          XT-PIC  ide0
NMI:          0 
LOC:          0 
ERR:          0
MIS:          0
jamin-mini:/# cat /proc/interrupts 
           CPU0       
  0:    1190144          XT-PIC  timer
  1:       1219          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  3:       5371          XT-PIC  orinoco_cs
  8:          3          XT-PIC  rtc
  9:    1926132          XT-PIC  acpi, usb-uhci, Ricoh Co Ltd RL5c475,
YMFPCI
 11:         55          XT-PIC  sonypi
 12:       9081          XT-PIC  PS/2 Mouse
 14:      89782          XT-PIC  ide0
NMI:          0 
LOC:          0 
ERR:          0
MIS:          0


**** ASD stopped ****

jamin-mini:/# cat /proc/interrupts 
           CPU0       
  0:    1203690          XT-PIC  timer
  1:       1219          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  3:       5406          XT-PIC  orinoco_cs
  8:          3          XT-PIC  rtc
  9:    1950699          XT-PIC  acpi, usb-uhci, Ricoh Co Ltd RL5c475,
YMFPCI
 11:         55          XT-PIC  sonypi
 12:       9081          XT-PIC  PS/2 Mouse
 14:      89806          XT-PIC  ide0
NMI:          0 
LOC:          0 
ERR:          0
MIS:          0
jamin-mini:/# cat /proc/interrupts 
           CPU0       
  0:    1203843          XT-PIC  timer
  1:       1219          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  3:       5423          XT-PIC  orinoco_cs
  8:          3          XT-PIC  rtc
  9:    1950699          XT-PIC  acpi, usb-uhci, Ricoh Co Ltd RL5c475,
YMFPCI
 11:         55          XT-PIC  sonypi
 12:       9081          XT-PIC  PS/2 Mouse
 14:      89806          XT-PIC  ide0
NMI:          0 
LOC:          0 
ERR:          0
MIS:          0
jamin-mini:/# cat /proc/interrupts 
           CPU0       
  0:    1203985          XT-PIC  timer
  1:       1219          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  3:       5439          XT-PIC  orinoco_cs
  8:          3          XT-PIC  rtc
  9:    1950699          XT-PIC  acpi, usb-uhci, Ricoh Co Ltd RL5c475,
YMFPCI
 11:         55          XT-PIC  sonypi
 12:       9081          XT-PIC  PS/2 Mouse
 14:      89808          XT-PIC  ide0
NMI:          0 
LOC:          0 
ERR:          0
MIS:          0
jamin-mini:/# cat /proc/interrupts 
           CPU0       
  0:    1204728          XT-PIC  timer
  1:       1219          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  3:       5453          XT-PIC  orinoco_cs
  8:          3          XT-PIC  rtc
  9:    1950699          XT-PIC  acpi, usb-uhci, Ricoh Co Ltd RL5c475,
YMFPCI
 11:         55          XT-PIC  sonypi
 12:       9081          XT-PIC  PS/2 Mouse
 14:      89814          XT-PIC  ide0
NMI:          0 
LOC:          0 
ERR:          0
MIS:          0


-- 
Jamin W. Collins


-------------------------------------------------------
This sf.net email is sponsored by: OSDN - Tired of that same old
cell phone?  Get a new here for FREE!
https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390

Re: C1VN, ACPI and ESD/ASD = const CPU

[Carlos Morgado]

On 2002.08.27 15:18:44 +0100 Jamin W. Collins wrote:

> I'm able to run either ESD or ASD with a non-ACPI kernel without these
> symptoms and likewise, running the ACPI kernel without ESD or ASD results
> in normal CPU utilization.
> 

check the irq distribution on an apm and an acpi kernel

-- 
Carlos Morgado - chbm(at)chbm(dot)nu - http://chbm.nu/ -- gpgkey: 0x1FC57F0A
http://wwwkeys.pgp.net/ FP:0A27 35D3 C448 3641 0573 6876 2A37 4BB2 1FC5 7F0A
Software is like sex; it's better when it's free. - Linus Torvalds


-------------------------------------------------------
This sf.net email is sponsored by: OSDN - Tired of that same old
cell phone?  Get a new here for FREE!
https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390

Re: C1VN, ACPI and ESD/ASD = const CPU

[Jamin W. Collins]

On Tue, 27 Aug 2002 16:53:20 +0100
Carlos Morgado <chbm-tNiY1ywYjSU@public.gmane.org> wrote:

> check the irq distribution on an apm and an acpi kernel

ACPI Kernel - 2.4.18 - with ACPI and SWSUP patches
  0:    1204728          XT-PIC  timer
  1:       1219          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  3:       5453          XT-PIC  orinoco_cs
  8:          3          XT-PIC  rtc
  9:    1950699          XT-PIC  acpi, usb-uhci, Ricoh Co Ltd RL5c475,
YMFPCI
 11:         55          XT-PIC  sonypi
 12:       9081          XT-PIC  PS/2 Mouse
 14:      89814          XT-PIC  ide0

APM Kernel - 2.4.18 - stock Debian 586tsc 
  0:      34547          XT-PIC  timer
  1:          5          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  3:        190          XT-PIC  orinoco_cs
  8:          3          XT-PIC  rtc
  9:         31          XT-PIC  usb-uhci, Ricoh Co Ltd RL5c475, YMFPCI
 11:          3          XT-PIC  sonypi
 12:         15          XT-PIC  PS/2 Mouse
 14:      68288          XT-PIC  ide0

With the exception of acpi not existing in the APM kernel, they look
exactly the same.

-- 
Jamin W. Collins


-------------------------------------------------------
This sf.net email is sponsored by: OSDN - Tired of that same old
cell phone?  Get a new here for FREE!
https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390