ACPI system load

fraris · n00b Joined: 24 Feb 2003 Posts: 7

Hi all,
i have a little problem with my Toshiba Satellite 5200-701.
With kernel gentoo-2.4.20-rc1 and ACPI compiled anything works fine
(i refer to LCD display brightness, Alsa sound support for intel-8x0) but
my system load is always at 100% (about 98% of system time and 2% of user).
This problem disappears if i boot without ACPI... but no control on LCD display is still available and my sound card ... fails to start

anyone experienced the same problem?
is any kernel patch available for this issue ?

thx in advace
/f/

CodeHacker84 · Posted: Sun Mar 09, 2003 4:35 am Post subject: 100% CPU load

I must say that I have experienced the same problem and I don't like the way this looks. I first noticed this happening with the 2.4.20-acpi-r8 kernel, and the same thing happens with r9. I just compiled with the 2.4.20-gentoo-r1 sources and discovered the same thing. If this is a bug, it showed up a long time ago, and it hasn't been fixed yet. I'd give it a shot, but I have zero kernel hacking experience, and I don't even know where to start.

P.S. I'm running one of those horribly-designed Toshiba Satellite 5005-S504's with the "legacy-free" architecture and the indequately-cooled desktop processors, so this is a big issue for me as leaving the damn thing up for too long could kill it.
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

CodeHacker84 · Posted: Sun Mar 09, 2003 4:46 pm Post subject: Just thought of this...

Ok, so we all know that Toshiba laptops have a strange and often impractical design :evil:

could it be that this system load problem is unique to the Toshiba architecture? Or maybe I'm doing something wrong.

In either case, I've not seen this particular problem mentioned before (and yes, my system actually behaves as if it's under 100% load, even though I don't have any processes that are shown to be taking up that much processor time in top) on any of the forums. I suppose that if I get my wlan card working soon I'll devote some time to tracking it down, but I'm not a kernel hacker :cry:

and I would feel a helluva lot more comfortable if we could get someone who knows what they're doing to work on this...
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

avendesora · Posted: Sun Mar 09, 2003 5:07 pm Post subject:

What process/thread is eating up all your CPU?

CodeHacker84 · Posted: Sun Mar 09, 2003 5:16 pm Post subject:

There isn't one, at least according to top. There is just a mysterious drain on the CPU that appears to be coming from nowhere. I'd debug it if I knew where to start looking, but I haven't the faintest clue. I can only guess that it has something to do with a patch that is included in both 2.4.20-gentoo-r1 and all >2.4.20-acpi-r6.
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

avendesora · Posted: Sun Mar 09, 2003 5:21 pm Post subject:

no processes with a high 'TIME' value show up with "ps aux" ?

CodeHacker84 · Posted: Sun Mar 09, 2003 9:09 pm Post subject:

How high is high? The highest time in the list is 1:48 for keventd. The machine's been up almost five hours.
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

CodeHacker84 · Posted: Sun Mar 09, 2003 11:04 pm Post subject:

K, looks like I'm on my own

I'll post some updates periodically. Anyone with ideas, feel free to help. Guess I'm gonna start by trying to cross-reference patches in the kernel ebuilds

maybe I can narrow it down to about fifty likely offenders :cry:

_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

BradN · Posted: Mon Mar 10, 2003 2:39 am Post subject:

Just an idea... can you disable all but the bare minimum ACPI stuff to make your sound card and such work, until the problem can be fixed completely?

CodeHacker84 · Posted: Mon Mar 10, 2003 8:05 pm Post subject:

Ok, well here's the story on MY problem at least...I can't speak for fraris :oops:

. I've compiled the damn kernel with everything compiled as modules except for basic ACPI support (General Setup->ACPI support->[*] ACPI support). None of these modules are listed as being loaded in my lsmod output, but I'm still getting the ghost CPU drain. Can't come up with an explanation...any one else have ideas?

[edit] The modules I'm referring to are the ACPI modules, everything else is compiled the same as I've always compiled it. Just thought some clarification was in order.
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

CodeHacker84 · Posted: Mon Mar 10, 2003 9:26 pm Post subject:

Alright, I've isolated the problem down to one menuconfig option: CONFIG_ACPI. I don't know what this option entails, but one or more of the resulting compiles has a bug

. I'm looking for the source file now...anyone who wants to can feel free to help.
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

CodeHacker84 · Posted: Mon Mar 10, 2003 10:10 pm Post subject:

Alright, I'm gonna do what I should've done about for months ago and submit a bug report to bugzilla. After that, I'm gonna continue to do what I can from my end, I know everyone's busy, maybe I'll find it.
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

avendesora · Posted: Mon Mar 10, 2003 10:12 pm Post subject:

Did you check out http://acpi.sourceforge.net/ before heading that way?
Chances are other people have the same issues.

CodeHacker84 · Posted: Tue Mar 11, 2003 12:00 am Post subject:

No, unfortunately I didn't think of checking there before I annoyed our already overworked bughunters...but after checking there I found one other reference to this problem mentioning something about a high number of interrupts. This person also was using a Toshiba "Legacy-Free" :twisted:

laptop.
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

CodeHacker84 · Posted: Wed Mar 12, 2003 12:36 am Post subject:

Alright, so the debugger guys seem to think it's just bad GKrellm output due to idle calls (at least if I understood the response correctly). I find this a little hard to believe when /proc/acpi/thermal_zone/THRM/temperature reports my CPU running at 71*C and the fan is running full-bore. I STILL think it's a kernel bug. Is there anyone out there with a Toshiba laptop running with a newer kernel version that DOESN'T have the problem described in this thread?
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

snkmoorthy · Guru Joined: 19 Nov 2002 Posts: 376

here is my 1 cent..

is there an APIC in your laptop, is it enabled or disabled in your kernel. look what dmesg says about ACPI. Turn on ACPI debug messages, check those messages.

I do remember another APIC problem, that is APIC turned on in kernel, but no real APIC, the system hangs. However, I have see this issue resolved, with "simulating APIC" in ACPI code. may be you don't want that, may be you do.

CodeHacker84 · Posted: Thu Mar 13, 2003 9:50 pm Post subject:

Ok, well, I've tried both with no success. I have enabled APIC in the kernel, and I have disabled it, but it seems to not have an effect. I've checked my dmesg output and I see things like (but not exactly), "IRQ's 7 and 10 disabled and routed through IRQ 6," but now because of my wireless card and the kernel's "eth1: Error -110 writing Tx descriptor to BAP" messages, I can't see all of my dmesg output anymore.

Also, I've seen something in the ACPI forums about another guy with a Toshiba laptop and he claims that he's getting excessive interrupt requests, which would lead me to believe that there was a change to the ACPI kernel code that now generates these interrupt requests. As for which source code module it is, I don't know, but I have compiled a list of source code modules that have changed since the last version that works, and (by cross-referencing the Makefile(s) ) that are in the right directories to have caused the problem. I'm currently checking through this list for code that makes me suspicious. The problem shouldn't be in /usr/src/linux/drivers/acpi/bus.c.

That's all I know.
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

azote · Guru Joined: 17 Sep 2002 Posts: 415

If I use ACPI on my toshiba laptop 5205-s703 the Nvidia card will not work properly!...
example:

normal boot with acpi : ~500 Frames Per Second

boot with acpi=off: ~3000 Frames Per Second

but the bad thing about the acpi=off .. is that I dont get sound neither wireless going...

paying without sound ... sucks...
_________________
contact me -> azote@mail.i2p

What do you want to emerge today?

if you think that a person is normal, it is because you do not know them well

CodeHacker84 · Posted: Fri Mar 14, 2003 7:33 pm Post subject:

Azote, there's another thing that could affect your video performance like that: a system plagued by extraneous interrupt requests. What does the output of "top" look like?
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

Noth · Posted: Sat Mar 15, 2003 9:40 pm Post subject: Re: ACPI system load

azote · Guru Joined: 17 Sep 2002 Posts: 415

fraris · n00b Joined: 24 Feb 2003 Posts: 7

i'm sorry... but I was out for vacation (!) ...

I read everything on the forum.. and i'll start working in 3 minutes.
Stay tuned... i'll post some results

thx for all

/f/

CodeHacker84 · Posted: Mon Mar 17, 2003 3:57 am Post subject:

Yep, that's about what I figured, azote...I have the same problem. I think that the possibility of excessive interrupt requests could do this, since the processor would be running the code for the interrupt about every 10 nanoseconds. I have a list of all the source files that have changed since the last version of the patch that worked properly, but I don't think I'm qualified to go looking through them for the problem, especially since I have reason to believe that the problem is a patch-kernel interfacing problem. I don't hack kernels...maybe it's time to learn.
_________________
Athlon64 3200
1GB Corsair PC3200 C2
MSI K8T Neo FIS2R
CoolerMaster WaveMaster (Black)
Antec TruePower 550W
120 GB WD SATA Drive
160 GB WD PATA Drive
54x32x54 CD-R/RW
Dual head: ATI 9800 AiW/LG1720P && Voodoo3 PCI/IBM G40 CRT

azote · Guru Joined: 17 Sep 2002 Posts: 415

Good news!!!

I just compiled kernel 2.5.64 !!!

and every thing works great!!

even my GLX acceleration works Great ... I have fixed all of my problems!!!

You really need to trieyed...

I havent got a crash yet.. (is a beta kernel)
the
Cpu(s): 5.0% user, 95.0% system, 0.0% nice, 0.0% idle
is fix! now
_________________
contact me -> azote@mail.i2p

What do you want to emerge today?

if you think that a person is normal, it is because you do not know them well

Liathus · Posted: Fri Mar 21, 2003 5:30 pm Post subject:

I am having the same issue on my toshiba 5105-s501 laptop. The issue appears if I use the current gentoo-sources or the ck-sources. Does anyone have any ideas about this? The stable kernels should work IMHO. I would rather not have to use a beta 2.5 series kernel.