Need info on acpi, IRQ, vesafb (aka Nobody cared!)

[VinzC]

Hi.

I have a couple of questions so forgive me. I've been struggling with Gentoo on my machine for quite a long time though.

I'm using gentoo-dev-sources-2.6.9-r9. My machine is a dual PIII (900) with an embedded Promise FastTrack (ATA100) controller. I have an Adaptec AHA 7850 SCSI adapter, 2 Realtek 8139 NIC, an nVidia FX5200 and a SB Live! Chipset is VIA 82xxxx. Mother board is an MSI 694D Pro2 Ir. Should be ACPI compliant.

I've recently come across messages like "Nobody cared! - Disabling IRQ #18". IRQ 18 was used by my sound card. I've fixed this problem by appending pci=noacpi to my kernel command line. (now my sound card uses IRQ... 26 .) Is this equivalent to removing acpi support from the kernel? Does that mean I won't profit from powering off with the button or from video suspend features in X?

Next with IRQ steering (is that correct?): I've seen a couple of threads which mentionned the message "Nobody cared!" was mainly due to incorrect handling of IRQ steering, if I've got it right. What's this? Is it usefull? When should I use or *not* use it?

My system is rather heavily "loaded", i.e. almost all my PCI slots are filled. Can it happen that hardware conflicts occur?

If I enable framebuffer console (e.g. for bootsplash), Linux hangs while loading vesafb. Knowing that my nVidia card (GeForce FX5200) shares its interrupt with a Realtek NIC, is it a possible cause for Linux to hang loading vesafb?

Somebody in another thread advised me to disable ACPI. But at that time it didn't solve my problem. I've reinstalled Gentoo since then but I haven't tried again. Would disabling ACPI allow me to use a bootsplash? Or will it be impossible as long as the video card shares its interrupt with another hardware?

Should I remove some devices? Or just swap some of them? Should I get back to an SMP kernel version 2.4?

Code: Select all

# lspci -tv
-[00]-+-00.0  VIA Technologies, Inc. VT82C693A/694x [Apollo PRO133x]
      +-01.0-[01]----00.0  nVidia Corporation NV34 [GeForce FX 5200]
      +-07.0  VIA Technologies, Inc. VT82C686 [Apollo Super South]
      +-07.1  VIA Technologies, Inc. VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE
      +-07.2  VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller
      +-07.3  VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller
      +-07.4  VIA Technologies, Inc. VT82C686 [Apollo Super ACPI]
      +-0c.0  Promise Technology, Inc. PDC20265 (FastTrak100 Lite/Ultra100)
      +-0e.0  Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+
      +-0f.0  Adaptec AHA-7850
      +-10.0  Creative Labs SB Live! EMU10k1
      +-10.1  Creative Labs SB Live! MIDI/Game Port
      \-12.0  Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+

Code: Select all

# cat /proc/interrupts
           CPU0       CPU1
  0:    5584604         36    IO-APIC-edge  timer
  1:      16892          0    IO-APIC-edge  i8042
  2:          0          0          XT-PIC  cascade
  8:          1          1    IO-APIC-edge  rtc
 11:          0          0          XT-PIC  acpi
 12:         67          0    IO-APIC-edge  i8042
 14:      10786          0    IO-APIC-edge  ide0
 15:      49301          0    IO-APIC-edge  ide1
 21:     320077          0   IO-APIC-level  eth0, nvidia
 25:     125531          0   IO-APIC-level  uhci_hcd, uhci_hcd
 26:          0          0   IO-APIC-level  EMU10K1
 27:        803          1   IO-APIC-level  aic7xxx, eth1
NMI:          0          0
LOC:    5584447    5584498
ERR:          0
MIS:          0

Thanks a lot in advance.

[truekaiser]

sorry i can't help but i have to give you a little bit of advice here. when it comes to serious problems people here know jack****. they only thing they do know is how to get xine to play or at the most get a wrapper to work for there wireless card.

[VinzC]

I'm not sure I understand you quite well, truekaiser. What do you mean?

[planet-admin]

Hi.

I have a couple of questions so forgive me. I've been struggling with Gentoo on my machine for quite a long time though.

I'm using gentoo-dev-sources-2.6.9-r9. My machine is a dual PIII (900) with an embedded Promise FastTrack (ATA100) controller. I have an Adaptec AHA 7850 SCSI adapter, 2 Realtek 8139 NIC, an nVidia FX5200 and a SB Live! Chipset is VIA 82xxxx. Mother board is an MSI 694D Pro2 Ir. Should be ACPI compliant.

I've recently come across messages like "Nobody cared! - Disabling IRQ #18". IRQ 18 was used by my sound card. I've fixed this problem by appending pci=noacpi to my kernel command line. (now my sound card uses IRQ... 26 .) Is this equivalent to removing acpi support from the kernel? Does that mean I won't profit from powering off with the button or from video suspend features in X?

Next with IRQ steering (is that correct?): I've seen a couple of threads which mentionned the message "Nobody cared!" was mainly due to incorrect handling of IRQ steering, if I've got it right. What's this? Is it usefull? When should I use or *not* use it?

My system is rather heavily "loaded", i.e. almost all my PCI slots are filled. Can it happen that hardware conflicts occur?

If I enable framebuffer console (e.g. for bootsplash), Linux hangs while loading vesafb. Knowing that my nVidia card (GeForce FX5200) shares its interrupt with a Realtek NIC, is it a possible cause for Linux to hang loading vesafb?

Somebody in another thread advised me to disable ACPI. But at that time it didn't solve my problem. I've reinstalled Gentoo since then but I haven't tried again. Would disabling ACPI allow me to use a bootsplash? Or will it be impossible as long as the video card shares its interrupt with another hardware?

Should I remove some devices? Or just swap some of them? Should I get back to an SMP kernel version 2.4?

Code: Select all

# lspci -tv
-[00]-+-00.0  VIA Technologies, Inc. VT82C693A/694x [Apollo PRO133x]
      +-01.0-[01]----00.0  nVidia Corporation NV34 [GeForce FX 5200]
      +-07.0  VIA Technologies, Inc. VT82C686 [Apollo Super South]
      +-07.1  VIA Technologies, Inc. VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE
      +-07.2  VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller
      +-07.3  VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller
      +-07.4  VIA Technologies, Inc. VT82C686 [Apollo Super ACPI]
      +-0c.0  Promise Technology, Inc. PDC20265 (FastTrak100 Lite/Ultra100)
      +-0e.0  Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+
      +-0f.0  Adaptec AHA-7850
      +-10.0  Creative Labs SB Live! EMU10k1
      +-10.1  Creative Labs SB Live! MIDI/Game Port
      \-12.0  Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+

Code: Select all

# cat /proc/interrupts
           CPU0       CPU1
  0:    5584604         36    IO-APIC-edge  timer
  1:      16892          0    IO-APIC-edge  i8042
  2:          0          0          XT-PIC  cascade
  8:          1          1    IO-APIC-edge  rtc
 11:          0          0          XT-PIC  acpi
 12:         67          0    IO-APIC-edge  i8042
 14:      10786          0    IO-APIC-edge  ide0
 15:      49301          0    IO-APIC-edge  ide1
 21:     320077          0   IO-APIC-level  eth0, nvidia
 25:     125531          0   IO-APIC-level  uhci_hcd, uhci_hcd
 26:          0          0   IO-APIC-level  EMU10K1
 27:        803          1   IO-APIC-level  aic7xxx, eth1
NMI:          0          0
LOC:    5584447    5584498
ERR:          0
MIS:          0

Thanks a lot in advance.

pci=noacpi shouldn't(I stand to be corrected) kill off other power-saving features that are in acpi, but acpi does have this nasty habit of disabling an irq, for no reason that I've ever been able to track down, and sometimes, it only does it upon a warm reboot, which I found odd, keeping in mind acpi, in linux isn't completely up to speed YET, but will be soon.

Vesafb has an alternate nowadays called vesafb-tng, which has support for more hardware, and resolutions, and I highly recommend you try both, also try rivafb support as well. I doubt that it's the IRQ sharing which is causing the lockup.

ACPI shouldn't have any effect on framebuffer, but then, anything IS possible with these monstrous machines, the only way to really find out is to try it, it's as simple as unselecting it in the kernel, and rebuilding. Again, I don't think that IRQ sharing is the cause of your lockup, please post your bootloader config file, and let us take a look at that as well.

Stick with the 2.6.x series kernel, though you might be interested in an experimental kernel, i.e. nitro, love, etc, as those all have some intriguing framebuffer hacks, etc.


Hope at least one of these things helps,
Michael

[VinzC]

Thanks, Michael.

In fact I had done tests with vesa-tng first then with vesa. Both hung. But I'll do the test once again and post my Grub config file whenever I get into trouble again.

I'll recompile my kernel and will make tests without ACPI support first, to be sure. It'll take me a little time and I won't probably be able to post before Wednesday or thursday as I'll be busy next week.

BTW I didn't have that problem with Gentoo Live CD and its SMP kernel (2.6.1). Can I use that kernel on my system to check vesa framebuffer?

One more question: rivafb cannot be used with nVidia acceleration features in X, am I right? HOWTO's say rivafb support should be disabled in the kernel when working with framebuffer.

Thanks a lot for your help.

[planet-admin]

Thanks, Michael.

In fact I had done tests with vesa-tng first then with vesa. Both hung. But I'll do the test once again and post my Grub config file whenever I get into trouble again.

I'll recompile my kernel and will make tests without ACPI support first, to be sure. It'll take me a little time and I won't probably be able to post before Wednesday or thursday as I'll be busy next week.

BTW I didn't have that problem with Gentoo Live CD and its SMP kernel (2.6.1). Can I use that kernel on my system to check vesa framebuffer?

One more question: rivafb cannot be used with nVidia acceleration features in X, am I right? HOWTO's say rivafb support should be disabled in the kernel when working with framebuffer.

Thanks a lot for your help.

The problem you're having could be only with the kernel sources you're using, you might want to try one of the experimentals, i.e. nitro, mm, ck, love, something. You can TRY to boot to the kernel from the live cd, but that's likely going to present a number of problems, and you can't test framebuffer with it unless framebuffer support is already built in, which I'm not certain about, it depends on the live cd really.

As to using rivafb with the nvidia acceleration features, I'm not completely certain on that, but I don't have any problem using the two in conjuction with an old Geforce2 card.

Michael

[mcdermottpa]

The problem you're having could be only with the kernel sources you're using, you might want to try one of the experimentals, i.e. nitro, mm, ck, love, something. You can TRY to boot to the kernel from the live cd, but that's likely going to present a number of problems, and you can't test framebuffer with it unless framebuffer support is already built in, which I'm not certain about, it depends on the live cd really.

As to using rivafb with the nvidia acceleration features, I'm not completely certain on that, but I don't have any problem using the two in conjuction with an old Geforce2 card.

Michael

I've found that using newer revisions of the gentoo-dev-sources when they are put in the Portage tree sometimes fix problems and sometimes break things.

ACPI has always been a bear for me. To get it working on my laptop I did an absolute barebones compile with acpi, checked that the kernel was working without errors, then started adding features (firewire/usb/etc.) over a bunch of compiles. Not everything seems to play well with acpi. I have run into kernel panics dropped ticks (slow clock) and more with some combinations. So I guess I'm just saying YMMV

My 2c.

[mcdermottpa]

sorry i can't help but i have to give you a little bit of advice here. when it comes to serious problems people here know jack****. they only thing they do know is how to get xine to play or at the most get a wrapper to work for there wireless card.

Thought I'd just make an observasion, it seems truekaiser "know[s] jack****" about solving the problems presented. If you have nothing good to say don't, unless your just trolling of course. A lot of us don't claim to be gurus but we do want to help where we can and get help in return.

[VinzC]

You can TRY to boot to the kernel from the live cd, but that's likely going to present a number of problems, and you can't test framebuffer with it unless framebuffer support is already built in, which I'm not certain about, it depends on the live cd really.

And how about extracting the LiveCD's config (less /proc/config.gz), doing make oldconfig and recompiling the kernel?

[VinzC]

So I guess I'm just saying YMMV

I then guess it'll take me between a little more than a short while and a little bit less than an eternity

[mcdermottpa]

So I guess I'm just saying YMMV

I then guess it'll take me between a little more than a short while and a little bit less than an eternity

lol. It took me several hours to get things right the first time. If you can isolate the exact conflict there is probably already a thread or two about it on the forum or some place else (google).

Turning on ccache for non-portage compiling just requires an addition to root's PATH (see howto in Gentoo Documentation) and will speed up kernel recompiles (adding/removing some things like 'premptive kernel' or acpi will force a complete compile with no caching). Also making a few scripts to copy/backup/restore kernel images and configs is helpful too.

[VinzC]

lol. It took me several hours to get things right the first time. If you can isolate the exact conflict there is probably already a thread or two about it on the forum or some place else (google).

Well, it already took me a couple of hours to get something stable. But at least everything I (me, myself and I, not a dark guru from a company earning more money every day) proudly installed is working . Plus another couple of days to fine tune my server. It's hard but thanks to your help I have a clear idea of where to start.

Also making a few scripts to copy/backup/restore kernel images and configs is helpful too.

I have written a small bash script to run menuconfig, kernel compilation and copy with config files. At that time I had written it for 2.4 now it's updated for 2.6. Oo, yes, it does help It just doesn't backup for I do it manually.

[VinzC]

Yesssssss!

Bootsplash now works! See this thread I started once about a specific problem. Thanks to all.

[VinzC]

Yesssssss!

Bootsplash now works! See this thread I started once about a specific problem. Thanks to all.

Noooooo... it was too beautiful. I've been trying and trying and trying some more, nothing worked. I just had to reboot once more to see my efforts vain.

I'm fed up.

[VinzC]

Noooooo... it was too beautiful. I've been trying and trying and trying some more, nothing worked. I just had to reboot once more to see my efforts vain.

I'm fed up.

Finally solved. It was a long way to go but here evrything works. See the thread I mentionned for more details.

Topic, chapter and book closed .