Bug 146696 - Toshiba Tecra A4 model PTA40E ACPI issue
Summary: Toshiba Tecra A4 model PTA40E ACPI issue
Status: RESOLVED DUPLICATE of bug 146098
Alias: None
Product: SUSE Linux 10.1
Classification: openSUSE
Component: Kernel (show other bugs)
Version: unspecified
Hardware: i686 SuSE Linux 10.1
: P3 - Medium : Major (vote)
Target Milestone: ---
Assignee: Thomas Renninger
QA Contact: E-mail List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2006-01-30 17:27 UTC by Pietro Dania
Modified: 2006-04-04 10:43 UTC (History)
0 users

See Also:
Found By: Beta-Customer
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments
Output of hwinfo: acpi enabled and acpi=off (81.31 KB, application/x-gzip)
2006-02-09 08:55 UTC, Pietro Dania
Details
Output of acpidmp (130.25 KB, text/plain)
2006-03-20 15:59 UTC, Pietro Dania
Details
Outputo of "cat /proc/interrupts" (657 bytes, text/plain)
2006-03-20 16:00 UTC, Pietro Dania
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Pietro Dania 2006-01-30 17:27:59 UTC
Booting with ACPI enable makes Ethernet, WLAN and PCMCIA not working.
Booting with acpi=off lets said hardware work, but kills battery though no CPU speedstepping is done, no suspend, no software poweroff.
Comment 1 Pietro Dania 2006-01-30 17:30:58 UTC
PS. no sound in both scenarios :(((
Comment 2 Greg Kroah-Hartman 2006-02-09 01:38:06 UTC
Does 10.1 beta 3 work better for you?

And can you attach the output of the hwinfo program?
Comment 3 Pietro Dania 2006-02-09 08:55:21 UTC
Created attachment 67134 [details]
Output of hwinfo: acpi enabled and acpi=off
Comment 4 Pietro Dania 2006-02-09 09:00:20 UTC
Can't try 10.1 beta3: PC is Production. 
Will try just booting the installer: if network work with acpi enable, I'll know.
Comment 5 Greg Kroah-Hartman 2006-02-09 21:53:03 UTC
Please let us know.
Comment 6 Pietro Dania 2006-02-10 09:17:46 UTC
I've tried the 10.1 beta 3 Rescue Environment from CD1:
same behaviour...
boot without options: no ethernet (supposedly, same for PCMCIA and WLAN);
boot with acpi=off: ethernet works perfectly
Comment 8 Thomas Renninger 2006-02-22 18:04:02 UTC
pci=noacpi should be the workaround you search.
If this does not work try noapic or both.

If you still have your machine booted you could attach /proc/interrupts, this might help later...
If you boot without acpi=off, please run acpidmp >/tmp/acpidmp and attach the output.

It is worth to also test 10.1 Beta4. All the ACPI patches of the last months appear there, maybe this is already fixed ... if it is an easy fix we could even backport it to 10.0.
Please also post dmesg of Beta4 insallation kernel, do you know how to save it to disk?

Andi, you proposed apic some kernels ago to get interrupts work. Is that because default changed to noapic on UP or has this been reverted, does apic boot option still make a difference?
Comment 9 Pietro Dania 2006-03-20 15:59:52 UTC
Created attachment 74010 [details]
Output of acpidmp
Comment 10 Pietro Dania 2006-03-20 16:00:40 UTC
Created attachment 74011 [details]
Outputo of "cat /proc/interrupts"
Comment 11 Pietro Dania 2006-03-20 16:06:24 UTC
Installed 10.1 beta6 and *EVERYTHING* worked nicely.
Upgraded to 10.1 beta8 and AGAIN NO ETHERNET!
AGAIN need acpi=off to get ethernet
AGAIN pci=noacpi doesn't help

Attached required output of acpidmp WITHOUT acpi=off and output of "cat /proc/interrupts": curiously, the count for IRQ 11 (usb, PCMCIA and network) hangs at 100000
Comment 12 Pietro Dania 2006-03-20 16:09:44 UTC
Moved BUG to SUSE 10.1 and raised priority
Comment 13 Thomas Renninger 2006-03-21 08:20:02 UTC
Can you try the "lapic" boot parameter, please.
Comment 14 Pietro Dania 2006-03-22 09:09:49 UTC
I've tried the "lapic" parameter:
no joy :(
I got the same IRQ 11 hang at 100000.
Comment 15 Thomas Renninger 2006-03-29 12:44:27 UTC
Summary:
booting without apci=off
-> no pcmcia, no network?
Or does it work until IRQ 11 get's it's 100000'th interrupt?

Hmm, this is /proc/interrupts with acpi:
    0:      39212          XT-PIC  timer
    1:        180          XT-PIC  i8042
    2:          0          XT-PIC  cascade
    3:          6          XT-PIC  serial
    4:          7          XT-PIC  serial
    6:          3          XT-PIC  ohci1394
    8:          2          XT-PIC  rtc
    9:         14          XT-PIC  acpi
   10:       4390          XT-PIC  Intel ICH6, uhci_hcd:usb1, ehci_hcd:usb5
   11:     100000          XT-PIC  uhci_hcd:usb2, uhci_hcd:usb3, uhci_hcd:usb4, eth0, yenta, ipw2200
   12:        119          XT-PIC  i8042
   14:      20317          XT-PIC  libata
   15:        790          XT-PIC  libata

and this is /proc/interrupts without acpi:
    0:     151527          XT-PIC  timer
    1:        815          XT-PIC  i8042
    2:          0          XT-PIC  cascade
    3:         31          XT-PIC  serial
    4:         28          XT-PIC  serial
    8:          2          XT-PIC  rtc
   10:       1443          XT-PIC  Intel ICH6
   11:      31914          XT-PIC  uhci_hcd:usb1, uhci_hcd:usb2, uhci_hcd:usb3, uhci_hcd:usb4, ehci_hcd:usb5, eth0, yenta, ipw2200, ohci1394
   12:       1745          XT-PIC  i8042
   14:      24832          XT-PIC  libata
   15:       3989          XT-PIC  libata
  NMI:          0 

Could you also post dmesg at least without apci=off, best both, please.

Adding olh (USB) and kkeil (network), maybe they can comment whether this is an (known?) IRQ problem.
Comment 16 Pietro Dania 2006-03-29 12:50:48 UTC
IRQ 11 gets to 100000 almost immediately
Comment 17 Thomas Renninger 2006-03-30 08:50:07 UTC
I see it:
You get this line with default kernel (non-working, right?):
<6>IO/L-APIC allowed because system is MP or new enough

and this line with debug kernel (working):
<6>Using APIC driver default

This is probably because of the different APCI configs set for default:
CONFIG_X86_UP_APIC=y
CONFIG_X86_APIC_OFF=y
CONFIG_X86_UP_IOAPIC=y
CONFIG_X86_LOCAL_APIC=y
CONFIG_X86_IO_APIC=y

and debug kernel:
CONFIG_X86_LOCAL_APIC=y
CONFIG_X86_IO_APIC=y

However the APIC configs are quite fragile... CC'ing Andi for help:
Andi: Here debug kernel works using APIC initialisation through mach-generic/probe.c (search for "Using APIC driver default")

and default kernel does not work (going through CONFIG_X86_APIC_OFF in arch/i386/kernel/apic dmi_check_apic (search for "IO/L-APIC allowed because system is MP or new enough").

Andi: Shouldn't we adjust the debug kernel configs to the default and then try to find a solution for this one to initialise the way it currently does with the debug kernel?
Comment 18 Andreas Kleen 2006-03-30 10:52:21 UTC
No, the debug kernel has to be a smp kernel. SMP kernels don't use UP_IOAPIC
They work slightly differently in how they initialize the APICs.
SMP always has APICs on

Why default doesn't work - might be a dup of #162076
Does it work with "apic" ?
Comment 19 Thomas Renninger 2006-03-30 12:27:13 UTC
Ohh sorry, I mixed up bugs/attachements forget comment #17, this comment should go into bug #156115, I will paste it there again....

Pietro, could you try to move yenta,sky2 and ipw2200 modules away from /lib/modules/${kern_ver}/kernel/drivers/...
Does USB work then (maybe it's possible to hit the 100000 border with a USB disk or whatever).
Try to somehow setup each of the devices that use IRQ11 and try to find out whether one of the drivers cause this(I somehow expect sky2, not sure at all, I just remember there were some bugs there recently?).
Comment 20 Thomas Renninger 2006-04-04 10:43:13 UTC
We have a duplicate... It's not the sky2 driver...

*** This bug has been marked as a duplicate of 146098 ***