Bug 112941

Please open such bugs on Product SUSE LINUX 10.0. 

Linuxrc problem.

No. Kernel problem with ahci module. 
It might be interesting to know the last kernel messages on console 4. 

The last lines on console 4 are:

<7>libata version 1.11 loaded.
ahci version 1.01
ahci: probe of 0000:00:1f.2 failed with error -12

Sorry for calling console 4 for screen 3 before, i'm a rookie at bugreports...  :-)

Probably a BIOS/ACPI issue. Can you switch to a console and do:

# lspci -v -s 00:1f.2

and paste that here? Thanks!

can't find the lspci program at this point, any suggestions where to look for it?

Hmm it should be there, are you switching to console 2 and running it from there?

No, I only get a shell promt at console 9. I can run ls, cd and every other
command under the /lbin directory, i have ls'ed around the filesystem without
finding lspci. Where at this part of the installation is the cd mounted? I
should be able to find lspci on the cd shouldn't I?

In case it isn't there, you can scp it from another machine on your network (I
guess the one from 9.3 should work) if you modprobe the module for your network
card and ifconfig (or dhcp) the IP on it afterwards. I've done that myself from
console 2 many times in the past :-)

Oh, and if you happen to have another Linux installed on the box already, you
can of course just run the lspci command from there.

Thanks for the tip, i got Knoppix 3.81 to boot in failsafe mode and this is the
lspci output. (not cut and paste thou)

0000:00.1f.2 IDE interface: Intel Corp. 82801FBM (ICH6M) SATA Controller (rev
04) (prog-if 80 [Master])
Subsystem: Acer Incorporated [ALI]: Unknown device 0070
Flags: bus master, 66Mhz, medium devsel, latency 0
I/O ports at <unassigned>
I/O ports at <unassigned>
I/O ports at <unassigned>
I/O ports at <unassigned>
I/O ports at <unassigned>
Capabilities: [70] Power Management version 2

Hope that is to some help.

Device doesn't have any resources assigned to it. See if you have any BIOS
settings for the sata controller - perhaps you can get it to work by changing
modes or otherwise modifying pnp settings?

Sorry for the delay but i've been busy with other things.
I just tryed the beta4 version with the same result as before.
I don't have any settings in my bios that i can change for the SATA controller.

I tryed the graphical installation again and now with a new error, before it was
completely dead now i get a small screen on the upper left with looking like this.

/----pstk----------rstk--------\
| 0: 51b11.    c| :            |
| 1: 91285658. 2| :            |
| 2: 91285658. 2| :   2a05.15  |
| 3:        0. 1| :   2a05. 5  |
|------------------------------|
| err 8                        |
| ip 385:         f.7          |
\------------------------------\

When i press any key it goes inte something looking like a mixture of text and
graphical install mode, i get the possibility to change the resolution with the
F keys. But as soon as i press any key again it hangs.
And i can't get into any console screens.

Have you checked for a BIOS update for your machine?

The settings may not have anything to do with the sata controller in particular,
look for acpi/pci settings as well. Your controller doesn't have any resources
assigned.

Sorry to say that my BIOS version is the latest and i don't have any settings
for acpi/pci in my BIOS either...

What kernel is knoppix 3.81 using? Did SUSE 9.3/2 work for you?

This is a new laptop for me so i have not installed any other release of linux
on it before, just had XP on it...  :-(

The Knoppix 3.8.1 runs with kernel 2.6.11 and I'm just trying the install of 9.3
and it looks like it's going ok. I had to start the install in failsafe mode to
get started, the graphics was all messed up otherwise. The installation went
fine but i still have som issues with the graphics, i'll try to fix that tomorrow.

I guess i can make an update to opensuse 10 beta4 from yast if everything works
ok and i can get online to get the update...  :-)

I would like to confirm this bug on my Travelmate 8101 with Beta 4.
Right now I'm running SuSE 9.3 with kernel 2.6.11.4-21.8-default. The relevant
section of dmesg is:

libata version 1.10 loaded.
ahci version 1.10
ACPI: PCI Interrupt Link [LNKD] enabled at IRQ 11
ACPI: PCI interrupt 0000:00:1f.2[B] -> GSI 11 (level, low) -> IRQ 11
ahci: probe of 0000:00:1f.2 failed with error -12
ata_piix version 1.03
ACPI: PCI interrupt 0000:00:1f.2[B] -> GSI 11 (level, low) -> IRQ 11

In addition I tried several other kernels up to 2.6.13-rc6-git11-default. In
principal they are all working as long as noapic is turned on. However, I don't
use any of these kernels because a) they don't work with the hotplug system and
b) there is a problem with the atkbd.c and/or the synaptics driver which
sometimes locks up the system during boot or keeps X from starting up. The
relevant section of dmesg for the mentioned kernel is:
libata version 1.11 loaded.
ahci version 1.01
ACPI: PCI Interrupt Link [LNKD] enabled at IRQ 11
PCI: setting IRQ 11 as level-triggered
ACPI: PCI Interrupt 0000:00:1f.2[B] -> Link [LNKD] -> GSI 11 (level, low) -> IRQ 11
ACPI: PCI interrupt for device 0000:00:1f.2 disabled
ahci: probe of 0000:00:1f.2 failed with error -12
ata_piix version 1.03
ACPI: PCI Interrupt 0000:00:1f.2[B] -> Link [LNKD] -> GSI 11 (level, low) -> IRQ 11

As ypu can see, although the error -12, occurs for both kernels, none of them
stop booting.

Frank.

Andi, this looks like pci/acpi issues, any chance you can take a look?

-12 is ENOMEM.

This cannot be in failsafe mode because failsafe mode disables ACPI. Can 
you add a full boot log for the failure? 

Please add acpidmp output too.

Anyways - ahci only returns -ENOMEM in a few cases (when kmalloc /
dma_alloc_coherent / ioremap) fail. The link disabling message just happens
in the cleanup of that.

So I don't think ACPI is to blame here. Or at least I don't see anything
wrong with what it does. You rather need to find out which
of the 6 ENOMEM places in ahci.c is failing and why it does that.

Back to Jens.

Frank, I'll add a debug patch for you to try.

Patch has been committed, I'll ask you to test a specific KOTD when it is
available. You should be able to install that kernel on your 9.3 for testing,
you probably want to back up your old kernel so you can recover it booting from
the 9.3 rescue CD or so.

Please try a KOTD kernel-default-2.6.13-20050906081835.i586.rpm or newer and
check dmesg for the ahci error when it fails.

(In reply to comment #24)
> Please try a KOTD kernel-default-2.6.13-20050906081835.i586.rpm or newer and
> check dmesg for the ahci error when it fails.

Due to a trip I'm not going to be able to test that before the weekend, will
this patch be included in the RC1 to be released on thursday?

The most recent kernel I can find on ftp.suse.com right now is
kernel-default-2.6.13-20050904232747.i586.rpm. However, to save everybody some
work. There are two more bug reports dealing with that problem: 113778 and
115432.  They both tracked it down to the yenta_socket module which is failing
just before the ahci error. The kernel also boots for me now with
BrokenModules=yenta_socket

As a side comment. There are three different people assigned to the same
problem. That is a waste of human resources. But also bug reporters should
search the database before submitting. However this thread was the first one
dealing with that problem ;-)

Thanks for your efforts,
Frank.

Ah indeed, thanks for noticing that! I'm marking this one as a duplicate.

*** This bug has been marked as a duplicate of 113778 ***