sata related hang with linux-2.6

sata related hang with linux-2.6

[Ramon de Ruiter]

Hello,

About once every three times i power-on my pc, it won't load the OS.
When the kernel is at the point of initializing my harddisk, it just
hangs with error message:

IRQ 10: Nobody cared!
(the following about 20 times and with different codes and messages for 
the "...":)
[c01061da>] ...
Disabling IRQ# 10 ata1: dev 0 ATA, max UDMA/100, 4d88960 sectors.

I'm not able to capture it decently but perhaps i
could make a decent photo of it if necessary.
When this happens, i reset the system and then it boots just fine.

I have a Abit motherboard with Nforce2 chipset and Silicon Image
Sata(raid) controller (CONFIG_SCSI_SATA_SIL) I have a 20G ide disk
attached to the Sata controller via a "serillel" converter and Sata
cable. I don't think the serillel part is the problem since i found
someone else, he's on CC, with native sata who expierences the same error.

I now use the 2.6.9 kernel but the problem is also reproducable with
2.6.7, 2.6.8 and a few kernels i tried before these.

Is such a bug already known, and what can i do to help?

Ramon.

Re: sata related hang with linux-2.6

[Paul Slootman]

Ramon de Ruiter  <won@home.nl> wrote:
>
>About once every three times i power-on my pc, it won't load the OS.
>When the kernel is at the point of initializing my harddisk, it just
>hangs with error message:
>
>IRQ 10: Nobody cared!
>(the following about 20 times and with different codes and messages for 
>the "...":)
>[c01061da>] ...
>Disabling IRQ# 10 ata1: dev 0 ATA, max UDMA/100, 4d88960 sectors.
>
>I'm not able to capture it decently but perhaps i
>could make a decent photo of it if necessary.
>When this happens, i reset the system and then it boots just fine.
>
>I have a Abit motherboard with Nforce2 chipset and Silicon Image
>Sata(raid) controller (CONFIG_SCSI_SATA_SIL) I have a 20G ide disk

IMHO there's something not quite right with the Silicon Image libata
driver; I've encountered this "IRQ xx: Nobody cared!" with such a
controller in a number of (slower) systems. In that case, it never works
for me.  In other systems, it doesn't give the "nobody cared" message
during module loading, and then it works fine.

Perhaps the driver is enabling the hardware to generate interrupts
before setting up the interrupt routine for it? Please excuse my
ignorance if I'm talking nonsense here :-)

I can perform tests if necessary...


Paul Slootman