Hello,
since Promise did not answer, I am asking here for help again.
We have a strange problem with the Promise SuperTrak SX6000 that leads
to a kernel oops:
When the system is powered up, the SuperTrak BIOS is initializing the
adapter. If we manually *abort* the initialization, Linux will boot
without problems and we can use the hardware raid.
However, if we let the controller initialze the adapter (that is the
default), the kernel will always Oops when I2O is loaded:
Oops: 0000
Call Trace:
[<c01f7fd6>][<c0107d6d>][<c0107ed6>][<c0105150>][<c0105150>][<c0109d48>][<c0105150>][<c0105150>][<c0105173>][<c01051d9>][<c0105000>][<c0105027>]
Warning (Oops_read): Code line not seen, dumping what data is available
Trace; c01f7fd6 <i2o_pci_interrupt+a/14>
Trace; c0107d6c <handle_IRQ_event+30/5c>
Trace; c0107ed6 <do_IRQ+6a/a8>
Trace; c0105150 <default_idle+0/28>
Trace; c0105150 <default_idle+0/28>
Trace; c0109d48 <call_do_IRQ+6/e>
Trace; c0105150 <default_idle+0/28>
Trace; c0105150 <default_idle+0/28>
Trace; c0105172 <default_idle+22/28>
Trace; c01051d8 <cpu_idle+40/54>
Trace; c0105000 <_stext+0/0>
Trace; c0105026 <rest_init+26/28>
This error is reproducable with all current kernels (2.4.9, 2.4.14, 2.4.17).
There are no other Promise controllers in the system. Changing PCI slots
or reassigning IRQs doesn't help either.
My guess is that the i2o module tries to initialize the board. When it
already was initialized by the BIOS, the system crashes.
Sebastian
> When the system is powered up, the SuperTrak BIOS is initializing the
> adapter. If we manually *abort* the initialization, Linux will boot
> without problems and we can use the hardware raid.
>
> However, if we let the controller initialze the adapter (that is the
> default), the kernel will always Oops when I2O is loaded:
Please try 2.4.18pre3-ac first of all. That has one small detail changed
that may matter.
> My guess is that the i2o module tries to initialize the board. When it
> already was initialized by the BIOS, the system crashes.
At the moment I've no idea. The i2o code is entitled to reinit the board
should it want to, and the supertrak 100 certainly works with 18pre3-ac3.
Alan Cox wrote:
> Please try 2.4.18pre3-ac first of all. That has one small detail
> changed that may matter.
Alan,
thanks for the quick reply. The good news: 2.4.18pre3-ac2 works.
I now get the message
"i2o/iop0 reset rejected, trying to clear"
(this message does not appear when initialization by the BIOS was
aborted manually at boot time.)
Whatever the problem is, the kernel seems to be able to handle the exception now.
Thanks again,
Sebastian