Sleeping thread (tid 100017, pid 27) owns a non-sleepable lock

[Bernd Walter]

On Fri, Aug 25, 2006 at 01:12:10PM +0200, Olivier Houchard wrote:
> 
> Here, something is deadly wrong. You get a page fault while scheduling the
> interrupt. Most likely something writes where it shouldn't and corrupts the
> ithread structures, or you're mixing INTR_FAST and non-INTR_FAST drivers, but
> reading at your dmesg it doesn't seem so.
> Can you reproduce it, or does it just happen randomly ?

It is reproduceable.
And contrary to the original boot problem I get excactly the same panic
when turning off one SDRAM chip.
But the panic is gone if I use a kernel without my rlswitch driver.
Now ukphy gets connected to all 6 PHY mappings.
Can't imagine at which place rlswitch could have corrupted memory, but
maybe it happens because of refusing multiple attachment.

-- 
B.Walter                http://www.bwct.de      http://www.fizon.de
bernd at bwct.de           info at bwct.de            support at fizon.de