Messages in this thread | | | Subject | Re: ioremap panic | From | Andi Kleen <> | Date | Wed, 20 Aug 2008 12:08:05 +0200 |
| |
Andrew Morton <akpm@linux-foundation.org> writes: > > Please work out what address qla2x00_probe_one() passed into ioremap(),
Note that ioremap doesn't get a virtual address, but a bus/phys one. The virtual address checked in the BUG above comes internally out of vmalloc's address space allocation.
> then see if you can work out what was already at that virtual address > (/proc/iomem, dmesg output, etc).
The bug is more likely that either the kernel page tables or the vmalloc vmlist are inconsistent somehow. In particularly it looks like vmalloc gave out an address for ioremap on which there was already a mapping.
I doubt it is qla2x00's fault.
-Andi
| |