Yoel Inbar wrote:
>
> Hello,
> I'm using lkcd 3.1.3 on a machine with 640 MB of RAM, and I'm dumping to
> a 512 MB swap partition. Sometimes, crash dumps will fail, and console
> messages like the following will be printed:
> attempt to access past end of device
> 08:02: rw=1, want=13439964, limit=498015
> I do have DUMP_LEVEL=4, so I'm afraid I might be running out of room on
> the partition I'm dumping to. However:
> 1. I also have DUMP_COMPRESS_PAGES=1.
> 2. Dumps don't fail consistently.
> So I think there might be something else going on. Any help would be
> greatly appreciated.
>
> Thanks,
>
> Yoel
Two things:
- I'm going to release a new patch on Wednesday that includes a
few changes, one to throw back in a smp_send_stop() mechanism,
addresses interrupts on CPU 0, turns off schedule(), and states
which CPU is dumping (in dump_execute()).
- There are now a number of individuals that will be working on
LKCD, thanks to developers at IBM, SGI and other locations.
We'll be starting to open up more discussion on this list as to
how to improve LKCD exponentially over its current level. More
on this to follow.
Now, as to what you're seeing, I suspect you are trying to write
beyond the end of your dump device. What's strange is, the return
value from brw_kiovec() wasn't < 0 (in fact, it returned the wrong
value). I'll have to look at this some more, but can you tell me
the size of your dump device? Just 'cat /proc/swaps', if you are
using your swap device.
I'll get on this as soon as I can. :)
--Matt
|