Re: 2.6.17-mm4 raid bugs & traces

[Andrew Morton <akpm@osdl.org>]

On Sat, 1 Jul 2006 13:11:53 +0200
Helge Hafting <helgehaf@aitel.hist.no> wrote:

> Raid-1 is not well in mm4.  Now, my raid devices may
> have some problems after running mm2, because mm2 tends to
> trip up on shutdown, which might cause linux to come up again
> with some failed mirrors. That shouldn't cause mm4 to go BUG
> and spit call traces though.
> 
> When I boot up mm4, dmesg tells me this:
> 
> [...]
> SCTP: Hash tables configured (established 16384 bind 16384)
> powernow-k8: Found 1 AMD Opteron(tm) Processor 244 processors (version 2.00.00)
> powernow-k8: BIOS error - no PSB or ACPI _PSS objects
> BIOS EDD facility v0.16 2004-Jun-25, 4 devices found
> Freeing unused kernel memory: 272k freed
> md: Autodetecting RAID arrays.
> md: autorun ...
> md: considering sde2 ...
> md:  adding sde2 ...
> md:  adding sdd2 ...
> md: sdc1 has different UUID to sde2
> md: sdb5 has different UUID to sde2
> md: sdb1 has different UUID to sde2
> md: sda5 has different UUID to sde2
> md: sda2 has different UUID to sde2
> md: created md2
> md: bind<sdd2>
> md: bind<sde2>
> md: running: <sde2><sdd2>
> md: kicking non-fresh sdd2 from array!
> md: unbind<sdd2>
> md: export_rdev(sdd2)

Yes, Reuben is hitting things like that too.

> BUG: warning at fs/block_dev.c:1109/__blkdev_put()

A flakey lockdep conversion.  -mm5 allegedly fixes this.