OOPS at mount

[David Woodhouse]

On Wed, 2007-04-25 at 19:39 +0200, Joakim Tjernlund wrote:
> On Wed, 2007-04-25 at 17:07 +0100, David Woodhouse wrote:
> > I've just committed something to the git tree which should fix the
> > symptoms --and I think it should apply OK to 2.6.20 which predates other
> > improvements in read_inode(). 
> > 
> > We should still work out the _cause_ though. Once you've confirmed that
> > this patch avoids the crash, you'll know the inode number (because it'll
> > bitch about it). If you don't want to wait for a full
> > CONFIG_JFFS2_FS_DEBUG=1 run, can you enable all the output in
> > jffs2_get_inode_nodes() _just_ for that inode, and show me what nodes
> > are found?
> 
> Seems to work, at least I can boot now but I get a few compliants:

Hm, so you've lost a _lot_ of data. I don't suppose there's any chance
they've been putting new images at the start of the flash without
actually _erasing_ the remainder of the space?

Are you getting lots of messages about bad CRCs?

This is with JFFS2 from 2.6.20, right? Not a bug in the read_inode code
I just committed a couple of days ago?

> Wonder how the lab manged to get that many corrupted nodes?
> One thing that is rather new in our system is that we trigger GC by
> sending HUP to the GC thread from a script in user space at startup
> and then every 24 hours.

That shouldn't make any difference. 

-- 
dwmw2