Looking at the kdb backtrace attached, it looks like you traced pid 7
(kupdated) several times - I assume that you have let the system run a
bit, re-entered kdb, and then re-traced pid 7?
If so, it looks like it's stuck in kupdated, at xfs_imap_to_bmap... but
if you have not left kdb, then you've just retraced the same frozen pid.
Can you let me know which it is?
If you did not hop in & out of kdb, perhaps you can try that once the
system "freezes," to see if the kupdated thread is moving or not.
Also, "bta" output from kdb (backtrace all) might be useful (and
verbose! private email is ok on that)
On Tue, 2003-12-16 at 08:28, Gustavo Rincon wrote:
> Hi, I did the HAVE_SECTOR_T definition in all the xfs Makefiles and the test
> failed in the same
> way that before. Here is some information gotten from kdb (Look into the
> All the help that you can get me it is highly appreciated.
> Thank you
> Gustavo Rincon
> -----Original Message-----
> From: Eric Sandeen [mailto:sandeen@xxxxxxx]
> Sent: Monday, December 15, 2003 5:06 PM
> To: Gustavo Rincon
> Cc: 'linux-xfs@xxxxxxxxxxx'; Tony Lambert; Ben McMillan
> Subject: Re: LBD patch and XFS problem!
> On Mon, 2003-12-15 at 14:46, Gustavo Rincon wrote:
> > Hi, I need some help with XFS and LBD patch.
> Whoohoo, we've just been discussing this... :)
> > I was doing some testing with the linux-2.4.24-pre1+xfs + LBD patch
> > from <http://www.gelato.unsw.edu.au/patches/lbd/>)
> Which LBD patch did you use? There is no 2.4.24-pre1 LBD patch.
> There may already be some problems there, the LBD patch used to make
> itself known via HAVE_SECTOR_T, but that's not there anymore. So now
> the conditional code in XFS for LBD doesn't work, and xfs has no way to
> know that LBD is present. Check your LBD patch for a definition of
> We're probably going to remove all the LBD bits from 2.4 xfs, and push
> the changes into Peter's patch, since xfs is now in 2.4.24 - I'll try to
> get an updated patch to him soon.
> In the meantime, you can probably fix things up by looking for
> #ifdef HAVE_SECTOR_T
> conditionals in XFS, and making them always true (either by removing the
> tests, or defining it yourself in, say, xfs_linux.h)
> That's just off the top of my head, no promises.
> Until that's fixed, I'm not inclined to dig through all the stacks
> posted below. :)
Eric Sandeen [C]XFS for Linux http://oss.sgi.com/projects/xfs
sandeen@xxxxxxx SGI, Inc. 651-683-3102