kern/62228: Kernel improperly identifies partition sizeofconcat
(grown) vinum volume
Bart Kus
eo at shell-server.com
Wed Feb 11 14:21:25 PST 2004
On Thu, 12 Feb 2004, Greg 'groggy' Lehey wrote:
> You can use the command 'i loc' to show all local variables.
Thanks, noted! I'll have to re-generate a core.
> I've taken a look at this PR, and I'm not sure quite what to do. As
> you've seen, le@ is currently looking at Vinum bugs, but I'm not even
> sure this is a Vinum bug; it could very well be growfs. There are
> also undoubtedly bugs remaining in fsck. If you can find a spare disk
> and install 4.9 on it, then use it to fsck this volume, the results
> could be interesting.
I did re-install the machine. Cut the disk down the middle, used
the 1st 1/2 for 4.9, ran a (semi-ugly) fsck, and got the file system back
to proper order. Have moved gigs and gigs onto it since then.
I can use the 2nd 1/2 of the machine to install 5.2 and keep
working on the PR. What should my next step be? Re-install 5.2, try an
fsck on the partition? If it core dumps, should I do anything more?
(Will dumpfs help here at all?) Maybe I should try a dd
if=/dev/vinum/media of=/dev/null under 5.2 to make sure the whole disk is
readable (can verify it reads the whole disk by counting records read).
If it can read the whole disk (as I suspect it can, since vinum reports
the right volume size), how could I generate more data on the file system
status itself?
I'll hold off posting a PR update until I get a reply to this
email, I think. No sense posting an update devoid of useful data.
--Bart
More information about the freebsd-bugs
mailing list