**HEADS UP** FreeBSD-5 boot code has been committed

YONETANI Tomokazu qhwt at myrealbox.com
Fri Nov 14 07:07:24 PST 2003


On Thu, Nov 13, 2003 at 02:01:35PM -0800, Matthew Dillon wrote:
>     Ok, I think the prelinking problem is due to if_fxp's dependancy on
>     miibus.ko.  The boot code is not preloading miibus.ko which causes the
>     kernel to attempt to load it from the not-yet-mounted root filesystem.
> 
>     I will commit fixes to prvent the panic and to figure out why the
>     miibus.ko dependancy is not being preloaded.

I've changed by loader.conf as others suggested and now it boots again.
One more thing I've noticed after having installed the new kernel/world is
that I can't reboot without seeing this:

  Waiting (max 60 seconds) for system thread vnlru to stop...stopped
  Waiting (max 60 seconds) for system thread bufdaemon to stop...stopped
  Waiting (max 60 seconds) for system thread syncer to stop...stopped
  
  syncing disks... 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10
  giving up on 10 buffers

If I drop into DDB(yes, I always compile DDB in the kernel) here, trace
command shows 'col_conv.342(...', which I've never seen before.
Then on the next boot, the system completely freezes in the middle of
fsck'ing /home. If I power-cycle the machine and boot into single-user mode,
and fsck on /home fails again, but this time it says some thing like
'DIRECTORY CORRUPTED', immediately followed by some bold messages

  ad0: READ command timeout tag=0 serv=0 - resetting
  ad0: read interrupt arrived earlyad0: read error detected (too) late

and fsck fails. But if I type fsck -p again here, it doesn't fail.
I'm suspecting that this might be a hardware problem because other people
doesn't complain about this, so I'm going to do some stuffs in this order:

- try a full fsck after booting into single-user mode.
- revert my root partition(which contains most of the base system)
  and compile pre-bootcode change and see if it fails.
- if it does, go and buy a new harddrive.
- if it fails, try another machine.
- if it fails, well, hold my breath and watch the list until things
  gets stabilized :)

Thanks.





More information about the Kernel mailing list