:Another problem is if A is mounting it read-only and then B tries to
:mount it read-write. This succeeds and is dangerous for the same
:reason as the last example. Since A can't write anything to the disk,
:I guess there is no way we can avoid this situation. (The only way I
:could think of avoiding a crash due to stale cache data was to have A
:check the clean flag before every read, but that seems excessively
:expensive.)
:
:Satoshi
You have to be able to mount an unclean filesystem read-only -- otherwise
the system would not be able to mount root and then fsck it from /etc/rc,
nor would you be able to mount a corrupted partition in order to attempt
to recover some of it.
The clean flag was never designed to handle multi-headed configurations
and should not be used for such. The filesystems - even read-only mounts,
are also not designed to handle multi-headed configurations.
You need a separate mechanism ( like a tcp connection or perhaps you can
even use the SCSI reservation stuff manually ) to control access to the
filesystems.
-Matt
Matthew Dillon
<dil...@backplane.com>
To Unsubscribe: send mail to majord...@freebsd.org
with "unsubscribe freebsd-current" in the body of the message
