Thank you for your insight and wisdom of your experiences, Al, Peter, and
Bryan.

I have yet another question.  I'm wondering something similar about i_sem.
What are the implications of intentionally dropping i_sem then re-acquiring
it in a layer between VFS and the underlying file system?  Is i_sem used to
protect state that must not change between the VFS's work and the file
system's work?  Or would it be okay to drop it as long as I re-acquire it
before invoking the file system?  I ask this because I may want to block a
thread in the intermediate layer while a helper thread enters the kernel to
perform operations with the same inode.

If I were to implement something of this nature, I would have to take care
not to invert the prescribed locking hierarchy.  For example, if I drop
i_sem, I must also drop i_zombie, then re-acquire them in the correct order.
Thus, I must extend my question to i_zombie as well.  What are the
implications of dropping and re-acquiring that in an intermediate layer?
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [EMAIL PROTECTED]

Reply via email to