>>>>> " " == Jeff Darcy <[EMAIL PROTECTED]> writes:

     > From: "Trond Myklebust" <[EMAIL PROTECTED]>
    >> ??? It'll be a completely useless form of lock if programs can
    >> end up blocking forever on a set of (from their perspective)
    >> valid operations.

     > Some would argue that if it could lead to deadlock, then it's
     > not a valid set of operations.

That's true for a single application. What if you have several
operating on the same file? There is no set of rules you can follow to
guarantee compatibility with all application out there.

That's why POSIX specifies deadlock detection as part of the
protocol. The alternative is be to make a more restrictive protocol
such as the case for flock().

     > That applies, of course, only to applications.  Servers acting
     > on behalf of multiple applications might end up issuing such an
     > "erroneous" sequence of operations, and it would be bad for a
     > server thread to lock up as a result.  In that case, wouldn't
     > it be better to implement asynchronous or non-blocking forms of
     > the locking calls than to insist on deadlock detection?

If servers are required to do this, and yet support blocking, then
you'll have to implement a sort of polling mechanism. It's hard to
make that scale.
Of course if you put the onus on the application to do the polling,
that's a different matter.

Cheers,
   Trond
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [EMAIL PROTECTED]

Reply via email to