On Wed, 9 Oct 2002, Alexander Viro wrote:
>
> Even aside of the problems with putting filesystems (and filesystem types)
> into driverfs (can_unload() for each fs module?), partitions _ARE_ reused.
> So logics with ->release() will be a killer.
Note that we don't actually do this yet, and when/if we do it it obviously
will require us to have the association data structures in place. And
clearly the rule would have to be that a partition can't be re-used while
a filesystem is busily bound to it - and that has nothing to do with
driverfs/kernel/sysfs/xxxfs.
(That rule pretty much exists already, although we obviously don't
_enforce_ the rule, since we don't even have the data structure linkages
in place).
As to the "can_unload()" thing, I really suspect that the reason it shows
up is because module unloading is fundamentally broken - again regardless
of any driverfs issues. Talk to Rusty some day about it ;)
(Side note: it may be that we could _fix_ module unloading by adding a
driverfs node to modules too, and getting rid of the "single module count"
thing, and replacing it with a more generic "list of nodes using it".
The reason module unloading is so painful largely is exactly the fact that
we only have a count, and the generic module layer has no idea what is
actually _using_ the module - except for other modules. In other words, we
get the inter-module dependencies right, but we don't see any other
dependencies).
Linus
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Oct 15 2002 - 22:00:33 EST