What is the correct/recommended way for one module to "plug in to" or
use another module? e.g. if I have one module (call it A) which
exports a symbol allowing a different module (call it B) to register
(and unregister) a function pointer (which gets called by A if it has
a valid value), it would be very good if A cannot unload while B is
using it (registered), and also that B can unload without A being
aware it has unregistered. The latter can simply be achieved by having
B call the unregister function of A when it unloads. The former I
suppose could be done via some sort of refcount scheme, but I'm sure
this type of problem has a proper correct solution in the Linux kernel
(e.g. all the modules that use the i2c module etc. etc.), and I'd
obviously like to avoid re-inventing the wheel (badly).
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ
