Re: [RFC] bloody mess with __attribute__() syntax

[Al Viro <viro@ftp.linux.org.uk>]

On Sat, Jul 07, 2007 at 03:11:52AM +0100, Al Viro wrote:
> Still not expressive enough...  Consider e.g.
> 
> struct foo *lookup_foo(char *s); // lookup by name, return NULL if failed
> 				 // or pointer to struct foo with ->mutex
> 				 // held.  Caller should unlock.
> 
> It's legitimate, not particulary rare and AFAICS can't be expressed.

Another fun problem: consider a structure with pair of methods - ->start()
and ->stop().  The only locking requirement is that calls are inverse wrt
locking (i.e. foo->start();foo->stop(); leaves the locking state unchanged).
Different instances may deal with different locks, different _kinds_ of
locks or no locks at all.  How do you annotate ->start() and ->stop()?

We have just such a beast in the kernel - seq_operations (see fs/seq_file.c
for code that calls these methods and grep for seq_operations to see users).