On Monday 21 January 2008 00:00:52 Tejun Heo wrote:
quoted text
> What should be do are
>
> * Check that the threadfn's argument fits into void *.

For everything but timer, you'll get a warning if the data isn't assignable to 
a void *, so you get a warning if you use a non-pointer already.

But it would be cool to allow functions which take an unsigned long.  To do 
this, I think that would need to be a special case for the data arg (which 
we'd really want to wrap in a macro), like:

	/* If fn expects an unsigned long, cast the data.  If not, we'll
	 * get a warning if data is not void * compatible. */
	__builtin_choose_expr(__builtin_compatible_p(typeof(1?(threadfn):NULL),
			      int (*)(unsigned long)),
			      (void *)(unsigned long)(data), (data))

quoted text
> * Trigger overflow in implicit constant conversion warning if the
> specified data is too large for the argument type.

Hmm, u64 on 32-bit platforms?  I think that will fail the above test: the type 
of the function ptr will be "int (*)(u64)" and so you'll end up passing data 
(a u64) to a void * argument, which will elicit a warning...

I'll test this out and see what I can make...

Thanks!
Rusty.
--
unsubscribe notice
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/