Re: [rfc/patch] wake_up_info() draft ...

[John Gardiner Myers <jgmyers@speakeasy.net>]

Minor issues:

I don't know why dup_wait_info() returns a value--it is always ignored.  
If duping can fail, the situation is not particularly recoverable.

I don't like that the dup method is responsible for copying the dup and 
dtor members of struct __wait_info.  It would be simpler for the common 
code in dup_wait_info() to always copy the dup and dtor function pointers:

void * (*dup)(void *);

static inline void dup_wait_info(wait_info_t *s, wait_info_t *d)
{
	close_wait_info(d);
	*d = *s;
	if (s->dup)
		d->data = s->dup(s->data);
}

I prefer the style where assignment functions, such as dup_wait_info(), 
place the destination argument to the left of the source, to mimic the 
assignment operator and functions such as strcpy().

remove_wait_queue_info() could be optimized slightly by transferring 
ownership of the wait queue info data instead of duping it.