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

[Manfred Spraul <manfred@colorfullife.com>]

Davide Libenzi wrote:

>On Fri, 2 Jan 2004, Manfred Spraul wrote:
>
>  
>
>>Hi Davide,
>>    
>>
>
>Hi Manfred,
>
>
>  
>
>>I think the patch adds unnecessary bloat, and mandates one particular 
>>use of the wait queue info interface.
>>    
>>
>
>why are you saying so?
>
>  
>
sizeof(waitqueue_t) increases.

>@@ -1658,6 +1659,8 @@
> 		unsigned flags;
> 		curr = list_entry(tmp, wait_queue_t, task_list);
> 		flags = curr->flags;
>+		if (info)
>+			dup_wait_info(&curr->info, info);
> 		if (curr->func(curr, mode, sync) &&
> 		    (flags & WQ_FLAG_EXCLUSIVE) &&
> 		    !--nr_exclusive)
>
IMHO these two lines belong into curr->func, perhaps with a reference 
implementation that uses

struct wait_queue_entry_info {
    wait_queue_t wait;
    struct wait_info info;
};

We have already a callback pointer, so why add special case code into 
the common codepaths? Custom callbacks could handle the special case of 
an info wakeup.

--
    Manfred