Re: [PATCH -mm 5/5][AIO] - Add listio syscall support

["Sébastien Dugué" <sebastien.dugue@bull.net>]

On Wed, 24 Jan 2007 10:52:58 +0100 Laurent Vivier <Laurent.Vivier@bull.net> wrote:

> Andrew Morton wrote:
> > On Wed, 17 Jan 2007 10:55:54 +0100
> > Sébastien Dugué <sebastien.dugue@bull.net> wrote:
> > 
> >> +struct lio_event *lio_create(struct sigevent __user *user_event,
> >> +			int mode)
> >> +{
> >> +	int ret = 0;
> >> +	struct lio_event *lio = NULL;
> >> +
> >> +	if (unlikely((mode == LIO_NOWAIT) && !user_event))
> >> +		return lio;
> >> +
> >> +	lio = kzalloc(sizeof(*lio), GFP_KERNEL);
> >> +
> >> +	if (!lio)
> >> +		return ERR_PTR(-EAGAIN);
> >> +
> > 
> > Why EAGAIN and not ENOMEM?
> 
> According to the POSIX AIO specifications:
> 
> "ERRORS:
> ...
> [EAGAIN]
>     The resources necessary to queue all the I/O requests were not available. "
> 
> I think memory is "a resource necessary to queue all the I/O requests"...
> 
> http://www.opengroup.org/onlinepubs/009695399/functions/lio_listio.html
> 
> Regards,
> Laurent

  Thanks Laurent,

  just forgot the whys here.

  Sébastien.