Re: aio-core why not using SuS? [Re: [rfc] aio-core for 2.5.29 (Re: async-io API registration for 2.5.29)]

[Benjamin LaHaise <bcrl@redhat.com>]

On Fri, Aug 16, 2002 at 01:54:59AM +0200, Andrea Arcangeli wrote:
> the window a kernel window inside a preempt_disable and a __cli() will
> have a goodness effect in a number of cases, but I don't think it
> matters significantly because you still need some gettimeofday in
> userspace (or clock_gettime if that matters, clock_gettime infact is
> even worse than gettimeofday due its certainly lower resolution).

Yeah, I've come full circle back to the relative timeout point of view.  
By grabbing a copy of jiffies at the beginning of the function the race 
with preempt can be avoided.

> Now reading the SuS specifications I also like less and less our current
> kernel API of this sumbit_io, the SuS does exactly what I suggested
> originally that is aio_read/aio_write/aio_fsync as separate calls. So
> the merging effect mentioned by Ben cannot be taken advantage of by the
> kernel anyways because userspace will issue separate calls for each
> command.

Read it again.  You've totally missed lio_listio.  Also keep in mind what 
happens with 4G/4G split for x86 which are needed to address the kernel 
virtual memory starvation issues.

		-ben
-- 
"You will be reincarnated as a toad; and you will be much happier."