O_DIRECT, fdatasync_area, mwrite

O_DIRECT, fdatasync_area, mwrite

[Reza Roboubi]

Linus Torvalds:
 > I suspect that this is about a few hundred lines of code (and a lot of
 > testing). And you can emulate O_DIRECT behavior with it, along with
 > splice (only for page-cache entities, though), and a lot of other
 > off-by-one uses.
( http://lwn.net/2002/0516/a/lt-async.php3 )

I have often read this message and wondered why no one has implemented 
it yet.  We would love to have such clean features instead of O_DIRECT. 
  At this very moment it's a bit hard for us to do this(given a 
prerequisites learning curve amongst other factors.)  We may at some 
point do it and I'd love to contribute.  What _are_ the issues with it, 
  and is there a technical reason it hasn't been implemented yet by anybody?

Further, is O_DIRECT to raw partition also unclean?

Thanks.

Reza.
-- 
Reza Roboubi
Chief Technical Officer
http://www.EarthDetails.com

Re: O_DIRECT, fdatasync_area, mwrite

[Reza Roboubi]

Reza Roboubi wrote:
> Linus Torvalds:
>  > I suspect that this is about a few hundred lines of code (and a lot of
>  > testing). And you can emulate O_DIRECT behavior with it, along with
>  > splice (only for page-cache entities, though), and a lot of other
>  > off-by-one uses.
> ( http://lwn.net/2002/0516/a/lt-async.php3 )
> 
> I have often read this message and wondered why no one has implemented 
> it yet.  We would love to have such clean features instead of O_DIRECT. 
>  At this very moment it's a bit hard for us to do this(given a 
> prerequisites learning curve amongst other factors.)  We may at some 
> point do it and I'd love to contribute.  What _are_ the issues with it, 
>  and is there a technical reason it hasn't been implemented yet by anybody?
> 
> Further, is O_DIRECT to raw partition also unclean?

I know the top guys are very busy;  but I'd appreciate just a brief 
comment on the present situation.

Reza.