Subject: Re: direct I/O again
To: None <firstname.lastname@example.org>
From: Juergen Hannken-Illjes <email@example.com>
Date: 03/27/2006 18:53:45
On Mon, Mar 27, 2006 at 07:57:50AM -0800, Chuck Silvers wrote:
> hi folks,
> I've finished the initial implementation of direct I/O,
> there's an updated diff in:
> this version does not allow concurrent writes to files, it's only
> intended to provide the ability to prevent caching fs data in the
> kernel. I'll add support for concurrency of non-overlapping writes
> in a separate change.
> if no one sees any problems with this part of the work,
> I'll check it in (along with associated doc changes) later this week.
* We only support direct I/O to user space for now.
Any chance to remove this restriction? fss(4) and vnd(4) would be
candidates to use it from kernel space.
Juergen Hannken-Illjes - firstname.lastname@example.org - TU Braunschweig (Germany)