Why is O_DIRECT checking deferred?

Why is O_DIRECT checking deferred?

[Andreas Dilger]

Andrea, Al,
I'm implementing O_DIRECT for our network filesystem, Lustre.  One
strange thing I noticed is that we do not check for a_ops->direct_IO
until deep in the read/write path instead of failing at open time.

Since not many filesystem implement O_DIRECT I was going to submit a
patch to generic_direct_IO() which printk's an error if a_ops->direct_IO
is not available, but it would really be better to do this at open
time before allocating the kiobuf.

Cheers, Andreas
--
Andreas Dilger
http://www-mddsp.enel.ucalgary.ca/People/adilger/
http://sourceforge.net/projects/ext2resize/

RE: Why is O_DIRECT checking deferred?

[Lever, Charles]

i wrote a patch that fixed this, and sent it to andrea months ago.
andrea, what is the status of getting that patch included?

> -----Original Message-----
> From: Andreas Dilger [mailto:adilger@clusterfs.com]
> Sent: Saturday, June 22, 2002 2:51 PM
> To: andrea@suse.de
> Cc: viro@math.psu.edu; linux-fsdevel@vger.kernel.org
> Subject: Why is O_DIRECT checking deferred?
> 
> 
> Andrea, Al,
> I'm implementing O_DIRECT for our network filesystem, Lustre.  One
> strange thing I noticed is that we do not check for a_ops->direct_IO
> until deep in the read/write path instead of failing at open time.
> 
> Since not many filesystem implement O_DIRECT I was going to submit a
> patch to generic_direct_IO() which printk's an error if 
> a_ops->direct_IO
> is not available, but it would really be better to do this at open
> time before allocating the kiobuf.
> 
> Cheers, Andreas
> --
> Andreas Dilger
> http://www-mddsp.enel.ucalgary.ca/People/adilger/
> http://sourceforge.net/projects/ext2resize/
> 
> -
> To unsubscribe from this list: send the line "unsubscribe 
> linux-fsdevel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 

Re: Why is O_DIRECT checking deferred?

[Andrea Arcangeli]

On Sat, Jun 22, 2002 at 12:50:35PM -0600, Andreas Dilger wrote:
> Andrea, Al,
> I'm implementing O_DIRECT for our network filesystem, Lustre.  One
> strange thing I noticed is that we do not check for a_ops->direct_IO
> until deep in the read/write path instead of failing at open time.
> 
> Since not many filesystem implement O_DIRECT I was going to submit a
> patch to generic_direct_IO() which printk's an error if a_ops->direct_IO
> is not available, but it would really be better to do this at open
> time before allocating the kiobuf.

This patch moves the check at open time:

	http://www.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.19pre10aa4/00_o_direct-open-check-1

Andrea