Re: [Xen-devel] [patch 0/6] xenblk: Add O_DIRECT and O_SYNC support.

[Joe Jin <joe.jin@xxxxxxxxxx>]

On 2008-11-04 07:24, Keir Fraser wrote:
> On 4/11/08 02:50, "Joe Jin" <joe.jin@xxxxxxxxxx> wrote:
> 
> > Resend.
> 
> I never saw this patchset before!
> 

Have sent it but could not deliveried to mailing list for I did not 
subscribed the mailing list :)

> > [patch 0/6] xenblk: Add O_DIRECT and O_SYNC support.
> > 
> > This patchset to add O_DIRECT and O_SYNC support to xenblk driver.
> > 
> > At present xenblk driver, a ring buffer have introduced in backend and
> > frontend, the cache got better for io performance. But at some conditions,
> > data need write date to disk as soon as possible, this ring buffer made
> > data could not sync to disk immediately, even open file with O_DIRECT or
> > O_SYNC flag. Obviously some potential risks at os for IO operation,
> > so sync data to disk asap which the file open with O_DIRECT/O_SYNC
> > is meaningful for xenblk driver.
> > 
> > Please review and welcome comment.
> 
> How is this better than waiting in the domU (blkfront) domain to collect the
> required asynchronous responses? Scattering changes into Linux filesystems
> is a dead end unless you can get the changes accepted upstream via lkml.
> 

Do you meant implement this feature just in domU (blkfront)? 
To direct-io answer is yes for request have marked if write request with
O_DIRECT flag, but to O_SYNC, filesystem implentation it at filesystem, 
layer, I think do not touch filesystem codes to implement O_SYNC support 
is impossible.

And, I think the changs related filesystem maybe hard to accepted by upsteam
for this is a special issue just related xenblk.

Thanks,
Joe




_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel