[PATCH v2 30/58] staging: lustre: create striped directory

James Simmons jsimmons at infradead.org
Tue Jul 26 18:09:33 UTC 2016


> On Tue, 2016-07-26 at 12:36 -0400, James Simmons wrote:
> > From: wang di <di.wang at intel.com>
> > 
> > 1. client send create request to the master MDT, which
> >   will allocate FIDs and create slaves. for all of slaves.
> > 
> > 2. Client needs to revalidate slaves during intent getattr
> >    and open request.
> > 
> > 3. lmv_stripe_md will include attributes(size, nlink etc)
> >    from all of stripe, which will be protected by UPDATE lock.
> >    client needs to merge these attributes when update inode.
> > 
> > 4. send create request to the MDT where the file is located,
> >    which can help creating master stripe of striped directory.
> > 
> > Changelog
> > 
> > v1) Original submitted patch
> > 
> > v2) fixed lmv_hash_fnv1a function to use do_div64 to fix
> >     __umoddi3 undefined bug on 32 bit platforms.
> > 
> > Signed-off-by: wang di <di.wang at intel.com>
> > Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3529
> > Reviewed-on: http://review.whamcloud.com/7196
> 
> Updated 	2 years, 5 months ago
> 
> how many of these old changes are yet to be integrated?

For the lnet/libcfs layer it is the lastest code so none.
For the lustre client code the further back you go the
more older patches are missing. The last 6 months I have
been filling in those gaps. Currently we are at the 2.5.58
version with this last code drop. As we approach the latest
release way less patches are needed to get caught up. Once
we get to the 2.6+ version of lustre the number of missing
patches should drop off greatly. Thats not to far away now.
I would estimate a few hundred patches are missing. Mind you
this is considering that larger patches need to be broken up.
For this last code drop 30 of the 50 patches was from an
original single large patch. 


More information about the devel mailing list