[pnfs] Layoutreturn Implementation

[Marc Eshel]

I think it should be async, first we can not do much if the call fails. We 
can not keep it until the server accept it back, and we need to remove the 
inode. If this is done as part of trimming inodes because we are low on 
memory we want it to be done as fast as possible so we can recover memory 
quickly.
Marc.

pnfs-bounces at linux-nfs.org wrote on 07/20/2006 01:51:59 PM:

> I think it has to be sync since we can't remove the inode until it 
> completes successfully.
> Dean
> 
> Iyer, Rahul wrote:
> > Hi,
> > Currently, free_layout() is being called from only one functioon: 
> > nfs4_clear_inode(). nfs4_clear_inode() is the .clear_inode pointer of 
> > super_operations. The .clear_inode pointer is called by clear_inode(). 

> > This is called mainly to remove the in memory inode reference.
> > 
> > The function that I spoke about, wwhich makes the LAYOUTRETURN rpc is 
> > pnfs_return_layout(). This function is, however called nowhere. All 
> > that needs to be done on the client is to call pnfs_return_layout() in 

> > nfs4_clear_inode().
> > So, I presume it's fairly simple. However, I had a question:
> > nfs4_proc_pnfs_layoutreturn() makes a call to rpc_call_sync(). Should 
> > LAYOUTRETURN be a synchronous RPC? Is it better asynchronous? Do we 
care?
> > 
> > Thanks
> > Rahul
> > 
> > 
> > 
------------------------------------------------------------------------
> >
> > _______________________________________________
> > pNFS mailing list
> > pNFS at linux-nfs.org
> > http://linux-nfs.org/cgi-bin/mailman/listinfo/pnfs
> 
> -- 
> Dean Hildebrand
> Ph.D. Candidate
> University of Michigan
> 
> _______________________________________________
> pNFS mailing list
> pNFS at linux-nfs.org
> http://linux-nfs.org/cgi-bin/mailman/listinfo/pnfs