Re: [PATCH 2/3] NFS: Ensure that we free the rpc_task after read and write cleanups are done

[Bruce Fields <bfields@fieldses.org>]

On Fri, Jan 04, 2013 at 06:52:12PM +0000, Myklebust, Trond wrote:
> On Fri, 2013-01-04 at 13:29 -0500, Bruce Fields wrote:
> > On Fri, Jan 04, 2013 at 01:15:01PM -0500, Trond Myklebust wrote:
> > > This patch ensures that we free the rpc_task after the cleanup callbacks
> > > are done in order to avoid a deadlock problem that can be triggered if
> > > the callback needs to wait for another workqueue item to complete.
> > 
> > Makes sense to me!
> > 
> > (Dumb question: so read and write data are the only two cases where the
> > calldata embeds an rpc task?  Why is that?)
> 
> nfs_commit_data and nfs_layoutcommit_data do too. The idea is to improve
> reliability when writing back dirty data in low memory conditions. The
> struct nfs_write_data and nfs_commit_data have their own mempool in
> order to guarantee a minimum number of available slots. By embedding the
> rpc_task, we can extend that guarantee to cover (part of) the RPC call
> too.
> 
> The only reason why nfs_read_data has the same embedding is for symmetry
> with nfs_write_data.

OK, thanks for the explanation.

--b.