Re: [PATCH 1/1] silence-call-timeout-printk

["J. Bruce Fields" <bfields@fieldses.org>]

On Thu, Mar 27, 2008 at 04:31:22PM -0400, Trond Myklebust wrote:
> 
> On Thu, 2008-03-27 at 16:23 -0400, J. Bruce Fields wrote:
> > On Thu, Mar 27, 2008 at 04:03:49PM -0400, Trond Myklebust wrote:
> > > On Thu, 2008-03-27 at 15:54 -0400, J. Bruce Fields wrote:
> > > > On Thu, Mar 27, 2008 at 02:49:12PM -0400, Olga Kornievskaia wrote:
> > > > > When the client's callback server goes away, the server's callback client
> > > > > tries to contact the server and times out. For nfsd, it is beneficial to
> > > > > printout a message when the client is unable to contact the server. For
> > > > > the callback server, the same message is printed yet it is really not
> > > > > an error. Thus we need a way to silence the message for the callback 
> > > > > and yet print it for other cases.
> > > 
> > > Bring, back,
> > > Bring, back,
> > > Oh, bring back my cl_chatty to me...
> > > 
> > >   http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=f518e35aec984036903c1003e867f833747a9d79
> > 
> > OK, OK, so, shall we apply a patch that basically just reverts Chuck's
> > patch there, except that it inverts the meaning and uses the name
> > cl_quiet instead of cl_chatty?  (On the theory that since the verbose
> > case is the common case, it's best to let this thing be initialized to 0
> > and only make the callback code mess with it?)  Or are you nostalgic for
> > cl_chatty?
> > 
> > --b.
> 
> Why have two names for the same thing? That will just cause confusion
> should we ever want to backport this stuff to older kernels.
> 
> In any case, Chuck has made these bit-flags private to the RPC layer, so
> you can still add RPC_CLNT_CREATE_QUIET to the rpc_create_args->flags,
> and have the default be cl_chatty = 1, when that create_arg flag isn't
> set.

OK, that sounds fine.

--b.