On Wed, 13 Nov 2013 01:26:56 +0000 "Myklebust, Trond" <Trond.Myklebust@xxxxxxxxxx> wrote: > On Wed, 2013-11-13 at 12:13 +1100, NeilBrown wrote: > > On Wed, 13 Nov 2013 00:30:53 +0000 "Myklebust, Trond" > > <Trond.Myklebust@xxxxxxxxxx> wrote: > > > > > On Wed, 2013-11-13 at 11:23 +1100, NeilBrown wrote: > > > > But back to my problem: Following Trond's suggestion I've come up with the > > > > following patch. Does it look right? > > > > > > > > The "fd = -1" is just to stop us trying to close a non-open fd in an error > > > > path. > > > > > > > > The change from testing ->servicename to ->prog stops us from repeating the > > > > failed DNS lookup on every request, not that the failure isn't fatal. > > > > > > > > The last stanza makes sure we always reply to an upcall, with EINVAL if > > > > nothing else seems appropriate. > > > > > > Wouldn't EACCES be more appropriate as a default? > > > > > > > Maybe. And that is what you suggested before and I mis-remembered - sorry. > > > > However EACCES is "Permission denied" which doesn't quite seem right to me. > > It isn't really "you aren't allowed to do that", but "your question doesn't > > make sense". > > > > However I'm not fussed. If you prefer EACCES, then I'll make it EACCES. > > If you look at gss_pipe_downcall(), then you'll note that it treats > EINVAL as a temporary error, and converts it to EAGAIN. That again > causes call_refreshresult to retry the upcall 2 more times before > failing with EACCES anyway... > Yes, I see now, thanks. I also see a 'BUG()' in there if the error code returned from user-space isn't in the known list. I suspect that should at most be a WARN, and probably removed altogether. Thanks, NeilBrown
Attachment:
signature.asc
Description: PGP signature
