On Fri, Feb 12, 2010 at 08:36:41PM -0500, Gordon Ross wrote: > Thanks for having a look. Yes, I'll be happy to take your changes. > (If I'm not careful, I may end up owning this "pile":)
Cool, I've attached a small patch... it's mostly minor stuff. (Actually two small patches: one against the gate and one against your "fixwarn" version. One of my changes hits an ifdef right next to one of your changes.) The patch does also include one change to lock test #12 that I needed when I was testing running the Mac OS X NFSv4 client against a server that was constantly rebooting. Test #12 has the child take a lock and then the parent kills the child and checks that the lock has been released after a couple seconds. The problem I hit was that if the server goes down before the child's lock can be released, then the parent will mistakenly report that the test failed when it's really just waiting on the server to come back up. To get around that I added some calls to the parent to open/close/unlink a separate file before it checks that the child's lock has been released. These operations should hold up the parent long enough to allow the child's lock to be released even if the server was down at the time. Feedback is welcome from any NFS experts. :-) Thanks! --macko
