On Thu, Sep 4, 2008 at 9:14 PM, Neil Brown <neilb@xxxxxxx> wrote: > On Thursday September 4, chuck.lever@xxxxxxxxxx wrote: >> >> o Return RPC_ANYSOCK on error (this is -1) instead of open-coding -1. > > I don't agree with this change. > > Everything else (including the last) looks fine, though I personally > wouldn't bother closing file descriptors and freeing memory just > before exit. If nothing else, it makes interpreting the output of valgrind easier. :-) Code in nfs-utils seems to last many many years, and is copied and pasted a lot without regard to context (ie code that may have been written for a single-shot application may suddenly find itself embedded in a long-running daemon). Considering that parts of nfs-utils are used in HA environments, I'd rather be safe then sorry. > RPC_ANYSOCK is an input value to various rpc library functions which > means "I haven't opened a socket, you do it for me". It is not really > a valid return value. Well, FWIW it is treated as a return value by utils/mount/network.c:get_socket() and it's callers. > Every time makesock is called, the return status is tested with > if (makesock(...) < 0) > > To me, that says that the returned value should be e.g. -1, not > RPC_ANYSOCK. OK, I'll drop it. Thanks for the review. -- Chuck Lever -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
