Anders Persson wrote: > Hey Peter, > > Peter Dunlap wrote: > >> http://cr.opensolaris.org/~pdunlap/iscsit-webrev/webrev/ >> >> >> > idm_so.c: > > line 383: > nit: comment is a bit off. > Will fix. > lines 410 - 421: > Is this function used anywhere? >
No. Will remove. > lines 448 & 497: > MSG_WAITALL is not a valid flag for sendmsg(). > Good catch. I see this same error in our initiator and the smb server from where much of our networking code is modeled. In the case of sorecvmsg(), we would like to get a complete iSCSI packet or error. Is there something analogous in sosendmsg() where we queue a complete packet or get an error? Or do we need to loop on sosendmsg until the resid is zero ? > line 1050: > It's checking for the wrong error code. solisten() returns a > positive integer (errno value) upon failure, and zero otherwise. > Will fix. > line 1142: > I would like to see a bit more robust error handling. You might get > a ECONNABORTED, in which case you ought to just 'continue'. As a > side note, you will not get ECONNABORTED in Nevada (see 6263346), > but it will be back with Volo. > Will fix. > lines 902 - 906: > You should hold so_conn->so_lock, otherwise you may end up with > garbage. > Will fix. Thanks for the feedback! -jim > > Cheers, > Anders > _______________________________________________ > iser-dev mailing list > [EMAIL PROTECTED] > http://mail.opensolaris.org/mailman/listinfo/iser-dev > _______________________________________________ networking-discuss mailing list [email protected]
