On Tue, Mar 31, 2020 at 4:13 PM Bart Van Assche <bvanassche@xxxxxxx> wrote: > > On 3/31/20 2:25 AM, Jinpu Wang wrote: > > On Sat, Mar 28, 2020 at 5:59 AM Bart Van Assche <bvanassche@xxxxxxx> wrote: > >> > >> On 2020-03-20 05:16, Jack Wang wrote: > >>> + /* > >>> + * Nothing was found, establish rtrs connection and proceed further. > >>> + */ > >>> + sess->rtrs = rtrs_clt_open(&rtrs_ops, sessname, > >>> + paths, path_cnt, RTRS_PORT, > >>> + sizeof(struct rnbd_iu), > >>> + RECONNECT_DELAY, BMAX_SEGMENTS, > >>> + MAX_RECONNECTS); > >> > >> Is the server port number perhaps hardcoded in the above code? > > > > Yes, we should have introduced a module parameter for rnbd-clt too, so > > if admin changes port_nr, it's possible to change it also on rnbd-clt. > > What if someone decides to use different port numbers for different rnbd > servers? Shouldn't the port number be configurable per connection > instead of making it a kernel module parameter? How about extracting the > destination port number from the address string like srp_parse_in() does? Probably we should do both, the per connection destination setting can over right the module parameter if both exist. > > Thanks, > > Bart. Thanks Bart!
