laforge has posted comments on this change. ( https://gerrit.osmocom.org/c/libosmocore/+/21243 )
Change subject: ns2: add support for frame relay ...................................................................... Patch Set 1: (9 comments) https://gerrit.osmocom.org/c/libosmocore/+/21243/1/include/osmocom/gprs/frame_relay.h File include/osmocom/gprs/frame_relay.h: https://gerrit.osmocom.org/c/libosmocore/+/21243/1/include/osmocom/gprs/frame_relay.h@1 PS1, Line 1: /*! \file frame_relay.h */ missing copyright / license statement (yes, my fault in the initial code, but I don't want to mess with your patchset now) https://gerrit.osmocom.org/c/libosmocore/+/21243/1/src/gb/frame_relay.c File src/gb/frame_relay.c: https://gerrit.osmocom.org/c/libosmocore/+/21243/1/src/gb/frame_relay.c@14 PS1, Line 14: missing copyright / license statement (yes, my fault in the initial code, but I don't want to mess with your patchset now) https://gerrit.osmocom.org/c/libosmocore/+/21243/1/src/gb/frame_relay.c@106 PS1, Line 106: /* RX Message: 14 [ 00 01 03 08 00 75 95 01 01 00 03 02 01 00 ] */ > this can be dropped right? what's bad about some examples of the messages in a comment? https://gerrit.osmocom.org/c/libosmocore/+/21243/1/src/gb/frame_relay.c@222 PS1, Line 222: ie[2] |= 0x02; > what are these 0x02 0x04 0x08? Please add defines for those. They are the bits we need to use for the "active" "add" and "del" situations. I find the existing code sufficiently clear, as the name of the function is visible from every "if" statement above. https://gerrit.osmocom.org/c/libosmocore/+/21243/1/src/gb/frame_relay.c@366 PS1, Line 366: link->succeed = link->succeed << 1; > link->succeed <<= 1; works either way, there is no rule that your version is superior or preferred, IMHO. Matter of taste. https://gerrit.osmocom.org/c/libosmocore/+/21243/1/src/gb/frame_relay.c@461 PS1, Line 461: LOGPFRL(link, LOGL_ERROR, "Could not create DLC %d\n", dlci); > return, or directly OSMO_ASERT. why? all the other DLCs in the message might just as well have been created? This function is executed in response to data received from the peer. I don't think we should assert here. https://gerrit.osmocom.org/c/libosmocore/+/21243/1/src/gb/frame_relay.c@468 PS1, Line 468: dlc->add = pvc->new; > not sure if this "new" here will cause problems when compiling with c++. we do not anticipate the compilation of the frame relay code in C++ programs. https://gerrit.osmocom.org/c/libosmocore/+/21243/1/src/gb/gprs_ns2_fr.c File src/gb/gprs_ns2_fr.c: https://gerrit.osmocom.org/c/libosmocore/+/21243/1/src/gb/gprs_ns2_fr.c@86 PS1, Line 86: struct iphdr > Don't we have this defined somewhere else? We should move this to some > osmocom compat header IMHO, s […] We don't really support FreeBSD, Apple or Cygwin platforms, so best to remove it altogether. Nobody ever even does compile testing on these. https://gerrit.osmocom.org/c/libosmocore/+/21243/1/src/gb/gprs_ns2_fr.c@301 PS1, Line 301: /* FIXME half writes */ > IIUC this should put stuff into a wqueue? yes, indeed. An osmo_wqueue is probably the best approach. I also think that for datagram oriented sockets you either can write a full packet or you cannot write, so you don't have to care about partial writes. A partial write would make no sense, as every write is one packet on the network (unless you play with CORK options). -- To view, visit https://gerrit.osmocom.org/c/libosmocore/+/21243 To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings Gerrit-Project: libosmocore Gerrit-Branch: master Gerrit-Change-Id: Id3b49f93d33c271f77cd9c9db03cde6b727a4d30 Gerrit-Change-Number: 21243 Gerrit-PatchSet: 1 Gerrit-Owner: lynxis lazus <[email protected]> Gerrit-Reviewer: Jenkins Builder Gerrit-Reviewer: pespin <[email protected]> Gerrit-CC: laforge <[email protected]> Gerrit-Comment-Date: Thu, 19 Nov 2020 17:44:56 +0000 Gerrit-HasComments: Yes Gerrit-Has-Labels: No Comment-In-Reply-To: pespin <[email protected]> Gerrit-MessageType: comment
