Harald Welte has posted comments on this change. ( https://gerrit.osmocom.org/13764 )
Change subject: LLC: Don't blindly assume all LLC frames have data payload ...................................................................... Patch Set 2: (1 comment) https://gerrit.osmocom.org/#/c/13764/2/src/gprs/gprs_llc.c File src/gprs/gprs_llc.c: https://gerrit.osmocom.org/#/c/13764/2/src/gprs/gprs_llc.c@961 PS2, Line 961: /* set l3 layer & remove the fcs */ > so do we need the fcs for kind of messages where there's not l3? […] there is FCS in any LLC frame, as the FCS covers both header and payload, right? The problam is not setting msg->l3h to NULL. The problem is to then call a msgb_l3trim() on a msgb that has l3h set to ?NULL. Of course we could also make msgb_l3trim() handle the l3 case == NULL. But I actually think that would be hiding unintended behavior. If at all, I would make msgb_l3trim() ASSERT in case somebody passes a msgb with l3h = NULL. The only thing that one *could* improve here is to remove the 3 byte FCS even for those messages without user data (i.e. llhp.data = NULL). Right now we don't do that, but then any downstream code (even future one) doesn't care on what's in the msgb, as the only relevant information is in the already-parsed llhp. -- To view, visit https://gerrit.osmocom.org/13764 To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings Gerrit-Project: osmo-sgsn Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I51bbd0f2c618d477a037af343ff41de1c8a5a3ae Gerrit-Change-Number: 13764 Gerrit-PatchSet: 2 Gerrit-Owner: Harald Welte <[email protected]> Gerrit-Reviewer: Harald Welte <[email protected]> Gerrit-Reviewer: Jenkins Builder (1000002) Gerrit-Reviewer: Keith Whyte <[email protected]> Gerrit-CC: Pau Espin Pedrol <[email protected]> Gerrit-Comment-Date: Wed, 24 Apr 2019 07:50:26 +0000 Gerrit-HasComments: Yes Gerrit-HasLabels: No
