On Wed, Jan 23, 2019 at 3:42 PM Marc Kleine-Budde <mkl@xxxxxxxxxxxxxx> wrote: > > On 1/23/19 3:34 PM, laurent vaudoit wrote: > > our client had some crash when using some simulation file, which send > > frame with the same source address as our j1939 board. > > > > after analysing, it seems it was related to some specific behaviour > > with rt-patch. > > > > so to avoid crash, and after some discussion on mailing list, > > confirming this was not some legal usecase, we decided to drop frame > > (in j1939/main.c) coming with > > the same source address as our board. > > And this is this filter who make the CM not working. > > Now il drop only frame with same source address, coming from external > > (not loopback) > > So you hardcoded a filter in j1939/main.c, after adjusting the filter it > works now. exact. > > Can you give us a dump of the illegal traffic to analyse and fix the > crash with the recent version of the j1939 stack? unfortunately, as it was some confidential vehicule log, i'm not allowed to share it, sorry. the analyze i made was that "Issue happens when we interconnect 2 can nodes with the same SA and both are exchanging multipacket broadcast message." also, the problem was not existing when we do not apply rt_patch. > > BTW: I've just pushed the latest iteration to the linux-can-next git. > > https://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can-next.git/log/?h=j1939 > https://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can-next.git/log/?h=j1939-individual > great! > Marc > Laurent > -- > Pengutronix e.K. | Marc Kleine-Budde | > Industrial Linux Solutions | Phone: +49-231-2826-924 | > Vertretung West/Dortmund | Fax: +49-5121-206917-5555 | > Amtsgericht Hildesheim, HRA 2686 | http://www.pengutronix.de | >
