Hi, On Wed, Jan 23, 2019 at 10:02 AM Oleksij Rempel <o.rempel@xxxxxxxxxxxxxx> wrote: > > Hi, > > On 23.01.19 08:45, laurent vaudoit wrote: > > Hi all, > > no feedback about this "problem" > > > > Regards > > Laurent > > > > On Wed, Jan 16, 2019 at 4:26 PM laurent vaudoit > > <laurent.vaudoit@xxxxxxxxx> wrote: > >> > >> Hi all, > >> i'm using the old J1939 kernel implementation (with iproute2 specific), > >> but it seems transport layer as not change a lot. > > Did you tested latest code? > > >> Here is my problem: > >> on my board i setup an interface, with 0x4a as SA. > >> > >> On a canalyzer i send a 0x0F segmented packet: > >> 0x18EC214Ax: 10 0F 00 03 FF CD FF 00 ==> RTS > >> 0x18EC4A21x: 11 03 01 FF FF CD FF 00 ==> CTS > >> 0x18EB214Ax: 01 B2 1E 78 4D 61 72 6B ==> DT > >> 0x18EB214Ax: 02 73 74 68 65 73 70 6F ==> DT > >> 0x18EB214Ax: 03 74 FF FF FF FF FF FF==> DT > >> > >> instead of sending an acknowledge, the board send an abort frame, with > >> 0x05 reason: > >> 0x18EC4A21x: FF 05 FF FF FF CD FF 00. > >> > >> After activating some debug log in the kernel, i see that when i > >> receive the first Data Transfer frame, i get the error message: > >> j1939xtp_rx_dat: last 10 > >> > >> if i look into the function j1939xtp_rx_dat, i see that the switch on > >> last_cmd wait BAM (0x20 or CTS 0x11), but for me it should wait RTS > >> (0x10, as last_cmd is not modified when sending the CTS (last_tx_cmd > >> is updated at this step) > > I don't know what code version are you using. Current code is updating > last_cmd for CTS. > > >> i've tried to modify but the behaviour is worst as there is some > >> retrasnmission of the CTS frame > >> > >> Would you have an idea on this behaviour? > > please use latest version. you are right i use an old j1939 stack, for 3.10 kernel wich needs iproute2 modified. i know there is some new implementation, but when i look on the kurt github, i see that latest commits are in 2017, and i do not see any change on the last_cmd updated for CTS. Did the code has changed of location? > > >> thanks in advance for your help > >> > >> Best regards > >> Laurent > > Laurent
