Hi Nicholas, [Cc'ing the list if I may, as it may be useful] Le mardi 16 février 2021 à 00:56 +0000, Nicholas Humfrey a écrit : > > It is a very small change that is basically activated on the “client”- > > side with: > > > > ipv6 ::, > > > > thus sending a zero identifier for our side. > > > I just tried applying your patch to Git master (5191399) and > successfully compiled it. Thanks for testing! > However when running this on my 'client': > ./pppd/pppd file ~/ppp-options ipv6 ::, ipv6cp-accept-local /dev/ttyAMA0 > 115200 > > Then it seemed to fail to negotiate the link-local address. Here is the > output: > https://gist.github.com/njh/437713988b108880a8fe23ed10168c0c It seems it keeps sending ConfReq for the zero identifier… strange. Here is the output with the patched version from last june when I wrote it, latest git at the time (some time after 2.4.8), which works: using channel 19 Using interface ppp0 Connect: ppp0 <--> /dev/pts/18 rcvd [LCP ConfReq id=0x1 <asyncmap 0x0> <magic 0xfc846b9d> <pcomp> <accomp>] sent [LCP ConfReq id=0x1 <asyncmap 0x0> <magic 0x29314916> <pcomp> <accomp>] sent [LCP ConfAck id=0x1 <asyncmap 0x0> <magic 0xfc846b9d> <pcomp> <accomp>] rcvd [LCP ConfAck id=0x1 <asyncmap 0x0> <magic 0x29314916> <pcomp> <accomp>] sent [LCP EchoReq id=0x0 magic=0x29314916] sent [CCP ConfReq id=0x1 <deflate 15> <deflate(old#) 15> <bsd v1 15>] sent [IPV6CP ConfReq id=0x1 <addr fe80::0000:0000:0000:0000>] rcvd [LCP EchoReq id=0x0 magic=0xfc846b9d] sent [LCP EchoRep id=0x0 magic=0x29314916] rcvd [CCP ConfReq id=0x1 <deflate 15> <deflate(old#) 15> <bsd v1 15>] sent [CCP ConfAck id=0x1 <deflate 15> <deflate(old#) 15> <bsd v1 15>] rcvd [IPV6CP ConfReq id=0x1 <addr fe80::0000:0000:0000:0001>] sent [IPV6CP ConfAck id=0x1 <addr fe80::0000:0000:0000:0001>] rcvd [LCP EchoRep id=0x0 magic=0xfc846b9d] rcvd [CCP ConfAck id=0x1 <deflate 15> <deflate(old#) 15> <bsd v1 15>] Deflate (15) compression enabled rcvd [IPV6CP ConfNak id=0x1 <addr fe80::0000:0000:0000:0002>] sent [IPV6CP ConfReq id=0x2 <addr fe80::0000:0000:0000:0002>] rcvd [IPV6CP ConfAck id=0x2 <addr fe80::0000:0000:0000:0002>] local LL address fe80::0000:0000:0000:0002 remote LL address fe80::0000:0000:0000:0001 Script /etc/ppp/ipv6-up started (pid 11628) Script /etc/ppp/ipv6-up finished (pid 11628), status = 0x0 ^CTerminating on signal 2 Script /etc/ppp/ipv6-down started (pid 11655) sent [LCP TermReq id=0x2 "User request"] Script /etc/ppp/ipv6-down finished (pid 11655), status = 0x0 Child process sudo /home/benoar/Dev/ppp/pppd/pppd notty noauth noip +ipv6 ipv6 ::1,::2 (pid 11617) terminated with signal 2 Modem hangup Connection terminated. Connect time 0.1 minutes. Sent 120 bytes, received 120 bytes. It correctly gets nacked when sending zero iid, and then acks the given identifier. BTW, I use a quick “self-test” for development testing: sudo ./pppd/pppd pty "sudo $(pwd)/pppd/pppd notty noauth noip +ipv6 ipv6 ::1,::2" noauth nodetach noip +ipv6 ipv6 ::, debug But with latest git and my patch, I indeed get the same trace as you. > In my hack, rather than changing the definition of what a valid > Interface Identifier is, I instead looked for the calls to eui64_magic() > and commented them out. The VALIDID macro was sparsely used and fitted my use-case well, so I went with that. > Looking at the git master branch, it looks like Pali Rohár is actively > working in this space: > > https://github.com/paulusmack/ppp/commits/master Well, I'll give a look to find what's wrong. Thanks for your feedback! Regards, -- benjamin