IRCaBot 2.1.0
GPLv3 © acetone, 2021-2022
#ls2
/2022/06/25
orignal will check msg 3
orignal I have found the problem, but why you didn't have invalid signature?
zzz version check is earlier
orignal fixed
orignal as for me I prefer to check signature first and then process content
orignal 18:59:55@271/info - SSU2: Peer test 4 error code 65
orignal do you check it when you pick Charlie?
orignal lol it's me who doesn't check it ))
orignal will fix
orignal SSU2: PeerTest msg=7 code=0
orignal finally. peer test was successive
zzz checked in a bunch of MTU fixes, I should go all the way to 1500 now
zzz I think i2pd is exceeding the MTU; I got a 1517 byte packet from ImQC
orignal thzt's bug
orignal because const size_t SSU2_MTU = 1488;
orignal do you know what was in that packet?
zzz I have several examples, as best as I can tell they each contained only a single first fragment block (I don't log padding anymore)
orignal that's enough
orignal I will look into it
orignal it's not just mtu problem but also buffer overflow
zzz FYI I'm now assuming full 1500 default MTU when published as "SSU2", as per the spec
zzz in each case of a big packet, I never got the other fragment
orignal what do you mean?
orignal you never get a follow-on packet or what?
zzz correct
orignal do you even receive follow-on fragments?
zzz ever?
zzz i certainly do from java... looking for i2pd...
orignal I guess I know what it is
orignal resends
zzz no, I haven't found a single follow-on fragment from i2pd
orignal thanks. will check
zzz I have this big cheat sheet of who is java and who is i2pd to make things easier :)
orignal do you see some error instead?
zzz nothing obvious, no
orignal maybe be endianess of msgid?
zzz doesn't really matter as long as you do it the same both places
zzz I suggest you do some tests to see if you're sending and handling fragmented messages correctly
orignal yes, that's what I'm going to do
zzz great. hard for me to prove anything from this side
orignal strange that you don't receive it at all
orignal that confuses me
zzz that's not 100% sure. If it's completely corrupt including the header it may not get logged
zzz hint: one easy way to test is to internally force a really low MTU, like 600
zzz and log an error if you ever send anything bigger than mtu
zzz or even 400 MTU. you want to test 3+ fragments, not just 2. Follow-on fragments have 5 bytes more overhead. I had a lot of mtu bugs too
zzz and I may have more, not sure