Wait, I figured it out. This isn't designed to work from the server
side. Sorry about the extra mail.
On 5/24/22 1:38 PM, rapier wrote:
On 5/20/22 9:08 PM, Damien Miller wrote:
Are you sure that it's the packet receive buffer in this case?
I'd recommend instrumenting sshbuf_read() first and seeing what what it's
doing, e.g. how much data the read() is returning.
You can set SSHBUF_DEBUG at compile time to get a lot more info.
A strace or similar might be helpful too.
So I tried this and I'm getting the following from the client
rapier@iztli:~$ ssh -p2289 192.168.1.102
kex_exchange_identification: No SSH version received in first 1024 lines
from server
Turns out the banner is getting filled with the SSHBUF_DBG lines:
debug1: kex_exchange_identification: banner line 0: sshbuf.c:34
sshbuf_check_sanity: sanity size 0 alloc 256 off 0 max 134217728
debug1: kex_exchange_identification: banner line 1: sshbuf.c:354
sshbuf_reserve: reserve buf = 0x556d1a798160 len = 3125
debug1: kex_exchange_identification: banner line 2: sshbuf.c:309
sshbuf_allocate: allocate buf = 0x556d1a798160 len = 3125
Am I missing a step? This is with 9.0p1 with the default sshd_config.
I've also tried it with 8.8 and 8.2 with the same results so I feel like
I am missing something. Client is 8.2p1.
Thanks,
Chris
_______________________________________________
openssh-unix-dev mailing list
openssh-unix-dev@xxxxxxxxxxx
https://lists.mindrot.org/mailman/listinfo/openssh-unix-dev