proposed change to ssh_connect_direct()

Peter Moody mindrot at hda3.com
Sat Jan 7 14:30:23 AEDT 2017


so I spent a bit of time looking at this and it seems like the only
way to go, at least if I want to keep it in ssh_connect_direct(), is
to use pthreads. further, it seems like getting that accepted is
something of a long shot:

https://github.com/openssh/openssh-portable/commit/328118aa798878a68398b92ba85adfb630bc5434

:)

so, approaching this from a different angle, what if I wanted to have
something else establish the tcp connection and then fork/dup2/exec
ssh and pass off the fd's for the network connection? This is how I
*sort of* understand -W to work, but that's the sshd code path, not
the client.

is something like this acceptable, at least in theory?

Cheers,
peter


On Thu, Jan 5, 2017 at 7:03 AM, Peter Moody <mindrot at hda3.com> wrote:
> if the remote hostname has multiple ip addresses, ssh_connect_direct
> will currently loop and try each address in sequence until one works.
>
> I'm interested in making ssh tries each address concurrently and
> return success on the first one that connects. in the land of host
> certs and ssh bastions, this can be incredibly effective.
>
> are there any objects to me working up a patch to implement this?
>
> Cheers,
> peter


More information about the openssh-unix-dev mailing list