[sslh] SSLH compiling for Cygwin o similar.

[Yves Rutschle]

On Fri, Sep 24, 2010 at 01:51:15PM +0200, juergen wagner wrote:
> i was able now not only to compile, but successfull run your program
> under windows and it works perfectly

Excellent -- would you be willing to document anything
"special" you may have done to get it running under cygwin?

> remove forking, also reprogram it to work with child process without
> forking, because under windows that are 4mb per process and by
> accessing the https site from many users it is terrible memory leak.

"memory overhead", not leak -- a memory leak is when the
program forgets to free up memory, and grows bigger and
bigger over time.

Under Linux all sslh processes would share their code space,
so the only overhead of the several forks is the data that
actually changes. I would sort of expect that to be the same
in Windows, is it not the case?

> May i ask you, did you already try to reprogram it to not work in fork
> mode? Any possibility to make it running with many processes in one
> copy of the program?

That is possible indeed -- it would come at a security price,
though: it means the one sslh process handles all data
flows. A vulnerability in sslh would then allow an attacker
to tamper with other dataflows, at least by crashing the
program and therefore dropping all connections at once.

Besides, I think cygwin is not very fast or efficient to
start with -- my guess is that if you are serious about
performance (memory, speed, etc), Cygwin under Windows is
not the way to go at all.

However, someone else requested something similar, so I'll
look into adding an option to make a fork-less version. I
can't commit to any deadline, though.

Cheers,
Y.