Bernd Edlinger wrote:
> However I fixed a fiew memory leaks with my enhancement patch for the FreeBSD
> See Bug# 1001656. Maybe you try to use this first.
> One major flaw was, when you do a IF_DOWN the unsent buffers must be cleaned
> up too.
Thanks, that could be it.
One test I'm running is pinging (I ported the ppp_test.c test app in the
no-longer-tested archive into a class for our app) various hosts and internet
sites to make sure routing works properly over the PPP link and there are a lot
of receive timeouts and I'm sure some of them take long enough to time out that
PPP is brought down in the tail end of them.
> By the way, how did you compile your target with PPP in the first place?
> This does somehow not compile, as far as I know.
Lots of swearing and fixing cryptic auth.c compile errors.
I posted about it on the list in late August:
http://comments.gmane.org/gmane.os.ecos.general/28914
Looks like someone broke the code base a little after 2004....
ken
--- Begin Message ---
Hi Ken,
> On our system, we bring PPP up and down to minimize connection time costs,
> but after running a while, we seem to get this error:
>
> socket returned: No buffer space available
OK. This must be a memory leak somewhere.
However I fixed a fiew memory leaks with my enhancement patch for the FreeBSD
See Bug# 1001656. Maybe you try to use this first.
One major flaw was, when you do a IF_DOWN the unsent buffers must be cleaned up
too.
By the way, how did you compile your target with PPP in the first place?
This does somehow not compile, as far as I know.
Regards
Bernd Edlinger
--- End Message ---
--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss
