Svante Signell, le Fri 11 Sep 2015 19:50:21 +0200, a écrit :
> Well, the Linux manpage for connect says:
The linux manpage is not supposed to be relevant.
> While POSIX says:
> http://pubs.opengroup.org/onlinepubs/9699919799/functions/connect.html
> [EADDRNOTAVAIL]
> The specified address
On Fri, 2015-09-11 at 17:43 +0200, Samuel Thibault wrote:
> Svante Signell, le Fri 11 Sep 2015 17:42:19 +0200, a écrit :
> > It seems like EADDRNOTAVAIL is an appropriate error,
>
> It is appropriate according to POSIX.
>
> > Otherwise I can create a patch for the socklog package. WDYT?
>
> That
Svante Signell, le Fri 11 Sep 2015 17:42:19 +0200, a écrit :
> It seems like EADDRNOTAVAIL is an appropriate error,
It is appropriate according to POSIX.
> Otherwise I can create a patch for the socklog package. WDYT?
That makes sense since POSIX allows it.
Samuel
On Fri, 2015-09-11 at 10:34 +0200, Svante Signell wrote:
> On Fri, 2015-09-11 at 02:10 +0200, Samuel Thibault wrote:
> > Hello,
> >
> > Thanks for the detailed description and testcase! I could fix it, this
> > case was actually apparently completely untested: reference counting was
> > never rea
On Fri, 2015-09-11 at 02:10 +0200, Samuel Thibault wrote:
> Hello,
>
> Thanks for the detailed description and testcase! I could fix it, this
> case was actually apparently completely untested: reference counting was
> never reaching zero, and thus the sock never actually shut down.
Nice to see
Hello,
Thanks for the detailed description and testcase! I could fix it, this
case was actually apparently completely untested: reference counting was
never reaching zero, and thus the sock never actually shut down.
SAmuel
Hi,
Looking into the socklog package test errors a bug was found in the
pflocal socket implementation, specifically in connect():
1)
PID1: create a local named socket using a server to receive data
PID2: connect to that socket with a client to send data
everything is fine :)
Terminal1:
./test_soc
