bug#5723: 23.1.94; make-network-process and emacs hangs

[YAMAMOTO Mitsuharu]

>>>>> On Thu, 18 Mar 2010 18:06:20 +0900, YAMAMOTO Mitsuharu 
>>>>> <mituharu@math.s.chiba-u.ac.jp> said:

>>> If your case blocks at `connect', then it can't be quit with C-g
>>> regardless of SYNC_INPUT on those ports/systems, because atimer is
>>> turned off during the `connect' call.

>> Why is the timer turned off around connect?  Handling SIGALARM
>> during connect doesn't seem much different from handling SIGIO.

> I think your argument makes sense.

The current code closes the socket and call `connect' again if
(blocking) `connect' is interrupted by a signal.

2004-11-09  Kim F. Storm  <storm@cua.dk>

        * process.c (Fmake_network_process): Remove kludge for interrupted
        connects on BSD.  If connect is interrupted, just close socket and
        start over rather than sleeping and retry with same socket.

(http://cvs.savannah.gnu.org/viewvc/emacs/src/process.c?root=emacs&r1=1.443&r2=1.444)

UNIX Network Programming (Richard Stevens et al.) says "What we must
do in this scenario is call /select/" (Section 16,5, Volume 1 third
edition).

Also, POSIX says:

  If connect() is interrupted by a signal that is caught while blocked
  waiting to establish a connection, connect() shall fail and set
  errno to [EINTR], but the connection request shall not be aborted,
  and the connection shall be established asynchronously.

  ...

  When the connection has been established asynchronously, select()
  and poll() shall indicate that the file descriptor for the socket is
  ready for writing.

(http://www.opengroup.org/onlinepubs/000095399/functions/connect.html)

Perhaps we should try this, not just removing `turn_on_atimers' calls.

                                     YAMAMOTO Mitsuharu
                                mituharu@math.s.chiba-u.ac.jp