This is the mail archive of the cygwin-developers@cygwin.com mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: what's left for 1.5.0


Corinna,

On Wed, Jun 11, 2003 at 09:05:42PM +0200, Corinna Vinschen wrote:
> On Wed, Jun 11, 2003 at 02:44:02PM -0400, Jason Tishler wrote:
> > > So CTRL_SHUTDOWN_EVENT actually is send also to the services child
> > > processes.

I believe that I have confirmed the above statement below.  And more,
importantly that normal applications do not receive CTRL_SHUTDOWN_EVENT
events.

> > I just tried running sshd directly under the LocalSystem account (using
> > the "at $TIME /interactive cmd" trick).  I got the following EventLog
> > entry:
> > 
> > [snip]
> > 
> > So, it appears that even normal apps receive CTRL_SHUTDOWN_EVENT
> > during shutdown.  Sigh...
> 
> That's not quite right.  When you're using `at', the application is
> started using the "Scheduler" or "Task Scheduler" service.  So the
> process is actually a service process.  Well, a child of a service
> process but from the CTRL_SHUTDOWN_EVENT point of view that's
> apparently the same.

OK, I tried again this time using fetchmail which usually runs as a
service under my account -- which makes it perfect for this kind of A/B
testing.

When fetchmail is started from the SCM via cygrunsrv, its log indicates
the following during shutdown:

    fetchmail: sleeping at Mon Mar 24 09:16:43 2003
    fetchmail: terminated with signal 15
    /* Windows reboots */
    fetchmail: starting fetchmail 6.2.2 daemon
    fetchmail: sleeping at Mon Mar 24 09:21:36 2003

When fetchmail is started as a normal application, its log indicates the
following during shutdown:

    fetchmail: sleeping at Thu Jun 12 07:35:38 2003
    /* Windows reboots */
    fetchmail: removing stale lockfile
    fetchmail: starting fetchmail 6.2.2 daemon
    fetchmail: sleeping at Thu Jun 12 07:40:06 2003

The above indicates that a CTRL_SHUTDOWN_EVENT event is *not* received
by fetchmail when running as a normal application.  IMO, we can
generalize from this test that only services and their children receive
CTRL_SHUTDOWN_EVENT events during shutdown -- never normal applications.
So, my confidence in this change has been restored.

Jason

-- 
PGP/GPG Key: http://www.tishler.net/jason/pubkey.asc or key servers
Fingerprint: 7A73 1405 7F2B E669 C19D  8784 1AFD E4CC ECF4 8EF6


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]