1

Unix/Linux 网络时间协议版本3 Network Time Protocol Version 3 (NTP) distribution for Unix systems

Replied: Wed, 19 Feb 1997 00:30:04 -0500
Replied: "Don Lewis <[e-mail]> "
Replied: Tue, 18 Feb 1997 18:25:35 -0500
Replied: "Don Lewis <[e-mail]> "
Received: from snow-white.ee.udel.edu by whimsy.udel.edu id aa22244;
          18 Feb 97 9:18 GMT
Received: from sunrise.gv.tsc.tdk.com ([e-mail] [192.168.241.191])
          by gatekeeper.tsc.tdk.com (8.8.4/8.8.4) with ESMTP
	  id BAA26998 for <[e-mail]>; Tue, 18 Feb 1997 01:18:52 -0800 (PST)
Received: from salsa.gv.tsc.tdk.com (salsa.gv.tsc.tdk.com [192.168.241.194])
          by sunrise.gv.tsc.tdk.com (8.8.4/8.8.4) with ESMTP
	  id BAA15181 for <[e-mail]>; Tue, 18 Feb 1997 01:18:51 -0800 (PST)
Received: (from gdonl@localhost)
          by salsa.gv.tsc.tdk.com (8.8.4/8.8.4)
	  id BAA03601; Tue, 18 Feb 1997 01:18:50 -0800 (PST)
From: Don Lewis <[e-mail]>
Message-Id: <[e-mail]>
Date: Tue, 18 Feb 1997 01:18:49 -0800
In-Reply-To: [e-mail]
       "Re: xntp3-5.89.6 will be available soon" (Feb 18,  1:56am)
X-Mailer: Mail User's Shell (7.2.6 alpha(3) 7/19/95)
To: [e-mail]
Subject: Re: xntp3-5.89.6 will be available soon
Cc: [e-mail]

I dig around in the freebsd-hackers archive, and it sounds like
signaled I/O won't work for the refclocks, at least without doing
TIOCSCTTY, and then you have to have dumped your previous controlling
tty and you're limited to one refclock.  It sounds like the symptoms
should be F_SETOWN failing, though.

I was digging through the FreeBSD source, and it looks like closing
the fd associated with the tty device calls ttyflush(), which will
post a signal to us if we somehow manange to get into signalled I/O
mode.  That will cause the handler to get invoked before the close()
returns.  Since the matching bit is still set in activefds, we'll select()
on it.  I'd expect it to tell us it's ready for reading, but who knows
what kind of wierd stuff will happen if we select() or read() on this
fd in this state.

I did notice that we're sometimes playing with activefds and maxactivefd
without signals being blocked.  We should wrap the entire insides of
close_socket() and close_file() between a BLOCKIO()/UNBLOCKIO() pair.
Also, we need to block SIGIO while we're fiddling with the refio list
in io_closeclock().  We could move the UNBLOCKIO() after the close_file()
and eliminate the signal stuff in close_file(), and we may want to
do the same with close_socket() and it's callers.

Not to say that there isn't a bug in FreeBSD ...

			---  Truck