Harlan Stenn [Wed, 5 Jan 2000 05:07:00 +0000 (05:07 -0000)]
ChangeLog, ntp_io.c, refclock_fg.c, refclock_wwv.c:
* ntpd/ntp_io.c (open_socket): TOS support
* ntpd/refclock_wwv.c: Declaration cleanup
* ntpd/refclock_fg.c (fg_receive): Not all sprintf's are created equal...
From: Marc.Brett@westgeo.com
Harlan Stenn [Wed, 5 Jan 2000 04:43:46 +0000 (04:43 -0000)]
ntp_loopfilter.c, refclock_wwv.c:
* ntpd/refclock_wwv.c: Dave Cleaned and Improved things.
* ntpd/ntp_loopfilter.c (local_clock): Dave fixed something.
Harlan Stenn [Tue, 4 Jan 2000 06:56:57 +0000 (06:56 -0000)]
ChangeLog, icom.c, refclock_oncore.c:
* libntp/icom.c (sndpkt): Make declaration and definition equivalent.
(doublefreq): Ditto.
* ntpd/refclock_oncore.c (MAP_FAILED): Some systems do not #define
this.
Harlan Stenn [Tue, 4 Jan 2000 06:01:09 +0000 (06:01 -0000)]
Many files:
* ntpd/refclock_wwv.c:
* ntpd/refclock_chu.c:
* libntp/icom.c:
* libntp/Makefile.am:
* include/icom.h:
* html/driver7.htm:
* html/driver36.htm:
Support for ICOM. The WWV/H driver, by the way, is getting truly
awesome. The CHU autotune function works okay as it is. I'd like
to find somebody else to test the audio drivers just to make sure
I haven't done something stupid. There is a new define ICOM
intended for the driver autotune function; however, I crafted the
thing in much the same way as the refclock_atom.c thing - it tries
to open /dev/icom and, if that fails, goes quietly to sleep.
From: Dave Mills <mills@udel.edu>
Harlan Stenn [Mon, 3 Jan 2000 21:40:40 +0000 (21:40 -0000)]
ChangeLog, refclock_oncore.c:
* ntpd/refclock_oncore.c (oncore_read_config): Patches and cleanup
From: Poul-Henning Kamp <phk@critter.freebsd.dk>
more isascii() stuff from HMS.
Harlan Stenn [Mon, 3 Jan 2000 20:35:03 +0000 (20:35 -0000)]
COPYRIGHT, ChangeLog, copyright.htm, map_vme.c:
* ntpd/map_vme.c (map_vme): tx.access_result indicates failure on
< 0, not <= 0. A fix that apparently did not get brought over
from the ntp3 base.
From: Michael Barone <michael.barone@lmco.com>
Harlan Stenn [Mon, 3 Jan 2000 06:47:51 +0000 (06:47 -0000)]
Many files:
* configure.in: Move the ONCORE_SHMEM_STATUS check and definition
here.
* ntpd/refclock_oncore.c (oncore_init_shmem): Some systems do not
have MAP_HASSEMAPHORE.
Harlan Stenn [Mon, 3 Jan 2000 02:58:49 +0000 (02:58 -0000)]
ChangeLog, refclock_oncore.c:
* ntpd/refclock_oncore.c: Drive ONCORE_SHMEM_STATUS based on
HAVE_SYS_MMAN_H . If this needs finer checks, do it in
configure.in .
(oncore_read_config): Add the isascii() checks; older versions of
Unix don't guarantee correct behavior of is*() without it.
* ntpd/refclock_oncore.c: Add proof-of-concept support for
delivering receiver data stream to other processes through
a memory-mapped file.
From: Poul-Henning Kamp <phk@FreeBSD.org>
Harlan Stenn [Sun, 2 Jan 2000 11:27:16 +0000 (11:27 -0000)]
ChangeLog, configure, configure.in:
* configure.in (ac_refclock_chu): Provide the CHU driver by
default, and by default prefer the AUDIO version. We used to
limit this to SunOS or Solaris; now we drive by the availability
of the audioio header file.
Select the IRIG and WWV audio drivers the same way.
Harlan Stenn [Thu, 30 Dec 1999 00:24:13 +0000 (00:24 -0000)]
ChangeLog, refclock_gpsvme.c:
* ntpd/refclock_gpsvme.c: Fixes
From: Michael Barone <michael.barone@lmco.com>
Patch severely hacked by HMS to "make it conform". I hope I
didn't break anything.
Harlan Stenn [Wed, 29 Dec 1999 23:00:33 +0000 (23:00 -0000)]
Many files:
* ntpd/refclock_fg.c:
* ntpd/refclock_conf.c:
* ntpd/ntp_control.c:
* ntpd/Makefile.am:
* libntp/clocktypes.c:
* include/ntp.h:
* html/refclock.htm:
* html/driver37.htm:
* configure.in:
* acconfig.h:
Forum Graphic GPS clock support
From: Dmitry Smirnov <das@online.nsk.su>
I cleaned up the order of clocks and #defines in configure.in and acconfig.h
Harlan Stenn [Wed, 29 Dec 1999 22:13:59 +0000 (22:13 -0000)]
Many files:
* configure.in: Default to compile the IRIG, CHU and WWV/H
drivers and compile the CHU driver for audio, not modem.
Requested by Dave Mills.
* html/audio.htm:
* html/driver36.htm:
* html/qth.htm:
Dave wrote these to go along with the changes I checked in yesterday.
Harlan Stenn [Wed, 29 Dec 1999 02:34:42 +0000 (02:34 -0000)]
Many files:
* ntpd/refclock_wwv.c:
* ntpd/refclock_conf.c:
* ntpd/refclock_chu.c:
* ntpd/ntp_refclock.c:
* ntpd/ntp_loopfilter.c:
* html/refclock.htm:
* html/pps.htm:
* html/index.htm:
* html/driver7.htm:
* html/driver6.htm:
* html/copyright.htm:
I indulged an old agenda to polish up some programs originally written
for a DSP evaluation board. The result is a really hot audio driver for
the NIST WWV/WWVH time/frequency station plus a makeover for the CHU
audio driver. Currently, they and their IRIG audio driver buddy use the
SunOS/Solaris audio interface, which is clean and simple. I hook the
line in jack to a shortwave radio and these drivers (driver 7 for CHU
and driver 36 for WWV) and the drivers do the rest. The WWV/H driver is
really hot - I am currently watching the ionosphere move from the
doppler that shows up on the tick modulation tracked by the
program. During midday when the F2 layer settles down, the program
closes in on a few tens of microseconds of spare jitter and wander.
This watch on whichever 15/20 MHz signal sounds the best. At night on
5/10 MHz and even 2.5 HMz where the multipath, interference and noise
are much worse, the driver bangs around somewhat more.
The CHU driver makeover discovered a few broken bones after all these
years, but its majority decoder is back in business. For various reasons
to icky to reveal here, its 103-modem demodulator is not quite as good
as the chip, but it comes close and hey, 2025/2125 FSK is just not a
good idea for HF radio. This driver is not nearly as sophisitcated as
the WWV/H driver, but here a few hundred miles south of Ottawa, it does
a passably good job.
I tell you this in the hopes of getting somebody interested in porting
the audio interface to other machines, in particular Linux, FreeBSD and
anything else with a sound card. When the ionosphere stays cool, the
WWV/H driver is as good as anything else around here and with much less
jitter than a serial port. The only downside is all the audio drivers
chew up a modest fraction of cycles for the DSP routines - a few percent
on the newer silicon, up to half the cycles on an old Sparc
IPC. Whattheheck, these old machines aren't doing anything else around
here except serving time, and even when sucking cycles for DSP they
still mind the PPS microseconds.
The audio driver documentation had a makeover, too.
From: Dave Mills <mills@udel.edu>
Harlan Stenn [Sun, 19 Dec 1999 07:18:03 +0000 (07:18 -0000)]
Many files:
* ntpd/ntp_refclock.c:
* ntpd/refclock_atom.c: Header cleanup
* html/driver30.htm: Cleanup and improvements.
From: Reg Clemens <reg@dwf.com>
* ntpd/ntp_refclock.c:
* ntpd/refclock_jupiter.c:
* ntpd/refclock_msfees.c:
* ntpd/refclock_mx4200.c:
Portability (Solaris) patches
* ntpd/refclock_mx4200.c:
Self-survey improvements, cleanup for, PPS re-activation
* libntp/syssignal.c:
Fix for "interrupted system call" (EINTR) failure of the PPS
ioctl(TIOCGPPSEV) call in Solaris. Not sure why it was
interrupted, but this works around the failure. Not sure if
the (now silent) interrupt affects the timing accuracy.
THERE IS A CHANCE THIS PART OF THE PATCH MAY ADVERSELY
AFFECT OTHER SYSTEMS!
* scripts/ntp-status: Added.
From: Marc.Brett@westgeo.com
* ntpdate/ntpdate.c: Deal with multiple addresses.
From: Jeffrey C Honig <jch@bsdi.com>
* ntpd/refclock_conf.c: Replaced macro SHM with CLOCK_SHM.
* ntpd/refclock_shm.c (shm_poll): Take clock time in UTC.
pp->day starts at 1, t->tm_yday starts at 0.
From: Jakob Stoklund Olesen <stoklund@taxidriver.dk>
Harlan Stenn [Sat, 11 Dec 1999 06:00:43 +0000 (06:00 -0000)]
ChangeLog, refclock_oncore.c:
* ntpd/refclock_oncore.c: #elif conversion. I can only hope I did
it right.
* ntpd/refclock_oncore.c: Various patches
From: Reg Clemens <reg@dwf.com> and Kamal A Mostafa <kamal@whence.com>
ChangeLog, refclock_wwv.c:
* ntpd/refclock_wwv.c: Cleanup/checkin of the current state of affairs.
From: Dave Mills <mills@udel.edu>
Harlan Stenn [Thu, 9 Dec 1999 20:23:41 +0000 (20:23 -0000)]
ChangeLog, ntp_proto.c:
* ntpd/ntp_proto.c (default_get_precision): Use the right arg type
to pass "sizeof freq" to sysctlbyname().
From: Ollivier Robert <roberto@eurocontrol.fr>
Harlan Stenn [Tue, 7 Dec 1999 00:21:36 +0000 (00:21 -0000)]
Many files:
* ntpd/Makefile.am (ntpd_SOURCES): Added refclock_wwv.c
* ntpd/refclock_wwvb.c:
* ntpd/refclock_wwv.c:
* ntpd/refclock_conf.c:
* ntpd/refclock_chu.c:
* libntp/clocktypes.c:
* include/ntp.h: Dave cleaned some things up
Dave cleaned some things up (WWVB -> SPECTRACOM, CHU -> CHU_AUDIO,
WWV_AUDIO)
* acconfig.h: REFCLOCK_WWVB -> REFCLOCK_SPECTRACOM, added REFCLOCK_WWV
* configure.in: Deal with it.
Harlan Stenn [Sun, 5 Dec 1999 05:53:20 +0000 (05:53 -0000)]
ChangeLog, ntpq.c, ntpq_ops.c:
* ntpq/ntpq.c:
* ntpq/ntpq_ops.c:
Make `ntpq -pn' independent of DNS, ad advertised.
From: Kamal A Mostafa <kamal@whence.com>
Sven Dietricht [Fri, 3 Dec 1999 23:12:12 +0000 (23:12 -0000)]
.del-mexit.c~3aed0663:
This was WINNT only code. Removed to ports/winnt directory
mexit.c:
mexit.c was originally in libntp, but is NT port specific, thus moved to
ports/winnt/libntp
Sven Dietricht [Fri, 3 Dec 1999 12:57:20 +0000 (12:57 -0000)]
systime.c:
Winnt adjtime() has moved to ports/winnt
nt_clockstuff.c:
Added adjtime routine for WINNT. This was too different to remain hacked
in with the version in adjime.
clockstuff.h:
Remove no longer used extern declarations resulting from moveing adjtime
ChangeLog:
Winnt cleanup of adjtime
Harlan Stenn [Thu, 2 Dec 1999 22:27:58 +0000 (22:27 -0000)]
Many files:
* libparse/Makefile.in: Manually hacked to remove the
libparse_kernel.a.[co] cruft
* libparse/Makefile.am (k*.o): Generate these form the Makefile,
not from separate .c files
Harlan Stenn [Thu, 2 Dec 1999 06:08:34 +0000 (06:08 -0000)]
Many files:
* html/tickadj.htm:
* html/notes.htm:
* html/hints/solaris.html:
Point to the new dosynctodr report.
* html/hints/solaris.xtra.S99ntpd:
Update. Should be rewritten to take advantage of the new -g
switch and perhaps a loop to wait until ntpd hits a reasonable
"state".
* html/hints/solaris-dosynctodr.html: New information
From: "John W. Sopko Jr." <sopko@cs.unc.edu>
Harlan Stenn [Thu, 2 Dec 1999 01:59:07 +0000 (01:59 -0000)]
ChangeLog, .del-plot_summary.pl~3aed0663, .del-summary.pl~3aed0663:
* scripts/plot_summary.pl: Improved option parsing. Fixed one
minor Perl compatibility error.
* scripts/summary.pl: Official revision for NTPv4: Parse new
statistic file formats correctly, provide error checking for bad
input files, and guard against negative arguments to sqrt()
because of numeric effects. Use one pattern to select valid
statistic files. Add selectable output directory
(``--output-directory'') and improved option parsing. Directory
with statistic files (now settable also with
``--input-directory'') defaults to ``/var/log/ntp''.
From: Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de>
Sven Dietricht [Wed, 1 Dec 1999 20:15:23 +0000 (20:15 -0000)]
recvbuff.c:
Receive buffers no longer have events associated with them.
Remove the functions to create and set these events.
ntpd.c:
Set Processor Affinity so NTP doesn't switch CPUs
Remove Receive packet events. Replaced MsgWait with
WaitForMultipleObjectsEx, which is interruptible.
Harlan Stenn [Wed, 1 Dec 1999 06:26:58 +0000 (06:26 -0000)]
ChangeLog, acconfig.h, config.h.in, configure, configure.in:
* acconfig.h: DECL_STIME_1
* configure.in (iCFLAGS): Use -std1 for alpha*-dec-osf* if we are
using "cc".
Reported by: Tom Smith <smith@cag.lkg.dec.com>
Harlan Stenn [Wed, 1 Dec 1999 05:38:02 +0000 (05:38 -0000)]
Many files:
* include/l_stdlib.h: DECL_SYSTIME_1 --- a long *
* configure.in: Use it for DEC OSF[45]
Reported by: Tom Smith <smith@cag.lkg.dec.com>
* ntpd/refclock_parse.c: Add missing declarations
* ntptrace/ntptrace.c: Ditto
* ntpd/ntp_proto.c: Ditto
* ntpd/refclock_palisade.c: Ditto
From: Jonathan Stone <jonathan@dsg.stanford.edu>
Sven Dietricht [Thu, 25 Nov 1999 13:27:43 +0000 (13:27 -0000)]
systime.c:
Applied Winnt adj_systime residual calc. corrections from Harry Pyle.
Also enclosed systime_10ms_ticks conditional which is specific
to reliant unix and SCO5 in their respective IFDEFs
nt_clockstuff.c:
Added global ppm_per_adjustment_unit for calculating adj_systime residuals
independently of OS tick frequency, as provided by Harry Pyle.
Sven Dietricht [Thu, 25 Nov 1999 13:06:03 +0000 (13:06 -0000)]
instsrv.c:
Changed NT service name from NetworkTimeProtocol to NTP. This allows
entering 'net stop ntp' instead of 'net stop networktimeprotocol'
to start/stop ntp.
Sven Dietricht [Wed, 17 Nov 1999 22:54:58 +0000 (22:54 -0000)]
nt_clockstuff.c:
NTP generates a syslog message when the tick counter used to interpolate the
time of day, rolls over.
This log message is now only printed if running the debug executable, since
this event seems to happen quite frequently on multiprocessor systems.
Sven Dietricht [Wed, 17 Nov 1999 22:12:26 +0000 (22:12 -0000)]
.del-winnt~3aed0663, winnt.htm:
Updated Windows NT documentation and converted to html format.
Removed NTP 3 documentation, this is still in the NTP 3 release,
and is no longer applicable to the new NT4 port.
build.htm:
Removed obsolete Windows NT docs. Replaced with a single link
to the hints/winnt.htm file, which combines all current NT docs
into one file.
Sven Dietricht [Wed, 17 Nov 1999 21:22:19 +0000 (21:22 -0000)]
ntp_io.c:
Removing some dead code from the old Windows NT port
ntpd.c:
Fixes IO to use the existing packet handlers instead of Windows events -
which get lost sometimes.
refclock_nmea.c:
Define COMx as device name if compiling WINNT
Harlan Stenn [Wed, 17 Nov 1999 07:17:14 +0000 (07:17 -0000)]
Many files:
* html/copyright.htm: Credit Jack for his work.
* html/pic/*: Save a *ton* of space on the pictures.
From: Jack Sasportas <jack@innovativeinternet.com>
Sven Dietricht [Tue, 16 Nov 1999 15:30:18 +0000 (15:30 -0000)]
win32_io.c:
Stubs for tcflush, cfsetispeed, cfsetospeed for refclocks
config.h:
Cleanup to battle entropy
ntp_iocompletionport.h:
Headers for new io functions
win32_io.h:
Implementing more general support for termios and friends
Sven Dietricht [Tue, 16 Nov 1999 07:58:58 +0000 (07:58 -0000)]
.del-ntpd.plg~29080509:
NTPD.PLG is a log file inadvertently added to the distribution.
This file is created by the Visual C compiler and should not be here.
.del-ntpdc.plg~29080509:
This is a compiler generated file and shouldn't be in the distribution
Harlan Stenn [Tue, 16 Nov 1999 05:57:54 +0000 (05:57 -0000)]
ChangeLog, configure, configure.in, ntp_update:
* ntp_update: Ignore stuff in any A.* directory.
* configure.in: Clean up header checks for sys/syssgi.h and
sys/select.h . Originally, we did not check for sys/select.h
under some versions of SCO (see xntp3-5). Somewhere in ntp4 we
dropped the SCO check and added the check for sys/syssgi.h,
exclusive of checking for sys/select.h. Marc Brett can't think of
a reason why we should not be checking for sys/select.h, so let's
look for it now.
Harlan Stenn [Sun, 14 Nov 1999 00:59:23 +0000 (00:59 -0000)]
ChangeLog, ntp_refclock.c, ntpdate.c:
* ntpdate/ntpdate.c: Add the ability for ntpdate to query a
multicast address. We use the response to the multicast address
to learn new servers which we then add to the peer list and query.
In changing this I made the list of peers a linked list.
* ntpd/ntp_refclock.c: sys/sio.h and SCO5_CLOCK cleanup
From: Kamal A Mostafa <kamal@whence.com>
Harlan Stenn [Sun, 14 Nov 1999 00:33:58 +0000 (00:33 -0000)]
ChangeLog, configure, configure.in, ntp_loopfilter.c:
* ntpd/ntp_loopfilter.c: Let -g do a "correct any" for the first
time adjustment.
From: Dave Mills <mills@udel.edu>
* configure.in: sys/sio.h needs to be checked independently.
Reported by: Kamal A Mostafa <kamal@whence.com>
Harlan Stenn [Thu, 11 Nov 1999 08:49:02 +0000 (08:49 -0000)]
Makefile.in, config.h.in, configure, configure.in:
* configure.in: alpha-dec-osf4* -> alpha*-dec-osf4*|alpha*-dec-osf5*
From: Tom Smith <smith@cag.lkg.dec.com>
* configure.in: Look for <sys/sio.h>. If TIOCDCDTIMESTAMP is
there, we have TTYCLK.
* acconfig.h: Lose old AIOCTIMESTAMP stuff
Reported by: Kamal A Mostafa <kamal@whence.com>
Harlan Stenn [Thu, 11 Nov 1999 08:23:06 +0000 (08:23 -0000)]
ChangeLog, acconfig.h, configure.in:
* configure.in: alpha-dec-osf4* -> alpha*-dec-osf4*|alpha*-dec-osf5*
From: Tom Smith <smith@cag.lkg.dec.com>
* configure.in: Look for <sys/sio.h>. If TIOCDCDTIMESTAMP is
there, we have TTYCLK.
* acconfig.h: Lose old AIOCTIMESTAMP stuff
Reported by: Kamal A Mostafa <kamal@whence.com>
Harlan Stenn [Mon, 8 Nov 1999 05:32:09 +0000 (05:32 -0000)]
Many files:
* configure.in: Lose AIOCTIMESTAMP tests
* ntpd/ntpd.c: lose select() EINTR debug warning
* ntpd/ntp_refclock.c: AIOCTIMESTAMP -> TIOCDCDTIMESTAMP. Watch
CLK_SETSTR.
* ntpd/refclock_atom.c: fdpps is only there for PPS or PPSAPI.
AIOCTIMESTAMP is gone now.
From: Kamal A Mostafa <kamal@whence.com>
Harlan Stenn [Mon, 8 Nov 1999 04:25:41 +0000 (04:25 -0000)]
Many files:
* configure.in (HAVE_MLOCKALL): Deal with dec-osf5 realities
* ntpd/refclock_ulink.c (ulink_poll): Fix cast.
* libntp/machines.c (ntp_set_tod): Use a long* for the argument to
stime().
Reported by: Tom Smith <smith@cag.lkg.dec.com>
* ntpd/ntpd.c (set_process_priority): Use whatever we have until
something works.
Harlan Stenn [Mon, 8 Nov 1999 01:12:40 +0000 (01:12 -0000)]
ChangeLog, ntp_loopfilter.c:
* ntpd/ntp_loopfilter.c: Keep clock_frequency changes in a temp
variable so we can record it to loopstats (near as HMS can tell).
From: Dave Mills <mills@udel.edu>
Harlan Stenn [Sat, 6 Nov 1999 04:35:47 +0000 (04:35 -0000)]
ChangeLog, configure.in, refclock_parse.c:
* ntpd/refclock_parse.c (parse_start): ASYNC_PPS_CD_NEG cleanup
* configure.in (ac_cv_make_ntptime): OK on Linux
From: <Ulrich.Windl@rz.uni-regensburg.de>
* configure.in: NetBSD has PPSAPI now
F_SETOWN is needed for NetBSD
From: Jonathan Stone <jonathan@dsg.stanford.edu>
projects / thirdparty / ntp.git / log
