dragonfly.git
18 years agoFix warnings.
Joerg Sonnenberger [Tue, 26 Apr 2005 10:08:41 +0000 (10:08 +0000)]
Fix warnings.

18 years agoFix warnings.
Joerg Sonnenberger [Tue, 26 Apr 2005 10:05:08 +0000 (10:05 +0000)]
Fix warnings.

18 years agoReally include patches, don't let them catch dust.
Joerg Sonnenberger [Tue, 26 Apr 2005 09:42:56 +0000 (09:42 +0000)]
Really include patches, don't let them catch dust.

Noticed-by: Mitja Horvat <Mitja.Horvat@klub-ksok.si>
18 years agoFix warnings. Use sysctlbyname.
Joerg Sonnenberger [Tue, 26 Apr 2005 09:40:58 +0000 (09:40 +0000)]
Fix warnings. Use sysctlbyname.

18 years agoInclude guard.
Joerg Sonnenberger [Tue, 26 Apr 2005 09:26:07 +0000 (09:26 +0000)]
Include guard.

18 years agoInclude sys/types.h to get uid_t and gid_t.
Joerg Sonnenberger [Tue, 26 Apr 2005 09:08:38 +0000 (09:08 +0000)]
Include sys/types.h to get uid_t and gid_t.

18 years agoEnsure that the directory fits into memory. Fix warnings.
Joerg Sonnenberger [Tue, 26 Apr 2005 08:54:59 +0000 (08:54 +0000)]
Ensure that the directory fits into memory. Fix warnings.

18 years agoANSIfy. Fix warnings.
Joerg Sonnenberger [Tue, 26 Apr 2005 08:48:19 +0000 (08:48 +0000)]
ANSIfy. Fix warnings.

18 years agoUse namespace mangler.
Joerg Sonnenberger [Tue, 26 Apr 2005 08:46:54 +0000 (08:46 +0000)]
Use namespace mangler.

18 years agouse uid_t / gid_t for prototype in stdlib.h, ANSIfy.
Joerg Sonnenberger [Tue, 26 Apr 2005 08:45:19 +0000 (08:45 +0000)]
use uid_t / gid_t for prototype in stdlib.h, ANSIfy.

18 years agoFix warnings.
Joerg Sonnenberger [Tue, 26 Apr 2005 08:41:44 +0000 (08:41 +0000)]
Fix warnings.

18 years agovfork can clobber the local stack frame, use fork(). We might also
Joerg Sonnenberger [Tue, 26 Apr 2005 08:37:24 +0000 (08:37 +0000)]
vfork can clobber the local stack frame, use fork(). We might also
want to remove the support for '+'. Fix warnings.

18 years agoFix warning.
Joerg Sonnenberger [Tue, 26 Apr 2005 08:29:52 +0000 (08:29 +0000)]
Fix warning.

18 years agoFix warnings, ANSIfy.
Joerg Sonnenberger [Tue, 26 Apr 2005 08:27:44 +0000 (08:27 +0000)]
Fix warnings, ANSIfy.

18 years agoANSIfy. Fix warnings.
Joerg Sonnenberger [Tue, 26 Apr 2005 08:21:34 +0000 (08:21 +0000)]
ANSIfy. Fix warnings.

18 years agoFix warnings. Sprinkle const. Use size_t in some places.
Joerg Sonnenberger [Tue, 26 Apr 2005 08:07:58 +0000 (08:07 +0000)]
Fix warnings. Sprinkle const. Use size_t in some places.

18 years agoDo not try to collect offset data if a prior offset correction is still
Matthew Dillon [Tue, 26 Apr 2005 07:01:43 +0000 (07:01 +0000)]
Do not try to collect offset data if a prior offset correction is still
in progress.  Since we do not compensate for the prior offset correction,
any corrections based on data accumulated while the prior correction is
active will overshoot or undershoot.

Offset calculations are based on compensated real time rather then
uncompensated real time, which is why this is a problem.  Frequency
corrections are based on uncompensated real time and do not have any
similar issue.  It would be even better, but more difficult, to adapt
the offset correction code to use uncompensated real time and then take
the compensation into account, but this is a considerably more complex
equation to get right.

There is still another problem and that is the offset error will continue
to build when the frequency is not corrected, potentially resulting in a
longer sync-up time.  On the positive side, the offset correction is based
on a standard deviation that only requires a minimum of 4 samples.

Correct a minor compiler warning.

18 years agoyp_maplist is one of those undocumented functions, resurrect and prototype it.
YONETANI Tomokazu [Tue, 26 Apr 2005 06:30:52 +0000 (06:30 +0000)]
yp_maplist is one of those undocumented functions, resurrect and prototype it.

Taken-from: FreeBSD include/rpcsvc/ypclnt.h,1.15

18 years agoFix warnings, remove unused headers. Remove additional { }.
Joerg Sonnenberger [Tue, 26 Apr 2005 06:16:29 +0000 (06:16 +0000)]
Fix warnings, remove unused headers. Remove additional { }.

18 years agoRemove more __NETBSD_SYSCALLS.
Joerg Sonnenberger [Tue, 26 Apr 2005 06:08:42 +0000 (06:08 +0000)]
Remove more __NETBSD_SYSCALLS.

18 years agoRemove compat junk, __getcwd always exists on DragonFly.
Joerg Sonnenberger [Tue, 26 Apr 2005 06:04:56 +0000 (06:04 +0000)]
Remove compat junk, __getcwd always exists on DragonFly.

18 years agoUnconditionalize HAS_UTRACE, we never have NetBSD syscalls.
Joerg Sonnenberger [Tue, 26 Apr 2005 05:51:00 +0000 (05:51 +0000)]
Unconditionalize HAS_UTRACE, we never have NetBSD syscalls.

18 years agoFix warnings.
Joerg Sonnenberger [Tue, 26 Apr 2005 05:48:21 +0000 (05:48 +0000)]
Fix warnings.

18 years agoReadd ypresp_allfn, now correctly typed.
Joerg Sonnenberger [Tue, 26 Apr 2005 05:30:12 +0000 (05:30 +0000)]
Readd ypresp_allfn, now correctly typed.

Noticed-by: yonetani
18 years agoAdd -n (not for real) option. The program goes through motions, but
Matthew Dillon [Tue, 26 Apr 2005 00:56:54 +0000 (00:56 +0000)]
Add -n (not for real) option.  The program goes through motions, but
doesn't actually update the system time.  Plus a few other cleanups.
Everything now works except -s/-S and leap second correction.

18 years agoOn MP systems, malloc's using R_NOWAIT are not supposed to block, so don't
Matthew Dillon [Tue, 26 Apr 2005 00:47:59 +0000 (00:47 +0000)]
On MP systems, malloc's using R_NOWAIT are not supposed to block, so don't
let them block on the BGL either.

18 years agoMinor rearrangement of an mpcount. This isn't actually a bug because the
Matthew Dillon [Tue, 26 Apr 2005 00:31:05 +0000 (00:31 +0000)]
Minor rearrangement of an mpcount.  This isn't actually a bug because the
code is safely in a critical section, but mpcount still ought not be
bumped until we actually intend to get the MP lock.

18 years agoFix an SMP bug. The SMP startup code waits for the APs to finish
Matthew Dillon [Tue, 26 Apr 2005 00:30:00 +0000 (00:30 +0000)]
Fix an SMP bug.  The SMP startup code waits for the APs to finish
initializing and goes through an interlock with the BGL.  Unfortunately,
it does not properly restore the BGL when it finishes.  This can cause
the system to lock-up on boot (spin trying to get a broken BGL) just
after finishing the AP initialization.  Additionally, this resulted in the
swapper thread running without the BGL.

18 years agoRedo the way the thread priority is displayed. Split out the critical
Matthew Dillon [Mon, 25 Apr 2005 22:26:22 +0000 (22:26 +0000)]
Redo the way the thread priority is displayed.  Split out the critical
section count from the priority field and also display the MP lock count
for each thread.

18 years agoAdd config file support, daemon() backgrounding, pidfile,
Matthew Dillon [Mon, 25 Apr 2005 20:50:59 +0000 (20:50 +0000)]
Add config file support, daemon() backgrounding, pidfile,
auto-kill-old-daemon on program restart, plus some cleanups.

18 years agoMake the trace code a bit smarter. It now looks around for something
Matthew Dillon [Mon, 25 Apr 2005 20:22:27 +0000 (20:22 +0000)]
Make the trace code a bit smarter.  It now looks around for something
that might be a frame pointer if the stack pointer given to it does
not appear to point at a frame.

18 years agoAdd information on interrupt preemptions by the current thread on any given
Matthew Dillon [Mon, 25 Apr 2005 20:05:09 +0000 (20:05 +0000)]
Add information on interrupt preemptions by the current thread on any given
cpu, and in the thread list, so we can see what is preempting what.

18 years agoconst correctness
Joerg Sonnenberger [Mon, 25 Apr 2005 19:36:57 +0000 (19:36 +0000)]
const correctness

18 years agoFix warnings, ANSIfy, constify.
Joerg Sonnenberger [Mon, 25 Apr 2005 19:03:46 +0000 (19:03 +0000)]
Fix warnings, ANSIfy, constify.

18 years agoFirst stab at WARNS=6 cleaning. More will be done once I figure out
Joerg Sonnenberger [Mon, 25 Apr 2005 18:52:28 +0000 (18:52 +0000)]
First stab at WARNS=6 cleaning. More will be done once I figure out
how to teach rpcgen const.

18 years agoAnsify getcwd and declare the syscall prototype.
Joerg Sonnenberger [Mon, 25 Apr 2005 18:19:48 +0000 (18:19 +0000)]
Ansify getcwd and declare the syscall prototype.

18 years agoImplement a course offset adjustment for large time steps.
Matthew Dillon [Mon, 25 Apr 2005 17:42:49 +0000 (17:42 +0000)]
Implement a course offset adjustment for large time steps.

Rework the logging code.  Add four logging levels (default 1, which logs
any corrections > 10 ppm made).

Log to syslog when not in debug mode.

18 years agoRemove inactive code for the maclabel parameter. DragonFly doesn't support
Sascha Wildner [Mon, 25 Apr 2005 17:33:26 +0000 (17:33 +0000)]
Remove inactive code for the maclabel parameter. DragonFly doesn't support
Mandatory Access Control currently.

18 years ago- ANSI
Liam J. Foy [Mon, 25 Apr 2005 16:10:25 +0000 (16:10 +0000)]
- ANSI
- WARNS 6

Submitted by: Peter Avalos <pavalos@theshell.com>

18 years agoCISS quirk.
Joerg Sonnenberger [Mon, 25 Apr 2005 15:14:54 +0000 (15:14 +0000)]
CISS quirk.

Obtained-from: FreeBSD

18 years agoBe consistent for preventing redundant header inclusion.
Hiten Pandya [Mon, 25 Apr 2005 14:27:03 +0000 (14:27 +0000)]
Be consistent for preventing redundant header inclusion.

18 years agoANSIfy.
Joerg Sonnenberger [Mon, 25 Apr 2005 08:33:25 +0000 (08:33 +0000)]
ANSIfy.

18 years agoANSIfy and fix function casts.
Joerg Sonnenberger [Mon, 25 Apr 2005 08:25:23 +0000 (08:25 +0000)]
ANSIfy and fix function casts.

18 years agoProperly create and destroy the DMA maps.
Joerg Sonnenberger [Mon, 25 Apr 2005 07:05:55 +0000 (07:05 +0000)]
Properly create and destroy the DMA maps.

18 years agoRemoved duplicate static variable, created when I merged compat.c and job.c
Max Okumoto [Mon, 25 Apr 2005 05:52:18 +0000 (05:52 +0000)]
Removed duplicate static variable, created when I merged compat.c and job.c

18 years agoDon't issue a frequency correction if it is the same as the last
Matthew Dillon [Mon, 25 Apr 2005 02:28:47 +0000 (02:28 +0000)]
Don't issue a frequency correction if it is the same as the last
frequency correction.

18 years agoImplement a variable polling rate capability.
Matthew Dillon [Sun, 24 Apr 2005 23:09:32 +0000 (23:09 +0000)]
Implement a variable polling rate capability.

Implement a polling state which runs through a fast-startup to get a rough
offset correction, a medium-polled acquisition mode to get a rough
frequency correction, and a long-polled maintainance mode (assuming the
time source is good enough) to get a fine frequency correction.

startup mode: 6 five-second polls
acquisition mode: 16 five-minute polls
maintainance mode: 30 minute polls (if the correllation is good enough)

Implement a short failure mode (3 failed polls in a row) which returns
to an aquisition state on recovery, and a long failure mode (10 failed polls
in a row) which returns to a startup state on recovery.

When resetting a regression, also remove any alternate regression further
on down the alt chain (since they won't be staggered any more).

18 years agoComment out description of the maclabel parameter.
Sascha Wildner [Sun, 24 Apr 2005 20:20:57 +0000 (20:20 +0000)]
Comment out description of the maclabel parameter.

18 years ago- ANSI
Liam J. Foy [Sun, 24 Apr 2005 15:31:30 +0000 (15:31 +0000)]
- ANSI
- WARNS 6

Submitted by: Peter Avalos <pavalos@theshell.com>

18 years ago- Add further functionality to check for invalid characters
Liam J. Foy [Sun, 24 Apr 2005 14:36:55 +0000 (14:36 +0000)]
- Add further functionality to check for invalid characters
- Remove keyword 'continue' for more indepth error reporting
  on each line
- WARNS 6 Clean

No objections: kernel@

18 years agoMove redirection of stdin into ProcExec().
Max Okumoto [Sun, 24 Apr 2005 12:44:50 +0000 (12:44 +0000)]
Move redirection of stdin into ProcExec().

18 years agoMove redirection of stdout into ProcExec().
Max Okumoto [Sun, 24 Apr 2005 12:44:08 +0000 (12:44 +0000)]
Move redirection of stdout into ProcExec().

18 years agoMoved manupulation of close-on-exec flags on std{in,out,err} into
Max Okumoto [Sun, 24 Apr 2005 12:43:43 +0000 (12:43 +0000)]
Moved manupulation of close-on-exec flags on std{in,out,err} into
ProcExec().  The old code was broken in that it cleared all the
flags, not just close-on-exec.

18 years agoUse a temp variable to move dups2() out of conditional.
Max Okumoto [Sun, 24 Apr 2005 12:43:09 +0000 (12:43 +0000)]
Use a temp variable to move dups2() out of conditional.

18 years agoMove clearing of close-on-exec of both stdin, stdout together.
Max Okumoto [Sun, 24 Apr 2005 12:42:38 +0000 (12:42 +0000)]
Move clearing of close-on-exec of both stdin, stdout together.

18 years agoMoved execvp() into ProcExec().
Max Okumoto [Sun, 24 Apr 2005 12:42:22 +0000 (12:42 +0000)]
Moved execvp() into ProcExec().

18 years agoMove redirection of stderr into ProcExec().
Max Okumoto [Sun, 24 Apr 2005 12:41:54 +0000 (12:41 +0000)]
Move redirection of stderr into ProcExec().

18 years agoMove common exec code into a ProcExec().
Max Okumoto [Sun, 24 Apr 2005 12:41:08 +0000 (12:41 +0000)]
Move common exec code into a ProcExec().

18 years agoMove OP_* #defines out of make.h and into GNode.h were they belong.
Max Okumoto [Sun, 24 Apr 2005 12:40:23 +0000 (12:40 +0000)]
Move OP_* #defines out of make.h and into GNode.h were they belong.

18 years agoCopy string into temp variable to quiet warning.
Max Okumoto [Sun, 24 Apr 2005 12:39:45 +0000 (12:39 +0000)]
Copy string into temp variable to quiet warning.

18 years agoMove variable declarations out of globals.h into var.h
Max Okumoto [Sun, 24 Apr 2005 12:39:17 +0000 (12:39 +0000)]
Move variable declarations out of globals.h into var.h

18 years agoRevert using execl(), it makes it difficult to extract common code.
Max Okumoto [Sun, 24 Apr 2005 12:38:26 +0000 (12:38 +0000)]
Revert using execl(), it makes it difficult to extract common code.

18 years agoGet rid off the wrappers around __va_copy(), they serve no real purpose.
Jeroen Ruigrok/asmodai [Sun, 24 Apr 2005 10:58:19 +0000 (10:58 +0000)]
Get rid off the wrappers around __va_copy(), they serve no real purpose.
Wrap va_copy() in __ISO_C_VISIBLE.

18 years ago__pure__ is supported from 2.96 onward, not 3.0.
Jeroen Ruigrok/asmodai [Sun, 24 Apr 2005 09:56:31 +0000 (09:56 +0000)]
__pure__ is supported from 2.96 onward, not 3.0.

18 years agoSynch our GNU_PREREQ() macro with FreeBSD:
Jeroen Ruigrok/asmodai [Sun, 24 Apr 2005 09:41:27 +0000 (09:41 +0000)]
Synch our GNU_PREREQ() macro with FreeBSD:

explicitly exclude the Intel C compiler and make the layout a tad bit
better to read.

18 years agoChange client_check() to calculate the best offset and the best frequency
Matthew Dillon [Sun, 24 Apr 2005 09:39:27 +0000 (09:39 +0000)]
Change client_check() to calculate the best offset and the best frequency
correction separately.

The best offset correction now utilizes a standard deviation of the offset
error with a minimum of 4 samples.  An offset correction occurs when the
standard deviation is less then 1/4 the average offset.  When an offset
correction is made, the offset data is cleared (since it becomes invalid
once a correction is made).  This calculation gives the code an almost
human perception of when the offset can be safely adjusted, regardless
of the 'mess' on the internet.

Change tv_delta_micro() into tv_delta_double() since most of the callers
were (previously) dividing the result by 1E6.

NOTE: the correllation for frequency correction is not working quite as
well as I had hoped, it's bouncing around a lot for internet ntp sources.
But the standard deviation on the offset appears to work very well.

18 years agoAllows 16 samples with a correllation >= 0.96 in addition to the
Matthew Dillon [Sun, 24 Apr 2005 06:24:42 +0000 (06:24 +0000)]
Allows 16 samples with a correllation >= 0.96 in addition to the
pre-existing 8 samples with a correllation >= 0.99 to satisfy the
requirements for a correction to be issued to the system.

18 years agoFix an overflow in the delta time calculation. A double as incorrectly
Matthew Dillon [Sun, 24 Apr 2005 06:23:56 +0000 (06:23 +0000)]
Fix an overflow in the delta time calculation.  A double as incorrectly
converted to a long.

18 years agoCleanup compiler warnings. Generate cleaner debug output. Change the
Matthew Dillon [Sun, 24 Apr 2005 05:04:28 +0000 (05:04 +0000)]
Cleanup compiler warnings.  Generate cleaner debug output.  Change the
default polling interval to 60 seconds.

Add a -T option to specify the polling interval.  Enhance test mode (-t).

18 years agoCorrect some packet ntohs/htons issues. Fix a documentation bug.
Matthew Dillon [Sun, 24 Apr 2005 03:10:53 +0000 (03:10 +0000)]
Correct some packet ntohs/htons issues.  Fix a documentation bug.
Adjust the command byte, rootdelay, and precision in the message to
conform to what ntpdate uses.

18 years agoInitial commit for the DragonFly home-made ntpd client. Why? Because
Matthew Dillon [Sun, 24 Apr 2005 02:36:50 +0000 (02:36 +0000)]
Initial commit for the DragonFly home-made ntpd client.  Why?  Because
xntpd is ridiculously huge, and OpenBSD's ntpd doesn't handle frequency
correction properly.

This is a work in progress, not yet ready for prime time.  Expect many
more commits.

18 years agoFix the structural type for kern.boottime and kern.basetime. Fix basetime
Matthew Dillon [Sun, 24 Apr 2005 02:01:08 +0000 (02:01 +0000)]
Fix the structural type for kern.boottime and kern.basetime.  Fix basetime
so the sysctl is properly sized with a NULL oldp.

Fix kern.ntp.delta.  It was doing a relative adjustment instead of an
absolute adjustment.

Document the ntp sysctl_*() PROC's.

18 years agoFix stupid order bug. The code should ignore the first sample(s),
Joerg Sonnenberger [Sat, 23 Apr 2005 20:43:34 +0000 (20:43 +0000)]
Fix stupid order bug. The code should ignore the first sample(s),
which means it should terminate before updating the statistics.

18 years agoMake access to basetime MP safe and interrupt-race safe by using a simple
Matthew Dillon [Sat, 23 Apr 2005 20:34:32 +0000 (20:34 +0000)]
Make access to basetime MP safe and interrupt-race safe by using a simple
tail-chasing FIFO for updates to basetime.

Reorganize the PROC sysctl's.  This actually undoes part of the last commit
and redoes it, though there was nothing wrong with the last commit.

Move the SYSCTL_OUT phase to *after* the SYSCTL_IN phase.

18 years agoAlways use SYSCTL_OUT, sysctl doesn't print the value otherwise.
Joerg Sonnenberger [Sat, 23 Apr 2005 19:59:05 +0000 (19:59 +0000)]
Always use SYSCTL_OUT, sysctl doesn't print the value otherwise.
Use CTLTYPE_QUAD to correctly descripe the various variables.

18 years agoAdd support for unsigned quads. Use strtoq and %qd / %qu.
Joerg Sonnenberger [Sat, 23 Apr 2005 19:57:35 +0000 (19:57 +0000)]
Add support for unsigned quads. Use strtoq and %qd / %qu.

18 years agoAdd forgotten SYSCTL_ADD_QUAD, move the SYSCTL_ADD_INT up where it belongs.
Joerg Sonnenberger [Sat, 23 Apr 2005 19:46:42 +0000 (19:46 +0000)]
Add forgotten SYSCTL_ADD_QUAD, move the SYSCTL_ADD_INT up where it belongs.
Add support for unsigned quads.

18 years agoMake kern.ntp.delta preemption save.
Joerg Sonnenberger [Sat, 23 Apr 2005 18:46:54 +0000 (18:46 +0000)]
Make kern.ntp.delta preemption save.

18 years agoGC more mips parts.
Joerg Sonnenberger [Sat, 23 Apr 2005 18:20:40 +0000 (18:20 +0000)]
GC more mips parts.

18 years agoRemove MIPS bits.
Joerg Sonnenberger [Sat, 23 Apr 2005 18:14:49 +0000 (18:14 +0000)]
Remove MIPS bits.

19 years agoRemove support for canbus (which is PC98 specific).
Sascha Wildner [Fri, 22 Apr 2005 22:02:58 +0000 (22:02 +0000)]
Remove support for canbus (which is PC98 specific).

19 years agoAdd some convenient targets for cvs maintanance by the cvs meister.
Matthew Dillon [Fri, 22 Apr 2005 20:51:37 +0000 (20:51 +0000)]
Add some convenient targets for cvs maintanance by the cvs meister.

19 years agoReally removed the hack. The last patch just added another
Max Okumoto [Fri, 22 Apr 2005 18:05:56 +0000 (18:05 +0000)]
Really removed the hack.  The last patch just added another
copy of the hack. sigh

19 years agoRemoved stdint.h hack
Max Okumoto [Fri, 22 Apr 2005 18:01:32 +0000 (18:01 +0000)]
Removed stdint.h hack

requested by joerg.

19 years agoDon't call cpu_mb1 after lwkt_setcpu_self, but call it internally
Joerg Sonnenberger [Fri, 22 Apr 2005 17:41:15 +0000 (17:41 +0000)]
Don't call cpu_mb1 after lwkt_setcpu_self, but call it internally
after the processing is done in lwkt_setcpu_self. The extern call
should automatically invalidate all non-local data and keeping it
in lwkt_setcpu_self ensures that it continues to work even with
IPO.

Requested-by: dillon
19 years ago- Check return value of setenv(3)
Liam J. Foy [Fri, 22 Apr 2005 16:59:46 +0000 (16:59 +0000)]
- Check return value of setenv(3)

Submitted by: Joe Talbott <josepht@cstone.net>

19 years agoremove spaces before tabs in this file
Max Okumoto [Fri, 22 Apr 2005 16:02:21 +0000 (16:02 +0000)]
remove spaces before tabs in this file

19 years agouse execl() instead of execv() so we can reduce the number of warnings.
Max Okumoto [Fri, 22 Apr 2005 16:01:56 +0000 (16:01 +0000)]
use execl() instead of execv() so we can reduce the number of warnings.

suggested by: joerg.

19 years agoMove prototype for Main_ParseArgLine() into make.h so we can delete
Max Okumoto [Fri, 22 Apr 2005 16:01:03 +0000 (16:01 +0000)]
Move prototype for Main_ParseArgLine() into make.h so we can delete
nonints.h

19 years agoMove some #defines related to parsing into parse.h
Max Okumoto [Fri, 22 Apr 2005 15:59:54 +0000 (15:59 +0000)]
Move some #defines related to parsing into parse.h

19 years agoMake shellName and shellPath static to job.c
Max Okumoto [Fri, 22 Apr 2005 15:59:32 +0000 (15:59 +0000)]
Make shellName and shellPath static to job.c

19 years agoMake the three version of the fork code more uniform.
Max Okumoto [Fri, 22 Apr 2005 15:59:07 +0000 (15:59 +0000)]
Make the three version of the fork code more uniform.

19 years ago- Use errno to report why freopen(3) failed
Liam J. Foy [Fri, 22 Apr 2005 15:36:39 +0000 (15:36 +0000)]
- Use errno to report why freopen(3) failed
- Minor cleanups

19 years agoBack out last commit, we are not there yet and I have to study
Joerg Sonnenberger [Fri, 22 Apr 2005 11:00:36 +0000 (11:00 +0000)]
Back out last commit, we are not there yet and I have to study
the interactions between the network RC scripts more.

19 years agoChange .EXPORT: to .EXPORTVAR: earlier versions of pmake used
Max Okumoto [Fri, 22 Apr 2005 10:38:30 +0000 (10:38 +0000)]
Change .EXPORT: to .EXPORTVAR: earlier versions of pmake used
.EXPORT: and .NOEXPORT: as special targets.

19 years agoMove to CPU #0 in settime() to prevent races.
Joerg Sonnenberger [Fri, 22 Apr 2005 10:12:26 +0000 (10:12 +0000)]
Move to CPU #0 in settime() to prevent races.

19 years agoUse normal variable names for rtsold, provide default values.
Joerg Sonnenberger [Fri, 22 Apr 2005 10:05:41 +0000 (10:05 +0000)]
Use normal variable names for rtsold, provide default values.

19 years agoRemove obsolete rcNG scripts as part of make upgrade.
Joerg Sonnenberger [Fri, 22 Apr 2005 09:14:08 +0000 (09:14 +0000)]
Remove obsolete rcNG scripts as part of make upgrade.

19 years agoDon't install wscons.
Joerg Sonnenberger [Fri, 22 Apr 2005 08:58:42 +0000 (08:58 +0000)]
Don't install wscons.

19 years agoAdd frequency correction support. The drift factor is currently
Joerg Sonnenberger [Fri, 22 Apr 2005 08:35:04 +0000 (08:35 +0000)]
Add frequency correction support. The drift factor is currently
determined via linear regression of the last 8 offset corrections.
It is bound to 128 ppm in one step to avoid extreme drifts when
the drift calculation is incorrect for some reason. With the
default values this can result in a maximum wrong drift of
8 * 60 * 8 * 128e-6 s ~= 0.5s (samples per update, update interval,
samples to trigger drift correction, max. drift error) until the
next iteration is done.