a2249ef75c
Make ntp, pmc, reboot, sysarch, time syscalls MPSAFE.
2008-04-21 12:56:30 +00:00
59c1cd1667
Disable the event counters by default, since they create terrible cache
...
behaviour on MP systems.
2008-02-10 13:56:17 +00:00
3615cf7715
Now that __HAVE_TIMECOUNTER and __HAVE_GENERIC_TODR are invariants,
...
remove the conditionals and the code associated with the undef case.
2008-01-20 18:09:03 +00:00
63baaccb6d
Serialize calls to tc_windup().
2008-01-05 18:00:37 +00:00
a72fb242fe
Add tc_detach() for detaching a timecounter. It is good enough to
...
detach the AMD Geode SC1100 27 MHz clock, geodecntr(4), but there
may be synchronization problems on MP machines.
2008-01-03 04:42:13 +00:00
5755384fd5
fix a comment.
2007-12-15 18:20:11 +00:00
b470ab628d
Use membar_*().
2007-11-30 23:05:43 +00:00
18558073a1
Kill another instance of KAUTH_GENERIC_ISSUSER.
2007-11-23 16:03:47 +00:00
85a2e7d924
Work around broken MD headers.
2007-11-16 01:21:24 +00:00
ab19d73932
Kill the memory barriers: force users of the outgoing timehand to move
...
on quickly by also incrementing its generation number.
2007-11-15 23:16:55 +00:00
d3c8987b25
Oops; fix thinko.
2007-11-15 22:28:05 +00:00
0239b151ae
Add a bit of locking around timecounter attachment / selection.
2007-11-15 20:12:04 +00:00
e7d1e5e9ea
Adjust a comment. In NetBSD the time-related kernel function prototypes
...
are in <sys/timevar.h>, not <sys/time.h> like they are in FreeBSD.
2007-10-21 14:55:09 +00:00
26c3495f7d
Timecounters are lockless. Add conservative memory barriers to ensure that
...
loads and stores occur in the correct order.
2007-08-17 21:20:24 +00:00
53524e44ef
Kill caddr_t; there will be some MI fallout, but it will be fixed shortly.
2007-03-04 05:59:00 +00:00
b07ec3fc38
Merge newlock2 to head.
2007-02-09 21:55:00 +00:00
1e70d64818
Consistent usage of KAUTH_GENERIC_ISSUSER.
2007-01-04 16:55:29 +00:00
1a7bc55dcc
remove some __unused from function parameters.
2006-11-01 10:17:58 +00:00
d094ff8f40
Avoid printf() the selected timecounter as it is called in the path of hardclock().
...
The information is available via sysctl anyway.
2006-10-21 07:16:54 +00:00
4d595fd7b1
- sprinkle __unused on function decls.
...
- fix a couple of unused bugs
- no more -Wno-unused for i386
2006-10-12 01:30:41 +00:00
182632b893
Reduce hardclock() code path length wrt/ multiply/divide by
...
re-calculating the th_scale value only when necessary.
As discussed with gdamore@. Tested at (sub-)usec level with
stratum 1 reference.
2006-09-24 06:39:28 +00:00
3134621b5b
Fix splclock()/splx() mismatch in previous.
2006-09-10 08:09:46 +00:00
a6e21ef05b
Guard the timecounter manipulations including the call to tc_windup()
...
in tc_init() with splclock(). Fixes doubled-up "selected timecounter"
messages on some architectures, including pc532 and cobalt.
Fix suggested by Frank Kardel.
2006-09-07 15:48:14 +00:00
28ea22fb52
use c99 initializers, per gimpy's request
2006-09-03 06:40:39 +00:00
e23bdf3c6e
move MAX_TCNAMELEN into timetc.h, use constant in tmp buffer allocation
2006-09-02 13:45:04 +00:00
bdd0dbdf82
add missing initializers
2006-09-02 06:35:12 +00:00
47ef6bac03
Using humanize_number() on clock speeds does more harm than good, and
...
precious little of either. Go back to displaying them with all their digits.
2006-08-06 13:29:42 +00:00
fc39059504
Clock frequencies tend to be big numbers -- use humanize_number() when
...
displaying them through printf.
2006-08-05 21:59:40 +00:00
f474dceb13
Use the LWP cached credentials where sane.
2006-07-23 22:06:03 +00:00
7e5391a22a
rename boottimebin to timebasebin as this
...
struct actually keeps the start of the UTC
time scale and not the boot time. the relationship
is: utc-time = up-time + timebase.
background: when doing an ACPI sleep the uptime
freezes and on wakeup the tc_setclock() leads to
a new timebasebin - this had no relationship with
a boottime as the structure was previously called.
discussed on tech-kern@
anomalies (moving boottime, uptime describing running time)
where discovered by Arnaud Lacombe.
2006-07-14 23:01:12 +00:00
54cd6fafa8
re-order initialization sequence to have real counters available during autoconfig
2006-06-09 22:47:56 +00:00
de4337ab21
merge FreeBSD timecounters from branch simonb-timecounters
...
- struct timeval time is gone
time.tv_sec -> time_second
- struct timeval mono_time is gone
mono_time.tv_sec -> time_uptime
- access to time via
{get,}{micro,nano,bin}time()
get* versions are fast but less precise
- support NTP nanokernel implementation (NTP API 4)
- further reading:
Timecounter Paper: http://phk.freebsd.dk/pubs/timecounter.pdf
NTP Nanokernel: http://www.eecis.udel.edu/~mills/ntp/html/kern.html
2006-06-07 22:33:33 +00:00
769d33c547
Import main FreeBSD timecounter support files as of 4th Feb 2006.
2006-02-04 08:06:23 +00:00
ad