Commit Graph

32 Commits

Author SHA1 Message Date
Tom Rhodes
5d8e279e37 Fix a typo higter->higher.
Spotted by:	ganbold
2008-12-24 09:17:30 +00:00
Alexander Motin
8cb16fdbea Restore original frequency on exit.
PR:		bin/113813
2008-11-18 15:48:23 +00:00
Alexander Motin
dbd3197760 Set of powerd enchancements:
1. Make it more SMP polite. Previous version uses average CPU load that
often leads to load underestimation. It make powerd with default
configuration unusable on systems with more then 2 CPUs. I propose to use
summary load instead of average one. IMO this is the best we can do without
specially tuned scheduler. Also as soon as measuring total load on SMP
systems is more useful then total idle, I have switched to it.

2. Make powerd's operation independent from number and size of frequency
levels. I have added internal frequency counter which translated into real
frequencies only on a last stage and only as good as gone. Some systems may
have only several power levels, while others - many of them, so adaptation
time with previous approach was completely different.

3. As part of previous I have changed adaptive mode to rise frequency on
demand up to 2 times and fall on 1/8 per time internal.

4. For desktop (AC-powered) systems I have added one more mode - "hiadaptive".
It rises frequency twice faster, drops it 4 times slower, prefers twice
lower CPU load and has additional delay before leaving the highest frequency
after the period of maximum load. This mode was specially made to improve
interactivity of the systems where operation capabilities are more
significant then power consumption, but keeping maximum frequency all the
time is not needed.

5. I have reduced default polling interval from 1/2 to 1/4 of second.
It is not so important for algorithm math now, but gives better system
interactivity.

Discussed on:	mobile@
2008-11-18 13:24:38 +00:00
David E. O'Brien
dc70a966e3 Add an abbreviation for adaptive mode, and document all the abreviations. 2008-06-22 17:52:57 +00:00
Kevin Lo
8f9872ccb3 getopt(3) returns -1, not EOF. 2008-02-18 03:19:25 +00:00
Dmitry Morozovsky
89a630d080 Failing to set new frequency should not lead to powerd exiting.
Change err(3) to warn(3) as three other cases.

Approved by:	njl, des
2007-06-13 19:05:11 +00:00
Dag-Erling Smørgrav
f1e58a9ac4 Back out revs 1.18 and 1.19; they were based on incorrect assumptions.
Requested by:	nate
2006-02-19 00:40:28 +00:00
Dag-Erling Smørgrav
af2cde52b3 Look for cpu/ichss instead of cpu/p4tcc, which is i386/amd64-only. 2006-02-18 16:01:07 +00:00
Dag-Erling Smørgrav
aa5cfd0420 Load cpufreq at startup if it isn't there already. 2006-02-18 11:35:12 +00:00
Dag-Erling Smørgrav
2f7a934c4f Don't use threads when there's no real reason to.
MFC after:	2 weeks
2006-01-03 21:32:02 +00:00
Joel Dahl
19797b2256 s/5.5/6.0/ in HISTORY section.
Discussed with:	ru
2005-11-24 09:25:10 +00:00
Ruslan Ermilov
78ad54210f -mdoc sweep. 2005-11-18 10:56:28 +00:00
Pawel Jakub Dawidek
177b9d3b7b Be more consistent with the rest of the manual page. 2005-11-06 20:01:47 +00:00
Nate Lawson
33a00fc0b1 Create the devd thread after we have called daemon(). Otherwise, it would
be killed when the parent exits.

Submitted by:	Rudolf Cejka <cejkar / fit.vutbr.cz>
2005-10-24 18:34:54 +00:00
Nate Lawson
b194bdc2ae Whitespace fix. 2005-10-23 00:55:56 +00:00
Nate Lawson
72699a2216 Set the signal handlers before creating the thread. This is so it inherits
the parent's signal mask.  Once daemon() forked, signals would be ignored
in the child thread.  While I'm here, check the return value of daemon().
This fixes termination in the daemon case (bug introduced in last commit).

Noticed by:	Frederik Lindberg
2005-10-20 05:12:20 +00:00
Nate Lawson
a1819ab598 Use a pipe for reading devd events as another method of getting the AC line
status.  Add a thread that waits for events on the named pipe instead of
polling the line status via apm or the ACPI sysctl.  Additionally, use
robust error handling in case devd goes away or we temporarily can't set
a frequency (i.e., passive cooling has preempted our setting).

Later, this should be improved so that if adaptive control is not being used,
we will block while waiting for AC line events, saving a little CPU.

Submitted by:	Frederik Lindberg <fli+freebsd-current/shapeshifter.se>
2005-10-19 04:48:44 +00:00
Dag-Erling Smørgrav
db20dc4dbb Accept "maximum" and "minimum" modes as well as their abbreviations.
MFC after:	2 weeks
2005-09-26 12:16:27 +00:00
Pawel Jakub Dawidek
a9ed1f7ffc Test last minute changes, as gcc isn't too smart.
Reported by:	Steve Kargl <sgk@troutmask.apl.washington.edu>
2005-08-24 19:42:40 +00:00
Pawel Jakub Dawidek
84148fd1da Add '-P' option which allows to specify pidfile. 2005-08-24 17:32:41 +00:00
Bruno Ducrot
7065e0e9f4 Change the default adaptive algorithm for phk's one. Thanks to phk
and Kevin Oberman for this work.

Submitted by:	Kevin Oberman
Reviewed by:	njl
Approved by:	njl
2005-08-24 07:52:59 +00:00
Nate Lawson
c996f586de powerd(8) requires root permissions to run so notify the user if they
try to start it without this.

Submitted by:	Philip S. Schulz
MFC after:	2 days
2005-07-25 16:43:12 +00:00
Hajimu UMEMOTO
f92a4b93c5 Ignore EPERM from sysctl(3). It is returned when passive cooling
is active.

Reviewed by:	njl
2005-07-18 20:15:31 +00:00
Ruslan Ermilov
83c94dadce Fixed markup.
Approved by:	re
2005-06-16 20:01:43 +00:00
Nate Lawson
2f8ad6c087 Drop the default running value to 60% from 80%. Testing shows that we
cycle too much, wasting power if we trigger a jump to full speed too
easily.  CPU-bound tasks quickly push idle below this level.
2005-04-10 20:42:55 +00:00
Nate Lawson
ab19351c39 Add support for recording the total energy used (in joules) when in
verbose mode.  This is useful for profiling new adaptive algorithms
in performance (via time(1)) and total energy consumed for a given
workload.
2005-04-10 19:02:29 +00:00
Christian Brueffer
7f2f06281c man page -> manual page 2005-02-27 15:49:07 +00:00
Nate Lawson
48bd710904 Move the AC line checking code into its own function, reducing the
ifdef __i386__ stuff for APM.  Tested as working correctly on i386 and
compile-tested on sparc64.
2005-02-27 01:58:49 +00:00
Marcel Moolenaar
5f4aa96715 Unbreak !i386 platforms. Only i386 has APM, so don't try to fallback
on APM on other platforms.
2005-02-27 00:23:19 +00:00
Nate Lawson
b3c9af5615 Change the default levels for adaptive mode to 80% and 90%. On a system
with moderate IO going on, system usage can hover around 65-77% even
though it would benefit from higher performance.  Typically, only truly
idle systems pass the 90% mark so only demote then.
2005-02-26 23:22:00 +00:00
Nate Lawson
a9039b10b7 Remove SRCS line since it isn't needed for single-source programs.
Pointed out by:	ru
2005-02-26 22:50:25 +00:00
Nate Lawson
5883360b2f Import powerd(8), a utility for managing power consumption. Currently, it
just uses cpufreq(4) but in the future it should also do things like spin
down disks.  This is based on the work by Colin Percival (called "estctrl").
2005-02-26 21:17:31 +00:00