Commit Graph

64 Commits

Author SHA1 Message Date
Jung-uk Kim
2be4e4713a Catch up with ACPI-CA 20070320 import. 2007-03-22 18:16:43 +00:00
Nate Lawson
7826bf983c Add missing function trace for debug prints. 2007-01-23 07:20:44 +00:00
Nate Lawson
bd82680347 Clean up some debug prints from last commit and move one under boot -v.
Reminded by:	bruno
2007-01-15 18:17:36 +00:00
Nate Lawson
30dd6af310 Fix LINT and ACPI_DEBUG builds and add print for use of flush cache inst. 2007-01-08 00:45:46 +00:00
Nate Lawson
907b6777c1 Re-work Cx handling to be per-cpu and asymmetrical, fixing support on
modern dual-core systems as well.

- Parse the _CST packages for each cpu and track all the states individually,
on a per-cpu basis.

- Revert to generic FADT/P_BLK based Cx control if the _CST package
is not present on all cpus. In that case, the new driver will
still support per-cpu Cx state handling. The driver will determine the
highest Cx level that can be supported by all the cpus and configure the
available Cx state based on that.

- Fixed the case where multiple cpus in the system share the same
registers for Cx state handling. To do that, added a new flag
parameter to the acpi_PkgGas and acpi_bus_alloc_gas functions that
enable the caller to add the RF_SHAREABLE flag.  This flag could also be
useful to other callers (acpi_throttle?) in the tree but this change is
not yet made.

- For Core Duo cpus, both cores seems to be taken out of C3 state when
any one of the cores need to transition out. This broke the short sleep
detection logic.  It is disabled now if there is more than one cpu in
the system for now as it fixed it in my case.  This quirk may need to
be re-enabled later differently.

- Added support to control cx_lowest on a per-cpu basis. There is still
a generic cx_lowest to enable changing cx_lowest for all cpus with a single
sysctl and for ease of use.  Sample output for the new sysctl:

dev.cpu.0.cx_supported: C1/1 C2/1 C3/57
dev.cpu.0.cx_lowest: C3
dev.cpu.0.cx_usage: 0.00% 43.16% 56.83%
dev.cpu.1.cx_supported: C1/1 C2/1 C3/57
dev.cpu.1.cx_lowest: C3
dev.cpu.1.cx_usage: 0.00% 45.65% 54.34%
hw.acpi.cpu.cx_lowest: C3

This work was done by Stephane E. Potvin with some simple reworking by
myself.  Thank you.

Submitted by:	Stephane E. Potvin <sepotvin / videotron.ca>
MFC after:	2 weeks
2007-01-07 21:53:42 +00:00
Nate Lawson
80f006a1e3 If we're trying to use C2/3 and reads from the register are returning
immediately, back off to the next higher Cx sleep state.  Some machines
with a Via chipset report a valid C3 but a register read doesn't actually
halt the CPU.  This would cause the machine to appear unresponsive as it
repeatedly called cpu_idle() which immediately returned.  Causing interrupts
(i.e. by pressing the power button) would cause the system to make forward
progress, showing that it wasn't actually hung.

Also, enable interrupts a little earlier.  We don't need them disabled
to calculate the delta time for the read.

Reported by:	silby
MFC after:	2 weeks
2005-10-25 21:15:47 +00:00
David E. O'Brien
2a191126de Canonize the include of acpi.h. 2005-09-11 18:39:03 +00:00
Nate Lawson
f2d942579b Advertise that we can handle unified SMP control of processor power
states, idling, etc.  This has been supported since the cpufreq import.
2005-04-10 19:21:42 +00:00
Nate Lawson
bce9288570 Fix support for _PDC by using the proper version/length format for the
buffer.  Also, reference the Intel document where the _PDC values were
found.  This now supports ACPI-assisted SpeedStep on my borrowed T42.
2005-04-10 19:07:08 +00:00
Nate Lawson
b29224c2a0 Add the acpi_get_features() method. This method is called on child drivers
to see what features they may support before calling identify/probe/attach.
This is necessary because the ACPI 3.0 spec requires driver support be
advertised before running any methods.  For now, the flags are as specified
in for the _PDC and _OSC methods but we can support private flags as needed.

Add an implementation of this for acpi_cpu.  It checks all its children
(notably cpufreq drivers) and calls the _PDC method to report the results.
2005-04-04 15:46:57 +00:00
Nate Lawson
43ce1c7762 If a device_add_child fails (i.e. low memory situation), be sure to free
the unused ivars also.

Submitted by:	pjd
Obtained from:	Coverity Prevent analysis
2005-03-27 03:37:43 +00:00
Nate Lawson
8b888c66d7 Remove handling _PSS notifies from acpi_cpu and let acpi_perf handle them. 2005-02-07 04:03:06 +00:00
Nate Lawson
8c5468e3f5 Remove acpi throttling support from the acpi_cpu(4) driver now that this
is supported by acpi_throttle(4).
2005-02-06 21:10:19 +00:00
Nate Lawson
3cc2f17689 Notify the OS that we're taking over Px states in acpi_perf(4) instead of
doing it in the cpu driver.  The previous code was incorrect anyway since
this value controls Px states, not throttling as the comment said.  Since
we didn't support Px states before, there was no impact.  Also, note that
we delay the write to SMI_CMD until after booting is complete since it
sometimes triggers a change in the frequency and we want to have all
drivers ready to detect/handle this.
2005-02-06 20:12:28 +00:00
Nate Lawson
3045c8af3f Staticize the legacy cpu devclasses and revert the name for the acpi_cpu
devclass.  As pointed out by dfr@, devclasses don't have to share the same
linkage if multiple drivers have the same name.  Newbus should match the
devclasses based on name and allocate non-conflicting unit numbers.
2005-02-06 07:36:08 +00:00
Nate Lawson
f4eb041868 Convert to the new GAS API so that we can free registers in the future. 2005-02-05 22:29:03 +00:00
Nate Lawson
7d3a06201d Make the devclass static for now until deciding whether to share them. 2005-02-04 07:25:44 +00:00
Nate Lawson
98aa9cd052 Update the CPU attachments to return CPU_IVAR_PCPU as well as pass on
appropriate requests to any children.
2005-02-04 05:36:40 +00:00
Nate Lawson
ae56b59f8b Enable throttling/C3 quirks for PIIX4 parts. Defer checking quirks until
after boot so that PCI is initialized and we can probe for the problem
chipsets.  Note that while probed but unusable states are disabled, they
aren't freed yet.  In the future, it may make sense to detach them.

Tested by:	Adam K Kirchoff <adamk at voicenet com>
MFC after:	2 days
2004-11-16 18:47:42 +00:00
Nate Lawson
f435261d6d Update C3 support when BM control is not present.
* Fix a bug where caches were flushed on non-C3 transitions.
* Be sure a working flush cache instruction is present before using it.
* Disable C3 completely if it isn't present.
2004-10-11 21:15:10 +00:00
Nate Lawson
18ececa0b8 If bus mastering control is not available (PM2_BLK), don't just disable
C3.  Instead, flush caches before entering C3.  This may be slower but
provides good power savings.
2004-10-11 06:06:42 +00:00
Nate Lawson
31ad3b8802 Move the code for halting the CPU (acpi_cpu_c1) into machdep files.
This removes the last MD portion of acpi_cpu.c.

MFC after:	2 weeks
2004-10-11 05:39:15 +00:00
Nate Lawson
d92a2ebdb2 MPSAFE locking
* Hold the ACPI lock over table register writes.
* Serialize calls to acpi_cpu_throttle_set() and the sysctls.
2004-08-13 06:21:47 +00:00
Nate Lawson
4a03551ddd Use uintmax_t for CPU statistics and add a cast to prevent truncation of
the statistics in a multiply.

Pointed out by:	YONETANI Tomokazu
2004-06-24 00:38:51 +00:00
Nate Lawson
3e7fa136ea Add more precision to the cx_usage sysctl output and special-case 0%.
Submitted by:	YONETANI Tomokazu <qhwt+freebsd-acpi AT les.ath.cx>
2004-06-19 02:27:23 +00:00
Nate Lawson
a2afe45a8a Rework acpi_cpu_idle() to select the next idle state before sleeping, not
after.  Unify the paths for all Cx states.  Remove cpu_idle_busy and
instead do the little profiling we need before re-enabling interrupts.
Use 1 quantum as estimate for C1 sleep duration since the timer interrupt
is the main reason we wake.

While here, change the cx_history sysctl to cx_usage and report statistics
for which idle states were used in terms of percent.  This seems more
intuitive than counters.  Remove the cx_stats structure since it's no
longer used.  Update the man page.

Change various types which do not need explicit size.
2004-06-05 07:02:18 +00:00
Peter Wemm
f2b6954343 Work around the preemption problem in acpi_cpu.c for shutting down.
Submitted by:  nate / jhb
2004-06-04 17:03:18 +00:00
Poul-Henning Kamp
fe12f24bb0 Add missing <sys/module.h> includes 2004-05-30 20:08:47 +00:00
Nate Lawson
ccc09458fa Change hw.acpi.cpu.cx_lowest to accept values in the form of C1,
C2, ...  Update power_profile to use the new format.  Update the
man page to reflect this and give more info on Cx states.
2004-05-07 05:22:38 +00:00
Nate Lawson
b0e2b625f2 Rename acpi_cpu to cpu. Change the probe routine to early on reject
devices it cannot attach to.  This gets rid of extraneous but harmless
device_probe_and_attach() errors.  While I'm here, make the device
description more useful.  The !acpi case for cpu is handled by legacy0.
2004-05-06 17:25:23 +00:00
Nate Lawson
eea17c34cd Move the timer difference convenience function from acpi_cpu.c to make it
globally available.  acpi_TimerDelta() subtracts two readings from the
ACPI PM timer and returns the difference.  It properly distinguishes between
24-bit and 32-bit timers and handles wraparound.
2004-04-21 00:36:15 +00:00
Nate Lawson
64278df5e0 Add MODULE_DEPEND entries so some of these drivers can eventually be
loaded separately from ACPI (i.e., embedded use).
2004-04-09 18:14:32 +00:00
Nate Lawson
e5ada020f1 Fix border error to allow systems that specify 100 for latency also use
C2 and 1000 to use C3.

Submitted by:	Bruno Ducrot <ducrot@poupinou.org>
Tested by:	Scott Lambert <lambert@lambertfam.org>
2004-03-17 21:49:47 +00:00
Nate Lawson
08b994c07e Document a sysctl.
Submitted by:	Craig Rodrigues <rodrigc@crodrigues.org>
2004-03-05 18:08:23 +00:00
Nate Lawson
c181b89bc1 Don't disable Cx support and throttling on machines with a P_BLK_LEN != 6
even though the spec mandates this.  Some have a value of 5 to indicate
throttling + C2 and some have 7 to indicate an extra C3 state.  Support
throttling if the value is >= 4, C2 for >= 5, and C3 for >= 6.
2004-03-04 05:17:52 +00:00
Nate Lawson
5016979369 Don't attach throttling if the P_BLK is 0, even if the P_BLK_LEN is 6.
This is more strict but no known systems have this problem.
2003-12-28 22:15:24 +00:00
Nate Lawson
21cea91f6e Remove the device_t parameter from package routines that only used it to
print an error message.  Update all callers of the package routines.
2003-12-23 18:26:53 +00:00
Nate Lawson
0d8fb61a49 Remove power profile support from acpi_cpu, it will be managed by a
script run from devd(8).
2003-12-18 04:38:45 +00:00
Nate Lawson
b279c35a4f Fix throttling to use the proper mask. The bug resulted in only two
throttling values being available regardless of the CPU's capabilities.
This has been broken since rev 1.1.  Also clarify a comment.

Submitted by:	Taku YAMAMATO <taku@cent.saitama-u.ac.jp>
2003-12-12 19:42:16 +00:00
John Baldwin
4d5f2cbbd6 Trim trailing whitespace. 2003-12-10 19:10:27 +00:00
Nate Lawson
73a34dd424 We don't need to call _INI on processor objects now that ACPI-CA does
this as it should.
2003-12-09 03:01:54 +00:00
John Baldwin
447a5fa113 Update this driver to be more module friendly:
- Dynamically allocate the cpu_softc[] array based on mp_maxid instead of
  using a statically sized array that depended on 'options SMP'.
- Use mp_maxid rather than MAXCPU when walking all the CPUs looking for a
  match.
- Always call smp_rendezvous() since UP kernels now provide this.
- Use mp_ncpus rather than cpu_ndevices when determining if we need to
  disable C3 for SMP machines.

Approved by:	re (rwatson)
Reviewed by:	njl
2003-12-03 20:27:48 +00:00
Nate Lawson
cd1f3db9ba * If a processor's softc is NULL, use C1 since there is no ACPI
processor object for this CPU.  This occurs for logical CPUs which
  do not have an associated processor object (e.g., HTT).

Approved by:	re (rwatson)
2003-11-27 16:32:46 +00:00
Nate Lawson
b6426963c4 * Add acpi_pcpu_get_id(idx, *acpi_id, *cpu_id) which fetches the
idx'th present CPU with pc_acpi_id equal to *acpi_id.  If *acpi_id
  does not match that processor's pc_acpi_id, return the value for
  ProcId derived from the MADT in *acpi_id.  If pc_acpi_id is 0xffffffff,
  always override it with the value of *acpi_id.  Finally, return
  pc_cpuid in *cpu_id and use that as our primary key.

* Use pc_cpuid as our unique key because we know it is valid since
  MD code set it.  The values for ProcId in the ASL and MADT don't
  match up on some machines (!), forcing us to fall back to ordered
  probing in that case.

* Remove some #ifdef SMP since the refcount doesn't hurt performance
  and will be needed for dynamic _CST objects.  Only one #ifdef SMP
  (for smp_rendezvous) remains.

* Hook up SMP in the compile flags in the Makefile.

Tested by:	marcel, truckman
Approved by:	re (scottl)
2003-11-26 19:01:33 +00:00
Nate Lawson
56a70eadf5 * Add a DEVMETHOD for acpi so that child detach methods get called. Add
an acpi_cpu method for shutdown that disables entry to acpi_cpu_idle
  and then IPIs/waits for threads to exit.  This fixes a panic late in
  reboot in the SMP case.

* In the !SMP case, don't use the processor id filled out by the MADT
  since there can only be one processor.  This was causing a panic in
  acpi_cpu_idle if the id was 1 since the data was being dereferenced from
  cpu_softc[1] even though the actual data was in cpu_softc[0] (which is
  correct).

* Rework the initialization functions so that cpu_idle_hook is written
  late in the boot process.

* Make the P_BLK, P_BLK_LEN, and cpu_cx_count all softc-local variables.
  This will help SMP boxes that have _CST or multiple P_BLKs.  No such
  boxes are known at this time.

* Always allocate the C1 state, even if the P_BLK is invalid.  This means
  we will always take over idling if enabled.  Remove the value -1 as
  valid for cx_lowest since this is redundant with machdep.cpu_idle_hlt.

* Reduce locking for the throttle initialization case to around the write
  to the smi_cmd port.  Add disabled code to write the CST_CNT.  It will
  be enabled once _CST re-evaluation is tested (post 5.2R).

Thank you:	dfr, imp, jhb, marcel, peter
Tested by:	rwatson, Harald Schmalzbauer <h@schmalzbauer.de>
Approved by:	re (rwatson)
2003-11-19 20:27:06 +00:00
Nate Lawson
6b74f9b7f5 Implement Cx CPU idle states and updated throttling support.
* Use the cpu_idle_hook() to do idling for C1-C3.
* Use both _CST and the FADT to detect Cx states.
* Use both _PTC and P_CNT for controlling throttling.
* Add a notify handler to detect changes in _CST and _PSS
* Call the _INI function for each processor if present.  This will be
  done by ACPI-CA in the future.
* Fix a bug on SMP systems where CPUs will attach multiple times if the
  bus is rescan.
* Document new sysctls for controlling idling.
2003-11-15 19:26:06 +00:00
Nate Lawson
be2b179704 Style and whitespace changes. Also, make the ivar functions non-inline
since inlining failed due to the size of BUS_*
2003-08-28 16:06:30 +00:00
David E. O'Brien
aad970f1fe Use __FBSDID().
Also some minor style cleanups.
2003-08-24 17:55:58 +00:00
Nate Lawson
a40f20c791 More useful announce message containing current speed of CPU 2003-01-23 22:18:14 +00:00
Nate Lawson
d6b992c723 For the cpu throttling message, s/enabled/available
Requested by:	many
2003-01-14 19:39:41 +00:00