Fix to return a proper cpu speed. it is caused by mistaking the meaning
of on and off.
Signed-Off-by: Hiroshi Miura <miura@da-cha.org>
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

From: "Pallipadi, Venkatesh" <venkatesh.pallipadi@intel.com>

The transient MSR values are for very short duration and so I feel reading
out the last set frequency is better than looping until reading something
expected.
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Dave Jones <davej@redhat.com>

From: "Pallipadi, Venkatesh" <venkatesh.pallipadi@intel.com>

In P4, CPU tsc rate won't change with CPU frequency change while using Enhanced
Speedstep Technology. Tested with both speedstep-centrino and acpi-cpufreq on
both i386 and x86-64.
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

Dothan is stepping 13  == 0x0D instead of 0x13 == 19

This bug crept bag in the shadows caused by the debug overhauling patch.
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

From: Bruno Ducrot <ducrot@poupinou.org>

dbs_timer_init() will schedule_work(&dbs_work) so that do_dbs_timer()
is likely called very early at same time slice scheduler, and
for each cpu, dbs_check_cpu(cpu) will be called, likely at the same time
slice than dbs_timer_init().
So far, we will get total_idle_ticks == idle_ticks == 0 unconditonaly,
and the processor will be put at full speed even though it's not needed
at startup.


Ack'ed by Venkatesh Pallipadi.
Signed-off-by: Bruno Ducrot <ducrot@poupinou.org>
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

Remove the deprecated /proc/cpufreq interface. The same input is available
using cpufreq-info --proc from cpufrequtils.
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

Remove the deprecated /proc/sys/cpu/ interface to cpufreq.
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

Patch to remove speedstep-centrino error messages getting printed by default.
Print them only when debug flags are enabled.

The reason for this patch is -
With the multiple drivers model that we have now, any installation will try
different drivers one after the other, until one of them succeeds. So,
failure to add speedstep-centrino alone doesn't mean error, as some other
driver (say acpi.ko) can succeed later and system will still be able to use
speedstep. Printing the error whenever speedstep-centrino fails can confuse
the enduser.
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

On some CPUs, we can see transient MSR values (which are not present in _PSS)
in IA32_PERF_STATUS MSR, while CPU is doing some automatic P-state transition
(like TM2).
Current code will return frequency as 0 in such cases. Fix it by retrying the
get after a delay and use lowest possible frequency as the current frequency
in worst case.

Thanks to Matt Domsch for identifying and root-causing this failure.
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

Venkatesh Pallipadi made me aware of a memory leak in speedstep-centrino:
centrino_model is allocated for all CPUs. There were two possibilities: either
make the centrino_model data per-CPU, or to share it across CPUs. I chose the
former variant, as ACPI data may be broken and/or different for multiple CPUs.

Additionally, I made centrino_cpu per-CPU, and removed a check whether setting
allowed_cpus to policy->cpus resulted in smp_processor_id() == policy->cpu:
if policy->cpus has more than one bit set, this check will fail very often.
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

From:  "Zou, Nanhai" <nanhai.zou@intel.com>

Export cpufreq transition information for drivers using the freq-table helpers
via sysfs.

Two minor updates from Dominik Brodowski:
s/translation/transition
Kconfig re-ordering
Signed-off-by: Dave Jones <davej@redhat.com>

From: Paul Devriendt

patch for powernow-k8 problem (Mobile Sempron 2800+, Acer Aspire 1362 )

If the initial frequency/voltage pair are not valid in the frequency table,
the first requested transition is to make them valid. Fix the code doing so.
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

Use the unified cpufreq debug infrastructure in the cpufreq-nforce2 driver.
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

Assert that the call to the cpufreq governor with CPUFREQ_GOV_STOP is really
the last. Without this patch, some strange in-kernel preemption combined with
the scheduler disliking the "removing" task may cause the opposite.
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

From: Pavel Machek <pavel (AT) ucw.cz>

These are very small cleanups / documentation additions. It avoids
using different names for same fields in different structures.

Updated to latest cpufreq-bk by Dominik Brodowski, and ack'ed by Mark
Langsdorf, Paul Devriendt and Pavel Machek.
Signed-off-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

Upon resuming from sleep or swsusp, the ACPI P-States driver can't determine
the current CPU frequency, as the ACPI specification doesn't contain a method
to determine the current P-State.

Therefore, _always_ re-set the CPU to the P-State it was before suspending,
and don't abort early if this is the same state as the CPU was put to
before (like it does make sense when using the ondemand governor, for example).
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

(if anyone has a brown spare paper bag, feel free to send it to me:)

The call to cpufreq_driver->resume() got lost in 2.6.6. Re-add it at the
proper place.
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

If something fails in the per-CPU initialization in powernow-k8, it should
unregister itself from the ACPI performance library.
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

The unregistration of the ACPI performance library should be done in the
CPU exit function, and the cleanup too.
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

The following patch makes ondemand governor aware of policy->cpus.

policy->cpus mask lets multiple cpu use the same policy (useful where cpus
share the frequency state), a recent change in cpufreq core.

Now ondemand governor looks at all cpus in policy->cpus and takes its
frequency increase/decrease decisions based on most lightly loaded cpu in
the group. This patch will not affect systems where policy->cpus contain
only one cpu.
Signed-off-by: "Venkatesh Pallipadi" <venkatesh.pallipadi@intel.com>
Signed-off-by: Dominik Brodowski <linux@brodo.de>
Signed-off-by: Dave Jones <davej@redhat.com>

Originally submitted by Mike Waychison <Michael.Waychison@sun.com> in
August but apparently lost.
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

SCSI ioctls can ask for a lot of memory and fail. We don't need to vomit
in the log file for this case. Again taken from the Red Hat minor
patches applied for FC3.

Original-patch: Arjan van de Ven <arjanv@redhat.com>
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Paul Laufer informed the list that he had changed address and his change
of address had been ignored so CREDITS was still wrong although other
files had been updated.

Fix this.
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Red Hat got some confused customers due to this message. The confused
user case is when they update the BIOS and all of a sudden we have "no
suitable data" yet we did before. We (Arjan van de Ven) thus changed it
to "No new microcode" which is much much clearer.
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Add some more funky AC97 knowledge to the intel8x0 driver. These come
from Red Hat and its partners and are included in our shipping code.
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

into ppc970.osdl.org:/home/torvalds/v2.6/linux

into ppc970.osdl.org:/home/torvalds/v2.6/linux

into pobox.com:/garz/repo/net-drivers-2.6

into pobox.com:/garz/repo/net-drivers-2.6

into pobox.com:/garz/repo/net-drivers-2.6

into nuts.davemloft.net:/disk1/BK/net-2.6

Merry Christmas everyone. Ho ho ho!

rth tells me that some versions of gcc may end up using the
SSE registers for data movement when you do that.

Use "-march=i686 -mtune=xxxx" instead.

(We do the same thing for march=pentium2/4 too, just for
consistency).

evdev, joydev, mousedev, tsdev - remove class device and devfs entry when
hardware driver disconnects instead of waiting for the last user to drop off. 
This way hardware drivers can be unloaded at any time.
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This patch adds missing lock_kernel()/unlock_kernel() pair in
reiserfs_get_dentry
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

mm/rmap.c contains an open-coded reference to swap_token_default_timeout
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

CFQ v2 has some spare queue logic that was never enabled.  It has an SMP
deadlock because it attempts to regrab the queue lock in the exit path, so
kill the spare queue stuff completely for 2.6.10.
Signed-off-by: Jens Axboe <axboe@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>