Add support for _PDC to the ACPI processor "Performance States library"
(perflib). If this field is empty, a bogus entry is passed to the _PDC
method so that the default (io) access is returned again. This patch
is partly based on David Moore's patch to
arch/i386/kernel/cpu/cpufreq/acpi.c, sent to the cpufreq mailing list on
June 24th, 2003.
[ACPI] Move the _PSS and _PCT access to drivers/acpi/processor.c
so that it can be used by various low-level drivers
(centrino, acpi-io, powernow-k{7,8}, ...)
from Dominik Brodowski
- remove unnecessary usage of flags.performance
- remove double check of _PPC in acpi-cpufreq driver as it's handled in
drivers/acpi/processor.c already
- remove unneeded EXPORT_SYMBOL
- allocation of memory only for probed CPUs
- add unregistration function to the core
- fix OOPS when PST has core_frequency values of zero
- fix cpufreq_get() output
- fix /proc/acpi/processor/*/performance write support [deprecated]
[ACPI] remove unnecessary check in acpi-cpufreq driver
from Dominik Brodowski
The acpi cpufreq driver includes a test at startup which detects whether
ACPI P-States are supported on any CPU, and whether transitions work.
However, this test is faulty: it is only run _after_ the acpi driver is
registered, causing race situations. Also, it doesn't save anything _as_ the
driver is already registered. So, it can safely be removed.
[ACPI] update passive cooling algorithm
from Dominik Brodowski
The current algorithm used by Linux ACPI for passive thermal management has
two shortcomings:
- if increasing the CPU processing power as a thermal situation goes away,
throttling states are decreased later than performance states. This is
not wise -- it should be the opposite ordering of going "up".
- only if the ACPI CPUfreq driver is used, performance states are used.
A generalized approach would offer passive cooling even if the ACPI
P-States cpufreq driver cannot be used (faulty BIOS, FixedHW access, etc.)
Jean Delvare [Tue, 27 Jan 2004 06:38:29 +0000 (22:38 -0800)]
[PATCH] I2C: Bring lm75 and lm78 in compliance with sysfs naming conventions
Here is a patch that brings the lm75 and lm78 drivers in compliance with
sysfs naming conventions. The drivers as found in existing 2.6 kernels
do not have a digit appended to the temperature-related files names as
the sysfs naming conversion recommends (obviously because they each have
a single temperature channel). As a result, libsensors won't find the
files.
It was discussed on the list wether a '1' should be appended in this
case, and our conclusion was that it would be better to do so because it
helps automatic processing of the sysfs exported files. Please apply if
you agree with this.
Alan Stern [Tue, 27 Jan 2004 06:37:40 +0000 (22:37 -0800)]
[PATCH] USB: Update sound/usb/usbaudio.c
On Tue, 27 Jan 2004, Greg KH wrote:
> Hm, can you send me a patch to fix up snd_usb_extigy_boot_quirk() in
> sound/usb/usbaudio.c now that this patch broke that code? :)
Here it is. The problem with changing things in the core is always that
you may miss some of the repercussions. In this case the patch will
restore the code's original functionality.
However the whole thing looks a bit creaky to me. Changing device
descriptors and who knows what else without informing usbcore isn't a good
idea. What this code _really_ appears to need is some form of
usb_device_reenumerate(). Such a function would fit very nicely into the
framework I've worked out for the revised usb_device_reset(); maybe I'll
add it in there. It would do essentially the same thing as device_reset
except for skipping the actual port reset.
Alan Stern [Mon, 26 Jan 2004 09:11:51 +0000 (01:11 -0800)]
[PATCH] USB: Don't dereference NULL actconfig
This patch fixes a simple error in a couple of utility routines. They
will no longer try to dereference a NULL actconfig pointer. Also, they
will work a little better if the configuration is changed while they are
running (which should never happen anyway).
Jean Delvare [Mon, 26 Jan 2004 08:58:46 +0000 (00:58 -0800)]
[PATCH] I2C: Add ADM1025EB support to i2c-parport
The following patch adds support for the ADM1025 evaluation board to the
i2c-parport (and i2c-parport-light) driver(s). In fact, it happens that
it was already supported as an ADM1032 evaluation board, so it is just a
matter of documenting it correctly.
James Simmons [Mon, 26 Jan 2004 06:49:42 +0000 (22:49 -0800)]
[PATCH] fbdev booting fix.
[FBCON] Fixed the order of which driver is used for the console. Before
the api change the last driver loaded became the default one. Now this is
not the case.
Andrew Morton [Mon, 26 Jan 2004 05:13:21 +0000 (21:13 -0800)]
[PATCH] kbuildL fix cscope index generation
From: Sam Ravnborg <sam@ravnborg.org>
cscope expect to find the list of files used for the database in a file
named cscope.files. Generate this file as part of 'make cscope'. This
solves http://bugme.osdl.org/show_bug.cgi?id=1948.
Andi Kleen [Mon, 26 Jan 2004 00:52:24 +0000 (16:52 -0800)]
[PATCH] Fix error checking in IPC_SET
The LSM changes broke the error checking for queue lengths in IPC_SET. The LSM check would
set set err to 0, but the next check expected it to still be -EPERM. Result was that
no error was reported, but the new parameters weren't correctly set.
Hirofumi Ogawa [Sat, 24 Jan 2004 10:57:20 +0000 (05:57 -0500)]
[netdrvr 8139cp] fix NAPI race
Andreas Happe <andreashappe@gmx.net> writes:
> my notebook (hp/compaq nx7000) still crashes when using 8139cp (runs
> rock solid with 8139too driver). The computer just locks up, there is no
> dmesg output. This has happened since I've got this laptop (around
> november '03).
It seems 8139cp.c has the race condition of rx_poll and interrupt.
NOTE, since I don't have this device, patch is untested. Sorry.
I got a Wacom Graphire3 for my birthday and unfortunately it didn't
work. After some playing around, I noticed the 2.6 kernel needs a few
small modifications to make it work.
This simple patch adds support for the Wacom Graphire 3.
Mark Haverkamp [Sat, 24 Jan 2004 04:54:17 +0000 (23:54 -0500)]
[PATCH] Fix for aacraid and high memory on 2.6.1
Here is an update to use pci_set_consistent_dma_mask. But since
dma_alloc_coherent uses dma_mask instead of consistent one, I left in
setting dma_mask as well until the alloc routine changes.
Andrew Morton [Sat, 24 Jan 2004 03:36:41 +0000 (19:36 -0800)]
[PATCH] The RAW_GETBIND compat_ioctl fails
From: James Cross <jscross@veritas.com>
The RAW_GETBIND compatibility ioctl call does convert properly between the
32bit/64bit version of raw_config_request due to a trivial error, and the
ioctl call fails.
projects / linux / log