From Dominik.

One big limitation of the ACPI specification is that it's impossible to
detect the current P-State by reading from ACPI-defined registers. And the
CPU isn't always at P0 when the system boots. So, try to "guess" the current
P-State by analyzing cpu_khz.

From Dominik.

If used as a bootparam, this would've become powernow-k7.powernow_acpi_force which looks silly.

Spotted by Charles Coffing <ccoffing@novell.com>

This breaks on x86-64 with the following warning.

drivers/cpufreq/cpufreq_userspace.c: In function `cpufreq_procctl':
drivers/cpufreq/cpufreq_userspace.c:170: warning: cast from pointer to integer of different size
drivers/cpufreq/cpufreq_userspace.c: In function `cpufreq_sysctl':
drivers/cpufreq/cpufreq_userspace.c:208: warning: cast from pointer to integer of different size

We don't need this, we can infer from CONFIG_ACPI_PROCESSOR

From Paul Devriendt

Decoding the legacy tables may have set these values.

As much as I like the idea of a 13GHz laptop, setting it to 1.3GHz is probably
for the best for the time being.

Due to the possible dependancy on ACPI.

acpi fallback is handled already in the init function.

More from Bruno Ducrot.

Warning: it will only half work on the ACER Aspire, though: there is
no pstate in the DSDT corresponding to the max frequency...
I'm looking how to handle that correctly, probably at the init stage,
if the max frequency is not given, then add it to powernow_table.

* Integrate acpi perflib from Dominik.
* Use acpi if the PST tables are known to be broken (ASUS Aspire match one PST,
  but give brain damaged values), or use ACPI if no PST found.

From: Dominik Brodowski <linux@dominikbrodowski.de>

Don't use the ACPI data on CPUs we don't know nothing about.

into delerium.codemonkey.org.uk:/mnt/nfs/sepia/bar/src/kernel/2.6/trees/cpufreq

From: Bruno Ducrot <ducrot@poupinou.org>

I think this patch is needed especially if the FSB is around 166MHz or
200MHz, or else I believe we get instabilities on some K7's motherboard
powernow capable (it's called Cool'n Quiet IIRC).
                                                                                         
                                                                                         
* Deduce fsb from cpu_khz and the max multiplier.  It will be given as kHz now,
  so that frequency associated to a multiplier will be computate more
  accurately.  Also, we need it for SGTC (see below).
* Fix how cpuid is computed in powernow_decode_bios().
* Be more restrictive for PST.  It may be possible (on desktop shipped with
  low power Athlon models) that FSB can be changed by dip switchs on
  motherboard for example.
* Fix computation for SGTC.  It use the bus timer (and then the bus
  frequency given by fsb).

From: Anton Blanchard <anton@samba.org>

We have to restore r13 when entering unrecoverable_exception.

From: Paul Mackerras <paulus@samba.org>

This patch changes PCI_DMA_TODEVICE to DMA_TO_DEVICE in a couple of
places in drivers/net/ibmveth.c, since it doesn't compile without this
change and it does compile with it.  It also reformats a couple of
over-long lines in the vicinity of the other changes.

From: Niraj Kumar <niraj17@iitbombay.org>

This is in continuation of the ufs2 read-only code that went into 2.6.5.

This patch fixes a bug where wrong content was being read off the disk
after around 4 MB mark.

From: Anton Blanchard <anton@samba.org>

It's been there since the kernel was first imported into bk.  We see no
reason for this.

From: Arjan van de Ven <arjanv@redhat.com>

root_nfs_name is called one in single threaded environment; can use static.

From: Jeff Mahoney <jeffm@suse.com>

This patch fixes a problem with the serial conversion to tiocm[sg]et.

The paste from rs_ioctl included the command sanity checking, but there's no
command for tiocm[sg]et.  The compile ends up failing.

From: "Luiz Fernando N. Capitulino" <lcapitulino@prefeitura.sp.gov.br>,
      me

drivers/pcmcia/rsrc_mgr.c: In function `find_io_region':
drivers/pcmcia/rsrc_mgr.c:604: warning: large integer implicitly truncated to unsigned type

We don't really know what underlying type an ioaddr_t has, so just use an
integer here and let the compiler promote it appropriately.

From: Samium Gromoff <deepfire@sic-elvis.zel.ru>

Without this one it fails to build, too.

From: Samium Gromoff <deepfire@sic-elvis.zel.ru>

Without this one it fails to build.

From: Nick Wellnhofer <wellnhofer@aevum.de>

hfs_fill_super in 2.6.5 returns -EIO instead of -EINVAL if a valid supe=
block isn't found.  So mount_block_root in init/do_mounts.c bails out before
trying to mount the root device as XFS.

From: <gerg@snapgear.com>

A few fixes for the 68328 "DragonBall" serial driver:

. use irqreturn_t for interrupt handlers
. correct a few variable types (stop compiler warnings)
. correctly use return values from put_user(), get_user() and copy_to_user()

Many of these originaly from kernel janitors.

From: <gerg@snapgear.com>

A whole bunch of fixes for the ColdFire serial driver:

. remove unused CONFIG_LEDMAN code
. reformat port definitions to new style structure init
. change "addr" field type to reduce casting in ColdFire serial driver
. cleanup locking problems in mcfrs_write().
. implement fraction baud rate clock support for hardware that
  supports it (namely the ColdFire 5272)
. implement wait_until_sent, some ColdFire parts of hardware support
  for this (again the 5272).
. correctly use return values from put_user(), get_user() and copy_to_user()

Many of these originaly from kernel janitors.

From: <gerg@snapgear.com>

Change "addr" field type to reduce casting in ColdFire serial driver.

From: Jeff Garzik <jgarzik@pobox.com>

DC390_init() takes a long, not an int.

40k reduction in my vmlinux.

Thanks to Denis Vlasenko <vda@port.imtp.ilyichevsk.odessa.ua> for performing
the analysis.

Saves 3.4k from my vmlinux.

Thanks to Denis Vlasenko <vda@port.imtp.ilyichevsk.odessa.ua> for performing
the analysis.

Shrinks my vmlinux by an astonishing 28k.

   text    data     bss     dec     hex filename
3038796  589890  150612 3779298b  39aae2 vmlinux.before
3009761  590107  150612 3750480  393a50 vmlinux.after

Thanks to Denis Vlasenko <vda@port.imtp.ilyichevsk.odessa.ua> for performing
the analysis.

From: Vincent Sanders <vince@kyllikki.org>

The vga16fb driver uses a direct ioremap on 0xa00000 to gain access to the
vga card.  This is wrong on architectures other than x86, every other driver
uses VGA_MAP_MEM macro from vga.h to ensure the correct memory mapping.

From: Matt Mackall <mpm@selenic.com>

Base hash sizes on available memory rather than total memory.  An
additional 50% above current used memory is considered reserved for the
purposes of hash sizing to compensate for the hashes themselves and the
remainder of kernel and userspace initialization.

From: Russell King <rmk@arm.linux.org.uk>

Cleanup the 4 duplicate "get_files_struct" implementations into one
get_files_struct() function to compliment put_files_struct().

From: William Lee Irwin III <wli@holomorphy.com>

This optimisation was reverted when I was removing all users of page->list.
Bill fixed it up, so unrevert it again.