GRU driver: minor updates

A few minor updates for the GRU driver.
- documentation changes found in code reviews
- changes to #ifdefs to make them recognized by "unifdef"
  (used in simulator testing)
- change GRU context load/unload to prefetch data

[akpm@linux-foundation.org: fix typo in comment]
Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Configure out AIO support

This patchs adds the CONFIG_AIO option which allows to remove support
for asynchronous I/O operations, that are not necessarly used by
applications, particularly on embedded devices. As this is a
size-reduction option, it depends on CONFIG_EMBEDDED. It allows to
save ~7 kilobytes of kernel code/data:

   text    data     bss     dec     hex filename
1115067  119180  217088 1451335  162547 vmlinux
1108025  119048  217088 1444161  160941 vmlinux.new
  -7042    -132       0   -7174   -1C06 +/-

This patch has been originally written by Matt Mackall
<mpm@selenic.com>, and is part of the Linux Tiny project.

[randy.dunlap@oracle.com: build fix]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Benjamin LaHaise <bcrl@kvack.org>
Cc: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Matt Mackall <mpm@selenic.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

w1: new driver. DS2431 chip

[akpm@linux-foundation.org: minor fixlets and cleanups]
Signed-off-by: Bernhard Weirich <bernhard.weirich@riedel.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Cc: Ben Gardner <bgardner@wabtec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds2490.c optimize ds_set_pullup

Optimize the ds_set_pullup function.  For a strong pullup to be sent the
ds2490 has to have both the strong pullup mode enabled, and the specific
write operation has to have the SPU bit enabled.  Previously the write
always had the SPU bit enabled and both the duration and model was set
when a strong pullup was requested.  Now the strong pullup mode is enabled
at initialization time, the delay is updated only when the value changes,
and the write SPU bit is set only when a strong pullup is required.  This
removes two or three bus transactions per strong pullup request.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: Documentation/w1/masters/ds2490 update

Provide some additional details about the status of the driver and the
ds2490 hardware.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds2490.c ds_write_block remove extra ds_wait_status

Drop the extra ds_wait_status() in ds_write_block().

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds2490.c magic number work

This replaces some magic numbers with marcos and corrects one marco.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds2490.c reset ds2490 in init

Reset the device in init as it can be in a bad state.  This is necessary
because a block write will wait for data to be placed in the output buffer
and block any later commands which will keep accumulating and the device
will not be idle.  Another case is removing the ds2490 module while a bus
search is in progress, somehow a few commands get through, but the input
transfers fail leaving data in the input buffer.  This will cause the next
read to fail see the note in ds_recv_data.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds2490.c ds_reset remove ds_wait_status

ds_reset no longer calls ds_wait_status, the result wasn't used and it
would only delay the following data operations.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds2490.c ds_dump_status rework

- add result register #defines
- rename ds_dump_status to ds_print_msg
- rename ds_recv_status to ds_dump_status
- ds_dump_status prints the requested status and no longer reads the
  status, this is because the second status read can return different
  data for example the result register
- the result register will be printed, though limited to detecting a
  new device, detecting other values such as a short would require
  additional reporting methods
- ST_EPOF was moved to ds_wait_status to clear the error condition
  sooner

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds2490.c simplify and fix ds_touch_bit

Simplify and fix ds_touch_bit.  If a device is attached in the middle of a
bus search the status register will return more than the default 16 bytes.
 The additional bytes indicate that it has detected a new device.  The way
ds_wait_status is coded, if it doesn't read 16 status bytes it returns an
error value.  ds_touch_bit then will detect that error and return an
error.  In that case it doesn't read the input buffer and returns
uninitialized data.  It doesn't stop there.  The next transaction will not
expect the extra byte in the input buffer and the short read will cause an
error and clear out both the old byte and new data in the input buffer.

Just ignore the value of ds_wait_status.  It is still required to wait
until ds2490 is again idle and there is data to read when ds_recv_data is
called.  This also removes the while loop.  None of the other commands
wait and verify that the issued command is in the status register.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds2490.c disable bit read and write

Don't export read and write bit operations, they didn't work, they weren't
used, and they can't be made to work.  The one wire low level bit
operations expect to set high or low levels, the ds2490 hardware only
supports complete read or write time slots, better to just comment them
out.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds2490.c ds_write_bit, grouping error, disable readback

ds_write_bit doesn't read the input buffer, so add COMM_ICP and a comment
that it will no longer generate a read back data byte.  If there is an
extra data byte later on then it will cause an error and discard what data
was there.  Corrected operator ordering for ds_send_control.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds2490.c add support for strong pullup

Add strong pullup support for ds2490 driver, also drop mdelay(750), which
busy waits, usage in favour of msleep for long delays.  Now with msleep
only being called when the strong pullup is active, one wire bus
operations are only taking minimal system overhead.

The new set_pullup will only enable the strong pullup when requested,
which is expected to be the only write operation that will benefit from a
strong pullup.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds2490.c correct print message

Corrected print message, it was writing not reading, this also prints the
endpoint used for the write instead of hardcoding it.  Failed to write
1-wire data to ep0x%x: err=%d.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: ds1wm.c msleep for reset

Like the previous w1_io.c reset coments and msleep patch, I don't have the
hardware to verify the change, but I think it is safe.  It also helps to
see a comment like this in the code.  "We'll wait a bit longer just to be
sure." If they are going to calculate delaying 324.9us, but actually delay
500us, why not just give up the CPU and sleep?  This is designed for a
battery powered ARM system, avoiding busywaiting has to be good for
battery life.

I sent a request for testers March 7, 2008 to the Linux kernel mailing
list and two developers who have patches for ds1wm.c, but I didn't get
any respons.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: w1_io.c reset comments and msleep

w1_reset_bus, added some comments about the timing and switched to msleep
for the later delay.  I don't have the hardware to test the sleep after
reset change.  The one wire doesn't have a timing requirement between
commands so it is fine.  I do have the USB hardware and it would be in big
trouble with 10ms interrupt transfers to find that the reset completed.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: w1.c s/printk/dev_dbg/

s/printk/dev_dbg/

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: w1_int.c use first available master number

Follow the example of other devices (like the joystick device).  Pick the
first available id for each detected device.  Currently for USB devices,
suspending and resuming would cause the number to increment.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: w1_therm consistent mutex access code cleanup

sl->master->mutex and dev->mutex refer to the same mutex variable, but be
consistent and use the same set of pointers for the lock and unlock calls.
 It is less confusing (and one less pointer dereference this way).

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: w1_family, remove unused variable need_exit

Removed the w1_family structure member variable need_exit.  It was only
being set and never used.  Even if it were to be used it is a polling type
operation.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: w1_therm fix user buffer overflow and cat

Fixed data reading bug by replacing binary attribute with device one.

Switching the sysfs read from bin_attribute to device_attribute.  The data
is far under PAGE_SIZE so the binary interface isn't required.  As the
device_attribute interface will make one call to w1_therm_read per file
open and buffer, the result is, the following problems go away.

buffer overflow:
Execute a short read on w1_slave and w1_therm_read_bin would still
return the full string size worth of data clobbering the user space
buffer when it returned.  Switching to device_attribute avoids the
buffer overflow problems.  With the snprintf formatted output dealing
with short reads without doing a conversion per read would have
been difficult.
bad behavior:
`cat w1_slave` would cause two temperature conversions to take place.
Previously the code assumed W1_SLAVE_DATA_SIZE would be returned with
each read.  It would not return 0 unless the offset was less
than W1_SLAVE_DATA_SIZE.  The result was the first read did a
temperature conversion, filled the buffer and returned, the
offset in the second read would be less than
W1_SLAVE_DATA_SIZE and also fill the buffer and return, the
third read would finnally have a big enough offset to return 0
and cause cat to stop.  Now w1_therm_read will be called at
most once per open.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: w1_slave_read_id read bug, use device_attribute

Fix bug reading the id sysfs file.  If less than the full 8 bytes were
read, the next read would start at the first byte instead of continuing.
It needed the offset added to memcpy, or the better solution was to
replace it with the device attribute instead of bin attribute.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: Document add, remove, search_count, and pullup.

Document w1_master_add, w1_master_remove, search_count, and pullup.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: new module parameter search_count

Added a new module parameter search_count which allows overriding the
default search count.  -1 continual, 0 disabled, N that many times.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: recode w1_slave_found logic

Simplified the logic in w1_slave_found by using the new
w1_attach_slave_device function to find a slave and mark it as active or
add the device if the crc checks.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: be able to manually add and remove slaves

sysfs entries were added to manually add and remove slave devices.  This
is useful if the automatic bus searching is disabled, and the device ids
are already known.

[akpm@linux-foundation.org: fix printk types]
Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: feature, w1_therm.c use strong pullup and documentation

Added strong pullup to thermal sensor driver and general documentation on
the sensor.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: feature, enable hardware strong pullup

Add a strong pullup option to the w1 system.  This supplies extra power
for parasite powered devices.  There is a w1_master_pullup sysfs entry and
enable_pullup module parameter to enable or disable the strong pullup.

The one wire bus requires at a minimum one wire and ground.  The common
wire is used for sending and receiving data as well as supplying power to
devices that are parasite powered of which temperature sensors can be one
example.  The bus must be idle and left high while a temperature
conversion is in progress, in addition the normal pullup resister on
larger networks or even higher temperatures might not supply enough power.
 The pullup resister can't provide too much pullup current, because
devices need to pull the bus down to write a value.  This enables the
strong pullup for supported hardware, which can supply more current when
requested.  Unsupported hardware will just delay with the bus high.

The hardware USB 2490 one wire bus master has a bit on some commands which
will enable the strong pullup as soon as the command finishes executing.
To use strong pullup, call the new w1_next_pullup function to register the
duration.  The next write command will call set_pullup before sending the
data, and reset the duration to zero once it returns.

Switched from simple_strtol to strict_strtol.

Signed-off-by: David Fries <david@fries.net>
Cc: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: w1_process, block or sleep

The w1_process thread's sleeping and termination has been modified.
msleep_interruptible was replaced by schedule_timeout and schedule to
allow for kthread_stop and wake_up_process to interrupt the sleep and the
unbounded sleeping when a bus search is disabled.  The W1_MASTER_NEED_EXIT
and flags variable were removed as they were redundant with
kthread_should_stop and kthread_stop.  If w1_process is sleeping,
requesting a search will immediately wake it up rather than waiting for
the end of msleep_interruptible previously.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: don't delay search start

Move the creation of the w1_process thread to after the device has been
initialized.  This way w1_process doesn't have to check to see if it has
been initialized and the bus search can proceed without sleeping.  That
also eliminates two checks in the w1_process loop.  The sleep now happens
at the end of the loop not the beginning.

Also added a comment for why the atomic_set was 2.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: abort search early on on exit

Early abort if the master driver or the hardware goes away in the middle
of a bus search operation.  The alternative is to spam the print buffer up
to 64*64 times with read errors in the case of USB.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

W1: fix deadlocks and remove w1_control_thread

w1_control_thread was removed which would wake up every second and process
newly registered family codes and complete some final cleanup for a
removed master.  Those routines were moved to the threads that were
previously requesting those operations.  A new function
w1_reconnect_slaves takes care of reconnecting existing slave devices when
a new family code is registered or removed.  The removal case was missing
and would cause a deadlock waiting for the family code reference count to
decrease, which will now happen.  A problem with registering a family code
was fixed.  A slave device would be unattached if it wasn't yet claimed,
then attached at the end of the list, two unclaimed slaves would cause an
infinite loop.

The struct w1_bus_master.search now takes a pointer to the struct
w1_master device to avoid searching for it, which would have caused a
lock ordering deadlock with the removal of w1_control_thread.

Signed-off-by: David Fries <david@fries.net>
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/char/tpm/tpm.c: fix error-path memory leak

tpm_register_hardware() leaks devname on an error path.

Addresses http://bugzilla.kernel.org/show_bug.cgi?id=11425

Reported-by: Daniel Marjamki <danielm77@spray.se>
Cc: Debora Velarde <debora@linux.vnet.ibm.com>
Cc: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
Cc: Marcel Selhorst <tpm@selhorst.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

TPM: include "moderated for non-subscribers" notation in MAINTAINERS

Cc: Eric Paris <eparis@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

tpm: work around bug in Broadcom BCM0102 chipset

Patch tpm-correct-tpm-timeouts-to-jiffies-conversion reveals a bug in the
Broadcom BCM0102 TPM chipset used in the Dell Latitude D820 - although
most of the timeouts are returned in usecs as per the spec, one is
apparently returned in msecs, which results in a too-small value leading
to a timeout when the code treats it as usecs.  To prevent a regression,
we check for the known too-short value and adjust it to a value that makes
things work.

Signed-off-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Cc: Marcin Obara <marcin_obara@users.sourceforge.net>
Cc: Marcel Selhorst <tpm@selhorst.net>
Cc: Kylene Jo Hall <kjhall@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

tpm: correct tpm timeouts to jiffies conversion

This patch fixes timeouts conversion to jiffies, by replacing
msecs_to_jiffies() calls with usecs_to_jiffies().  According to TCG TPM
Specification Version 1.2 Revision 103 (pages 166, 167) TPM timeouts and
durations are returned in microseconds (usec) not in miliseconds (msec).

This fixes a long hang while loading TPM driver, if TPM chip starts in
"Idle" state instead of "Ready" state.  Without this patch - 'modprobe'
may hang for 30 seconds or more.

Signed-off-by: Marcin Obara <marcin_obara@users.sourceforge.net>
Cc: Marcel Selhorst <tpm@selhorst.net>
Cc: Kylene Jo Hall <kjhall@us.ibm.com>
Cc: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

parport: remove CVS keywords

Remove CVS keywords that weren't updated for a long time from comments.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

afs: convert to new aops

Cannot assume writes will fully complete, so this conversion goes the easy
way and always brings the page uptodate before the write.

[dhowells@redhat.com: style tweaks]
Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

edac i5000: fix thermal issues

Make the Thermal messages (temperature got past Tmid) be displayed only
once because:

1) it's the BIOS job to configure and handle the memory throttling
2) if the BIOS is broken or is aware about the condition, flooding the
   system logs won't help anything.
3) According to the specification update for Intel 5000 MCHs, all the
   revisions of this MCH have problems on the thermal sensors, making
   not automatic (a.k.a. intelligent thermal throttling) impossible.

Signed-off-by: Aristeu Rozanski <aris@redhat.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

edac i5000: fix error messages

Update the i5000_edac messages, making everything pass through the EDAC
(so the log controls will work) and being more specific about the errors.
Also, it makes the miscellaneous errors optional and disabled by default.

As I didn't found anywhere information about M23ERR-M26ERR
(FERR_NF_THERMAL) on FERR_NF_FBD, I'm removing them.

Signed-off-by: Aristeu Rozanski <aris@redhat.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

edac mpc85xx: add support for mpc8572

This adds support for the dual-core MPC8572 processor.  We have
to support making SPR changes on each core.  Also, since we can
have multiple memory controllers sharing an interrupt, flag the
interrupts with IRQF_SHARED.

Signed-off-by: Andrew Kilkenny <akilkenny@xes-inc.com>
Signed-off-by: Nate Case <ncase@xes-inc.com>
Acked-by: Dave Jiang <djiang@mvista.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

edac: make i82443bxgx_edac coexist with intel_agp

Fix 443BX/GX MCH suppport in a EDAC.

It makes i82443bxgx_edac coexist with intel_agp using the same approach as
several other EDAC drivers.

Tested on Intel's L443GX with redhat's 2.6.18 with whole EDAC subsystem
backported a while ago.

[root@host ~]# dmesg|grep -iE '(AGP|EDAC)'
Linux agpgart interface v0.101 (c) Dave Jones
agpgart: Detected an Intel 440GX Chipset.
agpgart: AGP aperture is 64M @ 0xf8000000
EDAC MC: Ver: 2.1.0 Jun 27 2008
EDAC MC0: Giving out device to 'i82443bxgx_edac' 'I82443BXGX': DEV 0000:00:00.0
EDAC PCI0: Giving out device to module 'i82443bxgx_edac' controller 'EDAC PCI controller': DEV '0000:00:00.0' (POLLED)

Signed-off-by: Vladislav Bogdanov <slava@nsys.by>
Cc: Doug Thompson <norsk5@yahoo.com>
Cc: Dave Airlie <airlied@linux.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pid_ns: kill the now unused task_child_reaper()

task_child_reaper() has no callers anymore, kill it.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: Serge Hallyn <serue@us.ibm.com>
Acked-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pid_ns: de_thread: kill the now unneeded ->child_reaper change

de_thread() checks if the old leader was the ->child_reaper, this is not
possible any longer.  With the previous patch ->group_leader itself will
change ->child_reaper on exit.

Henceforth find_new_reaper() is the only function (apart from
initialization) which plays with ->child_reaper.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: Serge Hallyn <serue@us.ibm.com>
Acked-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

sysctl: simplify ->strategy

name and nlen parameters passed to ->strategy hook are unused, remove
them.  In general ->strategy hook should know what it's doing, and don't
do something tricky for which, say, pointer to original userspace array
may be needed (name).

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Acked-by: David S. Miller <davem@davemloft.net> [ networking bits ]
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: David Howells <dhowells@redhat.com>
Cc: Matt Mackall <mpm@selenic.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

proc: move sysrq-trigger out of fs/proc/

Move it into sysrq.c, along with the rest of the sysrq implementation.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

block: sanitize invalid partition table entries

We currently follow blindly what the partition table lies about the
disk, and let the kernel create block devices which can not be accessed.
Trying to identify the device leads to kernel logs full of:
  sdb: rw=0, want=73392, limit=28800
  attempt to access beyond end of device

Here is an example of a broken partition table, where sda2 starts
behind the end of the disk, and sdb3 is larger than the entire disk:
  Disk /dev/sdb: 14 MB, 14745600 bytes
  1 heads, 29 sectors/track, 993 cylinders, total 28800 sectors
     Device Boot      Start         End      Blocks   Id  System
  /dev/sdb1              29        7800        3886   83  Linux
  /dev/sdb2           37801       45601        3900+  83  Linux
  /dev/sdb3           15602       73402       28900+  83  Linux
  /dev/sdb4           23403       28796        2697   83  Linux

The kernel creates these completely invalid devices, which can not be
accessed, or may lead to other unpredictable failures:
  grep . /sys/class/block/sdb*/{start,size}
  /sys/class/block/sdb/size:28800
  /sys/class/block/sdb1/start:29
  /sys/class/block/sdb1/size:7772
  /sys/class/block/sdb2/start:37801
  /sys/class/block/sdb2/size:7801
  /sys/class/block/sdb3/start:15602
  /sys/class/block/sdb3/size:57801
  /sys/class/block/sdb4/start:23403
  /sys/class/block/sdb4/size:5394

With this patch, we ignore partitions which start behind the end of the disk,
and limit partitions to the end of the disk if they pretend to be larger:
  grep . /sys/class/block/sdb*/{start,size}
  /sys/class/block/sdb/size:28800
  /sys/class/block/sdb1/start:29
  /sys/class/block/sdb1/size:7772
  /sys/class/block/sdb3/start:15602
  /sys/class/block/sdb3/size:13198
  /sys/class/block/sdb4/start:23403
  /sys/class/block/sdb4/size:5394

These warnings are printed to the kernel log:
  sdb: p2 ignored, start 37801 is behind the end of the disk
  sdb: p3 size 57801 limited to end of disk

Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Cc: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/partitions/acorn.c: remove dead code

I missed this when I did the arm26 removal.

Reported-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

firmware: use dev_printk when possible

Convert printks to use dev_printk().

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Cc: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

char: moxa.c sparse annotation

The only use is to pass this to le16_to_cpu, declare as such
drivers/char/moxa.c:548:11: warning: cast to restricted __le16.

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: Jiri Slaby <jirislaby@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

applicom.c: fix apparently-broken code in do_ac_read()

The code scriblles over a local pointer whereas it appears to be trying
to write to the memory at which that pointer points.

Addresses http://bugzilla.kernel.org/show_bug.cgi?id=11397

Nobody we know can test this change.

Reported-by: Zvonimir Rakamaric <zrakamar@cs.ubc.ca>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Jiri Slaby <jirislaby@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Char: ds1286, eliminate busy waiting

ds1286_get_time(); is not called from atomic context, sleep for 20 ms is
better choice than a (home-made) busy waiting for such a situation.

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

COMPAT_BINFMT_ELF definition tweak

Don't repeat BINFMT_ELF definition, simply multiply COMPAT and BINFMT_ELF.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

binfmt_elf_fdpic: wire up AT_EXECFD, AT_EXECFN, AT_SECURE

These auxvec entries are the only ones left unhandled out of the current
base implementation. This syncs up binfmt_elf_fdpic with linux/auxvec.h
and current binfmt_elf.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

binfmt_elf_fdpic: convert initial stack alignment to arch_align_stack()

binfmt_elf_fdpic seems to have grabbed a hard-coded hack from an ancient
version of binfmt_elf in order to try and fix up initial stack alignment
on multi-threaded x86, which while in addition to being unused, was also
pushed down beyond the first set of operations on the stack pointer,
negating the entire purpose.

These days, we have an architecture independent arch_align_stack(), so we
switch to using that instead. Move the initial alignment up before the
initial stores while we're at it.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

binfmt_elf_fdpic: support auxvec base platform string

Commit 483fad1c3fa1060d7e6710e84a065ad514571739 ("ELF loader support for
auxvec base platform string") introduced AT_BASE_PLATFORM, but only
implemented it for binfmt_elf.

Given that AT_VECTOR_SIZE_BASE is unconditionally enlarged for us, and
it's only optionally added in for the platforms that set
ELF_BASE_PLATFORM, wire it up for binfmt_elf_fdpic, too.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

docbook: update procfs credits

Update Erik Mouw's email address & affiliation in DocBook.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Acked-by: Erik Mouw <mouw@nl.linux.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

quota: remove CVS keywords

Remove CVS keywords that weren't updated for a long time from comments.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/reiserfs: use an IS_ERR test rather than a NULL test

In case of error, the function open_xa_dir returns an ERR pointer, but
never returns a NULL pointer.  So a NULL test that comes after an IS_ERR
test should be deleted.

The semantic match that finds this problem is as follows:
(http://www.emn.fr/x-info/coccinelle/)

// <smpl>
@match_bad_null_test@
expression x, E;
statement S1,S2;
@@
x = open_xa_dir(...)
... when != x = E
(
*  if (x == NULL && ...) S1 else S2
|
*  if (x == NULL || ...) S1 else S2
)
// </smpl>

Signed-off-by: Julien Brunel <brunel@diku.dk>
Signed-off-by: Julia Lawall <julia@diku.dk>
Cc: Jeff Mahoney <jeffm@suse.com>
Cc: Jan Kara <jack@ucw.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

reiserfs/procfs.c: remove CVS keywords

Remove CVS keywords that weren't updated for a long time from comments.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

hfs: fix namelength memory corruption

Fix a stack corruption caused by a corrupted hfs filesystem.  If the
catalog name length is corrupted the memcpy overwrites the catalog btree
structure.  Since the field is limited to HFS_NAMELEN bytes in the
structure and the file format, we throw an error if it is too long.

Cc: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Eric Sesterhenn <snakebyte@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

hfsplus: check read_mapping_page() return value

While testing more corrupted images with hfsplus, i came across
one which triggered the following bug:

[15840.675016] BUG: unable to handle kernel paging request at fffffffb
[15840.675016] IP: [<c0116a4f>] kmap+0x15/0x56
[15840.675016] *pde = 00008067 *pte = 00000000
[15840.675016] Oops: 0000 [#1] PREEMPT DEBUG_PAGEALLOC
[15840.675016] Modules linked in:
[15840.675016]
[15840.675016] Pid: 11575, comm: ln Not tainted (2.6.27-rc4-00123-gd3ee1b4-dirty #29)
[15840.675016] EIP: 0060:[<c0116a4f>] EFLAGS: 00010202 CPU: 0
[15840.675016] EIP is at kmap+0x15/0x56
[15840.675016] EAX: 00000246 EBX: fffffffb ECX: 00000000 EDX: cab919c0
[15840.675016] ESI: 000007dd EDI: cab0bcf4 EBP: cab0bc98 ESP: cab0bc94
[15840.675016]  DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 0068
[15840.675016] Process ln (pid: 11575, ti=cab0b000 task=cab919c0 task.ti=cab0b000)
[15840.675016] Stack: 00000000 cab0bcdc c0231cfb 00000000 cab0bce0 00000800 ca9290c0 fffffffb
[15840.675016]        cab145d0 cab919c0 cab15998 22222222 22222222 22222222 00000001 cab15960
[15840.675016]        000007dd cab0bcf4 cab0bd04 c022cb3a cab0bcf4 cab15a6c ca9290c0 00000000
[15840.675016] Call Trace:
[15840.675016]  [<c0231cfb>] ? hfsplus_block_allocate+0x6f/0x2d3
[15840.675016]  [<c022cb3a>] ? hfsplus_file_extend+0xc4/0x1db
[15840.675016]  [<c022ce41>] ? hfsplus_get_block+0x8c/0x19d
[15840.675016]  [<c06adde4>] ? sub_preempt_count+0x9d/0xab
[15840.675016]  [<c019ece6>] ? __block_prepare_write+0x147/0x311
[15840.675016]  [<c0161934>] ? __grab_cache_page+0x52/0x73
[15840.675016]  [<c019ef4f>] ? block_write_begin+0x79/0xd5
[15840.675016]  [<c022cdb5>] ? hfsplus_get_block+0x0/0x19d
[15840.675016]  [<c019f22a>] ? cont_write_begin+0x27f/0x2af
[15840.675016]  [<c022cdb5>] ? hfsplus_get_block+0x0/0x19d
[15840.675016]  [<c0139ebe>] ? tick_program_event+0x28/0x4c
[15840.675016]  [<c013bd35>] ? trace_hardirqs_off+0xb/0xd
[15840.675016]  [<c022b723>] ? hfsplus_write_begin+0x2d/0x32
[15840.675016]  [<c022cdb5>] ? hfsplus_get_block+0x0/0x19d
[15840.675016]  [<c0161988>] ? pagecache_write_begin+0x33/0x107
[15840.675016]  [<c01879e5>] ? __page_symlink+0x3c/0xae
[15840.675016]  [<c019ad34>] ? __mark_inode_dirty+0x12f/0x137
[15840.675016]  [<c0187a70>] ? page_symlink+0x19/0x1e
[15840.675016]  [<c022e6eb>] ? hfsplus_symlink+0x41/0xa6
[15840.675016]  [<c01886a9>] ? vfs_symlink+0x99/0x101
[15840.675016]  [<c018a2f6>] ? sys_symlinkat+0x6b/0xad
[15840.675016]  [<c018a348>] ? sys_symlink+0x10/0x12
[15840.675016]  [<c01038bd>] ? sysenter_do_call+0x12/0x31
[15840.675016]  =======================
[15840.675016] Code: 00 00 75 10 83 3d 88 2f ec c0 02 75 07 89 d0 e8 12 56 05 00 5d c3 55 ba 06 00 00 00 89 e5 53 89 c3 b8 3d eb 7e c0 e8 16 74 00 00 <8b> 03 c1 e8 1e 69 c0 d8 02 00 00 05 b8 69 8e c0 2b 80 c4 02 00
[15840.675016] EIP: [<c0116a4f>] kmap+0x15/0x56 SS:ESP 0068:cab0bc94
[15840.675016] ---[ end trace 4fea40dad6b70e5f ]---

This happens because the return value of read_mapping_page() is passed on
to kmap unchecked.  The bug is triggered after the first
read_mapping_page() in hfsplus_block_allocate(), this patch fixes all
three usages in this functions but leaves the ones further down in the
file unchanged.

Signed-off-by: Eric Sesterhenn <snakebyte@gmx.de>
Cc: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

hfsplus: fix Buffer overflow with a corrupted image

When an hfsplus image gets corrupted it might happen that the catalog
namelength field gets b0rked.  If we mount such an image the memcpy() in
hfsplus_cat_build_key_uni() writes more than the 255 that fit in the name
field.  Depending on the size of the overwritten data, we either only get
memory corruption or also trigger an oops like this:

[  221.628020] BUG: unable to handle kernel paging request at c82b0000
[  221.629066] IP: [<c022d4b1>] hfsplus_find_cat+0x10d/0x151
[  221.629066] *pde = 0ea29163 *pte = 082b0160
[  221.629066] Oops: 0002 [#1] PREEMPT DEBUG_PAGEALLOC
[  221.629066] Modules linked in:
[  221.629066]
[  221.629066] Pid: 4845, comm: mount Not tainted (2.6.27-rc4-00123-gd3ee1b4-dirty #28)
[  221.629066] EIP: 0060:[<c022d4b1>] EFLAGS: 00010206 CPU: 0
[  221.629066] EIP is at hfsplus_find_cat+0x10d/0x151
[  221.629066] EAX: 00000029 EBX: 00016210 ECX: 000042c2 EDX: 00000002
[  221.629066] ESI: c82d70ca EDI: c82b0000 EBP: c82d1bcc ESP: c82d199c
[  221.629066]  DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 0068
[  221.629066] Process mount (pid: 4845, ti=c82d1000 task=c8224060 task.ti=c82d1000)
[  221.629066] Stack: c080b3c4 c82aa8f8 c82d19c2 00016210 c080b3be c82d1bd4 c82aa8f0 00000300
[  221.629066]        01000000 750008b1 74006e00 74006900 65006c00 c82d6400 c013bd35 c8224060
[  221.629066]        00000036 00000046 c82d19f0 00000082 c8224548 c8224060 00000036 c0d653cc
[  221.629066] Call Trace:
[  221.629066]  [<c013bd35>] ? trace_hardirqs_off+0xb/0xd
[  221.629066]  [<c013bca3>] ? trace_hardirqs_off_caller+0x14/0x9b
[  221.629066]  [<c013bd35>] ? trace_hardirqs_off+0xb/0xd
[  221.629066]  [<c013bca3>] ? trace_hardirqs_off_caller+0x14/0x9b
[  221.629066]  [<c013bd35>] ? trace_hardirqs_off+0xb/0xd
[  221.629066]  [<c0107aa3>] ? native_sched_clock+0x82/0x96
[  221.629066]  [<c01302d2>] ? __kernel_text_address+0x1b/0x27
[  221.629066]  [<c010487a>] ? dump_trace+0xca/0xd6
[  221.629066]  [<c0109e32>] ? save_stack_address+0x0/0x2c
[  221.629066]  [<c0109eaf>] ? save_stack_trace+0x1c/0x3a
[  221.629066]  [<c013b571>] ? save_trace+0x37/0x8d
[  221.629066]  [<c013b62e>] ? add_lock_to_list+0x67/0x8d
[  221.629066]  [<c013ea1c>] ? validate_chain+0x8a4/0x9f4
[  221.629066]  [<c013553d>] ? down+0xc/0x2f
[  221.629066]  [<c013f1f6>] ? __lock_acquire+0x68a/0x6e0
[  221.629066]  [<c013bd35>] ? trace_hardirqs_off+0xb/0xd
[  221.629066]  [<c013bca3>] ? trace_hardirqs_off_caller+0x14/0x9b
[  221.629066]  [<c013bd35>] ? trace_hardirqs_off+0xb/0xd
[  221.629066]  [<c0107aa3>] ? native_sched_clock+0x82/0x96
[  221.629066]  [<c013da5d>] ? mark_held_locks+0x43/0x5a
[  221.629066]  [<c013dc3a>] ? trace_hardirqs_on+0xb/0xd
[  221.629066]  [<c013dbf4>] ? trace_hardirqs_on_caller+0xf4/0x12f
[  221.629066]  [<c06abec8>] ? _spin_unlock_irqrestore+0x42/0x58
[  221.629066]  [<c013555c>] ? down+0x2b/0x2f
[  221.629066]  [<c022aa68>] ? hfsplus_iget+0xa0/0x154
[  221.629066]  [<c022b0b9>] ? hfsplus_fill_super+0x280/0x447
[  221.629066]  [<c0107aa3>] ? native_sched_clock+0x82/0x96
[  221.629066]  [<c013bca3>] ? trace_hardirqs_off_caller+0x14/0x9b
[  221.629066]  [<c013bca3>] ? trace_hardirqs_off_caller+0x14/0x9b
[  221.629066]  [<c013f1f6>] ? __lock_acquire+0x68a/0x6e0
[  221.629066]  [<c041c9e4>] ? string+0x2b/0x74
[  221.629066]  [<c041cd16>] ? vsnprintf+0x2e9/0x512
[  221.629066]  [<c010487a>] ? dump_trace+0xca/0xd6
[  221.629066]  [<c0109eaf>] ? save_stack_trace+0x1c/0x3a
[  221.629066]  [<c0109eaf>] ? save_stack_trace+0x1c/0x3a
[  221.629066]  [<c013b571>] ? save_trace+0x37/0x8d
[  221.629066]  [<c013b62e>] ? add_lock_to_list+0x67/0x8d
[  221.629066]  [<c013ea1c>] ? validate_chain+0x8a4/0x9f4
[  221.629066]  [<c01354d3>] ? up+0xc/0x2f
[  221.629066]  [<c013f1f6>] ? __lock_acquire+0x68a/0x6e0
[  221.629066]  [<c013bd35>] ? trace_hardirqs_off+0xb/0xd
[  221.629066]  [<c013bca3>] ? trace_hardirqs_off_caller+0x14/0x9b
[  221.629066]  [<c013bd35>] ? trace_hardirqs_off+0xb/0xd
[  221.629066]  [<c0107aa3>] ? native_sched_clock+0x82/0x96
[  221.629066]  [<c041cfb7>] ? snprintf+0x1b/0x1d
[  221.629066]  [<c01ba466>] ? disk_name+0x25/0x67
[  221.629066]  [<c0183960>] ? get_sb_bdev+0xcd/0x10b
[  221.629066]  [<c016ad92>] ? kstrdup+0x2a/0x4c
[  221.629066]  [<c022a7b3>] ? hfsplus_get_sb+0x13/0x15
[  221.629066]  [<c022ae39>] ? hfsplus_fill_super+0x0/0x447
[  221.629066]  [<c0183583>] ? vfs_kern_mount+0x3b/0x76
[  221.629066]  [<c0183602>] ? do_kern_mount+0x32/0xba
[  221.629066]  [<c01960d4>] ? do_new_mount+0x46/0x74
[  221.629066]  [<c0196277>] ? do_mount+0x175/0x193
[  221.629066]  [<c013dbf4>] ? trace_hardirqs_on_caller+0xf4/0x12f
[  221.629066]  [<c01663b2>] ? __get_free_pages+0x1e/0x24
[  221.629066]  [<c06ac07b>] ? lock_kernel+0x19/0x8c
[  221.629066]  [<c01962e6>] ? sys_mount+0x51/0x9b
[  221.629066]  [<c01962f9>] ? sys_mount+0x64/0x9b
[  221.629066]  [<c01038bd>] ? sysenter_do_call+0x12/0x31
[  221.629066]  =======================
[  221.629066] Code: 89 c2 c1 e2 08 c1 e8 08 09 c2 8b 85 e8 fd ff ff 66 89 50 06 89 c7 53 83 c7 08 56 57 68 c4 b3 80 c0 e8 8c 5c ef ff 89 d9 c1 e9 02 <f3> a5 89 d9 83 e1 03 74 02 f3 a4 83 c3 06 8b 95 e8 fd ff ff 0f
[  221.629066] EIP: [<c022d4b1>] hfsplus_find_cat+0x10d/0x151 SS:ESP 0068:c82d199c
[  221.629066] ---[ end trace e417a1d67f0d0066 ]---

Since hfsplus_cat_build_key_uni() returns void and only has one callsite,
the check is performed at the callsite.

Signed-off-by: Eric Sesterhenn <snakebyte@gmx.de>
Reviewed-by: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

hfsplus: quieten down mounting hfsplus journaled fs read only

Check whether the file system was to be mounted read only anyway before
warning about changing the mount to read only.

Signed-off-by: Mike Crowe <mac@mcrowe.com>
Cc: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

befs: annotate fs32 on tests for superblock endianness

Does compile-time byteswapping rather than runtime.

Noticed by sparse:
fs/befs/super.c:29:6: warning: cast to restricted __le32
fs/befs/super.c:29:6: warning: cast from restricted fs32
fs/befs/super.c:31:11: warning: cast to restricted __be32
fs/befs/super.c:31:11: warning: cast from restricted fs32
fs/befs/super.c:31:11: warning: cast to restricted __be32
fs/befs/super.c:31:11: warning: cast from restricted fs32
fs/befs/super.c:31:11: warning: cast to restricted __be32
fs/befs/super.c:31:11: warning: cast from restricted fs32
fs/befs/super.c:31:11: warning: cast to restricted __be32
fs/befs/super.c:31:11: warning: cast from restricted fs32
fs/befs/super.c:31:11: warning: cast to restricted __be32
fs/befs/super.c:31:11: warning: cast from restricted fs32
fs/befs/super.c:31:11: warning: cast to restricted __be32
fs/befs/super.c:31:11: warning: cast from restricted fs32
fs/befs/linuxvfs.c:811:7: warning: cast to restricted __le32
fs/befs/linuxvfs.c:811:7: warning: cast from restricted fs32
fs/befs/linuxvfs.c:812:7: warning: cast to restricted __be32
fs/befs/linuxvfs.c:812:7: warning: cast from restricted fs32
fs/befs/linuxvfs.c:812:7: warning: cast to restricted __be32
fs/befs/linuxvfs.c:812:7: warning: cast from restricted fs32
fs/befs/linuxvfs.c:812:7: warning: cast to restricted __be32
fs/befs/linuxvfs.c:812:7: warning: cast from restricted fs32
fs/befs/linuxvfs.c:812:7: warning: cast to restricted __be32
fs/befs/linuxvfs.c:812:7: warning: cast from restricted fs32
fs/befs/linuxvfs.c:812:7: warning: cast to restricted __be32
fs/befs/linuxvfs.c:812:7: warning: cast from restricted fs32
fs/befs/linuxvfs.c:812:7: warning: cast to restricted __be32
fs/befs/linuxvfs.c:812:7: warning: cast from restricted fs32

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: "Sergey S. Kostyliov" <rathamahata@php4.ru>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ext2: avoid printk floods in the face of directory corruption

A very large directory with many read failures (either due to storage
problems, or due to invalid size & blocks from corruption) will generate a
printk storm as the filesystem continues to try to read all the blocks.
This flood of messages can tie up the box until it is complete - which may
be a very long time, especially for very large corrupted values.

This is fixed by only reporting the corruption once each time we try to
read the directory.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: Eugene Teo <eugeneteo@kernel.sg>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ext2: fix ext2 block reservation early ENOSPC issue

We could run into ENOSPC error on ext2, even when there is free blocks on
the filesystem.

The problem is triggered in the case the goal block group has 0 free
blocks , and the rest block groups are skipped due to the check of
"free_blocks < windowsz/2".  Current code could fall back to non
reservation allocation to prevent early ENOSPC after examing all the block
groups with reservation on , but this code was bypassed if the reservation
window is turned off already, which is true in this case.

This patch fixed two issues:
1) We don't need to turn off block reservation if the goal block group has
0 free blocks left and continue search for the rest of block groups.

Current code the intention is to turn off the block reservation if the
goal allocation group has a few (some) free blocks left (not enough for
make the desired reservation window),to try to allocation in the goal
block group, to get better locality.  But if the goal blocks have 0 free
blocks, it should leave the block reservation on, and continues search for
the next block groups,rather than turn off block reservation completely.

2) we don't need to check the window size if the block reservation is off.

The problem was originally found and fixed in ext4.

Signed-off-by: Mingming Cao <cmm@us.ibm.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

telephony: remove CVS keywords

Remove CVS keywords that weren't updated for a long time from comments.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pnp: make the resource type an unsigned long

PnP encodes the resource type directly as its struct resource->flags value
which is an unsigned long.  Make it so...

Signed-off-by: Rene Herman <rene.herman@gmail.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Acked-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pnp: remove printk() with outdated version

There's no point in printing some ancient version number forever.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: Rene Herman <rene.herman@gmail.com>
Acked-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Acked-by: Adam M Belay <abelay@MIT.EDU>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fbcon_set_all_vcs: fix kernel crash when switching the rotated consoles

echo 3 >> /sys/class/graphics/fbcon/rotate_all, then switch to another
console. Result:

BUG: unable to handle kernel paging request at ffffc20005d00000
IP: [bitfill_aligned+149/265] bitfill_aligned+0x95/0x109
PGD 7e228067 PUD 7e229067 PMD 7bc1f067 PTE 0
Oops: 0002 [1] SMP
CPU 1
Modules linked in: [...a lot...]
Pid: 10, comm: events/1 Not tainted 2.6.26.5-45.fc9.x86_64 #1
RIP: 0010:[bitfill_aligned+149/265]  [bitfill_aligned+149/265] bitfill_aligned+0x95/0x109
RSP: 0018:ffff81007d811bc8  EFLAGS: 00010216
RAX: ffffc20005d00000 RBX: 0000000000000000 RCX: 0000000000000400
RDX: 0000000000000000 RSI: ffffc20005d00000 RDI: ffffffffffffffff
RBP: ffff81007d811be0 R08: 0000000000000400 R09: 0000000000000040
R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000010000
R13: ffffffff811632f0 R14: 0000000000000006 R15: ffff81007cb85400
FS:  0000000000000000(0000) GS:ffff81007e004780(0000) knlGS:0000000000000000
CS:  0010 DS: 0018 ES: 0018 CR0: 000000008005003b
CR2: ffffc20005d00000 CR3: 0000000000201000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process events/1 (pid: 10, threadinfo ffff81007d810000, task ffff81007d808000)
Stack:  ffff81007c9d75a0 0000000000000000 0000000000000000 ffff81007d811c80
 ffffffff81163a61 ffff810000000000 ffffffff8115f9c8 0000001000000000
 0000000100aaaaaa 000000007cd0d4a0 fffffd8a00000800 0001000000000000
Call Trace:
 [cfb_fillrect+523/798] cfb_fillrect+0x20b/0x31e
 [soft_cursor+416/436] ? soft_cursor+0x1a0/0x1b4
 [ccw_clear_margins+205/263] ccw_clear_margins+0xcd/0x107
 [fbcon_clear_margins+59/61] fbcon_clear_margins+0x3b/0x3d
 [fbcon_switch+1291/1466] fbcon_switch+0x50b/0x5ba
 [redraw_screen+261/481] redraw_screen+0x105/0x1e1
 [ccw_cursor+0/1869] ? ccw_cursor+0x0/0x74d
 [complete_change_console+48/190] complete_change_console+0x30/0xbe
 [change_console+115/120] change_console+0x73/0x78
 [console_callback+0/292] ? console_callback+0x0/0x124
 [console_callback+97/292] console_callback+0x61/0x124
 [schedule_delayed_work+25/30] ? schedule_delayed_work+0x19/0x1e
 [run_workqueue+139/282] run_workqueue+0x8b/0x11a
 [worker_thread+221/238] worker_thread+0xdd/0xee
 [autoremove_wake_function+0/56] ? autoremove_wake_function+0x0/0x38
 [worker_thread+0/238] ? worker_thread+0x0/0xee
 [kthread+73/118] kthread+0x49/0x76
 [child_rip+10/18] child_rip+0xa/0x12
 [kthread+0/118] ? kthread+0x0/0x76
 [child_rip+0/18] ? child_rip+0x0/0x12

Because fbcon_set_all_vcs()->FBCON_SWAP() uses display->rotate == 0 instead
of fbcon_ops->rotate, and vc_resize() has no effect because it is called with
new_cols/rows == ->vc_cols/rows.

Tested on 2.6.26.5-45.fc9.x86_64, but
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git seems to
have the same problem.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Cc: <stable@kernel.org> [2.6.27.x, 2.6.26.x, maybe 2.6.25.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fbdev: add new TMIO framebuffer driver

Add driver for TMIO framebuffer cells as found e.g. in Toshiba TC6393XB
chips.

Signed-off-by: Dmitry Baryshkov <dbaryshkov@gmail.com>
Cc: Ian Molton <spyro@f2s.com>
Acked-by: Samuel Ortiz <sameo@openedhand.com>
Acked-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fbdev: Epson S1D1 framebuffer maintainer

I'm stepping up as maintainer for the Epson S1D13XXXFB driver since I have
2 platforms currently using it.

Signed-off-by: Kristoffer Ericson <Kristoffer.Ericson@gmail.com>
Acked-by: Thibaut Varene <varenet@parisc-linux.org>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

framebuffer: fix carminefb section mismatch

Make alloc_carmine_fb() __devinit.

WARNING: drivers/video/carminefb.o(.text+0x81b): Section mismatch in reference from the function alloc_carmine_fb() to the variable .devinit.data:carminefb_fix

The function alloc_carmine_fb() references the variable __devinitdata
carminefb_fix.  This is often because alloc_carmine_fb lacks a
__devinitdata annotation or the annotation of carminefb_fix is wrong.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Sebastian Siewior <bigeasy@linutronix.de>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

matroxfb: support G200eV chip

Support the Matrox G200eV chip, based on timings that I found in the X.org
matrox driver.

Signed-off-by: Darrick J. Wong <djwong@us.ibm.com>
Acked-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: Petr Vandrovec <vandrove@vc.cvut.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fbdev: allow more chip revisions in Epson s1d13... video driver

The Epson s1d13xxx hardware is common in many handhelds, but our driver is
currently locked to a single chip revision.  This patch adds an array of
known to work revisions (which can be extended).

[akpm@linux-foundation.org: cleanups]
Signed-off-by: Kristoffer Ericson <Kristoffer.Ericson@gmail.com>
Acked-by: Thibaut Varène <varenet@parisc-linux.org>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fbdev: ignore VESA modes if framebuffer does not support them

Currently, it is possible to set a graphics VESA mode at boot time via the
vga= parameter even when no framebuffer driver supporting this is
configured.  This could lead to the system booting with a black screen,
without a usable console.

Fix this problem by only allowing to set graphics modes at boot time if a
supporting framebuffer driver is configured.

Signed-off-by: Michal Januszewski <spock@gentoo.org>
Acked-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

vgacon: vgacon_scrolldelta simplification

There's no point in checking diff == c->vc_rows, because it can be true
only when count == 0, but we already checked that.  Additionally move
variables used only in one block to this block.

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Cc: Antonino Daplas <adaplas@gmail.com>
Acked-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

vgacon: optimize scrolling

Join multiple scr_memcpyw into 1-3 calls (usually 2).  (benchmarked
average speedup: 1%)

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Cc: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fbmon: fix EDID parser to detect interlace modes

The current EDID parser in the linux kernel ignores interlace modes.  The
patch looks for the edid interlace flag and adjusts the vertical
resolution if it is found.

Signed-off-by: Jon Dufresne <jon.dufresne@gmail.com>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

MAINTAINERS: add entry for uvesafb

Add a maintainer entry for the uvesafb driver.

Signed-off-by: Michal Januszewski <spock@gentoo.org>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uvesafb: don't treat mode info retrieval failures as errors

Some BIOSes return error codes when queried for information about
modes from their own modelist.  uvesafb treats this as an error
case and bails out.

Change this behavior so that broken modes do not prevent the driver
from working.  Only the failure to retrieve information about any
usable video mode is considered to be an error case.

Signed-off-by: Michal Januszewski <spock@gentoo.org>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uvesafb: document mode to mode_option parameter change

Document the change from the old "mode" parameter to the "mode_option"
parameter.

Signed-off-by: Mike Pagano <mpagano@gentoo.org>
Cc: Krzysztof Halasa <khc@pm.waw.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

radeonfb: accelerate imageblit and other improvements

Implement support for HW color expansion of 1bpp images, along with some
improvements to the FIFO handling and other accel operations.

The offset fixup code is now unnecessary as the fbcon core will call our
set_par upon switch back from KD_GRAPHICS before anything else happens.  I
removed it as it would slow down accel operations.

The fifo wait has been improved to avoid hitting the HW register as often,
and the various accel ops are now performing better caching of register
values.

Overall, this improve accel performances.  The imageblit acceleration does
result in a small overall regression in performances on some machines (on
the order of 5% on some x86), probably becaus the SW path provides a
better bus utilisation, but I decided to ingnore that as the performances
is still very good, and on the other hand, some machines such as some
sparc64 get a 3 fold performance improvement.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: David S. Miller <davem@davemloft.net>
Cc: Krzysztof Halasa <khc@pm.waw.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

radeonfb: misc cleanup of engine and dst cache handling

Fix a couple of incomplete tests of the chip families in the engine
init/reset code and proper initialization of the destination cache mode.
The result should better match what the latest X radeon driver does.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: David S. Miller <davem@davemloft.net>
Cc: Krzysztof Halasa <khc@pm.waw.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

efifb/imacfb consolidation + hardware support

Remove imacfb entirely, merging its DMI table into the (otherwise very
similar) efifb driver.  This also adds hardware support for many of the
newer Intel Apple hardware.  This has been fairly well tested; we've been
shipping it in Fedora for some time.

Signed-off-by: Peter Jones <pjones@redhat.com>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Cc: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: Jaya Kumar <jayakumar.lkml@gmail.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

atmel_lcdfb: disallow setting larger resolution than the framebuffer memory can handle

Signed-off-by: Stanislaw Gruszka <stf_xl@wp.pl>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: Haavard Skinnemoen <hskinnemoen@atmel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cirrusfb: do not change MCLK for Alpine chips

A memory clock value (MCLK) is changed to a minimum required by a current
mode bandwidth.  This usually lowers the MCLK to its minimum (50 MHz) thus
decreasing the card performance.  Just leave the MCLK value set by card
BIOS.

The CL-GD5446 Technical Reference Manual point 9.9.1.3 states that if a
pixclock value is close (~1%) to the MCLK or MCLK/2 this may result in a
jitter on the screen.  A countermeasure is to use the MCLK as pixclock
source instead of a VCLK.  The patch implements this as well.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cirrusfb: fix 16bpp modes

The 16bpp mode did not work on the Cirrus cards as the visual type was set
to DIRECTCOLOR instead of TRUECOLOR.  The Alpine family used one incorrect
register setting so this 16bpp modes generated wrong horizontal frequency.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cirrusfb: add noaccel module parameter

The noaccel parameter is already handled if the driver is not built as
module.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cirrusfb: drop clock fields from cirrusfb_regs structure

Move call to pixclock calculation into the cirrusfb_set_par_foo().  It
makes copy of clock registers redundant.

Simplify clock calculations further.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cirrusfb: eliminate CRT registers from global structure

Move calculations of CRT register values into the cirrusfb_set_par_foo()
where the values are used.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cirrusfb: add __devinit attribute to probing functions

Add __devinit attribute to probing functions.  This fixed section mismatch
warning from my previous patch.

Kill one redundant forward declaration.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cirrusfb: use modedb and add mode_option parameter

Use modedb for initial mode instead of table of few predefined modes.

Add mode_option module parameter as a step toward unification of frame
buffers' parameters.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cirrusfb: drop device pointers from cirrusfb_info

The device pointer can be easily obtained from fb_info->device if needed.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cirrusfb: remove 24 bpp mode

The 24 bpp mode is not implemented.  Disallow it in the
cirrusfb_check_var() and remove it from checks.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cirrusfb: simplify clock calculation

Simplify clock calculation.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cirrusfb: remove information about memory size during mode change

Remove information about memory size displayed twice each time a display
mode change.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

intelfb: support 945GME (as used in ASUS Eee 901)

Add support for Intel's 945GME graphics chip to the intelfb driver.  I
have assumed that the 945GME is identical to the already-supported 945GM
apart from its PCI IDs; this is based on a quick look at the X driver for
these chips which seems to treat them identically.

The 945GME is used in the ASUS Eee 901, and I coded this in the hope that
I'd be able to use it to get a console at the native 1024x600 resolution
which is not known to the BIOS.  I realised too late that the intelfb
driver does not support mode changing on laptops, so it won't be any
use for me.

Signed-off-by: Phil Endecott <spam_from_intelfb@chezphil.org>
Acked-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>