[PATCH] vm: convert users of remap_page_range() under sound/ to use remap_pfn_range()

This patch converts all users of remap_page_range() under sound/ to use
remap_pfn_range(), with the exception of maestro3 changelogs, which are likely
expected to be preserved intact apart from additions (as most changelogs are),
regardless of API changes.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] vm: convert users of remap_page_range() under include/asm-*/ to use remap_pfn_range()

This patch converts uses of remap_page_range() via io_remap_page_range() in
include/asm-*/ to use remap_pfn_range().  io_remap_page_range() has a similar
physical address overflow issue that needs to be addressed later.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] vm: convert users of remap_page_range() under drivers/ and net/ to use remap_pfn_range()

This patch converts all callers of remap_page_range() under arch/ and net/ to
use remap_pfn_range() instead.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] vm: convert references to remap_page_range() under arch/ and Documentation/ to remap_pfn_range()

This patch converts all callers of remap_page_range() under arch/ and all
references in Documentation/ to use remap_pfn_range().

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] vm: introduce remap_pfn_range() to replace remap_page_range()

This patch introduces remap_pfn_range(), destined to replace
remap_page_range(), to which all callers of remap_page_range() are converted
in the sequel.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: move code to truncate on open to separate function

for OPEN with O_TRUNC, if the truncate fails, the open fails.  for
nfs4_open_upgrade, this means undo the get_write_access.  for new OPENs, this
means release the newly created stateid.

Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: fix putrootfh return

Thanks to Al Viro for noticing that putrootfh could return either a linux or
an nfs error.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd: clean up nfsd4_process_open2

Comments, dprintk cleanup

Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: move some nfsd4_process_open2 code to nfs4_new_open

Move most of the code in the new_open case of nfsd4_process_open2 to
nfs4_new_open.

Signed-off-by: Andy Adamson <andros@umich.edu>
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: move open_upgrade code into a separate function

Move most of the code for the open_upgrade case into a separate
nfs4_upgrade_open.

Signed-off-by: Andy Adamson <andros@umich.edu>
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: reorganize "if" in nfsd4_process_open2 to make test clearer

Reorganize "if" in nfsd4_process_open2 to make test clearer; there was no real
reason to test for the negative here.

Signed-off-by: Andy Adamson <andros@umich.edu>
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: move seqid decrement on reclaim to separate function

Move seqid decrement on reclaim to separate function

Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd: set OPEN_RESULT_LOCKTYPE_POSIX in open()

We should set the OPEN_RESULT_LOCKTYPE_POSIX flag in the return to open, to
indicate that we support locking.

Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: move open owner checks from nfsd4_process_open2 into new function

nfsd4_process_open2 has become a bit long and contorted.  The following
patches break nfsd4_process_open2 into smaller functions and add comments to
describe logic flow, in preparation for delegation state.

We begin by pulling out the code that searches for conflicting open owners
into a separate function.

Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd: remove incorrect stateid modification in nfsv4 open upgrade

When we upgrade an open, we keep the same stateid and bump only the seqid.

Signed-off-by: Andy Adamson <andros@umich.edu>
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: fix race in xdr encoding of lock_denied response.

We take a reference on the stateowner, and copy the clientid, instead of just
hoping no one destroys the stateowner before we reference it in
nfsd4_encode_lock_denied.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: revert awkward extension of state lock over xdr for replay encoding

With the addition of a reference count, we no longer need the code that
conditionally extended the nfs4 state lock over the xdr encode stage.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: take a reference to preserve stateowner through xdr replay code

Take a reference to preserve the stateowner through the xdr replay code, and
simplify nfsd4_proc_compound a little.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: reference count stateowners

Reference-counting the nfsd4 stateowner structs will let us fix a race and
simplify some of the xdr code a bit, and may also help us make the nfsd4
locking a little more fine-grained in the future.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd: make sure getxattr inode op is non-NULL before calling it

Make sure getxattr inode op is non-NULL before calling it.

Also, security hook should probably be called before calling the getxattr op
the first time.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: don't take i_sem around call to ->getxattr

The ->getxattr op doesn't take the i_sem (see
Documentation/filesystems/Locking)

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd: separate a little of logic from fh_verify into new function

Separate out the little bit of logic in fh_verify that checks the i_mode into
a separate function.  This is nicely encapsulated, fh_verify() is overly long
anyway, and this change helps make a subsequent nfs4 named attribute change
clearer.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nfsd4: nfsd oopsed when encountering a conflict with a local lock

Bug Fix: Non NFSD conflicting byte-range locks were causing an Oops.

Encode a zero length owner and zero clientid for non NFSD conflicting locks in
the lock_denied response.

Signed-off-by: Andy Adamson <andros@umich.edu>
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] select-cpio_list-or-source-directory-for-initramfs-image fix

Kconfig barfs if it doesn't have leading spaces in the help.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Select cpio_list or source directory for initramfs image

Attached is a patch that adds CONFIG_INITRAMFS_SOURCE, enabling you to
either specify a file as cpio_list, or a directory to generate a list from.

It depreciate the INITRAMFS_LIST environment variable introduced not long
ago.

There are some issues (suggestions/patches welcome) that I am not
sure about:
1) I put the menu entry under block devices, but I am not sure if
   this is the correct location?
2) There might be a better (or more correct) way to do this with
   kbuild?
3) Variable names and especially help text needs some love.
4) I am not sure if I am duplicating work in progress?

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] gen_init_cpio uses external file list

This patch makes gen_init_cpio generate the initramfs_data.cpio from a file
which contains a list of entries: file, dir, nod.  I swapped the order of
filename/location for the file arguments so that it would be more uniform
with the dir and node tyes.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] module_parm_array fixups

Fix up various things which Rusty's patch broke or missed.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] V4L: follow changes in saa7146

- [V4L] mxb, dpc7146, hexium_orion, hexium_gemini: follow latest changes in
  saa7146 driver

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: frontend updates

- [DVB] all: replace dvb_unregister_frontend_new() with
  dvb_unregister_frontend()

- [DVB] sp887x: fix firmware download, patch by Jose Alberto Reguero

- [DVB] tda1004x: add firmware loading via firmware_class()

- [DVB] dvb_frontend: without hierachical coding, code_rate_LP is
  irrelevant, so we tolerate the otherwise invalid FEC_NONE setting

- [DVB] ves1x93: fixed dropouts on older DVB cards, fix tuning issues
  (Andreas Share / Gregoire Favre),

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: misc driver updates

- [DVB] av7110: convert MODULE_PARM() to module_param(), replace home-brewn
  waiting stuff in osd code with wait_event_interruptible_timeout()

- [DVB] av7110: put a semaphore around osd calls to make sure they're
  properly serialized, timeout variable in arm_thread() must be int, not
  unsigned long

- [DVB] av7110: add additional OSD window types (patch by Jeremy Jones), new
  ioctl OSD_GET_CAPABILITY/OSD_CAP_MEMSIZE; returns size of OSD memory

- [DVB] av7110: put audio/video initialization into separate function
  init_av7110_av(); call this function after system initialization and after
  arm crash to restore the previous state; thanks to Soeren Sonnenburg
  <bugreports@nn7.de> for this patch.

- [DVB] av7110, budget, ttusb-budget: remove dvb i2c remains, support kernel
  i2c

- [DVB] av7110, budget: use msleep() instead of my_wait(), thanks to Kernel
  Janitors/Nishanth Aravamudan <nacc@us.ibm.com>

- [DVB] av7110, budget: fix videodev has no release callback

- [DVB] av7110: more sparse annotiations

- [DVB] budget: add support for TerraTec Cinergy 1200 DVB-S

- [DVB] budget: fix race condition in irq handler

- [DVB] skystar2, av7110, ttusb-budget, budget: make i2c
  client_(un)register() functions static

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: new driver for mobile USB Budget DVB-T devices

- [DVB] new driver for mobile USB Budget DVB-T devices, thanks to Patrick
  Boettcher

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: add frontend #2

- [DVB] add new driver for mobile DVB-T demodulator DiBcom 3000-MB
- [DVB] add new drivers to Makefile

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: add frontend

- [DVB] add new driver for Zarlink DVB-T MT352 frontend
- [DVB] add new driver for Conexant 22702 DVB OFDM frontend

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: frontend conversion #4

- [DVB] nxt6000, sp887x: convert from dvb-i2c to kernel-i2c, MODULE_PARM()
  to module_param(), dvb_delay() to mdelay()

- [DVB] sp887x: move from home-brewn firmware loading to firmware_class

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: frontend conversion #3

- [DVB] dvb_dummy_fe, grundig_29504-401, grundig_29504-491, mt312: convert
  from dvb-i2c to kernel-i2c, MODULE_PARM() to module_param(), dvb_delay() to
  mdelay()

- [DVB] update frontend Kconfig

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: frontend conversion #2

- [DVB] alps_tdlb7, alps_tdmb7, at76c651, cx24110, dst: convert from dvb-i2c
  to kernel-i2c, MODULE_PARM() to module_param(), dvb_delay() to mdelay()

- [DVB] alps_tdlb7: move from home-brewn firmware loading to firmware_class

- [DVB] dst: use sysfs attributes for type and flags for per-card parameters

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: frontend conversion

- [DVB] stv0299, tda1004x, ves1820, ves1x93: convert from dvb-i2c to
  kernel-i2c, MODULE_PARM() to module_param(), dvb_delay() to mdelay()

- [DVB] tda1004x: move from home-brewn firmware loading to firmware_class

- [DVB] stv0299: support Cinergy1200, patch by Uli Luckas

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: core update

- [DVB] remove non-linux compatibility stuff from dvb_functions.  rest in
  peace.

- [DVB] remove home-brewn dvb-i2c stuff. rest in peace.

- [DVB] convert MODULE_PARM() to module_param()

- [DVB] convert dvb_delay() to mdelay()

- [DVB] convert C++ comments to C comments

- [DVB] dvb_ca_en50221: fix for matrix CAMs from Sjoerd Simons, use c99
  initializers, Fix for aston CAM read timeout problems, Moved CAM CTRL IF
  reset to a better place, better debugging with multiple cards (Sjoerd
  Simons)

- [DVB] dvb-frontend: patch by Wolfgang Fritz: suppress spurious events
  during tuning, Do not allow write (and related) ioctls when frontend is
  opened RDONLY, Properly lock the frontend module on open/close, patch by
  Christopher Pascoe: remove bogus up(fe->sem) on fe thread exit, patch by
  Christopher Pascoe: remove bogus up(fe->sem) on fe thread exit

- [DVB] dvb-demux: using spin_lock instead of spin_lock_irq caused a race
  condition between irq/tasklet and user space task

- [DVB] dvb-core: add sysfs/udev support using "class_simple", prevent Oops
  when PES filter is set with invalid pes_type, protect feed_list with
  spin_locks

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: skystar2 dvb bt8xx update

- [DVB] convert drivers from dvb-i2c to kernel-i2c
- [DVB] convert MODULE_PARM() to module_param()
- [DVB] convert dvb_delay() to mdelay()
- [DVB] dvb-bt8xx: convert home-brewn bttv i2c access to a real bttv sub-driver

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: documentation update

- [DVB] add udev.txt which describes how to use dvb and udev/sysfs
- [DVB] add Visionplus VisionDTV USB-Ter DVB-T adapter documentation
- [DVB] update TT USB DEC documentation
- [DVB] update various Kconfig entries
- [DVB] remove obsolete firmware documentation

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB: update saa7146

- [DVB] fix videodev has no release callback

- [DVB] use PAGE_SIZE for pagetables, not home-brewn SAA7146_PGTABLE_SIZE

- [DVB] use cpu_to_le32() at various places for endianess independency

- [DVB] turn some error checks into BUG()s

- [DVB] make saa7146_i2c_adapter_prepare() support an adapter class

- [DVB] add support for V4L2_PIX_FMT_RGB32 pixelformat

- [DVB] replace generic saa7146 i2c name by card specific name, suggested by
  Uli Luckas <luckas@musoft.de>

Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] v4l: msp3400 cleanup

This is a minor cleanup for the msp3400.c module: use the new msleep()
function for delays.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] DVB/V4L dependency fix

The problem is that dvb is way to much work-in-progress right now, it's
almost impossible to build something which works on both mainline kernel +
latest dvb cvs.  Hope at least the interfaces are settled soon, for now I
gave up on getting it work in mainline in short-term.

The patch below compiles the DVB card configuration code only when
VIDEO_CX88_DVB is enabled (which is commented right now in Kconfig due to
being broken).

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] v4l: cx88 driver update

This is a major update of the cx88 driver.  Main new feature is support
for the MPEG PCI function of the cx2388x chips.  The changes in detail:

  * A bunch of code restructions, to allow multiple modules (one per PCI
    function) work on top of the core module.
  * Add a new module with common code for the mpeg PCI function.
  * Add a new module to support mpeg encoder cards (connexant
    "blackbird" reference design).  That one is not much tested yet and
    may have bugs.  They can be easily workarounded by not loading the
    module through, it isn't needed to capture uncompressed video.
  * Add a new module to support DVB cards.  That one is better tested
    than the blackbird one, but it needs additional cutting-edge stuff
    from the dvb project, thats why it is disabled in Kconfig for now.
  * Several cleanups along the way: use kthread, use msleep(), drop some
    2.4.x compatibility code, move insmod options to 2.6 style, ...
  * adapt code to the video-buf changes.
  * suspend fixes.
  * as usual some new tv cards.

Signed-off-by: Gerd Knorr <kraxel@bytesex.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] v4l: bttv driver update

This update for the bttv driver.  Changes:
 * adapt the driver to video-buf changes.
 * It also added sanity checks for the bt878 risc code buffer sizes.
 * adds support for new tv cards.
 * cleanup i2c driver autoload.
 * misc cleanups (msleep, ...).
 * fix IRQ bug when stopping vbi capture.
 * drop check for cx2388x (bt878 successor) and the printk saying bttv
   doesn't support these.
 * set i2c class correctly for dvb cards.

Signed-off-by: Gerd Knorr <kraxel@bytesex.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] v4l: adapt saa7146 to video-buf changes.

This patch adapts the saa7146 driver to the video-buf changes.

Signed-off-by: Gerd Knorr <kraxel@bytesex.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] v4l: avoid using struct file ptrs in video-buf

This patch makes the video-buf helper module pass through a void pointer
instead of a struct file pointer, that makes the code also usable when no
file pointer is available.  This is needed for when using the video-buf
infrastructure to manage DMA buffers for DVB cards.

The file pointer was used by the videobuf_queue_ops callbacks to get the
drivers private data via file->private_data, now a pointer to the drivers's
private data can be passed directly.

Adaptions of the drivers follow with separate patches.

Signed-off-by: Gerd Knorr <kraxel@bytesex.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] v4l: tuner update

This is a update for the analog tv tuner module, it adds support for two
new tuner chips.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] lighten mmlist_lock

Let's lighten the global spinlock mmlist_lock.

What's it for?
1. Its original role is to guard mmlist.
2. It later got a second role, to prevent get_task_mm from raising
   mm_users from the dead, just after it went down to 0.

Firstly consider the second: __exit_mm sets tsk->mm NULL while holding
task_lock before calling mmput; so mmlist_lock only guards against the
exceptional case, of get_task_mm on a kernel workthread which did AIO's
use_mm (which transiently sets its tsk->mm without raising mm_users) on an
mm now exiting.

Well, I don't think get_task_mm should succeed at all on use_mm tasks.
It's mainly used by /proc/pid and ptrace, seems at best confusing for those
to present the kernel thread as having a user mm, which it won't have a
moment later.  Define PF_BORROWED_MM, set in use_mm, clear in unuse_mm
(though we could just leave it), get_task_mm give NULL if set.

Secondly consider the first: and what's mmlist for?
1. Its original role was for swap_out to scan: rmap ended that in 2.5.27.
2. In 2.4.10 it got a second role, for try_to_unuse to scan for swapoff.

So, make mmlist a list of mms which maybe have pages on swap: add mm to
mmlist when first swap entry is assigned in try_to_unmap_one (pageout), or
in copy_page_range (fork); and mmput remove it from mmlist as before,
except usually list_empty and there's no need to lock.  drain_mmlist added
to swapoff, to empty out the mmlist if no swap is then in use.

mmput leave mm on mmlist until after its exit_mmap, so try_to_unmap_one can
still add mm to mmlist without worrying about the mm_users 0 case; but
try_to_unuse must avoid the mm_users 0 case (when an mm might be removed
from mmlist, and freed, while it's down in unuse_process): use
atomic_inc_return now all architectures support that.

Some of the detailed comments in try_to_unuse have grown out of date:
updated and trimmed some, but leave SWAP_MAP_MAX for another occasion.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] FAT: remove validity check of FAT first entry

The low byte of FAT's first entry must have same value with media-field.

So, I added that validity check for some case. In fact, in some case
this check detected non-FAT filesystem.  But in real world,
unfortunately too many devices is writing a wrong value.

This patch removes that validity check.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] FAT: removal of C[FT]_LE_[WL] macro

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] FAT: check free_clusters value

This patch make sure it's <= sbi->clusters.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] FAT: merge fix

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] FAT: remove debug_pr()

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] FAT: Fix the race bitween fat_free() and fat_get_cluster()

This patch fixes the race condition

  |      fat_free                |    fat_get_cluster
  +------------------------------+-----------------------
                                     fat_cache_lookup()
                                     (get the copy of cache)
   fat_cache_inval_inode()
(invalidate caches on inode)
                                     fat_cache_add()
                                     (update/add the getted cache)

The above race has possible that invalidated cache is added.

This patch fixes the race condition by adding the cache-id to copy of cache.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] FAT: the inode hash from per module to per sb

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] FAT: cache lock from per sb to per inode

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] FAT: rewrite the cache for file allocation table lookup

This rewrites the cache stuff for file allocation table (FAT).

This cache stocks the more pieces of FAT-chain by counting the number
of contiguous data blocks. And if cache hit, since a block number can
calculate without looking FAT-chain up, fat_get_block() become more
fast.

But if data blocks was fragmenting, unfortunately this cache is unuseful.

read from block device

[1st]# time dd if=/dev/hda6 of=/dev/null bs=1M count=2048
2147483648 bytes transferred in 229.524189 seconds (9356241 bytes/sec)
real    3m49.557s, user    0m0.026s, sys     1m20.414s
[2nd]# time dd if=/dev/hda6 of=/dev/null bs=1M count=2048
2147483648 bytes transferred in 229.539358 seconds (9355623 bytes/sec)
real    3m49.647s, user    0m0.036s, sys     1m20.144s

read from full contiguous file with this patch

[1st]# time dd if=data of=/dev/null bs=1M count=2048
2147483648 bytes transferred in 337.959477 seconds (6354264 bytes/sec)
real    5m37.970s, user    0m0.031s, sys     1m21.915s
[2nd]# time dd if=data of=/dev/null bs=1M count=2048
2147483648 bytes transferred in 225.401699 seconds (9527362 bytes/sec)
real    3m45.476s, user    0m0.027s, sys     1m19.286s

read from full fragmented file with this patch

[1st]# time dd if=data of=/dev/null bs=1M count=2048
2147483647 bytes transferred in 1146.529081 seconds (1873030 bytes/sec)
real    19m6.538s, user    0m0.020s, sys     1m32.774s
[2nd]# time dd if=data of=/dev/null bs=1M count=2048
2147483647 bytes transferred in 1045.084822 seconds (2054841 bytes/sec)
real    17m25.152s, user    0m0.022s, sys     1m34.801s

read from full contiguous file without this patch

[1st]# time dd if=data of=/dev/null bs=1M count=2048
2147483648 bytes transferred in 348.144707 seconds (6168365 bytes/sec)
real    5m48.169s, user    0m0.019s, sys     1m29.962s
[2nd]# time dd if=data of=/dev/null bs=1M count=2048
2147483648 bytes transferred in 324.017361 seconds (6627681 bytes/sec)
real    5m24.038s, user    0m0.023s, sys     1m20.602s

read from full fragmented file without this patch

[1st]# time dd if=data of=/dev/null bs=1M count=2048
2147483647 bytes transferred in 1156.845693 seconds (1856327 bytes/sec)
real    19m16.855s, user    0m0.031s, sys     1m32.172s
[2nd]# time dd if=data of=/dev/null bs=1M count=2048
2147483647 bytes transferred in 1066.518713 seconds (2013545 bytes/sec)
real    17m46.526s, user    0m0.023s, sys     1m33.630s

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] FAT: use hlist_head for fat_inode_hashtable

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] remove dead exports from fs/fat/

These were used by the defunct umsdos code only.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] sparc32: add atomic_sub_and_test()

Add atomic_sub_and_test() to sparc32, implemented in terms of
atomic_sub_return(), so reiser4 can be simultaneously microoptimized for
x86 and made to pass compilation testing on sparc32.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] atomic_inc_return() for sparc64

This patch declares atomic_add_return() as an alias of __atomic_add().
atomic64_add_return(),atomic_sub_return() and atomic64_sub_return() are
same.

Signed-off-by: KaiGai, Kohei <kaigai@ak.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] atomic_inc_return() for arm26

This patch implements atomic_inc_return() and so on for ARM26.  Because
Hugh says that SMP is not supported in arm26, it is implemented by normal
operations between local_irq_save() and local_irq_restore() like another
atomic operations.

Signed-off-by: KaiGai, Kohei <kaigai@ak.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] atomic_inc_return() for arm

This patch declares atomic_inc_return() as the alias of atomic_add_return()
and atomic_dec_return() as an alias of atomic_dec_return().

Signed-off-by: KaiGai, Kohei <kaigai@ak.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] atomic_inc_return() for x86_64

Signed-off-by: KaiGai, Kohei <kaigai@ak.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] atomic_inc_return() for i386

This patch implements atomic_inc_return() and so on for i386, and includes
runtime check whether CPU is legacy 386.

Signed-off-by: KaiGai, Kohei <kaigai@ak.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] JFFS2: work around uninitialised use of usercompr field by old code.

We didn't use the 'usercompr' field before recently.  Unfortunately we didn't
set it to zero either, so there's a lot of file systems out there with it set
to 0x5a.  Deal with that rather than getting confused.

Signed-Off-By: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64/io_apic init section fixups

Code section errors in i386/io_apic.c found by scripts/reference_init.pl.
Looks like they could cause problems for a few drivers or in a real hotplug
environment.

Error: ./arch/i386/kernel/io_apic.o .text refers to 000018ff R_386_PC32        .init.text

call chain:
  snd_mpu401_acpi_resource
    acpi_register_gsi
      mp_register_gsi
io_apic_set_pci_routing
{A}   ioapic_register_intr
    IO_APIC_irq_trigger
      find_irq_entry

Error: ./arch/i386/kernel/io_apic.o .text refers to 00001967 R_386_PC32        .init.text

(as above thru {A}, then:)
  IO_APIC_irq_trigger
    irq_trigger
      MPBIOS_trigger >> removing __init from this led to
   needing to remove __init from
   EISA_ELCR also.

Signed-off-by: Randy Dunlap <rddunlap@osdl.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: correct copy_user_generic return value when exception happens

Fix a bug that arch/x86_64/lib/copy_user:copy_user_generic will return a
wrong value when exception happens.

In the case the address is not 8-byte aligned (i.e.  go into
Lbad_alignment), if exception happens in Ls11, %rdx will be wrong number of
copied bytes, then copy_user_generic returns wrong value.  It also fixed a
bug of zeroing wrong number of bytes of destination at this situation.  (In
Lzero_rest)

Signed-off-by: Yanmin Zhang <yanmin.zhang@intel.com>
Signed-off-by: Nanhai Zou <nanhai.zou@intel.com>
Signed-off-by: Gordon Jin <gordon.jin@intel.com>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64 Kconfig: Split CONFIG_NUMA_EMU and CONFIG_K8_NUMA

Split CONFIG_NUMA_EMU and CONFIG_K8_NUMA because they are independent.

Signed-off-by: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: fix IOAPIC on Nvidia boards

Originally suggested by Zwane Mwaikumbo

Ignore all ACPI timer overrides on all Nvidia boards.  The fallback doesn't
work and no Nvidia boards needs a timer override.  But some buggy BIOS have
it anyways.

Also enable IO-APIC mode by default for Nvidia then.

Thanks to Andy Currid for confirming this.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: add an option to configure oops stack dump

Add an kstack= option to configure how much stack should be printed on a
oops.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: add notsc option

Add an notsc option so that it can be turned off again.

This may be useful on the Summit, but will only work when there is a HPET
fallback.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: use TSC on SMP EM64T machines

x86-64 MP kernels always used HPET timing when available.  This was done
because AMD systems don't have synchronized TSCs in all cases.  On Intel
this is not true (except Summit), so use faster TSC timing for those.

Also only enable HPET gettimeofday when there is more than one CPU online.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: intialize hpet char driver

Initialize HPET char driver

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: drop old APIC workaround

Suggested by Ingo Molnar.

Drop APIC level workaround for old IO-APICs on x86-64 because it is a
significant part of interrupt handling time.  x86-64 only runs on modern
chipsets that probably don't have this bug.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] module_param_array() should take a pointer

module_param_array() takes a variable to put the number of elements in.
Looking through the uses, many people don't care, so they declare a dummy
or share one variable between several parameters.  The latter is
problematic because sysfs uses that number to decide how many to display.

The solution is to change the variable arg to a pointer, and if the pointer
is NULL, use the "max" value.  This change is fairly small, but fixing up
the callers is a lot of (trivial) churn.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Merge http://lia64.bkbits.net/linux-ia64-release-2.6.10
into ppc970.osdl.org:/home/torvalds/v2.6/linux

Merge agluck-lia64.sc.intel.com:/data/home/aegl/BK/work/sgi-reorg
into agluck-lia64.sc.intel.com:/data/home/aegl/BK/linux-ia64-release-2.6.10

[IA64-SGI] Fixes calling arg1 for bte_crb_error_handler()

Signed-off-by: Patrick Gefre <pfg@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>

[IA64-SGI] Distribute useage of BTE interfaces.

During peak utilization periods, the first interface on a node has
an inordinately large amount of contention.  This is due to all
cpus starting their scan for an interface at 0.  This patch distributes
that based upon the slice the requesting cpu is attached to.

Signed-off-by: Robin Holt
Signed-off-by: Tony Luck <tony.luck@intel.com>

[IA64-SGI] Correct BTE notification timeouts on SN2.

The SN2 Block Transfer Engine occassionally fails to send a notification
that it has completed a transfer to the kernel.  This patch adds a
timeout mechanism which will detect the failure, reset the interface,
and then retry the transfer.

Signed-off-by: Robin Holt
Signed-off-by: Tony Luck <tony.luck@intel.com>

[IA64-SGI] BUG_ON test was backwards

Signed-off-by: Patrick Gefre <pfg@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>

[IA64-SGI] Double spin_unlock in bte.c

If all bte interfaces are in use, the current code will leave the
bte pointer set when exiting from the loop trying to locate an
interface to use.  This results in two processes using the same
interface and both trying to free the same one.  With a
premptible kernel, this results in the preempt count getting
off.

Signed-off-by: Robin Holt <holt@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>

Merge bk://kernel.bkbits.net/gregkh/linux/i2c-2.6
into ppc970.osdl.org:/home/torvalds/v2.6/linux

Merge bk://bk.arm.linux.org.uk/linux-2.6-serial
into ppc970.osdl.org:/home/torvalds/v2.6/linux

[SERIAL] serial_reg.h update.

This includes a fuller definition of the 8250 and compatible bits,
and adding notes where the definition varies between different
chips.

This also fixes the following build error:

drivers/serial/8250.c:185: error: `UART_FCR_R_TRIG_10'

[SERIAL] Keep trying to register our console device.

Some serial drivers receive their serial port device information via
the device model.  This unfortunately means that the selected port
may not be available when the console subsystem initialises, so we
must keep trying to register the console after each port is added.

[SERIAL] Convert 8250_pci to use new serial8250_register_port()

This allows 8250 PCI ports to register with their correct device
structures.

[SERIAL] Add new port registration/unregistration functions.

serial8250_register_port()/serial8250_unregister_port() has the
capability of registering ports with their struct device nodes,
which allows sysfs to indicate which tty devices belong to which
hardware devices.

We also add a serial8250 platform device driver in an initial
attempt at PM for ISA ports.  However, I'm leaving out the
platform device for the time being since adding that would cause
potential oops issues.

[SERIAL] Make port autoprobing set up->capabilities.

Convert port autoprobing to set up->capabilities as it discovers
various capabilities of the port.  Warn when the detected
capabilities do not match those in the uart_config table.

[SERIAL] Add explaination why we don't use RTS flow control.

[SERIAL] Clean up handling of LSR in receive function.

It's pointless accessing the LSR value via a pointer all the time -
it prevents the compiler optimising it.  Also, ensure that we
recognise a break sent during a kernel printk correctly.

[SERIAL] Add comment about frobbing the 950's ACR on TX disable.

Add a bit of explaination why we only turn off the transmitter if
we are called from uart_stop().

[SERIAL] Convert TI16C750 flow control into a port capability.

Add UART_CAP_AFE, and use this to enable TI16C750 flow control,
but only if we have 32 bytes or more of FIFO.

[SERIAL] Add FCR setting to serial8250_config structure.

This allows us to adapt the FCR for each port type in a much more
flexible way, and allows us to set the transmit trigger levels.

Merge PCI updates

I2C: convert from pci_module_init to pci_register_driver for all i2c drivers.

Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

[PATCH] I2C: i2c bus power management support

The patch below adds power management support to the i2c bus.
It adds just two small functions which call down to the devices
power management functions if they are present, so the i2c device
drivers will receive the suspend and resume events.

From: Gerd Knorr <kraxel@bytesex.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

[PATCH] I2C: lm87 driver ported to Linux 2.6

This is my port of the lm87 driver to Linux 2.6. It is based on the
preliminary work of Jeff Oliver. I then significantly improved the code,
added functionality, tested the whole thing on a real motherboard, fixed
a couple remaining bugs, and here we are.

I'll port a number of improvements and fixes back to the 2.4 version of
the driver after lm_sensors 2.8.8 is released (i.e. soon).

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>