54 files changed, 3661 insertions, 338 deletions
diff --git a/Documentation/DMA-API.txt b/Documentation/DMA-API.txt
index 1af0f2d50220..2ffb0d62f0fe 100644
--- a/Documentation/DMA-API.txt
+++ b/Documentation/DMA-API.txt
@@ -33,7 +33,9 @@ pci_alloc_consistent(struct pci_dev *dev, size_t size,
 Consistent memory is memory for which a write by either the device or
 the processor can immediately be read by the processor or device
-without having to worry about caching effects.
+without having to worry about caching effects.  (You may however need
+to make sure to flush the processor's write buffers before telling
+devices to read that memory.)
 This routine allocates a region of <size> bytes of consistent memory.
 it also returns a <dma_handle> which may be cast to an unsigned
@@ -304,12 +306,12 @@ dma address with dma_mapping_error(). A non zero return value means the mapping
 could not be created and the driver should take appropriate action (eg
 reduce current DMA mapping usage or delay and try again later).
-int
+        int
-dma_map_sg(struct device *dev, struct scatterlist *sg, int nents,
+        dma_map_sg(struct device *dev, struct scatterlist *sg,
-           enum dma_data_direction direction)
+                int nents, enum dma_data_direction direction)
-int
+        int
-pci_map_sg(struct pci_dev *hwdev, struct scatterlist *sg,
+        pci_map_sg(struct pci_dev *hwdev, struct scatterlist *sg,
-           int nents, int direction)
+                int nents, int direction)
 Maps a scatter gather list from the block layer.
@@ -327,12 +329,33 @@ critical that the driver do something, in the case of a block driver
 aborting the request or even oopsing is better than doing nothing and
 corrupting the filesystem.
-void
+With scatterlists, you use the resulting mapping like this:
-dma_unmap_sg(struct device *dev, struct scatterlist *sg, int nhwentries,
-             enum dma_data_direction direction)
+        int i, count = dma_map_sg(dev, sglist, nents, direction);
-void
+        struct scatterlist *sg;
-pci_unmap_sg(struct pci_dev *hwdev, struct scatterlist *sg,
-             int nents, int direction)
+        for (i = 0, sg = sglist; i < count; i++, sg++) {
+                hw_address[i] = sg_dma_address(sg);
+                hw_len[i] = sg_dma_len(sg);
+        }
+where nents is the number of entries in the sglist.
+The implementation is free to merge several consecutive sglist entries
+into one (e.g. with an IOMMU, or if several pages just happen to be
+physically contiguous) and returns the actual number of sg entries it
+mapped them to. On failure 0, is returned.
+Then you should loop count times (note: this can be less than nents times)
+and use sg_dma_address() and sg_dma_len() macros where you previously
+accessed sg->address and sg->length as shown above.
+        void
+        dma_unmap_sg(struct device *dev, struct scatterlist *sg,
+                int nhwentries, enum dma_data_direction direction)
+        void
+        pci_unmap_sg(struct pci_dev *hwdev, struct scatterlist *sg,
+                int nents, int direction)
 unmap the previously mapped scatter/gather list.  All the parameters
 must be the same as those and passed in to the scatter/gather mapping
diff --git a/Documentation/DMA-mapping.txt b/Documentation/DMA-mapping.txt
index ee4bb73683cd..7c717699032c 100644
--- a/Documentation/DMA-mapping.txt
+++ b/Documentation/DMA-mapping.txt
@@ -58,11 +58,15 @@ translating each of those pages back to a kernel address using
 something like __va().  [ EDIT: Update this when we integrate
 Gerd Knorr's generic code which does this. ]
-This rule also means that you may not use kernel image addresses
+This rule also means that you may use neither kernel image addresses
-(ie. items in the kernel's data/text/bss segment, or your driver's)
+(items in data/text/bss segments), nor module image addresses, nor
-nor may you use kernel stack addresses for DMA.  Both of these items
+stack addresses for DMA.  These could all be mapped somewhere entirely
-might be mapped somewhere entirely different than the rest of physical
+different than the rest of physical memory.  Even if those classes of
-memory.
+memory could physically work with DMA, you'd need to ensure the I/O
+buffers were cacheline-aligned.  Without that, you'd see cacheline
+sharing problems (data corruption) on CPUs with DMA-incoherent caches.
+(The CPU could write to one word, DMA would write to a different one
+in the same cache line, and one of them could be overwritten.)
 Also, this means that you cannot take the return of a kmap()
 call and DMA to/from that.  This is similar to vmalloc().
@@ -194,7 +198,7 @@ document for how to handle this case.
 Finally, if your device can only drive the low 24-bits of
 address during PCI bus mastering you might do something like:
-        if (pci_set_dma_mask(pdev, 0x00ffffff)) {
+        if (pci_set_dma_mask(pdev, DMA_24BIT_MASK)) {
                printk(KERN_WARNING
                       "mydev: 24-bit DMA addressing not available.\n");
                goto ignore_this_device;
@@ -212,7 +216,7 @@ functions (for example a sound card provides playback and record
 functions) and the various different functions have _different_
 DMA addressing limitations, you may wish to probe each mask and
 only provide the functionality which the machine can handle.  It
-is important that the last call to pci_set_dma_mask() be for the 
+is important that the last call to pci_set_dma_mask() be for the
 most specific mask.
 Here is pseudo-code showing how this might be done:
@@ -284,6 +288,11 @@ There are two types of DMA mappings:
             in order to get correct behavior on all platforms.
+             Also, on some platforms your driver may need to flush CPU write
+             buffers in much the same way as it needs to flush write buffers
+             found in PCI bridges (such as by reading a register's value
+             after writing it).
 - Streaming DMA mappings which are usually mapped for one DMA transfer,
  unmapped right after it (unless you use pci_dma_sync_* below) and for which
  hardware can optimize for sequential accesses.
@@ -303,6 +312,9 @@ There are two types of DMA mappings:
 Neither type of DMA mapping has alignment restrictions that come
 from PCI, although some devices may have such restrictions.
+Also, systems with caches that aren't DMA-coherent will work better
+when the underlying buffers don't share cache lines with other data.
                 Using Consistent DMA mappings.
diff --git a/Documentation/DocBook/Makefile b/Documentation/DocBook/Makefile
index 7d87dd73cbe4..5a2882d275ba 100644
--- a/Documentation/DocBook/Makefile
+++ b/Documentation/DocBook/Makefile
@@ -2,7 +2,7 @@
 # This makefile is used to generate the kernel documentation,
 # primarily based on in-line comments in various source files.
 # See Documentation/kernel-doc-nano-HOWTO.txt for instruction in how
-# to ducument the SRC - and how to read it.
+# to document the SRC - and how to read it.
 # To add a new book the only step required is to add the book to the
 # list of DOCBOOKS.
diff --git a/Documentation/DocBook/kernel-api.tmpl b/Documentation/DocBook/kernel-api.tmpl
index 8c9c6704e85b..ca02e04a906c 100644
--- a/Documentation/DocBook/kernel-api.tmpl
+++ b/Documentation/DocBook/kernel-api.tmpl
@@ -322,7 +322,6 @@ X!Earch/i386/kernel/mca.c
  <chapter id="sysfs">
     <title>The Filesystem for Exporting Kernel Objects</title>
 !Efs/sysfs/file.c
-!Efs/sysfs/dir.c
 !Efs/sysfs/symlink.c
 !Efs/sysfs/bin.c
  </chapter>
diff --git a/Documentation/DocBook/libata.tmpl b/Documentation/DocBook/libata.tmpl
index 5bcbb6ee3bc0..f869b03929db 100644
--- a/Documentation/DocBook/libata.tmpl
+++ b/Documentation/DocBook/libata.tmpl
@@ -705,7 +705,7 @@ and other resources, etc.
        <sect1><title>ata_scsi_error()</title>
        <para>
-        ata_scsi_error() is the current hostt->eh_strategy_handler()
+        ata_scsi_error() is the current transportt->eh_strategy_handler()
        for libata.  As discussed above, this will be entered in two
        cases - timeout and ATAPI error completion.  This function
        calls low level libata driver's eng_timeout() callback, the
diff --git a/Documentation/HOWTO b/Documentation/HOWTO
index 6c9e746267da..915ae8c986c6 100644
--- a/Documentation/HOWTO
+++ b/Documentation/HOWTO
@@ -603,7 +603,8 @@ start exactly where you are now.
 ----------
-Thanks to Paolo Ciarrocchi who allowed the "Development Process" section
+Thanks to Paolo Ciarrocchi who allowed the "Development Process"
+(http://linux.tar.bz/articles/2.6-development_process) section
 to be based on text he had written, and to Randy Dunlap and Gerrit
 Huizenga for some of the list of things you should and should not say.
 Also thanks to Pat Mochel, Hanna Linder, Randy Dunlap, Kay Sievers,
diff --git a/Documentation/acpi-hotkey.txt b/Documentation/acpi-hotkey.txt
index 744f1aec6553..38040fa37649 100644
--- a/Documentation/acpi-hotkey.txt
+++ b/Documentation/acpi-hotkey.txt
@@ -30,7 +30,7 @@ specific hotkey(event))
 echo "event_num:event_type:event_argument" > 
        /proc/acpi/hotkey/action.
 The result of the execution of this aml method is 
-attached to /proc/acpi/hotkey/poll_method, which is dnyamically
+attached to /proc/acpi/hotkey/poll_method, which is dynamically
 created.  Please use command "cat /proc/acpi/hotkey/polling_method" 
 to retrieve it.
diff --git a/Documentation/block/switching-sched.txt b/Documentation/block/switching-sched.txt
new file mode 100644
index 000000000000..5fa130a67531
--- /dev/null
+++ b/Documentation/block/switching-sched.txt
@@ -0,0 +1,22 @@
+As of the Linux 2.6.10 kernel, it is now possible to change the
+IO scheduler for a given block device on the fly (thus making it possible,
+for instance, to set the CFQ scheduler for the system default, but
+set a specific device to use the anticipatory or noop schedulers - which
+can improve that device's throughput).
+To set a specific scheduler, simply do this:
+echo SCHEDNAME > /sys/block/DEV/queue/scheduler
+where SCHEDNAME is the name of a defined IO scheduler, and DEV is the
+device name (hda, hdb, sga, or whatever you happen to have).
+The list of defined schedulers can be found by simply doing
+a "cat /sys/block/DEV/queue/scheduler" - the list of valid names
+will be displayed, with the currently selected scheduler in brackets:
+# cat /sys/block/hda/queue/scheduler
+noop anticipatory deadline [cfq]
+# echo anticipatory > /sys/block/hda/queue/scheduler
+# cat /sys/block/hda/queue/scheduler
+noop [anticipatory] deadline cfq
diff --git a/Documentation/cpu-freq/index.txt b/Documentation/cpu-freq/index.txt
index 5009805f9378..ffdb5323df37 100644
--- a/Documentation/cpu-freq/index.txt
+++ b/Documentation/cpu-freq/index.txt
@@ -53,4 +53,4 @@ the CPUFreq Mailing list:
 * http://lists.linux.org.uk/mailman/listinfo/cpufreq
 Clock and voltage scaling for the SA-1100:
-* http://www.lart.tudelft.nl/projects/scaling
+* http://www.lartmaker.nl/projects/scaling
diff --git a/Documentation/devices.txt b/Documentation/devices.txt
index 3c406acd4dfa..b369a8c46a73 100644
--- a/Documentation/devices.txt
+++ b/Documentation/devices.txt
@@ -1721,11 +1721,6 @@ Your cooperation is appreciated.
                These devices support the same API as the generic SCSI
                devices.
- 97 block       Packet writing for CD/DVD devices
-                  0 = /dev/pktcdvd0     First packet-writing module
-                  1 = /dev/pktcdvd1     Second packet-writing module
-                    ...
 98 char        Control and Measurement Device (comedi)
                  0 = /dev/comedi0      First comedi device
                  1 = /dev/comedi1      Second comedi device
diff --git a/Documentation/dvb/get_dvb_firmware b/Documentation/dvb/get_dvb_firmware
index 15fc8fbef67e..4820366b6ae8 100644
--- a/Documentation/dvb/get_dvb_firmware
+++ b/Documentation/dvb/get_dvb_firmware
@@ -259,9 +259,9 @@ sub dibusb {
 }
 sub nxt2002 {
-    my $sourcefile = "Broadband4PC_4_2_11.zip";
+    my $sourcefile = "Technisat_DVB-PC_4_4_COMPACT.zip";
    my $url = "http://www.bbti.us/download/windows/$sourcefile";
-    my $hash = "c6d2ea47a8f456d887ada0cfb718ff2a";
+    my $hash = "476befae8c7c1bb9648954060b1eec1f";
    my $outfile = "dvb-fe-nxt2002.fw";
    my $tmpdir = tempdir(DIR => "/tmp", CLEANUP => 1);
@@ -269,8 +269,8 @@ sub nxt2002 {
    wgetfile($sourcefile, $url);
    unzip($sourcefile, $tmpdir);
-    verify("$tmpdir/SkyNETU.sys", $hash);
+    verify("$tmpdir/SkyNET.sys", $hash);
-    extract("$tmpdir/SkyNETU.sys", 375832, 5908, $outfile);
+    extract("$tmpdir/SkyNET.sys", 331624, 5908, $outfile);
    $outfile;
 }
diff --git a/Documentation/feature-removal-schedule.txt b/Documentation/feature-removal-schedule.txt
index 495858b236b6..43ab119963d5 100644
--- a/Documentation/feature-removal-schedule.txt
+++ b/Documentation/feature-removal-schedule.txt
@@ -25,8 +25,9 @@ Who:	Adrian Bunk <bunk@stusta.de>
 ---------------------------
-What:   drivers depending on OBSOLETE_OSS_DRIVER
+What:   drivers that were depending on OBSOLETE_OSS_DRIVER
-When:   January 2006
+        (config options already removed)
+When:   before 2.6.19
 Why:    OSS drivers with ALSA replacements
 Who:    Adrian Bunk <bunk@stusta.de>
@@ -56,6 +57,15 @@ Who:	Jody McIntyre <scjody@steamballoon.com>
 ---------------------------
+What:   sbp2: module parameter "force_inquiry_hack"
+When:   July 2006
+Why:    Superceded by parameter "workarounds". Both parameters are meant to be
+        used ad-hoc and for single devices only, i.e. not in modprobe.conf,
+        therefore the impact of this feature replacement should be low.
+Who:    Stefan Richter <stefanr@s5r6.in-berlin.de>
+---------------------------
 What:   Video4Linux API 1 ioctls and video_decoder.h from Video devices.
 When:   July 2006
 Why:    V4L1 AP1 was replaced by V4L2 API. during migration from 2.4 to 2.6
@@ -71,14 +81,6 @@ Who:	Mauro Carvalho Chehab <mchehab@brturbo.com.br>
 ---------------------------
-What:   remove EXPORT_SYMBOL(panic_timeout)
-When:   April 2006
-Files:  kernel/panic.c
-Why:    No modular usage in the kernel.
-Who:    Adrian Bunk <bunk@stusta.de>
---------------------------
 What:   remove EXPORT_SYMBOL(insert_resource)
 When:   April 2006
 Files:  kernel/resource.c
@@ -127,13 +129,6 @@ Who:	Christoph Hellwig <hch@lst.de>
 ---------------------------
-What:   EXPORT_SYMBOL(lookup_hash)
-When:   January 2006
-Why:    Too low-level interface.  Use lookup_one_len or lookup_create instead.
-Who:    Christoph Hellwig <hch@lst.de>
---------------------------
 What:   CONFIG_FORCED_INLINING
 When:   June 2006
 Why:    Config option is there to see if gcc is good enough. (in january
@@ -241,3 +236,15 @@ Why:	The USB subsystem has changed a lot over time, and it has been
 Who:    Greg Kroah-Hartman <gregkh@suse.de>
 ---------------------------
+What:   find_trylock_page
+When:   January 2007
+Why:    The interface no longer has any callers left in the kernel. It
+        is an odd interface (compared with other find_*_page functions), in
+        that it does not take a refcount to the page, only the page lock.
+        It should be replaced with find_get_page or find_lock_page if possible.
+        This feature removal can be reevaluated if users of the interface
+        cannot cleanly use something else.
+Who:    Nick Piggin <npiggin@suse.de>
+---------------------------
diff --git a/Documentation/filesystems/sysfs.txt b/Documentation/filesystems/sysfs.txt
index c8bce82ddcac..89b1d196ca80 100644
--- a/Documentation/filesystems/sysfs.txt
+++ b/Documentation/filesystems/sysfs.txt
@@ -246,6 +246,7 @@ class/
 devices/
 firmware/
 net/
+fs/
 devices/ contains a filesystem representation of the device tree. It maps
 directly to the internal kernel device tree, which is a hierarchy of
@@ -264,6 +265,10 @@ drivers/ contains a directory for each device driver that is loaded
 for devices on that particular bus (this assumes that drivers do not
 span multiple bus types).
+fs/ contains a directory for some filesystems.  Currently each
+filesystem wanting to export attributes must create its own hierarchy
+below fs/ (see ./fuse.txt for an example).
 More information can driver-model specific features can be found in
 Documentation/driver-model/. 
diff --git a/Documentation/filesystems/vfs.txt b/Documentation/filesystems/vfs.txt
index adaa899e5c90..3a2e5520c1e3 100644
--- a/Documentation/filesystems/vfs.txt
+++ b/Documentation/filesystems/vfs.txt
@@ -694,7 +694,7 @@ struct file_operations
 ----------------------
 This describes how the VFS can manipulate an open file. As of kernel
-2.6.13, the following members are defined:
+2.6.17, the following members are defined:
 struct file_operations {
        loff_t (*llseek) (struct file *, loff_t, int);
@@ -723,6 +723,10 @@ struct file_operations {
        int (*check_flags)(int);
        int (*dir_notify)(struct file *filp, unsigned long arg);
        int (*flock) (struct file *, int, struct file_lock *);
+        ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, size_t, unsigned 
+int);
+        ssize_t (*splice_read)(struct file *, struct pipe_inode_info *, size_t, unsigned  
+int);
 };
 Again, all methods are called without any locks being held, unless
@@ -790,6 +794,12 @@ otherwise noted.
  flock: called by the flock(2) system call
+  splice_write: called by the VFS to splice data from a pipe to a file. This
+                method is used by the splice(2) system call
+  splice_read: called by the VFS to splice data from file to a pipe. This
+               method is used by the splice(2) system call
 Note that the file operations are implemented by the specific
 filesystem in which the inode resides. When opening a device node
 (character or block special) most filesystems will call special
diff --git a/Documentation/firmware_class/README b/Documentation/firmware_class/README
index 43e836c07ae8..e9cc8bb26f7d 100644
--- a/Documentation/firmware_class/README
+++ b/Documentation/firmware_class/README
@@ -105,20 +105,3 @@
   on the setup, so I think that the choice on what firmware to make
   persistent should be left to userspace.
- - Why register_firmware()+__init can be useful:
-        - For boot devices needing firmware.
-        - To make the transition easier:
-                The firmware can be declared __init and register_firmware()
-                called on module_init. Then the firmware is warranted to be
-                there even if "firmware hotplug userspace" is not there yet or
-                it doesn't yet provide the needed firmware.
-                Once the firmware is widely available in userspace, it can be
-                removed from the kernel. Or made optional (CONFIG_.*_FIRMWARE).
-        In either case, if firmware hotplug support is there, it can move the
-        firmware out of kernel memory into the real filesystem for later
-        usage.
-        Note: If persistence is implemented on top of initramfs,
-        register_firmware() may not be appropriate.
diff --git a/Documentation/firmware_class/firmware_sample_driver.c b/Documentation/firmware_class/firmware_sample_driver.c
index ad3edaba4533..87feccdb5c9f 100644
--- a/Documentation/firmware_class/firmware_sample_driver.c
+++ b/Documentation/firmware_class/firmware_sample_driver.c
@@ -5,8 +5,6 @@
 *
 * Sample code on how to use request_firmware() from drivers.
 *
- * Note that register_firmware() is currently useless.
- *
 */
 #include <linux/module.h>
@@ -17,11 +15,6 @@
 #include "linux/firmware.h"
-#define WE_CAN_NEED_FIRMWARE_BEFORE_USERSPACE_IS_AVAILABLE
-#ifdef WE_CAN_NEED_FIRMWARE_BEFORE_USERSPACE_IS_AVAILABLE
-char __init inkernel_firmware[] = "let's say that this is firmware\n";
-#endif
 static struct device ghost_device = {
        .bus_id    = "ghost0",
 };
@@ -104,10 +97,6 @@ static void sample_probe_async(void)
 static int sample_init(void)
 {
-#ifdef WE_CAN_NEED_FIRMWARE_BEFORE_USERSPACE_IS_AVAILABLE
-        register_firmware("sample_driver_fw", inkernel_firmware,
-                          sizeof(inkernel_firmware));
-#endif
        device_initialize(&ghost_device);
        /* since there is no real hardware insertion I just call the
         * sample probe functions here */
diff --git a/Documentation/fujitsu/frv/kernel-ABI.txt b/Documentation/fujitsu/frv/kernel-ABI.txt
index 0ed9b0a779bc..8b0a5fc8bfd9 100644
--- a/Documentation/fujitsu/frv/kernel-ABI.txt
+++ b/Documentation/fujitsu/frv/kernel-ABI.txt
@@ -1,17 +1,19 @@
-                                 =================================
+                        =================================
-                                 INTERNAL KERNEL ABI FOR FR-V ARCH
+                        INTERNAL KERNEL ABI FOR FR-V ARCH
-                                 =================================
+                        =================================
-The internal FRV kernel ABI is not quite the same as the userspace ABI. A number of the registers
+The internal FRV kernel ABI is not quite the same as the userspace ABI. A
-are used for special purposed, and the ABI is not consistent between modules vs core, and MMU vs
+number of the registers are used for special purposed, and the ABI is not
-no-MMU.
+consistent between modules vs core, and MMU vs no-MMU.
-This partly stems from the fact that FRV CPUs do not have a separate supervisor stack pointer, and
+This partly stems from the fact that FRV CPUs do not have a separate
-most of them do not have any scratch registers, thus requiring at least one general purpose
+supervisor stack pointer, and most of them do not have any scratch
-register to be clobbered in such an event. Also, within the kernel core, it is possible to simply
+registers, thus requiring at least one general purpose register to be
-jump or call directly between functions using a relative offset. This cannot be extended to modules
+clobbered in such an event. Also, within the kernel core, it is possible to
-for the displacement is likely to be too far. Thus in modules the address of a function to call
+simply jump or call directly between functions using a relative offset.
-must be calculated in a register and then used, requiring two extra instructions.
+This cannot be extended to modules for the displacement is likely to be too
+far. Thus in modules the address of a function to call must be calculated
+in a register and then used, requiring two extra instructions.
 This document has the following sections:
@@ -39,7 +41,8 @@ When a system call is made, the following registers are effective:
 CPU OPERATING MODES
 ===================
-The FR-V CPU has three basic operating modes. In order of increasing capability:
+The FR-V CPU has three basic operating modes. In order of increasing
+capability:
  (1) User mode.
@@ -47,42 +50,46 @@ The FR-V CPU has three basic operating modes. In order of increasing capability:
  (2) Kernel mode.
-      Normal kernel mode. There are many additional control registers available that may be
+      Normal kernel mode. There are many additional control registers
-      accessed in this mode, in addition to all the stuff available to user mode. This has two
+      available that may be accessed in this mode, in addition to all the
-      submodes:
+      stuff available to user mode. This has two submodes:
      (a) Exceptions enabled (PSR.T == 1).
-          Exceptions will invoke the appropriate normal kernel mode handler. On entry to the
+          Exceptions will invoke the appropriate normal kernel mode
-          handler, the PSR.T bit will be cleared.
+          handler. On entry to the handler, the PSR.T bit will be cleared.
      (b) Exceptions disabled (PSR.T == 0).
-          No exceptions or interrupts may happen. Any mandatory exceptions will cause the CPU to
+          No exceptions or interrupts may happen. Any mandatory exceptions
-          halt unless the CPU is told to jump into debug mode instead.
+          will cause the CPU to halt unless the CPU is told to jump into
+          debug mode instead.
  (3) Debug mode.
-      No exceptions may happen in this mode. Memory protection and management exceptions will be
+      No exceptions may happen in this mode. Memory protection and
-      flagged for later consideration, but the exception handler won't be invoked. Debugging traps
+      management exceptions will be flagged for later consideration, but
-      such as hardware breakpoints and watchpoints will be ignored. This mode is entered only by
+      the exception handler won't be invoked. Debugging traps such as
-      debugging events obtained from the other two modes.
+      hardware breakpoints and watchpoints will be ignored. This mode is
+      entered only by debugging events obtained from the other two modes.
-      All kernel mode registers may be accessed, plus a few extra debugging specific registers.
+      All kernel mode registers may be accessed, plus a few extra debugging
+      specific registers.
 =================================
 INTERNAL KERNEL-MODE REGISTER ABI
 =================================
-There are a number of permanent register assignments that are set up by entry.S in the exception
+There are a number of permanent register assignments that are set up by
-prologue. Note that there is a complete set of exception prologues for each of user->kernel
+entry.S in the exception prologue. Note that there is a complete set of
-transition and kernel->kernel transition. There are also user->debug and kernel->debug mode
+exception prologues for each of user->kernel transition and kernel->kernel
-transition prologues.
+transition. There are also user->debug and kernel->debug mode transition
+prologues.
        REGISTER        FLAVOUR USE
-        =============== ======= ====================================================
+        =============== ======= ==============================================
        GR1                     Supervisor stack pointer
        GR15                    Current thread info pointer
        GR16                    GP-Rel base register for small data
@@ -92,10 +99,12 @@ transition prologues.
        GR31            NOMMU   Destroyed by debug mode entry
        GR31            MMU     Destroyed by TLB miss kernel mode entry
        CCR.ICC2                Virtual interrupt disablement tracking
-        CCCR.CC3                Cleared by exception prologue (atomic op emulation)
+        CCCR.CC3                Cleared by exception prologue 
+                                (atomic op emulation)
        SCR0            MMU     See mmu-layout.txt.
        SCR1            MMU     See mmu-layout.txt.
-        SCR2            MMU     Save for EAR0 (destroyed by icache insns in debug mode)
+        SCR2            MMU     Save for EAR0 (destroyed by icache insns 
+                                               in debug mode)
        SCR3            MMU     Save for GR31 during debug exceptions
        DAMR/IAMR       NOMMU   Fixed memory protection layout.
        DAMR/IAMR       MMU     See mmu-layout.txt.
@@ -104,18 +113,21 @@ transition prologues.
 Certain registers are also used or modified across function calls:
        REGISTER        CALL                            RETURN
-        =============== =============================== ===============================
+        =============== =============================== ======================
        GR0             Fixed Zero                      -
        GR2             Function call frame pointer
        GR3             Special                         Preserved
        GR3-GR7         -                               Clobbered
-        GR8             Function call arg #1            Return value (or clobbered)
+        GR8             Function call arg #1            Return value 
-        GR9             Function call arg #2            Return value MSW (or clobbered)
+                                                        (or clobbered)
+        GR9             Function call arg #2            Return value MSW 
+                                                        (or clobbered)
        GR10-GR13       Function call arg #3-#6         Clobbered
        GR14            -                               Clobbered
        GR15-GR16       Special                         Preserved
        GR17-GR27       -                               Preserved
-        GR28-GR31       Special                         Only accessed explicitly
+        GR28-GR31       Special                         Only accessed 
+                                                        explicitly
        LR              Return address after CALL       Clobbered
        CCR/CCCR        -                               Mostly Clobbered
@@ -124,46 +136,53 @@ Certain registers are also used or modified across function calls:
 INTERNAL DEBUG-MODE REGISTER ABI
 ================================
-This is the same as the kernel-mode register ABI for functions calls. The difference is that in
+This is the same as the kernel-mode register ABI for functions calls. The
-debug-mode there's a different stack and a different exception frame. Almost all the global
+difference is that in debug-mode there's a different stack and a different
-registers from kernel-mode (including the stack pointer) may be changed.
+exception frame. Almost all the global registers from kernel-mode
+(including the stack pointer) may be changed.
        REGISTER        FLAVOUR USE
-        =============== ======= ====================================================
+        =============== ======= ==============================================
        GR1                     Debug stack pointer
        GR16                    GP-Rel base register for small data
-        GR31                    Current debug exception frame pointer (__debug_frame)
+        GR31                    Current debug exception frame pointer 
+                                (__debug_frame)
        SCR3            MMU     Saved value of GR31
-Note that debug mode is able to interfere with the kernel's emulated atomic ops, so it must be
+Note that debug mode is able to interfere with the kernel's emulated atomic
-exceedingly careful not to do any that would interact with the main kernel in this regard. Hence
+ops, so it must be exceedingly careful not to do any that would interact
-the debug mode code (gdbstub) is almost completely self-contained. The only external code used is
+with the main kernel in this regard. Hence the debug mode code (gdbstub) is
-the sprintf family of functions.
+almost completely self-contained. The only external code used is the
+sprintf family of functions.
-Futhermore, break.S is so complicated because single-step mode does not switch off on entry to an
+Futhermore, break.S is so complicated because single-step mode does not
-exception. That means unless manually disabled, single-stepping will blithely go on stepping into
+switch off on entry to an exception. That means unless manually disabled,
-things like interrupts. See gdbstub.txt for more information.
+single-stepping will blithely go on stepping into things like interrupts.
+See gdbstub.txt for more information.
 ==========================
 VIRTUAL INTERRUPT HANDLING
 ==========================
-Because accesses to the PSR is so slow, and to disable interrupts we have to access it twice (once
+Because accesses to the PSR is so slow, and to disable interrupts we have
-to read and once to write), we don't actually disable interrupts at all if we don't have to. What
+to access it twice (once to read and once to write), we don't actually
-we do instead is use the ICC2 condition code flags to note virtual disablement, such that if we
+disable interrupts at all if we don't have to. What we do instead is use
-then do take an interrupt, we note the flag, really disable interrupts, set another flag and resume
+the ICC2 condition code flags to note virtual disablement, such that if we
-execution at the point the interrupt happened. Setting condition flags as a side effect of an
+then do take an interrupt, we note the flag, really disable interrupts, set
-arithmetic or logical instruction is really fast. This use of the ICC2 only occurs within the
+another flag and resume execution at the point the interrupt happened.
+Setting condition flags as a side effect of an arithmetic or logical
+instruction is really fast. This use of the ICC2 only occurs within the
 kernel - it does not affect userspace.
 The flags we use are:
 (*) CCR.ICC2.Z [Zero flag]
-     Set to virtually disable interrupts, clear when interrupts are virtually enabled. Can be
+     Set to virtually disable interrupts, clear when interrupts are
-     modified by logical instructions without affecting the Carry flag.
+     virtually enabled. Can be modified by logical instructions without
+     affecting the Carry flag.
 (*) CCR.ICC2.C [Carry flag]
@@ -176,8 +195,9 @@ What happens is this:
        ICC2.Z is 0, ICC2.C is 1.
- (2) An interrupt occurs. The exception prologue examines ICC2.Z and determines that nothing needs
+ (2) An interrupt occurs. The exception prologue examines ICC2.Z and
-     doing. This is done simply with an unlikely BEQ instruction.
+     determines that nothing needs doing. This is done simply with an
+     unlikely BEQ instruction.
 (3) The interrupts are disabled (local_irq_disable)
@@ -187,48 +207,56 @@ What happens is this:
        ICC2.Z would be set to 0.
-     A TIHI #2 instruction (trap #2 if condition HI - Z==0 && C==0) would be used to trap if
+     A TIHI #2 instruction (trap #2 if condition HI - Z==0 && C==0) would
-     interrupts were now virtually enabled, but physically disabled - which they're not, so the
+     be used to trap if interrupts were now virtually enabled, but
-     trap isn't taken. The kernel would then be back to state (1).
+     physically disabled - which they're not, so the trap isn't taken. The
+     kernel would then be back to state (1).
- (5) An interrupt occurs. The exception prologue examines ICC2.Z and determines that the interrupt
+ (5) An interrupt occurs. The exception prologue examines ICC2.Z and
-     shouldn't actually have happened. It jumps aside, and there disabled interrupts by setting
+     determines that the interrupt shouldn't actually have happened. It
-     PSR.PIL to 14 and then it clears ICC2.C.
+     jumps aside, and there disabled interrupts by setting PSR.PIL to 14
+     and then it clears ICC2.C.
 (6) If interrupts were then saved and disabled again (local_irq_save):
-        ICC2.Z would be shifted into the save variable and masked off (giving a 1).
+        ICC2.Z would be shifted into the save variable and masked off 
+        (giving a 1).
-        ICC2.Z would then be set to 1 (thus unchanged), and ICC2.C would be unaffected (ie: 0).
+        ICC2.Z would then be set to 1 (thus unchanged), and ICC2.C would be
+        unaffected (ie: 0).
 (7) If interrupts were then restored from state (6) (local_irq_restore):
-        ICC2.Z would be set to indicate the result of XOR'ing the saved value (ie: 1) with 1, which
+        ICC2.Z would be set to indicate the result of XOR'ing the saved
-        gives a result of 0 - thus leaving ICC2.Z set.
+        value (ie: 1) with 1, which gives a result of 0 - thus leaving
+        ICC2.Z set.
        ICC2.C would remain unaffected (ie: 0).
-     A TIHI #2 instruction would be used to again assay the current state, but this would do
+     A TIHI #2 instruction would be used to again assay the current state,
-     nothing as Z==1.
+     but this would do nothing as Z==1.
 (8) If interrupts were then enabled (local_irq_enable):
-        ICC2.Z would be cleared. ICC2.C would be left unaffected. Both flags would now be 0.
+        ICC2.Z would be cleared. ICC2.C would be left unaffected. Both
+        flags would now be 0.
-     A TIHI #2 instruction again issued to assay the current state would then trap as both Z==0
+     A TIHI #2 instruction again issued to assay the current state would
-     [interrupts virtually enabled] and C==0 [interrupts really disabled] would then be true.
+     then trap as both Z==0 [interrupts virtually enabled] and C==0
+     [interrupts really disabled] would then be true.
- (9) The trap #2 handler would simply enable hardware interrupts (set PSR.PIL to 0), set ICC2.C to
+ (9) The trap #2 handler would simply enable hardware interrupts 
-     1 and return.
+     (set PSR.PIL to 0), set ICC2.C to 1 and return.
 (10) Immediately upon returning, the pending interrupt would be taken.
-(11) The interrupt handler would take the path of actually processing the interrupt (ICC2.Z is
+(11) The interrupt handler would take the path of actually processing the
-     clear, BEQ fails as per step (2)).
+     interrupt (ICC2.Z is clear, BEQ fails as per step (2)).
-(12) The interrupt handler would then set ICC2.C to 1 since hardware interrupts are definitely
+(12) The interrupt handler would then set ICC2.C to 1 since hardware
-     enabled - or else the kernel wouldn't be here.
+     interrupts are definitely enabled - or else the kernel wouldn't be here.
 (13) On return from the interrupt handler, things would be back to state (1).
-This trap (#2) is only available in kernel mode. In user mode it will result in SIGILL.
+This trap (#2) is only available in kernel mode. In user mode it will
+result in SIGILL.
diff --git a/Documentation/i2c/busses/i2c-parport b/Documentation/i2c/busses/i2c-parport
index d9f23c0763f1..77b995dfca22 100644
--- a/Documentation/i2c/busses/i2c-parport
+++ b/Documentation/i2c/busses/i2c-parport
@@ -12,18 +12,22 @@ meant as a replacement for the older, individual drivers:
                      teletext adapters)
 It currently supports the following devices:
- * Philips adapter
+ * (type=0) Philips adapter
- * home brew teletext adapter
+ * (type=1) home brew teletext adapter
- * Velleman K8000 adapter
+ * (type=2) Velleman K8000 adapter
- * ELV adapter
+ * (type=3) ELV adapter
- * Analog Devices evaluation boards (ADM1025, ADM1030, ADM1031, ADM1032)
+ * (type=4) Analog Devices ADM1032 evaluation board
- * Barco LPT->DVI (K5800236) adapter
+ * (type=5) Analog Devices evaluation boards: ADM1025, ADM1030, ADM1031
+ * (type=6) Barco LPT->DVI (K5800236) adapter
 These devices use different pinout configurations, so you have to tell
 the driver what you have, using the type module parameter. There is no
 way to autodetect the devices. Support for different pinout configurations
 can be easily added when needed.
+Earlier kernels defaulted to type=0 (Philips).  But now, if the type
+parameter is missing, the driver will simply fail to initialize.
 Building your own adapter
 -------------------------
diff --git a/Documentation/input/joystick-parport.txt b/Documentation/input/joystick-parport.txt
index 88a011c9f985..d537c48cc6d0 100644
--- a/Documentation/input/joystick-parport.txt
+++ b/Documentation/input/joystick-parport.txt
@@ -36,12 +36,12 @@ with them.
  All NES and SNES use the same synchronous serial protocol, clocked from
 the computer's side (and thus timing insensitive). To allow up to 5 NES
-and/or SNES gamepads connected to the parallel port at once, the output
+and/or SNES gamepads and/or SNES mice connected to the parallel port at once,
-lines of the parallel port are shared, while one of 5 available input lines
+the output lines of the parallel port are shared, while one of 5 available
-is assigned to each gamepad.
+input lines is assigned to each gamepad.
  This protocol is handled by the gamecon.c driver, so that's the one
-you'll use for NES and SNES gamepads.
+you'll use for NES, SNES gamepads and SNES mice.
  The main problem with PC parallel ports is that they don't have +5V power
 source on any of their pins. So, if you want a reliable source of power
@@ -106,7 +106,7 @@ A, Turbo B, Select and Start, and is connected through 5 wires, then it is
 either a NES or NES clone and will work with this connection. SNES gamepads
 also use 5 wires, but have more buttons. They will work as well, of course.
-Pinout for NES gamepads                 Pinout for SNES gamepads
+Pinout for NES gamepads                 Pinout for SNES gamepads and mice
           +----> Power                   +-----------------------\
           |                            7 | o  o  o  o |  x  x  o  | 1
@@ -454,6 +454,7 @@ uses the following kernel/module command line:
          6  | N64 pad
          7  | Sony PSX controller
          8  | Sony PSX DDR controller
+          9  | SNES mouse
  The exact type of the PSX controller type is autoprobed when used so
 hot swapping should work (but is not recomended).
diff --git a/Documentation/isdn/README.gigaset b/Documentation/isdn/README.gigaset
new file mode 100644
index 000000000000..85a64defd385
--- /dev/null
+++ b/Documentation/isdn/README.gigaset
@@ -0,0 +1,286 @@
+GigaSet 307x Device Driver
+==========================
+1.   Requirements
+     ------------
+1.1. Hardware
+     --------
+     This release supports the connection of the Gigaset 307x/417x family of
+     ISDN DECT bases via Gigaset M101 Data, Gigaset M105 Data or direct USB
+     connection. The following devices are reported to be compatible:
+     307x/417x:
+        Gigaset SX255isdn
+        Gigaset SX353isdn
+        Sinus 45 [AB] isdn (Deutsche Telekom)
+        Sinus 721X/XA
+        Vox Chicago 390 ISDN (KPN Telecom)
+     M101:
+        Sinus 45 Data 1 (Telekom)
+     M105:
+        Gigaset USB Adapter DECT
+        Sinus 45 Data 2 (Telekom)
+        Sinus 721 data
+        Chicago 390 USB (KPN)
+     See also http://www.erbze.info/sinus_gigaset.htm and
+              http://gigaset307x.sourceforge.net/
+     We had also reports from users of Gigaset M105 who could use the drivers
+     with SX 100 and CX 100 ISDN bases (only in unimodem mode, see section 2.4.)
+     If you have another device that works with our driver, please let us know.
+     For example, Gigaset SX205isdn/Sinus 721 X SE and Gigaset SX303isdn bases
+     are just versions without answering machine of models known to work, so
+     they should work just as well; but so far we are lacking positive reports
+     on these.
+     Chances of getting an USB device to work are good if the output of
+        lsusb
+     at the command line contains one of the following:
+        ID 0681:0001
+        ID 0681:0002
+        ID 0681:0009
+        ID 0681:0021
+        ID 0681:0022
+1.2. Software
+     --------
+     The driver works with ISDN4linux and so can be used with any software
+     which is able to use ISDN4linux for ISDN connections (voice or data).
+     CAPI4Linux support is planned but not yet available.
+     There are some user space tools available at
+     http://sourceforge.net/projects/gigaset307x/
+     which provide access to additional device specific functions like SMS,
+     phonebook or call journal.
+2.   How to use the driver
+     ---------------------
+2.1. Modules
+     -------
+     To get the device working, you have to load the proper kernel module. You
+     can do this using
+         modprobe modulename
+     where modulename is usb_gigaset (M105) or bas_gigaset (direct USB
+     connection to the base).
+2.2. Device nodes for user space programs
+     ------------------------------------
+     The device can be accessed from user space (eg. by the user space tools
+     mentioned in 1.2.) through the device nodes:
+     - /dev/ttyGU0 for M105 (USB data boxes)
+     - /dev/ttyGB0 for the base driver (direct USB connection)
+     You can also select a "default device" which is used by the frontends when
+     no device node is given as parameter, by creating a symlink /dev/ttyG to
+     one of them, eg.:
+        ln -s /dev/ttyGB0 /dev/ttyG
+2.3. ISDN4linux
+     ----------
+     This is the "normal" mode of operation. After loading the module you can
+     set up the ISDN system just as you'd do with any ISDN card.
+     Your distribution should provide some configuration utility.
+     If not, you can use some HOWTOs like
+         http://www.linuxhaven.de/dlhp/HOWTO/DE-ISDN-HOWTO-5.html
+     If this doesn't work, because you have some recent device like SX100 where
+     debug output (see section 3.2.) shows something like this when dialing
+         CMD Received: ERROR
+         Available Params: 0
+         Connection State: 0, Response: -1
+         gigaset_process_response: resp_code -1 in ConState 0 !
+         Timeout occurred
+     you might need to use unimodem mode:
+2.4. Unimodem mode
+     -------------
+     This is needed for some devices [e.g. SX100] as they have problems with
+     the "normal" commands.
+     If you have installed the command line tool gigacontr, you can enter
+     unimodem mode using
+         gigacontr --mode unimodem
+     You can switch back using
+         gigacontr --mode isdn
+     You can also load the driver using e.g.
+         modprobe usb_gigaset startmode=0
+     to prevent the driver from starting in "isdn4linux mode".
+     In this mode the device works like a modem connected to a serial port
+     (the /dev/ttyGU0, ... mentioned above) which understands the commands
+         ATZ                 init, reset
+             => OK or ERROR
+         ATD
+         ATDT                dial
+             => OK, CONNECT,
+                BUSY,
+                NO DIAL TONE,
+                NO CARRIER,
+                NO ANSWER
+         <pause>+++<pause>   change to command mode when connected
+         ATH                 hangup
+     You can use some configuration tool of your distribution to configure this
+     "modem" or configure pppd/wvdial manually. There are some example ppp
+     configuration files and chat scripts in the gigaset-VERSION/ppp directory.
+     Please note that the USB drivers are not able to change the state of the
+     control lines (the M105 driver can be configured to use some undocumented
+     control requests, if you really need the control lines, though). This means
+     you must use "Stupid Mode" if you are using wvdial or you should use the
+     nocrtscts option of pppd.
+     You must also assure that the ppp_async module is loaded with the parameter
+     flag_time=0. You can do this e.g. by adding a line like
+        options ppp_async flag_time=0
+     to /etc/modprobe.conf. If your distribution has some local module
+     configuration file like /etc/modprobe.conf.local,
+     using that should be preferred.
+2.5. Call-ID (CID) mode
+     ------------------
+     Call-IDs are numbers used to tag commands to, and responses from, the
+     Gigaset base in order to support the simultaneous handling of multiple
+     ISDN calls. Their use can be enabled ("CID mode") or disabled ("Unimodem
+     mode"). Without Call-IDs (in Unimodem mode), only a very limited set of
+     functions is available. It allows outgoing data connections only, but
+     does not signal incoming calls or other base events.
+     DECT cordless data devices (M10x) permanently occupy the cordless
+     connection to the base while Call-IDs are activated. As the Gigaset
+     bases only support one DECT data connection at a time, this prevents
+     other DECT cordless data devices from accessing the base.
+     During active operation, the driver switches to the necessary mode
+     automatically. However, for the reasons above, the mode chosen when
+     the device is not in use (idle) can be selected by the user.
+     - If you want to receive incoming calls, you can use the default
+       settings (CID mode).
+     - If you have several DECT data devices (M10x) which you want to use
+       in turn, select Unimodem mode by passing the parameter "cidmode=0" to
+       the driver ("modprobe usb_gigaset cidmode=0" or modprobe.conf).
+     If you want both of these at once, you are out of luck.
+     You can also use /sys/module/<name>/parameters/cidmode for changing
+     the CID mode setting (<name> is usb_gigaset or bas_gigaset).
+3.   Troubleshooting
+     ---------------
+3.1. Solutions to frequently reported problems
+     -----------------------------------------
+     Problem:
+        You have a slow provider and isdn4linux gives up dialing too early.
+     Solution:
+        Load the isdn module using the dialtimeout option. You can do this e.g.
+        by adding a line like
+           options isdn dialtimeout=15
+        to /etc/modprobe.conf. If your distribution has some local module
+        configuration file like /etc/modprobe.conf.local,
+        using that should be preferred.
+     Problem:
+        Your isdn script aborts with a message about isdnlog.
+     Solution:
+        Try deactivating (or commenting out) isdnlog. This driver does not
+        support it.
+     Problem:
+        You have two or more DECT data adapters (M101/M105) and only the
+        first one you turn on works.
+     Solution:
+        Select Unimodem mode for all DECT data adapters. (see section 2.4.)
+3.2. Telling the driver to provide more information
+     ----------------------------------------------
+     Building the driver with the "Gigaset debugging" kernel configuration
+     option (CONFIG_GIGASET_DEBUG) gives it the ability to produce additional
+     information useful for debugging.
+     You can control the amount of debugging information the driver produces by
+     writing an appropriate value to /sys/module/gigaset/parameters/debug, e.g.
+        echo 0 > /sys/module/gigaset/parameters/debug
+     switches off debugging output completely,
+        echo 0x10a020 > /sys/module/gigaset/parameters/debug
+     enables the standard set of debugging output messages. These values are
+     bit patterns where every bit controls a certain type of debugging output.
+     See the constants DEBUG_* in the source file gigaset.h for details.
+     The initial value can be set using the debug parameter when loading the
+     module "gigaset", e.g. by adding a line
+        options gigaset debug=0
+     to /etc/modprobe.conf, ...
+     Generated debugging information can be found
+     - as output of the command
+         dmesg
+     - in system log files written by your syslog daemon, usually
+       in /var/log/, e.g. /var/log/messages.
+3.3. Reporting problems and bugs
+     ---------------------------
+     If you can't solve problems with the driver on your own, feel free to
+     use one of the forums, bug trackers, or mailing lists on
+         http://sourceforge.net/projects/gigaset307x
+     or write an electronic mail to the maintainers.
+     Try to provide as much information as possible, such as
+     - distribution
+     - kernel version (uname -r)
+     - gcc version (gcc --version)
+     - hardware architecture (uname -m, ...)
+     - type and firmware version of your device (base and wireless module,
+       if any)
+     - output of "lsusb -v" (if using an USB device)
+     - error messages
+     - relevant system log messages (it would help if you activate debug
+       output as described in 3.2.)
+     For help with general configuration problems not specific to our driver,
+     such as isdn4linux and network configuration issues, please refer to the
+     appropriate forums and newsgroups.
+3.4. Reporting problem solutions
+     ---------------------------
+     If you solved a problem with our drivers, wrote startup scripts for your
+     distribution, ... feel free to contact us (using one of the places
+     mentioned in 3.3.). We'd like to add scripts, hints, documentation
+     to the driver and/or the project web page.
+4.   Links, other software
+     ---------------------
+     - Sourceforge project developing this driver and associated tools
+         http://sourceforge.net/projects/gigaset307x
+     - Yahoo! Group on the Siemens Gigaset family of devices
+         http://de.groups.yahoo.com/group/Siemens-Gigaset
+     - Siemens Gigaset/T-Sinus compatibility table
+         http://www.erbze.info/sinus_gigaset.htm
+5.   Credits
+     -------
+     Thanks to
+     Karsten Keil
+        for his help with isdn4linux
+     Deti Fliegl
+        for his base driver code
+     Dennis Dietrich
+        for his kernel 2.6 patches
+     Andreas Rummel
+        for his work and logs to get unimodem mode working
+     Andreas Degert
+        for his logs and patches to get cx 100 working
+     Dietrich Feist
+        for his generous donation of one M105 and two M101 cordless adapters
+     Christoph Schweers
+        for his generous donation of a M34 device
+     and all the other people who sent logs and other information.
diff --git a/Documentation/kbuild/modules.txt b/Documentation/kbuild/modules.txt
index fcccf2432f98..61fc079eb966 100644
--- a/Documentation/kbuild/modules.txt
+++ b/Documentation/kbuild/modules.txt
@@ -44,7 +44,7 @@ What is covered within this file is mainly information to authors
 of modules. The author of an external modules should supply
 a makefile that hides most of the complexity so one only has to type
 'make' to build the module. A complete example will be present in
-chapter �. Creating a kbuild file for an external module".
+chapter 4, "Creating a kbuild file for an external module".
 === 2. How to build external modules
diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index 78eb66b8619c..a853e8607865 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -1,4 +1,4 @@
-February 2003             Kernel Parameters                     v2.5.59
+                          Kernel Parameters
                          ~~~~~~~~~~~~~~~~~
 The following is a consolidated list of the kernel parameters as implemented
@@ -17,9 +17,17 @@ are specified on the kernel command line with the module name plus
        usbcore.blinkenlights=1
-The text in square brackets at the beginning of the description states the
+This document may not be entirely up to date and comprehensive. The command
-restrictions on the kernel for the said kernel parameter to be valid. The
+"modinfo -p ${modulename}" shows a current list of all parameters of a loadable
-restrictions referred to are that the relevant option is valid if:
+module. Loadable modules, after being loaded into the running kernel, also
+reveal their parameters in /sys/module/${modulename}/parameters/. Some of these
+parameters may be changed at runtime by the command
+"echo -n ${value} > /sys/module/${modulename}/parameters/${parm}".
+The parameters listed below are only valid if certain kernel build options were
+enabled and if respective hardware is present. The text in square brackets at
+the beginning of each description states the restrictions within which a
+parameter is applicable:
        ACPI    ACPI support is enabled.
        ALSA    ALSA sound support is enabled.
@@ -1049,10 +1057,10 @@ running once the system is up.
        noltlbs         [PPC] Do not use large page/tlb entries for kernel
                        lowmem mapping on PPC40x.
-        nomce           [IA-32] Machine Check Exception
        nomca           [IA-64] Disable machine check abort handling
+        nomce           [IA-32] Machine Check Exception
        noresidual      [PPC] Don't use residual data on PReP machines.
        noresume        [SWSUSP] Disables resume and restores original swap
@@ -1685,20 +1693,6 @@ running once the system is up.
 ______________________________________________________________________
-Changelog:
-2000-06-??      Mr. Unknown
-        The last known update (for 2.4.0) - the changelog was not kept before.
-2002-11-24      Petr Baudis <pasky@ucw.cz>
-                Randy Dunlap <randy.dunlap@verizon.net>
-        Update for 2.5.49, description for most of the options introduced,
-        references to other documentation (C files, READMEs, ..), added S390,
-        PPC, SPARC, MTD, ALSA and OSS category. Minor corrections and
-        reformatting.
-2005-10-19      Randy Dunlap <rdunlap@xenotime.net>
-        Lots of typos, whitespace, some reformatting.
 TODO:
diff --git a/Documentation/laptop-mode.txt b/Documentation/laptop-mode.txt
index b18e21675906..5696e879449b 100644
--- a/Documentation/laptop-mode.txt
+++ b/Documentation/laptop-mode.txt
@@ -919,11 +919,11 @@ int main(int argc, char **argv)
    int settle_time = 60;
    /* Parse the simple command-line */
-    if (ac == 2)
+    if (argc == 2)
-        disk = av[1];
+        disk = argv[1];
-    else if (ac == 4) {
+    else if (argc == 4) {
-        settle_time = atoi(av[2]);
+        settle_time = atoi(argv[2]);
-        disk = av[3];
+        disk = argv[3];
    } else
        usage();
diff --git a/Documentation/leds-class.txt b/Documentation/leds-class.txt
new file mode 100644
index 000000000000..8c35c0426110
--- /dev/null
+++ b/Documentation/leds-class.txt
@@ -0,0 +1,71 @@
+LED handling under Linux
+========================
+If you're reading this and thinking about keyboard leds, these are
+handled by the input subsystem and the led class is *not* needed.
+In its simplest form, the LED class just allows control of LEDs from
+userspace. LEDs appear in /sys/class/leds/. The brightness file will
+set the brightness of the LED (taking a value 0-255). Most LEDs don't
+have hardware brightness support so will just be turned on for non-zero
+brightness settings.
+The class also introduces the optional concept of an LED trigger. A trigger
+is a kernel based source of led events. Triggers can either be simple or
+complex. A simple trigger isn't configurable and is designed to slot into
+existing subsystems with minimal additional code. Examples are the ide-disk,
+nand-disk and sharpsl-charge triggers. With led triggers disabled, the code
+optimises away.
+Complex triggers whilst available to all LEDs have LED specific
+parameters and work on a per LED basis. The timer trigger is an example.
+You can change triggers in a similar manner to the way an IO scheduler
+is chosen (via /sys/class/leds/<device>/trigger). Trigger specific
+parameters can appear in /sys/class/leds/<device> once a given trigger is
+selected.
+Design Philosophy
+=================
+The underlying design philosophy is simplicity. LEDs are simple devices
+and the aim is to keep a small amount of code giving as much functionality
+as possible.  Please keep this in mind when suggesting enhancements.
+LED Device Naming
+=================
+Is currently of the form:
+"devicename:colour"
+There have been calls for LED properties such as colour to be exported as
+individual led class attributes. As a solution which doesn't incur as much
+overhead, I suggest these become part of the device name. The naming scheme
+above leaves scope for further attributes should they be needed.
+Known Issues
+============
+The LED Trigger core cannot be a module as the simple trigger functions
+would cause nightmare dependency issues. I see this as a minor issue
+compared to the benefits the simple trigger functionality brings. The
+rest of the LED subsystem can be modular.
+Some leds can be programmed to flash in hardware. As this isn't a generic
+LED device property, this should be exported as a device specific sysfs
+attribute rather than part of the class if this functionality is required.
+Future Development
+==================
+At the moment, a trigger can't be created specifically for a single LED.
+There are a number of cases where a trigger might only be mappable to a
+particular LED (ACPI?). The addition of triggers provided by the LED driver
+should cover this option and be possible to add without breaking the
+current interface.
diff --git a/Documentation/memory-barriers.txt b/Documentation/memory-barriers.txt
new file mode 100644
index 000000000000..4710845dbac4
--- /dev/null
+++ b/Documentation/memory-barriers.txt
@@ -0,0 +1,2133 @@
+                         ============================
+                         LINUX KERNEL MEMORY BARRIERS
+                         ============================
+By: David Howells <dhowells@redhat.com>
+Contents:
+ (*) Abstract memory access model.
+     - Device operations.
+     - Guarantees.
+ (*) What are memory barriers?
+     - Varieties of memory barrier.
+     - What may not be assumed about memory barriers?
+     - Data dependency barriers.
+     - Control dependencies.
+     - SMP barrier pairing.
+     - Examples of memory barrier sequences.
+     - Read memory barriers vs load speculation.
+ (*) Explicit kernel barriers.
+     - Compiler barrier.
+     - The CPU memory barriers.
+     - MMIO write barrier.
+ (*) Implicit kernel memory barriers.
+     - Locking functions.
+     - Interrupt disabling functions.
+     - Miscellaneous functions.
+ (*) Inter-CPU locking barrier effects.
+     - Locks vs memory accesses.
+     - Locks vs I/O accesses.
+ (*) Where are memory barriers needed?
+     - Interprocessor interaction.
+     - Atomic operations.
+     - Accessing devices.
+     - Interrupts.
+ (*) Kernel I/O barrier effects.
+ (*) Assumed minimum execution ordering model.
+ (*) The effects of the cpu cache.
+     - Cache coherency.
+     - Cache coherency vs DMA.
+     - Cache coherency vs MMIO.
+ (*) The things CPUs get up to.
+     - And then there's the Alpha.
+ (*) References.
+============================
+ABSTRACT MEMORY ACCESS MODEL
+============================
+Consider the following abstract model of the system:
+                            :                :
+                            :                :
+                            :                :
+                +-------+   :   +--------+   :   +-------+
+                |       |   :   |        |   :   |       |
+                |       |   :   |        |   :   |       |
+                | CPU 1 |<----->| Memory |<----->| CPU 2 |
+                |       |   :   |        |   :   |       |
+                |       |   :   |        |   :   |       |
+                +-------+   :   +--------+   :   +-------+
+                    ^       :       ^        :       ^
+                    |       :       |        :       |
+                    |       :       |        :       |
+                    |       :       v        :       |
+                    |       :   +--------+   :       |
+                    |       :   |        |   :       |
+                    |       :   |        |   :       |
+                    +---------->| Device |<----------+
+                            :   |        |   :
+                            :   |        |   :
+                            :   +--------+   :
+                            :                :
+Each CPU executes a program that generates memory access operations.  In the
+abstract CPU, memory operation ordering is very relaxed, and a CPU may actually
+perform the memory operations in any order it likes, provided program causality
+appears to be maintained.  Similarly, the compiler may also arrange the
+instructions it emits in any order it likes, provided it doesn't affect the
+apparent operation of the program.
+So in the above diagram, the effects of the memory operations performed by a
+CPU are perceived by the rest of the system as the operations cross the
+interface between the CPU and rest of the system (the dotted lines).
+For example, consider the following sequence of events:
+        CPU 1           CPU 2
+        =============== ===============
+        { A == 1; B == 2 }
+        A = 3;          x = A;
+        B = 4;          y = B;
+The set of accesses as seen by the memory system in the middle can be arranged
+in 24 different combinations:
+        STORE A=3,      STORE B=4,      x=LOAD A->3,    y=LOAD B->4
+        STORE A=3,      STORE B=4,      y=LOAD B->4,    x=LOAD A->3
+        STORE A=3,      x=LOAD A->3,    STORE B=4,      y=LOAD B->4
+        STORE A=3,      x=LOAD A->3,    y=LOAD B->2,    STORE B=4
+        STORE A=3,      y=LOAD B->2,    STORE B=4,      x=LOAD A->3
+        STORE A=3,      y=LOAD B->2,    x=LOAD A->3,    STORE B=4
+        STORE B=4,      STORE A=3,      x=LOAD A->3,    y=LOAD B->4
+        STORE B=4, ...
+        ...
+and can thus result in four different combinations of values:
+        x == 1, y == 2
+        x == 1, y == 4
+        x == 3, y == 2
+        x == 3, y == 4
+Furthermore, the stores committed by a CPU to the memory system may not be
+perceived by the loads made by another CPU in the same order as the stores were
+committed.
+As a further example, consider this sequence of events:
+        CPU 1           CPU 2
+        =============== ===============
+        { A == 1, B == 2, C = 3, P == &A, Q == &C }
+        B = 4;          Q = P;
+        P = &B          D = *Q;
+There is an obvious data dependency here, as the value loaded into D depends on
+the address retrieved from P by CPU 2.  At the end of the sequence, any of the
+following results are possible:
+        (Q == &A) and (D == 1)
+        (Q == &B) and (D == 2)
+        (Q == &B) and (D == 4)
+Note that CPU 2 will never try and load C into D because the CPU will load P
+into Q before issuing the load of *Q.
+DEVICE OPERATIONS
+-----------------
+Some devices present their control interfaces as collections of memory
+locations, but the order in which the control registers are accessed is very
+important.  For instance, imagine an ethernet card with a set of internal
+registers that are accessed through an address port register (A) and a data
+port register (D).  To read internal register 5, the following code might then
+be used:
+        *A = 5;
+        x = *D;
+but this might show up as either of the following two sequences:
+        STORE *A = 5, x = LOAD *D
+        x = LOAD *D, STORE *A = 5
+the second of which will almost certainly result in a malfunction, since it set
+the address _after_ attempting to read the register.
+GUARANTEES
+----------
+There are some minimal guarantees that may be expected of a CPU:
+ (*) On any given CPU, dependent memory accesses will be issued in order, with
+     respect to itself.  This means that for:
+        Q = P; D = *Q;
+     the CPU will issue the following memory operations:
+        Q = LOAD P, D = LOAD *Q
+     and always in that order.
+ (*) Overlapping loads and stores within a particular CPU will appear to be
+     ordered within that CPU.  This means that for:
+        a = *X; *X = b;
+     the CPU will only issue the following sequence of memory operations:
+        a = LOAD *X, STORE *X = b
+     And for:
+        *X = c; d = *X;
+     the CPU will only issue:
+        STORE *X = c, d = LOAD *X
+     (Loads and stores overlap if they are targetted at overlapping pieces of
+     memory).
+And there are a number of things that _must_ or _must_not_ be assumed:
+ (*) It _must_not_ be assumed that independent loads and stores will be issued
+     in the order given.  This means that for:
+        X = *A; Y = *B; *D = Z;
+     we may get any of the following sequences:
+        X = LOAD *A,  Y = LOAD *B,  STORE *D = Z
+        X = LOAD *A,  STORE *D = Z, Y = LOAD *B
+        Y = LOAD *B,  X = LOAD *A,  STORE *D = Z
+        Y = LOAD *B,  STORE *D = Z, X = LOAD *A
+        STORE *D = Z, X = LOAD *A,  Y = LOAD *B
+        STORE *D = Z, Y = LOAD *B,  X = LOAD *A
+ (*) It _must_ be assumed that overlapping memory accesses may be merged or
+     discarded.  This means that for:
+        X = *A; Y = *(A + 4);
+     we may get any one of the following sequences:
+        X = LOAD *A; Y = LOAD *(A + 4);
+        Y = LOAD *(A + 4); X = LOAD *A;
+        {X, Y} = LOAD {*A, *(A + 4) };
+     And for:
+        *A = X; Y = *A;
+     we may get either of:
+        STORE *A = X; Y = LOAD *A;
+        STORE *A = Y = X;
+=========================
+WHAT ARE MEMORY BARRIERS?
+=========================
+As can be seen above, independent memory operations are effectively performed
+in random order, but this can be a problem for CPU-CPU interaction and for I/O.
+What is required is some way of intervening to instruct the compiler and the
+CPU to restrict the order.
+Memory barriers are such interventions.  They impose a perceived partial
+ordering between the memory operations specified on either side of the barrier.
+They request that the sequence of memory events generated appears to other
+parts of the system as if the barrier is effective on that CPU.
+VARIETIES OF MEMORY BARRIER
+---------------------------
+Memory barriers come in four basic varieties:
+ (1) Write (or store) memory barriers.
+     A write memory barrier gives a guarantee that all the STORE operations
+     specified before the barrier will appear to happen before all the STORE
+     operations specified after the barrier with respect to the other
+     components of the system.
+     A write barrier is a partial ordering on stores only; it is not required
+     to have any effect on loads.
+     A CPU can be viewed as as commiting a sequence of store operations to the
+     memory system as time progresses.  All stores before a write barrier will
+     occur in the sequence _before_ all the stores after the write barrier.
+     [!] Note that write barriers should normally be paired with read or data
+     dependency barriers; see the "SMP barrier pairing" subsection.
+ (2) Data dependency barriers.
+     A data dependency barrier is a weaker form of read barrier.  In the case
+     where two loads are performed such that the second depends on the result
+     of the first (eg: the first load retrieves the address to which the second
+     load will be directed), a data dependency barrier would be required to
+     make sure that the target of the second load is updated before the address
+     obtained by the first load is accessed.
+     A data dependency barrier is a partial ordering on interdependent loads
+     only; it is not required to have any effect on stores, independent loads
+     or overlapping loads.
+     As mentioned in (1), the other CPUs in the system can be viewed as
+     committing sequences of stores to the memory system that the CPU being
+     considered can then perceive.  A data dependency barrier issued by the CPU
+     under consideration guarantees that for any load preceding it, if that
+     load touches one of a sequence of stores from another CPU, then by the
+     time the barrier completes, the effects of all the stores prior to that
+     touched by the load will be perceptible to any loads issued after the data
+     dependency barrier.
+     See the "Examples of memory barrier sequences" subsection for diagrams
+     showing the ordering constraints.
+     [!] Note that the first load really has to have a _data_ dependency and
+     not a control dependency.  If the address for the second load is dependent
+     on the first load, but the dependency is through a conditional rather than
+     actually loading the address itself, then it's a _control_ dependency and
+     a full read barrier or better is required.  See the "Control dependencies"
+     subsection for more information.
+     [!] Note that data dependency barriers should normally be paired with
+     write barriers; see the "SMP barrier pairing" subsection.
+ (3) Read (or load) memory barriers.
+     A read barrier is a data dependency barrier plus a guarantee that all the
+     LOAD operations specified before the barrier will appear to happen before
+     all the LOAD operations specified after the barrier with respect to the
+     other components of the system.
+     A read barrier is a partial ordering on loads only; it is not required to
+     have any effect on stores.
+     Read memory barriers imply data dependency barriers, and so can substitute
+     for them.
+     [!] Note that read barriers should normally be paired with write barriers;
+     see the "SMP barrier pairing" subsection.
+ (4) General memory barriers.
+     A general memory barrier gives a guarantee that all the LOAD and STORE
+     operations specified before the barrier will appear to happen before all
+     the LOAD and STORE operations specified after the barrier with respect to
+     the other components of the system.
+     A general memory barrier is a partial ordering over both loads and stores.
+     General memory barriers imply both read and write memory barriers, and so
+     can substitute for either.
+And a couple of implicit varieties:
+ (5) LOCK operations.
+     This acts as a one-way permeable barrier.  It guarantees that all memory
+     operations after the LOCK operation will appear to happen after the LOCK
+     operation with respect to the other components of the system.
+     Memory operations that occur before a LOCK operation may appear to happen
+     after it completes.
+     A LOCK operation should almost always be paired with an UNLOCK operation.
+ (6) UNLOCK operations.
+     This also acts as a one-way permeable barrier.  It guarantees that all
+     memory operations before the UNLOCK operation will appear to happen before
+     the UNLOCK operation with respect to the other components of the system.
+     Memory operations that occur after an UNLOCK operation may appear to
+     happen before it completes.
+     LOCK and UNLOCK operations are guaranteed to appear with respect to each
+     other strictly in the order specified.
+     The use of LOCK and UNLOCK operations generally precludes the need for
+     other sorts of memory barrier (but note the exceptions mentioned in the
+     subsection "MMIO write barrier").
+Memory barriers are only required where there's a possibility of interaction
+between two CPUs or between a CPU and a device.  If it can be guaranteed that
+there won't be any such interaction in any particular piece of code, then
+memory barriers are unnecessary in that piece of code.
+Note that these are the _minimum_ guarantees.  Different architectures may give
+more substantial guarantees, but they may _not_ be relied upon outside of arch
+specific code.
+WHAT MAY NOT BE ASSUMED ABOUT MEMORY BARRIERS?
+----------------------------------------------
+There are certain things that the Linux kernel memory barriers do not guarantee:
+ (*) There is no guarantee that any of the memory accesses specified before a
+     memory barrier will be _complete_ by the completion of a memory barrier
+     instruction; the barrier can be considered to draw a line in that CPU's
+     access queue that accesses of the appropriate type may not cross.
+ (*) There is no guarantee that issuing a memory barrier on one CPU will have
+     any direct effect on another CPU or any other hardware in the system.  The
+     indirect effect will be the order in which the second CPU sees the effects
+     of the first CPU's accesses occur, but see the next point:
+ (*) There is no guarantee that the a CPU will see the correct order of effects
+     from a second CPU's accesses, even _if_ the second CPU uses a memory
+     barrier, unless the first CPU _also_ uses a matching memory barrier (see
+     the subsection on "SMP Barrier Pairing").
+ (*) There is no guarantee that some intervening piece of off-the-CPU
+     hardware[*] will not reorder the memory accesses.  CPU cache coherency
+     mechanisms should propagate the indirect effects of a memory barrier
+     between CPUs, but might not do so in order.
+        [*] For information on bus mastering DMA and coherency please read:
+            Documentation/pci.txt
+            Documentation/DMA-mapping.txt
+            Documentation/DMA-API.txt
+DATA DEPENDENCY BARRIERS
+------------------------
+The usage requirements of data dependency barriers are a little subtle, and
+it's not always obvious that they're needed.  To illustrate, consider the
+following sequence of events:
+        CPU 1           CPU 2
+        =============== ===============
+        { A == 1, B == 2, C = 3, P == &A, Q == &C }
+        B = 4;
+        <write barrier>
+        P = &B
+                        Q = P;
+                        D = *Q;
+There's a clear data dependency here, and it would seem that by the end of the
+sequence, Q must be either &A or &B, and that:
+        (Q == &A) implies (D == 1)
+        (Q == &B) implies (D == 4)
+But! CPU 2's perception of P may be updated _before_ its perception of B, thus
+leading to the following situation:
+        (Q == &B) and (D == 2) ????
+Whilst this may seem like a failure of coherency or causality maintenance, it
+isn't, and this behaviour can be observed on certain real CPUs (such as the DEC
+Alpha).
+To deal with this, a data dependency barrier must be inserted between the
+address load and the data load:
+        CPU 1           CPU 2
+        =============== ===============
+        { A == 1, B == 2, C = 3, P == &A, Q == &C }
+        B = 4;
+        <write barrier>
+        P = &B
+                        Q = P;
+                        <data dependency barrier>
+                        D = *Q;
+This enforces the occurrence of one of the two implications, and prevents the
+third possibility from arising.
+[!] Note that this extremely counterintuitive situation arises most easily on
+machines with split caches, so that, for example, one cache bank processes
+even-numbered cache lines and the other bank processes odd-numbered cache
+lines.  The pointer P might be stored in an odd-numbered cache line, and the
+variable B might be stored in an even-numbered cache line.  Then, if the
+even-numbered bank of the reading CPU's cache is extremely busy while the
+odd-numbered bank is idle, one can see the new value of the pointer P (&B),
+but the old value of the variable B (1).
+Another example of where data dependency barriers might by required is where a
+number is read from memory and then used to calculate the index for an array
+access:
+        CPU 1           CPU 2
+        =============== ===============
+        { M[0] == 1, M[1] == 2, M[3] = 3, P == 0, Q == 3 }
+        M[1] = 4;
+        <write barrier>
+        P = 1
+                        Q = P;
+                        <data dependency barrier>
+                        D = M[Q];
+The data dependency barrier is very important to the RCU system, for example.
+See rcu_dereference() in include/linux/rcupdate.h.  This permits the current
+target of an RCU'd pointer to be replaced with a new modified target, without
+the replacement target appearing to be incompletely initialised.
+See also the subsection on "Cache Coherency" for a more thorough example.
+CONTROL DEPENDENCIES
+--------------------
+A control dependency requires a full read memory barrier, not simply a data
+dependency barrier to make it work correctly.  Consider the following bit of
+code:
+        q = &a;
+        if (p)
+                q = &b;
+        <data dependency barrier>
+        x = *q;
+This will not have the desired effect because there is no actual data
+dependency, but rather a control dependency that the CPU may short-circuit by
+attempting to predict the outcome in advance.  In such a case what's actually
+required is:
+        q = &a;
+        if (p)
+                q = &b;
+        <read barrier>
+        x = *q;
+SMP BARRIER PAIRING
+-------------------
+When dealing with CPU-CPU interactions, certain types of memory barrier should
+always be paired.  A lack of appropriate pairing is almost certainly an error.
+A write barrier should always be paired with a data dependency barrier or read
+barrier, though a general barrier would also be viable.  Similarly a read
+barrier or a data dependency barrier should always be paired with at least an
+write barrier, though, again, a general barrier is viable:
+        CPU 1           CPU 2
+        =============== ===============
+        a = 1;
+        <write barrier>
+        b = 2;          x = b;
+                        <read barrier>
+                        y = a;
+Or:
+        CPU 1           CPU 2
+        =============== ===============================
+        a = 1;
+        <write barrier>
+        b = &a;         x = b;
+                        <data dependency barrier>
+                        y = *x;
+Basically, the read barrier always has to be there, even though it can be of
+the "weaker" type.
+[!] Note that the stores before the write barrier would normally be expected to
+match the loads after the read barrier or data dependency barrier, and vice
+versa:
+        CPU 1                           CPU 2
+        ===============                 ===============
+        a = 1;           }----   --->{  v = c
+        b = 2;           }    \ /    {  w = d
+        <write barrier>        \        <read barrier>
+        c = 3;           }    / \    {  x = a;
+        d = 4;           }----   --->{  y = b;
+EXAMPLES OF MEMORY BARRIER SEQUENCES
+------------------------------------
+Firstly, write barriers act as a partial orderings on store operations.
+Consider the following sequence of events:
+        CPU 1
+        =======================
+        STORE A = 1
+        STORE B = 2
+        STORE C = 3
+        <write barrier>
+        STORE D = 4
+        STORE E = 5
+This sequence of events is committed to the memory coherence system in an order
+that the rest of the system might perceive as the unordered set of { STORE A,
+STORE B, STORE C } all occuring before the unordered set of { STORE D, STORE E
+}:
+        +-------+       :      :
+        |       |       +------+
+        |       |------>| C=3  |     }     /\
+        |       |  :    +------+     }-----  \  -----> Events perceptible
+        |       |  :    | A=1  |     }        \/       to rest of system
+        |       |  :    +------+     }
+        | CPU 1 |  :    | B=2  |     }
+        |       |       +------+     }
+        |       |   wwwwwwwwwwwwwwww }   <--- At this point the write barrier
+        |       |       +------+     }        requires all stores prior to the
+        |       |  :    | E=5  |     }        barrier to be committed before
+        |       |  :    +------+     }        further stores may be take place.
+        |       |------>| D=4  |     }
+        |       |       +------+
+        +-------+       :      :
+                           |
+                           | Sequence in which stores are committed to the
+                           | memory system by CPU 1
+                           V
+Secondly, data dependency barriers act as a partial orderings on data-dependent
+loads.  Consider the following sequence of events:
+        CPU 1                   CPU 2
+        ======================= =======================
+                { B = 7; X = 9; Y = 8; C = &Y }
+        STORE A = 1
+        STORE B = 2
+        <write barrier>
+        STORE C = &B            LOAD X
+        STORE D = 4             LOAD C (gets &B)
+                                LOAD *C (reads B)
+Without intervention, CPU 2 may perceive the events on CPU 1 in some
+effectively random order, despite the write barrier issued by CPU 1:
+        +-------+       :      :                :       :
+        |       |       +------+                +-------+  | Sequence of update
+        |       |------>| B=2  |-----       --->| Y->8  |  | of perception on
+        |       |  :    +------+     \          +-------+  | CPU 2
+        | CPU 1 |  :    | A=1  |      \     --->| C->&Y |  V
+        |       |       +------+       |        +-------+
+        |       |   wwwwwwwwwwwwwwww   |        :       :
+        |       |       +------+       |        :       :
+        |       |  :    | C=&B |---    |        :       :       +-------+
+        |       |  :    +------+   \   |        +-------+       |       |
+        |       |------>| D=4  |    ----------->| C->&B |------>|       |
+        |       |       +------+       |        +-------+       |       |
+        +-------+       :      :       |        :       :       |       |
+                                       |        :       :       |       |
+                                       |        :       :       | CPU 2 |
+                                       |        +-------+       |       |
+            Apparently incorrect --->  |        | B->7  |------>|       |
+            perception of B (!)        |        +-------+       |       |
+                                       |        :       :       |       |
+                                       |        +-------+       |       |
+            The load of X holds --->    \       | X->9  |------>|       |
+            up the maintenance           \      +-------+       |       |
+            of coherence of B             ----->| B->2  |       +-------+
+                                                +-------+
+                                                :       :
+In the above example, CPU 2 perceives that B is 7, despite the load of *C
+(which would be B) coming after the the LOAD of C.
+If, however, a data dependency barrier were to be placed between the load of C
+and the load of *C (ie: B) on CPU 2:
+        CPU 1                   CPU 2
+        ======================= =======================
+                { B = 7; X = 9; Y = 8; C = &Y }
+        STORE A = 1
+        STORE B = 2
+        <write barrier>
+        STORE C = &B            LOAD X
+        STORE D = 4             LOAD C (gets &B)
+                                <data dependency barrier>
+                                LOAD *C (reads B)
+then the following will occur:
+        +-------+       :      :                :       :
+        |       |       +------+                +-------+
+        |       |------>| B=2  |-----       --->| Y->8  |
+        |       |  :    +------+     \          +-------+
+        | CPU 1 |  :    | A=1  |      \     --->| C->&Y |
+        |       |       +------+       |        +-------+
+        |       |   wwwwwwwwwwwwwwww   |        :       :
+        |       |       +------+       |        :       :
+        |       |  :    | C=&B |---    |        :       :       +-------+
+        |       |  :    +------+   \   |        +-------+       |       |
+        |       |------>| D=4  |    ----------->| C->&B |------>|       |
+        |       |       +------+       |        +-------+       |       |
+        +-------+       :      :       |        :       :       |       |
+                                       |        :       :       |       |
+                                       |        :       :       | CPU 2 |
+                                       |        +-------+       |       |
+                                       |        | X->9  |------>|       |
+                                       |        +-------+       |       |
+          Makes sure all effects --->   \   ddddddddddddddddd   |       |
+          prior to the store of C        \      +-------+       |       |
+          are perceptible to              ----->| B->2  |------>|       |
+          subsequent loads                      +-------+       |       |
+                                                :       :       +-------+
+And thirdly, a read barrier acts as a partial order on loads.  Consider the
+following sequence of events:
+        CPU 1                   CPU 2
+        ======================= =======================
+                { A = 0, B = 9 }
+        STORE A=1
+        <write barrier>
+        STORE B=2
+                                LOAD B
+                                LOAD A
+Without intervention, CPU 2 may then choose to perceive the events on CPU 1 in
+some effectively random order, despite the write barrier issued by CPU 1:
+        +-------+       :      :                :       :
+        |       |       +------+                +-------+
+        |       |------>| A=1  |------      --->| A->0  |
+        |       |       +------+      \         +-------+
+        | CPU 1 |   wwwwwwwwwwwwwwww   \    --->| B->9  |
+        |       |       +------+        |       +-------+
+        |       |------>| B=2  |---     |       :       :
+        |       |       +------+   \    |       :       :       +-------+
+        +-------+       :      :    \   |       +-------+       |       |
+                                     ---------->| B->2  |------>|       |
+                                        |       +-------+       | CPU 2 |
+                                        |       | A->0  |------>|       |
+                                        |       +-------+       |       |
+                                        |       :       :       +-------+
+                                         \      :       :
+                                          \     +-------+
+                                           ---->| A->1  |
+                                                +-------+
+                                                :       :
+If, however, a read barrier were to be placed between the load of E and the
+load of A on CPU 2:
+        CPU 1                   CPU 2
+        ======================= =======================
+                { A = 0, B = 9 }
+        STORE A=1
+        <write barrier>
+        STORE B=2
+                                LOAD B
+                                <read barrier>
+                                LOAD A
+then the partial ordering imposed by CPU 1 will be perceived correctly by CPU
+2:
+        +-------+       :      :                :       :
+        |       |       +------+                +-------+
+        |       |------>| A=1  |------      --->| A->0  |
+        |       |       +------+      \         +-------+
+        | CPU 1 |   wwwwwwwwwwwwwwww   \    --->| B->9  |
+        |       |       +------+        |       +-------+
+        |       |------>| B=2  |---     |       :       :
+        |       |       +------+   \    |       :       :       +-------+
+        +-------+       :      :    \   |       +-------+       |       |
+                                     ---------->| B->2  |------>|       |
+                                        |       +-------+       | CPU 2 |
+                                        |       :       :       |       |
+                                        |       :       :       |       |
+          At this point the read ---->   \  rrrrrrrrrrrrrrrrr   |       |
+          barrier causes all effects      \     +-------+       |       |
+          prior to the storage of B        ---->| A->1  |------>|       |
+          to be perceptible to CPU 2            +-------+       |       |
+                                                :       :       +-------+
+To illustrate this more completely, consider what could happen if the code
+contained a load of A either side of the read barrier:
+        CPU 1                   CPU 2
+        ======================= =======================
+                { A = 0, B = 9 }
+        STORE A=1
+        <write barrier>
+        STORE B=2
+                                LOAD B
+                                LOAD A [first load of A]
+                                <read barrier>
+                                LOAD A [second load of A]
+Even though the two loads of A both occur after the load of B, they may both
+come up with different values:
+        +-------+       :      :                :       :
+        |       |       +------+                +-------+
+        |       |------>| A=1  |------      --->| A->0  |
+        |       |       +------+      \         +-------+
+        | CPU 1 |   wwwwwwwwwwwwwwww   \    --->| B->9  |
+        |       |       +------+        |       +-------+
+        |       |------>| B=2  |---     |       :       :
+        |       |       +------+   \    |       :       :       +-------+
+        +-------+       :      :    \   |       +-------+       |       |
+                                     ---------->| B->2  |------>|       |
+                                        |       +-------+       | CPU 2 |
+                                        |       :       :       |       |
+                                        |       :       :       |       |
+                                        |       +-------+       |       |
+                                        |       | A->0  |------>| 1st   |
+                                        |       +-------+       |       |
+          At this point the read ---->   \  rrrrrrrrrrrrrrrrr   |       |
+          barrier causes all effects      \     +-------+       |       |
+          prior to the storage of B        ---->| A->1  |------>| 2nd   |
+          to be perceptible to CPU 2            +-------+       |       |
+                                                :       :       +-------+
+But it may be that the update to A from CPU 1 becomes perceptible to CPU 2
+before the read barrier completes anyway:
+        +-------+       :      :                :       :
+        |       |       +------+                +-------+
+        |       |------>| A=1  |------      --->| A->0  |
+        |       |       +------+      \         +-------+
+        | CPU 1 |   wwwwwwwwwwwwwwww   \    --->| B->9  |
+        |       |       +------+        |       +-------+
+        |       |------>| B=2  |---     |       :       :
+        |       |       +------+   \    |       :       :       +-------+
+        +-------+       :      :    \   |       +-------+       |       |
+                                     ---------->| B->2  |------>|       |
+                                        |       +-------+       | CPU 2 |
+                                        |       :       :       |       |
+                                         \      :       :       |       |
+                                          \     +-------+       |       |
+                                           ---->| A->1  |------>| 1st   |
+                                                +-------+       |       |
+                                            rrrrrrrrrrrrrrrrr   |       |
+                                                +-------+       |       |
+                                                | A->1  |------>| 2nd   |
+                                                +-------+       |       |
+                                                :       :       +-------+
+The guarantee is that the second load will always come up with A == 1 if the
+load of B came up with B == 2.  No such guarantee exists for the first load of
+A; that may come up with either A == 0 or A == 1.
+READ MEMORY BARRIERS VS LOAD SPECULATION
+----------------------------------------
+Many CPUs speculate with loads: that is they see that they will need to load an
+item from memory, and they find a time where they're not using the bus for any
+other loads, and so do the load in advance - even though they haven't actually
+got to that point in the instruction execution flow yet.  This permits the
+actual load instruction to potentially complete immediately because the CPU
+already has the value to hand.
+It may turn out that the CPU didn't actually need the value - perhaps because a
+branch circumvented the load - in which case it can discard the value or just
+cache it for later use.
+Consider:
+        CPU 1                   CPU 2
+        ======================= =======================
+                                LOAD B
+                                DIVIDE          } Divide instructions generally
+                                DIVIDE          } take a long time to perform
+                                LOAD A
+Which might appear as this:
+                                                :       :       +-------+
+                                                +-------+       |       |
+                                            --->| B->2  |------>|       |
+                                                +-------+       | CPU 2 |
+                                                :       :DIVIDE |       |
+                                                +-------+       |       |
+        The CPU being busy doing a --->     --->| A->0  |~~~~   |       |
+        division speculates on the              +-------+   ~   |       |
+        LOAD of A                               :       :   ~   |       |
+                                                :       :DIVIDE |       |
+                                                :       :   ~   |       |
+        Once the divisions are complete -->     :       :   ~-->|       |
+        the CPU can then perform the            :       :       |       |
+        LOAD with immediate effect              :       :       +-------+
+Placing a read barrier or a data dependency barrier just before the second
+load:
+        CPU 1                   CPU 2
+        ======================= =======================
+                                LOAD B
+                                DIVIDE
+                                DIVIDE
+                                <read barrier>
+                                LOAD A
+will force any value speculatively obtained to be reconsidered to an extent
+dependent on the type of barrier used.  If there was no change made to the
+speculated memory location, then the speculated value will just be used:
+                                                :       :       +-------+
+                                                +-------+       |       |
+                                            --->| B->2  |------>|       |
+                                                +-------+       | CPU 2 |
+                                                :       :DIVIDE |       |
+                                                +-------+       |       |
+        The CPU being busy doing a --->     --->| A->0  |~~~~   |       |
+        division speculates on the              +-------+   ~   |       |
+        LOAD of A                               :       :   ~   |       |
+                                                :       :DIVIDE |       |
+                                                :       :   ~   |       |
+                                                :       :   ~   |       |
+                                            rrrrrrrrrrrrrrrr~   |       |
+                                                :       :   ~   |       |
+                                                :       :   ~-->|       |
+                                                :       :       |       |
+                                                :       :       +-------+
+but if there was an update or an invalidation from another CPU pending, then
+the speculation will be cancelled and the value reloaded:
+                                                :       :       +-------+
+                                                +-------+       |       |
+                                            --->| B->2  |------>|       |
+                                                +-------+       | CPU 2 |
+                                                :       :DIVIDE |       |
+                                                +-------+       |       |
+        The CPU being busy doing a --->     --->| A->0  |~~~~   |       |
+        division speculates on the              +-------+   ~   |       |
+        LOAD of A                               :       :   ~   |       |
+                                                :       :DIVIDE |       |
+                                                :       :   ~   |       |
+                                                :       :   ~   |       |
+                                            rrrrrrrrrrrrrrrrr   |       |
+                                                +-------+       |       |
+        The speculation is discarded --->   --->| A->1  |------>|       |
+        and an updated value is                 +-------+       |       |
+        retrieved                               :       :       +-------+
+========================
+EXPLICIT KERNEL BARRIERS
+========================
+The Linux kernel has a variety of different barriers that act at different
+levels:
+  (*) Compiler barrier.
+  (*) CPU memory barriers.
+  (*) MMIO write barrier.
+COMPILER BARRIER
+----------------
+The Linux kernel has an explicit compiler barrier function that prevents the
+compiler from moving the memory accesses either side of it to the other side:
+        barrier();
+This a general barrier - lesser varieties of compiler barrier do not exist.
+The compiler barrier has no direct effect on the CPU, which may then reorder
+things however it wishes.
+CPU MEMORY BARRIERS
+-------------------
+The Linux kernel has eight basic CPU memory barriers:
+        TYPE            MANDATORY               SMP CONDITIONAL
+        =============== ======================= ===========================
+        GENERAL         mb()                    smp_mb()
+        WRITE           wmb()                   smp_wmb()
+        READ            rmb()                   smp_rmb()
+        DATA DEPENDENCY read_barrier_depends()  smp_read_barrier_depends()
+All CPU memory barriers unconditionally imply compiler barriers.
+SMP memory barriers are reduced to compiler barriers on uniprocessor compiled
+systems because it is assumed that a CPU will be appear to be self-consistent,
+and will order overlapping accesses correctly with respect to itself.
+[!] Note that SMP memory barriers _must_ be used to control the ordering of
+references to shared memory on SMP systems, though the use of locking instead
+is sufficient.
+Mandatory barriers should not be used to control SMP effects, since mandatory
+barriers unnecessarily impose overhead on UP systems. They may, however, be
+used to control MMIO effects on accesses through relaxed memory I/O windows.
+These are required even on non-SMP systems as they affect the order in which
+memory operations appear to a device by prohibiting both the compiler and the
+CPU from reordering them.
+There are some more advanced barrier functions:
+ (*) set_mb(var, value)
+ (*) set_wmb(var, value)
+     These assign the value to the variable and then insert at least a write
+     barrier after it, depending on the function.  They aren't guaranteed to
+     insert anything more than a compiler barrier in a UP compilation.
+ (*) smp_mb__before_atomic_dec();
+ (*) smp_mb__after_atomic_dec();
+ (*) smp_mb__before_atomic_inc();
+ (*) smp_mb__after_atomic_inc();
+     These are for use with atomic add, subtract, increment and decrement
+     functions that don't return a value, especially when used for reference
+     counting.  These functions do not imply memory barriers.
+     As an example, consider a piece of code that marks an object as being dead
+     and then decrements the object's reference count:
+        obj->dead = 1;
+        smp_mb__before_atomic_dec();
+        atomic_dec(&obj->ref_count);
+     This makes sure that the death mark on the object is perceived to be set
+     *before* the reference counter is decremented.
+     See Documentation/atomic_ops.txt for more information.  See the "Atomic
+     operations" subsection for information on where to use these.
+ (*) smp_mb__before_clear_bit(void);
+ (*) smp_mb__after_clear_bit(void);
+     These are for use similar to the atomic inc/dec barriers.  These are
+     typically used for bitwise unlocking operations, so care must be taken as
+     there are no implicit memory barriers here either.
+     Consider implementing an unlock operation of some nature by clearing a
+     locking bit.  The clear_bit() would then need to be barriered like this:
+        smp_mb__before_clear_bit();
+        clear_bit( ... );
+     This prevents memory operations before the clear leaking to after it.  See
+     the subsection on "Locking Functions" with reference to UNLOCK operation
+     implications.
+     See Documentation/atomic_ops.txt for more information.  See the "Atomic
+     operations" subsection for information on where to use these.
+MMIO WRITE BARRIER
+------------------
+The Linux kernel also has a special barrier for use with memory-mapped I/O
+writes:
+        mmiowb();
+This is a variation on the mandatory write barrier that causes writes to weakly
+ordered I/O regions to be partially ordered.  Its effects may go beyond the
+CPU->Hardware interface and actually affect the hardware at some level.
+See the subsection "Locks vs I/O accesses" for more information.
+===============================
+IMPLICIT KERNEL MEMORY BARRIERS
+===============================
+Some of the other functions in the linux kernel imply memory barriers, amongst
+which are locking and scheduling functions.
+This specification is a _minimum_ guarantee; any particular architecture may
+provide more substantial guarantees, but these may not be relied upon outside
+of arch specific code.
+LOCKING FUNCTIONS
+-----------------
+The Linux kernel has a number of locking constructs:
+ (*) spin locks
+ (*) R/W spin locks
+ (*) mutexes
+ (*) semaphores
+ (*) R/W semaphores
+ (*) RCU
+In all cases there are variants on "LOCK" operations and "UNLOCK" operations
+for each construct.  These operations all imply certain barriers:
+ (1) LOCK operation implication:
+     Memory operations issued after the LOCK will be completed after the LOCK
+     operation has completed.
+     Memory operations issued before the LOCK may be completed after the LOCK
+     operation has completed.
+ (2) UNLOCK operation implication:
+     Memory operations issued before the UNLOCK will be completed before the
+     UNLOCK operation has completed.
+     Memory operations issued after the UNLOCK may be completed before the
+     UNLOCK operation has completed.
+ (3) LOCK vs LOCK implication:
+     All LOCK operations issued before another LOCK operation will be completed
+     before that LOCK operation.
+ (4) LOCK vs UNLOCK implication:
+     All LOCK operations issued before an UNLOCK operation will be completed
+     before the UNLOCK operation.
+     All UNLOCK operations issued before a LOCK operation will be completed
+     before the LOCK operation.
+ (5) Failed conditional LOCK implication:
+     Certain variants of the LOCK operation may fail, either due to being
+     unable to get the lock immediately, or due to receiving an unblocked
+     signal whilst asleep waiting for the lock to become available.  Failed
+     locks do not imply any sort of barrier.
+Therefore, from (1), (2) and (4) an UNLOCK followed by an unconditional LOCK is
+equivalent to a full barrier, but a LOCK followed by an UNLOCK is not.
+[!] Note: one of the consequence of LOCKs and UNLOCKs being only one-way
+    barriers is that the effects instructions outside of a critical section may
+    seep into the inside of the critical section.
+A LOCK followed by an UNLOCK may not be assumed to be full memory barrier
+because it is possible for an access preceding the LOCK to happen after the
+LOCK, and an access following the UNLOCK to happen before the UNLOCK, and the
+two accesses can themselves then cross:
+        *A = a;
+        LOCK
+        UNLOCK
+        *B = b;
+may occur as:
+        LOCK, STORE *B, STORE *A, UNLOCK
+Locks and semaphores may not provide any guarantee of ordering on UP compiled
+systems, and so cannot be counted on in such a situation to actually achieve
+anything at all - especially with respect to I/O accesses - unless combined
+with interrupt disabling operations.
+See also the section on "Inter-CPU locking barrier effects".
+As an example, consider the following:
+        *A = a;
+        *B = b;
+        LOCK
+        *C = c;
+        *D = d;
+        UNLOCK
+        *E = e;
+        *F = f;
+The following sequence of events is acceptable:
+        LOCK, {*F,*A}, *E, {*C,*D}, *B, UNLOCK
+        [+] Note that {*F,*A} indicates a combined access.
+But none of the following are:
+        {*F,*A}, *B,    LOCK, *C, *D,   UNLOCK, *E
+        *A, *B, *C,     LOCK, *D,       UNLOCK, *E, *F
+        *A, *B,         LOCK, *C,       UNLOCK, *D, *E, *F
+        *B,             LOCK, *C, *D,   UNLOCK, {*F,*A}, *E
+INTERRUPT DISABLING FUNCTIONS
+-----------------------------
+Functions that disable interrupts (LOCK equivalent) and enable interrupts
+(UNLOCK equivalent) will act as compiler barriers only.  So if memory or I/O
+barriers are required in such a situation, they must be provided from some
+other means.
+MISCELLANEOUS FUNCTIONS
+-----------------------
+Other functions that imply barriers:
+ (*) schedule() and similar imply full memory barriers.
+=================================
+INTER-CPU LOCKING BARRIER EFFECTS
+=================================
+On SMP systems locking primitives give a more substantial form of barrier: one
+that does affect memory access ordering on other CPUs, within the context of
+conflict on any particular lock.
+LOCKS VS MEMORY ACCESSES
+------------------------
+Consider the following: the system has a pair of spinlocks (M) and (Q), and
+three CPUs; then should the following sequence of events occur:
+        CPU 1                           CPU 2
+        =============================== ===============================
+        *A = a;                         *E = e;
+        LOCK M                          LOCK Q
+        *B = b;                         *F = f;
+        *C = c;                         *G = g;
+        UNLOCK M                        UNLOCK Q
+        *D = d;                         *H = h;
+Then there is no guarantee as to what order CPU #3 will see the accesses to *A
+through *H occur in, other than the constraints imposed by the separate locks
+on the separate CPUs. It might, for example, see:
+        *E, LOCK M, LOCK Q, *G, *C, *F, *A, *B, UNLOCK Q, *D, *H, UNLOCK M
+But it won't see any of:
+        *B, *C or *D preceding LOCK M
+        *A, *B or *C following UNLOCK M
+        *F, *G or *H preceding LOCK Q
+        *E, *F or *G following UNLOCK Q
+However, if the following occurs:
+        CPU 1                           CPU 2
+        =============================== ===============================
+        *A = a;
+        LOCK M          [1]
+        *B = b;
+        *C = c;
+        UNLOCK M        [1]
+        *D = d;                         *E = e;
+                                        LOCK M          [2]
+                                        *F = f;
+                                        *G = g;
+                                        UNLOCK M        [2]
+                                        *H = h;
+CPU #3 might see:
+        *E, LOCK M [1], *C, *B, *A, UNLOCK M [1],
+                LOCK M [2], *H, *F, *G, UNLOCK M [2], *D
+But assuming CPU #1 gets the lock first, it won't see any of:
+        *B, *C, *D, *F, *G or *H preceding LOCK M [1]
+        *A, *B or *C following UNLOCK M [1]
+        *F, *G or *H preceding LOCK M [2]
+        *A, *B, *C, *E, *F or *G following UNLOCK M [2]
+LOCKS VS I/O ACCESSES
+---------------------
+Under certain circumstances (especially involving NUMA), I/O accesses within
+two spinlocked sections on two different CPUs may be seen as interleaved by the
+PCI bridge, because the PCI bridge does not necessarily participate in the
+cache-coherence protocol, and is therefore incapable of issuing the required
+read memory barriers.
+For example:
+        CPU 1                           CPU 2
+        =============================== ===============================
+        spin_lock(Q)
+        writel(0, ADDR)
+        writel(1, DATA);
+        spin_unlock(Q);
+                                        spin_lock(Q);
+                                        writel(4, ADDR);
+                                        writel(5, DATA);
+                                        spin_unlock(Q);
+may be seen by the PCI bridge as follows:
+        STORE *ADDR = 0, STORE *ADDR = 4, STORE *DATA = 1, STORE *DATA = 5
+which would probably cause the hardware to malfunction.
+What is necessary here is to intervene with an mmiowb() before dropping the
+spinlock, for example:
+        CPU 1                           CPU 2
+        =============================== ===============================
+        spin_lock(Q)
+        writel(0, ADDR)
+        writel(1, DATA);
+        mmiowb();
+        spin_unlock(Q);
+                                        spin_lock(Q);
+                                        writel(4, ADDR);
+                                        writel(5, DATA);
+                                        mmiowb();
+                                        spin_unlock(Q);
+this will ensure that the two stores issued on CPU #1 appear at the PCI bridge
+before either of the stores issued on CPU #2.
+Furthermore, following a store by a load to the same device obviates the need
+for an mmiowb(), because the load forces the store to complete before the load
+is performed:
+        CPU 1                           CPU 2
+        =============================== ===============================
+        spin_lock(Q)
+        writel(0, ADDR)
+        a = readl(DATA);
+        spin_unlock(Q);
+                                        spin_lock(Q);
+                                        writel(4, ADDR);
+                                        b = readl(DATA);
+                                        spin_unlock(Q);
+See Documentation/DocBook/deviceiobook.tmpl for more information.
+=================================
+WHERE ARE MEMORY BARRIERS NEEDED?
+=================================
+Under normal operation, memory operation reordering is generally not going to
+be a problem as a single-threaded linear piece of code will still appear to
+work correctly, even if it's in an SMP kernel.  There are, however, three
+circumstances in which reordering definitely _could_ be a problem:
+ (*) Interprocessor interaction.
+ (*) Atomic operations.
+ (*) Accessing devices (I/O).
+ (*) Interrupts.
+INTERPROCESSOR INTERACTION
+--------------------------
+When there's a system with more than one processor, more than one CPU in the
+system may be working on the same data set at the same time.  This can cause
+synchronisation problems, and the usual way of dealing with them is to use
+locks.  Locks, however, are quite expensive, and so it may be preferable to
+operate without the use of a lock if at all possible.  In such a case
+operations that affect both CPUs may have to be carefully ordered to prevent
+a malfunction.
+Consider, for example, the R/W semaphore slow path.  Here a waiting process is
+queued on the semaphore, by virtue of it having a piece of its stack linked to
+the semaphore's list of waiting processes:
+        struct rw_semaphore {
+                ...
+                spinlock_t lock;
+                struct list_head waiters;
+        };
+        struct rwsem_waiter {
+                struct list_head list;
+                struct task_struct *task;
+        };
+To wake up a particular waiter, the up_read() or up_write() functions have to:
+ (1) read the next pointer from this waiter's record to know as to where the
+     next waiter record is;
+ (4) read the pointer to the waiter's task structure;
+ (3) clear the task pointer to tell the waiter it has been given the semaphore;
+ (4) call wake_up_process() on the task; and
+ (5) release the reference held on the waiter's task struct.
+In otherwords, it has to perform this sequence of events:
+        LOAD waiter->list.next;
+        LOAD waiter->task;
+        STORE waiter->task;
+        CALL wakeup
+        RELEASE task
+and if any of these steps occur out of order, then the whole thing may
+malfunction.
+Once it has queued itself and dropped the semaphore lock, the waiter does not
+get the lock again; it instead just waits for its task pointer to be cleared
+before proceeding.  Since the record is on the waiter's stack, this means that
+if the task pointer is cleared _before_ the next pointer in the list is read,
+another CPU might start processing the waiter and might clobber the waiter's
+stack before the up*() function has a chance to read the next pointer.
+Consider then what might happen to the above sequence of events:
+        CPU 1                           CPU 2
+        =============================== ===============================
+                                        down_xxx()
+                                        Queue waiter
+                                        Sleep
+        up_yyy()
+        LOAD waiter->task;
+        STORE waiter->task;
+                                        Woken up by other event
+        <preempt>
+                                        Resume processing
+                                        down_xxx() returns
+                                        call foo()
+                                        foo() clobbers *waiter
+        </preempt>
+        LOAD waiter->list.next;
+        --- OOPS ---
+This could be dealt with using the semaphore lock, but then the down_xxx()
+function has to needlessly get the spinlock again after being woken up.
+The way to deal with this is to insert a general SMP memory barrier:
+        LOAD waiter->list.next;
+        LOAD waiter->task;
+        smp_mb();
+        STORE waiter->task;
+        CALL wakeup
+        RELEASE task
+In this case, the barrier makes a guarantee that all memory accesses before the
+barrier will appear to happen before all the memory accesses after the barrier
+with respect to the other CPUs on the system.  It does _not_ guarantee that all
+the memory accesses before the barrier will be complete by the time the barrier
+instruction itself is complete.
+On a UP system - where this wouldn't be a problem - the smp_mb() is just a
+compiler barrier, thus making sure the compiler emits the instructions in the
+right order without actually intervening in the CPU.  Since there there's only
+one CPU, that CPU's dependency ordering logic will take care of everything
+else.
+ATOMIC OPERATIONS
+-----------------
+Whilst they are technically interprocessor interaction considerations, atomic
+operations are noted specially as some of them imply full memory barriers and
+some don't, but they're very heavily relied on as a group throughout the
+kernel.
+Any atomic operation that modifies some state in memory and returns information
+about the state (old or new) implies an SMP-conditional general memory barrier
+(smp_mb()) on each side of the actual operation.  These include:
+        xchg();
+        cmpxchg();
+        atomic_cmpxchg();
+        atomic_inc_return();
+        atomic_dec_return();
+        atomic_add_return();
+        atomic_sub_return();
+        atomic_inc_and_test();
+        atomic_dec_and_test();
+        atomic_sub_and_test();
+        atomic_add_negative();
+        atomic_add_unless();
+        test_and_set_bit();
+        test_and_clear_bit();
+        test_and_change_bit();
+These are used for such things as implementing LOCK-class and UNLOCK-class
+operations and adjusting reference counters towards object destruction, and as
+such the implicit memory barrier effects are necessary.
+The following operation are potential problems as they do _not_ imply memory
+barriers, but might be used for implementing such things as UNLOCK-class
+operations:
+        atomic_set();
+        set_bit();
+        clear_bit();
+        change_bit();
+With these the appropriate explicit memory barrier should be used if necessary
+(smp_mb__before_clear_bit() for instance).
+The following also do _not_ imply memory barriers, and so may require explicit
+memory barriers under some circumstances (smp_mb__before_atomic_dec() for
+instance)):
+        atomic_add();
+        atomic_sub();
+        atomic_inc();
+        atomic_dec();
+If they're used for statistics generation, then they probably don't need memory
+barriers, unless there's a coupling between statistical data.
+If they're used for reference counting on an object to control its lifetime,
+they probably don't need memory barriers because either the reference count
+will be adjusted inside a locked section, or the caller will already hold
+sufficient references to make the lock, and thus a memory barrier unnecessary.
+If they're used for constructing a lock of some description, then they probably
+do need memory barriers as a lock primitive generally has to do things in a
+specific order.
+Basically, each usage case has to be carefully considered as to whether memory
+barriers are needed or not.
+[!] Note that special memory barrier primitives are available for these
+situations because on some CPUs the atomic instructions used imply full memory
+barriers, and so barrier instructions are superfluous in conjunction with them,
+and in such cases the special barrier primitives will be no-ops.
+See Documentation/atomic_ops.txt for more information.
+ACCESSING DEVICES
+-----------------
+Many devices can be memory mapped, and so appear to the CPU as if they're just
+a set of memory locations.  To control such a device, the driver usually has to
+make the right memory accesses in exactly the right order.
+However, having a clever CPU or a clever compiler creates a potential problem
+in that the carefully sequenced accesses in the driver code won't reach the
+device in the requisite order if the CPU or the compiler thinks it is more
+efficient to reorder, combine or merge accesses - something that would cause
+the device to malfunction.
+Inside of the Linux kernel, I/O should be done through the appropriate accessor
+routines - such as inb() or writel() - which know how to make such accesses
+appropriately sequential.  Whilst this, for the most part, renders the explicit
+use of memory barriers unnecessary, there are a couple of situations where they
+might be needed:
+ (1) On some systems, I/O stores are not strongly ordered across all CPUs, and
+     so for _all_ general drivers locks should be used and mmiowb() must be
+     issued prior to unlocking the critical section.
+ (2) If the accessor functions are used to refer to an I/O memory window with
+     relaxed memory access properties, then _mandatory_ memory barriers are
+     required to enforce ordering.
+See Documentation/DocBook/deviceiobook.tmpl for more information.
+INTERRUPTS
+----------
+A driver may be interrupted by its own interrupt service routine, and thus the
+two parts of the driver may interfere with each other's attempts to control or
+access the device.
+This may be alleviated - at least in part - by disabling local interrupts (a
+form of locking), such that the critical operations are all contained within
+the interrupt-disabled section in the driver.  Whilst the driver's interrupt
+routine is executing, the driver's core may not run on the same CPU, and its
+interrupt is not permitted to happen again until the current interrupt has been
+handled, thus the interrupt handler does not need to lock against that.
+However, consider a driver that was talking to an ethernet card that sports an
+address register and a data register.  If that driver's core talks to the card
+under interrupt-disablement and then the driver's interrupt handler is invoked:
+        LOCAL IRQ DISABLE
+        writew(ADDR, 3);
+        writew(DATA, y);
+        LOCAL IRQ ENABLE
+        <interrupt>
+        writew(ADDR, 4);
+        q = readw(DATA);
+        </interrupt>
+The store to the data register might happen after the second store to the
+address register if ordering rules are sufficiently relaxed:
+        STORE *ADDR = 3, STORE *ADDR = 4, STORE *DATA = y, q = LOAD *DATA
+If ordering rules are relaxed, it must be assumed that accesses done inside an
+interrupt disabled section may leak outside of it and may interleave with
+accesses performed in an interrupt - and vice versa - unless implicit or
+explicit barriers are used.
+Normally this won't be a problem because the I/O accesses done inside such
+sections will include synchronous load operations on strictly ordered I/O
+registers that form implicit I/O barriers. If this isn't sufficient then an
+mmiowb() may need to be used explicitly.
+A similar situation may occur between an interrupt routine and two routines
+running on separate CPUs that communicate with each other. If such a case is
+likely, then interrupt-disabling locks should be used to guarantee ordering.
+==========================
+KERNEL I/O BARRIER EFFECTS
+==========================
+When accessing I/O memory, drivers should use the appropriate accessor
+functions:
+ (*) inX(), outX():
+     These are intended to talk to I/O space rather than memory space, but
+     that's primarily a CPU-specific concept. The i386 and x86_64 processors do
+     indeed have special I/O space access cycles and instructions, but many
+     CPUs don't have such a concept.
+     The PCI bus, amongst others, defines an I/O space concept - which on such
+     CPUs as i386 and x86_64 cpus readily maps to the CPU's concept of I/O
+     space.  However, it may also mapped as a virtual I/O space in the CPU's
+     memory map, particularly on those CPUs that don't support alternate
+     I/O spaces.
+     Accesses to this space may be fully synchronous (as on i386), but
+     intermediary bridges (such as the PCI host bridge) may not fully honour
+     that.
+     They are guaranteed to be fully ordered with respect to each other.
+     They are not guaranteed to be fully ordered with respect to other types of
+     memory and I/O operation.
+ (*) readX(), writeX():
+     Whether these are guaranteed to be fully ordered and uncombined with
+     respect to each other on the issuing CPU depends on the characteristics
+     defined for the memory window through which they're accessing. On later
+     i386 architecture machines, for example, this is controlled by way of the
+     MTRR registers.
+     Ordinarily, these will be guaranteed to be fully ordered and uncombined,,
+     provided they're not accessing a prefetchable device.
+     However, intermediary hardware (such as a PCI bridge) may indulge in
+     deferral if it so wishes; to flush a store, a load from the same location
+     is preferred[*], but a load from the same device or from configuration
+     space should suffice for PCI.
+     [*] NOTE! attempting to load from the same location as was written to may
+         cause a malfunction - consider the 16550 Rx/Tx serial registers for
+         example.
+     Used with prefetchable I/O memory, an mmiowb() barrier may be required to
+     force stores to be ordered.
+     Please refer to the PCI specification for more information on interactions
+     between PCI transactions.
+ (*) readX_relaxed()
+     These are similar to readX(), but are not guaranteed to be ordered in any
+     way. Be aware that there is no I/O read barrier available.
+ (*) ioreadX(), iowriteX()
+     These will perform as appropriate for the type of access they're actually
+     doing, be it inX()/outX() or readX()/writeX().
+========================================
+ASSUMED MINIMUM EXECUTION ORDERING MODEL
+========================================
+It has to be assumed that the conceptual CPU is weakly-ordered but that it will
+maintain the appearance of program causality with respect to itself.  Some CPUs
+(such as i386 or x86_64) are more constrained than others (such as powerpc or
+frv), and so the most relaxed case (namely DEC Alpha) must be assumed outside
+of arch-specific code.
+This means that it must be considered that the CPU will execute its instruction
+stream in any order it feels like - or even in parallel - provided that if an
+instruction in the stream depends on the an earlier instruction, then that
+earlier instruction must be sufficiently complete[*] before the later
+instruction may proceed; in other words: provided that the appearance of
+causality is maintained.
+ [*] Some instructions have more than one effect - such as changing the
+     condition codes, changing registers or changing memory - and different
+     instructions may depend on different effects.
+A CPU may also discard any instruction sequence that winds up having no
+ultimate effect.  For example, if two adjacent instructions both load an
+immediate value into the same register, the first may be discarded.
+Similarly, it has to be assumed that compiler might reorder the instruction
+stream in any way it sees fit, again provided the appearance of causality is
+maintained.
+============================
+THE EFFECTS OF THE CPU CACHE
+============================
+The way cached memory operations are perceived across the system is affected to
+a certain extent by the caches that lie between CPUs and memory, and by the
+memory coherence system that maintains the consistency of state in the system.
+As far as the way a CPU interacts with another part of the system through the
+caches goes, the memory system has to include the CPU's caches, and memory
+barriers for the most part act at the interface between the CPU and its cache
+(memory barriers logically act on the dotted line in the following diagram):
+            <--- CPU --->         :       <----------- Memory ----------->
+                                  :
+        +--------+    +--------+  :   +--------+    +-----------+
+        |        |    |        |  :   |        |    |           |    +--------+
+        |  CPU   |    | Memory |  :   | CPU    |    |           |    |        |
+        |  Core  |--->| Access |----->| Cache  |<-->|           |    |        |
+        |        |    | Queue  |  :   |        |    |           |--->| Memory |
+        |        |    |        |  :   |        |    |           |    |        |
+        +--------+    +--------+  :   +--------+    |           |    |        |
+                                  :                 | Cache     |    +--------+
+                                  :                 | Coherency |
+                                  :                 | Mechanism |    +--------+
+        +--------+    +--------+  :   +--------+    |           |    |        |
+        |        |    |        |  :   |        |    |           |    |        |
+        |  CPU   |    | Memory |  :   | CPU    |    |           |--->| Device |
+        |  Core  |--->| Access |----->| Cache  |<-->|           |    |        |
+        |        |    | Queue  |  :   |        |    |           |    |        |
+        |        |    |        |  :   |        |    |           |    +--------+
+        +--------+    +--------+  :   +--------+    +-----------+
+                                  :
+                                  :
+Although any particular load or store may not actually appear outside of the
+CPU that issued it since it may have been satisfied within the CPU's own cache,
+it will still appear as if the full memory access had taken place as far as the
+other CPUs are concerned since the cache coherency mechanisms will migrate the
+cacheline over to the accessing CPU and propagate the effects upon conflict.
+The CPU core may execute instructions in any order it deems fit, provided the
+expected program causality appears to be maintained.  Some of the instructions
+generate load and store operations which then go into the queue of memory
+accesses to be performed.  The core may place these in the queue in any order
+it wishes, and continue execution until it is forced to wait for an instruction
+to complete.
+What memory barriers are concerned with is controlling the order in which
+accesses cross from the CPU side of things to the memory side of things, and
+the order in which the effects are perceived to happen by the other observers
+in the system.
+[!] Memory barriers are _not_ needed within a given CPU, as CPUs always see
+their own loads and stores as if they had happened in program order.
+[!] MMIO or other device accesses may bypass the cache system.  This depends on
+the properties of the memory window through which devices are accessed and/or
+the use of any special device communication instructions the CPU may have.
+CACHE COHERENCY
+---------------
+Life isn't quite as simple as it may appear above, however: for while the
+caches are expected to be coherent, there's no guarantee that that coherency
+will be ordered.  This means that whilst changes made on one CPU will
+eventually become visible on all CPUs, there's no guarantee that they will
+become apparent in the same order on those other CPUs.
+Consider dealing with a system that has pair of CPUs (1 & 2), each of which has
+a pair of parallel data caches (CPU 1 has A/B, and CPU 2 has C/D):
+                    :
+                    :                          +--------+
+                    :      +---------+         |        |
+        +--------+  : +--->| Cache A |<------->|        |
+        |        |  : |    +---------+         |        |
+        |  CPU 1 |<---+                        |        |
+        |        |  : |    +---------+         |        |
+        +--------+  : +--->| Cache B |<------->|        |
+                    :      +---------+         |        |
+                    :                          | Memory |
+                    :      +---------+         | System |
+        +--------+  : +--->| Cache C |<------->|        |
+        |        |  : |    +---------+         |        |
+        |  CPU 2 |<---+                        |        |
+        |        |  : |    +---------+         |        |
+        +--------+  : +--->| Cache D |<------->|        |
+                    :      +---------+         |        |
+                    :                          +--------+
+                    :
+Imagine the system has the following properties:
+ (*) an odd-numbered cache line may be in cache A, cache C or it may still be
+     resident in memory;
+ (*) an even-numbered cache line may be in cache B, cache D or it may still be
+     resident in memory;
+ (*) whilst the CPU core is interrogating one cache, the other cache may be
+     making use of the bus to access the rest of the system - perhaps to
+     displace a dirty cacheline or to do a speculative load;
+ (*) each cache has a queue of operations that need to be applied to that cache
+     to maintain coherency with the rest of the system;
+ (*) the coherency queue is not flushed by normal loads to lines already
+     present in the cache, even though the contents of the queue may
+     potentially effect those loads.
+Imagine, then, that two writes are made on the first CPU, with a write barrier
+between them to guarantee that they will appear to reach that CPU's caches in
+the requisite order:
+        CPU 1           CPU 2           COMMENT
+        =============== =============== =======================================
+                                        u == 0, v == 1 and p == &u, q == &u
+        v = 2;
+        smp_wmb();                      Make sure change to v visible before
+                                         change to p
+        <A:modify v=2>                  v is now in cache A exclusively
+        p = &v;
+        <B:modify p=&v>                 p is now in cache B exclusively
+The write memory barrier forces the other CPUs in the system to perceive that
+the local CPU's caches have apparently been updated in the correct order.  But
+now imagine that the second CPU that wants to read those values:
+        CPU 1           CPU 2           COMMENT
+        =============== =============== =======================================
+        ...
+                        q = p;
+                        x = *q;
+The above pair of reads may then fail to happen in expected order, as the
+cacheline holding p may get updated in one of the second CPU's caches whilst
+the update to the cacheline holding v is delayed in the other of the second
+CPU's caches by some other cache event:
+        CPU 1           CPU 2           COMMENT
+        =============== =============== =======================================
+                                        u == 0, v == 1 and p == &u, q == &u
+        v = 2;
+        smp_wmb();
+        <A:modify v=2>  <C:busy>
+                        <C:queue v=2>
+        p = &v;         q = p;
+                        <D:request p>
+        <B:modify p=&v> <D:commit p=&v>
+                        <D:read p>
+                        x = *q;
+                        <C:read *q>     Reads from v before v updated in cache
+                        <C:unbusy>
+                        <C:commit v=2>
+Basically, whilst both cachelines will be updated on CPU 2 eventually, there's
+no guarantee that, without intervention, the order of update will be the same
+as that committed on CPU 1.
+To intervene, we need to interpolate a data dependency barrier or a read
+barrier between the loads.  This will force the cache to commit its coherency
+queue before processing any further requests:
+        CPU 1           CPU 2           COMMENT
+        =============== =============== =======================================
+                                        u == 0, v == 1 and p == &u, q == &u
+        v = 2;
+        smp_wmb();
+        <A:modify v=2>  <C:busy>
+                        <C:queue v=2>
+        p = &b;         q = p;
+                        <D:request p>
+        <B:modify p=&v> <D:commit p=&v>
+                        <D:read p>
+                        smp_read_barrier_depends()
+                        <C:unbusy>
+                        <C:commit v=2>
+                        x = *q;
+                        <C:read *q>     Reads from v after v updated in cache
+This sort of problem can be encountered on DEC Alpha processors as they have a
+split cache that improves performance by making better use of the data bus.
+Whilst most CPUs do imply a data dependency barrier on the read when a memory
+access depends on a read, not all do, so it may not be relied on.
+Other CPUs may also have split caches, but must coordinate between the various
+cachelets for normal memory accesss.  The semantics of the Alpha removes the
+need for coordination in absence of memory barriers.
+CACHE COHERENCY VS DMA
+----------------------
+Not all systems maintain cache coherency with respect to devices doing DMA.  In
+such cases, a device attempting DMA may obtain stale data from RAM because
+dirty cache lines may be resident in the caches of various CPUs, and may not
+have been written back to RAM yet.  To deal with this, the appropriate part of
+the kernel must flush the overlapping bits of cache on each CPU (and maybe
+invalidate them as well).
+In addition, the data DMA'd to RAM by a device may be overwritten by dirty
+cache lines being written back to RAM from a CPU's cache after the device has
+installed its own data, or cache lines simply present in a CPUs cache may
+simply obscure the fact that RAM has been updated, until at such time as the
+cacheline is discarded from the CPU's cache and reloaded.  To deal with this,
+the appropriate part of the kernel must invalidate the overlapping bits of the
+cache on each CPU.
+See Documentation/cachetlb.txt for more information on cache management.
+CACHE COHERENCY VS MMIO
+-----------------------
+Memory mapped I/O usually takes place through memory locations that are part of
+a window in the CPU's memory space that have different properties assigned than
+the usual RAM directed window.
+Amongst these properties is usually the fact that such accesses bypass the
+caching entirely and go directly to the device buses.  This means MMIO accesses
+may, in effect, overtake accesses to cached memory that were emitted earlier.
+A memory barrier isn't sufficient in such a case, but rather the cache must be
+flushed between the cached memory write and the MMIO access if the two are in
+any way dependent.
+=========================
+THE THINGS CPUS GET UP TO
+=========================
+A programmer might take it for granted that the CPU will perform memory
+operations in exactly the order specified, so that if a CPU is, for example,
+given the following piece of code to execute:
+        a = *A;
+        *B = b;
+        c = *C;
+        d = *D;
+        *E = e;
+They would then expect that the CPU will complete the memory operation for each
+instruction before moving on to the next one, leading to a definite sequence of
+operations as seen by external observers in the system:
+        LOAD *A, STORE *B, LOAD *C, LOAD *D, STORE *E.
+Reality is, of course, much messier.  With many CPUs and compilers, the above
+assumption doesn't hold because:
+ (*) loads are more likely to need to be completed immediately to permit
+     execution progress, whereas stores can often be deferred without a
+     problem;
+ (*) loads may be done speculatively, and the result discarded should it prove
+     to have been unnecessary;
+ (*) loads may be done speculatively, leading to the result having being
+     fetched at the wrong time in the expected sequence of events;
+ (*) the order of the memory accesses may be rearranged to promote better use
+     of the CPU buses and caches;
+ (*) loads and stores may be combined to improve performance when talking to
+     memory or I/O hardware that can do batched accesses of adjacent locations,
+     thus cutting down on transaction setup costs (memory and PCI devices may
+     both be able to do this); and
+ (*) the CPU's data cache may affect the ordering, and whilst cache-coherency
+     mechanisms may alleviate this - once the store has actually hit the cache
+     - there's no guarantee that the coherency management will be propagated in
+     order to other CPUs.
+So what another CPU, say, might actually observe from the above piece of code
+is:
+        LOAD *A, ..., LOAD {*C,*D}, STORE *E, STORE *B
+        (Where "LOAD {*C,*D}" is a combined load)
+However, it is guaranteed that a CPU will be self-consistent: it will see its
+_own_ accesses appear to be correctly ordered, without the need for a memory
+barrier.  For instance with the following code:
+        U = *A;
+        *A = V;
+        *A = W;
+        X = *A;
+        *A = Y;
+        Z = *A;
+and assuming no intervention by an external influence, it can be assumed that
+the final result will appear to be:
+        U == the original value of *A
+        X == W
+        Z == Y
+        *A == Y
+The code above may cause the CPU to generate the full sequence of memory
+accesses:
+        U=LOAD *A, STORE *A=V, STORE *A=W, X=LOAD *A, STORE *A=Y, Z=LOAD *A
+in that order, but, without intervention, the sequence may have almost any
+combination of elements combined or discarded, provided the program's view of
+the world remains consistent.
+The compiler may also combine, discard or defer elements of the sequence before
+the CPU even sees them.
+For instance:
+        *A = V;
+        *A = W;
+may be reduced to:
+        *A = W;
+since, without a write barrier, it can be assumed that the effect of the
+storage of V to *A is lost.  Similarly:
+        *A = Y;
+        Z = *A;
+may, without a memory barrier, be reduced to:
+        *A = Y;
+        Z = Y;
+and the LOAD operation never appear outside of the CPU.
+AND THEN THERE'S THE ALPHA
+--------------------------
+The DEC Alpha CPU is one of the most relaxed CPUs there is.  Not only that,
+some versions of the Alpha CPU have a split data cache, permitting them to have
+two semantically related cache lines updating at separate times.  This is where
+the data dependency barrier really becomes necessary as this synchronises both
+caches with the memory coherence system, thus making it seem like pointer
+changes vs new data occur in the right order.
+The Alpha defines the Linux's kernel's memory barrier model.
+See the subsection on "Cache Coherency" above.
+==========
+REFERENCES
+==========
+Alpha AXP Architecture Reference Manual, Second Edition (Sites & Witek,
+Digital Press)
+        Chapter 5.2: Physical Address Space Characteristics
+        Chapter 5.4: Caches and Write Buffers
+        Chapter 5.5: Data Sharing
+        Chapter 5.6: Read/Write Ordering
+AMD64 Architecture Programmer's Manual Volume 2: System Programming
+        Chapter 7.1: Memory-Access Ordering
+        Chapter 7.4: Buffering and Combining Memory Writes
+IA-32 Intel Architecture Software Developer's Manual, Volume 3:
+System Programming Guide
+        Chapter 7.1: Locked Atomic Operations
+        Chapter 7.2: Memory Ordering
+        Chapter 7.4: Serializing Instructions
+The SPARC Architecture Manual, Version 9
+        Chapter 8: Memory Models
+        Appendix D: Formal Specification of the Memory Models
+        Appendix J: Programming with the Memory Models
+UltraSPARC Programmer Reference Manual
+        Chapter 5: Memory Accesses and Cacheability
+        Chapter 15: Sparc-V9 Memory Models
+UltraSPARC III Cu User's Manual
+        Chapter 9: Memory Models
+UltraSPARC IIIi Processor User's Manual
+        Chapter 8: Memory Models
+UltraSPARC Architecture 2005
+        Chapter 9: Memory
+        Appendix D: Formal Specifications of the Memory Models
+UltraSPARC T1 Supplement to the UltraSPARC Architecture 2005
+        Chapter 8: Memory Models
+        Appendix F: Caches and Cache Coherency
+Solaris Internals, Core Kernel Architecture, p63-68:
+        Chapter 3.3: Hardware Considerations for Locks and
+                        Synchronization
+Unix Systems for Modern Architectures, Symmetric Multiprocessing and Caching
+for Kernel Programmers:
+        Chapter 13: Other Memory Models
+Intel Itanium Architecture Software Developer's Manual: Volume 1:
+        Section 2.6: Speculation
+        Section 4.4: Memory Access
diff --git a/Documentation/mtrr.txt b/Documentation/mtrr.txt
index b78af1c32996..c39ac395970e 100644
--- a/Documentation/mtrr.txt
+++ b/Documentation/mtrr.txt
@@ -138,19 +138,29 @@ Reading MTRRs from a C program using ioctl()'s:
 */
 #include <stdio.h>
+#include <stdlib.h>
 #include <string.h>
 #include <sys/types.h>
 #include <sys/stat.h>
 #include <fcntl.h>
 #include <sys/ioctl.h>
 #include <errno.h>
-#define MTRR_NEED_STRINGS
 #include <asm/mtrr.h>
 #define TRUE 1
 #define FALSE 0
 #define ERRSTRING strerror (errno)
+static char *mtrr_strings[MTRR_NUM_TYPES] =
+{
+    "uncachable",               /* 0 */
+    "write-combining",          /* 1 */
+    "?",                        /* 2 */
+    "?",                        /* 3 */
+    "write-through",            /* 4 */
+    "write-protect",            /* 5 */
+    "write-back",               /* 6 */
+};
 int main ()
 {
@@ -232,13 +242,22 @@ Creating MTRRs from a C programme using ioctl()'s:
 #include <fcntl.h>
 #include <sys/ioctl.h>
 #include <errno.h>
-#define MTRR_NEED_STRINGS
 #include <asm/mtrr.h>
 #define TRUE 1
 #define FALSE 0
 #define ERRSTRING strerror (errno)
+static char *mtrr_strings[MTRR_NUM_TYPES] =
+{
+    "uncachable",               /* 0 */
+    "write-combining",          /* 1 */
+    "?",                        /* 2 */
+    "?",                        /* 3 */
+    "write-through",            /* 4 */
+    "write-protect",            /* 5 */
+    "write-back",               /* 6 */
+};
 int main (int argc, char **argv)
 {
diff --git a/Documentation/networking/operstates.txt b/Documentation/networking/operstates.txt
new file mode 100644
index 000000000000..4a21d9bb836b
--- /dev/null
+++ b/Documentation/networking/operstates.txt
@@ -0,0 +1,161 @@
+1. Introduction
+Linux distinguishes between administrative and operational state of an
+interface. Admininstrative state is the result of "ip link set dev
+<dev> up or down" and reflects whether the administrator wants to use
+the device for traffic.
+However, an interface is not usable just because the admin enabled it
+- ethernet requires to be plugged into the switch and, depending on
+a site's networking policy and configuration, an 802.1X authentication
+to be performed before user data can be transferred. Operational state
+shows the ability of an interface to transmit this user data.
+Thanks to 802.1X, userspace must be granted the possibility to
+influence operational state. To accommodate this, operational state is
+split into two parts: Two flags that can be set by the driver only, and
+a RFC2863 compatible state that is derived from these flags, a policy,
+and changeable from userspace under certain rules.
+2. Querying from userspace
+Both admin and operational state can be queried via the netlink
+operation RTM_GETLINK. It is also possible to subscribe to RTMGRP_LINK
+to be notified of updates. This is important for setting from userspace.
+These values contain interface state:
+ifinfomsg::if_flags & IFF_UP:
+ Interface is admin up
+ifinfomsg::if_flags & IFF_RUNNING:
+ Interface is in RFC2863 operational state UP or UNKNOWN. This is for
+ backward compatibility, routing daemons, dhcp clients can use this
+ flag to determine whether they should use the interface.
+ifinfomsg::if_flags & IFF_LOWER_UP:
+ Driver has signaled netif_carrier_on()
+ifinfomsg::if_flags & IFF_DORMANT:
+ Driver has signaled netif_dormant_on()
+These interface flags can also be queried without netlink using the
+SIOCGIFFLAGS ioctl.
+TLV IFLA_OPERSTATE
+contains RFC2863 state of the interface in numeric representation:
+IF_OPER_UNKNOWN (0):
+ Interface is in unknown state, neither driver nor userspace has set
+ operational state. Interface must be considered for user data as
+ setting operational state has not been implemented in every driver.
+IF_OPER_NOTPRESENT (1):
+ Unused in current kernel (notpresent interfaces normally disappear),
+ just a numerical placeholder.
+IF_OPER_DOWN (2):
+ Interface is unable to transfer data on L1, f.e. ethernet is not
+ plugged or interface is ADMIN down.
+IF_OPER_LOWERLAYERDOWN (3):
+ Interfaces stacked on an interface that is IF_OPER_DOWN show this
+ state (f.e. VLAN).
+IF_OPER_TESTING (4):
+ Unused in current kernel.
+IF_OPER_DORMANT (5):
+ Interface is L1 up, but waiting for an external event, f.e. for a
+ protocol to establish. (802.1X)
+IF_OPER_UP (6):
+ Interface is operational up and can be used.
+This TLV can also be queried via sysfs.
+TLV IFLA_LINKMODE
+contains link policy. This is needed for userspace interaction
+described below.
+This TLV can also be queried via sysfs.
+3. Kernel driver API
+Kernel drivers have access to two flags that map to IFF_LOWER_UP and
+IFF_DORMANT. These flags can be set from everywhere, even from
+interrupts. It is guaranteed that only the driver has write access,
+however, if different layers of the driver manipulate the same flag,
+the driver has to provide the synchronisation needed.
+__LINK_STATE_NOCARRIER, maps to !IFF_LOWER_UP:
+The driver uses netif_carrier_on() to clear and netif_carrier_off() to
+set this flag. On netif_carrier_off(), the scheduler stops sending
+packets. The name 'carrier' and the inversion are historical, think of
+it as lower layer.
+netif_carrier_ok() can be used to query that bit.
+__LINK_STATE_DORMANT, maps to IFF_DORMANT:
+Set by the driver to express that the device cannot yet be used
+because some driver controlled protocol establishment has to
+complete. Corresponding functions are netif_dormant_on() to set the
+flag, netif_dormant_off() to clear it and netif_dormant() to query.
+On device allocation, networking core sets the flags equivalent to
+netif_carrier_ok() and !netif_dormant().
+Whenever the driver CHANGES one of these flags, a workqueue event is
+scheduled to translate the flag combination to IFLA_OPERSTATE as
+follows:
+!netif_carrier_ok():
+ IF_OPER_LOWERLAYERDOWN if the interface is stacked, IF_OPER_DOWN
+ otherwise. Kernel can recognise stacked interfaces because their
+ ifindex != iflink.
+netif_carrier_ok() && netif_dormant():
+ IF_OPER_DORMANT
+netif_carrier_ok() && !netif_dormant():
+ IF_OPER_UP if userspace interaction is disabled. Otherwise
+ IF_OPER_DORMANT with the possibility for userspace to initiate the
+ IF_OPER_UP transition afterwards.
+4. Setting from userspace
+Applications have to use the netlink interface to influence the
+RFC2863 operational state of an interface. Setting IFLA_LINKMODE to 1
+via RTM_SETLINK instructs the kernel that an interface should go to
+IF_OPER_DORMANT instead of IF_OPER_UP when the combination
+netif_carrier_ok() && !netif_dormant() is set by the
+driver. Afterwards, the userspace application can set IFLA_OPERSTATE
+to IF_OPER_DORMANT or IF_OPER_UP as long as the driver does not set
+netif_carrier_off() or netif_dormant_on(). Changes made by userspace
+are multicasted on the netlink group RTMGRP_LINK.
+So basically a 802.1X supplicant interacts with the kernel like this:
+-subscribe to RTMGRP_LINK
+-set IFLA_LINKMODE to 1 via RTM_SETLINK
+-query RTM_GETLINK once to get initial state
+-if initial flags are not (IFF_LOWER_UP && !IFF_DORMANT), wait until
+ netlink multicast signals this state
+-do 802.1X, eventually abort if flags go down again
+-send RTM_SETLINK to set operstate to IF_OPER_UP if authentication
+ succeeds, IF_OPER_DORMANT otherwise
+-see how operstate and IFF_RUNNING is echoed via netlink multicast
+-set interface back to IF_OPER_DORMANT if 802.1X reauthentication
+ fails
+-restart if kernel changes IFF_LOWER_UP or IFF_DORMANT flag
+if supplicant goes down, bring back IFLA_LINKMODE to 0 and
+IFLA_OPERSTATE to a sane value.
+A routing daemon or dhcp client just needs to care for IFF_RUNNING or
+waiting for operstate to go IF_OPER_UP/IF_OPER_UNKNOWN before
+considering the interface / querying a DHCP address.
+For technical questions and/or comments please e-mail to Stefan Rompf
+(stefan at loplof.de).
diff --git a/Documentation/networking/packet_mmap.txt b/Documentation/networking/packet_mmap.txt
index 4fc8e9874320..aaf99d5f0dad 100644
--- a/Documentation/networking/packet_mmap.txt
+++ b/Documentation/networking/packet_mmap.txt
@@ -254,7 +254,7 @@ and, the number of frames be
        <block number> * <block size> / <frame size>
-Suposse the following parameters, which apply for 2.6 kernel and an
+Suppose the following parameters, which apply for 2.6 kernel and an
 i386 architecture:
        <size-max> = 131072 bytes
diff --git a/Documentation/networking/tuntap.txt b/Documentation/networking/tuntap.txt
index ec3d109d787a..76750fb9151a 100644
--- a/Documentation/networking/tuntap.txt
+++ b/Documentation/networking/tuntap.txt
@@ -138,7 +138,7 @@ This means that you have to read/write IP packets when you are using tun and
 ethernet frames when using tap.
 5. What is the difference between BPF and TUN/TAP driver?
-BFP is an advanced packet filter. It can be attached to existing
+BPF is an advanced packet filter. It can be attached to existing
 network interface. It does not provide a virtual network interface.
 A TUN/TAP driver does provide a virtual network interface and it is possible
 to attach BPF to this interface.
diff --git a/Documentation/networking/xfrm_sync.txt b/Documentation/networking/xfrm_sync.txt
new file mode 100644
index 000000000000..8be626f7c0b8
--- /dev/null
+++ b/Documentation/networking/xfrm_sync.txt
@@ -0,0 +1,166 @@
+The sync patches work is based on initial patches from
+Krisztian <hidden@balabit.hu> and others and additional patches
+from Jamal <hadi@cyberus.ca>.
+The end goal for syncing is to be able to insert attributes + generate
+events so that the an SA can be safely moved from one machine to another
+for HA purposes.
+The idea is to synchronize the SA so that the takeover machine can do
+the processing of the SA as accurate as possible if it has access to it.
+We already have the ability to generate SA add/del/upd events.
+These patches add ability to sync and have accurate lifetime byte (to
+ensure proper decay of SAs) and replay counters to avoid replay attacks
+with as minimal loss at failover time.
+This way a backup stays as closely uptodate as an active member.
+Because the above items change for every packet the SA receives,
+it is possible for a lot of the events to be generated.
+For this reason, we also add a nagle-like algorithm to restrict
+the events. i.e we are going to set thresholds to say "let me
+know if the replay sequence threshold is reached or 10 secs have passed"
+These thresholds are set system-wide via sysctls or can be updated
+per SA.
+The identified items that need to be synchronized are:
+- the lifetime byte counter
+note that: lifetime time limit is not important if you assume the failover
+machine is known ahead of time since the decay of the time countdown
+is not driven by packet arrival.
+- the replay sequence for both inbound and outbound
+1) Message Structure
+----------------------
+nlmsghdr:aevent_id:optional-TLVs.
+The netlink message types are:
+XFRM_MSG_NEWAE and XFRM_MSG_GETAE.
+A XFRM_MSG_GETAE does not have TLVs.
+A XFRM_MSG_NEWAE will have at least two TLVs (as is
+discussed further below).
+aevent_id structure looks like:
+   struct xfrm_aevent_id {
+             struct xfrm_usersa_id           sa_id;
+             __u32                           flags;
+   };
+xfrm_usersa_id in this message layout identifies the SA.
+flags are used to indicate different things. The possible
+flags are:
+        XFRM_AE_RTHR=1, /* replay threshold*/
+        XFRM_AE_RVAL=2, /* replay value */
+        XFRM_AE_LVAL=4, /* lifetime value */
+        XFRM_AE_ETHR=8, /* expiry timer threshold */
+        XFRM_AE_CR=16, /* Event cause is replay update */
+        XFRM_AE_CE=32, /* Event cause is timer expiry */
+        XFRM_AE_CU=64, /* Event cause is policy update */
+How these flags are used is dependent on the direction of the
+message (kernel<->user) as well the cause (config, query or event).
+This is described below in the different messages.
+The pid will be set appropriately in netlink to recognize direction
+(0 to the kernel and pid = processid that created the event
+when going from kernel to user space)
+A program needs to subscribe to multicast group XFRMNLGRP_AEVENTS
+to get notified of these events.
+2) TLVS reflect the different parameters:
+-----------------------------------------
+a) byte value (XFRMA_LTIME_VAL)
+This TLV carries the running/current counter for byte lifetime since
+last event.
+b)replay value (XFRMA_REPLAY_VAL)
+This TLV carries the running/current counter for replay sequence since
+last event.
+c)replay threshold (XFRMA_REPLAY_THRESH)
+This TLV carries the threshold being used by the kernel to trigger events
+when the replay sequence is exceeded.
+d) expiry timer (XFRMA_ETIMER_THRESH)
+This is a timer value in milliseconds which is used as the nagle
+value to rate limit the events.
+3) Default configurations for the parameters:
+----------------------------------------------
+By default these events should be turned off unless there is
+at least one listener registered to listen to the multicast
+group XFRMNLGRP_AEVENTS.
+Programs installing SAs will need to specify the two thresholds, however,
+in order to not change existing applications such as racoon
+we also provide default threshold values for these different parameters
+in case they are not specified.
+the two sysctls/proc entries are:
+a) /proc/sys/net/core/sysctl_xfrm_aevent_etime
+used to provide default values for the XFRMA_ETIMER_THRESH in incremental
+units of time of 100ms. The default is 10 (1 second)
+b) /proc/sys/net/core/sysctl_xfrm_aevent_rseqth
+used to provide default values for XFRMA_REPLAY_THRESH parameter
+in incremental packet count. The default is two packets.
+4) Message types
+----------------
+a) XFRM_MSG_GETAE issued by user-->kernel.
+XFRM_MSG_GETAE does not carry any TLVs.
+The response is a XFRM_MSG_NEWAE which is formatted based on what
+XFRM_MSG_GETAE queried for.
+The response will always have XFRMA_LTIME_VAL and XFRMA_REPLAY_VAL TLVs.
+*if XFRM_AE_RTHR flag is set, then XFRMA_REPLAY_THRESH is also retrieved
+*if XFRM_AE_ETHR flag is set, then XFRMA_ETIMER_THRESH is also retrieved
+b) XFRM_MSG_NEWAE is issued by either user space to configure
+or kernel to announce events or respond to a XFRM_MSG_GETAE.
+i) user --> kernel to configure a specific SA.
+any of the values or threshold parameters can be updated by passing the
+appropriate TLV.
+A response is issued back to the sender in user space to indicate success
+or failure.
+In the case of success, additionally an event with
+XFRM_MSG_NEWAE is also issued to any listeners as described in iii).
+ii) kernel->user direction as a response to XFRM_MSG_GETAE
+The response will always have XFRMA_LTIME_VAL and XFRMA_REPLAY_VAL TLVs.
+The threshold TLVs will be included if explicitly requested in
+the XFRM_MSG_GETAE message.
+iii) kernel->user to report as event if someone sets any values or
+thresholds for an SA using XFRM_MSG_NEWAE (as described in #i above).
+In such a case XFRM_AE_CU flag is set to inform the user that
+the change happened as a result of an update.
+The message will always have XFRMA_LTIME_VAL and XFRMA_REPLAY_VAL TLVs.
+iv) kernel->user to report event when replay threshold or a timeout
+is exceeded.
+In such a case either XFRM_AE_CR (replay exceeded) or XFRM_AE_CE (timeout
+happened) is set to inform the user what happened.
+Note the two flags are mutually exclusive.
+The message will always have XFRMA_LTIME_VAL and XFRMA_REPLAY_VAL TLVs.
+Exceptions to threshold settings
+--------------------------------
+If you have an SA that is getting hit by traffic in bursts such that
+there is a period where the timer threshold expires with no packets
+seen, then an odd behavior is seen as follows:
+The first packet arrival after a timer expiry will trigger a timeout
+aevent; i.e we dont wait for a timeout period or a packet threshold
+to be reached. This is done for simplicity and efficiency reasons.
+-JHS
diff --git a/Documentation/pci.txt b/Documentation/pci.txt
index 711210b38f5f..66bbbf1d1ef6 100644
--- a/Documentation/pci.txt
+++ b/Documentation/pci.txt
@@ -259,7 +259,17 @@ on the bus need to be capable of doing it, so this is something which needs
 to be handled by platform and generic code, not individual drivers.
-8. Obsolete functions
+8. Vendor and device identifications
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+For the future, let's avoid adding device ids to include/linux/pci_ids.h.
+PCI_VENDOR_ID_xxx for vendors, and a hex constant for device ids.
+Rationale:  PCI_VENDOR_ID_xxx constants are re-used, but device ids are not.
+    Further, device ids are arbitrary hex numbers, normally used only in a
+    single location, the pci_device_id table.
+9. Obsolete functions
 ~~~~~~~~~~~~~~~~~~~~~
 There are several functions which you might come across when trying to
 port an old driver to the new PCI interface.  They are no longer present
diff --git a/Documentation/pcmcia/driver-changes.txt b/Documentation/pcmcia/driver-changes.txt
index 97420f08c786..4739c5c3face 100644
--- a/Documentation/pcmcia/driver-changes.txt
+++ b/Documentation/pcmcia/driver-changes.txt
@@ -1,5 +1,11 @@
 This file details changes in 2.6 which affect PCMCIA card driver authors:
+* New release helper (as of 2.6.17)
+   Instead of calling pcmcia_release_{configuration,io,irq,win}, all that's
+   necessary now is calling pcmcia_disable_device. As there is no valid
+   reason left to call pcmcia_release_io and pcmcia_release_irq, the
+   exports for them were removed.
 * Unify detach and REMOVAL event code, as well as attach and INSERTION
  code (as of 2.6.16)
       void (*remove)          (struct pcmcia_device *dev);
diff --git a/Documentation/power/video.txt b/Documentation/power/video.txt
index d18a57d1a531..43a889f8f08d 100644
--- a/Documentation/power/video.txt
+++ b/Documentation/power/video.txt
@@ -140,7 +140,7 @@ IBM TP T41p			s3_bios (2), switch to X after resume
 IBM TP T42                      s3_bios (2)
 IBM ThinkPad T42p (2373-GTG)    s3_bios (2)
 IBM TP X20                      ??? (*)
-IBM TP X30                      s3_bios (2)
+IBM TP X30                      s3_bios, s3_mode (4)
 IBM TP X31 / Type 2672-XXH      none (1), use radeontool (http://fdd.com/software/radeon/) to turn off backlight.
 IBM TP X32                      none (1), but backlight is on and video is trashed after long suspend. s3_bios,s3_mode (4) works too. Perhaps that gets better results?
 IBM Thinkpad X40 Type 2371-7JG  s3_bios,s3_mode (4)
diff --git a/Documentation/scsi/ChangeLog.megaraid b/Documentation/scsi/ChangeLog.megaraid
index 09f6300eda4b..c173806c91fa 100644
--- a/Documentation/scsi/ChangeLog.megaraid
+++ b/Documentation/scsi/ChangeLog.megaraid
@@ -1,3 +1,28 @@
+Release Date    : Mon Apr 11 12:27:22 EST 2006 - Seokmann Ju <sju@lsil.com>
+Current Version : 2.20.4.8 (scsi module), 2.20.2.6 (cmm module)
+Older Version   : 2.20.4.7 (scsi module), 2.20.2.6 (cmm module)
+1.      Fixed a bug in megaraid_reset_handler().
+        Customer reported "Unable to handle kernel NULL pointer dereference
+        at virtual address 00000000" when system goes to reset condition
+        for some reason. It happened randomly.
+        Root Cause: in the megaraid_reset_handler(), there is possibility not
+        returning pending packets in the pend_list if there are multiple
+        pending packets.
+        Fix: Made the change in the driver so that it will return all packets
+        in the pend_list.
+2.      Added change request.
+        As found in the following URL, rmb() only didn't help the
+        problem. I had to increase the loop counter to 0xFFFFFF. (6 F's)
+        http://marc.theaimsgroup.com/?l=linux-scsi&m=110971060502497&w=2
+        I attached a patch for your reference, too.
+        Could you check and get this fix in your driver?
+        Best Regards,
+        Jun'ichi Nomura
 Release Date    : Fri Nov 11 12:27:22 EST 2005 - Seokmann Ju <sju@lsil.com>
 Current Version : 2.20.4.7 (scsi module), 2.20.2.6 (cmm module)
 Older Version   : 2.20.4.6 (scsi module), 2.20.2.6 (cmm module)
diff --git a/Documentation/scsi/scsi_eh.txt b/Documentation/scsi/scsi_eh.txt
index 331afd791cbb..ce767b90bb0d 100644
--- a/Documentation/scsi/scsi_eh.txt
+++ b/Documentation/scsi/scsi_eh.txt
@@ -19,9 +19,9 @@ TABLE OF CONTENTS
        [2-1-1] Overview
        [2-1-2] Flow of scmds through EH
        [2-1-3] Flow of control
-    [2-2] EH through hostt->eh_strategy_handler()
+    [2-2] EH through transportt->eh_strategy_handler()
-        [2-2-1] Pre hostt->eh_strategy_handler() SCSI midlayer conditions
+        [2-2-1] Pre transportt->eh_strategy_handler() SCSI midlayer conditions
-        [2-2-2] Post hostt->eh_strategy_handler() SCSI midlayer conditions
+        [2-2-2] Post transportt->eh_strategy_handler() SCSI midlayer conditions
        [2-2-3] Things to consider
@@ -413,9 +413,9 @@ scmd->allowed.
            layer of failure of the scmds.
-[2-2] EH through hostt->eh_strategy_handler()
+[2-2] EH through transportt->eh_strategy_handler()
- hostt->eh_strategy_handler() is invoked in the place of
+ transportt->eh_strategy_handler() is invoked in the place of
 scsi_unjam_host() and it is responsible for whole recovery process.
 On completion, the handler should have made lower layers forget about
 all failed scmds and either ready for new commands or offline.  Also,
@@ -424,7 +424,7 @@ SCSI midlayer.  IOW, of the steps described in [2-1-2], all steps
 except for #1 must be implemented by eh_strategy_handler().
-[2-2-1] Pre hostt->eh_strategy_handler() SCSI midlayer conditions
+[2-2-1] Pre transportt->eh_strategy_handler() SCSI midlayer conditions
 The following conditions are true on entry to the handler.
@@ -437,7 +437,7 @@ except for #1 must be implemented by eh_strategy_handler().
 - shost->host_failed == shost->host_busy
-[2-2-2] Post hostt->eh_strategy_handler() SCSI midlayer conditions
+[2-2-2] Post transportt->eh_strategy_handler() SCSI midlayer conditions
 The following conditions must be true on exit from the handler.
diff --git a/Documentation/scsi/scsi_mid_low_api.txt b/Documentation/scsi/scsi_mid_low_api.txt
index 8bbae3e1abdf..75a535a975c3 100644
--- a/Documentation/scsi/scsi_mid_low_api.txt
+++ b/Documentation/scsi/scsi_mid_low_api.txt
@@ -804,7 +804,6 @@ Summary:
   eh_bus_reset_handler - issue SCSI bus reset
   eh_device_reset_handler - issue SCSI device reset
   eh_host_reset_handler - reset host (host bus adapter)
-   eh_strategy_handler - driver supplied alternate to scsi_unjam_host()
   info - supply information about given host
   ioctl - driver can respond to ioctls
   proc_info - supports /proc/scsi/{driver_name}/{host_no}
@@ -970,24 +969,6 @@ Details:
 /**
- *      eh_strategy_handler - driver supplied alternate to scsi_unjam_host()
- *      @shp: host on which error has occurred
- *
- *      Returns TRUE if host unjammed, else FALSE.
- *
- *      Locks: none
- *
- *      Calling context: kernel thread
- *
- *      Notes: Invoked from scsi_eh thread. LLD supplied alternate to 
- *      scsi_unjam_host() found in scsi_error.c
- *
- *      Optionally defined in: LLD
- **/
-     int eh_strategy_handler(struct Scsi_Host * shp)
-/**
 *      info - supply information about given host: driver name plus data
 *             to distinguish given host
 *      @shp: host to supply information about
diff --git a/Documentation/serial/driver b/Documentation/serial/driver
index 42ef9970bc86..88ad615dd338 100644
--- a/Documentation/serial/driver
+++ b/Documentation/serial/driver
@@ -3,14 +3,11 @@
                        --------------------
-   $Id: driver,v 1.10 2002/07/22 15:27:30 rmk Exp $
 This document is meant as a brief overview of some aspects of the new serial
 driver.  It is not complete, any questions you have should be directed to
 <rmk@arm.linux.org.uk>
-The reference implementation is contained within serial_amba.c.
+The reference implementation is contained within amba_pl011.c.
@@ -31,6 +28,11 @@ The serial core provides a few helper functions.  This includes identifing
 the correct port structure (via uart_get_console) and decoding command line
 arguments (uart_parse_options).
+There is also a helper function (uart_write_console) which performs a
+character by character write, translating newlines to CRLF sequences.
+Driver writers are recommended to use this function rather than implementing
+their own version.
 Locking
 -------
@@ -86,6 +88,7 @@ hardware.
                - TIOCM_DTR     DTR signal.
                - TIOCM_OUT1    OUT1 signal.
                - TIOCM_OUT2    OUT2 signal.
+                - TIOCM_LOOP    Set the port into loopback mode.
        If the appropriate bit is set, the signal should be driven
        active.  If the bit is clear, the signal should be driven
        inactive.
@@ -141,6 +144,10 @@ hardware.
  enable_ms(port)
        Enable the modem status interrupts.
+        This method may be called multiple times.  Modem status
+        interrupts should be disabled when the shutdown method is
+        called.
        Locking: port->lock taken.
        Interrupts: locally disabled.
        This call must not sleep
@@ -160,6 +167,8 @@ hardware.
        state.  Enable the port for reception.  It should not activate
        RTS nor DTR; this will be done via a separate call to set_mctrl.
+        This method will only be called when the port is initially opened.
        Locking: port_sem taken.
        Interrupts: globally disabled.
@@ -169,6 +178,11 @@ hardware.
        RTS nor DTR; this will have already been done via a separate
        call to set_mctrl.
+        Drivers must not access port->info once this call has completed.
+        This method will only be called when there are no more users of
+        this port.
        Locking: port_sem taken.
        Interrupts: caller dependent.
@@ -200,12 +214,13 @@ hardware.
        The interaction of the iflag bits is as follows (parity error
        given as an example):
        Parity error    INPCK   IGNPAR
-        None            n/a     n/a     character received
+        n/a             0       n/a     character received, marked as
-        Yes             n/a     0       character discarded
+                                        TTY_NORMAL
-        Yes             0       1       character received, marked as
+        None            1       n/a     character received, marked as
                                        TTY_NORMAL
-        Yes             1       1       character received, marked as
+        Yes             1       0       character received, marked as
                                        TTY_PARITY
+        Yes             1       1       character discarded
        Other flags may be used (eg, xon/xoff characters) if your
        hardware supports hardware "soft" flow control.
diff --git a/Documentation/sound/alsa/ALSA-Configuration.txt b/Documentation/sound/alsa/ALSA-Configuration.txt
index 1def6049784c..0ee2c7dfc482 100644
--- a/Documentation/sound/alsa/ALSA-Configuration.txt
+++ b/Documentation/sound/alsa/ALSA-Configuration.txt
@@ -120,6 +120,34 @@ Prior to version 0.9.0rc4 options had a 'snd_' prefix. This was removed.
    enable      - enable card
                - Default: enabled, for PCI and ISA PnP cards
+  Module snd-adlib
+  ----------------
+    Module for AdLib FM cards.
+    port        - port # for OPL chip
+    This module supports multiple cards. It does not support autoprobe, so
+    the port must be specified. For actual AdLib FM cards it will be 0x388.
+    Note that this card does not have PCM support and no mixer; only FM
+    synthesis.
+    Make sure you have "sbiload" from the alsa-tools package available and,
+    after loading the module, find out the assigned ALSA sequencer port
+    number through "sbiload -l". Example output:
+      Port     Client name                       Port name
+      64:0     OPL2 FM synth                     OPL2 FM Port
+    Load the std.sb and drums.sb patches also supplied by sbiload:
+      sbiload -p 64:0 std.sb drums.sb
+    If you use this driver to drive an OPL3, you can use std.o3 and drums.o3
+    instead. To have the card produce sound, use aplaymidi from alsa-utils:
+      aplaymidi -p 64:0 foo.mid
  Module snd-ad1816a
  ------------------
@@ -190,6 +218,15 @@ Prior to version 0.9.0rc4 options had a 'snd_' prefix. This was removed.
    The power-management is supported.
+  Module snd-als300
+  -----------------
+    Module for  Avance Logic ALS300 and ALS300+
+    This module supports multiple cards.
+    The power-management is supported.
  Module snd-als4000
  ------------------
@@ -701,6 +738,7 @@ Prior to version 0.9.0rc4 options had a 'snd_' prefix. This was removed.
          uniwill       3-jack
          F1734         2-jack
          lg            LG laptop (m1 express dual)
+          lg-lw         LG LW20 laptop
          test          for testing/debugging purpose, almost all controls can be
                        adjusted.  Appearing only when compiled with
                        $CONFIG_SND_DEBUG=y
@@ -1013,6 +1051,23 @@ Prior to version 0.9.0rc4 options had a 'snd_' prefix. This was removed.
    The power-management is supported.
+  Module snd-miro
+  ---------------
+    Module for Miro soundcards: miroSOUND PCM 1 pro, 
+                                miroSOUND PCM 12,
+                                miroSOUND PCM 20 Radio.
+    port        - Port # (0x530,0x604,0xe80,0xf40)
+    irq         - IRQ # (5,7,9,10,11)
+    dma1        - 1st dma # (0,1,3)
+    dma2        - 2nd dma # (0,1)
+    mpu_port    - MPU-401 port # (0x300,0x310,0x320,0x330)
+    mpu_irq     - MPU-401 irq # (5,7,9,10)
+    fm_port     - FM Port # (0x388)
+    wss         - enable WSS mode
+    ide         - enable onboard ide support
  Module snd-mixart
  -----------------
@@ -1202,6 +1257,20 @@ Prior to version 0.9.0rc4 options had a 'snd_' prefix. This was removed.
    The power-management is supported.
+  Module snd-riptide
+  ------------------
+    Module for Conexant Riptide chip
+      joystick_port     - Joystick port # (default: 0x200)
+      mpu_port          - MPU401 port # (default: 0x330)
+      opl3_port         - OPL3 port # (default: 0x388)
+    This module supports multiple cards.
+    The driver requires the firmware loader support on kernel.
+    You need to install the firmware file "riptide.hex" to the standard
+    firmware path (e.g. /lib/firmware).
  Module snd-rme32
  ----------------
diff --git a/Documentation/sound/alsa/Audiophile-Usb.txt b/Documentation/sound/alsa/Audiophile-Usb.txt
index 4692c8e77dc1..b535c2a198f8 100644
--- a/Documentation/sound/alsa/Audiophile-Usb.txt
+++ b/Documentation/sound/alsa/Audiophile-Usb.txt
@@ -1,4 +1,4 @@
-        Guide to using M-Audio Audiophile USB with ALSA and Jack        v1.2
+        Guide to using M-Audio Audiophile USB with ALSA and Jack        v1.3
        ========================================================
            Thibault Le Meur <Thibault.LeMeur@supelec.fr>
@@ -22,16 +22,16 @@ The device has 4 audio interfaces, and 2 MIDI ports:
 * Midi In (Mi)
 * Midi Out (Mo)
-The internal DAC/ADC has the following caracteristics:
+The internal DAC/ADC has the following characteristics:
 * sample depth of 16 or 24 bits
 * sample rate from 8kHz to 96kHz
-* Two ports can't use different sample depths at the same time.Moreover, the 
+* Two ports can't use different sample depths at the same time. Moreover, the 
 Audiophile USB documentation gives the following Warning: "Please exit any 
 audio application running before switching between bit depths"
 Due to the USB 1.1 bandwidth limitation, a limited number of interfaces can be 
 activated at the same time depending on the audio mode selected:
- * 16-bit/48kHz ==> 4 channels in/ 4 channels out
+ * 16-bit/48kHz ==> 4 channels in/4 channels out
   - Ai+Ao+Di+Do
 * 24-bit/48kHz ==> 4 channels in/2 channels out, 
                    or 2 channels in/4 channels out
@@ -41,8 +41,8 @@ activated at the same time depending on the audio mode selected:
 Important facts about the Digital interface:
 --------------------------------------------
- * The Do port additionnaly supports surround-encoded AC-3 and DTS passthrough, 
+ * The Do port additionally supports surround-encoded AC-3 and DTS passthrough, 
-though I haven't tested it under linux
+though I haven't tested it under Linux
   - Note that in this setup only the Do interface can be enabled
 * Apart from recording an audio digital stream, enabling the Di port is a way 
 to synchronize the device to an external sample clock
@@ -60,24 +60,23 @@ synchronization error (for instance sound played at an odd sample rate)
 The Audiophile USB MIDI ports will be automatically supported once the 
 following modules have been loaded:
 * snd-usb-audio
- * snd-seq
 * snd-seq-midi
-No additionnal setting is required.
+No additional setting is required.
 2.2 - Audio ports
 -----------------
 Audio functions of the Audiophile USB device are handled by the snd-usb-audio 
 module. This module can work in a default mode (without any device-specific 
-parameter), or in an advanced mode with the device-specific parameter called 
+parameter), or in an "advanced" mode with the device-specific parameter called 
 "device_setup".
 2.2.1 - Default Alsa driver mode
-The default behaviour of the snd-usb-audio driver is to parse the device 
+The default behavior of the snd-usb-audio driver is to parse the device 
 capabilities at startup and enable all functions inside the device (including 
-all ports at any sample rates and any sample depths supported). This approach 
+all ports at any supported sample rates and sample depths). This approach 
 has the advantage to let the driver easily switch from sample rates/depths 
 automatically according to the need of the application claiming the device.
@@ -114,9 +113,9 @@ gain).
 For people having this problem, the snd-usb-audio module has a new module 
 parameter called "device_setup".
-2.2.2.1 - Initializing the working mode of the Audiohile USB
+2.2.2.1 - Initializing the working mode of the Audiophile USB
-As far as the Audiohile USB device is concerned, this value let the user 
+As far as the Audiophile USB device is concerned, this value let the user 
 specify:
 * the sample depth
 * the sample rate
@@ -174,20 +173,20 @@ The parameter can be given:
 IMPORTANT NOTE WHEN SWITCHING CONFIGURATION:
 -------------------------------------------
- * You may need to _first_ intialize the module with the correct device_setup 
+ * You may need to _first_ initialize the module with the correct device_setup 
   parameter and _only_after_ turn on the Audiophile USB device
 * This is especially true when switching the sample depth:
-   - first trun off the device
+   - first turn off the device
-   - de-register the snd-usb-audio module
+   - de-register the snd-usb-audio module (modprobe -r)
-   - change the device_setup parameter (by either manually reprobing the module 
+   - change the device_setup parameter by changing the device_setup
-     or changing modprobe.conf)
+     option in /etc/modprobe.conf 
   - turn on the device
 2.2.2.3 - Audiophile USB's device_setup structure
 If you want to understand the device_setup magic numbers for the Audiophile 
 USB, you need some very basic understanding of binary computation. However, 
-this is not required to use the parameter and you may skip thi section.
+this is not required to use the parameter and you may skip this section.
 The device_setup is one byte long and its structure is the following:
@@ -231,11 +230,11 @@ Caution:
 2.2.3 -  USB implementation details for this device
-You may safely skip this section if you're not interrested in driver 
+You may safely skip this section if you're not interested in driver 
 development.
-This section describes some internals aspect of the device and summarize the 
+This section describes some internal aspects of the device and summarize the 
-data I got by usb-snooping the windows and linux drivers.
+data I got by usb-snooping the windows and Linux drivers.
 The M-Audio Audiophile USB has 7 USB Interfaces:
 a "USB interface":
@@ -277,9 +276,9 @@ Here is a short description of the AltSettings capabilities:
  - 16-bit depth, 8-48kHz sample mode
  - Synch playback (Do), audio format type III IEC1937_AC-3
-In order to ensure a correct intialization of the device, the driver 
+In order to ensure a correct initialization of the device, the driver 
 _must_know_ how the device will be used:
- * if DTS is choosen, only Interface 2 with AltSet nb.6 must be
+ * if DTS is chosen, only Interface 2 with AltSet nb.6 must be
   registered
 * if 96KHz only AltSets nb.1 of each interface must be selected
 * if samples are using 24bits/48KHz then AltSet 2 must me used if
@@ -290,7 +289,7 @@ _must_know_ how the device will be used:
   is not connected
 When device_setup is given as a parameter to the snd-usb-audio module, the 
-parse_audio_enpoint function uses a quirk called 
+parse_audio_endpoints function uses a quirk called 
 "audiophile_skip_setting_quirk" in order to prevent AltSettings not 
 corresponding to device_setup from being registered in the driver.
@@ -317,9 +316,8 @@ However you may see the following warning message:
 using the "default" ALSA device. This is less efficient than it could be. 
 Consider using a hardware device instead rather than using the plug layer."
 3.2 - Patching alsa to use direct pcm device
-------------------------------------------
+--------------------------------------------
 A patch for Jack by Andreas Steinmetz adds support for Big Endian devices. 
 However it has not been included in the CVS tree.
@@ -331,3 +329,32 @@ After having applied the patch you can run jackd with the following command
 line:
  % jackd -R -dalsa -Phw:1,0 -r48000 -p128 -n2 -D -Chw:1,1
+3.2 - Getting 2 input and/or output interfaces in Jack
+------------------------------------------------------
+As you can see, starting the Jack server this way will only enable 1 stereo
+input (Di or Ai) and 1 stereo output (Ao or Do).
+This is due to the following restrictions:
+* Jack can only open one capture device and one playback device at a time
+* The Audiophile USB is seen as 2 (or three) Alsa devices: hw:1,0, hw:1,1
+  (and optionally hw:1,2)
+If you want to get Ai+Di and/or Ao+Do support with Jack, you would need to
+combine the Alsa devices into one logical "complex" device.
+If you want to give it a try, I recommend reading the information from
+this page: http://www.sound-man.co.uk/linuxaudio/ice1712multi.html
+It is related to another device (ice1712) but can be adapted to suit
+the Audiophile USB.
+Enabling multiple Audiophile USB interfaces for Jackd will certainly require:
+* patching Jack with the previously mentioned "Big Endian" patch
+* patching Jackd with the MMAP_COMPLEX patch (see the ice1712 page)
+* patching the alsa-lib/src/pcm/pcm_multi.c file (see the ice1712 page)
+* define a multi device (combination of hw:1,0 and hw:1,1) in your .asoundrc
+  file 
+* start jackd with this device
+I had no success in testing this for now, but this may be due to my OS
+configuration. If you have any success with this kind of setup, please
+drop me an email.
diff --git a/Documentation/sound/alsa/DocBook/writing-an-alsa-driver.tmpl b/Documentation/sound/alsa/DocBook/writing-an-alsa-driver.tmpl
index 6feef9e82b63..1faf76383bab 100644
--- a/Documentation/sound/alsa/DocBook/writing-an-alsa-driver.tmpl
+++ b/Documentation/sound/alsa/DocBook/writing-an-alsa-driver.tmpl
@@ -1123,8 +1123,8 @@
          if ((err = pci_enable_device(pci)) < 0)
                  return err;
          /* check PCI availability (28bit DMA) */
-          if (pci_set_dma_mask(pci, 0x0fffffff) < 0 ||
+          if (pci_set_dma_mask(pci, DMA_28BIT_MASK) < 0 ||
-              pci_set_consistent_dma_mask(pci, 0x0fffffff) < 0) {
+              pci_set_consistent_dma_mask(pci, DMA_28BIT_MASK) < 0) {
                  printk(KERN_ERR "error to set 28bit mask DMA\n");
                  pci_disable_device(pci);
                  return -ENXIO;
@@ -1172,7 +1172,7 @@
  }        
  /* PCI IDs */
-  static struct pci_device_id snd_mychip_ids[] = {
+  static struct pci_device_id snd_mychip_ids[] __devinitdata = {
          { PCI_VENDOR_ID_FOO, PCI_DEVICE_ID_BAR,
            PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0, },
          ....
@@ -1216,7 +1216,7 @@
        The allocation of PCI resources is done in the
      <function>probe()</function> function, and usually an extra
      <function>xxx_create()</function> function is written for this
-      purpose. 
+      purpose.
      </para>
      <para>
@@ -1225,7 +1225,7 @@
      allocating resources. Also, you need to set the proper PCI DMA
      mask to limit the accessed i/o range. In some cases, you might
      need to call <function>pci_set_master()</function> function,
-      too. 
+      too.
      </para>
      <para>
@@ -1236,8 +1236,8 @@
 <![CDATA[
  if ((err = pci_enable_device(pci)) < 0)
          return err;
-  if (pci_set_dma_mask(pci, 0x0fffffff) < 0 ||
+  if (pci_set_dma_mask(pci, DMA_28BIT_MASK) < 0 ||
-      pci_set_consistent_dma_mask(pci, 0x0fffffff) < 0) {
+      pci_set_consistent_dma_mask(pci, DMA_28BIT_MASK) < 0) {
          printk(KERN_ERR "error to set 28bit mask DMA\n");
          pci_disable_device(pci);
          return -ENXIO;
@@ -1256,13 +1256,13 @@
      functions. Unlike ALSA ver.0.5.x., there are no helpers for
      that. And these resources must be released in the destructor
      function (see below). Also, on ALSA 0.9.x, you don't need to
-      allocate (pseudo-)DMA for PCI like ALSA 0.5.x. 
+      allocate (pseudo-)DMA for PCI like ALSA 0.5.x.
      </para>
      <para>
        Now assume that this PCI device has an I/O port with 8 bytes
        and an interrupt. Then struct <structname>mychip</structname> will have the
-        following fields: 
+        following fields:
        <informalexample>
          <programlisting>
@@ -1565,7 +1565,7 @@
        <informalexample>
          <programlisting>
 <![CDATA[
-  static struct pci_device_id snd_mychip_ids[] = {
+  static struct pci_device_id snd_mychip_ids[] __devinitdata = {
          { PCI_VENDOR_ID_FOO, PCI_DEVICE_ID_BAR,
            PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0, },
          ....
diff --git a/Documentation/spi/pxa2xx b/Documentation/spi/pxa2xx
new file mode 100644
index 000000000000..9c45f3df2e18
--- /dev/null
+++ b/Documentation/spi/pxa2xx
@@ -0,0 +1,234 @@
+PXA2xx SPI on SSP driver HOWTO
+===================================================
+This a mini howto on the pxa2xx_spi driver.  The driver turns a PXA2xx
+synchronous serial port into a SPI master controller
+(see Documentation/spi/spi_summary). The driver has the following features
+- Support for any PXA2xx SSP
+- SSP PIO and SSP DMA data transfers.
+- External and Internal (SSPFRM) chip selects.
+- Per slave device (chip) configuration.
+- Full suspend, freeze, resume support.
+The driver is built around a "spi_message" fifo serviced by workqueue and a
+tasklet. The workqueue, "pump_messages", drives message fifo and the tasklet
+(pump_transfer) is responsible for queuing SPI transactions and setting up and
+launching the dma/interrupt driven transfers.
+Declaring PXA2xx Master Controllers
+-----------------------------------
+Typically a SPI master is defined in the arch/.../mach-*/board-*.c as a
+"platform device".  The master configuration is passed to the driver via a table
+found in include/asm-arm/arch-pxa/pxa2xx_spi.h:
+struct pxa2xx_spi_master {
+        enum pxa_ssp_type ssp_type;
+        u32 clock_enable;
+        u16 num_chipselect;
+        u8 enable_dma;
+};
+The "pxa2xx_spi_master.ssp_type" field must have a value between 1 and 3 and
+informs the driver which features a particular SSP supports.
+The "pxa2xx_spi_master.clock_enable" field is used to enable/disable the
+corresponding SSP peripheral block in the "Clock Enable Register (CKEN"). See
+the "PXA2xx Developer Manual" section "Clocks and Power Management".
+The "pxa2xx_spi_master.num_chipselect" field is used to determine the number of
+slave device (chips) attached to this SPI master.
+The "pxa2xx_spi_master.enable_dma" field informs the driver that SSP DMA should
+be used.  This caused the driver to acquire two DMA channels: rx_channel and
+tx_channel.  The rx_channel has a higher DMA service priority the tx_channel.
+See the "PXA2xx Developer Manual" section "DMA Controller".
+NSSP MASTER SAMPLE
+------------------
+Below is a sample configuration using the PXA255 NSSP.
+static struct resource pxa_spi_nssp_resources[] = {
+        [0] = {
+                .start  = __PREG(SSCR0_P(2)), /* Start address of NSSP */
+                .end    = __PREG(SSCR0_P(2)) + 0x2c, /* Range of registers */
+                .flags  = IORESOURCE_MEM,
+        },
+        [1] = {
+                .start  = IRQ_NSSP, /* NSSP IRQ */
+                .end    = IRQ_NSSP,
+                .flags  = IORESOURCE_IRQ,
+        },
+};
+static struct pxa2xx_spi_master pxa_nssp_master_info = {
+        .ssp_type = PXA25x_NSSP, /* Type of SSP */
+        .clock_enable = CKEN9_NSSP, /* NSSP Peripheral clock */
+        .num_chipselect = 1, /* Matches the number of chips attached to NSSP */
+        .enable_dma = 1, /* Enables NSSP DMA */
+};
+static struct platform_device pxa_spi_nssp = {
+        .name = "pxa2xx-spi", /* MUST BE THIS VALUE, so device match driver */
+        .id = 2, /* Bus number, MUST MATCH SSP number 1..n */
+        .resource = pxa_spi_nssp_resources,
+        .num_resources = ARRAY_SIZE(pxa_spi_nssp_resources),
+        .dev = {
+                .platform_data = &pxa_nssp_master_info, /* Passed to driver */
+        },
+};
+static struct platform_device *devices[] __initdata = {
+        &pxa_spi_nssp,
+};
+static void __init board_init(void)
+{
+        (void)platform_add_device(devices, ARRAY_SIZE(devices));
+}
+Declaring Slave Devices
+-----------------------
+Typically each SPI slave (chip) is defined in the arch/.../mach-*/board-*.c
+using the "spi_board_info" structure found in "linux/spi/spi.h". See
+"Documentation/spi/spi_summary" for additional information.
+Each slave device attached to the PXA must provide slave specific configuration
+information via the structure "pxa2xx_spi_chip" found in
+"include/asm-arm/arch-pxa/pxa2xx_spi.h".  The pxa2xx_spi master controller driver
+will uses the configuration whenever the driver communicates with the slave
+device.
+struct pxa2xx_spi_chip {
+        u8 tx_threshold;
+        u8 rx_threshold;
+        u8 dma_burst_size;
+        u32 timeout_microsecs;
+        u8 enable_loopback;
+        void (*cs_control)(u32 command);
+};
+The "pxa2xx_spi_chip.tx_threshold" and "pxa2xx_spi_chip.rx_threshold" fields are
+used to configure the SSP hardware fifo.  These fields are critical to the
+performance of pxa2xx_spi driver and misconfiguration will result in rx
+fifo overruns (especially in PIO mode transfers). Good default values are
+        .tx_threshold = 12,
+        .rx_threshold = 4,
+The "pxa2xx_spi_chip.dma_burst_size" field is used to configure PXA2xx DMA
+engine and is related the "spi_device.bits_per_word" field.  Read and understand
+the PXA2xx "Developer Manual" sections on the DMA controller and SSP Controllers
+to determine the correct value. An SSP configured for byte-wide transfers would
+use a value of 8.
+The "pxa2xx_spi_chip.timeout_microsecs" fields is used to efficiently handle
+trailing bytes in the SSP receiver fifo.  The correct value for this field is
+dependent on the SPI bus speed ("spi_board_info.max_speed_hz") and the specific
+slave device.  Please note the the PXA2xx SSP 1 does not support trailing byte
+timeouts and must busy-wait any trailing bytes.
+The "pxa2xx_spi_chip.enable_loopback" field is used to place the SSP porting
+into internal loopback mode.  In this mode the SSP controller internally
+connects the SSPTX pin the the SSPRX pin.  This is useful for initial setup
+testing.
+The "pxa2xx_spi_chip.cs_control" field is used to point to a board specific
+function for asserting/deasserting a slave device chip select.  If the field is
+NULL, the pxa2xx_spi master controller driver assumes that the SSP port is
+configured to use SSPFRM instead.
+NSSP SALVE SAMPLE
+-----------------
+The pxa2xx_spi_chip structure is passed to the pxa2xx_spi driver in the
+"spi_board_info.controller_data" field. Below is a sample configuration using
+the PXA255 NSSP.
+/* Chip Select control for the CS8415A SPI slave device */
+static void cs8415a_cs_control(u32 command)
+{
+        if (command & PXA2XX_CS_ASSERT)
+                GPCR(2) = GPIO_bit(2);
+        else
+                GPSR(2) = GPIO_bit(2);
+}
+/* Chip Select control for the CS8405A SPI slave device */
+static void cs8405a_cs_control(u32 command)
+{
+        if (command & PXA2XX_CS_ASSERT)
+                GPCR(3) = GPIO_bit(3);
+        else
+                GPSR(3) = GPIO_bit(3);
+}
+static struct pxa2xx_spi_chip cs8415a_chip_info = {
+        .tx_threshold = 12, /* SSP hardward FIFO threshold */
+        .rx_threshold = 4, /* SSP hardward FIFO threshold */
+        .dma_burst_size = 8, /* Byte wide transfers used so 8 byte bursts */
+        .timeout_microsecs = 64, /* Wait at least 64usec to handle trailing */
+        .cs_control = cs8415a_cs_control, /* Use external chip select */
+};
+static struct pxa2xx_spi_chip cs8405a_chip_info = {
+        .tx_threshold = 12, /* SSP hardward FIFO threshold */
+        .rx_threshold = 4, /* SSP hardward FIFO threshold */
+        .dma_burst_size = 8, /* Byte wide transfers used so 8 byte bursts */
+        .timeout_microsecs = 64, /* Wait at least 64usec to handle trailing */
+        .cs_control = cs8405a_cs_control, /* Use external chip select */
+};
+static struct spi_board_info streetracer_spi_board_info[] __initdata = {
+        {
+                .modalias = "cs8415a", /* Name of spi_driver for this device */
+                .max_speed_hz = 3686400, /* Run SSP as fast a possbile */
+                .bus_num = 2, /* Framework bus number */
+                .chip_select = 0, /* Framework chip select */
+                .platform_data = NULL; /* No spi_driver specific config */
+                .controller_data = &cs8415a_chip_info, /* Master chip config */
+                .irq = STREETRACER_APCI_IRQ, /* Slave device interrupt */
+        },
+        {
+                .modalias = "cs8405a", /* Name of spi_driver for this device */
+                .max_speed_hz = 3686400, /* Run SSP as fast a possbile */
+                .bus_num = 2, /* Framework bus number */
+                .chip_select = 1, /* Framework chip select */
+                .controller_data = &cs8405a_chip_info, /* Master chip config */
+                .irq = STREETRACER_APCI_IRQ, /* Slave device interrupt */
+        },
+};
+static void __init streetracer_init(void)
+{
+        spi_register_board_info(streetracer_spi_board_info,
+                                ARRAY_SIZE(streetracer_spi_board_info));
+}
+DMA and PIO I/O Support
+-----------------------
+The pxa2xx_spi driver support both DMA and interrupt driven PIO message
+transfers.  The driver defaults to PIO mode and DMA transfers must enabled by
+setting the "enable_dma" flag in the "pxa2xx_spi_master" structure and and
+ensuring that the "pxa2xx_spi_chip.dma_burst_size" field is non-zero.  The DMA
+mode support both coherent and stream based DMA mappings.
+The following logic is used to determine the type of I/O to be used on
+a per "spi_transfer" basis:
+if !enable_dma or dma_burst_size == 0 then
+        always use PIO transfers
+if spi_message.is_dma_mapped and rx_dma_buf != 0 and tx_dma_buf != 0 then
+        use coherent DMA mode
+if rx_buf and tx_buf are aligned on 8 byte boundary then
+        use streaming DMA mode
+otherwise
+        use PIO transfer
+THANKS TO
+---------
+David Brownell and others for mentoring the development of this driver.
diff --git a/Documentation/spi/spi-summary b/Documentation/spi/spi-summary
index a5ffba33a351..068732d32276 100644
--- a/Documentation/spi/spi-summary
+++ b/Documentation/spi/spi-summary
@@ -414,7 +414,33 @@ to get the driver-private data allocated for that device.
 The driver will initialize the fields of that spi_master, including the
 bus number (maybe the same as the platform device ID) and three methods
 used to interact with the SPI core and SPI protocol drivers.  It will
-also initialize its own internal state.
+also initialize its own internal state.  (See below about bus numbering
+and those methods.)
+After you initialize the spi_master, then use spi_register_master() to
+publish it to the rest of the system.  At that time, device nodes for
+the controller and any predeclared spi devices will be made available,
+and the driver model core will take care of binding them to drivers.
+If you need to remove your SPI controller driver, spi_unregister_master()
+will reverse the effect of spi_register_master().
+BUS NUMBERING
+Bus numbering is important, since that's how Linux identifies a given
+SPI bus (shared SCK, MOSI, MISO).  Valid bus numbers start at zero.  On
+SOC systems, the bus numbers should match the numbers defined by the chip
+manufacturer.  For example, hardware controller SPI2 would be bus number 2,
+and spi_board_info for devices connected to it would use that number.
+If you don't have such hardware-assigned bus number, and for some reason
+you can't just assign them, then provide a negative bus number.  That will
+then be replaced by a dynamically assigned number. You'd then need to treat
+this as a non-static configuration (see above).
+SPI MASTER METHODS
    master->setup(struct spi_device *spi)
        This sets up the device clock rate, SPI mode, and word sizes.
@@ -431,6 +457,9 @@ also initialize its own internal state.
        state it dynamically associates with that device.  If you do that,
        be sure to provide the cleanup() method to free that state.
+SPI MESSAGE QUEUE
 The bulk of the driver will be managing the I/O queue fed by transfer().
 That queue could be purely conceptual.  For example, a driver used only
@@ -440,6 +469,9 @@ But the queue will probably be very real, using message->queue, PIO,
 often DMA (especially if the root filesystem is in SPI flash), and
 execution contexts like IRQ handlers, tasklets, or workqueues (such
 as keventd).  Your driver can be as fancy, or as simple, as you need.
+Such a transfer() method would normally just add the message to a
+queue, and then start some asynchronous transfer engine (unless it's
+already running).
 THANKS TO
diff --git a/Documentation/video4linux/CARDLIST.saa7134 b/Documentation/video4linux/CARDLIST.saa7134
index 8c7195455963..bca50903233f 100644
--- a/Documentation/video4linux/CARDLIST.saa7134
+++ b/Documentation/video4linux/CARDLIST.saa7134
@@ -52,7 +52,7 @@
 51 -> ProVideo PV952                           [1540:9524]
 52 -> AverMedia AverTV/305                     [1461:2108]
 53 -> ASUS TV-FM 7135                          [1043:4845]
- 54 -> LifeView FlyTV Platinum FM               [5168:0214,1489:0214]
+ 54 -> LifeView FlyTV Platinum FM / Gold        [5168:0214,1489:0214,5168:0304]
 55 -> LifeView FlyDVB-T DUO                    [5168:0306]
 56 -> Avermedia AVerTV 307                     [1461:a70a]
 57 -> Avermedia AVerTV GO 007 FM               [1461:f31f]
@@ -84,7 +84,7 @@
 83 -> Terratec Cinergy 250 PCI TV              [153b:1160]
 84 -> LifeView FlyDVB Trio                     [5168:0319]
 85 -> AverTV DVB-T 777                         [1461:2c05]
- 86 -> LifeView FlyDVB-T                        [5168:0301]
+ 86 -> LifeView FlyDVB-T / Genius VideoWonder DVB-T [5168:0301,1489:0301]
 87 -> ADS Instant TV Duo Cardbus PTV331        [0331:1421]
 88 -> Tevion/KWorld DVB-T 220RF                [17de:7201]
 89 -> ELSA EX-VISION 700TV                     [1048:226c]
@@ -92,3 +92,4 @@
 91 -> AVerMedia A169 B                         [1461:7360]
 92 -> AVerMedia A169 B1                        [1461:6360]
 93 -> Medion 7134 Bridge #2                    [16be:0005]
+ 94 -> LifeView FlyDVB-T Hybrid Cardbus         [5168:3306,5168:3502]
diff --git a/Documentation/usb/et61x251.txt b/Documentation/video4linux/et61x251.txt
index 29340282ab5f..29340282ab5f 100644
--- a/Documentation/usb/et61x251.txt
+++ b/Documentation/video4linux/et61x251.txt
diff --git a/Documentation/usb/ibmcam.txt b/Documentation/video4linux/ibmcam.txt
index c25003644131..4a40a2e99451 100644
--- a/Documentation/usb/ibmcam.txt
+++ b/Documentation/video4linux/ibmcam.txt
@@ -122,7 +122,7 @@ WHAT YOU NEED:
 - A Linux box with USB support (2.3/2.4; 2.2 w/backport may work)
 - A Video4Linux compatible frame grabber program such as xawtv.
-  
 HOW TO COMPILE THE DRIVER:
 You need to compile the driver only if you are a developer
diff --git a/Documentation/usb/ov511.txt b/Documentation/video4linux/ov511.txt
index a7fc0432bff1..142741e3c578 100644
--- a/Documentation/usb/ov511.txt
+++ b/Documentation/video4linux/ov511.txt
@@ -9,7 +9,7 @@ INTRODUCTION:
 This is a driver for the OV511, a USB-only chip used in many "webcam" devices.
 Any camera using the OV511/OV511+ and the OV6620/OV7610/20/20AE should work.
-Video capture devices that use the Philips SAA7111A decoder also work. It 
+Video capture devices that use the Philips SAA7111A decoder also work. It
 supports streaming and capture of color or monochrome video via the Video4Linux
 API. Most V4L apps are compatible with it. Most resolutions with a width and
 height that are a multiple of 8 are supported.
@@ -52,15 +52,15 @@ from it:
        chmod 666 /dev/video
        chmod 666 /dev/video0 (if necessary)
-        
 Now you are ready to run a video app! Both vidcat and xawtv work well for me
 at 640x480.
-        
 [Using vidcat:]
        vidcat -s 640x480 -p c > test.jpg
        xview test.jpg
-        
 [Using xawtv:]
 From the main xawtv directory:
@@ -70,7 +70,7 @@ From the main xawtv directory:
        make
        make install
-Now you should be able to run xawtv. Right click for the options dialog. 
+Now you should be able to run xawtv. Right click for the options dialog.
 MODULE PARAMETERS:
@@ -286,4 +286,3 @@ Randy Dunlap, and others. Big thanks to them for their pioneering work on that
 and the USB stack. Thanks to Bret Wallach for getting camera reg IO, ISOC, and
 image capture working. Thanks to Orion Sky Lawlor, Kevin Moore, and Claudio
 Matsuoka for their work as well.
diff --git a/Documentation/usb/se401.txt b/Documentation/video4linux/se401.txt
index 7b9d1c960a10..7b9d1c960a10 100644
--- a/Documentation/usb/se401.txt
+++ b/Documentation/video4linux/se401.txt
diff --git a/Documentation/usb/sn9c102.txt b/Documentation/video4linux/sn9c102.txt
index b957beae5607..142920bc011f 100644
--- a/Documentation/usb/sn9c102.txt
+++ b/Documentation/video4linux/sn9c102.txt
@@ -174,7 +174,7 @@ Module parameters are listed below:
 -------------------------------------------------------------------------------
 Name:           video_nr
 Type:           short array (min = 0, max = 64)
-Syntax:         <-1|n[,...]> 
+Syntax:         <-1|n[,...]>
 Description:    Specify V4L2 minor mode number:
                -1 = use next available
                 n = use minor number n
@@ -187,7 +187,7 @@ Default:        -1
 -------------------------------------------------------------------------------
 Name:           force_munmap
 Type:           bool array (min = 0, max = 64)
-Syntax:         <0|1[,...]> 
+Syntax:         <0|1[,...]>
 Description:    Force the application to unmap previously mapped buffer memory
                before calling any VIDIOC_S_CROP or VIDIOC_S_FMT ioctl's. Not
                all the applications support this feature. This parameter is
@@ -206,7 +206,7 @@ Default:        2
 -------------------------------------------------------------------------------
 Name:           debug
 Type:           ushort
-Syntax:         <n> 
+Syntax:         <n>
 Description:    Debugging information level, from 0 to 3:
                0 = none (use carefully)
                1 = critical errors
@@ -267,7 +267,7 @@ The sysfs interface also provides the "frame_header" entry, which exports the
 frame header of the most recent requested and captured video frame. The header
 is always 18-bytes long and is appended to every video frame by the SN9C10x
 controllers. As an example, this additional information can be used by the user
-application for implementing auto-exposure features via software. 
+application for implementing auto-exposure features via software.
 The following table describes the frame header:
@@ -441,7 +441,7 @@ blue pixels in one video frame. Each pixel is associated with a 8-bit long
 value and is disposed in memory according to the pattern shown below:
 B[0]   G[1]    B[2]    G[3]    ...   B[m-2]         G[m-1]
-G[m]   R[m+1]  G[m+2]  R[m+2]  ...   G[2m-2]        R[2m-1] 
+G[m]   R[m+1]  G[m+2]  R[m+2]  ...   G[2m-2]        R[2m-1]
 ...
 ...                                  B[(n-1)(m-2)]  G[(n-1)(m-1)]
 ...                                  G[n(m-2)]      R[n(m-1)]
@@ -472,12 +472,12 @@ The pixel reference value is calculated as follows:
 The algorithm purely describes the conversion from compressed Bayer code used
 in the SN9C10x chips to uncompressed Bayer. Additional steps are required to
 convert this to a color image (i.e. a color interpolation algorithm).
- 
 The following Huffman codes have been found:
-0: +0 (relative to reference pixel value) 
+0: +0 (relative to reference pixel value)
 100: +4
 101: -4?
-1110xxxx: set absolute value to xxxx.0000 
+1110xxxx: set absolute value to xxxx.0000
 1101: +11
 1111: -11
 11001: +20
diff --git a/Documentation/usb/stv680.txt b/Documentation/video4linux/stv680.txt
index 6448041e7a37..4f8946f32f51 100644
--- a/Documentation/usb/stv680.txt
+++ b/Documentation/video4linux/stv680.txt
@@ -5,15 +5,15 @@ Copyright, 2001, Kevin Sisson
 INTRODUCTION:
-STMicroelectronics produces the STV0680B chip, which comes in two 
+STMicroelectronics produces the STV0680B chip, which comes in two
-types, -001 and -003. The -003 version allows the recording and downloading 
+types, -001 and -003. The -003 version allows the recording and downloading
-of sound clips from the camera, and allows a flash attachment. Otherwise, 
+of sound clips from the camera, and allows a flash attachment. Otherwise,
-it uses the same commands as the -001 version. Both versions support a 
+it uses the same commands as the -001 version. Both versions support a
-variety of SDRAM sizes and sensors, allowing for a maximum of 26 VGA or 20 
+variety of SDRAM sizes and sensors, allowing for a maximum of 26 VGA or 20
-CIF pictures. The STV0680 supports either a serial or a usb interface, and 
+CIF pictures. The STV0680 supports either a serial or a usb interface, and
 video is possible through the usb interface.
-The following cameras are known to work with this driver, although any 
+The following cameras are known to work with this driver, although any
 camera with Vendor/Product codes of 0553/0202 should work:
 Aiptek Pencam (various models)
@@ -34,15 +34,15 @@ http://www.linux-usb.org
 MODULE OPTIONS:
 When the driver is compiled as a module, you can set a "swapRGB=1"
-option, if necessary, for those applications that require it 
+option, if necessary, for those applications that require it
-(such as xawtv). However, the driver should detect and set this 
+(such as xawtv). However, the driver should detect and set this
 automatically, so this option should not normally be used.
 KNOWN PROBLEMS:
-The driver seems to work better with the usb-ohci than the usb-uhci host 
+The driver seems to work better with the usb-ohci than the usb-uhci host
-controller driver. 
+controller driver.
 HELP:
@@ -50,6 +50,4 @@ The latest info on this driver can be found at:
 http://personal.clt.bellsouth.net/~kjsisson or at
 http://stv0680-usb.sourceforge.net
-Any questions to me can be send to:  kjsisson@bellsouth.net
+Any questions to me can be send to:  kjsisson@bellsouth.net
+\ No newline at end of file
diff --git a/Documentation/usb/w9968cf.txt b/Documentation/video4linux/w9968cf.txt
index 9d46cd0b19e3..3b704f2aae6d 100644
--- a/Documentation/usb/w9968cf.txt
+++ b/Documentation/video4linux/w9968cf.txt
@@ -1,5 +1,5 @@
-                   W996[87]CF JPEG USB Dual Mode Camera Chip 
+                   W996[87]CF JPEG USB Dual Mode Camera Chip
                     Driver for Linux 2.6 (basic version)
                   =========================================
@@ -115,7 +115,7 @@ additional testing and full support, would be much appreciated.
 ======================
 For it to work properly, the driver needs kernel support for Video4Linux, USB
 and I2C, and the "ovcamchip" module for the image sensor. Make sure you are not
-actually using any external "ovcamchip" module, given that the W996[87]CF 
+actually using any external "ovcamchip" module, given that the W996[87]CF
 driver depends on the version of the module present in the official kernels.
 The following options of the kernel configuration file must be enabled and
@@ -197,16 +197,16 @@ Note:            The kernel must be compiled with the CONFIG_KMOD option
                 enabled for the 'ovcamchip' module to be loaded and for
                 this parameter to be present.
 -------------------------------------------------------------------------------
-Name:           simcams 
+Name:           simcams
-Type:           int 
+Type:           int
-Syntax:         <n> 
+Syntax:         <n>
 Description:    Number of cameras allowed to stream simultaneously.
                n may vary from 0 to 32.
 Default:        32
 -------------------------------------------------------------------------------
 Name:           video_nr
 Type:           int array (min = 0, max = 32)
-Syntax:         <-1|n[,...]> 
+Syntax:         <-1|n[,...]>
 Description:    Specify V4L minor mode number.
                -1 = use next available
                 n = use minor number n
@@ -219,7 +219,7 @@ Default:        -1
 -------------------------------------------------------------------------------
 Name:           packet_size
 Type:           int array (min = 0, max = 32)
-Syntax:         <n[,...]> 
+Syntax:         <n[,...]>
 Description:    Specify the maximum data payload size in bytes for alternate
                settings, for each device. n is scaled between 63 and 1023.
 Default:        1023
@@ -234,7 +234,7 @@ Default:        2
 -------------------------------------------------------------------------------
 Name:           double_buffer
 Type:           bool array (min = 0, max = 32)
-Syntax:         <0|1[,...]> 
+Syntax:         <0|1[,...]>
 Description:    Hardware double buffering: 0 disabled, 1 enabled.
                It should be enabled if you want smooth video output: if you
                obtain out of sync. video, disable it, or try to
@@ -243,13 +243,13 @@ Default:        1 for every device.
 -------------------------------------------------------------------------------
 Name:           clamping
 Type:           bool array (min = 0, max = 32)
-Syntax:         <0|1[,...]> 
+Syntax:         <0|1[,...]>
 Description:    Video data clamping: 0 disabled, 1 enabled.
 Default:        0 for every device.
 -------------------------------------------------------------------------------
 Name:           filter_type
 Type:           int array (min = 0, max = 32)
-Syntax:         <0|1|2[,...]> 
+Syntax:         <0|1|2[,...]>
 Description:    Video filter type.
                0 none, 1 (1-2-1) 3-tap filter, 2 (2-3-6-3-2) 5-tap filter.
                The filter is used to reduce noise and aliasing artifacts
@@ -258,13 +258,13 @@ Default:        0 for every device.
 -------------------------------------------------------------------------------
 Name:           largeview
 Type:           bool array (min = 0, max = 32)
-Syntax:         <0|1[,...]> 
+Syntax:         <0|1[,...]>
 Description:    Large view: 0 disabled, 1 enabled.
 Default:        1 for every device.
 -------------------------------------------------------------------------------
 Name:           upscaling
 Type:           bool array (min = 0, max = 32)
-Syntax:         <0|1[,...]> 
+Syntax:         <0|1[,...]>
 Description:    Software scaling (for non-compressed video only):
                0 disabled, 1 enabled.
                Disable it if you have a slow CPU or you don't have enough
@@ -341,8 +341,8 @@ Default:        50 for every device.
 -------------------------------------------------------------------------------
 Name:           bandingfilter
 Type:           bool array (min = 0, max = 32)
-Syntax:         <0|1[,...]> 
+Syntax:         <0|1[,...]>
-Description:    Banding filter to reduce effects of fluorescent 
+Description:    Banding filter to reduce effects of fluorescent
                lighting:
                0 disabled, 1 enabled.
                This filter tries to reduce the pattern of horizontal
@@ -374,7 +374,7 @@ Default:        0 for every device.
 -------------------------------------------------------------------------------
 Name:           monochrome
 Type:           bool array (min = 0, max = 32)
-Syntax:         <0|1[,...]> 
+Syntax:         <0|1[,...]>
 Description:    The image sensor is monochrome:
                0 = no, 1 = yes
 Default:        0 for every device.
@@ -400,19 +400,19 @@ Default:        32768 for every device.
 -------------------------------------------------------------------------------
 Name:           contrast
 Type:           long array (min = 0, max = 32)
-Syntax:         <n[,...]> 
+Syntax:         <n[,...]>
 Description:    Set picture contrast (0-65535).
 Default:        50000 for every device.
 -------------------------------------------------------------------------------
 Name:           whiteness
 Type:           long array (min = 0, max = 32)
-Syntax:         <n[,...]> 
+Syntax:         <n[,...]>
 Description:    Set picture whiteness (0-65535).
 Default:        32768 for every device.
 -------------------------------------------------------------------------------
 Name:           debug
 Type:           int
-Syntax:         <n> 
+Syntax:         <n>
 Description:    Debugging information level, from 0 to 6:
                0 = none (use carefully)
                1 = critical errors
diff --git a/Documentation/usb/zc0301.txt b/Documentation/video4linux/zc0301.txt
index f55262c6733b..f55262c6733b 100644
--- a/Documentation/usb/zc0301.txt
+++ b/Documentation/video4linux/zc0301.txt
diff --git a/Documentation/vm/hugetlbpage.txt b/Documentation/vm/hugetlbpage.txt
index 1ad9af1ca4d0..687104bfd09a 100644
--- a/Documentation/vm/hugetlbpage.txt
+++ b/Documentation/vm/hugetlbpage.txt
@@ -27,12 +27,21 @@ number of free hugetlb pages at any time.  It also displays information about
 the configured hugepage size - this is needed for generating the proper
 alignment and size of the arguments to the above system calls.
-The output of "cat /proc/meminfo" will have output like:
+The output of "cat /proc/meminfo" will have lines like:
 .....
 HugePages_Total: xxx
 HugePages_Free:  yyy
-Hugepagesize:    zzz KB
+HugePages_Rsvd:  www
+Hugepagesize:    zzz kB
+where:
+HugePages_Total is the size of the pool of hugepages.
+HugePages_Free is the number of hugepages in the pool that are not yet
+allocated.
+HugePages_Rsvd is short for "reserved," and is the number of hugepages
+for which a commitment to allocate from the pool has been made, but no
+allocation has yet been made. It's vaguely analogous to overcommit.
 /proc/filesystems should also show a filesystem of type "hugetlbfs" configured
 in the kernel.
@@ -42,11 +51,11 @@ pages in the kernel.  Super user can dynamically request more (or free some
 pre-configured) hugepages.
 The allocation (or deallocation) of hugetlb pages is possible only if there are
 enough physically contiguous free pages in system (freeing of hugepages is
-possible only if there are enough hugetlb pages free that can be transfered
+possible only if there are enough hugetlb pages free that can be transferred
 back to regular memory pool).
-Pages that are used as hugetlb pages are reserved inside the kernel and can
+Pages that are used as hugetlb pages are reserved inside the kernel and cannot
-not be used for other purposes.
+be used for other purposes.
 Once the kernel with Hugetlb page support is built and running, a user can
 use either the mmap system call or shared memory system calls to start using
@@ -60,7 +69,7 @@ Use the following command to dynamically allocate/deallocate hugepages:
 This command will try to configure 20 hugepages in the system.  The success
 or failure of allocation depends on the amount of physically contiguous
 memory that is preset in system at this time.  System administrators may want
-to put this command in one of the local rc init file.  This will enable the
+to put this command in one of the local rc init files.  This will enable the
 kernel to request huge pages early in the boot process (when the possibility
 of getting physical contiguous pages is still very high).
@@ -78,8 +87,8 @@ the uid and gid of the current process are taken.  The mode option sets the
 mode of root of file system to value & 0777.  This value is given in octal.
 By default the value 0755 is picked. The size option sets the maximum value of
 memory (huge pages) allowed for that filesystem (/mnt/huge). The size is
-rounded down to HPAGE_SIZE.  The option nr_inode sets the maximum number of
+rounded down to HPAGE_SIZE.  The option nr_inodes sets the maximum number of
-inodes that /mnt/huge can use.  If the size or nr_inode options are not
+inodes that /mnt/huge can use.  If the size or nr_inodes options are not
 provided on command line then no limits are set.  For size and nr_inodes
 options, you can use [G|g]/[M|m]/[K|k] to represent giga/mega/kilo. For
 example, size=2K has the same meaning as size=2048. An example is given at
@@ -88,7 +97,7 @@ the end of this document.
 read and write system calls are not supported on files that reside on hugetlb
 file systems.
-A regular chown, chgrp and chmod commands (with right permissions) could be
+Regular chown, chgrp, and chmod commands (with right permissions) could be
 used to change the file attributes on hugetlbfs.
 Also, it is important to note that no such mount command is required if the
@@ -96,8 +105,8 @@ applications are going to use only shmat/shmget system calls.  Users who
 wish to use hugetlb page via shared memory segment should be a member of
 a supplementary group and system admin needs to configure that gid into
 /proc/sys/vm/hugetlb_shm_group.  It is possible for same or different
-applications to use any combination of mmaps and shm* calls.  Though the
+applications to use any combination of mmaps and shm* calls, though the
-mount of filesystem will be required for using mmaps.
+mount of filesystem will be required for using mmap calls.
 *******************************************************************
diff --git a/Documentation/watchdog/watchdog-api.txt b/Documentation/watchdog/watchdog-api.txt
index c5beb548cfc4..21ed51173662 100644
--- a/Documentation/watchdog/watchdog-api.txt
+++ b/Documentation/watchdog/watchdog-api.txt
@@ -36,6 +36,9 @@ timeout or margin.  The simplest way to ping the watchdog is to write
 some data to the device.  So a very simple watchdog daemon would look
 like this:
+#include <stdlib.h>
+#include <fcntl.h>
 int main(int argc, const char *argv[]) {
        int fd=open("/dev/watchdog",O_WRONLY);
        if (fd==-1) {
diff --git a/Documentation/x86_64/boot-options.txt b/Documentation/x86_64/boot-options.txt
index 1921353259ae..f2cd6ef53ff3 100644
--- a/Documentation/x86_64/boot-options.txt
+++ b/Documentation/x86_64/boot-options.txt
@@ -151,6 +151,11 @@ NUMA
  numa=fake=X   Fake X nodes and ignore NUMA setup of the actual machine.
+  numa=hotadd=percent
+                Only allow hotadd memory to preallocate page structures upto
+                percent of already available memory.
+                numa=hotadd=0 will disable hotadd memory.
 ACPI
  acpi=off      Don't enable ACPI