openwrt/staging/blogic.git
17 years agodm io: delay dec_count
Heinz Mauelshagen [Wed, 9 May 2007 09:33:00 +0000 (02:33 -0700)]
dm io: delay dec_count

Delay decrementing the 'struct io' reference count until after the bio has
been freed so that a bio destructor function may reference it.  Required by a
later patch.

Signed-off-by: Heinz Mauelshagen <hjm@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: Milan Broz <mbroz@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm raid1: add handle_errors feature flag
Jonathan E Brassow [Wed, 9 May 2007 09:32:59 +0000 (02:32 -0700)]
dm raid1: add handle_errors feature flag

This patch adds the ability to specify desired features in the mirror
constructor/mapping table.

The first feature of interest is "handle_errors".  Currently, mirroring will
ignore any I/O errors from the devices.  Subsequent patches will check for
this flag and handle the errors.  If flag/feature is not present, mirror will
do nothing - maintaining backwards compatibility.

Signed-off-by: Jonathan E Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm log: report fault status
Jonathan E Brassow [Wed, 9 May 2007 09:32:58 +0000 (02:32 -0700)]
dm log: report fault status

This patch reports the status of the log device so that userspace can detect
the error and take appropriate action.

Signed-off-by: Jonathan E Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm log: fault detection
Jonathan E Brassow [Wed, 9 May 2007 09:32:57 +0000 (02:32 -0700)]
dm log: fault detection

This patch gives the disk logging code the ability to store the fact that an
error occured on the log device.  In addition, an event is raised when an
error is encountered during I/O to the log device.

Signed-off-by: Jonathan E Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm: allow offline devices
Mike Anderson [Wed, 9 May 2007 09:32:57 +0000 (02:32 -0700)]
dm: allow offline devices

Allow check_device_area to succeed if a device has an i_size of zero.  This
addresses an issue seen on DASD devices setting up a multipath table for paths
in online and offline state.

Signed-off-by: Mike Anderson <andmike@us.ibm.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm mpath: log device name
Edward Goggin [Wed, 9 May 2007 09:32:56 +0000 (02:32 -0700)]
dm mpath: log device name

Make the mapped device structure accessible to hardware handlers so error
messages can include the device name.

Signed-off-by: Edward Goggin <egoggin@emc.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm crypt: add null iv
Ludwig Nussel [Wed, 9 May 2007 09:32:55 +0000 (02:32 -0700)]
dm crypt: add null iv

Add a new IV generation method 'null' to read old filesystem images created
with SuSE's loop_fish2 module.

Signed-off-by: Ludwig Nussel <ludwig.nussel@suse.de>
Acked-By: Christophe Saout <christophe@saout.de>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm crypt: use smaller bvecs in clones
Olaf Kirch [Wed, 9 May 2007 09:32:54 +0000 (02:32 -0700)]
dm crypt: use smaller bvecs in clones

Allocate smaller clones

With the previous dm-crypt fixes, there is no need for the clone bios to have
the same bvec size as the original - we just need to make them big enough for
the remaining number of pages.  The only requirement is that we clear the
"out" index in convert_context, so that crypt_convert starts storing data at
the right position within the clone bio.

Signed-off-by: Olaf Kirch <olaf.kirch@oracle.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm crypt: fix remove first_clone
Olaf Kirch [Wed, 9 May 2007 09:32:53 +0000 (02:32 -0700)]
dm crypt: fix remove first_clone

Get rid of first_clone in dm-crypt

This gets rid of first_clone, which is not really needed.  Apparently, cloned
bios used to share their bvec some time way in the past - this is no longer
the case.  Contrarily, this even hurts us if we try to create a clone off
first_clone after it has completed, and crypt_endio has destroyed its bvec.

Signed-off-by: Olaf Kirch <olaf.kirch@oracle.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm crypt: fix avoid cloned bio ref after free
Olaf Kirch [Wed, 9 May 2007 09:32:52 +0000 (02:32 -0700)]
dm crypt: fix avoid cloned bio ref after free

Do not access the bio after generic_make_request

We should never access a bio after generic_make_request - there's no guarantee
it still exists.

Signed-off-by: Olaf Kirch <olaf.kirch@oracle.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm crypt: fix call to clone_init
Olaf Kirch [Wed, 9 May 2007 09:32:52 +0000 (02:32 -0700)]
dm crypt: fix call to clone_init

Call clone_init early

We need to call clone_init as early as possible - at least before call
bio_put(clone) in any error path.  Otherwise, the destructor will try to
dereference bi_private, which may still be NULL.

Signed-off-by: Olaf Kirch <olaf.kirch@oracle.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm crypt: disable barriers
Milan Broz [Wed, 9 May 2007 09:32:51 +0000 (02:32 -0700)]
dm crypt: disable barriers

Disable barriers in dm-crypt because of current workqueue processing can
reorder requests.

This must be addresed later but for now disabling barriers is needed to
prevent data corruption.

Signed-off-by: Milan Broz <mbroz@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodm raid1: one kmirrord per mirror
Holger Smolinski [Wed, 9 May 2007 09:32:50 +0000 (02:32 -0700)]
dm raid1: one kmirrord per mirror

This patch replaces the single instance of kmirrord by one instance per mirror
set.  This change is required to avoid a deadlock in kmirrord when the
persistent dirty log of a mirror itself resides on a mirror.  The single
instance of kmirrord then issues a sync write to the dirty log in write_bits
which gets deferred to kmirrord itself later in the call chain.  But kmirrord
never does the deferred work because it is still waiting for the sync
write_bits.

_mirror_sets is removed as it no longer needed, and we always flush the
workqueue before destroying it to ensure all work is complete before
destroying it.

Signed-off-by: Holger Smolinski <smolinski@de.ibm.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoFRV: Replace pgd management via slabs through quicklists
Christoph Lameter [Wed, 9 May 2007 09:32:48 +0000 (02:32 -0700)]
FRV: Replace pgd management via slabs through quicklists

This is done in order to be able to run SLUB which expects no modifications
to its page structs.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoFix spellings of slab allocator section in init/Kconfig
Christoph Lameter [Wed, 9 May 2007 09:32:47 +0000 (02:32 -0700)]
Fix spellings of slab allocator section in init/Kconfig

Fix some of the spelling issues. Fix sentences. Discourage SLOB use
since SLUB can pack objects denser.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Cc: Matt Mackall <mpm@selenic.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agokrealloc: fix kerneldoc comments
Pekka J Enberg [Wed, 9 May 2007 09:32:46 +0000 (02:32 -0700)]
krealloc: fix kerneldoc comments

No "blank" (or "*") line is allowed between the function name and lines for
it parameter(s).

Cc: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: rework slab order determination
Christoph Lameter [Wed, 9 May 2007 09:32:46 +0000 (02:32 -0700)]
SLUB: rework slab order determination

In some cases SLUB is creating uselessly slabs that are larger than
slub_max_order. Also the layout of some of the slabs was not satisfactory.

Go to an iterarive approach.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: include lifetime stats and sets of cpus / nodes in tracking output
Christoph Lameter [Wed, 9 May 2007 09:32:45 +0000 (02:32 -0700)]
SLUB: include lifetime stats and sets of cpus / nodes in tracking output

We have information about how long an object existed and about the nodes and
cpus where the allocations and frees took place.  Add that information to the
tracking output in /sys/slab/xx/alloc_calls and /sys/slab/free_calls

This will then enable slabinfo to output nice reports like this:

  christoph@qirst:~/slub$ ./slabinfo kmalloc-128

  Slabcache: kmalloc-128           Aliases:  0 Order :  0

  Sizes (bytes)     Slabs              Debug                Memory
  ------------------------------------------------------------------------
  Object :     128  Total  :      12   Sanity Checks : On   Total:   49152
  SlabObj:     200  Full   :       7   Redzoning     : On   Used :   24832
  SlabSiz:    4096  Partial:       4   Poisoning     : On   Loss :   24320
  Loss   :      72  CpuSlab:       1   Tracking      : On   Lalig:   13968
  Align  :       8  Objects:      20   Tracing       : Off  Lpadd:    1152

  kmalloc-128 has no kmem_cache operations

  kmalloc-128: Kernel object allocation
  -----------------------------------------------------------------------
        6 param_sysfs_setup+0x71/0x130 age=284512/284512/284512 pid=1 nodes=0-1,3
       11 percpu_populate+0x39/0x80 age=283914/284428/284512 pid=1 nodes=0
       21 __register_chrdev_region+0x31/0x170 age=282896/284347/284473 pid=1-1705 nodes=0-2
        1 sys_inotify_init+0x76/0x1c0 age=283423 pid=1004 nodes=0
       19 as_get_io_context+0x32/0xd0 age=6/247567/283988 pid=1-11782 nodes=0,2
       10 ida_pre_get+0x4a/0x80 age=277666/283773/284526 pid=0-2177 nodes=0,2
       24 kobject_kset_add_dir+0x37/0xb0 age=282727/283860/284472 pid=1-1723 nodes=0-2
        1 acpi_ds_build_internal_buffer_obj+0xd3/0x11d age=284508 pid=1 nodes=0
       24 con_insert_unipair+0xd7/0x110 age=284438/284438/284438 pid=1 nodes=0,2
        1 uart_open+0x2d2/0x4b0 age=283896 pid=1 nodes=0
       26 dma_pool_create+0x73/0x1a0 age=282762/282833/282916 pid=1705-1723 nodes=0
        1 neigh_table_init_no_netlink+0xd2/0x210 age=284461 pid=1 nodes=0
        2 neigh_parms_alloc+0x2b/0xe0 age=284410/284411/284412 pid=1 nodes=2
        2 neigh_resolve_output+0x1e1/0x280 age=276289/276291/276293 pid=0-2443 nodes=0
        1 netlink_kernel_create+0x90/0x170 age=284472 pid=1 nodes=0
        4 xt_alloc_table_info+0x39/0xf0 age=283958/283958/283959 pid=1 nodes=1
        3 fn_hash_insert+0x473/0x720 age=277653/277661/277666 pid=2177-2185 nodes=0
        1 get_mtrr_state+0x285/0x2a0 age=284526 pid=0 nodes=0
        1 cacheinfo_cpu_callback+0x26d/0x3e0 age=284458 pid=1 nodes=0
       29 kernel_param_sysfs_setup+0x25/0x90 age=284511/284511/284512 pid=1 nodes=0-1,3
        5 process_zones+0x5e/0x170 age=284546/284546/284546 pid=0 nodes=0
        1 drm_core_init+0x48/0x160 age=284421 pid=1 nodes=2

  kmalloc-128: Kernel object freeing
  ------------------------------------------------------------------------
      163 <not-available> age=4295176847 pid=0 nodes=0-3
        1 __vunmap+0x6e/0xf0 age=282907 pid=1723 nodes=0
       28 free_as_io_context+0x12/0x90 age=9243/262197/283474 pid=42-11754 nodes=0
        1 acpi_get_object_info+0x1b7/0x1d4 age=284475 pid=1 nodes=0
        1 do_acpi_find_child+0x45/0x4e age=284475 pid=1 nodes=0

  NUMA nodes           :    0    1    2    3
  ------------------------------------------
  All slabs                 7    2    2    1
  Partial slabs             2    2    0    0

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: add CONFIG_SLUB_DEBUG
Christoph Lameter [Wed, 9 May 2007 09:32:44 +0000 (02:32 -0700)]
SLUB: add CONFIG_SLUB_DEBUG

CONFIG_SLUB_DEBUG can be used to switch off the debugging and sysfs components
of SLUB.  Thus SLUB will be able to replace SLOB.  SLUB can arrange objects in
a denser way than SLOB and the code size should be minimal without debugging
and sysfs support.

Note that CONFIG_SLUB_DEBUG is materially different from CONFIG_SLAB_DEBUG.
CONFIG_SLAB_DEBUG is used to enable slab debugging in SLAB.  SLUB enables
debugging via a boot parameter.  SLUB debug code should always be present.

CONFIG_SLUB_DEBUG can be modified in the embedded config section.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: move tracking definitions and check_valid_pointer() away from debug code
Christoph Lameter [Wed, 9 May 2007 09:32:43 +0000 (02:32 -0700)]
SLUB: move tracking definitions and check_valid_pointer() away from debug code

Move the tracking definitions and the check_valid_pointer() function away from
the debugging related functions.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: consolidate trace code
Christoph Lameter [Wed, 9 May 2007 09:32:42 +0000 (02:32 -0700)]
SLUB: consolidate trace code

Trace in both slab_alloc and slab_free has a lot of common code.  Use a single
function for both.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: introduce DebugSlab(page)
Christoph Lameter [Wed, 9 May 2007 09:32:42 +0000 (02:32 -0700)]
SLUB: introduce DebugSlab(page)

This replaces the PageError() checking.  DebugSlab is clearer and allows for
future changes to the page bit used.  We also need it to support
CONFIG_SLUB_DEBUG.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: move resiliency check into SYSFS section
Christoph Lameter [Wed, 9 May 2007 09:32:41 +0000 (02:32 -0700)]
SLUB: move resiliency check into SYSFS section

Move the resiliency check into the SYSFS section after validate_slab that is
used by the resiliency check.  This will avoid a forward declaration.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: add macros for scanning objects in a slab
Christoph Lameter [Wed, 9 May 2007 09:32:40 +0000 (02:32 -0700)]
SLUB: add macros for scanning objects in a slab

Scanning of objects happens in a number of functions.  Consolidate that code.
DECLARE_BITMAP instead of coding the declaration for bitmaps.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: update comments
Christoph Lameter [Wed, 9 May 2007 09:32:39 +0000 (02:32 -0700)]
SLUB: update comments

Update comments throughout SLUB to reflect the new developments.  Fix up
various awkward sentences.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: get rid of finish_bootstrap
Christoph Lameter [Wed, 9 May 2007 09:32:39 +0000 (02:32 -0700)]
SLUB: get rid of finish_bootstrap

Its only purpose was to bring some sort of symmetry to sysfs usage when
dealing with bootstrapping per cpu flushing.  Since we do not time out slabs
anymore we have no need to run finish_bootstrap even without sysfs.  Fold it
back into slab_sysfs_init and drop the initcall for the !SYFS case.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: clean up krealloc
Christoph Lameter [Wed, 9 May 2007 09:32:38 +0000 (02:32 -0700)]
SLUB: clean up krealloc

We really do not need all this gaga there.

ksize gives us all the information we need to figure out if the object can
cope with the new size.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: use check_valid_pointer in kmem_ptr_validate
Christoph Lameter [Wed, 9 May 2007 09:32:37 +0000 (02:32 -0700)]
SLUB: use check_valid_pointer in kmem_ptr_validate

We needlessly duplicate code. Also make check_valid_pointer inline.

Signed-off-by: Christoph LAemter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: slabinfo upgrade
Christoph Lameter [Wed, 9 May 2007 09:32:37 +0000 (02:32 -0700)]
SLUB: slabinfo upgrade

-e Show empty slabs
-d Modification of slab debug options at runtime
-o Operations. Display of ctor / dtor etc.
-r Report: Display all available information about a slabcache.

Cleanup tracking display and make it work right.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: after object padding only needed for Redzoning
Christoph Lameter [Wed, 9 May 2007 09:32:36 +0000 (02:32 -0700)]
SLUB: after object padding only needed for Redzoning

If no redzoning is selected then we do not need padding before the next
object.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSLUB: add support for dynamic cacheline size determination
Christoph Lameter [Wed, 9 May 2007 09:32:35 +0000 (02:32 -0700)]
SLUB: add support for dynamic cacheline size determination

SLUB currently assumes that the cacheline size is static.  However, i386 f.e.
supports dynamic cache line size determination.

Use cache_line_size() instead of L1_CACHE_BYTES in the allocator.

That also explains the purpose of SLAB_HWCACHE_ALIGN.  So we will need to keep
that one around to allow dynamic aligning of objects depending on boot
determination of the cache line size.

[akpm@linux-foundation.org: need to define it before we use it]
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoDeclare {compat_}sys_utimensat
Stephen Rothwell [Wed, 9 May 2007 09:32:35 +0000 (02:32 -0700)]
Declare {compat_}sys_utimensat

This is needed before Powerpc can wire up the syscall.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agommc build fix
Andrew Morton [Wed, 9 May 2007 09:32:34 +0000 (02:32 -0700)]
mmc build fix

Cc: Pierre Ossman <drzeus@drzeus.cx>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoDocumentation/i386/boot.txt: update and correct
H. Peter Anvin [Wed, 9 May 2007 03:37:02 +0000 (20:37 -0700)]
Documentation/i386/boot.txt: update and correct

In the process of rewriting the x86 setup code, I found a number of
inaccuracies and outdated recommendations in the boot protocol
documentation.  Revamp to make it more up to date.

In particular, the common use of the heap actually requires (slightly)
more than 4K of heap plus stack, which is the recommended amount in
the document; currently the code compensates by being smaller than
specified, but we can't assume that will be true forever.  Thus,
recommend that if we have a modern bzImage kernel, that the bootloader
maximizes the available space.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agorevert 'sched: redundant reschedule when set_user_nice() boosts a prio of a task...
Andrew Morton [Wed, 9 May 2007 03:27:06 +0000 (20:27 -0700)]
revert 'sched: redundant reschedule when set_user_nice() boosts a prio of a task from the "expired" array'

Revert commit bd53f96ca54a21c07e7a0ae1886fa623d370b85f.

Con says:

This is no good, sorry. The one I saw originally was with the staircase
deadline cpu scheduler in situ and was different.

  #define TASK_PREEMPTS_CURR(p, rq) \
     ((p)->prio < (rq)->curr->prio)
     (((p)->prio < (rq)->curr->prio) && ((p)->array == (rq)->active))

This will fail to wake up a runqueue for a task that has been migrated to the
expired array of a runqueue which is otherwise idle which can happen with smp
balancing,

Cc: Dmitry Adamushko <dmitry.adamushko@gmail.com>
Cc: Con Kolivas <kernel@kolivas.org>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoapplesmc: Use standard sysfs names for labels
Jean Delvare [Wed, 9 May 2007 03:27:05 +0000 (20:27 -0700)]
applesmc: Use standard sysfs names for labels

We have a standard suffix to associate a designation string to a sensor:
_label.  Use it instead of _position so that libsensors will catch it.
(This isn't implemented yet, but should be soon.)

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Nicolas Boichat <nicolas@boichat.ch>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoapplesmc: Use the address as platform device ID
Jean Delvare [Wed, 9 May 2007 03:27:04 +0000 (20:27 -0700)]
applesmc: Use the address as platform device ID

Let the applesmc device export its address to userspace.  libsensors needs
this to recognize the device and give it a unique ID.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Nicolas Boichat <nicolas@boichat.ch>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agousb_gigaset: don't kmalloc(0)
Tilman Schmidt [Wed, 9 May 2007 03:27:03 +0000 (20:27 -0700)]
usb_gigaset: don't kmalloc(0)

Zero-sized allocations are pointless anyway, and the SLUB allocator
complains about them, so stop doing that.

Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: Hansjoerg Lipp <hjlipp@web.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68k: <asm/scatterlist.h> needs <linux/types.h>
Geert Uytterhoeven [Wed, 9 May 2007 03:27:03 +0000 (20:27 -0700)]
m68k: <asm/scatterlist.h> needs <linux/types.h>

The recent <linux/pci.h> cleanup uncovered that include/asm-m68k/scatterlist.h
needs to include <linux/types.h>

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoFRV: Miscellaneous fixes
David Howells [Wed, 9 May 2007 03:27:02 +0000 (20:27 -0700)]
FRV: Miscellaneous fixes

Miscellaneous fixes to bring FRV up to date:

 (1) Copy the new syscall numbers from i386 to asm-frv/unistd.h and fill out
     the syscall table in entry.S too.

 (2) Mark __frv_uart0 and __frv_uart1 __pminitdata rather than __initdata so
     that determine_clocks() can access them when CONFIG_PM=y.

 (3) Make arch/frv/mm/elf-fdpic.c include asm/mman.h so that MAP_FIXED is
     available (fixes commit 2fd3bebaad9da3b3b99c46a3389099424bf7ee35).

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agosmaps: only define clear_refs for CONFIG_MMU
David Rientjes [Wed, 9 May 2007 03:27:01 +0000 (20:27 -0700)]
smaps: only define clear_refs for CONFIG_MMU

/proc/pid/clear_refs is only defined in the CONFIG_MMU case, so make sure we
don't have any references to clear_refs_smap() in generic procfs code.

Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoFix unnecesary meminit
Yasunori Goto [Wed, 9 May 2007 03:27:01 +0000 (20:27 -0700)]
Fix unnecesary meminit

This is to fix unnecessary __meminit definition.  These are exported for
kernel modules.

I compiled on ia64/x86-64 with memory hotplug on/off.

Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Wed, 9 May 2007 03:33:59 +0000 (20:33 -0700)]
Merge /pub/scm/linux/kernel/git/herbert/crypto-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  [CRYPTO] cryptomgr: Fix use after free

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Wed, 9 May 2007 03:32:43 +0000 (20:32 -0700)]
Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Optimize fault kprobe handling just like powerpc.
  [SPARC]: Wire up utimensat syscall.
  [SPARC64]: Fix request_irq() ignored result warnings in PCI controller code.
  [SPARC64]: Kill asm-sparc64/pbm.h
  [ATYFB]: Fix sparc includes.
  [QLA2XXX]: Fix build on sparc.
  [SPARC64]: Removal of trivial pci_controller_info uses.
  [SPARC64]: Move index info pci_pbm_info.
  [SPARC64]: Move {setup,teardown}_msi_irq into pci_pbm_info.
  [SPARC64]: Move pci_ops into pci_pbm_info.
  [SPARC64] SBUS: Error interrupt registry cleanups.
  [SPARC64] PCI: Use root list of pbm's instead of pci_controller_info's
  [SPARC64] PCI: Kill PROM_PCIRNG_MAX and PROM_PCIIMAP_MAX.
  [SPARC64] PCI: Use common routine to fetch PBM properties.

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6
Linus Torvalds [Wed, 9 May 2007 03:32:16 +0000 (20:32 -0700)]
Merge /pub/scm/linux/kernel/git/jejb/scsi-misc-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6: (58 commits)
  [SCSI] zfcp: clear boxed flag on unit reopen.
  [SCSI] zfcp: clear adapter failed flag if an fsf request times out.
  [SCSI] zfcp: rework request ID management.
  [SCSI] zfcp: Fix deadlock between zfcp ERP and SCSI
  [SCSI] zfcp: Locking for req_no and req_seq_no
  [SCSI] zfcp: print S_ID and D_ID with 3 bytes
  [SCSI] ipr: Use PCI-E reset API for new ipr adapter
  [SCSI] qla2xxx: Update version number to 8.01.07-k7.
  [SCSI] qla2xxx: Add MSI support.
  [SCSI] qla2xxx: Correct pci_set_msi() usage semantics.
  [SCSI] qla2xxx: Attempt to stop firmware only if it had been previously executed.
  [SCSI] qla2xxx: Honor NVRAM port-down-retry-count settings.
  [SCSI] qla2xxx: Error-out during probe() if we're unable to complete HBA initialization.
  [SCSI] zfcp: Stop system after memory corruption
  [SCSI] mesh: cleanup variable usage in interrupt handler
  [SCSI] megaraid: replace yield() with cond_resched()
  [SCSI] megaraid: fix warnings when CONFIG_PROC_FS=n
  [SCSI] aacraid: correct SUN products to README
  [SCSI] aacraid: superfluous adapter reset for IBM 8 series ServeRAID controllers
  [SCSI] aacraid: kexec fix (reset interrupt handler)
  ...

17 years agoRevert "fbdev: ignore VESA modes if framebuffer is disabled"
Linus Torvalds [Wed, 9 May 2007 03:12:30 +0000 (20:12 -0700)]
Revert "fbdev: ignore VESA modes if framebuffer is disabled"

This reverts commit 464bdd33e9baad9806c7adbd8dfc37081a55f27e.

Peter Anvin correctly points out that VESA modes have nothing to do with
frame buffers per se - they are often just regular extended text modes.
Disabling them just because we don't have frame buffer support is very
wrong.

Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Antonino A. Daplas <adaplas@gmail.com>,
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoRemove suid/sgid bits on [f]truncate()
Linus Torvalds [Wed, 9 May 2007 03:10:00 +0000 (20:10 -0700)]
Remove suid/sgid bits on [f]truncate()

.. to match what we do on write().  This way, people who write to files
by using [f]truncate + writable mmap have the same semantics as if they
were using the write() family of system calls.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[CRYPTO] cryptomgr: Fix use after free
Herbert Xu [Wed, 9 May 2007 03:04:39 +0000 (13:04 +1000)]
[CRYPTO] cryptomgr: Fix use after free

By the time kthread_run returns the param may have already been freed
so writing the returned thread_struct pointer to param is wrong.

In fact, we don't need it in param anyway so this patch simply puts it
on the stack.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
17 years ago[SPARC64]: Optimize fault kprobe handling just like powerpc.
David S. Miller [Wed, 9 May 2007 01:25:14 +0000 (18:25 -0700)]
[SPARC64]: Optimize fault kprobe handling just like powerpc.

And eliminate DIE_GPF while we're at it.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC]: Wire up utimensat syscall.
David S. Miller [Wed, 9 May 2007 00:50:14 +0000 (17:50 -0700)]
[SPARC]: Wire up utimensat syscall.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Fix request_irq() ignored result warnings in PCI controller code.
David S. Miller [Wed, 9 May 2007 00:23:31 +0000 (17:23 -0700)]
[SPARC64]: Fix request_irq() ignored result warnings in PCI controller code.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Kill asm-sparc64/pbm.h
David S. Miller [Tue, 8 May 2007 07:43:56 +0000 (00:43 -0700)]
[SPARC64]: Kill asm-sparc64/pbm.h

Everything it contains can be hidden in pci_impl.h

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[ATYFB]: Fix sparc includes.
David S. Miller [Tue, 8 May 2007 07:37:23 +0000 (00:37 -0700)]
[ATYFB]: Fix sparc includes.

No need to use asm/pbm.h here.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[QLA2XXX]: Fix build on sparc.
David S. Miller [Tue, 8 May 2007 07:36:49 +0000 (00:36 -0700)]
[QLA2XXX]: Fix build on sparc.

We now use pci_device_to_OF_node() to get properties
and of_get_property() returns const pointers.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Removal of trivial pci_controller_info uses.
David S. Miller [Tue, 8 May 2007 07:19:02 +0000 (00:19 -0700)]
[SPARC64]: Removal of trivial pci_controller_info uses.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Move index info pci_pbm_info.
David S. Miller [Tue, 8 May 2007 06:49:01 +0000 (23:49 -0700)]
[SPARC64]: Move index info pci_pbm_info.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Move {setup,teardown}_msi_irq into pci_pbm_info.
David S. Miller [Tue, 8 May 2007 06:28:50 +0000 (23:28 -0700)]
[SPARC64]: Move {setup,teardown}_msi_irq into pci_pbm_info.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Move pci_ops into pci_pbm_info.
David S. Miller [Tue, 8 May 2007 06:24:05 +0000 (23:24 -0700)]
[SPARC64]: Move pci_ops into pci_pbm_info.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64] SBUS: Error interrupt registry cleanups.
David S. Miller [Tue, 8 May 2007 06:08:39 +0000 (23:08 -0700)]
[SPARC64] SBUS: Error interrupt registry cleanups.

Do not use IRQF_SHARED, these interrupt numbers should all
be unique.

Also use name strings without spaces in them just like
PCI controller drivers do, for consistency.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64] PCI: Use root list of pbm's instead of pci_controller_info's
David S. Miller [Tue, 8 May 2007 06:06:27 +0000 (23:06 -0700)]
[SPARC64] PCI: Use root list of pbm's instead of pci_controller_info's

The idea is to move more and more things into the pbm,
with the eventual goal of eliminating the pci_controller_info
entirely as there really isn't any need for it.

This stage of the transformations requires some reworking of
the PCI error interrupt handling.

It might be tricky to get rid of the pci_controller_info parenting for
a few reasons:

1) When we get an uncorrectable or correctable error we want
   to interrogate the IOMMU and streaming cache of both
   PBMs for error status.  These errors come from the UPA
   front-end which is shared between the two PBM PCI bus
   segments.

   Historically speaking this is why I choose the datastructure
   hierarchy of pci_controller_info-->pci_pbm_info

2) The probing does a portid/devhandle match to look for the
   'other' pbm, but this is entirely an artifact and can be
   eliminated trivially.

What we could do to solve #1 is to have a "buddy" pointer from one pbm
to another.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64] PCI: Kill PROM_PCIRNG_MAX and PROM_PCIIMAP_MAX.
David S. Miller [Tue, 8 May 2007 04:54:28 +0000 (21:54 -0700)]
[SPARC64] PCI: Kill PROM_PCIRNG_MAX and PROM_PCIIMAP_MAX.

They are totally unused.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64] PCI: Use common routine to fetch PBM properties.
David S. Miller [Tue, 8 May 2007 04:51:41 +0000 (21:51 -0700)]
[SPARC64] PCI: Use common routine to fetch PBM properties.

Namely bus-range and ino-bitmap.

This allows us also to eliminate pci_controller_info's
pci_{first,last}_busno fields as only the pbm ones are
used now.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agommc: use lock instead of claim in debug check
Pierre Ossman [Tue, 8 May 2007 20:35:17 +0000 (22:35 +0200)]
mmc: use lock instead of claim in debug check

As mmc_detect_change() can be called from irq context, using
claim (which can sleep) is inherently unsafe. Use the host
spinlock instead, which also is faster.

Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
17 years agommc: Use menuconfig objects
Jan Engelhardt [Tue, 8 May 2007 20:30:32 +0000 (22:30 +0200)]
mmc: Use menuconfig objects

Change Kconfig objects from "menu, config" into "menuconfig" so
that the user can disable the whole feature without having to
enter the menu first.

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
17 years agodisable socket power in adapter driver instead of media one
Alex Dubov [Wed, 2 May 2007 03:14:55 +0000 (20:14 -0700)]
disable socket power in adapter driver instead of media one

Socket power must be fully controlled by adapter driver. This also prevents
unnecessary power-off of the socket when media driver is unloaded, yet
media remains in the socket.

Signed-off-by: Alex Dubov <oakad@yahoo.com>
Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
17 years agoMerge branch 'hwmon-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6
Linus Torvalds [Tue, 8 May 2007 19:07:28 +0000 (12:07 -0700)]
Merge branch 'hwmon-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6

* 'hwmon-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6: (32 commits)
  Use menuconfig objects - hwmon
  hwmon/smsc47b397: Use dynamic sysfs callbacks
  hwmon/smsc47b397: Convert to a platform driver
  hwmon/w83781d: Deprecate W83627HF support
  hwmon/w83781d: Use dynamic sysfs callbacks
  hwmon/w83781d: Be less i2c_client-centric
  hwmon/w83781d: Clean up conversion macros
  hwmon/w83781d: No longer use i2c-isa
  hwmon/ams: Do not print error on systems without apple motion sensor
  hwmon/ams: Fix I2C read retry logic
  hwmon: New AD7416, AD7417 and AD7418 driver
  hwmon/coretemp: Add documentation
  hwmon: New coretemp driver
  i386: Use functions from library in msr driver
  i386: Add safe variants of rdmsr_on_cpu and wrmsr_on_cpu
  hwmon/lm75: Use dynamic sysfs callbacks
  hwmon/lm78: Use dynamic sysfs callbacks
  hwmon/lm78: Be less i2c_client-centric
  hwmon/lm78: No longer use i2c-isa
  hwmon: New max6650 driver
  ...

17 years agoFix sunrpc warning noise
Geert Uytterhoeven [Tue, 8 May 2007 09:37:26 +0000 (11:37 +0200)]
Fix sunrpc warning noise

Commit c5a4dd8b7c15927a8fbff83171b57cad675a79b9 introduced the following
compiler warnings:

net/sunrpc/sched.c:766: warning: format '%u' expects type 'unsigned int', but argument 3 has type 'size_t'
net/sunrpc/sched.c:785: warning: format '%u' expects type 'unsigned int', but argument 2 has type 'size_t'

  - Use %zu to format size_t
  - Kill 2 useless casts

Signed-off-by: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge git://oss.sgi.com:8090/xfs/xfs-2.6
Linus Torvalds [Tue, 8 May 2007 18:59:33 +0000 (11:59 -0700)]
Merge git://oss.sgi.com:8090/xfs/xfs-2.6

* git://oss.sgi.com:8090/xfs/xfs-2.6:
  [XFS] Add lockdep support for XFS
  [XFS] Fix race in xfs_write() b/w dmapi callout and direct I/O checks.
  [XFS] Get rid of redundant "required" in msg.
  [XFS] Export via a function xfs_buftarg_list for use by kdb/xfsidbg.
  [XFS] Remove unused ilen variable and references.
  [XFS] Fix to prevent the notorious 'NULL files' problem after a crash.
  [XFS] Fix race condition in xfs_write().
  [XFS] Fix uquota and oquota enforcement problems.
  [XFS] propogate return codes from flush routines
  [XFS] Fix quotaon syscall failures for group enforcement requests.
  [XFS] Invalidate quotacheck when mounting without a quota type.
  [XFS] reducing the number of random number functions.
  [XFS] remove more misc. unused args
  [XFS] the "aendp" arg to xfs_dir2_data_freescan is always NULL, remove it.
  [XFS] The last argument "lsn" of xfs_trans_commit() is always called with

17 years agoMerge branch 'reset-seq' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libat...
Linus Torvalds [Tue, 8 May 2007 18:58:20 +0000 (11:58 -0700)]
Merge branch 'reset-seq' of /linux/kernel/git/jgarzik/libata-dev

* 'reset-seq' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  [libata reset-seq] build and merge fixes
  libata: reimplement reset sequencing
  libata: improve ata_std_prereset()
  libata: improve 0xff status handling
  libata: add deadline support to prereset and reset methods

17 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Tue, 8 May 2007 18:57:17 +0000 (11:57 -0700)]
Merge branch 'upstream-linus' of /linux/kernel/git/jgarzik/netdev-2.6

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6: (40 commits)
  [netdrvr] atl1: fix build
  pasemi_mac: Use local-mac-address instead of mac-address if available
  pasemi_mac: PHY support
  pasemi_mac: Add msglevel support and "debug" module param
  pasemi_mac: Logic cleanup / rx performance improvements
  pasemi_mac: Minor cleanup / define fixes
  pasemi_mac: Add SKB reuse / copy-break
  pasemi_mac: Timer and interrupt fixes
  pasemi_mac: Abstract and fix up interrupt restart routines
  pasemi_mac: Move the IRQ mapping from the PCI layer to the driver
  tc35815: Remove unnecessary skb->dev assignment
  drivers/net/dm9000: Convert to generic boolean
  AT91RM9200 Ethernet: Fix multicast addressing
  AT91RM9200 Ethernet: Support additional PHYs
  PCMCIA-NETDEV : xirc2ps_cs: bugfix of multicast code
  sky2: re-enable 88E8056 for most motherboards
  MIPS: Drop unnecessary CONFIG_ISA from RBTX49XX
  ne: MIPS: Use platform_driver for ne on RBTX49XX
  ne: Add NEEDS_PORTLIST to control ISA auto-probe
  ne: Misc fixes for platform driver.
  ...

Fix conflict in drivers/net/pasemi_mac.c (get_property() got renamed to
of_get_property()) manually.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge branch 'drm-patches' of master.kernel.org:/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Tue, 8 May 2007 18:52:35 +0000 (11:52 -0700)]
Merge branch 'drm-patches' of /linux/kernel/git/airlied/drm-2.6

* 'drm-patches' of master.kernel.org:/pub/scm/linux/kernel/git/airlied/drm-2.6:
  via: Make sure we flush write-combining using a follow-up read.
  via: Try to improve command-buffer chaining.
  drm: remove old taskqueue remnant
  drm: rename badly named define and cleanup ioctl code spacing
  radeon: Don't mess up page flipping when a file descriptor is closed.
  drm/radeon: upgrade to 1.27 - make PCI GART more flexible

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Tue, 8 May 2007 18:51:43 +0000 (11:51 -0700)]
Merge /pub/scm/linux/kernel/git/dtor/input

* master.kernel.org:/pub/scm/linux/kernel/git/dtor/input:
  Input: move USB miscellaneous devices under drivers/input/misc
  Input: move USB mice under drivers/input/mouse
  Input: move USB gamepads under drivers/input/joystick
  Input: move USB touchscreens under drivers/input/touchscreen
  Input: move USB tablets under drivers/input/tablet
  Input: i8042 - fix AUX port detection with some chips
  Input: aaed2000_kbd - convert to use polldev library
  Input: drivers/usb/input - usb_buffer_free() cleanup
  Input: synaptics - don't complain about failed resets
  Input: pull input.h into uinpit.h
  Input: drivers/usb/input - fix sparse warnings (signedness)
  Input: evdev - fix some sparse warnings (signedness, shadowing)
  Input: drivers/joystick - fix various sparse warnings
  Input: force feedback - make sure effect is present before playing

17 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Tue, 8 May 2007 18:50:19 +0000 (11:50 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/paulus/powerpc

* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc: (77 commits)
  [POWERPC] Abolish powerpc_flash_init()
  [POWERPC] Early serial debug support for PPC44x
  [POWERPC] Support for the Ebony 440GP reference board in arch/powerpc
  [POWERPC] Add device tree for Ebony
  [POWERPC] Add powerpc/platforms/44x, disable platforms/4xx for now
  [POWERPC] MPIC U3/U4 MSI backend
  [POWERPC] MPIC MSI allocator
  [POWERPC] Enable MSI mappings for MPIC
  [POWERPC] Tell Phyp we support MSI
  [POWERPC] RTAS MSI implementation
  [POWERPC] PowerPC MSI infrastructure
  [POWERPC] Rip out the existing powerpc msi stubs
  [POWERPC] Remove use of 4level-fixup.h for ppc32
  [POWERPC] Add powerpc PCI-E reset API implementation
  [POWERPC] Holly bootwrapper
  [POWERPC] Holly DTS
  [POWERPC] Holly defconfig
  [POWERPC] Add support for 750CL Holly board
  [POWERPC] Generalize tsi108 PCI setup
  [POWERPC] Generalize tsi108 PHY types
  ...

Fixed conflict in include/asm-powerpc/kdebug.h manually

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoRemove unused variable in get_unmapped_area
Roland McGrath [Tue, 8 May 2007 18:19:38 +0000 (11:19 -0700)]
Remove unused variable in get_unmapped_area

Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge branch 'for-2.6.22' of git://git.kernel.dk/data/git/linux-2.6-block
Linus Torvalds [Tue, 8 May 2007 18:34:52 +0000 (11:34 -0700)]
Merge branch 'for-2.6.22' of git://git.kernel.dk/data/git/linux-2.6-block

* 'for-2.6.22' of git://git.kernel.dk/data/git/linux-2.6-block:
  [PATCH] ll_rw_blk: fix missing bounce in blk_rq_map_kern()
  [PATCH] splice: always call into page_cache_readahead()
  [PATCH] splice(): fix interaction with readahead

17 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/shaggy...
Linus Torvalds [Tue, 8 May 2007 18:32:30 +0000 (11:32 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/shaggy/jfs-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/shaggy/jfs-2.6:
  JFS: Fix race waking up jfsIO kernel thread
  JFS: use __set_current_state()
  Copy i_flags to jfs inode flags on write
  JFS: document uid, gid, and umask mount options in jfs.txt

17 years agosm501fb printk warning fixes
Andrew Morton [Tue, 8 May 2007 07:40:22 +0000 (00:40 -0700)]
sm501fb printk warning fixes

drivers/video/sm501fb.c: In function 'sm501fb_cursor':
drivers/video/sm501fb.c:992: warning: format '%08x' expects type 'unsigned int', but argument 3 has type 'long unsigned int'
drivers/video/sm501fb.c:992: warning: format '%08x' expects type 'unsigned int', but argument 4 has type 'long unsigned int'

Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoxilinxfb: xilinx framebuffer device driver
Andrei Konovalov [Tue, 8 May 2007 07:40:18 +0000 (00:40 -0700)]
xilinxfb: xilinx framebuffer device driver

Add support for the video controller IP block included into Xilinx ML300 and
ML403 reference designs.

Signed-off-by: Andrei Konovalov <akonovalov@ru.mvista.com>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopm2fb: fix of jumps in pm2fb_probe
Krzysztof Helt [Tue, 8 May 2007 07:40:16 +0000 (00:40 -0700)]
pm2fb: fix of jumps in pm2fb_probe

This patch fixes incorrect targets of jumps when an error occurs in the
pm2fb_probe.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoskeletonfb: improvements
Krzysztof Helt [Tue, 8 May 2007 07:40:14 +0000 (00:40 -0700)]
skeletonfb: improvements

This patch adds a macro to register PCI ids table and corrects type of
xxxfb_fix variable to avoid modpost warnings.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopm2fb: removal of pm2fb_par fields
Krzysztof Helt [Tue, 8 May 2007 07:40:12 +0000 (00:40 -0700)]
pm2fb: removal of pm2fb_par fields

This patch removes two redundant fields in the pm2fb_par structure.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoskeletonfb: more corrections
Krzysztof Helt [Tue, 8 May 2007 07:40:10 +0000 (00:40 -0700)]
skeletonfb: more corrections

More corrections to skeletonfb.c file:
- fixed types of more arguments
- removed returned values in void functions
- removed not existing fb_poll function
- fixed closing comment typo
- corrected misleading constant name ADDR

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Acked-By: James Simmons <jsimmons@infradead.org>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agovga16fb: actually support widths in multiples of 8
Antonino A. Daplas [Tue, 8 May 2007 07:40:08 +0000 (00:40 -0700)]
vga16fb: actually support widths in multiples of 8

vga16fb does not just support 8-pixel wide rectangles, but rectangles with
widths in multiples of 8.

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agovgacon: disallow console operations when in KD_GRAPHICS mode
Antonino A. Daplas [Tue, 8 May 2007 07:40:06 +0000 (00:40 -0700)]
vgacon: disallow console operations when in KD_GRAPHICS mode

Reported by James Pearson as:

 boot to run level 3

 if not root, then make sure /dev/console is writeable

 login and type:

 setterm -blank 0

 start X

 type into an xterm:

 while true; do echo "" > /dev/console; usleep 100000; done

 while the above loop is running switch to the text console and back
 again (Ctrl-Alt-F1 then Ctrl-Alt-F7)

 ... and the screen will be shifting (and wrapping) to the left.

This problem stems from continuously writing text to the system console (which
is in KD_TEXT mode) while the foreground console is in KD_GRAPHICS
mode. Somewhere along the way, console printing got confused and omitted the
KD_GRAPHICS/KD_TEXT test.  Thus, vgacon attempted to scroll the screen of X,
which causes X to shift.

Fix by disallowing vgacon to touch the hardware when the vc is in KD_GRAPHICS
mode. A definitive fix entails a full audit of the console code.

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUse menuconfig objects II: video/logo
Jan Engelhardt [Tue, 8 May 2007 07:40:04 +0000 (00:40 -0700)]
Use menuconfig objects II: video/logo

Change Kconfig objects from "menu, config" into "menuconfig" so that the user
can disable the whole feature without having to enter the menu first.

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopm2fb: accelerated fillrect and copyarea
Krzysztof Helt [Tue, 8 May 2007 07:40:02 +0000 (00:40 -0700)]
pm2fb: accelerated fillrect and copyarea

This is a port of accelerated functions from 2.4 kernel.  Only fillrect and
copyarea are accelerated.  Fillrect is not accelerated in 24-bit mode.

[adaplas]
Add appropriate flags

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agos3fb: updates
Ondrej Zajicek [Tue, 8 May 2007 07:40:00 +0000 (00:40 -0700)]
s3fb: updates

Move s3fb_get_tilemax to svgalib.c as svga_get_tilemax, because it reports
limitation of other code from svgalib (svga_settile, svga_tilecopy, ...)

Limit font width to 8 pixels in 4 bpp mode.

Signed-off-by: Ondrej Zajicek <santiago@crfreenet.org>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopm2fb: memclock setting corrections
Krzysztof Helt [Tue, 8 May 2007 07:39:57 +0000 (00:39 -0700)]
pm2fb: memclock setting corrections

This patch disables a memory clock setting if a board has been initialized by
BIOS.  This allows using the memory clock set by manufacturer of the board.

This patch also sets default clock for 3dlabs Permedia 2V reference board's
clock to 75MHz (BIOS setting for EONtronic Permedia 2Vboard), because the
default 83MHz can be too high.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopm2fb: reset transparency settings
krzysztof.h1@wp.pl [Tue, 8 May 2007 07:39:56 +0000 (00:39 -0700)]
pm2fb: reset transparency settings

This patch resets transparency settings when depth changes.  Otherwise the 16
and 24-bit modes work incorrectly after switching from 32-bit mode.

Signed-off-by: Krzysztof Helt < krzysztof.h1@wp.pl>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agofbcon: check console-fb mapping in fbcon_get_requirement
Antonino A. Daplas [Tue, 8 May 2007 07:39:54 +0000 (00:39 -0700)]
fbcon: check console-fb mapping in fbcon_get_requirement

- Check the console-to-fb mapping in fbcon_get_requirement(), otherwise the
  value returned may not be valid for the driver.

- Minor cleanup

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agofbdev: clean up exit patch of fb_set_var
Antonino A. Daplas [Tue, 8 May 2007 07:39:52 +0000 (00:39 -0700)]
fbdev: clean up exit patch of fb_set_var

Clean up exit patch of fb_set_var():

- consolidate all return values into a single local variable
- ensure that return values are valid error codes
- fix fb_set_var() returning success when fb_check_caps() failed

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodrivers/video/sis/: remove more kernel 2.4 code
Adrian Bunk [Tue, 8 May 2007 07:39:50 +0000 (00:39 -0700)]
drivers/video/sis/: remove more kernel 2.4 code

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Cc: Thomas Winischhofer <thomas@winischhofer.net>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agouse mutex instead of semaphore in virtual console driver
Matthias Kaehlcke [Tue, 8 May 2007 07:39:49 +0000 (00:39 -0700)]
use mutex instead of semaphore in virtual console driver

The virtual console driver uses a semaphore as mutex.  Use the mutex API
instead of the (binary) semaphore.

Signed-off-by: Matthias Kaehlcke <matthias.kaehlcke@gmail.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoatyfb: halve XCLK with Mobility and 32bit memory
Ville Syrjala [Tue, 8 May 2007 07:39:47 +0000 (00:39 -0700)]
atyfb: halve XCLK with Mobility and 32bit memory

Laptops with Rage Mobility and 32bit memory interface seem to require halved
XCLK to operate correctly.

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoatyfb: reorganize clock init
Ville Syrjala [Tue, 8 May 2007 07:39:45 +0000 (00:39 -0700)]
atyfb: reorganize clock init

Reorganize atyfb clock init code so command line clock overrides are effective
for all chips.  The old code would silently ignore some of the command line
clock overrides with some chips.

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoatyfb: increase SPLL delay
Ville Syrjala [Tue, 8 May 2007 07:39:44 +0000 (00:39 -0700)]
atyfb: increase SPLL delay

Wait 5 ms instead of 500 us for the SPLL to lock.  This matches the
recommendation in mach64 programmer's guide.

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agonvidiafb: fix return value of nvidiafb_open()
Antonino A. Daplas [Tue, 8 May 2007 07:39:42 +0000 (00:39 -0700)]
nvidiafb: fix return value of nvidiafb_open()

Fix return value of nvidiafb_open().

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agovga: vgastate fix
Ondrej Zajicek [Tue, 8 May 2007 07:39:41 +0000 (00:39 -0700)]
vga: vgastate fix

1) sets 'palette access disabled' during read from AR10.
   This is usually documented as needed for access AR01-AR0F,
   but on ARK Logic card it is needed for AR10 (otherwise read
   returns some nonsence and save_vga_text() returns before do
   anything).

2) do not restore AR10, because it was not changed during

3) remove modification of misc reg:
        /* force graphics mode */
        vga_w(state->vgabase, VGA_MIS_W, misc | 1);

   as comment is misleading - LSB of misc reg does not set/reset graphics
   mode, but set color/mono adresses of CRT and some other regs.
   but these regs are not used during save/restore fonts.
   (it worked even when (misc | 1) was replaced by (misc & ~1) ).

Signed-off-by: Ondrej Zajicek <santiago@crfreenet.org>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agos3fb: implement fb_get_caps
Antonino A. Daplas [Tue, 8 May 2007 07:39:39 +0000 (00:39 -0700)]
s3fb: implement fb_get_caps

Implement fb_get_caps().  This will allow the driver to tell upper layer that
it can only support 256 8x16 bitmaps when in tileblitting mode
(var->bits_per_pixel == 0);

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agofbdev: fbcon: check if mode can handle new screen
Antonino A. Daplas [Tue, 8 May 2007 07:39:37 +0000 (00:39 -0700)]
fbdev: fbcon: check if mode can handle new screen

Check if the mode can properly display the screen.  This will be needed by
drivers where the capability is not constant with each mode.  The function
fb_set_var() will query fbcon the requirement, then it will query the driver
(via a new hook fb_get_caps()) its capability.  If the driver's capability
cannot handle fbcon's requirement, then fb_set_var() will fail.

For example, if a particular driver supports 2 modes where:

mode1 = can only display 8x16 bitmaps
mode2 = can display any bitmap

then if current mode = mode2 and current font = 12x22

fbset <mode1> /* mode1 cannot handle 12x22 */
fbset will fail

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>