8 years agoMerge branch 'omap-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind...
Linus Torvalds [Thu, 26 May 2011 19:11:54 +0000]
Merge branch 'omap-for-linus' of git://git./linux/kernel/git/tmlind/linux-omap-2.6

* 'omap-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap-2.6: (33 commits)
  OMAP3: PM: Boot message is not an error, and not helpful, remove it
  OMAP3: cpuidle: change the power domains modes determination logic
  OMAP3: cpuidle: code rework for improved readability
  OMAP3: cpuidle: re-organize the C-states data
  OMAP3: clean-up mach specific cpuidle data structures
  OMAP3 cpuidle: remove useless SDP specific timings
  usb: otg: OMAP4430: Powerdown the internal PHY when USB is disabled
  usb: otg: OMAP4430: Fixing the omap4430_phy_init function
  usb: musb: am35x: fix compile error when building am35x
  usb: musb: OMAP4430: Power down the PHY during board init
  omap: drop board-igep0030.c
  omap: igep0020: add support for IGEP3
  omap: igep0020: minor refactoring
  omap: igep0020: name refactoring for future merge with IGEP3
  omap: Remove support for omap2evm
  arm: omap2plus: GPIO cleanup
  omap: musb: introduce default board config
  omap: move detection of NAND CS to common-board-devices
  omap: use common initialization for PMIC i2c bus
  omap: consolidate touch screen initialization among different boards
  ...

8 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
Linus Torvalds [Thu, 26 May 2011 19:11:17 +0000]
Merge branch 'merge' of git://git./linux/kernel/git/benh/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc:
  powerpc/4xx: Adding PCIe MSI support
  powerpc: Fix irq_free_virt by adjusting bounds before loop
  powerpc/irq: Protect irq_radix_revmap_lookup against irq_free_virt
  powerpc/irq: Check desc in handle_one_irq and expand generic_handle_irq
  powerpc/irq: Always free duplicate IRQ_LEGACY hosts
  powerpc/irq: Remove stale and misleading comment
  powerpc/cell: Rename ipi functions to match current abstractions
  powerpc/cell: Use common smp ipi actions
  Remove unused MSG_ flags in linux/smp.h
  powerpc/pseries: Update MAX_HCALL_OPCODE to reflect page coalescing
  powerpc/oprofile: Handle events that raise an exception without overflowing
  powerpc/ftrace: Implement raw syscall tracepoints on PowerPC

8 years agoFix build with !HUGETLBFS
Linus Torvalds [Thu, 26 May 2011 19:03:50 +0000]
Fix build with !HUGETLBFS

I stupidly broke the case of CONFIG_HUGETLBFS=n when doing the
conversion to vm_flags_t in commit ca16d140af91 ("mm: don't access
vm_flags as 'int'").  And my 'allyesconfig' build didn't find it, for
obvious reasons..

Include <linux/mm_types.h> in <linux/hugetlb.h>.  The problem could have
been avoided by just turning the hugetlb_file_setup() error wrapper into
a macro, but mm_types.h is a reasonable include in this file.

Reported-by: Richard -rw- Weinberger <richard.weinberger@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

8 years agoMerge branch 'linux-next' of git://git.kernel.org/pub/scm/linux/kernel/git/jlbec...
Linus Torvalds [Thu, 26 May 2011 17:55:15 +0000]
Merge branch 'linux-next' of git://git./linux/kernel/git/jlbec/ocfs2

* 'linux-next' of git://git.kernel.org/pub/scm/linux/kernel/git/jlbec/ocfs2: (28 commits)
  Ocfs2: Teach local-mounted ocfs2 to handle unwritten_extents correctly.
  ocfs2/dlm: Do not migrate resource to a node that is leaving the domain
  ocfs2/dlm: Add new dlm message DLM_BEGIN_EXIT_DOMAIN_MSG
  Ocfs2/move_extents: Set several trivial constraints for threshold.
  Ocfs2/move_extents: Let defrag handle partial extent moving.
  Ocfs2/move_extents: move/defrag extents within a certain range.
  Ocfs2/move_extents: helper to calculate the defraging length in one run.
  Ocfs2/move_extents: move entire/partial extent.
  Ocfs2/move_extents: helpers to update the group descriptor and global bitmap inode.
  Ocfs2/move_extents: helper to probe a proper region to move in an alloc group.
  Ocfs2/move_extents: helper to validate and adjust moving goal.
  Ocfs2/move_extents: find the victim alloc group, where the given #blk fits.
  Ocfs2/move_extents: defrag a range of extent.
  Ocfs2/move_extents: move a range of extent.
  Ocfs2/move_extents: lock allocators and reserve metadata blocks and data clusters for extents moving.
  Ocfs2/move_extents: Add basic framework and source files for extent moving.
  Ocfs2/move_extents: Adding new ioctl code 'OCFS2_IOC_MOVE_EXT' to ocfs2.
  Ocfs2/refcounttree: Publicize couple of funcs from refcounttree.c
  Ocfs2: Add a new code 'OCFS2_INFO_FREEFRAG' for o2info ioctl.
  Ocfs2: Add a new code 'OCFS2_INFO_FREEINODE' for o2info ioctl.
  ...

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/djm/tmem
Linus Torvalds [Thu, 26 May 2011 17:50:56 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/djm/tmem

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/djm/tmem:
  xen: cleancache shim to Xen Transcendent Memory
  ocfs2: add cleancache support
  ext4: add cleancache support
  btrfs: add cleancache support
  ext3: add cleancache support
  mm/fs: add hooks to support cleancache
  mm: cleancache core ops functions and config
  fs: add field to superblock to support cleancache
  mm/fs: cleancache documentation

Fix up trivial conflict in fs/btrfs/extent_io.c due to includes

8 years agoMerge branch 'for-linus' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Thu, 26 May 2011 17:49:11 +0000]
Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs

* 'for-linus' of git://oss.sgi.com/xfs/xfs:
  xfs: correctly decrement the extent buffer index in xfs_bmap_del_extent
  xfs: check for valid indices in xfs_iext_get_ext and xfs_iext_idx_to_irec
  xfs: fix up asserts in xfs_iflush_fork
  xfs: do not do pointer arithmetic on extent records
  xfs: do not use unchecked extent indices in xfs_bunmapi
  xfs: do not use unchecked extent indices in xfs_bmapi
  xfs: do not use unchecked extent indices in xfs_bmap_add_extent_*
  xfs: remove if_lastex
  xfs: remove the unused XFS_BMAPI_RSVBLOCKS flag
  xfs: do not discard alloc btree blocks
  xfs: add online discard support

8 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Linus Torvalds [Thu, 26 May 2011 16:53:20 +0000]
Merge branch 'for_linus' of git://git./linux/kernel/git/tytso/ext4

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4: (61 commits)
  jbd2: Add MAINTAINERS entry
  jbd2: fix a potential leak of a journal_head on an error path
  ext4: teach ext4_ext_split to calculate extents efficiently
  ext4: Convert ext4 to new truncate calling convention
  ext4: do not normalize block requests from fallocate()
  ext4: enable "punch hole" functionality
  ext4: add "punch hole" flag to ext4_map_blocks()
  ext4: punch out extents
  ext4: add new function ext4_block_zero_page_range()
  ext4: add flag to ext4_has_free_blocks
  ext4: reserve inodes and feature code for 'quota' feature
  ext4: add support for multiple mount protection
  ext4: ensure f_bfree returned by ext4_statfs() is non-negative
  ext4: protect bb_first_free in ext4_trim_all_free() with group lock
  ext4: only load buddy bitmap in ext4_trim_fs() when it is needed
  jbd2: Fix comment to match the code in jbd2__journal_start()
  ext4: fix waiting and sending of a barrier in ext4_sync_file()
  jbd2: Add function jbd2_trans_will_send_data_barrier()
  jbd2: fix sending of data flush on journal commit
  ext4: fix ext4_ext_fiemap_cb() to handle blocks before request range correctly
  ...

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6
Linus Torvalds [Thu, 26 May 2011 16:52:14 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6: (25 commits)
  cifs: remove unnecessary dentry_unhash on rmdir/rename_dir
  ocfs2: remove unnecessary dentry_unhash on rmdir/rename_dir
  exofs: remove unnecessary dentry_unhash on rmdir/rename_dir
  nfs: remove unnecessary dentry_unhash on rmdir/rename_dir
  ext2: remove unnecessary dentry_unhash on rmdir/rename_dir
  ext3: remove unnecessary dentry_unhash on rmdir/rename_dir
  ext4: remove unnecessary dentry_unhash on rmdir/rename_dir
  btrfs: remove unnecessary dentry_unhash in rmdir/rename_dir
  ceph: remove unnecessary dentry_unhash calls
  vfs: clean up vfs_rename_other
  vfs: clean up vfs_rename_dir
  vfs: clean up vfs_rmdir
  vfs: fix vfs_rename_dir for FS_RENAME_DOES_D_MOVE filesystems
  libfs: drop unneeded dentry_unhash
  vfs: update dentry_unhash() comment
  vfs: push dentry_unhash on rename_dir into file systems
  vfs: push dentry_unhash on rmdir into file systems
  vfs: remove dget() from dentry_unhash()
  vfs: dentry_unhash immediately prior to rmdir
  vfs: Block mmapped writes while the fs is frozen
  ...

8 years agomm: don't access vm_flags as 'int'
KOSAKI Motohiro [Thu, 26 May 2011 10:16:19 +0000]
mm: don't access vm_flags as 'int'

The type of vma->vm_flags is 'unsigned long'. Neither 'int' nor
'unsigned int'. This patch fixes such misuse.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
[ Changed to use a typedef - we'll extend it to cover more cases
  later, since there has been discussion about making it a 64-bit
  type..                      - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

8 years agoxen: cleancache shim to Xen Transcendent Memory
Dan Magenheimer [Thu, 26 May 2011 16:02:21 +0000]
xen: cleancache shim to Xen Transcendent Memory

This patch provides a shim between the kernel-internal cleancache
API (see Documentation/mm/cleancache.txt) and the Xen Transcendent
Memory ABI (see http://oss.oracle.com/projects/tmem).

Xen tmem provides "hypervisor RAM" as an ephemeral page-oriented
pseudo-RAM store for cleancache pages, shared cleancache pages,
and frontswap pages.  Tmem provides enterprise-quality concurrency,
full save/restore and live migration support, compression
and deduplication.

A presentation showing up to 8% faster performance and up to 52%
reduction in sectors read on a kernel compile workload, despite
aggressive in-kernel page reclamation ("self-ballooning") can be
found at:

http://oss.oracle.com/projects/tmem/dist/documentation/presentations/TranscendentMemoryXenSummit2010.pdf

Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Reviewed-by: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik Van Riel <riel@redhat.com>
Cc: Jan Beulich <JBeulich@novell.com>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Andreas Dilger <adilger@sun.com>
Cc: Ted Ts'o <tytso@mit.edu>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <joel.becker@oracle.com>
Cc: Nitin Gupta <ngupta@vflare.org>

8 years agoocfs2: add cleancache support
Dan Magenheimer [Thu, 26 May 2011 16:02:08 +0000]
ocfs2: add cleancache support

This eighth patch of eight in this cleancache series "opts-in"
cleancache for ocfs2.  Clustered filesystems must explicitly enable
cleancache by calling cleancache_init_shared_fs anytime an instance
of the filesystem is mounted.  Ocfs2 is currently the only user of
the clustered filesystem interface but nevertheless, the cleancache
hooks in the VFS layer are sufficient for ocfs2 including the matching
cleancache_flush_fs hook which must be called on unmount.

Details and a FAQ can be found in Documentation/vm/cleancache.txt

[v8: trivial merge conflict update]
[v5: jeremy@goop.org: simplify init hook and any future fs init changes]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Reviewed-by: Jeremy Fitzhardinge <jeremy@goop.org>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik Van Riel <riel@redhat.com>
Cc: Jan Beulich <JBeulich@novell.com>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Andreas Dilger <adilger@sun.com>
Cc: Ted Tso <tytso@mit.edu>
Cc: Nitin Gupta <ngupta@vflare.org>

8 years agoext4: add cleancache support
Dan Magenheimer [Thu, 26 May 2011 16:02:03 +0000]
ext4: add cleancache support

This seventh patch of eight in this cleancache series "opts-in"
cleancache for ext4.  Filesystems must explicitly enable cleancache
by calling cleancache_init_fs anytime an instance of the filesystem
is mounted. For ext4, all other cleancache hooks are in
the VFS layer including the matching cleancache_flush_fs
hook which must be called on unmount.

Details and a FAQ can be found in Documentation/vm/cleancache.txt

[v6-v8: no changes]
[v5: jeremy@goop.org: simplify init hook and any future fs init changes]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Reviewed-by: Jeremy Fitzhardinge <jeremy@goop.org>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Acked-by: Andreas Dilger <adilger@sun.com>
Cc: Ted Ts'o <tytso@mit.edu>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik Van Riel <riel@redhat.com>
Cc: Jan Beulich <JBeulich@novell.com>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <joel.becker@oracle.com>
Cc: Nitin Gupta <ngupta@vflare.org>

8 years agobtrfs: add cleancache support
Dan Magenheimer [Thu, 26 May 2011 16:01:56 +0000]
btrfs: add cleancache support

This sixth patch of eight in this cleancache series "opts-in"
cleancache for btrfs.  Filesystems must explicitly enable
cleancache by calling cleancache_init_fs anytime an instance
of the filesystem is mounted.  Btrfs uses its own readpage
which must be hooked, but all other cleancache hooks are in
the VFS layer including the matching cleancache_flush_fs hook
which must be called on unmount.

Details and a FAQ can be found in Documentation/vm/cleancache.txt

[v6-v8: no changes]
[v5: jeremy@goop.org: simplify init hook and any future fs init changes]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Reviewed-by: Jeremy Fitzhardinge <jeremy@goop.org>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik Van Riel <riel@redhat.com>
Cc: Jan Beulich <JBeulich@novell.com>
Cc: Andreas Dilger <adilger@sun.com>
Cc: Ted Ts'o <tytso@mit.edu>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <joel.becker@oracle.com>
Cc: Nitin Gupta <ngupta@vflare.org>

8 years agoext3: add cleancache support
Dan Magenheimer [Thu, 26 May 2011 16:01:49 +0000]
ext3: add cleancache support

This fifth patch of eight in this cleancache series "opts-in"
cleancache for ext3.  Filesystems must explicitly enable
cleancache by calling cleancache_init_fs anytime an instance
of the filesystem is mounted. For ext3, all other cleancache
hooks are in the VFS layer including the matching cleancache_flush_fs
hook which must be called on unmount.

Details and a FAQ can be found in Documentation/vm/cleancache.txt

[v6-v8: no changes]
[v5: jeremy@goop.org: simplify init hook and any future fs init changes]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Reviewed-by: Jeremy Fitzhardinge <jeremy@goop.org>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Acked-by: Andreas Dilger <adilger@sun.com>
Cc: Ted Ts'o <tytso@mit.edu>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik Van Riel <riel@redhat.com>
Cc: Jan Beulich <JBeulich@novell.com>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <joel.becker@oracle.com>
Cc: Nitin Gupta <ngupta@vflare.org>

8 years agomm/fs: add hooks to support cleancache
Dan Magenheimer [Thu, 26 May 2011 16:01:43 +0000]
mm/fs: add hooks to support cleancache

This fourth patch of eight in this cleancache series provides the
core hooks in VFS for: initializing cleancache per filesystem;
capturing clean pages reclaimed by page cache; attempting to get
pages from cleancache before filesystem read; and ensuring coherency
between pagecache, disk, and cleancache.  Note that the placement
of these hooks was stable from 2.6.18 to 2.6.38; a minor semantic
change was required due to a patchset in 2.6.39.

All hooks become no-ops if CONFIG_CLEANCACHE is unset, or become
a check of a boolean global if CONFIG_CLEANCACHE is set but no
cleancache "backend" has claimed cleancache_ops.

Details and a FAQ can be found in Documentation/vm/cleancache.txt

[v8: minchan.kim@gmail.com: adapt to new remove_from_page_cache function]
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Reviewed-by: Jeremy Fitzhardinge <jeremy@goop.org>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik Van Riel <riel@redhat.com>
Cc: Jan Beulich <JBeulich@novell.com>
Cc: Andreas Dilger <adilger@sun.com>
Cc: Ted Ts'o <tytso@mit.edu>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <joel.becker@oracle.com>
Cc: Nitin Gupta <ngupta@vflare.org>

8 years agomm: cleancache core ops functions and config
Dan Magenheimer [Thu, 26 May 2011 16:01:36 +0000]
mm: cleancache core ops functions and config

This third patch of eight in this cleancache series provides
the core code for cleancache that interfaces between the hooks in
VFS and individual filesystems and a cleancache backend.  It also
includes build and config patches.

Two new files are added: mm/cleancache.c and include/linux/cleancache.h.

Note that CONFIG_CLEANCACHE can default to on; in systems that do
not provide a cleancache backend, all hooks devolve to a simple
check of a global enable flag, so performance impact should
be negligible but can be reduced to zero impact if config'ed off.
However for this first commit, it defaults to off.

Details and a FAQ can be found in Documentation/vm/cleancache.txt

Credits: Cleancache_ops design derived from Jeremy Fitzhardinge
design for tmem

[v8: dan.magenheimer@oracle.com: fix exportfs call affecting btrfs]
[v8: akpm@linux-foundation.org: use static inline function, not macro]
[v7: dan.magenheimer@oracle.com: cleanup sysfs and remove cleancache prefix]
[v6: JBeulich@novell.com: robustly handle buggy fs encode_fh actor definition]
[v5: jeremy@goop.org: clean up global usage and static var names]
[v5: jeremy@goop.org: simplify init hook and any future fs init changes]
[v5: hch@infradead.org: cleaner non-global interface for ops registration]
[v4: adilger@sun.com: interface must support exportfs FS's]
[v4: hch@infradead.org: interface must support 64-bit FS on 32-bit kernel]
[v3: akpm@linux-foundation.org: use one ops struct to avoid pointer hops]
[v3: akpm@linux-foundation.org: document and ensure PageLocked reqts are met]
[v3: ngupta@vflare.org: fix success/fail codes, change funcs to void]
[v2: viro@ZenIV.linux.org.uk: use sane types]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Reviewed-by: Jeremy Fitzhardinge <jeremy@goop.org>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Acked-by: Al Viro <viro@ZenIV.linux.org.uk>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Acked-by: Minchan Kim <minchan.kim@gmail.com>
Acked-by: Andreas Dilger <adilger@sun.com>
Acked-by: Jan Beulich <JBeulich@novell.com>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik Van Riel <riel@redhat.com>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Ted Ts'o <tytso@mit.edu>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <joel.becker@oracle.com>

8 years agofs: add field to superblock to support cleancache
Dan Magenheimer [Thu, 26 May 2011 16:01:19 +0000]
fs: add field to superblock to support cleancache

This second patch of eight in this cleancache series adds a field to
the generic superblock to squirrel away a pool identifier that is
dynamically provided by cleancache-enabled filesystems at mount time
to uniquely identify files and pages belonging to this mounted filesystem.

Details and a FAQ can be found in Documentation/vm/cleancache.txt

[v8: trivial merge conflict update]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Reviewed-by: Jeremy Fitzhardinge <jeremy@goop.org>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik Van Riel <riel@redhat.com>
Cc: Jan Beulich <JBeulich@novell.com>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Andreas Dilger <adilger@sun.com>
Cc: Ted Ts'o <tytso@mit.edu>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <joel.becker@oracle.com>
Cc: Nitin Gupta <ngupta@vflare.org>

8 years agomm/fs: cleancache documentation
Dan Magenheimer [Thu, 26 May 2011 16:00:56 +0000]
mm/fs: cleancache documentation

This patchset introduces cleancache, an optional new feature exposed
by the VFS layer that potentially dramatically increases page cache
effectiveness for many workloads in many environments at a negligible
cost.  It does this by providing an interface to transcendent memory,
which is memory/storage that is not otherwise visible to and/or directly
addressable by the kernel.

Instead of being discarded, hooks in the reclaim code "put" clean
pages to cleancache.  Filesystems that "opt-in" may "get" pages
from cleancache that were previously put, but pages in cleancache are
"ephemeral", meaning they may disappear at any time. And the size
of cleancache is entirely dynamic and unknowable to the kernel.
Filesystems currently supported by this patchset include ext3, ext4,
btrfs, and ocfs2.  Other filesystems (especially those built entirely
on VFS) should be easy to add, but should first be thoroughly tested to
ensure coherency.

Details and a FAQ are provided in Documentation/vm/cleancache.txt

This first patch of eight in this cleancache series only adds two
new documentation files.

[v8: minor documentation changes by author]
[v3: akpm@linux-foundation.org: document sysfs API]
[v3: hch@infradead.org: move detailed description to Documentation/vm]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Reviewed-by: Jeremy Fitzhardinge <jeremy@goop.org>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik Van Riel <riel@redhat.com>
Cc: Jan Beulich <JBeulich@novell.com>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Andreas Dilger <adilger@sun.com>
Cc: Ted Ts'o <tytso@mit.edu>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <joel.becker@oracle.com>
Cc: Nitin Gupta <ngupta@vflare.org>

8 years agojbd2: Add MAINTAINERS entry
Theodore Ts'o [Thu, 26 May 2011 13:53:09 +0000]
jbd2: Add MAINTAINERS entry

Create a separate MAINTAINERS entry for jbd2

Cc: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>

8 years agocifs: remove unnecessary dentry_unhash on rmdir/rename_dir
Sage Weil [Tue, 24 May 2011 20:06:22 +0000]
cifs: remove unnecessary dentry_unhash on rmdir/rename_dir

Cifs has no problems with lingering references to unlinked directory
inodes.

CC: Steve French <sfrench@samba.org>
CC: linux-cifs@vger.kernel.org
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agoocfs2: remove unnecessary dentry_unhash on rmdir/rename_dir
Sage Weil [Tue, 24 May 2011 20:06:21 +0000]
ocfs2: remove unnecessary dentry_unhash on rmdir/rename_dir

Ocfs2 has no issues with lingering references to unlinked directory inodes.

CC: Mark Fasheh <mfasheh@suse.com>
CC: ocfs2-devel@oss.oracle.com
Acked-by: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agoexofs: remove unnecessary dentry_unhash on rmdir/rename_dir
Sage Weil [Tue, 24 May 2011 20:06:20 +0000]
exofs: remove unnecessary dentry_unhash on rmdir/rename_dir

Exofs has no problems with lingering references to unlinked directory
inodes.

CC: Benny Halevy <bhalevy@panasas.com>
CC: osd-dev@open-osd.org
Acked-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agonfs: remove unnecessary dentry_unhash on rmdir/rename_dir
Sage Weil [Tue, 24 May 2011 20:06:19 +0000]
nfs: remove unnecessary dentry_unhash on rmdir/rename_dir

NFS has no problems with lingering references to unlinked directory
inodes.

CC: Trond Myklebust <Trond.Myklebust@netapp.com>
CC: linux-nfs@vger.kernel.org
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agoext2: remove unnecessary dentry_unhash on rmdir/rename_dir
Sage Weil [Tue, 24 May 2011 20:06:18 +0000]
ext2: remove unnecessary dentry_unhash on rmdir/rename_dir

ext2 has no problems with lingering references to unlinked directory
inodes.

CC: Jan Kara <jack@suse.cz>
CC: linux-ext4@vger.kernel.org
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agoext3: remove unnecessary dentry_unhash on rmdir/rename_dir
Sage Weil [Tue, 24 May 2011 20:06:17 +0000]
ext3: remove unnecessary dentry_unhash on rmdir/rename_dir

ext3 has no problems with lingering references to unlinked directory
inodes.

CC: Jan Kara <jack@suse.cz>
CC: Andrew Morton <akpm@linux-foundation.org>
CC: Andreas Dilger <adilger.kernel@dilger.ca>
CC: linux-ext4@vger.kernel.org
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agoext4: remove unnecessary dentry_unhash on rmdir/rename_dir
Sage Weil [Tue, 24 May 2011 20:06:16 +0000]
ext4: remove unnecessary dentry_unhash on rmdir/rename_dir

ext4 has no problems with lingering references to unlinked directory
inodes.

CC: "Theodore Ts'o" <tytso@mit.edu>
CC: Andreas Dilger <adilger.kernel@dilger.ca>
CC: linux-ext4@vger.kernel.org
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agobtrfs: remove unnecessary dentry_unhash in rmdir/rename_dir
Sage Weil [Tue, 24 May 2011 20:06:15 +0000]
btrfs: remove unnecessary dentry_unhash in rmdir/rename_dir

Btrfs has no problems with lingering references to unlinked directory
inodes.

CC: Chris Mason <chris.mason@oracle.com>
CC: linux-btrfs@vger.kernel.org
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agoceph: remove unnecessary dentry_unhash calls
Sage Weil [Tue, 24 May 2011 20:06:14 +0000]
ceph: remove unnecessary dentry_unhash calls

Ceph does not need these, and they screw up our use of the dcache as a
consistent cache.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agovfs: clean up vfs_rename_other
Sage Weil [Tue, 24 May 2011 20:06:13 +0000]
vfs: clean up vfs_rename_other

Simplify control flow to match vfs_rename_dir.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agovfs: clean up vfs_rename_dir
Sage Weil [Tue, 24 May 2011 20:06:12 +0000]
vfs: clean up vfs_rename_dir

Simplify control flow through vfs_rename_dir.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agovfs: clean up vfs_rmdir
Sage Weil [Tue, 24 May 2011 20:06:11 +0000]
vfs: clean up vfs_rmdir

Simplify the control flow with an out label.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agovfs: fix vfs_rename_dir for FS_RENAME_DOES_D_MOVE filesystems
Miklos Szeredi [Tue, 24 May 2011 20:06:10 +0000]
vfs: fix vfs_rename_dir for FS_RENAME_DOES_D_MOVE filesystems

vfs_rename_dir() doesn't properly account for filesystems with
FS_RENAME_DOES_D_MOVE.  If new_dentry has a target inode attached, it
unhashes the new_dentry prior to the rename() iop and rehashes it after,
but doesn't account for the possibility that rename() may have swapped
{old,new}_dentry.  For FS_RENAME_DOES_D_MOVE filesystems, it rehashes
new_dentry (now the old renamed-from name, which d_move() expected to go
away), such that a subsequent lookup will find it.  Currently all
FS_RENAME_DOES_D_MOVE filesystems compensate for this by failing in
d_revalidate.

The bug was introduced by: commit 349457ccf2592c14bdf13b6706170ae2e94931b1
"[PATCH] Allow file systems to manually d_move() inside of ->rename()"

Fix by not rehashing the new dentry.  Rehashing used to be needed by
d_move() but isn't anymore.

Reported-by: Sage Weil <sage@newdream.net>
Signed-off-by: Miklos Szeredi <miklos@szeredi.hu>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agolibfs: drop unneeded dentry_unhash
Sage Weil [Tue, 24 May 2011 20:06:09 +0000]
libfs: drop unneeded dentry_unhash

There are no libfs issues with dangling references to empty directories.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agovfs: update dentry_unhash() comment
Sage Weil [Tue, 24 May 2011 20:06:08 +0000]
vfs: update dentry_unhash() comment

The helper is now only called by file systems, not the VFS.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agovfs: push dentry_unhash on rename_dir into file systems
Sage Weil [Tue, 24 May 2011 20:06:07 +0000]
vfs: push dentry_unhash on rename_dir into file systems

Only a few file systems need this.  Start by pushing it down into each
rename method (except gfs2 and xfs) so that it can be dealt with on a
per-fs basis.

Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agovfs: push dentry_unhash on rmdir into file systems
Sage Weil [Tue, 24 May 2011 20:06:06 +0000]
vfs: push dentry_unhash on rmdir into file systems

Only a few file systems need this.  Start by pushing it down into each
fs rmdir method (except gfs2 and xfs) so it can be dealt with on a per-fs
basis.

This does not change behavior for any in-tree file systems.

Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agovfs: remove dget() from dentry_unhash()
Sage Weil [Tue, 24 May 2011 20:06:05 +0000]
vfs: remove dget() from dentry_unhash()

This serves no useful purpose that I can discern.  All callers (rename,
rmdir) hold their own reference to the dentry.

A quick audit of all file systems showed no relevant checks on the value
of d_count in vfs_rmdir/vfs_rename_dir paths.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agovfs: dentry_unhash immediately prior to rmdir
Sage Weil [Tue, 24 May 2011 20:06:04 +0000]
vfs: dentry_unhash immediately prior to rmdir

This presumes that there is no reason to unhash a dentry if we fail because
it is a mountpoint or the LSM check fails, and that the LSM checks do not
depend on the dentry being unhashed.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agovfs: Block mmapped writes while the fs is frozen
Jan Kara [Mon, 23 May 2011 22:23:35 +0000]
vfs: Block mmapped writes while the fs is frozen

We should not allow file modification via mmap while the filesystem is
frozen. So block in block_page_mkwrite() while the filesystem is frozen.
We cannot do the blocking wait in __block_page_mkwrite() since e.g. ext4
will want to call that function with transaction started in some cases
and that would deadlock. But we can at least do the non-blocking reliable
check in __block_page_mkwrite() which is the hardest part anyway.

We have to check for frozen filesystem with the page marked dirty and under
page lock with which we then return from ->page_mkwrite(). Only that way we
cannot race with writeback done by freezing code - either we mark the page
dirty after the writeback has started, see freezing in progress and block, or
writeback will wait for our page lock which is released only when the fault is
done and then writeback will writeout and writeprotect the page again.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agovfs: Create __block_page_mkwrite() helper passing error values back
Jan Kara [Mon, 23 May 2011 22:23:34 +0000]
vfs: Create __block_page_mkwrite() helper passing error values back

Create __block_page_mkwrite() helper which does all what block_page_mkwrite()
does except that it passes back errors from __block_write_begin /
block_commit_write calls.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agofs/namespace.c: bound mount propagation fix
Roman Borisov [Wed, 25 May 2011 23:26:48 +0000]
fs/namespace.c: bound mount propagation fix

This issue was discovered by users of busybox.  And the bug is actual for
busybox users, I don't know how it affects others.  Apparently, mount is
called with and without MS_SILENT, and this affects mount() behaviour.
But MS_SILENT is only supposed to affect kernel logging verbosity.

The following script was run in an empty test directory:

mkdir -p mount.dir mount.shared1 mount.shared2
touch mount.dir/a mount.dir/b
mount -vv --bind         mount.shared1 mount.shared1
mount -vv --make-rshared mount.shared1
mount -vv --bind         mount.shared2 mount.shared2
mount -vv --make-rshared mount.shared2
mount -vv --bind mount.shared2 mount.shared1
mount -vv --bind mount.dir     mount.shared2
ls -R mount.dir mount.shared1 mount.shared2
umount mount.dir mount.shared1 mount.shared2 2>/dev/null
umount mount.dir mount.shared1 mount.shared2 2>/dev/null
umount mount.dir mount.shared1 mount.shared2 2>/dev/null
rm -f mount.dir/a mount.dir/b mount.dir/c
rmdir mount.dir mount.shared1 mount.shared2

mount -vv was used to show the mount() call arguments and result.
Output shows that flag argument has 0x00008000 = MS_SILENT bit:

mount: mount('mount.shared1','mount.shared1','(null)',0x00009000,'(null)'):0
mount: mount('','mount.shared1','',0x0010c000,''):0
mount: mount('mount.shared2','mount.shared2','(null)',0x00009000,'(null)'):0
mount: mount('','mount.shared2','',0x0010c000,''):0
mount: mount('mount.shared2','mount.shared1','(null)',0x00009000,'(null)'):0
mount: mount('mount.dir','mount.shared2','(null)',0x00009000,'(null)'):0
mount.dir:
a
b

mount.shared1:

mount.shared2:
a
b

After adding --loud option to remove MS_SILENT bit from just one mount cmd:

mkdir -p mount.dir mount.shared1 mount.shared2
touch mount.dir/a mount.dir/b
mount -vv --bind         mount.shared1 mount.shared1 2>&1
mount -vv --make-rshared mount.shared1               2>&1
mount -vv --bind         mount.shared2 mount.shared2 2>&1
mount -vv --loud --make-rshared mount.shared2               2>&1  # <-HERE
mount -vv --bind mount.shared2 mount.shared1         2>&1
mount -vv --bind mount.dir     mount.shared2         2>&1
ls -R mount.dir mount.shared1 mount.shared2      2>&1
umount mount.dir mount.shared1 mount.shared2 2>/dev/null
umount mount.dir mount.shared1 mount.shared2 2>/dev/null
umount mount.dir mount.shared1 mount.shared2 2>/dev/null
rm -f mount.dir/a mount.dir/b mount.dir/c
rmdir mount.dir mount.shared1 mount.shared2

The result is different now - look closely at mount.shared1 directory listing.
Now it does show files 'a' and 'b':

mount: mount('mount.shared1','mount.shared1','(null)',0x00009000,'(null)'):0
mount: mount('','mount.shared1','',0x0010c000,''):0
mount: mount('mount.shared2','mount.shared2','(null)',0x00009000,'(null)'):0
mount: mount('','mount.shared2','',0x00104000,''):0
mount: mount('mount.shared2','mount.shared1','(null)',0x00009000,'(null)'):0
mount: mount('mount.dir','mount.shared2','(null)',0x00009000,'(null)'):0

mount.dir:
a
b

mount.shared1:
a
b

mount.shared2:
a
b

The analysis shows that MS_SILENT flag which is ON by default in any
busybox-> mount operations cames to flags_to_propagation_type function and
causes the error return while is_power_of_2 checking because the function
expects only one bit set.  This doesn't allow to do busybox->mount with
any --make-[r]shared, --make-[r]private etc options.

Moreover, the recently added flags_to_propagation_type() function doesn't
allow us to do such operations as --make-[r]private --make-[r]shared etc.
when MS_SILENT is on.  The idea or clearing the MS_SILENT flag came from
to Denys Vlasenko.

Signed-off-by: Roman Borisov <ext-roman.borisov@nokia.com>
Reported-by: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Chuck Ebbert <cebbert@redhat.com>
Cc: Alexander Shishkin <virtuoso@slind.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agoexportfs: reallow building as a module
Jonas Gorski [Tue, 24 May 2011 18:12:08 +0000]
exportfs: reallow building as a module

Commit 990d6c2d7aee921e3bce22b2d6a750fd552262be ("vfs: Add name to file
handle conversion support") changed EXPORTFS to be a bool.
This was needed for earlier revisions of the original patch, but the actual
commit put the code needing it into its own file that only gets compiled
when FHANDLE is selected which in turn selects EXPORTFS.
So EXPORTFS can be safely compiled as a module when not selecting FHANDLE.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
Acked-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agomerge handle_reval_dot and nameidata_drop_rcu_last
Al Viro [Fri, 25 Mar 2011 15:00:12 +0000]
merge handle_reval_dot and nameidata_drop_rcu_last

new helper: complete_walk().  Done on successful completion
of walk, drops out of RCU mode, does d_revalidate of final
result if that hadn't been done already.

handle_reval_dot() and nameidata_drop_rcu_last() subsumed into
that one; callers converted to use of complete_walk().

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agoconsolidate nameidata_..._drop_rcu()
Al Viro [Fri, 25 Mar 2011 14:32:48 +0000]
consolidate nameidata_..._drop_rcu()

Merge these into a single function (unlazy_walk(nd, dentry)),
kill ..._maybe variants

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

8 years agopowerpc/4xx: Adding PCIe MSI support
Rupjyoti Sarmah [Tue, 29 Mar 2011 23:10:24 +0000]
powerpc/4xx: Adding PCIe MSI support

This patch adds MSI support for 440SPe, 460Ex, 460Sx and 405Ex.

Signed-off-by: Rupjyoti Sarmah <rsarmah@apm.com>
Signed-off-by: Tirumala R Marri <tmarri@apm.com>
Acked-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agoMerge branch 'move_extents' of git://oss.oracle.com/git/tye/linux-2.6 into ocfs2...
Joel Becker [Thu, 26 May 2011 04:51:55 +0000]
Merge branch 'move_extents' of git://oss.oracle.com/git/tye/linux-2.6 into ocfs2-merge-window

Conflicts:
fs/ocfs2/ioctl.c

8 years agoOcfs2: Teach local-mounted ocfs2 to handle unwritten_extents correctly.
Tristan Ye [Mon, 23 May 2011 07:57:26 +0000]
Ocfs2: Teach local-mounted ocfs2 to handle unwritten_extents correctly.

Oops, local-mounted of 'ocfs2_fops_no_plocks' is just missing the support
of unwritten_extents/punching-hole due to no func pointer was given correctly
to '.follocate' field.

Signed-off-by: Tristan Ye <tristan.ye@oracle.com>

8 years agoocfs2/dlm: Do not migrate resource to a node that is leaving the domain
Sunil Mushran [Thu, 19 May 2011 21:34:12 +0000]
ocfs2/dlm: Do not migrate resource to a node that is leaving the domain

During dlm domain shutdown, o2dlm has to free all the lock resources. Ones that
have no locks and references are freed. Ones that have locks and/or references
are migrated to another node.

The first task in migration is finding a target. Currently we scan the lock
resource and find one node that either has a lock or a reference. This is not
very efficient in a parallel umount case as we might end up migrating the
lock resource to a node which itself may have to migrate it to a third node.

The patch scans the dlm->exit_domain_map to ensure the target node is not
leaving the domain. If no valid target node is found, o2dlm does not migrate
the resource but instead waits for the unlock and deref messages that will
allow it to free the resource.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <jlbec@evilplan.org>

8 years agoocfs2/dlm: Add new dlm message DLM_BEGIN_EXIT_DOMAIN_MSG
Sunil Mushran [Thu, 19 May 2011 21:34:11 +0000]
ocfs2/dlm: Add new dlm message DLM_BEGIN_EXIT_DOMAIN_MSG

This patch adds a new dlm message DLM_BEGIN_EXIT_DOMAIN_MSG and ups the dlm
protocol to 1.2.

o2dlm sends this new message in dlm_unregister_domain() to mark the beginning
of the exit domain. This message is sent to all nodes in the domain.

Currently o2dlm has no way of informing other nodes of its impending exit.
This information is useful as the other nodes could disregard the exiting
node in certain operations. For example, in resource migration. If two or
more nodes were umounting in parallel, it would be more efficient if o2dlm
were to choose a non-exiting node to be the new master node rather than an
exiting one.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <jlbec@evilplan.org>

8 years agopowerpc: Fix irq_free_virt by adjusting bounds before loop
Milton Miller [Tue, 24 May 2011 20:34:18 +0000]
powerpc: Fix irq_free_virt by adjusting bounds before loop

Instead of looping over each irq and checking against the irq array
bounds, adjust the bounds before looping.

The old code will not free any irq if the irq + count is above
irq_virq_count because the test in the loop is testing irq + count
instead of irq + i.

This code checks the limits to avoid unsigned integer overflows.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agopowerpc/irq: Protect irq_radix_revmap_lookup against irq_free_virt
Milton Miller [Tue, 24 May 2011 20:34:18 +0000]
powerpc/irq: Protect irq_radix_revmap_lookup against irq_free_virt

The radix-tree code uses call_rcu when freeing internal elements.
We must protect against the elements being freed while we traverse
the tree, even if the returned pointer will still be valid.

While preparing a patch to expand the context in which
irq_radix_revmap_lookup will be called, I realized that the
radix tree was not locked.

When asked

    For a normal call_rcu usage, is it allowed to read the structure in
    irq_enter / irq_exit, without additional rcu_read_lock?  Could an
    element freed with call_rcu advance with the cpu still between
    irq_enter/irq_exit (and irq_disabled())?

Paul McKenney replied:

    Absolutely illegal to do so. OK for call_rcu_sched(), but a
    flaming bug for call_rcu().

    And thank you very much for finding this!!!

Further analysis:

In the current CONFIG_TREE_RCU implementation. CONFIG_TREE_PREEMPT_RCU
(and CONFIG_TINY_PREEMPT_RCU) uses explicit counters.

These counters are reflected from per-CPU to global in the
scheduling-clock-interrupt handler, so disabling irq does prevent the
grace period from completing. But there are real-time implementations
(such as the one use by the Concurrent guys) where disabling irq
does -not- prevent the grace period from completing.

While an alternative fix would be to switch radix-tree to rcu_sched, I
don't want to audit the other users of radix trees (nor put alternative
freeing in the library).  The normal overhead for rcu_read_lock and
unlock are a local counter increment and decrement.

This does not show up in the rcu lockdep because in 2.6.34 commit
2676a58c98 (radix-tree: Disable RCU lockdep checking in radix tree)
deemed it too hard to pass the condition of the protecting lock
to the library.

Signed-off-by: Milton Miller <miltonm@bga.com>
Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agopowerpc/irq: Check desc in handle_one_irq and expand generic_handle_irq
Milton Miller [Tue, 24 May 2011 20:34:18 +0000]
powerpc/irq: Check desc in handle_one_irq and expand generic_handle_irq

Look up the descriptor and check that it is found in handle_one_irq
before checking if we are on the irq stack, and call the handler
directly using the descriptor if we are on the stack.

We need check irq_to_desc finds the descriptor to avoid a NULL
pointer dereference.  It could have failed because the number from
ppc_md.get_irq was above NR_IRQS, or various exceptional conditions
with sparse irqs (eg race conditions while freeing an irq if its was
not shutdown in the controller).

fe12bc2c99 (genirq: Uninline and sanity check generic_handle_irq())
moved generic_handle_irq out of line to allow its use by interrupt
controllers in modules.  However, handle_one_irq is core arch code.
It already knows the details of struct irq_desc and handling irqs in
the nested irq case.  This will avoid the extra stack frame to return
the value we don't check.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agopowerpc/irq: Always free duplicate IRQ_LEGACY hosts
Milton Miller [Tue, 24 May 2011 20:34:17 +0000]
powerpc/irq: Always free duplicate IRQ_LEGACY hosts

Since kmem caches are allocated before init_IRQ as noted in 3af259d155
(powerpc: Radix trees are available before init_IRQ), we now call
kmalloc in all cases and can can always call kfree if we are asked
to allocate a duplicate or conflicting IRQ_HOST_MAP_LEGACY host.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agopowerpc/irq: Remove stale and misleading comment
Milton Miller [Tue, 24 May 2011 20:34:18 +0000]
powerpc/irq: Remove stale and misleading comment

The comment claims we will call host->ops->map() to update the flags if
we find a previously established mapping, but we never did.  We used
to call remap, but that call was removed in da05198002 (powerpc: Remove
irq_host_ops->remap hook).

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agopowerpc/cell: Rename ipi functions to match current abstractions
Milton Miller [Tue, 24 May 2011 20:34:18 +0000]
powerpc/cell: Rename ipi functions to match current abstractions

Rename functions and arguments to reflect current usage.  iic_cause_ipi
becomes iic_message_pass and iic_ipi_to_irq becomes iic_msg_to_irq,
and iic_request_ipi now takes a message (msg) instead of an ipi number.
Also mesg is renamed to msg.

Commit f1072939b6 (powerpc: Remove checks for MSG_ALL and
MSG_ALL_BUT_SELF) connected the smp_message_pass hook for cell to the
underlying iic_cause_IPI, a platform unique name.  Later 23d72bfd8f
(powerpc: Consolidate ipi message mux and demux) added a cause_ipi
hook to the smp_ops, also used in message passing, but for controllers
that can not send 4 unique messages and require multiplexing.  It is
even more confusing that the both take two arguments, but one is the
small message ordinal and the other is an opaque long data associated
with the cpu.

Since cell iic maps messages one to one to ipi irqs, rename the
function and argument to translate from ipi to message.  Also make it
clear that iic_request_ipi takes a message number as the argument
for which ipi to create and request.

No functionional change, just renames to avoid future confusion.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agopowerpc/cell: Use common smp ipi actions
Milton Miller [Tue, 24 May 2011 20:34:18 +0000]
powerpc/cell: Use common smp ipi actions

The cell iic interrupt controller has enough software caused interrupts
to use a unique interrupt for each of the 4 messages powerpc uses.
This means each interrupt gets its own irq action/data combination.

Use the seperate, optimized, arch common ipi action functions
registered via the helper smp_request_message_ipi instead passing the
message as action data to a single action that then demultipexes to
the required acton via a switch statement.

smp_request_message_ipi will register the action as IRQF_PER_CPU
and IRQF_DISABLED, and WARN if the allocation fails for some reason,
so no need to print on that failure.  It will return positive if
the message will not be used by the kernel, in which case we can
free the virq.

In addition to elimiating inefficient code, this also corrects the
error that a kernel built with kexec but without a debugger would
not register the ipi for kdump to notify the other cpus of a crash.

This also restores the debugger action to be static to kernel/smp.c.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agoRemove unused MSG_ flags in linux/smp.h
Milton Miller [Tue, 10 May 2011 19:29:13 +0000]
Remove unused MSG_ flags in linux/smp.h

Now that powerpc has removed its use of MSG_ALL_BUT_SELF and MSG_ALL
all these MSG_ flags are unused.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agopowerpc/pseries: Update MAX_HCALL_OPCODE to reflect page coalescing
Brian King [Tue, 24 May 2011 03:40:54 +0000]
powerpc/pseries: Update MAX_HCALL_OPCODE to reflect page coalescing

When page coalescing support was added recently, the MAX_HCALL_OPCODE
define was not updated for the newly added H_GET_MPP_X hcall.

Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agopowerpc/oprofile: Handle events that raise an exception without overflowing
Eric B Munson [Mon, 23 May 2011 04:22:40 +0000]
powerpc/oprofile: Handle events that raise an exception without overflowing

Commit 0837e3242c73566fc1c0196b4ec61779c25ffc93 fixes a situation on POWER7
where events can roll back if a specualtive event doesn't actually complete.
This can raise a performance monitor exception.  We need to catch this to ensure
that we reset the PMC.  In all cases the PMC will be less than 256 cycles from
overflow.

This patch lifts Anton's fix for the problem in perf and applies it to oprofile
as well.

Signed-off-by: Eric B Munson <emunson@mgebm.net>
Cc: <stable@kernel.org> # as far back as it applies cleanly
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agopowerpc/ftrace: Implement raw syscall tracepoints on PowerPC
Ian Munsie [Wed, 2 Feb 2011 17:27:24 +0000]
powerpc/ftrace: Implement raw syscall tracepoints on PowerPC

This patch implements the raw syscall tracepoints on PowerPC and exports
them for ftrace syscalls to use.

To minimise reworking existing code, I slightly re-ordered the thread
info flags such that the new TIF_SYSCALL_TRACEPOINT bit would still fit
within the 16 bits of the andi. instruction's UI field. The instructions
in question are in /arch/powerpc/kernel/entry_{32,64}.S to and the
_TIF_SYSCALL_T_OR_A with the thread flags to see if system call tracing
is enabled.

In the case of 64bit PowerPC, arch_syscall_addr and
arch_syscall_match_sym_name are overridden to allow ftrace syscalls to
work given the unusual system call table structure and symbol names that
start with a period.

Signed-off-by: Ian Munsie <imunsie@au1.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

8 years agotmpfs: fix XATTR N overriding POSIX_ACL Y
Eric Paris [Thu, 26 May 2011 02:49:18 +0000]
tmpfs: fix XATTR N overriding POSIX_ACL Y

Choosing TMPFS_XATTR default N was switching off TMPFS_POSIX_ACL,
even if it had been Y in oldconfig; and Linus reports that PulseAudio
goes subtly wrong unless it can use ACLs on /dev/shm.

Make TMPFS_POSIX_ACL select TMPFS_XATTR (and depend upon TMPFS),
and move the TMPFS_POSIX_ACL entry before the TMPFS_XATTR entry,
to avoid asking unnecessary questions then ignoring their answers.

Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

8 years agovideo: mb862xx: udelay need linux/delay.h
Stephen Rothwell [Thu, 26 May 2011 01:09:10 +0000]
video: mb862xx: udelay need linux/delay.h

Fix this:

  drivers/video/mb862xx/mb862xx-i2c.c: In function 'mb862xx_i2c_wait_event':
  drivers/video/mb862xx/mb862xx-i2c.c:25: error: implicit declaration of function 'udelay'

caused by commit f8a6b1f44833 ("video: mb862xx: add support for
controller's I2C bus adapter").

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/linux-2.6-nsfd
Linus Torvalds [Thu, 26 May 2011 01:10:16 +0000]
Merge git://git./linux/kernel/git/ebiederm/linux-2.6-nsfd

* git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/linux-2.6-nsfd:
  net: fix get_net_ns_by_fd for !CONFIG_NET_NS
  ns proc: Return -ENOENT for a nonexistent /proc/self/ns/ entry.
  ns: Declare sys_setns in syscalls.h
  net: Allow setting the network namespace by fd
  ns proc: Add support for the ipc namespace
  ns proc: Add support for the uts namespace
  ns proc: Add support for the network namespace.
  ns: Introduce the setns syscall
  ns: proc files for namespace naming policy.

8 years agoslub: remove no-longer used 'unlock_out' label
Linus Torvalds [Thu, 26 May 2011 01:06:54 +0000]
slub: remove no-longer used 'unlock_out' label

Commit a71ae47a2cbf ("slub: Fix double bit unlock in debug mode")
removed the only goto to this label, resulting in

  mm/slub.c: In function '__slab_alloc':
  mm/slub.c:1834: warning: label 'unlock_out' defined but not used

fixed trivially by the removal of the label itself too.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Thu, 26 May 2011 00:00:17 +0000]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (89 commits)
  bonding: documentation and code cleanup for resend_igmp
  bonding: prevent deadlock on slave store with alb mode (v3)
  net: hold rtnl again in dump callbacks
  Add Fujitsu 1000base-SX PCI ID to tg3
  bnx2x: protect sequence increment with mutex
  sch_sfq: fix peek() implementation
  isdn: netjet - blacklist Digium TDM400P
  via-velocity: don't annotate MAC registers as packed
  xen: netfront: hold RTNL when updating features.
  sctp: fix memory leak of the ASCONF queue when free asoc
  net: make dev_disable_lro use physical device if passed a vlan dev (v2)
  net: move is_vlan_dev into public header file (v2)
  bug.h: Fix build with CONFIG_PRINTK disabled.
  wireless: fix fatal kernel-doc error + warning in mac80211.h
  wireless: fix cfg80211.h new kernel-doc warnings
  iwlagn: dbg_fixed_rate only used when CONFIG_MAC80211_DEBUGFS enabled
  dst: catch uninitialized metrics
  be2net: hash key for rss-config cmd not set
  bridge: initialize fake_rtable metrics
  net: fix __dst_destroy_metrics_generic()
  ...

Fix up trivial conflicts in drivers/staging/brcm80211/brcmfmac/wl_cfg80211.c

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cjb/mmc
Linus Torvalds [Wed, 25 May 2011 23:55:55 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/cjb/mmc

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cjb/mmc: (75 commits)
  mmc: core: eMMC bus width may not work on all platforms
  mmc: sdhci: Auto-CMD23 fixes.
  mmc: sdhci: Auto-CMD23 support.
  mmc: core: Block CMD23 support for UHS104/SDXC cards.
  mmc: sdhci: Implement MMC_CAP_CMD23 for SDHCI.
  mmc: core: Use CMD23 for multiblock transfers when we can.
  mmc: quirks: Add/remove quirks conditional support.
  mmc: Add new VUB300 USB-to-SD/SDIO/MMC driver
  mmc: sdhci-pxa: Add quirks for DMA/ADMA to match h/w
  mmc: core: duplicated trial with same freq in mmc_rescan_try_freq()
  mmc: core: add support for eMMC Dual Data Rate
  mmc: core: eMMC signal voltage does not use CMD11
  mmc: sdhci-pxa: add platform code for UHS signaling
  mmc: sdhci: add hooks for setting UHS in platform specific code
  mmc: core: clear MMC_PM_KEEP_POWER flag on resume
  mmc: dw_mmc: fixed wrong regulator_enable in suspend/resume
  mmc: sdhi: allow powering down controller with no card inserted
  mmc: tmio: runtime suspend the controller, where possible
  mmc: sdhi: support up to 3 interrupt sources
  mmc: sdhi: print physical base address and clock rate
  ...

8 years agoMerge branch 'kconfig-for-40' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek...
Linus Torvalds [Wed, 25 May 2011 23:54:01 +0000]
Merge branch 'kconfig-for-40' of git://git./linux/kernel/git/mmarek/kbuild-2.6

* 'kconfig-for-40' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild-2.6:
  xconfig: merge code path to conf_write()
  kconfig: do not record timestamp in .config
  gconfig: Hide unused left treeview when start up the interface
  gconfig: enable rules hint for main treeviews
  MAINTAINERS: Update KCONFIG entry
  kconfig-language: add to hints
  kconfig: Document the new "visible if" syntax
  kconfig: quiet commands when V=0
  kconfig: change update-po-config to reflect new layout of arch/um
  kconfig: make update-po-config work in KBUILD_OUTPUT
  kconfig: rearrange clean-files
  kconfig: change gconf to modify hostprogs-y like nconf and mconf
  kconfig: change qconf to modify hostprogs-y like nconf and mconf
  kconfig: only build kxgettext when needed
  nconfig: Silence unused return values from wattrset
  kconfig: Do not record timestamp in auto.conf and autoconf.h
  kconfig: get rid of unused flags
  kconfig: allow multiple inclusion of the same file
  kconfig: Avoid buffer underrun in choice input

8 years agoMerge branch 'for-2.6.40' of git://git.kernel.org/pub/scm/linux/kernel/git/oleg/misc
Linus Torvalds [Wed, 25 May 2011 23:53:14 +0000]
Merge branch 'for-2.6.40' of git://git./linux/kernel/git/oleg/misc

* 'for-2.6.40' of git://git.kernel.org/pub/scm/linux/kernel/git/oleg/misc:
  signal: sys_pause() should check signal_pending()
  ptrace: ptrace_resume() shouldn't wake up !TASK_TRACED thread

8 years agoMerge branch 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelv...
Linus Torvalds [Wed, 25 May 2011 23:52:50 +0000]
Merge branch 'hwmon-for-linus' of git://git./linux/kernel/git/jdelvare/staging

* 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvare/staging:
  hwmon: New driver for the SMSC EMC6W201
  hwmon: (abituguru) Depend on DMI
  hwmon: (it87) Use request_muxed_region
  hwmon: (sch5627) Trigger Vbat measurements
  hwmon: (sch5627) Add sch5627_send_cmd function
  i8k: Integrate with the hwmon subsystem
  hwmon: (max6650) Properly support the MAX6650
  hwmon: (max6650) Drop device detection
  Move ACPI power meter driver to hwmon
  hwmon: (f71882fg) Add support for F71808A
  hwmon: (f71882fg) Split has_beep in fan_has_beep and temp_has_beep
  hwmon: (asc7621) Drop duplicate dependency
  hwmon: (jc42) Change detection class
  hwmon: Add driver for AMD family 15h processor power information
  hwmon: (k10temp) Add support for Fam15h (Bulldozer)
  hwmon: Use helper functions to set and get driver data
  i8k: Avoid lahf in 64-bit code

8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile
Linus Torvalds [Wed, 25 May 2011 22:35:32 +0000]
Merge git://git./linux/kernel/git/cmetcalf/linux-tile

* git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile: (26 commits)
  arch/tile: prefer "tilepro" as the name of the 32-bit architecture
  compat: include aio_abi.h for aio_context_t
  arch/tile: cleanups for tilegx compat mode
  arch/tile: allocate PCI IRQs later in boot
  arch/tile: support signal "exception-trace" hook
  arch/tile: use better definitions of xchg() and cmpxchg()
  include/linux/compat.h: coding-style fixes
  tile: add an RTC driver for the Tilera hypervisor
  arch/tile: finish enabling support for TILE-Gx 64-bit chip
  compat: fixes to allow working with tile arch
  arch/tile: update defconfig file to something more useful
  tile: do_hardwall_trap: do not play with task->sighand
  tile: replace mm->cpu_vm_mask with mm_cpumask()
  tile,mn10300: add device parameter to dma_cache_sync()
  audit: support the "standard" <asm-generic/unistd.h>
  arch/tile: clarify flush_buffer()/finv_buffer() function names
  arch/tile: kernel-related cleanups from removing static page size
  arch/tile: various header improvements for building drivers
  arch/tile: disable GX prefetcher during cache flush
  arch/tile: tolerate disabling CONFIG_BLK_DEV_INITRD
  ...

8 years agoMerge branch 'for-torvalds' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw...
Linus Torvalds [Wed, 25 May 2011 22:35:03 +0000]
Merge branch 'for-torvalds' of git://git./linux/kernel/git/linusw/linux-stericsson

* 'for-torvalds' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-stericsson:
  mach-ux500: voltage domain regulators for DB8500
  cpufreq: make DB8500 cpufreq driver compile
  cpufreq: update DB8500 cpufreq driver
  mach-ux500: move CPUfreq driver to cpufreq subsystem
  mfd: add DB5500 PRCMU driver
  mfd: update DB8500 PRCMU driver
  mach-ux500: move the DB8500 PRCMU driver to MFD
  mach-ux500: make PRCMU base address dynamic
  mach-ux500: rename PRCMU driver per SoC
  mach-ux500: update ASIC version detection
  mach-ux500: update SoC and board IRQ handling
  mach-ux500: update the DB5500 register file
  mach-ux500: update the DB8500 register file

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier...
Linus Torvalds [Wed, 25 May 2011 22:34:14 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/vapier/blackfin

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier/blackfin: (37 commits)
  Blackfin: use new common PERCPU_INPUT define
  MAINTAINERS: Fix Analog Devices mailinglist address
  Blackfin: boards: update ASoC resources after machine driver overhaul
  Blackfin: work around anomaly 05000480
  Blackfin: fix addr type with bfin_write_{or,and} helpers
  Blackfin: convert /proc/sram to seq_file
  Blackfin: switch /proc/gpio to seq_file
  Blackfin: fix indentation with bfin_read() helper
  Blackfin: convert old cpumask API to new one
  Blackfin: don't touch task->cpus_allowed directly
  Blackfin: don't touch cpu_possible_map and cpu_present_map directly
  Blackfin: bf548-ezkit/bf561-ezkit: update nor flash layout
  Blackfin: initial perf_event support
  Blackfin: update anomaly lists to latest public info
  Blackfin: use on-chip reset func with newer parts
  Blackfin: bf533-stamp/bf537-stamp: drop ad1980 from defconfigs
  Blackfin: optimize MMR reads during startup a bit
  Blackfin: bf537: demux port H mask A and emac rx ints
  Blackfin: bf537: fix excessive gpio int demuxing
  Blackfin: bf54x: drop unused pm gpio handling
  ...

8 years agoMerge branch 'rmobile-latest' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal...
Linus Torvalds [Wed, 25 May 2011 22:33:25 +0000]
Merge branch 'rmobile-latest' of git://git./linux/kernel/git/lethal/sh-2.6

* 'rmobile-latest' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6: (34 commits)
  ARM: mach-shmobile: mackerel: add renesas_usbhs support for USB1
  ARM: mach-shmobile: Correct the G4EVM SDHI0 I/O range.
  ARM: arch-shmobile: sh7372: add renesas_usbhs irq support
  ARM: mach-shmobile: sh73a0: mark DMA slave ID 0 as invalid
  ARM: mach-shmobile: mark DMA slave ID 0 as invalid
  ARM: mach-shmobile: Enable DMAEngine for SDHI on AG5EVM
  ARM: mach-shmobile: Enable DMAEngine for MMCIF on AG5EVM
  ARM: mach-shmobile: sh73a0 DMA Engine support for SY-DMAC
  dmaengine: shdma: Update SH_DMAC_MAX_CHANNELS to 20
  dmaengine: shdma: Fix SH_DMAC_MAX_CHANNELS handling
  dmaengine: shdma: Make second memory window optional
  ARM: mach-shmobile: Tidy up after SH7372 pm changes.
  ARM: mach-shmobile: sh7372 Core Standby CPUIdle
  ARM: mach-shmobile: CPUIdle support
  ARM: mach-shmobile: sh7372 Core Standby Suspend-to-RAM
  ARM: mach-shmobile: Suspend-to-RAM support
  mailmap: Add entry for Damian Hobson-Garcia.
  ARM: switch mackerel to dynamically manage the platform camera
  ARM: mach-shmobile: Add SDHI support for AG5EVM and sh73a0
  ARM: arch-shmobile: Use multiple irq vectors for SDHI
  ...

8 years agohrtimers: Fix typo causing erratic timers
Thomas Gleixner [Wed, 25 May 2011 21:08:17 +0000]
hrtimers: Fix typo causing erratic timers

commit 9ec2690758a5 ("timerfd: Manage cancelable timers in timerfd")
introduced a CONFIG_HIGHRES_TIMERS (should be CONFIG_HIGH_RES_TIMERS)
typo, which caused applications depending on CLOCK_REALTIME timers to
become sluggy due to the fact that the time base of the realtime
timers was not updated when the wall clock time was set.

This causes anything from 100% CPU use for some applications to odd
delays and hickups.

Reported-bisected-and-tested-by: Anca Emanuel <anca.emanuel@gmail.com>
Tested-by: Linus Torvalds <torvalds@linux-foundation.org>
Fatfingered-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

8 years agobonding: documentation and code cleanup for resend_igmp
Flavio Leitner [Wed, 25 May 2011 08:38:58 +0000]
bonding: documentation and code cleanup for resend_igmp

Improves the documentation about how IGMP resend parameter
works, fix two missing checks and coding style issues.

Signed-off-by: Flavio Leitner <fbl@redhat.com>
Acked-by: Rick Jones <rick.jones2@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agobonding: prevent deadlock on slave store with alb mode (v3)
Neil Horman [Wed, 25 May 2011 08:13:01 +0000]
bonding: prevent deadlock on slave store with alb mode (v3)

This soft lockup was recently reported:

[root@dell-per715-01 ~]# echo +bond5 > /sys/class/net/bonding_masters
[root@dell-per715-01 ~]# echo +eth1 > /sys/class/net/bond5/bonding/slaves
bonding: bond5: doing slave updates when interface is down.
bonding bond5: master_dev is not up in bond_enslave
[root@dell-per715-01 ~]# echo -eth1 > /sys/class/net/bond5/bonding/slaves
bonding: bond5: doing slave updates when interface is down.

BUG: soft lockup - CPU#12 stuck for 60s! [bash:6444]
CPU 12:
Modules linked in: bonding autofs4 hidp rfcomm l2cap bluetooth lockd sunrpc
be2d
Pid: 6444, comm: bash Not tainted 2.6.18-262.el5 #1
RIP: 0010:[<ffffffff80064bf0>]  [<ffffffff80064bf0>]
.text.lock.spinlock+0x26/00
RSP: 0018:ffff810113167da8  EFLAGS: 00000286
RAX: ffff810113167fd8 RBX: ffff810123a47800 RCX: 0000000000ff1025
RDX: 0000000000000000 RSI: ffff810123a47800 RDI: ffff81021b57f6f8
RBP: ffff81021b57f500 R08: 0000000000000000 R09: 000000000000000c
R10: 00000000ffffffff R11: ffff81011d41c000 R12: ffff81021b57f000
R13: 0000000000000000 R14: 0000000000000282 R15: 0000000000000282
FS:  00002b3b41ef3f50(0000) GS:ffff810123b27940(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 00002b3b456dd000 CR3: 000000031fc60000 CR4: 00000000000006e0

Call Trace:
 [<ffffffff80064af9>] _spin_lock_bh+0x9/0x14
 [<ffffffff886937d7>] :bonding:tlb_clear_slave+0x22/0xa1
 [<ffffffff8869423c>] :bonding:bond_alb_deinit_slave+0xba/0xf0
 [<ffffffff8868dda6>] :bonding:bond_release+0x1b4/0x450
 [<ffffffff8006457b>] __down_write_nested+0x12/0x92
 [<ffffffff88696ae4>] :bonding:bonding_store_slaves+0x25c/0x2f7
 [<ffffffff801106f7>] sysfs_write_file+0xb9/0xe8
 [<ffffffff80016b87>] vfs_write+0xce/0x174
 [<ffffffff80017450>] sys_write+0x45/0x6e
 [<ffffffff8005d28d>] tracesys+0xd5/0xe0

It occurs because we are able to change the slave configuarion of a bond while
the bond interface is down.  The bonding driver initializes some data structures
only after its ndo_open routine is called.  Among them is the initalization of
the alb tx and rx hash locks.  So if we add or remove a slave without first
opening the bond master device, we run the risk of trying to lock/unlock a
spinlock that has garbage for data in it, which results in our above softlock.

Note that sometimes this works, because in many cases an unlocked spinlock has
the raw_lock parameter initialized to zero (meaning that the kzalloc of the
net_device private data is equivalent to calling spin_lock_init), but thats not
true in all cases, and we aren't guaranteed that condition, so we need to pass
the relevant spinlocks through the spin_lock_init function.

Fix it by moving the spin_lock_init calls for the tx and rx hashtable locks to
the ndo_init path, so they are ready for use by the bond_store_slaves path.

Change notes:
v2) Based on conversation with Jay and Nicolas it seems that the ability to
enslave devices while the bond master is down should be safe to do.  As such
this is an outlier bug, and so instead we'll just initalize the errant spinlocks
in the init path rather than the open path, solving the problem.  We'll also
remove the warnings about the bond being down during enslave operations, since
it should be safe

v3) Fix spelling error

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Reported-by: jtluka@redhat.com
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
CC: nicolas.2p.debian@gmail.com
CC: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agonet: hold rtnl again in dump callbacks
Eric Dumazet [Wed, 25 May 2011 07:34:04 +0000]
net: hold rtnl again in dump callbacks

Commit e67f88dd12f6 (dont hold rtnl mutex during netlink dump callbacks)
missed fact that rtnl_fill_ifinfo() must be called with rtnl held.

Because of possible deadlocks between two mutexes (cb_mutex and rtnl),
its not easy to solve this problem, so revert this part of the patch.

It also forgot one rcu_read_unlock() in FIB dump_rules()

Add one ASSERT_RTNL() in rtnl_fill_ifinfo() to remind us the rule.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Patrick McHardy <kaber@trash.net>
CC: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agoAdd Fujitsu 1000base-SX PCI ID to tg3
Meelis Roos [Wed, 25 May 2011 05:43:47 +0000]
Add Fujitsu 1000base-SX PCI ID to tg3

This patch adds the PCI ID of Fujitsu 1000base-SX NIC to tg3 driver.
Tested to detect the card, MAC and serdes, not tested with link at the
moment since I have no fiber switch here. I did not add new constants to
the pci_ids.h header file since these constants are used only here.

Signed-off-by: Meelis Roos <mroos@linux.ee>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agobnx2x: protect sequence increment with mutex
Dmitry Kravkov [Wed, 25 May 2011 04:55:51 +0000]
bnx2x: protect sequence increment with mutex

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agosch_sfq: fix peek() implementation
Eric Dumazet [Wed, 25 May 2011 04:40:11 +0000]
sch_sfq: fix peek() implementation

Since commit eeaeb068f139 (sch_sfq: allow big packets and be fair),
sfq_peek() can return a different skb that would be normally dequeued by
sfq_dequeue() [ if current slot->allot is negative ]

Use generic qdisc_peek_dequeued() instead of custom implementation, to
get consistent result.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Jarek Poplawski <jarkao2@gmail.com>
CC: Patrick McHardy <kaber@trash.net>
CC: Jesper Dangaard Brouer <hawk@diku.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agoisdn: netjet - blacklist Digium TDM400P
Prarit Bhargava [Wed, 25 May 2011 02:12:23 +0000]
isdn: netjet - blacklist Digium TDM400P

[2nd try ... 1st attempt didn't make it to netdev mailing list]

A quick google search reveals that people with this card are blacklisting it
in the initramfs and in the module blacklist based on a statement that it
is unsupported. Since the older Digium is also unsupported I'm pretty
confident that this newer card is also not supported.

lspci -xxx -vv shows

04:07.0 Communication controller: Tiger Jet Network Inc. Tiger3XX Modem/ISDN interface
        Subsystem: Device b100:0003
P.

----8<----
The Asterisk Voice Card, DIGIUM TDM400P is unsupported by the netjet driver.
Blacklist it like the Digium X100P/X101P card.

Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agovia-velocity: don't annotate MAC registers as packed
Ulrich Hecht [Wed, 25 May 2011 01:07:22 +0000]
via-velocity: don't annotate MAC registers as packed

On ARM, memory accesses through packed pointers behave in unexpected
ways in GCC releases 4.3 and higher; see https://lkml.org/lkml/2011/2/2/163
for discussion.

In this particular case, 32-bit I/O registers are accessed bytewise,
causing incorrect setting of the DMA address registers which in turn
leads to an error interrupt storm that brings the system to a halt.

Since the mac_regs structure does not need any packing anyway, this patch
simply removes the attribute to fix the issue.

Signed-off-by: Ulrich Hecht <uli@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agoxen: netfront: hold RTNL when updating features.
Ian Campbell [Tue, 24 May 2011 21:56:02 +0000]
xen: netfront: hold RTNL when updating features.

Konrad reports:
[    0.930811] RTNL: assertion failed at /home/konrad/ssd/linux/net/core/dev.c (5258)
[    0.930821] Pid: 22, comm: xenwatch Not tainted 2.6.39-05193-gd762f43 #1
[    0.930825] Call Trace:
[    0.930834]  [<ffffffff8143bd0e>] __netdev_update_features+0xae/0xe0
[    0.930840]  [<ffffffff8143dd41>] netdev_update_features+0x11/0x30
[    0.930847]  [<ffffffffa0037105>] netback_changed+0x4e5/0x800 [xen_netfront]
[    0.930854]  [<ffffffff8132a838>] xenbus_otherend_changed+0xa8/0xb0
[    0.930860]  [<ffffffff8157ca99>] ? _raw_spin_unlock_irqrestore+0x19/0x20
[    0.930866]  [<ffffffff8132adfe>] backend_changed+0xe/0x10
[    0.930871]  [<ffffffff8132875a>] xenwatch_thread+0xba/0x180
[    0.930876]  [<ffffffff810a8ba0>] ? wake_up_bit+0x40/0x40
[    0.930881]  [<ffffffff813286a0>] ? split+0xf0/0xf0
[    0.930886]  [<ffffffff810a8646>] kthread+0x96/0xa0
[    0.930891]  [<ffffffff815855a4>] kernel_thread_helper+0x4/0x10
[    0.930896]  [<ffffffff815846b3>] ? int_ret_from_sys_call+0x7/0x1b
[    0.930901]  [<ffffffff8157cf61>] ? retint_restore_args+0x5/0x6
[    0.930906]  [<ffffffff815855a0>] ? gs_change+0x13/0x13

This update happens in xenbus watch callback context and hence does not already
hold the rtnl. Take the lock as necessary.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Tested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agosctp: fix memory leak of the ASCONF queue when free asoc
Wei Yongjun [Tue, 24 May 2011 21:48:02 +0000]
sctp: fix memory leak of the ASCONF queue when free asoc

If an ASCONF chunk is outstanding, then the following ASCONF
chunk will be queued for later transmission. But when we free
the asoc, we forget to free the ASCONF queue at the same time,
this will cause memory leak.

Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agonet: make dev_disable_lro use physical device if passed a vlan dev (v2)
Neil Horman [Tue, 24 May 2011 08:31:09 +0000]
net: make dev_disable_lro use physical device if passed a vlan dev (v2)

If the device passed into dev_disable_lro is a vlan, then repoint the dev
poniter so that we actually modify the underlying physical device.

Signed-of-by: Neil Horman <nhorman@tuxdriver.com>
CC: davem@davemloft.net
CC: bhutchings@solarflare.com

Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agonet: move is_vlan_dev into public header file (v2)
Neil Horman [Tue, 24 May 2011 08:31:08 +0000]
net: move is_vlan_dev into public header file (v2)

Migrate is_vlan_dev() to if_vlan.h so that core networkig can use it

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: davem@davemloft.net
CC: bhutchings@solarflare.com
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years agojbd2: fix a potential leak of a journal_head on an error path
Ding Dinghua [Wed, 25 May 2011 21:43:48 +0000]
jbd2: fix a potential leak of a journal_head on an error path

drop jh->b_jcount in error path

Signed-off-by: Ding Dinghua <dingdinghua@nrchpc.ac.cn>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>

8 years agoext4: teach ext4_ext_split to calculate extents efficiently
Yongqiang Yang [Wed, 25 May 2011 21:41:48 +0000]
ext4: teach ext4_ext_split to calculate extents efficiently

Make ext4_ext_split() get extents to be moved by calculating in a statement
instead of counting in a loop.

Signed-off-by: Yongqiang Yang <xiaoqiangnk@gmail.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>

8 years agoext4: Convert ext4 to new truncate calling convention
Jan Kara [Wed, 25 May 2011 21:39:48 +0000]
ext4: Convert ext4 to new truncate calling convention

Trivial conversion.  Fixup one error handling case calling vmtruncate()
and remove ->truncate callback. We also fix a bug that IS_IMMUTABLE and
IS_APPEND files could not be truncated during failed writes. In fact, the
test can be completely removed as upper layers do necessary permission
checks for truncate in do_sys_[f]truncate() and may_open() anyway.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>

8 years agommc: core: eMMC bus width may not work on all platforms
Philip Rakity [Wed, 25 May 2011 01:14:58 +0000]
mmc: core: eMMC bus width may not work on all platforms

CMD19 -- The offical way to validate bus widths from the JEDEC spec
does not work on all platforms.  Some platforms that use PCI/PCIe
to connect their SD controllers are known to fail.

If the quirk MMC_BUS_WIDTH_TEST is not defined we try to figure out
the bus width by reading the ext_csd at different bus widths and
compare this against the ext_csd read in 1 bit mode.  If no ext_csd
is available we default to 1 bit operations.

Code has been tested on mmp2 against 8 bit eMMC and Transcend 2GB
card that is known to not work in 4 bit mode.  The physical pins
on the card are not present to support 4 bit operation.

Signed-off-by: Philip Rakity <prakity@marvell.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

8 years agommc: sdhci: Auto-CMD23 fixes.
Andrei Warkentin [Wed, 25 May 2011 14:42:50 +0000]
mmc: sdhci: Auto-CMD23 fixes.

Fixes bugs in Auto-CMD23 feature enable decision. Auto-CMD23
should be enabled if host is >= v3, and SDMA is not in use.

USE_ADMA | USE_SDMA | Auto-CMD23
---------+----------+-----------
    0    |    0     |     1
---------+----------+-----------
    0    |    1     |     0
---------+----------+-----------
    1    |    0     |     1
---------+----------+-----------
    1    |    1     |     1

Signed-off-by: Andrei Warkentin <andreiw@motorola.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

8 years agommc: sdhci: Auto-CMD23 support.
Andrei Warkentin [Mon, 23 May 2011 20:06:39 +0000]
mmc: sdhci: Auto-CMD23 support.

Enables Auto-CMD23 support where available (SDHCI 3.0 controllers)

Signed-off-by: Andrei Warkentin <andreiw@motorola.com>
Tested-by: Arindam Nath <arindam.nath@amd.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

8 years agommc: core: Block CMD23 support for UHS104/SDXC cards.
Andrei Warkentin [Mon, 23 May 2011 20:06:38 +0000]
mmc: core: Block CMD23 support for UHS104/SDXC cards.

SD cards operating at UHS104 or better support SET_BLOCK_COUNT.

Signed-off-by: Andrei Warkentin <andreiw@motorola.com>
Reviewed-by: Arindam Nath <arindam.nath@amd.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

8 years agommc: sdhci: Implement MMC_CAP_CMD23 for SDHCI.
Andrei Warkentin [Mon, 23 May 2011 20:06:37 +0000]
mmc: sdhci: Implement MMC_CAP_CMD23 for SDHCI.

Implements support for multiblock transfers bounded
by SET_BLOCK_COUNT (CMD23).

Signed-off-by: Andrei Warkentin <andreiw@motorola.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

8 years agommc: core: Use CMD23 for multiblock transfers when we can.
Andrei Warkentin [Mon, 23 May 2011 20:06:36 +0000]
mmc: core: Use CMD23 for multiblock transfers when we can.

CMD23-prefixed instead of open-ended multiblock transfers
have a performance advantage on some MMC cards.

Signed-off-by: Andrei Warkentin <andreiw@motorola.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

8 years agoarch/tile: prefer "tilepro" as the name of the 32-bit architecture
Chris Metcalf [Wed, 25 May 2011 19:24:00 +0000]
arch/tile: prefer "tilepro" as the name of the 32-bit architecture

With this change, you can (and should) build with ARCH=tilepro for the
current 32-bit chips.  Building with ARCH=tile continues to work, but
we've renamed the defconfig file to tilepro_defconfig for consistency.

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>

8 years agoMerge branch 'misc' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild-2.6
Linus Torvalds [Wed, 25 May 2011 19:04:15 +0000]
Merge branch 'misc' of git://git./linux/kernel/git/mmarek/kbuild-2.6

* 'misc' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild-2.6:
  export_report: use warn() to issue WARNING, so they go to stderr
  export_report: sort SECTION 2 output
  export_report: do collectcfiles work in perl itself
  kbuild: make versioncheck work in KBUILD_OUTDIR
  kbuild: make includecheck work in KBUILD_OUTDIR
  kbuild: make headerdep work in KBUILD_OUTDIR
  kbuild: add targets to PHONY
  kbuild: don't warn about include/linux/version.h not including itself
  eradicate bashisms in scripts/patch-kernel

8 years agoMerge branch 'packaging' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek...
Linus Torvalds [Wed, 25 May 2011 19:03:47 +0000]
Merge branch 'packaging' of git://git./linux/kernel/git/mmarek/kbuild-2.6

* 'packaging' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild-2.6:
  kbuild: Create a kernel-headers RPM
  rpm-pkg: Fix when current directory is a symlink
  Replace '-' in kernel version with '_'

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph...
Linus Torvalds [Wed, 25 May 2011 18:46:31 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/sage/ceph-client

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client: (23 commits)
  ceph: fix cap flush race reentrancy
  libceph: subscribe to osdmap when cluster is full
  libceph: handle new osdmap down/state change encoding
  rbd: handle online resize of underlying rbd image
  ceph: avoid inode lookup on nfs fh reconnect
  ceph: use LOOKUPINO to make unconnected nfs fh more reliable
  rbd: use snprintf for disk->disk_name
  rbd: cleanup: make kfree match kmalloc
  rbd: warn on update_snaps failure on notify
  ceph: check return value for start_request in writepages
  ceph: remove useless check
  libceph: add missing breaks in addr_set_port
  libceph: fix TAG_WAIT case
  ceph: fix broken comparison in readdir loop
  libceph: fix osdmap timestamp assignment
  ceph: fix rare potential cap leak
  libceph: use snprintf for unknown addrs
  libceph: use snprintf for formatting object name
  ceph: use snprintf for dirstat content
  libceph: fix uninitialized value when no get_authorizer method is set
  ...

8 years agohwmon: New driver for the SMSC EMC6W201
Jean Delvare [Wed, 25 May 2011 18:43:33 +0000]
hwmon: New driver for the SMSC EMC6W201

This is a new driver for the SMSC EMC6W201 hardware monitoring device.
The device is functionally close to the EMC6D100 series, but is
register-incompatible.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Tested-by: Harry G McGavran Jr <w5pny@arrl.net>
Tested-by: Jeff Rickman <jrickman@myamigos.us>
Acked-by: Guenter Roeck <guenter.roeck@ericsson.com>