GitHub/exynos8895/android_kernel_samsung_universal8895.git
12 years agoCIFS: Move informational tcon calls to ops struct
Pavel Shilovsky [Sun, 27 May 2012 16:48:35 +0000 (20:48 +0400)]
CIFS: Move informational tcon calls to ops struct

and rename variables in cifs_mount.

Signed-off-by: Pavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Move getting dfs referalls to ops struct
Pavel Shilovsky [Sun, 27 May 2012 16:21:53 +0000 (20:21 +0400)]
CIFS: Move getting dfs referalls to ops struct

Signed-off-by: Pavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Process reconnects for SMB2 shares
Pavel Shilovsky [Tue, 27 Dec 2011 12:23:34 +0000 (16:23 +0400)]
CIFS: Process reconnects for SMB2 shares

Signed-off-by: Pavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Add tree connect/disconnect capability for SMB2
Pavel Shilovsky [Tue, 27 Dec 2011 12:04:00 +0000 (16:04 +0400)]
CIFS: Add tree connect/disconnect capability for SMB2

Signed-off-by: Pavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Add session setup/logoff capability for SMB2
Pavel Shilovsky [Tue, 27 Dec 2011 12:22:00 +0000 (16:22 +0400)]
CIFS: Add session setup/logoff capability for SMB2

Signed-off-by: Pavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Add capability to send SMB2 negotiate message
Pavel Shilovsky [Tue, 27 Dec 2011 12:12:43 +0000 (16:12 +0400)]
CIFS: Add capability to send SMB2 negotiate message

and add negotiate request type to let set_credits know that
we are only on negotiate stage and no need to make a decision
about disabling echos and oplocks.

Signed-off-by: Pavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Respect SMB2 header/max header size
Pavel Shilovsky [Thu, 12 Jan 2012 18:40:50 +0000 (22:40 +0400)]
CIFS: Respect SMB2 header/max header size

Use SMB2 header size values for allocation and memset because they
are bigger and suitable for both CIFS and SMB2.

Signed-off-by: Pavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Make demultiplex_thread work with SMB2 code
Pavel Shilovsky [Wed, 8 Jun 2011 11:51:07 +0000 (15:51 +0400)]
CIFS: Make demultiplex_thread work with SMB2 code

Now we can process SMB2 messages: check message, get message id
and wakeup awaiting routines.

Signed-off-by: Pavel Shilovsky <piastryyy@gmail.com>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Fix a wrong pointer in atomic_open
Pavel Shilovsky [Mon, 9 Jul 2012 12:09:23 +0000 (16:09 +0400)]
CIFS: Fix a wrong pointer in atomic_open

Commit 30d904947459cca2beb69e0110716f5248b31f2a caused a regression
in cifs open codepath.

Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Pavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Add SMB2 credits support
Pavel Shilovsky [Wed, 23 May 2012 12:18:00 +0000 (16:18 +0400)]
CIFS: Add SMB2 credits support

For SMB2 protocol we can add more than one credit for one received
request: it depends on CreditRequest field in SMB2 response header.
Also we divide all requests by type: echoes, oplocks and others.
Each type uses its own slot pull.

Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Pavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Make transport routines work with SMB2
Pavel Shilovsky [Mon, 26 Dec 2011 18:53:34 +0000 (22:53 +0400)]
CIFS: Make transport routines work with SMB2

Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Pavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Map SMB2 status codes to POSIX errors
Steve French [Tue, 15 Mar 2011 02:08:48 +0000 (02:08 +0000)]
CIFS: Map SMB2 status codes to POSIX errors

Add mapping table for 32 bit SMB2 status codes to linux errors.
Note that SMB2 does not use DOS/OS2 errors (ever) so mapping to
DOS/OS2 errors as a common network subset (as we do for cifs)
doesn't help. And note that the set of status codes is much more
complete here.

Signed-off-by: Steve French <sfrench@us.ibm.com>
Signed-off-by: Pavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Add SMB2 status codes
Pavel Shilovsky [Sun, 25 Dec 2011 09:27:35 +0000 (13:27 +0400)]
CIFS: Add SMB2 status codes

Reviewed-by: Jeff Layton <jlayton@samba.org>
Signed-off-by: Pavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Rename 7 error codes to NT_ style
Pavel Shilovsky [Wed, 20 Jun 2012 13:50:01 +0000 (17:50 +0400)]
CIFS: Rename 7 error codes to NT_ style

and consider such codes as CIFS errors.

Reviewed-by: Jeff Layton <jlayton@samba.org>
Signed-off-by: Pavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Rename Get/FreeXid and make them work with unsigned int
Pavel Shilovsky [Wed, 20 Jun 2012 07:21:16 +0000 (11:21 +0400)]
CIFS: Rename Get/FreeXid and make them work with unsigned int

Acked-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Pavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Move protocol specific tcon/tdis code to ops struct
Pavel Shilovsky [Fri, 25 May 2012 07:11:39 +0000 (11:11 +0400)]
CIFS: Move protocol specific tcon/tdis code to ops struct

and rename variables around the code changes.

Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Pavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Move protocol specific session setup/logoff code to ops struct
Pavel Shilovsky [Fri, 25 May 2012 06:54:49 +0000 (10:54 +0400)]
CIFS: Move protocol specific session setup/logoff code to ops struct

Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Pavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Move protocol specific negotiate code to ops struct
Pavel Shilovsky [Fri, 25 May 2012 06:43:58 +0000 (10:43 +0400)]
CIFS: Move protocol specific negotiate code to ops struct

Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Pavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Extend credit mechanism to process request type
Pavel Shilovsky [Wed, 23 May 2012 12:14:34 +0000 (16:14 +0400)]
CIFS: Extend credit mechanism to process request type

Split all requests to echos, oplocks and others - each group uses
its own credit slot. This is indicated by new flags

CIFS_ECHO_OP and CIFS_OBREAK_OP

that are not used now for CIFS. This change is required to support
SMB2 protocol because of different processing of these commands.

Acked-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Pavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agoCIFS: Move trans2 processing to ops struct
Pavel Shilovsky [Wed, 23 May 2012 10:31:03 +0000 (14:31 +0400)]
CIFS: Move trans2 processing to ops struct

Reviewed-by: Jeff Layton <jlayton@samba.org>
Signed-off-by: Pavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agocifs: reinstate sec=ntlmv2 mount option
Jeff Layton [Tue, 24 Jul 2012 00:34:17 +0000 (20:34 -0400)]
cifs: reinstate sec=ntlmv2 mount option

sec=ntlmv2 as a mount option got dropped in the mount option overhaul.

Cc: Sachin Prabhu <sprabhu@redhat.com>
Cc: <stable@vger.kernel.org> # 3.4+
Reported-by: Günter Kukkukk <linux@kukkukk.com>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agocifs: rename cifs_sign_smb2 to cifs_sign_smbv
Jeff Layton [Mon, 23 Jul 2012 17:28:38 +0000 (13:28 -0400)]
cifs: rename cifs_sign_smb2 to cifs_sign_smbv

"smb2" makes me think of the SMB2.x protocol, which isn't at all what
this function is for...

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agocifs: remove bogus reset of smb_buf_length in smb_send routines
Jeff Layton [Mon, 23 Jul 2012 17:28:37 +0000 (13:28 -0400)]
cifs: remove bogus reset of smb_buf_length in smb_send routines

There's a comment here about how we don't want to modify this length,
but nothing in this function actually does.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agocifs: move file_lock off stack in cifs_push_posix_locks
Jeff Layton [Mon, 23 Jul 2012 17:28:37 +0000 (13:28 -0400)]
cifs: move file_lock off stack in cifs_push_posix_locks

struct file_lock is pretty large, so we really don't want that on the
stack in a potentially long call chain. Reorganize the arguments to
CIFSSMBPosixLock to eliminate the need for that.

Eliminate the get_flag and simply use a non-NULL pLockInfo to indicate
that this is a "get" operation. In order to do that, need to add a new
loff_t argument for the start_offset.

Reported-by: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agocifs: remove extraneous newlines from cERROR and cFYI calls
Jeff Layton [Mon, 23 Jul 2012 17:14:28 +0000 (13:14 -0400)]
cifs: remove extraneous newlines from cERROR and cFYI calls

Those macros add a newline on their own, so there's not any need to
embed one in the message itself.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agocifs: after upcalling for krb5 creds, invalidate key rather than revoking it
Jeff Layton [Mon, 23 Jul 2012 17:14:28 +0000 (13:14 -0400)]
cifs: after upcalling for krb5 creds, invalidate key rather than revoking it

Calling key_revoke here isn't ideal as further requests for the key will
end up returning -EKEYREVOKED until it gets purged from the cache. What we
really intend here is to force a new upcall on the next request_key.

Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
12 years agolocks: fix checking of fcntl_setlease argument
J. Bruce Fields [Mon, 23 Jul 2012 19:17:17 +0000 (15:17 -0400)]
locks: fix checking of fcntl_setlease argument

The only checks of the long argument passed to fcntl(fd,F_SETLEASE,.)
are done after converting the long to an int.  Thus some illegal values
may be let through and cause problems in later code.

[ They actually *don't* cause problems in mainline, as of Dave Jones's
  commit 8d657eb3b438 "Remove easily user-triggerable BUG from
  generic_setlease", but we should fix this anyway.  And this patch will
  be necessary to fix real bugs on earlier kernels. ]

Cc: stable@vger.kernel.org
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoMerge branch 'for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Mon, 23 Jul 2012 19:27:27 +0000 (12:27 -0700)]
Merge branch 'for-linus-2' of git://git./linux/kernel/git/viro/vfs

Pull the big VFS changes from Al Viro:
 "This one is *big* and changes quite a few things around VFS.  What's in there:

   - the first of two really major architecture changes - death to open
     intents.

     The former is finally there; it was very long in making, but with
     Miklos getting through really hard and messy final push in
     fs/namei.c, we finally have it.  Unlike his variant, this one
     doesn't introduce struct opendata; what we have instead is
     ->atomic_open() taking preallocated struct file * and passing
     everything via its fields.

     Instead of returning struct file *, it returns -E...  on error, 0
     on success and 1 in "deal with it yourself" case (e.g.  symlink
     found on server, etc.).

     See comments before fs/namei.c:atomic_open().  That made a lot of
     goodies finally possible and quite a few are in that pile:
     ->lookup(), ->d_revalidate() and ->create() do not get struct
     nameidata * anymore; ->lookup() and ->d_revalidate() get lookup
     flags instead, ->create() gets "do we want it exclusive" flag.

     With the introduction of new helper (kern_path_locked()) we are rid
     of all struct nameidata instances outside of fs/namei.c; it's still
     visible in namei.h, but not for long.  Come the next cycle,
     declaration will move either to fs/internal.h or to fs/namei.c
     itself.  [me, miklos, hch]

   - The second major change: behaviour of final fput().  Now we have
     __fput() done without any locks held by caller *and* not from deep
     in call stack.

     That obviously lifts a lot of constraints on the locking in there.
     Moreover, it's legal now to call fput() from atomic contexts (which
     has immediately simplified life for aio.c).  We also don't need
     anti-recursion logics in __scm_destroy() anymore.

     There is a price, though - the damn thing has become partially
     asynchronous.  For fput() from normal process we are guaranteed
     that pending __fput() will be done before the caller returns to
     userland, exits or gets stopped for ptrace.

     For kernel threads and atomic contexts it's done via
     schedule_work(), so theoretically we might need a way to make sure
     it's finished; so far only one such place had been found, but there
     might be more.

     There's flush_delayed_fput() (do all pending __fput()) and there's
     __fput_sync() (fput() analog doing __fput() immediately).  I hope
     we won't need them often; see warnings in fs/file_table.c for
     details.  [me, based on task_work series from Oleg merged last
     cycle]

   - sync series from Jan

   - large part of "death to sync_supers()" work from Artem; the only
     bits missing here are exofs and ext4 ones.  As far as I understand,
     those are going via the exofs and ext4 trees resp.; once they are
     in, we can put ->write_super() to the rest, along with the thread
     calling it.

   - preparatory bits from unionmount series (from dhowells).

   - assorted cleanups and fixes all over the place, as usual.

  This is not the last pile for this cycle; there's at least jlayton's
  ESTALE work and fsfreeze series (the latter - in dire need of fixes,
  so I'm not sure it'll make the cut this cycle).  I'll probably throw
  symlink/hardlink restrictions stuff from Kees into the next pile, too.
  Plus there's a lot of misc patches I hadn't thrown into that one -
  it's large enough as it is..."

* 'for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (127 commits)
  ext4: switch EXT4_IOC_RESIZE_FS to mnt_want_write_file()
  btrfs: switch btrfs_ioctl_balance() to mnt_want_write_file()
  switch dentry_open() to struct path, make it grab references itself
  spufs: shift dget/mntget towards dentry_open()
  zoran: don't bother with struct file * in zoran_map
  ecryptfs: don't reinvent the wheels, please - use struct completion
  don't expose I_NEW inodes via dentry->d_inode
  tidy up namei.c a bit
  unobfuscate follow_up() a bit
  ext3: pass custom EOF to generic_file_llseek_size()
  ext4: use core vfs llseek code for dir seeks
  vfs: allow custom EOF in generic_file_llseek code
  vfs: Avoid unnecessary WB_SYNC_NONE writeback during sys_sync and reorder sync passes
  vfs: Remove unnecessary flushing of block devices
  vfs: Make sys_sync writeout also block device inodes
  vfs: Create function for iterating over block devices
  vfs: Reorder operations during sys_sync
  quota: Move quota syncing to ->sync_fs method
  quota: Split dquot_quota_sync() to writeback and cache flushing part
  vfs: Move noop_backing_dev_info check from sync into writeback
  ...

12 years agoMerge tag 'mmc-merge-for-3.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 22 Jul 2012 23:36:08 +0000 (16:36 -0700)]
Merge tag 'mmc-merge-for-3.6-rc1' of git://git./linux/kernel/git/cjb/mmc

Pull MMC updates from Chris Ball:
 "MMC highlights for 3.6:

  Core:
   - Rename cd-gpio to slot-gpio and extend it to support more slot GPIO
     functions, such as write-protect.
   - Add a function to get regulators (Vdd and Vccq) for a host.

  Drivers:
   - sdhci-pxav2, sdhci-pxav3: Add device tree support.
   - sdhi: Add device tree support.
   - sh_mmcif: Add support for regulators, device tree, slot-gpio.
   - tmio: Add regulator support, use slot-gpio."

* tag 'mmc-merge-for-3.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/cjb/mmc: (62 commits)
  mmc: sdhci-dove: Prepare for common clock framework
  mmc: sdhci-dove: Add SDHCI_QUIRK_NO_HISPD_BIT
  mmc: omap_hsmmc: ensure probe returns error upon resource failure
  mmc: mxs-mmc: Add wp-inverted property
  mmc: esdhc: Fix DMA_MASK to not break mx25 DMA access
  mmc: core: reset signal voltage on power up
  mmc: sd: Fix sd current limit setting
  mmc: omap_hsmmc: add clk_prepare and clk_unprepare
  mmc: sdhci: When a UHS switch fails, cycle power if regulator is used
  mmc: atmel-mci: modify CLKDIV displaying in debugfs
  mmc: atmel-mci: fix incorrect setting of host->data to NULL
  mmc: sdhci: poll for card even when card is logically unremovable
  mmc: sdhci: Introduce new flag SDHCI_USING_RETUNING_TIMER
  mmc: sdio: Change pr_warning to pr_warn_ratelimited
  mmc: core: Simplify and fix for SD switch processing
  mmc: sdhci: restore host settings when card is removed
  mmc: sdhci: fix incorrect command used in tuning
  mmc: sdhci-pci: CaFe has broken card detection
  mmc: sdhci: Report failure reasons for all cases in sdhci_add_host()
  mmc: s3cmci: Convert s3cmci driver to gpiolib API
  ...

12 years agoMerge branch 'x86-mce-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 22 Jul 2012 23:07:45 +0000 (16:07 -0700)]
Merge branch 'x86-mce-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86/mce changes from Ingo Molnar:
 "This tree improves the AMD thresholding bank code and includes a
  memory fault signal handling fixlet."

* 'x86-mce-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/mce: Fix siginfo_t->si_addr value for non-recoverable memory faults
  x86, MCE, AMD: Update copyrights and boilerplate
  x86, MCE, AMD: Give proper names to the thresholding banks
  x86, MCE, AMD: Make error_count read only
  x86, MCE, AMD: Cleanup reading of error_count
  x86, MCE, AMD: Print decimal thresholding values
  x86, MCE, AMD: Move shared bank to node descriptor
  x86, MCE, AMD: Remove local_allocate_... wrapper
  x86, MCE, AMD: Remove shared banks sysfs linking
  x86, amd_nb: Export model 0x10 and later PCI id

12 years agommc: sdhci-dove: Prepare for common clock framework
Sebastian Hesselbarth [Thu, 5 Jul 2012 10:14:01 +0000 (12:14 +0200)]
mmc: sdhci-dove: Prepare for common clock framework

As mach-dove is moving towards common clock framework prepare
the sdhci driver to grab its clock.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@googlemail.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: sdhci-dove: Add SDHCI_QUIRK_NO_HISPD_BIT
Sebastian Hesselbarth [Sat, 21 Jul 2012 23:26:19 +0000 (19:26 -0400)]
mmc: sdhci-dove: Add SDHCI_QUIRK_NO_HISPD_BIT

The sdio controller on dove doesn't have a bit to indicate
high-speed. With the quirk set it fixes accessing high-speed
sdcards.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@googlemail.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: omap_hsmmc: ensure probe returns error upon resource failure
Kevin Hilman [Tue, 10 Jul 2012 23:40:56 +0000 (16:40 -0700)]
mmc: omap_hsmmc: ensure probe returns error upon resource failure

If platform_get_resource_by_name() fails, driver probe is aborted an
should return an error so the driver is not bound to the device.

However, in the current error path of platform_get_resource_by_name(),
probe returns zero since the return value (ret) is not properly set.
With a zero return value, the driver core assumes probe was successful
and will bind the driver to the device.

Fix this by ensuring that probe returns an error code in this failure
path.

Signed-off-by: Kevin Hilman <khilman@ti.com>
Acked-by: Venkatraman S <svenkatr@ti.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: mxs-mmc: Add wp-inverted property
Marek Vasut [Thu, 19 Jul 2012 15:11:39 +0000 (11:11 -0400)]
mmc: mxs-mmc: Add wp-inverted property

The write-protect GPIO is inverted on some boards. Handle such case.

Signed-off-by: Marek Vasut <marex@denx.de>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: esdhc: Fix DMA_MASK to not break mx25 DMA access
Wilson Callan [Thu, 19 Jul 2012 06:49:16 +0000 (02:49 -0400)]
mmc: esdhc: Fix DMA_MASK to not break mx25 DMA access

Patch to not set reserved bits in i.MX25 PROCTL register.  DMA stops
working if those bits get set.

Signed-off-by: Wilson Callan <wilson.callan@savantsystems.com>
Acked-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agoMerge tag 'pm-for-3.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Sun, 22 Jul 2012 20:36:52 +0000 (13:36 -0700)]
Merge tag 'pm-for-3.6-rc1' of git://git./linux/kernel/git/rafael/linux-pm

Pull power management updates from Rafael Wysocki:

 - ACPI conversion to PM handling based on struct dev_pm_ops.
 - Conversion of a number of platform drivers to PM handling based on
   struct dev_pm_ops and removal of empty legacy PM callbacks from a
   couple of PCI drivers.
 - Suspend-to-both for in-kernel hibernation from Bojan Smojver.
 - cpuidle fixes and cleanups from ShuoX Liu, Daniel Lezcano and Preeti
   Murthy.
 - cpufreq bug fixes from Jonghwa Lee and Stephen Boyd.
 - Suspend and hibernate fixes from Srivatsa Bhat and Colin Cross.
 - Generic PM domains framework updates.
 - RTC CMOS wakeup signaling update from Paul Fox.
 - sparse warnings fixes from Sachin Kamat.
 - Build warnings fixes for the generic PM domains framework and PM
   sysfs code.
 - sysfs switch for printing device suspend times from Sameer Nanda.
 - Documentation fix from Oskar Schirmer.

* tag 'pm-for-3.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (70 commits)
  cpufreq: Fix sysfs deadlock with concurrent hotplug/frequency switch
  EXYNOS: bugfix on retrieving old_index from freqs.old
  PM / Sleep: call early resume handlers when suspend_noirq fails
  PM / QoS: Use NULL pointer instead of plain integer in qos.c
  PM / QoS: Use NULL pointer instead of plain integer in pm_qos.h
  PM / Sleep: Require CAP_BLOCK_SUSPEND to use wake_lock/wake_unlock
  PM / Sleep: Add missing static storage class specifiers in main.c
  cpuilde / ACPI: remove time from acpi_processor_cx structure
  cpuidle / ACPI: remove usage from acpi_processor_cx structure
  cpuidle / ACPI : remove latency_ticks from acpi_processor_cx structure
  rtc-cmos: report wakeups from interrupt handler
  PM / Sleep: Fix build warning in sysfs.c for CONFIG_PM_SLEEP unset
  PM / Domains: Fix build warning for CONFIG_PM_RUNTIME unset
  olpc-xo15-sci: Use struct dev_pm_ops for power management
  PM / Domains: Replace plain integer with NULL pointer in domain.c file
  PM / Domains: Add missing static storage class specifier in domain.c file
  PM / crypto / ux500: Use struct dev_pm_ops for power management
  PM / IPMI: Remove empty legacy PCI PM callbacks
  tpm_nsc: Use struct dev_pm_ops for power management
  tpm_tis: Use struct dev_pm_ops for power management
  ...

12 years agoMerge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target...
Linus Torvalds [Sun, 22 Jul 2012 20:31:57 +0000 (13:31 -0700)]
Merge branch 'for-next' of git://git./linux/kernel/git/nab/target-pending

Pull target updates from Nicholas Bellinger:
 "There have been lots of work in a number of areas this past round.
  The highlights include:

   - Break out target_core_cdb.c emulation into SPC/SBC ops (hch)
   - Add a parse_cdb method to target backend drivers (hch)
   - Move sync_cache + write_same + unmap into spc_ops (hch)
   - Use target_execute_cmd for WRITEs in iscsi_target + srpt (hch)
   - Offload WRITE I/O backend submission in tcm_qla2xxx + tcm_fc (hch +
     nab)
   - Refactor core_update_device_list_for_node() into enable/disable
     funcs (agrover)
   - Replace the TCM processing thread with a TMR work queue (hch)
   - Fix regression in transport_add_device_to_core_hba from TMR
     conversion (DanC)
   - Remove racy, now-redundant check of sess_tearing_down with qla2xxx
     (roland)
   - Add range checking, fix reading of data len + possible underflow in
     UNMAP (roland)
   - Allow for target_submit_cmd() returning errors + convert fabrics
     (roland + nab)
   - Drop bogus struct file usage for iSCSI/SCTP (viro)"

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending: (54 commits)
  iscsi-target: Drop bogus struct file usage for iSCSI/SCTP
  target: NULL dereference on error path
  target: Allow for target_submit_cmd() returning errors
  target: Check number of unmap descriptors against our limit
  target: Fix possible integer underflow in UNMAP emulation
  target: Fix reading of data length fields for UNMAP commands
  target: Add range checking to UNMAP emulation
  target: Add generation of LOGICAL BLOCK ADDRESS OUT OF RANGE
  target: Make unnecessarily global se_dev_align_max_sectors() static
  target: Remove se_session.sess_wait_list
  qla2xxx: Remove racy, now-redundant check of sess_tearing_down
  target: Check sess_tearing_down in target_get_sess_cmd()
  sbp-target: Consolidate duplicated error path code in sbp_handle_command()
  target: Un-export target_get_sess_cmd()
  qla2xxx: Get rid of redundant qla_tgt_sess.tearing_down
  target: Make core_disable_device_list_for_node use pre-refactoring lock ordering
  target: refactor core_update_device_list_for_node()
  target: Eliminate else using boolean logic
  target: Misc retval cleanups
  target: Remove hba param from core_dev_add_lun
  ...

12 years agoMerge tag 'regulator-3.6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie...
Linus Torvalds [Sun, 22 Jul 2012 20:05:23 +0000 (13:05 -0700)]
Merge tag 'regulator-3.6' of git://git./linux/kernel/git/broonie/regulator

Pull regulator updates from Mark Brown:
 "Lots and lots of fixes from Axel and some others here, plus some
  framework enhancements which continue the theme of factoring code out
  of the drivers and into the core.

   - Initial framework support for GPIO controlled enable signals,
     saving a bunch of code in drivers.
   - Move fixed regulator enable time and voltage mapping table
     specifications to data.
   - Used some of the recent framework enhancements to make voltage
     change notifications more useful, passing the voltage in as an
     argument to the notification.
   - Fixed the pattern used for finding individual regulators on a
     device to not rely on the node name, supporting the use of multiple
     PMICs of the same type in the system.
   - New drivers for Maxim MAX77686, TI LP872x and LP8788, Samsung
     S2MPS11, and Wolfson Arizona microphone supplies and LDOs."

* tag 'regulator-3.6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator: (176 commits)
  regulator: add new lp8788 regulator driver
  regulator: mc13xxx: Remove extern function declaration for mc13xxx_sw_regulator
  regulator: tps65910: set input_supply on desc unconditionally
  regulator: palmas: Fix calcuating selector in palmas_map_voltage_smps
  regulator: lp872x: Simplify implementation of lp872x_find_regulator_init_data()
  regulator: twl: Fix list_voltate for twl6030ldo_ops
  regulator: twl: Convert twl6030ldo_ops to [get|set]_voltage_sel
  regulator: twl: Fix the formula to calculate vsel and voltage for twl6030ldo
  regulator: s5m8767: Properly handle gpio_request failure
  regulator: max8997: Properly handle gpio_request failure
  regulator: tps62360: use devm_* for gpio request
  regulator: tps6586x: add support for input supply
  regulator: tps65217: Add device tree support
  regulator: aat2870: Remove unused min_uV and max_uV from struct aat2870_regulator
  regulator: aat2870: Convert to regulator_list_voltage_table
  regulator: da9052: initialize of_node param for regulator register
  regulator: Add REGULATOR_STATUS_UNDEFINED.
  regulator: Fix a typo in regulator_mode_to_status() core function.
  regulator: s2mps11: Use sec_reg_write rather than sec_reg_update when mask is 0xff
  regulator: s2mps11: Fix wrong setting for config.dev
  ...

12 years agoMerge tag 'regmap-3.6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap
Linus Torvalds [Sun, 22 Jul 2012 20:03:14 +0000 (13:03 -0700)]
Merge tag 'regmap-3.6' of git://git./linux/kernel/git/broonie/regmap

Pull regmap updates from Mark Brown:
 "A few fixes plus a few features, the most generally useful thing being
  the register paging support which can be used by quite a few devices:

   - Support for wake IRQs in regmap-irq
   - Support for register paging
   - Support for explicitly specified endianness, mostly for MMIO."

* tag 'regmap-3.6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
  regmap: Fix incorrect arguments to kzalloc() call
  regmap: Add hook for printk logging for debugging during early init
  regmap: Fix work_buf switching for page update during virtual range access.
  regmap: Add support for register indirect addressing.
  regmap: Move lock out from internal function _regmap_update_bits().
  regmap: mmio: Staticize regmap_mmio_gen_context()
  regmap: Remove warning on stubbed dev_get_regmap()
  regmap: Implement support for wake IRQs
  regmap: Don't try to map non-existant IRQs
  regmap: Constify regmap_irq_chip
  regmap: mmio: request native endian formatting
  regmap: allow busses to request formatting with specific endianness

12 years agoext4: switch EXT4_IOC_RESIZE_FS to mnt_want_write_file()
Al Viro [Thu, 19 Jul 2012 07:19:07 +0000 (11:19 +0400)]
ext4: switch EXT4_IOC_RESIZE_FS to mnt_want_write_file()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agobtrfs: switch btrfs_ioctl_balance() to mnt_want_write_file()
Al Viro [Thu, 19 Jul 2012 07:17:49 +0000 (11:17 +0400)]
btrfs: switch btrfs_ioctl_balance() to mnt_want_write_file()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agoswitch dentry_open() to struct path, make it grab references itself
Al Viro [Tue, 26 Jun 2012 17:58:53 +0000 (21:58 +0400)]
switch dentry_open() to struct path, make it grab references itself

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agospufs: shift dget/mntget towards dentry_open()
Al Viro [Mon, 25 Jun 2012 07:46:13 +0000 (11:46 +0400)]
spufs: shift dget/mntget towards dentry_open()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agozoran: don't bother with struct file * in zoran_map
Al Viro [Sat, 14 Jul 2012 09:49:40 +0000 (13:49 +0400)]
zoran: don't bother with struct file * in zoran_map

all we need it for is file->private_data, which is assign-once, already
assigned by that point and, incidentally, its value is already in use
by zoran ->mmap() anyway.  So just store that pointer instead...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agoecryptfs: don't reinvent the wheels, please - use struct completion
Al Viro [Mon, 25 Jun 2012 07:38:56 +0000 (11:38 +0400)]
ecryptfs: don't reinvent the wheels, please - use struct completion

... and keep the sodding requests on stack - they are small enough.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agodon't expose I_NEW inodes via dentry->d_inode
Al Viro [Thu, 19 Jul 2012 05:18:15 +0000 (09:18 +0400)]
don't expose I_NEW inodes via dentry->d_inode

d_instantiate(dentry, inode);
unlock_new_inode(inode);

is a bad idea; do it the other way round...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agotidy up namei.c a bit
Al Viro [Wed, 18 Jul 2012 16:43:19 +0000 (20:43 +0400)]
tidy up namei.c a bit

locking/unlocking for rcu walk taken to a couple of inline helpers

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agounobfuscate follow_up() a bit
Al Viro [Wed, 18 Jul 2012 13:32:50 +0000 (17:32 +0400)]
unobfuscate follow_up() a bit

really convoluted test in there has grown up during struct mount
introduction; what it checks is that we'd reached the root of
mount tree.

12 years agoext3: pass custom EOF to generic_file_llseek_size()
Eric Sandeen [Mon, 30 Apr 2012 18:16:04 +0000 (13:16 -0500)]
ext3: pass custom EOF to generic_file_llseek_size()

Use the new custom EOF argument to generic_file_llseek_size so
that SEEK_END will go to the max hash value for htree dirs
in ext3 rather than to i_size_read()

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agoext4: use core vfs llseek code for dir seeks
Eric Sandeen [Mon, 30 Apr 2012 18:14:03 +0000 (13:14 -0500)]
ext4: use core vfs llseek code for dir seeks

Use the new functionality in generic_file_llseek_size() to
accept a custom EOF position, and un-cut-and-paste all the
vfs llseek code from ext4.

Also fix up comments on ext4_llseek() to reflect reality.

Signed-off-by: Eric Sandeen <sandeen@redaht.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agovfs: allow custom EOF in generic_file_llseek code
Eric Sandeen [Mon, 30 Apr 2012 18:11:29 +0000 (13:11 -0500)]
vfs: allow custom EOF in generic_file_llseek code

For ext3/4 htree directories, using the vfs llseek function with
SEEK_END goes to i_size like for any other file, but in reality
we want the maximum possible hash value.  Recent changes
in ext4 have cut & pasted generic_file_llseek() back into fs/ext4/dir.c,
but replicating this core code seems like a bad idea, especially
since the copy has already diverged from the vfs.

This patch updates generic_file_llseek_size to accept
both a custom maximum offset, and a custom EOF position.  With this
in place, ext4_dir_llseek can pass in the appropriate maximum hash
position for both maxsize and eof, and get what it wants.

As far as I know, this does not fix any bugs - nfs in the kernel
doesn't use SEEK_END, and I don't know of any user who does.  But
some ext4 folks seem keen on doing the right thing here, and I can't
really argue.

(Patch also fixes up some comments slightly)

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agovfs: Avoid unnecessary WB_SYNC_NONE writeback during sys_sync and reorder sync passes
Jan Kara [Tue, 3 Jul 2012 14:45:34 +0000 (16:45 +0200)]
vfs: Avoid unnecessary WB_SYNC_NONE writeback during sys_sync and reorder sync passes

wakeup_flusher_threads(0) will queue work doing complete writeback for each
flusher thread. Thus there is not much point in submitting another work doing
full inode WB_SYNC_NONE writeback by writeback_inodes_sb().

After this change it does not make sense to call nonblocking ->sync_fs and
block device flush before calling sync_inodes_sb() because
wakeup_flusher_threads() is completely asynchronous and thus these functions
would be called in parallel with inode writeback running which will effectively
void any work they do. So we move sync_inodes_sb() call before these two
functions.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agovfs: Remove unnecessary flushing of block devices
Jan Kara [Tue, 3 Jul 2012 14:45:33 +0000 (16:45 +0200)]
vfs: Remove unnecessary flushing of block devices

It is not necessary to write block devices twice. The reason why we first did
flush and then proper sync is that
  for_each_bdev() {
    write_bdev()
    wait_for_completion()
  }
is much slower than
  for_each_bdev()
    write_bdev()
  for_each_bdev()
    wait_for_completion()
when there is bigger amount of data. But as is seen in the above, there's no real
need to scan pages and submit them twice. We just need to separate the submission
and waiting part. This patch does that.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agovfs: Make sys_sync writeout also block device inodes
Jan Kara [Tue, 3 Jul 2012 14:45:32 +0000 (16:45 +0200)]
vfs: Make sys_sync writeout also block device inodes

In case block device does not have filesystem mounted on it, sys_sync will just
ignore it and doesn't writeout its dirty pages. This is because writeback code
avoids writing inodes from superblock without backing device and
blockdev_superblock is such a superblock.  Since it's unexpected that sync
doesn't writeout dirty data for block devices be nice to users and change the
behavior to do so. So now we iterate over all block devices on blockdev_super
instead of iterating over all superblocks when syncing block devices.

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>
12 years agovfs: Create function for iterating over block devices
Jan Kara [Tue, 3 Jul 2012 14:45:31 +0000 (16:45 +0200)]
vfs: Create function for iterating over block devices

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agovfs: Reorder operations during sys_sync
Jan Kara [Tue, 3 Jul 2012 14:45:30 +0000 (16:45 +0200)]
vfs: Reorder operations during sys_sync

Change the order of operations during sync from

for_each_sb {
        writeback_inodes_sb();
        sync_fs(nowait);
        __sync_blockdev(nowait);
}
for_each_sb {
        sync_inodes_sb();
        sync_fs(wait);
        __sync_blockdev(wait);
}

to

for_each_sb
        writeback_inodes_sb();
for_each_sb
        sync_fs(nowait);
for_each_sb
        __sync_blockdev(nowait);
for_each_sb
        sync_inodes_sb();
for_each_sb
        sync_fs(wait);
for_each_sb
        __sync_blockdev(wait);

This is a preparation for the following patches in this series.

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>
12 years agoquota: Move quota syncing to ->sync_fs method
Jan Kara [Tue, 3 Jul 2012 14:45:29 +0000 (16:45 +0200)]
quota: Move quota syncing to ->sync_fs method

Since the moment writes to quota files are using block device page cache and
space for quota structures is reserved at the moment they are first accessed we
have no reason to sync quota before inode writeback. In fact this order is now
only harmful since quota information can easily change during inode writeback
(either because conversion of delayed-allocated extents or simply because of
allocation of new blocks for simple filesystems not using page_mkwrite).

So move syncing of quota information after writeback of inodes into ->sync_fs
method. This way we do not have to use ->quota_sync callback which is primarily
intended for use by quotactl syscall anyway and we get rid of calling
->sync_fs() twice unnecessarily. We skip quota syncing for OCFS2 since it does
proper quota journalling in all cases (unlike ext3, ext4, and reiserfs which
also support legacy non-journalled quotas) and thus there are no dirty quota
structures.

CC: "Theodore Ts'o" <tytso@mit.edu>
CC: Joel Becker <jlbec@evilplan.org>
CC: reiserfs-devel@vger.kernel.org
Acked-by: Steven Whitehouse <swhiteho@redhat.com>
Acked-by: Dave Kleikamp <shaggy@kernel.org>
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>
12 years agoquota: Split dquot_quota_sync() to writeback and cache flushing part
Jan Kara [Tue, 3 Jul 2012 14:45:28 +0000 (16:45 +0200)]
quota: Split dquot_quota_sync() to writeback and cache flushing part

Split off part of dquot_quota_sync() which writes dquots into a quota file
to a separate function. In the next patch we will use the function from
filesystems and we do not want to abuse ->quota_sync quotactl callback more
than necessary.

Acked-by: Steven Whitehouse <swhiteho@redhat.com>
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>
12 years agovfs: Move noop_backing_dev_info check from sync into writeback
Jan Kara [Tue, 3 Jul 2012 14:45:27 +0000 (16:45 +0200)]
vfs: Move noop_backing_dev_info check from sync into writeback

In principle, a filesystem may want to have ->sync_fs() called during sync(1)
although it does not have a bdi (i.e. s_bdi is set to noop_backing_dev_info).
Only writeback code really needs bdi set to something reasonable. So move the
checks where they are more logical.

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>
12 years agofs/ufs: get rid of write_super
Artem Bityutskiy [Thu, 12 Jul 2012 13:28:08 +0000 (16:28 +0300)]
fs/ufs: get rid of write_super

This patch makes UFS stop using the VFS '->write_super()' method along with
the 's_dirt' superblock flag, because they are on their way out.

The way we implement this is that we schedule a delay job instead relying on
's_dirt' and '->write_super()'.

The whole "superblock write-out" VFS infrastructure is served by the
'sync_supers()' kernel thread, which wakes up every 5 (by default) seconds and
writes out all dirty superblocks using the '->write_super()' call-back.  But the
problem with this thread is that it wastes power by waking up the system every
5 seconds, even if there are no diry superblocks, or there are no client
file-systems which would need this (e.g., btrfs does not use
'->write_super()'). So we want to kill it completely and thus, we need to make
file-systems to stop using the '->write_super()' VFS service, and then remove
it together with the kernel thread.

Tested using fsstress from the LTP project.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agofs/ufs: re-arrange the code a bit
Artem Bityutskiy [Thu, 12 Jul 2012 13:28:07 +0000 (16:28 +0300)]
fs/ufs: re-arrange the code a bit

This patch does not do any functional changes. It only moves 3 functions
in fs/ufs/super.c a little bit up in order to prepare for further changes
where I'll need this new arrangement to avoid forward declarations.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agofs/ufs: remove extra superblock write on unmount
Artem Bityutskiy [Thu, 12 Jul 2012 13:28:06 +0000 (16:28 +0300)]
fs/ufs: remove extra superblock write on unmount

UFS calls 'ufs_write_super()' from 'ufs_put_super()' in order to write the
superblocks to the media. However, it is not needed because VFS calls
'->sync_fs()' before calling '->put_super()' - so by the time we are in
'ufs_write_super()', the superblocks are already synchronized.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agofs/sysv: stop using write_super and s_dirt
Artem Bityutskiy [Tue, 3 Jul 2012 13:43:28 +0000 (16:43 +0300)]
fs/sysv: stop using write_super and s_dirt

It does not look like sysv FS needs 'write_super()' at all, because all it
does is a timestamp update. I cannot test this patch, because this
file-system is so old and probably has not been used by anyone for years,
so there are no tools to create it in Linux. But from the code I see that
marking the superblock as dirty is basically marking the superblock buffers as
drity and then setting the s_dirt flag. And when 'write_super()' is executed to
handle the s_dirt flag, we just update the timestamp and again mark the
superblock buffer as dirty. Seems pointless.

It looks like we can update the timestamp more opprtunistically - on unmount
or remount of sync, and nothing should change.

Thus, this patch removes 'sysv_write_super()' and 's_dirt'.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agofs/sysv: remove another useless write_super call
Artem Bityutskiy [Tue, 3 Jul 2012 13:43:27 +0000 (16:43 +0300)]
fs/sysv: remove another useless write_super call

We do not need to call 'sysv_write_super()' from 'sysv_remount()',
because VFS has called 'sysv_sync_fs()' before calling '->remount()'.
So remove it. Remove also '(un)lock_super()' which obvioulsy is becoming
useless in this function.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agofs/sysv: remove useless write_super call
Artem Bityutskiy [Tue, 3 Jul 2012 13:43:26 +0000 (16:43 +0300)]
fs/sysv: remove useless write_super call

We do not need to call 'sysv_write_super()' from 'sysv_put_super()',
because VFS has called 'sysv_sync_fs()' before calling '->put_super()'.
So remove it.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agohfs: get rid of hfs_sync_super
Artem Bityutskiy [Thu, 12 Jul 2012 14:28:49 +0000 (17:28 +0300)]
hfs: get rid of hfs_sync_super

This patch makes hfs stop using the VFS '->write_super()' method along with
the 's_dirt' superblock flag, because they are on their way out.

The whole "superblock write-out" VFS infrastructure is served by the
'sync_supers()' kernel thread, which wakes up every 5 (by default) seconds and
writes out all dirty superblocks using the '->write_super()' call-back.  But the
problem with this thread is that it wastes power by waking up the system every
5 seconds, even if there are no diry superblocks, or there are no client
file-systems which would need this (e.g., btrfs does not use
'->write_super()'). So we want to kill it completely and thus, we need to make
file-systems to stop using the '->write_super()' VFS service, and then remove
it together with the kernel thread.

Tested using fsstress from the LTP project.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agohfs: introduce VFS superblock object back-reference
Artem Bityutskiy [Thu, 12 Jul 2012 14:28:48 +0000 (17:28 +0300)]
hfs: introduce VFS superblock object back-reference

Add an 'sb' VFS superblock back-reference to the 'struct hfs_sb_info' data
structure - we will need to find the VFS superblock from a
'struct hfs_sb_info' object in the next patch, so this change is jut a
preparation.

Remove few useless newlines while on it.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agohfs: simplify a bit checking for R/O
Artem Bityutskiy [Thu, 12 Jul 2012 14:28:47 +0000 (17:28 +0300)]
hfs: simplify a bit checking for R/O

We have the following pattern in 2 places in HFS

if (!RDONLY)
hfs_mdb_commit();

This patch pushes the RDONLY check down to 'hfs_mdb_commit()'. This will
make the following patches a bit simpler.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agohfs: remove extra mdb write on unmount
Artem Bityutskiy [Thu, 12 Jul 2012 14:28:46 +0000 (17:28 +0300)]
hfs: remove extra mdb write on unmount

HFS calls 'hfs_write_super()' from 'hfs_put_super()' in order to write the MDB
to the media. However, it is not needed because VFS calls '->sync_fs()' before
calling '->put_super()' - so by the time we are in 'hfs_write_super()', the MDB
is already synchronized.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agohfs: get rid of lock_super
Artem Bityutskiy [Thu, 12 Jul 2012 14:28:45 +0000 (17:28 +0300)]
hfs: get rid of lock_super

Stop using lock_super for serializing the MDB changes - use the buffer-head own
lock instead. Tested with fsstress.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agohfs: push lock_super down
Artem Bityutskiy [Thu, 12 Jul 2012 14:28:44 +0000 (17:28 +0300)]
hfs: push lock_super down

HFS uses 'lock_super()'/'unlock_super()' around 'hfs_mdb_commit()' in order
to serialize MDB (Master Directory Block) changes. Push it down to
'hfs_mdb_commit()' in order to simplify the code a bit.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agohfsplus: get rid of write_super
Artem Bityutskiy [Thu, 12 Jul 2012 14:26:31 +0000 (17:26 +0300)]
hfsplus: get rid of write_super

This patch makes hfsplus stop using the VFS '->write_super()' method along with
the 's_dirt' superblock flag, because they are on their way out.

The whole "superblock write-out" VFS infrastructure is served by the
'sync_supers()' kernel thread, which wakes up every 5 (by default) seconds and
writes out all dirty superblocks using the '->write_super()' call-back.  But the
problem with this thread is that it wastes power by waking up the system every
5 seconds, even if there are no diry superblocks, or there are no client
file-systems which would need this (e.g., btrfs does not use
'->write_super()'). So we want to kill it completely and thus, we need to make
file-systems to stop using the '->write_super()' VFS service, and then remove
it together with the kernel thread.

Tested using fsstress from the LTP project.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agohfsplus: remove useless check
Artem Bityutskiy [Thu, 12 Jul 2012 14:26:30 +0000 (17:26 +0300)]
hfsplus: remove useless check

This check is useless because we always have 'sb->s_fs_info' to be non-NULL.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agohfsplus: amend debugging print
Artem Bityutskiy [Thu, 12 Jul 2012 14:26:29 +0000 (17:26 +0300)]
hfsplus: amend debugging print

Print correct function name in the debugging print of the
'hfsplus_sync_fs()' function.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agohfsplus: make hfsplus_sync_fs static
Artem Bityutskiy [Thu, 12 Jul 2012 14:26:28 +0000 (17:26 +0300)]
hfsplus: make hfsplus_sync_fs static

... because it is used only in fs/hfsplus/super.c.

Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agohold task_lock around checks in keyctl
Al Viro [Sat, 30 Jun 2012 07:55:24 +0000 (11:55 +0400)]
hold task_lock around checks in keyctl

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agoget rid of ->scm_work_list
Al Viro [Sun, 24 Jun 2012 06:03:05 +0000 (10:03 +0400)]
get rid of ->scm_work_list

recursion in __scm_destroy() will be cut by delaying final fput()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agoaio: now fput() is OK from interrupt context; get rid of manual delayed __fput()
Al Viro [Sun, 24 Jun 2012 06:00:10 +0000 (10:00 +0400)]
aio: now fput() is OK from interrupt context; get rid of manual delayed __fput()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agoswitch fput to task_work_add
Al Viro [Sun, 24 Jun 2012 05:56:45 +0000 (09:56 +0400)]
switch fput to task_work_add

... and schedule_work() for interrupt/kernel_thread callers
(and yes, now it *is* OK to call from interrupt).

We are guaranteed that __fput() will be done before we return
to userland (or exit).  Note that for fput() from a kernel
thread we get an async behaviour; it's almost always OK, but
sometimes you might need to have __fput() completed before
you do anything else.  There are two mechanisms for that -
a general barrier (flush_delayed_fput()) and explicit
__fput_sync().  Both should be used with care (as was the
case for fput() from kernel threads all along).  See comments
in fs/file_table.c for details.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agodeal with task_work callbacks adding more work
Al Viro [Wed, 27 Jun 2012 07:33:29 +0000 (11:33 +0400)]
deal with task_work callbacks adding more work

It doesn't matter on normal return to userland path (we'll recheck the
NOTIFY_RESUME flag anyway), but in case of exit_task_work() we'll
need that as soon as we get callbacks capable of triggering more
task_work_add().

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agomove exit_task_work() past exit_files() et.al.
Al Viro [Wed, 27 Jun 2012 07:31:24 +0000 (11:31 +0400)]
move exit_task_work() past exit_files() et.al.

... and get rid of PF_EXITING check in task_work_add().

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agomerge task_work and rcu_head, get rid of separate allocation for keyring case
Al Viro [Wed, 27 Jun 2012 07:07:19 +0000 (11:07 +0400)]
merge task_work and rcu_head, get rid of separate allocation for keyring case

task_work and rcu_head are identical now; merge them (calling the result
struct callback_head, rcu_head #define'd to it), kill separate allocation
in security/keys since we can just use cred->rcu now.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agotrim task_work: get rid of hlist
Al Viro [Wed, 27 Jun 2012 05:24:13 +0000 (09:24 +0400)]
trim task_work: get rid of hlist

layout based on Oleg's suggestion; single-linked list,
task->task_works points to the last element, forward pointer
from said last element points to head.  I'd still prefer
much more regular scheme with two pointers in task_work,
but...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agotrimming task_work: kill ->data
Al Viro [Tue, 26 Jun 2012 18:10:04 +0000 (22:10 +0400)]
trimming task_work: kill ->data

get rid of the only user of ->data; this is _not_ the final variant - in the
end we'll have task_work and rcu_head identical and just use cred->rcu,
at which point the separate allocation will be gone completely.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agosignal: make sure we don't get stopped with pending task_work
Al Viro [Sun, 15 Jul 2012 10:10:52 +0000 (14:10 +0400)]
signal: make sure we don't get stopped with pending task_work

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agouse __lookup_hash() in kern_path_parent()
Al Viro [Sun, 22 Jul 2012 19:46:21 +0000 (23:46 +0400)]
use __lookup_hash() in kern_path_parent()

No need to bother with lookup_one_len() here - it's an overkill

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

12 years agoMerge branch 'x86-build-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 22 Jul 2012 19:46:27 +0000 (12:46 -0700)]
Merge branch 'x86-build-for-linus' of git://git./linux/kernel/git/tip/tip

Pull a x86/build change from Ingo Molnar.

This makes the default stack alignment on x86-64 be just 8, allowing for
improved code generation (it can avoid some unnecessary extra alignment
logic and use just pure push/pop sequences) and smaller stack frames.

We can't generally do SSE with 16-byte alignment issues in the kernel anyway.

* 'x86-build-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86-64, gcc: Use -mpreferred-stack-boundary=3 if supported

12 years agoMerge branch 'x86-uv-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 22 Jul 2012 19:37:15 +0000 (12:37 -0700)]
Merge branch 'x86-uv-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86/uv changes from Ingo Molnar:
 "UV2 BAU productization fixes.

  The BAU (Broadcast Assist Unit) is SGI's fancy out of line way on UV
  hardware to do TLB flushes, instead of the normal APIC IPI methods.
  The commits here fix / work around hangs in their latest hardware
  iteration (UV2).

  My understanding is that the main purpose of the out of line
  signalling channel is to improve scalability: the UV APIC hardware
  glue does not handle broadcasting to many CPUs very well, and this
  matters most for TLB shootdowns.

  [ I don't agree with all aspects of the current approach: in hindsight
    it would have been better to link the BAU at the IPI/APIC driver
    level instead of the TLB shootdown level, where TLB flushes are
    really just one of the uses of broadcast SMP messages.  Doing that
    would improve scalability in some other ways and it would also
    remove a few uglies from the TLB path.  It would also be nice to
    push more is_uv_system() tests into proper x86_init or x86_platform
    callbacks.  Cliff? ]"

* 'x86-uv-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/uv: Work around UV2 BAU hangs
  x86/uv: Implement UV BAU runtime enable and disable control via /proc/sgi_uv/
  x86/uv: Fix the UV BAU destination timeout period

12 years agommc: core: reset signal voltage on power up
Aaron Lu [Tue, 10 Jul 2012 08:55:37 +0000 (16:55 +0800)]
mmc: core: reset signal voltage on power up

Add a call to mmc_set_signal_voltage() to set signal voltage to 3.3v in
mmc_power_up so that we do not need to touch signal voltage setting in
mmc/sd/sdio init functions and rescan function.

For mmc/sd cards, when doing a suspend/resume cycle, consider the unsafe
resume case, the card will lose its power and when powered on again, we
will set signal voltage to 3.3v in mmc_power_up before its resume function
gets called, which will re-init the card.

And for sdio cards, when doing a suspend/resume cycle, consider the unsafe
resume case, the card will either lose its power or not depending on if it
wants to wakeup the host. If power is not maintained, it is the same case as
mmc/sd cards. If power is maintained, mmc_power_up will not be called and
the card's signal voltage will remain at the last setting.

Signed-off-by: Aaron Lu <aaron.lu@amd.com>
Tested-by: Venkatraman S <svenkatr@ti.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: sd: Fix sd current limit setting
Aaron Lu [Wed, 4 Jul 2012 05:31:48 +0000 (13:31 +0800)]
mmc: sd: Fix sd current limit setting

Host has different current capabilities at different voltages, we need
to record these settings seperately. The defined voltages are 1.8/3.0/3.3.
For other voltages, we do not touch current limit setting.

Before we set the current limit for the sd card, find out the host's
operating voltage first and then find out the current capabilities of
the host at that voltage to set the current limit.

Signed-off-by: Aaron Lu <aaron.lu@amd.com>
Reviewed-by: Philip Rakity <prakity@marvell.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: omap_hsmmc: add clk_prepare and clk_unprepare
Rajendra Nayak [Wed, 27 Jun 2012 08:49:54 +0000 (14:19 +0530)]
mmc: omap_hsmmc: add clk_prepare and clk_unprepare

In preparation for OMAP moving to the Common Clock Framework (CCF)
add clk_prepare() and clk_unprepare() for the hsmmc clocks.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Acked-by: Venkatraman S <svenkatr@ti.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: sdhci: When a UHS switch fails, cycle power if regulator is used
Philip Rakity [Wed, 27 Jun 2012 16:15:00 +0000 (09:15 -0700)]
mmc: sdhci: When a UHS switch fails, cycle power if regulator is used

Power needs to be removed from the card when switching to 1.8v fails.

If a regulator is used to control vmmc we need to turn the
regulator off and then back on otherwise power will not be
removed from the card.

Signed-off-by: Philip Rakity <prakity@marvell.com>
Reviewed-by: Aaron Lu <aaron.lu@amd.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: atmel-mci: modify CLKDIV displaying in debugfs
Nicolas Ferre [Fri, 6 Jul 2012 10:11:51 +0000 (12:11 +0200)]
mmc: atmel-mci: modify CLKDIV displaying in debugfs

Modify clock division displaying in debugfs for matching
the new CLKDIV,CLKODD user interface arrangement.
Is using the has_odd_clk_div property to choose the proper format.

Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Acked-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: atmel-mci: fix incorrect setting of host->data to NULL
Nicolas Ferre [Fri, 6 Jul 2012 09:58:33 +0000 (11:58 +0200)]
mmc: atmel-mci: fix incorrect setting of host->data to NULL

Setting host->data to NULL is incorrect sequence in STATE_SENDING_STOP
state of FSM: This early setting leads to the skip of dma_unmap_sg()
in atmci_dma_cleanup() which is a bug.

Idea taken from dw_mmc by Seungwon Jeon.

Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: Seungwon Jeon <tgih.jun@samsung.com>
Acked-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: sdhci: poll for card even when card is logically unremovable
Daniel Drake [Thu, 5 Jul 2012 21:06:13 +0000 (22:06 +0100)]
mmc: sdhci: poll for card even when card is logically unremovable

The Marvell CaFe is now marked as having bad card detection to fix
a problem during system resume.

Now on the OLPC XO-1 we are facing the issue that the card is marked
as logically unremovable (via MMC_UNSAFE_RESUME), which means that
mmc_card_is_removable considers the card non-removable. The existing
code logic decides not to poll for card presence in this case, and
card detection is also disabled because of the quirk being set.
This means that no SD cards are detected when inserted after boot.

Refine the logic to enable card presence polling in the case when
a card is logically unremovable, only avoiding the poll in the case
when the card is physically non-removable (denoted with
MMC_CAP_NONREMOVABLE).

Signed-off-by: Daniel Drake <dsd@laptop.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: sdhci: Introduce new flag SDHCI_USING_RETUNING_TIMER
Aaron Lu [Wed, 4 Jul 2012 05:29:09 +0000 (13:29 +0800)]
mmc: sdhci: Introduce new flag SDHCI_USING_RETUNING_TIMER

Add a new flag of SDHCI_USING_RETUNING_TIMER to represent if the host
is using a retuning timer for the card inserted.

This flag is set when the host does tuning the first time for the card
and the host's retuning mode is 1. This flag is used afterwards whenever
needs to decide if the host is currently using a retuning timer.

This flag is cleared when the card is removed in sdhci_reinit.

The set/clear of the flag and the start/stop of the retuning timer is
associated with the card's init/remove time, so there is no need to
touch it when the host is to be removed as at that time the card should
have already been removed.

Signed-off-by: Aaron Lu <aaron.lu@amd.com>
Reviewed-by: Girish K S <girish.shivananjappa@linaro.org>
Reviewed-by: Philip Rakity <prakity@marvell.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: sdio: Change pr_warning to pr_warn_ratelimited
Liu Chuansheng [Mon, 9 Jul 2012 10:26:42 +0000 (06:26 -0400)]
mmc: sdio: Change pr_warning to pr_warn_ratelimited

When debugging one bad issue, got lots of pr_warning messages
"queuing unknown CIS tuple" which caused a printk storm and
flooded the console.

This patch changes the pr_warning to use pr_warn_ratelimited.

Signed-off-by: Liu Chuansheng <chuansheng.liu@intel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agoMerge branch 'x86-reboot-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 22 Jul 2012 19:25:47 +0000 (12:25 -0700)]
Merge branch 'x86-reboot-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86/reboot changes from Ingo Molnar:
 "Now that the revampted x86 real-mode trampoline code is upstream and
  seems to be working well, we can extend the 64-bit reboot code to be
  as capable as the 32-bit one."

* 'x86-reboot-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86-64, reboot: Be more paranoid in 64-bit reboot=bios
  x86, reboot: Drop redundant write of reboot_mode
  x86-64, reboot: Allow reboot=bios and reboot-cpu override on x86-64

12 years agommc: core: Simplify and fix for SD switch processing
Aaron Lu [Tue, 3 Jul 2012 06:16:11 +0000 (14:16 +0800)]
mmc: core: Simplify and fix for SD switch processing

In mmc_read_switch, just do a one time mode 0 switch command to get the
support bits information, no need to do multiple times as the support
bits do not change with different arguments.

And no need to check current limit support bits, as these bits are
fixed according to the signal voltage. If the signal voltage is 1.8V,
the support bits would be 0xf and if the signal voltage is 3.3V, the
support bits would be 0x01. We will check host's ability to set the
current limit.

Signed-off-by: Aaron Lu <aaron.lu@amd.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
12 years agommc: sdhci: restore host settings when card is removed
Aaron Lu [Fri, 29 Jun 2012 08:17:31 +0000 (16:17 +0800)]
mmc: sdhci: restore host settings when card is removed

Some of the host settings are affected by different cards inserted, e.g.
when an UHS-I card is inserted, the SDHCI_NEEDS_RETUING flag might be
set when the tuning timer expired and host's max_blk_count will be
reduced to make sure the data transfer for a command does not exceed 4MiB
to meet the retuning mode 1's requirement.

When the card is removed, we should restore the original setting of the
host since we can't be sure the next card being inserted will still be
an UHS-I card that needs tuning. The original setting include its
max_blk_count and no set of the flag of SDHCI_NEEDS_RETUNING.

Signed-off-by: Aaron Lu <aaron.lu@amd.com>
Signed-off-by: Chris Ball <cjb@laptop.org>