openwrt/staging/blogic.git
13 years agoMerge branches 'cma', 'cxgb4', 'ipath' and 'qib' into for-next
Roland Dreier [Tue, 15 Mar 2011 17:58:04 +0000 (10:58 -0700)]
Merge branches 'cma', 'cxgb4', 'ipath' and 'qib' into for-next

13 years agoRDMA/cma: Replace global lock in rdma_destroy_id() with id-specific one
Sean Hefty [Wed, 23 Feb 2011 17:05:39 +0000 (09:05 -0800)]
RDMA/cma: Replace global lock in rdma_destroy_id() with id-specific one

rdma_destroy_id currently uses the global rdma cm 'lock' to test if an
rdma_cm_id has been bound to a device.  This prevents an active
address resolution callback handler from assigning a device to the
rdma_cm_id after rdma_destroy_id checks for one.

Instead, we can replace the use of the global lock around the check to
the rdma_cm_id device pointer by setting the id state to destroying,
then flushing all active callbacks.  The latter is accomplished by
acquiring and releasing the handler_mutex.  Any active handler will
complete first, and any newly scheduled handlers will find the
rdma_cm_id in an invalid state.

In addition to optimizing the current locking scheme, the use of the
rdma_cm_id mutex is a more intuitive synchronization mechanism than
that of the global lock.  These changes are based on feedback from
Doug Ledford <dledford@redhat.com> while he was trying to debug a
crash in the rdma cm destroy path.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoIB/cm: Cancel pending LAP message when exiting IB_CM_ESTABLISH state
Sean Hefty [Thu, 3 Mar 2011 23:31:06 +0000 (23:31 +0000)]
IB/cm: Cancel pending LAP message when exiting IB_CM_ESTABLISH state

This problem was reported by Moni Shoua <monis@mellanox.com> and Amir
Vadai <amirv@mellanox.com>:

When destroying a cm_id from a context of a work queue and if
the lap_state of this cm_id is IB_CM_LAP_SENT, we need to
release the reference of this id that was taken upon the send
of the LAP message.  Otherwise, if the expected APR message
gets lost, it is only after a long time that the reference
will be released, while during that the work handler thread is
not available to process other things.

It turns out that we need to cancel any pending LAP messages whenever
we transition out of the IB_CM_ESTABLISH state.  This occurs when
disconnecting - either sending or receiving a DREQ.  It can also
happen in a corner case where we receive a REJ message after sending
an RTU, followed by a LAP.  Add checks and cancel any outstanding LAP
messages in these three cases.

Canceling the LAP when sending a DREQ fixes the destroy problem
reported by Moni.  When a cm_id is destroyed in the IB_CM_ESTABLISHED
state, it sends a DREQ to the remote side to notify the peer that the
connection is going away.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoIB/cm: Bump reference count on cm_id before invoking callback
Sean Hefty [Wed, 23 Feb 2011 16:17:40 +0000 (08:17 -0800)]
IB/cm: Bump reference count on cm_id before invoking callback

When processing a SIDR REQ, the ib_cm allocates a new cm_id.  The
refcount of the cm_id is initialized to 1.  However, cm_process_work
will decrement the refcount after invoking all callbacks.  The result
is that the cm_id will end up with refcount set to 0 by the end of the
sidr req handler.

If a user tries to destroy the cm_id, the destruction will proceed,
under the incorrect assumption that no other threads are referencing
the cm_id.  This can lead to a crash when the cm callback thread tries
to access the cm_id.

This problem was noticed as part of a larger investigation with kernel
crashes in the rdma_cm when running on a real time OS.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Acked-by: Doug Ledford <dledford@redhat.com>
Cc: <stable@kernel.org>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoRDMA/cma: Fix crash in request handlers
Sean Hefty [Wed, 23 Feb 2011 16:11:32 +0000 (08:11 -0800)]
RDMA/cma: Fix crash in request handlers

Doug Ledford and Red Hat reported a crash when running the rdma_cm on
a real-time OS.  The crash has the following call trace:

    cm_process_work
       cma_req_handler
          cma_disable_callback
          rdma_create_id
             kzalloc
             init_completion
          cma_get_net_info
          cma_save_net_info
          cma_any_addr
             cma_zero_addr
          rdma_translate_ip
             rdma_copy_addr
          cma_acquire_dev
             rdma_addr_get_sgid
             ib_find_cached_gid
             cma_attach_to_dev
          ucma_event_handler
             kzalloc
             ib_copy_ah_attr_to_user
          cma_comp

[ preempted ]

    cma_write
        copy_from_user
        ucma_destroy_id
           copy_from_user
           _ucma_find_context
           ucma_put_ctx
           ucma_free_ctx
              rdma_destroy_id
                 cma_exch
                 cma_cancel_operation
                 rdma_node_get_transport

        rt_mutex_slowunlock
        bad_area_nosemaphore
        oops_enter

They were able to reproduce the crash multiple times with the
following details:

    Crash seems to always happen on the:
            mutex_unlock(&conn_id->handler_mutex);
    as conn_id looks to have been freed during this code path.

An examination of the code shows that a race exists in the request
handlers.  When a new connection request is received, the rdma_cm
allocates a new connection identifier.  This identifier has a single
reference count on it.  If a user calls rdma_destroy_id() from another
thread after receiving a callback, rdma_destroy_id will proceed to
destroy the id and free the associated memory.  However, the request
handlers may still be in the process of running.  When control returns
to the request handlers, they can attempt to access the newly created
identifiers.

Fix this by holding a reference on the newly created rdma_cm_id until
the request handler is through accessing it.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Acked-by: Doug Ledford <dledford@redhat.com>
Cc: <stable@kernel.org>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoIB/ipath: Don't reset disabled devices
Nicolas Kaiser [Tue, 26 Oct 2010 15:47:52 +0000 (15:47 +0000)]
IB/ipath: Don't reset disabled devices

The comment some lines above states that disabled devices must not reset.

Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
13 years agoIB/qib: Fix M_Key field in SubnGet and SubnGetResp MADs
Mitko Haralanov [Fri, 4 Mar 2011 18:53:17 +0000 (18:53 +0000)]
IB/qib: Fix M_Key field in SubnGet and SubnGetResp MADs

Set the M_Key field in SubnGet and SugnGetResp MADs based on correctly
interpreting the protection level specified in the M_KeyProtBits field.

Signed-off-by: Mitko Haralanov <mitko@qlogic.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoIB/qib: Set default LE2 value for active cables to 0
Mitko Haralanov [Mon, 28 Feb 2011 13:39:49 +0000 (13:39 +0000)]
IB/qib: Set default LE2 value for active cables to 0

For active and far-EQ cables use an LE2 value of 0 for improved SI.

Signed-off-by: Mitko Haralanov <mitko@qlogic.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoRDMA/cxgb4: Debugfs dump_qp() updates
Steve Wise [Fri, 11 Mar 2011 22:29:50 +0000 (22:29 +0000)]
RDMA/cxgb4: Debugfs dump_qp() updates

- Show whether the SQ is in onchip memory or not.
- Dump both SQ and RQ QIDs.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoRDMA/cxgb4: Dispatch FATAL event on EEH errors
Steve Wise [Fri, 11 Mar 2011 22:30:53 +0000 (22:30 +0000)]
RDMA/cxgb4: Dispatch FATAL event on EEH errors

This at least kicks the user mode applications that are watching for
device events.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoRDMA/cxgb4: Use ULP_MODE_TCPDDP
Steve Wise [Fri, 11 Mar 2011 22:30:21 +0000 (22:30 +0000)]
RDMA/cxgb4: Use ULP_MODE_TCPDDP

Set the ULP mode for initial RDMA connection setup to the proper DDP
mode.  This avoids wasting some HW resources while in streaming mode.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoRDMA/cxgb4: Enable on-chip SQ support by default
Steve Wise [Fri, 11 Mar 2011 22:30:11 +0000 (22:30 +0000)]
RDMA/cxgb4: Enable on-chip SQ support by default

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoRDMA/cxgb4: Do CIDX_INC updates every 1/16 CQ depth CQE reaps
Steve Wise [Fri, 11 Mar 2011 22:30:42 +0000 (22:30 +0000)]
RDMA/cxgb4: Do CIDX_INC updates every 1/16 CQ depth CQE reaps

This avoids the CIDX_INC overflow issue with T4A2 when running
kernel RDMA applications.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoRDMA/cxgb4: Remove db_drop_task
Steve Wise [Fri, 11 Mar 2011 22:30:32 +0000 (22:30 +0000)]
RDMA/cxgb4: Remove db_drop_task

Unloading iw_cxgb4 can crash due to the unload code trying to use
db_drop_task, which is uninitialized.  So remove this dead code.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoRDMA/cxgb4: Turn on delayed ACK
Steve Wise [Fri, 11 Mar 2011 22:30:01 +0000 (22:30 +0000)]
RDMA/cxgb4: Turn on delayed ACK

Set the default to on.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoIB/qib: Return correct MAD when setting link width to 255
Mitko Haralanov [Wed, 23 Feb 2011 00:56:37 +0000 (16:56 -0800)]
IB/qib: Return correct MAD when setting link width to 255

Fix a bug which causes the driver to return incorrect MADs as a
response to Set(PortInfo) which sets the link width to 0xFF or link
speed to 0xF.

Signed-off-by: Mitko Haralanov <mitko@qlogic.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoMerge branch 'for-linus/bugfixes' of git://xenbits.xen.org/people/ianc/linux-2.6
Linus Torvalds [Fri, 18 Feb 2011 20:44:41 +0000 (12:44 -0800)]
Merge branch 'for-linus/bugfixes' of git://xenbits.xen.org/people/ianc/linux-2.6

* 'for-linus/bugfixes' of git://xenbits.xen.org/people/ianc/linux-2.6:
  xen: suspend and resume system devices when running PVHVM

13 years agoMerge branch 'fixes-2.6.38' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq
Linus Torvalds [Fri, 18 Feb 2011 20:36:06 +0000 (12:36 -0800)]
Merge branch 'fixes-2.6.38' of git://git./linux/kernel/git/tj/wq

* 'fixes-2.6.38' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
  workqueue: make sure MAYDAY_INITIAL_TIMEOUT is at least 2 jiffies long
  workqueue, freezer: unify spelling of 'freeze' + 'able' to 'freezable'
  workqueue: wake up a worker when a rescuer is leaving a gcwq

13 years agoExpand CONFIG_DEBUG_LIST to several other list operations
Linus Torvalds [Fri, 18 Feb 2011 19:32:28 +0000 (11:32 -0800)]
Expand CONFIG_DEBUG_LIST to several other list operations

When list debugging is enabled, we aim to readably show list corruption
errors, and the basic list_add/list_del operations end up having extra
debugging code in them to do some basic validation of the list entries.

However, "list_del_init()" and "list_move[_tail]()" ended up avoiding
the debug code due to how they were written. This fixes that.

So the _next_ time we have list_move() problems with stale list entries,
we'll hopefully have an easier time finding them..

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branch 'pm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/suspe...
Linus Torvalds [Fri, 18 Feb 2011 01:52:36 +0000 (17:52 -0800)]
Merge branch 'pm-fixes' of git://git./linux/kernel/git/rafael/suspend-2.6

* 'pm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/suspend-2.6:
  PM / Hibernate: Return error code when alloc_image_page() fails

13 years agoMerge branch 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Fri, 18 Feb 2011 01:52:17 +0000 (17:52 -0800)]
Merge branch 'drm-fixes' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm/radeon/kms: add missing frac fb div flag for dce4+
  drm/radeon/kms: do not reject X16 and Y16X16 floating-point formats on r300
  drm/nouveau: fix suspend/resume on GPUs that don't have PM support
  drm/nouveau: flips/flipd need to always set 'evict' for move_accel_cleanup()
  drm/nv40: fix tiling-related setup for a number of chipsets
  drm/nouveau: fix non-EDIDful native mode selection
  drm/nouveau: Fix detection of DDC-based LVDS on DCB15 boards.
  drm/nv04-nv40: Fix NULL dereference when we fail to find an LVDS native mode.
  drm/nv10: Fix crash when allocating a BO larger than half the available VRAM.

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Fri, 18 Feb 2011 01:51:52 +0000 (17:51 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IB/qib: Prevent double completions after a timeout or RNR error
  IB/qib: Fix double add_timer()
  RDMA/nes: Don't generate async events for unregistered devices

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Fri, 18 Feb 2011 01:51:27 +0000 (17:51 -0800)]
Merge git://git./linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  sparc64: Fix NMI startup bug which also breaks perf.
  sparc: fix size argument to find_next_zero_bit()
  sparc: use bitmap_set()
  sparc32: unaligned memory access (MNA) trap handler bug

13 years agofs/partitions: Validate map_count in Mac partition tables
Timo Warns [Thu, 17 Feb 2011 21:27:40 +0000 (22:27 +0100)]
fs/partitions: Validate map_count in Mac partition tables

Validate number of blocks in map and remove redundant variable.

Signed-off-by: Timo Warns <warns@pre-sense.de>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branches 'nes' and 'qib' into for-next
Roland Dreier [Thu, 17 Feb 2011 22:04:59 +0000 (14:04 -0800)]
Merge branches 'nes' and 'qib' into for-next

13 years agoIB/qib: Prevent double completions after a timeout or RNR error
Mike Marciniszyn [Wed, 16 Feb 2011 15:48:25 +0000 (15:48 +0000)]
IB/qib: Prevent double completions after a timeout or RNR error

There is a double completion associated with error handling for RC QPs.

The sequence is:

 - The do_rc_ack() routine fields an RNR nack and there are 0
   rnr_retries configured on the QP.
 - qib_error_qp() stops the pending timer
 - qib_rc_send_complete() is called from sdma_complete()
 - qib_rc_send_complete() starts the timer because the msb of the psn
   just completed says an ack is needed.
 - a bunch of flushes occur as ipoib posts WQEs to an error'ed QP
 - rc_timeout() calls qib_restart_rc()
 - qib_restart_rc() calls qib_send_complete() with a
   IB_WC_RETRY_EXC_ERR on a wqe that has already been completed in the
   past

The fix avoids starting the timer since another packet will never
arrive.

Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
13 years agoxen: suspend and resume system devices when running PVHVM
Ian Campbell [Thu, 17 Feb 2011 10:31:20 +0000 (10:31 +0000)]
xen: suspend and resume system devices when running PVHVM

Otherwise we fail to properly suspend/resume all of the emulated devices.

Something between 2.6.38-rc2 and rc3 appears to have exposed this
issue, but it's always been wrong not to do this.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Jeremy Fitzhardinge <jeremy@goop.org>
13 years agoMerge branch 'for-2.6.38' of git://linux-nfs.org/~bfields/linux
Linus Torvalds [Thu, 17 Feb 2011 05:53:41 +0000 (21:53 -0800)]
Merge branch 'for-2.6.38' of git://linux-nfs.org/~bfields/linux

* 'for-2.6.38' of git://linux-nfs.org/~bfields/linux:
  nfsd: correctly handle return value from nfsd_map_name_to_*

13 years agoMerge remote branch 'nouveau/drm-nouveau-next' of /ssd/git/drm-nouveau-next into...
Dave Airlie [Thu, 17 Feb 2011 03:56:35 +0000 (13:56 +1000)]
Merge remote branch 'nouveau/drm-nouveau-next' of /ssd/git/drm-nouveau-next into drm-fixes

* 'nouveau/drm-nouveau-next' of /ssd/git/drm-nouveau-next:
  drm/nouveau: fix suspend/resume on GPUs that don't have PM support
  drm/nouveau: flips/flipd need to always set 'evict' for move_accel_cleanup()
  drm/nv40: fix tiling-related setup for a number of chipsets
  drm/nouveau: fix non-EDIDful native mode selection
  drm/nouveau: Fix detection of DDC-based LVDS on DCB15 boards.
  drm/nv04-nv40: Fix NULL dereference when we fail to find an LVDS native mode.
  drm/nv10: Fix crash when allocating a BO larger than half the available VRAM.

13 years agodrm/radeon/kms: add missing frac fb div flag for dce4+
Alex Deucher [Thu, 17 Feb 2011 02:17:04 +0000 (21:17 -0500)]
drm/radeon/kms: add missing frac fb div flag for dce4+

The fixed ref/post dividers are set by the AdjustPll table
rather than the ss info table on dce4+.  Make sure we enable
the fractional feedback dividers when using a fixed post
or ref divider on them as well.

Fixes:
https://bugzilla.kernel.org/show_bug.cgi?id=29272

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Cc: stable@kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agodrm/radeon/kms: do not reject X16 and Y16X16 floating-point formats on r300
Marek Olšák [Wed, 16 Feb 2011 01:26:08 +0000 (02:26 +0100)]
drm/radeon/kms: do not reject X16 and Y16X16 floating-point formats on r300

Signed-off-by: Marek Olšák <maraeo@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agodrm/nouveau: fix suspend/resume on GPUs that don't have PM support
Ben Skeggs [Thu, 17 Feb 2011 01:11:28 +0000 (11:11 +1000)]
drm/nouveau: fix suspend/resume on GPUs that don't have PM support

This has been broken since 2.6.37, and fixes resume on a couple of fermi
boards I have access to.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
13 years agoMerge branch 'spi/merge' of git://git.secretlab.ca/git/linux-2.6
Linus Torvalds [Thu, 17 Feb 2011 00:48:36 +0000 (16:48 -0800)]
Merge branch 'spi/merge' of git://git.secretlab.ca/git/linux-2.6

* 'spi/merge' of git://git.secretlab.ca/git/linux-2.6:
  spi/pxa2xx pci: fix the release - remove race

13 years agoblock: revert block_dev read-only check
Chuck Ebbert [Wed, 16 Feb 2011 23:11:53 +0000 (18:11 -0500)]
block: revert block_dev read-only check

This reverts commit 75f1dc0d076d ("block: check bdev_read_only() from
blkdev_get()").  That commit added stricter checking to make sure
devices that were being used read-only were actually opened in that
mode.

It turns out that the change breaks a bunch of kernel code that opens
block devices.  Affected systems include dm, md, and the loop device.
Because strict checking for read-only opens of block devices was not
done before this, the code that opens the devices was opening them
read-write even if they were being used read-only.  Auditing all that
code will take time, and new userspace packages for dm, mdadm, etc.
will also be required.

Signed-off-by: Chuck Ebbert <cebbert@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrm/nouveau: flips/flipd need to always set 'evict' for move_accel_cleanup()
Ben Skeggs [Mon, 14 Feb 2011 03:51:28 +0000 (13:51 +1000)]
drm/nouveau: flips/flipd need to always set 'evict' for move_accel_cleanup()

We free the temporary binding before leaving this function, so we also have
to wait for the move to actually complete.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
13 years agodrm/nv40: fix tiling-related setup for a number of chipsets
Ben Skeggs [Mon, 7 Feb 2011 00:49:39 +0000 (10:49 +1000)]
drm/nv40: fix tiling-related setup for a number of chipsets

Due to the default case handling the older chipsets, a bunch of the newer
ones ended up having the wrong tiling regs used.  This commit switches the
default case to handle the newest chipsets.

This also makes nv4e touch the "extra" tiling regs.  "nv" doesn't touch
them for C51 but traces of the NVIDIA binary driver show it being done
there.

I couldn't find NV41/NV45 traces to confirm the behaviour there, but an
educated guess was taken at each of them.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
13 years agodrm/nouveau: fix non-EDIDful native mode selection
Ben Skeggs [Sun, 6 Feb 2011 22:41:18 +0000 (08:41 +1000)]
drm/nouveau: fix non-EDIDful native mode selection

The DRM core fills this value, but at too late a stage for this to work,
possibly resulting in an undesirable mode being selected.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
13 years agodrm/nouveau: Fix detection of DDC-based LVDS on DCB15 boards.
Francisco Jerez [Thu, 3 Feb 2011 00:56:32 +0000 (01:56 +0100)]
drm/nouveau: Fix detection of DDC-based LVDS on DCB15 boards.

Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
13 years agodrm/nv04-nv40: Fix NULL dereference when we fail to find an LVDS native mode.
Francisco Jerez [Thu, 3 Feb 2011 00:53:18 +0000 (01:53 +0100)]
drm/nv04-nv40: Fix NULL dereference when we fail to find an LVDS native mode.

Reported-by: Alex Buell <alex.buell@munted.org.uk>
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
13 years agodrm/nv10: Fix crash when allocating a BO larger than half the available VRAM.
Francisco Jerez [Thu, 3 Feb 2011 00:49:33 +0000 (01:49 +0100)]
drm/nv10: Fix crash when allocating a BO larger than half the available VRAM.

Reported-by: Alex Buell <alex.buell@munted.org.uk>
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
13 years agonfsd: correctly handle return value from nfsd_map_name_to_*
NeilBrown [Wed, 16 Feb 2011 02:08:35 +0000 (13:08 +1100)]
nfsd: correctly handle return value from nfsd_map_name_to_*

These functions return an nfs status, not a host_err.  So don't
try to convert  before returning.

This is a regression introduced by
3c726023402a2f3b28f49b9d90ebf9e71151157d; I fixed up two of the callers,
but missed these two.

Cc: stable@kernel.org
Reported-by: Herbert Poetzl <herbert@13thfloor.at>
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
13 years agoPM / Hibernate: Return error code when alloc_image_page() fails
Stanislaw Gruszka [Sat, 12 Feb 2011 20:06:51 +0000 (21:06 +0100)]
PM / Hibernate: Return error code when alloc_image_page() fails

Currently we return 0 in swsusp_alloc() when alloc_image_page() fails.
Fix that.  Also remove unneeded "error" variable since the only
useful value of error is -ENOMEM.

[rjw: Fixed up the changelog and changed subject.]

Signed-off-by: Stanislaw Gruszka <stf_xl@wp.pl>
Cc: stable@kernel.org
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
13 years agoworkqueue: make sure MAYDAY_INITIAL_TIMEOUT is at least 2 jiffies long
Tejun Heo [Wed, 16 Feb 2011 17:10:19 +0000 (18:10 +0100)]
workqueue: make sure MAYDAY_INITIAL_TIMEOUT is at least 2 jiffies long

MAYDAY_INITIAL_TIMEOUT is defined as HZ / 100 and depending on
configuration may end up 0 or 1.  Even when it's 1, depending on when
the mayday timer is added in the current jiffy interval, it may expire
way before a jiffy has passed.

Make sure MAYDAY_INITIAL_TIMEOUT is at least two to guarantee that at
least a full jiffy has passed before calling rescuers.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Ray Jui <rjui@broadcom.com>
Cc: stable@kernel.org
13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6
Linus Torvalds [Wed, 16 Feb 2011 17:07:00 +0000 (09:07 -0800)]
Merge git://git./linux/kernel/git/jejb/scsi-rc-fixes-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6:
  [SCSI] qla2xxx: Return DID_NO_CONNECT when FC device is lost.
  [SCSI] mptfusion: Bump version 03.04.18
  [SCSI] mptfusion: Fix Incorrect return value in mptscsih_dev_reset
  [SCSI] mptfusion: mptctl_release is required in mptctl.c
  [SCSI] target: fix use after free detected by SLUB poison
  [SCSI] target: Remove procfs based target_core_mib.c code
  [SCSI] target: Fix SCF_SCSI_CONTROL_SG_IO_CDB breakage
  [SCSI] target: Fix top-level configfs_subsystem default_group shutdown breakage
  [SCSI] target: fixed missing lock drop in error path
  [SCSI] target: Fix demo-mode MappedLUN shutdown UA/PR breakage
  [SCSI] target/iblock: Fix failed bd claim NULL pointer dereference
  [SCSI] target: iblock/pscsi claim checking for NULL instead of IS_ERR
  [SCSI] scsi_debug: Fix 32-bit overflow in do_device_access causing memory corruption
  [SCSI] qla2xxx: Change from irq to irqsave with host_lock
  [SCSI] qla2xxx: Fix race that could hang kthread_stop()

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Wed, 16 Feb 2011 17:04:41 +0000 (09:04 -0800)]
Merge git://git./linux/kernel/git/herbert/crypto-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: sha-s390 - Reset index after processing partial block

13 years agovfs: fix BUG_ON() in fs/namei.c:1461
Linus Torvalds [Wed, 16 Feb 2011 16:56:55 +0000 (08:56 -0800)]
vfs: fix BUG_ON() in fs/namei.c:1461

When Al moved the nameidata_dentry_drop_rcu_maybe() call into the
do_follow_link function in commit 844a391799c2 ("nothing in
do_follow_link() is going to see RCU"), he mistakenly left the

BUG_ON(inode != path->dentry->d_inode);

behind.  Which would otherwise be ok, but that BUG_ON() really needs to
be _after_ dropping RCU, since the dentry isn't necessarily stable
otherwise.

So complete the code movement in that commit, and move the BUG_ON() into
do_follow_link() too.  This means that we need to pass in 'inode' as an
argument (just for this one use), but that's a small thing.  And
eventually we may be confident enough in our path lookup that we can
just remove the BUG_ON() and the unnecessary inode argument.

Reported-and-tested-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoworkqueue, freezer: unify spelling of 'freeze' + 'able' to 'freezable'
Tejun Heo [Wed, 16 Feb 2011 08:25:31 +0000 (09:25 +0100)]
workqueue, freezer: unify spelling of 'freeze' + 'able' to 'freezable'

There are two spellings in use for 'freeze' + 'able' - 'freezable' and
'freezeable'.  The former is the more prominent one.  The latter is
mostly used by workqueue and in a few other odd places.  Unify the
spelling to 'freezable'.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Alan Stern <stern@rowland.harvard.edu>
Acked-by: "Rafael J. Wysocki" <rjw@sisk.pl>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Acked-by: Dmitry Torokhov <dtor@mail.ru>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Alex Dubov <oakad@yahoo.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Steven Whitehouse <swhiteho@redhat.com>
13 years agoLinux 2.6.38-rc5
Linus Torvalds [Wed, 16 Feb 2011 03:23:45 +0000 (19:23 -0800)]
Linux 2.6.38-rc5

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu
Linus Torvalds [Wed, 16 Feb 2011 01:51:18 +0000 (17:51 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/gerg/m68knommu

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu:
  m68knommu: set flow handler for secondary interrupt controller of 5249
  m68knommu: remove use of IRQ_FLG_LOCK from 68360 platform support
  m68knommu: fix dereference of port.tty
  m68knommu: add missing linker __modver section
  m68knommu: fix mis-named variable int set_irq_chip loop
  m68knommu: add optimize memmove() function
  m68k: remove arch specific non-optimized memcmp()
  m68knommu: fix use of un-defined _TIF_WORK_MASK
  m68knommu: Rename m548x_wdt.c to m54xx_wdt.c
  m68knommu: fix m548x_wdt.c compilation after headers renaming
  m68knommu: Remove dependencies on nonexistent M68KNOMMU

13 years agom68knommu: set flow handler for secondary interrupt controller of 5249
Greg Ungerer [Wed, 9 Feb 2011 03:43:58 +0000 (13:43 +1000)]
m68knommu: set flow handler for secondary interrupt controller of 5249

The secondary interrupt controller of the ColdFire 5249 code is not
setting the edge triggered flow handler. Set it.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
13 years agom68knommu: remove use of IRQ_FLG_LOCK from 68360 platform support
Greg Ungerer [Tue, 8 Feb 2011 11:40:11 +0000 (21:40 +1000)]
m68knommu: remove use of IRQ_FLG_LOCK from 68360 platform support

The m68knommu arch does not define or use IRQ_FLG_LOCK in its irq
subsystem. Remove obsolete use of it.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
13 years agom68knommu: fix dereference of port.tty
Greg Ungerer [Tue, 8 Feb 2011 11:32:36 +0000 (21:32 +1000)]
m68knommu: fix dereference of port.tty

The struct_tty associated with a port is now a direct pointer
from within the local private driver info struct. So fix all uses
of it.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
13 years agom68knommu: add missing linker __modver section
Greg Ungerer [Tue, 8 Feb 2011 04:45:59 +0000 (14:45 +1000)]
m68knommu: add missing linker __modver section

Add missing linker section __modver to fix:

  LD      vmlinux
/usr/local/bin/../m68k-uclinux/bin/ld.real: error: no memory region specified for loadable section `__modver'

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
13 years agom68knommu: fix mis-named variable int set_irq_chip loop
Greg Ungerer [Tue, 8 Feb 2011 04:40:44 +0000 (14:40 +1000)]
m68knommu: fix mis-named variable int set_irq_chip loop

Compiling for 68360 targets gives:

  CC      arch/m68knommu/platform/68360/ints.o
arch/m68knommu/platform/68360/ints.c: In function ‘init_IRQ’:
arch/m68knommu/platform/68360/ints.c:135:16: error: ‘irq’ undeclared (first use in this function)
arch/m68knommu/platform/68360/ints.c:135:16: note: each undeclared identifier is reported only once for each function it appears in

Fix variable name used.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
13 years agom68knommu: add optimize memmove() function
Greg Ungerer [Thu, 3 Feb 2011 11:58:39 +0000 (21:58 +1000)]
m68knommu: add optimize memmove() function

Add an m68k/coldfire optimized memmove() function for the m68knommu arch.
This is the same function as used by m68k. Simple speed tests show this
is faster once buffers are larger than 4 bytes, and significantly faster
on much larger buffers (4 times faster above about 100 bytes).

This also goes part of the way to fixing a regression caused by commit
ea61bc461d09e8d331a307916530aaae808c72a2 ("m68k/m68knommu: merge MMU and
non-MMU string.h"), which breaks non-coldfire non-mmu builds (which is
the 68x328 and 68360 families). They currently have no memmove() fucntion
defined, since there was none in the m68knommu/lib functions.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
13 years agom68k: remove arch specific non-optimized memcmp()
Greg Ungerer [Thu, 3 Feb 2011 11:31:20 +0000 (21:31 +1000)]
m68k: remove arch specific non-optimized memcmp()

The m68k arch implements its own memcmp() function. It is not optimized
in any way (it is the most strait forward coding of memcmp you can get).
Remove it and use the kernels standard memcmp() implementation.

This also goes part of the way to fixing a regression caused by commit
ea61bc461d09e8d331a307916530aaae808c72a2 ("m68k/m68knommu: merge MMU and
non-MMU string.h"), which breaks non-coldfire non-mmu builds (which is
the 68x328 and 68360 families). They currently have no memcmp() function
defined, since there is none in the m68knommu/lib functions.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>
13 years agoMerge branch 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Tue, 15 Feb 2011 23:25:33 +0000 (15:25 -0800)]
Merge branch 'drm-fixes' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (27 commits)
  drm/radeon/kms: hopefully fix pll issues for real (v3)
  drm/radeon/kms: add bounds checking to avivo pll algo
  drm: fix wrong usages of drm_device in DRM Developer's Guide
  drm/radeon/kms: fix a few more atombios endian issues
  drm/radeon/kms: improve 6xx/7xx CS error output
  drm/radeon/kms: check AA resolve registers on r300
  drm/radeon/kms: fix tracking of BLENDCNTL, COLOR_CHANNEL_MASK, and GB_Z on r300
  drm/radeon/kms: use linear aligned for evergreen/ni bo blits
  drm/radeon/kms: use linear aligned for 6xx/7xx bo blits
  drm/radeon: fix race between GPU reset and TTM delayed delete thread.
  drm/radeon/kms: evergreen/ni big endian fixes (v2)
  drm/radeon/kms: 6xx/7xx big endian fixes
  drm/radeon/kms: atombios big endian fixes
  drm/radeon: 6xx/7xx non-kms endian fixes
  drm/radeon/kms: optimize CS state checking for r100->r500
  drm: do not leak kernel addresses via /proc/dri/*/vma
  drm/radeon/kms: add connector table for mac g5 9600
  radeon mkregtable: Add missing fclose() calls
  drm/radeon/kms: fix interlaced modes on dce4+
  drm/radeon: fix memory debugging since d961db75ce86a84f1f04e91ad1014653ed7d9f46
  ...

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Tue, 15 Feb 2011 23:25:11 +0000 (15:25 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jmorris/security-testing-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/security-testing-2.6:
  pci: use security_capable() when checking capablities during config space read

13 years agothp: prevent hugepages during args/env copying into the user stack
Andrea Arcangeli [Tue, 15 Feb 2011 18:02:45 +0000 (19:02 +0100)]
thp: prevent hugepages during args/env copying into the user stack

Transparent hugepages can only be created if rmap is fully
functional. So we must prevent hugepages to be created while
is_vma_temporary_stack() is true.

This also optmizes away some harmless but unnecessary setting of
khugepaged_scan.address and it switches some BUG_ON to VM_BUG_ON.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Tue, 15 Feb 2011 23:19:45 +0000 (15:19 -0800)]
Merge branch 'release' of git://git./linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6:
  ACPI / Video: Probe for output switch method when searching video devices.
  ACPI / Wakeup: Enable button GPEs unconditionally during initialization
  ACPI / ACPICA: Avoid crashing if _PRW is defined for the root object
  ACPI: Fix acpi_os_read_memory() and acpi_os_write_memory() (v2)

13 years agoMerge branch 'tools-release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb...
Linus Torvalds [Tue, 15 Feb 2011 23:19:22 +0000 (15:19 -0800)]
Merge branch 'tools-release' of git://git./linux/kernel/git/lenb/linux-idle-2.6

* 'tools-release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-idle-2.6:
  tools: turbostat: style updates
  tools: turbostat: fix bitwise and operand

13 years agosparc64: Fix NMI startup bug which also breaks perf.
David S. Miller [Tue, 15 Feb 2011 23:04:07 +0000 (15:04 -0800)]
sparc64: Fix NMI startup bug which also breaks perf.

Doing NMI startup as an early initcall doesn't work because we need
to have SMP started up by then.

So we'd only NMI startup one cpu, which causes perf PMU grab to
BUG because the nmi_active count isn't what it's supposed to be.

This also points out that we don't have proper CPU up/down notifiers
for the NMI code which will need to be fixed at some point.

Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoMerge branch 'iomem' into release
Len Brown [Tue, 15 Feb 2011 20:25:48 +0000 (15:25 -0500)]
Merge branch 'iomem' into release

13 years agospi/pxa2xx pci: fix the release - remove race
Sebastian Andrzej Siewior [Wed, 2 Feb 2011 19:01:21 +0000 (00:31 +0530)]
spi/pxa2xx pci: fix the release - remove race

Right now the platform device and its platform data is included in one big
struct which requires its custom ->release function. The problem with the
release function within the driver is that it might be called after the
driver was removed because someone was holding a reference to it and it
was not called right after platform_device_unregister(). So we also free
the platform device memory to which one might hold a reference.

This patch uses the normal pdev functions so this kind of race does not
occur.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
13 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx
Linus Torvalds [Tue, 15 Feb 2011 20:07:35 +0000 (12:07 -0800)]
Merge branch 'fixes' of git://git./linux/kernel/git/djbw/async_tx

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx: (21 commits)
  dmaengine: add slave-dma maintainer
  dma: ipu_idmac: do not lose valid received data in the irq handler
  dmaengine: imx-sdma: fix up param for the last BD in sdma_prep_slave_sg()
  dmaengine: imx-sdma: correct sdmac->status in sdma_handle_channel_loop()
  dmaengine: imx-sdma: return sdmac->status in sdma_tx_status()
  dmaengine: imx-sdma: set sdmac->status to DMA_ERROR in err_out of sdma_prep_slave_sg()
  dmaengine: imx-sdma: remove IMX_DMA_SG_LOOP handling in sdma_prep_slave_sg()
  dmaengine i.MX dma: initialize dma capabilities outside channel loop
  dmaengine i.MX DMA: do not initialize chan_id field
  dmaengine i.MX dma: check sg entries for valid addresses and lengths
  dmaengine i.MX dma: set maximum segment size for our device
  dmaengine i.MX SDMA: reserve channel 0 by not registering it
  dmaengine i.MX SDMA: initialize dma capabilities outside channel loop
  dmaengine i.MX SDMA: do not initialize chan_id field
  dmaengine i.MX sdma: check sg entries for valid addresses and lengths
  dmaengine i.MX sdma: set maximum segment size for our device
  DMA: PL08x: fix channel pausing to timeout rather than lockup
  DMA: PL08x: fix infinite wait when terminating transfers
  dmaengine: imx-sdma: fix inconsistent naming in sdma_assign_cookie()
  dmaengine: imx-sdma: propagate error in sdma_probe() instead of returning 0
  ...

13 years agoMerge branch 'for-2.6.38' of git://linux-nfs.org/~bfields/linux
Linus Torvalds [Tue, 15 Feb 2011 20:06:38 +0000 (12:06 -0800)]
Merge branch 'for-2.6.38' of git://linux-nfs.org/~bfields/linux

* 'for-2.6.38' of git://linux-nfs.org/~bfields/linux:
  nfsd: break lease on unlink due to rename
  nfsd4: acquire only one lease per file
  nfsd4: modify fi_delegations under recall_lock
  nfsd4: remove unused deleg dprintk's.
  nfsd4: split lease setting into separate function
  nfsd4: fix leak on allocation error
  nfsd4: add helper function for lease setup
  nfsd4: split up nfsd_break_deleg_cb
  NFSD: memory corruption due to writing beyond the stat array
  NFSD: use nfserr for status after decode_cb_op_status
  nfsd: don't leak dentry count on mnt_want_write failure

13 years agoMerge branches 'core-fixes-for-linus' and 'timers-fixes-for-linus' of git://git.kerne...
Linus Torvalds [Tue, 15 Feb 2011 18:19:18 +0000 (10:19 -0800)]
Merge branches 'core-fixes-for-linus' and 'timers-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  Revert "lockdep, timer: Fix del_timer_sync() annotation"

* 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  timer debug: Hide kernel addresses via %pK in /proc/timer_list

13 years agoMerge branch 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 15 Feb 2011 18:18:48 +0000 (10:18 -0800)]
Merge branch 'perf-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: Fix text_poke_smp_batch() deadlock
  perf tools: Fix thread_map event synthesizing in top and record
  watchdog, nmi: Lower the severity of error messages
  ARM: oprofile: Fix backtraces in timer mode
  oprofile: Fix usage of CONFIG_HW_PERF_EVENTS for oprofile_perf_init and friends

13 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 15 Feb 2011 18:18:29 +0000 (10:18 -0800)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86, dmi, debug: Log board name (when present) in dmesg/oops output
  x86, ioapic: Don't warn about non-existing IOAPICs if we have none
  x86: Fix mwait_usable section mismatch
  x86: Readd missing irq_to_desc() in fixup_irq()
  x86: Fix section mismatch in LAPIC initialization

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Tue, 15 Feb 2011 17:40:27 +0000 (09:40 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: matrix_keypad - increase the limit of rows and columns
  Input: wacom - fix error path in wacom_probe()
  Input: ads7846 - check proper condition when freeing gpio
  Revert "Input: do not pass injected events back to the originating handler"
  Input: sysrq - rework re-inject logic
  Input: serio - clear pending rescans after sysfs driver rebind
  Input: rotary_encoder - use proper irqflags
  Input: wacom_w8001 - report resolution to userland

13 years ago[SCSI] qla2xxx: Return DID_NO_CONNECT when FC device is lost.
Andrew Vasquez [Fri, 28 Jan 2011 23:17:55 +0000 (15:17 -0800)]
[SCSI] qla2xxx: Return DID_NO_CONNECT when FC device is lost.

If the target device gets lost, this fix is needed, as it causes
negative unintended responses on basic I/O tests. If the target device
gets lost, the upstream qla2xxx driver returns
SCSI_MLQUEUE_TARGET_BUSY which causes an immediate retry without drop
in the number of allowed retries. This semantic change, as a result of
removing FC_DEVICE_LOST check is reasonable, as it only extends a
short transitional period, until the transport is called to notify
that the rport as lost (fc_remote_port_delete()). Once transport
notification is done, fc_remote_port_chkready() check will take over.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: Madhuranath Iyengar <Madhu.Iyengar@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6
Linus Torvalds [Tue, 15 Feb 2011 16:06:36 +0000 (08:06 -0800)]
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:
  get rid of nameidata_dentry_drop_rcu() calling nameidata_drop_rcu()
  drop out of RCU in return_reval
  split do_revalidate() into RCU and non-RCU cases
  in do_lookup() split RCU and non-RCU cases of need_revalidate
  nothing in do_follow_link() is going to see RCU

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable
Linus Torvalds [Tue, 15 Feb 2011 16:00:35 +0000 (08:00 -0800)]
Merge git://git./linux/kernel/git/mason/btrfs-unstable

* git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable:
  Btrfs: check return value of alloc_extent_map()
  Btrfs - Fix memory leak in btrfs_init_new_device()
  btrfs: prevent heap corruption in btrfs_ioctl_space_info()
  Btrfs: Fix balance panic
  Btrfs: don't release pages when we can't clear the uptodate bits
  Btrfs: fix page->private races

13 years agos390: remove task_show_regs
Martin Schwidefsky [Tue, 15 Feb 2011 08:43:32 +0000 (09:43 +0100)]
s390: remove task_show_regs

task_show_regs used to be a debugging aid in the early bringup days
of Linux on s390. /proc/<pid>/status is a world readable file, it
is not a good idea to show the registers of a process. The only
correct fix is to remove task_show_regs.

Reported-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agopci: use security_capable() when checking capablities during config space read
Chris Wright [Tue, 15 Feb 2011 01:21:49 +0000 (17:21 -0800)]
pci: use security_capable() when checking capablities during config space read

This reintroduces commit 47970b1b which was subsequently reverted
as f00eaeea.  The original change was broken and caused X startup
failures and generally made privileged processes incapable of reading
device dependent config space.  The normal capable() interface returns
true on success, but the LSM interface returns 0 on success.  This thinko
is now fixed in this patch, and has been confirmed to work properly.

So, once again...Eric Paris noted that commit de139a3 ("pci: check caps
from sysfs file open to read device dependent config space") caused the
capability check to bypass security modules and potentially auditing.
Rectify this by calling security_capable() when checking the open file's
capabilities for config space reads.

Reported-by: Eric Paris <eparis@redhat.com>
Tested-by: Dave Young <hidave.darkstar@gmail.com>
Acked-by: James Morris <jmorris@namei.org>
Cc: Dave Airlie <airlied@gmail.com>
Cc: Alex Riesen <raa.lkml@gmail.com>
Cc: Sedat Dilek <sedat.dilek@googlemail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: James Morris <jmorris@namei.org>
13 years agoget rid of nameidata_dentry_drop_rcu() calling nameidata_drop_rcu()
Al Viro [Tue, 15 Feb 2011 06:42:59 +0000 (01:42 -0500)]
get rid of nameidata_dentry_drop_rcu() calling nameidata_drop_rcu()

can't happen anymore and didn't work right anyway

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
13 years agodrop out of RCU in return_reval
Al Viro [Tue, 15 Feb 2011 06:35:28 +0000 (01:35 -0500)]
drop out of RCU in return_reval

... thus killing the need to handle drop-from-RCU in d_revalidate()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
13 years agosplit do_revalidate() into RCU and non-RCU cases
Al Viro [Tue, 15 Feb 2011 06:32:55 +0000 (01:32 -0500)]
split do_revalidate() into RCU and non-RCU cases

fixing oopsen in lookup_one_len()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
13 years agoin do_lookup() split RCU and non-RCU cases of need_revalidate
Al Viro [Tue, 15 Feb 2011 06:26:22 +0000 (01:26 -0500)]
in do_lookup() split RCU and non-RCU cases of need_revalidate

and use unlikely() instead of gotos, for fsck sake...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
13 years agonothing in do_follow_link() is going to see RCU
Al Viro [Tue, 15 Feb 2011 05:38:26 +0000 (00:38 -0500)]
nothing in do_follow_link() is going to see RCU

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
13 years agox86, dmi, debug: Log board name (when present) in dmesg/oops output
Naga Chumbalkar [Mon, 14 Feb 2011 22:47:17 +0000 (22:47 +0000)]
x86, dmi, debug: Log board name (when present) in dmesg/oops output

The "Type 2" SMBIOS record that contains Board Name is not
strictly required and may be absent in the SMBIOS on some
platforms.

( Please note that Type 2 is not listed in Table 3 in Sec 6.2
  ("Required Structures and Data") of the SMBIOS v2.7
  Specification. )

Use the Manufacturer Name (aka System Vendor) name.
Print Board Name only when it is present.

Before the fix:
  (i) dmesg output: DMI: /ProLiant DL380 G6, BIOS P62 01/29/2011
 (ii) oops output:  Pid: 2170, comm: bash Not tainted 2.6.38-rc4+ #3 /ProLiant DL380 G6

After the fix:
  (i) dmesg output: DMI: HP ProLiant DL380 G6, BIOS P62 01/29/2011
 (ii) oops output:  Pid: 2278, comm: bash Not tainted 2.6.38-rc4+ #4 HP ProLiant DL380 G6

Signed-off-by: Naga Chumbalkar <nagananda.chumbalkar@hp.com>
Reviewed-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Cc: <stable@kernel.org> # .3x - good for debugging, please apply as far back as it applies cleanly
LKML-Reference: <20110214224423.2182.13929.sendpatchset@nchumbalkar.americas.hpqcorp.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
13 years agox86, ioapic: Don't warn about non-existing IOAPICs if we have none
Paul Bolle [Mon, 14 Feb 2011 21:52:38 +0000 (22:52 +0100)]
x86, ioapic: Don't warn about non-existing IOAPICs if we have none

mp_find_ioapic() prints errors like:

    ERROR: Unable to locate IOAPIC for GSI 13

if it can't find the IOAPIC that manages that specific GSI. I
see errors like that at every boot of a laptop that apparently
doesn't have any IOAPICs.

But if there are no IOAPICs it doesn't seem to be an error that
none can be found. A solution that gets rid of this message is
to directly return if nr_ioapics (still) is zero. (But keep
returning -1 in that case, so nothing breaks from this change.)

The call chain that generates this error is:

pnpacpi_allocated_resource()
    case ACPI_RESOURCE_TYPE_IRQ:
        pnpacpi_parse_allocated_irqresource()
            acpi_get_override_irq()
                 mp_find_ioapic()

Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
13 years agoMerge branch 'urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/rric/oprofile...
Ingo Molnar [Tue, 15 Feb 2011 03:10:35 +0000 (04:10 +0100)]
Merge branch 'urgent' of git://git./linux/kernel/git/rric/oprofile into perf/urgent

13 years agodrm/radeon/kms: hopefully fix pll issues for real (v3)
Alex Deucher [Mon, 14 Feb 2011 16:43:11 +0000 (11:43 -0500)]
drm/radeon/kms: hopefully fix pll issues for real (v3)

The problematic boards have a recommended reference divider
to be used when spread spectrum is enabled on the laptop panel.
Enable the use of the recommended reference divider along with
the new pll algo.

v2: testing options

v3: When using the fixed reference divider with LVDS, prefer
min m to max p and use fractional feedback dividers.

Fixes:
https://bugzilla.kernel.org/show_bug.cgi?id=28852
https://bugzilla.kernel.org/show_bug.cgi?id=24462
https://bugzilla.kernel.org/show_bug.cgi?id=26552
MacbookPro issues reported by Justin Mattock <justinmattock@gmail.com>

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Cc: stable@kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agodrm/radeon/kms: add bounds checking to avivo pll algo
Alex Deucher [Mon, 14 Feb 2011 16:43:10 +0000 (11:43 -0500)]
drm/radeon/kms: add bounds checking to avivo pll algo

Prevent divider overflow.
Fixes:
https://bugzilla.kernel.org/show_bug.cgi?id=28932

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Cc: stable@kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agodrm: fix wrong usages of drm_device in DRM Developer's Guide
Xiao Jiang [Mon, 14 Feb 2011 09:58:23 +0000 (17:58 +0800)]
drm: fix wrong usages of drm_device in DRM Developer's Guide

A few wrong usages of drm_device, which should be drm_driver.

Signed-off-by: Xiao Jiang <jgq516@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agodrm/radeon/kms: fix a few more atombios endian issues
Alex Deucher [Mon, 14 Feb 2011 21:16:22 +0000 (16:16 -0500)]
drm/radeon/kms: fix a few more atombios endian issues

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agoMerge branch 'fixes' of master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Mon, 14 Feb 2011 22:49:29 +0000 (14:49 -0800)]
Merge branch 'fixes' of /home/rmk/linux-2.6-arm

* 'fixes' of master.kernel.org:/home/rmk/linux-2.6-arm:
  ARM: 6657/1: hw_breakpoint: fix ptrace breakpoint advertising on unsupported arch
  ARM: 6656/1: hw_breakpoint: avoid UNPREDICTABLE behaviour when reading DBGDSCR
  ARM: 6658/1: collie: do actually pass locomo_info to locomo driver
  ARM: 6659/1: Thumb-2: Make CONFIG_OABI_COMPAT depend on !CONFIG_THUMB2_KERNEL
  ARM: 6654/1: perf/oprofile: fix off-by-one in stack check
  ARM: fixup SMP alternatives in modules
  ARM: make SWP emulation explicit on !CPU_USE_DOMAINS
  ARM: Avoid building unsafe kernels on OMAP2 and MX3
  ARM: pxa: Properly configure PWM period for palm27x
  ARM: pxa: only save/restore registers when pm functions are defined
  ARM: pxa/colibri: use correct SD detect pin
  ARM: pxa: fix mfpr_sync to read from valid offset

13 years agoBtrfs: check return value of alloc_extent_map()
Tsutomu Itoh [Mon, 14 Feb 2011 00:45:29 +0000 (00:45 +0000)]
Btrfs: check return value of alloc_extent_map()

I add the check on the return value of alloc_extent_map() to several places.
In addition, alloc_extent_map() returns only the address or NULL.
Therefore, check by IS_ERR() is unnecessary. So, I remove IS_ERR() checking.

Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs - Fix memory leak in btrfs_init_new_device()
Ilya Dryomov [Sun, 6 Feb 2011 19:58:21 +0000 (19:58 +0000)]
Btrfs - Fix memory leak in btrfs_init_new_device()

Memory allocated by calling kstrdup() should be freed.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agobtrfs: prevent heap corruption in btrfs_ioctl_space_info()
Dan Rosenberg [Mon, 14 Feb 2011 21:04:23 +0000 (16:04 -0500)]
btrfs: prevent heap corruption in btrfs_ioctl_space_info()

Commit bf5fc093c5b625e4259203f1cee7ca73488a5620 refactored
btrfs_ioctl_space_info() and introduced several security issues.

space_args.space_slots is an unsigned 64-bit type controlled by a
possibly unprivileged caller.  The comparison as a signed int type
allows providing values that are treated as negative and cause the
subsequent allocation size calculation to wrap, or be truncated to 0.
By providing a size that's truncated to 0, kmalloc() will return
ZERO_SIZE_PTR.  It's also possible to provide a value smaller than the
slot count.  The subsequent loop ignores the allocation size when
copying data in, resulting in a heap overflow or write to ZERO_SIZE_PTR.

The fix changes the slot count type and comparison typecast to u64,
which prevents truncation or signedness errors, and also ensures that we
don't copy more data than we've allocated in the subsequent loop.  Note
that zero-size allocations are no longer possible since there is already
an explicit check for space_args.space_slots being 0 and truncation of
this value is no longer an issue.

Signed-off-by: Dan Rosenberg <drosenberg@vsecurity.com>
Signed-off-by: Josef Bacik <josef@redhat.com>
Reviewed-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: Fix balance panic
Yan, Zheng [Mon, 14 Feb 2011 21:00:03 +0000 (16:00 -0500)]
Btrfs: Fix balance panic

Mark the cloned backref_node as checked in clone_backref_node()

Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoMerge branch 'devicetree/merge' of git://git.secretlab.ca/git/linux-2.6
Linus Torvalds [Mon, 14 Feb 2011 18:10:37 +0000 (10:10 -0800)]
Merge branch 'devicetree/merge' of git://git.secretlab.ca/git/linux-2.6

* 'devicetree/merge' of git://git.secretlab.ca/git/linux-2.6:
  Revert "dt: add documentation of ARM dt boot interface"

13 years agoMerge branch 'rtc-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip...
Linus Torvalds [Mon, 14 Feb 2011 18:10:07 +0000 (10:10 -0800)]
Merge branch 'rtc-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'rtc-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  RTC: Fix minor compile warning
  RTC: Convert rtc drivers to use the alarm_irq_enable method
  RTC: Fix rtc driver ioctl specific shortcutting

13 years agoBtrfs: don't release pages when we can't clear the uptodate bits
Chris Mason [Mon, 14 Feb 2011 17:52:08 +0000 (12:52 -0500)]
Btrfs: don't release pages when we can't clear the uptodate bits

Btrfs tracks uptodate state in an rbtree as well as in the
page bits.  This is supposed to enable us to use block sizes other than
the page size, but there are a few parts still missing before that
completely works.

But, our readpage routine trusts this additional range based tracking
of uptodateness, much in the same way the buffer head up to date bits
are trusted for the other filesystems.

The problem is that sometimes we need to allocate memory in order to
split records in the rbtree, even when we are just clearing bits.  This
can be difficult when our clearing function is called GFP_ATOMIC, which
can happen in the releasepage path.

So, what happens today looks like this:

releasepage called with GFP_ATOMIC
btrfs_releasepage calls clear_extent_bit
clear_extent_bit fails to allocate ram, leaving the up to date bit set
btrfs_releasepage returns success

The end result is the page being gone, but btrfs thinking the range is
up to date.   Later on if someone tries to read that same page, the
btrfs readpage code will return immediately thinking the page is already
up to date.

This commit fixes things to fail the releasepage when we can't clear the
extent state bits.  It covers both data pages and metadata tree blocks.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: fix page->private races
Chris Mason [Thu, 10 Feb 2011 17:35:00 +0000 (12:35 -0500)]
Btrfs: fix page->private races

There is a race where btrfs_releasepage can drop the
page->private contents just as alloc_extent_buffer is setting
up pages for metadata.  Because of how the Btrfs page flags work,
this results in us skipping the crc on the page during IO.

This patch sovles the race by waiting until after the extent buffer
is inserted into the radix tree before it sets page private.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agonfsd: break lease on unlink due to rename
J. Bruce Fields [Sun, 6 Feb 2011 21:46:30 +0000 (16:46 -0500)]
nfsd: break lease on unlink due to rename

4795bb37effb7b8fe77e2d2034545d062d3788a8 "nfsd: break lease on unlink,
link, and rename", only broke the lease on the file that was being
renamed, and didn't handle the case where the target path refers to an
already-existing file that will be unlinked by a rename--in that case
the target file should have any leases broken as well.

Signed-off-by: J. Bruce Fields <bfields@redhat.com>
13 years agonfsd4: acquire only one lease per file
J. Bruce Fields [Tue, 1 Feb 2011 00:20:39 +0000 (19:20 -0500)]
nfsd4: acquire only one lease per file

Instead of acquiring one lease each time another client opens a file,
nfsd can acquire just one lease to represent all of them, and reference
count it to determine when to release it.

This fixes a regression introduced by
c45821d263a8a5109d69a9e8942b8d65bcd5f31a "locks: eliminate fl_mylease
callback": after that patch, only the struct file * is used to determine
who owns a given lease.  But since we recently converted the server to
share a single struct file per open, if we acquire multiple leases on
the same file from nfsd, it then becomes impossible on unlocking a lease
to determine which of those leases (all of whom share the same struct
file *) we meant to remove.

Thanks to Takashi Iwai <tiwai@suse.de> for catching a bug in a previous
version of this patch.

Tested-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
13 years agonfsd4: modify fi_delegations under recall_lock
J. Bruce Fields [Mon, 7 Feb 2011 21:53:46 +0000 (16:53 -0500)]
nfsd4: modify fi_delegations under recall_lock

Modify fi_delegations only under the recall_lock, allowing us to use
that list on lease breaks.

Also some trivial cleanup to simplify later changes.

Signed-off-by: J. Bruce Fields <bfields@redhat.com>
13 years agonfsd4: remove unused deleg dprintk's.
J. Bruce Fields [Mon, 7 Feb 2011 20:44:12 +0000 (15:44 -0500)]
nfsd4: remove unused deleg dprintk's.

These aren't all that useful, and get in the way of the next steps.

Signed-off-by: J. Bruce Fields <bfields@redhat.com>