openwrt/staging/blogic.git
14 years agoMerge branch 'nes' into for-next
Roland Dreier [Tue, 2 Mar 2010 07:51:57 +0000 (23:51 -0800)]
Merge branch 'nes' into for-next

14 years agoMerge branch 'mlx4' into for-next
Roland Dreier [Tue, 2 Mar 2010 07:51:56 +0000 (23:51 -0800)]
Merge branch 'mlx4' into for-next

14 years agoMerge branch 'iser' into for-next
Roland Dreier [Tue, 2 Mar 2010 07:51:56 +0000 (23:51 -0800)]
Merge branch 'iser' into for-next

14 years agoMerge branch 'ipoib' into for-next
Roland Dreier [Tue, 2 Mar 2010 07:51:56 +0000 (23:51 -0800)]
Merge branch 'ipoib' into for-next

14 years agoMerge branch 'ehca' into for-next
Roland Dreier [Tue, 2 Mar 2010 07:51:55 +0000 (23:51 -0800)]
Merge branch 'ehca' into for-next

14 years agoMerge branch 'cxgb3' into for-next
Roland Dreier [Tue, 2 Mar 2010 07:51:55 +0000 (23:51 -0800)]
Merge branch 'cxgb3' into for-next

14 years agoMerge branch 'cma' into for-next
Roland Dreier [Tue, 2 Mar 2010 07:51:54 +0000 (23:51 -0800)]
Merge branch 'cma' into for-next

14 years agoRDMA/nes: Add support for KR device id 0x0110
Chien Tung [Tue, 23 Feb 2010 17:52:10 +0000 (17:52 +0000)]
RDMA/nes: Add support for KR device id 0x0110

Add support for KR device id 0x0110.  While at it, cleanup
nes_init_phy() by splitting it into nes_init_1g_phy() and
nes_init_2025_phy().

Remove support for NES_PHY_TYPE_IRIS, which was used on an XFP board
that was only manufactured in small quantities and given out for evals
in even smaller quantities.

Signed-off-by: Chien Tung <chien.tin.tung@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoRDMA/cxgb3: Mark RDMA device with CXIO_ERROR_FATAL when removing
Steve Wise [Mon, 22 Feb 2010 22:07:22 +0000 (22:07 +0000)]
RDMA/cxgb3: Mark RDMA device with CXIO_ERROR_FATAL when removing

If cxgb3 calls the iw_cxgb3 t3cclient remove function due to a device
removal event, then the iwch device must be marked with CXIO_ERROR_FATAL
since the device below us is going away.  Otherwise, we can get stuck in
a deadlock as RDMA ULPs try and deallocate objects (like MRs, QPs, etc).
So always mark the device with CXIO_ERROR_FATAL when removing.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoRDMA/cxgb3: Don't allocate the SW queue for user mode CQs
Steve Wise [Wed, 27 Jan 2010 20:22:34 +0000 (20:22 +0000)]
RDMA/cxgb3: Don't allocate the SW queue for user mode CQs

Only kernel mode CQs need the SW queue memory allocated.  The SW queue
for user mode CQs is allocated in userspace by libcxgb3.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoRDMA/cxgb3: Increase the max CQ depth
Steve Wise [Fri, 29 Jan 2010 00:55:49 +0000 (00:55 +0000)]
RDMA/cxgb3: Increase the max CQ depth

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoRDMA/cxgb3: Doorbell overflow avoidance and recovery
Steve Wise [Wed, 27 Jan 2010 17:03:34 +0000 (17:03 +0000)]
RDMA/cxgb3: Doorbell overflow avoidance and recovery

T3 hardware doorbell FIFO overflows can cause application stalls due
to lost doorbell ring events.  This has been seen when running large
NP IMB alltoall MPI jobs.  The T3 hardware supports an xon/xoff-type
flow control mechanism to help avoid overflowing the HW doorbell FIFO.

This patch uses these interrupts to disable RDMA QP doorbell rings
when we near an overflow condition, and then turn them back on (and
ring all the active QP doorbells) when when the doorbell FIFO empties
out.  In addition if an doorbell ring is dropped by the hardware, the
code will now recover.

Design:

cxgb3:
- enable these DB interrupts
- in the interrupt handler, schedule work tasks to call the ULPs event
  handlers with the new events.
- ring all the qset txqs when an overflow is detected.

iw_cxgb3:
- disable db ringing on all active qps when we get the DB_FULL event
- enable db ringing on all active qps and ring all active dbs when we get
  the DB_EMPTY event
- On DB_DROP event:
       - disable db rings in the event handler
       - delay-schedule a work task which rings and enables the dbs on
         all active qps.
- in post_send and post_recv logic, don't ring the db if it's disabled.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/core: Pack struct ib_device a little tighter
Alexander Chiang [Tue, 2 Feb 2010 19:09:16 +0000 (19:09 +0000)]
IB/core: Pack struct ib_device a little tighter

A small change to reduce the size of ib_device to 1112 bytes
(from 1128).

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/ucm: Clean whitespace errors
Alexander Chiang [Tue, 2 Feb 2010 19:09:11 +0000 (19:09 +0000)]
IB/ucm: Clean whitespace errors

As shown when 'let c_space_errors=1' is set in vim.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/ucm: Increase maximum devices supported
Alexander Chiang [Tue, 2 Feb 2010 19:09:06 +0000 (19:09 +0000)]
IB/ucm: Increase maximum devices supported

Some large systems may support more than IB_UCM_MAX_DEVICES
(currently 32).

This change allows us to support more devices in a backwards-compatible
manner. the first IB_UCM_MAX_DEVICES keep the same major/minor device
numbers they've always had.

If there are more than IB_UCM_MAX_DEVICES, then we dynamically request
a new major device number (new minors start at 0).

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/ucm: Use stack variable 'base' in ib_ucm_add_one
Alexander Chiang [Tue, 2 Feb 2010 19:09:00 +0000 (19:09 +0000)]
IB/ucm: Use stack variable 'base' in ib_ucm_add_one

This change is not useful by itself, but sets us up for a future
change that allows us to support more than IB_UCM_MAX_DEVICES.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/ucm: Use stack variable 'devnum' in ib_ucm_add_one
Alexander Chiang [Tue, 2 Feb 2010 19:08:55 +0000 (19:08 +0000)]
IB/ucm: Use stack variable 'devnum' in ib_ucm_add_one

This change is not useful by itself, but sets us up for a future
change that allows us to dynamically allocate device numbers in case
we have more than IB_UCM_MAX_DEVICES in the system.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/umad: Clean whitespace
Alexander Chiang [Tue, 2 Feb 2010 19:08:50 +0000 (19:08 +0000)]
IB/umad: Clean whitespace

Clean errors as shown when 'let c_space_errors=1' is set in vim.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/umad: Increase maximum devices supported
Alexander Chiang [Tue, 2 Feb 2010 19:08:45 +0000 (19:08 +0000)]
IB/umad: Increase maximum devices supported

Some large systems may support more than IB_UMAD_MAX_PORTS
(currently 64).

This change allows us to support more ports in a backwards-compatible
manner.  The first IB_UMAD_MAX_PORTS keep the same major/minor device
numbers they've always had.

If there are more than IB_UMAD_MAX_PORTS, we then dynamically request
a new major device number (new minors start at 0).

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/umad: Use stack variable 'base' in ib_umad_init_port
Alexander Chiang [Tue, 2 Feb 2010 19:08:40 +0000 (19:08 +0000)]
IB/umad: Use stack variable 'base' in ib_umad_init_port

This change is not useful by itself, but sets us up for a future change
that allows us to support more than IB_UMAD_MAX_PORTS in a system.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/umad: Use stack variable 'devnum' in ib_umad_init_port
Alexander Chiang [Tue, 2 Feb 2010 19:08:35 +0000 (19:08 +0000)]
IB/umad: Use stack variable 'devnum' in ib_umad_init_port

This change is not useful by itself, but sets us up for a future
change that allows us to dynamically allocate device numbers in case
we have more than IB_UMAD_MAX_PORTS in the system.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/umad: Remove port_table[]
Alexander Chiang [Tue, 2 Feb 2010 19:08:30 +0000 (19:08 +0000)]
IB/umad: Remove port_table[]

We no longer need this data structure, as it was used to associate an
inode back to a struct ib_umad_port during ->open().  But now that
we're embedding a struct cdev in struct ib_umad_port, we can use the
container_of() macro to go from the inode back to the device instead.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/umad: Convert *cdev to cdev in struct ib_umad_port
Alexander Chiang [Tue, 2 Feb 2010 19:08:25 +0000 (19:08 +0000)]
IB/umad: Convert *cdev to cdev in struct ib_umad_port

Instead of storing pointers to cdev and sm_cdev, embed the full
structures instead.

This change allows us to use the container_of() macro in ib_umad_open()
and ib_umad_sm_open() in a future patch.

This change increases the size of struct ib_umad_port to 320 bytes
from 128.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/uverbs: Whitespace cleanup
Alexander Chiang [Tue, 2 Feb 2010 19:08:19 +0000 (19:08 +0000)]
IB/uverbs: Whitespace cleanup

Clean up the errors as shown when 'let c_space_errors=1' is set in vim.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/uverbs: Pack struct ib_uverbs_event_file tighter
Alexander Chiang [Tue, 2 Feb 2010 19:08:14 +0000 (19:08 +0000)]
IB/uverbs: Pack struct ib_uverbs_event_file tighter

Eliminate some padding in the structure by rearranging the members.
sizeof(struct ib_uverbs_event_file) is now 72 bytes (from 80) and
more members now fit in the first cacheline.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/uverbs: Increase maximum devices supported
Alexander Chiang [Tue, 2 Feb 2010 19:08:09 +0000 (19:08 +0000)]
IB/uverbs: Increase maximum devices supported

Some large systems may support more than IB_UVERBS_MAX_DEVICES
(currently 32).

This change allows us to support more devices in a backwards-compatible
manner.  The first IB_UVERBS_MAX_DEVICES keep the same major/minor
device numbers that they've always had.

If there are more than IB_UVERBS_MAX_DEVICES, we then dynamically
request a new major device number (new minors start at 0).

This change increases the maximum number of HCAs to 64 (from 32).

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/uverbs: use stack variable 'base' in ib_uverbs_add_one
Alexander Chiang [Tue, 2 Feb 2010 19:08:04 +0000 (19:08 +0000)]
IB/uverbs: use stack variable 'base' in ib_uverbs_add_one

This change is not useful by itself, but sets us up for a future change
that allows us to support more than IB_UVERBS_MAX_DEVICES in a system.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/uverbs: Use stack variable 'devnum' in ib_uverbs_add_one
Alexander Chiang [Tue, 2 Feb 2010 19:07:59 +0000 (19:07 +0000)]
IB/uverbs: Use stack variable 'devnum' in ib_uverbs_add_one

This change is not useful by itself, but it sets us up for a future
change that allows us to dynamically allocate device numbers in case
we have more than IB_UVERBS_MAX_DEVICES in the system.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/uverbs: Remove dev_table
Alexander Chiang [Tue, 2 Feb 2010 19:07:54 +0000 (19:07 +0000)]
IB/uverbs: Remove dev_table

dev_table's raison d'etre was to associate an inode back to a struct
ib_uverbs_device.

However, now that we've converted ib_uverbs_device to contain an
embedded cdev (instead of a *cdev), we can use the container_of()
macro and cast back to the containing device.

There's no longer any need for dev_table, so get rid of it.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/uverbs: Convert *cdev to cdev in struct ib_uverbs_device
Alexander Chiang [Tue, 2 Feb 2010 19:07:49 +0000 (19:07 +0000)]
IB/uverbs: Convert *cdev to cdev in struct ib_uverbs_device

Instead of storing a pointer to a cdev, embed the entire struct cdev.

This change allows us to use the container_of() macro in
ib_uverbs_open() in a future patch.

This change increases the size of struct ib_uverbs_device to 168 bytes
across 3 cachelines from 80 bytes in 2 cachelines.  However, we
rearrange the members so that everything fits into the first cacheline
except for the struct cdev. Finally, we don't touch the cdev in any
fastpaths, so this change shouldn't negatively affect performance.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/iser: Remove redundant locking from iser scsi command response flow
Or Gerlitz [Mon, 8 Feb 2010 13:23:11 +0000 (13:23 +0000)]
IB/iser: Remove redundant locking from iser scsi command response flow

Currently the iSER receive completion flow takes the session lock
twice.  Optimize it to avoid the first one by letting
iser_task_rdma_finalize() be called only from the cleanup_task
callback invoked by iscsi_free_task, thus reducing the contention on
the session lock between the scsi command submission to the scsi
command completion flows.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/iser: Use libiscsi passthrough mode
Or Gerlitz [Mon, 8 Feb 2010 13:22:34 +0000 (13:22 +0000)]
IB/iser: Use libiscsi passthrough mode

libiscsi passthrough mode invokes the transport xmit calls directly
without first going through an internal queue, unlike the other mode,
which uses a queue and a xmitworker thread.  Now that the "cant_sleep"
prerequisite of iscsi_host_alloc is met, move to use it.  Handling
xmit errors is now done by the passthrough flow of libiscsi.  Since
the queue/worker aren't used in this mode, the code that schedules the
xmitworker is removed.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/iser: Remove unnecessary connection checks
Or Gerlitz [Mon, 8 Feb 2010 13:21:18 +0000 (13:21 +0000)]
IB/iser: Remove unnecessary connection checks

Remove unnecessary checks for the IB connection state and for QP
overflow, as conn state changes are reported by iSER to libiscsi and
handled there. QP overflow is theoretically possible only when
unsolicited data-outs are used; anyway it's being checked and handled
by HW drivers.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/iser: Use atomic allocations
Or Gerlitz [Mon, 8 Feb 2010 13:20:43 +0000 (13:20 +0000)]
IB/iser: Use atomic allocations

Two minor flows in iSER's data path still use allocations; move them
to be atomic as a preperation step towards moving to use libiscsi
passthrough mode.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/iser: Simplify send flow/descriptors
Or Gerlitz [Mon, 8 Feb 2010 13:19:56 +0000 (13:19 +0000)]
IB/iser: Simplify send flow/descriptors

Simplify and shrink the logic/code used for the send descriptors.
Changes include removing struct iser_dto (an unnecessary abstraction),
using struct iser_regd_buf only for handling SCSI commands, using
dma_sync instead of dma_map/unmap, etc.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/iser: Use different CQ for send completions
Or Gerlitz [Mon, 8 Feb 2010 13:19:21 +0000 (13:19 +0000)]
IB/iser: Use different CQ for send completions

Use a different CQ for send completions, where send completions are
polled by the interrupt-driven receive completion handler.  Therefore,
interrupts aren't used for the send CQ.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/iser: Remove atomic counter for posted receive buffers
Or Gerlitz [Mon, 8 Feb 2010 13:18:39 +0000 (13:18 +0000)]
IB/iser: Remove atomic counter for posted receive buffers

Now that both the posting and reaping of receive buffers is done in
the completion path, the counter of outstanding buffers not be atomic.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/iser: New receive buffer posting logic
Or Gerlitz [Mon, 8 Feb 2010 13:17:42 +0000 (13:17 +0000)]
IB/iser: New receive buffer posting logic

Currently, the recv buffer posting logic is based on the transactional
nature of iSER which allows for posting a buffer before sending a PDU.
Change this to post only when the number of outstanding recv buffers
is below a water mark and in a batched manner, thus simplifying and
optimizing the data path.  Use a pre-allocated ring of recv buffers
instead of allocating from kmem cache.  A special treatment is given
to the login response buffer whose size must be 8K unlike the size of
buffers used for any other purpose which is 128 bytes.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/iser: Revert commit bba7ebb "avoid recv buffer exhaustion"
Or Gerlitz [Mon, 8 Feb 2010 13:17:11 +0000 (13:17 +0000)]
IB/iser: Revert commit bba7ebb "avoid recv buffer exhaustion"

We will make a major change in the recv buffer posting logic, after
which the problem commit bba7ebb "avoid recv buffer exhaustion caused
by unexpected PDUs" comes to solve doesn't exist any more, so revert it.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoRDMA/nes: Change WQ overflow return code
Or Gerlitz [Thu, 4 Feb 2010 11:25:07 +0000 (11:25 +0000)]
RDMA/nes: Change WQ overflow return code

Change the nes driver to return -ENOMEM on SQ/RQ overflow to match the
return code of other RDMA HW drivers (e.g cxgb3, ehca, mlx4, mthca).

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Acked-by: Chien Tung <chien.tin.tung@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoRDMA/nes: Multiple disconnects cause crash during AE handling
Faisal Latif [Fri, 12 Feb 2010 19:58:05 +0000 (19:58 +0000)]
RDMA/nes: Multiple disconnects cause crash during AE handling

There is a double disconnect during AE processing, causing crashes.
While fixing the crash, also simplify the AE handling code.

Signed-off-by: Faisal Latif <faisal.latif@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoRDMA/nes: Fix crash when listener destroyed during loopback setup
Faisal Latif [Fri, 12 Feb 2010 19:57:04 +0000 (19:57 +0000)]
RDMA/nes: Fix crash when listener destroyed during loopback setup

When a listener is destroyed and there is an MPA response pending for
loopback connection, the active side cm_node gets destroyed twice:
once in cm_event_connect_error() and again in nes_accept()/nes_reject().

Increment the cm_node's refcount so it's not destroyed by
cm_event_connect_error().

Signed-off-by: Faisal Latif <faisal.latif@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoRDMA/nes: Use atomic counters for CM listener create and destroy
Faisal Latif [Fri, 12 Feb 2010 19:55:03 +0000 (19:55 +0000)]
RDMA/nes: Use atomic counters for CM listener create and destroy

After running long iterative MPI tests, sometimes ethtool reports a
"CM Destroy Listener" count more than the "CM Create Listener" count.
This inconsistency is fixed by making counter variables atomic.

Signed-off-by: Faisal Latif <faisal.latif@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/ehca: Require in_wc in process_mad()
Alexander Schmidt [Tue, 16 Feb 2010 08:59:50 +0000 (08:59 +0000)]
IB/ehca: Require in_wc in process_mad()

If the caller does not pass a valid in_wc to process_mad(), return MAD
failure status, as it is not possible to generate a valid MAD redirect
response (and redirects are the only MAD responses ehca generates).

Signed-off-by: Alexander Schmidt <alexs@linux.vnet.ibm.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/ehca: Allow access for ib_query_qp()
Alexander Schmidt [Wed, 27 Jan 2010 17:03:08 +0000 (17:03 +0000)]
IB/ehca: Allow access for ib_query_qp()

The max_dest_rd_atomic and max_qp_rd_atomic values are properly
returned by query_qp(), so there should not be an error returned when
they are queried.

Signed-off-by: Alexander Schmidt <alexs@linux.vnet.ibm.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/ehca: Do not turn off irqs in tasklet context
Alexander Schmidt [Wed, 27 Jan 2010 17:01:56 +0000 (17:01 +0000)]
IB/ehca: Do not turn off irqs in tasklet context

The irq_spinlock is only taken in tasklet context, so it is safe not to
disable hardware interrupts.

Signed-off-by: Alexander Schmidt <alexs@linux.vnet.ibm.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIB/mlx4: Simplify retrieval of ib_device
Eli Cohen [Wed, 27 Jan 2010 13:57:03 +0000 (13:57 +0000)]
IB/mlx4: Simplify retrieval of ib_device

struct ib_qp  already holds a pointer to the ib device. No need to dive to the
hw device object to retrieve it.

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoIPoIB: Remove TX moderation settings from ethtool support
Or Gerlitz [Thu, 11 Feb 2010 23:40:40 +0000 (15:40 -0800)]
IPoIB: Remove TX moderation settings from ethtool support

As of commit f56bcd8 ("IPoIB: Use separate CQ for UD send
completions"), there are no TX interrupts.  Change the ethtool code
not to report TX moderation settings, so users will not be misled to
think they can control TX interrupt moderation.  Pointed out by Alex
Vainman <alexv@voltaire.com>

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoRDMA/cxgb3: Remove BUG_ON() on CQ rearm failure
Steve Wise [Thu, 11 Feb 2010 23:40:29 +0000 (15:40 -0800)]
RDMA/cxgb3: Remove BUG_ON() on CQ rearm failure

Failure to rearm a CQ means the cxgb3 device is wedged, but we shouldn't
kill the whole system with a BUG_ON() if this happens.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoRDMA/cm: Remove unused definition of RDMA_PS_SCTP
Sean Hefty [Thu, 11 Feb 2010 23:40:25 +0000 (15:40 -0800)]
RDMA/cm: Remove unused definition of RDMA_PS_SCTP

The defined SCTP number is incorrect (0x83, rather than 0x84), and
since it is not used anywhere, simply remove the definition.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp
Linus Torvalds [Thu, 11 Feb 2010 22:07:13 +0000 (14:07 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/bp/bp

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp:
  amd64_edac: Do not falsely trigger kerneloops

14 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Thu, 11 Feb 2010 22:07:00 +0000 (14:07 -0800)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block

* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
  cciss: Make cciss_seq_show handle holes in the h->drv[] array
  cfq-iosched: split seeky coop queues after one slice

14 years agoMerge branch 'bugfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6
Linus Torvalds [Thu, 11 Feb 2010 22:06:28 +0000 (14:06 -0800)]
Merge branch 'bugfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6

* 'bugfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6:
  NFS: Fix the mapping of the NFSERR_SERVERFAULT error
  NFS: Remove a redundant check for PageFsCache in nfs_migrate_page()
  NFS: Fix a bug in nfs_fscache_release_page()

14 years agoMerge git://git.infradead.org/users/cbou/battery-2.6.33
Linus Torvalds [Thu, 11 Feb 2010 22:06:15 +0000 (14:06 -0800)]
Merge git://git.infradead.org/users/cbou/battery-2.6.33

* git://git.infradead.org/users/cbou/battery-2.6.33:
  wm97xx_battery: Handle missing platform data gracefully

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6
Linus Torvalds [Thu, 11 Feb 2010 22:05:55 +0000 (14:05 -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: Obtain proper host structure during response-queue processing.
  [SCSI] compat_ioct: fix bsg SG_IO
  [SCSI] qla2xxx: make msix interrupt handler safe for irq
  [SCSI] zfcp: Report FC BSG errors in correct field
  [SCSI] mptfusion : mptscsih_abort return value should be SUCCESS instead of value 0.

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Thu, 11 Feb 2010 22:03:42 +0000 (14:03 -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: psmouse - make sure we don't schedule reconnects after cleanup

14 years agoMerge branch 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Thu, 11 Feb 2010 22:03:28 +0000 (14:03 -0800)]
Merge branch 'drm-linus' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (30 commits)
  vgaarb: fix incorrect dereference of userspace pointer.
  drm/radeon/kms: retry auxch on 0x20 timeout value.
  drm/radeon: Skip dma copy test in benchmark if card doesn't have dma engine.
  drm/vmwgfx: Fix a circular locking dependency bug.
  drm/vmwgfx: Drop scanout flag compat and add execbuf ioctl parameter members. Bumps major.
  drm/vmwgfx: Report propper framebuffer_{max|min}_{width|height}
  drm/vmwgfx: Update the user-space interface.
  drm/radeon/kms: fix screen clearing before fbcon.
  nouveau: fix state detection with switchable graphics
  drm/nouveau: move dereferences after null checks
  drm/nv50: make the pgraph irq handler loop like the pre-nv50 version
  drm/nv50: delete ramfc object after disabling fifo, not before
  drm/nv50: avoid unloading pgraph context when ctxprog is running
  drm/nv50: align size of buffer object to the right boundaries.
  drm/nv50: disregard dac outputs in nv50_sor_dpms()
  drm/nv50: prevent multiple init tables being parsed at the same time
  drm/nouveau: make dp auxch xfer len check for reads only
  drm/nv40: make INIT_COMPUTE_MEM a NOP, just like nv50
  drm/nouveau: Add proper vgaarb support.
  drm/nouveau: Fix fbcon on mixed pre-NV50 + NV50 multicard.
  ...

14 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx
Linus Torvalds [Thu, 11 Feb 2010 22:02:27 +0000 (14:02 -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:
  drivers/dma: Correct NULL test
  async-tx: fix buffer submission error handling in ipu_idma.c
  dmaengine: correct onstack wait_queue_head declaration
  ioat: fix infinite timeout checking in ioat2_quiesce
  dmaengine: fix memleak in dma_async_device_unregister

14 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Thu, 11 Feb 2010 22:01:46 +0000 (14:01 -0800)]
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  MIPS: Don't probe reserved EntryHi bits.
  MIPS: SNI: Correct NULL test
  MIPS: Fix __devinit __cpuinit confusion in cpu_cache_init
  MIPS: IP27: Make defconfig useful again.
  MIPS: Fixup of the r4k timer

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Thu, 11 Feb 2010 22:01:25 +0000 (14:01 -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:
  RDMA/cm: Revert association of an RDMA device when binding to loopback

14 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 11 Feb 2010 22:01:10 +0000 (14:01 -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, apic: Don't use logical-flat mode when CPU hotplug may exceed 8 CPUs
  x86-32: Make AT_VECTOR_SIZE_ARCH=2
  x86/agp: Fix amd64-agp module initialization regression
  x86, doc: Fix minor spelling error in arch/x86/mm/gup.c

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Thu, 11 Feb 2010 22:00:27 +0000 (14:00 -0800)]
Merge git://git./linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  sparc32: Fix thinko in previous change.
  sparc: Align clone and signal stacks to 16 bytes.

14 years agofs/exec.c: restrict initial stack space expansion to rlimit
Michael Neuling [Wed, 10 Feb 2010 21:56:42 +0000 (13:56 -0800)]
fs/exec.c: restrict initial stack space expansion to rlimit

When reserving stack space for a new process, make sure we're not
attempting to expand the stack by more than rlimit allows.

This fixes a bug caused by b6a2fea39318e43fee84fa7b0b90d68bed92d2ba ("mm:
variable length argument support") and unmasked by
fc63cf237078c86214abcb2ee9926d8ad289da9b ("exec: setup_arg_pages() fails
to return errors").

This bug means that when limiting the stack to less the 20*PAGE_SIZE (eg.
80K on 4K pages or 'ulimit -s 79') all processes will be killed before
they start.  This is particularly bad with 64K pages, where a ulimit below
1280K will kill every process.

To test, do:

  'ulimit -s 15; ls'

before and after the patch is applied.  Before it's applied, 'ls' should
be killed.  After the patch is applied, 'ls' should no longer be killed.

A stack limit of 15KB since it's small enough to trigger 20*PAGE_SIZE.
Also 15KB not a multiple of PAGE_SIZE, which is a trickier case to handle
correctly with this code.

4K pages should be fine to test with.

[kosaki.motohiro@jp.fujitsu.com: cleanup]
[akpm@linux-foundation.org: cleanup cleanup]
Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Americo Wang <xiyou.wangcong@gmail.com>
Cc: Anton Blanchard <anton@samba.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: James Morris <jmorris@namei.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Serge Hallyn <serue@us.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agocompat_ioctl: add compat handler for TIOCGSID ioctl
Andreas Schwab [Wed, 10 Feb 2010 21:56:40 +0000 (13:56 -0800)]
compat_ioctl: add compat handler for TIOCGSID ioctl

This is used by tcgetsid(3).

Signed-off-by: Andreas Schwab <schwab@linux-m68k.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMAINTAINERS: changed LTP maintainership responsibilities
Rishikesh [Wed, 10 Feb 2010 21:56:40 +0000 (13:56 -0800)]
MAINTAINERS: changed LTP maintainership responsibilities

Change the LTP maintainer responsibities from 2010.

Ref: http://marc.info/?l=ltp-list&m=126502242912536&w=2

Signed-off-by : Rishikesh K Rajak <risrajak@linux.vnet.ibm.com>
Cc: Subrata Modak <subrata@linux.vnet.ibm.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Cc: Garrett Cooper <yanegomi@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoedac: mpc85xx fix build regression by removing unused debug code
Peter Tyser [Wed, 10 Feb 2010 21:56:37 +0000 (13:56 -0800)]
edac: mpc85xx fix build regression by removing unused debug code

Some unused, unsupported debug code existed in the mpc85xx EDAC driver
that resulted in a build failure when CONFIG_EDAC_DEBUG was defined:

  drivers/edac/mpc85xx_edac.c: In function 'mpc85xx_mc_err_probe':
  drivers/edac/mpc85xx_edac.c:1031: error: implicit declaration of function 'edac_mc_register_mcidev_debug'
  drivers/edac/mpc85xx_edac.c:1031: error: 'debug_attr' undeclared (first use in this function)
  drivers/edac/mpc85xx_edac.c:1031: error: (Each undeclared identifier is reported only once
  drivers/edac/mpc85xx_edac.c:1031: error: for each function it appears in.)

Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoedac: mpc85xx fix bad page calculation
Peter Tyser [Wed, 10 Feb 2010 21:56:36 +0000 (13:56 -0800)]
edac: mpc85xx fix bad page calculation

Commit b4846251727a38a7f248e41308c060995371dd05 ("edac: mpc85xx add
mpc83xx support") accidentally broke how a chip select's first and last
page addresses are calculated.  The page addresses are being shifted too
far right by PAGE_SHIFT.  This results in errors such as:

  EDAC MPC85xx MC1: Err addr: 0x003075c0
  EDAC MPC85xx MC1: PFN: 0x00000307
  EDAC MPC85xx MC1: PFN out of range!
  EDAC MC1: INTERNAL ERROR: row out of range (4 >= 4)
  EDAC MC1: CE - no information available: INTERNAL ERROR

The vaule of PAGE_SHIFT is already being taken into consideration during
the calculation of the 'start' and 'end' variables, thus it is not
necessary to account for it again when setting a chip select's first and
last page address.

Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Cc: Ira W. Snyder <iws@ovro.caltech.edu>
Cc: Kumar Gala <galak@gate.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agommc_test: block addressed cards
Johan Kristell [Wed, 10 Feb 2010 21:56:34 +0000 (13:56 -0800)]
mmc_test: block addressed cards

This patch fixes a bug in the multiblock write tests where the written
data is read back for verifying one block at a time.  The tests in
mmc_test assumes that all cards are byte addressable.

This will cause the multi block write tests to fail, leading the user of
the mmc_test driver thinking there is something wrong with the sdhci
driver they are testing.

The start address for the block is calculated as: blocknum * 512. For
block addressable cards the blocknum alone should be used.

Signed-off-by: Johan Kristell <johan.kristell@axis.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agotpm_infineon: fix suspend/resume handler for pnp_driver
Marcel Selhorst [Wed, 10 Feb 2010 21:56:32 +0000 (13:56 -0800)]
tpm_infineon: fix suspend/resume handler for pnp_driver

When suspending, tpm_infineon calls the generic suspend function of the
TPM framework.  However, the TPM framework does not return and the system
hangs upon suspend.  When sending the necessary command "TPM_SaveState"
directly within the driver, suspending and resuming works fine.

Signed-off-by: Marcel Selhorst <m.selhorst@sirrix.com>
Cc: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Cc: Debora Velarde <debora@linux.vnet.ibm.com>
Cc: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
Cc: <stable@kernel.org> [2.6.32.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agosparc32: Fix thinko in previous change.
David S. Miller [Thu, 11 Feb 2010 20:29:16 +0000 (12:29 -0800)]
sparc32: Fix thinko in previous change.

Should mask stack with 0xf not "0x15".

Noticed by Blue Swirl <blauwirbel@gmail.com>

Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoamd64_edac: Do not falsely trigger kerneloops
Borislav Petkov [Thu, 11 Feb 2010 16:15:57 +0000 (17:15 +0100)]
amd64_edac: Do not falsely trigger kerneloops

An unfortunate "WARNING" in the message amd64_edac dumps when the system
doesn't support DRAM ECC or ECC checking is not enabled in the BIOS
used to trigger kerneloops which qualified the message as an OOPS thus
misleading the users. See, e.g.

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/422536
http://bugzilla.kernel.org/show_bug.cgi?id=15238

Downgrade the message level to KERN_NOTICE and fix the formulation.

Cc: stable@kernel.org # .32.x
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
Acked-by: Doug Thompson <dougthompson@xmission.com>
14 years agovgaarb: fix incorrect dereference of userspace pointer.
Andy Getzendanner [Thu, 11 Feb 2010 04:04:48 +0000 (14:04 +1000)]
vgaarb: fix incorrect dereference of userspace pointer.

This patch corrects a userspace pointer dereference in the VGA arbiter
in 2.6.32.1.

copy_from_user() is used at line 822 to copy the contents of buf into
kbuf, but a call to strncmp() on line 964 uses buf rather than kbuf.  This
problem led to a GPF in strncmp() when X was started on my x86_32 systems.
 X triggered the behavior with a write of "target PCI:0000:01:00.0" to
/dev/vga_arbiter.

The patch has been tested against 2.6.32.1 and observed to correct the GPF
observed when starting X or manually writing the string "target
PCI:0000:01:00.0" to /dev/vga_arbiter.

Signed-off-by: Andy Getzendanner <james.getzendanner@students.olin.edu>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agoMerge branch 'drm-radeon-linus' of ../drm-next
Dave Airlie [Thu, 11 Feb 2010 04:03:51 +0000 (14:03 +1000)]
Merge branch 'drm-radeon-linus' of ../drm-next

* 'drm-radeon-linus' of ../drm-next:
  drm/radeon/kms: retry auxch on 0x20 timeout value.
  drm/radeon: Skip dma copy test in benchmark if card doesn't have dma engine.
  drm/radeon/kms: fix screen clearing before fbcon.
  drm/radeon/kms: add quirk for VGA without DDC on rv730 XFX card.
  drm/radeon/kms: don't crash if no DDC bus on VGA/DVI connector.
  drm/radeon/kms: change Kconfig text to reflect the new option.
  drm/radeon/kms: suspend and resume audio stuff

14 years agodrm/radeon/kms: retry auxch on 0x20 timeout value.
Dave Airlie [Wed, 10 Feb 2010 06:52:45 +0000 (16:52 +1000)]
drm/radeon/kms: retry auxch on 0x20 timeout value.

ATOM appears to return 0x20 which seems to mean some sort of timeout.

retry the transaction up to 10 times before failing, this
makes DP->VGA convertor we bought work at least a bit more predictably.

Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agodrm/radeon: Skip dma copy test in benchmark if card doesn't have dma engine.
Pauli Nieminen [Wed, 10 Feb 2010 22:10:33 +0000 (00:10 +0200)]
drm/radeon: Skip dma copy test in benchmark if card doesn't have dma engine.

radeon_copy_dma is only available for r200 or newer cards.
Call to radeon_copy_dma would result to NULL pointer
dereference if benchmarking asic without dma engine.

Signed-off-by: Pauli Nieminen <suokkos@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agoMerge remote branch 'nouveau/for-airlied' of nouveau-2.6
Dave Airlie [Thu, 11 Feb 2010 02:10:48 +0000 (12:10 +1000)]
Merge remote branch 'nouveau/for-airlied' of nouveau-2.6

* 'nouveau/for-airlied' of /home/airlied/kernel/drm-next:
  nouveau: fix state detection with switchable graphics
  drm/nouveau: move dereferences after null checks
  drm/nv50: make the pgraph irq handler loop like the pre-nv50 version
  drm/nv50: delete ramfc object after disabling fifo, not before
  drm/nv50: avoid unloading pgraph context when ctxprog is running
  drm/nv50: align size of buffer object to the right boundaries.
  drm/nv50: disregard dac outputs in nv50_sor_dpms()
  drm/nv50: prevent multiple init tables being parsed at the same time
  drm/nouveau: make dp auxch xfer len check for reads only
  drm/nv40: make INIT_COMPUTE_MEM a NOP, just like nv50
  drm/nouveau: Add proper vgaarb support.
  drm/nouveau: Fix fbcon on mixed pre-NV50 + NV50 multicard.
  drivers/gpu/drm/nouveau/nouveau_grctx.c: correct NULL test
  drm/nouveau: call ttm_bo_wait with the bo lock held to prevent hang
  drm/nouveau: Fixup semaphores on pre-nv50 cards.
  drm/nouveau: Add getparam to get available PGRAPH units.
  drm/nouveau: Add module options to disable acceleration.
  drm/nouveau: fix non-vram notifier blocks

14 years agodrm/vmwgfx: Fix a circular locking dependency bug.
Thomas Hellstrom [Mon, 8 Feb 2010 09:57:25 +0000 (09:57 +0000)]
drm/vmwgfx: Fix a circular locking dependency bug.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agodrm/vmwgfx: Drop scanout flag compat and add execbuf ioctl parameter members. Bumps...
Jakob Bornecrantz [Tue, 9 Feb 2010 21:29:47 +0000 (21:29 +0000)]
drm/vmwgfx: Drop scanout flag compat and add execbuf ioctl parameter members. Bumps major.

Even if this bumps the version to 1 it does not mean the driver is
out of staging. From what we know this is the last backwards
incompatible change to the driver.

Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agodrm/vmwgfx: Report propper framebuffer_{max|min}_{width|height}
Jakob Bornecrantz [Tue, 9 Feb 2010 19:41:57 +0000 (19:41 +0000)]
drm/vmwgfx: Report propper framebuffer_{max|min}_{width|height}

Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agodrm/vmwgfx: Update the user-space interface.
Thomas Hellstrom [Tue, 9 Feb 2010 19:41:55 +0000 (19:41 +0000)]
drm/vmwgfx: Update the user-space interface.

When time-based throttling is implemented, we need to bump minor.
When the old way of detecting scanout is removed, we need to bump major.
In the meantime, this change should not break existing user-space.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agodrm/radeon/kms: fix screen clearing before fbcon.
Dave Airlie [Tue, 9 Feb 2010 02:31:08 +0000 (12:31 +1000)]
drm/radeon/kms: fix screen clearing before fbcon.

This memset_io was added to debug something way back and got
left behind, memset the fb to black so the borders don't be all white.

Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agonouveau: fix state detection with switchable graphics
Matthew Garrett [Wed, 10 Feb 2010 10:50:34 +0000 (20:50 +1000)]
nouveau: fix state detection with switchable graphics

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
14 years agodrm/nouveau: move dereferences after null checks
Marcin Slusarz [Sat, 30 Jan 2010 14:41:00 +0000 (15:41 +0100)]
drm/nouveau: move dereferences after null checks

Reported-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Signed-off-by: Maarten Maathuis <madman2003@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
14 years agoMIPS: Don't probe reserved EntryHi bits.
David Daney [Mon, 8 Feb 2010 20:27:00 +0000 (12:27 -0800)]
MIPS: Don't probe reserved EntryHi bits.

The patch that adds cpu_probe_vmbits is erroneously writing to reserved
bit 12.  Since we are really only probing high bits, don't write this bit
with a one.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Cc: Guenter Roeck <guenter.roeck@ericsson.com>
Patchwork: http://patchwork.linux-mips.org/patch/949/
Acked-by: Guenter Roeck <guenter.roeck@ericsson.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: SNI: Correct NULL test
Julia Lawall [Sat, 6 Feb 2010 08:42:16 +0000 (09:42 +0100)]
MIPS: SNI: Correct NULL test

Test the value that was just allocated rather than the previously tested one.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r@
expression *x;
expression e;
identifier l;
@@

if (x == NULL || ...) {
    ... when forall
    return ...; }
... when != goto l;
    when != x = e
    when != &x
*x == NULL
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
To: linux-mips@linux-mips.org
To: linux-kernel@vger.kernel.org
To: kernel-janitors@vger.kernel.org
Patchwork: http://patchwork.linux-mips.org/patch/945/
Acked-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Fix __devinit __cpuinit confusion in cpu_cache_init
David Daney [Thu, 4 Feb 2010 23:48:49 +0000 (15:48 -0800)]
MIPS: Fix __devinit __cpuinit confusion in cpu_cache_init

cpu_cache_init and the things it calls should all be __cpuinit instead
of __devinit.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/938/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: IP27: Make defconfig useful again.
Ralf Baechle [Wed, 3 Feb 2010 18:16:34 +0000 (19:16 +0100)]
MIPS: IP27: Make defconfig useful again.

RTC support was rewritten but the defconfig files were not updated.  Enable
IPv6 support which for some folks already is a must have.  Assign useful
values to other new options.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Fixup of the r4k timer
Wu Zhangjin [Mon, 1 Feb 2010 09:10:55 +0000 (17:10 +0800)]
MIPS: Fixup of the r4k timer

As reported by Maxime Bizon, the commit "MIPS: PowerTV: Fix support for
timer interrupts with > 64 external IRQs" have broken the r4k timer
since it didn't initialize the cp0_compare_irq_shift variable used in
c0_compare_int_pending() on the architectures whose cpu_has_mips_r2 is
false.

This patch fixes it via initializing the cp0_compare_irq_shift as the
cp0_compare_irq used in the old c0_compare_int_pending().

Reported-by: Maxime Bizon <mbizon@freebox.fr>
Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: David VomLehn <dvomlehn@cisco.com>
Cc: mbizon@freebox.fr
Cc: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/922/
Tested-by: Shane McDonald <mcdonald.shane@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoRDMA/cm: Revert association of an RDMA device when binding to loopback
Sean Hefty [Tue, 9 Feb 2010 00:41:15 +0000 (16:41 -0800)]
RDMA/cm: Revert association of an RDMA device when binding to loopback

Revert the following change from commit 6f8372b6 ("RDMA/cm: fix
loopback address support")

   The defined behavior of rdma_bind_addr is to associate an RDMA
   device with an rdma_cm_id, as long as the user specified a non-
   zero address.  (ie they weren't just trying to reserve a port)
   Currently, if the loopback address is passed to rdma_bind_addr,
   no device is associated with the rdma_cm_id.  Fix this.

It turns out that important apps such as Open MPI depend on
rdma_bind_addr() NOT associating any RDMA device when binding to a
loopback address.  Open MPI is being updated to deal with this, but at
least until a new Open MPI release is available, maintain the previous
behavior: allow rdma_bind_addr() to succeed, but do not bind to a
device.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Acked-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agodrivers/dma: Correct NULL test
Julia Lawall [Sat, 6 Feb 2010 08:43:41 +0000 (09:43 +0100)]
drivers/dma: Correct NULL test

cohd_fin has already been verified not to be NULL, so the argument to
BUG_ON cannot be true.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r@
expression *x;
expression e;
identifier l;
@@

if (x == NULL || ...) {
    ... when forall
    return ...; }
... when != goto l;
    when != x = e
    when != &x
*x == NULL
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Acked-by: Linus Walleij <linus.walleij@stericsson.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
14 years agoasync-tx: fix buffer submission error handling in ipu_idma.c
Guennadi Liakhovetski [Wed, 10 Feb 2010 16:32:38 +0000 (17:32 +0100)]
async-tx: fix buffer submission error handling in ipu_idma.c

If submitting new buffer failed, a wrong descriptor gets completed and it
doesn't check, if a callback is at all defined, which can lead to an Oops. Fix
these bugs and make ipu_update_channel_buffer() void, because it never fails.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
14 years agocompat_ioctl: ignore RAID_VERSION ioctl
Arnd Bergmann [Wed, 10 Feb 2010 15:14:04 +0000 (16:14 +0100)]
compat_ioctl: ignore RAID_VERSION ioctl

md ioctls are now handled by the md driver itself, but mdadm
may call RAID_VERSION on other devices as well. Mark the command
as IGNORE_IOCTL so this fails silently rather than printing
an annoying message.

Reported-by: "Michael S. Tsirkin" <m.s.tsirkin@gmail.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge branch 'i2c-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvar...
Linus Torvalds [Wed, 10 Feb 2010 15:34:46 +0000 (07:34 -0800)]
Merge branch 'i2c-for-linus' of git://git./linux/kernel/git/jdelvare/staging

* 'i2c-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvare/staging:
  i2c-tiny-usb: Fix on big-endian systems

14 years agoMerge branch 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6
Linus Torvalds [Wed, 10 Feb 2010 15:19:07 +0000 (07:19 -0800)]
Merge branch 'for-linus' of git://git390.marist.edu/linux-2.6

* 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6:
  [S390] Fix struct _lowcore layout.
  [S390] qdio: prevent call trace if CHPID is offline
  [S390] qdio: continue polling for buffer state ERROR

14 years agoMerge branch 'kvm-updates/2.6.33' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Wed, 10 Feb 2010 15:18:15 +0000 (07:18 -0800)]
Merge branch 'kvm-updates/2.6.33' of git://git./virt/kvm/kvm

* 'kvm-updates/2.6.33' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: PIT: control word is write-only
  kvmclock: count total_sleep_time when updating guest clock
  Export the symbol of getboottime and mmonotonic_to_bootbased

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/hskinnemoen/avr32-2.6
Linus Torvalds [Wed, 10 Feb 2010 15:17:54 +0000 (07:17 -0800)]
Merge git://git./linux/kernel/git/hskinnemoen/avr32-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/hskinnemoen/avr32-2.6:
  avr32: clean up memory allocation in at32_add_device_mci
  arch/avr32: Fix build failure for avr32 caused by typo

14 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
Linus Torvalds [Wed, 10 Feb 2010 15:16:58 +0000 (07:16 -0800)]
Merge branch 'merge' of git://git./linux/kernel/git/benh/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc:
  powerpc: Fix address masking bug in hpte_need_flush()

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Wed, 10 Feb 2010 15:16:44 +0000 (07:16 -0800)]
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6:
  cifs: fix dentry hash calculation for case-insensitive mounts
  [CIFS] Don't cache timestamps on utimes due to coarse granularity
  [CIFS] Maximum username length check in session setup does not match
  cifs: fix length calculation for converted unicode readdir names
  [CIFS] Add support for TCP_NODELAY

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Wed, 10 Feb 2010 15:15:21 +0000 (07:15 -0800)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (29 commits)
  drivers/net: Correct NULL test
  MAINTAINERS: networking drivers - Add git net-next tree
  net/sched: Fix module name in Kconfig
  cxgb3: fix GRO checksum check
  dst: call cond_resched() in dst_gc_task()
  netfilter: nf_conntrack: fix hash resizing with namespaces
  netfilter: xtables: compat out of scope fix
  netfilter: nf_conntrack: restrict runtime expect hashsize modifications
  netfilter: nf_conntrack: per netns nf_conntrack_cachep
  netfilter: nf_conntrack: fix memory corruption with multiple namespaces
  Bluetooth: Keep a copy of each HID device's report descriptor
  pktgen: Fix freezing problem
  igb: make certain to reassign legacy interrupt vectors after reset
  irda: add missing BKL in irnet_ppp ioctl
  irda: unbalanced lock_kernel in irnet_ppp
  ixgbe: Fix return of invalid txq
  ixgbe: Fix ixgbe_tx_map error path
  netxen: protect resource cleanup by rtnl lock
  netxen: fix tx timeout recovery for NX2031 chip
  Bluetooth: Enter active mode before establishing a SCO link.
  ...

14 years agox86, apic: Don't use logical-flat mode when CPU hotplug may exceed 8 CPUs
Suresh Siddha [Wed, 10 Feb 2010 02:01:44 +0000 (18:01 -0800)]
x86, apic: Don't use logical-flat mode when CPU hotplug may exceed 8 CPUs

We need to fall back from logical-flat APIC mode to physical-flat mode
when we have more than 8 CPUs.  However, in the presence of CPU
hotplug(with bios listing not enabled but possible cpus as disabled cpus in
MADT), we have to consider the number of possible CPUs rather than
the number of current CPUs; otherwise we may cross the 8-CPU boundary
when CPUs are added later.

32bit apic code can use more cleanups (like the removal of vendor checks in
32bit default_setup_apic_routing()) and more unifications with 64bit code.
Yinghai has some patches in works already. This patch addresses the boot issue
that is reported in the virtualization guest context.

[ hpa: incorporated function annotation feedback from Yinghai Lu ]

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
LKML-Reference: <1265767304.2833.19.camel@sbs-t61.sc.intel.com>
Acked-by: Shaohui Zheng <shaohui.zheng@intel.com>
Reviewed-by: Yinghai Lu <yinghai@kernel.org>
Cc: <stable@kernel.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>