~shefty/rdma-dev.git
8 years agoMerge branches 'core', 'cxgb4', 'ipoib', 'iser', 'misc', 'mlx4', 'qib' and 'srp'...
Roland Dreier [Tue, 26 Feb 2013 17:17:56 +0000 (09:17 -0800)]
Merge branches 'core', 'cxgb4', 'ipoib', 'iser', 'misc', 'mlx4', 'qib' and 'srp' into for-next

8 years agoIB/mlx4: Advertise MW support
Shani Michaeli [Wed, 6 Feb 2013 16:19:16 +0000 (16:19 +0000)]
IB/mlx4: Advertise MW support

Indicate memory windows support through device capabilities, kernel
verb entries and the relevant uverbs command mask entries.

Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shani Michaeli <shanim@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/mlx4: Support memory window binding
Shani Michaeli [Wed, 6 Feb 2013 16:19:15 +0000 (16:19 +0000)]
IB/mlx4: Support memory window binding

* Implement memory windows binding in mlx4_ib_post_send.

* Implement mlx4_ib_bind_mw by deferring to mlx4_ib_post_send.

* Rename MLX4_WQE_FMR_PERM_* flags to MLX4_WQE_FMR_AND_BIND_PERM_*,
  indicating that they are used both for fast registration work
  requests, and for memory window bind work requests.

Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shani Michaeli <shanim@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agomlx4: Implement memory windows allocation and deallocation
Shani Michaeli [Wed, 6 Feb 2013 16:19:14 +0000 (16:19 +0000)]
mlx4: Implement memory windows allocation and deallocation

Implement MW allocation and deallocation in mlx4_core and mlx4_ib.
Pass down the enable bind flag when registering memory regions.

Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shani Michaeli <shanim@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agomlx4_core: Enable memory windows in {INIT, QUERY}_HCA
Shani Michaeli [Wed, 6 Feb 2013 16:19:11 +0000 (16:19 +0000)]
mlx4_core: Enable memory windows in {INIT, QUERY}_HCA

Add memory windows-related code to INIT_HCA and QUERY_HCA.

Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shani Michaeli <shanim@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agomlx4_core: Disable memory windows for virtual functions
Shani Michaeli [Wed, 6 Feb 2013 16:19:10 +0000 (16:19 +0000)]
mlx4_core: Disable memory windows for virtual functions

Do not enable memory windows allocation for virtual functions.

In addition, add a few safety checks, such as:

* Verifying the PD of a new MPT matches the VF.
* Making sure binding memory window isn't enabled for FMRs, and
  that new memory windows are not FMR themselves.

Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shani Michaeli <shanim@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIPoIB: Free ipoib neigh on path record failure so path rec queries are retried
Roland Dreier [Mon, 25 Feb 2013 17:42:15 +0000 (09:42 -0800)]
IPoIB: Free ipoib neigh on path record failure so path rec queries are retried

If IPoIB fails to look up a path record (eg if it tries during an SM
failover when one SM is dead but the new one hasn't taken over yet), the
driver ends up with a neighbour structure but no address handle (AH).
There's no mechanism to recover from this: any further packets sent to
this destination will be silently dumped in ipoib_start_xmit().

Fix this by freeing the neighbour structures when a path rec query
fails, so that the next packet queued to be sent will trigger a new path
record query.

Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/srp: Fail I/O requests if the transport is offline
Bart Van Assche [Thu, 21 Feb 2013 17:20:00 +0000 (17:20 +0000)]
IB/srp: Fail I/O requests if the transport is offline

If an SRP target is no longer reachable and srp_reset_host() fails to
reconnect then ib_srp will invoke scsi_remove_host().  That function
will invoke __scsi_remove_device() for each LUN.  And that last
function will change the device state from SDEV_TRANSPORT_OFFLINE into
SDEV_CANCEL.  Certain user space software, e.g. older versions of
multipathd, continue queueing I/O to SCSI devices that are in the
SDEV_CANCEL state.

If these I/O requests are submitted as SG_IO that means that the
REQ_PREEMPT flag will be set and hence that these requests will be
passed to srp_queuecommand().  These requests will time out.  If new
requests are queued fast enough from user space these active requests
will prevent __scsi_remove_device() to finish.

Avoid this by failing I/O requests in the SDEV_CANCEL state if the
transport is offline.  Introduce a new variable to keep track of the
transport state instead of failing requests if (!target->connected ||
target->qp_in_error), so that the SCSI error handler has a chance to
retry commands after a transport layer failure occurred.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Cc: <stable@vger.kernel.org> # 3.8
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/srp: Avoid endless SCSI error handling loop
Bart Van Assche [Thu, 21 Feb 2013 17:19:04 +0000 (17:19 +0000)]
IB/srp: Avoid endless SCSI error handling loop

If a SCSI command times out it is passed to the SCSI error
handler. The SCSI error handler will try to abort the commands that
timed out.  If aborting fails, a device reset will be attempted.  If
the device reset also fails a host reset will be attempted.  If the
host reset also fails the whole procedure will be repeated.

srp_abort() and srp_reset_device() fail for a QP in the error state.
srp_reset_host() fails after host removal has started.  Hence if the
SCSI error handler gets invoked after host removal has started and
with the QP in the error state an endless loop will be triggered.

Modify the SCSI error handling functions in ib_srp as follows:
- Abort SCSI commands properly even if the QP is in the error state.
- Make srp_reset_host() reset SCSI requests even after host removal
  has already started or if reconnecting fails.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: David Dillow <dave@thedillows.org>
Cc: <stable@vger.kernel.org> # 3.8
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/srp: Avoid sending a task management function needlessly
Bart Van Assche [Thu, 21 Feb 2013 17:18:00 +0000 (17:18 +0000)]
IB/srp: Avoid sending a task management function needlessly

Do not send a task management function if sending will fail anyway
because either there is no RDMA/RC connection or the QP is in the
error state.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: David Dillow <dave@thedillows.org>
Cc: <stable@vger.kernel.org> # 3.8
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/srp: Track connection state properly
Bart Van Assche [Thu, 21 Feb 2013 17:16:54 +0000 (17:16 +0000)]
IB/srp: Track connection state properly

Remove an assignment that incorrectly overwrites the connection state
update by srp_connect_target().

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: David Dillow <dave@thedillows.org>
Cc: <stable@vger.kernel.org> # 3.8
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/mlx4: Remove redundant NULL check before kfree
Syam Sidhardhan [Sun, 24 Feb 2013 23:20:05 +0000 (23:20 +0000)]
IB/mlx4: Remove redundant NULL check before kfree

kfree on NULL pointer is a no-op.

Signed-off-by: Syam Sidhardhan <s.syam@samsung.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/mlx4: Fix compiler warning about uninitialized 'vlan' variable
Paul Bolle [Mon, 25 Feb 2013 17:17:13 +0000 (09:17 -0800)]
IB/mlx4: Fix compiler warning about uninitialized 'vlan' variable

Building qp.o triggers this gcc warning:

    drivers/infiniband/hw/mlx4/qp.c: In function ‘mlx4_ib_post_send’:
    drivers/infiniband/hw/mlx4/qp.c:1862:62: warning: ‘vlan’ may be used uninitialized in this function [-Wmaybe-uninitialized]
    drivers/infiniband/hw/mlx4/qp.c:1752:6: note: ‘vlan’ was declared here

Looking at the code it is clear 'vlan' is only set and used if 'is_eth'
is non-zero. But by initializing 'vlan' to 0xffff, on

    gcc (Ubuntu 4.7.2-22ubuntu1) 4.7.2

on x86-64 at least, we fix the warning, and the compiler was already
setting 'vlan' to 0 in the generated code, so there's no real downside.

Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
[ Get rid of unnecessary move of 'is_vlan' initialization.  - Roland ]

Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/mlx4: Convert is_xxx variables in build_mlx_header() to bool
Roland Dreier [Mon, 25 Feb 2013 17:02:03 +0000 (09:02 -0800)]
IB/mlx4: Convert is_xxx variables in build_mlx_header() to bool

Matches the way they're used, and actually lets at least x86-64 generate
better code:

    add/remove: 0/0 grow/shrink: 0/1 up/down: 0/-38 (-38)
    function                                     old     new   delta
    mlx4_ib_post_send                           4416    4378     -38

Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/iser: Enable iser when FMRs are not supported
Or Gerlitz [Thu, 21 Feb 2013 14:50:10 +0000 (14:50 +0000)]
IB/iser: Enable iser when FMRs are not supported

Reuse the "SG unaligned for FMR" driver flow to make the initiator
functional when running over driver instance which doesn't support
FMRs, such as a mlx4 virtual function.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Alex Tabachnik <alext@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/iser: Avoid error prints on EAGAIN registration failures
Or Gerlitz [Thu, 21 Feb 2013 14:50:09 +0000 (14:50 +0000)]
IB/iser: Avoid error prints on EAGAIN registration failures

Under IO/CPU stress its possible that the FMR pool might not have a
free FMR mapping element for iSER to use because of incomplete
background unmapping processing.  In that case we get -EAGAIN and the
IO is pushed back to the SCSI layer which soon retries it.  No need to
be so verbose about that.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/iser: Use proper define for the commands per LUN value advertised to SCSI ML
Or Gerlitz [Thu, 21 Feb 2013 14:50:08 +0000 (14:50 +0000)]
IB/iser: Use proper define for the commands per LUN value advertised to SCSI ML

ISER_DEF_CMD_PER_LUN was meant to be ISCSI_DEF_XMIT_CMDS_MAX, not plain 128

Signed-off-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/uverbs: Implement memory windows support in uverbs
Shani Michaeli [Wed, 6 Feb 2013 16:19:13 +0000 (16:19 +0000)]
IB/uverbs: Implement memory windows support in uverbs

The existing user/kernel uverbs API has IB_USER_VERBS_CMD_ALLOC/DEALLOC_MW.
Implement these calls, along with destroying user memory windows during
process cleanup.

Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shani Michaeli <shanim@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/core: Add "type 2" memory windows support
Shani Michaeli [Wed, 6 Feb 2013 16:19:12 +0000 (16:19 +0000)]
IB/core: Add "type 2" memory windows support

This patch enhances the IB core support for Memory Windows (MWs).

MWs allow an application to have better/flexible control over remote
access to memory.

Two types of MWs are supported, with the second type having two flavors:

    Type 1  - associated with PD only
    Type 2A - associated with QPN only
    Type 2B - associated with PD and QPN

Applications can allocate a MW once, and then repeatedly bind the MW
to different ranges in MRs that are associated to the same PD. Type 1
windows are bound through a verb, while type 2 windows are bound by
posting a work request.

The 32-bit memory key is composed of a 24-bit index and an 8-bit
key. The key is changed with each bind, thus allowing more control
over the peer's use of the memory key.

The changes introduced are the following:

* add memory window type enum and a corresponding parameter to ib_alloc_mw.
* type 2 memory window bind work request support.
* create a struct that contains the common part of the bind verb struct
  ibv_mw_bind and the bind work request into a single struct.
* add the ib_inc_rkey helper function to advance the tag part of an rkey.

Consumer interface details:

* new device capability flags IB_DEVICE_MEM_WINDOW_TYPE_2A and
  IB_DEVICE_MEM_WINDOW_TYPE_2B are added to indicate device support
  for these features.

  Devices can set either IB_DEVICE_MEM_WINDOW_TYPE_2A or
  IB_DEVICE_MEM_WINDOW_TYPE_2B if it supports type 2A or type 2B
  memory windows. It can set neither to indicate it doesn't support
  type 2 windows at all.

* modify existing provides and consumers code to the new param of
  ib_alloc_mw and the ib_mw_bind_info structure

Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shani Michaeli <shanim@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agomlx4_core: Propagate MR deregistration failures to caller
Shani Michaeli [Wed, 6 Feb 2013 16:19:09 +0000 (16:19 +0000)]
mlx4_core: Propagate MR deregistration failures to caller

MR deregistration fails when memory windows are bound to the MR.
Handle such failures by propagating them to the caller ULP.

Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shani Michaeli <shanim@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agomlx4_core: Rename MPT-related functions to have mpt_ prefix
Shani Michaeli [Wed, 6 Feb 2013 16:19:08 +0000 (16:19 +0000)]
mlx4_core: Rename MPT-related functions to have mpt_ prefix

The MPT - Memory Protection Table - is used by both memory windows and
memory regions.  Hence, all MPT references are relevant for both types
of memory objects.  Rename the relevant functions to start with mpt_
instead of the current mr_ prefix.

Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shani Michaeli <shanim@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/mlx4_ib: Remove local invalidate segment unused fields
Shani Michaeli [Wed, 6 Feb 2013 16:19:07 +0000 (16:19 +0000)]
IB/mlx4_ib: Remove local invalidate segment unused fields

Remove unused fields from the local invalidate WQE segment structure.

Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shani Michaeli <shanim@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIPoIB: Don't attempt to release resources on error flow
Itai Garbi [Tue, 19 Feb 2013 15:40:24 +0000 (15:40 +0000)]
IPoIB: Don't attempt to release resources on error flow

If the ipoib client info isn't found on the _remove_one callback, we
must not attempt to scan the returned null list.  Found by Coverity.

Signed-off-by: Itai Garbi <igarbi@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIPoIB: Add version and firmware info to ethtool reporting
Yan Burman [Tue, 19 Feb 2013 15:40:23 +0000 (15:40 +0000)]
IPoIB: Add version and firmware info to ethtool reporting

Implement version info as well as report firmware version and bus info
of the underlying IB HW device.

Signed-off-by: Yan Burman <yanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIPoIB: Fix ipoib_neigh hashing to use the correct daddr octets
Shlomo Pongratz [Tue, 19 Feb 2013 15:40:22 +0000 (15:40 +0000)]
IPoIB: Fix ipoib_neigh hashing to use the correct daddr octets

The hash function introduced in commit b63b70d877 ("IPoIB: Use a
private hash table for path lookup in xmit path") was designd to use
the 3 octets of the IPoIB HW address that holds the remote QPN.
However, this currently isn't the case on little-endian machines,
because the the code there uses the flags part (octet[0]) and not the
last octet of the QPN (octet[3]).  Fix this.

The fix caused a checkpatch warning on line over 80 characters, to
solve that changed the name of the temp variable that holds the daddr.

Signed-off-by: Shlomo Pongratz <shlomop@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/mlx4: Adjust duplicate test
Julia Lawall [Mon, 21 Jan 2013 13:02:58 +0000 (13:02 +0000)]
IB/mlx4: Adjust duplicate test

Delete successive tests to the same location.  The code tested the result
of a previous allocation, that itself was already tested.  It is changed to
test the result of the most recent allocation.

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

// <smpl>
@s exists@
local idexpression y;
expression x,e;
@@

*if ( \(x == NULL\|IS_ERR(x)\|y != 0\) )
 { ... when forall
   return ...; }
... when != \(y = e\|y += e\|y -= e\|y |= e\|y &= e\|y++\|y--\|&y\)
    when != \(XT_GETPAGE(...,y)\|WMI_CMD_BUF(...)\)
*if ( \(x == NULL\|IS_ERR(x)\|y != 0\) )
 { ... when forall
   return ...; }
// </smpl>

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/mlx4: Fix bug unwinding on error in mlx4_ib_init_sriov()
Dan Carpenter [Mon, 4 Feb 2013 11:22:36 +0000 (11:22 +0000)]
IB/mlx4: Fix bug unwinding on error in mlx4_ib_init_sriov()

We have to decrement "i" before calling mlx4_ib_free_demux_ctx() or we
free something that wasn't allocated.  That's fine for free_pv_object()
but it would lead to a NULL dereference calling mlx4_ib_free_demux_ctx().
The null dereference is because ->tun is NULL when we check:

if (!ctx->tun[i])

Also we didn't free ->sriov.demux[0] so it was a small leak.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/amso1100: Use module_pci_driver() to simplify the code
Wei Yongjun [Wed, 10 Oct 2012 13:10:43 +0000 (13:10 +0000)]
RDMA/amso1100: Use module_pci_driver() to simplify the code

Use the module_pci_driver() macro to make the code simpler by
eliminating module_init and module_exit calls.

dpatch engine is used to auto generate this patch.
(https://github.com/weiyj/dpatch)

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Reviewed-by: Steve WIse <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Fix cast warning
Stefan Hasko [Sat, 22 Dec 2012 02:29:21 +0000 (02:29 +0000)]
RDMA/cxgb4: Fix cast warning

Fix compile warning about cast to pointer from integer of different size.

Signed-off-by: Stefan Hasko <hasko.stevo@gmail.com>
Acked-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoIB/qib: Fix QP locate/remove race
Mike Marciniszyn [Thu, 7 Feb 2013 20:47:51 +0000 (20:47 +0000)]
IB/qib: Fix QP locate/remove race

remove_qp() can execute concurrently with a qib_lookup_qpn() on
another CPU, which in of itself, is ok, given the RCU locking.

The issue is that remove_qp() NULLs out the qp->next field so that a
qib_lookup_qpn() might fail to find a qp if it occurs after the one
that is being deleted.  This is a momentary issue and subsequent
qib_lookup_qpn() calls would find the qp's since the search restarts
from the bucket head.  At scale, the issue might causes dropped
packets and unnecessary retransmissions.

The fix just deletes the qp->next NULL assignment to prevent the
remove_qp() from hiding qp's from qib_lookup_qpn().

Reviewed-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: "cookie" can stay in host endianness
Paul Bolle [Tue, 5 Feb 2013 20:51:30 +0000 (20:51 +0000)]
RDMA/cxgb4: "cookie" can stay in host endianness

Work requests are passed between the host and the firmware with a
"cookie".  This cookie is swapped to big-endian when passed to the
firmware and back to host endianness on return.  This swapping seems
to be implemented incorrectly.  Moreover, the byte swapping triggers
GCC warnings on 32 bit:

    drivers/infiniband/hw/cxgb4/cm.c: In function ‘passive_ofld_conn_reply’:
    drivers/infiniband/hw/cxgb4/cm.c:2803:12: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
    drivers/infiniband/hw/cxgb4/cm.c: In function ‘send_fw_pass_open_req’:
    drivers/infiniband/hw/cxgb4/cm.c:2941:16: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
    [...]

But byte swapping isn't needed as the firmware doesn't actually touch
the cookie.  Dropping byte swapping makes the warnings go away too.

Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Address sparse warnings
Vipul Pandya [Mon, 7 Jan 2013 13:12:00 +0000 (13:12 +0000)]
RDMA/cxgb4: Address sparse warnings

Fixe the following types of sparse warnings
- cast to pointer from integer of different size
- cast from pointer to integer of different size
- incorrect type in assignment (different base types)
- incorrect type in argument 1 (different base types)
- cast from restricted __be64
- cast from restricted __be32

Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Insert hwtid in pass_accept_req instead in pass_establish
Vipul Pandya [Mon, 7 Jan 2013 13:11:59 +0000 (13:11 +0000)]
RDMA/cxgb4: Insert hwtid in pass_accept_req instead in pass_establish

CPL_ABORT_REQ_RSS can come before TCP connection is established.  In
such case peer_abort was trying to remove the hwtid, which was not
inserted.  To avoid this we insert the hwtid when we are sure that we
are surely going to send passive accept request.

Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Don't wakeup threads for MPAv2
Vipul Pandya [Mon, 7 Jan 2013 13:11:58 +0000 (13:11 +0000)]
RDMA/cxgb4: Don't wakeup threads for MPAv2

Don't wakeup threads blocked in rdma_init/rdma_fini if we are on
MPAv2, and want to retry connection with MPAv1.

Stop ep-timer on getting MPA version mismatch, before doing the
abort_connection - in process_mpa_request.

Take care to stop ep-timer in error paths for process_mpa_request.

Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Don't reconnect on abort for mpa_rev 1
Vipul Pandya [Mon, 7 Jan 2013 13:11:57 +0000 (13:11 +0000)]
RDMA/cxgb4: Don't reconnect on abort for mpa_rev 1

Only reconnect if the endpoint wasn't freed.

peer_abort() should only attempt to reconnect if the endpoint wasn't
freed.  Also remove hwtid from the debugfs idr.

Add missing check for peer2peer in MPAv2 code

Use correct mpa version on reject.

Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Fix endpoint timeout race condition
Vipul Pandya [Mon, 7 Jan 2013 13:11:56 +0000 (13:11 +0000)]
RDMA/cxgb4: Fix endpoint timeout race condition

The endpoint timeout logic had a race that could cause an endpoint
object to be freed while it was still on the timedout list.  This
can happen if the timer is stopped after it had fired, but before
the timedout thread processed the endpoint timeout.

Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Only log rx_data warnings if cpl status is non-zero
Vipul Pandya [Mon, 7 Jan 2013 13:11:55 +0000 (13:11 +0000)]
RDMA/cxgb4: Only log rx_data warnings if cpl status is non-zero

With newer firmware, we can get streaming data due to connection
errors before the driver moves the QP out of RTS.

Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Always log async errors
Vipul Pandya [Mon, 7 Jan 2013 13:11:54 +0000 (13:11 +0000)]
RDMA/cxgb4: Always log async errors

Log AEs even if the QP isn't in RTS.  It is useful information.

Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Keep QP referenced until TID released
Vipul Pandya [Mon, 7 Jan 2013 13:11:53 +0000 (13:11 +0000)]
RDMA/cxgb4: Keep QP referenced until TID released

The driver is currently releasing the last ref on the QP too early.
This can cause bus errors due to HW still fetching WRs from the HW
queue.  The fix is to keep a qp ref until we release the HW TID.

Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Display streaming mode error only if detected in RTS
Vipul Pandya [Mon, 7 Jan 2013 13:11:52 +0000 (13:11 +0000)]
RDMA/cxgb4: Display streaming mode error only if detected in RTS

With later firmware, the chances of getting streaming mode data after
we exit RTS is likely, so we don't need to warn for it.  The only real
case where we don't expect it is when the QP is in RTS.

Move QP to ERROR when streaming mode data received.

Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Abort connections when moving to ERROR state
Vipul Pandya [Mon, 7 Jan 2013 13:11:51 +0000 (13:11 +0000)]
RDMA/cxgb4: Abort connections when moving to ERROR state

If a FINI operation fails, then we need to ABORT instead of CLOSE.
Also, if we ABORT due to unexpected STREAMING data, then wake up
anybody blocked in FINI...

Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoRDMA/cxgb4: Abort connections that receive unexpected streaming mode data
Vipul Pandya [Mon, 7 Jan 2013 13:11:50 +0000 (13:11 +0000)]
RDMA/cxgb4: Abort connections that receive unexpected streaming mode data

This error means the RDMA connection was knocked out of RDMA mode,
probably due to an error on the connection.

Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
8 years agoLinux 3.8-rc7
Linus Torvalds [Fri, 8 Feb 2013 21:20:39 +0000 (08:20 +1100)]
Linux 3.8-rc7

8 years agoMerge branch 'fixes' of git://git.linaro.org/people/rmk/linux-arm
Linus Torvalds [Fri, 8 Feb 2013 21:01:18 +0000 (08:01 +1100)]
Merge branch 'fixes' of git://git.linaro.org/people/rmk/linux-arm

Pull ARM fixes from Russell King:
 "I was going to hold these off until v3.8 was out, and send them with a
  stable tag, but as everyone else is pushing much bigger fixes which
  Linus is accepting, let's save people from the hastle of having to
  patch v3.8 back into working or use a stable kernel.

  Looking at the diffstat, this really is high value for its size; this
  is miniscule compared to how the -rc6 to tip diffstat currently looks.

  So, four patches in this set:
   - Punit Agrawal reports that the kernel no longer boots on MPCore due
     to a new assumption made in the GIC code which isn't true of
     earlier GIC designs.  This is the biggest change in this set.
   - Punit's boot log also revealed a bunch of WARN_ON() dumps caused by
     the DT-ification of the GIC support without fixing up non-DT
     Realview - which now sees a greater number of interrupts than it
     did before.
   - A fix for the DMA coherent code from Marek which uses the wrong
     check for atomic allocations; this can result in spinlock lockups
     or other nasty effects.
   - A fix from Will, which will affect all Android based platforms if
     not applied (which use the 2G:2G VM split) - this causes
     particularly 'make' to misbehave unless this bug is fixed."

* 'fixes' of git://git.linaro.org/people/rmk/linux-arm:
  ARM: 7641/1: memory: fix broken mmap by ensuring TASK_UNMAPPED_BASE is aligned
  ARM: DMA mapping: fix bad atomic test
  ARM: realview: ensure that we have sufficient IRQs available
  ARM: GIC: fix GIC cpumask initialization

8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Fri, 8 Feb 2013 20:55:24 +0000 (07:55 +1100)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Revert iwlwifi reclaimed packet tracking, it causes problems for a
    bunch of folks.  From Emmanuel Grumbach.

 2) Work limiting code in brcmsmac wifi driver can clear tx status
    without processing the event.  From Arend van Spriel.

 3) rtlwifi USB driver processes wrong SKB, fix from Larry Finger.

 4) l2tp tunnel delete can race with close, fix from Tom Parkin.

 5) pktgen_add_device() failures are not checked at all, fix from Cong
    Wang.

 6) Fix unintentional removal of carrier off from tun_detach(),
    otherwise we confuse userspace, from Michael S.  Tsirkin.

 7) Don't leak socket reference counts and ubufs in vhost-net driver,
    from Jason Wang.

 8) vmxnet3 driver gets it's initial carrier state wrong, fix from Neil
    Horman.

 9) Protect against USB networking devices which spam the host with 0
    length frames, from Bjørn Mork.

10) Prevent neighbour overflows in ipv6 for locally destined routes,
    from Marcelo Ricardo.  This is the best short-term fix for this, a
    longer term fix has been implemented in net-next.

11) L2TP uses ipv4 datagram routines in it's ipv6 code, whoops.  This
    mistake is largely because the ipv6 functions don't even have some
    kind of prefix in their names to suggest they are ipv6 specific.
    From Tom Parkin.

12) Check SYN packet drops properly in tcp_rcv_fastopen_synack(), from
    Yuchung Cheng.

13) Fix races and TX skb freeing bugs in via-rhine's NAPI support, from
    Francois Romieu and your's truly.

14) Fix infinite loops and divides by zero in TCP congestion window
    handling, from Eric Dumazet, Neal Cardwell, and Ilpo Järvinen.

15) AF_PACKET tx ring handling can leak kernel memory to userspace, fix
    from Phil Sutter.

16) Fix error handling in ipv6 GRE tunnel transmit, from Tommi Rantala.

17) Protect XEN netback driver against hostile frontend putting garbage
    into the rings, don't leak pages in TX GOP checking, and add proper
    resource releasing in error path of xen_netbk_get_requests().  From
    Ian Campbell.

18) SCTP authentication keys should be cleared out and released with
    kzfree(), from Daniel Borkmann.

19) L2TP is a bit too clever trying to maintain skb->truesize, and ends
    up corrupting socket memory accounting to the point where packet
    sending is halted indefinitely.  Just remove the adjustments
    entirely, they aren't really needed.  From Eric Dumazet.

20) ATM Iphase driver uses a data type with the same name as the S390
    headers, rename to fix the build.  From Heiko Carstens.

21) Fix a typo in copying the inner network header offset from one SKB
    to another, from Pravin B Shelar.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (56 commits)
  net: sctp: sctp_endpoint_free: zero out secret key data
  net: sctp: sctp_setsockopt_auth_key: use kzfree instead of kfree
  atm/iphase: rename fregt_t -> ffreg_t
  net: usb: fix regression from FLAG_NOARP code
  l2tp: dont play with skb->truesize
  net: sctp: sctp_auth_key_put: use kzfree instead of kfree
  netback: correct netbk_tx_err to handle wrap around.
  xen/netback: free already allocated memory on failure in xen_netbk_get_requests
  xen/netback: don't leak pages on failure in xen_netbk_tx_check_gop.
  xen/netback: shutdown the ring if it contains garbage.
  net: qmi_wwan: add more Huawei devices, including E320
  net: cdc_ncm: add another Huawei vendor specific device
  ipv6/ip6_gre: fix error case handling in ip6gre_tunnel_xmit()
  tcp: fix for zero packets_in_flight was too broad
  brcmsmac: rework of mac80211 .flush() callback operation
  ssb: unregister gpios before unloading ssb
  bcma: unregister gpios before unloading bcma
  rtlwifi: Fix scheduling while atomic bug
  net: usbnet: fix tx_dropped statistics
  tcp: ipv6: Update MIB counters for drops
  ...

8 years agoMerge branch 'sctp_keys'
David S. Miller [Fri, 8 Feb 2013 19:55:08 +0000 (14:55 -0500)]
Merge branch 'sctp_keys'

Daniel Borkmann says:

====================
Cryptographically used keys should be zeroed out when our session
ends resp. memory is freed, thus do not leave them somewhere in the
memory.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: sctp: sctp_endpoint_free: zero out secret key data
Daniel Borkmann [Fri, 8 Feb 2013 03:04:35 +0000 (03:04 +0000)]
net: sctp: sctp_endpoint_free: zero out secret key data

On sctp_endpoint_destroy, previously used sensitive keying material
should be zeroed out before the memory is returned, as we already do
with e.g. auth keys when released.

Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Vlad Yasevich <vyasevic@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: sctp: sctp_setsockopt_auth_key: use kzfree instead of kfree
Daniel Borkmann [Fri, 8 Feb 2013 03:04:34 +0000 (03:04 +0000)]
net: sctp: sctp_setsockopt_auth_key: use kzfree instead of kfree

In sctp_setsockopt_auth_key, we create a temporary copy of the user
passed shared auth key for the endpoint or association and after
internal setup, we free it right away. Since it's sensitive data, we
should zero out the key before returning the memory back to the
allocator. Thus, use kzfree instead of kfree, just as we do in
sctp_auth_key_put().

Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoatm/iphase: rename fregt_t -> ffreg_t
Heiko Carstens [Fri, 8 Feb 2013 00:19:11 +0000 (00:19 +0000)]
atm/iphase: rename fregt_t -> ffreg_t

We have conflicting type qualifiers for "freg_t" in s390's ptrace.h and the
iphase atm device driver, which causes the compile error below.
Unfortunately the s390 typedef can't be renamed, since it's a user visible api,
nor can I change the include order in s390 code to avoid the conflict.

So simply rename the iphase typedef to a new name. Fixes this compile error:

In file included from drivers/atm/iphase.c:66:0:
drivers/atm/iphase.h:639:25: error: conflicting type qualifiers for 'freg_t'
In file included from next/arch/s390/include/asm/ptrace.h:9:0,
                 from next/arch/s390/include/asm/lowcore.h:12,
                 from next/arch/s390/include/asm/thread_info.h:30,
                 from include/linux/thread_info.h:54,
                 from include/linux/preempt.h:9,
                 from include/linux/spinlock.h:50,
                 from include/linux/seqlock.h:29,
                 from include/linux/time.h:5,
                 from include/linux/stat.h:18,
                 from include/linux/module.h:10,
                 from drivers/atm/iphase.c:43:
next/arch/s390/include/uapi/asm/ptrace.h:197:3: note: previous declaration of 'freg_t' was here

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Acked-by: chas williams - CONTRACTOR <chas@cmf.nrl.navy.mil>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoARM: 7641/1: memory: fix broken mmap by ensuring TASK_UNMAPPED_BASE is aligned
Will Deacon [Fri, 8 Feb 2013 11:52:29 +0000 (12:52 +0100)]
ARM: 7641/1: memory: fix broken mmap by ensuring TASK_UNMAPPED_BASE is aligned

We have received multiple reports of mmap failures when running with a
2:2 vm split. These manifest as either -EINVAL with a non page-aligned
address (ending 0xaaa) or a SEGV, depending on the application. The
issue is commonly observed in children of make, which appears to use
bottom-up mmap (assumedly because it changes the stack rlimit).

Further investigation reveals that this regression was triggered by
394ef6403abc ("mm: use vm_unmapped_area() on arm architecture"), whereby
TASK_UNMAPPED_BASE is no longer page-aligned for bottom-up mmap, causing
get_unmapped_area to choke on misaligned addressed.

This patch fixes the problem by defining TASK_UNMAPPED_BASE in terms of
TASK_SIZE and explicitly aligns the result to 16M, matching the other
end of the heap.

Acked-by: Nicolas Pitre <nico@linaro.org>
Reported-by: Steve Capper <steve.capper@arm.com>
Reported-by: Jean-Francois Moine <moinejf@free.fr>
Reported-by: Christoffer Dall <cdall@cs.columbia.edu>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
8 years agoARM: DMA mapping: fix bad atomic test
Russell King [Wed, 30 Jan 2013 23:55:35 +0000 (23:55 +0000)]
ARM: DMA mapping: fix bad atomic test

Realview fails to boot with this warning:
BUG: spinlock lockup suspected on CPU#0, init/1
 lock: 0xcf8bde10, .magic: dead4ead, .owner: init/1, .owner_cpu: 0
Backtrace:
[<c00185d8>] (dump_backtrace+0x0/0x10c) from [<c03294e8>] (dump_stack+0x18/0x1c) r6:cf8bde10 r5:cf83d1c0 r4:cf8bde10 r3:cf83d1c0
[<c03294d0>] (dump_stack+0x0/0x1c) from [<c018926c>] (spin_dump+0x84/0x98)
[<c01891e8>] (spin_dump+0x0/0x98) from [<c0189460>] (do_raw_spin_lock+0x100/0x198)
[<c0189360>] (do_raw_spin_lock+0x0/0x198) from [<c032cbac>] (_raw_spin_lock+0x3c/0x44)
[<c032cb70>] (_raw_spin_lock+0x0/0x44) from [<c01c9224>] (pl011_console_write+0xe8/0x11c)
[<c01c913c>] (pl011_console_write+0x0/0x11c) from [<c002aea8>] (call_console_drivers.clone.7+0xdc/0x104)
[<c002adcc>] (call_console_drivers.clone.7+0x0/0x104) from [<c002b320>] (console_unlock+0x2e8/0x454)
[<c002b038>] (console_unlock+0x0/0x454) from [<c002b8b4>] (vprintk_emit+0x2d8/0x594)
[<c002b5dc>] (vprintk_emit+0x0/0x594) from [<c0329718>] (printk+0x3c/0x44)
[<c03296dc>] (printk+0x0/0x44) from [<c002929c>] (warn_slowpath_common+0x28/0x6c)
[<c0029274>] (warn_slowpath_common+0x0/0x6c) from [<c0029304>] (warn_slowpath_null+0x24/0x2c)
[<c00292e0>] (warn_slowpath_null+0x0/0x2c) from [<c0070ab0>] (lockdep_trace_alloc+0xd8/0xf0)
[<c00709d8>] (lockdep_trace_alloc+0x0/0xf0) from [<c00c0850>] (kmem_cache_alloc+0x24/0x11c)
[<c00c082c>] (kmem_cache_alloc+0x0/0x11c) from [<c00bb044>] (__get_vm_area_node.clone.24+0x7c/0x16c)
[<c00bafc8>] (__get_vm_area_node.clone.24+0x0/0x16c) from [<c00bb7b8>] (get_vm_area_caller+0x48/0x54)
[<c00bb770>] (get_vm_area_caller+0x0/0x54) from [<c0020064>] (__alloc_remap_buffer.clone.15+0x38/0xb8)
[<c002002c>] (__alloc_remap_buffer.clone.15+0x0/0xb8) from [<c0020244>] (__dma_alloc+0x160/0x2c8)
[<c00200e4>] (__dma_alloc+0x0/0x2c8) from [<c00204d8>] (arm_dma_alloc+0x88/0xa0)[<c0020450>] (arm_dma_alloc+0x0/0xa0) from [<c00beb00>] (dma_pool_alloc+0xcc/0x1a8)
[<c00bea34>] (dma_pool_alloc+0x0/0x1a8) from [<c01a9d14>] (pl08x_fill_llis_for_desc+0x28/0x568)
[<c01a9cec>] (pl08x_fill_llis_for_desc+0x0/0x568) from [<c01aab8c>] (pl08x_prep_slave_sg+0x258/0x3b0)
[<c01aa934>] (pl08x_prep_slave_sg+0x0/0x3b0) from [<c01c9f74>] (pl011_dma_tx_refill+0x140/0x288)
[<c01c9e34>] (pl011_dma_tx_refill+0x0/0x288) from [<c01ca748>] (pl011_start_tx+0xe4/0x120)
[<c01ca664>] (pl011_start_tx+0x0/0x120) from [<c01c54a4>] (__uart_start+0x48/0x4c)
[<c01c545c>] (__uart_start+0x0/0x4c) from [<c01c632c>] (uart_start+0x2c/0x3c)
[<c01c6300>] (uart_start+0x0/0x3c) from [<c01c795c>] (uart_write+0xcc/0xf4)
[<c01c7890>] (uart_write+0x0/0xf4) from [<c01b0384>] (n_tty_write+0x1c0/0x3e4)
[<c01b01c4>] (n_tty_write+0x0/0x3e4) from [<c01acfe8>] (tty_write+0x144/0x240)
[<c01acea4>] (tty_write+0x0/0x240) from [<c01ad17c>] (redirected_tty_write+0x98/0xac)
[<c01ad0e4>] (redirected_tty_write+0x0/0xac) from [<c00c371c>] (vfs_write+0xbc/0x150)
[<c00c3660>] (vfs_write+0x0/0x150) from [<c00c39c0>] (sys_write+0x4c/0x78)
[<c00c3974>] (sys_write+0x0/0x78) from [<c0014460>] (ret_fast_syscall+0x0/0x3c)

This happens because the DMA allocation code is not respecting atomic
allocations correctly.

GFP flags should not be tested for GFP_ATOMIC to determine if an
atomic allocation is being requested.  GFP_ATOMIC is not a flag but
a value.  The GFP bitmask flags are all prefixed with __GFP_.

The rest of the kernel tests for __GFP_WAIT not being set to indicate
an atomic allocation.  We need to do the same.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
8 years agoARM: realview: ensure that we have sufficient IRQs available
Russell King [Wed, 30 Jan 2013 23:54:26 +0000 (23:54 +0000)]
ARM: realview: ensure that we have sufficient IRQs available

Realview EB with a rev B MPcore tile results in lots of warnings at
boot because it can't allocate enough IRQs.  Fix this by increasing
the number of available IRQs.

WARNING: at /home/rmk/git/linux-rmk/arch/arm/common/gic.c:757 gic_init_bases+0x12c/0x2ec()
Cannot allocate irq_descs @ IRQ96, assuming pre-allocated
Modules linked in:
Backtrace:
[<c00185d8>] (dump_backtrace+0x0/0x10c) from [<c03294e8>] (dump_stack+0x18/0x1c) r6:000002f5 r5:c042c62c r4:c044ff40 r3:c045f240
[<c03294d0>] (dump_stack+0x0/0x1c) from [<c00292c8>] (warn_slowpath_common+0x54/0x6c)
[<c0029274>] (warn_slowpath_common+0x0/0x6c) from [<c0029384>] (warn_slowpath_fmt+0x38/0x40)
[<c002934c>] (warn_slowpath_fmt+0x0/0x40) from [<c042c62c>] (gic_init_bases+0x12c/0x2ec)
[<c042c500>] (gic_init_bases+0x0/0x2ec) from [<c042cdc8>] (gic_init_irq+0x8c/0xd8)
[<c042cd3c>] (gic_init_irq+0x0/0xd8) from [<c042827c>] (init_IRQ+0x1c/0x24)
[<c0428260>] (init_IRQ+0x0/0x24) from [<c04256c8>] (start_kernel+0x1a4/0x300)
[<c0425524>] (start_kernel+0x0/0x300) from [<70008070>] (0x70008070)
---[ end trace 1b75b31a2719ed1c ]---
------------[ cut here ]------------
WARNING: at /home/rmk/git/linux-rmk/kernel/irq/irqdomain.c:234 irq_domain_add_legacy+0x80/0x140()
Modules linked in:
Backtrace:
[<c00185d8>] (dump_backtrace+0x0/0x10c) from [<c03294e8>] (dump_stack+0x18/0x1c) r6:000000ea r5:c0081a38 r4:00000000 r3:c045f240
[<c03294d0>] (dump_stack+0x0/0x1c) from [<c00292c8>] (warn_slowpath_common+0x54/0x6c)
[<c0029274>] (warn_slowpath_common+0x0/0x6c) from [<c0029304>] (warn_slowpath_null+0x24/0x2c)
[<c00292e0>] (warn_slowpath_null+0x0/0x2c) from [<c0081a38>] (irq_domain_add_legacy+0x80/0x140)
[<c00819b8>] (irq_domain_add_legacy+0x0/0x140) from [<c042c64c>] (gic_init_bases+0x14c/0x2ec)
[<c042c500>] (gic_init_bases+0x0/0x2ec) from [<c042cdc8>] (gic_init_irq+0x8c/0xd8)
[<c042cd3c>] (gic_init_irq+0x0/0xd8) from [<c042827c>] (init_IRQ+0x1c/0x24)
[<c0428260>] (init_IRQ+0x0/0x24) from [<c04256c8>] (start_kernel+0x1a4/0x300)
[<c0425524>] (start_kernel+0x0/0x300) from [<70008070>] (0x70008070)
---[ end trace 1b75b31a2719ed1d ]---
------------[ cut here ]------------
WARNING: at /home/rmk/git/linux-rmk/arch/arm/common/gic.c:762 gic_init_bases+0x170/0x2ec()
Modules linked in:
Backtrace:
[<c00185d8>] (dump_backtrace+0x0/0x10c) from [<c03294e8>] (dump_stack+0x18/0x1c) r6:000002fa r5:c042c670 r4:00000000 r3:c045f240
[<c03294d0>] (dump_stack+0x0/0x1c) from [<c00292c8>] (warn_slowpath_common+0x54/0x6c)
[<c0029274>] (warn_slowpath_common+0x0/0x6c) from [<c0029304>] (warn_slowpath_null+0x24/0x2c)
[<c00292e0>] (warn_slowpath_null+0x0/0x2c) from [<c042c670>] (gic_init_bases+0x170/0x2ec)
[<c042c500>] (gic_init_bases+0x0/0x2ec) from [<c042cdc8>] (gic_init_irq+0x8c/0xd8)
[<c042cd3c>] (gic_init_irq+0x0/0xd8) from [<c042827c>] (init_IRQ+0x1c/0x24)
[<c0428260>] (init_IRQ+0x0/0x24) from [<c04256c8>] (start_kernel+0x1a4/0x300)
[<c0425524>] (start_kernel+0x0/0x300) from [<70008070>] (0x70008070)
---[ end trace 1b75b31a2719ed1e ]---

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
8 years agoARM: GIC: fix GIC cpumask initialization
Russell King [Wed, 30 Jan 2013 23:49:57 +0000 (23:49 +0000)]
ARM: GIC: fix GIC cpumask initialization

Punit Agrawal reports:
> I was trying to boot 3.8-rc5 on Realview EB 11MPCore using
> realview-smp_defconfig as a starting point but the kernel failed to
> progress past the log below (config attached).
>
> Pawel suggested I try reverting 384a290283f - "ARM: gic: use a private
> mapping for CPU target interfaces" that you've authored. With this
> commit reverted the kernel boots.
>
> I am not quite sure why the commit breaks 11MPCore but Pawel (cc'd)
> might be able to shed light on that.

Some early GIC implementations return zero for the first distributor
CPU routing register.  This means we can't rely on that telling us
which CPU interface we're connected to.  We know that these platforms
implement PPIs for IRQs 29-31 - but we shouldn't assume that these
will always be populated.

So, instead, scan for a non-zero CPU routing register in the first
32 IRQs and use that as our CPU mask.

Reported-by: Punit Agrawal <punit.agrawal@arm.com>
Reviewed-by: Nicolas Pitre <nico@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
8 years agoMerge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Fri, 8 Feb 2013 08:46:26 +0000 (19:46 +1100)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull drm regression fix from Dave Airlie:
 "This one fixes a sleep while locked regression that was introduced
  earlier in 3.8."

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm/ttm: fix fence locking in ttm_buffer_object_transfer, 2nd try

8 years agonet: usb: fix regression from FLAG_NOARP code
Lucas Stach [Thu, 7 Feb 2013 16:18:39 +0000 (16:18 +0000)]
net: usb: fix regression from FLAG_NOARP code

In commit 6509141f9c2ba74df6cc72ec35cd1865276ae3a4 ("usbnet: add new
flag FLAG_NOARP for usb net devices"), the newly added flag NOARP was
using an already defined value, which broke drivers using flag
MULTI_PACKET.

Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agol2tp: dont play with skb->truesize
Eric Dumazet [Thu, 7 Feb 2013 14:00:34 +0000 (14:00 +0000)]
l2tp: dont play with skb->truesize

Andrew Savchenko reported a DNS failure and we diagnosed that
some UDP sockets were unable to send more packets because their
sk_wmem_alloc was corrupted after a while (tx_queue column in
following trace)

$ cat /proc/net/udp
  sl  local_address rem_address   st tx_queue rx_queue tr tm->when retrnsmt   uid  timeout inode ref pointer drops
...
  459: 00000000:0270 00000000:0000 07 00000000:00000000 00:00000000 00000000     0        0 4507 2 ffff88003d612380 0
  466: 00000000:0277 00000000:0000 07 00000000:00000000 00:00000000 00000000     0        0 4802 2 ffff88003d613180 0
  470: 076A070A:007B 00000000:0000 07 FFFF4600:00000000 00:00000000 00000000   123        0 5552 2 ffff880039974380 0
  470: 010213AC:007B 00000000:0000 07 00000000:00000000 00:00000000 00000000     0        0 4986 2 ffff88003dbd3180 0
  470: 010013AC:007B 00000000:0000 07 00000000:00000000 00:00000000 00000000     0        0 4985 2 ffff88003dbd2e00 0
  470: 00FCA8C0:007B 00000000:0000 07 FFFFFB00:00000000 00:00000000 00000000     0        0 4984 2 ffff88003dbd2a80 0
...

Playing with skb->truesize is tricky, especially when
skb is attached to a socket, as we can fool memory charging.

Just remove this code, its not worth trying to be ultra
precise in xmit path.

Reported-by: Andrew Savchenko <bircoph@gmail.com>
Tested-by: Andrew Savchenko <bircoph@gmail.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: James Chapman <jchapman@katalix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: sctp: sctp_auth_key_put: use kzfree instead of kfree
Daniel Borkmann [Thu, 7 Feb 2013 00:55:37 +0000 (00:55 +0000)]
net: sctp: sctp_auth_key_put: use kzfree instead of kfree

For sensitive data like keying material, it is common practice to zero
out keys before returning the memory back to the allocator. Thus, use
kzfree instead of kfree.

Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswit...
David S. Miller [Fri, 8 Feb 2013 04:37:36 +0000 (23:37 -0500)]
Merge branch 'fixes' of git://git./linux/kernel/git/jesse/openvswitch into openvswitch

Jesse Gross says:

====================
One bug fix for net/3.8 for a long standing problem that was reported a few
times recently.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'netback'
David S. Miller [Fri, 8 Feb 2013 04:31:47 +0000 (23:31 -0500)]
Merge branch 'netback'

Ian Campbell says:

====================
The Xen netback implementation contains a couple of flaws which can
allow a guest to cause a DoS in the backend domain, potentially
affecting other domains in the system.

CVE-2013-0216 is a failure to sanity check the ring producer/consumer
pointers which can allow a guest to cause netback to loop for an
extended period preventing other work from occurring.

CVE-2013-0217 is a memory leak on an error path which is guest
triggerable.

The following series contains the fixes for these issues, as previously
included in Xen Security Advisory 39:
http://lists.xen.org/archives/html/xen-announce/2013-02/msg00001.html

Changes in v2:
 - Typo and block comment format fixes
 - Added stable Cc
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonetback: correct netbk_tx_err to handle wrap around.
Ian Campbell [Wed, 6 Feb 2013 23:41:38 +0000 (23:41 +0000)]
netback: correct netbk_tx_err to handle wrap around.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Jan Beulich <JBeulich@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoxen/netback: free already allocated memory on failure in xen_netbk_get_requests
Ian Campbell [Wed, 6 Feb 2013 23:41:37 +0000 (23:41 +0000)]
xen/netback: free already allocated memory on failure in xen_netbk_get_requests

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoxen/netback: don't leak pages on failure in xen_netbk_tx_check_gop.
Matthew Daley [Wed, 6 Feb 2013 23:41:36 +0000 (23:41 +0000)]
xen/netback: don't leak pages on failure in xen_netbk_tx_check_gop.

Signed-off-by: Matthew Daley <mattjd@gmail.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Jan Beulich <JBeulich@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoxen/netback: shutdown the ring if it contains garbage.
Ian Campbell [Wed, 6 Feb 2013 23:41:35 +0000 (23:41 +0000)]
xen/netback: shutdown the ring if it contains garbage.

A buggy or malicious frontend should not be able to confuse netback.
If we spot anything which is not as it should be then shutdown the
device and don't try to continue with the ring in a potentially
hostile state. Well behaved and non-hostile frontends will not be
penalised.

As well as making the existing checks for such errors fatal also add a
new check that ensures that there isn't an insane number of requests
on the ring (i.e. more than would fit in the ring). If the ring
contains garbage then previously is was possible to loop over this
insane number, getting an error each time and therefore not generating
any more pending requests and therefore not exiting the loop in
xen_netbk_tx_build_gops for an externded period.

Also turn various netdev_dbg calls which no precipitate a fatal error
into netdev_err, they are rate limited because the device is shutdown
afterwards.

This fixes at least one known DoS/softlockup of the backend domain.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Acked-by: Jan Beulich <JBeulich@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty...
Linus Torvalds [Fri, 8 Feb 2013 01:22:30 +0000 (12:22 +1100)]
Merge tag 'fixes-for-linus' of git://git./linux/kernel/git/rusty/linux

Pull virtio fix from Rusty Russell:
 "Obviously I forgot to push this before linux.conf.au..."

* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux:
  virtio_console: Don't access uninitialized data.

8 years agoMerge tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Fri, 8 Feb 2013 01:15:14 +0000 (12:15 +1100)]
Merge tag 'rdma-for-linus' of git://git./linux/kernel/git/roland/infiniband

Pull IB regression fixes from Roland Dreier:

 - Fix mlx4 VFs not working on old guests because of 64B CQE changes

 - Fix ill-considered sparse fix for qib

 - Fix IPoIB crash due to skb double destruct introduced in 3.8-rc1

* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IB/qib: Fix for broken sparse warning fix
  mlx4_core: Fix advertisement of wrong PF context behaviour
  IPoIB: Fix crash due to skb double destruct

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux...
Linus Torvalds [Fri, 8 Feb 2013 01:06:46 +0000 (12:06 +1100)]
Merge branch 'for-linus' of git://git./linux/kernel/git/mason/linux-btrfs

Pull btrfs fixes from Chris Mason:
 "We've got corner cases for updating i_size that ceph was hitting,
  error handling for quotas when we run out of space, a very subtle
  snapshot deletion race, a crash while removing devices, and one
  deadlock between subvolume creation and the sb_internal code (thanks
  lockdep)."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs: move d_instantiate outside the transaction during mksubvol
  Btrfs: fix EDQUOT handling in btrfs_delalloc_reserve_metadata
  Btrfs: fix possible stale data exposure
  Btrfs: fix missing i_size update
  Btrfs: fix race between snapshot deletion and getting inode
  Btrfs: fix missing release of the space/qgroup reservation in start_transaction()
  Btrfs: fix wrong sync_writers decrement in btrfs_file_aio_write()
  Btrfs: do not merge logged extents if we've removed them from the tree
  btrfs: don't try to notify udev about missing devices

8 years agoMerge tag 'pinctrl-for-v3.8-late' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 8 Feb 2013 01:00:44 +0000 (12:00 +1100)]
Merge tag 'pinctrl-for-v3.8-late' of git://git./linux/kernel/git/linusw/linux-pinctrl

Pull late pinctrl fixes from Linus Walleij:
 "Two patches appeared as of late, one was completely news to me, the
  other one was rotated in -next for the next merge window but turned
  out to be a showstopper.

   - Exynos Kconfig fixup
   - SIRF DT translation bug"

* tag 'pinctrl-for-v3.8-late' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
  pinctrl: sirf: replace of_gpio_simple_xlate by sirf specific of_xlate
  pinctrl: exynos: change PINCTRL_EXYNOS option

8 years agoMerge tag 'stable/for-linus-3.8-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 8 Feb 2013 00:55:27 +0000 (11:55 +1100)]
Merge tag 'stable/for-linus-3.8-rc6-tag' of git://git./linux/kernel/git/konrad/xen

Pull Xen fixes from Konrad Rzeszutek Wilk:
 "This has two fixes.  One is a security fix wherein we would spam the
  kernel printk buffer if one of the guests was misbehaving.  The other
  is much tamer and it was us only checking for one type of error from
  the IRQ subsystem (when allocating new IRQs) instead of for all of
  them.

   - Fix an IRQ allocation where we only check for a specific error (-1).
   - CVE-2013-0231 / XSA-43.  Make xen-pciback rate limit error messages
     from xen_pcibk_enable_msi{,x}()"

* tag 'stable/for-linus-3.8-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
  xen: fix error handling path if xen_allocate_irq_dynamic fails
  xen-pciback: rate limit error messages from xen_pcibk_enable_msi{,x}()

8 years agoMerge tag 'regulator-v3.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/broon...
Linus Torvalds [Fri, 8 Feb 2013 00:52:53 +0000 (11:52 +1100)]
Merge tag 'regulator-v3.8-rc6' of git://git./linux/kernel/git/broonie/regulator

Pull regulator fixes from Mark Brown:
 "Mostly driver specific fixes here, though one of them uncovered the
  issue Stephen Warren fixed with multiple OF matches getting upset due
  to a lack of cleanup."

* tag 'regulator-v3.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
  regulator: s2mps11: fix incorrect register for buck10
  regulator: clear state each invocation of of_regulator_match
  regulator: max8997: Fix using wrong dev argument at various places
  regulator: max77686: Fix using wrong dev argument at various places
  regulator: max8907: Fix using wrong dev argument for calling of_regulator_match
  regulator: max8998: fix incorrect min_uV value for ldo10
  regulator: tps65910: Fix using wrong dev argument for calling of_regulator_match
  regulator: tps65217: Fix using wrong dev argument for calling of_regulator_match

8 years agodrm/ttm: fix fence locking in ttm_buffer_object_transfer, 2nd try
Daniel Vetter [Mon, 14 Jan 2013 14:08:14 +0000 (15:08 +0100)]
drm/ttm: fix fence locking in ttm_buffer_object_transfer, 2nd try

This fixes up

commit e8e89622ed361c46bf90ba4828e685a8b603f7e5
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Tue Dec 18 22:25:11 2012 +0100

    drm/ttm: fix fence locking in ttm_buffer_object_transfer

which leaves behind a might_sleep in atomic context, since the
fence_lock spinlock is held over a kmalloc(GFP_KERNEL) call. The fix
is to revert the above commit and only take the lock where we need it,
around the call to ->sync_obj_ref.

v2: Fixup things noticed by Maarten Lankhorst:
- Brown paper bag locking bug.
- No need for kzalloc if we clear the entire thing on the next line.
- check for bo->sync_obj (totally unlikely race, but still someone
  else could have snuck in) and clear fbo->sync_obj if it's cleared
  already.

Reported-by: Dave Airlie <airlied@gmail.com>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
8 years agovirtio_console: Don't access uninitialized data.
Sjur Brændeland [Mon, 21 Jan 2013 23:20:26 +0000 (09:50 +1030)]
virtio_console: Don't access uninitialized data.

Don't access uninitialized work-queue when removing device.
The work queue is initialized only if the device multi-queue.
So don't call cancel_work unless this is a multi-queue device.

This fixes the following panic:

Kernel panic - not syncing: BUG!
Call Trace:
62031b28:  [<6026085d>] panic+0x16b/0x2d3
62031b30:  [<6004ef5e>] flush_work+0x0/0x1d7
62031b60:  [<602606f2>] panic+0x0/0x2d3
62031b68:  [<600333b0>] memcpy+0x0/0x140
62031b80:  [<6002d58a>] unblock_signals+0x0/0x84
62031ba0:  [<602609c5>] printk+0x0/0xa0
62031bd8:  [<60264e51>] __mutex_unlock_slowpath+0x13d/0x148
62031c10:  [<6004ef5e>] flush_work+0x0/0x1d7
62031c18:  [<60050234>] try_to_grab_pending+0x0/0x17e
62031c38:  [<6004e984>] get_work_gcwq+0x71/0x8f
62031c48:  [<60050539>] __cancel_work_timer+0x5b/0x115
62031c78:  [<628acc85>] unplug_port+0x0/0x191 [virtio_console]
62031c98:  [<6005061c>] cancel_work_sync+0x12/0x14
62031ca8:  [<628ace96>] virtcons_remove+0x80/0x15c [virtio_console]
62031ce8:  [<628191de>] virtio_dev_remove+0x1e/0x7e [virtio]
62031d08:  [<601cf242>] __device_release_driver+0x75/0xe4
62031d28:  [<601cf2dd>] device_release_driver+0x2c/0x40
62031d48:  [<601ce0dd>] driver_unbind+0x7d/0xc6
62031d88:  [<601cd5d9>] drv_attr_store+0x27/0x29
62031d98:  [<60115f61>] sysfs_write_file+0x100/0x14d
62031df8:  [<600b737d>] vfs_write+0xcb/0x184
62031e08:  [<600b58b8>] filp_close+0x88/0x94
62031e38:  [<600b7686>] sys_write+0x59/0x88
62031e88:  [<6001ced1>] handle_syscall+0x5d/0x80
62031ea8:  [<60030a74>] userspace+0x405/0x531
62031f08:  [<600d32cc>] sys_dup+0x0/0x5e
62031f28:  [<601b11d6>] strcpy+0x0/0x18
62031f38:  [<600be46c>] do_execve+0x10/0x12
62031f48:  [<600184c7>] run_init_process+0x43/0x45
62031fd8:  [<60019a91>] new_thread_handler+0xba/0xbc

Signed-off-by: Sjur Brændeland <sjur.brandeland@stericsson.com>
Cc: stable@kernel.org
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
8 years agoMerge tag 'sound-3.8' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Linus Torvalds [Wed, 6 Feb 2013 21:43:30 +0000 (08:43 +1100)]
Merge tag 'sound-3.8' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "Just a couple of build regression fixes for ASoC fsl stuff.  It
  doesn't look too trivial, but neither intrusive, so hopefully I can
  avoid your curse..."

Hey, Takashi has a good track record, I think he gets a pass..

* tag 'sound-3.8' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ASoC: fsl: fix snd-soc-imx-pcm module build
  Revert "ASoC: fsl: fix multiple definition of init_module"

8 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-block
Linus Torvalds [Wed, 6 Feb 2013 21:38:33 +0000 (08:38 +1100)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-block

Pull block layer updates from Jens Axboe:
 "I've got a few bits pending for 3.8 final, that I better get sent out.
  It's all been sitting for a while, I consider it safe.

  It contains:

   - Two bug fixes for mtip32xx, fixing a driver hang and a crash.

   - A few-liner protocol error fix for drbd.

   - A few fixes for the xen block front/back driver, fixing a potential
     data corruption issue.

   - A race fix for disk_clear_events(), causing spurious warnings.  Out
     of the Chrome OS base.

   - A deadlock fix for disk_clear_events(), moving it to the a
     unfreezable workqueue.  Also from the Chrome OS base."

* 'for-linus' of git://git.kernel.dk/linux-block:
  drbd: fix potential protocol error and resulting disconnect/reconnect
  mtip32xx: fix for crash when the device surprise removed during rebuild
  mtip32xx: fix for driver hang after a command timeout
  block: prevent race/cleanup
  block: remove deadlock in disk_clear_events
  xen-blkfront: handle bvecs with partial data
  llist/xen-blkfront: implement safe version of llist_for_each_entry
  xen-blkback: implement safe iterator for the list of persistent grants

8 years agonet: qmi_wwan: add more Huawei devices, including E320
Bjørn Mork [Wed, 6 Feb 2013 05:22:08 +0000 (05:22 +0000)]
net: qmi_wwan: add more Huawei devices, including E320

Adding new class/subclass/protocol combinations based on the GPLed
out-of-tree Huawei driver. One of these has already appeared on a
device labelled as "E320".

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: cdc_ncm: add another Huawei vendor specific device
Bjørn Mork [Wed, 6 Feb 2013 05:21:53 +0000 (05:21 +0000)]
net: cdc_ncm: add another Huawei vendor specific device

Adding a new vendor specific class/subclass/protocol combination
for CDC NCM devices based on information from a GPLed out-of-tree
driver from Huawei.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoipv6/ip6_gre: fix error case handling in ip6gre_tunnel_xmit()
Tommi Rantala [Wed, 6 Feb 2013 03:24:02 +0000 (03:24 +0000)]
ipv6/ip6_gre: fix error case handling in ip6gre_tunnel_xmit()

ip6gre_tunnel_xmit() is leaking the skb when we hit this error branch,
and the -1 return value from this function is bogus. Use the error
handling we already have in place in ip6gre_tunnel_xmit() for this error
case to fix this.

Signed-off-by: Tommi Rantala <tt.rantala@gmail.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotcp: fix for zero packets_in_flight was too broad
Ilpo Järvinen [Mon, 4 Feb 2013 02:14:25 +0000 (02:14 +0000)]
tcp: fix for zero packets_in_flight was too broad

There are transients during normal FRTO procedure during which
the packets_in_flight can go to zero between write_queue state
updates and firing the resulting segments out. As FRTO processing
occurs during that window the check must be more precise to
not match "spuriously" :-). More specificly, e.g., when
packets_in_flight is zero but FLAG_DATA_ACKED is true the problematic
branch that set cwnd into zero would not be taken and new segments
might be sent out later.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Tested-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'wireless'
David S. Miller [Wed, 6 Feb 2013 19:51:14 +0000 (14:51 -0500)]
Merge branch 'wireless'

John W. Linville says:

====================
Please consider this pull request for the 3.8 stream...

Included is a bluetooth pull.  Gustavo says:

"Two simple fixes for 3.8. One of the patches fixes a situation
where the connection wasn't terminated if a timeout ocurrs for LE
an SCO connections.  The other fixes prevent NULL dereference in the
SMP code, it is a security fix as well."

Along with those...

Hauke Mehrtens provides a couple of ssb and bcma bus fixes that
prevent oopses when unloading those modules.

Larry Finger provides and rtlwifi fix to avoid a "scheduling while
atomic" bug.

Last but certainly not least, Arend van Spriel bring a brcmsmac fix that
reworks the mac80211 .flush() callback in order to avoid the dreaded
brcms_c_wait_for_tx_completion warnings.  This one looks a little
large, but I think it is safe and isolated to brcmsmac in any case.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
John W. Linville [Wed, 6 Feb 2013 18:55:44 +0000 (13:55 -0500)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless into for-davem

8 years agoBtrfs: move d_instantiate outside the transaction during mksubvol
Chris Mason [Wed, 6 Feb 2013 17:06:02 +0000 (12:06 -0500)]
Btrfs: move d_instantiate outside the transaction during mksubvol

Dave Sterba triggered a lockdep complaint about lock ordering
between the sb_internal lock and the cleaner semaphore.

btrfs_lookup_dentry() checks for orphans if we're looking up
the inode for a subvolume, and subvolume creation is triggering
the lookup with a transaction running.

This commit moves the d_instantiate after the transaction closes.

Signed-off-by: Chris Mason <chris.mason@fusionio.com>
8 years agoxen: fix error handling path if xen_allocate_irq_dynamic fails
Wei Liu [Thu, 31 Jan 2013 14:46:56 +0000 (14:46 +0000)]
xen: fix error handling path if xen_allocate_irq_dynamic fails

It is possible that the call to xen_allocate_irq_dynamic() returns negative
number other than -1.

Reviewed-by: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
8 years agoxen-pciback: rate limit error messages from xen_pcibk_enable_msi{,x}()
Jan Beulich [Wed, 6 Feb 2013 15:30:38 +0000 (10:30 -0500)]
xen-pciback: rate limit error messages from xen_pcibk_enable_msi{,x}()

... as being guest triggerable (e.g. by invoking
XEN_PCI_OP_enable_msi{,x} on a device not being MSI/MSI-X capable).

This is CVE-2013-0231 / XSA-43.

Also make the two messages uniform in both their wording and severity.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
8 years agoMerge branch 'fix/fsl' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound...
Takashi Iwai [Wed, 6 Feb 2013 14:16:58 +0000 (15:16 +0100)]
Merge branch 'fix/fsl' of git://git./linux/kernel/git/broonie/sound into for-linus-uncursed

We need this for fixing build error regressions in soc/fsl.

8 years agoBtrfs: fix EDQUOT handling in btrfs_delalloc_reserve_metadata
Jan Schmidt [Mon, 28 Jan 2013 06:26:00 +0000 (23:26 -0700)]
Btrfs: fix EDQUOT handling in btrfs_delalloc_reserve_metadata

When btrfs_qgroup_reserve returned a failure, we were missing a counter
operation for BTRFS_I(inode)->outstanding_extents++, leading to warning
messages about outstanding extents and space_info->bytes_may_use != 0.
Additionally, the error handling code didn't take into account that we
dropped the inode lock which might require more cleanup.

Luckily, all the cleanup code we need is already there and can be shared
with reserve_metadata_bytes, which is exactly what this patch does.

Reported-by: Lev Vainblat <lev@zadarastorage.com>
Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
8 years agoMerge remote-tracking branch 'regulator/fix/tps65910' into tmp
Mark Brown [Wed, 6 Feb 2013 11:28:53 +0000 (11:28 +0000)]
Merge remote-tracking branch 'regulator/fix/tps65910' into tmp

8 years agoMerge remote-tracking branch 'regulator/fix/tps65217' into tmp
Mark Brown [Wed, 6 Feb 2013 11:28:52 +0000 (11:28 +0000)]
Merge remote-tracking branch 'regulator/fix/tps65217' into tmp

8 years agoMerge remote-tracking branch 'regulator/fix/s2mps11' into tmp
Mark Brown [Wed, 6 Feb 2013 11:28:51 +0000 (11:28 +0000)]
Merge remote-tracking branch 'regulator/fix/s2mps11' into tmp

8 years agoMerge remote-tracking branch 'regulator/fix/of' into tmp
Mark Brown [Wed, 6 Feb 2013 11:28:51 +0000 (11:28 +0000)]
Merge remote-tracking branch 'regulator/fix/of' into tmp

8 years agoMerge remote-tracking branch 'regulator/fix/max8998' into tmp
Mark Brown [Wed, 6 Feb 2013 11:28:50 +0000 (11:28 +0000)]
Merge remote-tracking branch 'regulator/fix/max8998' into tmp

8 years agoMerge remote-tracking branch 'regulator/fix/max8997' into tmp
Mark Brown [Wed, 6 Feb 2013 11:28:49 +0000 (11:28 +0000)]
Merge remote-tracking branch 'regulator/fix/max8997' into tmp

8 years agoMerge remote-tracking branch 'regulator/fix/max8907' into tmp
Mark Brown [Wed, 6 Feb 2013 11:28:48 +0000 (11:28 +0000)]
Merge remote-tracking branch 'regulator/fix/max8907' into tmp

8 years agoMerge remote-tracking branch 'regulator/fix/max77686' into tmp
Mark Brown [Wed, 6 Feb 2013 11:28:48 +0000 (11:28 +0000)]
Merge remote-tracking branch 'regulator/fix/max77686' into tmp

8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/josef/btrfs-next.git for-chris...
Chris Mason [Wed, 6 Feb 2013 00:24:44 +0000 (19:24 -0500)]
Merge git://git./linux/kernel/git/josef/btrfs-next.git for-chris into for-linus

8 years agoMerge branch 'v4l_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Tue, 5 Feb 2013 21:36:12 +0000 (08:36 +1100)]
Merge branch 'v4l_for_linus' of git://git./linux/kernel/git/mchehab/linux-media

Pull media fixes from Mauro Carvalho Chehab:
 "For a regression fix on a few radio drivers that were preventing radio
  TX to work on those devices"

* 'v4l_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media:
  [media] radio: set vfl_dir correctly to fix modulator regression

8 years agoMerge tag 'usb-3.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Tue, 5 Feb 2013 21:32:32 +0000 (08:32 +1100)]
Merge tag 'usb-3.8-rc6' of git://git./linux/kernel/git/gregkh/usb

Pull USB fixes from Greg Kroah-Hartman:
 "Here are a few tiny USB fixes for 3.8-rc6.

  Nothing major here, some host controller bug fixes to resolve a number
  of bugs that people have reported, and a bunch of additional device
  ids are added to a number of drivers (which caused code to be deleted
  from the usb-storage driver, always nice)"

* tag 'usb-3.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (22 commits)
  USB: storage: optimize to match the Huawei USB storage devices and support new switch command
  USB: storage: Define a new macro for USB storage match rules
  USB: ftdi_sio: add Zolix FTDI PID
  USB: option: add Changhong CH690
  USB: ftdi_sio: add PID/VID entries for ELV WS 300 PC II
  USB: add OWL CM-160 support to cp210x driver
  USB: EHCI: fix bug in scheduling periodic split transfers
  USB: EHCI: fix for leaking isochronous data
  USB: option: add support for Telit LE920
  USB: qcserial: add Telit Gobi QDL device
  USB: EHCI: fix timer bug affecting port resume
  USB: UHCI: notify usbcore about port resumes
  USB: EHCI: notify usbcore about port resumes
  USB: add usb_hcd_{start,end}_port_resume
  USB: EHCI: unlink one async QH at a time
  USB: EHCI: remove ASS/PSS polling timeout
  usb: Using correct way to clear usb3.0 device's remote wakeup feature.
  usb: Prevent dead ports when xhci is not enabled
  USB: XHCI: fix memory leak of URB-private data
  drivers: xhci: fix incorrect bit test
  ...

8 years agoMerge branch 'fixes-for-v3.8-rc7' of git://git.linaro.org/people/mszyprowski/linux...
Linus Torvalds [Tue, 5 Feb 2013 21:23:47 +0000 (08:23 +1100)]
Merge branch 'fixes-for-v3.8-rc7' of git://git.linaro.org/people/mszyprowski/linux-dma-mapping

Pull DMA mapping fixes from Marek Szyprowski:
 "This pull request contains important bugfix patches for 9
  architectures, which finally fixes broken allmodconfig builds
  introduced in v3.8-rc1.  Those architectures don't use dma_map_ops
  based implementation and require manual update or additional dummy
  implementations of the missing new dma-mapping api functions:
  dma_mmap_coherent and dma_get_sgtable."

* 'fixes-for-v3.8-rc7' of git://git.linaro.org/people/mszyprowski/linux-dma-mapping:
  xtensa: Provide dummy dma_mmap_coherent() and dma_get_sgtable()
  parisc: Provide dummy dma_mmap_coherent() and dma_get_sgtable()
  mn10300: Provide dummy dma_mmap_coherent() and dma_get_sgtable()
  m68k: Provide dma_mmap_coherent() and dma_get_sgtable()
  frv: Provide dummy dma_mmap_coherent() and dma_get_sgtable()
  cris: Provide dma_mmap_coherent() and dma_get_sgtable()
  c6x: Provide dummy dma_mmap_coherent() and dma_get_sgtable()
  blackfin: Provide dma_mmap_coherent() and dma_get_sgtable()
  avr32: Provide dma_mmap_coherent() and dma_get_sgtable()

8 years agoBtrfs: fix possible stale data exposure
Josef Bacik [Wed, 30 Jan 2013 19:31:31 +0000 (14:31 -0500)]
Btrfs: fix possible stale data exposure

We specifically do not update the disk i_size if there are ordered extents
outstanding for any area between the current disk_i_size and our ordered
extent so that we do not expose stale data.  The problem is the check we
have only checks if the ordered extent starts at or after the current
disk_i_size, which doesn't take into account an ordered extent that starts
before the current disk_i_size and ends past the disk_i_size.  Fix this by
checking if the extent ends past the disk_i_size.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
8 years agoBtrfs: fix missing i_size update
Josef Bacik [Wed, 30 Jan 2013 19:17:31 +0000 (14:17 -0500)]
Btrfs: fix missing i_size update

If we have an ordered extent before the ordered extent we are currently
completing that is after the current disk_i_size we will put our i_size
update into that ordered extent so that we do not expose stale data.  The
problem is that if our disk i_size is updated past the previous ordered
extent we won't update the i_size with the pending i_size update.  So check
the pending i_size update and if its above the current disk i_size we need
to go ahead and try to update.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
8 years agoBtrfs: fix race between snapshot deletion and getting inode
Liu Bo [Tue, 29 Jan 2013 03:22:10 +0000 (03:22 +0000)]
Btrfs: fix race between snapshot deletion and getting inode

While running snapshot testscript created by Mitch and David,
the race between autodefrag and snapshot deletion can lead to
corruption of dead_root list so that we can get crash on
btrfs_clean_old_snapshots().

And besides autodefrag, scrub also does the same thing, ie. read
root first and get inode.

Here is the story(take autodefrag as an example):
(1) when we delete a snapshot or subvolume, it will set its root's
refs to zero and do a iput() on its own inode, and if this inode happens
to be the only active in-meory one in root's inode rbtree, it will add
itself to the global dead_roots list for later cleanup.

(2) after (1), the autodefrag thread may read another inode for defrag
and the inode is just in the deleted snapshot/subvolume, but all of these
are without checking if the root is still valid(refs > 0).  So the end up
result is adding the deleted snapshot/subvolume's root to the global
dead_roots list AGAIN.

Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu.

So all we need to do is to take the lock to protect 'read root and get inode',
since we synchronize to wait for the rcu grace period before adding something
to the global dead_roots list.

Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org>
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
8 years agoBtrfs: fix missing release of the space/qgroup reservation in start_transaction()
Miao Xie [Mon, 28 Jan 2013 12:36:22 +0000 (12:36 +0000)]
Btrfs: fix missing release of the space/qgroup reservation in start_transaction()

When we fail to start a transaction, we need to release the reserved free space
and qgroup space, fix it.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Reviewed-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>