firefly-linux-kernel-4.4.55.git
9 years agomacvtap: TUN_VNET_LE support
Michael S. Tsirkin [Sun, 23 Nov 2014 15:24:15 +0000 (17:24 +0200)]
macvtap: TUN_VNET_LE support

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
9 years agotun: TUN_VNET_LE support, fix sparse warnings for virtio headers
Michael S. Tsirkin [Thu, 23 Oct 2014 19:59:31 +0000 (22:59 +0300)]
tun: TUN_VNET_LE support, fix sparse warnings for virtio headers

Pretty straight-forward: convert all fields to/from
virtio endian-ness.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
9 years agotun: add VNET_LE flag
Michael S. Tsirkin [Tue, 18 Nov 2014 15:38:08 +0000 (17:38 +0200)]
tun: add VNET_LE flag

virtio 1.0 modified virtio net header format,
making all fields little endian.

Users can tweak header format before submitting it to tun,
but this means more data copies where none were necessary.
And if the iovec is in RO memory, this means we might
need to split iovec also means we might in theory overflow
iovec max size.

This patch adds a simpler way for applications to handle this,
using new "little endian" flag in tun.
As a result, tun simply byte-swaps header fields as appropriate.
This is a NOP on LE architectures.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agotun: drop most type defines
Michael S. Tsirkin [Wed, 19 Nov 2014 13:17:31 +0000 (15:17 +0200)]
tun: drop most type defines

It's just as easy to use IFF_ flags directly,
there's no point in adding our own defines.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agotun: move internal flag defines out of uapi
Michael S. Tsirkin [Wed, 19 Nov 2014 12:44:40 +0000 (14:44 +0200)]
tun: move internal flag defines out of uapi

TUN_ flags are internal and never exposed
to userspace. Any application using it is almost
certainly buggy.

Move them out to tun.c.

Note: we remove these completely in follow-up patches,
this code movement is split out for ease of review.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agovhost/net: enable virtio 1.0
Michael S. Tsirkin [Fri, 24 Oct 2014 11:25:03 +0000 (14:25 +0300)]
vhost/net: enable virtio 1.0

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agovhost/net: larger header for virtio 1.0
Michael S. Tsirkin [Fri, 24 Oct 2014 11:23:52 +0000 (14:23 +0300)]
vhost/net: larger header for virtio 1.0

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovhost/net: virtio 1.0 byte swap
Michael S. Tsirkin [Fri, 24 Oct 2014 11:19:48 +0000 (14:19 +0300)]
vhost/net: virtio 1.0 byte swap

I had to add an explicit tag to suppress compiler warning:
gcc isn't smart enough to notice that
len is always initialized since function is called with size > 0.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovhost: virtio 1.0 endian-ness support
Michael S. Tsirkin [Fri, 24 Oct 2014 11:04:47 +0000 (14:04 +0300)]
vhost: virtio 1.0 endian-ness support

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agovhost: switch to __get/__put_user exclusively
Michael S. Tsirkin [Mon, 1 Dec 2014 15:39:39 +0000 (17:39 +0200)]
vhost: switch to __get/__put_user exclusively

Most places in vhost can use __get/__put_user rather than
get/put_user since addresses are pre-validated.
This should be good for performance, but this also
will help make code sparse-clean: get/put_user macros
don't play well with __virtioXX bitwise tags.
Switch to get/put_user to __ variants everywhere in vhost.
There's one exception - for consistency switch that
as well, and add an explicit access_ok check.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agovhost/net: force len for TX to host endian
Michael S. Tsirkin [Fri, 24 Oct 2014 08:49:27 +0000 (11:49 +0300)]
vhost/net: force len for TX to host endian

vhost/net keeps a copy of the used ring in host memory but (ab)uses
the length field for internal house-keeping. This works because the
length in the used ring for tx is always 0. In order to suppress sparse
warnings, we force native endianness here.
Note that these values are never exposed to guests.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
9 years agovhost: add memory access wrappers
Michael S. Tsirkin [Fri, 24 Oct 2014 08:48:17 +0000 (11:48 +0300)]
vhost: add memory access wrappers

Add guest memory access wrappers to handle virtio endianness
conversions.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovhost: make features 64 bit
Michael S. Tsirkin [Fri, 24 Oct 2014 11:08:18 +0000 (14:08 +0300)]
vhost: make features 64 bit

We need to use bit 32 for virtio 1.0.
Make vhost_has_feature bool to avoid discarding high bits.

Cc: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Cc: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
9 years agovirtio_net: enable v1.0 support
Michael S. Tsirkin [Thu, 23 Oct 2014 21:32:24 +0000 (00:32 +0300)]
virtio_net: enable v1.0 support

Now that we have completed 1.0 support, enable it in our driver.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agovirtio_net: disable mac write for virtio 1.0
Michael S. Tsirkin [Wed, 26 Nov 2014 13:58:28 +0000 (15:58 +0200)]
virtio_net: disable mac write for virtio 1.0

The spec states that mac in config space is only driver-writable in the
legacy case.  Fence writing it in virtnet_set_mac_address() in the
virtio 1.0 case.

Suggested-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovirtio_net: bigger header when VERSION_1 is set
Michael S. Tsirkin [Thu, 23 Oct 2014 21:24:03 +0000 (00:24 +0300)]
virtio_net: bigger header when VERSION_1 is set

With VERSION_1 virtio_net uses same header size
whether mergeable buffers are enabled or not.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
9 years agovirtio_net: stricter short buffer length checks
Michael S. Tsirkin [Thu, 23 Oct 2014 21:22:11 +0000 (00:22 +0300)]
virtio_net: stricter short buffer length checks

Our buffer length check is not strict enough for mergeable
buffers: buffer can still be shorter that header + address
by 2 bytes.

Fix that up.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
9 years agovirtio_net: get rid of virtio_net_hdr/skb_vnet_hdr
Michael S. Tsirkin [Fri, 24 Oct 2014 13:55:57 +0000 (16:55 +0300)]
virtio_net: get rid of virtio_net_hdr/skb_vnet_hdr

virtio 1.0 doesn't use virtio_net_hdr anymore, and in fact, it's not
really useful since virtio_net_hdr_mrg_rxbuf includes that as the first
field anyway.

Let's drop it, precalculate header len and store within vi instead.

This way we can also remove struct skb_vnet_hdr.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
9 years agovirtio_net: pass vi around
Michael S. Tsirkin [Thu, 23 Oct 2014 21:12:10 +0000 (00:12 +0300)]
virtio_net: pass vi around

Too many places poke at [rs]q->vq->vdev->priv just to get
the vi structure.  Let's just pass the pointer around: seems
cleaner, and might even be faster.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovirtio_blk: fix race at module removal
Michael S. Tsirkin [Thu, 23 Oct 2014 15:57:19 +0000 (18:57 +0300)]
virtio_blk: fix race at module removal

If a device appears while module is being removed,
driver will get a callback after we've given up
on the major number.

In theory this means this major number can get reused
by something else, resulting in a conflict.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovirtio_blk: make serial attribute static
Michael S. Tsirkin [Thu, 23 Oct 2014 13:08:44 +0000 (16:08 +0300)]
virtio_blk: make serial attribute static

It's never declared so no need to make it extern.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agoKVM: s390: enable virtio-ccw revision 1
Cornelia Huck [Tue, 7 Oct 2014 14:39:52 +0000 (16:39 +0200)]
KVM: s390: enable virtio-ccw revision 1

Now that virtio-ccw has everything needed to support virtio 1.0 in
place, try to enable it if the host supports it.

Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agoKVM: s390 allow virtio_ccw status writes to fail
Michael S. Tsirkin [Thu, 23 Oct 2014 15:40:30 +0000 (18:40 +0300)]
KVM: s390 allow virtio_ccw status writes to fail

Gracefully handle failure to write device status.
We really should handle other errors as well, but this one is needed for
virtio 1.0 compliance.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agoKVM: s390: virtio-ccw revision 1 SET_VQ
Cornelia Huck [Tue, 7 Oct 2014 14:39:51 +0000 (16:39 +0200)]
KVM: s390: virtio-ccw revision 1 SET_VQ

The CCW_CMD_SET_VQ command has a different format for revision 1+
devices, allowing to specify a more complex virtqueue layout. For
now, we stay however with the old layout and simply use the new
command format for virtio-1 devices.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
9 years agoKVM: s390: Set virtio-ccw transport revision
Thomas Huth [Tue, 7 Oct 2014 14:39:50 +0000 (16:39 +0200)]
KVM: s390: Set virtio-ccw transport revision

With the new SET-VIRTIO-REVISION command of the virtio 1.0 standard, we
can now negotiate the virtio-ccw revision after setting a channel online.

Note that we don't negotiate version 1 yet.

[Cornelia Huck: reworked revision loop a bit]
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agovirtio_blk: v1.0 support
Michael S. Tsirkin [Tue, 7 Oct 2014 14:39:49 +0000 (16:39 +0200)]
virtio_blk: v1.0 support

Based on patch by Cornelia Huck.

Note: for consistency, and to avoid sparse errors,
      convert all fields, even those no longer in use
      for virtio v1.0.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agovirtio_net: v1.0 endianness
Michael S. Tsirkin [Tue, 7 Oct 2014 14:39:48 +0000 (16:39 +0200)]
virtio_net: v1.0 endianness

Based on patches by Rusty Russell, Cornelia Huck.
Note: more code changes are needed for 1.0 support
(due to different header size).
So we don't advertize support for 1.0 yet.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agovirtio: add legacy feature table support
Michael S. Tsirkin [Thu, 23 Oct 2014 15:07:47 +0000 (18:07 +0300)]
virtio: add legacy feature table support

virtio-blk has some legacy feature bits that modern drivers
must not negotiate, but are needed for old legacy hosts
(that e.g. don't support virtio-scsi).
Allow a separate legacy feature table for such cases.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovirtio: simplify feature bit handling
Michael S. Tsirkin [Thu, 23 Oct 2014 14:57:30 +0000 (17:57 +0300)]
virtio: simplify feature bit handling

Now that we use u64 for bits, we can simply & them together.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovirtio: set FEATURES_OK
Michael S. Tsirkin [Wed, 22 Oct 2014 14:41:38 +0000 (17:41 +0300)]
virtio: set FEATURES_OK

set FEATURES_OK as per virtio 1.0 spec

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovirtio: allow transports to get avail/used addresses
Cornelia Huck [Tue, 7 Oct 2014 14:39:47 +0000 (16:39 +0200)]
virtio: allow transports to get avail/used addresses

For virtio-1, we can theoretically have a more complex virtqueue
layout with avail and used buffers not on a contiguous memory area
with the descriptor table. For now, it's fine for a transport driver
to stay with the old layout: It needs, however, a way to access
the locations of the avail/used rings so it can register them with
the host.

Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agovirtio_config: endian conversion for v1.0
Michael S. Tsirkin [Wed, 22 Oct 2014 13:59:01 +0000 (16:59 +0300)]
virtio_config: endian conversion for v1.0

We (ab)use virtio conversion functions for device-specific
config space accesses.

Based on original patches by Cornelia and Rusty.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.com>
9 years agovirtio_ring: switch to new memory access APIs
Michael S. Tsirkin [Wed, 22 Oct 2014 12:42:09 +0000 (15:42 +0300)]
virtio_ring: switch to new memory access APIs

Use virtioXX_to_cpu and friends for access to
all multibyte structures in memory.

Note: this is intentionally mechanical.
A follow-up patch will split long lines etc.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovirtio: memory access APIs
Michael S. Tsirkin [Wed, 22 Oct 2014 12:35:56 +0000 (15:35 +0300)]
virtio: memory access APIs

virtio 1.0 makes all memory structures LE, so
we need APIs to conditionally do a byteswap on BE
architectures.

To make it easier to check code statically,
add virtio specific types for multi-byte integers
in memory.

Add low level wrappers that do a byteswap conditionally, these will be
useful e.g. for vhost.  Add high level wrappers that
query device endian-ness and act accordingly.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovirtio: add virtio 1.0 feature bit
Michael S. Tsirkin [Wed, 22 Oct 2014 12:49:34 +0000 (15:49 +0300)]
virtio: add virtio 1.0 feature bit

Based on original patches by Rusty Russell, Thomas Huth
and Cornelia Huck.

Note: at this time, we do not negotiate this feature bit
in core, drivers have to declare VERSION_1 support explicitly.

For this reason we treat this bit as a device bit
and not as a transport bit for now.

After all drivers are converted, we will be able to
move VERSION_1 to core and drop it from all drivers.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovirtio_ccw: add support for 64 bit features.
Michael S. Tsirkin [Thu, 27 Nov 2014 11:54:28 +0000 (13:54 +0200)]
virtio_ccw: add support for 64 bit features.

Negotiate full 64 bit features.
Change u32 to u64, make sure to use 1ULL everywhere.

Note: devices guarantee that VERSION_1 is clear unless
revision 1 is negotiated.

Note: We don't need to re-setup the ccw, but we do it
for clarity.

Based on patches by Rusty, Thomas Huth and Cornelia.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildebrand <dahi@linux.vnet.ibm.com>
9 years agovirtio: assert 32 bit features in transports
Michael S. Tsirkin [Thu, 27 Nov 2014 11:45:58 +0000 (13:45 +0200)]
virtio: assert 32 bit features in transports

At this point, no transports set any of the high 32 feature bits.
Since transports generally can't (yet) cope with such bits, add BUG_ON
checks to make sure they are not set by mistake.

Based on rproc patch by Rusty.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovirtio: add support for 64 bit features.
Michael S. Tsirkin [Tue, 7 Oct 2014 14:39:43 +0000 (16:39 +0200)]
virtio: add support for 64 bit features.

Change u32 to u64, and use BIT_ULL and 1ULL everywhere.

Note: transports are unchanged, and only set low 32 bit.
This guarantees that no transport sets e.g. VERSION_1
by mistake without proper support.

Based on patch by Rusty.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agomic_virtio: robust feature array size calculation
Michael S. Tsirkin [Thu, 27 Nov 2014 11:29:24 +0000 (13:29 +0200)]
mic_virtio: robust feature array size calculation

mic reads sizeof(vdev->features) bits from device, but in fact it stores
bits in local features variable. use sizeof(features) to make code
robust against future changes extending sizeof(vdev->features).

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
9 years agovirtio: use u32, not bitmap for features
Michael S. Tsirkin [Tue, 7 Oct 2014 14:39:42 +0000 (16:39 +0200)]
virtio: use u32, not bitmap for features

It seemed like a good idea to use bitmap for features
in struct virtio_device, but it's actually a pain,
and seems to become even more painful when we get more
than 32 feature bits.  Just change it to a u32 for now.

Based on patch by Rusty.

Suggested-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agovirtio: add low-level APIs for feature bits
Michael S. Tsirkin [Thu, 27 Nov 2014 19:19:02 +0000 (21:19 +0200)]
virtio: add low-level APIs for feature bits

Add low level APIs to test/set/clear feature bits.
For use by transports, to make it easier to
write code independent of feature bit array format.

Note: APIs is prefixed with __ and has _bit suffix
to stress its low level nature.  It's for use by transports only:
drivers should use virtio_has_feature and never need to set/clear
features.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
9 years agoLinux 3.18
Linus Torvalds [Sun, 7 Dec 2014 22:21:05 +0000 (14:21 -0800)]
Linux 3.18

9 years agoMerge branch 'for-3.18-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj...
Linus Torvalds [Sun, 7 Dec 2014 20:00:14 +0000 (12:00 -0800)]
Merge branch 'for-3.18-fixes' of git://git./linux/kernel/git/tj/libata

Pull libata fixes from Tejun Heo:
 "Three libata fixes for v3.18.  Nothing too interesting.  PCI ID ID and
  quirk additions to ahci and an error handling path fix in sata_fsl"

* 'for-3.18-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/libata:
  ahci: disable MSI on SAMSUNG 0xa800 SSD
  sata_fsl: fix error handling of irq_of_parse_and_map
  AHCI: Add DeviceIDs for Sunrise Point-LP SATA controller

9 years agoMerge git://www.linux-watchdog.org/linux-watchdog
Linus Torvalds [Sat, 6 Dec 2014 19:27:25 +0000 (11:27 -0800)]
Merge git://www.linux-watchdog.org/linux-watchdog

Pull watchdog fix from Wim Van Sebroeck:
 "Fix the watchdog mask bit offset for Exynos7"

* git://www.linux-watchdog.org/linux-watchdog:
  watchdog: s3c2410_wdt: Fix the mask bit offset for Exynos7

9 years agoMerge branch 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Sat, 6 Dec 2014 19:26:01 +0000 (11:26 -0800)]
Merge branch 'i2c/for-current' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:
 "Here are two more driver bugfixes for I2C which would be good to have"

* 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: cadence: Set the hardware time-out register to maximum value
  i2c: davinci: generate STP always when NACK is received

9 years agowatchdog: s3c2410_wdt: Fix the mask bit offset for Exynos7
Abhilash Kesavan [Fri, 17 Oct 2014 16:12:53 +0000 (21:42 +0530)]
watchdog: s3c2410_wdt: Fix the mask bit offset for Exynos7

The watchdog mask bit offset listed for Exynos7 is incorrect.
Fix this.

Signed-off-by: Abhilash Kesavan <a.kesavan@samsung.com>
Acked-by: Naveen Krishna Chatradhi <naveenkrishna.ch@gmail.com
Reviewd-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
9 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 5 Dec 2014 18:47:19 +0000 (10:47 -0800)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Thomas Gleixner:
 "Two final fixlets for 3.18:
   - Prevent microcode reload wreckage on 32bit
   - Unbreak cross compilation"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, microcode: Limit the microcode reloading to 64-bit for now
  x86: Use $(OBJDUMP) instead of plain objdump

9 years agoMerge tag 'sound-3.18' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Linus Torvalds [Fri, 5 Dec 2014 18:39:49 +0000 (10:39 -0800)]
Merge tag 'sound-3.18' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixlet from Takashi Iwai:
 "Just one commit for adding a copule of HD-audio quirk entries"

* tag 'sound-3.18' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: hda/realtek - Add headset Mic support for new Dell machine

9 years agoMerge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Fri, 5 Dec 2014 06:39:37 +0000 (22:39 -0800)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull drm intel fixes from Dave Airlie:
 "Two intel stable fixes, that should be it from me for this round"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm/i915: Unlock panel even when LVDS is disabled
  drm/i915: More cautious with pch fifo underruns

9 years agoMerge tag 'pm+acpi-3.18-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Fri, 5 Dec 2014 01:48:13 +0000 (17:48 -0800)]
Merge tag 'pm+acpi-3.18-rc8' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI backlight fix from Rafael Wysocki:
 "This is a simple fix for an ACPI backlight regression introduced by a
  recent commit that overlooked a corner case which should have been
  taken into account"

* tag 'pm+acpi-3.18-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI / video: update condition to check if device is in _DOD list

9 years agoMerge tag 'drm-intel-fixes-2014-12-04' of git://anongit.freedesktop.org/drm-intel...
Dave Airlie [Fri, 5 Dec 2014 01:12:29 +0000 (11:12 +1000)]
Merge tag 'drm-intel-fixes-2014-12-04' of git://anongit.freedesktop.org/drm-intel into drm-fixes

Silence some pch fifo underrun reports and panel locking backtraces,
both cc: stable.

* tag 'drm-intel-fixes-2014-12-04' of git://anongit.freedesktop.org/drm-intel:
  drm/i915: Unlock panel even when LVDS is disabled
  drm/i915: More cautious with pch fifo underruns

9 years agoMerge tag 'media/v3.18-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Fri, 5 Dec 2014 00:06:02 +0000 (16:06 -0800)]
Merge tag 'media/v3.18-rc8' of git://git./linux/kernel/git/mchehab/linux-media

Pull media fixes from Mauro Carvalho Chehab:
 "A core fix and some driver fixes:
   - regression fix in Remote Controller core affecting RC6 protocol
     handling
   - fix video buffer handling in cx23885
   - race fix in solo6x10
   - fix image selection in smiapp
   - fix reported payload size on s2255drv
   - two updates for MAINTAINERS file"

* tag 'media/v3.18-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media:
  [media] rc-core: fix toggle handling in the rc6 decoder
  MAINTAINERS: Update mchehab's addresses
  [media] cx23885: use sg = sg_next(sg) instead of sg++
  [media] s2255drv: fix payload size for JPG, MJPEG
  [media] Update MAINTAINERS for solo6x10
  [media] solo6x10: fix a race in IRQ handler
  [media] smiapp: Only some selection targets are settable

9 years agouapi: fix to export linux/vm_sockets.h
Masahiro Yamada [Thu, 4 Dec 2014 22:42:25 +0000 (14:42 -0800)]
uapi: fix to export linux/vm_sockets.h

A typo "header=y" was introduced by commit 7071cf7fc435 ("uapi: add
missing network related headers to kbuild").

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoi2c: cadence: Set the hardware time-out register to maximum value
Vishnu Motghare [Wed, 3 Dec 2014 12:35:25 +0000 (18:05 +0530)]
i2c: cadence: Set the hardware time-out register to maximum value

Cadence I2C controller has bug wherein it generates invalid read transactions
after timeout in master receiver mode. This driver does not use the HW
timeout and this interrupt is disabled but the feature itself cannot be
disabled. Hence, this patch writes the maximum value (0xFF) to this register.
This is one of the workarounds to this bug and it will not avoid the issue
completely but reduces the chances of error.

Signed-off-by: Vishnu Motghare <vishnum@xilinx.com>
Signed-off-by: Harini Katakam <harinik@xilinx.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Cc: stable@kernel.org
9 years agoi2c: davinci: generate STP always when NACK is received
Grygorii Strashko [Mon, 1 Dec 2014 15:34:04 +0000 (17:34 +0200)]
i2c: davinci: generate STP always when NACK is received

According to I2C specification the NACK should be handled as follows:
"When SDA remains HIGH during this ninth clock pulse, this is defined as the Not
Acknowledge signal. The master can then generate either a STOP condition to
abort the transfer, or a repeated START condition to start a new transfer."
[I2C spec Rev. 6, 3.1.6: http://www.nxp.com/documents/user_manual/UM10204.pdf]

Currently the Davinci i2c driver interrupts the transfer on receipt of a
NACK but fails to send a STOP in some situations and so makes the bus
stuck until next I2C IP reset (idle/enable).

For example, the issue will happen during SMBus read transfer which
consists from two i2c messages write command/address and read data:

S Slave Address Wr A Command Code A Sr Slave Address Rd A D1..Dn A P
<--- write -----------------------> <--- read --------------------->

The I2C client device will send NACK if it can't recognize "Command Code"
and it's expected from I2C master to generate STP in this case.
But now, Davinci i2C driver will just exit with -EREMOTEIO and STP will
not be generated.

Hence, fix it by generating Stop condition (STP) always when NACK is received.

This patch fixes Davinci I2C in the same way it was done for OMAP I2C
commit cda2109a26eb ("i2c: omap: query STP always when NACK is received").

Reviewed-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reported-by: Hein Tibosch <hein_tibosch@yahoo.es>
Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Cc: stable@kernel.org
9 years agoahci: disable MSI on SAMSUNG 0xa800 SSD
Tejun Heo [Thu, 4 Dec 2014 18:13:28 +0000 (13:13 -0500)]
ahci: disable MSI on SAMSUNG 0xa800 SSD

Just like 0x1600 which got blacklisted by 66a7cbc303f4 ("ahci: disable
MSI instead of NCQ on Samsung pci-e SSDs on macbooks"), 0xa800 chokes
on NCQ commands if MSI is enabled.  Disable MSI.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Dominik Mierzejewski <dominik@greysector.net>
Link: https://bugzilla.kernel.org/show_bug.cgi?id=89171
Cc: stable@vger.kernel.org
9 years agocontext_tracking: Restore previous state in schedule_user
Andy Lutomirski [Wed, 3 Dec 2014 23:37:08 +0000 (15:37 -0800)]
context_tracking: Restore previous state in schedule_user

It appears that some SCHEDULE_USER (asm for schedule_user) callers
in arch/x86/kernel/entry_64.S are called from RCU kernel context,
and schedule_user will return in RCU user context.  This causes RCU
warnings and possible failures.

This is intended to be a minimal fix suitable for 3.18.

Reported-and-tested-by: Dave Jones <davej@redhat.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge branch 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Wed, 3 Dec 2014 22:27:07 +0000 (14:27 -0800)]
Merge branch 'i2c/for-current' of git://git./linux/kernel/git/wsa/linux

Pull i2c bugfixes from Wolfram Sang:
 "A few driver bugfixes for 3.18"

* 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: omap: fix i207 errata handling
  i2c: designware: prevent early stop on TX FIFO empty
  i2c: omap: fix NACK and Arbitration Lost irq handling

9 years agoMerge tag 'pci-v3.18-fixes-4' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaa...
Linus Torvalds [Wed, 3 Dec 2014 22:24:10 +0000 (14:24 -0800)]
Merge tag 'pci-v3.18-fixes-4' of git://git./linux/kernel/git/helgaas/pci

Pull PCI fix from Bjorn Helgaas:
 "This fixes a Tegra20 regression that we introduced during the v3.18
  merge window"

* tag 'pci-v3.18-fixes-4' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  PCI: tegra: Use physical range for I/O mapping

9 years agoMerge tag 'devicetree-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 3 Dec 2014 22:08:50 +0000 (14:08 -0800)]
Merge tag 'devicetree-for-linus' of git://git./linux/kernel/git/glikely/linux

Pull devicetree bugfix from Grant Likely:
 "One more bug fix for v3.18.  I debated whether or not to send you this
  merge request because we're at such a late rc.  The bug isn't critical
  in that there is only one system known to be affected and the patch is
  easy to backport.  The codepath is used by pretty much every DT based
  system, so there is risk a of regression (it /should/ be safe, but
  I've been bitten by stuff that should be safe before).  I've had it in
  linux-next for a week and haven't received any complaints.

  I think it probably should just be merged right away rather than
  waiting for the merge window and backporting.  It does fix a real bug
  and the code is theoretically safer after the change.  I can't think
  of any situation where it would be dangerous to reserve the DT memory
  an extra time.

  Summary from tag:

    Single bugfix for boot failure seen in the wild.  The memory reserve
    code tries to be clever about reserving the FDT, but it should just
    go ahead and reserve it unconditionally to avoid the problem of
    partial overlap described in the patch"

* tag 'devicetree-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/glikely/linux:
  of/fdt: memblock_reserve /memreserve/ regions in the case of partial overlap

9 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-block
Linus Torvalds [Wed, 3 Dec 2014 22:05:24 +0000 (14:05 -0800)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-block

Pull block core regression fix from Jens Axboe:
 "Single fix for a regression introduced in this development cycle,
  where dm on top of dif/dix is broken.  From Darrick Wong"

* 'for-linus' of git://git.kernel.dk/linux-block:
  block: fix regression where bio_integrity_process uses wrong bio_vec iterator

9 years agoMerge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Wed, 3 Dec 2014 21:51:50 +0000 (13:51 -0800)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull drm fixes from Dave Airlie:
 "Radeon and Nouveau fixes:

  So nouveau had a few regression introduced, Ben and Maarten finally
  tracked down the one that was causing problems on my MacBookPro, also
  nvidia gave some info on the an engine we were using incorrectly, so
  disable our use of it, and one regresion with pci hotplug affecting
  optimus users.

  Radeon has an oops fixs, sync fix, and one workaround to avoid broken
  functionality on 32-bit x86, this needs better root causing and a
  better fix, but the bandaid is a lot safer at this point"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm/radeon: kernel panic in drm_calc_vbltimestamp_from_scanoutpos with 3.18.0-rc6
  drm/radeon: Ignore RADEON_GEM_GTT_WC on 32-bit x86
  drm/radeon: sync all BOs involved in a CS v2
  nouveau: move the hotplug ignore to correct place.
  drm/nouveau/gf116: remove copy1 engine
  drm/nouveau: prevent stale fence->channel pointers, and protect with rcu
  drm/nouveau/fifo/g84-: ack non-stall interrupt before handling it

9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Wed, 3 Dec 2014 21:35:18 +0000 (13:35 -0800)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Fill in ethtool link parameters for all link types in cxgb4, from
    Hariprasad Shenai.

 2) Fix probe regressions in stmmac driver, from Huacai Chen.

 3) Network namespace leaks on errirs in rtnetlink, from Nicolas
    Dichtel.

 4) Remove erroneous BUG check which can actually trigger legitimately,
    in xen-netfront.  From Seth Forshee.

 5) Validate length of IFLA_BOND_ARP_IP_TARGET netlink attributes, from
    Thomas Grag.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
  cxgb4: Fill in supported link mode for SFP modules
  xen-netfront: Remove BUGs on paged skb data which crosses a page boundary
  sh_eth: Fix sleeping function called from invalid context
  stmmac: platform: Move plat_dat checking earlier
  sh_eth: Fix skb alloc size and alignment adjust rule.
  rtnetlink: release net refcnt on error in do_setlink()
  bond: Check length of IFLA_BOND_ARP_IP_TARGET attributes

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Wed, 3 Dec 2014 18:56:12 +0000 (10:56 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jmorris/linux-security

Pull keyring/nfs fixes from James Morris:
 "From David Howells:

  The first one fixes the handling of maximum buffer size for key
  descriptions, fixing the size at 4095 + NUL char rather than whatever
  PAGE_SIZE happens to be and permits you to read back the full
  description without it getting clipped because some extra information
  got prepended.

  The second and third fix a bug in NFS idmapper handling whereby a key
  representing a mapping between an id and a name expires and causing
  EKEYEXPIRED to be seen internally in NFS (which prevents the mapping
  from happening) rather than re-looking up the mapping"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/linux-security:
  KEYS: request_key() should reget expired keys rather than give EKEYEXPIRED
  KEYS: Simplify KEYRING_SEARCH_{NO,DO}_STATE_CHECK flags
  KEYS: Fix the size of the key description passed to/from userspace

9 years agoMerge branch 'akpm' (patches from Andrew Morton)
Linus Torvalds [Wed, 3 Dec 2014 17:36:20 +0000 (09:36 -0800)]
Merge branch 'akpm' (patches from Andrew Morton)

Merge misc fixes from Andrew Morton:
 "10 fixes"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  slab: fix nodeid bounds check for non-contiguous node IDs
  lib/genalloc.c: export devm_gen_pool_create() for modules
  mm: fix anon_vma_clone() error treatment
  mm: fix swapoff hang after page migration and fork
  fat: fix oops on corrupted vfat fs
  ipc/sem.c: fully initialize sem_array before making it visible
  drivers/input/evdev.c: don't kfree() a vmalloc address
  mm/vmpressure.c: fix race in vmpressure_work_fn()
  mm: frontswap: invalidate expired data on a dup-store failure
  mm: do not overwrite reserved pages counter at show_mem()

9 years agoslab: fix nodeid bounds check for non-contiguous node IDs
Paul Mackerras [Tue, 2 Dec 2014 23:59:48 +0000 (15:59 -0800)]
slab: fix nodeid bounds check for non-contiguous node IDs

The bounds check for nodeid in ____cache_alloc_node gives false
positives on machines where the node IDs are not contiguous, leading to
a panic at boot time.  For example, on a POWER8 machine the node IDs are
typically 0, 1, 16 and 17.  This means that num_online_nodes() returns
4, so when ____cache_alloc_node is called with nodeid = 16 the VM_BUG_ON
triggers, like this:

  kernel BUG at /home/paulus/kernel/kvm/mm/slab.c:3079!
  Call Trace:
    .____cache_alloc_node+0x5c/0x270 (unreliable)
    .kmem_cache_alloc_node_trace+0xdc/0x360
    .init_list+0x3c/0x128
    .kmem_cache_init+0x1dc/0x258
    .start_kernel+0x2a0/0x568
    start_here_common+0x20/0xa8

To fix this, we instead compare the nodeid with MAX_NUMNODES, and
additionally make sure it isn't negative (since nodeid is an int).  The
check is there mainly to protect the array dereference in the get_node()
call in the next line, and the array being dereferenced is of size
MAX_NUMNODES.  If the nodeid is in range but invalid (for example if the
node is off-line), the BUG_ON in the next line will catch that.

Fixes: 14e50c6a9bc2 ("mm: slab: Verify the nodeid passed to ____cache_alloc_node")
Signed-off-by: Paul Mackerras <paulus@samba.org>
Reviewed-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Reviewed-by: Pekka Enberg <penberg@kernel.org>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agolib/genalloc.c: export devm_gen_pool_create() for modules
Michal Simek [Tue, 2 Dec 2014 23:59:45 +0000 (15:59 -0800)]
lib/genalloc.c: export devm_gen_pool_create() for modules

Modules can use this function for creating pool.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Cc: Laura Abbott <lauraa@codeaurora.org>
Cc: Olof Johansson <olof@lixom.net>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com>
Cc: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: fix anon_vma_clone() error treatment
Daniel Forrest [Tue, 2 Dec 2014 23:59:42 +0000 (15:59 -0800)]
mm: fix anon_vma_clone() error treatment

Andrew Morton noticed that the error return from anon_vma_clone() was
being dropped and replaced with -ENOMEM (which is not itself a bug
because the only error return value from anon_vma_clone() is -ENOMEM).

I did an audit of callers of anon_vma_clone() and discovered an actual
bug where the error return was being lost.  In __split_vma(), between
Linux 3.11 and 3.12 the code was changed so the err variable is used
before the call to anon_vma_clone() and the default initial value of
-ENOMEM is overwritten.  So a failure of anon_vma_clone() will return
success since err at this point is now zero.

Below is a patch which fixes this bug and also propagates the error
return value from anon_vma_clone() in all cases.

Fixes: ef0855d334e1 ("mm: mempolicy: turn vma_set_policy() into vma_dup_policy()")
Signed-off-by: Daniel Forrest <dan.forrest@ssec.wisc.edu>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tim Hartrick <tim@edgecast.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Michel Lespinasse <walken@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org> [3.12+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: fix swapoff hang after page migration and fork
Hugh Dickins [Tue, 2 Dec 2014 23:59:39 +0000 (15:59 -0800)]
mm: fix swapoff hang after page migration and fork

I've been seeing swapoff hangs in recent testing: it's cycling around
trying unsuccessfully to find an mm for some remaining pages of swap.

I have been exercising swap and page migration more heavily recently,
and now notice a long-standing error in copy_one_pte(): it's trying to
add dst_mm to swapoff's mmlist when it finds a swap entry, but is doing
so even when it's a migration entry or an hwpoison entry.

Which wouldn't matter much, except it adds dst_mm next to src_mm,
assuming src_mm is already on the mmlist: which may not be so.  Then if
pages are later swapped out from dst_mm, swapoff won't be able to find
where to replace them.

There's already a !non_swap_entry() test for stats: move that up before
the swap_duplicate() and the addition to mmlist.

Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Kelley Nielsen <kelleynnn@gmail.com>
Cc: <stable@vger.kernel.org> [2.6.18+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agofat: fix oops on corrupted vfat fs
Al Viro [Tue, 2 Dec 2014 23:59:37 +0000 (15:59 -0800)]
fat: fix oops on corrupted vfat fs

a) don't bother with ->d_time for positives - we only check it for
   negatives anyway.

b) make sure to set it at unlink and rmdir time - at *that* point
   soon-to-be negative dentry matches then-current directory contents

c) don't go into renaming of old alias in vfat_lookup() unless it
   has the same parent (which it will, unless we are seeing corrupted
   image)

[hirofumi@mail.parknet.co.jp: make change minimum, don't call d_move() for dir]
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Cc: <stable@vger.kernel.org> [3.17.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoipc/sem.c: fully initialize sem_array before making it visible
Manfred Spraul [Tue, 2 Dec 2014 23:59:34 +0000 (15:59 -0800)]
ipc/sem.c: fully initialize sem_array before making it visible

ipc_addid() makes a new ipc identifier visible to everyone.  New objects
start as locked, so that the caller can complete the initialization
after the call.  Within struct sem_array, at least sma->sem_base and
sma->sem_nsems are accessed without any locks, therefore this approach
doesn't work.

Thus: Move the ipc_addid() to the end of the initialization.

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Reported-by: Rik van Riel <riel@redhat.com>
Acked-by: Rik van Riel <riel@redhat.com>
Acked-by: Davidlohr Bueso <dave@stgolabs.net>
Acked-by: Rafael Aquini <aquini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agodrivers/input/evdev.c: don't kfree() a vmalloc address
Andrew Morton [Tue, 2 Dec 2014 23:59:31 +0000 (15:59 -0800)]
drivers/input/evdev.c: don't kfree() a vmalloc address

If kzalloc() failed and then evdev_open_device() fails, evdev_open()
will pass a vmalloc'ed pointer to kfree.

This might fix https://bugzilla.kernel.org/show_bug.cgi?id=88401, where
there was a crash in kfree().

Reported-by: Christian Casteyde <casteyde.christian@free.fr>
Belatedly-Acked-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: Henrik Rydberg <rydberg@euromail.se>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agocxgb4: Fill in supported link mode for SFP modules
Hariprasad Shenai [Fri, 28 Nov 2014 13:05:14 +0000 (18:35 +0530)]
cxgb4: Fill in supported link mode for SFP modules

Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoxen-netfront: Remove BUGs on paged skb data which crosses a page boundary
Seth Forshee [Wed, 26 Nov 2014 02:28:24 +0000 (20:28 -0600)]
xen-netfront: Remove BUGs on paged skb data which crosses a page boundary

These BUGs can be erroneously triggered by frags which refer to
tail pages within a compound page. The data in these pages may
overrun the hardware page while still being contained within the
compound page, but since compound_order() evaluates to 0 for tail
pages the assertion fails. The code already iterates through
subsequent pages correctly in this scenario, so the BUGs are
unnecessary and can be removed.

Fixes: f36c374782e4 ("xen/netfront: handle compound page fragments on transmit")
Cc: <stable@vger.kernel.org> # 3.7+
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Reviewed-by: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agomm/vmpressure.c: fix race in vmpressure_work_fn()
Andrew Morton [Tue, 2 Dec 2014 23:59:28 +0000 (15:59 -0800)]
mm/vmpressure.c: fix race in vmpressure_work_fn()

In some android devices, there will be a "divide by zero" exception.
vmpr->scanned could be zero before spin_lock(&vmpr->sr_lock).

Addresses https://bugzilla.kernel.org/show_bug.cgi?id=88051

[akpm@linux-foundation.org: neaten]
Reported-by: ji_ang <ji_ang@163.com>
Cc: Anton Vorontsov <anton.vorontsov@linaro.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: frontswap: invalidate expired data on a dup-store failure
Weijie Yang [Tue, 2 Dec 2014 23:59:25 +0000 (15:59 -0800)]
mm: frontswap: invalidate expired data on a dup-store failure

If a frontswap dup-store failed, it should invalidate the expired page
in the backend, or it could trigger some data corruption issue.
Such as:
 1. use zswap as the frontswap backend with writeback feature
 2. store a swap page(version_1) to entry A, success
 3. dup-store a newer page(version_2) to the same entry A, fail
 4. use __swap_writepage() write version_2 page to swapfile, success
 5. zswap do shrink, writeback version_1 page to swapfile
 6. version_2 page is overwrited by version_1, data corrupt.

This patch fixes this issue by invalidating expired data immediately
when meet a dup-store failure.

Signed-off-by: Weijie Yang <weijie.yang@samsung.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Seth Jennings <sjennings@variantweb.net>
Cc: Dan Streetman <ddstreet@ieee.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Bob Liu <bob.liu@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: do not overwrite reserved pages counter at show_mem()
Rafael Aquini [Tue, 2 Dec 2014 23:59:22 +0000 (15:59 -0800)]
mm: do not overwrite reserved pages counter at show_mem()

Minor fixlet to perform the reserved pages counter aggregation for each
node, at show_mem()

Signed-off-by: Rafael Aquini <aquini@redhat.com>
Acked-by: Rik van Riel <riel@redhat.com>
Acked-by: Johannes Weiner <jweiner@redhat.com>
Acked-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge branch 'drm-fixes-3.18' of git://people.freedesktop.org/~agd5f/linux into drm...
Dave Airlie [Wed, 3 Dec 2014 00:48:54 +0000 (10:48 +1000)]
Merge branch 'drm-fixes-3.18' of git://people.freedesktop.org/~agd5f/linux into drm-fixes

A few more small fixes for 3.18.

* 'drm-fixes-3.18' of git://people.freedesktop.org/~agd5f/linux:
  drm/radeon: kernel panic in drm_calc_vbltimestamp_from_scanoutpos with 3.18.0-rc6
  drm/radeon: Ignore RADEON_GEM_GTT_WC on 32-bit x86
  drm/radeon: sync all BOs involved in a CS v2

9 years agodrm/radeon: kernel panic in drm_calc_vbltimestamp_from_scanoutpos with 3.18.0-rc6
Petr Mladek [Thu, 27 Nov 2014 15:57:21 +0000 (16:57 +0100)]
drm/radeon: kernel panic in drm_calc_vbltimestamp_from_scanoutpos with 3.18.0-rc6

I was unable too boot 3.18.0-rc6 because of the following kernel
panic in drm_calc_vbltimestamp_from_scanoutpos():

    [drm] Initialized drm 1.1.0 20060810
    [drm] radeon kernel modesetting enabled.
    [drm] initializing kernel modesetting (RV100 0x1002:0x515E 0x15D9:0x8080).
    [drm] register mmio base: 0xC8400000
    [drm] register mmio size: 65536
    radeon 0000:0b:01.0: VRAM: 128M 0x00000000D0000000 - 0x00000000D7FFFFFF (16M used)
    radeon 0000:0b:01.0: GTT: 512M 0x00000000B0000000 - 0x00000000CFFFFFFF
    [drm] Detected VRAM RAM=128M, BAR=128M
    [drm] RAM width 16bits DDR
    [TTM] Zone  kernel: Available graphics memory: 3829346 kiB
    [TTM] Zone   dma32: Available graphics memory: 2097152 kiB
    [TTM] Initializing pool allocator
    [TTM] Initializing DMA pool allocator
    [drm] radeon: 16M of VRAM memory ready
    [drm] radeon: 512M of GTT memory ready.
    [drm] GART: num cpu pages 131072, num gpu pages 131072
    [drm] PCI GART of 512M enabled (table at 0x0000000037880000).
    radeon 0000:0b:01.0: WB disabled
    radeon 0000:0b:01.0: fence driver on ring 0 use gpu addr 0x00000000b0000000 and cpu addr 0xffff8800bbbfa000
    [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
    [drm] Driver supports precise vblank timestamp query.
    [drm] radeon: irq initialized.
    [drm] Loading R100 Microcode
    radeon 0000:0b:01.0: Direct firmware load for radeon/R100_cp.bin failed with error -2
    radeon_cp: Failed to load firmware "radeon/R100_cp.bin"
    [drm:r100_cp_init] *ERROR* Failed to load firmware!
    radeon 0000:0b:01.0: failed initializing CP (-2).
    radeon 0000:0b:01.0: Disabling GPU acceleration
    [drm] radeon: cp finalized
    BUG: unable to handle kernel NULL pointer dereference at 000000000000025c
    IP: [<ffffffff8150423b>] drm_calc_vbltimestamp_from_scanoutpos+0x4b/0x320
    PGD 0
    Oops: 0000 [#1] SMP
    Modules linked in:
    CPU: 1 PID: 1 Comm: swapper/0 Not tainted 3.18.0-rc6-4-default #2649
    Hardware name: Supermicro X7DB8/X7DB8, BIOS 6.00 07/26/2006
    task: ffff880234da2010 ti: ffff880234da4000 task.ti: ffff880234da4000
    RIP: 0010:[<ffffffff8150423b>]  [<ffffffff8150423b>] drm_calc_vbltimestamp_from_scanoutpos+0x4b/0x320
    RSP: 0000:ffff880234da7918  EFLAGS: 00010086
    RAX: ffffffff81557890 RBX: 0000000000000000 RCX: ffff880234da7a48
    RDX: ffff880234da79f4 RSI: 0000000000000000 RDI: ffff880232e15000
    RBP: ffff880234da79b8 R08: 0000000000000000 R09: 0000000000000000
    R10: 000000000000000a R11: 0000000000000001 R12: ffff880232dda1c0
    R13: ffff880232e1518c R14: 0000000000000292 R15: ffff880232e15000
    FS:  0000000000000000(0000) GS:ffff88023fc40000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
    CR2: 000000000000025c CR3: 0000000002014000 CR4: 00000000000007e0
    Stack:
     ffff880234da79d8 0000000000000286 ffff880232dcbc00 0000000000002480
     ffff880234da7958 0000000000000296 ffff880234da7998 ffffffff8151b51d
     ffff880234da7a48 0000000032dcbeb0 ffff880232dcbc00 ffff880232dcbc58
    Call Trace:
     [<ffffffff8151b51d>] ? drm_vma_offset_remove+0x1d/0x110
     [<ffffffff8152dc98>] radeon_get_vblank_timestamp_kms+0x38/0x60
     [<ffffffff8152076a>] ? ttm_bo_release_list+0xba/0x180
     [<ffffffff81503751>] drm_get_last_vbltimestamp+0x41/0x70
     [<ffffffff81503933>] vblank_disable_and_save+0x73/0x1d0
     [<ffffffff81106b2f>] ? try_to_del_timer_sync+0x4f/0x70
     [<ffffffff81505245>] drm_vblank_cleanup+0x65/0xa0
     [<ffffffff815604fa>] radeon_irq_kms_fini+0x1a/0x70
     [<ffffffff8156c07e>] r100_init+0x26e/0x410
     [<ffffffff8152ae3e>] radeon_device_init+0x7ae/0xb50
     [<ffffffff8152d57f>] radeon_driver_load_kms+0x8f/0x210
     [<ffffffff81506965>] drm_dev_register+0xb5/0x110
     [<ffffffff8150998f>] drm_get_pci_dev+0x8f/0x200
     [<ffffffff815291cd>] radeon_pci_probe+0xad/0xe0
     [<ffffffff8141a365>] local_pci_probe+0x45/0xa0
     [<ffffffff8141b741>] pci_device_probe+0xd1/0x130
     [<ffffffff81633dad>] driver_probe_device+0x12d/0x3e0
     [<ffffffff8163413b>] __driver_attach+0x9b/0xa0
     [<ffffffff816340a0>] ? __device_attach+0x40/0x40
     [<ffffffff81631cd3>] bus_for_each_dev+0x63/0xa0
     [<ffffffff8163378e>] driver_attach+0x1e/0x20
     [<ffffffff81633390>] bus_add_driver+0x180/0x240
     [<ffffffff81634914>] driver_register+0x64/0xf0
     [<ffffffff81419cac>] __pci_register_driver+0x4c/0x50
     [<ffffffff81509bf5>] drm_pci_init+0xf5/0x120
     [<ffffffff821dc871>] ? ttm_init+0x6a/0x6a
     [<ffffffff821dc908>] radeon_init+0x97/0xb5
     [<ffffffff810002fc>] do_one_initcall+0xbc/0x1f0
     [<ffffffff810e3278>] ? __wake_up+0x48/0x60
     [<ffffffff8218e256>] kernel_init_freeable+0x18a/0x215
     [<ffffffff8218d983>] ? initcall_blacklist+0xc0/0xc0
     [<ffffffff818a78f0>] ? rest_init+0x80/0x80
     [<ffffffff818a78fe>] kernel_init+0xe/0xf0
     [<ffffffff818c0c3c>] ret_from_fork+0x7c/0xb0
     [<ffffffff818a78f0>] ? rest_init+0x80/0x80
    Code: 45 ac 0f 88 a8 01 00 00 3b b7 d0 01 00 00 49 89 ff 0f 83 99 01 00 00 48 8b 47 20 48 8b 80 88 00 00 00 48 85 c0 0f 84 cd 01 00 00 <41> 8b b1 5c 02 00 00 41 8b 89 58 02 00 00 89 75 98 41 8b b1 60
    RIP  [<ffffffff8150423b>] drm_calc_vbltimestamp_from_scanoutpos+0x4b/0x320
     RSP <ffff880234da7918>
    CR2: 000000000000025c
    ---[ end trace ad2c0aadf48e2032 ]---
    Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000009

It has helped me to add a NULL pointer check that was suggested at
http://lists.freedesktop.org/archives/dri-devel/2014-October/070663.html

I am not familiar with the code. But the change looks sane
and we need something fast at this stage of 3.18 development.

Suggested-by: Helge Deller <deller@gmx.de>
Signed-off-by: Petr Mladek <pmladek@suse.cz>
Tested-by: Petr Mladek <pmladek@suse.cz>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
9 years agodrm/radeon: Ignore RADEON_GEM_GTT_WC on 32-bit x86
Michel Dänzer [Thu, 27 Nov 2014 09:00:54 +0000 (18:00 +0900)]
drm/radeon: Ignore RADEON_GEM_GTT_WC on 32-bit x86

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=84627
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
9 years agodrm/radeon: sync all BOs involved in a CS v2
Christian König [Thu, 27 Nov 2014 12:12:58 +0000 (13:12 +0100)]
drm/radeon: sync all BOs involved in a CS v2

Not just the userspace relocs, otherwise we won't wait
for a swapped out page tables to be swapped in again.

v2: rebased on Alex current drm-fixes-3.18

Signed-off-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
9 years agosata_fsl: fix error handling of irq_of_parse_and_map
Dmitry Torokhov [Fri, 14 Nov 2014 21:39:05 +0000 (13:39 -0800)]
sata_fsl: fix error handling of irq_of_parse_and_map

irq_of_parse_and_map() returns 0 on error (the result is unsigned int),
so testing for negative result never works.

Signed-off-by: Dmitry Torokhov <dtor@chromium.org>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: stable@vger.kernel.org
9 years agoAHCI: Add DeviceIDs for Sunrise Point-LP SATA controller
Devin Ryles [Fri, 7 Nov 2014 22:59:05 +0000 (17:59 -0500)]
AHCI: Add DeviceIDs for Sunrise Point-LP SATA controller

This patch adds DeviceIDs for Sunrise Point-LP.

Signed-off-by: Devin Ryles <devin.ryles@intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: stable@vger.kernel.org
9 years agoblock: fix regression where bio_integrity_process uses wrong bio_vec iterator
Darrick J. Wong [Wed, 26 Nov 2014 01:40:25 +0000 (17:40 -0800)]
block: fix regression where bio_integrity_process uses wrong bio_vec iterator

bio integrity handling is broken on a system with LVM layered atop a
DIF/DIX SCSI drive because device mapper clones the bio, modifies the
clone, and sends the clone to the lower layers for processing.
However, the clone bio has bi_vcnt == 0, which means that when the sd
driver calls bio_integrity_process to attach DIX data, the
for_each_segment_all() call (which uses bi_vcnt) returns immediately
and random garbage is sent to the disk on a disk write.  The disk of
course returns an error.

Therefore, teach bio_integrity_process() to use bio_for_each_segment()
to iterate the bio_vecs, since the per-bio iterator tracks which
bio_vecs are associated with that particular bio.  The integrity
handling code is effectively part of the "driver" (it's not the bio
owner), so it must use the correct iterator function.

v2: Fix a compiler warning about abandoned local variables.  This
patch supersedes "block: bio_integrity_process uses wrong bio_vec
iterator".  Patch applies against 3.18-rc6.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Acked-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
9 years agodrm/i915: Unlock panel even when LVDS is disabled
Daniel Vetter [Mon, 1 Dec 2014 16:56:54 +0000 (17:56 +0100)]
drm/i915: Unlock panel even when LVDS is disabled

Otherwise we'll have backtraces in assert_panel_unlocked because the
BIOS locks the register. In the reporter's case this regression was
introduced in

commit c31407a3672aaebb4acddf90944a114fa5c8af7b
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Thu Oct 18 21:07:01 2012 +0100

    drm/i915: Add no-lvds quirk for Supermicro X7SPA-H

Reported-by: Alexey Orishko <alexey.orishko@gmail.com>
Cc: Alexey Orishko <alexey.orishko@gmail.com>
Cc: stable@vger.kernel.org
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Francois Tigeot <ftigeot@wolfpond.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Tested-by: Alexey Orishko <alexey.orishko@gmail.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: More cautious with pch fifo underruns
Daniel Vetter [Mon, 24 Nov 2014 16:02:45 +0000 (17:02 +0100)]
drm/i915: More cautious with pch fifo underruns

Apparently PCH fifo underruns are tricky, we have plenty reports that
we see the occasional underrun (especially at boot-up).

So for a change let's see what happens when we don't re-enable pch
fifo underrun reporting when the pipe is disabled. This means that the
kernel can't catch pch fifo underruns when they happen (except when
all pipes are on on the pch). But we'll still catch underruns when
disabling the pipe again. So not a terrible reduction in test
coverage.

Since the DRM_ERROR is new and hence a regression plan B would be to
revert it back to a debug output. Which would be a lot worse than this
hack for underrun test coverage in the wild. See the referenced
discussions for more.

References: http://mid.gmane.org/CA+gsUGRfGe3t4NcjdeA=qXysrhLY3r4CEu7z4bjTwxi1uOfy+g@mail.gmail.com
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=85898
References: https://bugs.freedesktop.org/show_bug.cgi?id=85898
References: https://bugs.freedesktop.org/show_bug.cgi?id=86233
References: https://bugs.freedesktop.org/show_bug.cgi?id=86478
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Tested-by: lu hua <huax.lu@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agoALSA: hda/realtek - Add headset Mic support for new Dell machine
Kailang Yang [Tue, 2 Dec 2014 06:50:45 +0000 (14:50 +0800)]
ALSA: hda/realtek - Add headset Mic support for new Dell machine

Dell has new machines. It supports headset Mic and Headphone Mic.

Signed-off-by: Kailang Yang <kailang@realtek.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
9 years agoMerge tag 'keys-fixes-20141201' of git://git.kernel.org/pub/scm/linux/kernel/git...
James Morris [Tue, 2 Dec 2014 07:06:12 +0000 (18:06 +1100)]
Merge tag 'keys-fixes-20141201' of git://git./linux/kernel/git/dhowells/linux-fs into for-linus

9 years agonouveau: move the hotplug ignore to correct place.
Dave Airlie [Tue, 2 Dec 2014 06:27:25 +0000 (16:27 +1000)]
nouveau: move the hotplug ignore to correct place.

Introduced in b440bde74f, however it was added to
the wrong function in nouveau.

https://bugzilla.kernel.org/show_bug.cgi?id=86011
Cc: Bjorn Helgaas <bhelgaas@google.com>
CC: stable@vger.kernel.org # v3.15+
Signed-off-by: Dave Airlie <airlied@redhat.com>
9 years agoMerge branch 'linux-3.18' of git://anongit.freedesktop.org/git/nouveau/linux-2.6...
Dave Airlie [Tue, 2 Dec 2014 05:40:16 +0000 (15:40 +1000)]
Merge branch 'linux-3.18' of git://anongit.freedesktop.org/git/nouveau/linux-2.6 into drm-fixes

Just a couple of fixes for the fallout from the fence rework.

* 'linux-3.18' of git://anongit.freedesktop.org/git/nouveau/linux-2.6:
  drm/nouveau/gf116: remove copy1 engine
  drm/nouveau: prevent stale fence->channel pointers, and protect with rcu
  drm/nouveau/fifo/g84-: ack non-stall interrupt before handling it

9 years agodrm/nouveau/gf116: remove copy1 engine
Ilia Mirkin [Sun, 30 Nov 2014 17:56:18 +0000 (12:56 -0500)]
drm/nouveau/gf116: remove copy1 engine

Indications are that no GF116's actually have a copy engine there, but
actually have the decompression engine. This engine can be made to do
copies, but that should be done separately.

Unclear why this didn't turn up on all GF116's, but perhaps the
non-mobile ones came with enough VRAM to not trigger ttm migrations in
test scenarios.

Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=85465
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=59168
Cc: stable@vger.kernel.org
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
9 years agodrm/nouveau: prevent stale fence->channel pointers, and protect with rcu
Maarten Lankhorst [Mon, 1 Dec 2014 09:11:06 +0000 (19:11 +1000)]
drm/nouveau: prevent stale fence->channel pointers, and protect with rcu

Tested-by: Alexandre Courbot <acourbot@nvidia.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
9 years agodrm/nouveau/fifo/g84-: ack non-stall interrupt before handling it
Ben Skeggs [Mon, 1 Dec 2014 01:44:27 +0000 (11:44 +1000)]
drm/nouveau/fifo/g84-: ack non-stall interrupt before handling it

Closes a very unlikely race that can occur if another NonStallInterrupt
method passes between checking fences and acking the previous interrupt.

With this change, the interrupt will re-fire under such conditions.

Tested-by: Tobias Klausmann <tobias.johannes.klausmann@mni.thm.de>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
9 years agoMerge tag 'ext4_for_linus_urgent' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 2 Dec 2014 04:11:49 +0000 (20:11 -0800)]
Merge tag 'ext4_for_linus_urgent' of git://git./linux/kernel/git/tytso/ext4

Pull ext4 bugfix from Ted Ts'o:
 "Fix an ext4 metadata checksum regression introduced in v3.18-rc3"

* tag 'ext4_for_linus_urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  jbd2: fix regression where we fail to initialize checksum seed when loading

9 years agojbd2: fix regression where we fail to initialize checksum seed when loading
Darrick J. Wong [Tue, 2 Dec 2014 00:22:23 +0000 (16:22 -0800)]
jbd2: fix regression where we fail to initialize checksum seed when loading

When we're enabling journal features, we cannot use the predicate
jbd2_journal_has_csum_v2or3() because we haven't yet set the sb
feature flag fields!  Moreover, we just finished loading the shash
driver, so the test is unnecessary; calculate the seed always.

Without this patch, we fail to initialize the checksum seed the first
time we turn on journal_checksum, which means that all journal blocks
written during that first mount are corrupt.  Transactions written
after the second mount will be fine, since the feature flag will be
set in the journal superblock.  xfstests generic/{034,321,322} are the
regression tests.

(This is important for 3.18.)

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.coM>
Reported-by: Eric Whitney <enwlinux@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
9 years agoPCI: tegra: Use physical range for I/O mapping
Thierry Reding [Thu, 27 Nov 2014 08:54:09 +0000 (09:54 +0100)]
PCI: tegra: Use physical range for I/O mapping

Commit 0b0b0893d49b ("of/pci: Fix the conversion of IO ranges into IO
resources") changed how I/O resources are parsed from DT.  Rather than
containing the physical address of the I/O region, the addresses will now
be in I/O address space.

On Tegra the union of all ranges is used to expose a top-level memory-
mapped resource for the PCI host bridge.  This helps to make /proc/iomem
more readable.

Combining both of the above, the union would now include the I/O space
region.  This causes a regression on Tegra20, where the physical base
address of the PCIe controller (and therefore of the union) is located at
physical address 0x80000000.  Since I/O space starts at 0, the union will
now include all of system RAM which starts at 0x00000000.

This commit fixes this by keeping two copies of the I/O range: one that
represents the range in the CPU's physical address space, the other for the
range in the I/O address space.  This allows the translation setup within
the driver to reuse the physical addresses.  The code registering the I/O
region with the PCI core uses both ranges to establish the mapping.

Fixes: 0b0b0893d49b ("of/pci: Fix the conversion of IO ranges into IO resources")
Reported-by: Marc Zyngier <marc.zyngier@arm.com>
Tested-by: Marc Zyngier <marc.zyngier@arm.com>
Suggested-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
9 years agoKEYS: request_key() should reget expired keys rather than give EKEYEXPIRED
David Howells [Mon, 1 Dec 2014 22:52:53 +0000 (22:52 +0000)]
KEYS: request_key() should reget expired keys rather than give EKEYEXPIRED

Since the keyring facility can be viewed as a cache (at least in some
applications), the local expiration time on the key should probably be viewed
as a 'needs updating after this time' property rather than an absolute 'anyone
now wanting to use this object is out of luck' property.

Since request_key() is the main interface for the usage of keys, this should
update or replace an expired key rather than issuing EKEYEXPIRED if the local
expiration has been reached (ie. it should refresh the cache).

For absolute conditions where refreshing the cache probably doesn't help, the
key can be negatively instantiated using KEYCTL_REJECT_KEY with EKEYEXPIRED
given as the error to issue.  This will still cause request_key() to return
EKEYEXPIRED as that was explicitly set.

In the future, if the key type has an update op available, we might want to
upcall with the expired key and allow the upcall to update it.  We would pass
a different operation name (the first column in /etc/request-key.conf) to the
request-key program.

request_key() returning EKEYEXPIRED is causing an NFS problem which Chuck
Lever describes thusly:

After about 10 minutes, my NFSv4 functional tests fail because the
ownership of the test files goes to "-2". Looking at /proc/keys
shows that the id_resolv keys that map to my test user ID have
expired. The ownership problem persists until the expired keys are
purged from the keyring, and fresh keys are obtained.

I bisected the problem to 3.13 commit b2a4df200d57 ("KEYS: Expand
the capacity of a keyring"). This commit inadvertantly changes the
API contract of the internal function keyring_search_aux().

The root cause appears to be that b2a4df200d57 made "no state check"
the default behavior. "No state check" means the keyring search
iterator function skips checking the key's expiry timeout, and
returns expired keys.  request_key_and_link() depends on getting
an -EAGAIN result code to know when to perform an upcall to refresh
an expired key.

This patch can be tested directly by:

keyctl request2 user debug:fred a @s
keyctl timeout %user:debug:fred 3
sleep 4
keyctl request2 user debug:fred a @s

Without the patch, the last command gives error EKEYEXPIRED, but with the
command it gives a new key.

Reported-by: Carl Hetherington <cth@carlh.net>
Reported-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: Chuck Lever <chuck.lever@oracle.com>
9 years agoKEYS: Simplify KEYRING_SEARCH_{NO,DO}_STATE_CHECK flags
David Howells [Mon, 1 Dec 2014 22:52:50 +0000 (22:52 +0000)]
KEYS: Simplify KEYRING_SEARCH_{NO,DO}_STATE_CHECK flags

Simplify KEYRING_SEARCH_{NO,DO}_STATE_CHECK flags to be two variations of the
same flag.  They are effectively mutually exclusive and one or the other
should be provided, but not both.

Keyring cycle detection and key possession determination are the only things
that set NO_STATE_CHECK, except that neither flag really does anything there
because neither purpose makes use of the keyring_search_iterator() function,
but rather provides their own.

For cycle detection we definitely want to check inside of expired keyrings,
just so that we don't create a cycle we can't get rid of.  Revoked keyrings
are cleared at revocation time and can't then be reused, so shouldn't be a
problem either way.

For possession determination, we *might* want to validate each keyring before
searching it: do you possess a key that's hidden behind an expired or just
plain inaccessible keyring?  Currently, the answer is yes.  Note that you
cannot, however, possess a key behind a revoked keyring because they are
cleared on revocation.

keyring_search() sets DO_STATE_CHECK, which is correct.

request_key_and_link() currently doesn't specify whether to check the key
state or not - but it should set DO_STATE_CHECK.

key_get_instantiation_authkey() also currently doesn't specify whether to
check the key state or not - but it probably should also set DO_STATE_CHECK.

Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: Chuck Lever <chuck.lever@oracle.com>
9 years agoKEYS: Fix the size of the key description passed to/from userspace
David Howells [Mon, 1 Dec 2014 22:52:45 +0000 (22:52 +0000)]
KEYS: Fix the size of the key description passed to/from userspace

When a key description argument is imported into the kernel from userspace, as
happens in add_key(), request_key(), KEYCTL_JOIN_SESSION_KEYRING,
KEYCTL_SEARCH, the description is copied into a buffer up to PAGE_SIZE in size.
PAGE_SIZE, however, is a variable quantity, depending on the arch.  Fix this at
4096 instead (ie. 4095 plus a NUL termination) and define a constant
(KEY_MAX_DESC_SIZE) to this end.

When reading the description back with KEYCTL_DESCRIBE, a PAGE_SIZE internal
buffer is allocated into which the information and description will be
rendered.  This means that the description will get truncated if an extremely
long description it has to be crammed into the buffer with the stringified
information.  There is no particular need to copy the description into the
buffer, so just copy it directly to userspace in a separate operation.

Reported-by: Christian Kastner <debian@kvr.at>
Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: Christian Kastner <debian@kvr.at>
9 years agos390: fix machine check handling
Sebastian Ott [Fri, 28 Nov 2014 14:40:57 +0000 (15:40 +0100)]
s390: fix machine check handling

Commit eb7e7d76 "s390: Replace __get_cpu_var uses" broke machine check
handling.

We copy machine check information from per-cpu to a stack variable for
local processing. Next we should zap the per-cpu variable, not the
stack variable.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Acked-by: Christoph Lameter <cl@linux.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>