8 years agoMerge commit 'v3.4.4' into android-tegra-nv-3.4
Varun Wadekar [Mon, 9 Jul 2012 10:47:52 +0000]
Merge commit 'v3.4.4' into android-tegra-nv-3.4

8 years agoARM: vfp: only clear vfp state for current cpu in vfp_pm_suspend
Colin Cross [Sat, 28 Apr 2012 01:04:18 +0000]
ARM: vfp: only clear vfp state for current cpu in vfp_pm_suspend

vfp_pm_suspend runs on each cpu, only clear the hardware state
pointer for the current cpu.  Prevents a possible crash if one
cpu clears the hw state pointer when another cpu has already
checked if it is valid.

Change-Id: I997ab1554944eba86730818ff242d7ebe1b32736
Signed-off-by: Colin Cross <ccross@android.com>
Reviewed-on: http://git-master/r/114168
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Prashant Gaikwad <pgaikwad@nvidia.com>
Tested-by: Prashant Gaikwad <pgaikwad@nvidia.com>
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoarm: vfp: Fix memory corruption on PM suspend
Ido Yariv [Sat, 14 Apr 2012 20:20:30 +0000]
arm: vfp: Fix memory corruption on PM suspend

Commit 36af2a47 ("ARM: vfp: Always save VFP state in vfp_pm_suspend")
introduced a potential use-after-free bug. On SMP systems,
vfp_current_hw_state might hold dangling pointers in case a task which
used the VFP last migrates to another CPU and then exits. If
vfp_pm_suspend is called while vfp_current_hw_state still holds a
pointer to the freed thread_info, that memory location will be written,
potentially overwriting a new object allocated there.

The original problem is only relevant to UP systems in which the VFP
state is stored lazily.

Fix this by only storing the VFP state on UP systems, and avoid doing so
on SMP ones.

Change-Id: I8f7026eb735b340fcef4cf12fbd12b9a0ea08d3f
Signed-off-by: Ido Yariv <ido@wizery.com>
Signed-off-by: Eyal Shapira <eyal@wizery.com>
Signed-off-by: Colin Cross <ccross@android.com>
Reviewed-on: http://git-master/r/114167
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Prashant Gaikwad <pgaikwad@nvidia.com>
Tested-by: Prashant Gaikwad <pgaikwad@nvidia.com>
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoRevert "ARM: vfp: Always save VFP state in vfp_pm_suspend"
Prashant Gaikwad [Mon, 9 Jul 2012 08:19:53 +0000]
Revert "ARM: vfp: Always save VFP state in vfp_pm_suspend"

This reverts commit 342a48aab5112922a701461b40e143b86465a0a7.

Change-Id: I14d31098f6d172c79aeb5e823813ebe0c0b76e14
Reviewed-on: http://git-master/r/114166
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Prashant Gaikwad <pgaikwad@nvidia.com>
Tested-by: Prashant Gaikwad <pgaikwad@nvidia.com>
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoARM: tegra: fix build for smc with 4.6 toolchain
Dima Zavin [Mon, 30 Apr 2012 23:23:56 +0000]
ARM: tegra: fix build for smc with 4.6 toolchain

Reviewed-on: http://git-master/r/99741
(cherry picked from commit 828c895ed9e74bffdb55a18d6a010350a4dd7c50)

Change-Id: Icf806dc87ec538b8b3604f8c545fd3fc21ec0a3e
Signed-off-by: Dima Zavin <dima@android.com>
Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Reviewed-on: http://git-master/r/113525
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agomedia: video: tegra: fix uninitialized variable error
Nitin Kumbhar [Wed, 4 Jul 2012 01:14:55 +0000]
media: video: tegra: fix uninitialized variable error

data_ptr may be used uninitialized in ad5816_param_rd().
Initialize data_ptr variable with NULL to fix the error.

Change-Id: I60af359ebffaef4f244ad49751b486424cb521c8
Signed-off-by: Nitin Kumbhar <nkumbhar@nvidia.com>
Reviewed-on: http://git-master/r/113421
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>
Tested-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agotty: serial: tegra: deadlock with tty_wakeup
Pradeep Kumar [Wed, 20 Jun 2012 11:17:06 +0000]
tty: serial: tegra: deadlock with tty_wakeup

Tegra hsuart calls uart write wakeup which inturn calls tty_wakeup
from irq context while holding device spin lock. If line discipline
driver  calls a write call then system ends up deadlock on device
spinlock. Using tasklet to call uart write wakeup solves the problem.

Bug 989309
Signed-off-by: Pradeep Kumar <pgoudagunta@nvidia.com>

Change-Id: I11fb5f691356da1146bde0796d5b1c247eaca68a
Reviewed-on: http://git-master/r/110029
Tested-by: Pradeep Goudagunta <pgoudagunta@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>
Reviewed-by: Laxman Dewangan <ldewangan@nvidia.com>

8 years agoLinux 3.4.4
Greg Kroah-Hartman [Fri, 22 Jun 2012 18:37:50 +0000]
Linux 3.4.4

8 years agotarget: Return error to initiator if SET TARGET PORT GROUPS emulation fails
Roland Dreier [Tue, 5 Jun 2012 06:24:51 +0000]
target: Return error to initiator if SET TARGET PORT GROUPS emulation fails

commit 59e4f541baf728dbb426949bfa9f6862387ffd0e upstream.

The error paths in target_emulate_set_target_port_groups() are all
essentially "rc = -EINVAL; goto out;" but the code at "out:" ignores
rc and always returns success.  This means that even if eg explicit
ALUA is turned off, the initiator will always see a good SCSI status
for SET TARGET PORT GROUPS.

Fix this by returning rc as is intended.  It appears this bug was
added by the following patch:

commit 05d1c7c0d0db4cc25548d9aadebb416888a82327
Author: Andy Grover <agrover@redhat.com>
Date:   Wed Jul 20 19:13:28 2011 +0000

    target: Make all control CDBs scatter-gather

Signed-off-by: Roland Dreier <roland@purestorage.com>
Cc: Andy Grover <agrover@redhat.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
[bwh: Backported to 3.2: we have transport_complete_task()
 and not target_complete_cmd()]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoiwlwifi: fix TX power antenna access
Johannes Berg [Wed, 30 May 2012 08:36:12 +0000]
iwlwifi: fix TX power antenna access

commit a5fdde28b4f5fb756032e7ad2c6fcdcffde20958 upstream.

Since my commit
  iwlwifi: use valid TX/RX antenna from hw_params
the config values are pure overrides, not the
real values for all hardware. Therefore, the
EEPROM TX power reading code checks the wrong
values, it should check the hw_params values.

Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoiwlwifi: use correct supported firmware for 6035 and 6000g2
Meenakshi Venkataraman [Tue, 5 Jun 2012 18:24:37 +0000]
iwlwifi: use correct supported firmware for 6035 and 6000g2

commit d2c8b15d0cb486f4938ba7f2af349d9d1220cb10 upstream.

My patch

   iwlwifi: use correct released ucode version

did not correctly report supported firmware
for the 6035 device. This patch fixes it. The
minimum supported firmware version for 6035
is v6.

Also correct the minimum supported firmware
version for the 6000g2 series of devices.

Signed-off-by: Meenakshi Venkataraman <meenakshi.venkataraman@intel.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoiwlwifi: fix the Transmit Frame Descriptor rings
Emmanuel Grumbach [Wed, 16 May 2012 20:35:58 +0000]
iwlwifi: fix the Transmit Frame Descriptor rings

commit ebed633c61c023e5d1aa4ed159cd67406e9e37c2 upstream.

The logic that allows to have a short TFD queue was completely wrong.
We do maintain 256 Transmit Frame Descriptors, but they point to
recycled buffers. We used to attach and de-attach different TFDs for
the same buffer and it worked since they pointed to the same buffer.

Also zero the number of BDs after unmapping a TFD. This seems not
necessary since we don't reclaim the same TFD twice, but I like
housekeeping.

This patch solves this warning:

[ 6427.079855] WARNING: at lib/dma-debug.c:866 check_unmap+0x727/0x7a0()
[ 6427.079859] Hardware name: Latitude E6410
[ 6427.079865] iwlwifi 0000:02:00.0: DMA-API: device driver tries to free DMA memory it has not allocated [device address=0x00000000296d393c] [size=8 bytes]
[ 6427.079870] Modules linked in: ...
[ 6427.079950] Pid: 6613, comm: ifconfig Tainted: G           O 3.3.3 #5
[ 6427.079954] Call Trace:
[ 6427.079963]  [<c10337a2>] warn_slowpath_common+0x72/0xa0
[ 6427.079982]  [<c1033873>] warn_slowpath_fmt+0x33/0x40
[ 6427.079988]  [<c12dcb77>] check_unmap+0x727/0x7a0
[ 6427.079995]  [<c12dcdaa>] debug_dma_unmap_page+0x5a/0x80
[ 6427.080024]  [<fe2312ac>] iwlagn_unmap_tfd+0x12c/0x180 [iwlwifi]
[ 6427.080048]  [<fe231349>] iwlagn_txq_free_tfd+0x49/0xb0 [iwlwifi]
[ 6427.080071]  [<fe228e37>] iwl_tx_queue_unmap+0x67/0x90 [iwlwifi]
[ 6427.080095]  [<fe22d221>] iwl_trans_pcie_stop_device+0x341/0x7b0 [iwlwifi]
[ 6427.080113]  [<fe204b0e>] iwl_down+0x17e/0x260 [iwlwifi]
[ 6427.080132]  [<fe20efec>] iwlagn_mac_stop+0x6c/0xf0 [iwlwifi]
[ 6427.080168]  [<fd8480ce>] ieee80211_stop_device+0x5e/0x190 [mac80211]
[ 6427.080198]  [<fd833208>] ieee80211_do_stop+0x288/0x620 [mac80211]
[ 6427.080243]  [<fd8335b7>] ieee80211_stop+0x17/0x20 [mac80211]
[ 6427.080250]  [<c148dac1>] __dev_close_many+0x81/0xd0
[ 6427.080270]  [<c148db3d>] __dev_close+0x2d/0x50
[ 6427.080276]  [<c148d152>] __dev_change_flags+0x82/0x150
[ 6427.080282]  [<c148e3e3>] dev_change_flags+0x23/0x60
[ 6427.080289]  [<c14f6320>] devinet_ioctl+0x6a0/0x770
[ 6427.080296]  [<c14f8705>] inet_ioctl+0x95/0xb0
[ 6427.080304]  [<c147a0f0>] sock_ioctl+0x70/0x270

Reported-by: Antonio Quartulli <ordex@autistici.org>
Tested-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Reviewed-by: Wey-Yi W Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agontp: Correct TAI offset during leap second
Richard Cochran [Thu, 26 Apr 2012 12:11:32 +0000]
ntp: Correct TAI offset during leap second

commit dd48d708ff3e917f6d6b6c2b696c3f18c019feed upstream.

When repeating a UTC time value during a leap second (when the UTC
time should be 23:59:60), the TAI timescale should not stop. The kernel
NTP code increments the TAI offset one second too late. This patch fixes
the issue by incrementing the offset during the leap second itself.

Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoe1000e: Remove special case for 82573/82574 ASPM L1 disablement
Chris Boot [Tue, 24 Apr 2012 07:24:58 +0000]
e1000e: Remove special case for 82573/82574 ASPM L1 disablement

commit 59aed95263bdd0e2b48eb9be5a94346d2d4abf90 upstream.

For the 82573, ASPM L1 gets disabled wholesale so this special-case code
is not required. For the 82574 the previous patch does the same as for
the 82573, disabling L1 on the adapter. Thus, this code is no longer
required and can be removed.

Signed-off-by: Chris Boot <bootc@bootc.net>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoe1000e: Disable ASPM L1 on 82574
Chris Boot [Tue, 24 Apr 2012 07:24:52 +0000]
e1000e: Disable ASPM L1 on 82574

commit d4a4206ebbaf48b55803a7eb34e330530d83a889 upstream.

ASPM on the 82574 causes trouble. Currently the driver disables L0s for
this NIC but only disables L1 if the MTU is >1500. This patch simply
causes L1 to be disabled regardless of the MTU setting.

Signed-off-by: Chris Boot <bootc@bootc.net>
Cc: "Wyborny, Carolyn" <carolyn.wyborny@intel.com>
Cc: Nix <nix@esperi.org.uk>
Link: https://lkml.org/lkml/2012/3/19/362
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoremoteproc: fix missing fault indication in error-path
Sjur Brændeland [Sun, 10 Jun 2012 11:37:51 +0000]
remoteproc: fix missing fault indication in error-path

commit 30338cf09f82523d8747670f7363cc8af347c79f upstream.

If rproc_find_rsc_table() fails, rproc_fw_boot() must set
return-value before jumping to clean_up label. Otherwise no
error value is returned.

Signed-off-by: Sjur Brændeland <sjur.brandeland@stericsson.com>
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoremoteproc: fix print format warnings
Sjur Brændeland [Sun, 10 Jun 2012 11:37:07 +0000]
remoteproc: fix print format warnings

commit e981f6d41acda2ae8c05e60feb2cb97772b4a6e6 upstream.

Fix compile warnings from GCC 4.6.1 when printing values of type size_t.

drivers/remoteproc/remoteproc_core.c:251:6:
warning: format ‘%x’ expects argument of type ‘unsigned int’,
but argument 4 has type ‘size_t’ [-Wformat]
drivers/remoteproc/remoteproc_core.c:938:9:
warning: format ‘%u’ expects argument of type ‘unsigned int’,
but argument 4 has type ‘size_t’ [-Wformat]
drivers/remoteproc/remoteproc_core.c:1023:2:
warning: format ‘%d’ expects argument of type ‘int’,
but argument 4 has type ‘size_t’ [-Wformat]

Signed-off-by: Sjur Brændeland <sjur.brandeland@stericsson.com>
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoremoteproc/omap: fix dev_err typo
Ohad Ben-Cohen [Mon, 21 May 2012 13:31:12 +0000]
remoteproc/omap: fix dev_err typo

commit 6b03976288538a94e072bbfcd12d69a20daea8aa upstream.

For some reason one of the dev_err invocations is using a wrong
device so fix that.

Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years ago9p: BUG before corrupting memory
Sasha Levin [Mon, 11 Jun 2012 15:18:13 +0000]
9p: BUG before corrupting memory

commit 5fcb08befaf57faa1b00e514915c1660252b8c26 upstream.

The BUG_ON() in pack_sg_list() would get triggered only one time after we've
corrupted some memory by sg_set_buf() into an invalid sg buffer.

I'm still working on figuring out why I manage to trigger that bug...

Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoedac: fix the error about memory type detection on SandyBridge
Chen Gong [Mon, 14 May 2012 08:51:26 +0000]
edac: fix the error about memory type detection on SandyBridge

commit 2cbb587d3bc41a305168e91b4f3c5b6944a12566 upstream.

On SandyBridge, DDRIOA(Dev: 17 Func: 0 Offset: 328) is used
to detect whether DIMM is RDIMM/LRDIMM, not TA(Dev: 15 Func: 0).

Signed-off-by: Chen Gong <gong.chen@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoedac: avoid mce decoding crash after edac driver unloaded
Chen Gong [Tue, 8 May 2012 23:40:12 +0000]
edac: avoid mce decoding crash after edac driver unloaded

commit e35fca4791fcdd43dc1fd769797df40c562ab491 upstream.

Some edac drivers register themselves as mce decoders via
notifier_chain. But in current notifier_chain implementation logic,
it doesn't accept same notifier registered twice. If so, it will be
wrong when adding/removing the element from the list. For example,
on one SandyBridge platform, remove module sb_edac and then trigger
one error, it will hit oops because it has no mce decoder registered
but related notifier_chain still points to an invalid callback
function. Here is an example:

Call Trace:
 [<ffffffff8150ef6a>] atomic_notifier_call_chain+0x1a/0x20
 [<ffffffff8102b936>] mce_log+0x46/0x180
 [<ffffffff8102eaea>] apei_mce_report_mem_error+0x4a/0x60
 [<ffffffff812e19d2>] ghes_do_proc+0x192/0x210
 [<ffffffff812e2066>] ghes_proc+0x46/0x70
 [<ffffffff812e20d8>] ghes_notify_sci+0x48/0x80
 [<ffffffff8150ef05>] notifier_call_chain+0x55/0x80
 [<ffffffff81076f1a>] __blocking_notifier_call_chain+0x5a/0x80
 [<ffffffff812aea11>] ? acpi_os_wait_events_complete+0x23/0x23
 [<ffffffff81076f56>] blocking_notifier_call_chain+0x16/0x20
 [<ffffffff812ddc4d>] acpi_hed_notify+0x19/0x1b
 [<ffffffff812b16bd>] acpi_device_notify+0x19/0x1b
 [<ffffffff812beb38>] acpi_ev_notify_dispatch+0x67/0x7f
 [<ffffffff812aea3a>] acpi_os_execute_deferred+0x29/0x36
 [<ffffffff81069dc2>] process_one_work+0x132/0x450
 [<ffffffff8106bbcb>] worker_thread+0x17b/0x3c0
 [<ffffffff8106ba50>] ? manage_workers+0x120/0x120
 [<ffffffff81070aee>] kthread+0x9e/0xb0
 [<ffffffff81514724>] kernel_thread_helper+0x4/0x10
 [<ffffffff81070a50>] ? kthread_freezable_should_stop+0x70/0x70
 [<ffffffff81514720>] ? gs_change+0x13/0x13
Code: f3 49 89 d4 45 85 ed 4d 89 c6 48 8b 0f 74 48 48 85 c9 75 17 eb 41
0f 1f 80 00 00 00 00 41 83 ed 01 4c 89 f9 74 22 4d 85 ff 74 1d <4c> 8b
79 08 4c 89 e2 48 89 de 48 89 cf ff 11 4d 85 f6 74 04 41
RIP  [<ffffffff8150eef6>] notifier_call_chain+0x46/0x80
 RSP <ffff88042868fb20>
CR2: ffffffffa01af838
---[ end trace 0100930068e73e6f ]---
BUG: unable to handle kernel paging request at fffffffffffffff8
IP: [<ffffffff810705b0>] kthread_data+0x10/0x20
PGD 1a0d067 PUD 1a0e067 PMD 0
Oops: 0000 [#2] SMP

Only i7core_edac and sb_edac have such issues because they have more
than one memory controller which means they have to register mce
decoder many times.

Signed-off-by: Chen Gong <gong.chen@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agohwrng: atmel-rng - fix data valid check
Peter Korsgaard [Tue, 12 Jun 2012 00:27:04 +0000]
hwrng: atmel-rng - fix data valid check

commit c475c06f4bb689d6ad87d7512e036d6dface3160 upstream.

Brown paper bag: Data valid is LSB of the ISR (status register), and NOT
of ODATA (current random data word)!

With this, rngtest is a lot happier. Before:

rngtest 3
Copyright (c) 2004 by Henrique de Moraes Holschuh
This is free software; see the source for copying conditions.  There is NO warr.

rngtest: starting FIPS tests...
rngtest: bits received from input: 20000032
rngtest: FIPS 140-2 successes: 3
rngtest: FIPS 140-2 failures: 997
rngtest: FIPS 140-2(2001-10-10) Monobit: 604
rngtest: FIPS 140-2(2001-10-10) Poker: 996
rngtest: FIPS 140-2(2001-10-10) Runs: 36
rngtest: FIPS 140-2(2001-10-10) Long run: 0
rngtest: FIPS 140-2(2001-10-10) Continuous run: 117
rngtest: input channel speed: (min=622.371; avg=23682.481; max=28224.350)Kibitss
rngtest: FIPS tests speed: (min=12.361; avg=12.718; max=12.861)Mibits/s
rngtest: Program run time: 2331696 microsecondsx

After:
rngtest 3
Copyright (c) 2004 by Henrique de Moraes Holschuh
This is free software; see the source for copying conditions.  There is NO warr.

rngtest: starting FIPS tests...
rngtest: bits received from input: 20000032
rngtest: FIPS 140-2 successes: 999
rngtest: FIPS 140-2 failures: 1
rngtest: FIPS 140-2(2001-10-10) Monobit: 0
rngtest: FIPS 140-2(2001-10-10) Poker: 0
rngtest: FIPS 140-2(2001-10-10) Runs: 1
rngtest: FIPS 140-2(2001-10-10) Long run: 0
rngtest: FIPS 140-2(2001-10-10) Continuous run: 0
rngtest: input channel speed: (min=777.363; avg=43588.270; max=47870.711)Kibitss
rngtest: FIPS tests speed: (min=11.943; avg=12.716; max=12.844)Mibits/s
rngtest: Program run time: 1955282 microseconds

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Reported-by: George Pontis <GPontis@z9.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoASoC: wm8904: Fix GPIO and MICBIAS initialisation for regmap conversion
Mark Brown [Sat, 9 Jun 2012 03:38:12 +0000]
ASoC: wm8904: Fix GPIO and MICBIAS initialisation for regmap conversion

commit 433897f7408b556f7dfbb98c94deea02e634d2a7 upstream.

We no longer have a flat ASoC cache so can't peer directly into the array
any more but should instead use the register I/O functions to update the
cache.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: fix gathering of interface associations
Daniel Mack [Tue, 12 Jun 2012 18:23:52 +0000]
USB: fix gathering of interface associations

commit b3a3dd074f7053ef824ad077e5331b52220ceba1 upstream.

TEAC's UD-H01 (and probably other devices) have a gap in the interface
number allocation of their descriptors:

  Configuration Descriptor:
    bLength                 9
    bDescriptorType         2
    wTotalLength          220
    bNumInterfaces          3
    [...]
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        0
      bAlternateSetting       0
      [...]
    Interface Association:
      bLength                 8
      bDescriptorType        11
      bFirstInterface         2
      bInterfaceCount         2
      bFunctionClass          1 Audio
      bFunctionSubClass       0
      bFunctionProtocol      32
      iFunction               4
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        2
      bAlternateSetting       0
      [...]

Once a configuration is selected, usb_set_configuration() walks the
known interfaces of a given configuration and calls find_iad() on
each of them to set the interface association pointer the interface
is included in.

The problem here is that the loop variable is taken for the interface
number in the comparison logic that gathers the association. Which is
fine as long as the descriptors are sane.

In the case above, however, the logic gets out of sync and the
interface association fields of all interfaces beyond the interface
number gap are wrong.

Fix this by passing the interface's bInterfaceNumber to find_iad()
instead.

Signed-off-by: Daniel Mack <zonque@gmail.com>
Reported-by: bEN <ml_all@circa.be>
Reported-by: Ivan Perrone <ivanperrone@hotmail.com>
Tested-by: ivan perrone <ivanperrone@hotmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: serial: Enforce USB driver and USB serial driver match
Bjørn Mork [Wed, 30 May 2012 08:00:14 +0000]
USB: serial: Enforce USB driver and USB serial driver match

commit 954c3f8a5f1b7716be9eee978b3bc85bae92d7c8 upstream.

We need to make sure that the USB serial driver we find
matches the USB driver whose probe we are currently
executing. Otherwise we will end up with USB serial
devices bound to the correct serial driver but wrong
USB driver.

An example of such cross-probing, where the usbserial_generic
USB driver has found the sierra serial driver:

May 29 18:26:15 nemi kernel: [ 4442.559246] usbserial_generic 4-4:1.0: Sierra USB modem converter detected
May 29 18:26:20 nemi kernel: [ 4447.556747] usbserial_generic 4-4:1.2: Sierra USB modem converter detected
May 29 18:26:25 nemi kernel: [ 4452.557288] usbserial_generic 4-4:1.3: Sierra USB modem converter detected

sysfs view of the same problem:

bjorn@nemi:~$ ls -l /sys/bus/usb/drivers/sierra/
total 0
--w------- 1 root root 4096 May 29 18:23 bind
lrwxrwxrwx 1 root root    0 May 29 18:23 module -> ../../../../module/usbserial
--w------- 1 root root 4096 May 29 18:23 uevent
--w------- 1 root root 4096 May 29 18:23 unbind
bjorn@nemi:~$ ls -l /sys/bus/usb-serial/drivers/sierra/
total 0
--w------- 1 root root 4096 May 29 18:23 bind
lrwxrwxrwx 1 root root    0 May 29 18:23 module -> ../../../../module/sierra
-rw-r--r-- 1 root root 4096 May 29 18:23 new_id
lrwxrwxrwx 1 root root    0 May 29 18:32 ttyUSB0 -> ../../../../devices/pci0000:00/0000:00:1d.7/usb4/4-4/4-4:1.0/ttyUSB0
lrwxrwxrwx 1 root root    0 May 29 18:32 ttyUSB1 -> ../../../../devices/pci0000:00/0000:00:1d.7/usb4/4-4/4-4:1.2/ttyUSB1
lrwxrwxrwx 1 root root    0 May 29 18:32 ttyUSB2 -> ../../../../devices/pci0000:00/0000:00:1d.7/usb4/4-4/4-4:1.3/ttyUSB2
--w------- 1 root root 4096 May 29 18:23 uevent
--w------- 1 root root 4096 May 29 18:23 unbind

bjorn@nemi:~$ ls -l /sys/bus/usb/drivers/usbserial_generic/
total 0
lrwxrwxrwx 1 root root    0 May 29 18:33 4-4:1.0 -> ../../../../devices/pci0000:00/0000:00:1d.7/usb4/4-4/4-4:1.0
lrwxrwxrwx 1 root root    0 May 29 18:33 4-4:1.2 -> ../../../../devices/pci0000:00/0000:00:1d.7/usb4/4-4/4-4:1.2
lrwxrwxrwx 1 root root    0 May 29 18:33 4-4:1.3 -> ../../../../devices/pci0000:00/0000:00:1d.7/usb4/4-4/4-4:1.3
--w------- 1 root root 4096 May 29 18:33 bind
lrwxrwxrwx 1 root root    0 May 29 18:33 module -> ../../../../module/usbserial
--w------- 1 root root 4096 May 29 18:22 uevent
--w------- 1 root root 4096 May 29 18:33 unbind
bjorn@nemi:~$ ls -l /sys/bus/usb-serial/drivers/generic/
total 0
--w------- 1 root root 4096 May 29 18:33 bind
lrwxrwxrwx 1 root root    0 May 29 18:33 module -> ../../../../module/usbserial
-rw-r--r-- 1 root root 4096 May 29 18:33 new_id
--w------- 1 root root 4096 May 29 18:22 uevent
--w------- 1 root root 4096 May 29 18:33 unbind

So we end up with a mismatch between the USB driver and the
USB serial driver.  The reason for the above is simple: The
USB driver probe will succeed if *any* registered serial
driver matches, and will use that serial driver for all
serial driver functions.

This makes ref counting go wrong. We count the USB driver
as used, but not the USB serial driver.  This may result
in Oops'es as demonstrated by Johan Hovold <jhovold@gmail.com>:

[11811.646396] drivers/usb/serial/usb-serial.c: get_free_serial 1
[11811.646443] drivers/usb/serial/usb-serial.c: get_free_serial - minor base = 0
[11811.646460] drivers/usb/serial/usb-serial.c: usb_serial_probe - registering ttyUSB0
[11811.646766] usb 6-1: pl2303 converter now attached to ttyUSB0
[11812.264197] USB Serial deregistering driver FTDI USB Serial Device
[11812.264865] usbcore: deregistering interface driver ftdi_sio
[11812.282180] USB Serial deregistering driver pl2303
[11812.283141] pl2303 ttyUSB0: pl2303 converter now disconnected from ttyUSB0
[11812.283272] usbcore: deregistering interface driver pl2303
[11812.301056] USB Serial deregistering driver generic
[11812.301186] usbcore: deregistering interface driver usbserial_generic
[11812.301259] drivers/usb/serial/usb-serial.c: usb_serial_disconnect
[11812.301823] BUG: unable to handle kernel paging request at f8e7438c
[11812.301845] IP: [<f8e38445>] usb_serial_disconnect+0xb5/0x100 [usbserial]
[11812.301871] *pde = 357ef067 *pte = 00000000
[11812.301957] Oops: 0000 [#1] PREEMPT SMP
[11812.301983] Modules linked in: usbserial(-) [last unloaded: pl2303]
[11812.302008]
[11812.302019] Pid: 1323, comm: modprobe Tainted: G        W    3.4.0-rc7+ #101 Dell Inc. Vostro 1520/0T816J
[11812.302115] EIP: 0060:[<f8e38445>] EFLAGS: 00010246 CPU: 1
[11812.302130] EIP is at usb_serial_disconnect+0xb5/0x100 [usbserial]
[11812.302141] EAX: f508a180 EBX: f508a180 ECX: 00000000 EDX: f8e74300
[11812.302151] ESI: f5050800 EDI: 00000001 EBP: f5141e78 ESP: f5141e58
[11812.302160]  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
[11812.302170] CR0: 8005003b CR2: f8e7438c CR3: 34848000 CR4: 000007d0
[11812.302180] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
[11812.302189] DR6: ffff0ff0 DR7: 00000400
[11812.302199] Process modprobe (pid: 1323, ti=f5140000 task=f61e2bc0 task.ti=f5140000)
[11812.302209] Stack:
[11812.302216]  f8e3be0f f8e3b29c f8e3ae00 00000000 f513641c f5136400 f513641c f507a540
[11812.302325]  f5141e98 c133d2c1 00000000 00000000 f509c400 f513641c f507a590 f5136450
[11812.302372]  f5141ea8 c12f0344 f513641c f507a590 f5141ebc c12f0c67 00000000 f507a590
[11812.302419] Call Trace:
[11812.302439]  [<c133d2c1>] usb_unbind_interface+0x51/0x190
[11812.302456]  [<c12f0344>] __device_release_driver+0x64/0xb0
[11812.302469]  [<c12f0c67>] driver_detach+0x97/0xa0
[11812.302483]  [<c12f001c>] bus_remove_driver+0x6c/0xe0
[11812.302500]  [<c145938d>] ? __mutex_unlock_slowpath+0xcd/0x140
[11812.302514]  [<c12f0ff9>] driver_unregister+0x49/0x80
[11812.302528]  [<c1457df6>] ? printk+0x1d/0x1f
[11812.302540]  [<c133c50d>] usb_deregister+0x5d/0xb0
[11812.302557]  [<f8e37c55>] ? usb_serial_deregister+0x45/0x50 [usbserial]
[11812.302575]  [<f8e37c8d>] usb_serial_deregister_drivers+0x2d/0x40 [usbserial]
[11812.302593]  [<f8e3a6e2>] usb_serial_generic_deregister+0x12/0x20 [usbserial]
[11812.302611]  [<f8e3acf0>] usb_serial_exit+0x8/0x32 [usbserial]
[11812.302716]  [<c1080b48>] sys_delete_module+0x158/0x260
[11812.302730]  [<c110594e>] ? mntput+0x1e/0x30
[11812.302746]  [<c145c3c3>] ? sysenter_exit+0xf/0x18
[11812.302746]  [<c107777c>] ? trace_hardirqs_on_caller+0xec/0x170
[11812.302746]  [<c145c390>] sysenter_do_call+0x12/0x36
[11812.302746] Code: 24 02 00 00 e8 dd f3 20 c8 f6 86 74 02 00 00 02 74 b4 8d 86 4c 02 00 00 47 e8 78 55 4b c8 0f b6 43 0e 39 f8 7f a9 8b 53 04 89 d8 <ff> 92 8c 00 00 00 89 d8 e8 0e ff ff ff 8b 45 f0 c7 44 24 04 2f
[11812.302746] EIP: [<f8e38445>] usb_serial_disconnect+0xb5/0x100 [usbserial] SS:ESP 0068:f5141e58
[11812.302746] CR2: 00000000f8e7438c

Fix by only evaluating serial drivers pointing back to the
USB driver we are currently probing.  This still allows two
or more drivers to match the same device, running their
serial driver probes to sort out which one to use.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Reviewed-by: Felipe Balbi <balbi@ti.com>
Tested-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: serial: sierra: Add support for Sierra Wireless AirCard 320U modem
Tom Cassidy [Wed, 6 Jun 2012 07:08:48 +0000]
USB: serial: sierra: Add support for Sierra Wireless AirCard 320U modem

commit 19a3dd1575e954e8c004413bee3e12d3962f2525 upstream.

Add support for Sierra Wireless AirCard 320U modem

Signed-off-by: Tomas Cassidy <tomas.cassidy@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agousb: cdc-acm: fix devices not unthrottled on open
Otto Meta [Wed, 6 Jun 2012 16:46:21 +0000]
usb: cdc-acm: fix devices not unthrottled on open

commit 6c4707f3f8c44ec18282e1c014c80e1c257042f9 upstream.

Currently CDC-ACM devices stay throttled when their TTY is closed while
throttled, stalling further communication attempts after the next open.

Unthrottling during open/activate got lost starting with kernel
3.0.0 and this patch reintroduces it.

Signed-off-by: Otto Meta <otto.patches@sister-shadow.de>
Acked-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: cdc-wdm: Add Vodafone/Huawei K5005 support
Bjørn Mork [Sat, 19 May 2012 17:19:48 +0000]
USB: cdc-wdm: Add Vodafone/Huawei K5005 support

commit de102ef41f24a4c251c4a3838796bb27557d4d93 upstream.

Tested-by: Thomas Schäfer <tschaefer@t-online.de>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: add NO_D3_DURING_SLEEP flag and revert 151b61284776be2
Alan Stern [Wed, 13 Jun 2012 15:20:19 +0000]
USB: add NO_D3_DURING_SLEEP flag and revert 151b61284776be2

commit c2fb8a3fa25513de8fedb38509b1f15a5bbee47b upstream.

This patch (as1558) fixes a problem affecting several ASUS computers:
The machine crashes or corrupts memory when going into suspend if the
ehci-hcd driver is bound to any controllers.  Users have been forced
to unbind or unload ehci-hcd before putting their systems to sleep.

After extensive testing, it was determined that the machines don't
like going into suspend when any EHCI controllers are in the PCI D3
power state.  Presumably this is a firmware bug, but there's nothing
we can do about it except to avoid putting the controllers in D3
during system sleep.

The patch adds a new flag to indicate whether the problem is present,
and avoids changing the controller's power state if the flag is set.
Runtime suspend is unaffected; this matters only for system suspend.
However as a side effect, the controller will not respond to remote
wakeup requests while the system is asleep.  Hence USB wakeup is not
functional -- but of course, this is already true in the current state
of affairs.

A similar patch has already been applied as commit
151b61284776be2d6f02d48c23c3625678960b97 (USB: EHCI: fix crash during
suspend on ASUS computers).  The patch supersedes that one and reverts
it.  There are two differences:

The old patch added the flag at the USB level; this patch
adds it at the PCI level.

The old patch applied to all chipsets with the same vendor,
subsystem vendor, and product IDs; this patch makes an
exception for a known-good system (based on DMI information).

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Tested-by: Dâniel Fraga <fragabr@gmail.com>
Tested-by: Andrey Rahmatullin <wrar@wrar.name>
Tested-by: Steven Rostedt <rostedt@goodmis.org>
Reviewed-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: ftdi-sio: Add support for RT Systems USB-RTS01 serial adapter
Evan McNabb [Sat, 26 May 2012 02:46:14 +0000]
USB: ftdi-sio: Add support for RT Systems USB-RTS01 serial adapter

commit e00a54d772210d450e5c1a801534c3c8a448549f upstream.

Add support for RT Systems USB-RTS01 USB to Serial adapter:
http://www.rtsystemsinc.com/Photos/USBRTS01.html

Tested by controlling Icom IC-718 amateur radio transceiver via hamlib.

Signed-off-by: Evan McNabb <evan@mcnabbs.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: serial: cp210x: add Optris MS Pro usb id
Mikko Tuumanen [Fri, 1 Jun 2012 08:28:55 +0000]
USB: serial: cp210x: add Optris MS Pro usb id

commit 5bbfa6f427c1d7244a5ee154ab8fa37265a5e049 upstream.

Signed-off-by: Mikko Tuumanen <mikko.tuumanen@qemsoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: fix PS3 EHCI systems
Ricardo Martins [Tue, 22 May 2012 17:02:03 +0000]
USB: fix PS3 EHCI systems

commit 4f7a67e2dd49fbfba002c453bc24bf00e701cc71 upstream.

After commit aaa0ef289afe9186f81e2340114ea413eef0492a "PS3 EHCI QH
read work-around", Terratec Grabby (em28xx) stopped working with AMD
Geode LX 800 (USB controller AMD CS5536). Since this is a PS3 only
fix, the following patch adds a conditional block around it.

Signed-off-by: Ricardo Martins <rasm@fe.up.pt>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agousb: musb_gadget: fix crash caused by dangling pointer
Grazvydas Ignotas [Fri, 25 May 2012 21:21:33 +0000]
usb: musb_gadget: fix crash caused by dangling pointer

commit 08f75bf14fadaa81fe362d5acda9b77b113dd0a2 upstream.

usb_ep_ops.disable must clear external copy of the endpoint descriptor,
otherwise musb crashes after loading/unloading several gadget modules
in a row:

Unable to handle kernel paging request at virtual address bf013730
pgd = c0004000
[bf013730] *pgd=8f26d811, *pte=00000000, *ppte=00000000
Internal error: Oops: 7 [#1]
Modules linked in: g_cdc [last unloaded: g_file_storage]
CPU: 0    Not tainted  (3.2.17 #647)
PC is at musb_gadget_enable+0x4c/0x24c
LR is at _raw_spin_lock_irqsave+0x4c/0x58
[<c027c030>] (musb_gadget_enable+0x4c/0x24c) from [<bf01b760>] (gether_connect+0x3c/0x19c [g_cdc])
[<bf01b760>] (gether_connect+0x3c/0x19c [g_cdc]) from [<bf01ba1c>] (ecm_set_alt+0x15c/0x180 [g_cdc])
[<bf01ba1c>] (ecm_set_alt+0x15c/0x180 [g_cdc]) from [<bf01ecd4>] (composite_setup+0x85c/0xac4 [g_cdc])
[<bf01ecd4>] (composite_setup+0x85c/0xac4 [g_cdc]) from [<c027b744>] (musb_g_ep0_irq+0x844/0x924)
[<c027b744>] (musb_g_ep0_irq+0x844/0x924) from [<c027a97c>] (musb_interrupt+0x79c/0x864)
[<c027a97c>] (musb_interrupt+0x79c/0x864) from [<c027aaa8>] (generic_interrupt+0x64/0x7c)
[<c027aaa8>] (generic_interrupt+0x64/0x7c) from [<c00797cc>] (handle_irq_event_percpu+0x28/0x178)
...

Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agousb: musb: davinci: Fix build breakage
Jon Povey [Fri, 25 May 2012 01:50:18 +0000]
usb: musb: davinci: Fix build breakage

commit 6594b2d7b1ef8260e6e36ddc96bd37a40e39ba80 upstream.

This appears to have been broken by
commit 5cfb19ac604a68c030b245561f575c2d1bac1d49
(ARM: davinci: streamline sysmod access)

For now, fix by hardcoding USB_PHY_CTRL and DM355_DEEPSLEEP

Tested on DM365 with defconfig changes.

Signed-off-by: Jon Povey <jon.povey@racelogic.co.uk>
Acked-by: Sekhar Nori <nsekhar@ti.com>
CC: Felipe Balbi <balbi@ti.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: mct_u232: Fix incorrect TIOCMSET return
Alan Cox [Tue, 22 May 2012 19:45:13 +0000]
USB: mct_u232: Fix incorrect TIOCMSET return

commit 1aa3c63cf0a79153ee13c8f82e4eb6c40b66a161 upstream.

The low level helper returns 1 on success. The ioctl should however return
0. As this is the only user of the helper return, make the helper return 0 or
an error code.

Resolves-bug: https://bugzilla.kernel.org/show_bug.cgi?id=43009
Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: qcserial: Add Sierra Wireless device IDs
Bjørn Mork [Thu, 24 May 2012 09:19:04 +0000]
USB: qcserial: Add Sierra Wireless device IDs

commit c41444ccfa33a1c20efa319e554cb531576e64a2 upstream.

Some additional IDs found in the BSD/GPL licensed out-of-tree
GobiSerial driver from Sierra Wireless.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: mos7840: Fix compilation of usb serial driver
Tony Zelenoff [Tue, 5 Jun 2012 13:58:04 +0000]
USB: mos7840: Fix compilation of usb serial driver

commit b9c87663eead64c767e72a373ae6f8a94bead459 upstream.

The __devinitconst section can't be referenced
from usb_serial_device structure. Thus removed it as
it done in other mos* device drivers.

Error itself:
WARNING: drivers/usb/serial/mos7840.o(.data+0x8): Section mismatch in reference
from the variable moschip7840_4port_device to the variable
.devinit.rodata:id_table
The variable moschip7840_4port_device references
the variable __devinitconst id_table

[v2] no attach now

Signed-off-by: Tony Zelenoff <antonz@parallels.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agousb-storage: Add 090c:1000 to unusal-devs
Hans de Goede [Wed, 13 Jun 2012 09:44:58 +0000]
usb-storage: Add 090c:1000 to unusal-devs

commit afff07e61a5243e14ee3f0a272a0380cd744a8a3 upstream.

This device gives a bogus answer to get_capacity(16):
[ 8628.278614] scsi 8:0:0:0: Direct-Access     USB 2.0  USB Flash Drive  1100 PQ: 0 ANSI: 4
[ 8628.279452] sd 8:0:0:0: Attached scsi generic sg4 type 0
[ 8628.280338] sd 8:0:0:0: [sdd] 35747322042253313 512-byte logical blocks: (18.3 EB/15.8 EiB)

So set the quirk flag to avoid using get_capacity(16) with it:
[11731.386014] usb-storage 2-1.6:1.0: Quirks match for vid 090c pid 1000: 80000
[11731.386075] scsi9 : usb-storage 2-1.6:1.0
[11731.386172] usbcore: registered new interface driver usb-storage
[11731.386175] USB Mass Storage support registered.
[11732.387394] scsi 9:0:0:0: Direct-Access     USB 2.0  USB Flash Drive  1100 PQ: 0 ANSI: 4
[11732.388462] sd 9:0:0:0: Attached scsi generic sg3 type 0
[11732.389432] sd 9:0:0:0: [sdc] 7975296 512-byte logical blocks: (4.08 GB/3.80 GiB)

Which makes the capacity look a lot more sane :)

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Tested-by: Simon Raffeiner <sturmflut@lieberbiber.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoxHCI: Increase the timeout for controller save/restore state operation
Andiry Xu [Wed, 13 Jun 2012 02:51:57 +0000]
xHCI: Increase the timeout for controller save/restore state operation

commit 622eb783fe6ff4c1baa47db16c3a5db97f9e6e50 upstream.

When system software decides to power down the xHC with the intent of
resuming operation at a later time, it will ask xHC to save the internal
state and restore it when resume to correctly recover from a power event.
Two bits are used to enable this operation: Save State and Restore State.

xHCI spec 4.23.2 says software should "Set the Controller Save/Restore
State flag in the USBCMD register and wait for the Save/Restore State
Status flag in the USBSTS register to transition to '0'". However, it does
not define how long software should wait for the SSS/RSS bit to transition
to 0.

Currently the timeout is set to 1ms. There is bug report
(https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1002697)
indicates that the timeout is too short for ASMedia ASM1042 host controller
to save/restore the state successfully. Increase the timeout to 10ms helps to
resolve the issue.

This patch should be backported to stable kernels as old as 2.6.37, that
contain the commit 5535b1d5f8885695c6ded783c692e3c0d0eda8ca "USB: xHCI:
PCI power management implementation"

Signed-off-by: Andiry Xu <andiry.xu@gmail.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: Ming Lei <ming.lei@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoxhci: Don't free endpoints in xhci_mem_cleanup()
Takashi Iwai [Fri, 1 Jun 2012 08:06:24 +0000]
xhci: Don't free endpoints in xhci_mem_cleanup()

commit 32f1d2c536d0c26c5814cb0e6a0606c42d02fac1 upstream.

This patch fixes a few issues introduced in the recent fix
[f8a9e72d: USB: fix resource leak in xhci power loss path]

- The endpoints listed in bw table are just links and each entry is an
 array member of dev->eps[].  But the commit above adds a kfree() call
 to these instances, and thus it results in memory corruption.

- It clears only the first entry of rh_bw[], but there can be multiple
  ports.

- It'd be safer to clear the list_head of ep as well, not only
  removing from the list, as it's checked in
  xhci_discover_or_reset_device().

This patch should be backported to kernels as old as 3.2, that contain
the commit 839c817ce67178ca3c7c7ad534c571bba1e69ebe "xhci: Store
information about roothubs and TTs."

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Reviewed-by: Oliver Neukum <oneukum@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoxhci: Fix invalid loop check in xhci_free_tt_info()
Takashi Iwai [Fri, 1 Jun 2012 08:06:23 +0000]
xhci: Fix invalid loop check in xhci_free_tt_info()

commit 46ed8f00d8982e49f8fe2c1a9cea192f640cb3ba upstream.

xhci_free_tt_info() may access the invalid memory when it removes the
last entry but the list is not empty.  Then tt_next reaches to the
list head but it still tries to check the tt_info of that entry.

This patch fixes the bug and cleans up the messy code by rewriting
with a simple list_for_each_entry_safe().

This patch should be backported to kernels as old as 3.2, that contain
the commit 839c817ce67178ca3c7c7ad534c571bba1e69ebe "xhci: Store
information about roothubs and TTs."

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Reviewed-by: Oliver Neukum <oneukum@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoMake hard_irq_disable() actually hard-disable interrupts
Paul Mackerras [Fri, 15 Jun 2012 04:51:39 +0000]
Make hard_irq_disable() actually hard-disable interrupts

commit f948501b36c6b3d9352ce212a197098a7e958971 upstream.

At present, hard_irq_disable() does nothing on powerpc because of
this code in include/linux/interrupt.h:

    #ifndef hard_irq_disable
    #define hard_irq_disable()      do { } while(0)
    #endif

So we need to make our hard_irq_disable be a macro.  It was previously
a macro until commit 7230c56441 ("powerpc: Rework lazy-interrupt
handling") changed it to a static inline function.

Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
--
 arch/powerpc/include/asm/hw_irq.h |    3 +++
 1 file changed, 3 insertions(+)

8 years agohfsplus: fix bless ioctl when used with hardlinks
Matthew Garrett [Sun, 17 Jun 2012 21:05:25 +0000]
hfsplus: fix bless ioctl when used with hardlinks

commit 7dea9665fee828fb56db3bae5b9685d9fa006d33 upstream.

HFS+ doesn't really implement hard links - instead, hardlinks are indicated
by a magic file type which refers to an indirect node in a hidden
directory. The spec indicates that stat() should return the inode number
of the indirect node, but it turns out that this doesn't satisfy the
firmware when it's looking for a bootloader - it wants the catalog ID of
the hardlink file instead. Fix up this case.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agohfsplus: fix overflow in sector calculations in hfsplus_submit_bio
Janne Kalliomäki [Sun, 17 Jun 2012 21:05:24 +0000]
hfsplus: fix overflow in sector calculations in hfsplus_submit_bio

commit a6dc8c04218eb752ff79cdc24a995cf51866caed upstream.

The variable io_size was unsigned int, which caused the wrong sector number
to be calculated after aligning it. This then caused mount to fail with big
volumes, as backup volume header information was searched from a
wrong sector.

Signed-off-by: Janne Kalliomäki <janne@tuxera.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agokdump: Execute kmsg_dump(KMSG_DUMP_PANIC) after smp_send_stop()
Seiji Aguchi [Tue, 15 May 2012 21:35:09 +0000]
kdump: Execute kmsg_dump(KMSG_DUMP_PANIC) after smp_send_stop()

commit 62be73eafaa045d3233337303fb140f7f8a61135 upstream.

This patch moves kmsg_dump(KMSG_DUMP_PANIC) below smp_send_stop(),
to serialize the crash-logging process via smp_send_stop() and to
thus retrieve a more stable crash image of all CPUs stopped.

Signed-off-by: Seiji Aguchi <seiji.aguchi@hds.com>
Acked-by: Don Zickus <dzickus@redhat.com>
Cc: dle-develop@lists.sourceforge.net <dle-develop@lists.sourceforge.net>
Cc: Satoru Moriya <satoru.moriya@hds.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: a.p.zijlstra@chello.nl <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/5C4C569E8A4B9B42A84A977CF070A35B2E4D7A5CE2@USINDEVS01.corp.hds.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: option: fix port-data abuse
Johan Hovold [Tue, 29 May 2012 15:57:52 +0000]
USB: option: fix port-data abuse

commit 4273f9878b0a8271df055e3c8f2e7f08c6a4a2f4 upstream.

Commit 8b4c6a3ab596961b78465 ("USB: option: Use generic USB wwan code")
moved option port-data allocation to usb_wwan_startup but still cast the
port data to the old struct...

Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: option: fix memory leak
Johan Hovold [Tue, 29 May 2012 16:22:48 +0000]
USB: option: fix memory leak

commit b9c3aab315b51f81649a0d737c4c73783fbd8de0 upstream.

Fix memory leak introduced by commit 383cedc3bb435de7a2 ("USB: serial:
full autosuspend support for the option driver") which allocates
usb-serial data but never frees it.

Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: option: add more YUGA device ids
说不得 [Mon, 28 May 2012 13:31:29 +0000]
USB: option: add more YUGA device ids

commit 0ef0be15fd2564767f114c249fc4af704d8e16f4 upstream.

Signed-off-by: gavin zhu <gavin.zhu@qq.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: option: Updated Huawei K4605 has better id
Andrew Bird [Mon, 28 May 2012 11:43:06 +0000]
USB: option: Updated Huawei K4605 has better id

commit 42ca7da1c2363dbef4ba1b6917c4c02274b6a5e2 upstream.

Later firmwares for this device now have proper subclass and
protocol info so we can identify it nicely without needing to use
the blacklist. I'm not removing the old 0xff matching as there
may be devices in the field that still need that.

Signed-off-by: Andrew Bird <ajb@spheresystems.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoUSB: option: Add Vodafone/Huawei K5005 support
Bjørn Mork [Sat, 19 May 2012 17:20:50 +0000]
USB: option: Add Vodafone/Huawei K5005 support

commit 4cbbb039a9719fb3bba73d255c6a95bc6dc6428b upstream.

Tested-by: Thomas Schäfer <tschaefer@t-online.de>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoswap: fix shmem swapping when more than 8 areas
Hugh Dickins [Sat, 16 Jun 2012 00:55:50 +0000]
swap: fix shmem swapping when more than 8 areas

commit 9b15b817f3d62409290fd56fe3cbb076a931bb0a upstream.

Minchan Kim reports that when a system has many swap areas, and tmpfs
swaps out to the ninth or more, shmem_getpage_gfp()'s attempts to read
back the page cannot locate it, and the read fails with -ENOMEM.

Whoops.  Yes, I blindly followed read_swap_header()'s pte_to_swp_entry(
swp_entry_to_pte()) technique for determining maximum usable swap
offset, without stopping to realize that that actually depends upon the
pte swap encoding shifting swap offset to the higher bits and truncating
it there.  Whereas our radix_tree swap encoding leaves offset in the
lower bits: it's swap "type" (that is, index of swap area) that was
truncated.

Fix it by reducing the SWP_TYPE_SHIFT() in swapops.h, and removing the
broken radix_to_swp_entry(swp_to_radix_entry()) from read_swap_header().

This does not reduce the usable size of a swap area any further, it
leaves it as claimed when making the original commit: no change from 3.0
on x86_64, nor on i386 without PAE; but 3.0's 512GB is reduced to 128GB
per swapfile on i386 with PAE.  It's not a change I would have risked
five years ago, but with x86_64 supported for ten years, I believe it's
appropriate now.

Hmm, and what if some architecture implements its swap pte with offset
encoded below type? That would equally break the maximum usable swap
offset check.  Happily, they all follow the same tradition of encoding
offset above type, but I'll prepare a check on that for next.

Reported-and-Reviewed-and-Tested-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoSCSI: mpt2sas: Fix unsafe using smp_processor_id() in preemptible
nagalakshmi.nandigama@lsi.com [Tue, 17 Apr 2012 05:55:04 +0000]
SCSI: mpt2sas: Fix unsafe using smp_processor_id() in preemptible

commit a2c658505bf5c75516ee0a79287223e86a2474af upstream.

When CONFIG_DEBUG_PREEMPT is enabled, bug is observed in the smp_processor_id().
This is because smp_processor_id() is not called in preempt safe condition.

To fix this issue, use raw_smp_processor_id instead of smp_processor_id.

Signed-off-by: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agorpc_pipefs: allow rpc_purge_list to take a NULL waitq pointer
Jeff Layton [Mon, 11 Jun 2012 14:03:42 +0000]
rpc_pipefs: allow rpc_purge_list to take a NULL waitq pointer

commit 92123e068efa310b09e9943ac1cfd10ff6b6d2e4 upstream.

In the event that we don't have a dentry for a rpc_pipefs pipe, we still
need to allow the queue_timeout job to clean out the queue. There's just
no waitq to wake up in that event.

Reported-by: Hans de Bruin <jmdebruin@xmsnet.nl>
Reported-by: Joerg Platte <jplatte@naasa.net>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agotracing: Have tracing_off() actually turn tracing off
Steven Rostedt [Wed, 6 Jun 2012 23:50:40 +0000]
tracing: Have tracing_off() actually turn tracing off

commit f2bf1f6f5f89d031245067512449fc889b2f4bb2 upstream.

A recent update to have tracing_on/off() only affect the ftrace ring
buffers instead of all ring buffers had a cut and paste error.
The tracing_off() did the exact same thing as tracing_on() and
would not actually turn off tracing. Unfortunately, tracing_off()
is more important to be working than tracing_on() as this is a key
development tool, as it lets the developer turn off tracing as soon
as a problem is discovered. It is also used by panic and oops code.

This bug also breaks the 'echo func:traceoff > set_ftrace_filter'

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agonfsd4: BUG_ON(!is_spin_locked()) no good on UP kernels
J. Bruce Fields [Tue, 12 Jun 2012 12:28:48 +0000]
nfsd4: BUG_ON(!is_spin_locked()) no good on UP kernels

commit bc2df47a408f2d64cf81bcfd0f6e3e14c84cb0ab upstream.

Most frequent symptom was a BUG triggering in expire_client, with the
server locking up shortly thereafter.

Introduced by 508dc6e110c6dbdc0bbe84298ccfe22de7538486 "nfsd41:
free_session/free_client must be called under the client_lock".

Cc: Benny Halevy <bhalevy@tonian.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoNFSv4: Fix unnecessary delegation returns in nfs4_do_open
Trond Myklebust [Fri, 8 Jun 2012 14:58:09 +0000]
NFSv4: Fix unnecessary delegation returns in nfs4_do_open

commit 2d0dbc6ae8a5194aaecb9cfffb9053f38fce8b86 upstream.

While nfs4_do_open() expects the fmode argument to be restricted to
combinations of FMODE_READ and FMODE_WRITE, both nfs4_atomic_open()
and nfs4_proc_create will pass the nfs_open_context->mode,
which contains the full fmode_t.

This patch ensures that nfs4_do_open strips the other fmode_t bits,
fixing a problem in which the nfs4_do_open call would result in an
unnecessary delegation return.

Reported-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoNFSv4.1: Fix a request leak on the back channel
Trond Myklebust [Thu, 31 May 2012 19:26:38 +0000]
NFSv4.1: Fix a request leak on the back channel

commit b3b02ae5865c2dcd506322e0fc6def59a042e72f upstream.

If the call to svc_process_common() fails, then the request
needs to be freed before we can exit bc_svc_process.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoxen/setup: filter APERFMPERF cpuid feature out
Andre Przywara [Tue, 29 May 2012 11:07:31 +0000]
xen/setup: filter APERFMPERF cpuid feature out

commit 5e626254206a709c6e937f3dda69bf26c7344f6f upstream.

Xen PV kernels allow access to the APERF/MPERF registers to read the
effective frequency. Access to the MSRs is however redirected to the
currently scheduled physical CPU, making consecutive read and
compares unreliable. In addition each rdmsr traps into the hypervisor.
So to avoid bogus readouts and expensive traps, disable the kernel
internal feature flag for APERF/MPERF if running under Xen.
This will
a) remove the aperfmperf flag from /proc/cpuinfo
b) not mislead the power scheduler (arch/x86/kernel/cpu/sched.c) to
   use the feature to improve scheduling (by default disabled)
c) not mislead the cpufreq driver to use the MSRs

This does not cover userland programs which access the MSRs via the
device file interface, but this will be addressed separately.

Signed-off-by: Andre Przywara <andre.przywara@amd.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoxen/hvc: Check HVM_PARAM_CONSOLE_[EVTCHN|PFN] for correctness.
Konrad Rzeszutek Wilk [Wed, 23 May 2012 16:56:59 +0000]
xen/hvc: Check HVM_PARAM_CONSOLE_[EVTCHN|PFN] for correctness.

commit 5842f5768599094758931b74190cdf93641a8e35 upstream.

We need to make sure that those parameters are setup to be correct.
As such the value of 0 is deemed invalid and we find that we
bail out. The hypervisor sets by default all of them to be zero
and when the hypercall is done does a simple:

 a.value = d->arch.hvm_domain.params[a.index];

Which means that if the Xen toolstack forgot to setup the proper
HVM_PARAM_CONSOLE_EVTCHN (or the PFN one), we would get the
default value of 0 and use that.

Fixes-Oracle-Bug: 14091238
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoxen/hvc: Fix error cases around HVM_PARAM_CONSOLE_PFN
Konrad Rzeszutek Wilk [Wed, 23 May 2012 16:55:38 +0000]
xen/hvc: Fix error cases around HVM_PARAM_CONSOLE_PFN

commit a32c88b9386ce3df87f28dd46bdc3776cd6edf75 upstream.

We weren't resetting the parameter to be passed in to a
known default. Nor were we checking the return value of
hvm_get_parameter.

Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoxen/hvc: Collapse error logic.
Konrad Rzeszutek Wilk [Wed, 23 May 2012 16:53:11 +0000]
xen/hvc: Collapse error logic.

commit 2e5ad6b9c45d43cc4e7b8ac5ded1c55a7c4a3893 upstream.

All of the error paths are doing the same logic. In which
case we might as well collapse them in one path.

Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agodrm sis: initialize object_idr
Márton Németh [Mon, 11 Jun 2012 17:09:25 +0000]
drm sis: initialize object_idr

commit 648ccc7d35e3416fdc739d2e520e85de3125361b upstream.

The filed object_idr of struct drm_sis_private was introduced with
commit http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commitdiff;h=6de8a748881f1cd9d795454da2b6db616d5ca3d7 .

The idr_init(&dev->object_name_idr) is called instead of
idr_init(&dev_priv->object_idr) by mistake, leaving object_idr
uninitialized. Correct this.

This patch was not tested because of lack of hardware.

Signed-off-by: Márton Németh <nm127@freemail.hu>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agodrm/udl: only bind to the video devices on the hub.
Dave Airlie [Sat, 16 Jun 2012 06:41:28 +0000]
drm/udl: only bind to the video devices on the hub.

commit e5a867a51d9b009f90d5dca6a320608e4e8a37ec upstream.

This is ported from udlfb.

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=832188
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agodrm via: initialize object_idr
Márton Németh [Sun, 10 Jun 2012 21:39:55 +0000]
drm via: initialize object_idr

commit ce020ea53264f1460ae619cfc12f968dbd0b8974 upstream.

The field obejct_idr of struct drm_via_private was introduced with the
commit http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commitdiff;h=77ee8f3825054f23b17e9c8f728f061defd86cdc .
In that patch idr_init(&dev->object_name_idr) was called instead of
idr_init(&dev_priv->object_idr) by mistake, leaving the dev_priv->object_idr
uninitialized. To be more exact, the object_idr buffer is filled with zeros
because of kzalloc(), but the dev_priv->object_idr.lock spinlock can cause
system freeze at lib/idr.c:move_to_free_list() when spin_lock_irqsave()
is called on this spinlock.

The patch was tested on Clevo D4J, model D410J laptop, on the following
hardware, without AGP kernel module loaded:

  # lspci -s 01:00.0 -n
  01:00.0 0300: 1106:3108 (rev 01)
  # lspci -s 01:00.0 -v
  01:00.0 VGA compatible controller: VIA Technologies, Inc. K8M800/K8N800/K8N800A [S3 UniChrome Pro] (rev 01) (prog-if 00 [VGA controller])
          Subsystem: CLEVO/KAPOK Computer Device 4702
          Flags: bus master, 66MHz, medium devsel, latency 64, IRQ 16
          Memory at f0000000 (32-bit, prefetchable) [size=64M]
          Memory at d1000000 (32-bit, non-prefetchable) [size=16M]
          Expansion ROM at <unassigned> [disabled]
          Capabilities: [60] Power Management version 2
          Capabilities: [70] AGP version 3.0

Signed-off-by: Márton Németh <nm127@freemail.hu>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agodrm/radeon: add some additional 6xx/7xx/EG register init
Alex Deucher [Thu, 14 Jun 2012 20:06:36 +0000]
drm/radeon: add some additional 6xx/7xx/EG register init

commit b866d1334ba2d544bc575d75357dea6bdcdc7f46 upstream.

- SMX_SAR_CTL0 needs to be programmed correctly to prevent
problems with memory exports in certain cases.
- VC_ENHANCE needs to be initialized on 6xx/7xx.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoRevert "drm/i915/dp: Use auxch precharge value of 5 everywhere"
Daniel Vetter [Thu, 14 Jun 2012 20:15:00 +0000]
Revert "drm/i915/dp: Use auxch precharge value of 5 everywhere"

commit 6b4e0a93ff6e45714c72bdce193f719ed94810e3 upstream.

This reverts commit 092945e11c5b84f66dd08f0b87fb729715d377bc.

This commit prevents a DP screen from properly training the link.
Oddly enough it works, once the machine has been warm-booted with an
older kernel.

According to DP docs this _should_ have been the right precharge time.
Also, the commit that originally introduces this was just general snb
DP enabling and didn't mention any specific reason for this special
value. Whatever, trust the reporter that this makes things worse and
let's just revert it.

v2: Less spelling fail.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Reported-by: "Wouter M. Koolen" <W.M.Koolen-Wijkstra@cwi.nl>
Buglink: https://lkml.org/lkml/2012/6/14/301
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoARM i.MX imx21ads: Fix overlapping static i/o mappings
Jaccon Bastiaansen [Mon, 30 Apr 2012 09:53:43 +0000]
ARM i.MX imx21ads: Fix overlapping static i/o mappings

commit 350ab15bb2ffe7103bc6bf6c634f3c5b286eaf2a upstream.

The statically defined I/O memory regions for the i.MX21 on chip
peripherals and the on board I/O peripherals of the i.MX21ADS board
overlap. This results in a kernel crash during startup. This is fixed
by reducing the memory range for the on board I/O peripherals to the
actually required range.

Signed-off-by: Jaccon Bastiaansen <jaccon.bastiaansen@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoARM: imx6: exit coherency when shutting down a cpu
Shawn Guo [Tue, 22 May 2012 14:13:46 +0000]
ARM: imx6: exit coherency when shutting down a cpu

commit 602bf40971d7f9a1ec0b7ba2b7e6427849828651 upstream.

There is a system hang issue on imx6q which can easily be seen with
running a cpu hotplug stress testing (hotplug secondary cores from
user space via sysfs interface for thousands iterations).

It turns out that the issue is caused by coherency of the cpu that
is being shut down.  When shutting down a cpu, we need to have the
cpu exit coherency to prevent it from receiving cache, TLB, or BTB
maintenance operations broadcast by other CPUs in the cluster.

Copy cpu_enter_lowpower() and cpu_leave_lowpower() from mach-vexpress
to have coherency properly handled in platform_cpu_die(), thus fix
the issue.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoARM i.MX53: Fix PLL4 base address
Sascha Hauer [Mon, 4 Jun 2012 12:58:07 +0000]
ARM i.MX53: Fix PLL4 base address

commit cdd781ab1906d039c2a93078385645d2d5af8491 upstream.

MX53_DPLL4_BASE accidently returned the base address of PLL3.
Fix this.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoarm: tegra: pcie: Fix PCIE setup
Jay Agarwal [Thu, 21 Jun 2012 05:29:01 +0000]
arm: tegra: pcie: Fix PCIE setup

1. Remove resource allocation in setup
2. Reorganize memory/IO base/size macros.

Bug 1001291

Change-Id: Idee6db11aa761311aa2743daf8287e84a8a1f199
Signed-off-by: Jay Agarwal <jagarwal@nvidia.com>
Reviewed-on: http://git-master/r/110247
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Preetham Chandru <pchandru@nvidia.com>
Tested-by: Preetham Chandru <pchandru@nvidia.com>
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoUSB: tegra: otg: Enable host by default
Krishna Yarlagadda [Wed, 13 Jun 2012 12:50:16 +0000]
USB: tegra: otg: Enable host by default

Host support should be disabled only when there is platform
data specifying proper setting.

Bug 998001

Reviewed-on: http://git-master/r/108541
(cherry picked from commit a2a54795bdf8c01f712b66a436eb3d598c9d78b0)

Change-Id: If4c21236b14b7db947dbeb9ee5b61e78b9557dee
Signed-off-by: Krishna Yarlagadda <kyarlagadda@nvidia.com>
Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Reviewed-on: http://git-master/r/109794
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>
Tested-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agosecurity: tf_driver integrate TL's latest SDK
Karan Jhavar [Tue, 5 Jun 2012 23:23:35 +0000]
security: tf_driver integrate TL's latest SDK

Update TL's SDK to ver 01.12. New SDK adds -
1) LP2 exit time optimization
2) L2 cache optimization - stop only, no flush
3) PL310 - set dynamic clock gate
4) Support for TEEE client api's for drivers

bug 996822

Change-Id: Id46b7dd153ef05cffeed76558fa7a8c50cae5bd7
Signed-off-by: Karan Jhavar <kjhavar@nvidia.com>
Reviewed-on: http://git-master/r/108025
(cherry picked from commit 3f2b434827ef9456b12dab23339de19afa1ff77c)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: I44399a9c79dba6439858d1bcdf8cd8add1fb3a8b
Reviewed-on: http://git-master/r/109535
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Karan Jhavar <kjhavar@nvidia.com>
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoAvoid aliasing mappings in DMA coherent allocator
Manoj Chourasia [Mon, 4 Jun 2012 11:55:43 +0000]
Avoid aliasing mappings in DMA coherent allocator

Avoid multiple mappings with DMA coherent/writecombine allocator by pre-
allocating the mappings, and removing that memory from the system memory
mapping.  (See previous discussions on linux-arm-kernel as to why this
is bad.)

NB1: By default, we preallocate 2MB for DMA coherent, and 2MB for write
combine memory, rather than 1MB for each in case 1MB is not sufficient
for existing platform usage.  Platforms have the option of shrinking
this down to 1MB DMA / 1MB WC (or even 2MB DMA / 0MB WC) if they so wish.
The DMA memory must be a multiple of 1MB, the write combine memory must
also be a multiple of 1MB, and the two together must be a multiple of
2MB.

NB2: On ARMv6/7 where we use 'normal uncacheable' memory for both DMA
and WC, the two pools are combined into one, as was the case with the
previous implementation.

The down side to this change is that the memory is permanently set aside
for DMA purposes, but I believe that to be unavoidable if we are to
avoid the possibility of the cache getting in the way on VIPT CPUs.

This removes the last known offender (at this time) from the kernel.

Given that DMA memory is fully coherent by this patch, cache
invalidation/clean is not required and so, we skip cache related
activities for the memory managed by the DMA layer. The bus
address -> virtual address conversion normally used in the calling
path and the fact that we remove kernel static mapping corresponding
to the DMA buffers leads to exceptions otherwise.

bug 876019
bug 965047
bug 987589

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Manoj Chourasia <mchourasia@nvidia.com>
Reviewed-on: http://git-master/r/106212
(cherry picked from commit 017d8bd2c4bd03afe04721476dd26388a4bfe7f6)
Change-Id: Ia1452729971e7dfee70a1d7f3861f7e13563dcd7
Reviewed-on: http://git-master/r/108515
(cherry picked from commit 81187d5bca3c27b0737ec3f0f3ff1c4b384c8524)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: I72172941923247019815738aa92eb939eadb586d
Reviewed-on: http://git-master/r/109770
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agonet: bluetooth: prevent multiple l2cap_chan_destroy calls
Nagarjuna Kristam [Tue, 19 Jun 2012 07:23:06 +0000]
net: bluetooth: prevent multiple l2cap_chan_destroy calls

During l2cap channel disconnection, l2cap_chan_destroy is called twice,
when socket is released and when l2cap channel is closed. This causes
kernel panic.

add SOCK_DEAD check inside l2cap_sock_kill to prevent multiple
l2cap_chan_destroy function call

bug 1002888
Signed-off-by: Nagarjuna Kristam <nkristam@nvidia.com>

Change-Id: I1997ebe21351c034636f7133577a454a9e7bcedf
Reviewed-on: http://git-master/r/109731
Tested-by: Nagarjuna Kristam <nkristam@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Rakesh Kumar <krakesh@nvidia.com>
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoarm: tegra: p1852: Pinmux changes for TDM2 (I2S4)
Nitin Pai [Wed, 13 Jun 2012 13:17:00 +0000]
arm: tegra: p1852: Pinmux changes for TDM2 (I2S4)

The pinmux changes are wrt to P1852-SKU8 to get
TDM2(I2S4) controller working with DAP3.

Bug 948478

Change-Id: I4a073aae1358ff813a9822139024997ba3d941d5
Signed-off-by: Nitin Pai <npai@nvidia.com>
Reviewed-on: http://git-master/r/108544
(cherry picked from commit 1d30efaec8a1b4fe80f2f73677ca4a93a7408740)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: Ibca8f619410055bc98588fad47d9519103e1baf7
Reviewed-on: http://git-master/r/109549
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Nitin Pai <npai@nvidia.com>
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoARM: tegra: usb_phy: Fixing HSIC port power code
Vinod Atyam [Tue, 12 Jun 2012 16:00:02 +0000]
ARM: tegra: usb_phy: Fixing HSIC port power code

Fixing the port power code for HSIC. During USB driver re-designing,
port power code was incorrectly merged. Fixing the code now.

Bug 889618
Bug 951061

Change-Id: I1b65662ed864343911abaffb58439ca2996543e7
Signed-off-by: Vinod Atyam <vatyam@nvidia.com>
(cherry picked from commit 86d28cd375e4a86b311024f5f8d5fd74c92ba900)
Reviewed-on: http://git-master/r/108534
(cherry picked from commit fe7a3279c89f02e23da62e8e159824a301cc4b59)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: Idd4736d8e7806a9d1b36d1428170dfcebb6e6fec
Reviewed-on: http://git-master/r/109548
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoARM: tegra: usb_phy: PMC changes for HSIC remote wakeup
Vinod Atyam [Fri, 8 Jun 2012 12:18:39 +0000]
ARM: tegra: usb_phy: PMC changes for HSIC remote wakeup

These are the changes for supporting HSIC remote wakeup
and phy-off in auto-suspend.

Bug 889618
Bug 951061

Change-Id: Ifd8144739c5dea49d8019b42b1a608dd13cc29be
Signed-off-by: Vinod Atyam <vatyam@nvidia.com>
(cherry picked from commit ca33cc4c711b9f41c953ed0971b5a304815d00fa)
Reviewed-on: http://git-master/r/108527
(cherry picked from commit 5036a7bcf4e80d62a254f086038f468b4331c64b)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: I09e5ffad240f55dc400444b22626740d94d49e82
Reviewed-on: http://git-master/r/109547
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoarm: tegra: p1852: Remove unwanted audio clocks
Nitin Pai [Fri, 8 Jun 2012 07:40:52 +0000]
arm: tegra: p1852: Remove unwanted audio clocks

Audio Sync clocks should be set at driver level.
Removing this hack as the TDM driver sets the clocks now.

Bug 948478

Signed-off-by: Nitin Pai <npai@nvidia.com>
Reviewed-on: http://git-master/r/107534
(cherry picked from commit 90b40e3b6250226f775b72aa977888b7f358b9a2)
Change-Id: I0ed9fc9ab268b365b73c4170f53e6316a74b742c
Reviewed-on: http://git-master/r/108518
(cherry picked from commit ac5d08254a503086d77ebc52ebd3aabe4e830edd)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: I780eeb5ad92c637524f7a4ddae41ba63f96e5873
Reviewed-on: http://git-master/r/109546
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Nitin Pai <npai@nvidia.com>
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoARM: tegra: p1852: defconfig: Enable PL310_ERRATA_727915
Manoj Chourasia [Fri, 8 Jun 2012 11:17:35 +0000]
ARM: tegra: p1852: defconfig: Enable PL310_ERRATA_727915

bug 820324

Signed-off-by: Manoj Chourasia <mchourasia@nvidia.com>
Reviewed-on: http://git-master/r/107592
(cherry picked from commit 2bbf11be6d0b5ee8c47f370fc8e72ea0e0790ea6)

Change-Id: I85d47cb34ac4dbb9b66a4ba6e3d8771ffe716b60
Reviewed-on: http://git-master/r/108517
(cherry picked from commit c627b0bdca87ecf108a78c4bcf260fcc7f05b56a)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: Ia57e4e495170812256825bb9b16d00235a7bd7e7
Reviewed-on: http://git-master/r/109545
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agomtd: tegra-nor: Fix dma/cpu coherency issues
Manoj Chourasia [Wed, 9 May 2012 11:15:06 +0000]
mtd: tegra-nor: Fix dma/cpu coherency issues

Call dma sync single api's to maintain coherency between
CPU, dma and device in data transfers.

bug 984029
bug 980884

Signed-off-by: Manoj Chourasia <mchourasia@nvidia.com>
Reviewed-on: http://git-master/r/105917
(cherry picked from commit ff5d25f4a5847c664a123ca02ad981df699d50b6)

Change-Id: Id77d8239186b70e731f0c93f6e52ce81fb7e43f4
Reviewed-on: http://git-master/r/108507
(cherry picked from commit d739cbdfebec5331ac88e44d99d04ebfc20cef06)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: Ibe9ec568fdbf34a1039eb61a9e8fba5d85d9829e
Reviewed-on: http://git-master/r/109544
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoarm: tegra: xmm: removing L3 state transition
Vinayak Pane [Thu, 17 May 2012 01:45:02 +0000]
arm: tegra: xmm: removing L3 state transition

Remove L3 state from xmm power driver. New usb phy code
allows pm state transition L0->L2 and L2->L0. When the
system goes in LP0/LP1 suspend then the modem power
state is kept at L2 and bus state in suspend controlled by
PMC. At system resume the hsic bus is given back to USB.

Bug 991709

Change-Id: Ib3b74d7da30a5103eddd747726ac7c4423472c76
Signed-off-by: Vinayak Pane <vpane@nvidia.com>
Reviewed-on: http://git-master/r/108440
(cherry picked from commit 2c34eb5496ce163a7857877220a403bd9e84c811)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: I2e85796595a042777412d933405f6440fa21b2e7
Reviewed-on: http://git-master/r/109543
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Vinayak Pane <vpane@nvidia.com>
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoarm: tegra: usb_phy: disable all interrupt for 2ls
BH Hsieh [Wed, 13 Jun 2012 04:43:20 +0000]
arm: tegra: usb_phy: disable all interrupt for 2ls

Disable all interrupts on current CPU
while doing 2ls to make sure the 3 us
delay being executed precisely

bug 944998

Change-Id: I1810de454067926793a6e6922676cf2381d195e3
Reviewed-on: http://git-master/r/108194
(cherry picked from commit 302880f7edb789f7e031cd2f5d0f80512ed30725)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: Ida60379aebe82e9793c8b65b4d707314c5cda223
Reviewed-on: http://git-master/r/109542
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agousb: tegra: otg: fix unbalanced suspend/resume
Xin Xie [Wed, 13 Jun 2012 01:23:31 +0000]
usb: tegra: otg: fix unbalanced suspend/resume

Becasue of a previous workaround, suspend/resume pair is implemented as
suspend/complete pair in the driver's dev_pm_ops field.

If any driver suspend routine (which is called before otg suspend)
failed, then tegra_otg_resume() will be called even tegra_otg_suspend()
is not called. This will cause disable the OTG clock, and OTG will not
generate any more IRQ.

Ideal fix would be use suspend/resume pair to implement the dev_pm_ops,
but this might break previous workaround. This patch check suspend()
routine is called or not in the resume routine; even we use suspend/
resume pair in the future, it will not cause bad behaviour.

BUG 999759

Change-Id: I88f9c28bb76c5ff9e32917904017f291456cb933
Signed-off-by: Xin Xie <xxie@nvidia.com>
Reviewed-on: http://git-master/r/108444
(cherry picked from commit c751aeea65f126d8e2df7f8a9ba02a631f958dc1)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: Ic196eb09aab8d1d4aab5f95a801c0aa873e331e8
Reviewed-on: http://git-master/r/109541
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Xin Xie <xxie@nvidia.com>
Tested-by: Xin Xie <xxie@nvidia.com>
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agousb: ehci: tegra: fix hsic bus reset issue
Vinayak Pane [Wed, 13 Jun 2012 01:18:19 +0000]
usb: ehci: tegra: fix hsic bus reset issue

Repair wIndex to start the bus reset sequence properly.
HSIC device will fail to accept the address without tegra
bus reset.

Bug 991485

Change-Id: Iacd19aae1943e30d1c9a56c80c4cf9318279a773
Signed-off-by: Vinayak Pane <vpane@nvidia.com>
Reviewed-on: http://git-master/r/108441
(cherry picked from commit 47ea5ee9c39f61ba6ba3c977f059d2dfc0128390)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: Icf022c4fabefe565691dd25853e149fcfdf856ef
Reviewed-on: http://git-master/r/109540
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Vinayak Pane <vpane@nvidia.com>
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoarm: tegra: p1852: revert previous nor freq change
Bob Johnston [Tue, 12 Jun 2012 11:43:21 +0000]
arm: tegra: p1852: revert previous nor freq change

nor frequency was changed inadvertently in a conflict
resolution.  Need to revert it back to proper value.

Bug 948478

Reviewed-on: http://git-master/r/108247
Signed-off-by: Bob Johnston <bjohnston@nvidia.com>
(cherry picked from commit f0ebce2d266c256ef19d1f5b47e281d2d461d8b3)
Change-Id: I2117c84180d53b14d3f272e46b139b06ee753b1d
Reviewed-on: http://git-master/r/108376
(cherry picked from commit 64dc5f503961e7046ef6541d66b704741517dc1b)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: Id2a5bfdfcf622bc38ed4594fa92ff32cbf08c15d
Reviewed-on: http://git-master/r/109539
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoHACK: ARM: tegra: cardhu: 720p panel dsi workaround
Rakesh Iyer [Thu, 7 Jun 2012 21:54:20 +0000]
HACK: ARM: tegra: cardhu: 720p panel dsi workaround

Workaround to increase dsi maximum frequency to 500Mhz for 720p panel.

Bug 978305.

Change-Id: I71d7722513bc42bbf02769c30f687cb608b2adbd
Signed-off-by: Rakesh Iyer <riyer@nvidia.com>
Reviewed-on: http://git-master/r/108374
(cherry picked from commit a6b6d75d5c77451e57940a321d8d0a011f075a74)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: I701be8cb386a5fb2ea9ce694bb4bb58f1b68c46f
Reviewed-on: http://git-master/r/109538
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoARM: tegra: cardhu: enable DSI panel support
Rakesh Iyer [Sun, 27 May 2012 18:55:33 +0000]
ARM: tegra: cardhu: enable DSI panel support

Allow for DSI panel support to be present in release builds.
Removed CONFIG_TEGRA_CARDHU_DSI config variable.

Bug 978305.

Change-Id: I2b34fa5c5331de897d2b753da90e26278a6b7870
Signed-off-by: Rakesh Iyer <riyer@nvidia.com>
Reviewed-on: http://git-master/r/106919
(cherry picked from commit 3168303db4130d4405d7a56c92ecf382c091e699)
Reviewed-on: http://git-master/r/108372
(cherry picked from commit 90009fb5e08722bd8625f06ef64fbc7ddde5a464)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: Ic96eda2deb63ac25bc1ac58ca696ed8a4ab52cf7
Reviewed-on: http://git-master/r/109537
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoarm: tegra: cardhu: fix copyright information
Rakesh Iyer [Tue, 12 Jun 2012 20:43:58 +0000]
arm: tegra: cardhu: fix copyright information

Fix copyright information for board files.

Bug 978305.

Change-Id: I1322892c5752c5855edecfd7ef06e88663e1f5cd
Signed-off-by: Rakesh Iyer <riyer@nvidia.com>
Reviewed-on: http://git-master/r/108371
(cherry picked from commit db95bb83cbdf396b3c4918701af994ab481607b9)

Signed-off-by: Pritesh Raithatha <praithatha@nvidia.com>
Change-Id: I04c57a256ab811f134c217967469f93b976fbfcc
Reviewed-on: http://git-master/r/109536
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

8 years agoARM: tegra: enterprise: enable auto hotplug
Rakesh Bodla [Tue, 12 Jun 2012 14:26:49 +0000]
ARM: tegra: enterprise: enable auto hotplug

Enabling hotplug detection support for baseband
when it is crashed.

Bug 995784

Reviewed on: http://git-master/r/#change,108284
(cherry picked from commit 0cc4e037d1253ce65a404fe6429709d9172d6ea6)

Change-Id: If95f000e77c7f83a2120233b91a0c099e9bce204
Signed-off-by: Rakesh Bodla <rbodla@nvidia.com>
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>
Reviewed-on: http://git-master/r/109449
Reviewed-by: Automatic_Commit_Validation_User

8 years agoLinux 3.4.3
Greg Kroah-Hartman [Sun, 17 Jun 2012 18:21:44 +0000]
Linux 3.4.3

8 years agoata_piix: defer disks to the Hyper-V drivers by default
Andy Whitcroft [Fri, 4 May 2012 21:15:11 +0000]
ata_piix: defer disks to the Hyper-V drivers by default

commit cd006086fa5d91414d8ff9ff2b78fbb593878e3c upstream.

When we are hosted on a Microsoft Hyper-V hypervisor the guest disks
are exposed both via the Hyper-V paravirtualised drivers and via an
emulated SATA disk drive.  In this case we want to use the paravirtualised
drivers if we can as they are much more efficient.  Note that the Hyper-V
paravirtualised drivers only expose the virtual hard disk devices, the
CDROM/DVD devices must still be enumerated.

Mark the host controller ATA_HOST_IGNORE_ATA to prevent enumeration of
disk devices.

BugLink: http://bugs.launchpad.net/bugs/929545
BugLink: http://bugs.launchpad.net/bugs/942316
Signed-off-by: Andy Whitcroft <apw@canonical.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Cc: Victor Miasnikov <vvm@tut.by>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agolibata: add a host flag to ignore detected ATA devices
Andy Whitcroft [Fri, 4 May 2012 21:15:10 +0000]
libata: add a host flag to ignore detected ATA devices

commit db63a4c8115a0bb904496e1cdd3e7488e68b0d06 upstream.

Where devices are visible via more than one host we sometimes wish to
indicate that cirtain devices should be ignored on a specific host.  Add a
host flag indicating that this host wishes to ignore ATA specific devices.

Signed-off-by: Andy Whitcroft <apw@canonical.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Cc: Victor Miasnikov <vvm@tut.by>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agofuse: fix stat call on 32 bit platforms
Pavel Shilovsky [Thu, 10 May 2012 15:49:38 +0000]
fuse: fix stat call on 32 bit platforms

commit 45c72cd73c788dd18c8113d4a404d6b4a01decf1 upstream.

Now we store attr->ino at inode->i_ino, return attr->ino at the
first time and then return inode->i_ino if the attribute timeout
isn't expired. That's wrong on 32 bit platforms because attr->ino
is 64 bit and inode->i_ino is 32 bit in this case.

Fix this by saving 64 bit ino in fuse_inode structure and returning
it every time we call getattr. Also squash attr->ino into inode->i_ino
explicitly.

Signed-off-by: Pavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agodrm/i915: Mark the ringbuffers as being in the GTT domain
Chris Wilson [Mon, 4 Jun 2012 16:05:40 +0000]
drm/i915: Mark the ringbuffers as being in the GTT domain

commit 3eef8918ff440837f6af791942d8dd07e1a268ee upstream.

By correctly describing the rinbuffers as being in the GTT domain, it
appears that we are more careful with the management of the CPU cache
upon resume and so prevent some coherency issue when submitting commands
to the GPU later. A secondary effect is that the debug logs are then
consistent with the actual usage (i.e. they no longer describe the
ringbuffers as being in the CPU write domain when we are accessing them
through an wc iomapping.)

Reported-and-tested-by: Daniel Gnoutcheff <daniel@gnoutcheff.name>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=41092
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agodrm/radeon: fix tiling and command stream checking on evergreen v3
Jerome Glisse [Sat, 9 Jun 2012 14:57:41 +0000]
drm/radeon: fix tiling and command stream checking on evergreen v3

commit d26098759cf6d32148649c165f87a7590bc25b89 upstream.

Fix regresson since the introduction of command stream checking on
evergreen (thread referenced below). Issue is cause by ddx allocating
bo with formula width*height*bpp while programming the GPU command
stream with ALIGN(height, 8). In some case (where page alignment does
not hide the extra size bo should be according to height alignment)
the kernel will reject the command stream.

This patch reprogram the command stream to slice - 1 (slice is
a derivative value from height) which avoid rejecting the command
stream while keeping the value of command stream checking from a
security point of view.

This patch also fix wrong computation of layer size for 2D tiled
surface. Which should fix issue when 2D color tiling is enabled.
This dump the radeon KMS_DRIVER_MINOR so userspace can know if
they are on a fixed kernel or not.

https://lkml.org/lkml/2012/6/3/80
https://bugs.freedesktop.org/show_bug.cgi?id=50892
https://bugs.freedesktop.org/show_bug.cgi?id=50857

!!! STABLE need a custom version of this patch for 3.4 !!!

v2: actually bump the minor version and add comment about stable
v3: do compute the height the ddx was trying to use

[airlied: drop left over debug]

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agosched: Fix the relax_domain_level boot parameter
Dimitri Sivanich [Tue, 5 Jun 2012 18:44:36 +0000]
sched: Fix the relax_domain_level boot parameter

commit a841f8cef4bb124f0f5563314d0beaf2e1249d72 upstream.

It does not get processed because sched_domain_level_max is 0 at the
time that setup_relax_domain_level() is run.

Simply accept the value as it is, as we don't know the value of
sched_domain_level_max until sched domain construction is completed.

Fix sched_relax_domain_level in cpuset.  The build_sched_domain() routine calls
the set_domain_attribute() routine prior to setting the sd->level, however,
the set_domain_attribute() routine relies on the sd->level to decide whether
idle load balancing will be off/on.

Signed-off-by: Dimitri Sivanich <sivanich@sgi.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120605184436.GA15668@sgi.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agoacpi_video: fix leaking PCI references
Alan Cox [Wed, 25 Apr 2012 13:33:33 +0000]
acpi_video: fix leaking PCI references

commit cfb46f433a4da97c31780e08a259fac2cb6bd61f upstream.

Signed-off-by: Alan Cox <alan@linux.intel.com>
Acked-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agogma500: don't register the ACPI video bus
Alan Cox [Wed, 25 Apr 2012 13:34:04 +0000]
gma500: don't register the ACPI video bus

commit 155689defc782b486a7e6776a57ecc4ebb37ed52 upstream.

We are not yet ready for this and it makes a mess on some devices.

Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agocan: c_can: fix race condition in c_can_open()
AnilKumar Ch [Wed, 23 May 2012 12:15:11 +0000]
can: c_can: fix race condition in c_can_open()

commit f461f27a4436dbe691908fe08b867ef888848cc3 upstream.

Fix the issue of C_CAN interrupts getting disabled forever when canconfig
utility is used multiple times. According to NAPI usage we disable all
the hardware interrupts in ISR and re-enable them in poll(). Current
implementation calls napi_enable() after hardware interrupts are enabled.
If we get any interrupts between these two steps then we do not process
those interrupts because napi is not enabled. Mostly these interrupts
come because of STATUS is not 0x7 or ERROR interrupts. If napi_enable()
happens before HW interrupts enabled then c_can_poll() function will be
called eventual re-enabling.

This patch moves the napi_enable() call before interrupts enabled.

Signed-off-by: AnilKumar Ch <anilkumar@ti.com>
Acked-by: Wolfgang Grandegger <wg@grandegger.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

8 years agocan: c_can: fix an interrupt thrash issue with c_can driver
AnilKumar Ch [Wed, 23 May 2012 12:15:10 +0000]
can: c_can: fix an interrupt thrash issue with c_can driver

commit 148c87c89e1a8863d3d965179f3ab1a06490569e upstream.

This patch fixes an interrupt thrash issue with c_can driver.

In c_can_isr() function interrupts are disabled and enabled only in
c_can_poll() function. c_can_isr() & c_can_poll() both read the
irqstatus flag. However, irqstatus is always read as 0 in c_can_poll()
because all C_CAN interrupts are disabled in c_can_isr(). This causes
all interrupts to be re-enabled in c_can_poll() which in turn causes
another interrupt since the event is not really handled. This keeps
happening causing a flood of interrupts.

To fix this, read the irqstatus register in isr and use the same cached
value in the poll function.

Signed-off-by: AnilKumar Ch <anilkumar@ti.com>
Acked-by: Wolfgang Grandegger <wg@grandegger.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>