A number of tx queue wake-up events went missing due to the
outlined scenario below. Start state is a pool of 16 tx URBs,
active tx_urbs count = 15, with the netdev tx queue open.
CPU #1 [softirq] CPU #2 [softirq]
start_xmit() tx_acknowledge()
................ ................
At the end, the correct state expected is a 15 tx_urbs count
value with the tx queue state _open_. Due to the race, we get
the same tx_urbs value but with the tx queue state _stopped_.
The wake-up event is completely lost.
Thus avoid hand-rolled concurrency mechanisms and use a proper
lock for contexts and tx queue protection.
Jacob Pan [Fri, 13 Mar 2015 10:48:56 +0000 (03:48 -0700)]
powercap / RAPL: handle domains with different energy units
The current driver assumes all RAPL domains within a CPU package
have the same energy unit. This is no longer true for HSW server
CPUs since DRAM domain has is own fixed energy unit which can be
different than the package energy unit enumerated by package
power MSR. In fact, the default HSW EP package power unit is 61uJ
whereas DRAM domain unit is 15.3uJ. The result is that DRAM power
consumption is counted 4x more than real power reported by energy
counters, similarly for max_energy_range_uj of DRAM domain.
This patch adds domain specific energy unit per cpu type, it allows
domain energy unit to override package energy unit if non zero.
Please see this document for details.
"Intel Xeon Processor E5-1600 and E5-2600 v3 Product Families, Volume 2 of 2.
Datasheet, September 2014, Reference Number: 330784-001 "
Linus Torvalds [Fri, 13 Mar 2015 21:30:38 +0000 (14:30 -0700)]
Merge tag 'pm+acpi-4.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull power management and ACPI fixes from Rafael Wysocki:
"Just two fixes, one for an ACPI LPSS driver issue introduced during
the 3.17 cycle and one revert of a recent commit that sort of broke
the cpupower tool.
Specifics:
- Fix an ACPI LPSS (Low-Power Subsystem) driver issue causing the
8250_dw driver to confuse an LPSS clock with another one it is
supposed to handle due to the lack of identification allowing it to
tell those clocks apart (Heikki Krogerus).
- Revert a recent commit that was supposed to improve the usability
of the cpupower tool, but clearly did the opposite (Josh Boyer)"
* tag 'pm+acpi-4.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
Revert "cpupower Makefile change to help run the tool without 'make install'"
ACPI / LPSS: provide con_id for the clkdev
Merge branch 'cpuidle/4.0-fixes' of http://git.linaro.org/people/daniel.lezcano/linux into pm-cpuidle
Pull ARM cpuidle fixes for v4.0 from Daniel Lezcano.
* 'cpuidle/4.0-fixes' of http://git.linaro.org/people/daniel.lezcano/linux:
cpuidle: mvebu: Update cpuidle thresholds for Armada XP SOCs
cpuidle: mvebu: Fix the CPU PM notifier usage
Dave could hit this assert consistently running btrfs/078. This is because
when we update the block groups we could truncate the free space, which would
try to delete the csums for that range and dirty the csum root. For this to
happen we have to have already written out the csum root so it's kind of hard to
hit this case. This patch fixes this by changing the logic to only write the
dirty block groups if the dirty_cowonly_roots list is empty. This will get us
the same effect as before since we add the extent root last, and will cover the
case that we dirty some other root again but not the extent root. Thanks,
Josef Bacik [Fri, 13 Mar 2015 19:12:23 +0000 (15:12 -0400)]
Btrfs: account for the correct number of extents for delalloc reservations
Direct IO can easily pass in an buffer that is greater than
BTRFS_MAX_EXTENT_SIZE, so take this into account when reserving extents in the
delalloc reservation code. Thanks,
Josef Bacik [Fri, 13 Mar 2015 19:12:08 +0000 (15:12 -0400)]
Btrfs: fix merge delalloc logic
My patch to properly count outstanding extents wrt MAX_EXTENT_SIZE introduced a
regression when re-dirtying already dirty areas. We have logic in split to make
sure we are taking the largest space into account but didn't have it for merge,
so it was sometimes making us think we were turning a tiny extent into a huge
extent, when in reality we already had a huge extent and needed to use the other
side in our logic. This fixes the regression that was reported by a user on
list. Thanks,
Liu Bo [Fri, 6 Mar 2015 12:23:44 +0000 (20:23 +0800)]
Btrfs: catch transaction abortion after waiting for it
This problem is uncovered by a test case: http://patchwork.ozlabs.org/patch/244297.
Fsync() can report success when it actually doesn't. When we
have several threads running fsync() at the same tiem and in one fsync() we
get a transaction abortion due to some problems(in the test case it's disk
failures), and other fsync()s may return successfully which makes userspace
programs think that data is now safely flushed into disk.
It's because that after fsyncs() fail btrfs_sync_log() due to disk failures,
they get to try btrfs_commit_transaction() where it finds that there is
already a transaction being committed, and they'll just call wait_for_commit()
and return. Note that we actually check "trans->aborted" in btrfs_end_transaction,
but it's likely that the error message is still not yet throwed out and only after
wait_for_commit() we're sure whether the transaction is committed successfully.
This add the necessary check and it now passes the test.
Fabian Frederick [Sat, 14 Feb 2015 12:10:22 +0000 (13:10 +0100)]
btrfs: fix sizeof format specifier in btrfs_check_super_valid()
This patch fixes mips compilation warning:
fs/btrfs/disk-io.c: In function 'btrfs_check_super_valid':
fs/btrfs/disk-io.c:3927:21: warning: format '%lu' expects argument
of type 'long unsigned int', but argument 3 has type 'unsigned int' [-Wformat]
Linus Torvalds [Fri, 13 Mar 2015 20:34:38 +0000 (13:34 -0700)]
Merge tag 'stable/for-linus-4.0-rc3-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip
Pull xen bug fixes from David Vrabel:
- fix a PV regression in 3.19.
- fix a dom0 crash on hosts with large numbers of PIRQs.
- prevent pcifront from disabling memory or I/O port access, which may
trigger host crashes.
* tag 'stable/for-linus-4.0-rc3-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
xen-pciback: limit guest control of command register
xen/events: avoid NULL pointer dereference in dom0 on large machines
xen: Remove trailing semicolon from xenbus_register_frontend() definition
x86/xen: correct bug in p2m list initialization
Linus Torvalds [Fri, 13 Mar 2015 20:30:00 +0000 (13:30 -0700)]
Merge tag 'sound-4.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Pull sound fixes from Takashi Iwai:
"This is a round of HD-audio fixes: there are a long-standing
regression fix and a few more device/codec-specific quirks.
In addition, a couple of FireWire regression fixes, a USB-audio quirk
for Roland UA-22 and a sanity check in API for user-defined control
elements"
* tag 'sound-4.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
ALSA: hda - Don't access stereo amps for mono channel widgets
ALSA: hda - Add workaround for MacBook Air 5,2 built-in mic
ALSA: hda - Set single_adc_amp flag for CS420x codecs
ALSA: snd-usb: add quirks for Roland UA-22
ALSA: control: Add sanity checks for user ctl id name string
ALSA: hda - Fix built-in mic on Compaq Presario CQ60
ALSA: firewire-lib: leave unit reference counting completely
Revert "ALSA: dice: fix wrong offsets for Dice interface"
ALSA: hda - Fix regression of HD-audio controller fallback modes
Linus Torvalds [Fri, 13 Mar 2015 18:10:10 +0000 (11:10 -0700)]
Merge tag 'devicetree-fixes-for-4.0' of git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux
Pull DeviceTree fixes from Rob Herring:
- fix for stdout-path option parsing with added unittest
- fix for stdout-path interaction with earlycon
- several DT unittest fixes
- fix Sparc allmodconfig build error on of_platform_register_reconfig_notifier
- several DT overlay kconfig and build warning fixes
- several DT binding documentation updates
* tag 'devicetree-fixes-for-4.0' of git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux:
of/platform: Fix sparc:allmodconfig build
of: unittest: Add options string testcase variants
of: fix handling of '/' in options for of_find_node_by_path()
of/unittest: Fix the wrong expected value in of_selftest_property_string
of/unittest: remove the duplicate of_changeset_init
dt: submitting-patches: clarify that DT maintainers are to be cced on bindings
of: unittest: fix I2C dependency
of/overlay: Remove unused variable
Documentation: DT: Renamed of-serial.txt to 8250.txt
of: Fix premature bootconsole disable with 'stdout-path'
serial: add device tree binding documentation for ETRAX FS UART
of/overlay: Directly include idr.h
of: Drop superfluous dependance for OF_OVERLAY
of: Add vendor prefix for Arasan
of: Add prompt for OF_OVERLAY config
Linus Torvalds [Fri, 13 Mar 2015 17:55:32 +0000 (10:55 -0700)]
Merge branch 'gadget' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull gadgetfs fixes from Al Viro:
"Assorted fixes around AIO on gadgetfs: leaks, use-after-free, troubles
caused by ->f_op flipping"
* 'gadget' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
gadgetfs: really get rid of switching ->f_op
gadgetfs: get rid of flipping ->f_op in ep_config()
gadget: switch ep_io_operations to ->read_iter/->write_iter
gadgetfs: use-after-free in ->aio_read()
gadget/function/f_fs.c: switch to ->{read,write}_iter()
gadget/function/f_fs.c: use put iov_iter into io_data
gadget/function/f_fs.c: close leaks
move iov_iter.c from mm/ to lib/
new helper: dup_iter()
The idle power consumption is negatively impacted (proportionally less
than the performance gain), and we are still performing better than
the vendor kernel here:
without patch with patch vendor kernel
- power consumption idle (W) ~2.4 ~3.2 ~4.4
- power consumption busy (W) ~8.6 ~8.3 ~8.6
There is still room for improvement regarding the value of these
thresholds, they were chosen to mimic the vendor kernel.
This patch only impacts Armada XP SOCs and was tested on Online Labs
C1 boards. A similar approach can be taken to improve the performances
of the Armada 370 and Armada 38x SOCs.
Thanks a lot to Thomas Petazzoni, Gregory Clement and Willy Tarreau
for the discussions and tips around this topic.
Gregory CLEMENT [Thu, 26 Feb 2015 17:20:48 +0000 (18:20 +0100)]
cpuidle: mvebu: Fix the CPU PM notifier usage
As stated in kernel/cpu_pm.c, "Platform is responsible for ensuring
that cpu_pm_enter is not called twice on the same CPU before
cpu_pm_exit is called.". In the current code in case of failure when
calling mvebu_v7_cpu_suspend, the function cpu_pm_exit() is never
called whereas cpu_pm_enter() was called just before.
This patch moves the cpu_pm_exit() in order to balance the
cpu_pm_enter() calls.
Alexey Kodanev [Fri, 13 Mar 2015 16:13:53 +0000 (19:13 +0300)]
vxlan: fix wrong usage of VXLAN_VID_MASK
commit dfd8645ea1bd9127 wrongly assumes that VXLAN_VDI_MASK includes
eight lower order reserved bits of VNI field that are using for remote
checksum offload.
Right now, when VNI number greater then 0xffff, vxlan_udp_encap_recv()
will always return with 'bad_flag' error, reducing the usable vni range
from 0..16777215 to 0..65535. Also, it doesn't really check whether RCO
bits processed or not.
Fix it by adding new VNI mask which has all 32 bits of VNI field:
24 bits for id and 8 bits for other usage.
Guenter Roeck [Wed, 11 Mar 2015 03:31:04 +0000 (20:31 -0700)]
of/platform: Fix sparc:allmodconfig build
sparc:allmodconfig fails to build with:
drivers/built-in.o: In function `platform_bus_init':
(.init.text+0x3684): undefined reference to `of_platform_register_reconfig_notifier'
of_platform_register_reconfig_notifier is only declared if both OF_ADDRESS
and OF_DYNAMIC are configured. Yet, the include file only declares a dummy
function if OF_DYNAMIC is not configured. The sparc architecture does not
configure OF_ADDRESS, but does configure OF_DYNAMIC, causing above error.
Wincy Van [Wed, 4 Mar 2015 06:31:56 +0000 (14:31 +0800)]
KVM: VMX: Set msr bitmap correctly if vcpu is in guest mode
In commit 3af18d9c5fe9 ("KVM: nVMX: Prepare for using hardware MSR bitmap"),
we are setting MSR_BITMAP in prepare_vmcs02 if we should use hardware. This
is not enough since the field will be modified by following vmx_set_efer.
Fix this by setting vmx_msr_bitmap_nested in vmx_set_msr_bitmap if vcpu is
in guest mode.
Axel Lin [Fri, 6 Mar 2015 07:55:10 +0000 (15:55 +0800)]
phy: core: Fixup return value of phy_exit when !pm_runtime_enabled
When phy_pm_runtime_get_sync() returns -ENOTSUPP, phy_exit() also returns
-ENOTSUPP if !phy->ops->exit. Fix it.
Also move the code to override ret close to the code we got ret.
I think it is less error prone this way.
Oleg Nesterov [Fri, 13 Mar 2015 08:53:10 +0000 (09:53 +0100)]
x86/fpu: Drop_fpu() should not assume that tsk equals current
drop_fpu() does clear_used_math() and usually this is correct
because tsk == current.
However switch_fpu_finish()->restore_fpu_checking() is called before
__switch_to() updates the "current_task" variable. If it fails,
we will wrongly clear the PF_USED_MATH flag of the previous task.
Oleg Nesterov [Fri, 13 Mar 2015 08:53:09 +0000 (09:53 +0100)]
x86/fpu: Avoid math_state_restore() without used_math() in __restore_xstate_sig()
math_state_restore() assumes it is called with irqs disabled,
but this is not true if the caller is __restore_xstate_sig().
This means that if ia32_fxstate == T and __copy_from_user()
fails, __restore_xstate_sig() returns with irqs disabled too.
This triggers:
BUG: sleeping function called from invalid context at kernel/locking/rwsem.c:41
dump_stack
___might_sleep
? _raw_spin_unlock_irqrestore
__might_sleep
down_read
? _raw_spin_unlock_irqrestore
print_vma_addr
signal_fault
sys32_rt_sigreturn
Change __restore_xstate_sig() to call set_used_math()
unconditionally. This avoids enabling and disabling interrupts
in math_state_restore(). If copy_from_user() fails, we can
simply do fpu_finit() by hand.
[ Note: this is only the first step. math_state_restore() should
not check used_math(), it should set this flag. While
init_fpu() should simply die. ]
Thierry Reding [Wed, 25 Feb 2015 15:16:29 +0000 (16:16 +0100)]
phy: Find the right match in devm_phy_destroy()
devm_phy_create() stores the pointer to the new PHY at the address
returned by devres_alloc(). The res parameter passed to devm_phy_match()
is therefore the location where the pointer to the PHY is stored, hence
it needs to be dereferenced before comparing to the match data in order
to find the correct match.
Stephan Mueller [Thu, 12 Mar 2015 08:17:51 +0000 (09:17 +0100)]
crypto: aesni - fix memory usage in GCM decryption
The kernel crypto API logic requires the caller to provide the
length of (ciphertext || authentication tag) as cryptlen for the
AEAD decryption operation. Thus, the cipher implementation must
calculate the size of the plaintext output itself and cannot simply use
cryptlen.
The RFC4106 GCM decryption operation tries to overwrite cryptlen memory
in req->dst. As the destination buffer for decryption only needs to hold
the plaintext memory but cryptlen references the input buffer holding
(ciphertext || authentication tag), the assumption of the destination
buffer length in RFC4106 GCM operation leads to a too large size. This
patch simply uses the already calculated plaintext size.
In addition, this patch fixes the offset calculation of the AAD buffer
pointer: as mentioned before, cryptlen already includes the size of the
tag. Thus, the tag does not need to be added. With the addition, the AAD
will be written beyond the already allocated buffer.
Note, this fixes a kernel crash that can be triggered from user space
via AF_ALG(aead) -- simply use the libkcapi test application
from [1] and update it to use rfc4106-gcm-aes.
Using [1], the changes were tested using CAVS vectors to demonstrate
that the crypto operation still delivers the right results.
dmaengine: at_hdmac: Fix calculation of the residual bytes
This patch fixes the following issues regarding to the calculation of the
residue:
1. The residue is always calculated for the current transfer even if the
cookie is associated to a pending transfer.
2. For scatter/gather DMA the calculation of the residue for the current
transfer doesn't include the bytes of the child descriptors that are already
transferred.
It only calculates the difference between the transfer's total length minus
the number of bytes that are already transferred for the current child
descriptor.
For example: There is a scatter/gather DMA transfer with a total length of
1 MByte. Getting the residue several times while the transfer is running shows
something like that:
3. The driver stores the residue but never resets it when starting a new
transfer.
For example: If there are two subsequent DMA transfers. The first one with
a total length of 1 MByte and the second one with a total length of 1 kByte.
Getting the residue for both transfers shows something like that:
transfer 1: residue = 975584
transfer 2: residue = 1048380
Changes from V1:
* Fixed coding style of the multi-line comments.
* Improved accuracy of the residue calculation when the transfer for the
first descriptor is active.
Changes from V2:
* Member 'tx_width' of 'struct at_desc' restored, because the transfer width
can't be derived from the source width when using "slave_sg".
The transfer width is needed for the calculation of the residue if either
the transfer of the first or the last descriptor is in progress.
In the case of a "memory_to_memory_sg" transfer (part of this patch
series) the transfer width of both descriptors may differ. Thus it is
required to additionally set 'tx_width' of the last descriptor.
* Added functions for multiply used calculations.
changed the locking logic in put_event() by replacing mutex_lock_nested()
with perf_event_ctx_lock_nested(), but didn't fix the subsequent
mutex_unlock() with a correct counterpart, perf_event_ctx_unlock().
Contexts are thus leaked as a result of incremented refcount
in perf_event_ctx_lock_nested().
The current HDA generic parser initializes / modifies the amp values
always in stereo, but this seems causing the problem on ALC3229 codec
that has a few mono channel widgets: namely, these mono widgets react
to actions for both channels equally.
In the driver code, we do care the mono channel and create a control
only for the left channel (as defined in HD-audio spec) for such a
node. When the control is updated, only the left channel value is
changed. However, in the resume, the right channel value is also
restored from the initial value we took as stereo, and this overwrites
the left channel value. This ends up being the silent output as the
right channel has been never touched and remains muted.
This patch covers the places where unconditional stereo amp accesses
are done and converts to the conditional accesses.
QEMU wants to use virtio scsi structures with
a different VIRTIO_SCSI_CDB_SIZE/VIRTIO_SCSI_SENSE_SIZE,
let's add ifdefs to allow overriding them.
Keep the old defines under new names:
VIRTIO_SCSI_CDB_DEFAULT_SIZE/VIRTIO_SCSI_SENSE_DEFAULT_SIZE,
since that's what these values really are:
defaults for cdb/sense size fields.
virtio_mmio currently lacks generation support which
makes multi-byte field access racy.
Fix by getting the value at offset 0xfc for version 2
devices. Nothing we can do for version 1, so return
generation id 0.
virtio spec requires that all drivers set DRIVER_OK
before using devices. While rpmsg isn't yet
included in the virtio 1 spec, previous spec versions
also required this.
virtio rpmsg violates this rule: is calls kick
before setting DRIVER_OK.
The fix isn't trivial since simply calling virtio_device_ready earlier
would mean we might get an interrupt in parallel with adding buffers.
Instead, split kick out to prepare+notify calls. prepare before
virtio_device_ready - when we know we won't get interrupts. notify right
afterwards.
Linus Torvalds [Fri, 13 Mar 2015 01:46:19 +0000 (18:46 -0700)]
Merge branch 'akpm' (patches from Andrew)
Merge misc fixes from Andrew Morton:
"13 fixes"
* emailed patches from Andrew Morton <[email protected]>:
memcg: disable hierarchy support if bound to the legacy cgroup hierarchy
mm: reorder can_do_mlock to fix audit denial
kasan, module: move MODULE_ALIGN macro into <linux/moduleloader.h>
kasan, module, vmalloc: rework shadow allocation for modules
fanotify: fix event filtering with FAN_ONDIR set
mm/nommu.c: export symbol max_mapnr
arch/c6x/include/asm/pgtable.h: define dummy pgprot_writecombine for !MMU
nilfs2: fix deadlock of segment constructor during recovery
mm: cma: fix CMA aligned offset calculation
mm, hugetlb: close race when setting PageTail for gigantic pages
mm, oom: do not fail __GFP_NOFAIL allocation if oom killer is disabled
drivers/rtc/rtc-s3c.c: add .needs_src_clk to s3c6410 RTC data
ocfs2: make append_dio an incompat feature
Vladimir Davydov [Thu, 12 Mar 2015 23:26:19 +0000 (16:26 -0700)]
memcg: disable hierarchy support if bound to the legacy cgroup hierarchy
If the memory cgroup controller is initially mounted in the scope of the
default cgroup hierarchy and then remounted to a legacy hierarchy, it will
still have hierarchy support enabled, which is incorrect. We should
disable hierarchy support if bound to the legacy cgroup hierarchy.
A userspace call to mmap(MAP_LOCKED) may result in the successful locking
of memory while also producing a confusing audit log denial. can_do_mlock
checks capable and rlimit. If either of these return positive
can_do_mlock returns true. The capable check leads to an LSM hook used by
apparmour and selinux which produce the audit denial. Reordering so
rlimit is checked first eliminates the denial on success, only recording a
denial when the lock is unsuccessful as a result of the denial.
Andrey Ryabinin [Thu, 12 Mar 2015 23:26:14 +0000 (16:26 -0700)]
kasan, module: move MODULE_ALIGN macro into <linux/moduleloader.h>
include/linux/moduleloader.h is more suitable place for this macro.
Also change alignment to PAGE_SIZE for CONFIG_KASAN=n as such
alignment already assumed in several places.
Andrey Ryabinin [Thu, 12 Mar 2015 23:26:11 +0000 (16:26 -0700)]
kasan, module, vmalloc: rework shadow allocation for modules
Current approach in handling shadow memory for modules is broken.
Shadow memory could be freed only after memory shadow corresponds it is no
longer used. vfree() called from interrupt context could use memory its
freeing to store 'struct llist_node' in it:
void vfree(const void *addr)
{
...
if (unlikely(in_interrupt())) {
struct vfree_deferred *p = this_cpu_ptr(&vfree_deferred);
if (llist_add((struct llist_node *)addr, &p->list))
schedule_work(&p->wq);
Later this list node used in free_work() which actually frees memory.
Currently module_memfree() called in interrupt context will free shadow
before freeing module's memory which could provoke kernel crash.
So shadow memory should be freed after module's memory. However, such
deallocation order could race with kasan_module_alloc() in module_alloc().
Free shadow right before releasing vm area. At this point vfree()'d
memory is not used anymore and yet not available for other allocations.
New VM_KASAN flag used to indicate that vm area has dynamically allocated
shadow memory so kasan frees shadow only if it was previously allocated.
With FAN_ONDIR set, the user can end up getting events, which it hasn't
marked. This was revealed with fanotify04 testcase failure on
Linux-4.0-rc1, and is a regression from 3.19, revealed with 66ba93c0d7fe6
("fanotify: don't set FAN_ONDIR implicitly on a marks ignored mask").
# /opt/ltp/testcases/bin/fanotify04
[ ... ]
fanotify04 7 TPASS : event generated properly for type 100000
fanotify04 8 TFAIL : fanotify04.c:147: got unexpected event 30
fanotify04 9 TPASS : No event as expected
The testcase sets the adds the following marks : FAN_OPEN | FAN_ONDIR for
a fanotify on a dir. Then does an open(), followed by close() of the
directory and expects to see an event FAN_OPEN(0x20). However, the
fanotify returns (FAN_OPEN|FAN_CLOSE_NOWRITE(0x10)). This happens due to
the flaw in the check for event_mask in fanotify_should_send_event() which
does:
if (event_mask & marks_mask & ~marks_ignored_mask)
return true;
Chen Gang [Thu, 12 Mar 2015 23:26:03 +0000 (16:26 -0700)]
arch/c6x/include/asm/pgtable.h: define dummy pgprot_writecombine for !MMU
When !MMU, asm-generic will not define default pgprot_writecombine, so c6x
needs to define it by itself. The related error:
CC [M] fs/pstore/ram_core.o
fs/pstore/ram_core.c: In function 'persistent_ram_vmap':
fs/pstore/ram_core.c:399:10: error: implicit declaration of function 'pgprot_writecombine' [-Werror=implicit-function-declaration]
prot = pgprot_writecombine(PAGE_KERNEL);
^
fs/pstore/ram_core.c:399:8: error: incompatible types when assigning to type 'pgprot_t {aka struct <anonymous>}' from type 'int'
prot = pgprot_writecombine(PAGE_KERNEL);
^
Ryusuke Konishi [Thu, 12 Mar 2015 23:26:00 +0000 (16:26 -0700)]
nilfs2: fix deadlock of segment constructor during recovery
According to a report from Yuxuan Shui, nilfs2 in kernel 3.19 got stuck
during recovery at mount time. The code path that caused the deadlock was
as follows:
nilfs_fill_super()
load_nilfs()
nilfs_salvage_orphan_logs()
* Do roll-forwarding, attach segment constructor for recovery,
and kick it.
nilfs_segctor_thread()
nilfs_segctor_thread_construct()
* A lock is held with nilfs_transaction_lock()
nilfs_segctor_do_construct()
nilfs_segctor_drop_written_files()
iput()
iput_final()
write_inode_now()
writeback_single_inode()
__writeback_single_inode()
do_writepages()
nilfs_writepage()
nilfs_construct_dsync_segment()
nilfs_transaction_lock() --> deadlock
This can happen if commit 7ef3ff2fea8b ("nilfs2: fix deadlock of segment
constructor over I_SYNC flag") is applied and roll-forward recovery was
performed at mount time. The roll-forward recovery can happen if datasync
write is done and the file system crashes immediately after that. For
instance, we can reproduce the issue with the following steps:
< nilfs2 is mounted on /nilfs (device: /dev/sdb1) >
# dd if=/dev/zero of=/nilfs/test bs=4k count=1 && sync
# dd if=/dev/zero of=/nilfs/test conv=notrunc oflag=dsync bs=4k
count=1 && reboot -nfh
< the system will immediately reboot >
# mount -t nilfs2 /dev/sdb1 /nilfs
The deadlock occurs because iput() can run segment constructor through
writeback_single_inode() if MS_ACTIVE flag is not set on sb->s_flags. The
above commit changed segment constructor so that it calls iput()
asynchronously for inodes with i_nlink == 0, but that change was
imperfect.
This fixes the another deadlock by deferring iput() in segment constructor
even for the case that mount is not finished, that is, for the case that
MS_ACTIVE flag is not set.
Danesh Petigara [Thu, 12 Mar 2015 23:25:57 +0000 (16:25 -0700)]
mm: cma: fix CMA aligned offset calculation
The CMA aligned offset calculation is incorrect for non-zero order_per_bit
values.
For example, if cma->order_per_bit=1, cma->base_pfn= 0x2f800000 and
align_order=12, the function returns a value of 0x17c00 instead of 0x400.
This patch fixes the CMA aligned offset calculation.
The previous calculation was wrong and would return too-large values for
the offset, so that when cma_alloc looks for free pages in the bitmap with
the requested alignment > order_per_bit, it starts too far into the bitmap
and so CMA allocations will fail despite there actually being plenty of
free pages remaining. It will also probably have the wrong alignment.
With this change, we will get the correct offset into the bitmap.
One affected user is powerpc KVM, which has kvm_cma->order_per_bit set to
KVM_CMA_CHUNK_ORDER - PAGE_SHIFT, or 18 - 12 = 6.
Michal Hocko [Thu, 12 Mar 2015 23:25:52 +0000 (16:25 -0700)]
mm, oom: do not fail __GFP_NOFAIL allocation if oom killer is disabled
Tetsuo Handa has pointed out that __GFP_NOFAIL allocations might fail
after OOM killer is disabled if the allocation is performed by a kernel
thread. This behavior was introduced from the very beginning by 7f33d49a2ed5 ("mm, PM/Freezer: Disable OOM killer when tasks are frozen").
This means that the basic contract for the allocation request is broken
and the context requesting such an allocation might blow up unexpectedly.
There are basically two ways forward.
1) move oom_killer_disable after kernel threads are frozen. This has a
risk that the OOM victim wouldn't be able to finish because it would
depend on an already frozen kernel thread. This would be really tricky
to debug.
2) do not fail GFP_NOFAIL allocation no matter what and risk a
potential Freezable kernel threads will loop and fail the suspend.
Incidental allocations after kernel threads are frozen will at least
dump a warning - if we are lucky and the serial console is still active
of course...
This patch implements the later option because it is safer. We would see
warning rather than allocation failures for the kernel threads which would
blow up otherwise and have a higher chances to identify __GFP_NOFAIL users
from deeper pm code.
drivers/rtc/rtc-s3c.c: add .needs_src_clk to s3c6410 RTC data
Commit df9e26d093d3 ("rtc: s3c: add support for RTC of Exynos3250 SoC")
added an "rtc_src" DT property to specify the clock used as a source to
the S3C real-time clock.
Not all SoCs needs this so commit eaf3a659086e ("drivers/rtc/rtc-s3c.c:
fix initialization failure without rtc source clock") changed to check
the struct s3c_rtc_data .needs_src_clk to conditionally grab the clock.
But that commit didn't update the data for each IP version so the RTC
broke on the boards that needs a source clock. This is the case of at
least Exynos5250 and Exynos5440 which uses the s3c6410 RTC IP block.
This commit fixes the S3C rtc on the Exynos5250 Snow and Exynos5420
Peach Pit and Pi Chromebooks.
Mark Fasheh [Thu, 12 Mar 2015 23:25:46 +0000 (16:25 -0700)]
ocfs2: make append_dio an incompat feature
It turns out that making this feature ro_compat isn't quite enough to
prevent accidental corruption on mount from older kernels. Ocfs2 (like
other file systems) will process orphaned inodes even when the user mounts
in 'ro' mode. So for the case of a filesystem not knowing the append_dio
feature, mounting the filesystem could result in orphaned-for-dio files
being deleted, which we clearly don't want.
So instead, turn this into an incompat flag.
Btw, this is kind of my fault - initially I asked that we add a flag to
cover the feature and even suggested that we use an ro flag. It wasn't
until I was looking through our commits for v4.0-rc1 that I realized we
actually want this to be incompat.
Dave Airlie [Thu, 12 Mar 2015 23:21:24 +0000 (09:21 +1000)]
Merge branch 'drm-fixes-4.0' of git://people.freedesktop.org/~agd5f/linux into drm-fixes
Some additional radeon fixes for 4.0
* 'drm-fixes-4.0' of git://people.freedesktop.org/~agd5f/linux:
drm/radeon: drop setting UPLL to sleep mode
drm/radeon: fix wait to actually occur after the signaling callback
Dave Airlie [Thu, 12 Mar 2015 23:15:56 +0000 (09:15 +1000)]
Merge branch 'vmwgfx-fixes-4.0' of git://people.freedesktop.org/~thomash/linux into drm-fixes
A couple of fixes for vmwgfx.
* 'vmwgfx-fixes-4.0' of git://people.freedesktop.org/~thomash/linux:
drm/vmwgfx: Fix an issue with the device losing its irq line on module unload
drm/vmwgfx: Correctly NULLify dma buffer pointer on failure
drm/vmwgfx: Reorder device takedown somewhat
drm/vmwgfx: Fix a couple of lock dependency violations
Dave Airlie [Thu, 12 Mar 2015 23:15:01 +0000 (09:15 +1000)]
Merge tag 'drm-intel-fixes-2015-03-12' of git://anongit.freedesktop.org/drm-intel into drm-fixes
More i915 fixes, three out of four are fixes to old bugs, cc: stable.
* tag 'drm-intel-fixes-2015-03-12' of git://anongit.freedesktop.org/drm-intel:
drm/i915: Prevent TLB error on first execution on SNB
drm/i915: Do both mt and gen6 style forcewake reset on ivb probe
drm/i915: Make WAIT_IOCTL negative timeouts be indefinite again
drm/i915: use in_interrupt() not in_irq() to check context
Mel Gorman [Sat, 7 Mar 2015 15:20:48 +0000 (15:20 +0000)]
mm: thp: Return the correct value for change_huge_pmd
The wrong value is being returned by change_huge_pmd since commit 10c1045f28e8 ("mm: numa: avoid unnecessary TLB flushes when setting
NUMA hinting entries") which allows a fallthrough that tries to adjust
non-existent PTEs. This patch corrects it.
Takashi Iwai [Thu, 12 Mar 2015 19:47:15 +0000 (20:47 +0100)]
ALSA: hda - Add workaround for MacBook Air 5,2 built-in mic
MacBook Air 5,2 has the same problem as MacBook Pro 8,1 where the
built-in mic records only the right channel. Apply the same
workaround as MBP8,1 to spread the mono channel via a Cirrus codec
vendor-specific COEF setup.
Takashi Iwai [Thu, 12 Mar 2015 19:28:04 +0000 (20:28 +0100)]
ALSA: hda - Set single_adc_amp flag for CS420x codecs
CS420x codecs seem to deal only the single amps of ADC nodes even
though the nodes receive multiple inputs. This leads to the
inconsistent amp value after S3/S4 resume, for example.
The fix is just to set codec->single_adc_amp flag. Then the driver
handles these ADC amps as if single connections.
This patch fixes this by doing this in virtnet_free_queues(). And also
don't delete napi in virtnet_freeze() since it will call
virtnet_free_queues() which has already did this.
Linus Torvalds [Thu, 12 Mar 2015 16:45:46 +0000 (09:45 -0700)]
Merge tag 'pci-v4.0-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci
Pull PCI fixes from Bjorn Helgaas:
"Here are a couple updates for v4.0.
One fixes a config accessor problem on APM X-Gene that we introduced
when switching to generic config accessors, and the other fixes an
older read-past-end-of-buffer problem in sysfs.
APM X-Gene host bridge driver
- Add register offset to config space base address (Feng Kan)
Miscellaneous
- Don't read past the end of sysfs "driver_override" buffer (Sasha Levin)"
* tag 'pci-v4.0-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
PCI: xgene: Add register offset to config space base address
PCI: Don't read past the end of sysfs "driver_override" buffer
Daniel J Blueman [Thu, 12 Mar 2015 15:55:13 +0000 (16:55 +0100)]
x86/apic/numachip: Fix sibling map with NumaChip
On NumaChip systems, the physical processor ID assignment wasn't
accounting for the number of nodes in AMD multi-module
processors, giving an incorrect sibling map:
Linus Torvalds [Thu, 12 Mar 2015 15:45:46 +0000 (08:45 -0700)]
mm: fix up numa read-only thread grouping logic
Dave Chinner reported that commit 4d9424669946 ("mm: convert
p[te|md]_mknonnuma and remaining page table manipulations") slowed down
his xfsrepair test enormously. In particular, it was using more system
time due to extra TLB flushing.
The ultimate reason turns out to be how the change to use the regular
page table accessor functions broke the NUMA grouping logic. The old
special mknuma/mknonnuma code accessed the page table present bit and
the magic NUMA bit directly, while the new code just changes the page
protections using PROT_NONE and the regular vma protections.
That sounds equivalent, and from a fault standpoint it really is, but a
subtle side effect is that the *other* protection bits of the page table
entries also change. And the code to decide how to group the NUMA
entries together used the writable bit to decide whether a particular
page was likely to be shared read-only or not.
And with the change to make the NUMA handling use the regular permission
setting functions, that writable bit was basically always cleared for
private mappings due to COW. So even if the page actually ends up being
written to in the end, the NUMA balancing would act as if it was always
shared RO.
This code is a heuristic anyway, so the fix - at least for now - is to
instead check whether the page is dirty rather than writable. The bit
doesn't change with protection changes.
NOTE! This also adds a FIXME comment to revisit this issue,
Not only should we probably re-visit the whole "is this a shared
read-only page" heuristic (we might want to take the vma permissions
into account and base this more on those than the per-page ones, and
also look at whether the particular access that triggers it is a write
or not), but the whole COW issue shows that we should think about the
NUMA fault handling some more.
For example, maybe we should do the early-COW thing that a regular fault
does. Or maybe we should accept that while using the same bits as
PROTNONE was a good thing (and got rid of the specual NUMA bit), we
might still want to just preseve the other protection bits across NUMA
faulting.
Those are bigger questions, left for later. This just fixes up the
heuristic so that it at least approximates working again. More analysis
and work needed.
Ian Wilson [Thu, 12 Mar 2015 09:37:58 +0000 (09:37 +0000)]
netfilter: Zero the tuple in nfnl_cthelper_parse_tuple()
nfnl_cthelper_parse_tuple() is called from nfnl_cthelper_new(),
nfnl_cthelper_get() and nfnl_cthelper_del(). In each case they pass
a pointer to an nf_conntrack_tuple data structure local variable:
struct nf_conntrack_tuple tuple;
...
ret = nfnl_cthelper_parse_tuple(&tuple, tb[NFCTH_TUPLE]);
The problem is that this local variable is not initialized, and
nfnl_cthelper_parse_tuple() only initializes two fields: src.l3num and
dst.protonum. This leaves all other fields with undefined values
based on whatever is on the stack:
Li, Aubrey [Wed, 11 Mar 2015 08:09:00 +0000 (16:09 +0800)]
x86/platform, acpi: Bypass legacy PIC and PIT in ACPI hardware reduced mode
On a platform in ACPI Hardware-reduced mode, the legacy PIC and
PIT may not be initialized even though they may be present in
silicon. Touching these legacy components causes unexpected
results on the system.
On the Bay Trail-T(ASUS-T100) platform, touching these legacy
components blocks platform hardware low idle power state(S0ix)
during system suspend. So we should bypass them in ACPI hardware
reduced mode.
Jakub Kicinski [Wed, 11 Mar 2015 17:35:36 +0000 (18:35 +0100)]
Revert "i2c: core: Dispose OF IRQ mapping at client removal time"
This reverts commit e4df3a0b6228
("i2c: core: Dispose OF IRQ mapping at client removal time")
Calling irq_dispose_mapping() will destroy the mapping and disassociate
the IRQ from the IRQ chip to which it belongs. Keeping it is OK, because
existent mappings are reused properly.
Also, this commit breaks drivers using devm* for IRQ management on
OF-based systems because devm* cleanup happens in device code, after
bus's remove() method returns.
Daniel Mack [Thu, 12 Mar 2015 08:41:32 +0000 (09:41 +0100)]
ALSA: snd-usb: add quirks for Roland UA-22
The device complies to the UAC1 standard but hides that fact with
proprietary descriptors. The autodetect quirk for Roland devices
catches the audio interface but misses the MIDI part, so a specific
quirk is needed.
Takashi Iwai [Wed, 11 Mar 2015 17:12:49 +0000 (18:12 +0100)]
ALSA: control: Add sanity checks for user ctl id name string
There was no check about the id string of user control elements, so we
accepted even a control element with an empty string, which is
obviously bogus. This patch adds more sanity checks of id strings.
If rockchip_usb_phy_power() fails, we need to call clk_disable_unprepare()
before return. This is to ensure we have balanced clk_enable/disable calls.
Also remove unneeded ret checking in rockchip_usb_phy_power_off.
Axel Lin [Tue, 3 Mar 2015 01:08:43 +0000 (09:08 +0800)]
phy: samsung-usb2: Remove NULL terminating entry from phys array
Current code uses num_phys settings to tell the number of entries in phys.
Thus remove the NULL terminating entry from phys array which is not necessary.
Axel Lin [Fri, 27 Feb 2015 04:34:10 +0000 (12:34 +0800)]
phy: exynos-dp-video: Kill exynos_dp_video_phy_pwr_isol function
If IS_ERR(state->regs) the .probe fails.
So IS_ERR(state->regs) test in exynos_dp_video_phy_pwr_isol() is not necessary.
exynos_dp_video_phy_pwr_isol() simply does a regmap_update_bits() call now,
just call regmap_update_bits() instead and return proper return value.
Arnd Bergmann [Wed, 11 Mar 2015 21:46:59 +0000 (22:46 +0100)]
rds: avoid potential stack overflow
The rds_iw_update_cm_id function stores a large 'struct rds_sock' object
on the stack in order to pass a pair of addresses. This happens to just
fit withint the 1024 byte stack size warning limit on x86, but just
exceed that limit on ARM, which gives us this warning:
net/rds/iw_rdma.c:200:1: warning: the frame size of 1056 bytes is larger than 1024 bytes [-Wframe-larger-than=]
As the use of this large variable is basically bogus, we can rearrange
the code to not do that. Instead of passing an rds socket into
rds_iw_get_device, we now just pass the two addresses that we have
available in rds_iw_update_cm_id, and we change rds_iw_get_mr accordingly,
to create two address structures on the stack there.
Willem de Bruijn [Wed, 11 Mar 2015 19:43:55 +0000 (15:43 -0400)]
sock: fix possible NULL sk dereference in __skb_tstamp_tx
Test that sk != NULL before reading sk->sk_tsflags.
Fixes: 49ca0d8bfaf3 ("net-timestamp: no-payload option") Reported-by: One Thousand Gnomes <[email protected]> Signed-off-by: Willem de Bruijn <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Moving the notify outside of the spin locks means it can be delayed a
long time (if the dealloc thread is descheduled or there is an
interrupt or softirq).
Alexey Kodanev [Wed, 11 Mar 2015 11:29:17 +0000 (14:29 +0300)]
net: sysctl_net_core: check SNDBUF and RCVBUF for min length
sysctl has sysctl.net.core.rmem_*/wmem_* parameters which can be
set to incorrect values. Given that 'struct sk_buff' allocates from
rcvbuf, incorrectly set buffer length could result to memory
allocation failures. For example, set them as follows:
Moreover, the possible minimum is 1, so we can get another kernel panic:
...
BUG: unable to handle kernel paging request at ffff88013caee5c0
IP: [<ffffffff815604cf>] __alloc_skb+0x12f/0x1f0
...
Shawn Guo [Wed, 25 Feb 2015 14:53:38 +0000 (22:53 +0800)]
ASoC: kirkwood: fix struct clk pointer comparing
Since commit 035a61c314eb ("clk: Make clk API return per-user struct clk
instances"), clk API users can no longer check if two struct clk
pointers are pointing to the same hardware clock, i.e. struct clk_hw, by
simply comparing two pointers. That's because with the per-user clk
change, a brand new struct clk is created whenever clients try to look
up the clock by calling clk_get() or sister functions like clk_get_sys()
and of_clk_get(). This changes the original behavior where the struct
clk is only created for once when clock driver registers the clock to
CCF in the first place. The net change here is before commit 035a61c314eb the struct clk pointer is unique for given hardware
clock, while after the commit the pointers returned by clk lookup calls
become different for the same hardware clock.
That said, the struct clk pointer comparing in the code doesn't work any
more. Call helper function clk_is_match() instead to fix the problem.
Shawn Guo [Wed, 25 Feb 2015 14:53:37 +0000 (22:53 +0800)]
ASoC: fsl_spdif: fix struct clk pointer comparing
Since commit 035a61c314eb ("clk: Make clk API return per-user struct clk
instances"), clk API users can no longer check if two struct clk
pointers are pointing to the same hardware clock, i.e. struct clk_hw, by
simply comparing two pointers. That's because with the per-user clk
change, a brand new struct clk is created whenever clients try to look
up the clock by calling clk_get() or sister functions like clk_get_sys()
and of_clk_get(). This changes the original behavior where the struct
clk is only created for once when clock driver registers the clock to
CCF in the first place. The net change here is before commit 035a61c314eb the struct clk pointer is unique for given hardware
clock, while after the commit the pointers returned by clk lookup calls
become different for the same hardware clock.
That said, the struct clk pointer comparing in the code doesn't work any
more. Call helper function clk_is_match() instead to fix the problem.
Shawn Guo [Wed, 25 Feb 2015 14:53:32 +0000 (22:53 +0800)]
ARM: imx: fix struct clk pointer comparing
Since commit 035a61c314eb ("clk: Make clk API return per-user struct clk
instances"), clk API users can no longer check if two struct clk
pointers are pointing to the same hardware clock, i.e. struct clk_hw, by
simply comparing two pointers. That's because with the per-user clk
change, a brand new struct clk is created whenever clients try to look
up the clock by calling clk_get() or sister functions like clk_get_sys()
and of_clk_get(). This changes the original behavior where the struct
clk is only created for once when clock driver registers the clock to
CCF in the first place. The net change here is before commit 035a61c314eb the struct clk pointer is unique for given hardware
clock, while after the commit the pointers returned by clk lookup calls
become different for the same hardware clock.
That said, the struct clk pointer comparing in the code doesn't work any
more. Call helper function clk_is_match() instead to fix the problem.
Some drivers compare struct clk pointers as a means of knowing
if the two pointers reference the same clock hardware. This behavior is
dubious (drivers must not dereference struct clk), but did not cause any
regressions until the per-user struct clk patch was merged. Now the test
for matching clk's will always fail with per-user struct clk's.
clk_is_match is introduced to fix the regression and prevent drivers
from comparing the pointers manually.
While the original commit makes it easier to run cpupower from the
local build directory, it also leaves the binary with a rather poor
rpath of './' in it after it is installed on a system via 'make install'.
This is considered bad practice and can cause cpupower to fail in
rpmbuild with the following error:
ERROR 0004: file '/usr/bin/cpupower' contains an insecure rpath './' in [./]
error: Bad exit status from /var/tmp/rpm-tmp.A6u26r (%install)
Bad exit status from /var/tmp/rpm-tmp.A6u26r (%install)
Developers should be able to use LD_LIBRARY_PATH to achieve the same
effect and not introduce rpath into the binary.
Neal Cardwell [Tue, 10 Mar 2015 21:17:04 +0000 (17:17 -0400)]
tcp: restore 1.5x per RTT limit to CUBIC cwnd growth in congestion avoidance
Commit 814d488c6126 ("tcp: fix the timid additive increase on stretch
ACKs") fixed a bug where tcp_cong_avoid_ai() would either credit a
connection with an increase of snd_cwnd_cnt, or increase snd_cwnd, but
not both, resulting in cwnd increasing by 1 packet on at most every
alternate invocation of tcp_cong_avoid_ai().
Although the commit correctly implemented the CUBIC algorithm, which
can increase cwnd by as much as 1 packet per 1 packet ACKed (2x per
RTT), in practice that could be too aggressive: in tests on network
paths with small buffers, YouTube server retransmission rates nearly
doubled.
This commit restores CUBIC to a maximum cwnd growth rate of 1 packet
per 2 packets ACKed (1.5x per RTT). In YouTube tests this restored
retransmit rates to low levels.
Testing: This patch has been tested in datacenter netperf transfers
and live youtube.com and google.com servers.
Neal Cardwell [Tue, 10 Mar 2015 21:17:03 +0000 (17:17 -0400)]
tcp: fix tcp_cong_avoid_ai() credit accumulation bug with decreases in w
The recent change to tcp_cong_avoid_ai() to handle stretch ACKs
introduced a bug where snd_cwnd_cnt could accumulate a very large
value while w was large, and then if w was reduced snd_cwnd could be
incremented by a large delta, leading to a large burst and high packet
loss. This was tickled when CUBIC's bictcp_update() sets "ca->cnt =
100 * cwnd".
This bug crept in while preparing the upstream version of 814d488c6126.
Testing: This patch has been tested in datacenter netperf transfers
and live youtube.com and google.com servers.
Fixes: 814d488c6126 ("tcp: fix the timid additive increase on stretch ACKs") Signed-off-by: Neal Cardwell <[email protected]> Signed-off-by: Yuchung Cheng <[email protected]> Signed-off-by: Eric Dumazet <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Arnd Bergmann [Wed, 11 Mar 2015 19:46:52 +0000 (20:46 +0100)]
Merge tag 'at91-fixes3' of git://git.kernel.org/pub/scm/linux/kernel/git/nferre/linux-at91 into fixes
Pull "Third fixes batch for AT91 on 4.0" from Nicolas Ferre:
- clock fixes for USB
- compatible string changes for handling USB IP differences
(+ needed AHB matrix syscon)
- fix of a compilation error in PM code
* tag 'at91-fixes3' of git://git.kernel.org/pub/scm/linux/kernel/git/nferre/linux-at91:
ARM: at91: pm_slowclock: fix the compilation error
ARM: at91/dt: fix USB high-speed clock to select UTMI
ARM: at91/dt: fix at91 udc compatible strings
ARM: at91/dt: declare matrix node as a syscon device
ARM: at91/dt: at91sam9261: fix clocks and clock-names in udc definition
projects / linux.git / log