parisc,metag: Do not hardcode maximum userspace stack size
This patch affects only architectures where the stack grows upwards
(currently parisc and metag only). On those do not hardcode the maximum
initial stack size to 1GB for 32-bit processes, but make it configurable
via a config option.
The main problem with the hardcoded stack size is, that we have two
memory regions which grow upwards: stack and heap. To keep most of the
memory available for heap in a flexmap memory layout, it makes no sense
to hard allocate up to 1GB of the memory for stack which can't be used
as heap then.
This patch makes the stack size for 32-bit processes configurable and
uses 80MB as default value which has been in use during the last few
years on parisc and which hasn't showed any problems yet.
James Hogan [Tue, 13 May 2014 22:58:24 +0000 (23:58 +0100)]
metag: Reduce maximum stack size to 256MB
Specify the maximum stack size for arches where the stack grows upward
(parisc and metag) in asm/processor.h rather than hard coding in
fs/exec.c so that metag can specify a smaller value of 256MB rather than
1GB.
This fixes a BUG on metag if the RLIMIT_STACK hard limit is increased
beyond a safe value by root. E.g. when starting a process after running
"ulimit -H -s unlimited" it will then attempt to use a stack size of the
maximum 1GB which is far too big for metag's limited user virtual
address space (stack_top is usually 0x3ffff000):
Mikulas Patocka [Thu, 8 May 2014 19:51:37 +0000 (15:51 -0400)]
metag: fix memory barriers
Volatile access doesn't really imply the compiler barrier. Volatile access
is only ordered with respect to other volatile accesses, it isn't ordered
with respect to general memory accesses. Gcc may reorder memory accesses
around volatile access, as we can see in this simple example (if we
compile it with optimization, both increments of *b will be collapsed to
just one):
void fn(volatile int *a, long *b)
{
(*b)++;
*a = 10;
(*b)++;
}
Consequently, we need the compiler barrier after a write to the volatile
variable, to make sure that the compiler doesn't reorder the volatile
write with something else.
Mike Snitzer [Tue, 13 May 2014 17:49:39 +0000 (13:49 -0400)]
dm mpath: fix lock order inconsistency in multipath_ioctl
Commit 3e9f1be1b40 ("dm mpath: remove process_queued_ios()") did not
consistently take the multipath device's spinlock (m->lock) before
calling dm_table_run_md_queue_async() -- which takes the q->queue_lock.
Found with code inspection using hint from reported lockdep warning.
If the pool runs out of data space, dm-thin can be configured to
either error IOs that would trigger provisioning, or hold those IOs
until the pool is resized. Unfortunately, holding IOs until the pool is
resized can result in a cascade of tasks hitting the hung_task_timeout,
which may render the system unavailable.
Add a fixed timeout so IOs can only be held for a maximum of 60 seconds.
If LVM is going to resize a thin-pool that is out of data space it needs
to be prompt about it.
Joe Thornber [Tue, 6 May 2014 15:28:14 +0000 (16:28 +0100)]
dm thin: allow metadata commit if pool is in PM_OUT_OF_DATA_SPACE mode
Commit 3e1a0699 ("dm thin: fix out of data space handling") introduced
a regression in the metadata commit() method by returning an error if
the pool is in PM_OUT_OF_DATA_SPACE mode. This oversight caused a thin
device to return errors even if the default queue_if_no_space ENOSPC
handling mode is used.
Fix commit() to only fail if pool is in PM_READ_ONLY or PM_FAIL mode.
Mikulas Patocka [Thu, 20 Feb 2014 23:01:01 +0000 (18:01 -0500)]
dm crypt: fix cpu hotplug crash by removing per-cpu structure
The DM crypt target used per-cpu structures to hold pointers to a
ablkcipher_request structure. The code assumed that the work item keeps
executing on a single CPU, so it didn't use synchronization when
accessing this structure.
If a CPU is disabled by writing 0 to /sys/devices/system/cpu/cpu*/online,
the work item could be moved to another CPU. This causes dm-crypt
crashes, like the following, because the code starts using an incorrect
ablkcipher_request:
Fix this bug by removing the per-cpu definition. The structure
ablkcipher_request is accessed via a pointer from convert_context.
Consequently, if the work item is rescheduled to a different CPU, the
thread still uses the same ablkcipher_request.
This change may undermine performance improvements intended by commit c0297721 ("dm crypt: scale to multiple cpus") on select hardware. In
practice no performance difference was observed on recent hardware. But
regardless, correctness is more important than performance.
Wolfram Sang [Mon, 5 May 2014 16:36:21 +0000 (18:36 +0200)]
i2c: rcar: bail out on zero length transfers
This hardware does not support zero length transfers. Instead, the
driver does one (random) byte transfers currently with undefined results
for the slaves. We now bail out.
Olof Johansson [Fri, 11 Apr 2014 22:19:41 +0000 (15:19 -0700)]
i2c: s3c2410: resume race fix
Don't unmark the device as suspended until after it's been re-setup.
The main race would be w.r.t. an i2c driver that gets resumed at the same
time (asyncronously), that is allowed to do a transfer since suspended
is set to 0 before reinit, but really should have seen the -EIO return
instead.
Du, Wenkai [Thu, 10 Apr 2014 23:03:19 +0000 (23:03 +0000)]
i2c: designware: Mask all interrupts during i2c controller enable
There have been "i2c_designware 80860F41:00: controller timed out" errors
on a number of Baytrail platforms. The issue is caused by incorrect value in
Interrupt Mask Register (DW_IC_INTR_MASK) when i2c core is being enabled.
This causes call to __i2c_dw_enable() to immediately start the transfer which
leads to timeout. There are 3 failure modes observed:
1. Failure in S0 to S3 resume path
The default value after reset for DW_IC_INTR_MASK is 0x8ff. When we start
the first transaction after resuming from system sleep, TX_EMPTY interrupt
is already unmasked because of the hardware default.
2. Failure in normal operational path
This failure happens rarely and is hard to reproduce. Debug trace showed that
DW_IC_INTR_MASK had value of 0x254 when failure occurred, which meant
TX_EMPTY was unmasked.
3. Failure in S3 to S0 suspend path
This failure also happens rarely and is hard to reproduce. Adding debug trace
that read DW_IC_INTR_MASK made this failure not reproducible. But from ISR
call trace we could conclude TX_EMPTY was unmasked when problem occurred.
The patch masks all interrupts before the controller is enabled to resolve the
faulty DW_IC_INTR_MASK conditions.
Steven J. Hill [Thu, 10 Apr 2014 19:06:17 +0000 (14:06 -0500)]
MIPS: mm: Fix broken microMIPS kernel regression.
Commit f4ae17aa0f2122b52f642985b46210a1f2eceb0a [MIPS: mm: Use scratch for
PGD when !CONFIG_MIPS_PGD_C0_CONTEXT] broke microMIPS kernel builds. This
patch refactors that code similar to what was done for the 'clear_page'
and 'copy_page' functions.
Dan Carpenter [Wed, 14 May 2014 13:32:21 +0000 (16:32 +0300)]
ALSA: sb_mixer: missing return statement
The if condition here was supposed to return on error but the return
statement is missing. The effect is that the ->mixername is set to
"???" instead of "DT019X".
Thomas Petazzoni [Wed, 14 May 2014 11:36:36 +0000 (13:36 +0200)]
of: make of_update_property() usable earlier in the boot process
Commit 75b57ecf9d1d1e17d099ab13b8f48e6e038676be ('of: Make device
nodes kobjects so they show up in sysfs') has turned Device Tree nodes
in kobjects and added a sysfs based representation for Device Tree
nodes. Since the sysfs logic is only available after the execution of
a core_initcall(), the patch took precautions in of_add_property() and
of_remove_property() to not do any sysfs related manipulation early in
the boot process.
However, it forgot to do the same for of_update_property(), which if
used early in the boot process (before core_initcalls have been
called), tries to call sysfs_remove_bin_file(), and crashes:
To fix this problem, we simply skip the sysfs related calls in
of_update_property(), and rely on of_init() to fix up things when it
will be called, exactly as is done in of_add_property() and
of_remove_property().
Signed-off-by: Thomas Petazzoni <[email protected]> Fixes: 75b57ecf9d1d ("of: Make device nodes kobjects so they show up in sysfs") Signed-off-by: Grant Likely <[email protected]>
Takashi Iwai [Wed, 14 May 2014 12:27:12 +0000 (14:27 +0200)]
Merge tag 'asoc-v3.15-rc5-intel' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-linus
ASoC: Intel fixes for v3.15
This is a relatively large batch of fixes for the newly added
Haswell/Baytrail drivers from Intel. It's a bit larger than is good for
this point in the cycle but it's all for a newly added driver so not so
worrying as it might otherwise be. Some of it's integration problems,
some of it's the sort of problem usually turned up in stress tests.
Takashi Iwai [Wed, 14 May 2014 12:23:48 +0000 (14:23 +0200)]
Merge tag 'asoc-v3.15-rc5-core' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-linus
ASoC: Core fixes for v3.15
A few things here:
- Fix the creation of spurious CODEC<->CODEC links which caused DAPM to
have audio paths which shouldn't be present causing spurious powerups
and potential audible issues for users.
- Ensure the suspend->off transition doesn't have spurious transitions
to prepare added to the sequence.
- Fix incorrect skipping of PCM suspension for active audio streams.
- Remove Timur Tabi from the CS4270 maintainers, Cirrus are now doing
this and Timur no longer has the boards that he was using.
MIPS: Add new AUDIT_ARCH token for the N32 ABI on MIPS64
A MIPS64 kernel may support ELF files for all 3 MIPS ABIs
(O32, N32, N64). Furthermore, the AUDIT_ARCH_MIPS{,EL}64 token
does not provide enough information about the ABI for the 64-bit
process. As a result of which, userland needs to use complex
seccomp filters to decide whether a syscall belongs to the o32 or n32
or n64 ABI. Therefore, a new arch token for MIPS64/n32 is added so it
can be used by seccomp to explicitely set syscall filters for this ABI.
x86, mm, hugetlb: Add missing TLB page invalidation for hugetlb_cow()
The invalidation is required in order to maintain proper semantics
under CoW conditions. In scenarios where a process clones several
threads, a thread operating on a core whose DTLB entry for a
particular hugepage has not been invalidated, will be reading from
the hugepage that belongs to the forked child process, even after
hugetlb_cow().
The thread will not see the updated page as long as the stale DTLB
entry remains cached, the thread attempts to write into the page,
the child process exits, or the thread gets migrated to a different
processor.
[media] V4L2: fix VIDIOC_CREATE_BUFS in 64- / 32-bit compatibility mode
If a struct contains 64-bit fields, it is aligned on 64-bit boundaries
within containing structs in 64-bit compilations. This is the case with
struct v4l2_window, which contains pointers and is embedded into struct
v4l2_format, and that one is embedded into struct v4l2_create_buffers.
Unlike some other structs, used as a part of the kernel ABI as ioctl()
arguments, that are packed, these structs aren't packed. This isn't a
problem per se, but the ioctl-compat code for VIDIOC_CREATE_BUFS contains
a bug, that triggers in such 64-bit builds. That code wrongly assumes,
that in struct v4l2_create_buffers, struct v4l2_format immediately follows
the __u32 memory field, which in fact isn't the case. This bug wasn't
visible until now, because until recently hardly any applications used
this ioctl() and mostly embedded 32-bit only drivers implemented it. This
is changing now with addition of this ioctl() to some USB drivers, e.g.
UVC. This patch fixes the bug by copying parts of struct
v4l2_create_buffers separately.
[media] V4L2: ov7670: fix a wrong index, potentially Oopsing the kernel from user-space
Commit 75e2bdad8901a0b599e01a96229be922eef1e488 "ov7670: allow
configuration of image size, clock speed, and I/O method" uses a wrong
index to iterate an array. Apart from being wrong, it also uses an
unchecked value from user-space, which can cause access to unmapped
memory in the kernel, triggered by a normal desktop user with rights to
use V4L2 devices.
Charles Keepax [Tue, 13 May 2014 12:45:15 +0000 (13:45 +0100)]
ASoC: wm8962: Update register CLASS_D_CONTROL_1 to be non-volatile
The register CLASS_D_CONTROL_1 is marked as volatile because it contains
a bit, DAC_MUTE, which is also mirrored in the ADC_DAC_CONTROL_1
register. This causes problems for the "Speaker Switch" control, which
will report an error if the CODEC is suspended because it relies on a
volatile register.
To resolve this issue mark CLASS_D_CONTROL_1 as non-volatile and
manually keep the register cache in sync by updating both bits when
changing the mute status.
Fix this by calling dma_coerce_mask_and_coherent() in sst_byt_init() with
the same dma_dev device what is now used in sst_fw_new() when allocating the
DMA buffer.
Dirk Brandewie [Thu, 8 May 2014 19:57:24 +0000 (12:57 -0700)]
intel_pstate: remove setting P state to MAX on init
Setting the P state of the core to max at init time is a hold over
from early implementation of intel_pstate where intel_pstate disabled
cpufreq and loaded VERY early in the boot sequence. This was to
ensure that intel_pstate did not affect boot time. This in not needed
now that intel_pstate is a cpufreq driver.
Removing this covers the case where a CPU has gone through a manual
CPU offline/online cycle and the P state is set to MAX on init and the
CPU immediately goes idle. Due to HW coordination the P state request
on the idle CPU will drag all cores to MAX P state until the load is
reevaluated when to core goes non-idle.
Tejun Heo [Tue, 13 May 2014 15:28:30 +0000 (11:28 -0400)]
cgroup: fix rcu_read_lock() leak in update_if_frozen()
While updating cgroup_freezer locking, 68fafb77d827 ("cgroup_freezer:
replace freezer->lock with freezer_mutex") introduced a bug in
update_if_frozen() where it returns with rcu_read_lock() held. Fix it
by adding rcu_read_unlock() before returning.
Tejun Heo [Thu, 8 May 2014 01:31:17 +0000 (21:31 -0400)]
cgroup_freezer: replace freezer->lock with freezer_mutex
After 96d365e0b86e ("cgroup: make css_set_lock a rwsem and rename it
to css_set_rwsem"), css task iterators requires sleepable context as
it may block on css_set_rwsem. I missed that cgroup_freezer was
iterating tasks under IRQ-safe spinlock freezer->lock. This leads to
errors like the following on freezer state reads and transitions.
BUG: sleeping function called from invalid context at /work
/os/work/kernel/locking/rwsem.c:20
in_atomic(): 0, irqs_disabled(): 0, pid: 462, name: bash
5 locks held by bash/462:
#0: (sb_writers#7){.+.+.+}, at: [<ffffffff811f0843>] vfs_write+0x1a3/0x1c0
#1: (&of->mutex){+.+.+.}, at: [<ffffffff8126d78b>] kernfs_fop_write+0xbb/0x170
#2: (s_active#70){.+.+.+}, at: [<ffffffff8126d793>] kernfs_fop_write+0xc3/0x170
#3: (freezer_mutex){+.+...}, at: [<ffffffff81135981>] freezer_write+0x61/0x1e0
#4: (rcu_read_lock){......}, at: [<ffffffff81135973>] freezer_write+0x53/0x1e0
Preemption disabled at:[<ffffffff81104404>] console_unlock+0x1e4/0x460
freezer->lock used to be used in hot paths but that time is long gone
and there's no reason for the lock to be IRQ-safe spinlock or even
per-cgroup. In fact, given the fact that a cgroup may contain large
number of tasks, it's not a good idea to iterate over them while
holding IRQ-safe spinlock.
Let's simplify locking by replacing per-cgroup freezer->lock with
global freezer_mutex. This also makes the comments explaining the
intricacies of policy inheritance and the locking around it as the
states are protected by a common mutex.
The conversion is mostly straight-forward. The followings are worth
mentioning.
* freezer_css_online() no longer needs double locking.
* freezer_attach() now performs propagation simply while holding
freezer_mutex. update_if_frozen() race no longer exists and the
comment is removed.
* freezer_fork() now tests whether the task is in root cgroup using
the new task_css_is_root() without doing rcu_read_lock/unlock(). If
not, it grabs freezer_mutex and performs the operation.
* freezer_read() and freezer_change_state() grab freezer_mutex across
the whole operation and pin the css while iterating so that each
descendant processing happens in sleepable context.
Fixes: 96d365e0b86e ("cgroup: make css_set_lock a rwsem and rename it to css_set_rwsem") Signed-off-by: Tejun Heo <[email protected]> Acked-by: Li Zefan <[email protected]>
Tejun Heo [Thu, 8 May 2014 01:31:17 +0000 (21:31 -0400)]
cgroup: introduce task_css_is_root()
Determining the css of a task usually requires RCU read lock as that's
the only thing which keeps the returned css accessible till its
reference is acquired; however, testing whether a task belongs to the
root can be performed without dereferencing the returned css by
comparing the returned pointer against the root one in init_css_set[]
which never changes.
Implement task_css_is_root() which can be invoked in any context.
This will be used by the scheduled cgroup_freezer change.
v2: cgroup no longer supports modular controllers. No need to export
init_css_set. Pointed out by Li.
Tejun Heo [Mon, 12 May 2014 17:56:27 +0000 (13:56 -0400)]
kernfs, sysfs, cgroup: restrict extra perm check on open to sysfs
The kernfs open method - kernfs_fop_open() - inherited extra
permission checks from sysfs. While the vfs layer allows ignoring the
read/write permissions checks if the issuer has CAP_DAC_OVERRIDE,
sysfs explicitly denied open regardless of the cap if the file doesn't
have any of the UGO perms of the requested access or doesn't implement
the requested operation. It can be debated whether this was a good
idea or not but the behavior is too subtle and dangerous to change at
this point.
After cgroup got converted to kernfs, this extra perm check also got
applied to cgroup breaking libcgroup which opens write-only files with
O_RDWR as root. This patch gates the extra open permission check with
a new flag KERNFS_ROOT_EXTRA_OPEN_PERM_CHECK and enables it for sysfs.
For sysfs, nothing changes. For cgroup, root now can perform any
operation regardless of the permissions as it was before kernfs
conversion. Note that kernfs still fails unimplemented operations
with -EINVAL.
While at it, add comments explaining KERNFS_ROOT flags.
Su Friendy [Wed, 7 May 2014 05:54:52 +0000 (13:54 +0800)]
iommu/amd: fix enabling exclusion range for an exact device
set_device_exclusion_range(u16 devid, struct ivmd_header *m) enables
exclusion range for ONE device. IOMMU does not translate the access
to the exclusion range from the device.
The device is specified by input argument 'devid'. But 'devid' is not
passed to the actual set function set_dev_entry_bit(), instead
'm->devid' is passed. 'm->devid' does not specify the exact device
which needs enable the exclusion range. 'm->devid' represents DeviceID
field of IVMD, which has different meaning depends on IVMD type.
The caller init_exclusion_range() sets 'devid' for ONE device. When
m->type is equal to ACPI_IVMD_TYPE_ALL or ACPI_IVMD_TYPE_RANGE,
'm->devid' is not equal to 'devid'.
Alex Williamson [Tue, 22 Apr 2014 16:08:40 +0000 (10:08 -0600)]
iommu/amd: Fix interrupt remapping for aliased devices
An apparent cut and paste error prevents the correct flags from being
set on the alias device resulting in MSI on conventional PCI devices
failing to work. This also produces error events from the IOMMU like:
Broadwell display controller has 3 stream DMA engines. DMA0 cannot update DMA
postion buffer properly while DMA1 and DMA2 can work well. So this patch masks
the buggy DMA0 by keeping it as opened.
This is a tentative workaround, so keep the change small as Takashi suggested.
Linus Torvalds [Tue, 13 May 2014 02:28:52 +0000 (11:28 +0900)]
Merge tag 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging
Pull hwmon fixes from Guenter Roeck:
"Fix resource leak as well as broken store function in emc1403 driver,
and add support for additional chip revisions"
* tag 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
hwmon: (emc1403) Support full range of known chip revision numbers
hwmon: (emc1403) Fix resource leak on module unload
hwmon: (emc1403) fix inverted store_hyst()
Linus Torvalds [Tue, 13 May 2014 02:25:56 +0000 (11:25 +0900)]
Merge branch 'for-3.15-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu
Pull a percpu fix from Tejun Heo:
"Fix for a percpu allocator bug where it could try to kfree() a memory
region allocated using vmalloc(). The bug has been there for years
now and is unlikely to have ever triggered given the size of struct
pcpu_chunk. It's still theoretically possible and the fix is simple
and safe enough, so the patch is marked with -stable"
* 'for-3.15-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu:
percpu: make pcpu_alloc_chunk() use pcpu_mem_free() instead of kfree()
Linus Torvalds [Tue, 13 May 2014 02:24:07 +0000 (11:24 +0900)]
Merge branch 'for-3.15-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq
Pull workqueue fixes from Tejun Heo:
"Fixes for two bugs in workqueue.
One is exiting with internal mutex held in a failure path of
wq_update_unbound_numa(). The other is a subtle and unlikely
use-after-possible-last-put in the rescuer logic. Both have been
around for quite some time now and are unlikely to have triggered
noticeably often. All patches are marked for -stable backport"
* 'for-3.15-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
workqueue: fix a possible race condition between rescuer and pwq-release
workqueue: make rescuer_thread() empty wq->maydays list before exiting
workqueue: fix bugs in wq_update_unbound_numa() failure path
Linus Torvalds [Tue, 13 May 2014 02:22:57 +0000 (11:22 +0900)]
Merge branch 'for-3.15-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup
Pull cgroup fixes from Tejun Heo:
"During recent restructuring, device_cgroup unified config input check
and enforcement logic; unfortunately, it turned out to share too much.
Aristeu's patches fix the breakage and marked for -stable backport.
The other two patches are fallouts from kernfs conversion. The blkcg
change is temporary and will go away once kernfs internal locking gets
simplified (patches pending)"
* 'for-3.15-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
blkcg: use trylock on blkcg_pol_mutex in blkcg_reset_stats()
device_cgroup: check if exception removal is allowed
device_cgroup: fix the comment format for recently added functions
device_cgroup: rework device access check and exception checking
cgroup: fix the retry path of cgroup_mount()
Linus Torvalds [Tue, 13 May 2014 02:21:01 +0000 (11:21 +0900)]
Merge tag 'stable/for-linus-3.15-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip
Pull xen bug fixes from David Vrabel:
- Fix arm64 crash on boot.
- Quiet a noisy arm build warning (virt_to_pfn() redefined).
* tag 'stable/for-linus-3.15-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
arm64: introduce virt_to_pfn
xen/events/fifo: correctly align bitops
arm/xen: Remove definiition of virt_to_pfn in asm/xen/page.h
Linus Torvalds [Tue, 13 May 2014 02:11:48 +0000 (11:11 +0900)]
Merge tag 'md/3.15-fixes' of git://neil.brown.name/md
Pull md bugfixes from Neil Brown:
"Two bugfixes for md in 3.15
Both tagged for -stable"
* tag 'md/3.15-fixes' of git://neil.brown.name/md:
md: avoid possible spinning md thread at shutdown.
md/raid10: call wait_barrier() for each request submitted.
Linus Torvalds [Tue, 13 May 2014 02:07:02 +0000 (11:07 +0900)]
Merge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Pull ARM SoC fixes from Olof Johansson:
"Seems like we've had more fixes than usual this release cycle, but
there's nothing in particular that we're doing differently. Perhaps
it's just one of those cycles where more people are finding more
regressions (and/or that the latency of when people actually test
what's been in the tree for a while is catching up so that we get the
bug reports now).
The bigger changes here are are for TI and Marvell platforms:
* Timing changes for GPMC (generic localbus) on OMAP causing some
largeish DTS deltas.
* Fixes to window allocation on PCI for mvebu touching drivers/
stuff. Patches have acks from subsystem maintainers where needed.
* A fix from Thomas for a botched DT conversion in drivers/edma.
There's a handful of other fixes for the above platforms as well as
sunxi, at91, i.MX. I also included a MAINTAINER update for Broadcom,
and a trivial move of a binding doc.
I know you said you'd be offline this week, but I might as well post
it for when you return. :)"
I'm not quite offline yet. Doing a few pulls in the last hour before my
internet goes away..
* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (31 commits)
MAINTAINERS: update Broadcom ARM tree location and add an SoC family
ARM: dts: i.MX53: Fix ipu register space size
ARM: dts: kirkwood: fix mislocated pcie-controller nodes
ARM: sunxi: Enable GMAC in sunxi_defconfig
ARM: common: edma: Fix xbar mapping
ARM: sun7i: Fix i2c4 base address
ARM: Kirkwood: T5325: Fix double probe of Codec
ARM: mvebu: enable the SATA interface on Armada 375 DB
ARM: mvebu: specify I2C bus frequency on Armada 370 DB
ARM: mvebu: use qsgmii phy-mode for Armada XP GP interfaces
ARM: mvebu: fix NOR bus-width in Armada XP OpenBlocks AX3 Device Tree
ARM: mvebu: fix NOR bus-width in Armada XP DB Device Tree
ARM: mvebu: fix NOR bus-width in Armada XP GP Device Tree
ARM: dts: AM3517: Disable absent IPs inherited from OMAP3
ARM: dts: OMAP2: Fix interrupts for OMAP2420 mailbox
ARM: dts: OMAP5: Add mailbox dt node to fix boot warning
ARM: OMAP5: Switch to THUMB mode if needed on secondary CPU
ARM: dts: am437x-gp-evm: Do not reset gpio5
ARM: dts: omap3-igep0020: use SMSC9221 timings
PCI: mvebu: split PCIe BARs into multiple MBus windows when needed
...
Ben Dooks [Tue, 15 Apr 2014 16:06:34 +0000 (17:06 +0100)]
clk: shmobile: clk-mstp: change to using clock-indices
With the addition of clock-indices, we need to change the renesas
clock implementation to use these instead of the local definition
of "renesas,clock-indices".
Since this will break booting with older device trees, we add a
simple auto-detection of which properties are present.
Aaro Koskinen [Thu, 3 Apr 2014 19:24:01 +0000 (22:24 +0300)]
MIPS/loongson2_cpufreq: Fix CPU clock rate setting
Loongson2 has been using (incorrectly) kHz for cpu_clk rate. This has
been unnoticed, as loongson2_cpufreq was the only place where the rate
was set/get. After commit 652ed95d5fa6074b3c4ea245deb0691f1acb6656
(cpufreq: introduce cpufreq_generic_get() routine) things however broke,
and now loops_per_jiffy adjustments are incorrect (1000 times too long).
The patch fixes this by changing cpu_clk rate to Hz.
Paul Bolle [Thu, 3 Apr 2014 09:10:20 +0000 (11:10 +0200)]
MIPS: Loongson: No need to select GENERIC_HARDIRQS_NO__DO_IRQ
Commit 0e476d91244e ("MIPS: Loongson: Add Loongson-3 Kconfig options")
added "select GENERIC_HARDIRQS_NO__DO_IRQ". But the Kconfig symbol
GENERIC_HARDIRQS_NO__DO_IRQ was already removed in v2.6.38, so that
select is a nop. Drop it.
This change reverts most of commit 60724ca59eda766a30be57aec6b49bc3e2bead91 [MIPS: IP checksums: Remove
unncessary .set pseudos] that introduced warnings with the
CPU_DADDI_WORKAROUNDS option set:
arch/mips/lib/csum_partial.S: Assembler messages:
arch/mips/lib/csum_partial.S:467: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:467: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:467: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:467: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:467: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:467: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:467: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:467: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:467: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:467: Warning: used $3 with ".set at=$3"
[...]
arch/mips/lib/csum_partial.S:577: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:577: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:577: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:601: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:601: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:601: Warning: used $3 with ".set at=$3"
arch/mips/lib/csum_partial.S:601: Warning: used $3 with ".set at=$3"
[and so on, and so on...]
The warnings are benign and good code is produced regardless because no
macros that'd use the assembler's temporary register are involved, however
the `.set noat' directives removed by the commit referred are crucial to
guarantee this is still going to be the case after any changes in the
future. Therefore they need to be brought back to place which this
change does.
This corrects assembler warnings and broken code generated in
__strncpy_from_user_asm:
arch/mips/lib/strncpy_user.S: Assembler messages:
arch/mips/lib/strncpy_user.S:52: Warning: Macro instruction expanded into
multiple instructions in a branch delay slot
with the CPU_DADDI_WORKAROUNDS option set. The function schedules delay
slots manually where there is really no need to as GAS is happy to do it
all itself, so undo it all and remove `.set noreorder'.
Consequently the function loops indefinitely, showing up prominently as a
hang in the delay loop calibration at bootstrap.
This change corrects the problem by forcing the immediate 1 into a
register while keeping code produced identical where CPU_DADDI_WORKAROUNDS
is disabled.
Commit 231a35d37293ab88d325a9cb94e5474c156282c0 [[MIPS] RM: Collected
changes] broke DECstation support by introducing an incompatible copy of
arch/mips/dec/prom/call_o32.S in arch/mips/fw/lib/, built unconditionally.
The copy happens to land earlier of the two among the modules used in the
link and is therefore chosen for the DECstation rather than the intended
original. As a result random kernel data is corrupted because a pointer
to the "%s" formatted output template is used as a temporary stack pointer
rather than being passed down to prom_printf. This also explains why
prom_printf still works, up to a point -- the next argument is the actual
string to output so it works just fine as the output template until enough
kernel data has been corrupted to cause a crash.
This change adjusts the modified wrapper in arch/mips/fw/lib/call_o32.S to
let callers request no stack switching by passing a null temporary stack
pointer in $a1, reworks the DECstation callers to work with the updated
interface and removes the old copy from arch/mips/dec/prom/call_o32.S. A
few minor readability adjustments are included as well, most importantly
O32_SZREG is now used throughout where applicable rather than hardcoded
multiplies of 4 and $fp is used to access the argument save area as a more
usual register to operate the stack with rather than $s0.
Finally an update is made to the temporary stack space used by the SNI
platform to guarantee 8-byte alignment as per o32 requirements.
MIPS: DEC: Bus error handler <asm/cpu-type.h> fixes
Commit 69f24d1784b631b81a54eb57c49bf46536dd2382 [MIPS: Optimize
current_cpu_type() for better code.] missed an update for two DECstation
bus error support files that now do not build, this is a fix.
Add mailing list and patchwork URL so patches now have a defined path.
Also add myself as a TURBOchannel maintainer. However Maciej is going
to the the resident TURBOchannel guru.
Nicolin Chen [Tue, 6 May 2014 08:56:02 +0000 (16:56 +0800)]
ASoC: fsl_esai: Set PCRC and PRRC registers at the end of hw_params()
According to Reference Manual -- ESAI Initialization chapter, as the
standard procedure of ESAI personal reset, the PCRC and PRRC registers
should be remained in its reset value and then configured after T/RCCR
and T/RCR configurations's done but before TE/RE's enabling.
So this patch moves PCRC and PRRC settings to the end of hw_params().
Nicolin Chen [Tue, 6 May 2014 08:55:59 +0000 (16:55 +0800)]
ASoC: fsl_esai: Fix incorrect condition within ratio range check for FP
The range here from 1 to 16 is confined to FP divider only while the
sck_div indicates if the calculation contains PSR and PM dividers. So
for the case using PSR and PM since the sck_div is true, the range of
ratio would simply become bigger than 16.
So this patch fixes the condition here and adds one line comments to
make the purpose here clear.
Currently when the DAPM context bias level is SUSPEND and the target bias level
is OFF dapm_pre_sequence_async() will first transition to PREPARE and
dapm_post_sequence_async() will then transition back from PREPARE to STANDBY and
then to OFF.
This patch makes sure that dapm_pre_sequence_async() only transitions to PREPARE
when either going to ON or away from ON. This avoids the extra unnecessary
transitions.
Mika Westerberg [Thu, 8 May 2014 14:30:31 +0000 (17:30 +0300)]
spi/pxa2xx: Prevent DMA from transferring too many bytes
In case we are doing DMA transfer and the size of the buffer is not multiple
of 4 bytes the driver truncates that to 4-byte boundary and tries to handle
remaining bytes using PIO.
Or that is what it tried to do. What actually happens is that it calls
ALIGN() to the buffer size which aligns it to the next 4-byte boundary
(doesn't truncate). Doing this results 1-3 bytes extra to be transferred.
Furthermore we handle remaining bytes using PIO which results one extra
byte to be transferred. In worst case the driver transfers 4 extra bytes.
While investigating this it turned out that the DMA hardware doesn't even
have such limitation so we can solve this by dropping the code that tries
to handle unaligned bytes.
ASoC: dapm: Skip CODEC<->CODEC links in connect_dai_link_widgets()
For CODEC to CODEC DAI links the paths are created in snd_soc_dapm_new_pcm().
Also for CODEC to CODEC links the widgets are connected cross-over via a DAI
link widget, meaning that the capture widget of one CODEC will be connected to
the playback widget of the other and vice versa. Whereas
snd_soc_dapm_connect_dai_link_widgets() directly connects the playback widget of
the CPU DAI to the playback widget of the CODEC DAI and the capture widget of
the CPU DAI to the capture widget of the CODEC DAI. So not skipping
CODEC<->CODEC links in snd_soc_dapm_connect_dai_link_widgets() will create
incorrect connections between the two CODECs which will cause DAPM to detect
active paths where there are none and unnecessarily power up widgets.
Fixes: b893ea5 ("ASoC: sapm: Automatically connect DAI link widgets in DAPM graph.") Cc: <[email protected]> (for 3.14+) Signed-off-by: Lars-Peter Clausen <[email protected]> Signed-off-by: Mark Brown <[email protected]>
Josef Gajdusek [Mon, 12 May 2014 11:48:26 +0000 (13:48 +0200)]
hwmon: (emc1403) Support full range of known chip revision numbers
The datasheet for EMC1413/EMC1414, which is fully compatible to
EMC1403/1404 and uses the same chip identification, references revision
numbers 0x01, 0x03, and 0x04. Accept the full range of revision numbers
from 0x01 to 0x04 to make sure none are missed.
Jean Delvare [Mon, 12 May 2014 09:44:51 +0000 (11:44 +0200)]
hwmon: (emc1403) Fix resource leak on module unload
Commit 454aee17f claims to convert driver emc1403 to use
devm_hwmon_device_register_with_groups, however the patch itself makes
use of hwmon_device_register_with_groups instead. As the driver remove
function was still dropped, the hwmon device is no longer unregistered
on driver removal, leading to a resource leak.
Josef Gajdusek [Sun, 11 May 2014 12:40:44 +0000 (14:40 +0200)]
hwmon: (emc1403) fix inverted store_hyst()
Attempts to set the hysteresis value to a temperature below the target
limit fails with "write error: Numerical result out of range" due to
an inverted comparison.
Sudeep Holla [Fri, 9 May 2014 16:37:44 +0000 (17:37 +0100)]
arm64: use cpu_online_mask when using forced irq_set_affinity
Commit 01f8fa4f01d8("genirq: Allow forcing cpu affinity of interrupts")
enabled the forced irq_set_affinity which previously refused to route an
interrupt to an offline cpu.
Commit ffde1de64012("irqchip: Gic: Support forced affinity setting")
implements this force logic and disables the cpu online check for GIC
interrupt controller.
When __cpu_disable calls migrate_irqs, it disables the current cpu in
cpu_online_mask and uses forced irq_set_affinity to migrate the IRQs
away from the cpu but passes affinity mask with the cpu being offlined
also included in it.
When calling irq_set_affinity with force == true in a cpu hotplug path,
the caller must ensure that the cpu being offlined is not present in the
affinity mask or it may be selected as the target CPU, leading to the
interrupt not being migrated.
This patch uses cpu_online_mask when using forced irq_set_affinity so
that the IRQs are properly migrated away.
virt_to_pfn has been defined in arch/arm/include/asm/memory.h by commit e26a9e0 "ARM: Better virt_to_page() handling" and Xen has come to rely
on it. Introduce virt_to_pfn on arm64 too.
Viresh Kumar [Mon, 12 May 2014 08:12:29 +0000 (13:42 +0530)]
hrtimer: Set expiry time before switch_hrtimer_base()
switch_hrtimer_base() calls hrtimer_check_target() which ensures that
we do not migrate a timer to a remote cpu if the timer expires before
the current programmed expiry time on that remote cpu.
But __hrtimer_start_range_ns() calls switch_hrtimer_base() before the
new expiry time is set. So the sanity check in hrtimer_check_target()
is operating on stale or even uninitialized data.
Update expiry time before calling switch_hrtimer_base().
drivers: sh: compile drivers/sh/pm_runtime.c if ARCH_SHMOBILE_MULTI
If the kernel is built to support multi-ARM configuration with shmobile
support built in, then drivers/sh is not built. This contains the PM
runtime code in drivers/sh/pm_runtime.c, which implicitly enables the
module clocks for all devices, and thus is quite essential.
Without this, the state of clocks depends on implicit reset state, or on
the bootloader.
If ARCH_SHMOBILE_MULTI then build the drivers/sh directory, but ensure that
bits that may conflict (drivers/sh/clk if the common clock framework is
enabled) or are not used (drivers/sh/intc), are not built.
Also, only enable the PM runtime code when actually running on a shmobile
SoCs that needs it.
ARCH_SHMOBILE_MULTI was added a while ago by commit efacfce5f8a523457e9419a25d52fe39db00b26a ("ARM: shmobile: Introduce
ARCH_SHMOBILE_MULTI"), but drivers/sh was compiled for both
ARCH_SHMOBILE_LEGACY and ARCH_SHMOBILE_MULTI until commit bf98c1eac1d4a6bcf00532e4fa41d8126cd6c187 ("ARM: Rename ARCH_SHMOBILE to
ARCH_SHMOBILE_LEGACY").
Anton Blanchard [Fri, 9 May 2014 07:47:12 +0000 (17:47 +1000)]
powerpc: irq work racing with timer interrupt can result in timer interrupt hang
I am seeing an issue where a CPU running perf eventually hangs.
Traces show timer interrupts happening every 4 seconds even
when a userspace task is running on the CPU. /proc/timer_list
also shows pending hrtimers have not run in over an hour,
including the scheduler.
Looking closer, decrementers_next_tb is getting set to
0xffffffffffffffff, and at that point we will never take
a timer interrupt again.
In __timer_interrupt() we set decrementers_next_tb to
0xffffffffffffffff and rely on ->event_handler to update it:
*next_tb = ~(u64)0;
if (evt->event_handler)
evt->event_handler(evt);
In this case ->event_handler is hrtimer_interrupt. This will eventually
call back through the clockevents code with the next event to be
programmed:
static int decrementer_set_next_event(unsigned long evt,
struct clock_event_device *dev)
{
/* Don't adjust the decrementer if some irq work is pending */
if (test_irq_work_pending())
return 0;
__get_cpu_var(decrementers_next_tb) = get_tb_or_rtc() + evt;
If irq work came in between these two points, we will return
before updating decrementers_next_tb and we never process a timer
interrupt again.
This looks to have been introduced by 0215f7d8c53f (powerpc: Fix races
with irq_work). Fix it by removing the early exit and relying on
code later on in the function to force an early decrementer:
/* We may have raced with new irq work */
if (test_irq_work_pending())
set_dec(1);
H. Peter Anvin [Mon, 12 May 2014 03:25:20 +0000 (20:25 -0700)]
x86, rdrand: When nordrand is specified, disable RDSEED as well
One can logically expect that when the user has specified "nordrand",
the user doesn't want any use of the CPU random number generator,
neither RDRAND nor RDSEED, so disable both.
Resetting root port has more stuff to do than that for PCIe switch
ports and we should have resetting root port done in firmware instead
of the kernel itself. The problem was introduced by commit 5b2e198e
("powerpc/powernv: Rework EEH reset").
Dirk Brandewie [Thu, 8 May 2014 19:57:23 +0000 (12:57 -0700)]
intel_pstate: Set turbo VID for BayTrail
A documentation update exposed that there is a separate set of VID
values that must be used in the turbo/boost P state range. Add
enumerating and setting the correct VID for P states in the turbo
range.
Chromebooks (at least Acer C720 and Pixel) implement an ACPI object
for TPM, but don't implement the _DSM method to support PPI. As
a result, the TPM driver fails to load on those machines after
commit 1569a4c4ceba (ACPI / TPM: detect PPI features by checking
availability of _DSM functions) which causes them to fail to
resume from system suspend, becuase they require the TPM hardware
to be put into the right state during resume and the TPM driver
is necessary for that.
Fix the problem by making tpm_add_ppi() return 0 when tpm_ppi_handle
is still NULL after walking the ACPI namespace in search for the PPI
_DSM, which allows the TPM driver to load and operate the hardware
(during system resume in particular), but avoid creating the PPI
sysfs group in that case.
This change is based on a prototype patch from Jiang Liu.
Fixes: 1569a4c4ceba (ACPI / TPM: detect PPI features by checking availability of _DSM functions)
References: https://bugzilla.kernel.org/show_bug.cgi?id=74021 Reported-by: James Duley <[email protected]> Reported-by: Phillip Dixon <[email protected]> Tested-by: Brandon Casey <[email protected]> Cc: 3.14+ <[email protected]> # 3.14+ Signed-off-by: Rafael J. Wysocki <[email protected]>
Linus Torvalds [Sun, 11 May 2014 09:06:13 +0000 (18:06 +0900)]
Merge branch 'for-3.15' of git://linux-nfs.org/~bfields/linux
Pull nfsd fixes from Bruce Fields.
* 'for-3.15' of git://linux-nfs.org/~bfields/linux:
NFSD: Call ->set_acl with a NULL ACL structure if no entries
NFSd: call rpc_destroy_wait_queue() from free_client()
NFSd: Move default initialisers from create_client() to alloc_client()
Linus Torvalds [Sun, 11 May 2014 08:56:53 +0000 (17:56 +0900)]
Merge branch 'akpm' (incoming from Andrew)
Merge misc fixes from Andrew Morton:
"4 fixes"
* emailed patches from Andrew Morton <[email protected]>:
mm, thp: close race between mremap() and split_huge_page()
mm: postpone the disabling of kmemleak early logging
MAINTAINERS: update maintainership of LTP
drivers/rtc/rtc-hym8563.c: set uie_unsupported
mm, thp: close race between mremap() and split_huge_page()
It's critical for split_huge_page() (and migration) to catch and freeze
all PMDs on rmap walk. It gets tricky if there's concurrent fork() or
mremap() since usually we copy/move page table entries on dup_mm() or
move_page_tables() without rmap lock taken. To get it work we rely on
rmap walk order to not miss any entry. We expect to see destination VMA
after source one to work correctly.
But after switching rmap implementation to interval tree it's not always
possible to preserve expected walk order.
It works fine for dup_mm() since new VMA has the same vma_start_pgoff()
/ vma_last_pgoff() and explicitly insert dst VMA after src one with
vma_interval_tree_insert_after().
But on move_vma() destination VMA can be merged into adjacent one and as
result shifted left in interval tree. Fortunately, we can detect the
situation and prevent race with rmap walk by moving page table entries
under rmap lock. See commit 38a76013ad80.
Problem is that we miss the lock when we move transhuge PMD. Most
likely this bug caused the crash[1].
Catalin Marinas [Fri, 9 May 2014 22:36:59 +0000 (15:36 -0700)]
mm: postpone the disabling of kmemleak early logging
Commit 8910ae896c8c ("kmemleak: change some global variables to int"),
in addition to the atomic -> int conversion, moved the disabling of
kmemleak_early_log to the beginning of the kmemleak_init() function,
before the full kmemleak tracing is actually enabled. In this small
window, kmem_cache_create() is called by kmemleak which triggers
additional memory allocation that are not traced. This patch restores
the original logic with kmemleak_early_log disabling when kmemleak is
fully functional.
Heiko Stuebner [Fri, 9 May 2014 22:36:57 +0000 (15:36 -0700)]
drivers/rtc/rtc-hym8563.c: set uie_unsupported
The alarm of the hym8563 only supports a minute accuracy, while the uie
wants an alarm one second in the future. Therefore things like the
select() syscall will fail with a timeout, because the next alarm will
happen in a worst case of 60 seconds.
Olof Johansson [Sun, 11 May 2014 03:25:07 +0000 (20:25 -0700)]
Merge tag 'sunxi-fixes-for-3.15' of https://github.com/mripard/linux into fixes
Merge 'Allwinner fixes for 3.15' from Maxime Ripard:
Set of fixes for the Allwinner support for 3.15
Some minor things, the major thing being the enabling of the GMAC driver in
sunxi_defconfig that will un-break Olof's autobooters.
* tag 'sunxi-fixes-for-3.15' of https://github.com/mripard/linux:
ARM: sunxi: Enable GMAC in sunxi_defconfig
ARM: sun7i: Fix i2c4 base address
ARM: sun7i: fix PLL4 clock and add PLL8
ACPI / proc: Do not say when /proc interfaces will be deleted in Kconfig
Do not tell people in the Kconfig help when exactly we are going to
remove the deprecated ACPI interfaces in /proc, because, honestly,
we don't know. We will remove them when they are not used any more.
In particular, do not tell them that the interfaces will be removed
in a kernel release that already happened long ago.
Linus Torvalds [Fri, 9 May 2014 19:24:20 +0000 (12:24 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Peter Anvin:
"A somewhat unpleasantly large collection of small fixes. The big ones
are the __visible tree sweep and a fix for 'earlyprintk=efi,keep'. It
was using __init functions with predictably suboptimal results.
Another key fix is a build fix which would produce output that simply
would not decompress correctly in some configuration, due to the
existing Makefiles picking up an unfortunate local label and mistaking
it for the global symbol _end.
Additional fixes include the handling of 64-bit numbers when setting
the vdso data page (a latent bug which became manifest when i386
started exporting a vdso with time functions), a fix to the new MSR
manipulation accessors which would cause features to not get properly
unblocked, a build fix for 32-bit userland, and a few new platform
quirks"
* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86, vdso, time: Cast tv_nsec to u64 for proper shifting in update_vsyscall()
x86: Fix typo in MSR_IA32_MISC_ENABLE_LIMIT_CPUID macro
x86: Fix typo preventing msr_set/clear_bit from having an effect
x86/intel: Add quirk to disable HPET for the Baytrail platform
x86/hpet: Make boot_hpet_disable extern
x86-64, build: Fix stack protector Makefile breakage with 32-bit userland
x86/reboot: Add reboot quirk for Certec BPC600
asmlinkage: Add explicit __visible to drivers/*, lib/*, kernel/*
asmlinkage, x86: Add explicit __visible to arch/x86/*
asmlinkage: Revert "lto: Make asmlinkage __visible"
x86, build: Don't get confused by local symbols
x86/efi: earlyprintk=efi,keep fix
Jani Nikula [Fri, 9 May 2014 11:52:34 +0000 (14:52 +0300)]
drm/i915/vlv: reset VLV media force wake request register
Media force wake get hangs the machine when the system is booted without
displays attached. The assumption is that (at least some versions of)
the firmware has skipped some initialization in that case.
Empirical evidence suggests we need to reset the media force wake
request register in addition to the render one to avoid hangs.
projects / linux.git / log