| Age | Commit message (Collapse) | Author |
|---|
|
While investigating why the load-balancer did funny I found that the
rq->cpu_load[] tables were completely screwy.. a bit more digging
revealed that the updates that got through were missing ticks followed
by a catchup of 2 ticks.
The catchup assumes the cpu was idle during that time (since only nohz
can cause missed ticks and the machine is idle etc..) this means that
esp. the higher indices were significantly lower than they ought to
be.
The reason for this is that its not correct to compare against jiffies
on every jiffy on any other cpu than the cpu that updates jiffies.
This patch cludges around it by only doing the catch-up stuff from
nohz_idle_balance() and doing the regular stuff unconditionally from
the tick.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: pjt@google.com
Cc: Venkatesh Pallipadi <venki@google.com>
Link: http://lkml.kernel.org/n/tip-tp4kj18xdd5aj4vvj0qg55s2@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
It's far too easy to get ridiculously large imbalance pct when you
scale it like that. Use a fixed 125% for now.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/n/tip-zsriaft1dv7hhboyrpvqjy6s@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
Patches c22402a2f ("sched/fair: Let minimally loaded cpu balance the
group") and 0ce90475 ("sched/fair: Add some serialization to the
sched_domain load-balance walk") are horribly broken so revert them.
The problem is that while it sounds good to have the minimally loaded
cpu do the pulling of more load, the way we walk the domains there is
absolutely no guarantee this cpu will actually get to the domain. In
fact its very likely it wont. Therefore the higher up the tree we get,
the less likely it is we'll balance at all.
The first of mask always walks up, while sucky in that it accumulates
load on the first cpu and needs extra passes to spread it out at least
guarantees a cpu gets up that far and load-balancing happens at all.
Since its now always the first and idle cpus should always be able to
balance so they get a task as fast as possible we can also do away
with the added serialization.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/n/tip-rpuhs5s56aiv1aw7khv9zkw6@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
Commit ad7687dde ("x86/numa: Check for nonsensical topologies on real
hw as well") is broken in that the condition can trigger for valid
setups but only changes the end result for invalid setups with no real
means of discerning between those.
Rewrite set_cpu_sibling_map() to make the code clearer and make sure
to only warn when the check changes the end result.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/n/tip-klcwahu3gx467uhfiqjyhdcs@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
There's no need to convert a node number to a node number by
pretending its a cpu number..
Reported-by: Yinghai Lu <yinghai@kernel.org>
Reported-and-Tested-by: Greg Pearson <greg.pearson@hp.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/n/tip-0sqhrht34phowgclj12dgk8h@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
Put a comment that clarifies the condition that handles both signed
and unsigned case for logical min/max in hid_add_field().
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
|
|
When logical maximum is 0xffffffff, the parser fails even if
logical minimum is more than 0.
By HID specification this is a valid combination.
Signed-off-by: srinivas pandruvada <srinivas.pandruvada@intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
|
|
Signed-off-by: Antonio Ospite <ospite@studenti.unina.it>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Move drivers/input/fixp-arith.h to include/linux so that the functions
defined there can be used by other subsystems, for instance some video
devices ISPs can control the output HUE value by setting registers for
sin(HUE) and cos(HUE).
Signed-off-by: Antonio Ospite <ospite@studenti.unina.it>
Acked-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Also merge the "COLORS" control into it as it was V4L2_CID_SATURATION
anyway.
Signed-off-by: Antonio Ospite <ospite@studenti.unina.it>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Enabling vflip leads to a much better image, with vflip disabled the
image looks washed out as if there is a too high brightness setting.
Since we don't know how to lower the brightness setting when not
vflipping, simply always vflip and tell userspace to flip the image back,
resulting in a much better (less washed out) image.
Since the image is now no longer too bright, also modify the luminance
level the auto-gain algorithm aims for.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Our init sequence was not setting the register page to point to bank 1
before setting what should be the control reg. This causes the camera to
sometimes have its LED on after init. First selecting register bank 1,
rather then assuming the current register bank is bank 1, fixes this.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Now that the pac207 driver has been converted to the control framework, there
are no remaining users.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
The pac207's exposure control is a clock-divider, so it goes with quite
big steps. So lets use an autogain algorithm optimised for that.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/uprobes
|
|
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
We intend to eventually port all sub-drivers to the control-framework. At
which point it will make more sense to have the ctrl_handler in gspca_dev
then to have it in the subdrivers. Lets move it there now, to avoid a lot
of work to move it later.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
In case CONFIG_X86_VSMP is not set, limit the number of CPUs to
the number of CPUs of the first board.
Also make CONFIG_X86_VSMP depend on CONFIG_SMP, as there's
little point in having a vsmp machine with a single CPU.
Signed-off-by: Shai Fultheim <shai@scalemp.com>
[ido@wizery.com: rebased, fixed minor coding-style issues]
Signed-off-by: Ido Yariv <ido@wizery.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Now that stv06xx is using the control framework it is no longer necessary
to have a (non const) copy of sd_desc inside the sd specific data struct.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
HdG:
1) Let the gspca-core cleanup the controls on control-init error, like
with the other converted sub drivers
2) Note this also fixes a bug in the hdcs1020 support which was wrongly
reporting an exposure range of 0-65535, even though the effective range was
only 0-255
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
fault_reason - 0x20 == ARRAY_SIZE(irq_remap_fault_reasons) is
one past the end of the array.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Joerg Roedel <joerg.roedel@amd.com>
Cc: Youquan Song <youquan.song@intel.com>
Cc: walter harms <wharms@bfs.de>
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: <stable@kernel.org>
Link: http://lkml.kernel.org/r/20120513170938.GA4280@elgon.mountain
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
HdG: Small fix: don't register some controls for sensors which don't
have an implementation for them.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
The initial version was done by HV, corrections were made by HdG, and some
final small changes again by HV.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Even with BRC the highest quality setting is not usable, BRC strips so
much data from each MCU that the quality becomes worse then using a lower
quality setting to begin with.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Always automatically adjust the Bit Rate Control setting as needed, independent
of the sensor type. BRC is needed to not run out of bandwidth with higher
quality settings independent of the sensor.
Also only automatically adjust BRC, and don't adjust the JPEG quality control
automatically, as that is not needed and leads to ugly flashes when it is
changed. Note that before this patch-set the quality was never changed
either due to the bugs in the quality handling fixed in previous patches in
this set.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
The current code is using bits 0-1 of register 8 of the zc3xx controller
to set the JPEG quality, but the correct bits are bits 1-2. Bit 0 selects
between truncation or rounding in the quantization phase of the compression,
since rounding generally gives better results it should thus always be 1.
This patch also corrects the quality percentages which belong to the 4
different settings.
Last this patch removes the different reg 8 defaults depending on the sensor
type. Some of them where going for a default quality setting of 50%, which
generally is not necessary in any way and results in poor image quality.
75% is a good default to use for all scenarios.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
When the user changes the JPEG quality while the camera is streaming, the
driver should not only change the JPEG headers send to userspace, but also
actually tell the camera to use a different quantization table.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
The sensor specific dev_post_unset_alt functions all try to write to the
bridge, and none free any memory, so they should be skipped if stop0
is called on disconnection.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Just a small cleanup.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
This is necessary to ensure that worker-threads accessing the device
are stopped before our disconnect handler returns.
This causes a problem with stream_off calling sd_stop0 a second time
when the device handle is closed. This is fixed by setting
gscpa_dev->streaming to 0 on disconnect.
Note that now stream_off will never be called on a disconnected device,
and the present check can thus be removed from stream_off.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
So that we don't start a read stream when an app is only polling for control
events.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Fixing i386 allnoconfig built errors:
arch/x86/built-in.o: In function `amd_pmu_hw_config':
perf_event_amd.c:(.text+0xc3e1): undefined reference to `get_ibs_caps'
Reported-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Due to latency concerns the VIDIOC_QBUF, DQBUF and QUERYBUF do not use the
core lock, instead they rely only on queue_lock.
Changes by HdG:
1) Change release from the video_device to the v4l2_device, to avoid a
race on disconnect.
2) Adjust for the V4L2 core changes which cause non ioctl fops to no longer
take the V4L2 core lock.
[mchehab@redhat.com: fix a merge conflict]
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
There are two bugs here: first the calls to stop0 (in gspca_suspend) and
gspca_init_transfer (in gspca_resume) need to be called with the usb_lock held.
That's true for the other places they are called and it is what subdrivers
expect. Quite a few will unlock the usb_lock in stop0 while waiting for a
worker thread to finish, and if usb_lock isn't held then that can cause a
kernel oops.
The other problem is that a worker thread needs to detect that it has to
halt due to a suspend. Otherwise it will just go on looping. So add tests
against gspca_dev->frozen in the worker threads that need it.
Hdg, 2 minor changes:
1) The finepix device is ok with stopping reading a frame halfway through,
so add frozen checks in all places where we also check if we're still
streaming
2) Use gspca_dev->dev instead of gspca_dev->present to check for disconnect
in all touched drivers. I plan to do this everywhere in the future, and
most relevant lines in the touched drivers are already modified by this
patch.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Add V4L2_CAP_DEVICE_CAPS support to querycap and replace -EINVAL by
-ENOTTY whenever an ioctl is not supported.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
In order to support control event gspca has to use struct v4l2_fh.
As a bonus feature this also gives priority handling for free.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Prepare for control events: free up file->private_data by using
video_drvdata(file) to get to the gspca_dev struct.
[mchehab@redhat.com: fix a compile error: ‘file’ undeclared]
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Make the necessary changes to allow subdrivers to use the control framework.
This does not add control event support, that comes later.
It add a init_control cam_op that is called after init in probe that allows
the subdriver to set up the controls.
HdG: Call v4l2_ctrl_handler_setup from resume instead of
gspca_set_default_mode, as we just want to resend the current ctrl values to
the device.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
My last locking rework for pwc mistakenly assumed that videbuf2 does its
own locking, but it does not! This patch fixes the missing locking by
moving over the the video_device lock, and introducing a separate lock
for the videobuf2_queue.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Various error paths in fileio_init where not setting the request-count
to 0 when unrequesting the buffers on error to init the fileio emulation.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
This used to be the default if the lock pointer was set, but now that lock is by
default only used for ioctl serialization. Those drivers that already used
core locking have this flag set explicitly, except for some drivers where
it was obvious that there was no need to serialize any file operations other
than ioctl.
The drivers that didn't need this flag were:
drivers/media/radio/dsbr100.c
drivers/media/radio/radio-isa.c
drivers/media/radio/radio-keene.c
drivers/media/radio/radio-miropcm20.c
drivers/media/radio/radio-mr800.c
drivers/media/radio/radio-tea5764.c
drivers/media/radio/radio-timb.c
drivers/media/video/vivi.c
sound/i2c/other/tea575x-tuner.c
The other drivers that use core locking and where it was not immediately
obvious that this flag wasn't needed were changed so that the flag is set
together with a comment that that driver needs work to avoid having to
set that flag. This will often involve taking the core lock in the fops
themselves.
Eventually this flag should go and it should not be used in new drivers.
There are a few reasons why we want to avoid core locking of non-ioctl
fops: in the case of mmap this can lead to a deadlock in rare situations
since when mmap is called the mmap_sem is held and it is possible for
other parts of the code to take that lock as well (copy_from_user()/copy_to_user()
perform a down_read(&mm->mmap_sem) when a page fault occurs).
It is very unlikely that that happens since the core lock serializes all
fops, but the kernel warns about it if lock validation is turned on.
For poll it is also undesirable to take the core lock as that can introduce
increased latency. The same is true for read/write.
While it was possible to make flags or something to turn on/off taking the
core lock for each file operation, in practice it is much simpler to just
not take it at all except for ioctl and leave it to the driver to take the
lock. There are only a handful fops compared to the zillion ioctls we have.
I also wanted to make it obvious which drivers still take the lock for all
fops, so that's why I chose to have drivers set it explicitly.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
|
Rather than checking the priority for each ioctl that needs to, just mark
such ioctls in the table and do it only once.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
|
