We've unfortunately started seeing a situation where percpu interrupts
are partitioned in the system: one arbitrary set of CPUs has an
interrupt connected to a type of device, while another disjoint
set of CPUs has the same interrupt connected to another type of device.
This makes it impossible to have a device driver requesting this interrupt
using the current percpu-interrupt abstraction, as the same interrupt number
is now potentially claimed by at least two drivers, and we forbid interrupt
sharing on per-cpu interrupt.
A solution to this is to turn things upside down. Let's assume that our
system describes all the possible partitions for a given interrupt, and
give each of them a unique identifier. It is then possible to create
a namespace where the affinity identifier itself is a form of interrupt
number. At this point, it becomes easy to implement a set of partitions
as a cascaded irqchip, each affinity identifier being the HW irq.
This allows us to keep a number of nice properties:
- Each partition results in a separate percpu-interrupt (with a restrictied
affinity), which keeps drivers happy.
- Because the underlying interrupt is still per-cpu, the overhead of
the indirection can be kept pretty minimal.
- The core code can ignore most of that crap.
For that purpose, we implement a small library that deals with some of
the boilerplate code, relying on platform-specific drivers to provide
a description of the affinity sets and a set of callbacks.
Conflicts:
drivers/irqchip/Kconfig
drivers/irqchip/Makefile
Change-Id: Ie6b2bc8c4c152f0dcd3fbcab8950fae781338322
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: devicetree@vger.kernel.org
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Rob Herring <robh+dt@kernel.org>
Link: http://lkml.kernel.org/r/1460365075-7316-4-git-send-email-marc.zyngier@arm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Caesar Wang <wxt@rock-chips.com>
(cherry picked from commit 9e2c986cb4)
When iterating over the irq domain list, we try to match a domain
either by calling a match() function or by comparing a number
of fields passed as parameters.
Both approaches are a bit restrictive:
- match() is DT specific and only takes a device node
- the fallback case only deals with the fwnode_handle
It would be useful if we had a per-domain function that would
actually perform the matching check on the whole of the
irq_fwspec structure. This would allow for a domain to triage
matching attempts that need to extend beyond the fwnode.
Let's introduce irq_find_matching_fwspec(), which takes a full
blown irq_fwspec structure, and call into a select() function
implemented by the irqdomain. irq_find_matching_fwnode() is
made a wrapper around irq_find_matching_fwspec in order to
preserve compatibility.
Change-Id: I07df9af068d114c80cd97b9cb987a70c0e24afda
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: devicetree@vger.kernel.org
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Rob Herring <robh+dt@kernel.org>
Link: http://lkml.kernel.org/r/1460365075-7316-2-git-send-email-marc.zyngier@arm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Caesar Wang <wxt@rock-chips.com>
(cherry picked from commit 651e8b54ab)
Isolate hardware abstraction (FDT) code to gic_of_init().
Rest of the logic goes to gic_init_bases() and expects well
defined data to initialize GIC properly. The same solution
is used for GICv2 driver.
This is needed for ACPI initialization later.
Change-Id: I61fcbd96ecd2dc8130cdd2d6ce79841eb184e87b
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Tomasz Nowicki <tomasz.nowicki@linaro.org>
Signed-off-by: Hanjun Guo <hanjun.guo@linaro.org>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Caesar Wang <wxt@rock-chips.com>
(cherry picked from commit db57d7460e)
The arm,gic-v3 binding was written with good intentions and doesn't
enforce interrupt-cells to be 3, therefore making it easy to extend
the irq description in future if necessary:
> Cells 4 and beyond are reserved for future use.
Unfortunately, this sentence is immediately followed up with:
> When the 1st cell has a value of 0 or 1, cells 4 and beyond act as
> padding, and may be ignored. It is recommended that padding cells
> have a value of 0.
Consequently, any extensions to the PPI or SPI interrupt specifiers must
be able to work with random crap from legacy DTs, effectively
necessitating a new interrupt type in the first cell. Sigh.
This patch fixes the text so that additional, reserved cells are
required to be zero. This looks like a reasonable thing to require and
is already satisifed by the .dts files in-tree.
Change-Id: Ia5b07ab4243c0a4492b7c4516af95b86974c42a0
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Caesar Wang <wxt@rock-chips.com>
(cherry picked from commit 4aff7b8546)
Let's take the (outlandish) example of an interrupt controller
capable of handling both wired interrupts and PCI MSIs.
With the current code, the PCI MSI domain is going to be tagged
with DOMAIN_BUS_PCI_MSI, and the wired domain with DOMAIN_BUS_ANY.
Things get hairy when we start looking up the domain for a wired
interrupt (typically when creating it based on some firmware
information - DT or ACPI).
In irq_create_fwspec_mapping(), we perform the lookup using
DOMAIN_BUS_ANY, which is actually used as a wildcard. This gives
us one chance out of two to end up with the wrong domain, and
we try to configure a wired interrupt with the MSI domain.
Everything grinds to a halt pretty quickly.
What we really need to do is to start looking for a domain that
would uniquely identify a wired interrupt domain, and only use
DOMAIN_BUS_ANY as a fallback.
In order to solve this, let's introduce a new DOMAIN_BUS_WIRED
token, which is going to be used exactly as described above.
Of course, this depends on the irqchip to setup the domain
bus_token, and nobody had to implement this so far.
Only so far.
Change-Id: Ia71c7475354eb38ab9b15423560aa3d28ae16381
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Frank Rowand <frowand.list@gmail.com>
Cc: Grant Likely <grant.likely@linaro.org>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Jiang Liu <jiang.liu@linux.intel.com>
Link: http://lkml.kernel.org/r/1453816347-32720-2-git-send-email-marc.zyngier@arm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Caesar Wang <wxt@rock-chips.com>
(cherry picked from commit 530cbe100e)
Since there will be several places checking if fwnode.type
is equal FWNODE_IRQCHIP, this patch adds a convenient function
for this purpose.
Change-Id: I65ab9e1350428de18864ba493256b959efc01f45
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Suravee Suthikulpanit <Suravee.Suthikulpanit@amd.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Caesar Wang <wxt@rock-chips.com>
(cherry picked from commit 75aba7b0e9)
Cortex-A72 has a PMUv3 implementation that is compatible with the PMU
implemented by Cortex-A57.
This patch hooks up the new compatible string so that the Cortex-A57
event mappings are used.
Change-Id: I06b39699fa019d61be81a1a275f7eb6eed17808a
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Huang, Tao <huangtao@rock-chips.com>
(cherry picked from commit 5d7ee87708)
It's all very well providing an events directory to userspace that
details our events in terms of "event=0xNN", but if we don't define how
to encode the "event" field in the perf attr.config, then it's a waste
of time.
This patch adds a single format entry to describe that the event field
occupies the bottom 10 bits of our config field on ARMv8 (PMUv3).
Change-Id: I71f9ebf92cd2f7083c10f20a8707a91d4517cbcb
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Huang, Tao <huangtao@rock-chips.com>
(cherry picked from commit 57d7412395)
Add additional information about the ARM architected hardware events
to make counters self describing. This makes the hardware PMUs easier
to use as perf list contains possible events instead of users having
to refer to documentation like the ARM TRMs.
Change-Id: Idb004bb6d9889f8e63f518d105e238d43956b561
Signed-off-by: Drew Richardson <drew.richardson@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Huang, Tao <huangtao@rock-chips.com>
(cherry picked from commit 9e9caa6a49)
The enums are not necessary and this allows the event values to be
used to construct static strings at compile time.
Change-Id: I01049434e5ddc5c51b7ae914e9c55a0ef6bf66d9
Signed-off-by: Drew Richardson <drew.richardson@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Huang, Tao <huangtao@rock-chips.com>
(cherry picked from commit 90381cba64)
Note: dec/enc dev name change to rockchip-vpu-dec/rockchip-vpu-enc
Change-Id: I35d168fa7ccf6df4465affd01f6c3c5456182897
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
Adjust to new v4l2 APIs and fix some debug logs.
Change-Id: Iafba102fa326c669efcbb0baeb8897fe660dcdd4
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
This patch adds the DMA_ATTR_ALLOC_SINGLE_PAGES attribute to the
DMA-mapping subsystem.
This attribute can be used as a hint to the DMA-mapping subsystem that
it's likely not worth it to try to allocate large pages behind the
scenes. Large pages are likely to make an IOMMU TLB work more
efficiently but may not be worth it. See the Documentation contained in
this patch for more details about this attribute and when to use it.
Note that the name of the hint (DMA_ATTR_ALLOC_SINGLE_PAGES) is loosely
based on the name MADV_NOHUGEPAGE. Just as there is MADV_NOHUGEPAGE
vs. MADV_HUGEPAGE we could also add an "opposite" attribute to
DMA_ATTR_ALLOC_SINGLE_PAGES. Without having the "opposite" attribute
the lack of DMA_ATTR_ALLOC_SINGLE_PAGES means "use your best judgement
about whether to use small pages or large pages".
BUG=chromium:570532
TEST=Stress memory and watch cat videos.
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Acked-by: Marek Szyprowski <m.szyprowski@samsung.com>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
(am from https://patchwork.kernel.org/patch/8007151/)
Reviewed-on: https://chromium-review.googlesource.com/322334
Reviewed-by: Dmitry Torokhov <dtor@chromium.org>
Change-Id: I9d0af41446b5c41d6f39a2b77e711179d0b40eca
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
When set to 1 for CAPTURE queues by the driver on calling vb2_queue_init(),
forces the buffers on the queue to be allocated/mapped with
DMA_BIDIRECTIONAL DMA direction flag, instead of DMA_FROM_DEVICE. This
allows the device not only to write to the buffers, but also read out from
them. This may be useful e.g. for codec hardware, which may be using
CAPTURE buffers as reference to decode other buffers.
This flag is ignored for OUTPUT queues, as we don't want to allow HW to
be able to write to OUTPUT buffers.
Signed-off-by: Pawel Osciak <posciak@chromium.org>
BUG=chrome-os-partner:45346
TEST=video playback
Reviewed-on: https://chromium-review.googlesource.com/300726
Commit-Ready: Pawel Osciak <posciak@chromium.org>
Tested-by: Pawel Osciak <posciak@chromium.org>
Reviewed-by: Tomasz Figa <tfiga@chromium.org>
Change-Id: Ifba955eef75ac23c9a13edab04bc1fe7f5375c70
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
Some drivers also need a control like
V4L2_CID_MPEG_MFC51_VIDEO_FORCE_FRAME_TYPE to force an encoder
key frame. Add a general V4L2_CID_MPEG_VIDEO_FORCE_KEY_FRAME
so the new drivers and applications can use it.
Signed-off-by: Wu-Cheng Li <wuchengli@chromium.org>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
(cherry picked from commit 0c8485ca3f2aaf7842d45ba24c667a9492c9900f)
Signed-off-by: Daniel Kurtz <djkurtz@chromium.org>
BUG=chromium:572825
TEST=Build and boot oak-rev5 to UI
TEST=emerge-smaug chromeos-kernel-3_18
Reviewed-on: https://chromium-review.googlesource.com/328870
Commit-Ready: Daniel Kurtz <djkurtz@chromium.org>
Tested-by: Daniel Kurtz <djkurtz@chromium.org>
Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>
Change-Id: I45de3048d41edbe443b3d202c17e79f2d448213b
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
We do video allocation all the time and we need it to be fast. Plus TLB
efficiency isn't terribly important for video.
That means we want to set DMA_ATTR_ALLOC_SINGLE_PAGES
See also the previous change ("ARM: dma-mapping: Use
DMA_ATTR_ALLOC_SINGLE_PAGES hint to optimize alloc")
BUG=chromium:570532
TEST=Memory pressure + cat videos is even smoother!
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/322336
Reviewed-by: Dmitry Torokhov <dtor@chromium.org>
Change-Id: I8bda3d9655daaa893c7bead7108b863607d1614f
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
On RK3288 there is an issue with certain hardware state being corrupted
while decoding certain streams, which affects encoding task run directly
after that decoding task. To reinitialize the state properly, a dummy
encoding of a single 64x64 pixels keyframe must be performed before the
real encoding is run.
This patch adds necessary workaround code to the driver, which makes it
execute an encoding task using dummy buffers with static parameters
manually selected for lowest performance overhead and to assure that
aforementioned hardware state is reinitialized.
BUG=chrome-os-partner:41585
TEST=AppRTC loopback
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/286284
Reviewed-by: Pawel Osciak <posciak@chromium.org>
Change-Id: I019d1983633ec2cf2818956a7bf988314d853cdf
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
Currently plane sizes calculation on encoder side is happening only in
vidioc_s_fmt() function, howerver as a prerequisite for further patch
adding further code which needs this operation, this patch adds a common
helper function, which performs this operation.
BUG=chrome-os-partner:41585
TEST=AppRTC loopback
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/289047
Reviewed-by: Pawel Osciak <posciak@chromium.org>
Change-Id: I6e5769667ae40b3fb5758ce9471b4ddd6866183f
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
Currently there is only one place in the code checking whether given
context is decoding or encoding. However as a prerequisite for further
patch adding more such checks, this patch adds a common helper function
which returns appropriate enum value depending on operating mode of
given context.
BUG=chrome-os-partner:41585
TEST=AppRTC loopback
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/288661
Reviewed-by: Pawel Osciak <posciak@chromium.org>
Change-Id: Ide145770ca77897ada3cb878c4b9c9787824a827
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
Currently, FRAME_HEADER_SIZE is 256, but i've saw
the hdr be larger then 256 when doing screen share.
so we should increase it.
This needs change the define in v4lplugin too.
(change id: Ia6c2271b727218692c4e0b1603d243f32d2f1d77)
BUG=chromium:497324
TEST=screen share through Hangouts
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Reviewed-on: https://chromium-review.googlesource.com/277398
Reviewed-by: Tomasz Figa <tfiga@chromium.org>
Change-Id: I06e7712420404c43661774e176a7b9333ddc3def
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
This patch adds implementations of VIDIOC_ENUM_FRAMESIZES for rk3288-vpu
encoder and decoder devices. This IOCTL lets the userspace learn about
frame size limits of the hardware.
BUG=chromium:485409
TEST=vda/veatests, Chrome with crrev.com/1097913002.
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/269867
Reviewed-by: Heng-ruey Hsu <henryhsu@google.com>
Change-Id: Ia23a89c2f380b16cc7ef8338d33946d62f8a68fe
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
As a prerequisite for using find_format() helpers from contexts in which
a v4l2_format struct is not available, this patch makes it take u32 fourcc
as its argument instead, since it was the only member of that struct it
actually used anyway.
BUG=chromium:485409
TEST=vda/veatests, Chrome with crrev.com/1097913002.
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/269866
Reviewed-by: Pawel Osciak <posciak@chromium.org>
Trybot-Ready: Pawel Osciak <posciak@chromium.org>
Change-Id: Ifa9c4e3e378fbeafb6453a01b9e4f7c11606025b
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
For rk3288-vpu, kernel mapping of video buffers is required only for
encoder bitstream output buffers for additional bistream formatting. Any
other buffers can be allocated without kernel mapping, greatly
conserving the limited pool of vmalloc memory.
This patch modifies the rk3288-vpu driver to use the newly added vb2-dc
interface to create two separate allocation contexts, one for
allocations with kernel mapping and one without.
BUG=chrome-os-partner:38873
TEST=vda/vea unit tests
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/265364
Reviewed-by: Pawel Osciak <posciak@chromium.org>
Change-Id: I4154802dda2329934dea675a242d67e80b925db0
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
Currently the driver does not implement suspend and resume PM ops.
However when system is entering suspend, the driver should prevent
submitting further runs to the hardware and wait for current run to be
finished. To resume playback after leaving sleep state, next run, if
available, must be submitted to the hardware.
This patch adds proper suspend and resume callbacks to handle this.
BUG=chrome-os-partner:38565
TEST=suspend and resume veyron_jerry several times with video playing
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/263662
Reviewed-by: Pawel Osciak <posciak@chromium.org>
Change-Id: Id07b7d681ef78655879ce77c9705b1c25231df9d
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
Even though H264 standard allows arbitrary order of DPB entries, the
hardware requires picture with given POC to use the same DPB entry for
its whole lifetime. This means that the driver needs to reorder DPB
array to suit this requirement.
This patch modifies the driver to reorder H264 DPB and should fix
corruption issues when DPB array received from userspace does not meet
hardware requirements.
BUG=chrome-os-partner:38416
TEST=http://www.youtube.com/embed/YE7VzlLtp-4
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/263370
Reviewed-by: Pawel Osciak <posciak@chromium.org>
Tested-by: Caesar Wang <wxt@rock-chips.com>
Change-Id: Ibbd0e2bc4e527aadd21ef08ec68866678bf8a659
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
Current code always assumed the maximum supported resolution to be
1920x1088, and minimum 8x4 however the real limits are 48x48 and 3840x2160
for decoder and 96x96 and 1920x1088 for encoder. This patch modifies the
driver to use correct limits and also fixes incorrect log message.
BUG=chrome-os-partner:38232,chromium:464920
TEST=Screen sharing of a window bigger than 1920x1088 to Jerry
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/261851
Reviewed-by: Pawel Osciak <posciak@chromium.org>
Change-Id: I0c51e5a9ad235716ee447e052455b97ed0c295de
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
If segmentation_enable flag is zero, no matter what
segmentation_update_flag is, we cannot set it to HW.
Before this patch, incorrect segmentation_update_flag
might be set to HW which caused hangout corruption.
BUG=chrome-os-partner:35531
TEST=make a video chat by hangout with other device.
Signed-off-by: ZhiChao Yu <zhichao.yu@rock-chips.com>
Reviewed-on: https://chromium-review.googlesource.com/242732
Reviewed-by: Tomasz Figa <tfiga@chromium.org>
Tested-by: Tomasz Figa <tfiga@chromium.org>
Commit-Queue: Tomasz Figa <tfiga@chromium.org>
Change-Id: I86c00c27d3c97854db8c4164289fa434b29819ff
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Yakir Yang <ykk@rock-chips.com>
