Dave Airlie [Thu, 23 Dec 2021 20:08:29 +0000 (06:08 +1000)]
Merge tag 'drm-intel-gt-next-2021-12-23' of git://anongit.freedesktop.org/drm/drm-intel into drm-next
Driver Changes:
- Added bits of DG2 support around page table handling (Stuart Summers, Matthew Auld)
- Fixed wakeref leak in PMU busyness during reset in GuC mode (Umesh Nerlige Ramappa)
- Fixed debugfs access crash if GuC failed to load (John Harrison)
- Bring back GuC error log to error capture, undoing accidental earlier breakage (Thomas Hellström)
- Fixed memory leak in error capture caused by earlier refactoring (Thomas Hellström)
- Exclude reserved stolen from driver use (Chris Wilson)
- Add memory region sanity checking and optional full test (Chris Wilson)
- Fixed buffer size truncation in TTM shmemfs backend (Robert Beckett)
- Use correct lock and don't overwrite internal data structures when stealing GuC context ids (Matthew Brost)
- Don't hog IRQs when destroying GuC contexts (John Harrison)
- Make GuC to Host communication more robust (Matthew Brost)
- Continuation of locking refactoring around VMA and backing store handling (Maarten Lankhorst)
- Improve performance of reading GuC log from debugfs (John Harrison)
- Log when GuC fails to reset an engine (John Harrison)
- Speed up GuC/HuC firmware loading by requesting RP0 (Vinay Belgaumkar)
- Further work on asynchronous VMA unbinding (Thomas Hellström, Christian König)
- Refactor GuC/HuC firmware handling to prepare for future platforms (John Harrison)
- Prepare for future different GuC/HuC firmware signing key sizes (Daniele Ceraolo Spurio, Michal Wajdeczko)
- Add noreclaim annotations (Matthew Auld)
- Remove racey GEM_BUG_ON between GPU reset and GuC communication handling (Matthew Brost)
- Refactor i915->gt with to_gt(i915) to prepare for future platforms (Michał Winiarski, Andi Shyti)
- Increase GuC log size for CONFIG_DEBUG_GEM (John Harrison)
- Fixed engine busyness in selftests when in GuC mode (Umesh Nerlige Ramappa)
- Make engine parking work with PREEMPT_RT (Sebastian Andrzej Siewior)
- Replace X86_FEATURE_PAT with pat_enabled() (Lucas De Marchi)
- Selftest for stealing of guc ids (Matthew Brost)
Dave Airlie [Thu, 23 Dec 2021 20:01:39 +0000 (06:01 +1000)]
Merge branch 'etnaviv/next' of https://git.pengutronix.de/git/lst/linux into drm-next
- make etnaviv work on IOMMU enabled systems
- fix mapping of command buffers on systems with more than 4GB RAM
- close a DoS vector
- fix spurious GPU resets
Lucas Stach [Wed, 22 Dec 2021 00:17:28 +0000 (01:17 +0100)]
drm/etnaviv: consider completed fence seqno in hang check
Some GPU heavy test programs manage to trigger the hangcheck quite often.
If there are no other GPU users in the system and the test program
exhibits a very regular structure in the commandstreams that are being
submitted, we can end up with two distinct submits managing to trigger
the hangcheck with the FE in a very similar address range. This leads
the hangcheck to believe that the GPU is stuck, while in reality the GPU
is already busy working on a different job. To avoid those spurious
GPU resets, also remember and consider the last completed fence seqno
in the hang check.
Reported-by: Joerg Albert <joerg.albert@iav.de> Signed-off-by: Lucas Stach <l.stach@pengutronix.de> Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Maarten Lankhorst [Wed, 22 Dec 2021 15:56:22 +0000 (16:56 +0100)]
drm/i915: Use trylock instead of blocking lock for __i915_gem_free_objects.
Convert free_work into delayed_work, similar to ttm to allow converting the
blocking lock in __i915_gem_free_objects to a trylock.
Unlike ttm, the object should already be idle, as it's kept alive
by a reference through struct i915_vma->active, which is dropped
after all vma's are idle.
Because of this, we can use a no wait by default, or when the lock
is contested, we use ttm's 10 ms.
The trylock should only fail when the object is sharing it's resv with
other objects, and typically objects are not kept locked for a long
time, so we can safely retry on failure.
Dave Airlie [Thu, 23 Dec 2021 02:40:45 +0000 (12:40 +1000)]
Merge tag 'exynos-drm-next-for-v5.17' of git://git.kernel.org/pub/scm/linux/kernel/git/daeinki/drm-exynos into drm-next
Four cleanups
- Replacing lagacy gpio interface of dsi driver with gpiod one.
- Implementing a generic GEM object mmap and use it instead of
exynos specific one.
- Dropping the use of label from dsi driver. Which also fixes
a build warning.
- Just trivial cleanup by dropping unnecessay code.
Dave Airlie [Thu, 23 Dec 2021 02:30:56 +0000 (12:30 +1000)]
Merge tag 'drm/tegra/for-5.17-rc1' of https://gitlab.freedesktop.org/drm/tegra into drm-next
drm/tegra: Changes for v5.17-rc1
This contains a fairly large rework that makes the buffer objects behave
more according to what the DMA-BUF infrastructure expects. A buffer
object cache is implemented on top of that to make certain operations
such as page-flipping more efficient by avoiding needless map/unmap
operations. This in turn is useful to implement asynchronous commits to
support legacy cursor updates.
Another fairly big addition is the NVDEC driver. This uses the updated
UABI introduced in v5.15-rc1 to provide access to the video decode
engines found on Tegra210 and later.
This also includes some power management improvements that are useful on
older devices in particular because they, together with a bunch of other
changes across the kernel, allow the system to scale down frequency and
voltages when mostly idle and prevent these devices from becoming
excessively hot.
The remainder of these changes is an assortment of cleanups and minor
fixes.
Thomas Hellström [Tue, 21 Dec 2021 20:00:50 +0000 (21:00 +0100)]
drm/i915: Require the vm mutex for i915_vma_bind()
Protect updates of struct i915_vma flags and async binding / unbinding
with the vm::mutex. This means that i915_vma_bind() needs to assert
vm::mutex held. In order to make that possible drop the caching of
kmap_atomic() maps around i915_vma_bind().
An alternative would be to use kmap_local() but since we block cpu
unplugging during sleeps inside kmap_local() sections this may have
unwanted side-effects. Particularly since we might wait for gpu while
holding the vm mutex.
This change may theoretically increase execbuf cpu-usage on snb, but
at least on non-highmem systems that increase should be very small.
Christian König [Tue, 21 Dec 2021 20:00:48 +0000 (21:00 +0100)]
drm/i915: remove questionable fence optimization during copy
First of all as discussed multiple times now kernel copies *must* always
wait for all fences in a BO before actually doing the copy. This is
mandatory.
Additional to that drop the handling when there can't be a shared slot
allocated on the source BO and just properly return an error code.
Otherwise this code path would only be tested under out of memory
conditions.
Thomas Hellström [Tue, 21 Dec 2021 20:00:47 +0000 (21:00 +0100)]
drm/i915: Avoid using the i915_fence_array when collecting dependencies
Since the gt migration code was using only a single fence for
dependencies, these were collected in a dma_fence_array. However, it
turns out that it's illegal to use some dma_fences in a dma_fence_array,
in particular other dma_fence_arrays and dma_fence_chains, and this
causes trouble for us moving forward.
Have the gt migration code instead take a const struct i915_deps for
dependencies. This means we can skip the dma_fence_array creation
and instead pass the struct i915_deps instead to circumvent the
problem.
v2:
- Make the prev_deps() function static. (kernel test robot <lkp@intel.com>)
- Update the struct i915_deps kerneldoc.
v4:
- Rebase.
Inki Dae [Wed, 1 Dec 2021 02:32:14 +0000 (11:32 +0900)]
drm/exynos: drop the use of label from exynos_dsi_register_te_irq
Dropped the use of 'out' label from exynos_dsi_register_te_irq function
because the label isn't needed. This patch returns an error in each
error case directly not going to 'out' label.
With this patch build warning[1] is also fixed, which was reported by
kernel test robot <lkp@intel.com>
Thomas Zimmermann [Mon, 8 Nov 2021 10:28:44 +0000 (11:28 +0100)]
drm/exynos: Implement mmap as GEM object function
Moving the driver-specific mmap code into a GEM object function allows
for using DRM helpers for various mmap callbacks.
The respective exynos functions are being removed. The file_operations
structure exynos_drm_driver_fops is now being created by the helper macro
DEFINE_DRM_GEM_FOPS().
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Fixed merge conflict. Signed-off-by: Inki Dae <inki.dae@samsung.com>
Maíra Canal [Tue, 2 Nov 2021 02:20:03 +0000 (23:20 -0300)]
drm/exynos: Replace legacy gpio interface for gpiod interface
Considering the current transition of the GPIO subsystem, remove all
dependencies of the legacy GPIO interface (linux/gpio.h and linux
/of_gpio.h) and replace it with the descriptor-based GPIO approach.
Vinay Belgaumkar [Thu, 16 Dec 2021 23:30:22 +0000 (15:30 -0800)]
drm/i915/guc: Request RP0 before loading firmware
By default, GT (and GuC) run at RPn. Requesting for RP0
before firmware load can speed up DMA and HuC auth as well.
In addition to writing to 0xA008, we also need to enable
swreq in 0xA024 so that Punit will pay heed to our request.
SLPC will restore the frequency back to RPn after initialization,
but we need to manually do that for the non-SLPC path.
We don't need a manual override in the SLPC disabled case, just
use the intel_rps_set function to ensure consistent RPS state.
John Harrison [Sat, 11 Dec 2021 06:58:58 +0000 (22:58 -0800)]
drm/i915/guc: Flag an error if an engine reset fails
If GuC encounters an error during engine reset, the i915 driver
promotes to full GT reset. This includes an info message about why the
reset is happening. However, that is not treated as a failure by any
of the CI systems because resets are an expected occurrance during
testing. This kind of failure is a major problem and should never
happen. So, complain more loudly and make sure CI notices.
John Harrison [Sat, 11 Dec 2021 06:58:57 +0000 (22:58 -0800)]
drm/i915/guc: Increase GuC log size for CONFIG_DEBUG_GEM
Lots of testing is done with the DEBUG_GEM config option enabled but
not the DEBUG_GUC option. That means we only get teeny-tiny GuC logs
which are not hugely useful. Enabling full DEBUG_GUC also spews lots
of other detailed output that is not generally desired. However,
bigger GuC logs are extremely useful for almost any regression debug.
So enable bigger logs for DEBUG_GEM builds as well.
John Harrison [Sat, 11 Dec 2021 06:58:56 +0000 (22:58 -0800)]
drm/i915/guc: Speed up GuC log dumps
Add support for telling the debugfs interface the size of the GuC log
dump in advance. Without that, the underlying framework keeps calling
the 'show' function with larger and larger buffer allocations until it
fits. That means reading the log from graphics memory many times - 16
times with the full 18MB log size.
v2: Don't return error codes from size query. Report overflow in the
error dump as well (review feedback from Daniele).
Lucas Stach [Fri, 17 Dec 2021 10:59:28 +0000 (11:59 +0100)]
drm/etnaviv: limit submit sizes
Currently we allow rediculous amounts of kernel memory being allocated
via the etnaviv GEM_SUBMIT ioctl, which is a pretty easy DoS vector. Put
some reasonable limits in to fix this.
The commandstream size is limited to 64KB, which was already a soft limit
on older kernels after which the kernel only took submits on a best effort
base, so there is no userspace that tries to submit commandstreams larger
than this. Even if the whole commandstream is a single incrementing address
load, the size limit also limits the number of potential relocs and
referenced buffers to slightly under 64K, so use the same limit for those
arguments. The performance monitoring infrastructure currently supports
less than 50 performance counter signals, so limiting them to 128 on a
single submit seems like a reasonably future-proof number for now. This
number can be bumped if needed without breaking the interface.
Cc: stable@vger.kernel.org Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Lucas Stach <l.stach@pengutronix.de> Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Maarten Lankhorst [Thu, 16 Dec 2021 14:27:39 +0000 (15:27 +0100)]
drm/i915: Ensure i915_vma tests do not get -ENOSPC with the locking changes.
Now that we require locking to evict, multiple vmas from the same object
might not be evicted. This is expected and required, because execbuf will
move to short-term pinning by using the lock only. This will cause these
tests to fail, because they create a ton of vma's for the same object.
Unbind manually to prevent spurious -ENOSPC in those mock tests.
Maarten Lankhorst [Thu, 16 Dec 2021 14:27:38 +0000 (15:27 +0100)]
drm/i915: Ensure gem_contexts selftests work with unbind changes, v2.
In the next commits, we may not evict when refcount = 0.
igt_vm_isolation() continuously tries to pin/unpin at same address,
but also calls put() on the object, which means the object may not
be unpinned in time.
Instead of this, re-use the same object over and over, so they can
be unbound as required.
Changes since v1:
- Fix cleaning up obj_b on failure. (Matt)
Maarten Lankhorst [Thu, 16 Dec 2021 14:27:35 +0000 (15:27 +0100)]
drm/i915: Remove pages_mutex and intel_gtt->vma_ops.set/clear_pages members, v3.
Big delta, but boils down to moving set_pages to i915_vma.c, and removing
the special handling, all callers use the defaults anyway. We only remap
in ggtt, so default case will fall through.
Because we still don't require locking in i915_vma_unpin(), handle this by
using xchg in get_pages(), as it's locked with obj->mutex, and cmpxchg in
unpin, which only fails if we race a against a new pin.
Changes since v1:
- aliasing gtt sets ZERO_SIZE_PTR, not -ENODEV, remove special case
from __i915_vma_get_pages(). (Matt)
Changes since v2:
- Free correct old pages in __i915_vma_get_pages(). (Matt)
Remove race of clearing vma->pages accidentally from put,
free it but leave it set, as only get has the lock.
Maarten Lankhorst [Thu, 16 Dec 2021 14:27:34 +0000 (15:27 +0100)]
drm/i915: Change shrink ordering to use locking around unbinding.
Call drop_pages with the gem object lock held, instead of the other
way around. This will allow us to drop the vma bindings with the
gem object lock held.
We plan to require the object lock for unpinning in the future,
and this is an easy target.
Michał Winiarski [Tue, 14 Dec 2021 19:33:31 +0000 (21:33 +0200)]
drm/i915: Store backpointer to GT in uncore
We now support a per-gt uncore, yet we're not able to infer which GT
we're operating upon. Let's store a backpointer for now.
At this point the early initialization of the gt needs to be
broken in two parts where the first is needed to assign to the gt
the i915 private data pointer and the uncore. A temporary
function has been made and the two parts are
__intel_gt_init_early() and intel_gt_init_early(). This split
will be fixed in the future with the multitile patch.
Dave Airlie [Fri, 17 Dec 2021 06:06:14 +0000 (16:06 +1000)]
Merge tag 'drm-misc-next-2021-12-16' of git://anongit.freedesktop.org/drm/drm-misc into drm-next
drm-misc-next for 5.17:
UAPI Changes:
* vmwgfx: Version bump to 2.20
Cross-subsystem Changes:
* of: Create simple-framebuffer devices in of_platform_default_init()
Core Changes:
* Replace include <linux/kernel.h> with more fine-grained includes
* Document DRM_IOCTL_MODE_GETFB2
* format-helper: Support XRGB2101010 source buffers
Driver Changes:
* amdgpu: Fix runtime PM on some configs
* ast: Fix I2C initialization
* bridge: ti-sn65dsi86: Set regmap max_register
* panel: Add Team Source Display TST043015CMHX plus DT bindings
* simpledrm: Add support for Apple M1
* sprd: Add various drivers plus DT bindings
* vc4: Support 10-bit YUV 4:2:0 output; Fix clock-rate updates
* vmwgfx: Implement GEM support; Implement GL 4.3 support
Dave Airlie [Fri, 17 Dec 2021 05:23:45 +0000 (15:23 +1000)]
Merge tag 'drm-intel-next-2021-12-14' of ssh://git.freedesktop.org/git/drm/drm-intel into drm-next
drm/i915 feature pull #2 for v5.17:
Features and functionality:
- Add eDP privacy screen support (Hans)
- Add Raptor Lake S (RPL-S) support (Anusha)
- Add CD clock squashing support (Mika)
- Properly support ADL-P without force probe (Clint)
- Enable pipe color support (10 bit gamma) for display 13 platforms (Uma)
- Update ADL-P DMC firmware to v2.14 (Madhumitha)
Refactoring and cleanups:
- More FBC refactoring preparing for multiple FBC instances (Ville)
- Plane register cleanups (Ville)
- Header refactoring and include cleanups (Jani)
- Crtc helper and vblank wait function cleanups (Jani, Ville)
- Move pipe/transcoder/abox masks under intel_device_info.display (Ville)
Fixes:
- Add a delay to let eDP source OUI write take effect (Lyude)
- Use div32 version of MPLLB word clock for UHBR on SNPS PHY (Jani)
- Fix DMC firmware loader overflow check (Harshit Mogalapalli)
- Fully disable FBC on FIFO underruns (Ville)
- Disable FBC with double wide pipe as mutually exclusive (Ville)
- DG2 workarounds (Matt)
- Non-x86 build fixes (Siva)
- Fix HDR plane max width for NV12 (Vidya)
- Disable IRQ for selftest timestamp calculation (Anshuman)
- ADL-P VBT DDC pin mapping fix (Tejas)
Merges:
- Backmerge drm-next for privacy screen plumbing (Jani)
Dmitry Osipenko [Sat, 4 Dec 2021 14:58:49 +0000 (17:58 +0300)]
drm/tegra: Add back arm_iommu_detach_device()
DMA buffers of 2D/3D engines aren't mapped properly when
CONFIG_ARM_DMA_USE_IOMMU=y. The memory management code of Tegra DRM driver
has a longstanding overhaul overdue and it's not obvious where the problem
is in this case. Hence let's add back the old workaround which we already
had sometime before. It explicitly detaches DRM devices from the offending
implicit IOMMU domain. This fixes a completely broken 2d/3d drivers in
case of ARM32 multiplatform kernel config.
Cc: stable@vger.kernel.org Fixes: fa6661b7aa0b ("drm/tegra: Optionally attach clients to the IOMMU") Signed-off-by: Dmitry Osipenko <digetx@gmail.com> Signed-off-by: Thierry Reding <treding@nvidia.com>
Dmitry Osipenko [Sat, 4 Dec 2021 14:58:48 +0000 (17:58 +0300)]
gpu: host1x: Add back arm_iommu_detach_device()
Host1x DMA buffer isn't mapped properly when CONFIG_ARM_DMA_USE_IOMMU=y.
The memory management code of Host1x driver has a longstanding overhaul
overdue and it's not obvious where the problem is in this case. Hence
let's add back the old workaround which we already had sometime before.
It explicitly detaches Host1x device from the offending implicit IOMMU
domain. This fixes a completely broken Host1x DMA in case of ARM32
multiplatform kernel config.
Cc: stable@vger.kernel.org Fixes: af1cbfb9bf0f ("gpu: host1x: Support DMA mapping of buffers") Signed-off-by: Dmitry Osipenko <digetx@gmail.com> Signed-off-by: Thierry Reding <treding@nvidia.com>
Dmitry Osipenko [Tue, 30 Nov 2021 23:23:23 +0000 (02:23 +0300)]
drm/tegra: nvdec: Stop channel on suspend
CDMA must be stopped before hardware is suspended. Add channel stopping
to RPM suspend callback. Add system level suspend-resume callbacks.
Runtime PM initialization is moved to host1x client init phase because
RPM callback now uses host1x channel that is available only when host1x
client is registered.
Dmitry Osipenko [Tue, 30 Nov 2021 23:23:22 +0000 (02:23 +0300)]
drm/tegra: vic: Stop channel on suspend
CDMA must be stopped before hardware is suspended. Add channel stopping
to RPM suspend callback. Add system level suspend-resume callbacks.
Runtime PM initialization is moved to host1x client init phase because
RPM callback now uses host1x channel that is available only when host1x
client is registered.
Dmitry Osipenko [Tue, 30 Nov 2021 23:23:19 +0000 (02:23 +0300)]
drm/tegra: hdmi: Add OPP support
The HDMI on Tegra belongs to the core power domain and we're going to
enable GENPD support for the core domain. Now HDMI driver must use
OPP API for driving the controller's clock rate because OPP API takes
care of reconfiguring the domain's performance state based on HDMI clock
rate. Add OPP support to the HDMI driver.
Dmitry Osipenko [Tue, 30 Nov 2021 23:23:16 +0000 (02:23 +0300)]
gpu: host1x: Add host1x_channel_stop()
Add host1x_channel_stop() which waits till channel becomes idle and then
stops the channel hardware. This is needed for supporting suspend/resume
by host1x drivers since the hardware state is lost after power-gating,
thus the channel needs to be stopped before client enters into suspend.
Tested-by: Peter Geis <pgwipeout@gmail.com> # Ouya T30 Tested-by: Paul Fertser <fercerpav@gmail.com> # PAZ00 T20 Tested-by: Nicolas Chauvet <kwizart@gmail.com> # PAZ00 T20 and TK1 T124 Tested-by: Matt Merhar <mattmerhar@protonmail.com> # Ouya T30 Signed-off-by: Dmitry Osipenko <digetx@gmail.com> Signed-off-by: Thierry Reding <treding@nvidia.com>
Dmitry Osipenko [Tue, 30 Nov 2021 23:23:15 +0000 (02:23 +0300)]
gpu: host1x: Add initial runtime PM and OPP support
Add runtime PM and OPP support to the Host1x driver. For the starter we
will keep host1x always-on because dynamic power management require a major
refactoring of the driver code since lot's of code paths are missing the
RPM handling and we're going to remove some of these paths in the future.
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> Tested-by: Peter Geis <pgwipeout@gmail.com> # Ouya T30 Tested-by: Paul Fertser <fercerpav@gmail.com> # PAZ00 T20 Tested-by: Nicolas Chauvet <kwizart@gmail.com> # PAZ00 T20 and TK1 T124 Tested-by: Matt Merhar <mattmerhar@protonmail.com> # Ouya T30 Signed-off-by: Dmitry Osipenko <digetx@gmail.com> Signed-off-by: Thierry Reding <treding@nvidia.com>
Robin Murphy [Fri, 10 Dec 2021 17:54:43 +0000 (17:54 +0000)]
gpu: host1x: Add missing DMA API include
Host1x seems to be relying on picking up dma-mapping.h transitively from
iova.h, which has no reason to include it in the first place. Fix the
former issue before we totally break things by fixing the latter one.
CC: Thierry Reding <thierry.reding@gmail.com> CC: Mikko Perttunen <mperttunen@nvidia.com> CC: dri-devel@lists.freedesktop.org Signed-off-by: Robin Murphy <robin.murphy@arm.com> Signed-off-by: Thierry Reding <treding@nvidia.com>
Robin Murphy [Fri, 10 Dec 2021 17:54:44 +0000 (17:54 +0000)]
drm/tegra: vic: Fix DMA API misuse
Upon failure, dma_alloc_coherent() returns NULL. If that does happen,
passing some uninitialised stack contents to dma_mapping_error() - which
belongs to a different API in the first place - has precious little
chance of detecting it.
Also include the correct header, because the fragile transitive
inclusion currently providing it is going to break soon.
Fixes: 20e7dce255e9 ("drm/tegra: Remove memory allocation from Falcon library") CC: Thierry Reding <thierry.reding@gmail.com> CC: Mikko Perttunen <mperttunen@nvidia.com> CC: dri-devel@lists.freedesktop.org Signed-off-by: Robin Murphy <robin.murphy@arm.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Thierry Reding <treding@nvidia.com>
Arnd Bergmann [Wed, 13 Oct 2021 14:41:36 +0000 (16:41 +0200)]
drm/tegra: Mark nvdec PM functions as __maybe_unused
The resume helper is called conditionally and causes a harmless warning
when stubbed out:
drivers/gpu/drm/tegra/nvdec.c:240:12: error: 'nvdec_runtime_resume' defined but not used [-Werror=unused-function]
240 | static int nvdec_runtime_resume(struct device *dev)
Mark both suspend and resume as __maybe_unused for consistency to avoid
this warning.
Arnd Bergmann [Wed, 13 Oct 2021 14:40:58 +0000 (16:40 +0200)]
drm/tegra: Mark nvdec_writel() as inline
Without CONFIG_IOMMU_API, the nvdec_writel() function is unused, causing
a warning:
drivers/gpu/drm/tegra/nvdec.c:48:13: error: 'nvdec_writel' defined but not used [-Werror=unused-function]
48 | static void nvdec_writel(struct nvdec *nvdec, u32 value, unsigned int offset)
| ^~~~~~~~~~~~
As this is a trivial wrapper around an inline function, mark it as
inline itself, which avoids the warning as well.
drm/tegra: dc: rgb: Allow changing PLLD rate on Tegra30+
Asus Transformer TF700T is a Tegra30 tablet device which uses RGB->DSI
bridge that requires a precise clock rate in order to operate properly.
Tegra30 has a dedicated PLL for each display controller, hence the PLL
rate can be changed freely. Allow PLL rate changes on Tegra30+ for RGB
output. Configure the clock rate before display controller is enabled
since DC itself may be running off this PLL and it's not okay to change
the rate of the active PLL that doesn't support dynamic frequency
switching since hardware will hang.
As of commit 4782c0a5dd88 ("clk: tegra: Don't deassert reset on enabling
clocks"), module resets are no longer automatically deasserted when the
module clock is enabled. To make sure that the gr2d module continues to
work, we need to explicitly control the module reset.
drm/tegra: dc: rgb: Move PCLK shifter programming to CRTC
Asus TF700T tablet uses TC358768 DPI->DSI bridge that sits between Tegra's
DPI output and display panel input. Bridge requires to have stable PCLK
output before RGB encoder is enabled because it uses PCLK by itself to
clock internal logic and bridge is programmed before Tegra's encoder is
enabled. Hence the PCLK clock shifter must be programmed when CRTC is
enabled, otherwise clock is unstable and bridge hangs because of it.
Move the shifter programming from RGB encoder into CRTC.
Mikko Perttunen [Thu, 16 Sep 2021 15:09:20 +0000 (18:09 +0300)]
drm/tegra: Bump VIC/NVDEC clock rates to Fmax
To get full performance out of these engines, bump their clock rates
to maximum. In the future we may want something smarter but this
should be fine for now.
Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com> Signed-off-by: Thierry Reding <treding@nvidia.com>
Mikko Perttunen [Thu, 16 Sep 2021 14:55:17 +0000 (17:55 +0300)]
drm/tegra: Add NVDEC driver
Add support for booting and using NVDEC on Tegra210, Tegra186
and Tegra194 to the Host1x and TegraDRM drivers. Booting in
secure mode is not currently supported.
Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com> Signed-off-by: Thierry Reding <treding@nvidia.com>
drm/tegra: Do not reference tegra_plane_funcs directly
Instead of referencing the tegra_plane_funcs struct directly, use each
plane's vtable instead. This makes it more future-proof in case any of
the planes ever use a different set of functions.
Thierry Reding [Fri, 7 Feb 2020 15:50:52 +0000 (16:50 +0100)]
drm/tegra: Implement buffer object cache
This cache is used to avoid mapping and unmapping buffer objects
unnecessarily. Mappings are cached per client and stay hot until
the buffer object is destroyed.
DMA-BUF requires that each device that accesses a DMA-BUF attaches to it
separately. To do so the host1x_bo_pin() and host1x_bo_unpin() functions
need to be reimplemented so that they can return a mapping, which either
represents an attachment or a map of the driver's own GEM object.
Only couple drivers need to get the -ENODEV error code and majority of
drivers need to explicitly initialize the performance state. Add new
common helper which sets up OPP table for these drivers.
Dmitry Osipenko [Tue, 30 Nov 2021 23:23:08 +0000 (02:23 +0300)]
soc/tegra: Enable runtime PM during OPP state-syncing
GENPD core now can set up domain's performance state properly while device
is RPM-suspended. Runtime PM of a device must be enabled during setup
because GENPD checks whether device is suspended and check doesn't work
while RPM is disabled. Instead of replicating the boilerplate RPM-enable
code around OPP helper for each driver, let's make OPP helper to take care
of enabling it.
Stephen Boyd [Wed, 15 Dec 2021 00:25:29 +0000 (16:25 -0800)]
drm/bridge: ti-sn65dsi86: Set max register for regmap
Set the maximum register to 0xff so we can dump the registers for this
device in debugfs.
Fixes: a095f15c00e2 ("drm/bridge: add support for sn65dsi86 bridge driver") Cc: Rob Clark <robdclark@chromium.org> Cc: Douglas Anderson <dianders@chromium.org> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Signed-off-by: Stephen Boyd <swboyd@chromium.org> Reviewed-by: Robert Foss <robert.foss@linaro.org> Signed-off-by: Robert Foss <robert.foss@linaro.org> Link: https://patchwork.freedesktop.org/patch/msgid/20211215002529.382383-1-swboyd@chromium.org
Signed-off-by: Marek Vasut <marex@denx.de> Cc: Rob Herring <robh+dt@kernel.org> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: devicetree@vger.kernel.org
To: dri-devel@lists.freedesktop.org Acked-by: Rob Herring <robh@kernel.org> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20211127031909.283383-2-marex@denx.de
Marek Vasut [Sat, 27 Nov 2021 03:19:07 +0000 (04:19 +0100)]
dt-bindings: Add Team Source Display Technology vendor prefix
Add vendor prefix for Team Source Display Technology Co., Ltd.
Signed-off-by: Marek Vasut <marex@denx.de> Cc: Rob Herring <robh+dt@kernel.org> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: devicetree@vger.kernel.org
To: dri-devel@lists.freedesktop.org Acked-by: Rob Herring <robh@kernel.org> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20211127031909.283383-1-marex@denx.de
Marek Vasut [Sun, 17 Oct 2021 00:12:04 +0000 (02:12 +0200)]
drm/bridge: lvds-codec: Add support for pixel data sampling edge select
The OnSemi FIN3385 Parallel-to-LVDS encoder has a dedicated input line to
select input pixel data sampling edge. Add DT property "pclk-sample", not
the same as the one used by display timings but rather the same as used by
media, and configure bus flags based on this DT property.
Signed-off-by: Marek Vasut <marex@denx.de> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Rob Herring <robh+dt@kernel.org> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: devicetree@vger.kernel.org
To: dri-devel@lists.freedesktop.org Reviewed-by: Sam Ravnborg <sam@ravnborg.org> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20211017001204.299940-2-marex@denx.de
Marek Vasut [Sun, 17 Oct 2021 00:12:03 +0000 (02:12 +0200)]
dt-bindings: display: bridge: lvds-codec: Document pixel data sampling edge select
The OnSemi FIN3385 Parallel-to-LVDS encoder has a dedicated input line to
select input pixel data sampling edge. Add DT property "pclk-sample", not
the same as the one used by display timings but rather the same as used by
media, to define the pixel data sampling edge.
Signed-off-by: Marek Vasut <marex@denx.de> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Rob Herring <robh+dt@kernel.org> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: devicetree@vger.kernel.org
To: dri-devel@lists.freedesktop.org Reviewed-by: Rob Herring <robh@kernel.org> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20211017001204.299940-1-marex@denx.de
Matthew Brost [Tue, 14 Dec 2021 17:05:00 +0000 (09:05 -0800)]
drm/i915/guc: Selftest for stealing of guc ids
Testing the stealing of guc ids is hard from user space as we have 64k
guc_ids. Add a selftest, which artificially reduces the number of guc
ids, and forces a steal.
The test creates a spinner which is used to block all subsequent
submissions until it completes. Next, a loop creates a context and a NOP
request each iteration until the guc_ids are exhausted (request creation
returns -EAGAIN). The spinner is ended, unblocking all requests created
in the loop. At this point all guc_ids are exhausted but are available
to steal. Try to create another request which should successfully steal
a guc_id. Wait on last request to complete, idle GPU, verify a guc_id
was stolen via a counter, and exit the test. Test also artificially
reduces the number of guc_ids so the test runs in a timely manner.
v2:
(John Harrison)
- s/stole/stolen
- Fix some wording in test description
- Rework indexing into context array
- Add test description to commit message
- Fix typo in commit message
(Checkpatch)
- s/guc/(guc) in NUMBER_MULTI_LRC_GUC_ID
v3:
(John Harrison)
- Set array value to NULL after extracting error
- Fix a few typos in comments / error messages
- Delete redundant comment in commit message
John Harrison [Tue, 14 Dec 2021 17:04:57 +0000 (09:04 -0800)]
drm/i915/guc: Don't hog IRQs when destroying contexts
While attempting to debug a CT deadlock issue in various CI failures
(most easily reproduced with gem_ctx_create/basic-files), I was seeing
CPU deadlock errors being reported. This were because the context
destroy loop was blocking waiting on H2G space from inside an IRQ
spinlock. There no was deadlock as such, it's just that the H2G queue
was full of context destroy commands and GuC was taking a long time to
process them. However, the kernel was seeing the large amount of time
spent inside the IRQ lock as a dead CPU. Various Bad Things(tm) would
then happen (heartbeat failures, CT deadlock errors, outstanding H2G
WARNs, etc.).
Re-working the loop to only acquire the spinlock around the list
management (which is all it is meant to protect) rather than the
entire destroy operation seems to fix all the above issues.
projects / users / jedix / linux-maple.git / log