Sean Paul [Wed, 18 Sep 2019 20:07:29 +0000 (16:07 -0400)]
drm: Measure Self Refresh Entry/Exit times to avoid thrashing
Currently the self refresh idle timer is a const set by the crtc. This
is fine if the self refresh entry/exit times are well-known for all
panels used on that crtc. However panels and workloads can vary quite a
bit, and a timeout which works well for one doesn't work well for
another.
In the extreme, if the timeout is too short we could get in a situation
where the self refresh exits are taking so long we queue up a self refresh
entry before the exit commit is even finished.
This patch changes the idle timeout to a moving average of the entry
times + a moving average of exit times + the crtc constant.
This patch was tested on rockchip, with a kevin CrOS panel the idle
delay averages out to about ~235ms (35 entry + 100 exit + 100 const). On
the same board, the bob panel idle delay lands around ~340ms (90 entry
+ 150 exit + 100 const).
WRT the dedicated mutex in self_refresh_data, it would be nice if we
could rely on drm_crtc.mutex to protect the average times, but there are
a few reasons why a separate lock is a better choice:
- We can't rely on drm_crtc.mutex being held if we're doing a nonblocking
commit
- We can't grab drm_crtc.mutex since drm_modeset_lock() doesn't tell us
whether the lock was already held in the acquire context (it eats
-EALREADY), so we can't tell if we should drop it or not
- We don't need such a heavy-handed lock for what we're trying to do,
commit ordering doesn't matter, so a point-of-use lock will be less
contentious
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link to v1: https://patchwork.freedesktop.org/patch/msgid/20190917200443.64481-2-sean@poorly.run Link: https://patchwork.freedesktop.org/patch/msgid/20190918200734.149876-2-sean@poorly.run
Changes in v2:
- Migrate locking explanation from comment to commit msg (Daniel)
- Turf constant entry delay and multiply the avg times by 2 (Daniel)
Daniel Vetter [Tue, 3 Sep 2019 19:06:42 +0000 (21:06 +0200)]
drm/atomic: Rename crtc_state->pageflip_flags to async_flip
It's the only flag anyone actually cares about. Plus if we're unlucky,
the atomic ioctl might need a different flag for async flips. So
better to abstract this away from the uapi a bit.
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Michel Dänzer <michel@daenzer.net> Cc: Alex Deucher <alexdeucher@gmail.com> Cc: Adam Jackson <ajax@redhat.com> Cc: Sean Paul <sean@poorly.run> Cc: David Airlie <airlied@linux.ie> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Maxime Ripard <maxime.ripard@bootlin.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Cc: Leo Li <sunpeng.li@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: David Francis <David.Francis@amd.com> Cc: Mario Kleiner <mario.kleiner.de@gmail.com> Cc: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: "Christian König" <christian.koenig@amd.com> Cc: Ilia Mirkin <imirkin@alum.mit.edu> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Link: https://patchwork.freedesktop.org/patch/msgid/20190903190642.32588-3-daniel.vetter@ffwll.ch
Daniel Vetter [Tue, 3 Sep 2019 19:06:41 +0000 (21:06 +0200)]
drm/atomic: Reject FLIP_ASYNC unconditionally
It's never been wired up. Only userspace that tried to use it (and
didn't actually check whether anything works, but hey it builds) is
the -modesetting atomic implementation. And we just shut that up.
If there's anyone else then we need to silently accept this flag no
matter what, and find a new one. Because once a flag is tainted, it's
lost.
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Michel Dänzer <michel@daenzer.net> Cc: Alex Deucher <alexdeucher@gmail.com> Cc: Adam Jackson <ajax@redhat.com> Cc: Sean Paul <sean@poorly.run> Cc: David Airlie <airlied@linux.ie> Cc: stable@vger.kernel.org Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190903190642.32588-2-daniel.vetter@ffwll.ch
Daniel Vetter [Thu, 5 Sep 2019 18:53:18 +0000 (20:53 +0200)]
drm/atomic: Take the atomic toys away from X
The -modesetting ddx has a totally broken idea of how atomic works:
- doesn't disable old connectors, assuming they get auto-disable like
with the legacy setcrtc
- assumes ASYNC_FLIP is wired through for the atomic ioctl
- not a single call to TEST_ONLY
Iow the implementation is a 1:1 translation of legacy ioctls to
atomic, which is a) broken b) pointless.
We already have bugs in both i915 and amdgpu-DC where this prevents us
from enabling neat features.
If anyone ever cares about atomic in X we can easily add a new atomic
level (req->value == 2) for X to get back the shiny toys.
Since these broken versions of -modesetting have been shipping,
there's really no other way to get out of this bind.
v2:
- add an informational dmesg output (Rob, Ajax)
- reorder after the DRIVER_ATOMIC check to avoid useless noise (Ilia)
- allow req->value > 2 so that X can do another attempt at atomic in
the future
v3: Go with paranoid, insist that the X should be first (suggested by
Rob)
Cc: Ilia Mirkin <imirkin@alum.mit.edu>
References: https://gitlab.freedesktop.org/xorg/xserver/issues/629
References: https://gitlab.freedesktop.org/xorg/xserver/merge_requests/180
References: abbc0697d5fb ("drm/fb: revert the i915 Actually configure untiled displays from master") Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> (v1) Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> (v1) Cc: Michel Dänzer <michel@daenzer.net> Cc: Alex Deucher <alexdeucher@gmail.com> Cc: Adam Jackson <ajax@redhat.com> Acked-by: Adam Jackson <ajax@redhat.com> Cc: Sean Paul <sean@poorly.run> Cc: David Airlie <airlied@linux.ie> Cc: Rob Clark <robdclark@gmail.com> Acked-by: Rob Clark <robdclark@gmail.com> Cc: stable@vger.kernel.org Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190905185318.31363-1-daniel.vetter@ffwll.ch
uncovered a bit a mess in dp drivers. Most drivers (from a quick look,
all except i915) register all the dp stuff in their init code, which
is too early. With CONFIG_DRM_DP_AUX_CHARDEV this will blow up,
because drm_dp_aux_register tries to add a child to a device in sysfs
(the connector) which doesn't even exist yet.
No one seems to have cared thus far. But with the above change I also
moved the setting of dev->registered after the ->load callback, in an
attempt to keep old drivers from hitting any WARN_ON backtraces. But
that moved radeon.ko from the "working, by accident" to "now also
broken" category.
Since this is a huge mess I figured a revert would be simplest. But
this check has already caught issues in i915:
drm/i915: Do not create a new max_bpc prop for MST connectors
Hence I'd like to retain it. Fix the radeon regression by moving the
setting of dev->registered back to were it was, and stop the
backtraces with an explicit check for dev->driver->load.
Everyone else will stay as broken with CONFIG_DRM_DP_AUX_CHARDEV. The
next patch will improve the kerneldoc and add a todo entry for this.
Fixes: 4f5368b5541a ("drm/kms: Catch mode_object lifetime errors") Cc: Sean Paul <sean@poorly.run> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Reported-by: Michel Dänzer <michel@daenzer.net> Reviewed-by: Michel Dänzer <mdaenzer@redhat.com> Tested-by: Michel Dänzer <mdaenzer@redhat.com> Cc: Michel Dänzer <michel@daenzer.net> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190917120936.7501-1-daniel.vetter@ffwll.ch
There are at least two DSI controller drivers which relies on the old
behaviour of adv7511 driver. To avoid platform breakage this patch
should be reverted. This is a temporary solution, as it blocks adv7511
usage with other platforms.
Assumption that DSI device driver (bridge/panel) should first expose
drm_bridge/drm_panel object, then look for DSI bus is just incorrect -
it can work with devices controlled via i2c but it cannot work with
devices controlled via DSI - they will not be able to probe.
To solve the issue following steps should be performed:
- rework reverted patch allowing co-operation with broken DSI controller
drivers - with simple/ugly workaround,
- fix controller drivers and then remove workaround.
There were bugs in the DSI transfer (read and write) function
as it was only tested with displays ever needing a single byte
to be written. Fixed it up and tested so we can now write
messages of up to 16 bytes and read up to 4 bytes from the
display.
Tested with a Sony ACX424AKP display: this display now self-
identifies and can control backlight in command mode.
With the introduction of the as_lock to serialize address space registers,
the hwaccess_lock is only used within the job code and is not protecting
anything. panfrost_job_hw_submit() only accesses registers for 1 job slot
and it's already serialized by drm_sched.
Fixes: 7282f7645d06 ("drm/panfrost: Implement per FD address spaces") Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Signed-off-by: Rob Herring <robh@kernel.org> Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Reviewed-by: Steven Price <steven.price@arm.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190826223317.28509-9-robh@kernel.org
Rob Herring [Mon, 26 Aug 2019 22:33:16 +0000 (17:33 -0500)]
drm/panfrost: Flush and disable address space when freeing page tables
Currently, page tables are freed without disabling the address space first.
This probably is fine as we'll switch to new page tables when the address
space is allocated again and runtime PM suspend will reset the GPU
clearing the registers. However, it's better to clean up after ourselves.
There is also a problem that we could be accessing the h/w in
tlb_inv_context() when suspended.
Rework the disable code to make sure we flush caches/TLBs and disable the
address space before freeing the page tables if we are not suspended. As
the tlb_inv_context() hook is only called when freeing the page tables and
we do a flush before disabling the AS, lets remove the flush from
tlb_inv_context and avoid any runtime PM issues.
Fixes: 7282f7645d06 ("drm/panfrost: Implement per FD address spaces") Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Signed-off-by: Rob Herring <robh@kernel.org> Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Reviewed-by: Robin Murphy <robin.murphy@arm.com> Reviewed-by: Steven Price <steven.price@arm.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190826223317.28509-8-robh@kernel.org
Rob Herring [Mon, 26 Aug 2019 22:33:13 +0000 (17:33 -0500)]
drm/panfrost: Rework page table flushing and runtime PM interaction
There is no point in resuming the h/w just to do flush operations and
doing so takes several locks which cause lockdep issues with the shrinker.
Rework the flush operations to only happen when the h/w is already awake.
This avoids taking any locks associated with resuming which trigger
lockdep warnings.
Fixes: 013b65101315 ("drm/panfrost: Add madvise and shrinker support") Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Signed-off-by: Rob Herring <robh@kernel.org> Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Reviewed-by: Robin Murphy <robin.murphy@arm.com> Reviewed-by: Steven Price <steven.price@arm.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190826223317.28509-5-robh@kernel.org
Rob Herring [Mon, 26 Aug 2019 22:33:11 +0000 (17:33 -0500)]
drm/panfrost: Hold runtime PM reference until jobs complete
Doing a pm_runtime_put as soon as a job is submitted is wrong as it should
not happen until the job completes. It works currently because we are
relying on the autosuspend timeout to keep the h/w enabled.
Fixes: f3ba91228e8e ("drm/panfrost: Add initial panfrost driver") Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: Steven Price <steven.price@arm.com> Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Signed-off-by: Rob Herring <robh@kernel.org> Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Reviewed-by: Steven Price <steven.price@arm.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190826223317.28509-3-robh@kernel.org
Rob Herring [Mon, 26 Aug 2019 22:33:10 +0000 (17:33 -0500)]
drm/panfrost: Rework runtime PM initialization
There's a few issues with the runtime PM initialization.
The documentation states pm_runtime_set_active() should be called before
pm_runtime_enable(). The pm_runtime_put_autosuspend() could suspend the GPU
before panfrost_perfcnt_init() is called which touches the h/w. The
autosuspend delay keeps things from breaking. There's no need explicitly
power off the GPU only to wake back up with pm_runtime_get_sync(). Just
delaying pm_runtime_enable to the end of probe is sufficient.
Lets move all the runtime PM calls into the probe() function so they are
all in one place and are done after all initialization.
Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: Steven Price <steven.price@arm.com> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Signed-off-by: Rob Herring <robh@kernel.org> Reviewed-by: Robin Murphy <robin.murphy@arm.com> Reviewed-by: Steven Price <steven.price@arm.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190826223317.28509-2-robh@kernel.org
Rob Herring [Fri, 23 Aug 2019 02:12:14 +0000 (21:12 -0500)]
drm/panfrost: Use mutex_trylock in panfrost_gem_purge
Lockdep reports a circular locking dependency with pages_lock taken in
the shrinker callback. The deadlock can't actually happen with current
users at least as a BO will never be purgeable when pages_lock is held.
To be safe, let's use mutex_trylock() instead and bail if a BO is locked
already.
Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Signed-off-by: Rob Herring <robh@kernel.org> Reviewed-by: Steven Price <steven.price@arm.com> Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190823021216.5862-7-robh@kernel.org
Rob Herring [Fri, 23 Aug 2019 02:12:13 +0000 (21:12 -0500)]
drm/shmem: Use mutex_trylock in drm_gem_shmem_purge
Lockdep reports a circular locking dependency with pages_lock taken in
the shrinker callback. The deadlock can't actually happen with current
users at least as a BO will never be purgeable when pages_lock is held.
To be safe, let's use mutex_trylock() instead and bail if a BO is locked
already.
WARNING: possible circular locking dependency detected
5.3.0-rc1+ #100 Tainted: G L
------------------------------------------------------
kswapd0/171 is trying to acquire lock: 000000009b9823fd (&shmem->pages_lock){+.+.}, at: drm_gem_shmem_purge+0x20/0x40
but task is already holding lock: 00000000f82369b6 (fs_reclaim){+.+.}, at: __fs_reclaim_acquire+0x0/0x40
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
Fixes: 17acb9f35ed7 ("drm/shmem: Add madvise state and purge helpers") Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Maxime Ripard <maxime.ripard@bootlin.com> Cc: Sean Paul <sean@poorly.run> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Signed-off-by: Rob Herring <robh@kernel.org> Reviewed-by: Steven Price <steven.price@arm.com> Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190823021216.5862-6-robh@kernel.org
Rob Herring [Fri, 23 Aug 2019 02:12:12 +0000 (21:12 -0500)]
drm/shmem: Do dma_unmap_sg before purging pages
Calling dma_unmap_sg() in drm_gem_shmem_free_object() is too late if the
backing pages have already been released by the shrinker. The result is
the following abort:
Fixes: 17acb9f35ed7 ("drm/shmem: Add madvise state and purge helpers") Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Maxime Ripard <maxime.ripard@bootlin.com> Cc: Sean Paul <sean@poorly.run> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Reviewed-by: Steven Price <steven.price@arm.com> Signed-off-by: Rob Herring <robh@kernel.org> Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190823021216.5862-5-robh@kernel.org
Rob Herring [Fri, 23 Aug 2019 02:12:09 +0000 (21:12 -0500)]
drm/panfrost: Fix possible suspend in panfrost_remove
Calls to panfrost_device_fini() access the h/w, but we already done a
pm_runtime_put_sync_autosuspend() beforehand. This only works if the
autosuspend delay is long enough. A 0ms delay will hang the system when
removing the device. Fix this by moving the pm_runtime_put_sync_suspend()
after the panfrost_device_fini() call.
Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Signed-off-by: Rob Herring <robh@kernel.org> Reviewed-by: Steven Price <steven.price@arm.com> Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190823021216.5862-2-robh@kernel.org
Dave Airlie [Tue, 27 Aug 2019 07:22:15 +0000 (17:22 +1000)]
Merge tag 'drm-next-5.4-2019-08-23' of git://people.freedesktop.org/~agd5f/linux into drm-next
drm-next-5.4-2019-08-23:
amdgpu:
- Enable power features on Navi12
- Enable power features on Arcturus
- RAS updates
- Initial Renoir APU support
- Enable power featyres on Renoir
- DC gamma fixes
- DCN2 fixes
- GPU reset support for Picasso
- Misc cleanups and fixes
Dave Airlie [Tue, 27 Aug 2019 07:10:30 +0000 (17:10 +1000)]
Merge tag 'drm-misc-next-2019-08-23' of git://anongit.freedesktop.org/drm/drm-misc into drm-next
drm-misc-next for 5.4:
UAPI Changes:
Cross-subsystem Changes:
Core Changes:
- dma-buf: dma-fence selftests
Driver Changes:
- kirin: Various cleanups and reworks
- komeda: Add support for DT memory-regions
- meson: Rely on the compatible to detect vpu features
- omap: Implement alpha and pixel blend mode properties
- panfrost: Implement per-fd address spaces, various fixes
- rockchip: DSI DT binding rework
- fbdev: Various cleanups
Dave Airlie [Tue, 27 Aug 2019 06:52:06 +0000 (16:52 +1000)]
Merge tag 'imx-drm-next-2019-08-23' of git://git.pengutronix.de/pza/linux into drm-next
drm/imx: IPUv3 image converter fixes and improvements
Fix image converter seam handling for 1024x1024 pixel hardware
limitation at the main processing section input, improve error
handling, and slightly optimize for 1:1 conversions.
Add support for newly defined 32-bit RGB V4L2 pixel formats.
Dave Airlie [Tue, 27 Aug 2019 06:36:41 +0000 (16:36 +1000)]
Merge tag 'drm-intel-next-2019-08-22' of git://anongit.freedesktop.org/drm/drm-intel into drm-next
- More TGL enabling work (Michel, Jose, Lucas)
- Fixes on DP MST (Ville)
- More GTT and Execlists fixes and improvements (Chris)
- Code style clean-up on hdmi and dp side (Jani)
- Fix null pointer dereferrence (Xiong)
- Fix a couple of missing serialization on selftests (Chris)
- More vm locking rework (Chris)
drm-intel-next-2019-08-20:
- GuC and HuC related fixes and improvements (Daniele, Michal)
- Improve debug with more engine information and rework on debugfs files (Chris, Stuart)
- Simplify appearture address handling (Chris)
- Other fixes and cleanups around engines and execlists (Chris)
- Selftests fixes (Matt, Chris)
- Gen11 cache flush related fixes and improvements (Mika)
- More work around requests, timelines and locks to allow removal of struct_mutex (Chris)
- Add missing CML PCI ID (Anusha)
- More work on the new i915 buddy allocator (Matt)
- More headers, files and directories reorg (Daniele)
- Improvements on ggtt’s get pdp (Mika)
- Fix GPU reset (Chris)
- Fix GPIO pins on gen11 (Matt)
- Fix HW readout for crtc_clock in HDMI mode (Imre)
- Sanitize display Phy during unitit to workaround messages of HW state change during suspend (Imre)
- Be defensive when starting vma activity (Chris)
- More Tiger Lake enabling work (Michel, Daniele, Lucas)
- Relax pd_used assertion (Chris)
drm-intel-next-2019-08-13:
- More Tiger Lake enabling work (Lucas, Jose, Tomasz, Michel, Jordan, Anusha, Vandita)
- More selftest organization reworks, fixes and improvements (Lucas, Chris)
- Simplifications on GEM code like context and cleanup_early (Chris, Daniele)
- GuC and HuC related fixes and improvements (Daniele, Michal, Chris)
- Some clean up and fixes on headers, Makefile, and generated files (Lucas, Jani)
- MOCS setup clean up (Tvrtko)
- More Elkhartlake enabling work (Jose, Matt)
- Fix engine reset by clearing in flight execlists requests (Chris)
- Fix possible memory leak on intel_hdcp_auth_downstream (Wei)
- Introduce intel_gt_runtime_suspend/resume (Daniele)
- PMU improvements (Tvrtko)
- Flush extra hard after writing relocations through the GTT (Chris)
- Documentations fixes (Michal, Chris)
- Report dma_reserv allocation failure (Chris)
- Improvements around shrinker (Chris)
- More improvements around engine handling (Chris)
- Also more s/dev_priv/i915 (Chris)
- Abstract display suspend/resume operations (Rodrigo/Jani)
- Drop VM_IO from GTT mappings (Chris)
- Fix some NULL vs IS_ERR conditions (Dan)
- General improvements on error state (Chris)
- Isolate i915_getparam_iocrtl to its own file (Chris)
- Perf OA object refactor (Umesh)
- Ignore central i915->kernel_context and allocate it directly (Chris)
- More fixes and improvements around wakerefs (Chris)
- Clean-up and improvements around debugfs (Chris)
- Free the imported shmemfs file for phys objects (Chris)
- Many other fix and cleanups around engines and execlists (Chris)
- Split out uncore_mmio_debug (Daniele)
- Memory management fixes for blk and gtt (Matt)
- Introduction of buddy allocator to handle huge-pages for GTT (Matt)
- Fix ICL and TGL PG3 power domains (Anshuman)
- Extract GT IRQ to gt/ (Andi)
- Drop last_fence tracking in favor of whole vma->active (Chris)
- Make overlay to use i915_active instead of i915_active_request (Chris)
- Move misc display IRQ handling to its own function (Jose)
- Introduce new _TRANS2() macro in preparation for some coming PSR related work (Jose)
Matthew Ruffell [Thu, 15 Aug 2019 04:26:40 +0000 (16:26 +1200)]
drm/hisilicon/hibmc: Make CONFIG_DRM_HISI_HIBMC depend on ARM64
Hisilicon developed hibmc_drm for their arm64 based soc and did not
intend for this driver to be used on any other architecture than arm64.
Using it on amd64 leads to incorrect video modes being used, making
the screen unreadable, forcing users to manually blacklist the module
on the kernel command line to use the d-i server installer or any
graphical sessions.
Make CONFIG_DRM_HISI_HIBMC firmly depend on ARM64 to ensure it is not
built for other architectures.
Signed-off-by: Matthew Ruffell <matthew.ruffell@canonical.com> Signed-off-by: Xinliang Liu <z.liuxinliang@hisilicon.com>
hersen wu [Fri, 2 Aug 2019 20:01:37 +0000 (16:01 -0400)]
drm/amd/display: flicking observed while installing driver on Navi10 CF
[WHY] value of dchub_ref_clock is decided by dchubbub global timer
settings which is programmed by vbios command table disp_init.
for multi-GPU case, vbios is posted only for primary GPU. without
vbios posted for the secondary GPU, value of dchub_ref_clock is not
set properly. this value will affect dcn bandwidth calcuation and
cause underflow. user will see screen flicking during driver
installation for dual GPU case.
[HOW] dc init_hw always call vbios command table disp_init to
make sure dchubbub global timer is configured and enable.
Signed-off-by: hersen wu <hersenxs.wu@amd.com> Reviewed-by: Jun Lei <Jun.Lei@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Joshua Aberback [Thu, 8 Aug 2019 17:22:36 +0000 (13:22 -0400)]
drm/amd/display: Properly read LVTMA_PWRSEQ_CNTL
[Why]
The register LVTMA_PWRSEQ_CNTL is used to determine the power state of the
embedded display. Currently we do not actually read this register's values,
so during power down we think that this display is already off, so we skip
calling into VBIOS to actually turn it off.
[How]
- add relevant fields to shift / mask initialization
Signed-off-by: Joshua Aberback <joshua.aberback@amd.com> Reviewed-by: Charlene Liu <Charlene.Liu@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Jun Lei [Tue, 23 Jul 2019 20:56:03 +0000 (16:56 -0400)]
drm/amd/display: revert wait in pipelock
[why]
Previous workaround to prevent a vsync flip to be converted
to immediate flip is no longer needed, and is risky because
there are cases where it can result in infinite loop.
[how]
Remove wait loop (which is potentially infinite) before locking
pipe
Signed-off-by: Jun Lei <Jun.Lei@amd.com> Reviewed-by: Charlene Liu <Charlene.Liu@amd.com> Reviewed-by: Eric Yang <eric.yang2@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Bayan Zabihiyan [Thu, 8 Aug 2019 15:08:52 +0000 (11:08 -0400)]
drm/amd/display: Expose OTG_V_TOTAL_MID for HW Diags
[Why]
Existing HW Features, HW Diags test requested that the
registers be exposed.
[How]
Add V_TOTAL_MID to existing DC structures.
Make sure values are passed down throughout DC
Add Register definition.
Program the additional registers
Add additional Logic for V_TOTAL_CONTROL.
Signed-off-by: Bayan Zabihiyan <bayan.zabihiyan@amd.com> Reviewed-by: Charlene Liu <Charlene.Liu@amd.com> Acked-by: Anthony Koo <Anthony.Koo@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Jun Lei [Wed, 7 Aug 2019 20:24:46 +0000 (16:24 -0400)]
drm/amd/display: fix DML not calculating delivery time
[why]
Calculating DCFCLK DS time requires calculating
delivery time for luma/chroma, but this value is
not calculated in DMLv2, it was inadvertently
removed when porting DMLv2
[how]
Add the calculation back
Signed-off-by: Jun Lei <Jun.Lei@amd.com> Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
David Francis [Thu, 25 Jul 2019 19:22:16 +0000 (15:22 -0400)]
drm/amd/display: MST topology debugfs
DRM provides drm_dp_mst_dump_topology, which prints
useful information about MST devices
Hook this up to a debugfs file named amdgpu_mst_topology
Signed-off-by: David Francis <David.Francis@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Charlene Liu [Wed, 7 Aug 2019 21:25:49 +0000 (17:25 -0400)]
drm/amd/display: set av_mute in hw_init for HDMI
[Description]
OS will reserve HW state in UEFI mode.
Driver init_hw reset to RGB which caused HDMI green in YCbCr mode.
read HW blank_color based on acc_mode.
Signed-off-by: Charlene Liu <charlene.liu@amd.com> Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Jaehyun Chung [Wed, 7 Aug 2019 15:20:16 +0000 (11:20 -0400)]
drm/amd/display: Enable HW rotation
[Why] HW rotation is not enabled. Calculations for cursor rotation
are wrong for the values passed to set_cursor_position.
[How] Swap Src rect and height and vertically mirror surface for
the correct surface rotation direction. Cursor position is rotated
according to angle. Offset calculations are tweaked for non-rotated
cursor hotspot and width/height.
Signed-off-by: Jaehyun Chung <jaehyun.chung@amd.com> Reviewed-by: Jun Lei <Jun.Lei@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Zi Yu Liao [Tue, 6 Aug 2019 15:58:09 +0000 (11:58 -0400)]
drm/amd/display: fix stuck test pattern on right half of display
[why]
With visual confirm enabled, displays where ODM combine is enabled
has a test pattern stuck on the right half of the display even
though the display is unblanked.
[how]
Add a condition to not show the colour ramp test pattern when the
display is unblanked.
Signed-off-by: Zi Yu Liao <ziyu.liao@amd.com> Reviewed-by: Eric Yang <eric.yang2@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Charlene Liu [Fri, 2 Aug 2019 18:49:58 +0000 (14:49 -0400)]
drm/amd/display: support spdif
[Description]
port spdif fix to staging:
spdif hardwired to afmt inst 1.
spdif func pointer
spdif resource allocation (reserve last audio endpoint for spdif only)
Signed-off-by: Charlene Liu <charlene.liu@amd.com> Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Ahmad Othman [Thu, 1 Aug 2019 19:05:27 +0000 (15:05 -0400)]
drm/amd/display: Refactoring VTEM
[Why]
Video Timing Extended Metadata packet (VTEM) is not
specific to freesync. So move it out of freesync module
[How]
- Moved VTEM from freesync module to info_packet module
- Created new structure for VTEM parameters that can be used for VRR
and FVA
Signed-off-by: Ahmad Othman <ahmad.othman@amd.com> Reviewed-by: Chris Park <Chris.Park@amd.com> Acked-by: Ahmad Othman <Ahmad.Othman@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Martin Leung [Fri, 2 Aug 2019 15:01:05 +0000 (11:01 -0400)]
drm/amd/display: cleaned up coding error in init_hw
[why]
during a refactor a redundant code that has unknown behaviour was added.
[how]
removed old bad code
Fixes: 8a31820b1218 ("drm/amd/display: Make init_hw and init_pipes generic for seamless boot") Signed-off-by: Martin Leung <martin.leung@amd.com> Reviewed-by: Anthony Koo <Anthony.Koo@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Nikola Cornij [Thu, 1 Aug 2019 19:52:58 +0000 (15:52 -0400)]
drm/amd/display: Zero-out dsc init regs
[why]
Before a statically allocated PPS data structure, that did
get zeroed-out at startup, had been re-used for making packed PPS
SDP. With S3 fix, using a non-initialized PPS data structure was
introduced, while wrongly assuming it'd get initialized before it's
populated. As a consequence 'vbr_enable' and perhaps some other
fields are left uninitialized when making packed PPS SDP. This can
affect 'simple_422' as well because of the way PPS SDP packing is
done (the fields are not masked first, only shifted). The behavior
will be different, depending on the content of uninitialized data.
[how]
Zero-out PPS data structure at initialization time before it's
populated
Fixes: 1a9e3d4569fc ("drm/amd/display: Set DSC before DIG front-end is connected to its back-end") Signed-off-by: Nikola Cornij <nikola.cornij@amd.com> Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Joseph Gravenor [Tue, 30 Jul 2019 20:37:35 +0000 (16:37 -0400)]
drm/amd/display: Implement voltage limitation stub
add new function to get the voltage at the end of
dcn_validate_bandwidth, to check against the
highest voltage we allow.
Created a stub to allow for optimizations
Signed-off-by: Joseph Gravenor <joseph.gravenor@amd.com> Reviewed-by: Eric Yang <eric.yang2@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Acked-by: Sun peng Li <Sunpeng.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Josip Pavic [Tue, 30 Jul 2019 20:56:14 +0000 (16:56 -0400)]
drm/amd/display: load iram for abm 2.3
[Why]
ABM 2.3 firmware expects information in iRAM that differs from previous
versions of ABM, so a mechanism is required to provide it with that
information.
[How]
Extend the existing iRAM definition to include parameters added by
ABM 2.3, and load it if DMCU is running ABM 2.3.
Signed-off-by: Josip Pavic <Josip.Pavic@amd.com> Reviewed-by: Anthony Koo <Anthony.Koo@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: fix MPO HUBP underflow with Scatter Gather
[why]
With Scatter Gather enabled, HUBP underflows during MPO enabled video
playback. hubp_init has a register write that fixes this problem, but
the register is cleared when HUBP gets power gated.
[how]
Make a call to hubp_init during enable_plane, so that the fix can
be applied after HUBP powers back on again.
Signed-off-by: Zi Yu Liao <ziyu.liao@amd.com> Reviewed-by: Tony Cheng <Tony.Cheng@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: fix audio endpoint not getting disabled issue
[Why]
Disable_audio_stream gets enum option as a paramenter which will decide
if we free acquired resources or not. However checks for the option is
guarded by the other condition which check if audio stream is getting
diabled more than once. With both conditions combined, if we attempt to
disable audio stream twice in a row, first with keep and second with
free as an option, we will never free any resources, which will make
system think there is audio endpoint connected even after we plug out
the device
[How]
Get rid of option as parameter to disable_audio_stream and move the part
of the code that free acquired resources to outside where to keep or to
free resources is actually determined
Signed-off-by: Su Sung Chung <Su.Chung@amd.com> Reviewed-by: Jun Lei <Jun.Lei@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: Delete dead code in command_table_helper
[Why]
dig_encoder_sel_to_atom will always return zero on any ASIC version
past DCE80 since programming of the FE selection is handled by
driver, but the translation code was left in the function, making
it look like a coding error.
[How]
Remove code that has no effect, and replace with a comment describing
why it returns zero.
Signed-off-by: Julian Parkin <julian.parkin@amd.com> Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: re structure odm to allow 4 to 1 support
Currently odm is handled using top_bottom pipe by special casing
the differing opps to differentiate from mpc combine.
Since top/bottom pipe list was made to track mpc muxing this creates
difficulties in adding a 4 pipe odm case support.
Rather than continue using mpc combine list, this change reworks odm
to use it's own linked list to keep track of odm combine pipes. This
also opens up options for using mpo with odm, if a practical use case
is ever found.
Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com> Reviewed-by: Charlene Liu <Charlene.Liu@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Deucher [Fri, 23 Aug 2019 14:42:33 +0000 (09:42 -0500)]
drm/amdgpu: set adev->num_vmhubs for gmc6,7,8
So that we properly handle them on older asics.
Fixes: 3ff985485b29 ("drm/amdgpu: Export function to flush TLB of specific vm hub") Tested-by: Tom St Denis <tom.stdenis@amd.com> Reviewed-by: Xiaojie Yuan <xiaojie.yuan@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Deucher [Thu, 22 Aug 2019 19:17:57 +0000 (14:17 -0500)]
drm/amdgpu/display: add flag for multi-display mclk switching
Add a dcfeaturemask flag for mclk switching. Disable by default;
enable once the feature has seen more testing.
Set amdgpu.dcfeaturemask=2 on the kernel command line in grub
to enable this.
Acked-by: Evan Quan <evan.quan@amd.com> Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Deucher [Fri, 23 Aug 2019 16:32:37 +0000 (11:32 -0500)]
drm/amd/display: update bw_calcs to take pipe sync into account (v3)
Properly set all_displays_in_sync so that when the data is
propagated to powerplay, it's set properly and we can enable
mclk switching when all monitors are in sync.
v2: fix logic, clean up
v3: check for blending chains, simplify logic
Acked-by: Evan Quan <evan.quan@amd.com> Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Kent Russell [Fri, 23 Aug 2019 13:13:18 +0000 (09:13 -0400)]
drm/powerplay: Fix Vega20 power reading again
For the 40.46 SMU release, they changed CurrSocketPower to
AverageSocketPower, but this was changed back in 40.47 so just check if
it's 40.46 and make the appropriate change
Tested with 40.45, 40.46 and 40.47 successfully
Signed-off-by: Kent Russell <kent.russell@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Kevin Wang [Wed, 21 Aug 2019 02:58:19 +0000 (10:58 +0800)]
drm/amd/powerpaly: fix navi series custom peak level value error
fix other navi asic set peak performance level error.
because the navi10_ppt.c will handle navi12 14 asic,
it will use navi10 peak value to set other asic, it is not correct.
after patch:
only navi10 use custom peak value, other asic will used default value.
Signed-off-by: Kevin Wang <kevin1.wang@amd.com> Reviewed-by: Evan Quan <evan.quan@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
omapdrm: no need to check return value of debugfs_create functions
When calling debugfs functions, there is no need to ever check the
return value. The function can work or not, but the code logic should
never do something different based on this.
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Sebastian Reichel <sebastian.reichel@collabora.com> Cc: Jyri Sarha <jsarha@ti.com> Cc: Tony Lindgren <tony@atomide.com> Cc: zhong jiang <zhongjiang@huawei.com> Cc: dri-devel@lists.freedesktop.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190613115749.GC26335@kroah.com Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190704023557.4551-1-huangfq.daxian@gmail.com
Dave Airlie [Fri, 23 Aug 2019 03:23:38 +0000 (13:23 +1000)]
Merge branch 'linux-5.4' of git://github.com/skeggsb/linux into drm-next
This is mostly just the stuff I missed last round. Various cleanup
patches + fixes, improvements to display colour management, and some
code to avoid loading when power cables aren't properly attached.
Mark Menzynski [Fri, 2 Aug 2019 09:21:00 +0000 (11:21 +0200)]
drm/nouveau/volt: Fix for some cards having 0 maximum voltage
Some, mostly Fermi, vbioses appear to have zero max voltage. That causes Nouveau to not parse voltage entries, thus users not being able to set higher clocks.
When changing this value Nvidia driver still appeared to ignore it, and I wasn't able to find out why, thus the code is ignoring the value if it is zero.
CC: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Mark Menzynski <mmenzyns@redhat.com> Reviewed-by: Karol Herbst <kherbst@redhat.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
It's not safe to iterate the entire list of CRTCs in
nv50_disp_atomic_commit(), as we could be doing a non-blocking modeset
on one CRTC in parallel with one or more other CRTCs. Likewise, this
means it's also not safe to do so in order to track runtime PM state.
While this code is certainly wrong, so far the only issues I've seen
this cause in the wild is the occasional PM ref unbalance after an
atomic check failure + module reloading (since the PCI device will
outlive nouveau in such scenarios).
So, do this far more elegantly: grab a runtime PM ref across the modeset
and commit tail, then grab/put references for each CRTC enable/disable.
This also ends up being much simpler then the previous broken solution
we had.
Finally, since we've removed all it's users: get rid of
nouveau_drm->have_disp_power_ref.
Signed-off-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Lyude Paul [Wed, 7 Aug 2019 23:47:05 +0000 (19:47 -0400)]
drm/nouveau/dispnv04: Remove runtime PM
Originally when trying to fix the issue of runtime PM references with
non-blocking CRTCs on nv50, I ended up stumbling on this code when
trying to remove nouveau_drm->have_disp_power_ref, and attempted to fix
it to remove the dependency on have_disp_power_ref. However, Ilia Mirkin
pointed out that this code is actually completely useless, as pre-nv50
never had runtime PM support in the first place! Go figure.
So, since it's useless just get rid of it. Note that since the only
thing nouveau_crtc_set_config() was doing was grabbing a runtime PM ref,
calling drm_crtc_helper_set_config() then dropping the ref; we can just
remove the function entirely and just call drm_crtc_helper_set_config()
directly.
Signed-off-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Mark Menzynski [Thu, 18 Jul 2019 08:07:40 +0000 (10:07 +0200)]
drm/nouveau/gpio: check the gpio function 16 in the power check as well
Added GPIO is "Thermal and External Power Detect". It's uncertain if this
GPIO is set on GPU initialization or only if a change is detected by the
GPU at runtime.
This GPIO can be found in Rankine and Curie and rarely on Tesla GPUs
VBIOS.
Untested, wrote according to documentation.
Signed-off-by: Mark Menzynski <mmenzyns@redhat.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 2 Jul 2019 04:19:12 +0000 (14:19 +1000)]
drm/nouveau/therm: don't attempt fan control where PMU is already managing it
There's already a condition in place which attempts to detect this, but
since we've begun to require a PMU subdev even on boards where we don't
load a custom FW, it's become inaccurate.
This will prevent unnecessarily running a periodic fan update thread on
GP100 and newer, where we don't yet override the default PMU FW.
Ilia Mirkin [Tue, 11 Jun 2019 02:32:45 +0000 (22:32 -0400)]
drm/nouveau/kms/nv50-: enable modern color management properties
For GF119:GV100, we can enable DEGAMMA/CTM/GAMMA. For earlier GPUs, as
there is no CTM, having both degamma and gamma is a bit pointless. Later
GPUs currently lack an implementation.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
This opcode signals that the software reset sequence has completed.
Ordinarily, no actual operations are performed by the opcode.
However it allows for possible software work arounds by devinit
engines in software agents other than the VBIOS, such as the resman,
FCODE, and EFI driver.
Signed-off-by: Rhys Kidd <rhyskidd@gmail.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>