drm_fb_helper_hotplug_event() should tolerate the fb_helper argument being
NULL. Commit 03a9606e7fee ("drm/fb-helper: Avoid race with DRM userspace")
introduced a fb_helper dereference before the NULL check.
Fixup by moving the dereference after the NULL check.
Fixes: 03a9606e7fee ("drm/fb-helper: Avoid race with DRM userspace") Reported-by: kbuild test robot <lkp@intel.com> Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Noralf Trønnes <noralf@tronnes.org> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/msgid/20190515132925.48867-1-noralf@tronnes.org
drm/hisilicon: Convert hibmc-drm driver to VRAM MM
The data structure |struct drm_vram_mm| and its helpers replace hibmc's
TTM-based memory manager. It's the same implementation; except for the
type names.
v5:
* set .llseek via DRM_VRAM_MM_FILE_OPERATIONS
v4:
* don't select DRM_TTM or DRM_VRAM_MM_HELPER
v3:
* use drm_gem_vram_mm_funcs
* convert driver to drm_device-based instance
v2:
* implement hibmc_mmap() with drm_vram_mm_mmap()
drm/vboxvideo: Convert vboxvideo driver to VRAM MM
The data structure |struct drm_vram_mm| and its helpers replace vboxvideo's
TTM-based memory manager. It's the same implementation; except for the type
names.
v4:
* don't select DRM_TTM or DRM_VRAM_MM_HELPER
v3:
* use drm_gem_vram_mm_funcs
* convert driver to drm_device-based instance
v2:
* implement vbox_mmap() with drm_vram_mm_mmap()
drm/vboxvideo: Convert vboxvideo driver to |struct drm_gem_vram_object|
This patch replaces |struct vbox_bo| and its helpers with the generic
implementation of |struct drm_gem_vram_object|. The only change in
semantics is that &ttm_bo_driver.verify_access() now does the actual
verification.
drm/mgag200: Replace mapping code with drm_gem_vram_{kmap/kunmap}()
The mgag200 driver establishes several memory mappings for frame buffers
and cursors. This patch converts the driver to use the equivalent
drm_gem_vram_kmap() functions. It removes the dependencies on TTM
and cleans up the code.
The data structure |struct drm_vram_mm| and its helpers replace mgag200's
TTM-based memory manager. It's the same implementation; except for the type
names.
v4:
* don't select DRM_TTM or DRM_VRAM_MM_HELPER
v3:
* use drm_gem_vram_mm_funcs
* convert driver to drm_device-based instance
v2:
* implement mgag200_mmap() with drm_vram_mm_mmap()
The data structure |struct drm_vram_mm| and its helpers replace bochs'
TTM-based memory manager. It's the same implementation; except for the
type names.
v5:
* set .llseek via DRM_VRAM_MM_FILE_OPERATIONS
v4:
* don't select DRM_TTM or DRM_VRAM_MM_HELPER
v3:
* use drm_gem_vram_mm_funcs
* convert driver to drm_device-based instance
v2:
* implement bochs_mmap() with drm_vram_mm_mmap()
drm/ast: Replace mapping code with drm_gem_vram_{kmap/kunmap}()
The AST driver establishes several memory mappings for frame buffers
and cursors. This patch converts the driver to use the equivalent
drm_gem_vram_kmap() functions. It removes the dependencies on TTM
and cleans up the code.
The data structure |struct drm_vram_mm| and its helpers replace ast's
TTM-based memory manager. It's the same implementation; except for the
type names.
v4:
* don't select DRM_TTM or DRM_VRAM_MM_HELPER
v3:
* use drm_gem_vram_mm_funcs
* convert driver to drm_device-based instance
v2:
* implement ast_mmap() with drm_vram_mm_mmap()
There's now a pointer to struct drm_vram_mm stored in struct drm_device.
DRM drivers that use VRAM MM should use this field to refer to their
instance of the data structure. Appropriate helpers are now provided as
well.
Adding struct drm_vram_mm to struct drm_device further avoids wrappers
and boilerplate code in drivers. This patch implements default functions
for callbacks in struct drm_driver and struct file_operations that use
the struct drm_vram_mm stored in struct drm_device. Drivers that need to
provide their own implementations can still do so.
The patch also adds documentation for the VRAM helper library in general.
v5:
* set .llseek to no_llseek() from DRM_VRAM_MM_FILE_OPERATIONS
v4:
* cleanups from checkpatch.pl
* document VRAM helper library
drm: Add VRAM MM, a simple memory manager for dedicated VRAM
The VRAM MM memory manager is a helper library that manages dedicated video
memory of simple framebuffer devices. It is supported to be used with
struct drm_gem_vram_object, but does not depend on it.
The implementation is based on the respective code from ast, bochs, and
mgag200. These drivers share the exact same implementation except for type
names. The helpers are currently build with TTM. This may change in future
revisions.
v4:
* cleanups from checkpatch.pl
v2:
* renamed to struct drm_vram_mm
* add drm_vram_mm_mmap() helper
* documentation fixes
These basic helper functions for GEM VRAM allow for pinning and mapping
GEM VRAM objects via the PRIME interfaces. It's not a full implementation,
but complete enough for generic fbcon.
drm: Add drm_gem_vram_fill_create_dumb() to create dumb buffers
The helper function drm_gem_vram_fill_create_dumb() implements most of
struct drm_driver.dumb_create() for GEM-VRAM buffer objects. It's not a
full implementation of the callback, as several driver-specific parameters
are still required.
v4:
* cleanups from checkpatch.pl
v2:
* documentation fixes
The type |struct drm_gem_vram_object| implements a GEM object for simple
framebuffer devices with dedicated video memory. The BO is either located
in VRAM or system memory.
The implementation has been created from the respective code in ast,
bochs and mgag200. These drivers copy their implementation from each
other; except for the names of several data types. The helpers are
currently build with TTM, but this is considered an implementation
detail and may change in future updates.
v5:
* do WARN_ON_ONCE for pin-count mismatches
* allocate only 2 entries in placements array
v4:
* cleanups from checkpatch.pl
* removed several fixed-size types from interfaces
* DRM_VRAM_HELPER now selects DRM_TTM
* remove separate config option for GEM VRAM
v2:
* rename to |struct drm_gem_vram_object|
* move drm_is_gem_ttm() to a later patch in the series
* add drm_gem_vram_kmap_at()
* return is_iomem from kmap functions
* redefine TTM placement flags for public interface
* documentation fixes
Noralf Trønnes [Mon, 6 May 2019 18:01:30 +0000 (20:01 +0200)]
drm/fb-helper: Avoid race with DRM userspace
drm_fb_helper_is_bound() is used to check if DRM userspace is in control.
This is done by looking at the fb on the primary plane. By the time
fb-helper gets around to committing, it's possible that the facts have
changed.
Avoid this race by holding the drm_device->master_mutex lock while
committing. When DRM userspace does its first open, it will now wait
until fb-helper is done. The helper will stay away if there's a master.
Two igt tests fail with the new 'bail out if master' rule. Work around
this by relaxing this rule for drm_fb_helper_restore_fbdev_mode_unlocked()
until the tests have been fixed. Add todo entry for this.
Locking rule: Always take the fb-helper lock first.
v5: drm_fb_helper_restore_fbdev_mode_unlocked(): Use
restore_fbdev_mode_force()
v2:
- Remove drm_fb_helper_is_bound() (Daniel Vetter)
- No need to check fb_helper->dev->master in
drm_fb_helper_single_fb_probe(), restore_fbdev_mode() has the check.
Daniel Vetter [Thu, 2 May 2019 13:56:03 +0000 (15:56 +0200)]
drm: Some ocd in drm_file.c
Move the open helper around to avoid the forward decl, and give
drm_setup a drm_legacy_ prefix since it's all legacy stuff in there.
v2: Move drm_legacy_setup into drm_legacy_misc.c (Chris). The
counterpart in the form of drm_legacy_dev_reinit is there already too,
plus it fits perfectly into Dave's work of making DRIVER_LEGACY code
compile-time optional.
Dave Airlie [Thu, 9 May 2019 01:03:54 +0000 (11:03 +1000)]
Merge tag 'drm-misc-next-fixes-2019-05-08' of git://anongit.freedesktop.org/drm/drm-misc into drm-next
- A handful of fixes from -next that just missed feature freeze
- More panfrost fixes that went directly in -misc-next-fixes (various)
- Fix searchpaths during build (Masahiro)
- msm patch to fix the driver for chips without zap shader (Rob)
- Fix freeing imported buffers in drm_gem_cma_free_object() (Noralf)
Dave Airlie [Thu, 9 May 2019 00:43:44 +0000 (10:43 +1000)]
Merge branch 'for-upstream/mali-dp' of git://linux-arm.org/linux-ld into drm-next
This is the 2nd pull request for the malidp-next. The new patches add
additional support for Arm Mali D71 so that it can now be enabled
correctly and brought up on any SoC that contains the IP. From now on
we will start focusing on adding writeback, scaling and other useful
features to bring the driver to the same level of maturity as mali-dp.
Rob Clark [Wed, 8 May 2019 13:06:52 +0000 (06:06 -0700)]
drm/msm/a6xx: No zap shader is not an error
Depending on platform firmware, a zap shader may not be required to take
the GPU out of secure mode on boot, in which case we can just write
RBBM_SECVID_TRUST_CNTL directly. Which we *mostly* handled, but missed
clearing 'ret' resulting that hw_init() returned an error on these
devices.
The logic for freeing an imported buffer with a virtual address is
broken. It will free the buffer instead of unmapping the dma buf.
Fix by reversing the if ladder and first check if the buffer is imported.
HD702E lcd is FriendlyELEC developed eDP LCD panel with 800x1280
resolution. It has built in Goodix, GT9271 captive touchscreen
with backlight adjustable via PWM.
Add support for it.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com> Reviewed-by: Rob Herring <robh@kernel.org> Cc: Thierry Reding <thierry.reding@gmail.com> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: dri-devel@lists.freedesktop.org Signed-off-by: Sam Ravnborg <sam@ravnborg.org> Link: https://patchwork.freedesktop.org/patch/msgid/20190507130708.11255-2-jagan@amarulasolutions.com
HD702E lcd is FriendlyELEC developed eDP LCD panel with 800x1280
resolution. It has built in Goodix, GT9271 captive touchscreen
with backlight adjustable via PWM.
Add dt-bindings documentation for it.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com> Reviewed-by: Rob Herring <robh@kernel.org> Cc: Thierry Reding <thierry.reding@gmail.com> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: dri-devel@lists.freedesktop.org Signed-off-by: Sam Ravnborg <sam@ravnborg.org> Link: https://patchwork.freedesktop.org/patch/msgid/20190507130708.11255-1-jagan@amarulasolutions.com
Mario Kleiner [Thu, 18 Apr 2019 06:01:57 +0000 (08:01 +0200)]
drm: Fix timestamp docs for variable refresh properties.
As discussed with Nicholas and Daniel Vetter (patchwork
link to discussion below), the VRR timestamping behaviour
produced utterly useless and bogus vblank/pageflip
timestamps. We have found a way to fix this and provide
sane behaviour.
As of Linux 5.2, the amdgpu driver will be able to
provide exactly the same vblank / pageflip timestamp
semantic in variable refresh rate mode as in standard
fixed refresh rate mode. This is achieved by deferring
core vblank handling (drm_crtc_handle_vblank()) until
the end of front porch, and also defer the sending of
pageflip completion events until end of front porch,
when we can safely compute correct pageflip/vblank
timestamps.
The same approach will be possible for other VRR
capable kms drivers, so we can actually have sane
and useful timestamps in VRR mode.
This patch removes the section of the docs that
describes the broken timestamp behaviour present
in Linux 5.0/5.1.
drm/komeda: Fixed warning: Function parameter or member not described
Fixed the warnings: Function parameter or member 'xxx' not described
when make htmldocs
This patch depends on:
- https://patchwork.freedesktop.org/series/54448/
- https://patchwork.freedesktop.org/series/54449/
- https://patchwork.freedesktop.org/series/54450/
v2: Rebase and add reporter
Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com> Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
drm/komeda: Add sysfs attribute: core_id and config_id
Add two sysfs node: core_id, config_id, user can read them to fetch the
HW product information.
Also, use memset to initialize config_id, rather than quirky C syntax.
Courtesy of Nathan Chancellor <natechancellor@gmail.com>.
Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com> Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
[Merged Nathan's patch that uses memset to initialize config_id into
original patch as the fixes tag changed due to rebase, reworded the
commit to reference the merged patch] Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
It was changed to GFP_ATOMIC in commit ec2f0577c (add & use
virtio_gpu_queue_fenced_ctrl_buffer) because the allocation happened
with a spinlock held. That was no longer true after commit 9fdd90c0f (add virtio_gpu_alloc_fence()).
Signed-off-by: Chia-I Wu <olvaffe@gmail.com> Reviewed-by: Robert Foss <robert.foss@collabora.com> Reviewed-by: Emil Velikov <emil.velikov@collabora.com> Link: http://patchwork.freedesktop.org/patch/msgid/20190429221021.159784-1-olvaffe@gmail.com Cc: Gerd Hoffmann <kraxel@redhat.com> Cc: Gustavo Padovan <gustavo.padovan@collabora.com> Cc: Robert Foss <robert.foss@collabora.com> Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
For most drivers, drm_fence_init is followed by drm_fence_emit
immediately. But for our driver, they are done separately. We also
don't know the fence seqno until drm_fence_emit.
and have no such issue. But in our driver, because most ioctls
queue commands into ctrlq, we do not want to grab a lock. Instead,
we set seqno to 0 when a fence is created, and update it when the
command is finally queued and the seqno is known.
Robin Murphy [Fri, 3 May 2019 15:31:45 +0000 (16:31 +0100)]
drm/panfrost: Show stored feature registers
Re-reading the feature registers for the sake of displaying the raw
values seems pointless, and in fact showing the copies that we've
already read and stored is arguably more useful in terms of giving
exposure to any potential bugs in that part of the process.
Robin Murphy [Fri, 3 May 2019 15:31:43 +0000 (16:31 +0100)]
drm/panfrost: Disable PM on probe failure
Make sure to disable runtime PM again if probe fails after we've enabled
it. Otherwise, any subsequent attempt to re-probe starts triggering
"Unbalanced pm_runtime_enable!" assertions from the driver core.
Robin Murphy [Fri, 3 May 2019 15:31:42 +0000 (16:31 +0100)]
drm/panfrost: Set DMA masks earlier
The DMA masks need to be set correctly before any DMA API activity kicks
off, and the current point in panfrost_probe() is way too late in that
regard. since panfrost_mmu_init() has already set up a live address
space and DMA-mapped MMU pagetables. We can't set masks until we've
queried the appropriate value from MMU_FEATURES, but as soon as
reasonably possible after that should suffice.
Lucas Stach [Wed, 19 Dec 2018 14:16:41 +0000 (15:16 +0100)]
drm/etnaviv: initialize idle mask before querying the HW db
If there is a match in the HW DB, the function is left early, before
inititalizing the idle mask. Fix this by doing the init earlier, as
only old GPUs, not present in the HW DB need a different idle mask.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de> Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
drm/amdgpu: Avoid HW reset if guilty job already signaled.
Also reject TDRs if another one already running.
v2:
Stop all schedulers across device and entire XGMI hive before
force signaling HW fences.
Avoid passing job_signaled to helper fnctions to keep all the decision
making about skipping HW reset in one place.
v3:
Fix SW sched. hang after non HW reset. sched.hw_rq_count has to be balanced
against it's decrement in drm_sched_stop in non HW reset case.
v4: rebase
v5: Revert v3 as we do it now in sceduler code.
drm/scheduler: Add flag to hint the release of guilty job.
Problem:
Sched thread's cleanup function races against TO handler
and removes the guilty job from mirror list and we
have no way of differentiating if the job was removed from within the
TO handler or from the sched thread's clean-up function.
Fix:
Add a flag to scheduler to hint the TO handler that the guilty job needs
to be explicitly released.
Christian König [Thu, 18 Apr 2019 15:00:21 +0000 (11:00 -0400)]
drm/scheduler: rework job destruction
We now destroy finished jobs from the worker thread to make sure that
we never destroy a job currently in timeout processing.
By this we avoid holding lock around ring mirror list in drm_sched_stop
which should solve a deadlock reported by a user.
v2: Remove unused variable.
v4: Move guilty job free into sched code.
v5:
Move sched->hw_rq_count to drm_sched_start to account for counter
decrement in drm_sched_stop even when we don't call resubmit jobs
if guily job did signal.
v6: remove unused variable
drm: introduce a capability flag for syncobj timeline support
Unfortunately userspace users of this API cannot be publicly disclosed
yet.
This commit effectively disables timeline syncobj ioctls for all
drivers. Each driver wishing to support this feature will need to
expose DRIVER_SYNCOBJ_TIMELINE.
v2: Add uAPI capability check (Christian)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by: Christian König <christian.koenig@amd.com> (v1) Cc: Dave Airlie <airlied@redhat.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Christian König <christian.koenig@amd.com> Cc: Chunming Zhou <david1.zhou@amd.com> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190416125750.31370-1-lionel.g.landwerlin@intel.com
drm: report consistent errors when checking syncobj capibility
We've been somewhat inconsistent when adding the new ioctl and
returned ENODEV instead of EOPNOTSUPPORTED upon failing the syncobj
capibility.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Fixes: ea569910cbab98 ("drm/syncobj: add transition iotcls between binary and timeline v2") Fixes: 01d6c357837918 ("drm/syncobj: add support for timeline point wait v8") Cc: Dave Airlie <airlied@redhat.com> Cc: Christian König <christian.koenig@amd.com> Cc: Chunming Zhou <david1.zhou@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> for the series. Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190416123048.2913-1-lionel.g.landwerlin@intel.com
Tobias Klausmann [Thu, 28 Mar 2019 13:17:31 +0000 (14:17 +0100)]
drm/nouveau/nouveau: forward error generated while resuming objects tree
On a failed resume we may experience unrecoverable errors. Plumb the error code
through to actually let the driver fail. On a reverse-prime setup this helps the
drm subsystem to at least recover the integrated gpu.
This can especially happen with secboot timing out, leaving the hardware in a
non-functioning state.
Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann@mni.thm.de> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
There is a spelling mistake in a nvkm_debug message. Fix it.
Signed-off-by: Colin Ian King <colin.king@canonical.com> Reviewed-by: Mukesh Ojha <mojha@codeaurora.org> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Lyude Paul [Tue, 9 Apr 2019 20:23:30 +0000 (16:23 -0400)]
drm/nouveau/i2c: Disable i2c bus access after ->fini()
For a while, we've had the problem of i2c bus access not grabbing
a runtime PM ref when it's being used in userspace by i2c-dev, resulting
in nouveau spamming the kernel log with errors if anything attempts to
access the i2c bus while the GPU is in runtime suspend. An example:
[ 130.078386] nouveau 0000:01:00.0: i2c: aux 000d: begin idle timeout ffffffff
Since the GPU is in runtime suspend, the MMIO region that the i2c bus is
on isn't accessible. On x86, the standard behavior for accessing an
unavailable MMIO region is to just return ~0.
Except, that turned out to be a lie. While computers with a clean
concious will return ~0 in this scenario, some machines will actually
completely hang a CPU on certian bad MMIO accesses. This was witnessed
with someone's Lenovo ThinkPad P50, where sensors-detect attempting to
access the i2c bus while the GPU was suspended would result in a CPU
hang:
Yikes! While I wanted to try to make it so that accessing an i2c bus on
nouveau would wake up the GPU as needed, airlied pointed out that pretty
much any usecase for userspace accessing an i2c bus on a GPU (mainly for
the DDC brightness control that some displays have) is going to only be
useful while there's at least one display enabled on the GPU anyway, and
the GPU never sleeps while there's displays running.
Since teaching the i2c bus to wake up the GPU on userspace accesses is a
good deal more difficult than it might seem, mostly due to the fact that
we have to use the i2c bus during runtime resume of the GPU, we instead
opt for the easiest solution: don't let userspace access i2c busses on
the GPU at all while it's in runtime suspend.
Changes since v1:
* Also disable i2c busses that run over DP AUX
Signed-off-by: Lyude Paul <lyude@redhat.com> Cc: stable@vger.kernel.org Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Commit 3a6536c51d5d ("drm/nouveau: Intercept ACPI_VIDEO_NOTIFY_PROBE")
added a definition of ACPI_VIDEO_NOTIFY_PROBE because <acpi/video.h> didn't
supply one. Later, commit eff4a751cce5 ("ACPI / video: Move
ACPI_VIDEO_NOTIFY_* defines to acpi/video.h") moved ACPI_VIDEO_NOTIFY_PROBE
and other definitions to <acpi/video.h>, so the copy in nouveau_display.c
is now unnecessary.
Remove the unnecessary definition from nouveau_display.c.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> CC: Hans de Goede <hdegoede@redhat.com> Acked-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Jon Derrick [Sat, 16 Mar 2019 00:05:18 +0000 (18:05 -0600)]
drm/nouveau/mmu: qualify vmm during dtor
If the BAR initialization failed it may leave the vmm structure in an
unitialized state, leading to a null-pointer-dereference when the vmm is
dereferenced during teardown.
Signed-off-by: Jon Derrick <jonathan.derrick@intel.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Jon Derrick [Sat, 16 Mar 2019 00:05:17 +0000 (18:05 -0600)]
drm/nouveau/bar/gf100: ensure BAR is mapped
If the BAR is zero size, it indicates it was never successfully mapped.
Ensure that the BAR is valid during initialization before attempting to
use it.
Signed-off-by: Jon Derrick <jonathan.derrick@intel.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Jon Derrick [Sat, 16 Mar 2019 00:05:16 +0000 (18:05 -0600)]
drm/nouveau/bar/nv50: ensure BAR is mapped
If the BAR is zero size, it indicates it was never successfully mapped.
Ensure that the BAR is valid during initialization before attempting to
use it.
Signed-off-by: Jon Derrick <jonathan.derrick@intel.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
The 50 ms default timeout waiting for vblanks is too small
for the first vblank from the ST-Ericsson MCDE display
controller over DSI. Presumably this is because the DSI
display is command-mode only and the state machine will
take some time setting up its state for the first display
update, and we hit a timeout. 100 ms makes it pass without
problems.
Evan Quan [Thu, 25 Apr 2019 07:38:25 +0000 (15:38 +0800)]
drm/amdgpu: power down the Vega20 VCE engine on request
Power down the engine also along with disabling its DPM
functionality.
Signed-off-by: Evan Quan <evan.quan@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Evan Quan [Thu, 25 Apr 2019 07:35:17 +0000 (15:35 +0800)]
drm/amdgpu: expose VCE 4.0 powergate interface
SMU will use this interface to power down the VCE engine.
Signed-off-by: Evan Quan <evan.quan@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Mario Kleiner [Fri, 26 Apr 2019 21:40:16 +0000 (23:40 +0200)]
drm/amd/display: Compensate for pre-DCE12 BTR-VRR hw limitations. (v3)
Pre-DCE12 needs special treatment for BTR / low framerate
compensation for more stable behaviour:
According to comments in the code and some testing on DCE-8
and DCE-11, DCE-11 and earlier only apply VTOTAL_MIN/MAX
programming with a lag of one frame, so the special BTR hw
programming for intermediate fixed duration frames must be
done inside the current frame at flip submission in atomic
commit tail, ie. one vblank earlier, and the fixed refresh
intermediate frame mode must be also terminated one vblank
earlier on pre-DCE12 display engines.
To achieve proper termination on < DCE-12 shift the point
when the switch-back from fixed vblank duration to variable
vblank duration happens from the start of VBLANK (vblank irq,
as done on DCE-12+) to back-porch or end of VBLANK (handled
by vupdate irq handler). We must leave the switch-back code
inside VBLANK irq for DCE12+, as before.
Doing this, we get much better behaviour of BTR for up-sweeps,
ie. going from short to long frame durations (~high to low fps)
and for constant framerate flips, as tested on DCE-8 and
DCE-11. Behaviour is still not quite as good as on DCN-1
though.
On down-sweeps, going from long to short frame durations
(low fps to high fps) < DCE-12 is a little bit improved,
although by far not as much as for up-sweeps and constant
fps.
v2: Fix some wrong locking, as pointed out by Nicholas.
v3: Simplify if-condition in vupdate-irq - nit by Nicholas. Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com> Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Mario Kleiner [Fri, 26 Apr 2019 21:40:14 +0000 (23:40 +0200)]
drm/amd/display: Fix and simplify apply_below_the_range()
The comparison of inserted_frame_duration_in_us against a
duration calculated from max_refresh_in_uhz is both wrong
in its math and not needed, as the min_duration_in_us value
is already cached in in_out_vrr for reuse. No need to
recalculate it wrongly at each invocation.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com> Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: Expose DRM_FORMAT_RGB565 on overlay planes
RGB565 support isn't restricted to just the primary plane in DC, so
also expose support for it on overlays.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: David Francis <david.francis@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: Do VRR transition before enable_crc_interrupts
[Why]
Originally we did the amdgpu_dm_handle_vrr_transition call before
interrupts were enabled. After the interrupt toggling logic was
moved around for support enabling CRTCs with no primary planes
active this was no longer being called in the case where there
wasn't a modeset.
This fixes failures in igt@kms_vrr@* with error
"Timed out: Waiting for vblank event".
[How]
Shift them back into the loop that always ran before interrupts were
enabled.
Pull out the logic that updated VRR state into the same loop since
there's no reason these need to be split.
In the case where we're going from VRR off, no planes to VRR on, some
active planes we'll still be covered for having the VRR vupdate
handler enabled - vblank will be re-enabled at this point, it will
see that VRR is active and set the vupdate interrupt on there.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Reviewed-by: David Francis <David.Francis@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
John Barberiz [Wed, 3 Apr 2019 23:22:55 +0000 (19:22 -0400)]
drm/amd/display: Refactor dp vendor parsing logic to a function
Refactor dp vendor parsing int to a new function, and call it before
get_active_converter_info().
Also, add a flag to skip parsing of Display ID 2.0. Some devices fail on
readind DID2, but we shouldn't fail EDID read because of it. Add this
flag to facilitate the logic.
Signed-off-by: John Barberiz <John.Barberiz@amd.com> Reviewed-by: Aric Cyr <Aric.Cyr@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Charlene Liu [Thu, 11 Apr 2019 01:43:23 +0000 (21:43 -0400)]
drm/amd/display: Add hubp_init entry to hubp vtable
Different HW will need to init HUBP differently. For now, add a vtable
entry, and hook a NO-OP for DCN1.
In addition, future HW will need to access the HUBPREQ_DEBUG register
for hubp_init. Add it to the reg list.
Signed-off-by: Charlene Liu <charlene.liu@amd.com> Reviewed-by: Jun Lei <Jun.Lei@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
The new interface now replaces the old interface for all known
configurations.
Signed-off-by: Eric Yang <Eric.Yang2@amd.com> Reviewed-by: Tony Cheng <Tony.Cheng@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Thomas Lim [Wed, 10 Apr 2019 21:06:07 +0000 (17:06 -0400)]
drm/amd/display: Add power down display on boot flag
[Why]
Due to the generic introduction of seamless boot, the display is no
longer blanked upon boot. However, this causes corruption on some
systems that does not lock the memory in the non-secure boot case,
resulting in brief corruption on boot due to garbage being written into
the frame buffer.
[How]
Add a flag, read during DC init, to determine whether display should be
blanked on boot. Default to true.
Signed-off-by: Thomas Lim <Thomas.Lim@amd.com> Reviewed-by: Aric Cyr <Aric.Cyr@amd.com> Acked-by: Anthony Koo <Anthony.Koo@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Yongqiang Sun [Fri, 29 Mar 2019 18:36:57 +0000 (14:36 -0400)]
drm/amd/display: Refactor watermark programming
* Replace certain register writes with register sets that overwrites the
the entire register, instead of only a field within the register.
* Add program_watermarks() entry to hubbub vtable. Hook it up to
existing functions that program watermarks.
* Add additional watermark registers.
Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com> Reviewed-by: Tony Cheng <Tony.Cheng@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: Allow commits with no planes active
[Why]
Many userspace applications (and IGT) seem to expect that most drivers
can keep a CRTC active and enabled if there are no primary or overlay
planes.
DC is setup to handle this but only in the case where there are
absolutely no planes on the CRTC - no primary, cursor, or overlay.
[How]
Add a check to reject commits that have cursor planes enabled and
nothing else on CRTCs since we can't handle that. The new helper
does_crtc_have_active_cursor is used for this.
In atomic commit tail, we need to let DC know that there are zero
planes enabled when doing stream updates to let it disable and blank
pipes as appropriate.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Reviewed-by: David Francis <David.Francis@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: Split enabling CRTC interrupts into two passes
[Why]
When disabling all the pipes for a CRTC the page-flip interrupt also
gets disabled on Raven. We can't re-enable the page-flip interrupt
unless we give DC at least one active DC plane.
We currently enable interrupts after the call to dc_commit_state since
there's currently no valid sequence that should disable all the planes
or re-enable planes for a CRTC without first going through
dc_commit_state.
If we were to allow for a CRTC to be enabled with no primary plane this
would not be the case - the call to dc_commit_updates_for_stream would
enable the planes when going from zero to at least one active plane,
but manage_dm_interrupts would have been called too early.
This results in a page-flip timeout on any subsequent commits since we
think the page-flip are now enabled when they're actually disabled.
We need to enable interrupts after the call to
dc_commit_updates_for_stream.
[How]
Split enabling interrupts into two passes. One pass before
dc_commit_updates_for_stream and one after it.
Shifting all the interrupts to be strictly below the call doesn't
currently work even though it should in theory. We end up queuing
off the vblank event to be handle by the flip handler before it's
actually enabled in some cases, particularly:
The framebuffer states haven't changed and we can technically still
do a "pageflip" in this case and send back the event.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Reviewed-by: David Francis <David.Francis@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: Fix CRC vblank refs when changing interrupts
[Why]
We only currently drop the vblank reference when the stream is
being removed from the context. We should be dropping it whenever we
disable interrupts and reaquiring it after we re-enable them.
We also never get the extra reference correctly when re-enabling
interrupts, since grabbing the reference has the following condition:
if (!crtc_state->crc_enabled && enable)
drm_crtc_vblank_get(crtc);
This means that crc_enabled must be *false* in order to grab the extra
reference.
[How]
Always drop the ref whenever we're disabling interrupts.
Only disable CRC capture when the stream is being removed.
Always grab the ref by setting dm_new_crtc_state->crc_enabled = false
before the call to re-enable CRC capture.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Reviewed-by: David Francis <David.Francis@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: Disable cursors before disabling planes
[Why]
We can't do cursor programming after the planes have been disabled
since there won't be any pipes - leading to lock warnings and the wrong
cursor state being left in the registers.
When we re-enable the planes after the previous cursor state will also
remain if we don't have a cursor plane.
[How]
If we're disabling the planes then do the cursor programming first.
If we're not disabling the planes then do the cursor programming after.
Introduce the amdgpu_dm_commit_cursors helper to avoid code duplication
for both of these cases.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Reviewed-by: David Francis <David.Francis@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
The vblank and pageflip interrupts should only be enabled for a CRTC
that's enabled and has active planes.
The current logic takes care of this, but isn't setup to handle the case
where the active plane count goes to zero but the stream remains
enabled.
We currently block this case since we don't allow commits that enable a
CRTC with no active planes, but shouldn't be any reason we can't support
this from a hardware perspective and many userspace applications expect
to be able to do it (like IGT).
[How]
The count_crtc_active_planes function fills in the number of
"active_planes" on the dm_crtc_state. This should be the same as
DC's plane_count on the stream_status but easier to access since we
don't need to lock the private atomic state with the DC context.
Add the "interrupts_enabled" flag to the dm_crtc_state and set it based
on whether the stream exists and if there are active planes on the
stream.
Update the disable and enable logic to make use of this new flag.
There shouldn't be any functional change (yet) with this patch.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Reviewed-by: David Francis <David.Francis@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amdgpu: value of amdgpu_sriov_vf cannot be set into F32_POLL_ENABLE
amdgpu_sriov_vf would return 0x0 or 0x4 to indicate if sriov.
but F32_POLL_ENABLE need 0x0 or 0x1 to determine if enabled.
set 0x4 into F32_POLL_ENABLE would make SDMA0_GFX_RB_WPTR_POLL_CNTL not working.
Signed-off-by: Wentao Lou <Wentao.Lou@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Evan Quan [Thu, 25 Apr 2019 03:20:38 +0000 (11:20 +0800)]
drm/amdgpu: update Vega20 sdma golden settings
Update Vega20 sdma golden settings.
Signed-off-by: Evan Quan <evan.quan@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Colin Ian King [Fri, 26 Apr 2019 21:48:11 +0000 (22:48 +0100)]
drm/amd/display: fix incorrect null check on pointer
Currently an allocation is being made but the allocation failure
check is being performed on another pointer. Fix this by checking
the correct pointer. Also use the normal kernel idiom for null
pointer checks.
Addresses-Coverity: ("Resource leak") Fixes: 43e3ac8389ef ("drm/amd/display: Add function to copy DC streams") Signed-off-by: Colin Ian King <colin.king@canonical.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Implement komeda_kms_check to add all affected_planes (even unchanged) to
drm_atomic_state. since komeda need to re-calculate the resources
assumption in every commit.
v2: Rebase
Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com> Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>