Dave Airlie [Wed, 22 Mar 2017 22:53:41 +0000 (08:53 +1000)]
Merge tag 'drm-misc-next-2017-03-21' of git://anongit.freedesktop.org/git/drm-misc into drm-next
drm-misc for 4.12, 2nd attempt this week:
- topic branch from Jon Corbet for the new graph kerneldoc support
- lots of graphs for kms/atomic things using the above
- some vblank query tuning from Chris
- gem/cma_fops macros
- moar docs
Driver stuff:
- vc4 hdmi audio, yay (Eric)
- dw-hdmi polish from a bunch of people
- some rockchip dp updates that didn't make last week (Chris Zhong)
- misc bridge&driver updates
* tag 'drm-misc-next-2017-03-21' of git://anongit.freedesktop.org/git/drm-misc: (37 commits)
drm/edid: detect SCDC support in HF-VSDB
drm/edid: detect SCDC support in HF-VSDB
drm/edid: check for HF-VSDB block
drm: Add SCDC helpers
drm: bridge: dw-hdmi: add HDMI vendor specific infoframe config
drm/bridge: dw_hdmi: support i2c extended read mode
drm/msm: add stubs for msm_{perf,rd}_debugfs_cleanup
drm: bochs: Don't remove uninitialized fbdev framebuffer
drm: vc4: remove redundant check of plane being non-null
drm/vc4: use platform_register_drivers
dma-fence: add dma_fence_match_context helper
drm/vc4: Add HDMI audio support
dt-bindings: Document the dmas and dma-names properties for VC4 HDMI
drm/atmel-hlcdc: Fix suspend/resume implementation
drm: Skip the waitqueue setup for vblank queries
drm: Defer disabling the vblank IRQ until the next interrupt (for instant-off)
drm/doc: atomic overview, with graph
drm/doc: diagram for mode objects and properties
drm/doc: Consistent kerneldoc include order
drm/doc: Add KMS overview graphs
...
Dave Airlie [Wed, 22 Mar 2017 22:47:23 +0000 (08:47 +1000)]
Merge tag 'drm-intel-next-2017-03-20' of git://anongit.freedesktop.org/git/drm-intel into drm-next
More in i915 for 4.12:
- designware i2c fixes from Hans de Goede, in a topic branch shared
with other subsystems (maybe, they didn't confirm, but requested the
pull)
- drop drm_panel usage from the intel dsi vbt panel (Jani)
- vblank evasion improvements and tracing (Maarten and Ville)
- clarify spinlock irq semantics again a bit (Tvrtko)
- new ->pwrite backend hook (right now just for shmem pageche writes),
from Chris
- more planar/ccs work from Ville
- hotplug safe connector iterators everywhere
- userptr fixes (Chris)
- selftests for cache coloring eviction (Matthew Auld)
- extend debugfs drop_caches interface for shrinker testing (Chris)
- baytrail "the rps kills the machine" fix (Chris)
- use new atomic state iterators, a lot (Maarten)
- refactor guc/huc code some (Arkadiusz Hiler)
- tighten breadcrumbs rbtree a bit (Chris)
- improve wrap-around and time handling in rps residency counters
(Mika)
- split reset-in-progress in two flags, backoff and handoff (Chris)
- other misc reset improvements from a few people
- bunch of vgpu interaction fixes with recent code changes
- misc stuff all over, as usual
* tag 'drm-intel-next-2017-03-20' of git://anongit.freedesktop.org/git/drm-intel: (144 commits)
drm/i915: Update DRIVER_DATE to 20170320
drm/i915: Initialise i915_gem_object_create_from_data() directly
drm/i915: Correct error handling for i915_gem_object_create_from_data()
drm/i915: i915_gem_object_create_from_data() doesn't require struct_mutex
drm/i915: Retire an active batch pool object rather than allocate new
drm/i915: Add i810/i815 pci-ids for completeness
drm/i915: Skip execlists_dequeue() early if the list is empty
drm/i915: Stop using obj->obj_exec_link outside of execbuf
drm/i915: Squelch WARN for VLV_COUNTER_CONTROL
drm/i915/glk: Enable pooled EUs for Geminilake
drm/i915: Remove superfluous i915_add_request_no_flush() helper
drm/i915/vgpu: Neuter forcewakes for VGPU more thoroughly
drm/i915: Fix vGPU balloon for ggtt guard page
drm/i915: Avoid use-after-free of ctx in request tracepoints
drm/i915: Assert that the context pin_counts do not overflow
drm/i915: Wait for reset to complete before returning from debugfs/i915_wedged
drm/i915: Restore engine->submit_request before unwedging
drm/i915: Move engine->submit_request selection to a vfunc
drm/i915: Split I915_RESET_IN_PROGRESS into two flags
drm/i915: make context status notifier head be per engine
...
Shashank Sharma [Mon, 13 Mar 2017 11:24:02 +0000 (16:54 +0530)]
drm/edid: detect SCDC support in HF-VSDB
This patch does following:
- Adds a new structure (drm_hdmi_info) in drm_display_info.
This structure will be used to save and indicate if sink
supports advanced HDMI 2.0 features
- Adds another structure drm_scdc within drm_hdmi_info, to
reflect scdc support and capabilities in connected HDMI 2.0 sink.
- Checks the HF-VSDB block for presence of SCDC, and marks it
in scdc structure
- If SCDC is present, checks if sink is capable of generating
SCDC read request, and marks it in scdc structure.
V2: Addressed review comments
Thierry:
- Fix typos in commit message and make abbreviation consistent
across the commit message.
- Change structure object name from hdmi_info -> hdmi
- Fix typos and abbreviations in description of structure drm_hdmi_info
end the description with a full stop.
- Create a structure drm_scdc, and keep all information related to SCDC
register set (supported, read request supported) etc in it.
Ville:
- Change rr -> read_request
- Call drm_detect_scrambling function drm_parse_hf_vsdb so that all
of HF-VSDB parsing can be kept in same function, in incremental
patches.
V3: Rebase.
V4: Rebase.
V5: Rebase.
V6: Addressed review comments from Ville
- Add clock rate calculations for 1/10 and 1/40 ratios
- Remove leftovers from old patchset
V7: Added R-B from Jose.
V8: Rebase.
V9: Rebase.
V10: Rebase.
Shashank Sharma [Mon, 13 Mar 2017 11:24:01 +0000 (16:54 +0530)]
drm/edid: detect SCDC support in HF-VSDB
This patch does following:
- Adds a new structure (drm_hdmi_info) in drm_display_info.
This structure will be used to save and indicate if sink
supports advanced HDMI 2.0 features
- Adds another structure drm_scdc within drm_hdmi_info, to
reflect scdc support and capabilities in connected HDMI 2.0 sink.
- Checks the HF-VSDB block for presence of SCDC, and marks it
in scdc structure
- If SCDC is present, checks if sink is capable of generating
SCDC read request, and marks it in scdc structure.
V2: Addressed review comments
Thierry:
- Fix typos in commit message and make abbreviation consistent
across the commit message.
- Change structure object name from hdmi_info -> hdmi
- Fix typos and abbreviations in description of structure drm_hdmi_info
end the description with a full stop.
- Create a structure drm_scdc, and keep all information related to SCDC
register set (supported, read request supported) etc in it.
Ville:
- Change rr -> read_request
- Call drm_detect_scrambling function drm_parse_hf_vsdb so that all
of HF-VSDB parsing can be kept in same function, in incremental
patches.
Thierry Reding [Mon, 13 Mar 2017 11:23:59 +0000 (16:53 +0530)]
drm: Add SCDC helpers
SCDC is a mechanism defined in the HDMI 2.0 specification that allows
the source and sink devices to communicate.
This commit introduces helpers to access the SCDC and provides the
symbolic names for the various registers defined in the specification.
V2: Rebase.
V3: Added R-B from Jose.
V4: Rebase
V5: Addressed review comments from Ville
- Handle the I2c return values in a better way (dp_dual_mode)
- Make the macros for SCDC Major/Minor more readable, by adding
a 'GET' in the macro names
V6: Rebase
V7: Rebase
V8: Rebase
V9: Rebase
V10: Rebase
Nickey Yang [Mon, 20 Mar 2017 02:57:31 +0000 (10:57 +0800)]
drm/bridge: dw_hdmi: support i2c extended read mode
"I2C Master Interface Extended Read Mode" implements a segment
pointer-based read operation using the Special Register configuration.
This patch fix https://patchwork.kernel.org/patch/7098101/ mentioned
"The current implementation does not support "I2C Master Interface
Extended Read Mode" to read data addressed by non-zero segment
pointer, this means that if EDID has more than 1 extension blocks,
EDID reading operation won't succeed"
With this patch, dw-hdmi can read EDID data with 1/2/4 blocks.
Arnd Bergmann [Mon, 20 Mar 2017 09:39:25 +0000 (10:39 +0100)]
drm/msm: add stubs for msm_{perf,rd}_debugfs_cleanup
We now call those two functions even when they are not defined
or declared anywhere because DEBUG_FS is disabled:
drivers/gpu/drm/msm/msm_drv.c: In function 'msm_drm_uninit':
drivers/gpu/drm/msm/msm_drv.c:244:2: error: implicit declaration of function 'msm_perf_debugfs_cleanup';did you mean 'msm_framebuffer_cleanup'? [-Werror=implicit-function-declaration]
drivers/gpu/drm/msm/msm_drv.c:245:2: error: implicit declaration of function 'msm_rd_debugfs_cleanup';did you mean 'msm_framebuffer_cleanup'? [-Werror=implicit-function-declaration]
This adds empty stub implementations for that case.
In the same spirit of the fix for QXL in commit 861078381ba5 ("drm: qxl:
Don't alloc fbdev if emulation is not supported"), prevent the Oops in
the unbind path of Bochs if fbdev emulation is disabled.
Dave Airlie [Mon, 20 Mar 2017 06:49:20 +0000 (16:49 +1000)]
Merge tag 'imx-drm-next-2017-03-17' of git://git.pengutronix.de/git/pza/linux into drm-next
imx-drm PRE/PRG support, deferred plane disabling, separate alpha support
- Initial support for the Prefetch Resolve Engine/Gasket on i.MX6QP,
improving linear scanout buffer memory bandwidth utilization. This
will in the future grow reordering support and allow direct scanout
of Vivante tiled renderbuffers from the GPU.
- Deferred plane disabling gets rid of some busy waiting in the atomic
plane disable and crtc disable paths that lead to wait_for_vblank
timeouts.
- Add support for RGBA formats with a separate alpha plane, that can
reduce memory bandwidth utilization for mostly transparent overlay
planes by skipping color reads for completely transparent regions.
- Allow moving an active overlay plane without enforcing a modeset.
- Add 8-bit and 16-bit bayer formats to ipu_cpmem_set_image.
- Set the base address in ipu_cpmem_set_image even for invalid formats
to increase robustness against errors.
- Use drm_plane_helper_check_state in plane atomic_check.
- Some cleanup.
* tag 'imx-drm-next-2017-03-17' of git://git.pengutronix.de/git/pza/linux: (22 commits)
drm/imx: Remove unneeded definition for structure imx_drm_component
drm/imx: use PRG/PRE when possible
drm/imx: enable/disable PRG on CRTC enable/disable
gpu: ipu-v3: only set non-zero AXI ID for IC when PRG is absent
gpu: ipu-v3: hook up PRG unit
gpu: ipu-v3: document valid IPUv3 compatibles and extend for i.MX6 QuadPlus
gpu: ipu-v3: add driver for Prefetch Resolve Gasket
gpu: ipu-v3: add DT binding for the Prefetch Resolve Gasket
gpu: ipu-v3: add driver for Prefetch Resolve Engine
gpu: ipu-v3: add DT binding for the Prefetch Resolve Engine
drm/imx: ipuv3-plane: add support for separate alpha planes
drm/imx: extend drm_plane_state_to_eba for separate channel support
gpu: ipu-v3: add support for separate alpha channels
drm: add RGB formats with separate alpha plane
drm/imx: add deferred plane disabling
drm/imx: don't wait for vblank and stop calling cleanup_planes in commit_tail
gpu: ipu-v3: add unsynchronised DP channel disabling
gpu: ipu-v3: remove IRQ dance on DC channel disable
gpu: ipu-cpmem: add bayer formats to ipu_cpmem_set_image
gpu: ipu-cpmem: set image base address even for incorrect formats
...
Colin Ian King [Thu, 16 Mar 2017 18:54:18 +0000 (18:54 +0000)]
drm: vc4: remove redundant check of plane being non-null
The pointer plane is always null on the error path at label 'fail'
hence the check if it is non-null is redundant. We can therefore
remove the check and the destruction of plane as well as the fail
error path and instead just return an -ENOMEM ERR_PTR.
Detected by CoverityScan, CID#1339532 ("Logically Dead Code")
Philipp Zabel [Fri, 17 Mar 2017 16:34:49 +0000 (17:34 +0100)]
dma-fence: add dma_fence_match_context helper
Add a helper to check if all fences in a fence array are from a given
context. For convenience, the function can also handle being given a
non-array fence.
Chris Wilson [Thu, 16 Mar 2017 13:19:53 +0000 (13:19 +0000)]
drm/i915: Retire an active batch pool object rather than allocate new
Since obj->active_count is only updated upon retirement, if we see an
active object in the batch pool, double check that is still active
before deciding to allocate a new object.
Chris Wilson [Mon, 13 Mar 2017 11:28:10 +0000 (11:28 +0000)]
drm/i915: Add i810/i815 pci-ids for completeness
To improve our historical record and to simplify userspace that wants to
include i915_pciids.h as its canonical breakdown of PCI IDs and their
respective generations, include the gen1 ids for i810 and i815.
Chris Wilson [Fri, 17 Mar 2017 12:07:16 +0000 (12:07 +0000)]
drm/i915: Skip execlists_dequeue() early if the list is empty
Do an early read of the execlists' queue before we take the spinlock and
start checking. This is safe as the first writer to the execlists queue
will cause the tasklet to be run again after a memory barrier.
v2: Keep guc in sync with execlists queue changes
v3: Explain the mb between the tasklet running on one cpu and the
execlist_first update and schedule from a second cpu.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Michał Winiarski <michal.winiarski@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Reviewed-by: Michał Winiarski <michal.winiarski@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170317120716.17191-1-chris@chris-wilson.co.uk
Chris Wilson [Thu, 16 Mar 2017 13:19:57 +0000 (13:19 +0000)]
drm/i915: Stop using obj->obj_exec_link outside of execbuf
i915_gem_stolen_list_info() sneakily takes advantage of the
obj->obj_exec_link to save itself from having to allocate. Enough of the
subterfuge, just allocate an array of pointers and sort them instead of
the list.
Chris Wilson [Fri, 17 Mar 2017 12:59:18 +0000 (12:59 +0000)]
drm/i915: Squelch WARN for VLV_COUNTER_CONTROL
Before rc6 is initialised (after driver load or resume), the value inside
VLV_COUNTER_CONTROL is undefined so we cannot make an assertion that is
in HIGH_RANGE mode.
Fixes: 6b7f6aa75e38 ("drm/i915: Use coarse grained residency counter with byt")
Testcase: igt/drv_suspend/debugfs-reader Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170317125918.11351-1-chris@chris-wilson.co.uk Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
It is unclear if the recommendation to disable it for 2x6 configurations
from commit e015dd69b2cf ("drm/i915/bxt: Add WaEnablePooledEuFor2x6")
should also apply to GLK, but it is applied anyway to be on the safe
side. That restriction can be lifted later if determined not to impact
performance.
The extra restriction should not impact user space either. The only user
space that uses this feature is Beignet, and it only does so for 3x6
devices. See See Beignet's commit 6901899ec90a ("Runtime: set the sub
slice according to kernel pooled EU configure.").
v2: Improve commit message. (Mika, Roy)
Cc: Arun Siluvery <arun.siluvery@intel.com> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Yang Rong <rong.r.yang@intel.com> Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com> Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com> Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170317140436.24645-1-ander.conselvan.de.oliveira@intel.com
The only time we need to emit a flush inside request emission is after
an execbuffer, for which we can use the full __i915_add_request(). All
other instances want the simpler i915_add_request() without flushing, so
remove the useless helper.
Tvrtko Ursulin [Fri, 10 Mar 2017 09:57:47 +0000 (09:57 +0000)]
drm/i915/vgpu: Neuter forcewakes for VGPU more thoroughly
If we avoid initializing forcewake domains when running as
a guest, and also use gen2 mmio accessors in that case, we
can avoid the timer traffic and any looping through the
forcewake code which is currently just so it can end up in
the no-op forcewake implementation.
Zhenyu Wang [Fri, 10 Mar 2017 02:22:38 +0000 (10:22 +0800)]
drm/i915: Fix vGPU balloon for ggtt guard page
From commit a6508ded2a66 ("drm/i915: Use page coloring to provide the guard
page at the end of the GTT"), we no longer explicitly subtract guard page
at end for GGTT address space init, so shouldn't subtract that for vGPU
balloon too, as that will leave that end page to be available for
vGPU. Change balloon to cover full range too.
This fixes to use recent drm-intel tip kernel for guest OS. Found by GVT-g
cmd parser that guest kernel uses end page as scratch then try to run
MI_STORE_REG_MEM onto it.
Chris Wilson [Thu, 16 Mar 2017 20:42:35 +0000 (20:42 +0000)]
drm/i915: Avoid use-after-free of ctx in request tracepoints
trace_i915_gem_request_out may be used after the request is completed,
and so the request may have been retired on another thread, invalidating
the rq->ctx. Avoid dereferencing rq->ctx in the tracepoint by switching
to the fence context id instead, updating all tracepoints to match.
Eric Anholt [Mon, 27 Feb 2017 20:28:02 +0000 (12:28 -0800)]
drm/vc4: Add HDMI audio support
The HDMI encoder IP embeds all needed blocks to output audio, with a
custom DAI called MAI moving audio between the two parts of the HDMI
core. This driver now exposes a sound card to let users stream audio
to their display.
Using the hdmi-codec driver has been considered here, but MAI meant
having to significantly rework hdmi-codec, and it would have left
little shared code with the I2S mode anyway.
The encoder requires that the audio be SPDIF-formatted frames only,
which alsalib will format-convert for us.
This patch is the combined work of Eric Anholt (initial register setup
with a separate dmaengine driver and using simple-audio-card) and
Boris Brezillon (moving it all into HDMI, massive debug to get it
actually working), and which Eric has the permission to release.
v2: Drop "-audio" from sound card name, since that's already implied
(suggestion by Boris)
Chris Wilson [Thu, 16 Mar 2017 17:13:04 +0000 (17:13 +0000)]
drm/i915: Restore engine->submit_request before unwedging
When we wedge the device, we override engine->submit_request with a nop
to ensure that all in-flight requests are marked in error. However, igt
would like to unwedge the device to test -EIO handling. This requires us
to flush those in-flight requests and restore the original
engine->submit_request.
v2: Use a vfunc to unify enabling request submission to engines
v3: Split new vfunc to a separate patch.
v4: Make the wait interruptible -- the third party fences we wait upon
may be indefinitely broken, so allow the reset to be aborted.
Fixes: 821ed7df6e2a ("drm/i915: Update reset path to fix incomplete requests")
Testcase: igt/gem_eio Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> #v3 Link: http://patchwork.freedesktop.org/patch/msgid/20170316171305.12972-3-chris@chris-wilson.co.uk
Chris Wilson [Thu, 16 Mar 2017 17:13:03 +0000 (17:13 +0000)]
drm/i915: Move engine->submit_request selection to a vfunc
It turns out that we may want to restore the original
engine->submit_request (and engine->schedule) callbacks from more than
just the guc <-> execlists transition. Move this to a vfunc so we can
have a common interface.
v2: Move initial selection to intel_engines_init_common(), repaint vfunc
with engine->set_default_submission (and a similar colour for the
helper).
Chris Wilson [Thu, 16 Mar 2017 17:13:02 +0000 (17:13 +0000)]
drm/i915: Split I915_RESET_IN_PROGRESS into two flags
I915_RESET_IN_PROGRESS is being used for both signaling the requirement
to i915_mutex_lock_interruptible() to avoid taking the struct_mutex and
to instruct a waiter (already holding the struct_mutex) to perform the
reset. To allow for a little more coordination, split these two meaning
into a couple of distinct flags. I915_RESET_BACKOFF tells
i915_mutex_lock_interruptible() not to acquire the mutex and
I915_RESET_HANDOFF tells the waiter to call i915_reset().
Changbin Du [Mon, 13 Mar 2017 02:47:11 +0000 (10:47 +0800)]
drm/i915: make context status notifier head be per engine
GVTg has introduced the context status notifier to schedule the GVTg
workload. At that time, the notifier is bound to GVTg context only,
so GVTg is not aware of host workloads.
Now we are going to improve GVTg's guest workload scheduler policy,
and add Guc emulation support for new Gen graphics. Both these two
features require acknowledgment for all contexts running on hardware.
(But will not alter host workload.) So here try to make some change.
The change is simple:
1. Move the context status notifier head from i915_gem_context to
intel_engine_cs. Which means there is a notifier head per engine
instead of per context. Execlist driver still call notifier for
each context sched-in/out events of current engine.
2. At GVTg side, it binds a notifier_block for each physical engine
at GVTg initialization period. Then GVTg can hear all context
status events.
In this patch, GVTg do nothing for host context event, but later
will add a function there. But in any case, the notifier callback is
a noop if this is no active vGPU.
Since intel_gvt_init() is called at early initialization stage and
require the status notifier head has been initiated, I initiate it in
intel_engine_setup().
v2: remove a redundant newline. (chris)
Fixes: 3c7ba6359d70 ("drm/i915: Introduce execlist context status change notification")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=100232 Signed-off-by: Changbin Du <changbin.du@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Zhi Wang <zhi.a.wang@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/20170313024711.28591-1-changbin.du@intel.com Acked-by: Zhenyu Wang <zhenyuw@linux.intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Thu, 16 Mar 2017 12:56:18 +0000 (12:56 +0000)]
drm/i915/scheduler: emulate a scheduler for guc
This emulates execlists on top of the GuC in order to defer submission of
requests to the hardware. This deferral allows time for high priority
requests to gazump their way to the head of the queue, however it nerfs
the GuC by converting it back into a simple execlist (where the CPU has
to wake up after every request to feed new commands into the GuC).
v2: Drop hack status - though iirc there is still a lockdep inversion
between fence and engine->timeline->lock (which is impossible as the
nesting only occurs on different fences - hopefully just requires some
judicious lockdep annotation)
v3: Apply lockdep nesting to enabling signaling on the request, using
the pattern we already have in __i915_gem_request_submit();
v4: Replaying requests after a hang also now needs the timeline
spinlock, to disable the interrupts at least
v5: Hold wq lock for completeness, and emit a tracepoint for enabling signal
v6: Reorder interrupt checking for a happier gcc.
v7: Only signal the tasklet after a user-interrupt if using guc scheduling
v8: Restore lost update of rq through the i915_guc_irq_handler (Tvrtko)
v9: Avoid re-initialising the engine->irq_tasklet from inside a reset
v10: Hook up the execlists-style tracepoints
v11: Clear the execlists irq_posted bit after taking over the interrupt/tasklet
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170316125619.6856-1-chris@chris-wilson.co.uk Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Chris Wilson [Tue, 14 Mar 2017 11:14:52 +0000 (11:14 +0000)]
drm/i915: Replace irq_seqno_barrier on hws write with a clflush
When manually overwriting the HWS, rather than assume irq_seqno_barrier
does the right thing, we can explicitly flush the cacheline instead.
This avoids us calling the engine->irq_seqno_barrier() from an illegal
context:
Mika Kuoppala [Wed, 15 Mar 2017 16:12:59 +0000 (18:12 +0200)]
drm/i915: Use coarse grained residency counter with byt
Set byt rc residency counters high level as chv does by
default. We lose some accuracy on byt but we can do the calculation
without extra hw read on both platforms, as now they behave
identically in this respect.
Mika Kuoppala [Wed, 15 Mar 2017 15:43:03 +0000 (17:43 +0200)]
drm/i915: Use ktime to calculate rc0 residency
We have used cz timestamp register to gain a reference time wrt
to residency calculations. The residency counts are in cz clk ticks
(333Mhz clock) but for some reason the cz timestamp register gives
100us units. Perhaps for some other usage, the base-ten based values
are easier, but in residency calculations raw units would have been
the easiest.
As there is not much advantage of using base-ten clock through
a more costly punit access, take our reference times directly from
kernel clock.
v2: use ktime (Chris, Ville)
Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
References: https://bugs.freedesktop.org/show_bug.cgi?id=94852 Reported-by: Len Brown <len.brown@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Mika Kuoppala [Wed, 15 Mar 2017 15:43:00 +0000 (17:43 +0200)]
drm/i915: Return residency as microseconds
Change the granularity from milliseconds to microseconds
when returning rc6 residencies. This is in preparation
for increased resolution on some platforms.
v2: use 64bit div macro (Chris)
Cc: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
The current suspend resume implementation is assuming register values are
kept when entering suspend, which is no longer the case with the
suspend-to-RAM on the sama5d2.
While at it, switch to the generic infrastructure to enter suspend mode
(drm_atomic_helper_suspend/resume()).
lockdep doesn't like us taking the mm->mmap_sem inside the get_pages
callback for a couple of reasons. The straightforward deadlock:
[13755.434059] =============================================
[13755.434061] [ INFO: possible recursive locking detected ]
[13755.434064] 4.11.0-rc1-CI-CI_DRM_297+ #1 Tainted: G U
[13755.434066] ---------------------------------------------
[13755.434068] gem_userptr_bli/8398 is trying to acquire lock:
[13755.434070] (&mm->mmap_sem){++++++}, at: [<ffffffffa00c988a>] i915_gem_userptr_get_pages+0x5a/0x2e0 [i915]
[13755.434096]
but task is already holding lock:
[13755.434098] (&mm->mmap_sem){++++++}, at: [<ffffffff8104d485>] __do_page_fault+0x105/0x560
[13755.434105]
other info that might help us debug this:
[13755.434108] Possible unsafe locking scenario:
By checking the status of the gup worker (serialized by the
obj->mm.lock) we can determine whether it is still active, has failed or
has succeeded. If the worker is still active (or failed), we know that
it cannot be bound and so we can skip taking struct_mutex (risking
potential recursion). As we check the worker status, we mark it to
discard any partial results, forcing us to restart on the next
get_pages.
Reported-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Fixes: 1c8782dd313e ("drm/i915/userptr: Disallow wrapping GTT into a userptr")
Testcase: igt/gem_userptr_blits/map-fixed-invalidate-gup
Testcase: igt/gem_userptr_blits/dmabuf-sync Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Michał Winiarski <michal.winiarski@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170315140150.19432-1-chris@chris-wilson.co.uk Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Lucas Stach [Wed, 8 Mar 2017 11:13:21 +0000 (12:13 +0100)]
drm/imx: use PRG/PRE when possible
Allow the planes to use the PRG/PRE units as linear prefetchers when
possible. This improves DRAM efficiency a bit and reduces the chance
for display underflow when the memory subsystem is under load.
This does not yet support scanning out tiled buffers directly, as this
needs more work, but it already wires up the basic interaction between
imx-drm, the IPUv3 driver and the PRG and PRE drivers.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Lucas Stach [Wed, 8 Mar 2017 11:13:20 +0000 (12:13 +0100)]
drm/imx: enable/disable PRG on CRTC enable/disable
On i.MX6 QuadPlus the PRG needs to be clocked in order to pass
through the data access requests from the IDMAC. This call is a
no-op for other all other SoCs.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Lucas Stach [Wed, 8 Mar 2017 11:13:17 +0000 (12:13 +0100)]
gpu: ipu-v3: document valid IPUv3 compatibles and extend for i.MX6 QuadPlus
Document the valid compatible strings for the IPUv3.
On i.MX6 QuadPlus the IPU needs to know which PRG has to be
used for this IPU instance. Add a "fsl,prg" property containing
a phandle pointing to the correct PRG device.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Michal Wajdeczko [Wed, 15 Mar 2017 13:37:41 +0000 (13:37 +0000)]
drm/i915: Make intel_uc_sanitize_options() more robust
After negative guc fw selection we could leave guc
submission flag still turned on. Reorder some checks
to cover this case. While here, fix info message and
return early if there is no Guc.
Chris Wilson [Wed, 15 Mar 2017 20:40:26 +0000 (20:40 +0000)]
drm: Skip the waitqueue setup for vblank queries
Avoid adding to the waitqueue and reprobing the current vblank if the
caller is only querying the current vblank sequence and timestamp, where
we know that the wait would return immediately.
v2: Add CRTC identifier to debug messages
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Michel Dänzer <michel@daenzer.net> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Dave Airlie <airlied@redhat.com>, Cc: Mario Kleiner <mario.kleiner.de@gmail.com> Reviewed-by: Michel Dänzer <michel@daenzer.net> Reviewed-and-tested-by: Mario Kleiner <mario.kleiner.de@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20170315204027.20160-2-chris@chris-wilson.co.uk
Chris Wilson [Wed, 15 Mar 2017 20:40:25 +0000 (20:40 +0000)]
drm: Defer disabling the vblank IRQ until the next interrupt (for instant-off)
On vblank instant-off systems, we can get into a situation where the cost
of enabling and disabling the vblank IRQ around a drmWaitVblank query
dominates. And with the advent of even deeper hardware sleep state,
touching registers becomes ever more expensive. However, we know that if
the user wants the current vblank counter, they are also very likely to
immediately queue a vblank wait and so we can keep the interrupt around
and only turn it off if we have no further vblank requests queued within
the interrupt interval.
After vblank event delivery, this patch adds a shadow of one vblank where
the interrupt is kept alive for the user to query and queue another vblank
event. Similarly, if the user is using blocking drmWaitVblanks, the
interrupt will be disabled on the IRQ following the wait completion.
However, if the user is simply querying the current vblank counter and
timestamp, the interrupt will be disabled after every IRQ and the user
will enabled it again on the first query following the IRQ.
v2: Mario Kleiner -
After testing this, one more thing that would make sense is to move
the disable block at the end of drm_handle_vblank() instead of at the
top.
Turns out that if high precision timestaming is disabled or doesn't
work for some reason (as can be simulated by echo 0 >
/sys/module/drm/parameters/timestamp_precision_usec), then with your
delayed disable code at its current place, the vblank counter won't
increment anymore at all for instant queries, ie. with your other
"instant query" patches. Clients which repeatedly query the counter
and wait for it to progress will simply hang, spinning in an endless
query loop. There's that comment in vblank_disable_and_save:
"* Skip this step if there isn't any high precision timestamp
* available. In that case we can't account for this and just
* hope for the best.
*/
With the disable happening after leading edge of vblank (== hw counter
increment already happened) but before the vblank counter/timestamp
handling in drm_handle_vblank, that step is needed to keep the counter
progressing, so skipping it is bad.
Now without high precision timestamping support, a kms driver must not
set dev->vblank_disable_immediate = true, as this would cause problems
for clients, so this shouldn't matter, but it would be good to still
make this robust against a future kms driver which might have
unreliable high precision timestamping, e.g., high precision
timestamping that intermittently doesn't work.
v3: Patch before coffee needs extra coffee.
Testcase: igt/kms_vblank Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Michel Dänzer <michel@daenzer.net> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Dave Airlie <airlied@redhat.com>, Cc: Mario Kleiner <mario.kleiner.de@gmail.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20170315204027.20160-1-chris@chris-wilson.co.uk
Chris Wilson [Wed, 15 Mar 2017 21:07:26 +0000 (21:07 +0000)]
drm/i915: Only attempt to signal the request once from the interrupt handler
Check that request has not been signaled before acquiring a reference to
the request for signaling later in the interrupt handler.
The loading of the cacheline (for request->fence.flags) should be "free"
when followed by the locked increment of the request->fence.refcount
(which then sets the cacheline to exclusive mode), i.e. the cost of
test_bit prior to an atomic_inc should be negligible. This should
benefit us when we have a pile of bare breadcrumbs (interrupted execbuf)
where we may get interrupts faster than we can get rid of the
intel_wait, or if the device is too slow to run the bottom-half between
interrupts.
Chris Wilson [Wed, 15 Mar 2017 21:07:25 +0000 (21:07 +0000)]
drm/i915/breadcrumbs: Assert that we do not shortcut the current bottom-half
We need to ensure that we always serialize updates to the bottom-half
using the breadcrumbs.irq_lock so that we don't race with a concurrent
interrupt handler. This is most important just prior to leaving the
waiter (when the intel_wait will be overwritten), so make sure we are
not the current bottom-half when skipping the irq locks.
Chris Wilson [Wed, 15 Mar 2017 21:07:24 +0000 (21:07 +0000)]
drm/i915/breadcrumbs: Disable interrupt bottom-half first on idling
Before walking the rbtree of waiters (marking them as complete and waking
them), decouple the interrupt handler. This prevents a race between the
missed waiter waking up and removing its intel_wait (which skips
checking the lock) and the interrupt handler dereferencing the
intel_wait. (Though we do not expect to encounter waiters during idle!)
Chris Wilson [Wed, 15 Mar 2017 21:07:23 +0000 (21:07 +0000)]
drm/i915/breadcrumbs: Update bottom-half before marking as complete
When adding a new request to the breadcrumb rbtree, we mark all those
requests inside the rbtree that are already completed as complete. This
wakes those waiters up and allows them to skip the spinlock before
returning to userspace. If one of those is the current bottom-half and
allocated its intel_wait on the stack, it may then overwrite the
b->irq_wait upon exiting i915_wait_request() just as the interrupt handler
dereferences it.
Fixes: 56299fb7d904 ("drm/i915: Signal first fence from irq handler if complete") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170315210726.12095-2-chris@chris-wilson.co.uk
Chris Wilson [Wed, 15 Mar 2017 21:07:22 +0000 (21:07 +0000)]
drm/i915/breadcrumbs: Use booleans for intel_breadcrumbs_busy()
Since commit 9b6586ae9f6b ("drm/i915: Keep a global seqno per-engine")
converted intel_breadcrumbs_busy() to reporting a single boolean, we
need only compute a boolean internally (and not needlessly compute the
flag).
Chris Wilson [Tue, 14 Mar 2017 17:18:40 +0000 (17:18 +0000)]
drm/i915: Extend rpm wakelock during i915_handle_error()
We take the runtime pm wakelock during i915_handle_error() to ensure
that all paths that reach the error handler keep the device awake during
the hw reads. However, we need to extend that from the reset handler to
include the earlier capture routines.
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Oscar Mateo <oscar.mateo@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/20170314133309.126432-1-michal.wajdeczko@intel.com Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Lucas Stach [Wed, 8 Mar 2017 11:13:15 +0000 (12:13 +0100)]
gpu: ipu-v3: add DT binding for the Prefetch Resolve Gasket
This adds the the devicetree binding for the Prefetch Resolve Gasket,
as found on i.MX6 QuadPlus.
The PRG is fairly simple in that it only has a configuration register
range and two clocks, one for the AHB slave port and one for the AXI
ports and the functional units.
The PRE connections need to be described in the DT, as the PRE<->PRG
assignment is a mix between fixed and muxable connections.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de> Acked-by: Rob Herring <robh@kernel.org> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Lucas Stach [Wed, 8 Mar 2017 11:13:14 +0000 (12:13 +0100)]
gpu: ipu-v3: add driver for Prefetch Resolve Engine
This adds support for the i.MX6 QuadPlus PRE units. Currently only
linear prefetch into SRAM is supported, other modes of operation
like the tiled-to-linear conversion will be added later.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Lucas Stach [Wed, 8 Mar 2017 11:13:13 +0000 (12:13 +0100)]
gpu: ipu-v3: add DT binding for the Prefetch Resolve Engine
The Prefetch Resolve Engine is a prefetch and tile resolve engine
which prefetches display data from DRAM to an internal SRAM region.
It has a single clock for configuration register access and the
functional units. A single shared interrupt is used for status and
error signaling.
The only external dependency is the SRAM region to use for the
prefetch double buffer.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Philipp Zabel [Fri, 22 Jul 2016 10:02:45 +0000 (12:02 +0200)]
drm/imx: ipuv3-plane: add support for separate alpha planes
The IPUv3 can read 8-bit alpha values from a separate plane buffer using
a companion IDMAC channel driven by the Alpha Transparency Controller
(ATC) for the graphics channels. The conditional read mechanism allows
to reduce memory bandwidth by skipping reads of color data for
completely transparent bursts.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Philipp Zabel [Fri, 9 Jan 2015 10:03:13 +0000 (11:03 +0100)]
gpu: ipu-v3: add support for separate alpha channels
The IPUv3 can read 8-bit alpha values from a separate IDMAC channel driven
by the Alpha Transparency Controller (ATC) for the graphics IDMAC channels.
This allows to reduce memory bandwidth via a conditional read mechanism or
to support planar YUV formats with alpha transparency.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Philipp Zabel [Fri, 9 Jan 2015 10:05:13 +0000 (11:05 +0100)]
drm: add RGB formats with separate alpha plane
Some hardware can read the alpha components separately and then
conditionally fetch color components only for non-zero alpha values.
This patch adds fourcc definitions for two-plane RGB formats with an
8-bit alpha channel on a second plane.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Philipp Zabel [Fri, 24 Feb 2017 17:31:05 +0000 (18:31 +0100)]
drm/imx: add deferred plane disabling
The DP (display processor) channel disable code tried to busy wait for
the DP sync flow end interrupt status bit when disabling the partial
plane without a full modeset. That never worked reliably, and it was
disabled completely by the recent "gpu: ipu-v3: remove IRQ dance on DC
channel disable" patch, causing ipu_wait_interrupt to always time out
after 50 ms, which in turn would trigger a timeout in
drm_atomic_helper_wait_for_vblanks.
This patch changes ipu_plane_atomic_disable to only queue a DP channel
register update at the next frame boundary and set a flag, which can be
done without any waiting whatsoever. The imx_drm_atomic_commit_tail then
calls a new ipu_plane_disable_deferred function that does the actual
IDMAC teardown of the planes that are flagged for deferred disabling,
after waiting for the vblank.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de> Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Philipp Zabel [Fri, 24 Feb 2017 17:16:47 +0000 (18:16 +0100)]
drm/imx: don't wait for vblank and stop calling cleanup_planes in commit_tail
drm_atomic_helper_cleanup_planes only calls the cleanup_fb plane
helpers, which we don't implement as a CMA framebuffer based driver.
There is no reason to wait for vblanks in commit_tail only to do nothing
afterwards.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de> Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
When disabling the foreground DP channel during a modeset, the DC is
already disabled without waiting for end of frame. There is no reason
to wait for a frame boundary before updating the DP registers in that
case.
Add support to apply updates immediately. No functional changes, yet.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de> Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Lucas Stach [Tue, 8 Nov 2016 15:13:25 +0000 (16:13 +0100)]
gpu: ipu-v3: remove IRQ dance on DC channel disable
This has never worked properly, as the IRQ got retriggered immediately
on unmask. Remove the IRQ wait dance, as it is apparently safe to disable
the DC channel at any point in time.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Philipp Zabel [Mon, 6 Feb 2017 11:44:08 +0000 (12:44 +0100)]
gpu: ipu-cpmem: add bayer formats to ipu_cpmem_set_image
The IPU does not natively understand bayer formats, but it can pass them
through unchanged. Add support for setting the image base address and
cropping offset to ipu_cpmem_set_image.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Philipp Zabel [Mon, 6 Feb 2017 11:39:01 +0000 (12:39 +0100)]
gpu: ipu-cpmem: set image base address even for incorrect formats
Otherwise, if the image base address is kept at zero, and if the user
ignores the error return value, the IPU may be configured to write into
the dma-apbh@00110000 region for large frames, which will lock up the
system.
Reported-by: Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Philipp Zabel [Thu, 20 Oct 2016 13:37:52 +0000 (15:37 +0200)]
drm/imx: ipuv3-plane: update overlay plane position also without modeset
Previously, the overlay plane position would only be updated when the
plane was first enabled or during a modeset. We can instruct the DP to
move the plane also when just updating the EBA.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Philipp Zabel [Wed, 19 Oct 2016 08:50:26 +0000 (10:50 +0200)]
drm/imx: ipuv3-plane: use drm_plane_helper_check_state, clipped coordinates
Use drm_plane_helper_check_state to clip raw user coordinates to crtc
bounds. This checks for full plane coverage and scaling already, so
we can drop some custom checks. Use the clipped coordinates everywhere.
Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Arkadiusz Hiler [Tue, 14 Mar 2017 14:28:12 +0000 (15:28 +0100)]
drm/i915/uc: Simplify firmware path handling
Currently fw->path values can represent one of three possible states:
1) NULL - device without the uC
2) '\0' - device with the uC but have no firmware
3) else - device with the uC and we have firmware
Second case is used only to WARN at a later stage.
We can WARN right away and merge cases 1 and 2.
Code can be even further simplified and common (HuC/GuC logic) happening
right before the fetch can be offloaded to the common function.
v2: fewer temporary variables, more straightforward flow (M. Wajdeczko)
v3: DRM_ERROR instead of WARN (M. Wajdeczko)
v4: coding standard (J. Lahtinen)
v5: non-trivial rebase
v6: remove path check, we are checking fetch status (M. Wajdeczko)
Arkadiusz Hiler [Tue, 14 Mar 2017 14:28:11 +0000 (15:28 +0100)]
drm/i915/guc: Simplify intel_guc_init_hw()
Current version of intel_guc_init_hw() does a lot:
- cares about submission
- loads huc
- implement WA
This change offloads some of the logic to intel_uc_init_hw(), which now
cares about the above.
v2: rename guc_hw_reset and fix typo in define name (M. Wajdeczko)
v3: rename once again
v4: remove spurious comments and add some style (J. Lahtinen)
v5: flow changes, got rid of dead checks (M. Wajdeczko)
v6: rebase
v7: rebase & onion teardown (J. Lahtinen)
Arkadiusz Hiler [Tue, 14 Mar 2017 14:28:10 +0000 (15:28 +0100)]
drm/i915/guc: Extract param logic form guc_init_fw()
Let intel_guc_init_fw() focus on determining and fetching the correct
firmware.
This patch introduces intel_uc_sanitize_options() that is called from
intel_sanitize_options().
Then, if we have GuC, we can call intel_guc_init_fw() conditionally
and we do not have to do the internal checks.
v2: fix comment, notify when nuking GuC explicitly enabled (M. Wajdeczko)
v3: fix comment again, change the nuke message (M. Wajdeczko)
v4: update title to reflect new function name + rebase
v5: text && remove 2 uneccessary checks (M. Wajdeczko)
Arkadiusz Hiler [Tue, 14 Mar 2017 14:28:09 +0000 (15:28 +0100)]
drm/i915/uc: Introduce intel_uc_init_fw()
Instead of calling intel_guc_init() and intel_huc_init() one by one this
patch introduces intel_uc_init_fw() function that calls them both.
Called functions are renamed accordingly.
Trying to have subject_verb_object ordering and more descriptive names,
the intel_huc_init() and intel_guc_init() functions are renamed.
For guc_init():
* `intel_guc` is the subject, so those functions now take intel_guc
structure, instead of the dev_priv
* init is the verb
* fw is the object which better describes the function's role
huc_init() change follows the same reasoning.
v2: settle on intel_uc_fetch_fw name (M. Wajdeczko)
v3: yet another rename - intel_uc_init_fw (J. Lahtinen)
v4: non-trivial rebase
Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Michal Winiarski <michal.winiarski@intel.com> Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Arkadiusz Hiler [Tue, 14 Mar 2017 14:28:08 +0000 (15:28 +0100)]
drm/i915/uc: Move intel_uc_fw_fetch() to intel_uc.c
The file fits better.
Additionally rename it to intel_uc_prepare_fw(), as the function does
more than simple fetch.
`obj` cleanup in the function is also fixed (i.e. removed). In the fail
scenario it was always 'put' but there's no possible flow that
initializes the obj properly and then goes to the fail label.
v2: remove second declaration, reorder (M. Wajdeczko)
v3: non-trivial rebase
v4: remove obj cleanup in the fail scenario (C. Wilson)
Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com> Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com> Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Arkadiusz Hiler [Tue, 14 Mar 2017 14:28:07 +0000 (15:28 +0100)]
drm/i915/uc: Rename intel_?uc_{setup, load}() to _init_hw()
GuC historically has two "startup" functions called _init() and _setup()
Then HuC came with it's _init() and _load().
This commit renames intel_guc_setup() and intel_huc_load() to
*uc_init_hw() as they called from the i915_gem_init_hw().
The aim is to be consistent in that entry points called during
particular driver init phases (e.g. init_hw) are all suffixed by that
phase. When reading the leaf functions, it should be clear at what stage
during the driver load it is called and therefore what operations are
legal at that point.
Also, since the functions start with intel_guc and intel_huc they take
appropiate structure.
v2: commit message update (Chris Wilson)
v3: change taken parameters to be more "semantic" (M. Wajdeczko)
Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Michal Winiarski <michal.winiarski@intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Dave Airlie [Wed, 15 Mar 2017 01:32:01 +0000 (11:32 +1000)]
Merge tag 'drm-misc-next-2017-03-12' of git://anongit.freedesktop.org/git/drm-misc into drm-next
More drm-misc stuff for 4.12:
- drm_platform removal from Laurent
- more dw-hdmi bridge driver updates (Laurent, Kieran, Neil)
- more header cleanup and documentation
- more drm_debugs_remove_files removal (Noralf)
- minor qxl updates (Gerd)
- edp crc support in helper + analogix_dp (Tomeu) for more igt
testing!
- old/new iterator roll-out (Maarten)
- new bridge drivers: lvds (Laurent), megachips-something (Peter
Senna)
* tag 'drm-misc-next-2017-03-12' of git://anongit.freedesktop.org/git/drm-misc: (51 commits)
drm: bridge: dw-hdmi: Move the driver to a separate directory.
drm: bridge: dw-hdmi: Switch to regmap for register access
drm: bridge: dw-hdmi: Remove device type from platform data
drm: bridge: dw-hdmi: Add support for custom PHY configuration
drm: bridge: dw-hdmi: Create PHY operations
drm: bridge: dw-hdmi: Fix the PHY power up sequence
drm: bridge: dw-hdmi: Fix the PHY power down sequence
drm: bridge: dw-hdmi: Enable CSC even for DVI
drm: bridge: dw-hdmi: Move CSC configuration out of PHY code
drm: bridge: dw-hdmi: Remove unused functions
drm: Extract drm_file.h
drm: Remove DRM_MINOR_CNT
drm: rename drm_fops.c to drm_file.c
drm/doc: document fallback behaviour for atomic events
drm: Remove drmP.h include from drm_kms_helper_common.c
drm: Extract drm_pci.h
drm: Move drm_lock_data out of drmP.h
drm: Extract drm_prime.h
drm/doc: Add todo about connector_list_iter
drm/qxl: Remove qxl_debugfs_remove_files()
...
Chris Wilson [Tue, 14 Mar 2017 11:50:19 +0000 (11:50 +0000)]
drm/i915: Avoid rcu_barrier() from reclaim paths (shrinker)
The rcu_barrier() takes the cpu_hotplug mutex which itself is not
reclaim-safe, and so rcu_barrier() is illegal from inside the shrinker.
[ 309.661373] =========================================================
[ 309.661376] [ INFO: possible irq lock inversion dependency detected ]
[ 309.661380] 4.11.0-rc1-CI-CI_DRM_2333+ #1 Tainted: G W
[ 309.661383] ---------------------------------------------------------
[ 309.661386] gem_exec_gttfil/6435 just changed the state of lock:
[ 309.661389] (rcu_preempt_state.barrier_mutex){+.+.-.}, at: [<ffffffff81100731>] _rcu_barrier+0x31/0x160
[ 309.661399] but this lock took another, RECLAIM_FS-unsafe lock in the past:
[ 309.661402] (cpu_hotplug.lock){+.+.+.}
[ 309.661404]
and interrupts could create inverse lock ordering between them.
[ 309.661410]
other info that might help us debug this:
[ 309.661414] Possible interrupt unsafe locking scenario:
Daniel Vetter [Thu, 2 Mar 2017 15:16:38 +0000 (16:16 +0100)]
drm/doc: atomic overview, with graph
I want to split up a few more things and document some details better
(like how exactly to subclass drm_atomic_state). And maybe also split
up the helpers a bit per-topic, but this should be a ok-ish start for
better atomic overview.
v2: Spelling and clarifications (Eric).
v3: Implement suggestion from Gabriel to fix the graph.
v4: Review from Laurent.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Gabriel Krisman Bertazi <krisman@collabora.co.uk> Acked-by: Eric Anholt <eric@anholt.net> Cc: Eric Anholt <eric@anholt.net> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170302151638.1882-6-daniel.vetter@ffwll.ch
Daniel Vetter [Thu, 2 Mar 2017 15:16:36 +0000 (16:16 +0100)]
drm/doc: Consistent kerneldoc include order
First overview text (if there is any), then headers (since generally
you want to start out with the data structures), then all the other
stuff with functions.
Most of this is pre-shpinx, since with the old docbook only the
overview stuff was pulled in directly. Everything else was put in a
per-section index, so include order didn't really matter.
drm/i915/glk: Improve rounding caused by pre-CSC gamma tables
The 33rd entry in the pre-CSC gamma table in Geminilake can represent a
value of 1.0 as 17 bits fixed point with one integer bit. However, the
table was generated such that the value of 1.0 would be 0.ffff with
all the intervals scaled accordingly. For instance, 0.5 mapped to
0.7fff instead of 0.8000.
For a reason that is not clear to the author, the rounding seems to be
different when a cursor plane is used, leading to some seemingly random
failures of the kms_cursor_crc igt tests. The differences weren't
perceptible at 8bpc with images captured by a Chamelium device, but did
cause CRC mismatches.