Driver Changes:
- Prepare for local/device memory support on DG1 by starting
to use it for kernel internal allocations: context, ring
and engine scratch (Matt A, CQ, Abdiel, Imre)
- Sandybridge fix to avoid hard hang on ring resume (Chris)
- Limit imported dma-buf size to int32 (Matt A)
- Double check heartbeat timeout before resetting (Chris)
- Use new tasklet API for execution list (Emil)
- Fix SPDX checkpats warnings (Chris)
- Fixes for various checkpatch warnings (Chris)
- Selftest improvements (Chris)
- Move the defer_request waiter active assertion to correct spot (Chris)
- Make local-memory probing a GT operation (Matt, Tvrtko)
- Protect against request freeing during cancellation on wedging (Chris)
- Retire unexpected starting state error dumping (Chris)
- Distinction of memory regions in debugging (Zbigniew)
- Always flush the submission queue on checking for idle (Chris)
- Consolidate 2big error check to helper (Matt)
- Decrease number of subplatform bits (Tvrtko)
- Remove unused internal request priority levels (Chris)
- Document the unused internal header bits in buddy allocator (Matt)
- Cleanup the region class/instance encoding (Matt)
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/YGxksaZGXHnFxlwg@jlahtine-mobl.ger.corp.intel.com
Currently the FRL training mode (Concurrent, Sequential) and
training type (Normal, Extended) are not defined properly and
are passed as bool values in drm_helpers for pcon
configuration for FRL training.
This patch:
-Add register masks for Sequential and Normal FRL training options.
-Fixes the drm_helpers for FRL Training configuration to use the
appropriate masks.
-Modifies the calls to the above drm_helpers in i915/intel_dp as per
the above change.
v2: Re-used the register masks for these options, instead of enum. (Ville)
Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210323112422.1211-2-ankit.k.nautiyal@intel.com
An upcoming platform has a restriction that the FB stride must be
power-of-two aligned. To support framebuffer layouts that are not in
this layout add a logic that pads the tile rows to the POT aligned size.
The HW won't read the padding PTEs, so these don't have to point to an
allocated address, or even have their valid flag set. So use a NULL PTE
instead for instance the scratch page, which is simple and keeps the SG
table compact.
v2:
- Simplify plane_view_dst_stride(). (Ville)
- Pass pitch_tiles as unsigned int.
v3:
- Drop unintentional s/plane_state->rotation/plane_config->rotation/
change.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210325214808.2071517-24-imre.deak@intel.com
Save some place in the GTT VMAs by using a u16 instead of unsigned int
to store the view dimensions. The maximum FB stride is 256kB which is
4096 tiles in the worst case (yf-tiles), the maximum FB height is 16k
pixels, which is 16384 tiles in the worst case (linear 4x1 tiled FB).
v2:
- Fix worst case tile height formula in commit log. (Ville)
- Add an assign_chk_ovf helper to simplify the related assignments.
v3:
- Enclose params of the assign_chk_ovf macro in parentheses.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210325214808.2071517-21-imre.deak@intel.com
Instead of copying separately the GTT remapped and color plane view info
from the FB to the plane state, do this by copying the whole
intel_fb_view struct. For this we make sure the FB view state is fully
inited (that is also including the view type) already during FB
creation, so this init is not required during atomic check time. This
also means the we don't need to reset the unused color plane info during
atomic check, as these are already reset during FB creation.
I noticed that initial FBs will only work atm if they are page aligned
(which BIOS most probably always ensures), but add a comment to sanitize
this part once. Also we won't disable the plane if
get_initial_plane_config() failed for some reason (for instance due to
unsupported rotation), add a TODO: comment for this too.
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210325214808.2071517-19-imre.deak@intel.com
To allow the simplification of FB/plane view computation in the
follow-up patches, unify the corresponding state in the
intel_framebuffer and intel_plane_state structs into a new intel_fb_view
struct.
This adds some overhead to intel_framebuffer as the rotated view will
have now space for 4 color planes instead of the required 2 and it'll
also contain the unused offset for each color_plane info. Imo this is an
acceptable trade-off to get a simplified way of the remap computation.
Use the new intel_fb_view struct for the FB normal view as well, so (in
the follow-up patches) we can remove the special casing for normal view
calculation wrt. the calculation of remapped/rotated views. This also
adds an overhead to the intel_framebuffer struct, as the gtt remap info
and per-color plane offset/pitch is not required for the normal view,
but imo this is an acceptable trade-off as above. The per-color plane
pitch filed will be used by a follow-up patch, so we can retrieve the
pitch for each view in the same way.
No functional changes in this patch.
v2:
- Make the patch have _no functional change_.
(fix skl_check_nv12_aux_surface() and skl_check_main_surface()).
- s/i915_color_plane_view::pitch/stride/ (Ville)
Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210325214808.2071517-17-imre.deak@intel.com
Factor out to a new function the logic to calculate an FB plane's
normal-view size.
Instead of using intel_remapped_plane_info, which is related only to
remapping, add a helper to get the tile pitch and rows for an FB plane,
so these helpers can be used both by the normal size calculation and the
remapping code.
Also add a new fb_plane_view_dims struct in which we can pass around the
view (either FB plane or plane source) and tile dimensions conveniently
to functions calculating further view parameters.
v2:
- Add back the +1 tile adjustment for x!=0 in calc_plane_normal_size(). (Ville)
- s/pages/tiles/ in calc_plane_normal_size(). (Ville)
- Add a helper for the plane view width calculation. (Ville)
- Return tiles as unsigned int from calc_plane_normal_size().
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210325214808.2071517-16-imre.deak@intel.com
The HW plane state is cleared and inited after we store the rotation to
it, so store it instead to the uapi state to match what we do with all
other plane state until intel_plane_copy_uapi_to_hw_state() is called.
Rotation for initial FBs is not supported atm, but let's still fix the
plane state setup here.
While at it remove the redundant intel_state->uapi.src/dst init, which
will be done in intel_plane_copy_uapi_to_hw_state().
v2: Remove redundant intel_state->uapi.src/dst init. (Ville)
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210325214808.2071517-2-imre.deak@intel.com
The dbuf bandwidth calculations don't need the planes to be
added to the state. Each plane's data rate has already been
precalculated and stored in the crtc state, and that with
the dbuf slice usage for each plane is all the dbuf bandwidth
code needs to figure out what the minimum cdclk is.
What we're trying to do here is make sure each plane recalculates
its minimum cdclk (ie. plane->min_cdclk()) on those platforms where
the number of active planes affects the result of said calculation.
Nothing to do with any dbuf cdclk requirements.
Not sure if we had stuff in slightly different order or what,
but at least in the current scheme this is not necessary.
Cc: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210325004415.17432-2-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Add new runtime info field for stepping. Add new helpers for accessing
them. As we'll be switching platforms over to the new scheme
incrementally, check for non-initialized steppings.
In case a platform does not have separate display and gt steppings, it's
okay to use a common shorthand. However, in this case the display
stepping must not be initialized, and gt stepping is the single point of
truth.
v3: Remove IS_STEP() (José)
v2: Rename stepping->step
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/bb4275733fa390ea3dbf6f62794d55b616665230.1616764798.git.jani.nikula@intel.com
We've discussed a bit how to get the gem/gt team better integrated
and collaborate more with the wider community and agreed to the
following:
- all gem/gt patches are reviewed on dri-devel for now. That's
overkill, but in the past there was definitely too little of that.
- i915-gem folks are encouraged to cross review core patches from
other teams
- big features (especially uapi changes) need to be discussed in an
rfc patch that documents the interface and big picture design,
before we get lost in the details of the code
- Also a rough TODO (can be refined as we go ofc) to get gem/gt back
on track, like we've e.g. done with DAL/DC to get that in shape.
v2:
- add dma_fence annotations (Dave)
- tasklet helpers (Jani on irc)
There was also a discussion about moving these into gitlab issues, or
gitlab issues as additional discussion place at least. For now it's
just the TODO file
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Jason Ekstrand <jason@jlekstrand.net>
Cc: Dave Airlie <airlied@redhat.com>
Acked-by: Dave Airlie <airlied@redhat.com>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210324211041.1354941-1-daniel.vetter@ffwll.ch
A new Kconfig option CONFIG_DRM_I915_REQUEST_TIMEOUT is added, defaulting
to 20s, and this timeout is applied to all users contexts using the
previously added watchdog facility.
Result of this is that any user submission will simply fail after this
timeout, either causing a reset (for non-preemptable), or incomplete
results.
This can have an effect that workloads which used to work fine will
suddenly start failing. Even workloads comprised of short batches but in
long dependency chains can be terminated.
And because of lack of agreement on usefulness and safety of fence error
propagation this partial execution can be invisible to userspace even if
it is "listening" to returned fence status.
Another interaction is with hangcheck where care needs to be taken timeout
is not set lower or close to three times the heartbeat interval. Otherwise
a hang in any application can cause complete termination of all
submissions from unrelated clients. Any users modifying the per engine
heartbeat intervals therefore need to be aware of this potential denial of
service to avoid inadvertently enabling it.
Given all this I am personally not convinced the scheme is a good idea.
Intuitively it feels object importers would be better positioned to
enforce the time they are willing to wait for something to complete.
v2:
* Improved commit message and Kconfig text.
* Pull in some helper code from patch which got dropped.
v3:
* Bump timeout to 20s to see if it helps Tigerlake.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210324121335.2307063-7-tvrtko.ursulin@linux.intel.com
Prepares the plumbing for setting request/fence expiration time. All code
is put in place but is never activated due yet missing ability to actually
configure the timer.
Outline of the basic operation:
A timer is started when request is ready for execution. If the request
completes (retires) before the timer fires, timer is cancelled and nothing
further happens.
If the timer fires request is added to a lockless list and worker queued.
Purpose of this is twofold: a) It allows request cancellation from a more
friendly context and b) coalesces multiple expirations into a single event
of consuming the list.
Worker locklessly consumes the list of expired requests and cancels them
all using previous added i915_request_cancel().
Associated timeout value is stored in rq->context.watchdog.timeout_us.
v2:
* Log expiration.
v3:
* Include more information about user timeline in the log message.
v4:
* Remove obsolete comment and fix formatting. (Matt)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210324121335.2307063-6-tvrtko.ursulin@linux.intel.com
Disallow sentinel requests follow previous sentinels to make request
cancellation work better when faced with a chain of requests which have
all been marked as in error.
Because in cases where we end up with a stream of cancelled requests we
want to turn off request coalescing so they each will get individually
skipped by the execlists_schedule_in (which is called per ELSP port, not
per request).
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
[danvet: Fix typo in the commit message that Matthew spotted.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210324121335.2307063-4-tvrtko.ursulin@linux.intel.com
Currently, we cancel outstanding requests within a context when the
context is closed. We may also want to cancel individual requests using
the same graceful preemption mechanism.
v2 (Tvrtko):
* Cancel waiters carefully considering no timeline lock and RCU.
* Fixed selftests.
v3 (Tvrtko):
* Remove error propagation to waiters for now.
v4 (Tvrtko):
* Rebase for extracted i915_request_active_engine. (Matt)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
[danvet: Resolve conflict because intel_engine_flush_scheduler is
still called intel_engine_flush_submission]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210324121335.2307063-3-tvrtko.ursulin@linux.intel.com
