drm/i915: Apply batch location restrictions before pinning

We special case the position of the batch within the GTT to prevent
negative self-relocation deltas from underflowing. However, that
restriction is being applied after a trial pin of the batch in its
current position. Thus we are not rejecting an invalid location if the
batch has been used before, leading to an assertion if we happen to need
to rearrange the entire payload. In the worst case, this may cause a GPU
hang on gen7 or perhaps missing state.

References: https://bugs.freedesktop.org/show_bug.cgi?id=105720
Fixes: 2889caa92321 ("drm/i915: Eliminate lots of iterations over the execobjects array")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Martin Peres <martin.peres@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180610194325.13467-2-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/i915: Refactor unsettting obj->mm.pages

As i915_gem_object_phys_attach() wants to play dirty and mess around
with obj->mm.pages itself (replacing the shmemfs with a DMA allocation),
refactor the gubbins so into i915_gem_object_unset_pages() that we don't
have to duplicate all the secrets.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180611075532.26534-1-chris@chris-wilson.co.uk
Link: https://patchwork.freedesktop.org/patch/msgid/152871104647.1718.8796913290418060204@jlahtine-desk.ger.corp.intel.com

drm/i915: Squash GEM load failure message (again)

Due to a silent conflict (silent because we are trying to fix the CI
test that is meant to exercising these failures!) between commit
51e645b6652c ("drm/i915: Mark the GPU as wedged without error on fault
injection") and commit 8571a05a9dd0 ("drm/i915: Use GEM suspend when
aborting initialisation"), we failed to actually squash the error
message after injecting the load failure.

Rearrange the code to export i915_load_failure() for better logging of
real errors (and quiet logging of injected errors).

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180609111058.2660-1-chris@chris-wilson.co.uk

drm/i915: fix PCH_NOP setting for non-PCH platforms

Setting PCH type to PCH_NOP before checking whether we actually have a
PCH ends up returning true for HAS_PCH_SPLIT() on all non-PCH split
platforms. Fix this by using PCH_NOP only for platforms that actually
have a PCH.

Cc: Ville Syrjala <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180608123330.31003-6-jani.nikula@intel.com

drm/i915: be more strict about HAS_PCH_NOP() usage

HAS_PCH_NOP() implies a PCH platform without south display, not generic
disabled display. Prefer num_pipes == 0 for PCH independent checks.

Cc: Ville Syrjala <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180608123330.31003-5-jani.nikula@intel.com

drm/i915: clean up virtual PCH special case handling

Use intel_pch_type() also for mapping the no PCH case (PCH id 0) to
PCH_NONE to simplify code.

Also make sure that intel_pch_type() knows all the PCH ids returned by
intel_virt_detect_pch(). Loudly fail if this isn't the case; this
shouldn't happen anyway.

Cc: Colin Xu <Colin.Xu@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Colin Xu <Colin.Xu@intel.com>
Reviewed-by: Colin Xu <Colin.Xu@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180608123330.31003-4-jani.nikula@intel.com

drm/i915: document PCH_NOP

There's a difference between PCH_NONE and PCH_NOP: the former means we
don't have a PCH while in the latter we do, but it doesn't have the
south display.

Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180608123330.31003-3-jani.nikula@intel.com

drm/i915: fix guest virtual PCH detection on non-PCH systems

Virtualized non-PCH systems such as Broxton or Geminilake should use
PCH_NONE to indicate no PCH rather than PCH_NOP. The latter is a
specific case to indicate a PCH system without south display.

Reported-by: Colin Xu <Colin.Xu@intel.com>
Cc: Colin Xu <Colin.Xu@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Colin Xu <Colin.Xu@intel.com>
Reviewed-by: Colin Xu <Colin.Xu@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180608123330.31003-2-jani.nikula@intel.com

drm/i915/gtt: Reorder aliasing_ppgtt fini

To allow ourselves to use a first class vma for the aliasing_ppgtt page
directory, we have to reorder the shutdown on module unload to remove
and unpin the aliasing_ppgtt before complaining about any objects left
in the GGTT.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180609090151.22007-1-chris@chris-wilson.co.uk

drm/i915/gtt: Remove vgpu check for gen6

Since vgpu is not supported on Haswell or any other gen6/7, we do not
need to check and act upon it's enablement.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180608150435.15010-2-chris@chris-wilson.co.uk

drm/i915/gtt: Remove redundant hsw_mm_switch()

hsw_mm_switch() and gen7_mm_switch() are identical, so let's remove the
redundant specialism.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180608150435.15010-1-chris@chris-wilson.co.uk

drm/i915/gtt: Fix unwind length passed to gen6_ppgtt_clear_range

When we want to unwind an error when allocating the PD for gen6, we call
gen6_ppgtt_clear_range() telling to clear upto the PD we've previously
allocated. However, we passed it an incorrect length, passing it the
endpoint instead. Fortunately, as the start was always 0, this has no
impact today, but tomorrow we want to start using non-zero origins.

Reported-by: Matthew Auld <matthew.william.auld@gmail.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180608173221.10455-1-chris@chris-wilson.co.uk

drm/i915: Fix sprite destination colorkeying on SKL+

On SKL+ the dst colorkey must be configured on the lower
plane that contains the colorkey. This is in contrast to
most earlier platforms where the dst colorkey is configured
on the plane above.

The hardware will peform dst keying only between two immediately
adjacent (in zorder) planes. Plane 2 will be keyed against plane 1,
plane 3 againts plane 2, and so on. There is no way to key arbitrary
planes against plane 1. Thus offering dst color keying on plane 3+
is pointless. In fact it can be harmful since enabling dst keying on
more than one plane on the same pipe leads to only the top-most of
the planes performing the keying. For any plane lower in zorder the
dst key enable is simply ignored.

v2: s/plane 0/plane 1/ etc. since the hw plane names start from 1
    Don't break dst colorkey on pre-SKL sprites (hunk ended in the
    wrong patch)

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180529182804.8571-1-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com> #v1

drm/i915: Mark the GPU as wedged without error on fault injection

If we have been instructed (by CI) to inject a fault to load the module
with a wedged GPU, do so quietly less we upset CI.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180607134558.31150-1-chris@chris-wilson.co.uk

drm/i915: Add WaKBLVECSSemaphoreWaitPoll

There is a problem with kbl up to rev E0 where a heavy
memory/fabric traffic from adjacent engine(s) can cause an engine
reset to fail. This traffic can be from normal memory accesses
or it can be from heavy polling on a semaphore wait.

For engine hogging causing a fail, we already fallback to
full reset. Which effectively stops all engines and thus
we only add a workaround documentation.

For the semaphore wait loop poll case, we add one microsecond
poll interval to semaphore wait to guarantee bandwidth for
the reset preration. The side effect is that we make semaphore
completion latencies also 1us longer.

v2: Let full reset handle the adjacent engine idling (Chris)
v3: Skip render engine (Joonas), please checkpatch on define (Mika)

References: https://bugs.freedesktop.org/show_bug.cgi?id=106684
References: VTHSD#2227190, HSDES#1604216706, BSID#0917
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180607172444.17080-1-mika.kuoppala@linux.intel.com

drm/i915: Cancel reset preparations on failed resets

Our reset handling has a retry layer further up in the
chain. As we have told the engine to prepare for reset,
and failed it, make sure to remove that preparation so
that the next attempted reset has a clean slate by triggering
another full prepare cycle for the engines.

v2: ret as int, simplified cleanup (Chris)

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180605160357.32591-1-mika.kuoppala@linux.intel.com

drm/i915/gtt: Push allocation to hw ppgtt constructor

In the next patch, we will subclass the gen6 hw_ppgtt. In order, for the
two different generations of hw ppgtt stucts to be of different size,
push the allocation down to the constructor.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180607163040.9781-1-chris@chris-wilson.co.uk

drm/i915: Decouple vma vfuncs from vm

To allow for future non-object backed vma, we need to be able to
specialise the callbacks for binding, et al, the vma. For example,
instead of calling vma->vm->bind_vma(), we now call
vma->ops->bind_vma(). This gives us the opportunity to later override the
operation for a custom vma.

v2: flip order of unbind/bind

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180607154047.9171-2-chris@chris-wilson.co.uk

drm/i915: Prepare for non-object vma

In order to allow ourselves to use VMA to wrap other entities other than
GEM objects, we need to allow for the vma->obj backpointer to be NULL.
In most cases, we know we are operating on a GEM object and its vma, but
we need the core code (such as i915_vma_pin/insert/bind/unbind) to work
regardless of the innards.

The remaining eyesore here is vma->obj->cache_level and related (but
less of an issue) vma->obj->gt_ro. With a bit of care we should mirror
those on the vma itself.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180607154047.9171-1-chris@chris-wilson.co.uk

drm/i915: Change i915_gem_fault() to return vm_fault_t

In preparation for vm_fault_t becoming a distinct type, convert the
fault handler (i915_gem_fault()) over to the new interface.

Based on a patch by Souptick Joarder

References: 1c8f422059ae ("mm: change return type to vm_fault_t")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Souptick Joarder <jrdr.linux@gmail.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180606214520.20220-1-chris@chris-wilson.co.uk

drm/i915: Use GEM suspend when aborting initialisation

As part of our GEM initialisation now, we send a request to the hardware
in order to record the initial GPU state. This coupled with deferred
idle workers, makes aborting on error tricky. We already have the
mechanism in place to wait on the GPU and cancel all the deferred
workers for suspend, so let's reuse it during the error teardown. It is
already used in places for later init error handling, but doing so at
this point is slightly ugly due to the mutex dance (it's ok, the module
load is still single threaded).

Testcase: igt/drv_module_reload/basic-reload-inject
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180606145441.4460-1-chris@chris-wilson.co.uk

drm/i915: Update DRIVER_DATE to 20180606

Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/i915/gtt: Fix typo in fill_px() macro

The macro declared the ppgtt parameter but implicitly used the local vm
instead.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180606205128.25952-1-chris@chris-wilson.co.uk

drm/i915: Mark i915.inject_load_failure as being hit

When we reach the magic value and do inject a fault into our module load,
mark the module option as being hit. Since we fail from inside pci
probe, the module load isn't actually aborted and the module (and
parameters) are left lingering. igt can then inspect the parameter on its
synchronous completion of modprobe to see if the fault injection was
successful, and will keeping on injecting new faults until the module
succeeds in loading having surpassed the number of fault points.

v2: Reset to 0 after being hit;

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180606144153.4244-1-chris@chris-wilson.co.uk

drm/i915/gtt: Rename i915_hw_ppgtt base member

In the near future, I want to subclass gen6_hw_ppgtt as it contains a
few specialised members and I wish to add more. To avoid the ugliness of
using ppgtt->base.base, rename the i915_hw_ppgtt base member
(i915_address_space) as vm, which is our common shorthand for an
i915_address_space local.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180605153758.18422-1-chris@chris-wilson.co.uk

drm/i915/error: Fixup inactive/active counting

The inactive counter was over the active list, and vice versa.
Fortuitously this should not cause a problem in practice as they shared
the same array and clamped the number of entries they would write.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180605160623.30163-1-chris@chris-wilson.co.uk

drm/i915/guc: Don't leak stage descriptor pool on init failure

In case of failure during GuC clients creation, we forget to
cleanup earlier pool allocation. Use proper teardown to fix that.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michal Winiarski <michal.winiarski@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180605120547.16468-1-michal.wajdeczko@intel.com

drm/i915/pmu: Do not assume fixed hrtimer period

As Chris has discovered on his Ivybridge, and later automated test runs
have confirmed, on most of our platforms hrtimer faced with heavy GPU load
can occasionally become sufficiently imprecise to affect PMU sampling
calculations.

This means we cannot assume sampling frequency is what we asked for, but
we need to measure the interval ourselves.

This patch is similar to Chris' original proposal for per-engine counters,
but instead of introducing a new set to work around the problem with
frequency sampling, it swaps around the way internal frequency accounting
is done. Instead of accumulating current frequency and dividing by
sampling frequency on readout, it accumulates frequency scaled by each
period.

v2:
 * Typo in commit message, comment on period calculation and USEC_PER_SEC.
   (Chris Wilson)

Testcase: igt/perf_pmu/*busy* # snb, ivb, hsw
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180605140253.3541-1-tvrtko.ursulin@linux.intel.com

drm/i915/trace: Context field needs to be 64-bit wide

Underlaying field is u64 so the tracepoint needs to be as well.

v2:
 * Re-order binary packet for 64-bit alignment. (Chris Wilson)

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180605134124.25672-1-tvrtko.ursulin@linux.intel.com

drm/i915/trace: Remove engine out of the context sandwich

In the string tracepoint representation we ended up with the engine
sandwiched between context hardware id and context fence id.

Move the two pieces of context data together for redability.

Binary records are left as is, that is both fields remaing under the
existing name and ordering.

v2:
 * Do not consolidate the printk format, just reorder. (Lionel)

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180525082642.18246-2-tvrtko.ursulin@linux.intel.com

drm/i915/trace: Describe engines as class:instance pairs

Instead of using the engine->id, use uabi_class:instance pairs in trace-
points including engine info.

This will be more readable, more future proof and more stable for
userspace consumption.

v2:
 * Use u16 for class and instance. (Chris Wilson)

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: svetlana.kukanova@intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180525082642.18246-1-tvrtko.ursulin@linux.intel.com

drm/i915: Swap magics and use SZ_1M

Since the kernel provides SZ_1M, use it in preference of 1 << 20.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180605135746.8020-1-chris@chris-wilson.co.uk

drm/i915: Correctly handle error path in i915_gem_init_hw

In function gem_init_hw() we are calling uc_init_hw() but in case
of error later in function, we missed to call matching uc_fini_hw()

v2: pulled out from the series

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180605122443.23776-1-michal.wajdeczko@intel.com

drm/i915/vma: Move the bind_count vs pin_count assertion to a helper

To spare ourselves a long line later, refactor the repeated check of
bind_count vs pin_count to a helper.

v2: Fix up the commentary!

Suggested-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180605094107.31367-1-chris@chris-wilson.co.uk

drm/i915/gtt: Teach restore-gtt to walk the ggtt vma list not the object list

In preparation, for having non-vma objects stored inside the ggtt, to
handle restoration of the GGTT following resume, we need to walk over
the ggtt address space rebinding vma, as opposed to walking over bound
objects looking for ggtt entries.

v2: Skip objects only bound for the aliasing_ppgtt

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> #v1
Link: https://patchwork.freedesktop.org/patch/msgid/20180605082856.19221-1-chris@chris-wilson.co.uk

drm/i915/ringbuffer: Make context pin/unpin symmetric

Currently, we have a special routine for pinning the context state at
the start of activity tracking, but lack the complementary unpin
routine. Create it to to ease later patches that want to do partial
teardown on error, and, not least, to improve the readability of the
code.

Suggested-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180605085348.3018-1-chris@chris-wilson.co.uk

drm/i915: Add new vGPU cap info bit VGT_CAPS_HUGE_GTT

This adds a new vGPU cap info bit VGT_CAPS_HUGE_GTT, which is to detect
whether the host supports shadowing of huge gtt pages. If host does
support it, remove the page sizes restriction for vGPU.

Signed-off-by: Changbin Du <changbin.du@intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1525770425-5373-1-git-send-email-changbin.du@intel.com

drm/i915/icl: Don't update enabled dbuf slices struct until updated in hw

Do not update number of enabled dbuf slices in dev_priv struct until we
actually enable/disable dbuf slice in hw. This is leading to never
updating dbuf slices and resulting in DBuf slice mismatch warning.

Fixes: aa9664ffe863 ("drm/i915/icl: Enable 2nd DBuf slice only when needed")
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Mahesh Kumar <mahesh1.kumar@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180517132626.5885-1-mahesh1.kumar@intel.com

drm/i915/perf: fix ctx_id read with GuC & ICL

One thing we didn't really understand about the OA report is that the
ContextID field (dword 2) is copy of the context descriptor (dword 1).

On Gen8->10 and without using GuC we didn't notice the issue because
we only checked the 21bits of the ContextID field in the OA reports
which matches exactly the hw_id stored into the context descriptor.

When using GuC submission we have an issue of a non matching hw_id
because GuC uses bit 20 of the hw_id to signal proxy submission. This
change introduces a mask to compare only the relevant bits.

On ICL the context descriptor format has changed and we failed to
address this. On top of using a mask we also need to shift the bits
properly.

v2: Reuse lrc_desc rather than recomputing part of it (Chris/Michel)

v3: Always pin the context we're filtering with (Chris)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 1de401c08fa805 ("drm/i915/perf: enable perf support on ICL")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104252
BSpec: 1237
Testcase: igt/perf/gen8-unprivileged-single-ctx-counters
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180602112946.30803-3-lionel.g.landwerlin@intel.com
Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: intel-gfx@lists.freedesktop.org

drm/i915: drop one bit on the hw_id when using guc

We currently using GuC as a proxy to the hardware. When Guc is used in
such mode, it consumes the bit 20 of the hw_id to indicate that the
workload was submitted by proxy.

So far we probably haven't seen the issue because we need to allocate
1048576+ contexts to hit this issue. Still, we should avoid allocating
the hw_id on that bit and restriction to bits [0:19] (i.e 20bits
instead of 21).

v2: Leave the max hw_id computation in i915_gem_context.c (Michel)

v3: Be consistent on if/else usage (Chris)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
BSpec: 1237
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180602112946.30803-2-lionel.g.landwerlin@intel.com

drm/i915/gtt: Remove obsolete switch_mm hooks for gen8+

As the ppgtt for execlists is tightly coupled to the executing context,
and not switch separately, we no longer use the ppgtt->switch_mm hooks
on gen8+. Remove them.

References: 79e6770cb1f5 ("drm/i915: Remove obsolete ringbuffer emission for gen8+")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180604131552.29370-1-chris@chris-wilson.co.uk

drm/i915: Move i915_gem_fini to i915_gem.c

We should keep i915_gem_init/fini functions together for easier
tracking of their symmetry.

v2: rebased, pulled out from the series

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180604090032.20840-1-michal.wajdeczko@intel.com

drm/i915/icl: Calculate link clock using the new registers

Start using the new registers for ICL and on.

Cc: Manasi Navare <manasi.d.navare@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180522002558.29262-13-paulo.r.zanoni@intel.com

drm/i915/icl: Get DDI clock for ICL based on PLLs.

PLLs are the source clocks for the DDIs so in order
to determine the ddi clock we need to check the PLL
configuration.

This gets a little tricky for ICL since there is
no register bit that maps directly to the link clock.
So this patch creates a separate function in intel_dpll_mgr.c
to obtain the write array PLL Params and compares the set
pll_params with the table to get the corresponding link
clock.

v2:
  - Fix the encoder type check (DK).
  - Improve our error checking, return a sane value (Mika, Paulo).
  - Fix table entries (Paulo).

Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Mika Kahola <mika.kahola@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Signed-off-by: Manasi Navare <manasi.d.navare@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
[Paulo: implement v2]
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180523224444.19017-1-paulo.r.zanoni@intel.com

drm/i915/icl: Add Icelake PCH detection

This patch adds the support to detect PCH_ICP.

Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Suggested-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Signed-off-by: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180522002558.29262-10-paulo.r.zanoni@intel.com

drm/i915/icl: Map VBT DDC Pin to BSpec DDC Pin

On ICL we need to map VBT DDC Pin to BSpec DDC Pin.
Adding ICL Pin Values.

According to VBT
Block 2 (General Bytes Definition)
DDC Bus

+----------+-----------+--------------------+
| DDI Type | VBT Value | BSpec Mapped Value |
+----------+-----------+--------------------+
| DDI-A    | 0x1       | 0x1                |
| DDI-B    | 0x2       | 0x2                |
| PORT-1   | 0x4       | 0x9                |
| PORT-2   | 0x5       | 0xA                |
| PORT-3   | 0x6       | 0xB                |
| PORT-4   | 0x7       | 0xC                |
+----------+-----------+--------------------+

Cc: James Ausmus <james.ausmus@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Anusha Srivatsa <anusha.srivatsa@intel.com>
Cc: Clinton Taylor <clinton.a.taylor@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: James Ausmus <james.ausmus@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
[Paulo: checkpatch fixes.]
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180522002558.29262-9-paulo.r.zanoni@intel.com

drm/i915/icl: Add register definition for DFLEXDPMLE

DFLEXDPMLE register is required to tell the FIA hardware which
main links of DP are enabled on TCC Connectors. FIA uses this
information to program PHY to Controller signal mapping.
This register is applicable in both TC connector's Alternate mode
as well as DP connector mode.

v2:
* Remove _ICL prefix since the reg is first introduced
in ICL (Paulo)
* s/ICL/icl in commit message (Lucas)

Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Animesh Manna <animesh.manna@intel.com>
Cc: Madhav Chauhan <madhav.chauhan@intel.com>
Cc: Anusha Srivatsa <anusha.srivatsa@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Manasi Navare <manasi.d.navare@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1527275032-4555-1-git-send-email-manasi.d.navare@intel.com

drm/i915/icl: introduce tc_port

Add and enum for TC ports and auxiliary functions to handle them.
Icelake brings a lot of registers and other things that only apply to
the TC ports and are indexed starting from 0, so having an enum for
tc_ports that starts at 0 really helps the indexing.

This patch is based on previous patches written by Dhinakaran Pandiyan
and Mahesh Kumar.

Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Cc: Mahesh Kumar <mahesh1.kumar@intel.com>
Reviewed-by: Mahesh Kumar <mahesh1.kumar@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180522002558.29262-4-paulo.r.zanoni@intel.com

drm/i915/icl: Extend AUX F interrupts to ICL

ICL has AUX F.

Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Anusha Srivatsa <anusha.srivatsa@intel.com>
Reviewed-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180522002558.29262-2-paulo.r.zanoni@intel.com

drm/i915/icl: fix icl_unmap/map_plls_to_ports

All connectors may not have best_encoder attached, so don't dereference
encoder pointer for each connector.

Fixes: c27e917e2bda ("drm/i915/icl: add basic support for the ICL clocks")
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Mahesh Kumar <mahesh1.kumar@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180525155238.7054-1-lucas.demarchi@intel.com

drm/i915: Apply the full CPU domain markup before freezing

Let's not take any chances by using a shortcut to mark the objects as in
the CPU domain upon freezing (all pages will be written to disk and so
on restore all objects will start from the CPU domain). Currently, we
simply mark the objects as being in the CPU domain, bypassing the
flushes. Let's call the full domain transfer function so that we have
less special case code (and symmetry with the suspend path) even though
it will be mostly redundant.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180601144125.18026-2-chris@chris-wilson.co.uk

drm/i915: Flush all writes before suspend

As we have already suspended the device, this should be a no-op except
for marking that all writes are indeed complete. The downside is that
we then have to walk all the lists of objects for what should be a no-op
(in some cases they will be mmio read to ensure the GGTT writes are
indeed flushed, and clflushes to ensure that cpu writes are in memory).

It seems prudent and the safer course for us to ensure all writes are
flushed to memory before suspend.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180601144125.18026-1-chris@chris-wilson.co.uk

drm/i915: Configure SKL+ scaler initial phase correctly

Set up the SKL+ scaler initial phase registers correctly. Otherwise
we start fetching the data from the center of the first pixel instead
of the top-left corner, which obviously then leads to right/bottom edges
replicating data excessively as the data runs out half a pixel too soon.

Cc: Vidya Srinivas <vidya.srinivas@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180521185613.5097-2-ville.syrjala@linux.intel.com
Reviewed-By: Vidya Srinivas <vidya.srinivas@intel.com>

drm/i915: Remove bogus NV12 PLANE_COLOR_CTL setup

We already handle the color encoding mode properly. Remove the broken
NV12 special case.

Cc: Vidya Srinivas <vidya.srinivas@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Fixes: 8ed30ab6aced ("drm/i915: Enable YUV to RGB for Gen10 in Plane Ctrl Reg")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180521185613.5097-1-ville.syrjala@linux.intel.com
Reviewed-By: Vidya Srinivas <vidya.srinivas@intel.com>

drm/i915: Fix tabs vs. spaces in sprite code

The sprite code has a bunch of spaces where tabs should be used. Fix it
up.

v2: Make the patch subject more specific (Jani)

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180530165933.11424-3-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915: s/plane/i9xx_plane/

Call the enum i9xx_plane_id variable i9xx_plane like we do elsewhere.

Cc: Hans de Goede <j.w.r.degoede@gmail.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180130203807.13721-7-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>

drm/i915: Disable trickle feed for SNB/IVB cursors

We disable trickle feed whenever possible, except for the cursors
on SNB/IVB. Let's try disabling it there too if for no other reason
than consistency.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180130203807.13721-5-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>

drm/i915: Clean up cursor defines

Use MCURSOR_ instead of CURSOR_ as the prefix for the non-845/865
cursor defines consistently, and move the pipe CSC enable bit next
to the other non-845/865 cursor defines.

v2: Take care of gvt uses as well
v3: Another gvt use popped up

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180131143709.875-1-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com> #v2

drm/i915: Have plane->get_hw_state() return the current pipe

Like we do for encoder let's make the plane->get_hw_state() return
the pipe to which the plane is currently attached. We don't currently
allow planes to move between the pipes, but perhaps one day we will.

In either case this makes the code more uniform and perhaps makes
intel_plane_mapping_ok() slightly more clear.

Note that for i965 and g4x planes A and B still have pipe select bits
but they're hardwired to pipe A and B respectively. This means we can
safely interpret those bits just like on gen2/3. This allows the
same readout code work for plane C (which can still be assigned
to eiter pipe on i965) should we ever expose it.

g4x no longer allows moving the cursor planes between the pipes,
but the pipe select bits can still be set in the register. Thus
we have to ignore those bits. OTOH i965 still allows the cursors
to move between pipes thus we have to trust the bits there.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180130203807.13721-3-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>

drm/i915/gtt: Don't restore the non-existent PDE for GGTT

On resume, we have to rewrite all the PDE entries for gen7 ppgtts. If we
switch on full-ppgtt, there is then one address space with no PDE, the
GGTT. Currently under aliasing-ppgtt, the GGTT address space does have
an associated ppgtt and so the restore works just fine. We would have a
similar problem if we tried disabling aliasing-ppgtt
(i915.enable_ppgtt=0). So skip the empty ppgtt, as being non-existent it
doesn't need restoring.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180601093554.13083-2-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/i915/gtt: Avoid calling non-existent allocate_va_range

On hsw and older, we do not need to allocate the ppgtt on the fly and so
ppgtt->allocate_va_range() is NULL. Fixup ppgtt_bind_vma not to call it,
in that case!

v2: PIN_UPDATE still exists.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180601093554.13083-1-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180601093554.13083-2-chris@chris-wilson.co.uk

drm/i915: Check intel_contexts to avoid one extra pointer chase

As we store the intel_context on the request (rq->hw_context), we can
simply compare that against the local intel_context for the
i915->kernel_context rather than using the rq->gem_context.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180601094002.13329-1-chris@chris-wilson.co.uk

drm/i915: Assert we idle in the kernel context

Now that we always switch to the kernel context upon idling, we can
make that assertion.

References: 4dfacb0bcbee ("drm/i915: Switch to kernel context before idling at runtime")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180531224057.6036-1-chris@chris-wilson.co.uk

drm/i915: Only sanitize GEM from late suspend

During testing we encounter a conflict between SUSPEND_TEST_DEVICES and
disabling reset (gem_eio/suspend). This results in the device continuing
on without being reset, but since it has gone through HW sanitization to
account for the suspend/resume cycle, we have to assume the device has
been reset to its defaults. A simple way around this is to skip the
sanitize phase for SUSPEND_TEST_DEVICES by moving it to suspend-late.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180531082246.9763-4-chris@chris-wilson.co.uk

drm/i915: After reset on sanitization, reset the engine backends

As we reset the GPU on suspend/resume, we also do need to reset the
engine state tracking so call into the engine backends. This is
especially important so that we can also sanitize the state tracking
across resume.

References: https://bugs.freedesktop.org/show_bug.cgi?id=106702
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180531082246.9763-3-chris@chris-wilson.co.uk

drm/i915: "Race-to-idle" after switching to the kernel context

During suspend we want to flush out all active contexts and their
rendering. To do so we queue a request from the kernel's context, once
we know that request is done, we know the GPU is completely idle. To
speed up that switch bump the GPU clocks.

Switching to the kernel context prior to idling is also used to enforce
a barrier before changing OA properties, and when evicting active
rendering from the global GTT. All cases where we do want to
race-to-idle.

v2: Limit the boosting to only the switch before suspend.
v3: Limit it to the wait-for-idle on suspend.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: David Weinehall <david.weinehall@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Tested-by: David Weinehall <david.weinehall@linux.intel.com> #v1
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180531082246.9763-2-chris@chris-wilson.co.uk

drm/i915: Switch to kernel context before idling at runtime

We can reduce our exposure to random neutrinos by resting on the kernel
context having flushed out the user contexts to system memory and
beyond. The corollary is that we then we require two passes through the
idle handler to go to sleep, which on a truly idle system involves an
extra pass through the slow and irregular retire work handler.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180531082246.9763-1-chris@chris-wilson.co.uk

drm/i915/guc: Don't read SOFT_SCRATCH(15) on MMIO error

SOFT_SCRATCH(15) is used by GuC for sending MMIO GuC events to host and
those events are now handled by intel_guc_to_host_event_handler_mmio().

We should not try to read it on MMIO action error as 1) we may be using
different set of registers for GuC MMIO communication, and 2) GuC may
use CTB mechanism for sending events to host.

While here, upgrade error message to DRM_ERROR.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180528171618.10436-1-michal.wajdeczko@intel.com

drm/i915: Call intel_opregion_notify_encoder in intel_sanitize_encoder, v2.

Normally this is called on a modeset, but the call is missing when
we inherit the mode from the BIOS, so make sure it's called somewhere
in hardware readout.

Changes since v1:
- Unconditionally call intel_opregion_notify_encoder. (Ville)

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180516085038.36785-1-maarten.lankhorst@linux.intel.com
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>

drm/i915: Update DRIVER_DATE to 20180530

Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/i915: Remove stale asserts from i915_gem_find_active_request()

Since we use i915_gem_find_active_request() from inside
intel_engine_dump() and may call that at any time, we do not guarantee
that the engine is paused nor that the signal kthreads and irq handler
are suspended, so we cannot assert that the breadcrumb doesn't advance
and that the irq hasn't happened on another CPU signaling the request we
believe to be idle.

The second assert removed (that request->engine == engine) remains
valid, but is now more rigorously checked during retirement.

Fixes: f636edb214a5 ("drm/i915: Make i915_engine_info pretty printer to standalone")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180529132922.6831-1-chris@chris-wilson.co.uk
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

drm/i915/psr: Set idle frame count based on sink synchronization latency

DPCD 2009h "Synchronization latency in sink" has bits that tell us the
maximum number of frames sink can take to resynchronize to source timing
when exiting PSR. More importantly, as per eDP 1.4b, this is the "Minimum
number of frames following PSR exit that the Source device needs to
wait for PSR entry."

We currently use this value only to setup the number frames to wait before
PSR2 selective update. But, based on the above description it makes more
sense to use this to configure idle frames for both PSR1 and and PSR2. This
will ensure we wait the required number of frames before
activation whether it is PSR1 or PSR2.

The minimum number of idle frames remains 6, while allowing sink
synchronization latency and VBT to increase this value.

This also solves the flip-flop between sink and source frames that I
noticed on my Thinkpad X260 during PSR exit. This specific panel has a
value of 8h, which according to the spec means the "Source device must
wait for more than eight active frames after PSR exit before initiating PSR
entry. (In this case, should be provided by the panel supplier.)" VBT
however has a value of 0.

Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Jose Roberto de Souza <jose.souza@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180525033047.7596-1-dhinakaran.pandiyan@intel.com

drm/i915/icl: Wa_1406463099

Prevents an error in the GAM unit. Also known as WaGamTlbPendError

References: HSDES#1406463099
References: HSDES#1406465643
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1527285939-20113-12-git-send-email-oscar.mateo@intel.com

drm/i915/icl: WaEnableFloatBlendOptimization

Enables blend optimization for floating point RTs

v2: Rebased on top of the WA refactoring
v3: Added References (Mika)

References: HSDES#1406393558
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1527285939-20113-5-git-send-email-oscar.mateo@intel.com

drm/i915/icl: Wa_2006665173

Disable blend embellishment in RCC.

Also, some other registers style fixed in passing.

v2: Rebased on top of the WA refactoring
v3: Added References (Mika)
v4:
  - Fixed in B0
  - Mentioned style fixes in commit message

References: HSDES#2006665173
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1527285939-20113-4-git-send-email-oscar.mateo@intel.com

drm/i915/icl: WaEnableStateCacheRedirectToCS

Redirects the state cache to the CS Command buffer section for
performance reasons.

v2: Rebased
v3: Rebased on top of the WA refactoring
v3: Added References (Mika)

References: HSDES#1604325460
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1527285939-20113-3-git-send-email-oscar.mateo@intel.com

drm/i915/icl: WaDisableImprovedTdlClkGating

Revert to the legacy implementation.

v2: GEN7_ROW_CHICKEN2 is masked
v3:
  - Rebased
  - Renamed to Wa_2006611047
  - A0 and B0 only
v4:
  - Add spaces around '<<' (and fix the surrounding code as well)
  - Mark the WA as pre-prod
v5: Rebased on top of the WA refactoring
v6: Added References (Mika)
v7: Fixed in B0

References: HSDES#2006611047
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1527285939-20113-2-git-send-email-oscar.mateo@intel.com

drm/i915/uc: Trivial s/dev_priv/i915 in intel_uc.c

Some functions already use i915 name instead of dev_priv.
Let's rename this param in all remaining functions, except
those that still use legacy macros.

v2: don't forget about function descriptions (Sagar)
v3: rebased
v4: rebased
v5: rebased, pulled out from the series

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180525121858.53928-1-michal.wajdeczko@intel.com

drm/i915: Simplify ilk-ivb underrun suppression

Let's suppress the underruns around every modeset sequence instead
of trying to avoid it. Planes are disabled at this point anyway so
we don't really gain anything from keeping the underrun reporting
enabled. Also for PCH ports we already suppress all underruns here
anyway so trying avoid it for the CPU eDP doesn't seem all that
important.

Maybe this gets rid of some lingering spurious underruns?

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180524190406.2973-2-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>

drm/i915: Consult VBT "LVDS config" bits to determine whether internal LVDS is present

VBT seems to have some bits to tell us whether the internal LVDS port
has something hooked up. In theory one might expect the VBT to not have
a child device for the LVDS port if there's no panel hooked up, but
in practice many VBTs still add the child device. The "LVDS config" bits
seem more reliable though, so let's check those.

So far we've used the "LVDS config" bits to check for eDP support on
ILK+, and disable the internal LVDS when the value is 3. That value
is actually documented as "Both internal LVDS and SDVO LVDS", but in
practice it looks to mean "eDP" on all the ilk+ VBTs I've seen. So let's
keep that interpretation, but for pre-ILK we will consider the value
3 to also indicate the presence of the internal LVDS.

Currently we have 25 DMI matches for the "no internal LVDS" quirk. In an
effort to reduce that let's toss in a WARN when the DMI match and VBT
both tell us that the internal LVDS is not present. The hope is that
people will report a bug, and then we can just nuke the corresponding
entry from the DMI quirk list. Credits to Jani for this idea.

v2: Split the basic int_lvds_support thing to a separate patch (Jani)
v3: Rebase
v4: Limit this to VBT version >= 134

Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180518150138.18361-1-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915: Try to suppress more spurious PCH underruns on ILK-IVB

My ILK seems to generate a spurious PCH underrun with most interlaced
HDMI modes. Add a second vblank wait to avoid it.

We have seen some spurious PCH underruns still in CI as well, some
of which seem to be progressive DP. The logs also point towards some
spurious underrins with progressive HDMI on SNB. While I don't have
a solid explanation for those let's try to kill all the birds with one
stone and always do the double wait.

Buzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106387
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180524190406.2973-1-ville.syrjala@linux.intel.com
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>

drm/i915: Initialize panel_pipe to INVALID_PIPE

We can always figure out which pipe is affected by the panel power
sequencer lockout mechanism. So no need for the pipe A fallback
anymore. The only case we may have to worry about is an invalid
port select in the power sequencer, but INVALID_PIPE is just fine
in that case. We'll get the WARN about the bogus pps port select
anyway.

Cc: Jani Nikula <jani.nikula@intel.com>
Suggested-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180523145718.22932-1-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915: Prepare GEM for suspend earlier

In order to prepare the GPU for sleeping, we may want to submit commands
to it. This is a complicated process that may even require some swapping
in from shmemfs, if the GPU was in the wrong state. As such, we need to
do this preparation step synchronously before the rest of the system has
started to turn off (e.g. swapin fails if scsi is suspended).
Fortunately, we are provided with a such a hook, pm_ops.prepare().

v2: Compile cleanup
v3: Fewer asserts, fewer problems?

v4: Ville pointed out that in some circumstances (such as switching off
the overlay) the display code may issue a GPU request. This is
unexpected, and will result in us going to sleep with us believing the
GPU is still awake (though all user work has been saved). Add a comment
to remind our future selves of what trouble brews.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106640
Testcase: igt/drv_suspend after igt/gem_tiled_swapping
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180525092629.1456-1-chris@chris-wilson.co.uk
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>

drm/i915/execlists: Wait for ELSP submission on restart

After a reset, we will ensure that there is at least one request
submitted to HW to ensure that a context is loaded for powersaving.
Let's wait for this submission via a tasklet to complete before we drop
our forcewake, ensuring the system is ready for rc6 before we let it
possibly sleep.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180522101937.7738-1-chris@chris-wilson.co.uk
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>

drm/i915: Flush the ring stop bit after clearing RING_HEAD in reset

Inside the live_hangcheck (reset) selftests, we occasionally see
failures like

<7>[  239.094840] i915_gem_set_wedged rcs0
<7>[  239.094843] i915_gem_set_wedged  current seqno 19a98, last 19a9a, hangcheck 0 [5158 ms]
<7>[  239.094846] i915_gem_set_wedged  Reset count: 6239 (global 1)
<7>[  239.094848] i915_gem_set_wedged  Requests:
<7>[  239.095052] i915_gem_set_wedged  first  19a99 [e8c:5f] prio=1024 @ 5159ms: (null)
<7>[  239.095056] i915_gem_set_wedged  last   19a9a [e81:1a] prio=139 @ 5159ms: igt/rcs0[5977]/1
<7>[  239.095059] i915_gem_set_wedged  active 19a99 [e8c:5f] prio=1024 @ 5159ms: (null)
<7>[  239.095062] i915_gem_set_wedged  [head 0220, postfix 0280, tail 02a8, batch 0xffffffff_ffffffff]
<7>[  239.100050] i915_gem_set_wedged  ring->start:  0x00283000
<7>[  239.100053] i915_gem_set_wedged  ring->head:   0x000001f8
<7>[  239.100055] i915_gem_set_wedged  ring->tail:   0x000002a8
<7>[  239.100057] i915_gem_set_wedged  ring->emit:   0x000002a8
<7>[  239.100059] i915_gem_set_wedged  ring->space:  0x00000f10
<7>[  239.100085] i915_gem_set_wedged  RING_START: 0x00283000
<7>[  239.100088] i915_gem_set_wedged  RING_HEAD:  0x00000260
<7>[  239.100091] i915_gem_set_wedged  RING_TAIL:  0x000002a8
<7>[  239.100094] i915_gem_set_wedged  RING_CTL:   0x00000001
<7>[  239.100097] i915_gem_set_wedged  RING_MODE:  0x00000300 [idle]
<7>[  239.100100] i915_gem_set_wedged  RING_IMR: fffffefe
<7>[  239.100104] i915_gem_set_wedged  ACTHD:  0x00000000_0000609c
<7>[  239.100108] i915_gem_set_wedged  BBADDR: 0x00000000_0000609d
<7>[  239.100111] i915_gem_set_wedged  DMA_FADDR: 0x00000000_00283260
<7>[  239.100114] i915_gem_set_wedged  IPEIR: 0x00000000
<7>[  239.100117] i915_gem_set_wedged  IPEHR: 0x02800000
<7>[  239.100120] i915_gem_set_wedged  Execlist status: 0x00044052 00000002
<7>[  239.100124] i915_gem_set_wedged  Execlist CSB read 5 [5 cached], write 5 [5 from hws], interrupt posted? no, tasklet queued? no (enabled)
<7>[  239.100128] i915_gem_set_wedged  ELSP[0] count=1, ring->start=00283000, rq: 19a99 [e8c:5f] prio=1024 @ 5164ms: (null)
<7>[  239.100132] i915_gem_set_wedged  ELSP[1] count=1, ring->start=00257000, rq: 19a9a [e81:1a] prio=139 @ 5164ms: igt/rcs0[5977]/1
<7>[  239.100135] i915_gem_set_wedged  HW active? 0x5
<7>[  239.100250] i915_gem_set_wedged  E 19a99 [e8c:5f] prio=1024 @ 5164ms: (null)
<7>[  239.100338] i915_gem_set_wedged  E 19a9a [e81:1a] prio=139 @ 5164ms: igt/rcs0[5977]/1
<7>[  239.100340] i915_gem_set_wedged  Queue priority: 139
<7>[  239.100343] i915_gem_set_wedged  Q 0 [e98:19] prio=132 @ 5164ms: igt/rcs0[5977]/8
<7>[  239.100346] i915_gem_set_wedged  Q 0 [e84:19] prio=121 @ 5165ms: igt/rcs0[5977]/2
<7>[  239.100349] i915_gem_set_wedged  Q 0 [e87:19] prio=82 @ 5165ms: igt/rcs0[5977]/3
<7>[  239.100352] i915_gem_set_wedged  Q 0 [e84:1a] prio=44 @ 5164ms: igt/rcs0[5977]/2
<7>[  239.100356] i915_gem_set_wedged  Q 0 [e8b:19] prio=20 @ 5165ms: igt/rcs0[5977]/4
<7>[  239.100362] i915_gem_set_wedged  drv_selftest [5894] waiting for 19a99

where the GPU saw an arbitration point and idles; AND HAS NOT BEEN RESET!
The RING_MODE indicates that is idle and has the STOP_RING bit set, so
try clearing it.

v2: Only clear the bit on restarting the ring, as we want to be sure the
STOP_RING bit is kept if reset fails on wedging.
v3: Spot when the ring state doesn't make sense when re-initialising the
engine and dump it to the logs so that we don't have to wait for an
error later and try to guess what happened earlier.
v4: Prepare to print all the unexpected state, not just the first.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180518100933.2239-1-chris@chris-wilson.co.uk

drm/i915: Forward declare struct intel_context

This is to avoid an error with structure declared in parameter list if the
include ordering changes.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180524150621.17332-2-tvrtko.ursulin@linux.intel.com

drm/i915: Include i915_scheduler.h from i915_gem_context.h

struct i915_gem_context embeds struct i915_sched_attr so needs to include
the respective header.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180524150621.17332-1-tvrtko.ursulin@linux.intel.com

drm/i915: Look for an active kernel context before switching

We were not very carefully checking to see if an older request on the
engine was an earlier switch-to-kernel-context before deciding to emit a
new switch. The end result would be that we could get into a permanent
loop of trying to emit a new request to perform the switch simply to
flush the existing switch.

What we need is a means of tracking the completion of each timeline
versus the kernel context, that is to detect if a more recent request
has been submitted that would result in a switch away from the kernel
context. To realise this, we need only to look in our syncmap on the
kernel context and check that we have synchronized against all active
rings.

v2: Since all ringbuffer clients currently share the same timeline, we do
have to use the gem_context to distinguish clients.

As a bonus, include all the tracing used to debug the death inside
suspend.

v3: Test, test, test. Construct a selftest to exercise and assert the
expected behaviour that multiple switch-to-contexts do not emit
redundant requests.

Reported-by: Mika Kuoppala <mika.kuoppala@intel.com>
Fixes: a89d1f921c15 ("drm/i915: Split i915_gem_timeline into individual timelines")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180524081135.15278-1-chris@chris-wilson.co.uk

drm/i915/psr: Fix ALPM cap check for PSR2

While touching the code around this, I noticed that absence of ALPM
capability does not stop us from enabling PSR2. But, the spec
unambiguously states that ALPM is required for PSR2 and so does this
commit that introduced this code

drm/i915/psr: enable ALPM for psr2

    As per edp1.4 spec , alpm is required for psr2 operation as it's
    used for all psr2  main link power down management and alpm enable
    bit must be set for psr2 operation.

Cc: Jose Roberto de Souza <jose.souza@intel.com>
Cc: Vathsala Nagaraju <vathsala.nagaraju@intel.com>
Reviewed-by: Vathsala Nagaraju <vathsala.nagaraju@intel.com>
Reviewed-by: Tarun Vyas <tarun.vyas@intel.com>
Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180511195145.3829-6-dhinakaran.pandiyan@intel.com

drm/i915/psr: Fall back to max. synchronization latency if DPCD read fails

Noticed that we assume the best case of 0 latency when the DPCD read
fails, reasonable pessimism is safer.

eDP spec does say that if latency is greater than 8, the panel
supplier needs to provide it. I didn't see anything specific in the VBT
for this, so let's go with 8 frames as a fallback.

Cc: Jose Roberto de Souza <jose.souza@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180511195145.3829-5-dhinakaran.pandiyan@intel.com

drm/i915/psr: Avoid unnecessary DPCD read of DP_PSR_CAPS

intel_dp->psr_dpcd already has the required values.

Cc: Jose Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Tarun Vyas <tarun.vyas@intel.com>
Reviewed-by: Vathsala Nagaraju <vathsala.nagaraju@intel.com>
Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180511195145.3829-4-dhinakaran.pandiyan@intel.com

drm/i915/psr: Check for SET_POWER_CAPABLE bit at PSR init time.

By moving the check from psr_compute_config() to psr_init_dpcd(), we get
to set the dev_priv->psr.sink_support flag only when the panel is
capable of changing power state. An additional benefit is that the check
will be performed only at init time instead of every atomic_check.

This should change the psr_basic IGT failures on HSW to skips.

v2: Return early when SET_POWER_CAPABLE bit is 0 (Jose)

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106217
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106346
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
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/20180511195145.3829-2-dhinakaran.pandiyan@intel.com

drm/i915/psr: Avoid DPCD reads when panel does not support PSR

Ville noticed that we are unncessarily reading DPCD's after knowing
panel did not support PSR. Looks like this check that was present
earlier got removed unintentionally, let's put it back.

While we do this, add the PSR version number in the debug print.

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Tarun Vyas <tarun.vyas@intel.com>
Reviewed-by: Vathsala Nagaraju <vathsala.nagaraju@intel.com>
Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180511195145.3829-1-dhinakaran.pandiyan@intel.com

drm/i915/psr: Nuke PSR support for VLV and CHV

PSR hardware and hence the driver code for VLV and CHV deviates a lot from
their DDI counterparts. While the feature has been disabled for a long time
now, retaining support for these platforms is a maintenance burden. There
have been multiple refactoring commits to just keep the existing code for
these platforms in line with the rest. There are known issues that need to
be fixed to enable PSR on these platforms, and there is no PSR capable
platform in CI to ensure the code does not break again if we get around to
fixing the existing issues. On account of all these reasons, let's nuke
this code for now and bring it back if a need arises in the future.

Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180511230059.19387-1-dhinakaran.pandiyan@intel.com

drm/i915/psr: vbt change for psr

For psr block #9, the vbt description has moved to options [0-3] for
TP1,TP2,TP3 Wakeup time from decimal value without any change to vbt
structure. Since spec does not  mention from which VBT version this
change was added to vbt.bsf file, we cannot depend on bdb->version check
to change for all the platforms.

There is RCR inplace for GOP team to  provide the version number
to make generic change. Since Kabylake with bdb version 209 is having this
change, limiting this change to gen9_bc and version 209+ to unblock google.

Tested on skl(bdb version 203,without options) and
kabylake(bdb version 209,212) having new options.

bspec 20131

v2: (Jani and Rodrigo)
    move the 165 version check to intel_bios.c
v3: Jani
    Move the abstraction to intel_bios.
v4: Jani
    Rename tp*_wakeup_time to have "us" suffix.
    For values outside range[0-3],default to max 2500us.
    Old decimal value was wake up time in multiples of 100us.
v5: Jani and Rodrigo
    Handle option 2 in default condition.
    Print oustide range value.
    For negetive values default to 2500us.
v6: Jani
    Handle default first and then fall through for case 2.
v7: Rodrigo
    Apply this change for IS_GEN9_BC and vbt version > 209
v8: Puthik
    Add new function vbt_psr_to_us.
v9: Jani
    Change to v7 version as it's more readable.
    DK
    add comment /*fall through*/ after case2.

Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Puthikorn Voravootivat <puthik@chromium.org>
Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Maulik V Vaghela <maulik.v.vaghela@intel.com>
Signed-off-by: Vathsala Nagaraju <vathsala.nagaraju@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1526981243-2745-1-git-send-email-vathsala.nagaraju@intel.com

drm/i915: Implement WaProgramMgsrForL3BankSpecificMmioReads

L3Bank could be fused off in hardware for debug purpose, and it
is possible that subslice is enabled while its corresponding L3Bank pairs
are disabled. In such case, if MCR packet control register(0xFDC) is
programed to point to a disabled bank pair, a MMIO read into L3Bank range
will return 0 instead of correct values.

However, this is not going to be the case in any production silicon.
Therefore, we only check at initialization and issue a warning should
this really happen.

References: HSDES#1405586840

v2:
 - use fls instead of find_last_bit (Chris)
 - use is_power_of_2() instead of counting bit set (Chris)
v3:
 - rebase on latest tip
v5:
 - Added references (Mika)
 - Move local variable into scope where they are used (Ursulin)
 - use a new local variable to reduce long line of code (Ursulin)
v6:
 - Some coding style and use more local variables for clearer
   logic (Ursulin)

Cc: Oscar Mateo <oscar.mateo@intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Yunwei Zhang <yunwei.zhang@intel.com>
Reviewed-by: Oscar Mateo <oscar.mateo@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1526683285-24861-1-git-send-email-yunwei.zhang@intel.com

drm/i915/icl: Enable WaProgramMgsrForCorrectSliceSpecificMmioReads

WaProgramMgsrForCorrectSliceSpecificMmioReads applies for Icelake as
well.

References: HSD#1405586840, BSID#0575

v2:
 - GEN11 mask is different from its predecessors. (Oscar)
 - Better separate GEN10 and GEN11. (Oscar)

Cc: Oscar Mateo <oscar.mateo@intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Yunwei Zhang <yunwei.zhang@intel.com>
Reviewed-by: Oscar Mateo <oscar.mateo@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1526683232-24753-1-git-send-email-yunwei.zhang@intel.com

drm/i915/cnl: Implement WaProgramMgsrForCorrectSliceSpecificMmioReads

WaProgramMgsrForCorrectSliceSpecificMmioReads dictate that before any MMIO
read into Slice/Subslice specific registers, MCR packet control
register(0xFDC) needs to be programmed to point to any enabled
slice/subslice pair. Otherwise, incorrect value will be returned.

However, that means each subsequent MMIO read will be forwarded to a
specific slice/subslice combination as read is unicast. This is OK since
slice/subslice specific register values are consistent in almost all cases
across slice/subslice. There are rare occasions such as INSTDONE that this
value will be dependent on slice/subslice combo, in such cases, we need to
program 0xFDC and recover this after. This is already covered by
read_subslice_reg.

Also, 0xFDC will lose its information after TDR/engine reset/power state
change.

References: HSD#1405586840, BSID#0575

v2:
 - use fls() instead of find_last_bit() (Chris)
 - added INTEL_SSEU to extract sseu from device info. (Chris)
v3:
 - rebase on latest tip
v5:
 - Added references (Mika)
 - Change the ordered of passing arguments and etc. (Ursulin)
v7:
 - Moved WA explanation Comments(Oscar)
 - Rebased.
v8:
 - Renamed sanitize_mcr to calculate_s_ss_select. (Oscar)
 - calculate s/ss selector instead of whole mcr. (Oscar)
v9:
 - Updated function name (Oscar)
 - Remove redundant variables (Oscar)
v10:
 - Separate pre-GEN10 and GEN11 mask. (Oscar)

Cc: Oscar Mateo <oscar.mateo@intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Yunwei Zhang <yunwei.zhang@intel.com>
Reviewed-by: Oscar Mateo <oscar.mateo@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1526683197-24656-1-git-send-email-yunwei.zhang@intel.com

drm/i915/icl: Disable pipe CSC and gamma in cursor plane

'Pipe CSC enable' bit is more than just deprecated in ICL+, it was
disabled in commit 077ef1f09c25 ("drm/i915/icl: Don't set pipe
CSC/Gamma in PLANE_COLOR_CTL") for primary and sprite planes as it was
causing those planes to be rendered as always black but it was not
disabled in cursor plane, also causing it to be rendered as black.

As mentioned in the commit referenced above, this is a workaround
too and the CSC and gamma per plane values needs to be setup before
enable CSC and gamma again.

BSpec: 4278 and 7635

Cc: James Ausmus <james.ausmus@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180518201547.15793-1-jose.souza@intel.com

drm/i915/selftests: Pin the mock kernel context

The driver assumes that the kernel context is pinned and always
available for use from any process or atomic context. Make it so for
selftesting as well!

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180523142346.27953-1-chris@chris-wilson.co.uk