drm/i915: Call MG_DP_MODE() macro with the right parameters order

The commit that this patch fixes changed the order of the parameters
of MG_DP_MODE() but din't update the callers, breaking type-c on ICL.

Fixes: 58106b7d816e ("drm/i915: Make MG PHY macros semantically consistent")
Cc: Clint Taylor <clinton.a.taylor@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Aditya Swarup <aditya.swarup@intel.com>
Cc: Manasi navare <manasi.d.navare@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190222202437.6575-1-jose.souza@intel.com

drm/i915/pmu: Always sample an active ringbuffer

As we no longer have a precise indication of requests queued to an
engine, make no presumptions and just sample the ring registers to see
if the engine is busy.

v2: Report busy while the ring is idling on a semaphore/event.
v3: Give the struct a name!
v4: Always 0 outside the powerwell; trusting the powerwell is
accurate enough for our sampling pmu.
v5: Protect against gen7 mmio madness and try to improve grammar

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190223000102.14290-1-chris@chris-wilson.co.uk

drm/i915: Remove unused HAS_PCH_CNP_LP

Other than LPT, no other PCH needed to differentiate between
LP and HP. So let's remove this before we spread this mistake
to future platforms.

Cc: Anusha Srivatsa <anusha.srivatsa@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190221211716.9433-1-rodrigo.vivi@intel.com

drm/i915: Sort newer to older platforms.

No functional change. Just a reorg to match the preferred
behavior.

v2: missing else (Ville)

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190221214430.27095-1-rodrigo.vivi@intel.com

drm/i915: Sort ctx workarounds init from newer to older platforms.

No functional change. Just a reorg to match the preferred
behavior.

Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190221231452.21672-1-rodrigo.vivi@intel.com

drm/i915/guc: Flush the residual log capture irq on disabling

As we disable the log capture events, flush any residual interrupt
before we flush and disable the worker.

v2: Mika pointed out that it wasn't the worker re-queueing itself, but a
rogue irq.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109716
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/20190221163833.21393-1-chris@chris-wilson.co.uk

drm/i915: remove redundant likely/unlikely annotation

unlikely has already included in IS_ERR(), so just
remove redundant likely/unlikely annotation.

Signed-off-by: Chengguang Xu <cgxu519@gmx.com>
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/20190221020819.21832-1-cgxu519@gmx.com

drm/i915: Reorder struct_mutex-vs-reset_lock in i915_gem_fault()

Annoyingly, struct_mutex was not entirely eliminated from the reset
pathway; for reasons of its own, intel_display_resume() requires
struct_mutex to prepare the planes it already captured. To avoid the
immediate problem of a deadlock between the struct_mutex and the reset
srcu, we have to acquire the reset_lock before struct_mutex in
i915_gem_fault(). Now any wait underneath struct_mutex will result us in
having to forcibly reset all inflight rendering, less than ideal, but
better than a deadlock (and will do for the short term).

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/20190221102924.13442-1-chris@chris-wilson.co.uk

drm/i915/icl: Drop redundant gamma mode mask

gamma mode mask was not considering the 30th and 31st bits.
Due to this state readout was masking these bits, causing a
mismatch and false warning, even though the registers were
updated correctly. Dropped the gamma mode mask as it is
redundant and ideally entire register content should be
matching. This resolves the state mismatch warnings.

Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1550689519-6977-1-git-send-email-uma.shankar@intel.com
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109624

drm/i915/hdcp: Silence compiler critics

drivers/gpu/drm/i915/intel_hdcp.c:92 intel_hdcp2_capable() warn: inconsistent indenting
drivers/gpu/drm/i915/intel_hdcp.c:786:5: error: no previous prototype for ‘intel_hdcp_check_link’

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/20190221084833.19489-1-chris@chris-wilson.co.uk

drm/i915/guc: Calling guc_disable_communication in all suspend paths

This aim of this patch is to call guc_disable_communication in all
suspend paths. The reason to introduce this is to resolve a bug that
occurred due to suspend late not being called in the hibernate devices
path.

Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Signed-off-by: Sujaritha Sundaresan <sujaritha.sundaresan@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/20190220013927.9488-3-sujaritha.sundaresan@intel.com

drm/i915/guc: Splitting CT channel open/close functions

The aim of this patch is to allow enabling and disabling
of CTB without requiring the mutex lock.

v2: Phasing out ctch_is_enabled function and replacing it with
    ctch->enabled (Daniele)

Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Signed-off-by: Sujaritha Sundaresan <sujaritha.sundaresan@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/20190220013927.9488-2-sujaritha.sundaresan@intel.com

drm/i915: Prevent user context creation while wedged

Introduce a new ABI method for detecting a wedged driver by reporting
-EIO from DRM_IOCTL_I915_GEM_CONTEXT_CREATE.

This came up in considering how to handle context recovery from
userspace. There we wish to create a new context after the original is
banned (the clients opts into the no recovery after reset strategy) in
order to rebuild the mesa context from scratch. In doing so, if the
device was wedged and not the context banned, we would fall into a loop
of permanently trying to recreate the context and never making forward
progress. This patch would inform the client that we are no longer able
to create a context, and the client would have no choice but to abort
(or at least inform its callers about the lost device for anv).

References: https://lists.freedesktop.org/archives/mesa-dev/2019-February/215469.html
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/20190220225556.28715-1-chris@chris-wilson.co.uk

drm/i915: Extend skl+ crc sources with more planes

On skl the crc registers were extended to provide plane crcs
for up to 7 planes. Add the new crc sources.

The current code uses the ivb+ register definitions for skl+
which does happen to work as the plane1, plane2, and dmux/pf
bits happen the match what ivb+ had. So no bug in the current
code.

v2: Drop the unused set_wa parameter (DK)

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190214192219.3858-4-ville.syrjala@linux.intel.com
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/i915: Remove the broken DP CRC support for g4x

DP CRCs don't really work on g4x. If you want any CRCs on DP you must
select the CRC source before the port is enabled, otherwise the CRC
source select bits simply ignore any writes to them. And once the port
is enabled we mustn't change the CRC source select until the port is
disabled. That almost works, but not quite :( Eventually the CRC source
select bits get permanently stuck one way or the other, and after that
a reboot (or possibly a display reset) is needed to get working CRCs
on that pipe (not matter which CRC source we try to use).

Additionally the DFT scrambler reset bits we're trying to use don't
seem to exist on g4x. There are some potentially relevant looking bits
in the pipe registers, but when I tried it I got stable looking CRCs
without setting any bits for this.

If there is a way to make DP CRCs work reliably on g4x, I wasn't
able to find it. So let's just remove the broken code we have.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190214192219.3858-3-ville.syrjala@linux.intel.com
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>

drm/i915: Use named initializers for the crc source name array

We assume that the index of the string in the crc source names
array matches the enum value for the crc source. Let's use named
initializers to make sure that is indeed the case even if someone
rearranges either the enum or the array.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190214192219.3858-2-ville.syrjala@linux.intel.com
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/i915: Remove the "pf" crc source

The "pipe" and "pf" crc sources are in fact the same thing.
Remove the "pf" one.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190214192219.3858-1-ville.syrjala@linux.intel.com
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/i915: Reduce the RPS shock

Limit deboosting and boosting to keep ourselves at the extremes
when in the respective power modes (i.e. slowly decrease frequencies
while in the HIGH_POWER zone and slowly increase frequencies while
in the LOW_POWER zone). On idle, we will hit the timeout and drop
to the next level quickly, and conversely if busy we expect to
hit a waitboost and rapidly switch into max power.

This should improve the UX experience by keeping the GPU clocks higher
than they ostensibly should be (based on simple busyness) by switching
into the INTERACTIVE mode (due to waiting for pageflips) and increasing
clocks via waitboosting. This will incur some additional power, our
saving grace should be rc6 and powergating to keep the extra current
draw in check.

Food for future thought would be deadline scheduling? If we know certain
contexts (high priority compositors) absolutely must hit the next vblank
then we can raise the frequencies ahead of time. Part of this is covered
by per-context frequencies, where userspace is given control over the
frequency range they want the GPU to execute at (for largely the same
problem as this, where the workload is very latency sensitive but at the
EI level appears mostly idle). Indeed, the per-context series does
extend the modeset boosting to include a frequency range tweak which
seems applicable to solving this jittery UX behaviour.

Reported-by: Lyude Paul <lyude@redhat.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109408
References: 0d55babc8392 ("drm/i915: Drop stray clearing of rps->last_adj")
References: 60548c554be2 ("drm/i915: Interactive RPS mode")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Lyude Paul <lyude@redhat.com>
Cc: Eero Tamminen <eero.t.tamminen@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Quoting Lyude Paul:
> Before reverting 0d55babc8392754352f1058866dd4182ae587d11: [4.20]
>
> 35 measurements [of gnome-shell animations]
> Average: 33.65657142857143 FPS
> FPS observed: 20.8 - 46.87 FPS
> Percentage under 60 FPS: 100.0%
> Percentage under 55 FPS: 100.0%
> Percentage under 50 FPS: 100.0%
> Percentage under 45 FPS: 97.14285714285714%
> Percentage under 40 FPS: 97.14285714285714%
> Percentage under 35 FPS: 45.714285714285715%
> Percentage under 30 FPS: 11.428571428571429%
> Percentage under 25 FPS: 2.857142857142857%
>
> After reverting: [4.19 behaviour]
>
> 30 measurements
> Average: 49.833666666666666 FPS
> FPS observed: 33.85 - 60.0 FPS
> Percentage under 60 FPS: 86.66666666666667%
> Percentage under 55 FPS: 70.0%
> Percentage under 50 FPS: 53.333333333333336%
> Percentage under 45 FPS: 20.0%
> Percentage under 40 FPS: 6.666666666666667%
> Percentage under 35 FPS: 6.666666666666667%
> Percentage under 30 FPS: 0%
> Percentage under 25 FPS: 0%
>
> Patched:
> 42 measurements
> Average: 46.05428571428571 FPS
> FPS observed: 1.82 - 59.98 FPS
> Percentage under 60 FPS: 88.09523809523809%
> Percentage under 55 FPS: 61.904761904761905%
> Percentage under 50 FPS: 45.23809523809524%
> Percentage under 45 FPS: 35.714285714285715%
> Percentage under 40 FPS: 33.33333333333333%
> Percentage under 35 FPS: 19.047619047619047%
> Percentage under 30 FPS: 7.142857142857142%
> Percentage under 25 FPS: 4.761904761904762%

Tested-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190219122215.8941-13-chris@chris-wilson.co.uk

drm/i915: Fix KBL HDCP2.2 encrypt status signalling

HDCP transmitter is supposed to indicate the HDCP encryption status of
the link through enc_en signals in a window of time called "window of
opportunity" defined by HDCP HDMI spec.

But on KBL this timing of signalling has an issue. To fix the issue this
WA of resetting the signalling is required.

v2:
  WA is moved into the toggle_signalling [Daniel]
v3:
  Commit msg is rewritten with more information
v4:
  Reviewed-by Daniel.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-17-git-send-email-ramalingam.c@intel.com

drm/i915: CP_IRQ handling for DP HDCP2.2 msgs

Implements the
Waitqueue is created to wait for CP_IRQ
Signaling the CP_IRQ arrival through atomic variable.
For applicable DP HDCP2.2 msgs read wait for CP_IRQ.

As per HDCP2.2 spec "HDCP Transmitters must process CP_IRQ interrupts
when they are received from HDCP Receivers"

Without CP_IRQ processing, DP HDCP2.2 H_Prime msg was getting corrupted
while reading it based on corresponding status bit. This creates the
random failures in reading the DP HDCP2.2 msgs.

v2:
  CP_IRQ arrival is tracked based on the atomic val inc [daniel]
  Recording the reviewed-by Daniel from IRC.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-16-git-send-email-ramalingam.c@intel.com

drm/i915: Implement the HDCP2.2 support for HDMI

Implements the HDMI adaptation specific HDCP2.2 operations.

Basically these are DDC read and write for authenticating through
HDCP2.2 messages.

v2: Rebased.
v3:
  No more special handling of Gmbus burst read for AKE_SEND_CERT.
  Style fixed with few naming. [Uma]
  %s/PARING/PAIRING
v4:
  msg_sz is initialized at definition.
  Lookup table is defined for HDMI HDCP2.2 msgs [Daniel].
v5: Rebased.
v6:
  Make a function as inline [Uma]
  %s/uintxx_t/uxx
v7:
  Errors due to sinks are reported as DEBUG logs.
  Adjust to the new mei interface.
v8:
  ARRAY_SIZE for the # of array members [Jon & Daniel].
  hdcp adaptation is added as a const in the hdcp_shim [Daniel]

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-15-git-send-email-ramalingam.c@intel.com

drm/i915: Implement the HDCP2.2 support for DP

Implements the DP adaptation specific HDCP2.2 functions.

These functions perform the DPCD read and write for communicating the
HDCP2.2 auth message back and forth.

v2:
  wait for cp_irq is merged with this patch. Rebased.
v3:
  wait_queue is used for wait for cp_irq [Chris Wilson]
v4:
  Style fixed.
  %s/PARING/PAIRING
  Few style fixes [Uma]
v5:
  Lookup table for DP HDCP2.2 msg details [Daniel].
  Extra lines are removed.
v6: Rebased.
v7:
  Fixed some regression introduced at v5. [Ankit]
  Macro HDCP_2_2_RX_CAPS_VERSION_VAL is reused [Uma]
  Converted a function to inline [Uma]
  %s/uintxx_t/uxx
v8:
  Error due to the sinks are reported as DEBUG logs.
  Adjust to the new mei interface.
v9:
  ARRAY_SIZE for no of array members [Jon & Daniel]
  return of the wait_for_cp_irq is made as void [Daniel]
  Wait for HDCP2.2 msg is done based on polling the reg bit than
    CP_IRQ based. [Daniel]
  hdcp adaptation is added as a const in the hdcp_shim [Daniel]
v10:
  config_stream_type is redefined [Daniel]
  DP Errata specific defines are moved into intel_dp.c.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Signed-off-by: Ankit K Nautiyal <ankit.k.nautiyal@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-14-git-send-email-ramalingam.c@intel.com

drm: removing the DP Errata msg and its msg id

Since DP ERRATA message is not defined at spec, those structure
definition is removed from drm_hdcp.h

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Suggested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Acked-by: Dave Airlie <airlied@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-13-git-send-email-ramalingam.c@intel.com

drm/i915: Handle HDCP2.2 downstream topology change

When repeater notifies a downstream topology change, this patch
reauthenticate the repeater alone without disabling the hdcp
encryption. If that fails then complete reauthentication is executed.

v2:
  Rebased.
v3:
  Typo in commit msg is fixed [Uma]
v4:
  Rebased as part of patch reordering.
  Minor style fixes.
v5:
  Rebased.
v6:
  Rebased.
v7:
  Errors due to sinks are reported as DEBUG logs.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-12-git-send-email-ramalingam.c@intel.com

drm/i915: Implement HDCP2.2 link integrity check

Implements the link integrity check once in 500mSec.

Once encryption is enabled, an ongoing Link Integrity Check is
performed by the HDCP Receiver to check that cipher synchronization
is maintained between the HDCP Transmitter and the HDCP Receiver.

On the detection of synchronization lost, the HDCP Receiver must assert
the corresponding bits of the RxStatus register. The Transmitter polls
the RxStatus register and it may initiate re-authentication.

v2:
  Rebased.
v3:
  enum check_link_response is used check the link status [Uma]
v4:
  Rebased as part of patch reordering.
v5:
  Required members of intel_hdcp is defined [Sean Paul]
v6:
  hdcp2_check_link is cancelled at required places.
v7:
  Rebased for the component i/f changes.
  Errors due to the sinks are reported as DEBUG logs.
v8:
  hdcp_check_work is used for both hdcp1 and hdcp2 check_link [Daniel]
  hdcp2.2 encryption status check is put under WARN_ON [Daniel]
  drm_hdcp.h changes are moved into separate patch [Daniel]
v9:
  enum check_link_status is defined at intel_drv.h [Daniel]

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-11-git-send-email-ramalingam.c@intel.com

drm: HDCP2.2 link check period

Time period for HDCP2.2 link check.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Acked-by: Dave Airlie <airlied@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-10-git-send-email-ramalingam.c@intel.com

drm/i915: Implement HDCP2.2 repeater authentication

Implements the HDCP2.2 repeaters authentication steps such as verifying
the downstream topology and sending stream management information.

v2: Rebased.
v3:
  -EINVAL is returned for topology error and rollover scenario.
  Endianness conversion func from drm_hdcp.h is used [Uma]
v4:
  Rebased as part of patches reordering.
  Defined the mei service functions [Daniel]
v5:
  Redefined the mei service functions as per comp redesign.
v6:
  %s/uintxx_t/uxx
  Check for comp_master is removed.
v7:
  Adjust to the new mei interface.
  style issue fixed.
v8:
  drm_hdcp.h change is moved into separate patch [Daniel]
v9:
  %s/__swab16/cpu_to_be16. [Tomas]
  Reviewed-by Uma.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-9-git-send-email-ramalingam.c@intel.com

drm/i915: Implement HDCP2.2 receiver authentication

Implements HDCP2.2 authentication for hdcp2.2 receivers, with
following steps:
Authentication and Key exchange (AKE).
Locality Check (LC).
Session Key Exchange(SKE).
DP Errata for stream type configuration for receivers.

At AKE, the HDCP Receiver’s public key certificate is verified by the
HDCP Transmitter. A Master Key k m is exchanged.

At LC, the HDCP Transmitter enforces locality on the content by
requiring that the Round Trip Time (RTT) between a pair of messages
is not more than 20 ms.

At SKE, The HDCP Transmitter exchanges Session Key ks with
the HDCP Receiver.

In DP HDCP2.2 encryption and decryption logics use the stream type as
one of the parameter. So Before enabling the Encryption DP HDCP2.2
receiver needs to be communicated with stream type. This is added to
spec as ERRATA.

This generic implementation is complete only with the hdcp2 specific
functions defined at hdcp_shim.

v2: Rebased.
v3:
  %s/PARING/PAIRING
  Coding style fixing [Uma]
v4:
  Rebased as part of patch reordering.
  Defined the functions for mei services. [Daniel]
v5:
  Redefined the mei service functions as per comp redesign.
  Required intel_hdcp members are defined [Sean Paul]
v6:
  Typo of cipher is Fixed [Uma]
  %s/uintxx_t/uxx
  Check for comp_master is removed.
v7:
  Adjust to the new interface.
  Avoid using bool structure members. [Tomas]
v8: Rebased.
v9:
  bool is used in struct intel_hdcp [Daniel]
  config_stream_type is redesigned [Daniel]
  Reviewed-by Uma.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-8-git-send-email-ramalingam.c@intel.com

drm/i915: Enable and Disable of HDCP2.2

Considering that HDCP2.2 is more secure than HDCP1.4, When a setup
supports HDCP2.2 and HDCP1.4, HDCP2.2 will be enabled.

When HDCP2.2 enabling fails and HDCP1.4 is supported, HDCP1.4 is
enabled.

This change implements a sequence of enabling and disabling of
HDCP2.2 authentication and HDCP2.2 port encryption.

v2:
  Included few optimization suggestions [Chris Wilson]
  Commit message is updated as per the rebased version.
  intel_wait_for_register is used instead of wait_for. [Chris Wilson]
v3:
  Extra comment added and Style issue fixed [Uma]
v4:
  Rebased as part of patch reordering.
  HDCP2 encryption status is tracked.
  HW state check is moved into WARN_ON [Daniel]
v5:
  Redefined the mei service functions as per comp redesign.
  Merged patches related to hdcp2.2 enabling and disabling [Sean Paul].
  Required shim functionality is defined [Sean Paul]
v6:
  Return values are handles [Uma]
  Realigned the code.
  Check for comp_master is removed.
v7:
  HDCP2.2 is attempted only if mei interface is up.
  Adjust to the new interface
  Avoid bool usage in struct [Tomas]
v8:
  mei_binded status check is removed.
  %s/hdcp2_in_use/hdcp2_encrypted
v9:
  bool is used in struct intel_hdcp. [Daniel]
v10:
  panel is replaced with sink [Uma]
  Mei interface decided the hdcp2_capability.
  WARN_ON if hdcp_enable is called when hdcp state is ENABLED.
  Reviewed-by Uma.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-7-git-send-email-ramalingam.c@intel.com

drm/i915: hdcp1.4 CP_IRQ handling and SW encryption tracking

"hdcp_encrypted" flag is defined to denote the HDCP1.4 encryption status.
This SW tracking is used to determine the need for real hdcp1.4 disable
and hdcp_check_link upon CP_IRQ.

On CP_IRQ we filter the CP_IRQ related to the states like Link failure
and reauthentication req etc and handle them in hdcp_check_link.
CP_IRQ corresponding to the authentication msg availability are ignored.

WARN_ON is added for the abrupt stop of HDCP encryption of a port.

v2:
  bool is used in struct for the cleaner coding. [Daniel]
  check_link work_fn is scheduled for cp_irq handling [Daniel]
v3:
  rebased.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-6-git-send-email-ramalingam.c@intel.com

drm/i915: MEI interface implementation

Defining the mei-i915 interface functions and initialization of
the interface.

v2:
  Adjust to the new interface changes. [Tomas]
  Added further debug logs for the failures at MEI i/f.
  port in hdcp_port data is equipped to handle -ve values.
v3:
  mei comp is matched for global i915 comp master. [Daniel]
  In hdcp_shim hdcp_protocol() is replaced with const variable. [Daniel]
  mei wrappers are adjusted as per the i/f change [Daniel]
v4:
  port initialization is done only at hdcp2_init only [Danvet]
v5:
  I915 registers a subcomponent to be matched with mei_hdcp [Daniel]
v6:
  HDCP_disable for all connectors incase of comp_unbind.
  Tear down HDCP comp interface at i915_unload [Daniel]
v7:
  Component init and fini are moved out of connector ops [Daniel]
  hdcp_disable is not called from unbind. [Daniel]
v8:
  subcomponent name is dropped as it is already merged.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> [v11]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-5-git-send-email-ramalingam.c@intel.com

drm/i915: Initialize HDCP2.2

Add the HDCP2.2 initialization to the existing HDCP1.4 stack.

v2:
  mei interface handle is protected with mutex. [Chris Wilson]
v3:
  Notifiers are used for the mei interface state.
v4:
  Poll for mei client device state
  Error msg for out of mem [Uma]
  Inline req for init function removed [Uma]
v5:
  Rebase as Part of reordering.
  Component is used for the I915 and MEI_HDCP interface [Daniel]
v6:
  HDCP2.2 uses the I915 component master to communicate with mei_hdcp
- [Daniel]
  Required HDCP2.2 variables defined [Sean Paul]
v7:
  intel_hdcp2.2_init returns void [Uma]
  Realigning the codes.
v8:
  Avoid using bool structure members.
  MEI interface related changes are moved into separate patch.
  Commit msg is updated accordingly.
  intel_hdcp_exit is defined and used from i915_unload
v9:
  Movement of the hdcp_check_link is moved to new patch [Daniel]
  intel_hdcp2_exit is removed as mei_comp will be unbind in i915_unload.
v10:
  bool is used in struct to make coding simpler. [Daniel]
  hdmi hdcp init is placed correctly after encoder attachment.
v11:
  hdcp2_capability check is moved into hdcp.c [Tomas]

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-4-git-send-email-ramalingam.c@intel.com

drm/i915: Gathering the HDCP1.4 routines together

All HDCP1.4 routines are gathered together, followed by the generic
functions those can be extended for HDCP2.2 too.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Reviewed-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1550338640-17470-2-git-send-email-ramalingam.c@intel.com

drm/i915: Avoid reset lock in writing fence registers

The idea of taking the reset lock around writing the fence register was
to serialise the mmio write we also perform during the reset where those
registers get clobbered. However, the lock is overkill as write tearing
between reset and fence_update() is harmless; the final value of the
fence register is the same. A race between revoke_fences() and
fence_update() is also harmless at this point as on the fault path where
this is necessary, we acquire the reset lock to coordinate ourselves in
the upper layer.

The danger of acquiring the reset lock again in fence_update() is that
we may recurse from the shrinker along the i915_gem_fault() path.

<4> [125.739646] ============================================
<4> [125.739652] WARNING: possible recursive locking detected
<4> [125.739659] 5.0.0-rc6-ga6e4cbf00557-drmtip_223+ #1 Tainted: G     U
<4> [125.739666] --------------------------------------------
<4> [125.739672] gem_mmap_gtt/1017 is trying to acquire lock:
<4> [125.739679] 00000000a730190a (&dev_priv->gpu_error.reset_backoff_srcu){+.+.}, at: i915_reset_trylock+0x0/0x310 [i915]
<4> [125.739848]
but task is already holding lock:
<4> [125.739854] 00000000a730190a (&dev_priv->gpu_error.reset_backoff_srcu){+.+.}, at: i915_reset_trylock+0x192/0x310 [i915]
<4> [125.739918]
other info that might help us debug this:
<4> [125.739925]  Possible unsafe locking scenario:

<4> [125.739930]        CPU0
<4> [125.739934]        ----
<4> [125.739937]   lock(&dev_priv->gpu_error.reset_backoff_srcu);
<4> [125.739944]   lock(&dev_priv->gpu_error.reset_backoff_srcu);
<4> [125.739950]
 *** DEADLOCK ***

<4> [125.739958]  May be due to missing lock nesting notation

<4> [125.739966] 5 locks held by gem_mmap_gtt/1017:
<4> [125.739972]  #0: 00000000471f682c (&mm->mmap_sem){++++}, at: __do_page_fault+0x133/0x500
<4> [125.739987]  #1: 0000000026542685 (&dev->struct_mutex){+.+.}, at: i915_gem_fault+0x1f6/0x860 [i915]
<4> [125.740061]  #2: 00000000a730190a (&dev_priv->gpu_error.reset_backoff_srcu){+.+.}, at: i915_reset_trylock+0x192/0x310 [i915]
<4> [125.740126]  #3: 00000000c828eb4f (fs_reclaim){+.+.}, at: fs_reclaim_acquire.part.25+0x0/0x30
<4> [125.740140]  #4: 000000002d360d65 (shrinker_rwsem){++++}, at: shrink_slab+0x1cb/0x2c0
<4> [125.740151]
stack backtrace:
<4> [125.740159] CPU: 1 PID: 1017 Comm: gem_mmap_gtt Tainted: G     U            5.0.0-rc6-ga6e4cbf00557-drmtip_223+ #1
<4> [125.740170] Hardware name: Dell Inc.                 OptiPlex 745                 /0GW726, BIOS 2.3.1  05/21/2007
<4> [125.740180] Call Trace:
<4> [125.740189]  dump_stack+0x67/0x9b
<4> [125.740199]  __lock_acquire+0xc75/0x1b00
<4> [125.740209]  ? arch_tlb_finish_mmu+0x2a/0xa0
<4> [125.740216]  ? tlb_finish_mmu+0x1a/0x30
<4> [125.740222]  ? zap_page_range_single+0xe2/0x130
<4> [125.740230]  ? lock_acquire+0xa6/0x1c0
<4> [125.740237]  lock_acquire+0xa6/0x1c0
<4> [125.740296]  ? i915_clear_error_registers+0x280/0x280 [i915]
<4> [125.740357]  i915_reset_trylock+0x44/0x310 [i915]
<4> [125.740417]  ? i915_clear_error_registers+0x280/0x280 [i915]
<4> [125.740426]  ? lockdep_hardirqs_on+0xe0/0x1b0
<4> [125.740434]  ? _raw_spin_unlock_irqrestore+0x39/0x60
<4> [125.740499]  fence_update+0x218/0x470 [i915]
<4> [125.740571]  i915_vma_unbind+0xa6/0x550 [i915]
<4> [125.740640]  i915_gem_object_unbind+0xfa/0x190 [i915]
<4> [125.740711]  i915_gem_shrink+0x2dc/0x590 [i915]
<4> [125.740722]  ? ___preempt_schedule+0x16/0x18
<4> [125.740792]  ? i915_gem_shrinker_scan+0xc9/0x130 [i915]
<4> [125.740861]  i915_gem_shrinker_scan+0xc9/0x130 [i915]
<4> [125.740870]  do_shrink_slab+0x143/0x3f0
<4> [125.740878]  shrink_slab+0x228/0x2c0
<4> [125.740886]  shrink_node+0x167/0x450
<4> [125.740894]  do_try_to_free_pages+0xc4/0x340
<4> [125.740902]  try_to_free_pages+0xdc/0x2e0
<4> [125.740911]  __alloc_pages_nodemask+0x662/0x1110
<4> [125.740921]  ? reacquire_held_locks+0xb5/0x1b0
<4> [125.740928]  ? reacquire_held_locks+0xb5/0x1b0
<4> [125.740986]  ? i915_reset_trylock+0x192/0x310 [i915]
<4> [125.741045]  ? i915_memcpy_init_early+0x30/0x30 [i915]
<4> [125.741054]  pte_alloc_one+0x12/0x70
<4> [125.741060]  __pte_alloc+0x11/0xf0
<4> [125.741067]  apply_to_page_range+0x37e/0x440
<4> [125.741127]  remap_io_mapping+0x6c/0x100 [i915]
<4> [125.741196]  i915_gem_fault+0x5a9/0x860 [i915]
<4> [125.741204]  ? ptlock_alloc+0x15/0x30
<4> [125.741212]  __do_fault+0x2c/0xb0
<4> [125.741218]  __handle_mm_fault+0x8ee/0xfa0
<4> [125.741227]  handle_mm_fault+0x196/0x3a0
<4> [125.741235]  __do_page_fault+0x246/0x500
<4> [125.741243]  ? page_fault+0x8/0x30
<4> [125.741250]  page_fault+0x1e/0x30
<4> [125.741256] RIP: 0033:0x55d0cc456e12
<4> [125.741264] Code: b0 df ff ff 89 c2 8b 85 70 df ff ff 01 c2 8b 85 70 df ff ff 48 98 48 8d 0c 85 00 00 00 00 48 8b 85 e0 df ff ff 48 01 c8 f7 d2 <89> 10 83 85 70 df ff ff 01 81 bd 70 df ff ff ff 03 00 00 7e be 48
<4> [125.741280] RSP: 002b:00007ffc1bab7ab0 EFLAGS: 00010206
<4> [125.741287] RAX: 00007fc787cb6000 RBX: 0000000000000000 RCX: 0000000000000000
<4> [125.741295] RDX: 00000000ffffffff RSI: 0000000000005401 RDI: 0000000000000002
<4> [125.741303] RBP: 00007ffc1bab9b70 R08: 00007ffc1bab7920 R09: 000000000000001b
<4> [125.741310] R10: 7165722074736554 R11: 0000000000000246 R12: 000055d0cc454a80
<4> [125.741318] R13: 00007ffc1bab9f60 R14: 0000000000000000 R15: 0000000000000000

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109665
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/20190219122215.8941-4-chris@chris-wilson.co.uk

drm/i915: Beware temporary wedging when determining -EIO

At a few points in our uABI, we check to see if the driver is wedged and
report -EIO back to the user in that case. However, as we perform the
check and reset asynchronously (where once before they were both
serialised by the struct_mutex), we may instead see the temporary wedging
used to cancel inflight rendering to avoid a deadlock during reset
(caused by either us timing out in our reset handler,
i915_wedge_on_timeout or with malice aforethought in intel_reset_prepare
for a stuck modeset). If we suspect this is the case, that is we see a
wedged driver *and* reset in progress, then wait until the reset is
resolved before reporting upon the wedged status.

v2: might_sleep() (Mika)

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109580
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/20190220145637.23503-1-chris@chris-wilson.co.uk

drm/i915: Update DRIVER_DATE to 20190220

Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

Merge tag 'topic/mei-hdcp-2019-02-19' of git://anongit.freedesktop.org/drm/drm-intel into drm-intel-next-queued

Prep patches + headers for the mei-hdcp/i915 component interfaces

Also contains the prep work in the component helpers plus adjustements
for the snd-hda/i915 component interface.

Plus one small static inline in the drm_hdcp.h header that both i915
and mei_hdcp will need.

Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
From: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20190219071619.GA11016@phenom.ffwll.local

Merge drm/drm-next into drm-intel-next-queued

Doing a backmerge to be able to merge topic/mei-hdcp-2019-02-19 PR.

Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

Merge https://gitlab.freedesktop.org/drm/msm into drm-next

On the display side, cleanups and fixes to enabled modifiers
(QCOM_COMPRESSED).  And otherwise mostly misc fixes all around.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rob Clark <robdclark@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CAF6AEGuZ5uBKpf=fHvKpTiD10nychuEY8rnE+HeRz0QMvtY5_A@mail.gmail.com

Merge branch 'linux-5.1' of git://github.com/skeggsb/linux into drm-next

Various fixes/cleanups, along with initial support for SVM features
utilising HMM address-space mirroring and device memory migration.
There's a lot more work to do in these areas, both in terms of
features and efficiency, but these can slowly trickle in later down
the track.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Ben Skeggs <skeggsb@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CACAvsv5bsB4rRY1Gqa_Bp_KAd-v_q1rGZ4nYmOAQhceL0Nr-Xg@mail.gmail.com

drm/nouveau/dmem: use dma addresses during migration copies

Removes the need for temporary VMM mappings.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/dmem: use physical vram addresses during migration copies

Removes the need for temporary VMM mappings.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/dmem: extend copy function to allow direct use of physical addresses

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/svm: new ioctl to migrate process memory to GPU memory

This add an ioctl to migrate a range of process address space to the
device memory. On platform without cache coherent bus (x86, ARM, ...)
this means that CPU can not access that range directly, instead CPU
will fault which will migrate the memory back to system memory.

This is behind a staging flag so that we can evolve the API.

Signed-off-by: Jérôme Glisse <jglisse@redhat.com>

drm/nouveau/dmem: device memory helpers for SVM

Device memory can be use in SVM, in which case we do not have any of
the existing buffer object. This commit add infrastructure to allow
use of device memory without nouveau_bo. Again this is a temporary
solution until a rework of GPU memory management.

Signed-off-by: Jérôme Glisse <jglisse@redhat.com>

drm/nouveau/svm: initial support for shared virtual memory

This uses HMM to mirror a process' CPU page tables into a channel's page
tables, and keep them synchronised so that both the CPU and GPU are able
to access the same memory at the same virtual address.

While this code also supports Volta/Turing, it's only enabled for Pascal
GPUs currently due to channel recovery being unreliable right now on the
later GPUs.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: prepare for enabling svm with existing userspace interfaces

For a channel to make use of SVM features, it requires a different GPU MMU
configuration than we would normally use, which is not desirable to switch
to unless a client is actively going to use SVM.

In order to supporting SVM without more extensive changes to the userspace
interfaces, the SVM_INIT ioctl needs to replace the previous configuration
safely.

The only way we can currently do this safely, accounting for some unlikely
failure conditions, is to allocate the new VMM without destroying the last
one, and prioritising the SVM-enabled configuration in the code that cares.

This will get cleaned up again further down the track.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fault/gv100-: expose VoltaFaultBufferA

This nvclass exposes the replayable fault buffer, which will be used
by SVM to manage GPU page faults.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fault/gp100: expose MaxwellFaultBufferA

This nvclass exposes the replayable fault buffer, which will be used
by SVM to manage GPU page faults.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mmu/gp100-: support vmms with gcc/tex replayable faults enabled

Some GPU units are capable of supporting "replayable" page faults, where
the execution unit will wait for SW to fixup GPU page tables rather than
triggering a channel-fatal fault.

This feature isn't useful (it's harmful, even) unless something like HMM
is being used to manage events appearing in the replayable fault buffer,
so, it's disabled by default.

This commit allows a client to request it be enabled.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mmu/gp100-: add privileged methods for fault replay/cancel

Host methods exist to do at least some of what we need, but we are not
currently pushing replay/cancels through a channel like UVM does as it's
not clear whether it's necessary in our case (UVM also updates PTEs with
the GPU).

UVM also pushes a software method for fault cancels on Pascal, seemingly
because the host methods don't appear to be sufficient.  If/when we want
to push the replay/cancel on the GPU, we can re-purpose the cancellation
code here to implement that swmthd.

Keep it simple for now, until we figure out exactly what we need here.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mmu: add a privileged method to directly manage PTEs

This provides a somewhat more direct method of manipulating the GPU page
tables, which will be required to support SVM.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mmu: store mapped flag separately from memory pointer

This will be used to support a privileged client providing PTEs directly,
without a memory object to use as a reference.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mmu: support initialisation of client-managed address-spaces

NVKM is currently responsible for managing the allocation of a client's
GPU address-space, but there's various use-cases (ie. HMM address-space
mirroring) where giving a client more direct control is desirable.

This commit allows for a VMM to be created where the area allocated for
NVKM is limited to a client-specified window, the remainder of address-
space is controlled directly by the client.

Leaving a window is necessary to support various internal requirements,
but also to support existing allocation interfaces as not all of the HW
is capable of working with a HMM allocation.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: expose method to determine current context

MMU will need access to this info.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: expose fecs methods for pausing ctxsw

MMU will need access to these.

v2. Apply fix from Rhys Kidd to send correct FECS method for STOP_CTXSW.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/falcon: fix a few indentation issues

There are a few statements that are indented incorrectly. Fix these.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mmu/gf100-: virtualise setting pdb base address for invalidation

It appears that Pascal and newer need something different.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mmu/gf100-: make mmu invalidate function more general

Will want to reuse this for fault replay/cancellation swmthds.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: store fecs/gpccs falcon pointers in substructures

Future changes will want to add some additional things here, keep them
grouped together.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: move fecs bind_pointer into a function

Makes the code somewhat less magic.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: remove some unnecessary reg writes

This is already done during golden context creation.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: move fecs elpg setup into functions

Makes the code somewhat less magic.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: move fecs discover_pm_image_size into a function

Makes the code somewhat less magic.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: move fecs discover_zcull_image_size into a function

Makes the code somewhat less magic.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: move fecs discover_image_size into a function

Makes the code somewhat less magic.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: move fecs set_watchdog_timeout method into a function

Makes the code somewhat less magic.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: allow accelerated buffer moves even when gr isn't present

There's no need to avoid using copy engines if gr init fails for some
reason (usually missing FW, or incomplete bring-up).

It's not terribly useful for an end-user, but it'll slightly speed up
suspend/resume when saving fb contents, and allow for host/ce code to
be validated.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv04-nv4x: move resume code to dispnv04 init hook

It has no relevance to the atomic path used by newer GPUs.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv04-nv4x: move suspend code to dispnv04 fini hook

It has no relevance to the atomic path used by newer GPUs.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv04-nv4x: move a bunch of pre-nv50 page flip code to dispnv04

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms: display destroy/init/fini hooks can be static

Swapped order of functions in dispnv04 to allow this, but no code changes.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: allocate kernel channel(s) before initialising display

Some of the pre-NV50 depends on SW methods to implement synchronisation
for page flips, and we want to move this setup out of common code, thus
we require the channel to have been allocation before display init.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/disp/gf119-: decode exception reason to human-readable string

We also change the error strings to match NVIDIA's naming.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bios/init: handle INIT_GENERIC_CONDITION_ID_NO_PANEL_SEQ_DELAYS

As I currently understand it, this is related to features we have no
support for as of yet.

In theory, this change should be a noop, just without the warning.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bios/init: label existing INIT_GENERIC_CONDITION types

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/secboot: fix missing newline in error messages

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/sec2/tu102-: instantiate SEC2 falcon

Required for ACR.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/sec2: utilise engine PRI address from TOP

Turing has its SEC2 instance in an alternate location, and this avoids
needing to duplicate the code here for it.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/nvdec/tu102-: instantiate NVDEC0 falcon

Required to run VPR scrubber binary as part of secboot.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/nvdec/gp102-: utilise engine PRI address from TOP

Turing has its NVDEC instances in an alternate location.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gsp/gv100-: instantiate GSP falcon

We need this for Turing ACR, but it's present from Volta onwards.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/top/gv100-: translate entry for the GSP

So we're able to connect fault/interrupt handling to the GSP subdev.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/top: add function to lookup PRI address for devices

Will be using this in upcoming changes to avoid the need for entirely
new subdevs to deal with Turing register moves.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/core: define GSP subdev

Exact meaning of the acronym is unknown, but we need this for Turing ACR.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: fix missing break in switch statement

The NOUVEAU_GETPARAM_PCI_DEVICE case is missing a break statement and falls
through to the following NOUVEAU_GETPARAM_BUS_TYPE case and may end up
re-assigning the getparam->value to an undesired value. Fix this by adding
in the missing break.

Detected by CoverityScan, CID#1460507 ("Missing break in switch")

Fixes: 359088d5b8ec ("drm/nouveau: remove trivial cases of nvxx_device() usage")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

This patch fixes the following warning:

drivers/gpu/drm/nouveau/nouveau_bo.c:1434:53: warning: this statement may fall through [-Wimplicit-fallthrough=]

Warning level 3 was used: -Wimplicit-fallthrough=3

This patch is part of the ongoing efforts to enabling
-Wimplicit-fallthrough.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: Don't WARN_ON VCPI allocation failures

This is much louder then we want. VCPI allocation failures are quite
normal, since they will happen if any part of the modesetting process is
interrupted by removing the DP MST topology in question. So just print a
debugging message on VCPI failures instead.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Fixes: f479c0ba4a17 ("drm/nouveau/kms/nv50: initial support for DP 1.2 multi-stream")
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: dri-devel@lists.freedesktop.org
Cc: nouveau@lists.freedesktop.org
Cc: <stable@vger.kernel.org> # v4.10+
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/pmu: don't print reply values if exec is false

Currently the uninitialized values in the array reply are printed out
when exec is false and nvkm_pmu_send has not updated the array. Avoid
confusion by only dumping out these values if they have been actually
updated.

Detected by CoverityScan, CID#1271291 ("Uninitialized scaler variable")
Fixes: ebb58dc2ef8c ("drm/nouveau/pmu: rename from pwr (no binary change)")

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bios/ramcfg: fix missing parentheses when calculating RON

Currently, the expression for calculating RON is always going to result
in zero no matter the value of ram->mr[1] because the ! operator has
higher precedence than the shift >> operator.  I believe the missing
parentheses around the expression before appying the ! operator will
result in the desired result.

[ Note, not tested ]

Detected by CoveritScan, CID#1324005 ("Operands don't affect result")

Fixes: c25bf7b6155c ("drm/nouveau/bios/ramcfg: Separate out RON pull value")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bios/dp: make array vsoff static, shrinks object size

Don't populate the array vsoff on the stack but instead make it
static. Makes the object code smaller by 67 bytes:

Before:
   text    data     bss     dec     hex filename
   5753     112       0    5865    16e9 .../nouveau/nvkm/subdev/bios/dp.o

After:
   text    data     bss     dec     hex filename
   5622     176       0    5798    16a6 .../nouveau/nvkm/subdev/bios/dp.o

(gcc version 8.2.0 x86_64)

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/ce/tu102: rename implementation from tu104

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/tu102: rename implementation from tu104

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/disp/tu102: rename implementation from tu104

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fault/tu102: rename implementation from tu104

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bar/tu102: rename implementation from tu104

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mmu/tu102: rename implementation from tu104

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mc/tu102: rename implementation from tu104

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/devinit/tu102: rename implementation from tu104

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/volt/gf117: fix speedo readout register

GF117 appears to use the same register as GK104 (but still with the
general Fermi readout mechanism).

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108980
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>