drm/i915/pvc: skip all copy engines from aux table invalidate

As we have more copy engines now, mask all of them from aux table
invalidate.

v2 (MattR):
 - Use I915_MAX_BCS to determine mask rather than hardcoding BCS8.
   (Prathap)

Cc: Prathap Kumar Valsan <prathap.kumar.valsan@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Prathap Kumar Valsan <prathap.kumar.valsan@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505213812.3979301-12-matthew.d.roper@intel.com

drm/i915/pvc: Reset support for new copy engines

Add the reset support for new copy engines in PVC.

Bspec: 52549
Original-author: CQ Tang
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Stuart Summers <stuart.summers@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505213812.3979301-11-matthew.d.roper@intel.com

drm/i915/pvc: Interrupt support for new copy engines

Add the interrupt handler support for new copy engines.

Bspec: 54030
Original-author: CQ Tang
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Stuart Summers <stuart.summers@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505213812.3979301-10-matthew.d.roper@intel.com

drm/i915/pvc: Engine definitions for new copy engines

This patch adds the basic definitions needed to support
new copy engines. Also updating the cmd_info to accommodate
new engines, as the engine id's of legacy engines have been
changed.

v2:
 - Add _BCS(n) definition, similar to other engines.  (Tvrtko)
 - Add I915_MAX_BCS definition, similar to other engnes.  (Prathap)
 - Move GVT change to avoid u16 overflow to its own patch.  (Tvrtko)

Original-author: CQ Tang
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Prathap Kumar Valsan <prathap.kumar.valsan@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505213812.3979301-9-matthew.d.roper@intel.com

drm/i915/gvt: Use intel_engine_mask_t for ring mask

When i915 adds additional PVC blitter instances (in an upcoming patch),
the definition of VECS0 will change from bit(10) to bit(18), causing
GVT's R_ALL mask to overflow the u16 storage that's currently used.
Let's replace the u16 with an intel_engine_mask_t to ensure we avoid
this.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Zhi Wang <zhi.a.wang@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505213812.3979301-8-matthew.d.roper@intel.com

drm/i915/pvc: Reduce stack usage in reset selftest with extra blitter engine

PVC adds extra blitter engines (in the following patch). The reset
selftest has a local array on the stack which is sized by the number
of engines. The increase pushes the size of this array to the point
where it trips the 'stack too large' compile warning. This patch takes
the allocation of the stack and makes it dynamic instead.

v2 (MattR):
 - Minor cosmetic changes:  re-sort definition and allocate using
   kmalloc_array().  (Tvrtko)

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505213812.3979301-7-matthew.d.roper@intel.com

drm/i915/pvc: Read correct RP_STATE_CAP register

The SoC registers, including RP_STATE_CAP, have moved to a new location
in GTTMMADR on Ponte Vecchio.  We need to update the register offset
accordingly.

Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505213812.3979301-5-matthew.d.roper@intel.com

drm/i915/pvc: Define MOCS table for PVC

v2 (MattR):
 - Clarify comment above RING_CMD_CCTL programming.
 - Remove bspec reference from field definition.  (Lucas)
 - Add WARN if we try to use a (presumably uninitialized) wb_index of 0.
   On most platforms 0 is an invalid MOCS entry and even on the ones
   where it isn't, it isn't the right setting for wb_index.  (Lucas)

Bspec: 45101, 72161
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Ayaz A Siddiqui <ayaz.siddiqui@intel.com>
Signed-off-by: Fei Yang <fei.yang@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505213812.3979301-4-matthew.d.roper@intel.com

drm/i915/gem: Make drop_pages() return bool

Commit e4e806253003 ("drm/i915: Change shrink ordering to use locking
around unbinding.") changed the return type to int without changing the
return values or their meaning to "0 is success". Move it back to
boolean.

Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220503061556.513175-1-lucas.demarchi@intel.com

drm/i915: Only setup private tmpfs mount when needed and fix logging

If i915 does not want to use huge pages there is a) no point in setting up
the private mount and b) should former fail, it is misleading to log THP
support is disabled in the caller, which does not even know if callee
tried to enable it.

Fix both by restructuring the flow in i915_gemfs_init and at the same time
note the failure to set it up in all cases.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Eero Tamminen <eero.t.tamminen@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220429100414.647857-2-tvrtko.ursulin@linux.intel.com

drm/i915: Enable THP on Icelake and beyond

We have a statement from HW designers that the GPU read regression when
using 2M pages was fixed from Icelake onwards, which was also confirmed
by bencharking Eero did last year:

"""
When IOMMU is disabled, enabling THP causes following perf changes on
TGL-H (GT1):

    10-15% SynMark Batch[0-3]
    5-10% MemBW GPU texture, SynMark ShMapVsm
    3-5% SynMark TerrainFly* + Geom* + Fill* + CSCloth + Batch4
    1-3% GpuTest Triangle, SynMark TexMem* + DeferredAA + Batch[5-7]
          + few others
    -7% MemBW GPU blend

In the above 3D benchmark names, * means all the variants of tests with
the same prefix. For example "SynMark TexMem*", means both TexMem128 &
TexMem512 tests in the synthetic (Intel internal) SynMark test suite.

In the (public, but proprietary) GfxBench & GLB(enchmark) test suites,
there are both onscreen and offscreen variants of each test. Unless
explicitly stated otherwise, numbers are for both variants.

All tests are run with FullHD monitor. All tests are fullscreen except
for GLB and GpuTest ones, which are run in 1/2 screen window (GpuTest
triangle is run both in fullscreen and 1/2 screen window).
"""

Since the only regression is MemBW GPU blend, against many more gains,
it sounds it is time to enable THP on Gen11+.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
References: https://gitlab.freedesktop.org/drm/intel/-/issues/430
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Eero Tamminen <eero.t.tamminen@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220429100414.647857-1-tvrtko.ursulin@linux.intel.com

drm/i915: Drop has_psr from device info

No need to have this parameter in intel_device_info struct
as all platforms with display version 9 or newer has this feature.

As a side effect of the of removal this flag, it will not be printed
in dmesg during driver load anymore and developers will have to rely
on to check the macro and compare with platform being used and IP
versions of it.

Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505193524.276400-7-jose.souza@intel.com

drm/i915: Drop has_dp_mst from device info

No need to have this parameter in intel_device_info struct
as the requirement to support it is the DDI support.

As a side effect of the of removal this flag, it will not be printed
in dmesg during driver load anymore and developers will have to rely
on to check the macro and compare with platform being used and IP
versions of it.

Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505193524.276400-6-jose.souza@intel.com

drm/i915: Drop has_ddi from device info

No need to have this parameter in intel_device_info struct
as all platforms with display version 9 or newer, haswell or broadwell
supports it.

As a side effect of the of removal this flag, it will not be printed
in dmesg during driver load anymore and developers will have to rely
on to check the macro and compare with platform being used and IP
versions of it.

Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505193524.276400-5-jose.souza@intel.com

drm/i915: Drop has_logical_ring_elsq from device info

No need to have this parameter in intel_device_info struct
as all platforms with graphics version 11 or newer has this feature.

As a side effect of the of removal this flag, it will not be printed
in dmesg during driver load anymore and developers will have to rely
on to check the macro and compare with platform being used and IP
versions of it.

Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505193524.276400-4-jose.souza@intel.com

drm/i915: Drop has_reset_engine from device info

No need to have this parameter in intel_device_info struct
as all platforms with graphics version 7 or newer can reset engines.

As a side effect of the of removal this flag, it will not be printed
in dmesg during driver load anymore and developers will have to rely
on to check the macro and compare with platform being used and IP
versions of it.

Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505193524.276400-3-jose.souza@intel.com

drm/i915: Drop has_rc6 from device info

No need to have this parameter in intel_device_info struct
as all platforms with graphics version 6 or newer have software
support for this feature.

As a side effect of the of removal this flag, it will not be printed
in dmesg during driver load anymore and developers will have to rely
on to check the macro and compare with platform being used and IP
versions of it.

Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505193524.276400-2-jose.souza@intel.com

drm/i915: Drop has_gt_uc from device info

No need to have this parameter in intel_device_info struct
as all platforms with graphics version 9 or newer has graphics
microcontroller.

As a side effect of the of removal this flag, it will not be printed
in dmesg during driver load anymore and developers will have to rely
on to check the macro and compare with platform being used and IP
versions of it.

Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505193524.276400-1-jose.souza@intel.com

drm/i915/gt: Fix build error without CONFIG_PM

drivers/gpu/drm/i915/gt/intel_gt_sysfs_pm.c: In function ‘act_freq_mhz_show’:
drivers/gpu/drm/i915/gt/intel_gt_sysfs_pm.c:276:20: error: implicit declaration of function ‘sysfs_gt_attribute_r_max_func’ [-Werror=implicit-function-declaration]
  276 |  u32 actual_freq = sysfs_gt_attribute_r_max_func(dev, attr,
      |                    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Move sysfs_gt_attribute_* macros out of #ifdef block to fix this.

Fixes: 56a709cf7746 ("drm/i915/gt: Create per-tile RPS sysfs interfaces")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220506032652.1856-1-yuehaibing@huawei.com

drm/i915: Don't use DRM_DEBUG_WARN_ON for ring unexpectedly not idle

DRM_DEBUG_WARN_ON should only be used when we are certain CI is guaranteed
to exercise a certain code path, so in case of values coming from MMIO
reads we cannot be sure CI will have all the possible SKUs and parts, or
that it will catch all possible error conditions. Use drm_warn instead.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505110007.943449-2-tvrtko.ursulin@linux.intel.com

drm/i915: Don't use DRM_DEBUG_WARN_ON for unexpected l3bank/mslice config

DRM_DEBUG_WARN_ON should only be used when we are certain CI is guaranteed
to exercise a certain code path, so in case of values coming from MMIO
reads we cannot be sure CI will have all the possible SKUs and parts.

Use drm_warn instead and move logging to init phase while at it.

v2:
 * GEM_WARN_ON in intel_gt_get_valid_steering.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505110007.943449-1-tvrtko.ursulin@linux.intel.com

drm/i915/huc: Don't fail the probe if HuC init fails

The previous patch introduced new failure cases in the HuC init flow
that can be hit by simply changing the config, so we want to avoid
failing the probe in those scenarios. HuC load failure is already
considered a non-fatal error and we have a way to report to userspace
if the HuC is not available via a dedicated getparam, so no changes
in expectation there.
The error message in the HuC init code has also been lowered to info to
avoid throwing error message for an expected behavior.

Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220504204816.2082588-5-daniele.ceraolospurio@intel.com

drm/i915/huc: Prepare for GSC-loaded HuC

HuC loading via GSC is performed via a PXP command sent through the mei
modules, so we need both MEI_GSC and MEI_PXP to be available. Given that
the GSC will do both the transfer and the authentication, the legacy HuC
loading paths can be safely skipped.
Also note that the GSC-loaded HuC survives GT reset.

v2: move the huc_is_authenticated() function to this patch.

Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: Alan Previn <alan.previn.teres.alexis@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220504204816.2082588-4-daniele.ceraolospurio@intel.com

drm/i915/huc: Add fetch support for gsc-loaded HuC binary

On newer platforms (starting DG2 G10 B-step and G11 A-step), ownership of
HuC loading has been moved from the GuC to the GSC. As part of the
change, the header format of the HuC binary has been updated and does not
match the GuC anymore. The GSC will perform all the required checks on
the binary size, so we only need to check that the version matches.

Note that since we still haven't added any gsc-loaded FWs, the
loaded_via_gsc variable will always be kept to its initialization value
of zero.

v2: Add a note about loaded_via_gsc being zero (Alan)

Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: Alan Previn <alan.previn.teres.alexis@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220504204816.2082588-3-daniele.ceraolospurio@intel.com

drm/i915/huc: drop intel_huc_is_authenticated

The function name is confusing, because it doesn't check the actual auth
status in HW but the SW status. Given that there is only one user (the
huc_auth function itself), just get rid of it and use the FW status
checker directly.

Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: Alan Previn <alan.previn.teres.alexis@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220504204816.2082588-2-daniele.ceraolospurio@intel.com

drm/i915/guc: Support programming the EU priority in the GuC descriptor

In GuC submission mode the EU priority must be updated by the GuC rather
than the driver as the GuC owns the programming of the context descriptor.

Given that the GuC code uses the GuC priorities, we can't use a generic
function using i915 priorities for both execlists and GuC submission.
The existing function has therefore been pushed to the execlists
back-end while a new one has been added for GuC.

v2: correctly use the GuC prio.

Cc: John Harrison <john.c.harrison@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Aravind Iddamsetty <aravind.iddamsetty@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220504234636.2119794-1-daniele.ceraolospurio@intel.com

drm/i915: Fix race in __i915_vma_remove_closed

i915_vma_reopen checked if the vma is closed before without taking the
lock. So multiple threads could attempt removing the vma.

Instead the lock needs to be taken before actually checking.

v2: move struct declaration

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: intel-gfx@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Cc: <stable@vger.kernel.org> # v5.3+
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/5732
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Fixes: 155ab8836caa ("drm/i915: Move object close under its own lock")
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220420095720.3331609-1-kherbst@redhat.com

drm/i915: use IOMEM_ERR_PTR() directly

Use IOMEM_ERR_PTR() instead of self defined IO_ERR_PTR().

Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220503144937.679424-1-tvrtko.ursulin@linux.intel.com

drm/i915/pvc: add initial Ponte Vecchio definitions

Additional blitter and media engines will be enabled later.

Bspec: 44481, 44482
Signed-off-by: Stuart Summers <stuart.summers@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220502163417.2635462-2-matthew.d.roper@intel.com

drm/i915: Fix assert in i915_ggtt_pin

Use lockdep_assert_not_held to simplify and correct the code. Otherwise
false positive are hit if lock state is uknown like after a previous
taint.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220429140757.651406-1-tvrtko.ursulin@linux.intel.com

drm/i915: remove superfluous string helper include

Remove the duplicate and incorrect (uses "" instead of <>)
linux/string_helpers.h include.

Fixes: cc1338f259a2 ("drm/i915/xehp: Update topology dumps for Xe_HP")
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220425154754.990815-1-jani.nikula@intel.com

drm/i915/gt: Document the eviction of the Flat-CCS objects

Capture the eviction details for Flat-CCS capable, lmem objects.

v2:
  Fix the Flat-ccs capbility of lmem obj with smem residency
  possibility [Thomas]
v3:
  Fixed the suggestions [Matt]

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
cc: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220502142618.2704-4-ramalingam.c@intel.com

drm/i915/gt: optimize the ccs_sz calculation per chunk

Calculate the ccs_sz that needs to be emitted based on the src
and dst pages emitted per chunk. And handle the return value of emit_pte
for the ccs pages.

v2:
  ccs_sz moved to the reduced scope [Matt]

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220502142618.2704-3-ramalingam.c@intel.com

drm/i915/gt: Clear SET_PREDICATE_RESULT prior to executing the ring

Userspace may leave predication enabled upon return from the batch
buffer, which has the consequent of preventing all operation from the
ring from being executed, including all the synchronisation, coherency
control, arbitration and user signaling. This is more than just a local
gpu hang in one client, as the user has the ability to prevent the
kernel from applying critical workarounds and can cause a full GT reset.

We could simply execute MI_SET_PREDICATE upon return from the user
batch, but this has the repercussion of modifying the user's context
state. Instead, we opt to execute a fixup batch which by mixing
predicated operations can determine the state of the
SET_PREDICATE_RESULT register and restore it prior to the next userspace
batch. This allows us to protect the kernel's ring without changing the
uABI.

Suggested-by: Zbigniew Kempczynski <zbigniew.kempczynski@intel.com>
Signed-off-by: Chris Wilson <chris.p.wilson@intel.com>
Cc: Zbigniew Kempczynski <zbigniew.kempczynski@intel.com>
Cc: Thomas Hellstrom <thomas.hellstrom@intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220425152317.4275-4-ramalingam.c@intel.com

drm/i915/selftests: Skip poisoning SET_PREDICATE_RESULT on dg2

When predication is enabled all commands baring a few (such as MI_BB_END)
are nop'ed. If we accidentally enable predication while poisoning the
context, not only is the rest of the poisoning skipped (thus disabling
the test), but the closing instructions of the poison request are
nop'ed. Not only do we then not signal the waiting context, but we even
prevent re-enabling arbitration and the GPU will not perform a context
switch at the end of the request.

Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Suggested-by: CQ Tang <cq.tang@intel.com>
Signed-off-by: Chris Wilson <chris.p.wilson@intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220425152317.4275-3-ramalingam.c@intel.com

drm/i915/xehpsdv/dg1/tgl: Fix issue with LRI relative addressing

When bit 19 of MI_LOAD_REGISTER_IMM instruction opcode is set on tgl+
devices, HW does not care about certain register address offsets, but
instead check the following for valid address ranges on specific engines:
RCS && CCS: BITS(0 - 10)
BCS: BITS(0 - 11)
VECS && VCS: BITS(0 - 13)
Also, tgl+ now support relative addressing for BCS engine - So, this
patch fixes issue with live_gt_lrc selftest that is failing where there is
mismatch between LRC register layout generated during init and HW
default register offsets.

Signed-off-by: Akeem G Abodunrin <akeem.g.abodunrin@intel.com>
cc: Prathap Kumar Valsan <prathap.kumar.valsan@intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220425152317.4275-2-ramalingam.c@intel.com

drm/i915: Xe_HP SDV and DG2 have up to 4 CCS engines

Cc: Vinay Belgaumkar <vinay.belgaumkar@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Tested-by: Jordan Justen <jordan.l.justen@intel.com> # mesa anvil & iris
Link: https://patchwork.freedesktop.org/patch/msgid/20220428041926.1483683-5-matthew.d.roper@intel.com

drm/i915/xehp: Add compute engine ABI

We're now ready to start exposing compute engines to userspace.

v2:
 - Move kerneldoc for other engine classes to a separate patch.  (Andi)

Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Vinay Belgaumkar <vinay.belgaumkar@intel.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Szymon Morek <szymon.morek@intel.com>
UMD (mesa): https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14395
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Tested-by: Jordan Justen <jordan.l.justen@intel.com> # mesa anvil & iris
Link: https://patchwork.freedesktop.org/patch/msgid/20220428041926.1483683-4-matthew.d.roper@intel.com

drm/i915/xehp: Add register for compute engine's MMIO-based TLB invalidation

Compute engines have a separate register that the driver should use to
perform MMIO-based TLB invalidation.

Note that the term "context" in this register's bspec description is
used to refer to the engine instance (in the same way "context" is used
on bspec 46167).

Bspec: 43930
Cc: Prathap Kumar Valsan <prathap.kumar.valsan@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Prathap Kumar Valsan <prathap.kumar.valsan@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220428041926.1483683-3-matthew.d.roper@intel.com

drm/i915/uapi: Add kerneldoc for engine class enum

We'll be adding a new type of engine soon.  Let's document the existing
engine classes first to help make it clear what each type of engine is
used for.

Cc: Andi Shyti <andi.shyti@linux.intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220428041926.1483683-2-matthew.d.roper@intel.com

drm/i915/pmu: Use existing uncore helper to read gpm_timestamp

Use intel_uncore_read64_2x32 to read upper and lower fields of the GPM
timestamp.

v2: Fix compile error

Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220427003515.3944267-1-umesh.nerlige.ramappa@intel.com

drm/i915/dg2: Define GuC firmware version for DG2

First release of GuC for DG2.

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
CC: Tomasz Mistat <tomasz.mistat@intel.com>
CC: Ramalingam C <ramalingam.c@intel.com>
CC: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220427165550.3636686-3-John.C.Harrison@Intel.com

drm/i915/dg2: add gsc with special gsc bar offsets

DG2 uses different GSC offsets on memory bar
and uses PXP head (HECI1).

v2 (Daniele): Rebased to before the ATS patches

Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> #v1
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220419193314.526966-7-daniele.ceraolospurio@intel.com

mei: gsc: retrieve the firmware version

Add a hook to retrieve the firmware version of the
GSC devices to bus-fixup.
GSC has a different MKHI clients GUIDs but the same message structure
to retrieve the firmware version as MEI so mei_fwver() can be reused.

CC: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220419193314.526966-6-daniele.ceraolospurio@intel.com

mei: gsc: add runtime pm handlers

Implement runtime handlers for mei-gsc, to track
idle state of the device properly.

CC: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220419193314.526966-5-daniele.ceraolospurio@intel.com

mei: gsc: setup char driver alive in spite of firmware handshake failure

Setup char device in spite of firmware handshake failure.
In order to provide host access to the firmware status registers and other
information required for the manufacturing process.

Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220419193314.526966-4-daniele.ceraolospurio@intel.com

mei: add support for graphics system controller (gsc) devices

GSC is a graphics system controller, based on CSE, it provides
a chassis controller for graphics discrete cards, as well as it
supports media protection on selected devices.

mei_gsc binds to a auxiliary devices exposed by Intel discrete
driver i915.

v2: fix error check in mei_gsc_probe
v3: update MODULE_LICENSE ("GPL" is preferred over "GPL v2" and they
    both map to GPL version 2)

Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com> #v3
Link: https://patchwork.freedesktop.org/patch/msgid/20220419193314.526966-3-daniele.ceraolospurio@intel.com

drm/i915/gsc: add gsc as a mei auxiliary device

GSC is a graphics system controller, it provides
a chassis controller for graphics discrete cards.

There are two MEI interfaces in GSC: HECI1 and HECI2.

Both interfaces are on the BAR0 at offsets 0x00258000 and 0x00259000.
GSC is a GT Engine (class 4: instance 6). HECI1 interrupt is signaled
via bit 15 and HECI2 via bit 14 in the interrupt register.

This patch exports GSC as auxiliary device for mei driver to bind to
for HECI2 interface and prepares for HECI1 interface as
it will follow up soon.

CC: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Vitaly Lubart <vitaly.lubart@intel.com>
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220419193314.526966-2-daniele.ceraolospurio@intel.com

Merge drm/drm-next into drm-intel-gt-next

In order to get the GSC Support merged on drm-intel-gt-next
in a clean fashion we needed this ATS-M patch to avoid
conflict in i915_pci.c:

commit 412c942bdfae ("drm/i915/ats-m: add ATS-M platform info")

--

Fixing a silent conflict on drivers/gpu/drm/i915/gt/intel_gt_gmch.c:
-       if (!intel_vtd_active(i915))
+       if (!i915_vtd_active(i915))

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

drm/i915/selftests: tweak the misaligned_case

The compact-pt layout restrictions should only apply to the ppGTT. Also
make this play nice on platforms that only have the 64K GTT restriction,
and not the compact-pt thing.

Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Nirmoy Das <nirmoy.das@linux.intel.com>
Cc: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@linux.intel.com>
Reviewed-by: Ramalingam C <ramalingam.c@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220420181613.70033-4-matthew.auld@intel.com

drm/i915/selftests: fixup min_alignment usage

Trying to cast the region id into the region type doesn't work too well,
since the i915_vm_min_alignment() won't give us the correct value for
the stolen-lmem case.

Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Nirmoy Das <nirmoy.das@linux.intel.com>
Cc: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@linux.intel.com>
Reviewed-by: Ramalingam C <ramalingam.c@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220420181613.70033-3-matthew.auld@intel.com

drm/i915/buddy: sanity check the size

Ensure we check that the size is compatible with the requested
page_size. For tiny objects that are automatically annotated with
TTM_PL_FLAG_CONTIGUOUS(since they fit within a single page), we
currently end up silently overriding the min_page_size, which ends up
hiding bugs elsewhere.

Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Nirmoy Das <nirmoy.das@linux.intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220420181613.70033-2-matthew.auld@intel.com

drm/i915: consider min_page_size when migrating

We can only force migrate an object if the existing object size is
compatible with the new destinations min_page_size for the region.
Currently we blow up with something like:

[ 2857.497462] kernel BUG at drivers/gpu/drm/i915/gt/intel_migrate.c:431!
[ 2857.497497] invalid opcode: 0000 [#1] PREEMPT SMP NOPTI
[ 2857.497502] CPU: 1 PID: 8921 Comm: i915_selftest Tainted: G     U  W         5.18.0-rc1-drm-tip+ #27
[ 2857.497513] RIP: 0010:emit_pte.cold+0x11a/0x17e [i915]
[ 2857.497646] Code: 00 48 c7 c2 f0 cd c1 a0 48 c7 c7 e9 99 bd a0 e8 d2 77 5d e0 bf 01 00 00 00 e8 08 47 5d e0 31 f6 bf 09 00 00 00 e8 3c 7b 4d e0 <0f> 0b 48 c7 c1 e0 2a c5 a0 ba 34 00 00 00 48 c7 c6 00 ce c1 a0 48
[ 2857.497654] RSP: 0018:ffffc900000f7748 EFLAGS: 00010246
[ 2857.497658] RAX: 0000000000000000 RBX: ffffc900000f77c8 RCX: 0000000000000006
[ 2857.497662] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000009
[ 2857.497665] RBP: 0000000000000000 R08: 0000000000000001 R09: 0000000000000001
[ 2857.497668] R10: 0000000000022302 R11: ffff88846dea08f0 R12: 0000000000010000
[ 2857.497672] R13: 0000000001880000 R14: 000000000000081b R15: ffff888106b7c040
[ 2857.497675] FS:  00007f0d4c4e0600(0000) GS:ffff88845da80000(0000) knlGS:0000000000000000
[ 2857.497679] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 2857.497682] CR2: 00007f113966c088 CR3: 0000000211e60003 CR4: 00000000003706e0
[ 2857.497686] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 2857.497689] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 2857.497692] Call Trace:
[ 2857.497694]  <TASK>
[ 2857.497697]  intel_context_migrate_copy+0x1e5/0x4f0 [i915]

Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Nirmoy Das <nirmoy.das@linux.intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220420181613.70033-1-matthew.auld@intel.com

drm/i915/dg2: Add workaround 18019627453

A new DG2 workaround added to fix some corner cases hangs.

v2:
- implementing the second and preferred option for this workaround

BSpec: 54077
BSpec: 68173
BSpec: 71488
Cc: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220419182753.364237-2-jose.souza@intel.com

drm/i915/dg2: Enable Wa_22012727170/Wa_22012727685

The above two workaronuds regarding context isolation are implemented
by GuC. The KMD just needs to enable them.

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220415224025.3693037-7-umesh.nerlige.ramappa@intel.com

drm/i915/dg2: Enable Wa_14014475959 - RCS / CCS context exit

There is bug in DG2 where if the CCS contexts switches out while the RCS
is running it can cause memory corruption. To workaround this add an
atomic to a memory address with a value 1 and semaphore wait to the same
address for a value of 0. The GuC firmware is responsible for writing 0
to the memory address when it is safe for the context to switch out.

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220415224025.3693037-6-umesh.nerlige.ramappa@intel.com

drm/i915/guc: Apply Wa_16011777198

Enable GuC Wa to reset RCS/CCS before it goes into RC6.

Signed-off-by: Vinay Belgaumkar <vinay.belgaumkar@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220415224025.3693037-5-umesh.nerlige.ramappa@intel.com

drm/i915/guc: Enable GuC based workarounds for DG2

There are some workarounds for DG2 that are implemented in the GuC
firmware. However, the KMD is required to enable these by setting the
appropriate flag as GuC does not know what platform it is running on.
  Wa_16011759253
  Wa_14012630569
  Wa_14013746162

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
CC: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220415224025.3693037-4-umesh.nerlige.ramappa@intel.com

drm/i915/guc: Enable Wa_22011802037 for gen12 GuC based platforms

Initiating a reset when the command streamer is not idle or in the
middle of executing an MI_FORCE_WAKE can result in a hang. Multiple
command streamers can be part of a single reset domain, so resetting one
would mean resetting all command streamers in that domain.

To workaround this, before initiating a reset, ensure that all command
streamers within that reset domain are either IDLE or are not executing
a MI_FORCE_WAKE.

Enable GuC PRE_PARSER WA bit so that GuC follows the WA sequence when
initiating engine-resets.

For gt-resets, ensure that i915 applies the WA sequence.

Opens to address in future patches:
- The part of the WA to wait for pending forcewakes is also applicable
  to execlists backend.
- The WA also needs to be applied for gen11

Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220415224025.3693037-3-umesh.nerlige.ramappa@intel.com

drm/i915: Add Wa_22011802037 force cs halt

Prior to doing a reset, SW must ensure command streamer is stopped,
as a workaround, to eliminate a race condition in GPM flow.
Setting both the ring stop and prefetch disable bits, will cause the
command streamer to halt.

Signed-off-by: Tilak Tangudu <tilak.tangudu@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220415224025.3693037-2-umesh.nerlige.ramappa@intel.com

drm/i915/hwconfig: Add DG2 support

v2:
 * Jordan: Drop stepping/skew checking as suggested by John.

Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Tested-by: Jordan Justen <jordan.l.justen@intel.com>
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220331231737.315957-1-jordan.l.justen@intel.com

drm/i915: Don't show engine information in fdinfo with GuC submission

At present i915 does not fetch busyness information from GuC, resulting in
incorrect busyness values in fdinfo. Because engine information is coupled
with busyness in fdinfo, skip showing client engine information in fdinfo
with GuC submission till fetching busyness is supported in the i915 GuC
submission backend.

v2 (Daniele):
  Make commit title and description more precise
  Add FIXME with brief description at code change
  s/intel_guc_submission_is_used/intel_uc_uses_guc_submission/

v3 (Daniele):
  Drop FIXME in comment

Bug: https://gitlab.freedesktop.org/drm/intel/-/issues/5564
Fixes: 055634e4b62f ("drm/i915: Expose client engine utilisation via fdinfo")
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/b614c2c61a954de06fbe2c3a7c70d3a91804407e.1649982207.git.ashutosh.dixit@intel.com

drm/i915/guc/slpc: Use i915_probe_error instead of drm_err

This will ensure we don't have false positives when we run
error injection tests.

Signed-off-by: Vinay Belgaumkar <vinay.belgaumkar@intel.com>
Reviewed-by: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220412224852.21501-1-vinay.belgaumkar@intel.com

drm/i915/uapi: Add DRM_I915_QUERY_GEOMETRY_SUBSLICES

Newer platforms have DSS that aren't necessarily available for both
geometry and compute, two queries will need to exist. This introduces
the first, when passing a valid engine class and engine instance in the
flags returns a topology describing geometry.

Based on past discussion, we currently only support this new query item
on Xe_HP and beyond; earlier platforms do not need to worry about
geometry and compute pipelines having access to different topology and
should continue to use the existing topology query.

v2: fix white space errors
v3: change flags from hosting 2 8 bit numbers to holding a
i915_engine_class_instance struct
v4: add error if non rcs engine passed.
v5 (by MattR):
 - Improve kerneldoc and cross references to related structs/enums.
   (Daniel)
 - Clarify that geometry query is only supported on render engines
   (Francisco)
 - Clarify that the new query is only supported on Xe_HP+.
 - Fix checkpatch warnings.

Cc: Ashutosh Dixit <ashutosh.dixit@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Francisco Jerez <currojerez@riseup.net>
UMD (mesa): https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14143
Testcase: igt@i915_query@test-query-geometry-subslices
Signed-off-by: Matt Atwood <matthew.s.atwood@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20220414192230.749771-4-matthew.d.roper@intel.com

drm/i915/doc: Link query items to their uapi structs

Document the possible options for drm_i915_query_item.query_id with
links to the corresponding uapi structures.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20220414192230.749771-3-matthew.d.roper@intel.com

drm/i915/doc: Convert perf UAPI comments to kerneldoc

Convert the comments for  drm_i915_query_perf_config and
drm_i915_perf_oa_config to kerneldoc so that they will show up in the
generated documentation.  Also correct a couple places that referred to
query_id when they actually meant to refer to query_item.flags.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20220414192230.749771-2-matthew.d.roper@intel.com

drm/i915/doc: Convert drm_i915_query_topology_info comment to kerneldoc

This structure has a great comment describing the fields, but it's not
currently in kerneldoc form and does not show up in the generated
documentation.  Let's fix that and also clarify the description of what
"subslice" refers to on gen12 platforms and beyond and that "slice" is
no longer meaningful on Xe_HP and beyond.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20220414192230.749771-1-matthew.d.roper@intel.com

drm/i915/guc: Update to GuC version 70.1.1

The latest GuC firmware drops the context descriptor pool in favour of
passing all creation data in the create H2G. It also greatly simplifies
the work queue and removes the process descriptor used for multi-LRC
submission. So, remove all mention of LRC and process descriptors and
update the registration code accordingly.

Unfortunately, the new API also removes the ability to set default
values for the scheduling policies at context registration time.
Instead, a follow up H2G must be sent. The individual scheduling
policy update H2G commands are also dropped in favour of a single KLV
based H2G. So, change the update wrappers accordingly and call this
during context registration..

Of course, this second H2G per registration might fail due to being
backed up. The registration code has a complicated state machine to
cope with the actual registration call failing. However, if that works
then there is no support for unwinding if a further call should fail.
Unwinding would require sending a H2G to de-register - but that can't
be done because the CTB is already backed up.

So instead, add a new flag to say whether the context has a pending
policy update. This is set if the policy H2G fails at registration
time. The submission code checks for this flag and retries the policy
update if set. If that call fails, the submission path early exists
with a retry error. This is something that is already supported for
other reasons.

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220412225955.1802543-2-John.C.Harrison@Intel.com

drm/i915/migrate: Evict and restore the flatccs capable lmem obj

When we are swapping out the local memory obj on flat-ccs capable platform,
we need to capture the ccs data too along with main meory and we need to
restore it when we are swapping in the content.

When lmem object is swapped into a smem obj, smem obj will
have the extra pages required to hold the ccs data corresponding to the
lmem main memory. So main memory of lmem will be copied into the initial
pages of the smem and then ccs data corresponding to the main memory
will be copied to the subsequent pages of smem. ccs data is 1/256 of
lmem size.

Swapin happens exactly in reverse order. First main memory of lmem is
restored from the smem's initial pages and the ccs data will be restored
from the subsequent pages of smem.

Extracting and restoring the CCS data is done through a special cmd called
XY_CTRL_SURF_COPY_BLT

v2: Fixing the ccs handling
v3: Handle the ccs data at same loop as main memory [Thomas]
v4: changes for emit_copy_ccs
v5: handle non-flat-ccs scenario

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405150840.29351-10-ramalingam.c@intel.com

drm/i915/gem: Add extra pages in ttm_tt for ccs data

On Xe-HP and later devices, dedicated compression control state (CCS)
stored in local memory is used for each surface, to support the
3D and media compression formats.

The memory required for the CCS of the entire local memory is 1/256 of
the local memory size. So before the kernel boot, the required memory
is reserved for the CCS data and a secure register will be programmed
with the CCS base address

So when an object is allocated in local memory, dont need to explicitly
allocate the space for ccs data. But when the obj is evicted into the
smem, to hold the compression related data along with the obj extra space
is needed in smem. i.e obj_size + (obj_size/256).

Hence when a smem pages are allocated for an obj with lmem placement
possibility we create with the extra pages required for the ccs data for
the obj size.

v2:
  Used imperative wording [Thomas]
v3:
  Inflate the pages only when obj's placement is lmem only
v4:
  GEM_BUG_ON if the ttm->num_pages > obj page size [Thomas]

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
cc: Christian Koenig <christian.koenig@amd.com>
cc: Hellstrom Thomas <thomas.hellstrom@intel.com>
Reviewed-by: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405150840.29351-9-ramalingam.c@intel.com

drm/i915/selftest_migrate: Check CCS meta data clear

Extend the live migrate selftest, to verify the ccs surface clearing
during the Flat-CCS capable lmem obj clear.

v2:
  Look at right places for ccs data [Thomas]

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405150840.29351-8-ramalingam.c@intel.com

drm/i915/selftest_migrate: Consider the possible roundup of size

Consider the possible round up happened at obj size alignment to
min_page_size during the obj allocation.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405150840.29351-7-ramalingam.c@intel.com

drm/i915/gt: Clear compress metadata for Flat-ccs objects

Xe-HP and latest devices support Flat CCS which reserved a portion of
the device memory to store compression metadata, during the clearing of
device memory buffer object we also need to clear the associated
CCS buffer.

XY_CTRL_SURF_COPY_BLT is a BLT cmd used for reading and writing the
ccs surface of a lmem memory. So on Flat-CCS capable platform we use
XY_CTRL_SURF_COPY_BLT  to clear the CCS meta data.

v2: Fixed issues with platform naming [Lucas]
v3: Rebased [Ram]
    Used the round_up funcs [Bob]
v4: Fixed ccs blk calculation [Ram]
    Added Kdoc on flat-ccs.
v5: GENMASK is used [Matt]
    mocs fix [Matt]
    Comments Fix [Matt]
    Flush address programming [Ram]
v6: FLUSH_DW is fixed
    Few coding style fix
v7: Adopting the XY_FAST_COLOR_BLT (Thomas]
v8: XY_CTRL_SURF_COPY_BLT for ccs clearing.
v9: emit_copy_ccs is used.
v10: ctrl_surf cmds are filled in caller itself. [Thomas]
     only one ctrl surf cmd is used as size of lmem is <=8M [Thomas]

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Signed-off-by: Ayaz A Siddiqui <ayaz.siddiqui@intel.com>
Reviewed-by: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405150840.29351-6-ramalingam.c@intel.com

drm/i915/gt: Pass the -EINVAL when emit_pte doesn't update any PTE

When emit_pte doesn't update any PTE with return value as 0, interpret
it as -EINVAL.

v2:
  Add missing goto [Thomas]

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405150840.29351-5-ramalingam.c@intel.com

drm/i915/gt: Optimize the migration and clear loop

Move the static calculations out of the loops for copy and clear.

v2:
  Fix the loss of proper error code on emit_pte

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405150840.29351-4-ramalingam.c@intel.com

drm/i915/gt: Use XY_FAST_COLOR_BLT to clear obj on graphics ver 12+

Use faster XY_FAST_COLOR_BLT cmd on graphics version of 12 and more,
for clearing (Zero out) the pages of the newly allocated object.

XY_FAST_COLOR_BLT is faster than the older XY_COLOR_BLT.

v2:
  Typo fix at title [Thomas]
v3:
  XY_FAST_COLOR_BLT is used only for FLAT_CCS capable gen12+

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405150840.29351-3-ramalingam.c@intel.com

drm/i915/gt: use engine instance directly for offset

To make it uniform across copy and clear, use the engine offset directly
to calculate the offset in the cmd forming for emit_clear.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405150840.29351-2-ramalingam.c@intel.com

Merge tag 'drm-intel-next-2022-04-13-1' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

drm/i915 feature pull for v5.19:

Features and functionality:
- Add support for new Tile 4 format on DG2 (Stan)
- Add support for new CCS clear color compression on DG2 (Mika, Juha-Pekka)
- Add support for new render and media compression formats on DG2 (Matt)
- Support multiple eDP and LVDS native mode refresh rates (Ville)
- Support static DRRS (Ville)
- ATS-M platform info (Matt)
- RPL-S PCI IDs (Tejas)
- Extend DP HDR support to HSW+ (Uma)
- Bump ADL-P DMC version to v2.16 (Madhumitha)
- Let users disable PSR2 while enabling PSR1 (José)

Refactoring and cleanups:
- Massive DRRS and panel fixed mode refactoring and cleanups (Ville)
- Power well refactoring and cleanup (Imre)
- Clean up and refactor crtc readout and compute config (Ville)
- Use kernel string helpers (Lucas)
- Refactor gmbus pin lookups and allocation (Jani)
- PCH display cleanups (Ville)
- DPLL and DPLL manager refactoring (Ville)
- Include and header refactoring (Jani, Tvrtko)
- DMC abstractions (Jani)
- Non-x86 build refactoring (Casey)
- VBT parsing refactoring (Ville)
- Bigjoiner refactoring (Ville)
- Optimize plane, pfit, scaler, etc. programming using unlocked writes (Ville)
- Split several register writes in commit to noarm+arm pairs (Ville)
- Clean up SAGV handling (Ville)
- Clean up bandwidth and ddb allocation (Ville)
- FBC cleanups (Ville)

Fixes:
- Fix native HDMI and DP HDMI DFP clock limits on deep color/4:2:0 (Ville)
- Fix DMC firmware platform check (Lucas)
- Fix cursor coordinates on bigjoiner secondary (Ville)
- Fix MSO vs. bigjoiner timing confusion (Ville)
- Fix ADL-P eDP voltage swing (José)
- Fix VRR capability property update (Manasi)
- Log DG2 SNPS PHY calibration errors (Matt, Lucas)
- Fix PCODE request status checks (Stan)
- Fix uncore unclaimed access warnings (Lucas)
- Fix VBT new max TMDS clock parsing (Shawn)
- Fix ADL-P non-existent underrun recovery (Swathi Dhanavanthri)
- Fix ADL-N stepping info (Tejas)
- Fix DPT mapping flags to contiguous (Stan)
- Fix DG2 max display bandwidth (Vinod)
- Fix DP low voltage SKU checks (Ankit)
- Fix RPL-S VT-d translation enable via quirk (Tejas)
- Fixes to PSR2 (José)
- Fix PIPE_MBUS_DBOX_CTL programming (José)
- Fix LTTPR capability read/check on DP 1.2 (Imre)
- Fix ADL-P register corruption after DDI clock enabling (Imre)
- Fix ADL-P MBUS DBOX BW and B credits (Caz)

Merges:
- Backmerge drm-next (Rodrigo, Jani)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/874k2xgewe.fsf@intel.com

drm/i915/fbc: Call intel_fbc_activate() directly from frontbuffer flush

Remove the pointless detour via __intel_fbc_post_update() during
frontbuffer flush.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220315140001.1172-8-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>

drm/i915/fbc: Introduce intel_fbc_sanitize()

Let's make sure FBC is always disabled when we start to take
over the hardware state.

I suspect this should never really happen, since the only time
when we really should be taking over with the display already
active is when the previous state was progammed by the BIOS,
which likely shouldn't use FBC. This could be driver init,
or S4 resume when the boot kernel doesn't load i915. But I
suppose no harm in keeping this code around for exra safety
since it's quite trivial.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220315140001.1172-7-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>

drm/i915/fbc: Remove intel_fbc_global_disable()

By the time intel_fbc_global_disable() gets called during driver
teardown we should have already disabled all the crtcs, so
no way FBC should be enabled at this point.

And I have no idea what the other user (i915_restore_display())
is even trying to achieve.

So let's just throw intel_fbc_global_disable() into the bin.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220315140001.1172-6-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>

drm/i915: Remove remaining locks from i9xx plane udpates

Now that fbc no longer nukes while a flip is pending we can
remove the last uncore.lock from the i9xx plane code.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220315140001.1172-5-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>

drm/i915/fbc: Skip nuke when flip is pending

Don't issue a nuke from frontbuffer flush while a flip is pending.
This avoids the DSPADDR/DSPSURF rmw abuse from the pre-snb nuke
from racing with the DSPADDR/DSPSURF write being performed by
the flip/plane update. The flip itself will already cause the nuke
so a double nuke is redundant.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220315140001.1172-4-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>

drm/i915/fbc: Streamline frontbuffer busy bits handling

If the frontbuffer bits say this fbc instance isn't affected just
skip the whole thing.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220315140001.1172-3-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>

drm/i915/fbc: Eliminate possible_framebuffer_bits

Not sure what the point of this fbc->possible_frontbuffer_bits is.
And especially don't see why it's returning all the bits when
fbc is not even enabled. So let's just get rid of this and only
say we are interested in the plane's frontbuffer bits when fbc
is actually enabled.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220315140001.1172-2-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>

drm/i915/display/vrr: Reset VRR capable property on a long hpd

With some VRR panels, user can turn VRR ON/OFF on the fly from the panel settings.
When VRR is turned OFF ,sends a long HPD to the driver clearing the Ignore MSA bit
in the DPCD. Currently the driver parses that onevery HPD but fails to reset
the corresponding VRR Capable Connector property.
Hence the userspace still sees this as VRR Capable panel which is incorrect.

Fix this by explicitly resetting the connector property.

v2: Reset vrr capable if status == connector_disconnected
v3: Use i915 and use bool vrr_capable (Jani Nikula)
v4: Move vrr_capable to after update modes call (Jani N)
Remove the redundant comment (Jan N)
v5: Fixes the regression on older platforms by resetting the VRR
only if HAS_VRR
v6: Remove the checks from driver, add in drm core before
setting VRR prop (Ville)
v7: Move VRR set/reset to set/unset_edid (Ville)

Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Fixes: 9bc34b4d0f3c ("drm/i915/display/vrr: Reset VRR capable property on a long hpd")
Signed-off-by: Manasi Navare <manasi.d.navare@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220303233222.4698-1-manasi.d.navare@intel.com

drm/i915/dg2: Add support for DG2 clear color compression

Add support for the DG2 specific render compression with clear color
framebuffer format.

DG2 onwards discrete gfx has support for new flat CCS mapping,
which brings in display feature in to avoid Aux walk for compressed
surface. This support build on top of Flat CCS support added in XEHPSDV.
FLAT CCS surface base address should be 64k aligned,
Compressed displayable surfaces must use tile4 format.

HAS: 1407880786
B.Spec : 7655
B.Spec : 53902

v2: Merge all bits required for the support of functionality into this
    patch from the patch adding the corresponding modifier.

Cc: Mika Kahola <mika.kahola@intel.com>
Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Juha-Pekka Heikkilä <juha-pekka.heikkila@intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Acked-by: Anshuman Gupta <anshuman.gupta@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220411143405.1073845-5-imre.deak@intel.com

drm/fourcc: Introduce format modifier for DG2 clear color

DG2 clear color render compression uses Tile4 layout. Therefore, we need
to define a new format modifier for uAPI to support clear color rendering.

v2:
  Display version is fixed. [Imre]
  KDoc is enhanced for cc modifier. [Nanley & Lionel]
v3:
  Split out the modifier addition to a separate patch.
  Clarify the modifier layout description.

Cc: dri-devel@lists.freedesktop.org
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
cc: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Juha-Pekka Heikkilä <juha-pekka.heikkila@intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Acked-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Acked-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220411143405.1073845-4-imre.deak@intel.com

drm/i915/dg2: Add support for DG2 render and media compression

Add support for DG2 render and media compression, for the description of
buffer layouts see the previous patch adding the corresponding
frame buffer modifiers.

v2:
  Display version fix [Imre]
v3:
  Split out modifier addition to separate patch.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
cc: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
cc: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Juha-Pekka Heikkilä <juha-pekka.heikkila@intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220411143405.1073845-3-imre.deak@intel.com

drm/fourcc: Introduce format modifiers for DG2 render and media compression

The render/media engines on DG2 unify render compression and media
compression into a single format for the first time, using the Tile 4
layout for main surfaces. The compression algorithm is different from
any previous platform and the display engine must still be configured to
decompress either a render or media compressed surface; as such, we
need new RC and MC framebuffer modifiers to represent buffers in this
format.

v2: Clarify modifier layout description.

Cc: dri-devel@lists.freedesktop.org
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Acked-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Acked-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220411143405.1073845-2-imre.deak@intel.com

Merge drm/drm-next into drm-intel-gt-next

Pull in TTM changes needed for DG2 CCS enabling from Ram.

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

Merge tag 'drm-misc-next-2022-04-07' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next for 5.19:

UAPI Changes:

Cross-subsystem Changes:

Core Changes:
  - atomic: Add atomic_print_state to private objects
  - edid: Constify the EDID parsing API, rework of the API
  - dma-buf: Add dma_resv_replace_fences, dma_resv_get_singleton, make
    dma_resv_excl_fence private
  - format: Support monochrome formats
  - fbdev: fixes for cfb_imageblit and sys_imageblit, pagelist
    corruption fix
  - selftests: several small fixes
  - ttm: Rework bulk move handling

Driver Changes:
  - Switch all relevant drivers to drm_mode_copy or drm_mode_duplicate
  - bridge: conversions to devm_drm_of_get_bridge and panel_bridge,
    autosuspend for analogix_dp, audio support for it66121, DSI to DPI
    support for tc358767, PLL fixes and I2C support for icn6211
  - bridge_connector: Enable HPD if supported
  - etnaviv: fencing improvements
  - gma500: GEM and GTT improvements, connector handling fixes
  - komeda: switch to plane reset helper
  - mediatek: MIPI DSI improvements
  - omapdrm: GEM improvements
  - panel: DT bindings fixes for st7735r, few fixes for ssd130x, new
    panels: ltk035c5444t, B133UAN01, NV3052C
  - qxl: Allow to run on arm64
  - sysfb: Kconfig rework, support for VESA graphic mode selection
  - vc4: Add a tracepoint for CL submissions, HDMI YUV output,
    HDMI and clock improvements
  - virtio: Remove restriction of non-zero blob_flags,
  - vmwgfx: support for CursorMob and CursorBypass 4, various
    improvements and small fixes

[airlied: fixup conflict with newvision panel callbacks]
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20220407085940.pnflvjojs4qw4b77@houat

drm/i915/bios: Validate the panel_name table

In addition to the fp_timing,dvo_timing,panel_pnp_id tables
there also exists a panel_name table. Unlike the others this
is just one offset+table_size even though there are still 16
actual panel_names in the data block.

The panel_name table made its first appearance somewhere
around VBT version 156-163. The exact version is not known.
But we don't need to know that since we can just check whether
the pointers block has enough room for it or not.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405173410.11436-7-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915/bios: Trust the LFP data pointers

Now that we've sufficiently validated the LFP data pointers we
can trust them.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405173410.11436-6-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915/bios: Validate LFP data table pointers

Make sure the LFP data table pointers sane. Sensible looking
table entries, everything points correctly into the data block,
etc.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405173410.11436-5-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915/bios: Use the copy of the LFP data table always

Currently get_lvds_fp_timing() still returns a pointer to the original
data block rather than our copy. Let's convert the data pointer offsets
to be relative to the data block rather than the whole BDB. With that
we can make get_lvds_fp_timing() return a pointer to the copy.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405173410.11436-4-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915/bios: Make copies of VBT data blocks

Make a copy of each VBT data block with a guaranteed minimum
size. The extra (if any) will just be left zeroed.

This means we don't have to worry about going out of bounds
when accessing any of the structure members. Otherwise that
could easliy happen if we simply get the version check wrong,
or if the VBT is broken/malicious.

v2: Don't do arithmetic between bdb header and copy
    of the LFP data block (Jani)
v3: Make all the copies up front
v4: Only WARN about min_size==0 if we found the block

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220406133817.30652-1-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915/bios: Use the cached BDB version

We have the BDB version cached, use it. We're going to have to
start doing some of the BDB block parsing later, at which point
we may no longer have the VBT around anymore (we free it at the
end of intel_bios_init() when it didn't come via OpRegion).

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220405173410.11436-2-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915/dg2: Do not explode on phy calibration error

When the PHY fails on calibration we were previously skipping the ddi
initialization. However the driver is not really prepared for that,
ultimately leading to a NULL pointer dereference:

[   75.748348] i915 0000:03:00.0: [drm:intel_modeset_init_nogem [i915]] SNPS PHY A failed to calibrate; output will not be used.
...
[   75.750336] i915 0000:03:00.0: [drm:intel_modeset_setup_hw_state [i915]] [CRTC:80:pipe A] hw state readout: enabled
...

( no DDI A/PHY A )
[   75.753080] i915 0000:03:00.0: [drm:intel_modeset_setup_hw_state [i915]] [ENCODER:235:DDI B/PHY B] hw state readout: disabled, pipe A
[   75.753164] i915 0000:03:00.0: [drm:intel_modeset_setup_hw_state [i915]] [ENCODER:245:DDI C/PHY C] hw state readout: disabled, pipe A
...
[   75.754425] i915 0000:03:00.0: [drm] *ERROR* crtc 80: Can't calculate constants, dotclock = 0!
[   75.765558] i915 0000:03:00.0: drm_WARN_ON_ONCE(drm_drv_uses_atomic_modeset(dev))
[   75.765569] WARNING: CPU: 5 PID: 1759 at drivers/gpu/drm/drm_vblank.c:728 drm_crtc_vblank_helper_get_vblank_timestamp_internal+0x347/0x360
...
[   75.781230] BUG: kernel NULL pointer dereference, address: 000000000000007c
[   75.788198] #PF: supervisor read access in kernel mode
[   75.793347] #PF: error_code(0x0000) - not-present page
[   75.798480] PGD 0 P4D 0
[   75.801019] Oops: 0000 [#1] PREEMPT SMP NOPTI
[   75.805377] CPU: 5 PID: 1759 Comm: modprobe Tainted: G        W         5.18.0-rc1-demarchi+ #199
[   75.827613] RIP: 0010:icl_aux_power_well_disable+0x3b/0x200 [i915]
[   75.833890] Code: 83 ec 30 65 48 8b 04 25 28 00 00 00 48 89 44 24 28 48 8b 06 0f b6 70 1c f6 40 20 04 8d 56 fa 0f 45 f2 e8 88 bd ff ff 48 89 ef <8b> 70 7c e8 ed 67 ff ff 48 89 ef 89 c6 e8 73 67 ff ff 84 c0 75 0a
[   75.852629] RSP: 0018:ffffc90003a7fb30 EFLAGS: 00010246
[   75.857852] RAX: 0000000000000000 RBX: ffff8881145e8f10 RCX: 0000000000000000
[   75.864978] RDX: ffff888115220840 RSI: 0000000000000000 RDI: ffff888115220000
[   75.872106] RBP: ffff888115220000 R08: ffff88888effffe8 R09: 00000000fffdffff
[   75.879234] R10: ffff88888e200000 R11: ffff88888ed00000 R12: ffff8881145e8f10
[   75.886363] R13: 0000000000000001 R14: ffff888115223240 R15: 0000000000000000
[   75.893490] FS:  00007ff6e753a740(0000) GS:ffff88888f680000(0000) knlGS:0000000000000000
[   75.901573] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   75.907313] CR2: 000000000000007c CR3: 00000001216a6001 CR4: 0000000000770ee0
[   75.914446] PKRU: 55555554
[   75.917153] Call Trace:
[   75.919603]  <TASK>
[   75.921709]  intel_power_domains_sanitize_state+0x88/0xb0 [i915]
[   75.927814]  intel_modeset_init_nogem+0x317/0xef0 [i915]
[   75.933205]  i915_driver_probe+0x5f6/0xdf0 [i915]
[   75.937976]  i915_pci_probe+0x51/0x1d0 [i915]

We skip the initialization of PHY A, but later we try to find out what
is the phy for that power well and dereference dig_port, which is NULL.

Failing the PHY calibration could be left as a warning or error, like it
was before commit b4eb76d82a0e ("drm/i915/dg2: Skip output init on PHY
calibration failure"). However that often fails for outputs not being
used, which would make the warning/error appear on systems that have no
visible issues. Anyway, there is still a need to fix those failures,
but that is left for later.

Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220410061537.4187383-1-lucas.demarchi@intel.com

drm/i915: Swap ret and status returned from skl_pcode_request

If ret isn't zero, it is almost for sure ETIMEDOUT, because
we use it in wait_for macro which does continuous retries
until timeout is reached. If we still ran out of time and
retries, we most likely would be interested in getting status,
to understand what was the actual error propagated from PCode,
rather than to find out that we had a time out, which is anyway
quite obvious, if the function fails.

v2: Make it status ? status : ret(thanks Vinod for the hint)

Signed-off-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220411081343.18099-1-stanislav.lisovskiy@intel.com