drm/i915/dg2: Tile 4 plane format support

Tile4 in bspec format is 4K tile organized into
64B subtiles with same basic shape as for legacy TileY
which will be supported by Display13.

v2: - Moved Tile4 associating struct for modifier/display to
      the beginning(Imre Deak)
    - Removed unneeded case I915_FORMAT_MOD_4_TILED modifier
      checks(Imre Deak)
    - Fixed I915_FORMAT_MOD_4_TILED to be 9 instead of 12
      (Imre Deak)

v3: - Rebased patch on top of new changes related to plane_caps.
    - Added static assert to check that PLANE_CTL_TILING_YF
      matches PLANE_CTL_TILING_4(Nanley Chery)
    - Fixed naming and layout description for Tile 4 in drm uapi
      header(Nanley Chery)

v4: - Extracted drm_fourcc changes to separate patch(Nanley Chery)

Reviewed-by: Imre Deak <imre.deak@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Juha-Pekka Heikkilä <juha-pekka.heikkila@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220118115544.15116-3-stanislav.lisovskiy@intel.com

drm/i915: Introduce new Tile 4 format

This tiling layout uses 4KB tiles in a row-major layout. It has the same
shape as Tile Y at two granularities: 4KB (128B x 32) and 64B (16B x 4). It
only differs from Tile Y at the 256B granularity in between. At this
granularity, Tile Y has a shape of 16B x 32 rows, but this tiling has a shape
of 64B x 8 rows.

Reviewed-by: Imre Deak <imre.deak@intel.com>
Acked-by: Nanley Chery <nanley.g.chery@intel.com>
Signed-off-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220118115544.15116-2-stanislav.lisovskiy@intel.com

Merge tag 'drm-intel-gt-next-2022-02-17' of git://anongit.freedesktop.org/drm/drm-intel into drm-intel-next

UAPI Changes:

- Weak parallel submission support for execlists

  Minimal implementation of the parallel submission support for
  execlists backend that was previously only implemented for GuC.
  Support one sibling non-virtual engine.

Core Changes:

- Two backmerges of drm/drm-next for header file renames/changes and
  i915_regs reorganization

Driver Changes:

- Add new DG2 subplatform: DG2-G12 (Matt R)
- Add new DG2 workarounds (Matt R, Ram, Bruce)
- Handle pre-programmed WOPCM registers for DG2+ (Daniele)
- Update guc shim control programming on XeHP SDV+ (Daniele)
- Add RPL-S C0/D0 stepping information (Anusha)
- Improve GuC ADS initialization to work on ARM64 on dGFX (Lucas)

- Fix KMD and GuC race on accessing PMU busyness (Umesh)
- Use PM timestamp instead of RING TIMESTAMP for reference in PMU with GuC (Umesh)
- Report error on invalid reset notification from GuC (John)
- Avoid WARN splat by holding RPM wakelock during PXP unbind (Juston)
- Fixes to parallel submission implementation (Matt B.)
- Improve GuC loading status check/error reports (John)
- Tweak TTM LRU priority hint selection (Matt A.)
- Align the plane_vma to min_page_size of stolen mem (Ram)

- Introduce vma resources and implement async unbinding (Thomas)
- Use struct vma_resource instead of struct vma_snapshot (Thomas)
- Return some TTM accel move errors instead of trying memcpy move (Thomas)
- Fix a race between vma / object destruction and unbinding (Thomas)
- Remove short-term pins from execbuf (Maarten)
- Update to GuC version 69.0.3 (John, Michal Wa.)
- Improvements to GT reset paths in GuC backend (Matt B.)
- Use shrinker_release_pages instead of writeback in shmem object hooks (Matt A., Tvrtko)
- Use trylock instead of blocking lock when freeing GEM objects (Maarten)
- Allocate intel_engine_coredump_alloc with ALLOW_FAIL (Matt B.)
- Fixes to object unmapping and purging (Matt A)
- Check for wedged device in GuC backend (John)
- Avoid lockdep splat by locking dpt_obj around set_cache_level (Maarten)
- Allow dead vm to unbind vma's without lock (Maarten)
- s/engine->i915/i915/ for DG2 engine workarounds (Matt R)

- Use to_gt() helper for GGTT accesses (Michal Wi.)
- Selftest improvements (Matt B., Thomas, Ram)
- Coding style and compiler warning fixes (Matt B., Jasmine, Andi, Colin, Gustavo, Dan)

From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/Yg4i2aCZvvee5Eai@jlahtine-mobl.ger.corp.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
[Fixed conflicts while applying, using the fixups/drm-intel-gt-next.patch
from drm-rerere's 1f2b1742abdd ("2022y-02m-23d-16h-07m-57s UTC: drm-tip
rerere cache update")]

drm/i915: Extract intel_bw_check_data_rate()

Extract the data rate calculation loop out from
intel_bw_atomic_check() to make it a bit less confusing.

v2: Deal with 'bool changed'

Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com> #v1
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220218064039.12834-7-ville.syrjala@linux.intel.com

drm/i915: Extract icl_qgv_points_mask()

Declutter intel_bw_atomic_check() a bit by pulling
the max QGV mask calculation out.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220218064039.12834-6-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>

drm/i915: Properly clear crtc state when disabling it fully, again

Currently we just leave the old gunk lying around in the crtc
state when userspace asks us to fully disable the crtc. That
doesn't match what the state would be had we never even enabled
the crtc in the first place. So let's make this consistent and
call intel_crtc_prepare_cleared_state() for disabled crtcs as well
(excluding bigjoiner slaves of course which have had their state
copied from the master).

I actually already did this once in commit fff13e63a141 ("drm/i915:
Clear most of crtc state when disabling the crtc") but then
commit 19f65a3dbf75 ("drm/i915: Try to make bigjoiner work in atomic
check") undid it all :(

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220217103221.10405-5-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915: Remove odd any_ms=true assignment

For some reason we're flagging that we need to run through the
full modeset calculations (any_ms==true -> do cdclk/etc. checks)
if any crtc got initially flagged for a modeset and is not
enabled via the uapi. No idea why this is here since later on
(after all fastset handling) we do full run through the crtcs
and flag any_ms if anything still needs a full modeset. So let's
just throw out this early weirdo.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220217103221.10405-4-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915: Dump hw.enable and pipe_mode

Include hw.enable and pipe_mode in the crtc debugfs state dump.
These are fairly fundemental to the operation of the driver
so not seeing them leaves us in the dark.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220217103221.10405-3-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915: Dump the crtc hw state always

Let's just dump the "full" hw crtc state in debugs so that we can
see if some stale junk was left behind when the crtc is supposed
to be entirely off.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220217103221.10405-2-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

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

This reverts commit 9bc34b4d0f3cb368241684cc5e0445d435dded44.

Just oopses on most machines.

Cc: Manasi Navare <manasi.d.navare@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220223081810.19917-1-ville.syrjala@linux.intel.com
Acked-by: Jani Nikula <jani.nikula@intel.com>

drm/i915/adl-n: Add PCH Support for Alder Lake N

Add the PCH ID for ADL-N.

Signed-off-by: Tejas Upadhyay <tejaskumarx.surendrakumar.upadhyay@intel.com>
Reviewed-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220127103520.348015-1-tejaskumarx.surendrakumar.upadhyay@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)

Cc: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Manasi Navare <manasi.d.navare@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220215202601.22943-1-manasi.d.navare@intel.com

drm/i915/adlp: Add TypeC PHY TBT->DP-alt/legacy mode switch workaround

Add display workaround # 1309179469 , which fixes a PHY hang when
switching from TBT mode to DP-alt/legacy mode. The workaround also
requires an IFWI/PHY firmware change, before that this change has no
effect (the DKL_PCS_DW5/SOFTRESET flag is always cleared).

HSDES: 18018237866
HSDES: 16014473319

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220218122611.767974-1-imre.deak@intel.com

drm/i915/pxp: prefer forward declaration over includes

Always use forward declarations instead of includes in headers if
possible.

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/20220214173644.2097124-1-jani.nikula@intel.com

drm/i915/reg: split out icl_dsi_regs.h

The ICL DSI registers have fairly isolated usage. Split the register
macros to a separate file.

Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220217224023.3994777-4-jani.nikula@intel.com

drm/i915/reg: split out vlv_dsi_regs.h and vlv_dsi_pll_regs.h

The VLV (including CHV, BXT, and GLK) DSI registers have fairly isolated
usage. Split the register macros to separated files.

Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220217224023.3994777-3-jani.nikula@intel.com

drm/i915/dsi: add separate init timer mask definition for ICL DSI

Having a separate definition will be useful for splitting VLV and ICL
register files.

Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220217224023.3994777-2-jani.nikula@intel.com

drm/i915/dsi: disassociate VBT video transfer mode from register values

The VBT DSI video transfer mode field values have been defined in terms
of the VLV MIPI_VIDEO_MODE_FORMAT register. The ICL DSI code maps that
to ICL DSI_TRANS_FUNC_CONF() register. The values are the same, though
the shift is different.

Make a clean break and disassociate the values from each other. Assume
the values can be different, and translate the VBT value to VLV and ICL
register values as needed. Use the existing macros from intel_bios.h.

This will be useful in splitting the DSI register macros to files by DSI
implementation.

Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220217224023.3994777-1-jani.nikula@intel.com

iosys-map: Add a few more helpers

First the simplest ones:

- iosys_map_memset(): when abstracting system and I/O memory,
  just like the memcpy() use case, memset() also has dedicated
  functions to be called for using IO memory.
- iosys_map_memcpy_from(): we may need to copy data from I/O
  memory, not only to.

In certain situations it's useful to be able to read or write to an
offset that is calculated by having the memory layout given by a struct
declaration. Usually we are going to read/write a u8, u16, u32 or u64.

As a pre-requisite for the implementation, add iosys_map_memcpy_from()
to be the equivalent of iosys_map_memcpy_to(), but in the other
direction. Then add 2 pairs of macros:

- iosys_map_rd() / iosys_map_wr()
- iosys_map_rd_field() / iosys_map_wr_field()

The first pair takes the C-type and offset to read/write. The second
pair uses a struct describing the layout of the mapping in order to
calculate the offset and size being read/written.

We could use readb, readw, readl, readq and the write* counterparts,
however due to alignment issues this may not work on all architectures.
If alignment needs to be checked to call the right function, it's not
possible to decide at compile-time which function to call: so just leave
the decision to the memcpy function that will do exactly that.

Finally, in order to use the above macros with a map derived from
another, add another initializer: IOSYS_MAP_INIT_OFFSET().

v2:
  - Rework IOSYS_MAP_INIT_OFFSET() so it doesn't rely on aliasing rules
    within the union
  - Add offset to both iosys_map_rd_field() and iosys_map_wr_field() to
    allow the struct itself to be at an offset from the mapping
  - Add documentation to iosys_map_rd_field() with example and expected
    memory layout
v3:
  - Drop kernel.h include as it's not needed anymore

Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: Christian König <christian.koenig@amd.com>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Mauro Carvalho Chehab <mchehab@kernel.org>
Cc: dri-devel@lists.freedesktop.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220216174147.3073235-3-lucas.demarchi@intel.com

iosys-map: Add offset to iosys_map_memcpy_to()

In certain situations it's useful to be able to write to an
offset of the mapping. Add a dst_offset to iosys_map_memcpy_to().

Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: Christian König <christian.koenig@amd.com>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: dri-devel@lists.freedesktop.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220216174147.3073235-2-lucas.demarchi@intel.com

drm/i915/dg2: Enable 5th port

DG2 supports a 5th display output which the hardware refers to as "TC1,"
even though it isn't a Type-C output.  This behaves similarly to the TC1
on past platforms with just a couple minor differences:

 * DG2's TC1 bit in SDEISR is at bit 25 rather than 24 as it is on
   ICP/TGP/ADP.
 * DG2 doesn't need the hpd inversion setting that we had to use on DG1

v2:
  intel_ddi_init(dev_priv, PORT_TC1); [Matt]

Cc: Swathi Dhanavanthri <swathi.dhanavanthri@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220218010328.183423-3-lucas.demarchi@intel.com

drm/i915/dg2: Drop 38.4 MHz MPLLB tables

Our early understanding of DG2 was incorrect; since the 5th display
isn't actually a Type-C output, 38.4 MHz input clocks are never used on
this platform and we can drop the corresponding MPLLB tables.

Cc: Anusha Srivatsa <anusha.srivatsa@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220218010328.183423-2-lucas.demarchi@intel.com

drm/i915: Fix for PHY_MISC_TC1 offset

Currently ICL_PHY_MISC macro is returning offset 0x64C10 for PHY_E.
The PORT_TC1 port is not yet enabled properly in the driver, but
intel_phy_snps.c is relying on intel_phy_is_snps() to filter out
unavailable phys. That function was already considering the last phy as
available. Just correct the offset of the last phy to 0x64C14 as the
rest of the support for it is coming on next commits.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220218010328.183423-1-lucas.demarchi@intel.com

drm/i915: Pimp icl+ sagv pre/post update

Add some debugs on what exactly we're doing to the QGV point mask
in the icl+ sagv pre/post plane update hooks. Currently we're just
guessing.

v2: s/u32/u16/ for consistency with the mask sizes (Stan)

Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220218064039.12834-5-ville.syrjala@linux.intel.com

drm/i915: Split pre-icl vs. icl+ SAGV hooks apart

To further reduce the confusion between the pre-icl vs. icl+
SAGV codepaths let's do a full split.

Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220218064039.12834-4-ville.syrjala@linux.intel.com

drm/i915: Fix bw atomic check when switching between SAGV vs. no SAGV

If the only thing that is changing is SAGV vs. no SAGV but
the number of active planes and the total data rates end up
unchanged we currently bail out of intel_bw_atomic_check()
early and forget to actually compute the new WGV point
mask and thus won't actually enable/disable SAGV as requested.
This ends up poorly if we end up running with SAGV enabled
when we shouldn't. Usually ends up in underruns.

To fix this let's go through the QGV point mask computation
if either the data rates/number of planes, or the state
of SAGV is changing.

v2: Check more carefully if things are changing to avoid
    the extra calculations/debugs from introducing unwanted
    overhead

Cc: stable@vger.kernel.org
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com> #v1
Fixes: 20f505f22531 ("drm/i915: Restrict qgv points which don't have enough bandwidth.")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220218064039.12834-3-ville.syrjala@linux.intel.com

drm/i915: Correctly populate use_sagv_wm for all pipes

When changing between SAGV vs. no SAGV on tgl+ we have to
update the use_sagv_wm flag for all the crtcs or else
an active pipe not already in the state will end up using
the wrong watermarks. That is especially bad when we end up
with the tighter non-SAGV watermarks with SAGV enabled.
Usually ends up in underruns.

Cc: stable@vger.kernel.org
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Fixes: 7241c57d3140 ("drm/i915: Add TGL+ SAGV support")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220218064039.12834-2-ville.syrjala@linux.intel.com

drm/i915: Drop pointless i830 PIPECONF read

Reading the PIPECONF enable bit out from the hardware
in i9xx_set_pipeconf() on i830 is pointless as the bit should
always be set since we keep both pipes constantly running on
i830. Drop the pointless read and just always keep the bit set.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220202111616.1579-4-ville.syrjala@linux.intel.com
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>

drm/i915: Make the CHV CGM CSC register writes lockless

The CHV CGM CSC registers are single buffered and so we
may have to write them from the vblank worker, which
imposes very tight dealines. Drop the pointless locking
for the register accessess to reduce the overhead.
All the other registers we bash from the vblank worker
(LUTs) were already made lockless earlier.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220202111616.1579-3-ville.syrjala@linux.intel.com
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>

drm/i915: Make the pipe/output CSC register writes lockless

The pipe/output CSC register writes don't need to be locked
since all the registers are suitably isolated to their own
cachelines. So eliminate the locks to reduce the overhead
during the vblank evade critical section.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220202111616.1579-2-ville.syrjala@linux.intel.com
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>

drm/i915: Move PIPE_CHICKEN RMW out from the vblank evade critical section

We don't want any RMWs in the part of the commit that happens
under vblank evasion. Eventually we want to use the DSB to
handle that and it can't read registers at all. Also reads
are just slowing us down needlessly.

Let's move the whole PIPE_CHICKEN stuff out from the critical
section since we don't have anything there that needs to be
syncrhonized with other plane/pipe registers. If we ever need
to add such things then we have to move it back, but without
doing any reads.

TODO: should look into eliminating the RMW anyway...

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220202111616.1579-1-ville.syrjala@linux.intel.com
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>

drm/i915/display: Implement Wa_16013835468

PSR2 workaround required when mode has delayed vblank.

BSpec: 52890
BSpec: 49421
Cc: Jouni Högander <jouni.hogander@intel.com>
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220210185223.95399-2-jose.souza@intel.com

drm/i915/display: Group PSR2 prog sequences and workarounds

Grouping inside of the same if all the programing sequences and
workarounds of PSR2.
The order of programing changed in intel_psr_enable_source() but
it will not affect PSR2 as at this point PSR2_ENABLE is still disabled.

Cc: Jouni Högander <jouni.hogander@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220210185223.95399-1-jose.souza@intel.com

drm/i915/display/tgl+: Implement new PLL programming step

A new programming step was added to combo and TC PLL sequences.
If override_AFC_startup is set in VBT, driver should overwrite
AFC_startup value to 0x0 or 0x7 in PLL's div0 register.

The current understating is that only TGL needs this and all other
display 12 and newer platforms will have a older VBT or a newer VBT
with override_AFC_startup set to 0 but in any case there is a
drm_warn_on_once() to let us know if this is not true.

v2:
- specification updated, now AFC can be override to 0x0 or 0x7
- not using a union for div0 (Imre)
- following previous wrong vbt naming: bits instead of bytes (Imre)

BSpec: 49204
BSpec: 20122
BSpec: 49968
BSpec: 71360
Cc: Imre Deak <imre.deak@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220216134059.25348-1-jose.souza@intel.com

drm/i915: Disconnect PHYs left connected by BIOS on disabled ports

BIOS may leave a TypeC PHY in a connected state even though the
corresponding port is disabled. This will prevent any hotplug events
from being signalled (after the monitor deasserts and then reasserts its
HPD) until the PHY is disconnected and so the driver will not detect a
connected sink. Rebooting with the PHY in the connected state also
results in a system hang.

Fix the above by disconnecting TypeC PHYs on disabled ports.

Before commit 64851a32c463e5 the PHY connected state was read out even
for disabled ports and later the PHY got disconnected as a side effect
of a tc_port_lock/unlock() sequence (during connector probing), hence
recovering the port's hotplug functionality.

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/5014
Fixes: 64851a32c463 ("drm/i915/tc: Add a mode for the TypeC PHY's disconnected state")
Cc: <stable@vger.kernel.org> # v5.16+
Cc: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220217152237.670220-1-imre.deak@intel.com

drm/i915/dp: remove accidental static on what should be a local variable

The variable should obviously be local, not static.

Fixes: a421d8a99216 ("drm/i915/dp: rewrite DP 2.0 128b/132b link training based on errata")
Cc: Ville Syrjälä <ville.syrjala@linux.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/20220217083938.3587465-1-jani.nikula@intel.com

drm/i915: Polish ilk+ wm register bits

Use REG_GENMASK() & co. for ilk+ watermark registers.

v2: Stick to the current bitmask sizes (Jani)
    Fix "watermarm" typo (Jani)

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220216232806.6194-4-ville.syrjala@linux.intel.com

drm/i915: Introduce intel_crtc_planes_update_arm()

No reason the high level intel_update_crtc() needs to know
that there is something magical about the commit order of
planes between different platforms. So let's hide that
detail even better.

In order to keep to somewhat consistent naming between
things we shall call this intel_crtc_planes_update_arm()
to match the plane->update_arm() vfunc naming convention.
And let's rename the noarm counterpart to
intel_crtc_planes_update_noarm() to more clearly associate
it with the plane->update_noarm() vfunc.

v2: Change the naming convention a bit

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220216232806.6194-2-ville.syrjala@linux.intel.com

drm/i915: Clean up SSKPD/MLTR defines

Give names to the SSKPD/MLTR fields, and use the
REG_GENMASK* and REG_FIELD_GET*.

Also drop the bogus non-mirrored SSKP register define.

v2: Rebase due to intel_mchbar_regs.h
    Leave gen6_check_mch_setup() in place for the moment

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220216232806.6194-3-ville.syrjala@linux.intel.com

drm/i915/dg2: Print PHY name properly on calibration error

We need to use phy_name() to convert the PHY value into a human-readable
character in the error message.

Fixes: a6a128116e55 ("drm/i915/dg2: Wait for SNPS PHY calibration during display init")
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Swathi Dhanavanthri <swathi.dhanavanthri@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220215163545.2175730-1-matthew.d.roper@intel.com

drm/i915/lrc: replace include with forward declarations

Prefer forward declarations over includes if possible.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220214173810.2108975-2-jani.nikula@intel.com

drm/i915/lrc: move lrc_get_runtime() to intel_lrc.c

Move the static inline next to the only caller.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220214173810.2108975-1-jani.nikula@intel.com

drm/i915/fbdev: hide struct intel_fbdev in intel_fbdev.c

As all access to struct intel_fbdev guts is nicely stowed away in
intel_fbdev.c, we can hide the struct definition there too.

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220215122957.2755529-2-jani.nikula@intel.com

drm/i915/fbdev: add intel_fbdev_framebuffer() helper

Wrap accessing struct intel_fbdev guts in a helper.

v2: s/intel_fbdev_to_framebuffer/intel_fbdev_framebuffer/g (Ville)

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220215122957.2755529-1-jani.nikula@intel.com

drm/i915: fix build issue when using clang

drm/i915 adds some extra cflags, namely -Wall, which causes instances of
-Wformat-security to appear when building with clang, even though this
warning is turned off kernel-wide in the main Makefile:

drivers/gpu/drm/i915/gt/intel_gt.c:983:2: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]
        GEM_TRACE("ERROR\n");
        ^~~~~~~~~~~~~~~~~~~~
./drivers/gpu/drm/i915/i915_gem.h:76:24: note: expanded from macro 'GEM_TRACE'
 #define GEM_TRACE(...) trace_printk(__VA_ARGS__)
                       ^~~~~~~~~~~~~~~~~~~~~~~~~
./include/linux/kernel.h:369:3: note: expanded from macro 'trace_printk'
                do_trace_printk(fmt, ##__VA_ARGS__);    \
                ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
./include/linux/kernel.h:383:30: note: expanded from macro 'do_trace_printk'
                __trace_bprintk(_THIS_IP_, trace_printk_fmt, ##args);   \
                                          ^~~~~~~~~~~~~~~~
drivers/gpu/drm/i915/gt/intel_gt.c:983:2: note: treat the string as an argument to avoid this

This does not happen with GCC because it does not enable
-Wformat-security with -Wall. Disable -Wformat-security within the i915
Makefile so that these warnings do not show up with clang.

Signed-off-by: Tong Zhang <ztong0001@gmail.com>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220214195821.29809-1-ztong0001@gmail.com

drm/i915/gvt: #include drm_edid.h for drm_edid_block_valid()

As the excessive #includes from i915_drv.h were axed, kvmgt.c build
started failing. Add the necessary #include where needed.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Fixes: 14da21cc4671 ("drm/i915: axe lots of unnecessary includes from i915_drv.h")
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Zhenyu Wang <zhenyuw@linux.intel.com>
Cc: Zhi Wang <zhi.a.wang@intel.com>
Cc: intel-gvt-dev@lists.freedesktop.org
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Zhi Wang <zhi.a.wang@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220215122030.2741656-1-jani.nikula@intel.com

drm/i915: Add missing intel_mchbar_regs.h

The new header from the previous commit didn't get added after a
conflict resolution...let's add it now.

Fixes: e30e6c7b82a1 ("drm/i915: Move MCHBAR registers to their own header")
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220215061342.2055952-2-matthew.d.roper@intel.com

drm/i915: Move MCHBAR registers to their own header

Registers that exist within the MCH BAR and are mirrored into the GPU's
MMIO space are a good candidate to separate out into their own header.

For reference, the mirror of the MCH BAR starts at the following
locations in the graphics MMIO space (the end of the MCHBAR range
differs slightly on each platform):

 * Pre-gen6:           0x10000
 * Gen6-Gen11 + RKL:  0x140000

v2:
 - Create separate patch to swtich a few register definitions to be
   relative to the MCHBAR mirror base.
 - Drop upper bound of MCHBAR mirror from commit message; there are too
   many different combinations between various platforms to list out,
   and the documentation is spotty for the older pre-gen6 platforms
   anyway.

Bspec: 134, 51771
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220215061342.2055952-2-matthew.d.roper@intel.com

drm/i915: Define MCH registers relative to MCHBAR_MIRROR_BASE

A few of our MCH registers are defined with absolute register offsets.
For consistency, let's switch their definitions to be relative offsets
from MCHBAR_MIRROR_BASE.

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220215061342.2055952-1-matthew.d.roper@intel.com

drm/i915/gt: Order GT registers by MMIO offset

The random order of register definitions we have today causes a lot of
confusion and unintentional duplication when new registers/bits are
added to the driver.  Let's order the GT register file by MMIO offset

A couple duplicated/unused register definitions are dropped while doing
this re-order:  GEN11_GT_INTR_DW{0,1}, GEN11_IIR_REG{0,1}_SELECTOR, and
GEN11_INTR_IDENTITY_REG{0,1} aren't used anywhere in the driver because
we have other parameterized macros referencing those registers.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220209051140.1599643-7-matthew.d.roper@intel.com

drm/i915/gt: Use consistent offset notation in intel_gt_regs.h

Switch all register offsets to use lowercase hex values for consistency.
Also strip any unnecessary leading 0's.  For example, "_MMIO(0x0D08)"
becomes "_MMIO(0xd08)."

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220209051140.1599643-6-matthew.d.roper@intel.com

drm/i915/gt: Cleanup spacing of intel_gt_regs.h

There's a lot of inconsistent spacing and indentation in our register
definitions.  Let's clean things up a bit and follow some consistent
rules:

 * "#define" always starts in column 0
 * There's exactly one space between '#define' and the name of a
   register.
 * There's exactly three spaces between '#define' and the name of a
   bit/bitfield.
 * Tabs (no spaces) are used between a definition name and its value;
   the value starts on column 48 unless the name is too long, in which
   case a single tab is used.

Final diff for this patch is empty if whitespace is ignored:

        $ git diff -w
        $

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220209051140.1599643-5-matthew.d.roper@intel.com

drm/i915/gt: Use parameterized RING_MI_MODE

We have both a parameterized RING_MI_MODE() macro and an RCS-specific
MI_MODE; drop the latter and use the former everywhere.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220209051140.1599643-4-matthew.d.roper@intel.com

drm/i915/gt: Move SFC lock bits to intel_engine_regs.h

These SFC registers were defined in an unusual way, taking an engine as
a parameter rather than an engine MMIO base offset.  Let's adjust them
to match the style used by other per-engine registers and move them to
intel_engine_regs.h.

While doing this move, we can drop GEN12_HCP_SFC_FORCED_LOCK completely;
it was intended for use in an early version of a hardware workaround,
but was no longer necessary by the time the workaround was finalized.
It is not used anywhere in the driver.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220209051140.1599643-3-matthew.d.roper@intel.com

drm/i915/gt: Drop duplicate register definition for VDBOX_CGCTL3F18

Due to some mistaken merge conflict resolution, we wound up with a copy
of VDBOX_CGCTL3F18 in both intel_engine_regs.h and intel_gt_regs.h.
Since this is a per-engine register, referenced relative to an engine's
base offset, drop the copy from intel_gt_regs.h

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220209051140.1599643-2-matthew.d.roper@intel.com

drm/i915: Unconfuse pre-icl vs. icl+ intel_sagv_{pre,post}_plane_update()

intel_sagv_{pre,post}_plane_update() can accidentally forget
to bail out early on pre-icl and proceed down the icl+ codepath
at the end of the function. Fortunately it'll bail out before
it gets too far due to old_qgv_mask==new_qgv_mask==0 so no real
bug here. But lets make the code less confusing anyway.

Cc: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220214091811.13725-5-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>

drm/i915: Widen the QGV point mask

adlp+ adds some extra bits to the QGV point mask. The code attempts
to handle that but forgot to actually make sure we can store those
bits in the bw state. Fix it.

Cc: stable@vger.kernel.org
Cc: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Fixes: 192fbfb76744 ("drm/i915: Implement PSF GV point support")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220214091811.13725-4-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>

drm/i915: Add REG_GENMASK64() and REG_FIELD_GET64()

We treat SSKPD as a 64 bit register. Add the support macros
to define/extract bits in such registers.

v2: Fix 32bit builds

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220211182045.23555-1-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915: Use {active,scaled}_planes to compute ilk watermarks

Use the {active,scaled}_planes bitmasks from the crtc state
rather than poking at the plane state directly. One step
towards eliminating the last use of the somewhat questionble
intel_atomic_crtc_state_for_each_plane_state() macro which
peeks into the plane state without actually holding the plane
mutex.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220211090629.15555-5-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915: Introduce scaled_planes bitmask

Add another plane bitmask, this time tracking which planes are
scaled. This is going to be useful in ILK watermark computations,
and skl+ pipe scaler assignments.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220211090629.15555-4-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915: Move intel_plane_atomic_calc_changes() & co. out

Exfiltrate intel_plane_atomic_calc_changes() and its friends from
intel_display.c to intel_atomic_plane.c since that is a much better
fit.

While at it also nuke the official looking kernel docs for
intel_wm_need_update() and flag it for eventual destruction so
that people don't get any wrong ideas about using it in new code.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220211090629.15555-2-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915: Change bigjoiner state tracking to use the pipe bitmask

Get rid of the inflexible bigjoiner_linked_crtc pointer thing
and just track things as a bitmask of pipes instead. We can
also nuke the bigjoiner_slave boolean as the role of the pipe
can be determined from its position in the bitmask.

It might be possible to nuke the bigjoiner boolean as well
if we make encoder.compute_config() do the bitmask assignment
directly for the master pipe. But for now I left that alone so
that encoer.compute_config() will just flag the state as needing
bigjoiner, and the intel_atomic_check_bigjoiner() is still
responsible for determining the bitmask. But that may have to change
as the encoder may be in the best position to determine how
exactly we should populate the bitmask.

Most places that just looked at the single bigjoiner_linked_crtc
now iterate over the whole bitmask, eliminating the singular
slave pipe assumption.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220203183823.22890-11-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915: Return both master and slave pipes from enabled_bigjoiner_pipes()

Return both the master and slave pipe bitmasks from
enabled_bigjoiner_pipes(). We'll have use for both during
readout soon.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220203183823.22890-10-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915: Use for_each_intel_crtc_in_pipe_mask() more

Convert a few hand roller for_each_intel_crtc_in_pipe_mask()
to the real thing.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220203183823.22890-9-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915: Convert for_each_intel_crtc_mask() to take a pipe mask instead

Often using pipes is more convenient than crtc indices.
Convert the current for_each_intel_crtc_mask() to take a
pipe mask instead of a crtc index mask, and rename it to
for_each_intel_crtc_in_pipe_mask() to make it clear what
it does.

The current users of for_each_intel_crtc_mask() don't really
care which kind of mask we use, but for other uses a pipe
mask if better.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220203183823.22890-8-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915: Introduce intel_crtc_is_bigjoiner_{slave,master}()

Introduce helpers to query whether the crtc is the slave/master
for bigjoiner. This decouples most places from the exact
state layout we use to track this relationship, allowing us
to change and extend it more easily.

Performed with cocci:
@@
expression S, E;
@@
(
  S->bigjoiner_slave = E;
|
- S->bigjoiner_slave
+ intel_crtc_is_bigjoiner_slave(S)
)

@@
expression S, E;
@@
(
- E && S->bigjoiner && !intel_crtc_is_bigjoiner_slave(S)
+ E && intel_crtc_is_bigjoiner_master(S)
|
- S->bigjoiner && !intel_crtc_is_bigjoiner_slave(S)
+ intel_crtc_is_bigjoiner_master(S)
)

@@
expression S;
@@
- (intel_crtc_is_bigjoiner_master(S))
+ intel_crtc_is_bigjoiner_master(S)

@@
expression S, E1, E2, E3;
@@
- intel_crtc_is_bigjoiner_slave(S) ? E1 : S->bigjoiner ? E2 : E3
+ intel_crtc_is_bigjoiner_slave(S) ? E1 : intel_crtc_is_bigjoiner_master(S) ? E2 : E3

@@
typedef bool;
@@
+ bool intel_crtc_is_bigjoiner_slave(const struct intel_crtc_state *crtc_state)
+ {
+  return crtc_state->bigjoiner_slave;
+ }
+
  intel_master_crtc(...) {...}

@@
typedef bool;
@@
+ bool intel_crtc_is_bigjoiner_master(const struct intel_crtc_state *crtc_state)
+ {
+  return crtc_state->bigjoiner && !crtc_state->bigjoiner_slave;
+ }
+
  intel_master_crtc(...) {...}

@@
typedef bool;
identifier S;
@@
- bool is_trans_port_sync_mode(const struct intel_crtc_state *S);
+ bool is_trans_port_sync_mode(const struct intel_crtc_state *state);
+ bool intel_crtc_is_bigjoiner_slave(const struct intel_crtc_state *crtc_state);
+ bool intel_crtc_is_bigjoiner_master(const struct intel_crtc_state *crtc_state);

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220203183823.22890-7-ville.syrjala@linux.intel.com
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>

drm/i915/dg1: Update DMC_DEBUG3 register

Current DMC_DEBUG3(_MMIO(0x101090)) address is for TGL,
it is wrong for DG1. Just like commit 5bcc95ca382e
("drm/i915/dg1: Update DMC_DEBUG register"), correct
this issue for DG1 platform to avoid wrong register
being read.

BSpec: 49788

v2: fix "not wrong" typo. (Jani)

Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Chuansheng Liu <chuansheng.liu@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220211002933.84240-1-chuansheng.liu@intel.com

drm/i915: fix drm_i915.h include grouping and sorting

Group and sort includes in i915_drv.h similar to other places.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/679923380d4757fed4e3a4c4bed80e40b9fdaeec.1644507885.git.jani.nikula@intel.com

drm/i915: axe lots of unnecessary includes from i915_drv.h

It's fairly difficult to ensure these are actually not needed due to
indirect includes via other files. However, it's easier to add them back
as needed and, most importantly, where needed instead of exhaustively
proving they're unnecessary.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/bc2bbcd46b66e44e98e1ef76980dfabcfac700d5.1644507885.git.jani.nikula@intel.com

drm/i915: include some drm headers only where needed

Include drm_fourcc.h, drm_plane.h, and drm_color_mgmt.h where needed, so
we can drop the includes for drm_atomic.h and drm_fourcc.h from
i915_drv.h, reducing the build dependencies.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/b03711b2286396b2e9d5822f6adef4e7a6dc0f7b.1644507885.git.jani.nikula@intel.com

drm/i915: include shmem_fs.h only where needed

Don't include shmem_fs.h in i915_drv.h, reducing the build dependencies.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/44eade17f7ba1480d67c584466eeea3553f31506.1644507885.git.jani.nikula@intel.com

drm/i915: don't include drm_cache.h in i915_drv.h

Include it only in files that use it.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/14edab4a193ea3f73f387a88e3836c8555401871.1644507885.git.jani.nikula@intel.com

drm/i915: split out i915_file_private.h from i915_drv.h

Limit the scope of struct drm_i915_file_private to the files that
actually need it.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/e375859dc1729a1b988036e4103e5b1bd48caa00.1644507885.git.jani.nikula@intel.com

drm/i915: move i915_reset_count()/i915_reset_engine_count() out of i915_drv.h

It doesn't help much, as i915_drv.h includes i915_gpu_error.h, but it's
a step in the right direction.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/7af2f698a320c1efb0563f56a432c6d122d40b94.1644507885.git.jani.nikula@intel.com

drm/i915: move i915_gem_vm_lookup() where it's used

Move the function next to the only user. Arguably it's perhaps not the
best place, but it's much better than having a static inline in a
header.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/a080e401840a8b9d45946ff33fd63c7939a623ae.1644507885.git.jani.nikula@intel.com

drm/i915: move i915_cache_level_str() static in i915_debugfs.c

Move the function next to the only user.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/dc0901dbe424c21b3e03b875bf5b944b214d1af4.1644507885.git.jani.nikula@intel.com

drm/i915: split out gem/i915_gem_domain.h from i915_drv.h

We already have the gem/i915_gem_domain.c file.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/8208321ad09f1fb0d1a61dc0f2449cce8b23a9b9.1644507885.git.jani.nikula@intel.com

drm/i915: Nuke some dead code

Remove all the dead code from icl_ddi_bigjoiner_pre_enable().

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220203183823.22890-6-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915: Clean up the bigjoiner state copy logic

Currently the bigjoiner state copy logic is kind of
a byzantine mess.

Clean it up to operate in the following manner during a full
modeset:
1) master uapi -> hw state copy
2) master hw -> slave hw state copy

And during a non-modeset update we do:
1) master uapi -> hw state light copy
2) master hw -> slave hw state light copy

I think that is now easier to reason about since we never do
any kind of master uapi -> slave hw state copy short circuit
that could happen previously.

Obviously this does now depend on the master uapi->hw copy
always happening before the master hw -> slave hw copy, but
that is guaranteed by the fact that we always add both crtcs
to the state early, the crtcs are registered in pipe
order (so the compute_config loop happens in pipe order),
and the hardware requires the master pipe has to be lower
than the slave pipe as well. And for good measure we shall
add a check+WARN for this before doing the bigjoiner crtc
assignment.

v2: Fix uapi.ctm vs. hw.ctm copy-paste fail

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220204072049.1610-1-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915: Remove weird code from intel_atomic_check_bigjoiner()

There's some weird junk in intel_atomic_check_bigjoiner()
that's trying to look at the old crtc state's bigjoiner
usage for some reason. That code is totally unnecessary,
and maybe even actively harmful. Not entirely sure which
since it's such a mess that I can't actually wrap my brain
around what it ends up doing.

Either way, thanks to intel_bigjoiner_add_affected_crtcs()
all of the old bigjoiner crtcs are guaranteed to be in the
state already if any one of them is in the state. Also if
any one of those crtcs got flagged for a modeset, then all
of them will have been flagged, and the bigjoiner links
will have been detached via kill_bigjoiner_slave().

So there is no need to look examing any old bigjoiner
usage in intel_atomic_check_bigjoiner(). All we have to care
about is whether bigjoiner is needed for the new state,
and whether we can get the slave crtc we need.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220203183823.22890-4-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915: Fix bigjoiner state copy fails

We seem to be missing a few things from the bigjoiner state copy.
Namely hw.mode isn't getting copied (which probably causes PIPESRC
to be misconfigured), CTM/LUTs aren't getting copied (which could
cause the pipe to produced incorrect output), and we also forgot
to copy over the color_mgmt_changed flag so potentially we fail
to do the actual CTM/LUT programming (assuming we aren't doing
a full modeset or fastset). Fix it all.

v2: Fix uapi.ctm vs. hw.ctm copy-paste fail

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220204072009.1546-1-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915: Flag crtc scaling_filter changes as modeset

The core doesn't flag scaling_filter prop changes as needing
a modeset. That doesn't work for us since we only reprogram the
pipe scaler during full modesets and fastsets. So we need to
flag the prop change as a modeset ourselves. Assuming nothing else
has changed the operation will get promoted (demoted?) to a fastset
later.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220203183823.22890-2-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

drm/i915/dp: Reuse intel_hdmi_tmds_clock()

Reuse intel_hdmi_tmds_clock() for DP->HDMI TMDS clock calculations.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211015133921.4609-9-ville.syrjala@linux.intel.com
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>

drm/i915/hdmi: Simplify intel_hdmi_mode_clock_valid()

Just loop over the possible bpc values instead of
using an ugly if construct.

A slight change in behaviour is that we now call
intel_hdmi_{source,sink}_bpc_possible() even for 8bpc,
but that is fine since 8bpc is always supported.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211015133921.4609-8-ville.syrjala@linux.intel.com
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>

drm/i915: Extract skl_crtc_calc_dbuf_bw()

Extract the dbuf slice data_rate calculation into a small
helper. Should make it a bit easier to handle the different
color planes of planar formats correctly.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220118092354.11631-7-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>

drm/i915: Extract skl_allocate_plane_ddb()

Replace some copy-pasta with a function.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220118092354.11631-6-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>

drm/i915: Introduce skl_plane_ddb_iter

Collect a bit of the stuff used during the plane ddb
allocation into a struct we can pass around.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220118092354.11631-5-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>

drm/i915: Fix plane relative_data_rate calculation

We are currently computing the relative data rates as
src_size * scale_factor where scale_factor is src_size / dst_size.
Thus relative data rate is src_size * src_size / dst_size,
which is just utter nonsense. What we really seem to want is
just a reasonable estimate on how much data will be fetched
which is just src_size. So let's do that instead.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220118092354.11631-4-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>

drm/i915: Extract skl_ddb_entry_init()

Extract a small helper to populate a ddb entry.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220118092354.11631-3-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>

drm/i915: Drop pointless dev_priv argument

skl_ddb_entry_init_from_hw() has no need for dev_priv.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220118092354.11631-2-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>

drm/i915: split out gem/i915_gem_create.h from i915_drv.h

We already have the gem/i915_gem_create.c file.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/f4d5108498ce10fff8577520276d40e86b2d5a16.1644507885.git.jani.nikula@intel.com

drm/i915: split out gem/i915_gem_dmabuf.h from i915_drv.h

We already have the gem/i915_gem_dmabuf.c file.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/2f3fa0fb7cd78c204e27b2454410b6530289efdc.1644507885.git.jani.nikula@intel.com

drm/i915/dg2: Add Wa_22011100796

Whenever Full soft reset is required, reset all individual engines
first, and then do a full soft reset.

Signed-off-by: Bruce Chang <yu.bruce.chang@intel.com>
cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@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/20220128185209.18077-5-ramalingam.c@intel.com

drm/i915: More gt idling time with guc submission

On i915_selftest@live@gt_timelines, we create many contexts in loop and
create and submit request and then destoy contexts. Destroying the context
needs to disable scheduling, wait for G2H, deregister context and wait
for G2H to destroy each context. Idling of the gt has to wait for all
this to complete which is taking ~3sec for this test.

Hence we are increasing the igt_flush_test's timeout for gt idling to
3Sec.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220128185209.18077-4-ramalingam.c@intel.com

drm/i915: align the plane_vma to min_page_size of stolen mem

Align the plane vma size to the stolem memory regions' min_page_size.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
cc: Matthew Auld <matthew.auld@intel.com>
cc: Chris P Wilson <chris.p.wilson@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220128185209.18077-3-ramalingam.c@intel.com

drm/i915/dg2: Add Wa_22011450934

An indirect ctx wabb is implemented as per Wa_22011450934 to avoid rcs
restore hang during context restore of a preempted context in GPGPU mode

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
cc: Chris Wilson <chris.p.wilson@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220128185209.18077-2-ramalingam.c@intel.com

drm/i915/selftests: Disable runtime pm wakeref tracking for the mock device

commit c50df701d49e ("drm/i915: Enable rpm wakeref tracking
whether runtime pm is enabled or not") enabled wakeref tracking
even for the mock device. Turns out that has somewhat significant
overhead, and on the glacial Core m3's we have in CI the vma
selftests are now exceeding the allotted time budget.

So let's disable the wakeref tracking once again for the mock
device in order to avoid blowing up the selftest runtime.

Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Acked-by: Tomi Sarvela <tomi.p.sarvela@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220204171053.18409-1-ville.syrjala@linux.intel.com

drm/i915: remove leftover i915_gem_pm.h declarations from i915_drv.h

Remove the duplicates.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/48f5ace6393533372da5d13df3de0203c8db11b3.1644507885.git.jani.nikula@intel.com

drm/i915: split out i915_gem_internal.h from i915_drv.h

We already have the i915_gem_internal.c file.

Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/6715d1f3232c445990630bb3aac00f279f516fee.1644507885.git.jani.nikula@intel.com

drm/i915/mst: update slot information for 128b/132b

128b/132b supports using 64 slots starting from 0, while 8b/10b reserves
slot 0 for metadata.

Commit d6c6a76f80a1 ("drm: Update MST First Link Slot Information Based
on Encoding Format") added support for updating the topology state
accordingly, and commit 41724ea273cd ("drm/amd/display: Add DP 2.0 MST
DM Support") started using it in the amd driver.

This feels more than a little cumbersome, especially updating the
information in atomic check. For i915, add the update to MST connector
.compute_config hook rather than iterating over all MST managers and
connectors in global mode config .atomic_check. Fingers crossed.

v3:
- Propagate errors from intel_dp_mst_update_slots() (Ville)

v2:
- Update in .compute_config() not .atomic_check (Ville)

Cc: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Cc: Lyude Paul <lyude@redhat.com>
Cc: Uma Shankar <uma.shankar@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220208152317.3019070-1-jani.nikula@intel.com