Merge tag 'drm-xe-next-2025-07-15' of https://gitlab.freedesktop.org/drm/xe/kernel into drm-next

Driver Changes:
 - Create and use XE_DEVICE_WA infrastructure (Atwood)
 - SRIOV: Mark BMG as SR-IOV capable (Michal)
 - Dont skip TLB invalidations on VF (Tejas)
 - Fix migration copy direction in access_memory (Auld)
 - General code clean-up (Lucas, Brost, Dr. David, Xin)
 - More missing XeLP workarounds (Tvrtko)
 - SRIOV: Relax VF/PF version negotiation (Michal)
 - SRIOV: LMTT invalidation (Michal)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/aHacDvF9IaVHI61C@intel.com

drm/xe/pf: Invalidate LMTT after completing changes

Once we finish populating all leaf pages in the VF's LMTT we should
make sure that hardware will not access any stale data. Explicitly
force LMTT invalidation (as it was already planned in the past).

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Piotr Piórkowski <piotr.piorkowski@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://lore.kernel.org/r/20250711193316.1920-7-michal.wajdeczko@intel.com

drm/xe/pf: Invalidate LMTT during LMEM unprovisioning

Invalidate LMTT immediately after removing VF's LMTT page tables
and clearing root PTE in the LMTT PD to avoid any invalid access
by the hardware (and VF) due to stale data.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Piotr Piórkowski <piotr.piorkowski@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250711193316.1920-6-michal.wajdeczko@intel.com

drm/xe/pf: Force GuC virtualization mode

By default the GuC starts in the 'native' mode and enables the VGT
mode (aka 'virtualization' mode) only after it receives at least one
set of VF configuration data. While this happens naturally while PF
begins VFs provisioning, we might need this sooner as some actions,
like TLB_INVALIDATION_ALL(0x7002), is supported by the GuC only in
the VGT mode.

And this becomes a real problem if we would want to use above action
to invalidate the LMTT early during VFs auto-provisioning, before VFs
are enabled, as such H2G would be rejected:

 [ ] xe 0000:4d:00.0: [drm] *ERROR* GT0: FAST_REQ H2G fence 0x804e failed! e=0x30, h=0
 [ ] xe 0000:4d:00.0: [drm] *ERROR* GT0: Fence 0x804e was used by action 0x7002 sent at:
      h2g_write+0x33e/0x870 [xe]
      __guc_ct_send_locked+0x1e1/0x1110 [xe]
      guc_ct_send_locked+0x9f/0x740 [xe]
      xe_guc_ct_send_locked+0x19/0x60 [xe]
      send_tlb_invalidation+0xc2/0x470 [xe]
      xe_gt_tlb_invalidation_all_async+0x45/0xa0 [xe]
      xe_gt_tlb_invalidation_all+0x4b/0xa0 [xe]
      lmtt_invalidate_hw+0x64/0x1a0 [xe]
      xe_lmtt_invalidate_hw+0x5c/0x340 [xe]
      pf_update_vf_lmtt+0x398/0xae0 [xe]
      pf_provision_vf_lmem+0x350/0xa60 [xe]
      xe_gt_sriov_pf_config_bulk_set_lmem+0xe2/0x410 [xe]
      xe_gt_sriov_pf_config_set_fair_lmem+0x1c6/0x620 [xe]
      xe_gt_sriov_pf_config_set_fair+0xd5/0x3f0 [xe]
      xe_pci_sriov_configure+0x360/0x1200 [xe]
      sriov_numvfs_store+0xbc/0x1d0
      dev_attr_store+0x17/0x40
      sysfs_kf_write+0x4a/0x80
      kernfs_fop_write_iter+0x166/0x220
      vfs_write+0x2ba/0x580
      ksys_write+0x77/0x100
      __x64_sys_write+0x19/0x30
      x64_sys_call+0x2bf/0x2660
      do_syscall_64+0x93/0x7a0
      entry_SYSCALL_64_after_hwframe+0x76/0x7e
 [ ] xe 0000:4d:00.0: [drm] *ERROR* GT0: CT dequeue failed: -71
 [ ] xe 0000:4d:00.0: [drm] GT0: trying reset from receive_g2h [xe]

This could be mitigated by pushing earlier a PF self-configuration
with some hard-coded values that cover unlimited access to the GGTT,
use of all GuC contexts and doorbells.  This step is sufficient for
the GuC to switch into the VGT mode.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://lore.kernel.org/r/20250711193316.1920-5-michal.wajdeczko@intel.com

drm/xe/pf: Move GGTT config KLVs encoding to helper

In upcoming patch we will want to encode GGTT config KLVs based
on raw numbers, without relying on the allocated GGTT node.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://lore.kernel.org/r/20250711193316.1920-4-michal.wajdeczko@intel.com

drm/xe/pf: Resend PF provisioning after GT reset

If we reload the GuC due to suspend/resume or GT reset then we
have to resend not only any VFs provisioning data, but also PF
configuration, like scheduling parameters (EQ, PT), as otherwise
GuC will continue to use default values.

Fixes: 411220808cee ("drm/xe/pf: Restart VFs provisioning after GT reset")
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://lore.kernel.org/r/20250711193316.1920-3-michal.wajdeczko@intel.com

drm/xe/pf: Prepare to stop SR-IOV support prior GT reset

As part of the resume or GT reset, the PF driver schedules work
which is then used to complete restarting of the SR-IOV support,
including resending to the GuC configurations of provisioned VFs.

However, in case of short delay between those two actions, which
could be seen by triggering a GT reset on the suspened device:

 $ echo 1 > /sys/kernel/debug/dri/0000:00:02.0/gt0/force_reset

this PF worker might be still busy, which lead to errors due to
just stopped or disabled GuC CTB communication:

 [ ] xe 0000:00:02.0: [drm:xe_gt_resume [xe]] GT0: resumed
 [ ] xe 0000:00:02.0: [drm] GT0: trying reset from force_reset_show [xe]
 [ ] xe 0000:00:02.0: [drm] GT0: reset queued
 [ ] xe 0000:00:02.0: [drm] GT0: reset started
 [ ] xe 0000:00:02.0: [drm:guc_ct_change_state [xe]] GT0: GuC CT communication channel stopped
 [ ] xe 0000:00:02.0: [drm:guc_ct_send_recv [xe]] GT0: H2G request 0x5503 canceled!
 [ ] xe 0000:00:02.0: [drm] GT0: PF: Failed to push VF1 12 config KLVs (-ECANCELED)
 [ ] xe 0000:00:02.0: [drm] GT0: PF: Failed to push VF1 configuration (-ECANCELED)
 [ ] xe 0000:00:02.0: [drm:guc_ct_change_state [xe]] GT0: GuC CT communication channel disabled
 [ ] xe 0000:00:02.0: [drm] GT0: PF: Failed to push VF2 12 config KLVs (-ENODEV)
 [ ] xe 0000:00:02.0: [drm] GT0: PF: Failed to push VF2 configuration (-ENODEV)
 [ ] xe 0000:00:02.0: [drm] GT0: PF: Failed to push 2 of 2 VFs configurations
 [ ] xe 0000:00:02.0: [drm:pf_worker_restart_func [xe]] GT0: PF: restart completed

While this VFs reprovisioning will be successful during next spin
of the worker, to avoid those errors, make sure to cancel restart
worker if we are about to trigger next reset.

Fixes: 411220808cee ("drm/xe/pf: Restart VFs provisioning after GT reset")
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://lore.kernel.org/r/20250711193316.1920-2-michal.wajdeczko@intel.com

drm/xe/lrc: Add table with LRC layout

Add a table to document the LRC's BO layout to make it easier to
visualize how each region stacks on top of each other.

Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Link: https://lore.kernel.org/r/20250710-lrc-refactors-v2-4-a5e2ca03f6bd@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Waste fewer instructions in emit_wa_job()

I was debugging some unrelated issue and noticed the current code was
very verbose. We can improve it easily by using the more common batch
buffer building pattern.

Before:
                bb->cs[bb->len++] = MI_LOAD_REGISTER_REG | MI_LRR_DST_CS_MMIO;
     c4d:       41 8b 56 10             mov    0x10(%r14),%edx
     c51:       49 8b 4e 08             mov    0x8(%r14),%rcx
     c55:       8d 72 01                lea    0x1(%rdx),%esi
     c58:       41 89 76 10             mov    %esi,0x10(%r14)
     c5c:       c7 04 91 01 00 08 15    movl   $0x15080001,(%rcx,%rdx,4)
                        bb->cs[bb->len++] = entry->reg.addr;
     c63:       8b 08                   mov    (%rax),%ecx
     c65:       41 8b 56 10             mov    0x10(%r14),%edx
     c69:       49 8b 76 08             mov    0x8(%r14),%rsi
     c6d:       81 e1 ff ff 3f 00       and    $0x3fffff,%ecx
     c73:       8d 7a 01                lea    0x1(%rdx),%edi
     c76:       41 89 7e 10             mov    %edi,0x10(%r14)
     c7a:       89 0c 96                mov    %ecx,(%rsi,%rdx,4)
 ..etc..

After:
                *cs++ = MI_LOAD_REGISTER_REG | MI_LRR_DST_CS_MMIO;
     c52:       41 c7 04 24 01 00 08    movl   $0x15080001,(%r12)
     c59:       15
                        *cs++ = entry->reg.addr;
     c5a:       8b 10                   mov    (%rax),%edx
 ..etc..

Resulting in the following binary change:

add/remove: 0/0 grow/shrink: 0/2 up/down: 0/-348 (-348)
Function                                     old     new   delta
xe_gt_record_default_lrcs.cold               304     296      -8
xe_gt_record_default_lrcs                   2200    1860    -340
Total: Before=13554, After=13206, chg -2.57%

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250710-lrc-refactors-v2-7-a5e2ca03f6bd@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe/gt: Drop third submission for default context

There's no need to submit the nop job again on the first queue. Any
state needed is already saved when the first LRC is switched out. The
comment is a little misleading regarding indirect W/A: first of all
there's still no indirect W/A enabled and secondly, even after they are,
there's no need to submit this job again for having their state
propagated: the indirect W/A will actually run on every LRC switch.

Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250710-lrc-refactors-v2-6-a5e2ca03f6bd@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe/lrc: Remove leftover TODO/FIXME

There isn't anything to set for CTX_TIMESTAMP handling in the empty
LRC: that is set on every LRC init since it should always start from 0
rather than the value saved in the image after first submission.

The FIXME about perma-pinning also doesn't make much sense as we will
always going to pin the lrc and the GGTT mapping has nothing to do with
VM bind.

Nuke these leftover comments.

Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250710-lrc-refactors-v2-5-a5e2ca03f6bd@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe/gt: Extract emit_job_sync()

Both the nop and wa jobs are going through the same boiler plate calls
to emit the job with a timeout and handling error for both bb and job.
Extract emit_job_sync() so those functions create the bb, handling
possible errors and delegate the part about really emitting the job
and waiting for its completion.

Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Link: https://lore.kernel.org/r/20250710-lrc-refactors-v2-3-a5e2ca03f6bd@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Count dwords before allocating

The bb allocation in emit_wa_job() is wrong in 2 ways: first it's
allocating enough space for the 3DSTATE or hardcoding 4k depending on
the engine. In the first case it doesn't account for the WAs and in the
former it may not be sufficient. Secondly it's using the size instead of
number of dwords, causing the buffer to be 4x bigger than needed:
xe_bb_new() receives number of dwords as parameter and its declaration
was also not following its implementation.

Lastly, reword the debug message since it's not only about the LRC WAs
anymore as it also include the 3DSTATE for render.

While it's unlikely this is causing any real issue, let's calculate the
needed space and allocate just enough.

Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Link: https://lore.kernel.org/r/20250710-lrc-refactors-v2-2-a5e2ca03f6bd@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe/lrc: Reduce scope of empty lrc data

The only case in which new lrc data is created from scratch is when it's
called prior to recording the default lrc. There's no need to check for
NULL init_data since in that case the function already failed: just move
the allocation where it's needed.

Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250710-lrc-refactors-v2-1-a5e2ca03f6bd@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe/vf: Store negotiated VF/PF ABI version at device level

There is no need to maintain PF ABI version on per-GT level.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://lore.kernel.org/r/20250713103625.1964-8-michal.wajdeczko@intel.com

drm/xe/pf: Stop requiring VF/PF version negotiation on every GT

While some VF/PF relay actions must be handled on the GT level,
like query for runtime registers, it was clarified by the arch
team that initial version negotiation can be done by the VF just
once, by using any available GuC/GT.

Move handling of the VF/PF ABI version negotiation on the PF side
from the GT level functions to the device level functions.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20250713103625.1964-7-michal.wajdeczko@intel.com

drm/xe/pf: Expose basic info about VFs in debugfs

We already have function to print summary about VFs, but we missed
to add debugfs attribute to make it visible. Do it now.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://lore.kernel.org/r/20250713103625.1964-6-michal.wajdeczko@intel.com

drm/xe: Introduce xe_gt_is_main_type helper

Instead of checking for not being a media type GT provide a small
helper to explicitly express our intentions.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://lore.kernel.org/r/20250713103625.1964-5-michal.wajdeczko@intel.com

drm/xe: Introduce xe_tile_is_root helper

Instead of looking at the tile->id member provide a small helper
to explicitly express our intentions.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://lore.kernel.org/r/20250713103625.1964-4-michal.wajdeczko@intel.com

drm/xe: Move PF and VF device types to separate headers

We plan to add more PF and VF types and mixing them in a single
file is not desired.  Move them out to new dedicated files.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20250713103625.1964-3-michal.wajdeczko@intel.com

drm/xe: Combine PF and VF device data into union

There is no need to keep PF and VF data fields fully separate
since we can be only in one mode at the time. Move them into
a anonymous union to save few bytes.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://lore.kernel.org/r/20250713103625.1964-2-michal.wajdeczko@intel.com

drm/xe: Update register definitions in LRC layout header

Update the register definitions in xe_lrc_layout.h to align with the
official hardware specification (Bspec) terminology. Specifically:

- rename PVC_CTX_ACC_CTR_THOLD to CTX_ACC_CTR_THOLD
- rename PVC_CTX_ASID to CTX_ASID

Signed-off-by: Xin Wang <x.wang@intel.com>
Reviewed-by: Stuart Summers <stuart.summers@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250711060924.7373-1-x.wang@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Add plumbing for indirect context workarounds

Some upcoming workarounds need to be emitted from the indirect workaround
context so lets add some plumbing where they will be able to easily slot
in.

No functional changes for now since everything is still deactivated.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Bspec: 45954
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250711160153.49833-7-tvrtko.ursulin@igalia.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Allow specifying number of extra dwords at the end of wa bb emission

Indirect context setup will need more than one.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250711160153.49833-6-tvrtko.ursulin@igalia.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Track number of written dwords from workaround batch buffer emission

Indirect context setup will need to get to the number of written dwords.
Lets add it as an output parameter so it can be accessed from the finish
helper regardless of whether code is writing directly or via an shadow
buffer.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250711160153.49833-5-tvrtko.ursulin@igalia.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Rename utilization workaround emission function

Lucas suggested to consolidate to a slightly different naming scheme which
will align with the upcoming additions better.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Suggested-by: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250711160153.49833-4-tvrtko.ursulin@igalia.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Pass wa bb setup arguments in a struct

Group the function arguments in a struct for more readable code and easier
extending.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250711160153.49833-3-tvrtko.ursulin@igalia.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Generalize wa bb emission code

Generalize the wa bb emission by splitting it into three phases - setup,
emit and finish, and extract setup and finish steps into helpers.

This will enable using the same infrastructure for emitting the indirect
context workarounds.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250711160153.49833-2-tvrtko.ursulin@igalia.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Fix missing kernel-doc

Fix warning:

Warning: drivers/gpu/drm/xe/xe_device_types.h:658 struct member 'wa_active' not described in 'xe_device'

Fixes: 661a6950e061 ("drm/xe: Add infrastructure for Device OOB workarounds")
Cc: Matt Atwood <matthew.s.atwood@intel.com>
Reviewed-by: Jonathan Cavitt <joanthan.cavitt@intel.com>
Link: https://lore.kernel.org/r/20250711214911.2009714-2-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Remove unused functions

xe_bo_create_from_data() last use was removed in 2023 by
commit 0e1a47fcabc8 ("drm/xe: Add a helper for DRM device-lifetime BO
create")

xe_rtp_match_first_gslice_fused_off() last use was removed in 2023 by
commit 4e124151fcfc ("drm/xe/dg2: Drop pre-production workarounds")

Remove them, and xe_dss_mask_empty whose last use was by
xe_rtp_match_first_gslice_fused_off().

(Xe has a bunch ofother symbols that have been added but not used,
given how new it is, I've left those, as opposed to these that
had the code that used them removed).

Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org>
Link: https://lore.kernel.org/r/20250713152531.219326-1-linux@treblig.org
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

Merge tag 'amd-drm-next-6.17-2025-07-11' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-6.17-2025-07-11:

amdgpu:
- Clean up function signatures
- GC 10 KGQ reset fix
- SDMA reset cleanups
- Misc fixes
- LVDS fixes
- UserQ fix

amdkfd:
- Reset fix

Signed-off-by: Simona Vetter <simona.vetter@ffwll.ch>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250711205548.21052-1-alexander.deucher@amd.com

drm/xe: Normalize default param values

Document xe module params with the default values following a similar
strategy for all of them:

1) Define a DEFAULT_* macro with the default value. When the
   value can't be directly stringified, also define a *_STR
   variant
2) Use __stringify() or the _STR variant to make sure the
   default value shows up in the param description

This allows us to show the correct default according to the
configuration. max_vfs for example was wrongly documented for
CONFIG_DRM_XE_DEBUG and svm_notifier_size didn't have its default
documented.

Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://lore.kernel.org/r/20250626-guc-log-level-v3-1-c3ed8b452e91@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe/migrate: Fix alignment check

The check would fail if the address is unaligned, but not when
accounting the offset. Instead of `buf | offset` it should have
been `buf + offset`. To make it more readable and also drop the
uintptr_t, just use the IS_ALIGNED() macro.

Fixes: 270172f64b11 ("drm/xe: Update xe_ttm_access_memory to use GPU for non-visible access")
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://lore.kernel.org/r/20250710-migrate-aligned-v1-1-44003ef3c078@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Remove references to CONFIG_DRM_XE_DEVMEM_MIRROR

The prefetch code was referencing CONFIG_DRM_XE_DEVMEM_MIRROR, which has
been replaced by CONFIG_DRM_XE_PAGEMAP. As a result, prefetches were
limited to SRAM. Update the code to use CONFIG_DRM_XE_PAGEMAP instead of
the deprecated option.

Fixes: f86ad0ed620c ("drm/gpusvm, drm/pagemap: Move migration functionality to drm_pagemap")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20250710205413.1105595-1-matthew.brost@intel.com

drm/xe: Move page fault init after topology init

We need the topology to determine GT page fault queue size, move page
fault init after topology init.

Cc: stable@vger.kernel.org
Fixes: 3338e4f90c14 ("drm/xe: Use topology to determine page fault queue size")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Reviewed-by: Stuart Summers <stuart.summers@intel.com>
Link: https://lore.kernel.org/r/20250710191208.1040215-1-matthew.brost@intel.com

drm/xe/migrate: fix copy direction in access_memory

After we do the modification on the host side, ensure we write the
result back to VRAM and not the other way around, otherwise the
modification will be lost if treated like a read.

Fixes: 270172f64b11 ("drm/xe: Update xe_ttm_access_memory to use GPU for non-visible access")
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250710134128.800756-2-matthew.auld@intel.com

Merge tag 'drm-intel-next-2025-07-10' of https://gitlab.freedesktop.org/drm/i915/kernel into drm-next

Driver Changes:
- DSI panel's version 2 mipi-sequences fix (Hans)

Signed-off-by: Simona Vetter <simona.vetter@ffwll.ch>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/aHA-qrFQ00TaNsGr@intel.com

drm/xe: Dont skip TLB invalidations on VF

Skipping TLB invalidations on VF causing unrecoverable
faults. Probable reason for skipping TLB invalidations
on SRIOV could be lack of support for instruction
MI_FLUSH_DW_STORE_INDEX. Add back TLB flush with some
additional handling.

Helps in resolving,
[  704.913454] xe 0000:00:02.1: [drm:pf_queue_work_func [xe]]
                ASID: 0
                VFID: 0
                PDATA: 0x0d92
                Faulted Address: 0x0000000002fa0000
                FaultType: 0
                AccessType: 1
                FaultLevel: 0
                EngineClass: 3 bcs
                EngineInstance: 8
[  704.913551] xe 0000:00:02.1: [drm:pf_queue_work_func [xe]] Fault response: Unsuccessful -22

V2:
 - Use Xmas tree (MichalW)

Suggested-by: Matthew Brost <matthew.brost@intel.com>
Fixes: 97515d0b3ed92 ("drm/xe/vf: Don't emit access to Global HWSP if VF")
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250710045945.1023840-1-tejas.upadhyay@intel.com
Signed-off-by: Tejas Upadhyay <tejas.upadhyay@intel.com>

Merge tag 'drm-xe-next-2025-07-10' of https://gitlab.freedesktop.org/drm/xe/kernel into drm-next

UAPI Changes:
 - Documentation fixes (Shuicheng)

Cross-subsystem Changes:
 - MTD intel-dg driver for dgfx non-volatile memory device (Sasha)
 - i2c: designware changes to allow i2c integration with BMG (Heikki)

Core Changes:
 - Restructure migration in preparation for multi-device (Brost, Thomas)
 - Expose fan control and voltage regulator version on sysfs (Raag)

Driver Changes:
 - Add WildCat Lake support (Roper)
 - Add aux bus child device driver for NVM on DGFX (Sasha)
 - Some refactor and fixes to allow cleaner BMG w/a (Lucas, Maarten, Auld)
 - BMG w/a (Vinay)
 - Improve handling of aborted probe (Michal)
 - Do not wedge device on killed exec queues (Brost)
 - Init changes for flicker-free boot (Maarten)
 - Fix out-of-bounds field write in MI_STORE_DATA_IMM (Jia)
 - Enable the GuC Dynamic Inhibit Context Switch optimization (Daniele)
 - Drop bo->size (Brost)
 - Builds and KConfig fixes (Harry, Maarten)
 - Consolidate LRC offset calculations (Tvrtko)
 - Fix potential leak in hw_engine_group (Michal)
 - Future-proof for multi-tile + multi-GT cases (Roper)
 - Validate gt in pmu event (Riana)
 - SRIOV PF: Clear all LMTT pages on alloc (Michal)
 - Allocate PF queue size on pow2 boundary (Brost)
 - SRIOV VF: Make multi-GT migration less error prone (Tomasz)
 - Revert indirect ring state patch to fix random LRC context switches failures (Brost)
 - Fix compressed VRAM handling (Auld)
 - Add one additional BMG PCI ID (Ravi)
 - Recommend GuC v70.46.2 for BMG, LNL, DG2 (Julia)
 - Add GuC and HuC to PTL (Daniele)
 - Drop PTL force_probe requirement (Atwood)
 - Fix error flow in display suspend (Shuicheng)
 - Disable GuC communication on hardware initialization error (Zhanjun)
 - Devcoredump fixes and clean up (Shuicheng)
 - SRIOV PF: Downgrade some info to debug (Michal)
 - Don't allocate temporary GuC policies object (Michal)
 - Support for I2C attached MCUs (Heikki, Raag, Riana)
 - Add GPU memory bo trace points (Juston)
 - SRIOV VF: Skip some W/a (Michal)
 - Correct comment of xe_pm_set_vram_threshold (Shuicheng)
 - Cancel ongoing H2G requests when stopping CT (Michal)

Signed-off-by: Simona Vetter <simona.vetter@ffwll.ch>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/aHA7184UnWlONORU@intel.com

Merge tag 'drm-misc-next-2025-07-10' of https://gitlab.freedesktop.org/drm/misc/kernel into drm-next

drm-misc-next for 6.17:

UAPI Changes:

Cross-subsystem Changes:

Core Changes:

Driver Changes:
- amdgpu: debugfs improvements
- ast: Improve hardware generations implementation
- dma-buf heaps:
  - Give the CMA heap a stable name
- panthor: fix UAF in debugfs
- rockchip: Convert inno_hdmi to a bridge
- sti: Convert to devm_drm_bridge_alloc()
- vkms: Use faux_device

- bridge:
  - Improve CEC handling code, convertions to devm_drm_bridge_alloc()

Signed-off-by: Simona Vetter <simona.vetter@ffwll.ch>
From: Maxime Ripard <mripard@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250710-observant-elite-dingo-acfd6d@houat

drm/xe/sriov: Mark BMG as SR-IOV capable

Enable SR-IOV support for BMG platforms. Note that as other flags from
the platform descriptor, it only means it may have that capability: it
still depends on runtime checks for the proper support in HW and
firmware.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Tested-by: Jakub Kolakowski <jakub1.kolakowski@intel.com>
Signed-off-by: Jakub Kolakowski <jakub1.kolakowski@intel.com>
Reviewed-by: Satyanarayana K V P <satyanarayana.k.v.p@intel.com>
Link: https://lore.kernel.org/r/20250710103040.375610-3-jakub1.kolakowski@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: extend Wa_15015404425 to apply to PTL

Wa_15015404425 only needs to be applied on PTL platforms with an A step
compute die. There is no way to map PCI revid to the compute die
stepping. The easiest way to figure out compute die stepping our end is
to map the media IP's stepping to the compute die. For PTL, compute die
has an A stepping if and only if the media IP's stepping is also A-step
(This relationship is determined on a per platform basis and just
happens to be this way on PTL).

In addition this workaround is a chicken-and-egg problem. Wa_15015404425
requires that all register reads be preceded by four dummy MMIO writes
(including during early driver  init and even pre-OS firmware). The
driver needs to perform some MMIO reads during init which include the
GMD_ID register that contains the Media IPs stepping. To handle this in
the safest manner assume the workaround applies to all of PTL during
driver probe and deactivate the workaround after.

The overall solution becomes a set of two workarounds:

* 15015404425 - a Device OOB workaround that's always active for PTL
* 15015404425_disable - a GT OOB workaround that applies to PTL
  platfroms with a B0 or later stepping

The first of these workarounds issues dummy MMIO writes we do when
reading registers. The second guards logic that disables the first once
we have the necessary information later in the probe process.

v2: rename SoC to device, avoid null pointer dereference, update commit
message.
v3: rebase
v5: move disable check into xe_device_probe to avoid linking in xe_wa
into xe_pci, reword commit message
v6: squash extension and b0 support into 1 patch

Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://lore.kernel.org/r/20250709221605.172516-7-matthew.s.atwood@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Move Wa_15015404425 to use the new XE_DEVICE_WA macro

Move Wa_15015404425 to use the new implemented OOB macro XE_DEVICE_WA()

v2: rename from SoC to Device
v5: move workaround call back into the flush call
v6: remove redundant commenting

Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://lore.kernel.org/r/20250709221605.172516-6-matthew.s.atwood@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: Add infrastructure for Device OOB workarounds

Some workarounds need to be able to be applied ahead of any GT
initialization for example 15015404425. This patch creates XE_DEVICE_WA
macro, in the same vein as XE_WA. This macro can be used ahead of GT
initialization, and can be tracked in sysfs. This should alleviate some
of the complexities that exist in i915.

v2: name change SoC to Device, address style issues
v5: split into separate patch from RTP changes, put oob within a struct,
move the initiation of oob workarounds into xe_device_probe_early(),
clean up the comments around XE_WA.

Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://lore.kernel.org/r/20250709221605.172516-5-matthew.s.atwood@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: add new type to RTP context

Prepare the RTP context to be used before GT init. Add the xe device as
a type, put WARN_ONs to protect existing RTP_MATCHes.

v5: split out into separate patch, change definition order
v6: catch missing cases for checking gt init

Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://lore.kernel.org/r/20250709221605.172516-4-matthew.s.atwood@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: add xe_device_wa infrastructure

There are some workarounds that must be appplied before gt init,
wa_15015404425 for example. Instead of sprinking them conditionally
throughout the driver as we did for i915 generate an oob.rules file
reusing the RTP infrastructure to make these easier to track.

v2: rename xe_soc_wa to xe_device_wa
v5: derive prefix from argument rather than hard coding the values.
v6: split out xe_gen-wa_oob changes

Signed-off-by: Matt Atwood <matthew.s.atwood@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250709221605.172516-3-matthew.s.atwood@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe: prepare xe_gen_wa_oob to be multi-use

There is a need for additional oob rules files. Make the current gen
file more robust to support more files.

Signed-off-by: Matt Atwood <matthew.s.atwood@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250709221605.172516-2-matthew.s.atwood@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe/guc: Cancel ongoing H2G requests when stopping CT

Once we have started a GT reset sequence, which includes stopping
GuC CTB communication, we should also cancel all ongoing H2G send-
recv requests, as either GuC is already dead, or due to imminent
reset GuC will not be able to reply, or due to internal cleanup
we will lose pending fences. With this we will report dedicated
-ECANCELED error instead of misleading -ETIME.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Acked-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250709174038.1876-4-michal.wajdeczko@intel.com

drm/xe/guc: Move state change logger to helper

In the state change helper we are already doing extra stuff,
move debug state logger there to cover all state changes.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Acked-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250709174038.1876-3-michal.wajdeczko@intel.com

drm/xe/guc: Rename CT state change helper

In this helper we are already doing much more than just setting
a new CT state and its name was little misleading. Rename it.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Acked-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250709174038.1876-2-michal.wajdeczko@intel.com

drm/xe/pm: Correct comment of xe_pm_set_vram_threshold()

The parameter threshold is with size in MiB, not in bits.
Correct it to avoid any confusion.

v2: s/mb/MiB, s/vram/VRAM, fix return section. (Michal)

Fixes: 30c399529f4c ("drm/xe: Document Xe PM component")
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Shuicheng Lin <shuicheng.lin@intel.com>
Link: https://lore.kernel.org/r/20250708021450.3602087-2-shuicheng.lin@intel.com
Reviewed-by: Stuart Summers <stuart.summers@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/xe/bmg: Don't use WA 16023588340 and 22019338487 on VF

These workarounds are not applicable for use by the VFs.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Tested-by: Jakub Kolakowski <jakub1.kolakowski@intel.com>
Reviewed-by: Satyanarayana K V P <satyanarayana.k.v.p@intel.com>
Signed-off-by: Jakub Kolakowski <jakub1.kolakowski@intel.com>
Link: https://lore.kernel.org/r/20250710103040.375610-2-jakub1.kolakowski@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/i915/bios: Apply vlv_fixup_mipi_sequences() to v2 mipi-sequences too

It turns out that the fixup from vlv_fixup_mipi_sequences() is necessary
for some DSI panel's with version 2 mipi-sequences too.

Specifically the Acer Iconia One 8 A1-840 (not to be confused with the
A1-840FHD which is different) has the following sequences:

BDB block 53 (1284 bytes) - MIPI sequence block:
Sequence block version v2
Panel 0 *

Sequence 2 - MIPI_SEQ_INIT_OTP
GPIO index 9, source 0, set 0 (0x00)
Delay: 50000 us
GPIO index 9, source 0, set 1 (0x01)
Delay: 6000 us
GPIO index 9, source 0, set 0 (0x00)
Delay: 6000 us
GPIO index 9, source 0, set 1 (0x01)
Delay: 25000 us
Send DCS: Port A, VC 0, LP, Type 39, Length 5, Data ff aa 55 a5 80
Send DCS: Port A, VC 0, LP, Type 39, Length 3, Data 6f 11 00
...
Send DCS: Port A, VC 0, LP, Type 05, Length 1, Data 29
Delay: 120000 us

Sequence 4 - MIPI_SEQ_DISPLAY_OFF
Send DCS: Port A, VC 0, LP, Type 05, Length 1, Data 28
Delay: 105000 us
Send DCS: Port A, VC 0, LP, Type 05, Length 2, Data 10 00
Delay: 10000 us

Sequence 5 - MIPI_SEQ_ASSERT_RESET
Delay: 10000 us
GPIO index 9, source 0, set 0 (0x00)

Notice how there is no MIPI_SEQ_DEASSERT_RESET, instead the deassert
is done at the beginning of MIPI_SEQ_INIT_OTP, which is exactly what
the fixup from vlv_fixup_mipi_sequences() fixes up.

Extend it to also apply to v2 sequences, this fixes the panel not working
on the Acer Iconia One 8 A1-840.

Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/14605
Signed-off-by: Hans de Goede <hansg@kernel.org>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: https://lore.kernel.org/r/20250703143824.7121-1-hansg@kernel.org
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/xe/bo: add GPU memory trace points

Add TRACE_GPU_MEM tracepoints for tracking global GPU memory usage.

These are required by VSR on Android 12+ for reporting GPU driver memory
allocations.

v5:
 - Drop process_mem tracking
 - Set the gpu_id field to dev->primary->index (Lucas, Tvrtko)
 - Formatting cleanup under 80 columns

v3:
 - Use now configurable CONFIG_TRACE_GPU_MEM instead of adding a
   per-driver Kconfig (Lucas)

v2:
 - Use u64 as preferred by checkpatch (Tvrtko)
 - Fix errors in comments/Kconfig description (Tvrtko)
 - drop redundant "CONFIG" in Kconfig

Signed-off-by: Juston Li <justonli@chromium.org>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250709192313.479336-2-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/xe/xe_i2c: Add support for i2c in survivability mode

Initialize i2c in survivability mode to allow firmware
update of Add-In Management Controller (AMC) in
survivability mode.

Signed-off-by: Riana Tauro <riana.tauro@intel.com>
Signed-off-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Reviewed-by: Raag Jadav <raag.jadav@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://lore.kernel.org/r/20250701122252.2590230-6-heikki.krogerus@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/xe/pm: Wire up suspend/resume for I2C controller

Wire up suspend/resume handles for I2C controller to match its power
state with SGUnit.

Signed-off-by: Raag Jadav <raag.jadav@intel.com>
Signed-off-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Reviewed-by: Karthik Poosa <karthik.poosa@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://lore.kernel.org/r/20250701122252.2590230-5-heikki.krogerus@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/xe: Support for I2C attached MCUs

Adding adaption/glue layer where the I2C host adapter
(Synopsys DesignWare I2C adapter) and the I2C clients (the
microcontroller units) are enumerated.

The microcontroller units (MCU) that are attached to the GPU
depend on the OEM. The initially supported MCU will be the
Add-In Management Controller (AMC).

Co-developed-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://lore.kernel.org/r/20250701122252.2590230-4-heikki.krogerus@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
[Rodrigo fixed the co-developed tags and SPDX format in the .c file]

i2c: designware: Add quirk for Intel Xe

The regmap is coming from the parent also in case of Xe
GPUs. Reusing the Wangxun quirk for that.

Acked-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Co-developed-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Reviewed-by: Andi Shyti <andi.shyti@kernel.org>
Link: https://lore.kernel.org/r/20250701122252.2590230-3-heikki.krogerus@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
[Rodrigo fixed the co-developed tags while merging]

i2c: designware: Use polling by default when there is no irq resource

The irq resource itself can be used as a generic way to
determine when polling is needed.

This not only removes the need for special additional device
properties that would soon be needed when the platform may
or may not have the irq, but it also removes the need to
check the platform in the first place in order to determine
is polling needed or not.

Signed-off-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Reviewed-by: Andi Shyti <andi.shyti@kernel.org>
Link: https://lore.kernel.org/r/20250701122252.2590230-2-heikki.krogerus@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/xe/guc: Don't allocate temporary policies object

Since we are already using reusable buffer objects from the GuC
buffer cache, we can directly write into their CPU pointers and
spare unnecessary temporary allocation.

While around, also make sure to clear obtained buffer, to avoid
sending some stale data.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250702142504.1656-1-michal.wajdeczko@intel.com

drm/xe/pf: Print configuration KLVs using debug printer

While we print VF's configuration KLVs only under DEBUG_SRIOV
config, we should be doing it at debug level, not info level.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Lukasz Laguna <lukasz.laguna@intel.com>
Reviewed-by: Lukasz Laguna <lukasz.laguna@intel.com>
Link: https://lore.kernel.org/r/20250703145709.1832-1-michal.wajdeczko@intel.com

drm/xe/pf: Print runtime registers using debug printer

While we already print VF's runtime registers only under DEBUG_SRIOV
config, we should be still doing it at debug level, not info.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Lukasz Laguna <lukasz.laguna@intel.com>
Link: https://lore.kernel.org/r/20250604190021.725-2-michal.wajdeczko@intel.com

drm/panthor: Fix UAF in panthor_gem_create_with_handle() debugfs code

The object is potentially already gone after the drm_gem_object_put().
In general the object should be fully constructed before calling
drm_gem_handle_create(), except the debugfs tracking uses a separate
lock and list and separate flag to denotate whether the object is
actually initialized.

Since I'm touching this all anyway simplify this by only adding the
object to the debugfs when it's ready for that, which allows us to
delete that separate flag. panthor_gem_debugfs_bo_rm() already checks
whether we've actually been added to the list or this is some error
path cleanup.

v2: Fix build issues for !CONFIG_DEBUGFS (Adrián)

v3: Add linebreak and remove outdated comment (Liviu)

Fixes: a3707f53eb3f ("drm/panthor: show device-wide list of DRM GEM objects over DebugFS")
Cc: Adrián Larumbe <adrian.larumbe@collabora.com>
Cc: Boris Brezillon <boris.brezillon@collabora.com>
Cc: Steven Price <steven.price@arm.com>
Cc: Liviu Dudau <liviu.dudau@arm.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Signed-off-by: Simona Vetter <simona.vetter@intel.com>
Signed-off-by: Simona Vetter <simona.vetter@ffwll.ch>
Reviewed-by: Steven Price <steven.price@arm.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://lore.kernel.org/r/20250709135220.1428931-1-simona.vetter@ffwll.ch

drm/xe: Expose fan control and voltage regulator version

Add sysfs attributes for late binding features which expose bound version
to the user.

v2: Rework attribute and macro naming (Badal)
v3: Drop fancy formatting (Rodrigo)
v4: Form version string using local variables (Rodrigo)

Signed-off-by: Raag Jadav <raag.jadav@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20250709164224.2676086-1-raag.jadav@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

fbcon: Fix outdated registered_fb reference in comment

The variable was renamed to fbcon_registered_fb, but this comment was
not updated along with the change. Correct it to avoid confusion.

Signed-off-by: Shixiong Ou <oushixiong@kylinos.cn>
Fixes: efc3acbc105a ("fbcon: Maintain a private array of fb_info")
[sima: Add Fixes: line.]
Signed-off-by: Simona Vetter <simona.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20250709103438.572309-1-oushixiong1025@163.com

drm/ast: Gen7: Switch default registers to gen4+ state

Change the default register settings for Gen7 to mach Gen4 and
later. Gen7 currently uses the settings for Gen1, which is most
likely incorrect.

Using Gen4+ settings enables E2M linear-access modes in VGACRA2.
It appears to be related to the chip's PCIE2MBOX feature, which
is unused.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250706162816.211552-11-tzimmermann@suse.de

drm/ast: Gen7: Disable VGASR0[1] as on Gen4+

Set VGACRB6[5], which disables asynchronous sequencer resets via
VGASR0[1]. This was most likely an oversight when adding support
for Gen7. Aligns Gen7 with the earlier Gen4+.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250706162816.211552-10-tzimmermann@suse.de

drm/ast: Split ast_set_def_ext_reg() by chip generation

Duplicate ast_set_def_ext_reg() for individual chip generations
and move call it into per-chip source files. Remove the original
code. AST2100 and AST2500 reuse the function from earlier chips.
AST2600 appears to be incorrect as it uses an older function. Keep
this behavior for now.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250706162816.211552-9-tzimmermann@suse.de

drm/ast: Handle known struct ast_dramstruct with helpers

Most of struct ast_dramstruct stores hardware state. Some index
values have known or special meaning. The known values are

- 0xffff - Terminal entry in the array
- 0xff00 - Delays the programming for usecs
- 0x0004 - Sets the type of DRAM

Add constants and helper macros for these cases. Also add a helper
macro for testing. Update Gen1 and Gen2+ accordingly.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250706162816.211552-8-tzimmermann@suse.de

drm/ast: Move struct ast_dramstruct to ast_post.h

Declare struct ast_dramstruct in ast_post.h and remove its original
header file.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250706162816.211552-7-tzimmermann@suse.de

drm/ast: Move Gen2+ and Gen1 POST code to separate source files

Move POST code for Gen2+ and Gen1 to separate source files and
hide it in ast_2100_post() ans ast_2000_post(). With P2A
configuration, the POST logic for these chip generations has
been mingled in ast_init_dram_reg(). Hence, handle all generations
in a single change. The split simplifies both cases. Also move
the DRAM init tables for each Gen into the respective source
file. No changes to the overall logic.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250706162816.211552-6-tzimmermann@suse.de

drm/ast: Move Gen4+ POST code to separate source file

Move POST code for Gen4+ to separate source file and hide it in
ast_2300_post(). With P2A configuration, it performs a full board
POST and enables the transmitter chip; otherwise it only enables the
transmitter chip.

Also fix coding style in several places. No changes to the overall
logic.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250706162816.211552-5-tzimmermann@suse.de

drm/ast: Move Gen6+ POST code to separate source file

Move POST code for Gen6+ to separate source file and hide it in
ast_2500_post(). With P2A configuration, it performs a full board
POST; otherwise it enables the transmitter chip. No changes to the
overall logic.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250706162816.211552-4-tzimmermann@suse.de

drm/ast: Move Gen7+ POST code to separate source file

Move POST code for Gen7+ to separate source file and hide it in
ast_2600_post(). There's not much going on here except for enabling
the DP transmitter chip.

v2:
- simplify logic (Jocelyn)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250706162816.211552-3-tzimmermann@suse.de

drm/ast: Declare helpers for POST in header

Provide POST helpers in header file before splitting up the AST
POST code.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250706162816.211552-2-tzimmermann@suse.de

dma-buf: heaps: Give default CMA heap a fixed name

The CMA heap's name in devtmpfs can vary depending on how the heap is
defined. Its name defaults to "reserved", but if a CMA area is defined
in the devicetree, the heap takes on the devicetree node's name, such as
"default-pool" or "linux,cma". To simplify naming, unconditionally name
it "default_cma_region", but keep a legacy node in place backed by the
same underlying allocator for backwards compatibility.

Reviewed-by: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Jared Kangas <jkangas@redhat.com>
Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
Link: https://lore.kernel.org/r/20250610131231.1724627-4-jkangas@redhat.com

dma-buf: heaps: Parameterize heap name in __add_cma_heap()

Prepare for the introduction of a fixed-name CMA heap by replacing the
unused void pointer parameter in __add_cma_heap() with the heap name.

Reviewed-by: Maxime Ripard <mripard@kernel.org>
Acked-by: John Stultz <jstultz@google.com>
Signed-off-by: Jared Kangas <jkangas@redhat.com>
Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
Link: https://lore.kernel.org/r/20250610131231.1724627-3-jkangas@redhat.com

Documentation: dma-buf: heaps: Fix code markup

Code snippets should be wrapped in double backticks to follow
reStructuredText semantics; the use of single backticks uses the
:title-reference: role by default, which isn't quite what we want.
Add double backticks to code snippets to fix this.

Reviewed-by: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Jared Kangas <jkangas@redhat.com>
Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
Link: https://lore.kernel.org/r/20250610131231.1724627-2-jkangas@redhat.com

dma-buf: system_heap: No separate allocation for attachment sg_tables

struct dma_heap_attachment is a separate allocation from the struct
sg_table it contains, but there is no reason for this. Let's use the
slab allocator just once instead of twice for dma_heap_attachment.

Signed-off-by: T.J. Mercier <tjmercier@google.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
Link: https://lore.kernel.org/r/20250417180943.1559755-1-tjmercier@google.com

drm/sched: Consolidate drm_sched_rq_select_entity_rr

Extract out two copies of the identical code to
drm_sched_rq_select_entity_rr()'s epilogue to make it smaller and more
readable.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Danilo Krummrich <dakr@kernel.org>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Philipp Stanner <phasta@kernel.org>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
[phasta: commit message]
Signed-off-by: Philipp Stanner <phasta@kernel.org>
Link: https://lore.kernel.org/r/20250708122121.75689-1-tvrtko.ursulin@igalia.com

drm/sti: hda: convert to devm_drm_bridge_alloc() API

devm_drm_bridge_alloc() is the new API to be used for allocating (and
partially initializing) a private driver struct embedding a struct
drm_bridge.

This driver was missed during the automated conversion in commit
9c399719cfb9 ("drm: convert many bridge drivers from devm_kzalloc() to
devm_drm_bridge_alloc() API") and following commits.

The lack of conversion for this driver is a bug since commit a7748dd127ea
("drm/bridge: get/put the bridge reference in drm_bridge_add/remove()")
which is the first commmit having added a drm_bridge_get/put() pair and
thus exposing the incorrect initial refcount issue.

Fix this by switching the driver to the new API.

Reported-by: Marek Szyprowski <m.szyprowski@samsung.com>
Closes: https://lore.kernel.org/all/ce9c6aa3-5372-468f-a4bf-5a261259e459@samsung.com/
Fixes: a7748dd127ea ("drm/bridge: get/put the bridge reference in drm_bridge_add/remove()")
Reviewed-by: Maxime Ripard <mripard@kernel.org>
Link: https://lore.kernel.org/r/20250708-drm-bridge-convert-to-alloc-api-leftovers-v1-2-6285de8c3759@bootlin.com
Signed-off-by: Luca Ceresoli <luca.ceresoli@bootlin.com>

drm/sti: hdmi: convert to devm_drm_bridge_alloc() API

devm_drm_bridge_alloc() is the new API to be used for allocating (and
partially initializing) a private driver struct embedding a struct
drm_bridge.

This driver was missed during the automated conversion in commit
9c399719cfb9 ("drm: convert many bridge drivers from devm_kzalloc() to
devm_drm_bridge_alloc() API") and following commits.

The lack of conversion for this driver is a bug since commit a7748dd127ea
("drm/bridge: get/put the bridge reference in drm_bridge_add/remove()")
which is the first commmit having added a drm_bridge_get/put() pair and
thus exposing the incorrect initial refcount issue.

Fix this by switching the driver to the new API.

Reported-by: Marek Szyprowski <m.szyprowski@samsung.com>
Closes: https://lore.kernel.org/all/ce9c6aa3-5372-468f-a4bf-5a261259e459@samsung.com/
Fixes: a7748dd127ea ("drm/bridge: get/put the bridge reference in drm_bridge_add/remove()")
Reviewed-by: Maxime Ripard <mripard@kernel.org>
Link: https://lore.kernel.org/r/20250708-drm-bridge-convert-to-alloc-api-leftovers-v1-1-6285de8c3759@bootlin.com
Signed-off-by: Luca Ceresoli <luca.ceresoli@bootlin.com>

drm/amdgpu: fix the logic to validate fpriv and root bo

Fix the smatch warning,
smatch warnings:
drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c:2146 amdgpu_pt_info_read()
error: we previously assumed 'fpriv' could be null (see line 2146)

"if (!fpriv && !fpriv->vm.root.bo)", It has to be an OR condition
rather than an AND which makes an NULL dereference in case fpriv is NULL.

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202507090525.9rDWGhz3-lkp@intel.com/
Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Link: https://lore.kernel.org/r/20250709071618.591866-1-sunil.khatri@amd.com
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>

drm/amdgpu: fix MQD debugfs undefined symbol when DEBUG_FS=n

Fix undefined reference to amdgpu_mqd_info_fops during
debugfs_create_file if DEBUG_FS=n

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Link: https://lore.kernel.org/r/20250708101551.68033-1-sunil.khatri@amd.com
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>

drm/xe: Release runtime pm for error path of xe_devcoredump_read()

xe_pm_runtime_put() is missed to be called for the error path in
xe_devcoredump_read().
Add function description comments for xe_devcoredump_read() to help
understand it.

v2: more detail function comments and refine goto logic (Matt)

Fixes: c4a2e5f865b7 ("drm/xe: Add devcoredump chunking")
Cc: stable@vger.kernel.org
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Shuicheng Lin <shuicheng.lin@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250707004911.3502904-6-shuicheng.lin@intel.com

drm/xe: Remove unused code in devcoredump_snapshot()

The deleted code is no longer needed because patch "drm/xe/guc: Plumb
GuC-capture into dev coredump" has removed the related usage code.
Remove the code to tidy up the function.

v2: s/bacause/because

Reviewed-by: Zhanjun Dong <zhanjun.dong@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Shuicheng Lin <shuicheng.lin@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250707004911.3502904-5-shuicheng.lin@intel.com

drm/xe/uc: Disable GuC communication on hardware initialization error

Disable GuC communication on Xe micro controller hardware initialization
error.

Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/4917
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Zhanjun Dong <zhanjun.dong@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250707231108.3217573-1-zhanjun.dong@intel.com

drm/xe/pm: Restore display pm if there is error after display suspend

xe_bo_evict_all() is called after xe_display_pm_suspend(). So if there
is error with xe_bo_evict_all(), display pm should be restored.

Fixes: 51462211f4a9 ("drm/xe/pxp: add PXP PM support")
Fixes: cb8f81c17531 ("drm/xe/display: Make display suspend/resume work on discrete")
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Shuicheng Lin <shuicheng.lin@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Link: https://lore.kernel.org/r/20250708035424.3608190-2-shuicheng.lin@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

gpu/trace: make TRACE_GPU_MEM configurable

Move the source to a better place in Device Drivers -> Graphics support
now that its configurable.

v4:
 - Move source location (Tvrtko)

v3:
 - Patch introduced to replace per-driver config (Lucas)

Signed-off-by: Juston Li <justonli@chromium.org>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250611225145.1739201-1-justonli@chromium.org
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drm/rockchip: vop2: Fix the update of LAYER/PORT select registers when there are multi display output on rk3588/rk3568

The all video ports of rk3568/rk3588 share the same OVL_LAYER_SEL
and OVL_PORT_SEL registers, and the configuration of these two registers
can be set to take effect when the vsync signal arrives at a certain Video
Port.

If two threads for two display output choose to update these two registers
simultaneously to meet their own plane adjustment requirements(change plane
zpos or switch plane from one crtc to another), then no matter which Video
Port'svsync signal we choose to follow for these two registers, the display
output of the other Video Port will be abnormal.
This is because the configuration of this Video Port does not take
effect at the right time (its configuration should take effect when its
VSYNC signal arrives).

In order to solve this problem, when performing plane migration or
change the zpos of planes, there are two things to be observed and
followed:

1. When a plane is migrated from one VP to another, the configuration of
   the layer can only take effect after the Port mux configuration is
   enabled.

2. When change the zpos of planes, we must ensure that the change for
   the previous VP takes effect before we proceed to change the next VP.
   Otherwise, the new configuration might overwrite the previous one for
   the previous VP, or it could lead to the configuration of the previous
   VP being take effect along with the VSYNC of the new VP.

This issue only occurs in scenarios where multi-display output is enabled.

Fixes: c5996e4ab109 ("drm/rockchip: vop2: Make overlay layer select register configuration take effect by vsync")
Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20250421102156.424480-1-andyshrk@163.com

drm/rockchip: inno_hdmi: Simpify clk get/enable by devm_clk_get_enabled api

Make use of devm_clk_get_enabled() to replace devm_clk_get() and
clk_prepare_enable(), which will make the cleanup of clk code simpler.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20250512124615.2848731-9-andyshrk@163.com

drm/rockchip: inno_hdmi: switch i2c registration to devm functions

Switch from i2c_add_adapter() to resource managed
devm_i2c_add_adapter(), which will make the cleanup code more simpler.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20250512124615.2848731-8-andyshrk@163.com

drm/rockchip: inno_hdmi: Use sleep_range instead of udelay

usleep_range is preferred over udelay.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20250512124615.2848731-7-andyshrk@163.com

drm/rockchip: inno_hdmi: Move ddc/i2c configuration and HOTPLUG unmute to inno_hdmi_init_hw

Putting these scattered initialization code together is helpful
for the following migration to the DRM bridge driver mode.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20250512124615.2848731-6-andyshrk@163.com

drm/rockchip: inno_hdmi: Rename function inno_hdmi_reset to inno_hdmi_init_hw

This function not only configure hardware reset register, but also
do some other configurations. Therefore, it is more appropriate to
name it inno_hdmi_init_hw, which will also facilitate the addition
of other functions to this function in the following patch.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20250512124615.2848731-5-andyshrk@163.com

drm/rockchip: inno_hdmi: Remove unnecessary parentheses to make checkpatch happy

Remove unnecessary parentheses to make checkpatch happy.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20250512124615.2848731-4-andyshrk@163.com

drm/rockchip: inno_hdmi: Refactor register macros to make checkpatch happy

1. Prefer using the BIT macro
2. Macro argument 'n' as '(n)' to avoid precedence issues
3. Add a blank line after enum declarations

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20250512124615.2848731-3-andyshrk@163.com

drm/rockchip: inno_hdmi: Merge register definition to c file

Since this register definition is only use in one single c
file, there is no need to put it in a separate header.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20250512124615.2848731-2-andyshrk@163.com

drm/rockchip: dw_hdmi: Use dev_err_probe() to simplify code

In the probe path, dev_err() can be replaced with dev_err_probe()
which will check if error code is -EPROBE_DEFER and prints the
error name. It also sets the defer probe reason which can be
checked later through debugfs.

Signed-off-by: Yumeng Fang <fang.yumeng@zte.com.cn>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20250515203554564-j1jBXUXR6bdiN6zARicC@zte.com.cn

drm/rockchip: cdn-dp: Convert to drm bridge

Convert it to drm bridge driver, it will be convenient for us to
migrate the connector part to the display driver later.
Considering that some code depend on the connector, the following
changes have been made:
- Only process edid in &drm_bridge_funcs.edid_read(), so no need to
store additional edid info.
- Now cdn_dp_get_sink_capability() only focused on reading DPCD_REV.
- Update bpc info in cdn_dp_bridge_atomic_enable() instead of
cdn_dp_encoder_mode_set(). Actually, the bpc data will be used in
cdn_dp_bridge_atomic_enable().
- Switch to use DRM_BRIDGE_OP_DP_AUDIO helpers.

This patch also convert to use devm_drm_bridge_alloc() API.

Tested with RK3399 EVB IND board.

Signed-off-by: Chaoyi Chen <chaoyi.chen@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20250529071334.441-1-kernel@airkyi.com