Merge tag 'amd-drm-next-6.4-2023-04-14' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-6.4-2023-04-14:

amdgpu:
- S4 fixes for APUs
- GFX11 fixes
- Misc code cleanups
- DCN 3.2 fixes
- DCN 3.1.4 fixes
- FPO/FAMS work to improve display power savings
- DP fixes
- UMC 8.10 code cleanup
- SDMA v4 fix
- GPU clock counter fixes
- SMU 13 fixes
- Sdma v6 invalidation fix for preemption
- RAS fixes
- S0ix fix
- GC 9.4.3 updates

amdkfd:
- Fix user pointers with IOMMU
- Fix coherency flag handling

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230414204609.7942-1-alexander.deucher@amd.com

drm/amdgpu: add some basic elements for multiple XCD case

Add some basic definitions and structure member. Inscrease MAX_WB slots
to 1024 to support the increasing number of rings for multiple partitions.

v2: unify naming style

Signed-off-by: Le Ma <le.ma@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move vmhub out of amdgpu_ring_funcs (v4)

It looks better to place this field in ring
structure. Also drop the repeated ring funcs definitions
if there's no difference except for vmhub field.

v2: rename the field to vm_hub like others (Le)
v3: apply the changes to new ip blocks (Hawking)
v4: fix vcn sw ring (Alex)

Signed-off-by: Le Ma <le.ma@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amdgpu: enable ras for mp0 v13_0_10 on SRIOV"

This reverts commit fe120b9f5ce873516a2604e4ff0c19084be94e8c.
This patch impacts sriov multi-vf stability

Signed-off-by: Jane Jian <Jane.Jian@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add common ip block for GC 9.4.3

Add common IP handling for GC 9.4.3

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Le Ma <Le.Ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add logging when DP link training Clock recovery is Successful

Log when Clock recovery is successful, as part of
DP link training process.

Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Cc: Fangzhi Zuo <Jerry.Zuo@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add common early init support for GC 9.4.3

init asic funcs and cp/pg flags for GC 9.4.3

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Le Ma <Le.Ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: switch to v9_4_3 gfx_funcs callbacks for GC 9.4.3

add gfx_funcs callbacks implemenation based on
gc_v9_4_3 ip headers

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add logging when setting DP sink power state fails

Log if we fail to setup sink power states.

Cc: Fangzhi Zuo <Jerry.Zuo@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Add gfx_target_version for GC 9.4.3

Required for Thunk GFX version sysfs query.

Signed-off-by: Graham Sider <Graham.Sider@amd.com>
Reviewed-by: Amber Lin <Amber.Lin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Enable HW_UPDATE_RPTR on GC 9.4.3

GC 9.4.3 uses the hardware to update AQL queues read pointer, so
remove CP_HQD_PQ_CONTROL__NO_UPDATE_RPTR_MASK flag from MQD if it's
GC 9.4.3, and keep it for other existing gfx9 ASICs.

Signed-off-by: Amber Lin <Amber.Lin@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: reserve the old gc_11_0_*_mes.bin

Reserve the MOUDLE_FIRMWARE declaration of gc_11_0_*_mes.bin
to fix falling back to old mes bin on failure via autoload.

Fixes: 97998b893c30 ("drm/amd/amdgpu: introduce gc_*_mes_2.bin v2")
Signed-off-by: Li Ma <li.ma@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: change the reference clock for raven/raven2

Due to switch to golden tsc register to get clock counter for raven/ raven2.
Chang the reference clock from 25MHZ to 100MHZ.

Suggested-by: shanshengwang <shansheng.wang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Jesse Zhang <jesse.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: skip kfd-iommu suspend/resume for S0ix

GFX is in gfxoff mode during s0ix so we shouldn't need to
actually execute kfd_iommu_suspend/kfd_iommu_resume operation.

Signed-off-by: Aaron Liu <aaron.liu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add gc v9_4_3 rlc_funcs implementation

all the gc v9_4_3 registers fall in gc_rlcpdec address range
have different relative offsets and base_idx from the ones
defined in gc v9_0 ip headers. gc_v9_0_rlc_funcs can not be
reused anymore for gc v9_4_3

v2: drop unused handshake function (Alex)

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Le Ma <Le.Ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Merge tag 'drm-intel-next-fixes-2023-04-13' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

Just one Cc:stable fix for sampler indirect state in bindless heap.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZDfxo+PXyw9ivFLI@jlahtine-mobl.ger.corp.intel.com

drm/amdgpu: drop temp programming for pagefault handling

Was introduced as workaround. not needed anymore

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Jack Gui <Jack.Gui@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: include protection for doorbell.h

This patch adds double include protection for doorbell.h

Cc: Christian Koenig <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian Koenig <christian.koenig@amd.com>
Signed-off-by: Shashank Sharma <shashank.sharma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: rename num_doorbells

Rename doorbell.num_doorbells to doorbell.num_kernel_doorbells to
make it more readable.

Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Christian Koenig <christian.koenig@amd.com>
Acked-by: Christian Koenig <christian.koenig@amd.com>
Signed-off-by: Shashank Sharma <shashank.sharma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: switch to golden tsc registers for raven/raven2

Due to raven/raven2 maybe enable  sclk slow down,
they cannot get clock count by the RLC at the auto level of dpm performance.
So switch to golden tsc register.

Suggested-by: shanshengwang <shansheng.wang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Jesse Zhang <jesse.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: correct the pcie link state check for SMU13

Update the driver implementations to fit those data exposed
by PMFW.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add gfx v11_0_3 fed irq handling for sriov

Add gfx v11_0_3 fed irq handling for sriov.

Signed-off-by: YiPeng Chai <YiPeng.Chai@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Rework retry fault removal

Rework retry fault removal from the software filter by
storing an expired timestamp for a fault that is being removed.
When a new fault comes, and it matches an entry in the sw filter,
it will be added as a new fault only when its timestamp is greater
than the timestamp expiry of the fault in the sw filter.
This helps in avoiding stale faults being added back into the
filter and preventing legitimate faults from being handled.

Suggested-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Reviewed-by: Philip Yang <Philip.Yang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Enable IH retry CAM on GFX9

This patch enables the IH retry CAM on GFX9 series cards. This
retry filter is used to prevent sending lots of retry interrupts
in a short span of time and overflowing the IH ring buffer. This
will also help reduce CPU interrupt workload.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: remove unused num_of_active_display variable

clang with W=1 reports
drivers/gpu/drm/amd/amdgpu/../pm/swsmu/amdgpu_smu.c:1700:6: error: variable
  'num_of_active_display' set but not used [-Werror,-Wunused-but-set-variable]
        int num_of_active_display = 0;
            ^
This variable is not used so remove it.

Fixes: 75145aab7a0d ("drm/amdgpu/swsmu: clean up a bunch of stale interfaces")
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: simplify amdgpu_ras_eeprom.c

All chips that support RAS also support IP discovery, so
use the IP versions rather than a mix of IP versions and
asic types.  Checking the validity of the atom_ctx pointer
is not required as the vbios is already fetched at this
point.

v2: add comments to id asic types based on feedback from Luben

Reviewed-by: Luben Tuikov <luben.tuikov@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: Luben Tuikov <luben.tuikov@amd.com>

Merge tag 'drm-misc-next-2023-04-12' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next for v6.4-rc1:

Cross-subsystem Changes:
- Convert MIPI DSIM bridge dt to yaml.

Core Changes:
- Fix UAF race in drm scheduler.

Driver Changes:
- Add primary plane positioning support to VKMS.
- Convert omapdrm fbdev emulation to in-kernel client.
- Assorted small fixes to vkms, vc4, nouveau, vmwgfx.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/b7c37d4e-8f16-85dc-0f5f-3bd98f961395@linux.intel.com

drm/i915: disable sampler indirect state in bindless heap

By default the indirect state sampler data (border colors) are stored
in the same heap as the SAMPLER_STATE structure. For userspace drivers
that can be 2 different heaps (dynamic state heap & bindless sampler
state heap). This means that border colors have to copied in 2
different places so that the same SAMPLER_STATE structure find the
right data.

This change is forcing the indirect state sampler data to only be in
the dynamic state pool (more convenient for userspace drivers, they
only have to have one copy of the border colors). This is reproducing
the behavior of the Windows drivers.

BSpec: 46052

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: stable@vger.kernel.org
Reviewed-by: Haridhar Kalvala <haridhar.kalvala@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230407093237.3296286-1-lionel.g.landwerlin@intel.com
(cherry picked from commit 16fc9c08f0ec7b1c95f1ea4a16097acdb3fc943d)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/amdgpu: Enable GFX11 SDMA context empty interrupt

Enable SDMA queue empty context switching. SDMA context switch due to
quantum programming no longer done here (as of sdma v6), so re-name
sdma_v6_0_ctx_switch_enable to sdma_v6_0_ctxempty_int_enable to reflect
this.

Also program SDMAx_QUEUEx_SCHEDULE_CNTL for context switch due to
quantum in KFD. Set to amdgpu_sdma_phase_quantum (defaults to 32 i.e.
3200us).

Signed-off-by: Graham Sider <Graham.Sider@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Reviewed-by: Stanley Yang <Stanley.Yang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Check PCIe atomics support on GFX11 to set CP_HQD_HQ_STATUS0[29]

CP_HQD_HQ_STATUS0[29] bit will be used by CPFW to acknowledge whether
PCIe atomics are supported. The default value of this bit is set
to 0. Driver will check whether PCIe atomics are supported and set the
bit to 1 if supported. This will force CPFW to use real atomic ops.
If the bit is not set, CPFW will default to read/modify/write using the
firmware itself.

This is applicable only to GFX11 RS64 CP with MEC FW >= 509. If MEC
FW < 509 and for all GFX11 F32 CP, PCIe atomics needs to be supported
else it will skip the device.

This commit also involves moving amdgpu_amdkfd_device_probe() function
call after per-IP early_init loop in amdgpu_device_ip_early_init()
function so as to check for RS64 enabled device.

Signed-off-by: Sreekant Somasekharan <sreekant.somasekharan@amd.com>
Reviewed-by: Graham Sider <Graham.Sider@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add logging for DP link traning Test Pattern Seqeunces

Add some more logging for DP link traning test pattern seqeunces
for better debugging.

Cc: Fangzhi Zuo <Jerry.Zuo@amd.com>
Cc: Harry Wentland <harry.wentland@amd.com>
Cc: Leo Li <sunpeng.li@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Fangzhi Zuo <Jerry.Zuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: correct ras enabled flag

XGMI RAS should be according to the gmc xgmi physical nodes number,
XGMI RAS should not be enabled if xgmi num_physical_nodes is zero.

Signed-off-by: Stanley.Yang <Stanley.Yang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix unexpected block id

Aldebaran supports VCN and JPEG RAS, it reports unexpected
block id message during VCN and JPEG RAS initialization if VCN
and JPEG block id not defined.

Signed-off-by: Stanley.Yang <Stanley.Yang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: use sdma_v6 single packet invalidation

This achieves the same result as the sequence used in emit_flush_gpu_tlb
but the invalidation is now a single packet instead of the 3 packets
required to implement reg_write_reg_wait.

Signed-off-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Monk Liu <monk.liu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display : Log DP link training downspread info

Update the existing log with DP LT downspread info:

[Downstream devices shall support down spreading of the link clock.
The down-spread amplitude shall either be disabled (0.0%) or up to 0.5%,
as written by the upstream device to the DOWNSPREAD_CTRL register
(DPCD 00107h). The modulation frequency range shall be 30 to 33 kHz]

Besides, fix checkpatch warning:

CHECK: Alignment should match open parenthesis

Cc: Mario Limonciello <mario.limonciello@amd.com>
Cc: Harry Wentland <harry.wentland@amd.com>
Cc: Leo Li <sunpeng.li@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove unused matching_stream_ptrs variable

clang with W=1 reports
drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_link_enc_cfg.c:625:6: error:
  variable 'matching_stream_ptrs' set but not used [-Werror,-Wunused-but-set-variable]
        int matching_stream_ptrs = 0;
            ^
This variable is not used so remove it.

Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: set variables dml*_funcs storage-class-specifier to static

smatch reports
drivers/gpu/drm/amd/amdgpu/../display/dc/dml/display_mode_lib.c:44:24: warning: symbol
  'dml20_funcs' was not declared. Should it be static?
drivers/gpu/drm/amd/amdgpu/../display/dc/dml/display_mode_lib.c:51:24: warning: symbol
  'dml20v2_funcs' was not declared. Should it be static?
drivers/gpu/drm/amd/amdgpu/../display/dc/dml/display_mode_lib.c:58:24: warning: symbol
  'dml21_funcs' was not declared. Should it be static?
drivers/gpu/drm/amd/amdgpu/../display/dc/dml/display_mode_lib.c:65:24: warning: symbol
  'dml30_funcs' was not declared. Should it be static?
drivers/gpu/drm/amd/amdgpu/../display/dc/dml/display_mode_lib.c:72:24: warning: symbol
  'dml31_funcs' was not declared. Should it be static?
drivers/gpu/drm/amd/amdgpu/../display/dc/dml/display_mode_lib.c:79:24: warning: symbol
  'dml314_funcs' was not declared. Should it be static?
drivers/gpu/drm/amd/amdgpu/../display/dc/dml/display_mode_lib.c:86:24: warning: symbol
  'dml32_funcs' was not declared. Should it be static?

These variables are only used in one file so should be static.
Cleanup whitespace, use tabs consistently for indents.

Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: set variables aperture_default_system and context0_default_system storage-class-specifier to static

smatch reports
drivers/gpu/drm/amd/amdgpu/../display/dc/dcn10/dcn10_hubp.c:758:10: warning: symbol
  'aperture_default_system' was not declared. Should it be static?
drivers/gpu/drm/amd/amdgpu/../display/dc/dcn10/dcn10_hubp.c:759:10: warning: symbol
  'context0_default_system' was not declared. Should it be static?

These variables are only used in one file so should be static.

Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: set variable dcn3_14_soc storage-class-specifier to static

smatch reports
drivers/gpu/drm/amd/amdgpu/../display/dc/dml/dcn314/dcn314_fpu.c:100:37: warning: symbol
  'dcn3_14_soc' was not declared. Should it be static?

This variable is only used in one file so should be static.

Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix warnings

Fix below warning due to incompatible types in conditional operator

../pm/swsmu/smu13/smu_v13_0_6_ppt.c:315:17: sparse: sparse: incompatible
types in conditional expression (different base types):

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reported-by: kernel test robot <lkp@intel.com>
Reviewed-by: Luben Tuikov <luben.tuikov@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Link: https://lore.kernel.org/oe-kbuild-all/202303082135.NjdX1Bij-lkp@intel.com/
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: correct SMU13.0.7 max shader clock reporting

Correct the max shader clock reporting on SMU
13.0.7.

Signed-off-by: Horatio Zhang <Hongkun.Zhang@amd.com>
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: correct SMU13.0.7 pstate profiling clock settings

Correct the pstate standard/peak profiling mode clock
settings for SMU13.0.7.

Signed-off-by: Horatio Zhang <Hongkun.Zhang@amd.com>
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: refine get gpu clock counter method

[why]
regGOLDEN_TSC_COUNT_LOWER/regGOLDEN_TSC_COUNT_UPPER are protected and
unaccessible under sriov.
The clock counter high bit may update during reading process.

[How]
Replace regGOLDEN_TSC_COUNT_LOWER/regGOLDEN_TSC_COUNT_UPPER with
regCP_MES_MTIME_LO/regCP_MES_MTIME_HI to get gpu clock under sriov.
Refine get gpu clock counter method to make the result more precise.

Signed-off-by: Tong Liu01 <Tong.Liu01@amd.com>
Acked-by: Luben Tuikov <luben.tuikov@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: optimize redundant code in umc_v6_7

Optimize redundant code in umc_v6_7.

Signed-off-by: YiPeng Chai <YiPeng.Chai@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix sdma v4 sw fini error

Fix sdma v4 sw fini error for sdma 4.2.2 to
solve the following general protection fault

[  +0.108196] general protection fault, probably for non-canonical
address 0xd5e5a4ae79d24a32: 0000 [#1] PREEMPT SMP PTI
[  +0.000018] RIP: 0010:free_fw_priv+0xd/0x70
[  +0.000022] Call Trace:
[  +0.000012]  <TASK>
[  +0.000011]  release_firmware+0x55/0x80
[  +0.000021]  amdgpu_ucode_release+0x11/0x20 [amdgpu]
[  +0.000415]  amdgpu_sdma_destroy_inst_ctx+0x4f/0x90 [amdgpu]
[  +0.000360]  sdma_v4_0_sw_fini+0xce/0x110 [amdgpu]

Signed-off-by: lyndonli <Lyndon.Li@amd.com>
Reviewed-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: DROP redundant drm_prime_sg_to_dma_addr_array

For DMA-MAP userptr on other GPUs, the dma address array
will be populated in amdgpu_ttm_backend_bind.

Remove the redundant call from the driver.

v2:
  update the comment

Signed-off-by: Shane Xiao <shane.xiao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: optimize redundant code in umc_v8_10

Optimize redundant code in umc_v8_10

Signed-off-by: YiPeng Chai <YiPeng.Chai@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

amd/amdgpu: Inherit coherence flags base on original BO flags

For SG BO to DMA-map userptrs on other GPUs, the SG BO
need inherit MTYPEs in PTEs from original BO.

If we set the flags, the device can be coherent with
the CPUs and other GPUs.

v2:
  1. Drop unnecessary flags check
  2. Remove local variable align

Signed-off-by: Shane Xiao <shane.xiao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Fix incorrect comment about Vangogh power cap support

The comment mentions that power1 cap attributes are not supported on
Vangogh, but the opposite is indeed valid: for APUs, only Vangogh is
supported. While at it, also fixed the Renoir comment below (thanks
Melissa for noticing that!).

Cc: Lijo Lazar <lijo.lazar@amd.com>
Cc: Melissa Wen <mwen@igalia.com>
Signed-off-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add userptr bo support for mGPUs when iommu is on

For userptr bo with iommu on, multiple GPUs use same system
memory dma mapping address when both adev and bo_adev are in
identity mode or in the same iommu group.

If RAM direct map to one GPU, other GPUs can share the original
BO in order to reduce dma address array usage when RAM can
direct map to these GPUs. However, we should explicit check
whether RAM can direct map to all these GPUs.

This patch fixes a potential issue that where RAM is
direct mapped on some but not all GPUs.

v2:
  1. Update comment
  2. Add helper function reuse_dmamap

Signed-off-by: Shane Xiao <shane.xiao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu: Drop the hang limit parameter

The driver doesn't resubmit jobs on hangs any more, hence drop
the hang limit parameter - amdgpu_job_hang_limit, wherever it is used.

Suggested-by: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Mario Limonciello <mario.limonciello@amd.com>
Cc: Kent Russell <kent.russell@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix potential null dereference

The adev->dm.dc pointer can be NULL and dereferenced in amdgpu_dm_fini()
without checking.

Add a NULL pointer check before calling dc_dmub_srv_destroy().

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes: 9a71c7d31734 ("drm/amd/display: Register DMUB service with DC")
Signed-off-by: Igor Artemiev <Igor.A.Artemiev@mcst.ru>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.230

This DC version brings along:
- FW Release 0.0.161.0
- Improvements on FPO/FAMS
- Correction to DML calculation
- Fix to multiple clock related issues

Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Fix an out of bounds error in BIOS parser

The array is hardcoded to 8 in atomfirmware.h, but firmware provides
a bigger one sometimes. Deferencing the larger array causes an out
of bounds error.

commit 4fc1ba4aa589 ("drm/amd/display: fix array index out of bound error
in bios parser") fixed some of this, but there are two other cases
not covered by it.  Fix those as well.

Reported-by: erhard_f@mailbox.org
Link: https://bugzilla.kernel.org/show_bug.cgi?id=214853
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2473
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FW Promotion] Release 0.0.161.0

 - Add command to idle opt.
 - Rename d3 entry event and add idle trigger param on
   notify event.
 - Add bit to fw boot status to notify status when hardware
   is powered up.

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Improve robustness of FIXED_VS link training at DP1 rates

[WHY]
New sequence for transparent mode DP1.x link training was provided by LTTPR
vendor

[HOW]
Implement new FIXED_VS sequence, increase LT retry count to minimize
any potential intermittent lightup failures

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Michael Strauss <michael.strauss@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add MES KIQ clear to tell RLC that KIQ is dequeued

[Why]
As MES KIQ is dequeued, tell RLC that KIQ is inactive

[How]
Clear the RLC_CP_SCHEDULERS Active bit which RLC checks KIQ status
In addition, driver can halt MES under SRIOV when unloading driver

v2:
Use scheduler0 mask to clear KIQ portion of RLC_CP_SCHEDULERS

Signed-off-by: Yifan Zha <Yifan.Zha@amd.com>
Reviewed-by: Horace Chen <horace.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add dscclk instance offset check

[why]
based on dscclk instance offset check conditiona program dscclk

Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Charlene Liu <Charlene.Liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: On clock init, maintain DISPCLK freq

[Description]
- On init if a display is connected, we need to maintain the DISPCLK
  frequency
- Even though DPG_EN=1, the display still requires the correct
  timing or it could cause audio corruption (if DISPCLK freq
  is reduced)
- Read the current DISPCLK freq and request the same value to ensure
  the timing is valid and unchanged
- However, add option to do a full pipe power down (including link)
  which will also avoid audio related issues
- Disabled for the time being on dcn32

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add FPO + VActive support

[Description]
- When determining FPO support, include FPO + VActive support
- Support FPO + VActive if one display meets regular requirements
  for FPO and the second display is able to switch in VACTIVE with
  a given amount of margin

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Correct DML calculation to follow HW SPEC

[Why]
In 2560x1600@240p eDP panel, driver use lowest voltage level
to play 1080p video cause underflow. According to HW SPEC,
the senario should use high voltage level.

[How]
ChromaPre value is zero when bandwidth validation.
Correct ChromaPre calculation.

Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Paul Hsieh <Paul.Hsieh@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: prep work for root clock optimization enablement for DCN314

To enable root clock optimizations, we need a number of
register writes and need to account for the difference
in DPSTREAMCLK between DCN31 and DCN314. To prevent
issues, add a number of register writes to
DCCG_MASK_SH_LIST_DCN314_COMMON(), and define dccg314_init()
which is mostly in alignment with dccg31_init() but
accounts for the new DPSTREAMCLK sequence.

Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add scaler control for dcn32

[WHY]
It will introduce the extra warnning log on some asic
that doesn't register

[HOW]
Add the register on dcn32

Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Zhikai Zhai <zhikai.zhai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Clear FAMS flag if FAMS doesn't reduce vlevel

[Description]
- If we find that applying FAMS doesn't reduce the voltage level,
  we will not use it
- Ensure to clear the stream flags indicating FAMS if we hit this
  case

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add MES KIQ dequeue in MES hw fini

[Why]
Need dequeue MES KIQ under SRIOV when unloading driver

[How]
Modify mes_v11_0_kiq_dequeue_sched which was used to dequeue MES SCHED
to support veriable pipe.
Add MES KIQ dequeue in hw fini

Signed-off-by: Yifan Zha <Yifan.Zha@amd.com>
Reviewed-by: Horace Chen <horace.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove unused average_render_time_in_us and i variables

clang with W=1 reports
drivers/gpu/drm/amd/amdgpu/../display/modules/freesync/freesync.c:1132:15: error: variable
  'average_render_time_in_us' set but not used [-Werror,-Wunused-but-set-variable]
        unsigned int average_render_time_in_us = 0;
                     ^
This variable is not used so remove it, which caused i to be unused so remove that as well.

Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add new parameters in v11_struct

Added some new parameters defined for the gfx usermode queues
use cases in the v11_mqd_struct.

Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Christian Koenig <christian.koenig@amd.com>
Reviewed-by: Shashank Sharma <shashank.sharma@amd.com>
Signed-off-by: Arvind Yadav <Arvind.Yadav@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu: introduce gc_*_mes_2.bin v2

To avoid new mes fw running with old driver, rename
mes schq fw to gc_*_mes_2.bin.

v2: add MODULE_FIRMWARE declaration
v3: squash in fixup patch

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: allow more APUs to do mode2 reset when go to S4

Skip mode2 reset only for IMU enabled APUs when do S4.

This patch is to fix the regression issue
https://gitlab.freedesktop.org/drm/amd/-/issues/2483
It is generated by commit b589626674de ("drm/amdgpu: skip ASIC reset
for APUs when go to S4").

Fixes: b589626674de ("drm/amdgpu: skip ASIC reset for APUs when go to S4")
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2483
Tested-by: Yuan Perry <Perry.Yuan@amd.com>
Signed-off-by: Tim Huang <tim.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/vmwgfx: remove unused vmw_overlay function

clang with W=1 reports
drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c:56:35: error:
  unused function 'vmw_overlay' [-Werror,-Wunused-function]
static inline struct vmw_overlay *vmw_overlay(struct drm_device *dev)
                                  ^
This function is not used, so remove it.

Signed-off-by: Tom Rix <trix@redhat.com>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230321182414.1826372-1-trix@redhat.com

drm/vmwgfx: Fix Legacy Display Unit atomic drm support

Legacy Display Unit (LDU) fb dirty support used a custom fb dirty callback. Latter
handled only the DIRTYFB IOCTL presentation path but not the ADDFB2/PAGE_FLIP/RMFB
IOCTL path, common for Wayland compositors.

Get rid of the custom callback in favor of drm_atomic_helper_dirtyfb and unify the
handling of the presentation paths inside of vmw_ldu_primary_plane_atomic_update.
This also homogenizes the fb dirty callbacks across all DUs: LDU, SOU and STDU.

Signed-off-by: Martin Krastev <krastevm@vmware.com>
Reviewed-by: Maaz Mombasawala <mombasawalam@vmware.com>
Fixes: 2f5544ff0300 ("drm/vmwgfx: Use atomic helper function for dirty fb IOCTL")
Cc: <stable@vger.kernel.org> # v5.0+
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230321020949.335012-3-zack@kde.org

drm/vmwgfx: Print errors when running on broken/unsupported configs

virtualbox implemented an incomplete version of the svga device which
they decided to drop soon after the initial release. The device was
always broken in various ways and never supported by vmwgfx.

vmwgfx should refuse to load on those configurations but currently
drm has no way of reloading fbdev when the specific pci driver refuses
to load, which would leave users without a usable fb. Instead of
refusing to load print an error and disable a bunch of functionality
that virtualbox never implemented to at least get fb to work on their
setup.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Reviewed-by: Maaz Mombasawala <mombasawalam@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230321020949.335012-2-zack@kde.org

drm/vmwgfx: Drop mksstat_init_record fn as currently unused

This internal helper handles a type of mksstat event counter
which is currently unused. Remove the routine to avoid compile
warnings.

Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Martin Krastev <krastevm@vmware.com>
Reviewed-by: Maaz Mombasawala <mombasawalam@vmware.com>
Reviewed-by: Zack Rusin <zackr@vmware.com>
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230321020949.335012-1-zack@kde.org

drm/nouveau/disp: make gv100_disp_core_mthd_base static

The symbol is not used outside of the file, so mark it static.

Fixes the following warning:

./drivers/gpu/drm/nouveau/nvkm/engine/disp/gv100.c:591:1: warning:
symbol 'gv100_disp_core_mthd_base' was not declared. Should it be static?

Signed-off-by: ruanjinjie <ruanjinjie@huawei.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220924073957.4140388-1-ruanjinjie@huawei.com

drm/nouveau/mc/ga100: make ga100_mc_device static

Make ga100_mc_device static as it isn't exported, to
fix the following sparse warning:

drivers/gpu/drm/nouveau/nvkm/subdev/mc/ga100.c:51:1: warning: symbol 'ga100_mc_device' was not declared. Should it be static?

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221229155249.669436-1-ben-linux@fluff.org

drm/vc4: remove unused render_wait variable

smatch reports
drivers/gpu/drm/vc4/vc4_irq.c:60:1: warning: symbol
  'render_wait' was not declared. Should it be static?

This variable is not used so remove it.

Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20230406151203.1953812-1-trix@redhat.com

Merge tag 'mediatek-drm-next-6.4' of https://git./linux/kernel/git/chunkuang.hu/linux into drm-next

Mediatek DRM Next for Linux 6.4

1. Add support for 10-bit overlays
2. Add MediaTek SoC DRM (vdosys1) support for mt8195
3. Change mmsys compatible for mt8195 mediatek-drm
4. Only trigger DRM HPD events if bridge is attached
5. Change the aux retries times when receiving AUX_DEFER

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230410233005.2572-1-chunkuang.hu@kernel.org

Merge tag 'drm-msm-next-2023-04-10' of https://gitlab.freedesktop.org/drm/msm into drm-next

main pull request for v6.4

Core Display:
============
* Bugfixes for error handling during probe
* rework UBWC decoder programming
* prepare_commit cleanup
* bindings for SM8550 (MDSS, DPU), SM8450 (DP)
* timeout calculation fixup
* atomic: use drm_crtc_next_vblank_start() instead of our own
  custom thing to calculate the start of next vblank

DP:
==
* interrupts cleanup

DPU:
===
* DSPP sub-block flush on sc7280
* support AR30 in addition to XR30 format
* Allow using REC_0 and REC_1 to handle wide (4k) RGB planes
* Split the HW catalog into individual per-SoC files

DSI:
===
* rework DSI instance ID detection on obscure platforms

GPU:
===
* uapi C++ compatibility fix
* a6xx: More robust gdsc reset
* a3xx and a4xx devfreq support
* update generated headers
* various cleanups and fixes
* GPU and GEM updates to avoid allocations which could trigger
  reclaim (shrinker) in fence signaling path
* dma-fence deadline hint support and wait-boost
* a640 speedbin support
* a650 speedbin support

Conflicts in drivers/gpu/drm/msm/adreno/adreno_gpu.c:

Conflict between the 7fa5047a436b ("drm: Use of_property_present() for
testing DT property presence") and 9f251f934012 ("drm/msm/adreno: Use
OPP for every GPU generation"). The latter removed the of_ function
call outright, so I went with what's in the PR unchanged.

From: Rob Clark <robdclark@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CAF6AEGvwuj5tabyW910+N-B=5kFNAC7QNYoQ=0xi3roBjQvFFQ@mail.gmail.com
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Merge tag 'drm-habanalabs-next-2023-04-10' of https://git./linux/kernel/git/ogabbay/linux into drm-next

This tag contains additional habanalabs driver changes for v6.4:

- uAPI changes:
  - Add a definition of a new Gaudi2 server type. This is used by userspace
    to know what is the connectivity between the accelerators inside the
    server

- New features and improvements:
  - speedup h/w queues test in Gaudi2 to reduce device initialization times.

- Firmware related fixes:
  - Fixes to the handshake protocol during f/w initialization.
  - Sync f/w events interrupt in hard reset to avoid warning message.
  - Improvements to extraction of the firmware version.

- Misc bug fixes and code cleanups. Notable fixes are:
  - Multiple fixes for interrupt handling in Gaudi2.
  - Unmap mapped memory in case TLB invalidation fails.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Oded Gabbay <ogabbay@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230410124637.GA2441888@ogabbay-vm-u20.habana-labs.com

accel/habanalabs: add missing error flow in hl_sysfs_init()

hl_sysfs_fini() is called only if hl_sysfs_init() completes
successfully. Therefore if hl_sysfs_init() fails, need to remove any
sysfs group that was added until that point.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: speedup h/w queues test in Gaudi2

HW queues testing at driver load and after reset takes a substantial
amount of time.
This commit reduces the queues test time in Gaudi2 devices by running
all the tests in parallel instead of one after the other.
Time measurements on tests duration shows that the new method is almost
x100 faster than the serial approach.

Signed-off-by: Moti Haimovski <mhaimovski@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: fix handling of arc farm sei event

There is only single eq entry for arc farm sei event which aggregates
events from the four arc farms.
Fix the code to handle this event according to this behavior.

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: remove Gaudi1 multi MSI code

Multi MSI interrupts aren't working in Gaudi1 and because of that,
we are only using a single MSI interrupt. Therefore, let's remove this
dead code in order to avoid confusion.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs/uapi: new Gaudi2 server type

Add definition of a new Gaudi2 server type. This represents
the connectivity between the cards in that server type.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>

accel/habanalabs: fixes for unexpected error interrupt

Removing redundant asic prop variable as we don't need to expose this
to common code. In addition, fix some typos.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: don't wait for STS_OK after sending COMMS WFE

Sending COMMS_GOTO_WFE instructs the FW's CPU to halt (WFE state).
Once sent, FW's CPU isn't expected to continue communicating with LKD.
Therefore, the stage of waiting for COMMS_STS_OK should be skipped or
else waiting for COMMS_STS_OK will simply timeout, which will trigger
unexpected behavior.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: sync f/w events interrupt in hard reset

Receiving events from FW, while the device is in hard reset, causes
a warning message in Driver log. The message may point to a
problem in the Driver or FW. But It also can appear as a result
of events that have been sent from FW just before the hard reset.
In order to avoid receiving events from FW while the device is in reset
and is already in 'disabled' mode, sync the f/w events interrupt right
before setting the device to 'disabled'.

Signed-off-by: Tal Cohen <talcohen@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: fix wrong reset and event flags

During event handling, driver sets relevant reset and user event
notifier flags. Fix few wrong flags settings.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: fix events mask of decoder abnormal interrupts

The decoder IRQ status register may have several set bits upon an
abnormal interrupt. Therefore, when setting the events mask, need to
check all bits and not using if-else.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: remove completion from abnormal interrupt work name

Decoder abnormal interrupts are for errors and not for completion, so
rename the relevant work and work function to not include 'completion'.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: print raw binning masks in debug level

There are rare cases of failures when cards are initialized due to
wrong values in efuse mappings that are parsed by firmware.

To help debug those cases, print (in debug level) the raw binning masks
as fetched from the firmware during device initialization.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: fix HBM MMU interrupt handling

Current mapping between HMMU event and HMMU block is wrong.
In addition the captured address in case of a page fault or
an access error is scrambled, Hence we must call the descramble
function.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: improvements to FW ver extraction

1. Rename the func to hl_get_preboot_major_minor because we also set
   the extracted values in hdev fields.

2. Free the allocated string in the calling function which makes more
   sense

Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: fix access error clear event

The register which needs to be cleared is the valid register instead
of the address.

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: send disable pci when compute ctx is active

Fix an issue in hard reset flow in which the driver didn't send a
disable pci message if there was an active compute context.
In hard reset, disable pci message should be sent no matter if
a compute context exists or not.

Signed-off-by: Tal Cohen <talcohen@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>

accel/habanalabs: remove duplicated disable pci msg

The disable pci message is sent in reset device. It informs the FW not
to raise more EQs. The Driver may ignore received EQs, when the device
is in disabled mode.
The duplication happens when hard reset is scheduled during compute
reset and also performs 'escalate_reset_flow'.

Signed-off-by: Tal Cohen <talcohen@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>

accel/habanalabs: change COMMS warning messages to error level

COMMS protocol is used for LKD <--> FW communication, and any
communication failure between the two might turn out to be
destructive, hence, it should be well emphasized.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>

accel/habanalabs: check return value of add_va_block_locked

since the function might fail and we should propagate the failure.

Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>

accel/habanalabs: print event type when device is disabled

When the device is in disabled state, the driver isn't suppose to
receive any events from FW. Printing the event type, as part of the
message that was already printed, shall help to get more info if this
unexpected message is received.

Signed-off-by: Tal Cohen <talcohen@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>

accel/habanalabs: unmap mapped memory when TLB inv fails

Once a memory mapping is added to the page tables, it's followed by
a TLB invalidation request which could potentially fail (HW failure).
Removing the mapping is simply a part of this failure handling routine.
TLB invalidation failure prints were updated to be more accurate.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>