drm/amdgpu: init umc functions for arcturus umc ras

reuse vg20 umc functions for arcturus umc ras

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

drm/amdgpu: enable ras capablity check on arcturus

check hw ras capablity via atomfirmware

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

drm/amdkfd: Rename kfd_kernel_queue_*.c to kfd_packet_manager_*.c

After the recent cleanup, the functionalities provided by the previous
kfd_kernel_queue_*.c are actually all packet manager related. So rename
them to reflect that.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Eliminate ops_asic_specific in kernel queue

The ops_asic_specific function pointers are actually quite generic after
using a simple if condition. Eliminate it by code refactoring.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Merge CIK kernel queue functions into VI

The only difference that CIK kernel queue functions are different from
VI is avoid allocating eop_mem. We can achieve that by using a if
condition.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: remove set but not used variable 'us_mvdd'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c: In
function ‘vegam_populate_smc_acpi_level’:
drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c:1117:11:
warning: variable 'us_mvdd' set but not used [-Wunused-but-set-variable]

It is never used, so can be removed.

Fixes: ac7822b0026f ("drm/amd/powerplay: add smumgr support for VEGAM (v2)")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove set but not used variable 'invalid'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c: In function
‘amdgpu_amdkfd_evict_userptr’:
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c:1665:6: warning:
variable ‘invalid’ set but not used [-Wunused-but-set-variable]

'invalid' is never used, so can be removed. Thus 'atomic_inc_return'
can be replaced as 'atomic_inc'

Fixes: 5ae0283e831a ("drm/amdgpu: Add userptr support for KFD")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove set but not used variable 'count'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/amdkfd/kfd_device.c: In function
‘kgd2kfd_post_reset’:
drivers/gpu/drm/amd/amdkfd/kfd_device.c:745:11: warning:
variable ‘count’ set but not used [-Wunused-but-set-variable]

'count' is never used, so can be removed. Thus 'atomic_dec_return'
can be replaced as 'atomic_dec'

Fixes: e42051d2133b ("drm/amdkfd: Implement GPU reset handlers in KFD")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove set but not used variable 'amdgpu_connector'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/amdgpu/amdgpu_display.c: In function
‘amdgpu_display_crtc_scaling_mode_fixup’:
drivers/gpu/drm/amd/amdgpu/amdgpu_display.c:693:27: warning: variable
‘amdgpu_connector’ set but not used [-Wunused-but-set-variable]

Fixes: d38ceaf99ed0 ("drm/amdgpu: add core driver (v4)")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove set but not used variable 'mc_shared_chmap' from 'gfx_v6_0.c' and 'gfx_v7_0.c'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c: In function
‘gfx_v6_0_constants_init’:
drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c:1579:6: warning: variable
‘mc_shared_chmap’ set but not used [-Wunused-but-set-variable]

drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c: In function
‘gfx_v7_0_gpu_early_init’:
drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c:4262:6: warning: variable
‘mc_shared_chmap’ set but not used [-Wunused-but-set-variable]

Fixes: 2cd46ad22383 ("drm/amdgpu: add graphic pipeline implementation for si v8")
Fixes: d93f3ca706b8 ("drm/amdgpu/gfx7: rework gpu_init()")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove set but not used variable 'bpc'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_link.c: In function get_pbn_from_timing:
drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_link.c:2364:11: warning:
 variable bpc set but not used [-Wunused-but-set-variable]

It is not used since commit e49f69363adf ("drm/amd/display: use
proper formula to calculate bandwidth from timing"), this also
remove get_color_depth(), which is only used here.

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Stop using GFP_NOIO explicitly for two places

Adapt the change from: 1cd106ecfc1f04 ("drm/amdkfd: Stop using GFP_NOIO explicitly")

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Use QUEUE_IS_ACTIVE macro in mqd v10

This is done for other GFX in commit bb2d2128a54c4. Port it to GFX10.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Fix a bug when calculating save_area_used_size

workgroup context data writes from m->cp_hqd_cntl_stack_size, so we
should deduct it when calculating the used size.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Update get_wave_state() for GFX10

Given control stack is now in the userspace context save restore area
on GFX10, the same as GFX8, it is not needed to copy it back to userspace.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Implement queue priority controls for gfx10

Ported from gfx9.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Rename create_cp_queue() to init_user_queue()

create_cp_queue() could also work with SDMA queues, so we should rename
it. It only initialize the data values rather than creating queues.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Avoid using doorbell_off as offset in process doorbell pages

dorbell_off in the queue properties is mainly used for the doorbell dw
offset in pci bar. We should not set it to the doorbell byte offset in
process doorbell pages. This makes the code much easier to read.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Use better name to indicate the offset is in dwords

The doorbell offset could mean the byte offset or the dword offset,
and the 0 offset place is also different, sometimes the start of PCI
doorbell bar or the start of process doorbell pages. Use better name
to avoid confusion.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove set but not used variable 'ds_port'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_link_dp.c: In function dp_wa_power_up_0010FA:
drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_link_dp.c:2320:35: warning:
 variable ds_port set but not used [-Wunused-but-set-variable]

It is never used, so can be removed.

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: remove set but not used variable 'threshold', 'state'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c: In function fiji_populate_single_graphic_level:
drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c:943:11: warning: variable threshold set but not used [-Wunused-but-set-variable]
drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c: In function fiji_populate_memory_timing_parameters:
drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c:1504:8: warning: variable state set but not used [-Wunused-but-set-variable]

They are introduced by commit 2e112b4ae3ba ("drm/amd/pp:
remove fiji_smc/smumgr split."), but never used,
so remove it.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use static const, not const static

Move the static keyword to the front of declarations.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: remove set but not used variable 'data'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c: In function vega10_get_performance_level:
drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c:5217:23: warning: variable data set but not used [-Wunused-but-set-variable]

'data' is introduced by commit f688b614b643 ("drm/amd/pp:
Implement get_performance_level for legacy dgpu"), but never used,
so remove it.

Reviewed-by: Evan Quan <evan.quan@amd.com>
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: remove set but not used variable 'vbios_version', 'data'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c: In function smu7_check_mc_firmware:
drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c:4215:11: warning: variable vbios_version set but not used [-Wunused-but-set-variable]
drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c: In function smu7_get_performance_level:
drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c:5054:21: warning: variable data set but not used [-Wunused-but-set-variable]

'vbios_version' is introduced by commit 599a7e9fe1b6 ("drm/amd/powerplay:
implement smu7 hwmgr to manager asics with smu ip version 7."),
but never used, so remove it.

'data' is introduced by commit f688b614b643 ("drm/amd/pp:
Implement get_performance_level for legacy dgpu"), but never used,
so remove it.

Reviewed-by: Evan Quan <evan.quan@amd.com>
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove duplicated comparison expression

There is comparison expression that is duplicated and hence one
of the expressions can be removed.  Remove it.

Addresses-Coverity: ("Same on both sides")
Fixes: 12e2b2d4c65f ("drm/amd/display: add dcc programming for dual plane")
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix spelling mistake "exeuction" -> "execution"

There are spelling mistakes in a DC_ERROR message and a comment.
Fix these.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Simplify the mmap offset related bit operations

The new code uses straightforward bit shifts and thus has better readability.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Use kernel queue v9 functions for v10

The kernel queue functions for v9 and v10 are the same except
pm_map_process_v* which have small difference, so they should be reused.
This eliminates the need of reapplying several patches which were
applied on v9 but not on v10, such as bigger GWS and more than 2
SDMA engine support which were introduced on Arcturus.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Only keep release_mem function for Hawaii

release_mem is only used for Hawaii, but because GFX7 and GFX8 share the
same function pointer structure, so we only delete release_mem for GFX9
and GFX10.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Adjust function sequences to avoid unnecessary declarations

This is cleaner.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove redundant variable status

Variable status is redundant, it is being initialized with a value
that is over-written later and this is being returned immediately
after the assignment.  Clean up the code by removing status and
just returning the value returned from the call to function
dc->hwss.dmdata_status_done.

Addresses-Coverity: ("Unused value")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove duplicated assignment to grph_obj_type

Variable grph_obj_type is being assigned twice, one of these is
redundant so remove it.

Addresses-Coverity: ("Evaluation order violation")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove set but not used variable 'mc_shared_chmap'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c: In function
‘gfx_v8_0_gpu_early_init’:
drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c:1713:6: warning: variable
‘mc_shared_chmap’ set but not used [-Wunused-but-set-variable]

Fixes: 0bde3a95eaa9 ("drm/amdgpu: split gfx8 gpu init into sw and hw parts")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove always false comparison in 'amdgpu_atombios_i2c_process_i2c_ch'

Fixes gcc '-Wtype-limits' warning:

drivers/gpu/drm/amd/amdgpu/atombios_i2c.c: In function
‘amdgpu_atombios_i2c_process_i2c_ch’:
drivers/gpu/drm/amd/amdgpu/atombios_i2c.c:79:11: warning: comparison is
always false due to limited range of data type [-Wtype-limits]

'num' is 'u8', so it will never be greater than 'TOM_MAX_HW_I2C_READ',
which is defined as 255. Therefore, the comparison can be removed.

Fixes: d38ceaf99ed0 ("drm/amdgpu: add core driver (v4)")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove set but not used variable 'dig'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/amdgpu/atombios_dp.c: In function
‘amdgpu_atombios_dp_link_train’:
drivers/gpu/drm/amd/amdgpu/atombios_dp.c:716:34: warning: variable ‘dig’
set but not used [-Wunused-but-set-variable]

Fixes: d38ceaf99ed0 ("drm/amdgpu: add core driver (v4)")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove set but not used variable 'dig_connector'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/amdgpu/atombios_dp.c: In function
‘amdgpu_atombios_dp_get_panel_mode’:
drivers/gpu/drm/amd/amdgpu/atombios_dp.c:364:36: warning: variable
‘dig_connector’ set but not used [-Wunused-but-set-variable]

It is never used, so can be removed.

Fixes: d38ceaf99ed0 ("drm/amdgpu: add core driver (v4)")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add function parameter description in 'amdgpu_gart_bind'

Fixes gcc warning:

drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c:313: warning: Function
parameter or member 'flags' not described in 'amdgpu_gart_bind'

Fixes: d38ceaf99ed0 ("drm/amdgpu: add core driver (v4)")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add function parameter description in 'amdgpu_device_set_cg_state'

Fixes gcc warning:

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c:1954: warning: Function
parameter or member 'state' not described in 'amdgpu_device_set_cg_state'

Fixes: e3ecdffac9cc ("drm/amdgpu: add documentation for amdgpu_device.c")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove 4 set but not used variable in amdgpu_atombios_get_connector_info_from_object_table

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c: In function
'amdgpu_atombios_get_connector_info_from_object_table':
drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c:376:26: warning: variable
'grph_obj_num' set but not used [-Wunused-but-set-variable]
drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c:376:13: warning: variable
'grph_obj_id' set but not used [-Wunused-but-set-variable]
drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c:341:37: warning: variable
'con_obj_type' set but not used [-Wunused-but-set-variable]
drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c:341:24: warning: variable
'con_obj_num' set but not used [-Wunused-but-set-variable]

They are never used, so can be removed.

Fixes: d38ceaf99ed0 ("drm/amdgpu: add core driver (v4)")
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/display: fix warning when CONFIG_DRM_AMD_DC_DCN is not set

dm_dcn_crtc_high_irq() is only used when CONFIG_DRM_AMD_DC_DCN is set.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/display: fix the build when CONFIG_DRM_AMD_DC_DCN is not set

Need to protect some DSC functions.

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: rename DCN1_0 kconfig to DCN

Since dcn20 and dcn21 are under dcn1 it doesnt make sense to
have it named dcn1.

Change it to "dcn" to make it generic

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

drm/amd/display: Drop CONFIG_DRM_AMD_DC_DCN2_1 flag

[Why]

DCN21 is stable enough to be build by default. So drop the flags.

[How]

Remove them using the unifdef tool. The following commands were executed
in sequence:

$ find -name '*.c' -exec unifdef -m -DCONFIG_DRM_AMD_DC_DCN2_1 -UCONFIG_TRIM_DRM_AMD_DC_DCN2_1 '{}' ';'
$ find -name '*.h' -exec unifdef -m -DCONFIG_DRM_AMD_DC_DCN2_1 -UCONFIG_TRIM_DRM_AMD_DC_DCN2_1 '{}' ';'

In addition:

* Remove from kconfig, and replace any dependencies with DCN1_0.
* Remove from any makefiles.
* Fix and cleanup Renoir definitions in dal_asic_id.h
* Expand DCN1 ifdef to include DCN21 code in the following files:
    * clk_mgr/clk_mgr.c: dc_clk_mgr_create()
    * core/dc_resources.c: dc_create_resource_pool()
    * gpio/hw_factory.c: dal_hw_factory_init()
    * gpio/hw_translate.c: dal_hw_translate_init()

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

drm/amd/display: Drop CONFIG_DRM_AMD_DC_DCN2_0 and DSC_SUPPORTED

[Why]

DCN2 and DSC are stable enough to be build by default. So drop the flags.

[How]

Remove them using the unifdef tool. The following commands were executed
in sequence:

$ find -name '*.c' -exec unifdef -m -DCONFIG_DRM_AMD_DC_DSC_SUPPORT -DCONFIG_DRM_AMD_DC_DCN2_0 -UCONFIG_TRIM_DRM_AMD_DC_DCN2_0 '{}' ';'
$ find -name '*.h' -exec unifdef -m -DCONFIG_DRM_AMD_DC_DSC_SUPPORT -DCONFIG_DRM_AMD_DC_DCN2_0 -UCONFIG_TRIM_DRM_AMD_DC_DCN2_0 '{}' ';'

In addition:

* Remove from kconfig, and replace any dependencies with DCN1_0.
* Remove from any makefiles.
* Fix and cleanup NV defninitions in dal_asic_id.h
* Expand DCN1 ifdef to include DCN2 code in the following files:
    * clk_mgr/clk_mgr.c: dc_clk_mgr_create()
    * core/dc_resources.c: dc_create_resource_pool()
    * dce/dce_dmcu.c: dcn20_*lock_phy()
    * dce/dce_dmcu.c: dcn20_funcs
    * dce/dce_dmcu.c: dcn20_dmcu_create()
    * gpio/hw_factory.c: dal_hw_factory_init()
    * gpio/hw_translate.c: dal_hw_translate_init()

Signed-off-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Disable VUpdate interrupt for DCN hardware

[Why]

On DCN hardware, the crtc_high_irq handler makes vupdate_high_irq
handler redundant.

All the vupdate handler does is handle vblank events, and update vrr
for DCE hw (excluding VEGA, more on that later). As far as usermode is
concerned. vstartup happens close enough to vupdate on DCN that it can
be considered the "same". Handling vblank and updating vrr at vstartup
effectively replaces vupdate on DCN.

Vega is a bit special. Like DCN, the VRR registers on Vega are
double-buffered, and swapped at vupdate. But Unlike DCN, it lacks a
vstartup interrupt. This means we can't quite remove the vupdate handler
for it, since delayed user events due to vrr are sent off there.

[How]

Remove registration of vupdate interrupt handler for DCN. Disable
vupdate interrupt if asic family DCN, enable otherwise.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Send vblank and user events at vsartup for DCN

[Why]

For DCN hardware, the crtc_high_irq handler is assigned to the vstartup
interrupt. This is different from DCE, which has it assigned to vblank
start.

We'd like to send vblank and user events at vstartup because:

* It happens close enough to vupdate - the point of no return for HW.

* It is programmed as lines relative to vblank end - i.e. it is not in
  the variable portion when VRR is enabled. We should signal user
  events here.

* The pflip interrupt responsible for sending user events today only
  fires if the DCH HUBP component is not clock gated. In situations
  where planes are disabled - but the CRTC is enabled - user events won't
  be sent out, leading to flip done timeouts.

Consequently, this makes vupdate on DCN hardware redundant. It will be
removed in the next change.

[How]

Add a DCN-specific crtc_high_irq handler, and hook it to the VStartup
signal. Inside the DCN handler, we send off user events if the pflip
handler hasn't already done so.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: disable lttpr for invalid lttpr caps.

1-Read lttpr caps in 5-bytes
2-Parse caps
3-Validate caps and set lttpr_mode
4-Use hw default timeout when lttpr is disabled.

Signed-off-by: abdoulaye berthe <abdoulaye.berthe@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: use previous aux timeout val if no repeater.

[Why]
The aux timeout value is not default before reading link cap.
Setting it to default when lttpr is not enabled causes some monitor
not to light up.

[How]
Read the aux engine timeout value before setting it to extended.
Set the aux engine timeout to its previous value if no lttpr.

Signed-off-by: abdoulaye berthe <abdoulaye.berthe@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: implement lttpr logic

1-If at least one repeater is present in the link and we are in non
transparent mode, perform clock recovery then channel equalization
with all repeaters one by one before training DPRX.

2-Mark the end of LT with a repeater by setting training pattern 0
at the end of channel equalization with each repeater.

Signed-off-by: abdoulaye berthe <abdoulaye.berthe@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: configure lttpr mode

[Description]
1-Grant extended timeout request. Done once after detection
2-Configure lttpr mode based on lttpr support before LT
3-Account for lttpr cap when determining max link settings

Signed-off-by: abdoulaye berthe <abdoulaye.berthe@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: check for dp rev before reading lttpr regs

[Why]
LTTPR was introduced after DP1.2. Reading LTTPR registers 0xFXXXX
on some DP 1.2 display is causing an unexpected behavior.

[How]
Make sure that we don't read any lttpr registers on 1.2 displays.

Signed-off-by: abdoulaye berthe <abdoulaye.berthe@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: initialize lttpr

[Description]
When reading link, update the procedure as follows:
1-Set aux timeout to extended: 3.2ms
2-Start with reading lttpr caps
3-Determine if lttpr support should be enabled. Reset aux timeout to
400us if no repeater is found.

Signed-off-by: abdoulaye berthe <abdoulaye.berthe@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Unify all scaling when Integer Scaling enabled

[why]
We want to guarantee integer ratio scaling for all scaling modes.

[how]
Treat centered, fullscreen, preserve aspect ratio the same: scale
the view as many times as possible, and fill in the rest with a black
border.

Signed-off-by: Reza Amini <Reza.Amini@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add oem i2c implemenation in dc

[why]
Need it for some OEM I2C devices in Nv10

[how]
Link up code to parse OEM table and expose DC interface
to access the pins

Signed-off-by: Jun Lei <Jun.Lei@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add some hardware status in DTN log debugfs

[Why]
For debug purpose, we need to check the following hardware status
in DTN log debugfs:
1.dpp & hubp clock enable;
2.crtc blank enable;
3.link phy status;

[How]
Add the upper information in the amdgpu_dm_dtn_log debugfs.

For CRTC blanked status, since DCN2 and greater reports it on the OPP
instead of OTG, we patch it in after calling optc1_read_otg_states.
Ideally, this should be done in the DCN version specific function hooks.
It has been left as a TODO item.

Signed-off-by: Leo (Hanghong) Ma <hanghong.ma@amd.com>
Reviewed-by: Mikita Lipski <Mikita.Lipski@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.58

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: optimize bandwidth after commit streams.

[Why]
System is unable to enter S0i3 due to DISPLAY_OFF_MASK not asserted
in SMU.

[How]
Optimized bandwidth should be called paired and to resolve unplug
display underflow issue, optimize bandwidth after commit streams is
moved to next page flip, in case of S0i3, there is a change for no
flip coming causing display count is 1 in SMU side.
Add optimize bandwidth after commit stream.

Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Create debug option to disable v.active clock change policy.

[WHY]
It has been a useful option in debugging GFXOFF and P.State Change issues.
May be required as for platform specific workaround.

[HOW]
Create option in enum dc_debug_options, "disable_vactive_clock_change".
When it is set, dm_dram_clock_change_vactive, will translate into
p_state_change_support: false.

Signed-off-by: David Galiffi <David.Galiffi@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: set MSA MISC1 bit 6 while sending colorimetry in VSC SDP

[Why]
It is confusing to sinks if we send VSC SDP only on some format. Today we
signal colorimetry format using MSA while in formats like sRGB.
But when we switch to BT2020 we set the bit to ignore MSA  colorimetry and
instead use the colorimetry information in the VSC SDP.

But if sink supports signaling of colorimetry via VSC SDP we should always
set the MSA MISC1 bit 6, instead of doing so selectively.

[How]
If sink supports signaling of colorimetry via VSC SDP, and we are sending
the colorimetry info via VSC SDP with packet revision 05h, then always
set MSA MISC1 bit 6.

Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add a sanity check for DSC already enabled/disabled

[why]
If acquire/release DSC resource sequence is affected by a regression,
it can happen that the already-in-use DSC HW block is being wrongly
re-used for a different pipe. The reverse is also possible, i.e.
already-disabled DSC HW block could be disabled from other context.

[how]
Read back the enable state of DSC HW and report an error if duplicate
enable or disable was attempted.

Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use SIGNAL_TYPE_NONE in disable_output unless eDP

[WHY]
Currently made a change where disable_output is called using signal_type.
Using actual signal_type when calilng disable_output in power_down_encoders
would make DP to HDMI dongle not light up on boot. As it would have signal_type
SIGNAL_TYPE_DISPLAY_PORT.

[HOW]
Set signal_type to SIGNAL_TYPE_NONE unless it is eDP.

Signed-off-by: Sung Lee <sung.lee@amd.com>
Reviewed-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Renoir chroma viewport WA

[Why]
For unknown reason, immediate flip with host VM translation on NV12
surface will underflow on last row of PTE.

[How]
Hack chroma viewport height to make fetch one more row of PTE.
Note that this will cause hubp underflow on all video underlay
cases, but the underflow is not user visible since it is in
blank region.

Signed-off-by: Eric Yang <Eric.Yang2@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix assert observed when performing dummy p-state check

[WHY]
V.Active dram clock change workaround need a small modification for DMLv2
to ensure that the dummy p-state check doesn't fail.

Signed-off-by: David Galiffi <David.Galiffi@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.57

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Drop CONFIG_DRM_AMD_DC_DMUB guards

[Why]
Support for DMUB only depends on support for DC. It doesn't use floating
point so we don't need to guard it by any specific DCN revision.

[How]
Drop the guards and cleanup the newlines around each one.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Register DMUB service with DC

[Why]
DC can utilize the DMUB server to send commands to the DMUB but it's
the DM responsibility to pass it the service to use.

[How]
Create the dc_dmub_srv after we finish initializing the dmub_srv.
Cleanup the dc_dmub_srv before destroying the dmub_srv or dc.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add DMUB support to DC

DC will use DMUB for command submission and flow control during
initialization.

Register offloading as well as submitting some BIOS commands are part
of the DC internal interface but are guarded behind debug options.

It won't be functional in amdgpu_dm yet since we don't pass the
DMUB service to DC for use.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add DMCUB to firmware query interface

The DMCUB firmware version can be read using the AMDGPU_INFO ioctl
or the amdgpu_firmware_info debugfs entry.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Hook up the DMUB service in DM

[Why]
We need DMCUB on Renoir to support DMCU and PHY initialization.
The DMUB service provides a mechanism to load the DMCUB.

[How]
Include the DMUB service in amdgpu_dm.

Frontdoor loading of the DMCUB firmware needs to happen via PSP. To
pass the firmware to PSP we need to hand it off to the firmware list
in the base driver during software initialization.

Most of the DMUB service can technically be initialized at this point
in time, but we don't want to be allocating framebuffer memory for
hardware that doesn't support the DMCUB and in order to check that we
need to be able to read registers - something DM helpers aren't setup
to do in software initialization.

So everything but the service creation itself will get deferred to
hardware initialization.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add PSP FW version mask.

[Why]
PSP version format is AB.CD.EF.GH, where CD and GH is the main version.
current psp version check for dmcub loading dmcu check 0x00110029, in
case of some psp version eg: 0x00110227 which main version should be
0x00110027, will result in unexpeceted dmcub loading dmcu FW.

[How]
Add psp version mask 0x00FF00FF for checking version.

Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Change dmcu init sequence for dmcub loading dmcu FW.

[Why]
DMCU isn't intiliazed properly by dmcub loading due to dmcub initialize
sequence.

[How]
Change dmcu init sequece to meet dmcub initilize.

Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add the DMUB service

The DMUB service is the interface to the DMCUB.

It's required to support Renoir features so it will be enabled and
compiled automatically when the Renoir display engine is enabled via
CONFIG_DRM_AMD_DC_DCN2_1.

DMUB code will initially be guarded by CONFIG_DRM_AMD_DC_DMUB and later
switched to CONFIG_DRM_AMD_DC_DCN2_1 with the config option dropped.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Drop DMCUB from DCN21 resources

The interface to the DMCUB won't be through DC itself. DC will instead
call into the DMUB interface introduced with a future change.

The CONFIG_DRM_AMD_DC_DMUB defines will still be used for now but will
be dropped at the end of the series.

Since this define was never configurable in the first place this code
wasn't used.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add PSP loading support for DMCUB ucode

DMCUB ucode requires secure loading through PSP. This is already
supported in PSP as GFX_FW_TYPE_DMUB, it just needs to be mapped from
AMDGPU_UCODE_ID_DMCUB to GFX_FW_TYPE_DMUB.

DMUB is a shorthand name for DMCUB and can be used interchangeably.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add ucode support for DMCUB

The DMCUB is a secondary DMCU (Display MicroController Unit) that has
its own separate firmware. It's required for DMCU support on Renoir.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix dereference of pointer aconnector when it is null

Currently pointer aconnector is being dereferenced by the call to
to_dm_connector_state before it is being null checked, this could
lead to a null pointer dereference.  Fix this by checking that
aconnector is null before dereferencing it.

Addresses-Coverity: ("Dereference before null check")
Reviewed-by: Mikita Lipski <mikita.lipski@amd.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add MST atomic routines

- Adding encoder atomic check to find vcpi slots for a connector
- Using DRM helper functions to calculate PBN
- Adding connector atomic check to release vcpi slots if connector
loses CRTC
- Calculate  PBN and VCPI slots only once during atomic
check and store them on crtc_state to eliminate
redundant calculation
- Call drm_dp_mst_atomic_check to verify validity of MST topology
during state atomic check

v2: squashed previous 3 separate patches, removed DSC PBN calculation,
and added PBN and VCPI slots properties to amdgpu connector

v3:
- moved vcpi_slots and pbn properties to dm_crtc_state and dc_stream_state
- updates stream's vcpi_slots and pbn on commit
- separated patch from the DSC MST series

v4:
- set vcpi_slots and pbn properties to dm_connector_state
- copy porperties from connector state on to crtc state

v5:
- keep the pbn and vcpi values only on connnector state
- added a void pointer to the stream state instead on two ints,
because dc_stream_state is OS agnostic. Pointer points to the
current dm_connector_state.

v6:
- Remove new param from stream

v7:
- Fix error with using max capable bpc

v8:
- squash in fix from Gustavo A. R. Silva

Reviewed-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/vcn: finish delay work before release resources

flush/cancel delayed works before doing finalization
to avoid concurrently requests.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: read pcie speed/width info (v2)

sysfs interface to read pcie speed&width info on navi1x.

v2: fix warning (trivial)

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/smu_v11: Unify and fix power limits

[Why]
On Navi10, and presumably arcterus, updating pp_table via sysfs would
not re-scale the maximum possible power limit one can set. On navi10,
the SMU code ignored the power percentage overdrive setting entirely,
and would not allow you to exceed the default power limit at all.

[How]
Adding a function to the SMU interface to get the pptable version of the
default power limit allows ASIC-specific code to provide the correct
maximum-settable power limit for the current pptable.

v3: fix spelling (Alex)

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Matt Coffin <mcoffin13@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/navi10: Implement od clk printing

[Why]
Before this patch, navi10 overdrive settings could not be printed via
pp_od_clk_voltage

[How]
Implement printing for the overdrive settings for the following clocks
in navi10's ppt print_clk_levels implementation:

* SMU_OD_SCLK
* SMU_OD_MCLK
* SMU_OD_VDDC_CURVE

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Matt Coffin <mcoffin13@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/navi10: implement GFXCLK_CURVE overdrive

[Why]
Before this patch, there was no way to set the gfxclk voltage curve in
the overdrive settings for navi10 through pp_od_clk_voltage

[How]
Add the required implementation to navi10's ppt dpm table editing
implementation, similar to the vega20 implementation and interface.

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Matt Coffin <mcoffin13@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/navi10: implement sclk/mclk OD via pp_od_clk_voltage

[Why]
Before this patch, there was no way to use pp_od_clk_voltage on navi

[How]
Similar to the vega20 implementation, but using the common smc_v11_0
headers, implemented the pp_od_clk_voltage API for navi10's pptable
implementation

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Matt Coffin <mcoffin13@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu: finish delay works before release resources

flush/cancel delayed works before doing finalization
to avoid concurrently requests.

Signed-off-by: Jesse Zhang <zhexi.zhang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/radeon: Clean up code in radeon_pci_shutdown()

This fixes the formatting on one comment and consolidates the
pci_get_drvdata() into the radeon_suspend_kms().

Signed-off-by: Kyle Mahlkuch <kmahlkuc@linux.vnet.ibm.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/powerplay/smu7: fix AVFS handling with custom powerplay table

When a custom powerplay table is provided, we need to update
the OD VDDC flag to avoid AVFS being enabled when it shouldn't be.

Bug: https://bugzilla.kernel.org/show_bug.cgi?id=205393
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: avoid upload corrupted ta ucode to psp

xgmi, ras, hdcp and dtm ta are actually separated ucode and
need to handled case by case to upload to psp.

We support the case that ta binary have one or multiple of
them built-in. As a result, the driver should check each ta
binariy's availablity before decide to upload them to psp.

In the terminate (unload) case, the driver will check the
context readiness before perform unload activity. It's fine
to keep it as is.

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/powerplay: fix AVFS handling with custom powerplay table

When a custom powerplay table is provided, we need to update
the OD VDDC flag to avoid AVFS being enabled when it shouldn't be.

Bug: https://bugzilla.kernel.org/show_bug.cgi?id=205393
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: dynamically disable ds and ulv for compute

This is to improve the performance in the compute mode
for vega10. For example, the original performance for a rocm
bandwidth test: 2G internal GPU copy, is about 99GB/s.
With the idle power features disabled dynamically, the porformance
is promoted to about 215GB/s.

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: correct Arcturus OD support

OD is not supported on Arcturus. Thus the
pp_od_clk_voltage sysfs interface is also not supported.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: allow direct upload save restore list for raven2

It will cause modprobe atombios stuck problem in raven2 if it doesn't
allow direct upload save restore list from gfx driver.
So it needs to allow direct upload save restore list for raven2
temporarily.

Signed-off-by: changzhu <Changfeng.Zhu@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/sched: struct completion requires linux/completion.h inclusion

Fixes: 83a7772ba223 ("drm/sched: Use completion to wait for sched->thread idle v2.")
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Avoid accidental thread reactivation.

Problem:
During GPU reset we call the GPU scheduler to suspend it's
thread, those two functions in amdgpu also suspend and resume
the sceduler for their needs but this can collide with GPU
reset in progress and accidently restart a suspended thread
before time.

Fix:
Serialize with GPU reset.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/sched: Avoid job cleanup if sched thread is parked.

When the sched thread is parked we assume ring_mirror_list is
not accessed from here.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amdgpu: dont schedule jobs while in reset"

This reverts commit 89b3d86403f1025f6b430d8f9ffc590efbadce62.

We will do a proper fix in next patch.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/sched: Use completion to wait for sched->thread idle v2.

Removes thread park/unpark hack from drm_sched_entity_fini and
by this fixes reactivation of scheduler thread while the thread
is supposed to be stopped.

v2: Per sched entity completion.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix vega20 pstate status change

vega20 only requires all devices be set to same pstate level for low
pstate and not high.

Signed-off-by: Jonathan Kim <Jonathan.Kim@amd.com>
Reviewed-by: Evan Quan <Evan.Quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix sysfs interface pcie_replay_count error on navi asic

the asic callback function of get_pcie_replay_count is not implement on navi asic,
it will cause null pinter error when read this interface.

Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Kent Russell <kent.russell@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Need to disable msix when unloading driver

For driver reload test, it will report "can't enable
MSI (MSI-X already enabled)".

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

drm/amdgpu: Add comments to gmc structure

Explain fields like aper_base, agp_start etc. The definition
of those fields are confusing as they are from different view
(CPU or GPU). Add comments for easier understand.

Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
Reviewed-by: Alex Deucher <Alex.Deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>