drm/amdgpu: move amdgpu_device_get_job_timeout_settings

It's only used in amdgpu_device.c and the naming also
reflects that.  Move it there.

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

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

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

rivers/gpu/drm/amd/amdgpu/../display/modules/freesync/freesync.c:
 In function mod_freesync_get_settings:
drivers/gpu/drm/amd/amdgpu/../display/modules/freesync/freesync.c:984:24:
 warning: variable core_freesync set but not used [-Wunused-but-set-variable]

It is not used since commit 98e6436d3af5 ("drm/amd/display: Refactor FreeSync module")

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix typo in some comments

p and g are switched in 'amdpgu_dm'

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Print more sdma engine hqds in debug fs

Previously only PCIe-optimized SDMA engine hqds were
exposed in debug fs. Print all SDMA engine hqds.

Reported-by: Jonathan Kim <Jonathan.Kim@amd.com>
Signed-off-by: Jonathan Kim <Jonathan.Kim@amd.com>
Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Fix MQD size calculation

On device initialization, a chunk of GTT memory is pre-allocated for
HIQ and all SDMA queues mqd. The size of this allocation was wrong.
The correct sdma engine number should be PCIe-optimized SDMA engine
number plus xgmi SDMA engine number.

Reported-by: Jonathan Kim <Jonathan.Kim@amd.com>
Signed-off-by: Jonathan Kim <Jonathan.Kim@amd.com>
Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix error handling in amdgpu_ras_recovery_init

Don't set a struct pointer to NULL before freeing its members. It's
hard to see what's happening due to a local pointer-to-pointer data
aliasing con->eh_data.

Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Tested-by: Philip Cox <Philip.Cox@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove redundant variable r and redundant return statement

There is a return statement that is not reachable and a variable that
is not used.  Remove them.

Addresses-Coverity: ("Structurally dead code")
Fixes: de7b45babd9b ("drm/amdgpu: cleanup creating BOs at fixed location (v2)")
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix uninitialized variable pasid_mapping_needed

The boolean variable pasid_mapping_needed is not initialized and
there are code paths that do not assign it any value before it is
is read later.  Fix this by initializing pasid_mapping_needed to
false.

Addresses-Coverity: ("Uninitialized scalar variable")
Fixes: 6817bf283b2b ("drm/amdgpu: grab the id mgr lock while accessing passid_mapping")
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Make some functions static

Fix sparse warnings:

drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_hdcp.c:32:6: warning: symbol 'lp_write_i2c' was not declared. Should it be static?
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_hdcp.c:42:6: warning: symbol 'lp_read_i2c' was not declared. Should it be static?
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_hdcp.c:52:6: warning: symbol 'lp_write_dpcd' was not declared. Should it be static?
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_hdcp.c:59:6: warning: symbol 'lp_read_dpcd' was not declared. Should it be static?

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: fix struct init in update_bounding_box

dcn20_resource.c:2636:9: error: missing braces around initializer [-Werror=missing-braces]
  struct _vcs_dpi_voltage_scaling_st calculated_states[MAX_CLOCK_LIMIT_STATES] = {0};
         ^

Fixes: 7ed4e6352c16f ("drm/amd/display: Add DCN2 HW Sequencer and Resource")

Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add code comment in vcn_v2_5_hw_init

Add a comment to VCN 2.5 encode ring

Signed-off-by: Leo Liu <leo.liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: James Zhu <James.Zhu@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: David (ChunMing) Zhou <David1.Zhou@amd.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: amd-gfx@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Cc: linux-kernel@vger.kernel.org
Cc: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/vcn: use amdgpu_ring_test_helper

Instead of amdgpu_ring_test_ring, so the helper function determines
whether the ring is ready

Signed-off-by: Leo Liu <leo.liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: James Zhu <James.Zhu@amd.com>
Acked-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Cc: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: improve MSI-X handling (v3)

Check the number of supported vectors and fall back to MSI if
we return or error or 0 MSI-X vectors.

v2: only allocate one vector.  We can't currently use more than
one anyway.

v3: install the irq on vector 0.

Tested-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Shaoyun liu <shaoyun.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix spelling mistake AUTHENICATED -> AUTHENTICATED

There is a spelling mistake in the macros H1_A45_AUTHENICATED and
D1_A4_AUTHENICATED, fix these by adding the missing T.

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: display_mode_vba_21: remove uint typedef

The type definition for 'uint' clashes with the generic kernel
headers:

drivers/gpu/drm/amd/amdgpu/../display/dc/dml/dcn21/display_mode_vba_21.c:43:22: error: redefinition of typedef 'uint' is a C11 feature [-Werror,-Wtypedef-redefinition]
include/linux/types.h:92:23: note: previous definition is here

Just remove this type and use plain 'unsigned int' consistently,
as it is already use almost everywhere in this file.

Fixes: b04641a3f4c5 ("drm/amd/display: Add Renoir DML")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: hide another #warning

An earlier patch of mine disabled some #warning statements
that get in the way of build testing, but then another
instance was added around the same time.

Remove that as well.

Fixes: b5203d16aef4 ("drm/amd/amdgpu: hide #warning for missing DC config")
Fixes: e1c14c43395c ("drm/amdgpu: Enable DC on Renoir")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: make pmu support optional, again

When CONFIG_PERF_EVENTS is disabled, we cannot compile the pmu
portion of the amdgpu driver:

drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c:48:38: error: no member named 'hw' in 'struct perf_event'
        struct hw_perf_event *hwc = &event->hw;
                                     ~~~~~  ^
drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c:51:13: error: no member named 'attr' in 'struct perf_event'
        if (event->attr.type != event->pmu->type)
            ~~~~~  ^
...

The same bug was already fixed by commit d155bef0636e ("amdgpu: make pmu
support optional") but broken again by what looks like an incorrectly
rebased patch.

Fixes: 64f55e629237 ("drm/amdgpu: Add RAS EEPROM table.")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: memory leak

In dcn*_clock_source_create when dcn20_clk_src_construct fails allocated
clk_src needs release.

Signed-off-by: Navid Emamdoost <navid.emamdoost@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

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

rivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c: In function
‘amdgpu_gfx_graphics_queue_acquire’:
drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c:234:16: warning:
variable ‘pipe’ set but not used [-Wunused-but-set-variable]

It is never used, so can be removed.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: yu kuai <yukuai3@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Drop unused variable and statement

Even though 'smu8_smu' is declared, it is not used after below statement.

   smu8_smu = hwmgr->smu_backend;

So 'unused variable' could be safely removed
to stop warning message as below:

   drivers/gpu/drm/amd/amdgpu/../powerplay/smumgr/smu8_smumgr.c:180:22:
   warning: variable ‘smu8_smu’ set but not used
   [-Wunused-but-set-variable]

   struct smu8_smumgr *smu8_smu;
             ^
Signed-off-by: Austin Kim <austindh.kim@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix multiple memory leaks in acp_hw_init

In acp_hw_init there are some allocations that needs to be released in
case of failure:

1- adev->acp.acp_genpd should be released if any allocation attemp for
adev->acp.acp_cell, adev->acp.acp_res or i2s_pdata fails.
2- all of those allocations should be released if
mfd_add_hotplug_devices or pm_genpd_add_device fail.
3- Release is needed in case of time out values expire.

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

drm/amdgpu: don't increment vram lost if we are in hibernation

We reset the GPU as part of our hibernation sequence so we need
to make sure we don't mark vram as lost in that case.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=111879
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu : enable msix for amdgpu driver

We might used out of the msi resources in some cloud project
which have a lot gpu devices(including PF and VF), msix can
provide enough resources from system level view

Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Improve KFD IOCTL printing

The code use hex define, so should the printing.

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/amdgpu: Export setup_vm_pt_regs() logic for mmhub 2.0

The KFD code will call this function later.

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

drm/amdgpu: Add the HDP flush support for Navi

The HDP flush support code was missing in the nbio and nv files.

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 array to probe kfd2kgd_calls

This is the same idea as the kfd device info probe and move all the
probe control together for easy maintenance.

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: Delete unnecessary function declarations

Ajust the function sequences so that those function delcarations are not
needed any more.

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/amdgpu: Delete useless header file reference

Those header file includes are not needed.

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/amdgpu/sriov ip block setting of Arcturus

Add ip block setting for Arcturus SRIOV

1.PSP need to be initialized before IH.
2.SMU doesn't need to be initialized at kmd driver.
3.Arcturus doesn't support DCE hardware,it needs to skip
  register access to DCE.

Signed-off-by: Jack Zhang <Jack.Zhang1@amd.com>
Reviewed-by: Le Ma <Le.Ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: return tcc_disabled_mask to userspace

UMDs need this for correct programming of harvested chips.

Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/dm/mst: Use ->atomic_best_encoder

We are supposed to be atomic after all. We'll need this in a moment for
the next commit.

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

drm/amdgpu/dm/mst: Don't create MST topology managers for eDP ports

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

drm/amdgpu/dm/mst: Remove unnecessary NULL check

kfree() checks this automatically.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Iterate through DRM connectors correctly

Currently, every single piece of code in amdgpu that loops through
connectors does it incorrectly and doesn't use the proper list iteration
helpers, drm_connector_list_iter_begin() and
drm_connector_list_iter_end(). Yeesh.

So, do that.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/dm: Resume short HPD IRQs before resuming MST topology

Since we're going to be reprobing the entire topology state on resume
now using sideband transactions, we need to ensure that we actually have
short HPD irqs enabled before calling drm_dp_mst_topology_mgr_resume().
So, do that.

Changes since v4:
* Fix typo in comments

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: fix kgd2kfd_device_init() definition conflict error

The patch c670707 drm/amd: Pass drm_device to kfd introduced this issue and
fix the following compiler error.

  CC [M]  drivers/gpu/drm/amd/amdgpu//../powerplay/smumgr/fiji_smumgr.o
drivers/gpu/drm/amd/amdgpu//amdgpu_amdkfd.c:746:6: error: conflicting types for ‘kgd2kfd_device_init’
 bool kgd2kfd_device_init(struct kfd_dev *kfd,
      ^
In file included from drivers/gpu/drm/amd/amdgpu//amdgpu_amdkfd.c:23:0:
drivers/gpu/drm/amd/amdgpu//amdgpu_amdkfd.h:253:6: note: previous declaration of ‘kgd2kfd_device_init’ was here
 bool kgd2kfd_device_init(struct kfd_dev *kfd,
      ^
scripts/Makefile.build:273: recipe for target 'drivers/gpu/drm/amd/amdgpu//amdgpu_amdkfd.o' failed
make[1]: *** [drivers/gpu/drm/amd/amdgpu//amdgpu_amdkfd.o] Error 1

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

drm/amd/powerplay: initlialize smu->is_apu is false by default

the member of is_apu in smu_context need to initlialize by default.

set default value is false (dGPU)

for patch:
drm/amd/powerplay: bypass dpm_context null pointer check guard
for some smu series

Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Prike Liang <Prike.Liang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amdgpu: disable stutter mode for renoir"

This reverts commit 5813f97a5969bf1e7e723397a74e00b5de7278d6.

Since SBIOS WCD9925N, NMI printing disappeared. Hence enable stutter
mode.

Signed-off-by: Aaron Liu <aaron.liu@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: update the interface for getting dpm full scale clock frequency

Update get_dpm_uclk_limited to get more clock type full scale dpm frequency.

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

drm/amd/powerplay: implement the interface for setting sclk/uclk profile_peak level

Add the interface for setting sclk and uclk peak frequency.

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

drm/amd/powerplay: implement interface set_power_profile_mode() (v2)

v1:
Add set_power_profile_mode() for none manual dpm level case setting power profile mode.

v2:
After input power profile update successfully need can update the smu profile mode.

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

drm/amd/powerplay: add the interfaces for getting and setting profiling dpm clock level

implement get_profiling_clk_mask and force_clk_levels for forcing dpm clk to limit value.

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

drm/amd/powerplay: add interface for getting workload type

The workload type was got from the input of power profile mode.

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

drm/amd/powerplay: add interface for forcing and unforcing dpm limit value

That's base function for forcing and unforcing dpm limit value.

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

drm/amd/powerplay: implement the interface for setting soft freq range

The APU soft freq range set by different way from DGPU, thus need implement
the function respectively base on each common SMU part.

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

drm/amd/powerplay: bypass dpm_context null pointer check guard for some smu series

For now APU has no smu_dpm_context structure for containing default/current related dpm tables,
thus will needn't initialize smu_dpm_context to aviod APU null pointer issue.

Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu: add IH cg support on soc15 project

enable/disable IH clock gating on soc15 projects.

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: add sensor lock support for smu

when multithreading access sysfs of amdgpu_pm_info at the sametime.
the swsmu driver cause smu firmware hang.

eg:
single thread access:
Message A + Param A ==> right
Message B + Param B ==> right
Message C + Param C ==> right
multithreading access:
Message A + Param B ==> error
Message B + Param A ==> error
Message C + Param C ==> right

the patch will add sensor lock(mutex) to avoid this error.

Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: change metrics update period from 1ms to 100ms

v2:
change period from 10ms to 100ms (typo error)

too high frequence to update mertrics table will cause smu firmware
error,so change mertrics table update period from 1ms to 100ms
(navi10, 12, 14)

Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Fix race in gfx10 context restore handler

Missing synchronization with VGPR restore leads to intermittent
VGPR trashing in the user shader.

Signed-off-by: Jay Cornwall <jay.cornwall@amd.com>
Reviewed-by: Yong Zhao <Yong.Zhao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Use setup_vm_pt_regs function from base driver in KFD

This was done on GFX9 previously, now do it for 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/amdgpu: Export setup_vm_pt_regs() logic for gfxhub 2.0

The KFD code will call this function later.

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

drm/amdkfd: Eliminate get_atc_vmid_pasid_mapping_valid

get_atc_vmid_pasid_mapping_valid() is very similar to
get_atc_vmid_pasid_mapping_pasid(), so they can be merged into a new
function get_atc_vmid_pasid_mapping_info() to reduce register access
times. More importantly, getting the PASID and the valid bit atomically
with a single read fixes some potential race conditions where the
mapping changes between the two reads.

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: Query vmid pasid mapping through stored info for non HWS

Because we record the mapping under non HWS mode in the software,
we can query pasid through vmid using the stored mapping instead of
reading from ATC registers.

This also prepares for the defeatured ATC block in future ASICs.

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: Record vmid pasid mapping in the driver for non HWS mode

This makes possible the vmid pasid mapping query through software.

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 hex print format for pasid

Since KFD pasid starts from 0x8000 (32768 in decimal), it is better
perceived as a hex number. Meanwhile, change the pasid type from
unsigned int to uint16_t to be consistent throughout the code.

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: Delete unused defines

They are not used anywhere.

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: Move the control stack on GFX10 to userspace buffer

The GFX10 does not require the control stack to be right after mqd
buffer any more, so move it back to usersapce allocated CSWR buffer.

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: Pass drm_device to kfd

kfd needs drm_device to call into drm_cgroup functions

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

drm/amdkfd: Store kfd_dev in iolink and cache properties

This is required to check against cgroup permissions.

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

drm/amdkfd: use navi12 specific family id for navi12 code path

Keep the same use of CHIP_IDs for navi12 in kfd

Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add SMUIO values for other I2C controller v2

These are the offsets for CKSVII2C1, and match up with the values
already added for CKSVII2C

v2: Don't remove some of the CSKVII2C values

Signed-off-by: Kent Russell <kent.russell@amd.com>
Reviewed-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Use better name for sdma queue non HWS path

The old name is prone to confusion. The register offset is for a RLC queue
rather than a SDMA engine. The value is not a base address, but a
register offset.

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

drm/amdkfd: Delete useless SDMA register setting on non HWS path

HW folks have confirm that we should not touch RESUME_CTX of
SDMA*_GFX_CONTEXT_CNTL when manipulating RLC queues.

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

drm/amd/display: prevent memory leak

In dcn*_create_resource_pool the allocated memory should be released if
construct pool fails.

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

drm/amdkfd: Add NAVI12 support from kfd side

Add device info for both navi12 PF and VF

Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: restrict hotplug error message

We should print the error only when we are hotplugged and crash
basically all userspace applications.

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

drm/amdgpu: once more fix amdgpu_bo_create_kernel_at

When CPU access is needed we should tell that to
amdgpu_bo_create_reserved() or otherwise the access is denied later on.

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

drm/amd/powerplay: update arcturus smu-driver interaction header

To pair the latest SMU firmware.

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

drm/amdgpu: add comments in ras interrupt callback

add comments to clarify why checking GFX IP BLOCK for each ras interrupt callback

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: implement common gmc_ras_late_init

common gmc_ecc_late_init can be shared among all generations of gmc

v2: rename gmc_ecc_late_init to gmc_ras_late_init

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move xgmi ras fini to xgmi block

it's more suitable to put xgmi ras fini in xgmi block

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move mmhub ras fini to mmhub block

it's more suitable to put mmhub ras fini in mmhub block

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move umc ras fini to umc block

it's more suitable to put umc ras fini in umc block

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add ras fini for xgmi

add ras fini for xgmi to cleanup xgmi ras framework

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add ras fini for nbio

add a common nbio ras fini implementation to cleanup nbio ras framework

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: simplify the access to eeprom_control struct

simplify the code of accessing to eeprom_control struct

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove ih_info parameter of gfx_ras_late_init

gfx_ras_late_init can get the info by itself

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove ih_info parameter of umc_ras_late_init

umc_ras_late_init can get the info by itself

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add common sdma_ras_fini function

sdma_ras_fini can be shared among all generations of sdma

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add common gfx_ras_fini function

gfx_ras_fini can be shared among all generations of gfx

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add common gmc_ras_fini function

gmc_ras_fini can be shared among all generations of gmc

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move mmhub_ras_if from gmc to mmhub block

mmhub_ras_if is relevant to mmhub

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: replace mmhub_funcs with mmhub.funcs

remove mmhub_funcs in adev

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add common mmhub member for adev

put mmhub_funcs and ras_if pointer into mmhub struct

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move umc_ras_if from gmc to umc block

umc_ras_if is relevant to umc

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: refine sdma4 ras_data_cb

simplify code logic and refine return value

v2: remove unused error source code

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move sdma ecc functions to generic sdma file

sdma ras ecc functions can be reused among all sdma generations

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move gfx ecc functions to generic gfx file

gfx ras ecc common functions could be reused among all gfx generations

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move umc ras irq functions to umc block

move umc ras irq functions from gmc v9 to generic umc block, these
functions are relevant to umc and they can be shared among all
generations of umc

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: update parameter of ras_ih_cb

change struct ras_err_data *err_data to void *err_data, align with
umc code and the callback's declaration in each ras block could
pay no attention to the structure type

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix an UMC hw arbitrator bug(v3)

issue:
the UMC6 h/w bug is that when MCLK is doing the switch
in the middle of a page access being preempted by high
priority client (e.g. DISPLAY) then UMC and the mclk switch
would stuck there due to deadlock

how:
fixed by disabling auto PreChg for UMC to avoid high
priority client preempting other client's access on
the same page, thus the deadlock could be avoided

v2:
put the patch in callback of UMC6
v3:
rename the callback to "init_registers"

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Hawking Zhang <hawking.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove gfx9 NGG

Never used.

Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/atomfirmware: simplify the interface to get vram info

fetch both the vram type and width in one function call.  This
avoids having to parse the same data table twice to get the two
pieces of data.

Reviewed-by: Xiaojie Yuan <xiaojie.yuan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/atomfirmware: use proper index for querying vram type (v3)

The index is stored in scratch register 4 after asic init.  Use
that index.  No functional change since all asics in a family
use the same type of vram (G5, G6, HBM) and that is all we use
at the monent, but if we ever need to query other info, we will
now have the proper index.

v2: module array is variable sized, handle that.
v3: fix off by one in array handling

Reviewed-by: Xiaojie Yuan <xiaojie.yuan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/display: include slab.h in dcn21_resource.c

It's apparently needed in some configurations.

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

drm/amdgpu/psp: silence response status warning

log the response status related error to the driver's
debug log since  psp response status is not 0 even though
there was no problem while the command was submitted.

This warning misleads, hence this change.

Signed-off-by: Shirish S <shirish.s@amd.com>
Acked-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: remove duplicate macro of smu_get_uclk_dpm_states

remove duplicate macro of smu_get_uclk_dpm_states

fix commit:
drm/amd/powerplay: add the interface for getting ultimate frequency v3

Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Prike Liang <Prike.Liang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu:Fix compute ring unable to detect hang.

When compute fence did not signal, compute ring cannot detect hardware hang
because its timeout value is set to be infinite by default.

In SR-IOV and passthrough mode, if user does not declare custome timeout
value for compute ring, then use gfx ring timeout value as default. So
that when there is a ture hardware hang, compute ring can detect it.

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>