Problem:
amdgpu_ras_reserve_bad_pages was moved to amdgpu_ras_reset_gpu
because writing to EEPROM during ASIC reset was unstable.
But for ERREVENT_ATHUB_INTERRUPT amdgpu_ras_reset_gpu is called
directly from ISR context and so locking is not allowed. Also it's
irrelevant for this partilcular interrupt as this is generic RAS
interrupt and not memory errors specific.
Fix:
Avoid calling amdgpu_ras_reserve_bad_pages if not in task context.
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The table grows quickly during debug/development effort when
multiple RAS errors are injected. Allow to avoid this by setting
table header back to empty if needed.
v2: Switch to debugfs entry instead of load time parameter.
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Shirish S [Thu, 12 Sep 2019 06:33:55 +0000 (12:03 +0530)]
drm/amdgpu: remove needless usage of #ifdef
define sched_policy in case CONFIG_HSA_AMD is not
enabled, with this there is no need to check for CONFIG_HSA_AMD
else where in driver code.
Suggested-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Shirish S <shirish.s@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Shirish S [Tue, 10 Sep 2019 07:15:01 +0000 (12:45 +0530)]
drm/amdgpu: fix build error without CONFIG_HSA_AMD
If CONFIG_HSA_AMD is not set, build fails:
drivers/gpu/drm/amd/amdgpu/amdgpu_device.o: In function `amdgpu_device_ip_early_init':
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c:1626: undefined reference to `sched_policy'
Use CONFIG_HSA_AMD to guard this.
Fixes: 1abb680ad371 ("drm/amdgpu: disable gfxoff while use no H/W scheduling policy") Signed-off-by: Shirish S <shirish.s@amd.com> Reviewed-by: Huang Rui <ray.huang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Mon, 2 Sep 2019 14:39:40 +0000 (16:39 +0200)]
drm/amdgpu: cleanup PTE flag generation v3
Move the ASIC specific code into a new callback function.
v2: mask the flags for SI and CIK instead of a BUG_ON().
v3: remove last missed BUG_ON().
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Huang Rui <ray.huang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Fri, 6 Sep 2019 17:20:40 +0000 (13:20 -0400)]
drm/amdgpu: check if nbio->ras_if exist
To avoid NULL function pointer access. This happens on VG10, reboot
command hangs and have to power off/on to reboot the machine. This is
serial console log:
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amdgpu: disable gfxoff while use no H/W scheduling policy
While gfxoff is enabled, the mmVM_XXX registers will be 0xfffffff while the GFX
is in "off" state. KFD queue creattion doesn't use ring based method, so it will
trigger a VM fault.
Signed-off-by: Huang Rui <ray.huang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Yong Zhao [Fri, 30 Aug 2019 22:09:10 +0000 (18:09 -0400)]
drm/amdgpu: Add a kernel parameter for specifying the asic type
As more and more new asics start to reuse the old device IDs before
launch, there is a need to quickly override the existing asic type
corresponding to the reused device ID through a kernel parameter. With
this, engineers no longer need to rely on local hack patches,
facilitating cooperation across teams.
Signed-off-by: Yong Zhao <Yong.Zhao@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Bayan Zabihiyan [Mon, 19 Aug 2019 19:18:43 +0000 (15:18 -0400)]
drm/amd/display: Isolate DSC module from driver dependencies
[Why]
Edid Utility wishes to include DSC module from driver instead
of doing it's own logic which will need to be updated every time
someone modifies the driver logic.
[How]
Modify some functions such that we dont need to pass the entire
DC structure as parameter.
-Remove DC inclusion from module.
-Filter out problematic types and inclusions
Signed-off-by: Bayan Zabihiyan <bayan.zabihiyan@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>
Jaehyun Chung [Mon, 19 Aug 2019 20:45:05 +0000 (16:45 -0400)]
drm/amd/display: OTC underflow fix
[Why] Underflow occurs on some display setups(repro'd on 3x4K HDR) on boot,
mode set, and hot-plugs with. Underflow occurs because mem clk
is not set high after disabling pstate switching. This behaviour occurs
because some calculations assumed displays were synchronized.
[How] Add a condition to check if timing sync is disabled so that
synchronized vblank can be set to false.
Signed-off-by: Jaehyun Chung <jaehyun.chung@amd.com> Reviewed-by: Alvin Lee <Alvin.Lee2@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Jun Lei [Thu, 15 Aug 2019 19:22:34 +0000 (15:22 -0400)]
drm/amd/display: remove hw access from dc_destroy
[why]
dc_destroy should only clean up SW, this is because GPUs may be
removed before driver unload, leading to HW to be unavailable.
[how]
remove GPIO close as part of GPIO destroy, this is unnecessary because
GPIO is not shared, and GPIOs are generally closed after being opened
Add tracking to HW access during destructor to make future issues
easier to pinpoint, and block access to prevent hangs.
Signed-off-by: Jun Lei <Jun.Lei@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>
[Why]
- Need to add missing surface address register definitions.
- RGBE+A does not work in a stereo configuration because
surface addresses are no programmed correctly.
[How]
Added the register definitions and surface address programming.
Signed-off-by: Ilya Bakoulin <Ilya.Bakoulin@amd.com> Reviewed-by: Charlene Liu <Charlene.Liu@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Nikola Cornij [Fri, 16 Aug 2019 18:26:56 +0000 (14:26 -0400)]
drm/amd/display: config to override DSC start slice height
[why]
It's sometimes useful to have this option when debugging
[how]
Add a config flag. If the flag is not set, use driver default policy.
If the flag is set, use the value from the flag as the starting DSC slice
height.
Signed-off-by: Nikola Cornij <nikola.cornij@amd.com> Reviewed-by: Martin Leung <Martin.Leung@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Krunoslav Kovac [Fri, 9 Aug 2019 21:16:18 +0000 (17:16 -0400)]
drm/amd/display: Subsample mode suboptimal for YCbCr4:2:2
[Why&How]
Driver defaults to 1-tap subsample mode for 4:2:2.
DCE11.2 added 3-tap. The policy is:
DCE8-DCE11 - change to 2-tap, it's better than 1-tap.
DCE11.2+ - use 3-tap
Note that 4:2:0 was added in DCE11.2 and already uses 3-tap always.
Note 2 is that DCE not covered on Linux, only DCN+
Signed-off-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com> Reviewed-by: Charlene Liu <Charlene.Liu@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Lewis Huang [Wed, 7 Aug 2019 10:05:49 +0000 (18:05 +0800)]
drm/amd/display: refine i2c over aux
[Why]
When user mode use i2c over aux through ADL or DDI, the function
dal_ddc_service_query_ddc_data will be called. There are two issues.
1. When read/write length > 16byte, current always return false because
the DEFAULT_AUX_MAX_DATA_SIZE != length.
2. When usermode only need to read data through i2c, driver will write
mot = true at the same address and cause i2c sink confused. Therefore
the following read command will get garbage.
[How]
1. Add function dal_dcc_submit_aux_command to handle length > 16 byte.
2. Check read size and write size when query ddc data.
Signed-off-by: Lewis Huang <Lewis.Huang@amd.com> Reviewed-by: Charlene Liu <Charlene.Liu@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Tao Zhou [Fri, 30 Aug 2019 11:50:39 +0000 (19:50 +0800)]
drm/amdgpu: move the call of ras recovery_init and bad page reserve to proper place
ras recovery_init should be called after ttm init,
bad page reserve should be put in front of gpu reset since i2c
may be unstable during gpu reset.
add cleanup for recovery_init and recovery_fini
v2: add more comment and print.
remove cancel_work_sync in recovery_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>
Felix Kuehling [Fri, 30 Aug 2019 04:10:34 +0000 (00:10 -0400)]
drm/amdgpu: Disable page faults while reading user wptrs
These wptrs must be pinned and GPU accessible when this is called
from hqd_load functions. So they should never fault. This resolves
a circular lock dependency issue involving four locks including the
DQM lock and mmap_sem.
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Oak Zeng <Oak.Zeng@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Aaron Liu [Wed, 4 Sep 2019 03:47:48 +0000 (11:47 +0800)]
drm/amdgpu: disable stutter mode for renoir
With stutter mode enabled, NMI prints frequently.
Disable stutter for the moment because NMI warning storm, and will
enable it back till the issue is addressed
Signed-off-by: Aaron Liu <aaron.liu@amd.com> Acked-by: Huang Rui <ray.huang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Prike Liang <Prike.Liang@amd.com> Reviewed-by: Evan Quan <evan.quan@amd.com> Reviewed-by: Kevin Wang <kevin1.wang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Tao Zhou [Mon, 2 Sep 2019 11:27:23 +0000 (19:27 +0800)]
drm/amdgpu: change r type to int in gmc_v9_0_late_init
change r type from bool to int, suitable for both bool and int return
value
Signed-off-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Kevin Wang [Tue, 3 Sep 2019 08:02:33 +0000 (16:02 +0800)]
drm/amd/powerplay: replace smu->table_count with SMU_TABLE_COUNT in smu (v2)
fix bellow patch issue:
drm/amd/powerplay: introduce smu table id type to handle the smu table
for each asic
----
"This patch introduces new smu table type, it's to handle the
different smu table
defines for each asic with the same smu ip."
before:
use smu->table_count to represent the actual table count in smc firmware
use actual table count to check smu function parameter about smu table
after:
use logic table count "SMU_TABLE_COUNT" to check function parameter
because table id already mapped in smu driver,
and smu function will use logic table id not actual table id to check func parameter.
v2: squash in warning fix
Signed-off-by: Kevin Wang <kevin1.wang@amd.com> Reviewed-by: Evan Quan <evan.quan@amd.com> Acked-by: Huang Rui <ray.huang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Jack Zhang [Tue, 3 Sep 2019 02:15:23 +0000 (10:15 +0800)]
drm/amd/amdgpu: add sw_fini interface for df_funcs
add sw_fini interface of df_funcs.
This interface will remove sysfs file of df_cntr_avail
function.
The old behavior only create sysfs of df_cntr_avail
in sw_init, but never remove it for lack of sw_fini
interface. With this,driver will report create
sysfs fail when it's loaded for the second time.
Signed-off-by: Jack Zhang <Jack.Zhang1@amd.com> Reviewed-by: Jonathan Kim <Jonathan.Kim@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amdgpu/nbio: switch to amdgpu_nbio_ras_late_init helper function
amdgpu_nbio_ras_late_init is used to init nbio specfic
ras debugfs/sysfs node and nbio specific interrupt handler.
It can be shared among nbio generations
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amdgpu/mmhub: switch to amdgpu_mmhub_ras_late_init helper function
amdgpu_mmhub_ras_late_init is used to init mmhub specfic
ras debugfs/sysfs node and mmhub specific interrupt handler.
It can be shared among mmhub generations
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amdgpu/sdma: switch to amdgpu_sdma_ras_late_init helper function
amdgpu_sdma_ras_late_init is used to init sdma specfic
ras debugfs/sysfs node and sdma specific interrupt handler.
It can be shared among sdma generations
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amdgpu/gmc: switch to amdgpu_gmc_ras_late_init helper function
amdgpu_gmc_ras_late_init is used to init gmc specfic
ras debugfs/sysfs node and gmc specific interrupt handler.
It can be shared among gmc generations.
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
In case of RAS error allow user configure auto system
reboot through ras_ctrl.
This is also part of the temproray work around for the RAS
hang problem.
Problem:
Under certain conditions, when some IP bocks take a RAS error,
we can get into a situation where a GPU reset is not possible
due to issues in RAS in SMU/PSP.
Temporary fix until proper solution in PSP/SMU is ready:
When uncorrectable error happens the DF will unconditionally
broadcast error event packets to all its clients/slave upon
receiving fatal error event and freeze all its outbound queues,
err_event_athub interrupt will be triggered.
In such case and we use this interrupt
to issue GPU reset. THe GPU reset code is modified for such case to avoid HW
reset, only stops schedulers, deatches all in progress and not yet scheduled
job's fences, set error code on them and signals.
Also reject any new incoming job submissions from user space.
All this is done to notify the applications of the problem.
v2:
Extract amdgpu_amdkfd_pre/post_reset from amdgpu_device_lock/unlock_adev
Move amdgpu_job_stop_all_jobs_on_sched to amdgpu_job.c
Remove print param from amdgpu_ras_query_error_count
v3:
Update based on prevoius bug fixing patch to properly call amdgpu_amdkfd_pre_reset
for other XGMI hive memebers.
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com> Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amdgpu: Fix bugs in amdgpu_device_gpu_recover in XGMI case.
Issue 1:
In XGMI case amdgpu_device_lock_adev for other devices in hive
was called to late, after access to their repsective schedulers.
So relocate the lock to the begining of accessing the other devs.
Issue 2:
Using amdgpu_device_ip_need_full_reset to switch the device list from
all devices in hive to the single 'master' device who owns this reset
call is wrong because when stopping schedulers we iterate all the devices
in hive but when restarting we will only reactivate the 'master' device.
Also, in case amdgpu_device_pre_asic_reset conlcudes that full reset IS
needed we then have to stop schedulers for all devices in hive and not
only the 'master' but with amdgpu_device_ip_need_full_reset we
already missed the opprotunity do to so. So just remove this logic and
always stop and start all schedulers for all devices in hive.
Also minor cleanup and print fix.
v4: Minor coding style fix.
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com> Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Fri, 30 Aug 2019 12:42:10 +0000 (14:42 +0200)]
drm/amdgpu: remove amdgpu_cs_try_evict
Trying to evict things from the current working set doesn't work that
well anymore because of per VM BOs.
Rely on reserving VRAM for page tables to avoid contention.
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Fri, 30 Aug 2019 12:38:37 +0000 (14:38 +0200)]
drm/amdgpu: reserve at least 4MB of VRAM for page tables v2
This hopefully helps reduce the contention for page tables.
v2: adjust maximum reported VRAM size as well
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Fri, 13 Sep 2019 22:38:07 +0000 (17:38 -0500)]
drm/amdgpu: use moving fence instead of exclusive for VM updates
Make VM updates depend on the moving fence instead of the exclusive one.
Makes it less likely to actually have a dependency.
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Evan Quan [Wed, 21 Aug 2019 09:19:52 +0000 (17:19 +0800)]
drm/amd/powerplay: update cached feature enablement status V3
Need to update in cache feature enablement status after pp_feature
settings. Another fix for the commit below:
drm/amd/powerplay: implment sysfs feature status function in smu
V2: update smu_feature_update_enable_state() and relates
V3: use bitmap_or and bitmap_andnot
Signed-off-by: Evan Quan <evan.quan@amd.com> Reviewed-by: Jack Gui <Jack.Gui@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Evan Quan [Fri, 30 Aug 2019 09:30:46 +0000 (17:30 +0800)]
drm/amd/powerplay: guard manual mode prerequisite for clock level force
Force clock level is for dpm manual mode only.
Reported-by: Candice Li <candice.li@amd.com> Signed-off-by: Evan Quan <evan.quan@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Jack Gui <Jack.Gui@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This workaround is better handled in user mode in a way that doesn't
require allocating extra memory and breaking userptr BOs.
The TLB bug is a performance bug, not a functional or security bug.
Hence it is safe to remove this kernel part of the workaround to
allow a better workaround using only virtual address alignments in
user mode.
v2: Removed VI_BO_SIZE_ALIGN definition
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Mon, 26 Aug 2019 22:46:28 +0000 (18:46 -0400)]
drm/amdgpu: Use optimal mtypes and PTE bits for Arcturus
For compute VRAM allocations on Arturus use the new RW mtype
for non-coherent local memory, CC mtype for coherent local
memory and PTE_SNOOPED bit for invalidating non-dirty cache
lines on remote XGMI mappings.
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Acked-by: Christian König <christian.koenig@amd.com> Tested-by: Amber Lin <Amber.Lin@amd.com> Reviewed-by: Shaoyun Liu <Shaoyun.Liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>