drm/amdgpu: allow direct submission in the VM backends v2

This allows us to update page tables directly while in a page fault.

v2: use direct/delayed entities and still wait for moves

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

drm/amdgpu: split the VM entity into direct and delayed

For page fault handling we need to use a direct update which can't be
blocked by ongoing user CS.

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

drm/ttm: return -EBUSY on pipelining with no_gpu_wait (v2)

Setting the no_gpu_wait flag means that the allocate BO must be available
immediately and we can't wait for any GPU operation to finish.

v2: squash in mem leak fix, rebase

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

drm/amdgpu: grab the id mgr lock while accessing passid_mapping

Need to make sure that we actually dropping the right fence.
Could be done with RCU as well, but to complicated for a fix.

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>

drm/amdgpu/SRIOV: Navi12 SRIOV VF doesn't load TOC

In SRIOV case, the autoload sequence is the same

as bare metal, except VF won't load TOC.

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

drm/amdgpu/SRIOV: Navi10/12 VF doesn't support SMU

In SRIOV case, SMU and powerplay are handled in HV.

VF shouldn't have control over SMU and powerplay.

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

drm/amd/amdgpu: power up sdma engine when S3 resume back

The sdma_v4 should be ungated when the IP resume back,
otherwise it will hang up and resume time out error.

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/amdgpu: For Navi12 SRIOV VF, register mailbox functions

Mailbox functions and interrupts are only for Navi12 VF.

Register functions and irqs during initialization.

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

drm/amdgpu/sriov: add ring_stop before ring_create in psp v11 code

psp  v11 code missed ring stop in ring create function(VMR)
while psp v3.1 code had the code. This will cause VM destroy1
fail and psp ring create fail.

For SIOV-VF, ring_stop should not be deleted in ring_create
function.

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

drm/amd/powerplay: check SMU engine readiness before proceeding on S3 resume

This is especially needed for non-psp loading way.

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

drm/amd/powerplay: properly set mp1 state for SW SMU suspend/reset routine

Set mp1 state properly for SW SMU suspend/reset routine.

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

drm/amdgpu: Fix KFD-related kernel oops on Hawaii

Hawaii needs to flush caches explicitly, submitting an IB in a user
VMID from kernel mode. There is no s_fence in this case.

Fixes: eb3961a57424 ("drm/amdgpu: remove fence context from the job")
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>

drm/amdgpu: Fix mutex lock from atomic context.

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>

drm/amdgpu: Add SRIOV mailbox backend for Navi1x

Mimic the ones for Vega10, add mailbox backend for Navi1x

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

drm/amdgpu: implement ras query function for pcie bif

ras error query funtionality implementation

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

drm/amdgpu: add pcie bif ras related registers

These registers will be accessed for querying ras errors.

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

drm/amdgpu: support pcie bif ras query and inject

Call pcie bif ras query/inject in amdgpu ras.

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

drm/amdgpu: add ras error query count interface for nbio

Add the interface query_ras_error_count for nbio.

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

drm/amdgpu: fix CPDMA hang in PRT mode for VEGA10

add and_mask since the programming logic of golden setting changed

Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Tianci.Yin <tianci.yin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: enable error injection to XGMI block via debugfs

allow inject error to XGMI block via debugfs node ras_ctrl

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: initialize ras structures for xgmi block (v2)

init ras common interface and fs node for xgmi block

v2: remove unnecessary physical node number check before
invoking amdgpu_xgmi_ras_late_init

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: fix the missed asic name while inited renoir_device_info

This patch fixes null pointer issue below, I missed to init the asic renior name
while I rebase the patches.

[  106.004250] BUG: kernel NULL pointer dereference, address: 0000000000000000
[  106.004254] #PF: supervisor read access in kernel mode
[  106.004256] #PF: error_code(0x0000) - not-present page
[  106.004257] PGD 0 P4D 0
[  106.004261] Oops: 0000 [#1] SMP NOPTI
[  106.004264] CPU: 3 PID: 1422 Comm: modprobe Not tainted 5.2.0-rc1-custom #1
[  106.004266] Hardware name: AMD Celadon-RN/Celadon-RN, BIOS
WCD9814N_Weekly_19_08_1 08/14/2019
[  106.004272] RIP: 0010:strncpy+0x12/0x30
[  106.004274] Code: c1 c0 11 48 c1 c6 15 48 31 d0 48 c1 c2 20 31 c2 89 d0 31 f0
41 5c 5d c3 55 48 85 d2 48 89 f8 48 89 e5 74 1e 48 01 fa 48 89 f9 <44> 0f b6 06
41 80 f8 01 44 88 01 48 83 de ff 48 83 c1 01 48 39 d1
[  106.004278] RSP: 0018:ffffc092c1fd37a8 EFLAGS: 00010286
[  106.004281] RAX: ffff9e943466a28c RBX: 00000000000036ed RCX: ffff9e943466a28c
[  106.004283] RDX: ffff9e943466a2ac RSI: 0000000000000000 RDI: ffff9e943466a28c
[  106.004285] RBP: ffffc092c1fd37a8 R08: ffff9e943d100000 R09: 0000000000000228
[  106.004287] R10: ffff9e94418dc5a8 R11: ffff9e944746c0d0 R12: 0000000000000000
[  106.004289] R13: ffff9e943fa1ec00 R14: ffff9e943466a200 R15: ffff9e943466a200
[  106.004291] FS:  00007f7a022c5540(0000) GS:ffff9e9447ac0000(0000)
knlGS:0000000000000000
[  106.004294] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  106.004296] CR2: 0000000000000000 CR3: 00000001ff0b0000 CR4: 0000000000340ee0
[  106.004298] Call Trace:
[  106.004382]  kfd_topology_add_device+0x150/0x610 [amdgpu]
[  106.004445]  kgd2kfd_device_init+0x2e0/0x4f0 [amdgpu]
[  106.004509]  amdgpu_amdkfd_device_init+0x14c/0x1b0 [amdgpu]

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

drm/amd/display: Implement voltage limitation for dali

[Why]
we only want the lowest voltage to be available for dali.

[How]
Use the get_highest_allowed_voltage_level function
to return 0 for dali

Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add Asic ID for Dali

Dali is a new asic revision based on raven2

Add the ID and ASICREV_IS_DALI define

Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Allow to reset to EERPOM table.

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>

drm/amdgpu: Add amdgpu_ras_eeprom_reset_table

This will allow to reset the table on the fly.

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>

drm/amdgpu: rename umc ras_init to err_cnt_init

this interface is related to specific version of umc, distinguish it
from 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 umc ras init to umc block

move umc ras init from ras module to umc block, generic ras module
should pay less attention to specific ras 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 late init from gmc to umc block

umc late init is umc specific, it's more suitable to be put 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: remove duplicated header file include

amdgpu_ras.h is already included.

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

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>

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>

drm/amd/powerplay: update smu11_driver_if_arcturus.h

Also bump the SMU11_DRIVER_IF_VERSION_ARCT.

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

drm/amd/powerplay: issue DC-BTC for arcturus on SMU init

Need to perform DC-BTC for arcturus on bootup.

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

drm/amdgpu: Avoid RAS recovery init when no RAS support.

Fixes driver load regression on APUs.

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

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>

drm/amdgpu: cleanup mtype mapping

Unify how we map the UAPI flags to the PTE hardware flags for a mapping.

Only the MTYPE is actually ASIC dependent, all other flags should be
copied over 1 to 1 and ASIC differences are handled later on.

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

drm/amdgpu: add navi14 PCI ID for work station SKU

Add the navi14 PCI device id.

Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Tianci.Yin <tianci.yin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: Add the interface for geting dpm current power state

implement the sysfs power_dpm_state

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/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:

[  OK  ] Reached target Unmount All Filesystems.
[  OK  ] Reached target Final Step.
         Starting Reboot...
[  305.696271] systemd-shutdown[1]: Syncing filesystems and block
devices.
[  306.947328] systemd-shutdown[1]: Sending SIGTERM to remaining
processes...
[  306.963920] systemd-journald[1722]: Received SIGTERM from PID 1
(systemd-shutdow).
[  307.322717] systemd-shutdown[1]: Sending SIGKILL to remaining
processes...
[  307.336472] systemd-shutdown[1]: Unmounting file systems.
[  307.454202] EXT4-fs (sda2): re-mounted. Opts: errors=remount-ro
[  307.480523] systemd-shutdown[1]: All filesystems unmounted.
[  307.486537] systemd-shutdown[1]: Deactivating swaps.
[  307.491962] systemd-shutdown[1]: All swaps deactivated.
[  307.497624] systemd-shutdown[1]: Detaching loop devices.
[  307.504418] systemd-shutdown[1]: All loop devices detached.
[  307.510418] systemd-shutdown[1]: Detaching DM devices.
[  307.565907] sd 2:0:0:0: [sda] Synchronizing SCSI cache
[  307.731313] BUG: kernel NULL pointer dereference, address:
0000000000000000
[  307.738802] #PF: supervisor read access in kernel mode
[  307.744326] #PF: error_code(0x0000) - not-present page
[  307.749850] PGD 0 P4D 0
[  307.752568] Oops: 0000 [#1] SMP PTI
[  307.756314] CPU: 3 PID: 1 Comm: systemd-shutdow Not tainted
5.2.0-rc1-kfd-yangp #453
[  307.764644] Hardware name: ASUS All Series/Z97-PRO(Wi-Fi ac)/USB 3.1,
BIOS 9001 03/07/2016
[  307.773580] RIP: 0010:soc15_common_hw_fini+0x33/0xc0 [amdgpu]
[  307.779760] Code: 89 fb e8 60 f5 ff ff f6 83 50 df 01 00 04 75 3d 48
8b b3 90 7d 00 00 48 c7 c7 17 b8 530
[  307.799967] RSP: 0018:ffffac9483153d40 EFLAGS: 00010286
[  307.805585] RAX: 0000000000000000 RBX: ffff9eb299da0000 RCX:
0000000000000006
[  307.813261] RDX: 0000000000000000 RSI: ffff9eb29e3508a0 RDI:
ffff9eb29e350000
[  307.820935] RBP: ffff9eb299da0000 R08: 0000000000000000 R09:
0000000000000000
[  307.828609] R10: 0000000000000000 R11: 0000000000000000 R12:
ffff9eb299dbd1f8
[  307.836284] R13: ffffffffc04f8368 R14: ffff9eb29cebd130 R15:
0000000000000000
[  307.843959] FS:  00007f06721c9940(0000) GS:ffff9eb2a18c0000(0000)
knlGS:0000000000000000
[  307.852663] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  307.858842] CR2: 0000000000000000 CR3: 000000081d798005 CR4:
00000000001606e0
[  307.866516] Call Trace:
[  307.869169]  amdgpu_device_ip_suspend_phase2+0x80/0x110 [amdgpu]
[  307.875654]  ? amdgpu_device_ip_suspend_phase1+0x4d/0xd0 [amdgpu]
[  307.882230]  amdgpu_device_ip_suspend+0x2e/0x60 [amdgpu]
[  307.887966]  amdgpu_pci_shutdown+0x2f/0x40 [amdgpu]
[  307.893211]  pci_device_shutdown+0x31/0x60
[  307.897613]  device_shutdown+0x14c/0x1f0
[  307.901829]  kernel_restart+0xe/0x50
[  307.905669]  __do_sys_reboot+0x1df/0x210
[  307.909884]  ? task_work_run+0x73/0xb0
[  307.913914]  ? trace_hardirqs_off_thunk+0x1a/0x1c
[  307.918970]  do_syscall_64+0x4a/0x1c0
[  307.922904]  entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  307.928336] RIP: 0033:0x7f0671cf8373
[  307.932176] Code: 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00
00 0f 1f 44 00 00 89 fa be 69 19 128
[  307.952384] RSP: 002b:00007ffdd1723d68 EFLAGS: 00000202 ORIG_RAX:
00000000000000a9
[  307.960527] RAX: ffffffffffffffda RBX: 0000000001234567 RCX:
00007f0671cf8373
[  307.968201] RDX: 0000000001234567 RSI: 0000000028121969 RDI:
00000000fee1dead
[  307.975875] RBP: 00007ffdd1723dd0 R08: 0000000000000000 R09:
0000000000000000
[  307.983550] R10: 0000000000000002 R11: 0000000000000202 R12:
00007ffdd1723dd8
[  307.991224] R13: 0000000000000000 R14: 0000001b00000004 R15:
00007ffdd17240c8
[  307.998901] Modules linked in: xt_MASQUERADE nfnetlink iptable_nat
xt_addrtype xt_conntrack nf_nat nf_cos
[  308.026505] CR2: 0000000000000000
[  308.039998] RIP: 0010:soc15_common_hw_fini+0x33/0xc0 [amdgpu]
[  308.046180] Code: 89 fb e8 60 f5 ff ff f6 83 50 df 01 00 04 75 3d 48
8b b3 90 7d 00 00 48 c7 c7 17 b8 530
[  308.066392] RSP: 0018:ffffac9483153d40 EFLAGS: 00010286
[  308.072013] RAX: 0000000000000000 RBX: ffff9eb299da0000 RCX:
0000000000000006
[  308.079689] RDX: 0000000000000000 RSI: ffff9eb29e3508a0 RDI:
ffff9eb29e350000
[  308.087366] RBP: ffff9eb299da0000 R08: 0000000000000000 R09:
0000000000000000
[  308.095042] R10: 0000000000000000 R11: 0000000000000000 R12:
ffff9eb299dbd1f8
[  308.102717] R13: ffffffffc04f8368 R14: ffff9eb29cebd130 R15:
0000000000000000
[  308.110394] FS:  00007f06721c9940(0000) GS:ffff9eb2a18c0000(0000)
knlGS:0000000000000000
[  308.119099] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  308.125280] CR2: 0000000000000000 CR3: 000000081d798005 CR4:
00000000001606e0
[  308.135304] printk: systemd-shutdow: 3 output lines suppressed due to
ratelimiting
[  308.143518] Kernel panic - not syncing: Attempted to kill init!
exitcode=0x00000009
[  308.151798] Kernel Offset: 0x15000000 from 0xffffffff81000000
(relocation range: 0xffffffff80000000-0xff)
[  308.171775] ---[ end Kernel panic - not syncing: Attempted to kill
init! exitcode=0x00000009 ]---

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: fix null pointer deref in firmware header printing

v2: declare as (struct common_firmware_header *) type because
    struct xxx_firmware_header inherits from it

When CE's ucode_id(8) is used to get sdma_hdr, we will be accessing an
unallocated amdgpu_firmware_info instance.

This issue appears on rhel7.7 with gcc 4.8.5. Newer compilers might have
optimized out such 'defined but not referenced' variable.

[ 1120.798564] BUG: unable to handle kernel NULL pointer dereference at 000000000000000a
[ 1120.806703] IP: [<ffffffffc0e3c9b3>] psp_np_fw_load+0x1e3/0x390 [amdgpu]
[ 1120.813693] PGD 80000002603ff067 PUD 271b8d067 PMD 0
[ 1120.818931] Oops: 0000 [#1] SMP
[ 1120.822245] Modules linked in: amdgpu(OE+) amdkcl(OE) amd_iommu_v2 amdttm(OE) amd_sched(OE) xt_CHECKSUM ipt_MASQUERADE nf_nat_masquerade_ipv4 tun bridge stp llc devlink ip6t_rpfilter ip6t_REJECT nf_reject_ipv6 ipt_REJECT nf_reject_ipv4 xt_conntrack ebtable_nat ip6table_nat nf_conntrack_ipv6 nf_defrag_ipv6 nf_nat_ipv6 ip6table_mangle ip6table_security ip6table_raw iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat iptable_mangle iptable_security iptable_raw nf_conntrack libcrc32c ip_set nfnetlink ebtable_filter ebtables ip6table_filter ip6_tables iptable_filter sunrpc dm_mirror dm_region_hash dm_log dm_mod intel_pmc_core intel_powerclamp coretemp intel_rapl joydev kvm_intel eeepc_wmi asus_wmi kvm sparse_keymap iTCO_wdt irqbypass rfkill crc32_pclmul snd_hda_codec_realtek mxm_wmi ghash_clmulni_intel intel_wmi_thunderbolt iTCO_vendor_support snd_hda_codec_generic snd_hda_codec_hdmi aesni_intel lrw gf128mul glue_helper ablk_helper sg cryptd pcspkr snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd pinctrl_sunrisepoint pinctrl_intel soundcore acpi_pad mei_me wmi mei i2c_i801 pcc_cpufreq ip_tables ext4 mbcache jbd2 sd_mod crc_t10dif crct10dif_generic i915 i2c_algo_bit iosf_mbi drm_kms_helper e1000e syscopyarea sysfillrect sysimgblt fb_sys_fops ahci libahci drm ptp libata crct10dif_pclmul crct10dif_common crc32c_intel serio_raw pps_core drm_panel_orientation_quirks video i2c_hid
[ 1120.954136] CPU: 4 PID: 2426 Comm: modprobe Tainted: G           OE  ------------   3.10.0-1062.el7.x86_64 #1
[ 1120.964390] Hardware name: System manufacturer System Product Name/Z170-A, BIOS 1302 11/09/2015
[ 1120.973321] task: ffff991ef1e3c1c0 ti: ffff991ee625c000 task.ti: ffff991ee625c000
[ 1120.981020] RIP: 0010:[<ffffffffc0e3c9b3>]  [<ffffffffc0e3c9b3>] psp_np_fw_load+0x1e3/0x390 [amdgpu]
[ 1120.990483] RSP: 0018:ffff991ee625f950  EFLAGS: 00010202
[ 1120.995935] RAX: 0000000000000002 RBX: ffff991edf6b2d38 RCX: ffff991edf6a0000
[ 1121.003391] RDX: 0000000000000000 RSI: ffff991f01d13898 RDI: ffffffffc110afb3
[ 1121.010706] RBP: ffff991ee625f9b0 R08: 0000000000000000 R09: 0000000000000000
[ 1121.018029] R10: 00000000000004c4 R11: ffff991ee625f64e R12: ffff991edf6b3220
[ 1121.025353] R13: ffff991edf6a0000 R14: 0000000000000008 R15: ffff991edf6b2d30
[ 1121.032666] FS:  00007f97b0c0b740(0000) GS:ffff991f01d00000(0000) knlGS:0000000000000000
[ 1121.041000] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 1121.046880] CR2: 000000000000000a CR3: 000000025e604000 CR4: 00000000003607e0
[ 1121.054239] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 1121.061631] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 1121.068938] Call Trace:
[ 1121.071494]  [<ffffffffc0e3dba8>] psp_hw_init+0x218/0x270 [amdgpu]
[ 1121.077886]  [<ffffffffc0da3188>] amdgpu_device_fw_loading+0xe8/0x160 [amdgpu]
[ 1121.085296]  [<ffffffffc0e3b34c>] ? vega10_ih_irq_init+0x4bc/0x730 [amdgpu]
[ 1121.092534]  [<ffffffffc0da5c75>] amdgpu_device_init+0x1495/0x1c90 [amdgpu]
[ 1121.099675]  [<ffffffffc0da9cab>] amdgpu_driver_load_kms+0x8b/0x2f0 [amdgpu]
[ 1121.106888]  [<ffffffffc01b25cf>] drm_dev_register+0x12f/0x1d0 [drm]
[ 1121.113419]  [<ffffffffa4dcdfd8>] ? pci_enable_device_flags+0xe8/0x140
[ 1121.120183]  [<ffffffffc0da260a>] amdgpu_pci_probe+0xca/0x170 [amdgpu]
[ 1121.126919]  [<ffffffffa4dcf97a>] local_pci_probe+0x4a/0xb0
[ 1121.132622]  [<ffffffffa4dd10c9>] pci_device_probe+0x109/0x160
[ 1121.138607]  [<ffffffffa4eb4205>] driver_probe_device+0xc5/0x3e0
[ 1121.144766]  [<ffffffffa4eb4603>] __driver_attach+0x93/0xa0
[ 1121.150507]  [<ffffffffa4eb4570>] ? __device_attach+0x50/0x50
[ 1121.156422]  [<ffffffffa4eb1da5>] bus_for_each_dev+0x75/0xc0
[ 1121.162213]  [<ffffffffa4eb3b7e>] driver_attach+0x1e/0x20
[ 1121.167771]  [<ffffffffa4eb3620>] bus_add_driver+0x200/0x2d0
[ 1121.173590]  [<ffffffffa4eb4c94>] driver_register+0x64/0xf0
[ 1121.179345]  [<ffffffffa4dd0905>] __pci_register_driver+0xa5/0xc0
[ 1121.185593]  [<ffffffffc099f000>] ? 0xffffffffc099efff
[ 1121.190914]  [<ffffffffc099f0a4>] amdgpu_init+0xa4/0xb0 [amdgpu]
[ 1121.197101]  [<ffffffffa4a0210a>] do_one_initcall+0xba/0x240
[ 1121.202901]  [<ffffffffa4b1c90a>] load_module+0x271a/0x2bb0
[ 1121.208598]  [<ffffffffa4dad740>] ? ddebug_proc_write+0x100/0x100
[ 1121.214894]  [<ffffffffa4b1ce8f>] SyS_init_module+0xef/0x140
[ 1121.220698]  [<ffffffffa518bede>] system_call_fastpath+0x25/0x2a
[ 1121.226870] Code: b4 01 60 a2 00 00 31 c0 e8 83 60 33 e4 41 8b 47 08 48 8b 4d d0 48 c7 c7 b3 af 10 c1 48 69 c0 68 07 00 00 48 8b 84 01 60 a2 00 00 <48> 8b 70 08 31 c0 48 89 75 c8 e8 56 60 33 e4 48 8b 4d d0 48 c7
[ 1121.247422] RIP  [<ffffffffc0e3c9b3>] psp_np_fw_load+0x1e3/0x390 [amdgpu]
[ 1121.254432]  RSP <ffff991ee625f950>
[ 1121.258017] CR2: 000000000000000a
[ 1121.261427] ---[ end trace e98b35387ede75bd ]---

Signed-off-by: Xiaojie Yuan <xiaojie.yuan@amd.com>
Fixes: c5fb912653dae3f878 ("drm/amdgpu: add firmware header printing for psp fw loading (v2)")
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: enable renoir while device probes

This patch is to add asic flag to enable device probe during kfd init.

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>

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>

drm/amdkfd: add renoir kfd topology

This patch adds renoir kfd topology which is the same with Raven.

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>

drm/amdkfd: add package manager for renoir

Renoir use GFX v9, so adds v9 package manager.

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>

drm/amdkfd: init kernel queue for renoir

Renoir is GFX v9, so init v9 kernel queue.

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>

drm/amdkfd: init kfd apertures v9 for renoir

Renoir is GMC v9, so init v9 kfd apertures.

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>

drm/amdkfd: add renoir type for the workaround of iommu v2 (v2)

Renoir is the same with Raven, will enable iommu event in future.

v2: fix the checking (Thong)

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>

drm/amdkfd: enable kfd device queue manager v9 for renoir

Renoir is GFX9, so enable v9 devcie queue manager.

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>

drm/amdkfd: add renoir kfd device info (v2)

This patch inits renoir kfd device info, so we treat renoir as "dgpu"
(bypass iommu v2). Will enable needs_iommu_device till renoir iommu is ready.

v2: rebase and align the drm-next

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>

drm/amdkfd: add renoir cache info for CRAT (v2)

Renoir's cache info should be the same with raven and carrizo's.

v2: fix missed "break"

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>

drm/amdkfd: Support Navi14 in KFD

Initial support of Navi14 in KFD. The device IDs will be added later.

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: Disable retry faults in VMID0

There is no point retrying page faults in VMID0. Those faults are
always fatal.

Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-and-Tested-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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>

drm/amdgpu/irq: check if nbio funcs exist

We need to check if the nbios funcs exist before
checking the individual pointers.

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

drm/amd/display: replace FIXME with TODO

Signed-off-by: Qingqing Zhuo <qingqing.zhuo@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>

drm/amd/display: verify stream link before link test

[Why]
DP1.2 LL CTS test failure.

[How]
The failure is caused by not verify stream link is equal
to link, only check stream and link is not null.

Signed-off-by: Jing Zhou <Jing.Zhou@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: dce11.x /dce12 update formula input

[Description]
1. OUTSTANDING_REQUEST_LIMIT update from 0xFF to 0x1F (HW doc update)
2. using memory type to convert UMC's MCLK to Yclk.

Signed-off-by: Charlene Liu <charlene.liu@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: 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>

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>

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>

drm/amd/display: Reuse dcn2 registers

[Why & How]
Use dcn2 blender, shaper, 3dlut registers

Signed-off-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Acked-by: Vitaly Prosyak <Vitaly.Prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove temporary transition code

Remove code used to allow compilation error free
interface change.

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@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>

drm/amd/display: update navi to use new surface programming behaviour

New behaviour will track global updates and update any hw that isn't
related to current stream being updated.

This should fix any issues caused by pipe split pipes being taken
by other streams.

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@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>

drm/amd/display: Add missing surface address registers

[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>

drm/amd/display: 3.2.49

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

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>

drm/amd/display: Add back support for DSC 4:2:2 Simple

[why]
The requirement has been clarified and only DSC 4:2:2 Native has to
be disabled.

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

drm/amd/display: Don't allocate payloads if link lost

We should not allocate payloads if the link is lost until the link is retrained.
Some displays require this.

Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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>

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>

drm/amd/display: Fix DML tests

[Why]
DML diags tests are failing because the struct contents get
clobbered by a memcpy.

[How]
Remove the memcpy call.

Signed-off-by: Ilya Bakoulin <Ilya.Bakoulin@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/amdkfd: Fix a building error when KFD_SUPPORT_IOMMU_V2 is turned off

The issue was accidentally introduced recently.

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: 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>

drm/amdgpu: save umc error records

save umc error records to ras bad page array

v2: add bad pages before gpu reset
v3: add NULL check for adev->umc.funcs

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

drm/amdgpu: Hook EEPROM table to RAS

support eeprom records load and save for ras,
move EEPROM records storing to bad page reserving

v2: remove redundant check for con->eh_data

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

drm/amdgpu: change ras bps type to eeprom table record structure

change bps type from retired page to eeprom table record, prepare for
saving umc error records to eeprom

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/madgpu: Fix EEPROM Checksum calculation.

Fix typo which messed up the calculation.

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

drm/amdgpu: Remove clock gating restore.

Restoring clock gating break SMU opeartion afterwards, avoid
this until this further invistigated with SMU.

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

drm/amdgpu: Add smu lock around in pp_smu_i2c_bus_access

Protect from concurrent SMU accesses.

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

drm/amd/display: Add stereo mux and dig programming calls for dcn21

[Why]
The earlier patch "Hook up calls to do stereo mux and dig programming..."
doesn't include update for dcn21.

[How]
Align dcn21 gpio settings with updated stereo control interface.

Signed-off-by: Roman Li <Roman.Li@amd.com>
Acked-by: Aaron Liu <aaron.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Query kfd device info by CHIP id instead of pci device id

This optimizes out the pci device id usage in KFD and makes the code
more maintainable.

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: 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>

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>

drm/amd/display: update renoir_ip_offset.h

This patch updates MP1_BASE in renoir_ip_offset.h

Signed-off-by: Aaron Liu <aaron.liu@amd.com>
Acked-by: Roman Li <roman.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: implement sysfs for getting dpm clock

With the common interface print_clk_levels can get the following dpm clock:

-pp_dpm_dcefclk
-pp_dpm_fclk
-pp_dpm_mclk
-pp_dpm_sclk
-pp_dpm_socclk

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>

drm/amdgpu: clean up load TMR sequence

Removed redundant goto statement

Signed-off-by: John Clements <john.clements@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: enable TA load support in Arcturus

Add support for loading XGMI/RAS FW

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

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>

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>

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: init UMC & RSMU register base address

UMC RAS feature requires access to UMC & RSMU registers

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/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/gfx: switch to amdgpu_gfx_ras_late_init helper function

amdgpu_gfx_ras_late_init is used to init gfx specfic
ras debugfs/sysfs node and gfx specific interrupt handler.
It can be shared among gfx 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>

drm/amdgpu: set ip specific ras interface pointer to NULL after free it

to prevent access to dangling pointers

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>

dmr/amdgpu: Add system auto reboot to RAS.

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.

v4: Use latest kernel API for disk sync.

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

drm/amdgpu: Avoid HW GPU reset for RAS.

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>