Freeing memory was warned during suspend.
Move the self test out of suspend.

Link: https://bugzilla.redhat.com/show_bug.cgi?id=2151825
Cc: jfalempe@redhat.com
Signed-off-by: NJack Xiao <Jack.Xiao@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NFeifei Xu <Feifei.Xu@amd.com>
Reviewed-and-tested-by: NEvan Quan <evan.quan@amd.com>
Tested-by: NJocelyn Falempe <jfalempe@redhat.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.1.x

Some users have reported flickerng with S/G display.  We've
tried extensively to reproduce and debug the issue on a wide
variety of platform configurations (DRAM bandwidth, etc.) and
a variety of monitors, but so far have not been able to.  We
disabled S/G display on a number of platforms to address this
but that leads to failure to pin framebuffers errors and
blank displays when there is memory pressure or no displays
at all on systems with limited carveout (e.g., Chromebooks).
Add a option to disable this as a debugging option as a
way for users to disable this, depending on their use case,
and for us to help debug this further.

v2: fix typo
Reviewed-by: NHarry Wentland <harry.wentland@amd.com>
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

test ib function is not necessary on hw ring,
so remove it.

v2: squash in NULL check fix
Signed-off-by: NJesseZhang <Jesse.Zhang@amd.com>
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Currently amdgpu calls drm_sched_fini() from the fence driver sw fini
routine - such function is expected to be called only after the
respective init function - drm_sched_init() - was executed successfully.

Happens that we faced a driver probe failure in the Steam Deck
recently, and the function drm_sched_fini() was called even without
its counter-part had been previously called, causing the following oops:

amdgpu: probe of 0000:04:00.0 failed with error -110
BUG: kernel NULL pointer dereference, address: 0000000000000090
PGD 0 P4D 0
Oops: 0002 [#1] PREEMPT SMP NOPTI
CPU: 0 PID: 609 Comm: systemd-udevd Not tainted 6.2.0-rc3-gpiccoli #338
Hardware name: Valve Jupiter/Jupiter, BIOS F7A0113 11/04/2022
RIP: 0010:drm_sched_fini+0x84/0xa0 [gpu_sched]
[...]
Call Trace:
 <TASK>
 amdgpu_fence_driver_sw_fini+0xc8/0xd0 [amdgpu]
 amdgpu_device_fini_sw+0x2b/0x3b0 [amdgpu]
 amdgpu_driver_release_kms+0x16/0x30 [amdgpu]
 devm_drm_dev_init_release+0x49/0x70
 [...]

To prevent that, check if the drm_sched was properly initialized for a
given ring before calling its fini counter-part.

Notice ideally we'd use sched.ready for that; such field is set as the latest
thing on drm_sched_init(). But amdgpu seems to "override" the meaning of such
field - in the above oops for example, it was a GFX ring causing the crash, and
the sched.ready field was set to true in the ring init routine, regardless of
the state of the DRM scheduler. Hence, we ended-up using sched.ops as per
Christian's suggestion [0], and also removed the no_scheduler check [1].

[0] https://lore.kernel.org/amd-gfx/984ee981-2906-0eaf-ccec-9f80975cb136@amd.com/
[1] https://lore.kernel.org/amd-gfx/cd0e2994-f85f-d837-609f-7056d5fb7231@amd.com/

Fixes: 067f44c8 ("drm/amdgpu: avoid over-handle of fence driver fini in s3 test (v2)")
Suggested-by: NChristian König <christian.koenig@amd.com>
Cc: Guchun Chen <guchun.chen@amd.com>
Cc: Luben Tuikov <luben.tuikov@amd.com>
Cc: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: NLuben Tuikov <luben.tuikov@amd.com>
Signed-off-by: NGuilherme G. Piccoli <gpiccoli@igalia.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

The pid field corresponds to the result of gettid() in userspace.
However, userspace cannot reliably attribute PTE events to processes
with just the thread id. This patch allows userspace to easily
attribute PTE update events to specific processes by comparing this
field with the result of getpid().

For attributing events to specific threads, the thread id is also
contained in the common fields of each trace event.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NFriedrich Vock <friedrich.vock@gmx.de>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

enable athub cg on gc 11.0.3
Signed-off-by: NKenneth Feng <kenneth.feng@amd.com>
Reviewed-by: NLikun Gao <Likun.Gao@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

amdgpu_sync_get_fence deletes the returned fence from the
syncobj, so the refcount of fence needs to lowered to avoid
a memory leak.

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2360Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Tested-by: NMikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NBert Karwatzki <spasswolf@web.de>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/3b590ba0f11d24b8c6c39c3d38250129c1116af4.camel@web.de

A mistake has been made on some boards with NBIO 4.3.0 where some
NBIO registers aren't properly set by the hardware.

Ensure that they're set during initialization.

Cc: Natikar Basavaraj <Basavaraj.Natikar@amd.com>
Tested-by: NSatyanarayana ReddyTVN <Satyanarayana.ReddyTVN@amd.com>
Tested-by: NRutvij Gajjar <Rutvij.Gajjar@amd.com>
Signed-off-by: NMario Limonciello <mario.limonciello@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.1.x

Enable HDP clock gating control for gfx 11.0.3.
Signed-off-by: NEvan Quan <evan.quan@amd.com>
Reviewed-by: NFeifei Xu <Feifei.Xu@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

SQ_WAVE_INST_DW0 isn't present on gfx11 compared to gfx10, so update
wave data type to signify a difference.
Signed-off-by: NGraham Sider <Graham.Sider@amd.com>
Reviewed-by: NMukul Joshi <Mukul.Joshi@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.1.x

To support new mes ip block
Signed-off-by: NLi Ma <li.ma@amd.com>
Reviewed-by: NYifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

The GC 11.0.4 needs load IMU to power up GFX before loads GFX firmware.
Signed-off-by: NLi Ma <li.ma@amd.com>
Reviewed-by: NYifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Rebase of driver has incorrect unconditional trap enablement
for GFX11 when adding mes queues.
Reported-by: NGraham Sider <graham.sider@amd.com>
Signed-off-by: NJonathan Kim <jonathan.kim@amd.com>
Reviewed-by: NGraham Sider <graham.sider@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.1.x

Always enable multipipe policy on ASICs with GC VERSION > 9.0.0
instead of MEC number > 1.

This will allow multipipe policy on ASICs with one MEC,
e.g., gfx11 APUs.
Signed-off-by: NLang Yu <Lang.Yu@amd.com>
Reviewed-by: NAaron Liu <aaron.liu@amd.com>
Reviewed-by: NYifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.1.x

There is only one MEC on these APUs.
Signed-off-by: NLang Yu <Lang.Yu@amd.com>
Reviewed-by: NAaron Liu <aaron.liu@amd.com>
Reviewed-by: NYifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.1.x

It can be that neither fence were initialized when we run out of UVD
streams for example.

v2: fix typo breaking compile

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2324Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.1.x

We need to reset this or otherwise run into list corruption later on.

Fixes: e44a0fe6 ("drm/amdgpu: rework reserved VMID handling")
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Tested-by: NCandice Li <candice.li@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

The point bo->kfd_bo is NULL for queue's write pointer BO
when creating queue on mGPU. To avoid using the pointer
fixes the error.
Signed-off-by: NEric Huang <jinhuieric.huang@amd.com>
Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This fixes a potential memory leak of dma_fence objects in the CS code
as well as glitches in firefox because of missing pipeline sync.

v2: use the scheduler instead of the fence context
Signed-off-by: NChristian König <christian.koenig@amd.com>
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2323
Tested-by: Michal Kubecek mkubecek@suse.cz
Tested-by: NVlastimil Babka <vbabka@suse.cz>
Acked-by: NAlex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230109130120.73389-1-christian.koenig@amd.com

Fixed bug on error when unloading amdgpu.

The error message is as follows:
[  377.706202] kernel BUG at drivers/gpu/drm/drm_buddy.c:278!
[  377.706215] invalid opcode: 0000 [#1] PREEMPT SMP NOPTI
[  377.706222] CPU: 4 PID: 8610 Comm: modprobe Tainted: G          IOE      6.0.0-thomas #1
[  377.706231] Hardware name: ASUS System Product Name/PRIME Z390-A, BIOS 2004 11/02/2021
[  377.706238] RIP: 0010:drm_buddy_free_block+0x26/0x30 [drm_buddy]
[  377.706264] Code: 00 00 00 90 0f 1f 44 00 00 48 8b 0e 89 c8 25 00 0c 00 00 3d 00 04 00 00 75 10 48 8b 47 18 48 d3 e0 48 01 47 28 e9 fa fe ff ff <0f> 0b 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 41 54 55 48 89 f5 53
[  377.706282] RSP: 0018:ffffad2dc4683cb8 EFLAGS: 00010287
[  377.706289] RAX: 0000000000000000 RBX: ffff8b1743bd5138 RCX: 0000000000000000
[  377.706297] RDX: ffff8b1743bd5160 RSI: ffff8b1743bd5c78 RDI: ffff8b16d1b25f70
[  377.706304] RBP: ffff8b1743bd59e0 R08: 0000000000000001 R09: 0000000000000001
[  377.706311] R10: ffff8b16c8572400 R11: ffffad2dc4683cf0 R12: ffff8b16d1b25f70
[  377.706318] R13: ffff8b16d1b25fd0 R14: ffff8b1743bd59c0 R15: ffff8b16d1b25f70
[  377.706325] FS:  00007fec56c72c40(0000) GS:ffff8b1836500000(0000) knlGS:0000000000000000
[  377.706334] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  377.706340] CR2: 00007f9b88c1ba50 CR3: 0000000110450004 CR4: 00000000003706e0
[  377.706347] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  377.706354] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[  377.706361] Call Trace:
[  377.706365]  <TASK>
[  377.706369]  drm_buddy_free_list+0x2a/0x60 [drm_buddy]
[  377.706376]  amdgpu_vram_mgr_fini+0xea/0x180 [amdgpu]
[  377.706572]  amdgpu_ttm_fini+0x12e/0x1a0 [amdgpu]
[  377.706650]  amdgpu_bo_fini+0x22/0x90 [amdgpu]
[  377.706727]  gmc_v11_0_sw_fini+0x26/0x30 [amdgpu]
[  377.706821]  amdgpu_device_fini_sw+0xa1/0x3c0 [amdgpu]
[  377.706897]  amdgpu_driver_release_kms+0x12/0x30 [amdgpu]
[  377.706975]  drm_dev_release+0x20/0x40 [drm]
[  377.707006]  release_nodes+0x35/0xb0
[  377.707014]  devres_release_all+0x8b/0xc0
[  377.707020]  device_unbind_cleanup+0xe/0x70
[  377.707027]  device_release_driver_internal+0xee/0x160
[  377.707033]  driver_detach+0x44/0x90
[  377.707039]  bus_remove_driver+0x55/0xe0
[  377.707045]  pci_unregister_driver+0x3b/0x90
[  377.707052]  amdgpu_exit+0x11/0x6c [amdgpu]
[  377.707194]  __x64_sys_delete_module+0x142/0x2b0
[  377.707201]  ? fpregs_assert_state_consistent+0x22/0x50
[  377.707208]  ? exit_to_user_mode_prepare+0x3e/0x190
[  377.707215]  do_syscall_64+0x38/0x90
[  377.707221]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
Signed-off-by: NYiPeng Chai <YiPeng.Chai@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

Removing the firmware framebuffer from the driver means that even
if the driver doesn't support the IP blocks in a GPU it will no
longer be functional after the driver fails to initialize.

This change will ensure that unsupported IP blocks at least cause
the driver to work with the EFI framebuffer.

Cc: stable@vger.kernel.org
Suggested-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NLijo Lazar <lijo.lazar@amd.com>
Signed-off-by: NMario Limonciello <mario.limonciello@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Fix potential NULL dereference, in the case when "man", the resource manager
might be NULL, when/if we print debug information.

Cc: Alex Deucher <Alexander.Deucher@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: AMD Graphics <amd-gfx@lists.freedesktop.org>
Cc: Dan Carpenter <error27@gmail.com>
Cc: kernel test robot <lkp@intel.com>
Fixes: 7554886d ("drm/amdgpu: Fix size validation for non-exclusive domains (v4)")
Signed-off-by: NLuben Tuikov <luben.tuikov@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

When the fence can't be added we need to drop the reference.
Suggested-by: NBert Karwatzki <spasswolf@web.de>
Signed-off-by: NChristian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230105111703.52695-2-christian.koenig@amd.comReviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NLuben Tuikov <luben.tuikov@amd.com>

drm_sched_job_add_dependency() consumes the references of the gang
members. Only triggered by mesh shaders.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Fixes: 1728baa7 ("drm/amdgpu: use scheduler dependencies for CS")
Tested-by: NMike Lothian <mike@fireburn.co.uk>
Tested-by: NBert Karwatzki <spasswolf@web.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230105111703.52695-1-christian.koenig@amd.comReviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NLuben Tuikov <luben.tuikov@amd.com>

This reverts commit de05abe6.

The bug referenced below was bisected to this commit. There has been no
activity toward fixing it in 3 months, so let's revert for now.

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2162Signed-off-by: NMichel Dänzer <mdaenzer@redhat.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

That function consumes the reference.
Reviewed-by: NLuben Tuikov <luben.tuikov@amd.com>
Reported-by: NBorislav Petkov (AMD) <bp@alien8.de>
Tested-by: NBorislav Petkov (AMD) <bp@alien8.de>
Signed-off-by: NChristian König <christian.koenig@amd.com>
Fixes: aab9cf7b ("drm/amdgpu: use scheduler dependencies for VM updates")
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Enable VCN Dynamic Power Gating control for GC IP v11.0.4.
Signed-off-by: NSaleemkhan Jamadar <saleemkhan.jamadar@amd.com>
Reviewed-by: NVeerabadhran Gopalakrishnan <veerabadhran.gopalakrishnan@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.0, 6.1

MES is part of gfxoff and MES suspend and resume are skipped for S0i3.
But the mes_self_test call path is still in the amdgpu_device_ip_late_init.
it's should also be skipped for s0ix as no hardware re-initialization
happened.

Besides, mes_self_test will free the BO that triggers a lot of warning
messages while in the suspend state.

[   81.656085] WARNING: CPU: 2 PID: 1550 at drivers/gpu/drm/amd/amdgpu/amdgpu_object.c:425 amdgpu_bo_free_kernel+0xfc/0x110 [amdgpu]
[   81.679435] Call Trace:
[   81.679726]  <TASK>
[   81.679981]  amdgpu_mes_remove_hw_queue+0x17a/0x230 [amdgpu]
[   81.680857]  amdgpu_mes_self_test+0x390/0x430 [amdgpu]
[   81.681665]  mes_v11_0_late_init+0x37/0x50 [amdgpu]
[   81.682423]  amdgpu_device_ip_late_init+0x53/0x280 [amdgpu]
[   81.683257]  amdgpu_device_resume+0xae/0x2a0 [amdgpu]
[   81.684043]  amdgpu_pmops_resume+0x37/0x70 [amdgpu]
[   81.684818]  pci_pm_resume+0x5c/0xa0
[   81.685247]  ? pci_pm_thaw+0x90/0x90
[   81.685658]  dpm_run_callback+0x4e/0x160
[   81.686110]  device_resume+0xad/0x210
[   81.686529]  async_resume+0x1e/0x40
[   81.686931]  async_run_entry_fn+0x33/0x120
[   81.687405]  process_one_work+0x21d/0x3f0
[   81.687869]  worker_thread+0x4a/0x3c0
[   81.688293]  ? process_one_work+0x3f0/0x3f0
[   81.688777]  kthread+0xff/0x130
[   81.689157]  ? kthread_complete_and_exit+0x20/0x20
[   81.689707]  ret_from_fork+0x22/0x30
[   81.690118]  </TASK>
[   81.690380] ---[ end trace 0000000000000000 ]---

v2: make the comment clean and use adev->in_s0ix instead of
adev->suspend
Signed-off-by: NTim Huang <tim.huang@amd.com>
Reviewed-by: NMario Limonciello <mario.limonciello@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.0, 6.1

It's also part of gfxoff.

Cc: stable@vger.kernel.org # 6.0, 6.1
Reviewed-by: NMario Limonciello <mario.limonciello@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

If kfd_process_device_init_vm returns failure after vm is converted to
compute vm and vm->pasid set to compute pasid, KFD will not take
pdd->drm_file reference. As a result, drm close file handler maybe
called to release the compute pasid before KFD process destroy worker to
release the same pasid and set vm->pasid to zero, this generates below
WARNING backtrace and NULL pointer access.

Add helper amdgpu_amdkfd_gpuvm_set_vm_pasid and call it at the last step
of kfd_process_device_init_vm, to ensure vm pasid is the original pasid
if acquiring vm failed or is the compute pasid with pdd->drm_file
reference taken to avoid double release same pasid.

 amdgpu: Failed to create process VM object
 ida_free called for id=32770 which is not allocated.
 WARNING: CPU: 57 PID: 72542 at ../lib/idr.c:522 ida_free+0x96/0x140
 RIP: 0010:ida_free+0x96/0x140
 Call Trace:
  amdgpu_pasid_free_delayed+0xe1/0x2a0 [amdgpu]
  amdgpu_driver_postclose_kms+0x2d8/0x340 [amdgpu]
  drm_file_free.part.13+0x216/0x270 [drm]
  drm_close_helper.isra.14+0x60/0x70 [drm]
  drm_release+0x6e/0xf0 [drm]
  __fput+0xcc/0x280
  ____fput+0xe/0x20
  task_work_run+0x96/0xc0
  do_exit+0x3d0/0xc10

 BUG: kernel NULL pointer dereference, address: 0000000000000000
 RIP: 0010:ida_free+0x76/0x140
 Call Trace:
  amdgpu_pasid_free_delayed+0xe1/0x2a0 [amdgpu]
  amdgpu_driver_postclose_kms+0x2d8/0x340 [amdgpu]
  drm_file_free.part.13+0x216/0x270 [drm]
  drm_close_helper.isra.14+0x60/0x70 [drm]
  drm_release+0x6e/0xf0 [drm]
  __fput+0xcc/0x280
  ____fput+0xe/0x20
  task_work_run+0x96/0xc0
  do_exit+0x3d0/0xc10
Signed-off-by: NPhilip Yang <Philip.Yang@amd.com>
Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This reverts commit f9d00a4a.

This causes problem for KFD because when we overcommit we accidentially
bind the BO to GTT for moving it into VRAM. We also need to make sure
that this is done only as fallback after trying to evict first.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Remove the "domain" argument to amdgpu_bo_create_kernel_at() since this
function takes an "offset" argument which is the offset off of VRAM, and as
such allocation always takes place in VRAM. Thus, the "domain" argument is
unnecessary.

Cc: Alex Deucher <Alexander.Deucher@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: AMD Graphics <amd-gfx@lists.freedesktop.org>
Signed-off-by: NLuben Tuikov <luben.tuikov@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Fix amdgpu_bo_validate_size() to check whether the TTM domain manager for the
requested memory exists, else we get a kernel oops when dereferencing "man".

v2: Make the patch standalone, i.e. not dependent on local patches.
v3: Preserve old behaviour and just check that the manager pointer is not
    NULL.
v4: Complain if GTT domain requested and it is uninitialized--most likely a
    bug.

Cc: Alex Deucher <Alexander.Deucher@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: AMD Graphics <amd-gfx@lists.freedesktop.org>
Signed-off-by: NLuben Tuikov <luben.tuikov@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Always check if fru_addr is not NULL. This commit also fixes a "smatch"
warning.

v2: Add a Fixes tag.

Cc: Alex Deucher <Alexander.Deucher@amd.com>
Cc: Dan Carpenter <error27@gmail.com>
Cc: kernel test robot <lkp@intel.com>
Cc: AMD Graphics <amd-gfx@lists.freedesktop.org>
Fixes: afbe5d1e ("drm/amdgpu: Bug-fix: Reading I2C FRU data on newer ASICs")
Signed-off-by: NLuben Tuikov <luben.tuikov@amd.com>
Reviewed-by: NKent Russell <kent.russell@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Instead of reserving a VMID for a single process allow that many
processes use the reserved ID. This allows for proper isolation
between the processes.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This is completely pointless since the VMID always stays allocated until
the VM is idle.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This should probably not access job->vm and also emit the SPM switch
under the conditional execute.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Bas pointed out that this isn't working as expected and could cause
crashes. Fix the handling by storing the marker that a switch is needed
inside the job instead.
Reported-by: NBas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Add a per-process MMU notifier lock for processing notifiers from
userptrs. Use that lock to properly synchronize page table updates with
MMU notifiers.
Signed-off-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Xiaogang Chen<Xiaogang.Chen@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

When buffers are freed during suspend there is no guarantee that
they can be re-allocated during resume.

The PSP subsystem seems to be quite buggy regarding this, so add
a WARN_ON() to point out those bugs.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NAlex Deucher <alexdeucher@amd.com>
Tested-by: NGuilherme G. Piccoli <gpiccoli@igalia.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>