Convert the collision-retry lock around hmm_range_fault to use the one now
provided by the mmu_interval notifier.

Although this driver does not seem to use the collision retry lock that
hmm provides correctly, it can still be converted over to use the
mmu_interval_notifier api instead of hmm_mirror without too much trouble.

This also deletes another place where a driver is associating additional
data (struct amdgpu_mn) with a mmu_struct.

Link: https://lore.kernel.org/r/20191112202231.3856-13-jgg@ziepe.caSigned-off-by: NPhilip Yang <Philip.Yang@amd.com>
Reviewed-by: NPhilip Yang <Philip.Yang@amd.com>
Tested-by: NPhilip Yang <Philip.Yang@amd.com>
Signed-off-by: NJason Gunthorpe <jgg@mellanox.com>

find_vma() must be called under the mmap_sem, reorganize this code to
do the vma check after entering the lock.

Further, fix the unlocked use of struct task_struct's mm, instead use
the mm from hmm_mirror which has an active mm_grab. Also the mm_grab
must be converted to a mm_get before acquiring mmap_sem or calling
find_vma().

Fixes: 66c45500 ("drm/amdgpu: use new HMM APIs and helpers")
Fixes: 0919195f ("drm/amdgpu: Enable amdgpu_ttm_tt_get_user_pages in worker threads")
Link: https://lore.kernel.org/r/20191112202231.3856-11-jgg@ziepe.caAcked-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NPhilip Yang <Philip.Yang@amd.com>
Tested-by: NPhilip Yang <Philip.Yang@amd.com>
Signed-off-by: NJason Gunthorpe <jgg@mellanox.com>

stolen memory should be fixed in visible region.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NTianci.Yin <tianci.yin@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

amdgpu uses a need_dma32 flag to indicate to the drm core that some
allocations need to be done using GFP_DMA32, but it only checks the
device addressing capabilities to make that decision.  Unfortunately
PCIe root ports that have limited addressing exist as well.  Use the
dma_addressing_limited instead to also take those into account.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Be more consistent with the naming of the other DMA-buf objects.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/323401/

All users pass PAGE_SIZE here, and if we wanted to support single entries
for huge pages we should really just add a HMM_FAULT_HUGEPAGE flag instead
that uses the huge page size instead of having the caller calculate that
size once, just for the hmm code to verify it.

Link: https://lore.kernel.org/r/20190806160554.14046-8-hch@lst.deSigned-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NJason Gunthorpe <jgg@mellanox.com>
Signed-off-by: NJason Gunthorpe <jgg@mellanox.com>

The start, end and page_shift values are all saved in the range structure,
so we might as well use that for argument passing.

Link: https://lore.kernel.org/r/20190806160554.14046-7-hch@lst.deSigned-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NJason Gunthorpe <jgg@mellanox.com>
Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: NJason Gunthorpe <jgg@mellanox.com>

hmm_range_fault can only return -EAGAIN if called with the
HMM_FAULT_ALLOW_RETRY flag, which amdgpu never does.  Remove the handling
for the -EAGAIN case with its non-standard locking scheme.

Link: https://lore.kernel.org/r/20190806160554.14046-2-hch@lst.deSigned-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NJason Gunthorpe <jgg@mellanox.com>
Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: NJason Gunthorpe <jgg@mellanox.com>

Signed-off-by: NGerd Hoffmann <kraxel@redhat.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20190805140119.7337-14-kraxel@redhat.com

Drop drm_gem_object from amdgpu_bo, use the
ttm_buffer_object.base instead.

Build tested only.
Signed-off-by: NGerd Hoffmann <kraxel@redhat.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20190805140119.7337-6-kraxel@redhat.com

Wipe VRAM memory containing sensitive data when moving or releasing
BOs. Clearing the memory is pipelined to minimize any impact on
subsequent memory allocation latency. Use of a poison value should
help debug future use-after-free bugs.

When moving BOs, the existing ttm_bo_pipelined_move ensures that the
memory won't be reused before being wiped.

When releasing BOs, the BO is fenced with the memory fill operation,
which results in queuing the BO for a delayed delete.

v2: Move amdgpu_amdkfd_unreserve_memory_limit into
amdgpu_bo_release_notify so that KFD can use memory that's still
being cleared in the background
Signed-off-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This allows easier expansion to other flags, and also makes the callers a
little easier to read.

Link: https://lore.kernel.org/r/20190726005650.2566-4-rcampbell@nvidia.comSigned-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NRalph Campbell <rcampbell@nvidia.com>
Reviewed-by: NJason Gunthorpe <jgg@mellanox.com>
Signed-off-by: NJason Gunthorpe <jgg@mellanox.com>

Under memory pressure, buffer moves between RAM to VRAM  can
fail when there is no GTT space available. In those cases
amdgpu_bo_move falls back to ttm_bo_move_memcpy, which seems to
succeed, although it doesn't really support non-contiguous or
invisible VRAM. This manifests as VM faults with corrupted page
table entries in KFD eviction stress tests.

Print some helpful messages when lack of GTT space is causing buffer
moves to fail. Check that source and destination memory regions are
supported by ttm_bo_move_memcpy before taking that fallback.
Signed-off-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

When starting a new mm_node, the page_offset becomes 0.
Signed-off-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

With mm_nodes larger than 4GB, byte_count in amdgpu_fill_buffer would
overflow.
Signed-off-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

An upcoming change in the hmm_range_register API requires passing in
a pointer to an hmm_mirror instead of mm_struct. To access the
hmm_mirror we need pass bo instead of ttm to amdgpu_ttm_tt_get_user_pages
because mirror is part of amdgpu_mn structure, which is accessible from bo.

v2: fix building without CONFIG_HMM_MIRROR (Arnd)
Signed-off-by: NPhilip Yang <Philip.Yang@amd.com>
Signed-off-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

To differentiate the mtypes across asics.
Signed-off-by: NHawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

HMM provides new APIs and helps in kernel 5.2-rc1 to simplify driver
path. The old hmm APIs are deprecated and will be removed in future.

Below are changes in driver:

1. Change hmm_vma_fault to hmm_range_register and hmm_range_fault which
supports range with multiple vmas, remove the multiple vmas handle path
and data structure.
2. Change hmm_vma_range_done to hmm_range_unregister.
3. Use default flags to avoid pre-fill pfn arrays.
4. Use new hmm_device_ helpers.
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>

Drop use of drmP.h in all files named amdgpu*
in drm/amd/amdgpu/

Fix fallout.
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>
Reviewed-by: NAlex 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>
Link: https://patchwork.freedesktop.org/patch/msgid/20190609220757.10862-10-sam@ravnborg.org

Fix so no files in drm/amd/ depends on the
deprecated drm_os_linux.h header file.

It was done manually:
- remove drm_os_linux.h from drmP.h
- fix all build errros
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>
Reviewed-by: NAlex 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>
Link: https://patchwork.freedesktop.org/patch/msgid/20190609220757.10862-3-sam@ravnborg.org

Use unsigned long for number of pages.

Check that pfns are valid after hmm_vma_fault. If they are not,
return an error instead of continuing with invalid page pointers and
PTEs.
Signed-off-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NPhilip Yang <Philip.Yang@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

userptr may cross two VMAs if the forked child process (not call exec
after fork) malloc buffer, then free it, and then malloc larger size
buf, kerenl will create new VMA adjacent to old VMA which was cloned
from parent process, some pages of userptr are in the first VMA, the
rest pages are in the second VMA.

HMM expects range only have one VMA, loop over all VMAs in the address
range, create multiple ranges to handle this case. See
is_mergeable_anon_vma in mm/mmap.c for details.
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>

Only select HMM_MIRROR will get kernel config dependency warnings
if CONFIG_HMM is missing in the config. Add depends on HMM will
solve the issue.

Add conditional compilation to fix compilation errors if HMM_MIRROR
is not enabled as HMM config is not enabled.

Remove unused function amdgpu_ttm_tt_mark_user_pages.
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>

Use HMM helper function hmm_vma_fault() to get physical pages backing
userptr and start CPU page table update track of those pages. Then use
hmm_vma_range_done() to check if those pages are updated before
amdgpu_cs_submit for gfx or before user queues are resumed for kfd.

If userptr pages are updated, for gfx, amdgpu_cs_ioctl will restart
from scratch, for kfd, restore worker is rescheduled to retry.

HMM simplify the CPU page table concurrent update check, so remove
guptasklock, mmu_invalidations, last_set_pages fields from
amdgpu_ttm_tt struct.

HMM does not pin the page (increase page ref count), so remove related
operations like release_pages(), put_page(), mark_page_dirty().
Signed-off-by: NPhilip Yang <Philip.Yang@amd.com>
Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

As far as we know this was never used by userspace and so should be removed.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NMarek Olšák <marek.olsak@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This reverts commit 915d3eec.

This depends on an HMM fix which is not upstream yet.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This reverts commit 6b8f7e3d.

This depends on an HMM fix which is not upstream yet.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This reverts commit 5aeaccca.

This depends on an HMM fix which is not upstream yet.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

userptr may cross two VMAs if the forked child process (not call exec
after fork) malloc buffer, then free it, and then malloc larger size
buf, kerenl will create new VMA adjacent to old VMA which was cloned
from parent process, some pages of userptr are in the first VMA, the
rest pages are in the second VMA.

HMM expects range only have one VMA, loop over all VMAs in the address
range, create multiple ranges to handle this case. See
is_mergeable_anon_vma in mm/mmap.c for details.
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>

Otherwise we won't be able to cleanly handle page faults.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NChunming Zhou <david1.zhou@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

A BO's address has to be at least the minimum offset. Sharing this
test in ttm_bo_mmap() removes code from drivers. A full buffer-address
validation is still done within drm_vma_offset_lockup_locked().
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Acked-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

The parameter file_page_offset is a constant shared by all drivers. Just
replace it with the constant itself.
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Acked-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Most TTM drivers define the constant DRM_FILE_PAGE_OFFSET of the same
value. The only exception is vboxvideo, which is being converted to the
new offset by this patch. Unifying the constants in a single place
simplifies the driver code.
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Acked-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Only select HMM_MIRROR will get kernel config dependency warnings
if CONFIG_HMM is missing in the config. Add depends on HMM will
solve the issue.

Add conditional compilation to fix compilation errors if HMM_MIRROR
is not enabled as HMM config is not enabled.

Remove unused function amdgpu_ttm_tt_mark_user_pages.
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>

Use HMM helper function hmm_vma_fault() to get physical pages backing
userptr and start CPU page table update track of those pages. Then use
hmm_vma_range_done() to check if those pages are updated before
amdgpu_cs_submit for gfx or before user queues are resumed for kfd.

If userptr pages are updated, for gfx, amdgpu_cs_ioctl will restart
from scratch, for kfd, restore worker is rescheduled to retry.

HMM simplify the CPU page table concurrent update check, so remove
guptasklock, mmu_invalidations, last_set_pages fields from
amdgpu_ttm_tt struct.

HMM does not pin the page (increase page ref count), so remove related
operations like release_pages(), put_page(), mark_page_dirty().
Signed-off-by: NPhilip Yang <Philip.Yang@amd.com>
Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

- move all adjustments into one place
- specify GDS/GWS/OA alignment in basic units of the heaps
- it looks like GDS alignment was 1 instead of 4
Signed-off-by: NMarek Olšák <marek.olsak@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

if lru is changed, we cannot do bulk moving.
v2:
root bo isn't in bulk moving, skip its change.
Signed-off-by: NChunming Zhou <david1.zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Start using drm_gpu_scheduler.ready isntead.

v3:
Add helper function to run ring test and set
sched.ready flag status accordingly, clean explicit
sched.ready sets from the IP specific files.

v4: Add kerneldoc and rebase.
Signed-off-by: NAndrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Make sure that the global BO state is always correctly initialized.

This allows removing all the device code to initialize it.

v2: fix up vbox (Alex)
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NJunwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

As the name says we only need one global instance of ttm_mem_global.

Drop all the driver initialization and just use a single exported
instance which is initialized during BO global initialization.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NJunwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>