<feed xmlns='http://www.w3.org/2005/Atom'>
<title>linux/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c, branch v5.1</title>
<subtitle>Mirror of https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/
</subtitle>
<id>https://git.shady.money/linux/atom?h=v5.1</id>
<link rel='self' href='https://git.shady.money/linux/atom?h=v5.1'/>
<link rel='alternate' type='text/html' href='https://git.shady.money/linux/'/>
<updated>2019-02-13T22:50:54Z</updated>
<entry>
<title>drm/amdgpu: Simplify eviction fence handling</title>
<updated>2019-02-13T22:50:54Z</updated>
<author>
<name>Felix Kuehling</name>
<email>Felix.Kuehling@amd.com</email>
</author>
<published>2019-02-04T23:17:26Z</published>
<link rel='alternate' type='text/html' href='https://git.shady.money/linux/commit/?id=2d086fded1021d6b482dfc705bab8f085d10a496'/>
<id>urn:sha1:2d086fded1021d6b482dfc705bab8f085d10a496</id>
<content type='text'>
Temporarily removing eviction fences to avoid triggering them by
accident is no longer necessary due to the fence_owner logic in
amdgpu_sync_resv.

As a result the ef_list usage of amdgpu_amdkfd_remove_eviction_fence
and amdgpu_amdkfd_add_eviction_fence are no longer needed.

Signed-off-by: Felix Kuehling &lt;Felix.Kuehling@amd.com&gt;
Acked-by: Christian König &lt;christian.koenig@amd.com&gt;
Reviewed-by: Harish Kasiviswanathan &lt;Harish.Kasiviswanathan@amd.com&gt;
Signed-off-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
</content>
</entry>
<entry>
<title>drm/amdgpu: Replace ttm_bo_wait with amdgpu_bo_sync_wait</title>
<updated>2019-02-13T22:50:39Z</updated>
<author>
<name>Felix Kuehling</name>
<email>Felix.Kuehling@amd.com</email>
</author>
<published>2019-02-04T22:53:05Z</published>
<link rel='alternate' type='text/html' href='https://git.shady.money/linux/commit/?id=c60cd590cb7da0d7dbb423727bb67350182a371c'/>
<id>urn:sha1:c60cd590cb7da0d7dbb423727bb67350182a371c</id>
<content type='text'>
The fence_owner logic in amdgpu_sync_wait will allow waiting without
having to temporarily remove eviction fences.

Signed-off-by: Felix Kuehling &lt;Felix.Kuehling@amd.com&gt;
Reviewed-by: Christian König &lt;christian.koenig@amd.com&gt;
Signed-off-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
</content>
</entry>
<entry>
<title>drm/amdgpu: Simplify kgd2kfd interface</title>
<updated>2019-01-14T20:04:29Z</updated>
<author>
<name>Amber Lin</name>
<email>Amber.Lin@amd.com</email>
</author>
<published>2018-12-13T20:18:01Z</published>
<link rel='alternate' type='text/html' href='https://git.shady.money/linux/commit/?id=8e07e2676a42e7d3e5fe8eebac6262ec975664a1'/>
<id>urn:sha1:8e07e2676a42e7d3e5fe8eebac6262ec975664a1</id>
<content type='text'>
After amdkfd is merged into amdgpu module, amdgpu can call amdkfd
functions directly.

Signed-off-by: Amber Lin &lt;Amber.Lin@amd.com&gt;
Reviewed-by: Felix Kuehling &lt;Felix.Kuehling@amd.com&gt;
Acked-by: Christian König &lt;christian.koenig@amd.com&gt;
Signed-off-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
</content>
</entry>
<entry>
<title>drm/amdgpu: Relocate kgd2kfd function declaration</title>
<updated>2019-01-14T20:04:28Z</updated>
<author>
<name>Amber Lin</name>
<email>Amber.Lin@amd.com</email>
</author>
<published>2018-12-13T16:57:35Z</published>
<link rel='alternate' type='text/html' href='https://git.shady.money/linux/commit/?id=2d3d25b616a09c16c2506f23289532a31638620f'/>
<id>urn:sha1:2d3d25b616a09c16c2506f23289532a31638620f</id>
<content type='text'>
Since amdkfd is merged into amdgpu module and amdgpu can access amdkfd
directly, move declaration of kgd2kfd functions from kfd_priv.h to
amdgpu_amdkfd.h

Signed-off-by: Amber Lin &lt;Amber.Lin@amd.com&gt;
Reviewed-by: Felix Kuehling &lt;Felix.Kuehling@amd.com&gt;
Acked-by: Christian König &lt;christian.koenig@amd.com&gt;
Signed-off-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
</content>
</entry>
<entry>
<title>drm/amdkfd: Add support for doorbell BOs</title>
<updated>2018-12-07T23:14:00Z</updated>
<author>
<name>Felix Kuehling</name>
<email>Felix.Kuehling@amd.com</email>
</author>
<published>2018-11-21T02:44:27Z</published>
<link rel='alternate' type='text/html' href='https://git.shady.money/linux/commit/?id=b408a548846f2343716351d55a6c9af9e73ec32c'/>
<id>urn:sha1:b408a548846f2343716351d55a6c9af9e73ec32c</id>
<content type='text'>
This allows user mode to map doorbell pages into GPUVM address space.
That way GPUs can submit to user mode queues (self-dispatch).

Acked-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
Signed-off-by: Felix Kuehling &lt;Felix.Kuehling@amd.com&gt;
Signed-off-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
</content>
</entry>
<entry>
<title>drm/amdkfd: Add DMABuf import functionality</title>
<updated>2018-12-07T23:13:54Z</updated>
<author>
<name>Felix Kuehling</name>
<email>Felix.Kuehling@amd.com</email>
</author>
<published>2018-11-21T02:00:29Z</published>
<link rel='alternate' type='text/html' href='https://git.shady.money/linux/commit/?id=1dde0ea95b782425b95455d487cb44991525a1d1'/>
<id>urn:sha1:1dde0ea95b782425b95455d487cb44991525a1d1</id>
<content type='text'>
This is used for interoperability between ROCm compute and graphics
APIs. It allows importing graphics driver BOs into the ROCm SVM
address space for zero-copy GPU access.

The API is split into two steps (query and import) to allow user mode
to manage the virtual address space allocation for the imported buffer.

Acked-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
Signed-off-by: Felix Kuehling &lt;Felix.Kuehling@amd.com&gt;
Signed-off-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
</content>
</entry>
<entry>
<title>drm/amdgpu: Add KFD VRAM limit checking</title>
<updated>2018-12-07T23:13:15Z</updated>
<author>
<name>Felix Kuehling</name>
<email>Felix.Kuehling@amd.com</email>
</author>
<published>2018-11-20T01:05:54Z</published>
<link rel='alternate' type='text/html' href='https://git.shady.money/linux/commit/?id=611736d8447c0c48a172db0b968dddae60696a72'/>
<id>urn:sha1:611736d8447c0c48a172db0b968dddae60696a72</id>
<content type='text'>
We don't want KFD processes evicting each other over VRAM usage.
Therefore prevent overcommitting VRAM among KFD applications with
a per-GPU limit. Also leave enough room for page tables on top
of the application memory usage.

Acked-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
Signed-off-by: Felix Kuehling &lt;Felix.Kuehling@amd.com&gt;
Reviewed-by: Eric Huang &lt;JinHuiEric.Huang@amd.com&gt;
Signed-off-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
</content>
</entry>
<entry>
<title>drm/ttm: allow reserving more than one shared slot v3</title>
<updated>2018-12-07T22:53:44Z</updated>
<author>
<name>Christian König</name>
<email>christian.koenig@amd.com</email>
</author>
<published>2018-09-19T14:25:08Z</published>
<link rel='alternate' type='text/html' href='https://git.shady.money/linux/commit/?id=a9f34c70fd168b164aadffd46bb757ded52e25b9'/>
<id>urn:sha1:a9f34c70fd168b164aadffd46bb757ded52e25b9</id>
<content type='text'>
Let's support simultaneous submissions to multiple engines.

v2: rename the field to num_shared and fix up all users
v3: rebased

Signed-off-by: Christian König &lt;christian.koenig@amd.com&gt;
Reviewed-by: Michel Dänzer &lt;michel.daenzer@amd.com&gt;
Reviewed-by: Junwei Zhang &lt;Jerry.Zhang@amd.com&gt;
Reviewed-by: Huang Rui &lt;ray.huang@amd.com&gt;
Signed-off-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
</content>
</entry>
<entry>
<title>drm/amdkfd: change system memory overcommit limit</title>
<updated>2018-11-20T19:01:54Z</updated>
<author>
<name>Eric Huang</name>
<email>JinhuiEric.Huang@amd.com</email>
</author>
<published>2018-09-05T15:46:14Z</published>
<link rel='alternate' type='text/html' href='https://git.shady.money/linux/commit/?id=5d240da93edc29adb68320c5e475dc9c7fcad5dd'/>
<id>urn:sha1:5d240da93edc29adb68320c5e475dc9c7fcad5dd</id>
<content type='text'>
It is to improve system limit by:
1. replacing userptrlimit with a total memory limit that
conunts TTM memory usage and userptr usage.
2. counting acc size for all BOs.

Signed-off-by: Eric Huang &lt;JinHuiEric.Huang@amd.com&gt;
Reviewed-by: Felix Kuehling &lt;Felix.Kuehling@amd.com&gt;
Signed-off-by: Felix Kuehling &lt;Felix.Kuehling@amd.com&gt;
Signed-off-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
</content>
</entry>
<entry>
<title>drm/amdgpu: Fix KFD doorbell SG BO mapping</title>
<updated>2018-11-20T19:01:12Z</updated>
<author>
<name>Felix Kuehling</name>
<email>Felix.Kuehling@amd.com</email>
</author>
<published>2018-05-03T21:37:56Z</published>
<link rel='alternate' type='text/html' href='https://git.shady.money/linux/commit/?id=0f04e538580c086046e3b31cbc7ef4307a5b771d'/>
<id>urn:sha1:0f04e538580c086046e3b31cbc7ef4307a5b771d</id>
<content type='text'>
This change prepares for adding SG BOs that will be used for mapping
doorbells into GPUVM address space.

This type of BO would be mistaken for an invalid userptr BO. Improve
that check to test that it's actually a userptr BO so that SG BOs that
are still in the CPU domain can be validated and mapped correctly.

Signed-off-by: Felix Kuehling &lt;Felix.Kuehling@amd.com&gt;
Signed-off-by: Alex Deucher &lt;alexander.deucher@amd.com&gt;
</content>
</entry>
</feed>
