lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <3602500f-05f5-10b8-5ec6-0a6246e2bb6b@amd.com>
Date:   Wed, 18 Jan 2023 16:37:02 +0100
From:   Christian König <christian.koenig@....com>
To:     Danilo Krummrich <dakr@...hat.com>, daniel@...ll.ch,
        airlied@...hat.com, bskeggs@...hat.com, jason@...kstrand.net,
        tzimmermann@...e.de, mripard@...nel.org, corbet@....net
Cc:     dri-devel@...ts.freedesktop.org, nouveau@...ts.freedesktop.org,
        linux-doc@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH drm-next 00/14] [RFC] DRM GPUVA Manager & Nouveau VM_BIND
 UAPI

Am 18.01.23 um 16:34 schrieb Danilo Krummrich:
> Hi Christian,
>
> On 1/18/23 09:53, Christian König wrote:
>> Am 18.01.23 um 07:12 schrieb Danilo Krummrich:
>>> This patch series provides a new UAPI for the Nouveau driver in 
>>> order to
>>> support Vulkan features, such as sparse bindings and sparse residency.
>>>
>>> Furthermore, with the DRM GPUVA manager it provides a new DRM core 
>>> feature to
>>> keep track of GPU virtual address (VA) mappings in a more generic way.
>>>
>>> The DRM GPUVA manager is indented to help drivers implement 
>>> userspace-manageable
>>> GPU VA spaces in reference to the Vulkan API. In order to achieve 
>>> this goal it
>>> serves the following purposes in this context.
>>>
>>>      1) Provide a dedicated range allocator to track GPU VA 
>>> allocations and
>>>         mappings, making use of the drm_mm range allocator.
>>
>> This means that the ranges are allocated by the kernel? If yes that's 
>> a really really bad idea.
>
> No, it's just for keeping track of the ranges userspace has allocated.

Ok, that makes more sense.

So basically you have an IOCTL which asks kernel for a free range? Or 
what exactly is the drm_mm used for here?

Regards,
Christian.

>
> - Danilo
>
>>
>> Regards,
>> Christian.
>>
>>>
>>>      2) Generically connect GPU VA mappings to their backing 
>>> buffers, in
>>>         particular DRM GEM objects.
>>>
>>>      3) Provide a common implementation to perform more complex mapping
>>>         operations on the GPU VA space. In particular splitting and 
>>> merging
>>>         of GPU VA mappings, e.g. for intersecting mapping requests 
>>> or partial
>>>         unmap requests.
>>>
>>> The new VM_BIND Nouveau UAPI build on top of the DRM GPUVA manager, 
>>> itself
>>> providing the following new interfaces.
>>>
>>>      1) Initialize a GPU VA space via the new 
>>> DRM_IOCTL_NOUVEAU_VM_INIT ioctl
>>>         for UMDs to specify the portion of VA space managed by the 
>>> kernel and
>>>         userspace, respectively.
>>>
>>>      2) Allocate and free a VA space region as well as bind and 
>>> unbind memory
>>>         to the GPUs VA space via the new DRM_IOCTL_NOUVEAU_VM_BIND 
>>> ioctl.
>>>
>>>      3) Execute push buffers with the new DRM_IOCTL_NOUVEAU_EXEC ioctl.
>>>
>>> Both, DRM_IOCTL_NOUVEAU_VM_BIND and DRM_IOCTL_NOUVEAU_EXEC, make use 
>>> of the DRM
>>> scheduler to queue jobs and support asynchronous processing with DRM 
>>> syncobjs
>>> as synchronization mechanism.
>>>
>>> By default DRM_IOCTL_NOUVEAU_VM_BIND does synchronous processing,
>>> DRM_IOCTL_NOUVEAU_EXEC supports asynchronous processing only.
>>>
>>> The new VM_BIND UAPI for Nouveau makes also use of drm_exec 
>>> (execution context
>>> for GEM buffers) by Christian König. Since the patch implementing 
>>> drm_exec was
>>> not yet merged into drm-next it is part of this series, as well as a 
>>> small fix
>>> for this patch, which was found while testing this series.
>>>
>>> This patch series is also available at [1].
>>>
>>> There is a Mesa NVK merge request by Dave Airlie [2] implementing the
>>> corresponding userspace parts for this series.
>>>
>>> The Vulkan CTS test suite passes the sparse binding and sparse 
>>> residency test
>>> cases for the new UAPI together with Dave's Mesa work.
>>>
>>> There are also some test cases in the igt-gpu-tools project [3] for 
>>> the new UAPI
>>> and hence the DRM GPU VA manager. However, most of them are testing 
>>> the DRM GPU
>>> VA manager's logic through Nouveau's new UAPI and should be 
>>> considered just as
>>> helper for implementation.
>>>
>>> However, I absolutely intend to change those test cases to proper 
>>> kunit test
>>> cases for the DRM GPUVA manager, once and if we agree on it's 
>>> usefulness and
>>> design.
>>>
>>> [1] 
>>> https://gitlab.freedesktop.org/nouvelles/kernel/-/tree/new-uapi-drm-next 
>>> /
>>> https://gitlab.freedesktop.org/nouvelles/kernel/-/merge_requests/1
>>> [2] https://gitlab.freedesktop.org/nouveau/mesa/-/merge_requests/150/
>>> [3] 
>>> https://gitlab.freedesktop.org/dakr/igt-gpu-tools/-/tree/wip_nouveau_vm_bind
>>>
>>> I also want to give credit to Dave Airlie, who contributed a lot of 
>>> ideas to
>>> this patch series.
>>>
>>> Christian König (1):
>>>    drm: execution context for GEM buffers
>>>
>>> Danilo Krummrich (13):
>>>    drm/exec: fix memory leak in drm_exec_prepare_obj()
>>>    drm: manager to keep track of GPUs VA mappings
>>>    drm: debugfs: provide infrastructure to dump a DRM GPU VA space
>>>    drm/nouveau: new VM_BIND uapi interfaces
>>>    drm/nouveau: get vmm via nouveau_cli_vmm()
>>>    drm/nouveau: bo: initialize GEM GPU VA interface
>>>    drm/nouveau: move usercopy helpers to nouveau_drv.h
>>>    drm/nouveau: fence: fail to emit when fence context is killed
>>>    drm/nouveau: chan: provide nouveau_channel_kill()
>>>    drm/nouveau: nvkm/vmm: implement raw ops to manage uvmm
>>>    drm/nouveau: implement uvmm for user mode bindings
>>>    drm/nouveau: implement new VM_BIND UAPI
>>>    drm/nouveau: debugfs: implement DRM GPU VA debugfs
>>>
>>>   Documentation/gpu/driver-uapi.rst             |   11 +
>>>   Documentation/gpu/drm-mm.rst                  |   43 +
>>>   drivers/gpu/drm/Kconfig                       |    6 +
>>>   drivers/gpu/drm/Makefile                      |    3 +
>>>   drivers/gpu/drm/amd/amdgpu/Kconfig            |    1 +
>>>   drivers/gpu/drm/drm_debugfs.c                 |   56 +
>>>   drivers/gpu/drm/drm_exec.c                    |  294 ++++
>>>   drivers/gpu/drm/drm_gem.c                     |    3 +
>>>   drivers/gpu/drm/drm_gpuva_mgr.c               | 1323 
>>> +++++++++++++++++
>>>   drivers/gpu/drm/nouveau/Kbuild                |    3 +
>>>   drivers/gpu/drm/nouveau/Kconfig               |    2 +
>>>   drivers/gpu/drm/nouveau/include/nvif/if000c.h |   23 +-
>>>   drivers/gpu/drm/nouveau/include/nvif/vmm.h    |   17 +-
>>>   .../gpu/drm/nouveau/include/nvkm/subdev/mmu.h |   10 +
>>>   drivers/gpu/drm/nouveau/nouveau_abi16.c       |   23 +
>>>   drivers/gpu/drm/nouveau/nouveau_abi16.h       |    1 +
>>>   drivers/gpu/drm/nouveau/nouveau_bo.c          |  152 +-
>>>   drivers/gpu/drm/nouveau/nouveau_bo.h          |    2 +-
>>>   drivers/gpu/drm/nouveau/nouveau_chan.c        |   16 +-
>>>   drivers/gpu/drm/nouveau/nouveau_chan.h        |    1 +
>>>   drivers/gpu/drm/nouveau/nouveau_debugfs.c     |   24 +
>>>   drivers/gpu/drm/nouveau/nouveau_drm.c         |   25 +-
>>>   drivers/gpu/drm/nouveau/nouveau_drv.h         |   92 +-
>>>   drivers/gpu/drm/nouveau/nouveau_exec.c        |  310 ++++
>>>   drivers/gpu/drm/nouveau/nouveau_exec.h        |   55 +
>>>   drivers/gpu/drm/nouveau/nouveau_fence.c       |    7 +
>>>   drivers/gpu/drm/nouveau/nouveau_fence.h       |    2 +-
>>>   drivers/gpu/drm/nouveau/nouveau_gem.c         |   83 +-
>>>   drivers/gpu/drm/nouveau/nouveau_mem.h         |    5 +
>>>   drivers/gpu/drm/nouveau/nouveau_prime.c       |    2 +-
>>>   drivers/gpu/drm/nouveau/nouveau_sched.c       |  780 ++++++++++
>>>   drivers/gpu/drm/nouveau/nouveau_sched.h       |   98 ++
>>>   drivers/gpu/drm/nouveau/nouveau_svm.c         |    2 +-
>>>   drivers/gpu/drm/nouveau/nouveau_uvmm.c        |  575 +++++++
>>>   drivers/gpu/drm/nouveau/nouveau_uvmm.h        |   68 +
>>>   drivers/gpu/drm/nouveau/nouveau_vmm.c         |    4 +-
>>>   drivers/gpu/drm/nouveau/nvif/vmm.c            |   73 +-
>>>   .../gpu/drm/nouveau/nvkm/subdev/mmu/uvmm.c    |  168 ++-
>>>   .../gpu/drm/nouveau/nvkm/subdev/mmu/uvmm.h    |    1 +
>>>   drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.c |   32 +-
>>>   drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.h |    3 +
>>>   include/drm/drm_debugfs.h                     |   25 +
>>>   include/drm/drm_drv.h                         |    6 +
>>>   include/drm/drm_exec.h                        |  144 ++
>>>   include/drm/drm_gem.h                         |   75 +
>>>   include/drm/drm_gpuva_mgr.h                   |  527 +++++++
>>>   include/uapi/drm/nouveau_drm.h                |  216 +++
>>>   47 files changed, 5266 insertions(+), 126 deletions(-)
>>>   create mode 100644 drivers/gpu/drm/drm_exec.c
>>>   create mode 100644 drivers/gpu/drm/drm_gpuva_mgr.c
>>>   create mode 100644 drivers/gpu/drm/nouveau/nouveau_exec.c
>>>   create mode 100644 drivers/gpu/drm/nouveau/nouveau_exec.h
>>>   create mode 100644 drivers/gpu/drm/nouveau/nouveau_sched.c
>>>   create mode 100644 drivers/gpu/drm/nouveau/nouveau_sched.h
>>>   create mode 100644 drivers/gpu/drm/nouveau/nouveau_uvmm.c
>>>   create mode 100644 drivers/gpu/drm/nouveau/nouveau_uvmm.h
>>>   create mode 100644 include/drm/drm_exec.h
>>>   create mode 100644 include/drm/drm_gpuva_mgr.h
>>>
>>>
>>> base-commit: 0b45ac1170ea6416bc1d36798414c04870cd356d
>>
>

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ