Thread (8 messages) 8 messages, 4 authors, 2011-12-26

[RFC v3 0/2] Introduce DMA buffer sharing mechanism

From: Daniel Vetter <hidden>
Date: 2011-12-20 19:29:43
Also in: dri-devel, linux-media, linux-mm, lkml

On Mon, Dec 19, 2011 at 02:03:28PM +0530, Sumit Semwal wrote:
Hello Everyone,

This is RFC v3 for DMA buffer sharing mechanism - changes from v2 are in the
changelog below.

Various subsystems - V4L2, GPU-accessors, DRI to name a few - have felt the 
need to have a common mechanism to share memory buffers across different
devices - ARM, video hardware, GPU.

This need comes forth from a variety of use cases including cameras, image 
processing, video recorders, sound processing, DMA engines, GPU and display
buffers, and others.

This RFC is an attempt to define such a buffer sharing mechanism- it is the
result of discussions from a couple of memory-management mini-summits held by
Linaro to understand and address common needs around memory management. [1]

A new dma_buf buffer object is added, with operations and API to allow easy
sharing of this buffer object across devices.

The framework allows:
- a new buffer-object to be created with fixed size.
- different devices to 'attach' themselves to this buffer, to facilitate
  backing storage negotiation, using dma_buf_attach() API.
- association of a file pointer with each user-buffer and associated
   allocator-defined operations on that buffer. This operation is called the
   'export' operation.
- this exported buffer-object to be shared with the other entity by asking for
   its 'file-descriptor (fd)', and sharing the fd across.
- a received fd to get the buffer object back, where it can be accessed using
   the associated exporter-defined operations.
- the exporter and user to share the scatterlist using map_dma_buf and
   unmap_dma_buf operations.

Documentation present in the patch-set gives more details.

This is based on design suggestions from many people at the mini-summits,
most notably from Arnd Bergmann [off-list ref], Rob Clark [off-list ref] and
Daniel Vetter [off-list ref].

The implementation is inspired from proof-of-concept patch-set from
Tomasz Stanislawski [off-list ref], who demonstrated buffer sharing
between two v4l2 devices. [2]

References:
[1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement
[2]: http://lwn.net/Articles/454389

Patchset based on top of 3.2-rc3, the current version can be found at

http://git.linaro.org/gitweb?p=people/sumitsemwal/linux-3.x.git
Branch: dma-buf-upstr-v2

Earlier versions:
v2 at: https://lkml.org/lkml/2011/12/2/53
v1 at: https://lkml.org/lkml/2011/10/11/92

Best regards,
~Sumit Semwal
I think this is a really good v1 version of dma_buf. It contains all the
required bits (with well-specified semantics in the doc patch) to
implement some basic use-cases and start fleshing out the integration with
various subsystem (like drm and v4l). All the things still under
discussion like
- userspace mmap support
- more advanced (and more strictly specified) coherency models
- and shared infrastructure for implementing exporters
are imo much clearer once we have a few example drivers at hand and a
better understanding of some of the insane corner cases we need to be able
to handle.

And I think any risk that the resulting clarifications will break a basic
use-case is really minimal, so I think it'd be great if this could go into
3.3 (maybe as some kind of staging/experimental infrastructure).

Hence for both patches:
Reviewed-by: Daniel Vetter <redacted>
-- 
Daniel Vetter
Mail: daniel at ffwll.ch
Mobile: +41 (0)79 365 57 48
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help