[RFC,1/2] dma-buf: Introduce dma buffer sharing mechanism

Message ID 1318325033-32688-2-git-send-email-sumit.semwal@ti.com (mailing list archive)
State RFC, archived
Headers

Commit Message

Semwal, Sumit Oct. 11, 2011, 9:23 a.m. UTC
  This is the first step in defining a dma buffer sharing mechanism.

A new buffer object dma_buf 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 get_scatterlist and
   put_scatterlist operations.

Atleast one 'attach()' call is required to be made prior to calling the
get_scatterlist() operation.

Couple of building blocks in get_scatterlist() are added to ease introduction
of sync'ing across exporter and users, and late allocation by the exporter.

mmap() file operation is provided for the associated 'fd', as wrapper over the
optional allocator defined mmap(), to be used by devices that might need one.

More details are there in the documentation patch.

This is based on design suggestions from many people at the mini-summits[1],
most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and
Daniel Vetter <daniel@ffwll.ch>.

The implementation is inspired from proof-of-concept patch-set from
Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing
between two v4l2 devices. [2]

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

Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Sumit Semwal <sumit.semwal@ti.com>
---
 drivers/base/Kconfig    |   10 ++
 drivers/base/Makefile   |    1 +
 drivers/base/dma-buf.c  |  242 +++++++++++++++++++++++++++++++++++++++++++++++
 include/linux/dma-buf.h |  162 +++++++++++++++++++++++++++++++
 4 files changed, 415 insertions(+), 0 deletions(-)
 create mode 100644 drivers/base/dma-buf.c
 create mode 100644 include/linux/dma-buf.h
  

Comments

Dave Airlie Oct. 12, 2011, 12:41 p.m. UTC | #1
On Tue, Oct 11, 2011 at 10:23 AM, Sumit Semwal <sumit.semwal@ti.com> wrote:
> This is the first step in defining a dma buffer sharing mechanism.
>
> A new buffer object dma_buf 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 get_scatterlist and
>   put_scatterlist operations.
>
> Atleast one 'attach()' call is required to be made prior to calling the
> get_scatterlist() operation.
>
> Couple of building blocks in get_scatterlist() are added to ease introduction
> of sync'ing across exporter and users, and late allocation by the exporter.
>
> mmap() file operation is provided for the associated 'fd', as wrapper over the
> optional allocator defined mmap(), to be used by devices that might need one.

Why is this needed? it really doesn't make sense to be mmaping objects
independent of some front-end like drm or v4l.

how will you know what contents are in them, how will you synchronise
access. Unless someone has a hard use-case for this I'd say we drop it
until someone does.

Dave.
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Rob Clark Oct. 12, 2011, 1:28 p.m. UTC | #2
On Wed, Oct 12, 2011 at 7:41 AM, Dave Airlie <airlied@gmail.com> wrote:
> On Tue, Oct 11, 2011 at 10:23 AM, Sumit Semwal <sumit.semwal@ti.com> wrote:
>> This is the first step in defining a dma buffer sharing mechanism.
>>
>> A new buffer object dma_buf 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 get_scatterlist and
>>   put_scatterlist operations.
>>
>> Atleast one 'attach()' call is required to be made prior to calling the
>> get_scatterlist() operation.
>>
>> Couple of building blocks in get_scatterlist() are added to ease introduction
>> of sync'ing across exporter and users, and late allocation by the exporter.
>>
>> mmap() file operation is provided for the associated 'fd', as wrapper over the
>> optional allocator defined mmap(), to be used by devices that might need one.
>
> Why is this needed? it really doesn't make sense to be mmaping objects
> independent of some front-end like drm or v4l.

well, the mmap is actually implemented by the buffer allocator
(v4l/drm).. although not sure if this was the point

> how will you know what contents are in them, how will you synchronise
> access. Unless someone has a hard use-case for this I'd say we drop it
> until someone does.

The intent was that this is for well defined formats.. ie. it would
need to be a format that both v4l and drm understood in the first
place for sharing to make sense at all..

Anyways, the basic reason is to handle random edge cases where you
need sw access to the buffer.  For example, you are decoding video and
pull out a frame to generate a thumbnail w/ a sw jpeg encoder..

On gstreamer 0.11 branch, for example, there is already a map/unmap
virtual method on the gst buffer for sw access (ie. same purpose as
PrepareAccess/FinishAccess in EXA).  The idea w/ dmabuf mmap() support
is that we could implement support to mmap()/munmap() before/after sw
access.

With this current scheme, synchronization could be handled in
dmabufops->mmap() and vm_ops->close()..  it is perhaps a bit heavy to
require mmap/munmap for each sw access, but I suppose this isn't
really for the high-performance use case.  It is just so that some
random bit of sw that gets passed a dmabuf handle without knowing who
allocated it can have sw access if really needed.

BR,
-R

> Dave.
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Dave Airlie Oct. 12, 2011, 1:35 p.m. UTC | #3
>
> well, the mmap is actually implemented by the buffer allocator
> (v4l/drm).. although not sure if this was the point

Then why not use the correct interface? doing some sort of not-quite
generic interface isn't really helping anyone except adding an ABI
that we have to support.

If someone wants to bypass the current kernel APIs we should add a new
API for them not shove it into this generic buffer sharing layer.

> The intent was that this is for well defined formats.. ie. it would
> need to be a format that both v4l and drm understood in the first
> place for sharing to make sense at all..

How will you know the stride to take a simple example? The userspace
had to create this buffer somehow and wants to share it with
"something", you sound like
you really needs another API that is a simple accessor API that can
handle mmaps.

> Anyways, the basic reason is to handle random edge cases where you
> need sw access to the buffer.  For example, you are decoding video and
> pull out a frame to generate a thumbnail w/ a sw jpeg encoder..

Again, doesn't sound like it should be part of this API, and also
sounds like the sw jpeg encoder will need more info about the buffer
anyways like stride and format.

> With this current scheme, synchronization could be handled in
> dmabufops->mmap() and vm_ops->close()..  it is perhaps a bit heavy to
> require mmap/munmap for each sw access, but I suppose this isn't
> really for the high-performance use case.  It is just so that some
> random bit of sw that gets passed a dmabuf handle without knowing who
> allocated it can have sw access if really needed.

So I think thats fine, write a sw accessor providers, don't go
overloading the buffer sharing code.

This API will limit what people can use this buffer sharing for with
pure hw accessors, you might say, oh buts its okay to fail the mmap
then, but the chances of sw handling that I'm not so sure off.

Dave.
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Rob Clark Oct. 12, 2011, 1:50 p.m. UTC | #4
On Wed, Oct 12, 2011 at 8:35 AM, Dave Airlie <airlied@gmail.com> wrote:
>>
>> well, the mmap is actually implemented by the buffer allocator
>> (v4l/drm).. although not sure if this was the point
>
> Then why not use the correct interface? doing some sort of not-quite
> generic interface isn't really helping anyone except adding an ABI
> that we have to support.

But what if you don't know who allocated the buffer?  How do you know
what interface to use to mmap?

> If someone wants to bypass the current kernel APIs we should add a new
> API for them not shove it into this generic buffer sharing layer.
>
>> The intent was that this is for well defined formats.. ie. it would
>> need to be a format that both v4l and drm understood in the first
>> place for sharing to make sense at all..
>
> How will you know the stride to take a simple example? The userspace
> had to create this buffer somehow and wants to share it with
> "something", you sound like
> you really needs another API that is a simple accessor API that can
> handle mmaps.

Well, things like stride, width, height, color format, userspace needs
to know all this already, even for malloc()'d sw buffers.  The
assumption is userspace already has a way to pass this information
around so it was not required to be duplicated by dmabuf.

>> Anyways, the basic reason is to handle random edge cases where you
>> need sw access to the buffer.  For example, you are decoding video and
>> pull out a frame to generate a thumbnail w/ a sw jpeg encoder..
>
> Again, doesn't sound like it should be part of this API, and also
> sounds like the sw jpeg encoder will need more info about the buffer
> anyways like stride and format.
>
>> With this current scheme, synchronization could be handled in
>> dmabufops->mmap() and vm_ops->close()..  it is perhaps a bit heavy to
>> require mmap/munmap for each sw access, but I suppose this isn't
>> really for the high-performance use case.  It is just so that some
>> random bit of sw that gets passed a dmabuf handle without knowing who
>> allocated it can have sw access if really needed.
>
> So I think thats fine, write a sw accessor providers, don't go
> overloading the buffer sharing code.

But then we'd need a different set of accessors for every different
drm/v4l/etc driver, wouldn't we?

> This API will limit what people can use this buffer sharing for with
> pure hw accessors, you might say, oh buts its okay to fail the mmap
> then, but the chances of sw handling that I'm not so sure off.

I'm not entirely sure the case you are worried about.. sharing buffers
between multiple GPU's that understand same tiled formats?  I guess
that is a bit different from a case like a jpeg encoder that is passed
a dmabuf handle without any idea where it came from..

I guess if sharing a buffer between multiple drm devices, there is
nothing stopping you from having some NOT_DMABUF_MMAPABLE flag you
pass when the buffer is allocated, then you don't have to support
dmabuf->mmap(), and instead mmap via device and use some sort of
DRM_CPU_PREP/FINI ioctls for synchronization..

BR,
-R

> Dave.
>
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Dave Airlie Oct. 12, 2011, 2:01 p.m. UTC | #5
> But then we'd need a different set of accessors for every different
> drm/v4l/etc driver, wouldn't we?

Not any more different than you need for this, you just have a new
interface that you request a sw object from,
then mmap that object, and underneath it knows who owns it in the kernel.

mmap just feels wrong in this API, which is a buffer sharing API not a
buffer mapping API.

> I guess if sharing a buffer between multiple drm devices, there is
> nothing stopping you from having some NOT_DMABUF_MMAPABLE flag you
> pass when the buffer is allocated, then you don't have to support
> dmabuf->mmap(), and instead mmap via device and use some sort of
> DRM_CPU_PREP/FINI ioctls for synchronization..

Or we could make a generic CPU accessor that we don't have to worry about.

Dave.
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Rob Clark Oct. 12, 2011, 2:24 p.m. UTC | #6
On Wed, Oct 12, 2011 at 9:01 AM, Dave Airlie <airlied@gmail.com> wrote:
>> But then we'd need a different set of accessors for every different
>> drm/v4l/etc driver, wouldn't we?
>
> Not any more different than you need for this, you just have a new
> interface that you request a sw object from,
> then mmap that object, and underneath it knows who owns it in the kernel.

oh, ok, so you are talking about a kernel level interface, rather than
userspace..

but I guess in this case I don't quite see the difference.  It amounts
to which fd you call mmap (or ioctl[*]) on..  If you use the dmabuf fd
directly then you don't have to pass around a 2nd fd.

[*] there is nothing stopping defining some dmabuf ioctls (such as for
synchronization).. although the thinking was to keep it simple for
first version of dmabuf

BR,
-R

> mmap just feels wrong in this API, which is a buffer sharing API not a
> buffer mapping API.
>
>> I guess if sharing a buffer between multiple drm devices, there is
>> nothing stopping you from having some NOT_DMABUF_MMAPABLE flag you
>> pass when the buffer is allocated, then you don't have to support
>> dmabuf->mmap(), and instead mmap via device and use some sort of
>> DRM_CPU_PREP/FINI ioctls for synchronization..
>
> Or we could make a generic CPU accessor that we don't have to worry about.
>
> Dave.
>
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Dave Airlie Oct. 12, 2011, 2:34 p.m. UTC | #7
On Wed, Oct 12, 2011 at 3:24 PM, Rob Clark <robdclark@gmail.com> wrote:
> On Wed, Oct 12, 2011 at 9:01 AM, Dave Airlie <airlied@gmail.com> wrote:
>>> But then we'd need a different set of accessors for every different
>>> drm/v4l/etc driver, wouldn't we?
>>
>> Not any more different than you need for this, you just have a new
>> interface that you request a sw object from,
>> then mmap that object, and underneath it knows who owns it in the kernel.
>
> oh, ok, so you are talking about a kernel level interface, rather than
> userspace..
>
> but I guess in this case I don't quite see the difference.  It amounts
> to which fd you call mmap (or ioctl[*]) on..  If you use the dmabuf fd
> directly then you don't have to pass around a 2nd fd.
>
> [*] there is nothing stopping defining some dmabuf ioctls (such as for
> synchronization).. although the thinking was to keep it simple for
> first version of dmabuf
>

Yes a separate kernel level interface.

Well I'd like to keep it even simpler. dmabuf is a buffer sharing API,
shoehorning in a sw mapping API isn't making it simpler.

The problem I have with implementing mmap on the sharing fd, is that
nothing says this should be purely optional and userspace shouldn't
rely on it.

In the Intel GEM space alone you have two types of mapping, one direct
to shmem one via GTT, the GTT could be even be a linear view. The
intel guys initially did GEM mmaps direct to the shmem pages because
it seemed simple, up until they
had to do step two which was do mmaps on the GTT copy and ended up
having two separate mmap methods. I think the problem here is it seems
deceptively simple to add this to the API now because the API is
simple, however I think in the future it'll become a burden that we'll
have to workaround.

Dave.
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Daniel Vetter Oct. 12, 2011, 2:49 p.m. UTC | #8
On Wed, Oct 12, 2011 at 03:34:54PM +0100, Dave Airlie wrote:
> On Wed, Oct 12, 2011 at 3:24 PM, Rob Clark <robdclark@gmail.com> wrote:
> > On Wed, Oct 12, 2011 at 9:01 AM, Dave Airlie <airlied@gmail.com> wrote:
> >>> But then we'd need a different set of accessors for every different
> >>> drm/v4l/etc driver, wouldn't we?
> >>
> >> Not any more different than you need for this, you just have a new
> >> interface that you request a sw object from,
> >> then mmap that object, and underneath it knows who owns it in the kernel.
> >
> > oh, ok, so you are talking about a kernel level interface, rather than
> > userspace..
> >
> > but I guess in this case I don't quite see the difference.  It amounts
> > to which fd you call mmap (or ioctl[*]) on..  If you use the dmabuf fd
> > directly then you don't have to pass around a 2nd fd.
> >
> > [*] there is nothing stopping defining some dmabuf ioctls (such as for
> > synchronization).. although the thinking was to keep it simple for
> > first version of dmabuf
> >
> 
> Yes a separate kernel level interface.
> 
> Well I'd like to keep it even simpler. dmabuf is a buffer sharing API,
> shoehorning in a sw mapping API isn't making it simpler.
> 
> The problem I have with implementing mmap on the sharing fd, is that
> nothing says this should be purely optional and userspace shouldn't
> rely on it.
> 
> In the Intel GEM space alone you have two types of mapping, one direct
> to shmem one via GTT, the GTT could be even be a linear view. The
> intel guys initially did GEM mmaps direct to the shmem pages because
> it seemed simple, up until they
> had to do step two which was do mmaps on the GTT copy and ended up
> having two separate mmap methods. I think the problem here is it seems
> deceptively simple to add this to the API now because the API is
> simple, however I think in the future it'll become a burden that we'll
> have to workaround.

Yeah, that's my feeling, too. Adding mmap sounds like a neat, simple idea,
that could simplify things for simple devices like v4l. But as soon as
you're dealing with a real gpu, nothing is simple. Those who don't believe
this, just take a look at the data upload/download paths in the
open-source i915,nouveau,radeon drivers. Making this fast (and for gpus,
it needs to be fast) requires tons of tricks, special-cases and jumping
through loops.

You absolutely want the device-specific ioctls to do that. Adding a
generic mmap just makes matters worse, especially if userspace expects
this to work synchronized with everything else that is going on.

Cheers, Daniel
  
Rob Clark Oct. 12, 2011, 3:15 p.m. UTC | #9
On Wed, Oct 12, 2011 at 9:34 AM, Dave Airlie <airlied@gmail.com> wrote:
> On Wed, Oct 12, 2011 at 3:24 PM, Rob Clark <robdclark@gmail.com> wrote:
>> On Wed, Oct 12, 2011 at 9:01 AM, Dave Airlie <airlied@gmail.com> wrote:
>>>> But then we'd need a different set of accessors for every different
>>>> drm/v4l/etc driver, wouldn't we?
>>>
>>> Not any more different than you need for this, you just have a new
>>> interface that you request a sw object from,
>>> then mmap that object, and underneath it knows who owns it in the kernel.
>>
>> oh, ok, so you are talking about a kernel level interface, rather than
>> userspace..
>>
>> but I guess in this case I don't quite see the difference.  It amounts
>> to which fd you call mmap (or ioctl[*]) on..  If you use the dmabuf fd
>> directly then you don't have to pass around a 2nd fd.
>>
>> [*] there is nothing stopping defining some dmabuf ioctls (such as for
>> synchronization).. although the thinking was to keep it simple for
>> first version of dmabuf
>>
>
> Yes a separate kernel level interface.

I'm not against it, but if it is a device-independent interface, it
just seems like six of one, half-dozen of the other..

Ie. how does it differ if the dmabuf fd is the fd used for ioctl/mmap,
vs if some other /dev/buffer-sharer file that you open?

But I think maybe I'm misunderstanding what you have in mind?

BR,
-R

> Well I'd like to keep it even simpler. dmabuf is a buffer sharing API,
> shoehorning in a sw mapping API isn't making it simpler.
>
> The problem I have with implementing mmap on the sharing fd, is that
> nothing says this should be purely optional and userspace shouldn't
> rely on it.
>
> In the Intel GEM space alone you have two types of mapping, one direct
> to shmem one via GTT, the GTT could be even be a linear view. The
> intel guys initially did GEM mmaps direct to the shmem pages because
> it seemed simple, up until they
> had to do step two which was do mmaps on the GTT copy and ended up
> having two separate mmap methods. I think the problem here is it seems
> deceptively simple to add this to the API now because the API is
> simple, however I think in the future it'll become a burden that we'll
> have to workaround.
>
> Dave.
>
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Tomasz Stanislawski Oct. 14, 2011, 10 a.m. UTC | #10
Hi Mr. Sumit Semwal,
Thank you for taking care of the framework for buffer sharing.
The support of buffer sharing in V4L2, both exporting and importing was 
posted in shrbuf proof-of-concept patch. It should be easy to port it to 
dmabuf.

http://lists.linaro.org/pipermail/linaro-mm-sig/2011-August/000485.html

Please refer to the comments below:

On 10/11/2011 11:23 AM, Sumit Semwal wrote:
> This is the first step in defining a dma buffer sharing mechanism.
>
> A new buffer object dma_buf 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 get_scatterlist and
>     put_scatterlist operations.
>
> Atleast one 'attach()' call is required to be made prior to calling the
> get_scatterlist() operation.
>
> Couple of building blocks in get_scatterlist() are added to ease introduction
> of sync'ing across exporter and users, and late allocation by the exporter.
>
> mmap() file operation is provided for the associated 'fd', as wrapper over the
> optional allocator defined mmap(), to be used by devices that might need one.
>
> More details are there in the documentation patch.
>
> This is based on design suggestions from many people at the mini-summits[1],
> most notably from Arnd Bergmann<arnd@arndb.de>, Rob Clark<rob@ti.com>  and
> Daniel Vetter<daniel@ffwll.ch>.
>
> The implementation is inspired from proof-of-concept patch-set from
> Tomasz Stanislawski<t.stanislaws@samsung.com>, who demonstrated buffer sharing
> between two v4l2 devices. [2]
>
> [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement
> [2]: http://lwn.net/Articles/454389
>
> Signed-off-by: Sumit Semwal<sumit.semwal@linaro.org>
> Signed-off-by: Sumit Semwal<sumit.semwal@ti.com>
> ---
>   drivers/base/Kconfig    |   10 ++
>   drivers/base/Makefile   |    1 +
>   drivers/base/dma-buf.c  |  242 +++++++++++++++++++++++++++++++++++++++++++++++
>   include/linux/dma-buf.h |  162 +++++++++++++++++++++++++++++++
>   4 files changed, 415 insertions(+), 0 deletions(-)
>   create mode 100644 drivers/base/dma-buf.c
>   create mode 100644 include/linux/dma-buf.h
>
> diff --git a/drivers/base/Kconfig b/drivers/base/Kconfig
> index 21cf46f..07d8095 100644
> --- a/drivers/base/Kconfig
> +++ b/drivers/base/Kconfig
> @@ -174,4 +174,14 @@ config SYS_HYPERVISOR
>
>   source "drivers/base/regmap/Kconfig"
>
> +config DMA_SHARED_BUFFER
> +	bool "Buffer framework to be shared between drivers"
> +	default n
> +	depends on ANON_INODES
> +	help
> +	  This option enables the framework for buffer-sharing between
> +	  multiple drivers. A buffer is associated with a file using driver
> +	  APIs extension; the file's descriptor can then be passed on to other
> +	  driver.
> +
>   endmenu
> diff --git a/drivers/base/Makefile b/drivers/base/Makefile
> index 99a375a..d0df046 100644
> --- a/drivers/base/Makefile
> +++ b/drivers/base/Makefile
> @@ -8,6 +8,7 @@ obj-$(CONFIG_DEVTMPFS)	+= devtmpfs.o
>   obj-y			+= power/
>   obj-$(CONFIG_HAS_DMA)	+= dma-mapping.o
>   obj-$(CONFIG_HAVE_GENERIC_DMA_COHERENT) += dma-coherent.o
> +obj-$(CONFIG_DMA_SHARED_BUFFER) += dma-buf.o
>   obj-$(CONFIG_ISA)	+= isa.o
>   obj-$(CONFIG_FW_LOADER)	+= firmware_class.o
>   obj-$(CONFIG_NUMA)	+= node.o
> diff --git a/drivers/base/dma-buf.c b/drivers/base/dma-buf.c
> new file mode 100644
> index 0000000..58c51a0
> --- /dev/null
> +++ b/drivers/base/dma-buf.c
> @@ -0,0 +1,242 @@
> +/*
> + * Framework for buffer objects that can be shared across devices/subsystems.
> + *
> + * Copyright(C) 2011 Linaro Limited. All rights reserved.
> + * Author: Sumit Semwal<sumit.semwal@ti.com>
> + *
> + * Many thanks to linaro-mm-sig list, and specially
> + * Arnd Bergmann<arnd@arndb.de>, Rob Clark<rob@ti.com>  and
> + * Daniel Vetter<daniel@ffwll.ch>  for their support in creation and
> + * refining of this idea.
> + *
> + * This program is free software; you can redistribute it and/or modify it
> + * under the terms of the GNU General Public License version 2 as published by
> + * the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful, but WITHOUT
> + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
> + * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
> + * more details.
> + *
> + * You should have received a copy of the GNU General Public License along with
> + * this program.  If not, see<http://www.gnu.org/licenses/>.
> + */
> +
> +#include<linux/fs.h>
> +#include<linux/slab.h>
> +#include<linux/dma-buf.h>
> +#include<linux/anon_inodes.h>
> +
> +static inline int is_dma_buf_file(struct file *);
> +
> +static int dma_buf_mmap(struct file *file, struct vm_area_struct *vma)
> +{
> +	struct dma_buf *dmabuf;
> +
> +	if (!is_dma_buf_file(file))
> +		return -EINVAL;
> +
> +	dmabuf = file->private_data;
> +
> +	if (!dmabuf->ops->mmap)
> +		return -EINVAL;
> +
> +	return dmabuf->ops->mmap(dmabuf, vma);
> +}
> +
> +static int dma_buf_release(struct inode *inode, struct file *file)
> +{
> +	struct dma_buf *dmabuf;
> +
> +	if (!is_dma_buf_file(file))
> +		return -EINVAL;
> +
> +	dmabuf = file->private_data;
> +
> +	dmabuf->ops->release(dmabuf);
> +	kfree(dmabuf);
> +	return 0;
> +}
> +
> +static const struct file_operations dma_buf_fops = {
> +	.mmap		= dma_buf_mmap,
> +	.release	= dma_buf_release,
> +};
> +
> +/*
> + * is_dma_buf_file - Check if struct file* is associated with dma_buf
> + */
> +static inline int is_dma_buf_file(struct file *file)
> +{
> +	return file->f_op ==&dma_buf_fops;
> +}
> +
> +/**
> + * dma_buf_export - Creates a new dma_buf, and associates an anon file
> + * with this buffer,so it can be exported.
> + * Also connect the allocator specific data and ops to the buffer.
> + *
> + * @priv:	[in]	Attach private data of allocator to this buffer
> + * @ops:	[in]	Attach allocator-defined dma buf ops to the new buffer.
> + * @flags:	[in]	mode flags for the file.

What is the purpose of these flags? The file is not visible to any 
process by any file system, is it?

> + *
> + * Returns, on success, a newly created dma_buf object, which wraps the
> + * supplied private data and operations for dma_buf_ops. On failure to
> + * allocate the dma_buf object, it can return NULL.
> + *
> + */
> +struct dma_buf *dma_buf_export(void *priv, struct dma_buf_ops *ops,
> +				int flags)
> +{
> +	struct dma_buf *dmabuf;
> +	struct file *file;
> +

why priv is not allowed to be NULL?

> +	BUG_ON(!priv || !ops);
> +
> +	dmabuf = kzalloc(sizeof(struct dma_buf), GFP_KERNEL);
> +	if (dmabuf == NULL)
> +		return dmabuf;
> +
> +	dmabuf->priv = priv;
> +	dmabuf->ops = ops;
> +
> +	file = anon_inode_getfile("dmabuf",&dma_buf_fops, dmabuf, flags);
> +
> +	dmabuf->file = file;
> +
> +	mutex_init(&dmabuf->lock);
> +	INIT_LIST_HEAD(&dmabuf->attachments);
> +
> +	return dmabuf;
> +}
> +EXPORT_SYMBOL(dma_buf_export);
> +
> +
> +/**
> + * dma_buf_fd - returns a file descriptor for the given dma_buf
> + * @dmabuf:	[in]	pointer to dma_buf for which fd is required.
> + *
> + * On success, returns an associated 'fd'. Else, returns error.
> + */
> +int dma_buf_fd(struct dma_buf *dmabuf)
> +{
> +	int error, fd;
> +
> +	if (!dmabuf->file)
> +		return -EINVAL;
> +
> +	error = get_unused_fd_flags(0);
> +	if (error<  0)
> +		return error;
> +	fd = error;
> +
> +	fd_install(fd, dmabuf->file);
> +
> +	return fd;
> +}
> +EXPORT_SYMBOL(dma_buf_fd);
> +
> +/**
> + * dma_buf_get - returns the dma_buf structure related to an fd
> + * @fd:	[in]	fd associated with the dma_buf to be returned
> + *
> + * On success, returns the dma_buf structure associated with an fd; uses
> + * file's refcounting done by fget to increase refcount. returns ERR_PTR
> + * otherwise.
> + */
> +struct dma_buf *dma_buf_get(int fd)
> +{
> +	struct file *file;
> +
> +	file = fget(fd);
> +
> +	if (!file)
> +		return ERR_PTR(-EBADF);
> +
> +	if (!is_dma_buf_file(file)) {
> +		fput(file);
> +		return ERR_PTR(-EINVAL);
> +	}
> +
> +	return file->private_data;
> +}
> +EXPORT_SYMBOL(dma_buf_get);
> +
> +/**
> + * dma_buf_put - decreases refcount of the buffer
> + * @dmabuf:	[in]	buffer to reduce refcount of
> + *
> + * Uses file's refcounting done implicitly by fput()
> + */
> +void dma_buf_put(struct dma_buf *dmabuf)
> +{
> +	BUG_ON(!dmabuf->file);
> +
> +	fput(dmabuf->file);
> +

return is not needed

> +	return;
> +}
> +EXPORT_SYMBOL(dma_buf_put);
> +
> +/**
> + * dma_buf_attach - Add the device to dma_buf's attachments list; optionally,
> + * calls attach() of dma_buf_ops to allow device-specific attach functionality
> + * @dmabuf:	[in]	buffer to attach device to.
> + * @dev:	[in]	device to be attached.
> + *
> + * Returns struct dma_buf_attachment * for this attachment; may return NULL.
> + *
> + */
> +struct dma_buf_attachment *dma_buf_attach(struct dma_buf *dmabuf,
> +						struct device *dev)
> +{
> +	struct dma_buf_attachment *attach;
> +	int ret;
> +
> +	BUG_ON(!dmabuf || !dev);
> +
> +	mutex_lock(&dmabuf->lock);
> +

There is no need to call kzalloc inside critical section protected by 
dmabuf->lock. The code would be simpler if the allocation is moved 
outside the section.

> +	attach = kzalloc(sizeof(struct dma_buf_attachment), GFP_KERNEL);
> +	if (attach == NULL)
> +		goto err_alloc;
> +
> +	attach->dev = dev;
> +	if (dmabuf->ops->attach) {
> +		ret = dmabuf->ops->attach(dmabuf, dev, attach);
> +		if (!ret)
> +			goto err_attach;
> +	}
> +	list_add(&attach->node,&dmabuf->attachments);
> +
> +err_alloc:
> +	mutex_unlock(&dmabuf->lock);
> +	return attach;
> +err_attach:
> +	kfree(attach);
> +	mutex_unlock(&dmabuf->lock);
> +	return ERR_PTR(ret);
> +}
> +EXPORT_SYMBOL(dma_buf_attach);
> +
> +/**
> + * dma_buf_detach - Remove the given attachment from dmabuf's attachments list;
> + * optionally calls detach() of dma_buf_ops for device-specific detach
> + * @dmabuf:	[in]	buffer to detach from.
> + * @attach:	[in]	attachment to be detached; is free'd after this call.
> + *
> + */
> +void dma_buf_detach(struct dma_buf *dmabuf, struct dma_buf_attachment *attach)
> +{
> +	BUG_ON(!dmabuf || !attach);
> +
> +	mutex_lock(&dmabuf->lock);
> +	list_del(&attach->node);
> +	if (dmabuf->ops->detach)
> +		dmabuf->ops->detach(dmabuf, attach);
> +
[as above]
> +	kfree(attach);
> +	mutex_unlock(&dmabuf->lock);
> +	return;
> +}
> +EXPORT_SYMBOL(dma_buf_detach);
> diff --git a/include/linux/dma-buf.h b/include/linux/dma-buf.h
> new file mode 100644
> index 0000000..5bdf16a
> --- /dev/null
> +++ b/include/linux/dma-buf.h
> @@ -0,0 +1,162 @@
> +/*
> + * Header file for dma buffer sharing framework.
> + *
> + * Copyright(C) 2011 Linaro Limited. All rights reserved.
> + * Author: Sumit Semwal<sumit.semwal@ti.com>
> + *
> + * Many thanks to linaro-mm-sig list, and specially
> + * Arnd Bergmann<arnd@arndb.de>, Rob Clark<rob@ti.com>  and
> + * Daniel Vetter<daniel@ffwll.ch>  for their support in creation and
> + * refining of this idea.
> + *
> + * This program is free software; you can redistribute it and/or modify it
> + * under the terms of the GNU General Public License version 2 as published by
> + * the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful, but WITHOUT
> + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
> + * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
> + * more details.
> + *
> + * You should have received a copy of the GNU General Public License along with
> + * this program.  If not, see<http://www.gnu.org/licenses/>.
> + */
> +#ifndef __DMA_BUF_H__
> +#define __DMA_BUF_H__
> +
> +#include<linux/file.h>
> +#include<linux/err.h>
> +#include<linux/device.h>
> +#include<linux/scatterlist.h>
> +#include<linux/list.h>
> +#include<linux/dma-mapping.h>
> +
> +struct dma_buf;
> +
> +/**
> + * struct dma_buf_attachment - holds device-buffer attachment data
> + * @dmabuf: buffer for this attachment.
> + * @dev: device attached to the buffer.
> + * @node: list_head to allow manipulation of list of dma_buf_attachment.
> + * @priv: exporter-specific attachment data.
> + */
> +struct dma_buf_attachment {
> +	struct dma_buf *dmabuf;
> +	struct device *dev;
> +	struct list_head node;
> +	void *priv;
> +};
> +
> +/**
> + * struct dma_buf_ops - operations possible on struct dma_buf
> + * @create: creates a struct dma_buf of a fixed size. Actual allocation
> + *	    does not happen here.

The 'create' ops is not present in dma_buf_ops.

> + * @attach: allows different devices to 'attach' themselves to the given
> + *	    buffer. It might return -EBUSY to signal that backing storage
> + *	    is already allocated and incompatible with the requirements
> + *	    of requesting device. [optional]
> + * @detach: detach a given device from this buffer. [optional]
> + * @get_scatterlist: returns list of scatter pages allocated, increases
> + *		     usecount of the buffer. Requires atleast one attach to be
> + *		     called before. Returned sg list should already be mapped
> + *		     into _device_ address space.

You must add a comment that this call 'may sleep'.

I like the get_scatterlist idea. It allows the exported to create a 
valid scatterlist for a client in a elegant way.

I do not like this whole attachment idea. The problem is that currently 
there is no support in DMA framework for allocation for multiple 
devices. As long as no such a support exists, there is no generic way to 
handle attribute negotiations and buffer allocations that involve 
multiple devices. So the exporter drivers would have to implement more 
or less hacky solutions to handle memory requirements and choosing the 
device that allocated memory.

Currently, AFAIK there is even no generic way for a driver to acquire 
its own DMA memory requirements.

Therefore all logic hidden beneath 'attachment' is pointless. I think 
that support for attach/detach (and related stuff) should be postponed 
until support for multi-device allocation is added to DMA framework.

I don't say the attachment list idea is wrong but adding attachment 
stuff creates an illusion that problem of multi-device allocations is 
somehow magically solved. We should not force the developers of exporter 
drivers to solve the problem that is not solvable yet.

The other problem are the APIs. For example, the V4L2 subsystem assumes 
that memory is allocated after successful VIDIOC_REQBUFS with 
V4L2_MEMORY_MMAP memory type. Therefore attach would be automatically 
followed by get_scatterlist, blocking possibility of any buffer 
migrations in future.

The same situation happens if buffer sharing is added to framebuffer API.

The buffer sharing mechanism is dedicated to improve cooperation between 
multiple APIs. Therefore the common denominator strategy should be 
applied that is buffer-creation == buffer-allocation.

> + * @put_scatterlist: decreases usecount of buffer, might deallocate scatter
> + *		     pages.
> + * @mmap: memory map this buffer - optional.
> + * @release: release this buffer; to be called after the last dma_buf_put.
> + * @sync_sg_for_cpu: sync the sg list for cpu.
> + * @sync_sg_for_device: synch the sg list for device.
> + */
> +struct dma_buf_ops {
> +	int (*attach)(struct dma_buf *, struct device *,
> +			struct dma_buf_attachment *);
> +
> +	void (*detach)(struct dma_buf *, struct dma_buf_attachment *);
> +
> +	/* For {get,put}_scatterlist below, any specific buffer attributes
> +	 * required should get added to device_dma_parameters accessible
> +	 * via dev->dma_params.
> +	 */
> +	struct scatterlist * (*get_scatterlist)(struct dma_buf_attachment *,
> +						enum dma_data_direction,
> +						int *nents);
> +	void (*put_scatterlist)(struct dma_buf_attachment *,
> +						struct scatterlist *,
> +						int nents);
> +	/* TODO: Add interruptible and interruptible_timeout versions */

I don't agree the interruptible and interruptible_timeout versions are 
needed. I think that get_scatterlist should alway be interruptible. You 
can add try_get_scatterlist callback that returns ERR_PTR(-EBUSY) if the 
call would be blocking.

> +
> +	/* allow mmap optionally for devices that need it */
> +	int (*mmap)(struct dma_buf *, struct vm_area_struct *);

The mmap is not needed for inital version. It could be added at any time 
in the future. The dmabuf client should not be allowed to create mapping 
of the dmabuf from the scatterlist.

> +	/* after final dma_buf_put() */
> +	void (*release)(struct dma_buf *);
> +
> +	/* allow allocator to take care of cache ops */
> +	void (*sync_sg_for_cpu) (struct dma_buf *, struct device *);
> +	void (*sync_sg_for_device)(struct dma_buf *, struct device *);
> +};
> +
> +/**
> + * struct dma_buf - shared buffer object
> + * @file: file pointer used for sharing buffers across, and for refcounting.
> + * @attachments: list of dma_buf_attachment that denotes all devices attached.
> + * @ops: dma_buf_ops associated with this buffer object
> + * @priv: user specific private data
> + */
> +struct dma_buf {
> +	size_t size;
> +	struct file *file;
> +	struct list_head attachments;
> +	const struct dma_buf_ops *ops;
> +	/* mutex to serialize list manipulation and other ops */
> +	struct mutex lock;
> +	void *priv;
> +};
> +
> +#ifdef CONFIG_DMA_SHARED_BUFFER
> +struct dma_buf_attachment *dma_buf_attach(struct dma_buf *dmabuf,
> +							struct device *dev);
> +void dma_buf_detach(struct dma_buf *dmabuf,
> +				struct dma_buf_attachment *dmabuf_attach);
> +struct dma_buf *dma_buf_export(void *priv, struct dma_buf_ops *ops, int flags);
> +int dma_buf_fd(struct dma_buf *dmabuf);
> +struct dma_buf *dma_buf_get(int fd);
> +void dma_buf_put(struct dma_buf *dmabuf);
> +
> +#else
> +
> +static inline struct dma_buf_attachment *dma_buf_attach(struct dma_buf *dmabuf,
> +							struct device *dev)
> +{
> +	return ERR_PTR(-ENODEV);
> +}
> +
> +static inline void dma_buf_detach(struct dma_buf *dmabuf,
> +				  struct dma_buf_attachment *dmabuf_attach)
> +{
> +	return;
> +}
> +
> +static inline struct dma_buf *dma_buf_export(void *priv,
> +						struct dma_buf_ops *ops,
> +						int flags)
> +{
> +	return ERR_PTR(-ENODEV);
> +}
> +
> +static inline int dma_buf_fd(struct dma_buf *dmabuf)
> +{
> +	return -ENODEV;
> +}
> +
> +static inline struct dma_buf *dma_buf_get(int fd)
> +{
> +	return ERR_PTR(-ENODEV);
> +}
> +
> +static inline void dma_buf_put(struct dma_buf *dmabuf)
> +{
> +	return;
> +}
> +#endif /* CONFIG_DMA_SHARED_BUFFER */
> +
> +#endif /* __DMA_BUF_H__ */

I hope you find my comments useful.

Yours sincerely,
Tomasz Stanislawski
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Sumit Semwal Oct. 14, 2011, 2:13 p.m. UTC | #11
On 14 October 2011 15:30, Tomasz Stanislawski <t.stanislaws@samsung.com> wrote:
> Hi Mr. Sumit Semwal,
Hello Mr. Tomasz Stanislawski :),

> Thank you for taking care of the framework for buffer sharing.
> The support of buffer sharing in V4L2, both exporting and importing was
> posted in shrbuf proof-of-concept patch. It should be easy to port it to
> dmabuf.
>
> http://lists.linaro.org/pipermail/linaro-mm-sig/2011-August/000485.html
I should thank you for the wonderful proof-of-concept patch, and the
idea behind it! I am currently working on the V4L2 side patch for it,
and would send out the RFC soon.
Also, thanks for a good review and some pertinent points; replies inline.
>
> Please refer to the comments below:
>
> On 10/11/2011 11:23 AM, Sumit Semwal wrote:
>>
<snip>
>> +/**
>> + * dma_buf_export - Creates a new dma_buf, and associates an anon file
>> + * with this buffer,so it can be exported.
>> + * Also connect the allocator specific data and ops to the buffer.
>> + *
>> + * @priv:      [in]    Attach private data of allocator to this buffer
>> + * @ops:       [in]    Attach allocator-defined dma buf ops to the new
>> buffer.
>> + * @flags:     [in]    mode flags for the file.
>
> What is the purpose of these flags? The file is not visible to any process
> by any file system, is it?
These are the standard file mode flags, which can be used to do
file-level access-type control by the exporter, so for example
write-access can be denied for  a buffer exported as a read-only
buffer.
>
>> + *
>> + * Returns, on success, a newly created dma_buf object, which wraps the
>> + * supplied private data and operations for dma_buf_ops. On failure to
>> + * allocate the dma_buf object, it can return NULL.
>> + *
>> + */
>> +struct dma_buf *dma_buf_export(void *priv, struct dma_buf_ops *ops,
>> +                               int flags)
>> +{
>> +       struct dma_buf *dmabuf;
>> +       struct file *file;
>> +
>
> why priv is not allowed to be NULL?
priv will be used by the exporter to attach its own context to the dma
buf; I couldn't think of any case where it could be NULL?
>
>> +       BUG_ON(!priv || !ops);
>> +
<snip>
>> +       BUG_ON(!dmabuf->file);
>> +
>> +       fput(dmabuf->file);
>> +
>
> return is not needed
Right; will correct this.
>
>> +       return;
>> +}
>> +EXPORT_SYMBOL(dma_buf_put);
>> +
>> +/**
>> + * dma_buf_attach - Add the device to dma_buf's attachments list;
>> optionally,
>> + * calls attach() of dma_buf_ops to allow device-specific attach
>> functionality
>> + * @dmabuf:    [in]    buffer to attach device to.
>> + * @dev:       [in]    device to be attached.
>> + *
>> + * Returns struct dma_buf_attachment * for this attachment; may return
>> NULL.
>> + *
>> + */
>> +struct dma_buf_attachment *dma_buf_attach(struct dma_buf *dmabuf,
>> +                                               struct device *dev)
>> +{
>> +       struct dma_buf_attachment *attach;
>> +       int ret;
>> +
>> +       BUG_ON(!dmabuf || !dev);
>> +
>> +       mutex_lock(&dmabuf->lock);
>> +
>
> There is no need to call kzalloc inside critical section protected by
> dmabuf->lock. The code would be simpler if the allocation is moved outside
> the section.
Yes, you're right; will correct it.
>
>> +       attach = kzalloc(sizeof(struct dma_buf_attachment), GFP_KERNEL);
>> +       if (attach == NULL)
>> +               goto err_alloc;
>> +
>> +       attach->dev = dev;
>> +       if (dmabuf->ops->attach) {
>> +               ret = dmabuf->ops->attach(dmabuf, dev, attach);
>> +               if (!ret)
>> +                       goto err_attach;
>> +       }
>> +       list_add(&attach->node,&dmabuf->attachments);
>> +
>> +err_alloc:
>> +       mutex_unlock(&dmabuf->lock);
>> +       return attach;
>> +err_attach:
>> +       kfree(attach);
>> +       mutex_unlock(&dmabuf->lock);
>> +       return ERR_PTR(ret);
>> +}
>> +EXPORT_SYMBOL(dma_buf_attach);
>> +
>> +/**
>> + * dma_buf_detach - Remove the given attachment from dmabuf's attachments
>> list;
>> + * optionally calls detach() of dma_buf_ops for device-specific detach
>> + * @dmabuf:    [in]    buffer to detach from.
>> + * @attach:    [in]    attachment to be detached; is free'd after this
>> call.
>> + *
>> + */
>> +void dma_buf_detach(struct dma_buf *dmabuf, struct dma_buf_attachment
>> *attach)
>> +{
>> +       BUG_ON(!dmabuf || !attach);
>> +
>> +       mutex_lock(&dmabuf->lock);
>> +       list_del(&attach->node);
>> +       if (dmabuf->ops->detach)
>> +               dmabuf->ops->detach(dmabuf, attach);
>> +
>
> [as above]
Ok.
>>
>> +       kfree(attach);
>> +       mutex_unlock(&dmabuf->lock);
>> +       return;
>> +}
>> +EXPORT_SYMBOL(dma_buf_detach);
>> diff --git a/include/linux/dma-buf.h b/include/linux/dma-buf.h
>> new file mode 100644
>> index 0000000..5bdf16a
>> --- /dev/null
>> +++ b/include/linux/dma-buf.h
>> @@ -0,0 +1,162 @@
>> +/*
>> + * Header file for dma buffer sharing framework.
>> + *
>> + * Copyright(C) 2011 Linaro Limited. All rights reserved.
>> + * Author: Sumit Semwal<sumit.semwal@ti.com>
>> + *
>> + * Many thanks to linaro-mm-sig list, and specially
>> + * Arnd Bergmann<arnd@arndb.de>, Rob Clark<rob@ti.com>  and
>> + * Daniel Vetter<daniel@ffwll.ch>  for their support in creation and
>> + * refining of this idea.
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> it
>> + * under the terms of the GNU General Public License version 2 as
>> published by
>> + * the Free Software Foundation.
>> + *
>> + * This program is distributed in the hope that it will be useful, but
>> WITHOUT
>> + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
>> + * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
>> for
>> + * more details.
>> + *
>> + * You should have received a copy of the GNU General Public License
>> along with
>> + * this program.  If not, see<http://www.gnu.org/licenses/>.
>> + */
>> +#ifndef __DMA_BUF_H__
>> +#define __DMA_BUF_H__
>> +
>> +#include<linux/file.h>
>> +#include<linux/err.h>
>> +#include<linux/device.h>
>> +#include<linux/scatterlist.h>
>> +#include<linux/list.h>
>> +#include<linux/dma-mapping.h>
>> +
>> +struct dma_buf;
>> +
>> +/**
>> + * struct dma_buf_attachment - holds device-buffer attachment data
>> + * @dmabuf: buffer for this attachment.
>> + * @dev: device attached to the buffer.
>> + * @node: list_head to allow manipulation of list of dma_buf_attachment.
>> + * @priv: exporter-specific attachment data.
>> + */
>> +struct dma_buf_attachment {
>> +       struct dma_buf *dmabuf;
>> +       struct device *dev;
>> +       struct list_head node;
>> +       void *priv;
>> +};
>> +
>> +/**
>> + * struct dma_buf_ops - operations possible on struct dma_buf
>> + * @create: creates a struct dma_buf of a fixed size. Actual allocation
>> + *         does not happen here.
>
> The 'create' ops is not present in dma_buf_ops.
Yes, this is a copy-paste mistake; will correct.
>> + * @attach: allows different devices to 'attach' themselves to the given
>> + *         buffer. It might return -EBUSY to signal that backing storage
>> + *         is already allocated and incompatible with the requirements
>> + *         of requesting device. [optional]
>> + * @detach: detach a given device from this buffer. [optional]
>> + * @get_scatterlist: returns list of scatter pages allocated, increases
>> + *                  usecount of the buffer. Requires atleast one attach
>> to be
>> + *                  called before. Returned sg list should already be
>> mapped
>> + *                  into _device_ address space.
>
> You must add a comment that this call 'may sleep'.
Thanks, I will add this.
>
> I like the get_scatterlist idea. It allows the exported to create a valid
> scatterlist for a client in a elegant way.
>
> I do not like this whole attachment idea. The problem is that currently
> there is no support in DMA framework for allocation for multiple devices. As
> long as no such a support exists, there is no generic way to handle
> attribute negotiations and buffer allocations that involve multiple devices.
> So the exporter drivers would have to implement more or less hacky solutions
> to handle memory requirements and choosing the device that allocated memory.

You are right, there is currently no generic way for attribute
negotiation for buffers of multiple devices. However, since buffer
sharing idea itself is about allowing for buffer allocation and
migration based on 'some' negotiation, the attachment idea makes it
easier to do so.

What you choose to call 'more or less hacky solution', I would call
'platform-specific mechanism' for negotiation, allocation (and
migration if possible and required). :)
>
> Currently, AFAIK there is even no generic way for a driver to acquire its
> own DMA memory requirements.
>
> Therefore all logic hidden beneath 'attachment' is pointless. I think that
> support for attach/detach (and related stuff) should be postponed until
> support for multi-device allocation is added to DMA framework.
>
> I don't say the attachment list idea is wrong but adding attachment stuff
> creates an illusion that problem of multi-device allocations is somehow
> magically solved. We should not force the developers of exporter drivers to
> solve the problem that is not solvable yet.

I quite like what you said - the problem of having a 'generic'
mechanism of DMA attribute negotiation might not have been solved yet,
and so the exporter drivers shouldn't need to try to do that.
However, exporter drivers would most likely know about the negotiation
etc needed and possible on their given platform - and the attachment
mechanism gives them a way to gather requirements from participating
DMA users, and use it for such decisions.

>
> The other problem are the APIs. For example, the V4L2 subsystem assumes that
> memory is allocated after successful VIDIOC_REQBUFS with V4L2_MEMORY_MMAP
> memory type. Therefore attach would be automatically followed by
> get_scatterlist, blocking possibility of any buffer migrations in future.
That is correct for V4L2_MEMORY_MMAP; should that need to be a
requirement even for V4L2_MEMORY_DMABUF_USER memory type? [I am
deliberately distinguishing a V4L2 driver which is a shared buffer
exporter from a V4L2 shared-buffer-user driver] - could we not do a
get_scatterlist on each VIDIOC_QBUF, and put_scatterlist on each
VIDIOC_DQBUF?
>
> The same situation happens if buffer sharing is added to framebuffer API.
>
> The buffer sharing mechanism is dedicated to improve cooperation between
> multiple APIs. Therefore the common denominator strategy should be applied
> that is buffer-creation == buffer-allocation.

It might be prudent to do so in the media drivers. But the buffer
sharing mechanism is also useful for other, non-media usecases? I
would say that we should be able to leave enough flexibility.
>
>> + * @put_scatterlist: decreases usecount of buffer, might deallocate
>> scatter
>> + *                  pages.
>> + * @mmap: memory map this buffer - optional.
>> + * @release: release this buffer; to be called after the last
>> dma_buf_put.
>> + * @sync_sg_for_cpu: sync the sg list for cpu.
>> + * @sync_sg_for_device: synch the sg list for device.
>> + */
>> +struct dma_buf_ops {
>> +       int (*attach)(struct dma_buf *, struct device *,
>> +                       struct dma_buf_attachment *);
>> +
>> +       void (*detach)(struct dma_buf *, struct dma_buf_attachment *);
>> +
>> +       /* For {get,put}_scatterlist below, any specific buffer attributes
>> +        * required should get added to device_dma_parameters accessible
>> +        * via dev->dma_params.
>> +        */
>> +       struct scatterlist * (*get_scatterlist)(struct dma_buf_attachment
>> *,
>> +                                               enum dma_data_direction,
>> +                                               int *nents);
>> +       void (*put_scatterlist)(struct dma_buf_attachment *,
>> +                                               struct scatterlist *,
>> +                                               int nents);
>> +       /* TODO: Add interruptible and interruptible_timeout versions */
>
> I don't agree the interruptible and interruptible_timeout versions are
> needed. I think that get_scatterlist should alway be interruptible. You can
> add try_get_scatterlist callback that returns ERR_PTR(-EBUSY) if the call
> would be blocking.
Sure, that's a good idea - I could do that.
>
>> +
>> +       /* allow mmap optionally for devices that need it */
>> +       int (*mmap)(struct dma_buf *, struct vm_area_struct *);
>
> The mmap is not needed for inital version. It could be added at any time in
> the future. The dmabuf client should not be allowed to create mapping of the
> dmabuf from the scatterlist.
There's already a discussion between Rob and David on this mail-chain
earlier; I guess we'll need to wait a little before deciding to remove
this.
>
<snip>
>
> I hope you find my comments useful.
Like I said Tomasz, I thank you for a very good review! Hope my
replies satisfy some of the concerns you raised?
>
> Yours sincerely,
> Tomasz Stanislawski
>
  
Clark, Rob Oct. 14, 2011, 3:34 p.m. UTC | #12
On Fri, Oct 14, 2011 at 5:00 AM, Tomasz Stanislawski
<t.stanislaws@samsung.com> wrote:
>> + * @attach: allows different devices to 'attach' themselves to the given
>> + *         buffer. It might return -EBUSY to signal that backing storage
>> + *         is already allocated and incompatible with the requirements
>> + *         of requesting device. [optional]
>> + * @detach: detach a given device from this buffer. [optional]
>> + * @get_scatterlist: returns list of scatter pages allocated, increases
>> + *                  usecount of the buffer. Requires atleast one attach
>> to be
>> + *                  called before. Returned sg list should already be
>> mapped
>> + *                  into _device_ address space.
>
> You must add a comment that this call 'may sleep'.
>
> I like the get_scatterlist idea. It allows the exported to create a valid
> scatterlist for a client in a elegant way.
>
> I do not like this whole attachment idea. The problem is that currently
> there is no support in DMA framework for allocation for multiple devices. As
> long as no such a support exists, there is no generic way to handle
> attribute negotiations and buffer allocations that involve multiple devices.
> So the exporter drivers would have to implement more or less hacky solutions
> to handle memory requirements and choosing the device that allocated memory.
>
> Currently, AFAIK there is even no generic way for a driver to acquire its
> own DMA memory requirements.

dev->dma_params (struct device_dma_parameters).. for example

it would need to be expanded a bit to have a way to say "it needs to
be physically contiguous"..


> Therefore all logic hidden beneath 'attachment' is pointless. I think that
> support for attach/detach (and related stuff) should be postponed until
> support for multi-device allocation is added to DMA framework.
>
> I don't say the attachment list idea is wrong but adding attachment stuff
> creates an illusion that problem of multi-device allocations is somehow
> magically solved. We should not force the developers of exporter drivers to
> solve the problem that is not solvable yet.
>
> The other problem are the APIs. For example, the V4L2 subsystem assumes that
> memory is allocated after successful VIDIOC_REQBUFS with V4L2_MEMORY_MMAP
> memory type. Therefore attach would be automatically followed by
> get_scatterlist, blocking possibility of any buffer migrations in future.

But this problem really only applies if v4l is your buffer allocator.
I don't think a v4l limitation is a valid argument to remove the
attachment stuff.

> The same situation happens if buffer sharing is added to framebuffer API.
>
> The buffer sharing mechanism is dedicated to improve cooperation between
> multiple APIs. Therefore the common denominator strategy should be applied
> that is buffer-creation == buffer-allocation.

I think it would be sufficient if buffer creators that cannot defer
the allocation just take a worst-case approach and allocate physically
contiguous buffers.  No need to penalize other potential buffer
allocators.  This allows buffer creators with more flexibility the
option for deferring the allocation until it knows whether the buffer
really needs to be contiguous.

>> + * @put_scatterlist: decreases usecount of buffer, might deallocate
>> scatter
>> + *                  pages.
>> + * @mmap: memory map this buffer - optional.
>> + * @release: release this buffer; to be called after the last
>> dma_buf_put.
>> + * @sync_sg_for_cpu: sync the sg list for cpu.
>> + * @sync_sg_for_device: synch the sg list for device.
>> + */
>> +struct dma_buf_ops {
>> +       int (*attach)(struct dma_buf *, struct device *,
>> +                       struct dma_buf_attachment *);
>> +
>> +       void (*detach)(struct dma_buf *, struct dma_buf_attachment *);
>> +
>> +       /* For {get,put}_scatterlist below, any specific buffer attributes
>> +        * required should get added to device_dma_parameters accessible
>> +        * via dev->dma_params.
>> +        */
>> +       struct scatterlist * (*get_scatterlist)(struct dma_buf_attachment
>> *,
>> +                                               enum dma_data_direction,
>> +                                               int *nents);
>> +       void (*put_scatterlist)(struct dma_buf_attachment *,
>> +                                               struct scatterlist *,
>> +                                               int nents);
>> +       /* TODO: Add interruptible and interruptible_timeout versions */
>
> I don't agree the interruptible and interruptible_timeout versions are
> needed. I think that get_scatterlist should alway be interruptible. You can
> add try_get_scatterlist callback that returns ERR_PTR(-EBUSY) if the call
> would be blocking.
>
>> +
>> +       /* allow mmap optionally for devices that need it */
>> +       int (*mmap)(struct dma_buf *, struct vm_area_struct *);
>
> The mmap is not needed for inital version. It could be added at any time in
> the future. The dmabuf client should not be allowed to create mapping of the
> dmabuf from the scatterlist.

fwiw, this wasn't intended for allowing the client to create the
mapping.. the intention was that the buffer creator always be the one
that implements the mmap'ing.  This was just to implement fops->mmap()
for the dmabuf handle, ie. so userspace could mmap the buffer without
having to know *who* allocated it.  Otherwise you have to also pass
around the fd of the allocator and an offset.

BR,
-R
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Daniel Vetter Oct. 14, 2011, 3:35 p.m. UTC | #13
On Fri, Oct 14, 2011 at 12:00:58PM +0200, Tomasz Stanislawski wrote:
> >+/**
> >+ * struct dma_buf_ops - operations possible on struct dma_buf
> >+ * @create: creates a struct dma_buf of a fixed size. Actual allocation
> >+ *	    does not happen here.
> 
> The 'create' ops is not present in dma_buf_ops.
> 
> >+ * @attach: allows different devices to 'attach' themselves to the given
> >+ *	    buffer. It might return -EBUSY to signal that backing storage
> >+ *	    is already allocated and incompatible with the requirements
> >+ *	    of requesting device. [optional]
> >+ * @detach: detach a given device from this buffer. [optional]
> >+ * @get_scatterlist: returns list of scatter pages allocated, increases
> >+ *		     usecount of the buffer. Requires atleast one attach to be
> >+ *		     called before. Returned sg list should already be mapped
> >+ *		     into _device_ address space.
> 
> You must add a comment that this call 'may sleep'.
> 
> I like the get_scatterlist idea. It allows the exported to create a
> valid scatterlist for a client in a elegant way.
> 
> I do not like this whole attachment idea. The problem is that
> currently there is no support in DMA framework for allocation for
> multiple devices. As long as no such a support exists, there is no
> generic way to handle attribute negotiations and buffer allocations
> that involve multiple devices. So the exporter drivers would have to
> implement more or less hacky solutions to handle memory requirements
> and choosing the device that allocated memory.
> 
> Currently, AFAIK there is even no generic way for a driver to
> acquire its own DMA memory requirements.
> 
> Therefore all logic hidden beneath 'attachment' is pointless. I
> think that support for attach/detach (and related stuff) should be
> postponed until support for multi-device allocation is added to DMA
> framework.

Imo we clearly need this to make the multi-device-driver with insane dma
requirements work on arm. And rewriting the buffer handling in
participating subsystem twice isn't really a great plan. I envision that
on platforms where we need this madness, the driver must call back to the
dma subsytem to create a dma_buf. The dma subsytem should be already aware
of all the requirements and hence should be able to handle them..

> I don't say the attachment list idea is wrong but adding attachment
> stuff creates an illusion that problem of multi-device allocations
> is somehow magically solved. We should not force the developers of
> exporter drivers to solve the problem that is not solvable yet.

Well, this is why we need to create a decent support infrastructure for
platforms (= arm madness) that needs this, so that device drivers and
subsystem don't need to invent that wheel on their own. Which as you point
out, they actually can't.

> The other problem are the APIs. For example, the V4L2 subsystem
> assumes that memory is allocated after successful VIDIOC_REQBUFS
> with V4L2_MEMORY_MMAP memory type. Therefore attach would be
> automatically followed by get_scatterlist, blocking possibility of
> any buffer migrations in future.

Well, pardon to break the news, but v4l needs to rework the buffer
handling. If you want to share buffers with a gpu driver, you _have_ to
life with the fact that gpus do fully dynamic buffer management, meaning:
- buffers get allocated and destroyed on the fly, meaning static reqbuf
  just went out the window (we obviously cache buffer objects and reuse
  them for performance, as long as the processing pipeline doesn't really
  change).
- buffers get moved around in memory, meaning you either need full-blown
  sync-objects with a callback to drivers to tear-down mappings on-demand,
  or every driver needs to guarnatee to call put_scatterlist in a
  reasonable short time. The latter is probably the more natural thing for
  v4l devices.

> The same situation happens if buffer sharing is added to framebuffer API.

You can fix that by using the gem/ttm infrastructure of drm (or whatever
the blob gpu drivers are using). Which is why I think fb should just die,
please.

> The buffer sharing mechanism is dedicated to improve cooperation
> between multiple APIs. Therefore the common denominator strategy
> should be applied that is buffer-creation == buffer-allocation.

No.

Really, there's just no way gpu's will be moving back to static buffer
management. And I know, for many use-cases we could get away with a bunch
of static buffers (e.g. a video processing pipe). But in drm-land even
scanout-buffers can get moved around - currently only when they're not
being used, but strictly speaking nothing prevents us from copying the
scanout to a new location and issueing a pageflip and so even move the
buffer around even when it's in use.

But let's look quickly at an OpenCL usecase, moving Gb's of date per
second around between the cpu and a bunch of add-on gpus (or other special
purpose processing units). We'd also need to extend dma_buf with sync
objects to make this work well, but there's simply no way this is gonna
work with statically allocated objects.

Also, gem buffer objects that are currently unused can be swapped out.

Cheers, Daniel
  
Marek Szyprowski Nov. 3, 2011, 8:04 a.m. UTC | #14
Hello,

I'm sorry for a late reply, but after Kernel Summit/ELC I have some comments.

On Friday, October 14, 2011 5:35 PM Daniel Vetter wrote:

> On Fri, Oct 14, 2011 at 12:00:58PM +0200, Tomasz Stanislawski wrote:
> > >+/**
> > >+ * struct dma_buf_ops - operations possible on struct dma_buf
> > >+ * @create: creates a struct dma_buf of a fixed size. Actual allocation
> > >+ *	    does not happen here.
> >
> > The 'create' ops is not present in dma_buf_ops.
> >
> > >+ * @attach: allows different devices to 'attach' themselves to the given
> > >+ *	    buffer. It might return -EBUSY to signal that backing storage
> > >+ *	    is already allocated and incompatible with the requirements
> > >+ *	    of requesting device. [optional]
> > >+ * @detach: detach a given device from this buffer. [optional]
> > >+ * @get_scatterlist: returns list of scatter pages allocated, increases
> > >+ *		     usecount of the buffer. Requires atleast one attach to be
> > >+ *		     called before. Returned sg list should already be mapped
> > >+ *		     into _device_ address space.
> >
> > You must add a comment that this call 'may sleep'.
> >
> > I like the get_scatterlist idea. It allows the exported to create a
> > valid scatterlist for a client in a elegant way.
> >
> > I do not like this whole attachment idea. The problem is that
> > currently there is no support in DMA framework for allocation for
> > multiple devices. As long as no such a support exists, there is no
> > generic way to handle attribute negotiations and buffer allocations
> > that involve multiple devices. So the exporter drivers would have to
> > implement more or less hacky solutions to handle memory requirements
> > and choosing the device that allocated memory.
> >
> > Currently, AFAIK there is even no generic way for a driver to
> > acquire its own DMA memory requirements.
> >
> > Therefore all logic hidden beneath 'attachment' is pointless. I
> > think that support for attach/detach (and related stuff) should be
> > postponed until support for multi-device allocation is added to DMA
> > framework.
> 
> Imo we clearly need this to make the multi-device-driver with insane dma
> requirements work on arm. And rewriting the buffer handling in
> participating subsystem twice isn't really a great plan. I envision that
> on platforms where we need this madness, the driver must call back to the
> dma subsytem to create a dma_buf. The dma subsytem should be already aware
> of all the requirements and hence should be able to handle them..
> 
> > I don't say the attachment list idea is wrong but adding attachment
> > stuff creates an illusion that problem of multi-device allocations
> > is somehow magically solved. We should not force the developers of
> > exporter drivers to solve the problem that is not solvable yet.
> 
> Well, this is why we need to create a decent support infrastructure for
> platforms (= arm madness) that needs this, so that device drivers and
> subsystem don't need to invent that wheel on their own. Which as you point
> out, they actually can't.

The real question is whether it is possible to create any generic support
infrastructure. I really doubt. IMHO this is something that will be hacked for
each 'product release' and will never read the mainline...
 
> > The other problem are the APIs. For example, the V4L2 subsystem
> > assumes that memory is allocated after successful VIDIOC_REQBUFS
> > with V4L2_MEMORY_MMAP memory type. Therefore attach would be
> > automatically followed by get_scatterlist, blocking possibility of
> > any buffer migrations in future.
> 
> Well, pardon to break the news, but v4l needs to rework the buffer
> handling. If you want to share buffers with a gpu driver, you _have_ to
> life with the fact that gpus do fully dynamic buffer management, meaning:
> - buffers get allocated and destroyed on the fly, meaning static reqbuf
>   just went out the window (we obviously cache buffer objects and reuse
>   them for performance, as long as the processing pipeline doesn't really
>   change).
> - buffers get moved around in memory, meaning you either need full-blown
>   sync-objects with a callback to drivers to tear-down mappings on-demand,
>   or every driver needs to guarnatee to call put_scatterlist in a
>   reasonable short time. The latter is probably the more natural thing for
>   v4l devices.

I'm really not convinced if it is possible to go for the completely dynamic
buffer management, especially if we are implementing a proof-of-concept 
solution. Please notice the following facts:

1. all v4l2 drivers do the 'static' buffer management - memory is being 
allocated on REQBUF() call and then mapped permanently into both userspace 
and dma (io) address space.

2. dma-mapping api is very limited in the area of the dynamic buffer management,
this API has been designed definitely for static buffer allocation and mapping.

It looks that fully dynamic buffer management requires a complete change of 
v4l2 api principles (V4L3?) and a completely new DMA API interface. That's 
probably the reason by none of the GPU driver relies on the DMA-mapping API
and implements custom solution for managing the mappings. 

This reminds me one more issue I've noticed in the current dma buf proof-of-
concept. You assumed that the exporter will be responsible for mapping the 
buffer into io address space of all the client devices. What if the device 
needs additional custom hooks/hacks during the mappings? This will be a serious
problem for the current GPU drivers for example. IMHO the API will be much
clearer if each client driver will map the scatter list gathered from the
dma buf by itself. Only the client driver has the complete knowledge how 
to do this correctly for this particular device. This way it will also work
with devices that don't do the real DMA (like for example USB devices that
copy all data from usb packets to the target buffer with the cpu).

Best regards
  
Clark, Rob Nov. 8, 2011, 4:59 p.m. UTC | #15
On Thu, Nov 3, 2011 at 3:04 AM, Marek Szyprowski
<m.szyprowski@samsung.com> wrote:
> Hello,
>
> I'm sorry for a late reply, but after Kernel Summit/ELC I have some comments.
>
> On Friday, October 14, 2011 5:35 PM Daniel Vetter wrote:
>
>> On Fri, Oct 14, 2011 at 12:00:58PM +0200, Tomasz Stanislawski wrote:
>> > >+/**
>> > >+ * struct dma_buf_ops - operations possible on struct dma_buf
>> > >+ * @create: creates a struct dma_buf of a fixed size. Actual allocation
>> > >+ *            does not happen here.
>> >
>> > The 'create' ops is not present in dma_buf_ops.
>> >
>> > >+ * @attach: allows different devices to 'attach' themselves to the given
>> > >+ *            buffer. It might return -EBUSY to signal that backing storage
>> > >+ *            is already allocated and incompatible with the requirements
>> > >+ *            of requesting device. [optional]
>> > >+ * @detach: detach a given device from this buffer. [optional]
>> > >+ * @get_scatterlist: returns list of scatter pages allocated, increases
>> > >+ *                     usecount of the buffer. Requires atleast one attach to be
>> > >+ *                     called before. Returned sg list should already be mapped
>> > >+ *                     into _device_ address space.
>> >
>> > You must add a comment that this call 'may sleep'.
>> >
>> > I like the get_scatterlist idea. It allows the exported to create a
>> > valid scatterlist for a client in a elegant way.
>> >
>> > I do not like this whole attachment idea. The problem is that
>> > currently there is no support in DMA framework for allocation for
>> > multiple devices. As long as no such a support exists, there is no
>> > generic way to handle attribute negotiations and buffer allocations
>> > that involve multiple devices. So the exporter drivers would have to
>> > implement more or less hacky solutions to handle memory requirements
>> > and choosing the device that allocated memory.
>> >
>> > Currently, AFAIK there is even no generic way for a driver to
>> > acquire its own DMA memory requirements.
>> >
>> > Therefore all logic hidden beneath 'attachment' is pointless. I
>> > think that support for attach/detach (and related stuff) should be
>> > postponed until support for multi-device allocation is added to DMA
>> > framework.
>>
>> Imo we clearly need this to make the multi-device-driver with insane dma
>> requirements work on arm. And rewriting the buffer handling in
>> participating subsystem twice isn't really a great plan. I envision that
>> on platforms where we need this madness, the driver must call back to the
>> dma subsytem to create a dma_buf. The dma subsytem should be already aware
>> of all the requirements and hence should be able to handle them..
>>
>> > I don't say the attachment list idea is wrong but adding attachment
>> > stuff creates an illusion that problem of multi-device allocations
>> > is somehow magically solved. We should not force the developers of
>> > exporter drivers to solve the problem that is not solvable yet.
>>
>> Well, this is why we need to create a decent support infrastructure for
>> platforms (= arm madness) that needs this, so that device drivers and
>> subsystem don't need to invent that wheel on their own. Which as you point
>> out, they actually can't.
>
> The real question is whether it is possible to create any generic support
> infrastructure. I really doubt. IMHO this is something that will be hacked for
> each 'product release' and will never read the mainline...
>
>> > The other problem are the APIs. For example, the V4L2 subsystem
>> > assumes that memory is allocated after successful VIDIOC_REQBUFS
>> > with V4L2_MEMORY_MMAP memory type. Therefore attach would be
>> > automatically followed by get_scatterlist, blocking possibility of
>> > any buffer migrations in future.
>>
>> Well, pardon to break the news, but v4l needs to rework the buffer
>> handling. If you want to share buffers with a gpu driver, you _have_ to
>> life with the fact that gpus do fully dynamic buffer management, meaning:
>> - buffers get allocated and destroyed on the fly, meaning static reqbuf
>>   just went out the window (we obviously cache buffer objects and reuse
>>   them for performance, as long as the processing pipeline doesn't really
>>   change).
>> - buffers get moved around in memory, meaning you either need full-blown
>>   sync-objects with a callback to drivers to tear-down mappings on-demand,
>>   or every driver needs to guarnatee to call put_scatterlist in a
>>   reasonable short time. The latter is probably the more natural thing for
>>   v4l devices.
>
> I'm really not convinced if it is possible to go for the completely dynamic
> buffer management, especially if we are implementing a proof-of-concept
> solution. Please notice the following facts:
>
> 1. all v4l2 drivers do the 'static' buffer management - memory is being
> allocated on REQBUF() call and then mapped permanently into both userspace
> and dma (io) address space.

Is this strictly true if we are introducing a new 'enum v4l2_memory'
for dmabuf's?  Shouldn't that give us some flexibility, especially if
the v4l2 device is only the importer, not the allocator, of the
memory.

and a couple more notes:
1) for V4L2_MEMORY_DMABUF we don't want the importing v4l2 device to
create a userspace memory mapping
2) there is nothing stopping a dmabuf exporter from being more strict
about static allocation of buffers..  we just want the API to be
flexible enough so that exporting devices which are more dynamic can
continue to do so.

> 2. dma-mapping api is very limited in the area of the dynamic buffer management,
> this API has been designed definitely for static buffer allocation and mapping.
>
> It looks that fully dynamic buffer management requires a complete change of
> v4l2 api principles (V4L3?) and a completely new DMA API interface. That's
> probably the reason by none of the GPU driver relies on the DMA-mapping API
> and implements custom solution for managing the mappings.
>
> This reminds me one more issue I've noticed in the current dma buf proof-of-
> concept. You assumed that the exporter will be responsible for mapping the
> buffer into io address space of all the client devices. What if the device
> needs additional custom hooks/hacks during the mappings? This will be a serious
> problem for the current GPU drivers for example. IMHO the API will be much
> clearer if each client driver will map the scatter list gathered from the
> dma buf by itself. Only the client driver has the complete knowledge how
> to do this correctly for this particular device. This way it will also work
> with devices that don't do the real DMA (like for example USB devices that
> copy all data from usb packets to the target buffer with the cpu).

The exporter doesn't map.. it returns a scatterlist to the importer.
But the exporter does allocate and pin backing pages.  And it is
preferable if the exporter has the opportunity to wait until as much
is known about the various importing devices to know if it must
allocate contiguous pages, or pages in a certain range.

That said, on a platform where everything had iommu's or somehow
didn't have any particular memory requirements, or where the exporter
had the strictest requirements (or at least knew of the strictest
requirements), then the exporter is free to allocate/pin the backing
pages earlier, such as even before the buffer is exported.

BR,
-R

> Best regards
> --
> Marek Szyprowski
> Samsung Poland R&D Center
>
>
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Daniel Vetter Nov. 8, 2011, 5:42 p.m. UTC | #16
On Tue, Nov 08, 2011 at 10:59:56AM -0600, Clark, Rob wrote:
> On Thu, Nov 3, 2011 at 3:04 AM, Marek Szyprowski
> > 2. dma-mapping api is very limited in the area of the dynamic buffer management,
> > this API has been designed definitely for static buffer allocation and mapping.
> >
> > It looks that fully dynamic buffer management requires a complete change of
> > v4l2 api principles (V4L3?) and a completely new DMA API interface. That's
> > probably the reason by none of the GPU driver relies on the DMA-mapping API
> > and implements custom solution for managing the mappings.
> >
> > This reminds me one more issue I've noticed in the current dma buf proof-of-
> > concept. You assumed that the exporter will be responsible for mapping the
> > buffer into io address space of all the client devices. What if the device
> > needs additional custom hooks/hacks during the mappings? This will be a serious
> > problem for the current GPU drivers for example. IMHO the API will be much
> > clearer if each client driver will map the scatter list gathered from the
> > dma buf by itself. Only the client driver has the complete knowledge how
> > to do this correctly for this particular device. This way it will also work
> > with devices that don't do the real DMA (like for example USB devices that
> > copy all data from usb packets to the target buffer with the cpu).
> 
> The exporter doesn't map.. it returns a scatterlist to the importer.
> But the exporter does allocate and pin backing pages.  And it is
> preferable if the exporter has the opportunity to wait until as much
> is known about the various importing devices to know if it must
> allocate contiguous pages, or pages in a certain range.

Actually I think the importer should get a _mapped_ scatterlist when it
calls get_scatterlist. The simple reason is that for strange stuff like
memory remapped into e.g. omaps TILER doesn't have any sensible notion of
an address in physical memory. For the USB-example I think the right
approach is to attach the usb hci to the dma_buf, after all that is the
device that will read the data and move over the usb bus to the udl
device. Similar for any other device that sits behind a bus that can't do
dma (or it doesn't make sense to do dma).

Imo if there's a use-case where the client needs to frob the sg_list
before calling dma_map_sg, we have an issue with the dma subsystem in
general.

> That said, on a platform where everything had iommu's or somehow
> didn't have any particular memory requirements, or where the exporter
> had the strictest requirements (or at least knew of the strictest
> requirements), then the exporter is free to allocate/pin the backing
> pages earlier, such as even before the buffer is exported.

Yeah, I think the important thing is that the dma_buf api should allow
decent buffer management. If certain subsystems ignore that and just
allocate up-front, no problem for me. But given how all graphics drivers
for essentially all OS have moved to dynamic buffer management, I expect
decoders, encoders, v4l devices and whatever else might sit in a graphics
pipeline to follow.

Yours, Daniel
  
Russell King - ARM Linux Nov. 8, 2011, 5:55 p.m. UTC | #17
On Tue, Nov 08, 2011 at 06:42:27PM +0100, Daniel Vetter wrote:
> Actually I think the importer should get a _mapped_ scatterlist when it
> calls get_scatterlist. The simple reason is that for strange stuff like
> memory remapped into e.g. omaps TILER doesn't have any sensible notion of
> an address in physical memory. For the USB-example I think the right
> approach is to attach the usb hci to the dma_buf, after all that is the
> device that will read the data and move over the usb bus to the udl
> device. Similar for any other device that sits behind a bus that can't do
> dma (or it doesn't make sense to do dma).
> 
> Imo if there's a use-case where the client needs to frob the sg_list
> before calling dma_map_sg, we have an issue with the dma subsystem in
> general.

Let's clear something up about the DMA API, which I think is causing some
misunderstanding here.  For this purpose, I'm going to talk about
dma_map_single(), but the same applies to the scatterlist and _page
variants as well.

	dma = dma_map_single(dev, cpuaddr, size, dir);

dev := the device _performing_ the DMA operation.  You are quite correct
       that in the case of a USB peripheral device, the device is normally
       the USB HCI device.

dma := dma address to be programmed into 'dev' which corresponds (by some
       means) with 'cpuaddr'.  This may not be the physical address due
       to bus offset translations or mappings setup in IOMMUs.

Therefore, it is wrong to talk about a 'physical address' when talking
about the DMA API.

We can take this one step further.  Lets say that the USB HCI is not
capable of performing memory accesses itself, but it is connected to a
separate DMA engine device:

	mem <---> dma engine <---> usb hci <---> usb peripheral

(such setups do exist, but despite having such implementations I've never
tried to support it.)

In this case, the dma engine, in response to control signals from the
USB host controller, will generate the appropriate bus address to access
memory and transfer the data into the USB HCI device.

So, in this case, the struct device to be used for mapping memory for
transfers to the usb peripheral is the DMA engine device, not the USB HCI
device nor the USB peripheral device.
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Daniel Vetter Nov. 8, 2011, 6:43 p.m. UTC | #18
On Tue, Nov 08, 2011 at 05:55:17PM +0000, Russell King - ARM Linux wrote:
> On Tue, Nov 08, 2011 at 06:42:27PM +0100, Daniel Vetter wrote:
> > Actually I think the importer should get a _mapped_ scatterlist when it
> > calls get_scatterlist. The simple reason is that for strange stuff like
> > memory remapped into e.g. omaps TILER doesn't have any sensible notion of
> > an address in physical memory. For the USB-example I think the right
> > approach is to attach the usb hci to the dma_buf, after all that is the
> > device that will read the data and move over the usb bus to the udl
> > device. Similar for any other device that sits behind a bus that can't do
> > dma (or it doesn't make sense to do dma).
> > 
> > Imo if there's a use-case where the client needs to frob the sg_list
> > before calling dma_map_sg, we have an issue with the dma subsystem in
> > general.
> 
> Let's clear something up about the DMA API, which I think is causing some
> misunderstanding here.  For this purpose, I'm going to talk about
> dma_map_single(), but the same applies to the scatterlist and _page
> variants as well.
> 
> 	dma = dma_map_single(dev, cpuaddr, size, dir);
> 
> dev := the device _performing_ the DMA operation.  You are quite correct
>        that in the case of a USB peripheral device, the device is normally
>        the USB HCI device.
> 
> dma := dma address to be programmed into 'dev' which corresponds (by some
>        means) with 'cpuaddr'.  This may not be the physical address due
>        to bus offset translations or mappings setup in IOMMUs.
> 
> Therefore, it is wrong to talk about a 'physical address' when talking
> about the DMA API.
> 
> We can take this one step further.  Lets say that the USB HCI is not
> capable of performing memory accesses itself, but it is connected to a
> separate DMA engine device:
> 
> 	mem <---> dma engine <---> usb hci <---> usb peripheral
> 
> (such setups do exist, but despite having such implementations I've never
> tried to support it.)
> 
> In this case, the dma engine, in response to control signals from the
> USB host controller, will generate the appropriate bus address to access
> memory and transfer the data into the USB HCI device.
> 
> So, in this case, the struct device to be used for mapping memory for
> transfers to the usb peripheral is the DMA engine device, not the USB HCI
> device nor the USB peripheral device.

Thanks for the clarification. I think this is another reason why
get_scatterlist should return the sg_list already mapped into the device
address space - it's more consisten with the other dma apis. Another
reason to completely hide everything but mapped addresses is crazy stuff
like this

	mem <---> tiling iommu <-+-> gpu
	                         |
	                         +-> scanout engine
	                         |
				 +-> mpeg decoder

where it doesn't really make sense to talk about the memory backing the
dma buffer because that's smeared all over the place due to tiling. IIRC
for the case of omap these devices can also access memory through other
paths and iommut that don't tile (but just remap like a normal iommu)

-Daniel
  
Dave Airlie Nov. 25, 2011, 2:13 p.m. UTC | #19
On Tue, Oct 11, 2011 at 10:23 AM, Sumit Semwal <sumit.semwal@ti.com> wrote:
> This is the first step in defining a dma buffer sharing mechanism.
>
> A new buffer object dma_buf 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 get_scatterlist and
>   put_scatterlist operations.
>
> Atleast one 'attach()' call is required to be made prior to calling the
> get_scatterlist() operation.
>
> Couple of building blocks in get_scatterlist() are added to ease introduction
> of sync'ing across exporter and users, and late allocation by the exporter.
>
> mmap() file operation is provided for the associated 'fd', as wrapper over the
> optional allocator defined mmap(), to be used by devices that might need one.
>
> More details are there in the documentation patch.
>

Some questions, I've started playing around with using this framework
to do buffer sharing between DRM devices,

Why struct scatterlist and not struct sg_table? it seems like I really
want to use an sg_table,

I'm not convinced fd's are really useful over just some idr allocated
handle, so far I'm just returning the "fd" to userspace as a handle,
and passing it back in the other side, so I'm not really sure what an
fd wins us here, apart from the mmap thing which I think shouldn't be
here anyways.
(if fd's do win us more we should probably record that in the docs patch).

Dave.
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Daniel Vetter Nov. 25, 2011, 4:02 p.m. UTC | #20
On Fri, Nov 25, 2011 at 02:13:22PM +0000, Dave Airlie wrote:
> On Tue, Oct 11, 2011 at 10:23 AM, Sumit Semwal <sumit.semwal@ti.com> wrote:
> > This is the first step in defining a dma buffer sharing mechanism.
> >
> > A new buffer object dma_buf 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 get_scatterlist and
> >   put_scatterlist operations.
> >
> > Atleast one 'attach()' call is required to be made prior to calling the
> > get_scatterlist() operation.
> >
> > Couple of building blocks in get_scatterlist() are added to ease introduction
> > of sync'ing across exporter and users, and late allocation by the exporter.
> >
> > mmap() file operation is provided for the associated 'fd', as wrapper over the
> > optional allocator defined mmap(), to be used by devices that might need one.
> >
> > More details are there in the documentation patch.
> >
> 
> Some questions, I've started playing around with using this framework
> to do buffer sharing between DRM devices,
> 
> Why struct scatterlist and not struct sg_table? it seems like I really
> want to use an sg_table,

No reason at all besides that intel-gtt is using scatterlist internally
(and only kludges the sg_table together in an ad-hoc fashion) and so I
haven't noticed. sg_table for more consistency with the dma api sounds
good.

> I'm not convinced fd's are really useful over just some idr allocated
> handle, so far I'm just returning the "fd" to userspace as a handle,
> and passing it back in the other side, so I'm not really sure what an
> fd wins us here, apart from the mmap thing which I think shouldn't be
> here anyways.
> (if fd's do win us more we should probably record that in the docs patch).

Imo fds are nice because their known and there's already all the
preexisting infrastructure for them around. And if we ever get fancy with
e.g. sync objects we can easily add poll support (or some insane ioctls).
But I agree that "we can mmap" is bust as a reason and should just die.
-Daniel
  
Dave Airlie Nov. 25, 2011, 4:15 p.m. UTC | #21
> +struct dma_buf_attachment *dma_buf_attach(struct dma_buf *dmabuf,
> +                                               struct device *dev)
> +{
> +       struct dma_buf_attachment *attach;
> +       int ret;
> +
> +       BUG_ON(!dmabuf || !dev);
> +
> +       mutex_lock(&dmabuf->lock);
> +
> +       attach = kzalloc(sizeof(struct dma_buf_attachment), GFP_KERNEL);
> +       if (attach == NULL)
> +               goto err_alloc;
> +
> +       attach->dev = dev;
> +       if (dmabuf->ops->attach) {
> +               ret = dmabuf->ops->attach(dmabuf, dev, attach);
> +               if (!ret)
> +                       goto err_attach;
> +       }
> +       list_add(&attach->node, &dmabuf->attachments);
> +

I would assume at some point this needed at
attach->dmabuf = dmabuf;
added.

Dave.
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Dave Airlie Nov. 25, 2011, 4:28 p.m. UTC | #22
I've rebuilt my PRIME interface on top of dmabuf to see how it would work,

I've got primed gears running again on top, but I expect all my object
lifetime and memory ownership rules need fixing up (i.e. leaks like a
sieve).

http://cgit.freedesktop.org/~airlied/linux/log/?h=drm-prime-dmabuf

has the i915/nouveau patches for the kernel to produce the prime interface.

Dave.
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Daniel Vetter Nov. 26, 2011, 2 p.m. UTC | #23
On Fri, Nov 25, 2011 at 17:28, Dave Airlie <airlied@gmail.com> wrote:
> I've rebuilt my PRIME interface on top of dmabuf to see how it would work,
>
> I've got primed gears running again on top, but I expect all my object
> lifetime and memory ownership rules need fixing up (i.e. leaks like a
> sieve).
>
> http://cgit.freedesktop.org/~airlied/linux/log/?h=drm-prime-dmabuf
>
> has the i915/nouveau patches for the kernel to produce the prime interface.

I've noticed that your implementations for get_scatterlist (at least
for the i915 driver) doesn't return the sg table mapped into the
device address space. I've checked and the documentation makes it
clear that this should be the case (and we really need this to support
certain insane hw), but the get/put_scatterlist names are a bit
misleading. Proposal:

- use struct sg_table instead of scatterlist like you've already done
in you branch. Simply more consistent with the dma api.

- rename get/put_scatterlist into map/unmap for consistency with all
the map/unmap dma api functions. The attachement would then serve as
the abstract cookie to the backing storage, similar to how struct page
* works as an abstract cookie for dma_map/unmap_page. The only special
thing is that struct device * parameter because that's already part of
the attachment.

- add new wrapper functions dma_buf_map_attachment and
dma_buf_unmap_attachement to hide all the pointer/vtable-chasing that
we currently expose to users of this interface.

Comments?

Cheers, Daniel
  
Rob Clark Nov. 27, 2011, 6:59 a.m. UTC | #24
On Sat, Nov 26, 2011 at 8:00 AM, Daniel Vetter <daniel@ffwll.ch> wrote:
> On Fri, Nov 25, 2011 at 17:28, Dave Airlie <airlied@gmail.com> wrote:
>> I've rebuilt my PRIME interface on top of dmabuf to see how it would work,
>>
>> I've got primed gears running again on top, but I expect all my object
>> lifetime and memory ownership rules need fixing up (i.e. leaks like a
>> sieve).
>>
>> http://cgit.freedesktop.org/~airlied/linux/log/?h=drm-prime-dmabuf
>>
>> has the i915/nouveau patches for the kernel to produce the prime interface.
>
> I've noticed that your implementations for get_scatterlist (at least
> for the i915 driver) doesn't return the sg table mapped into the
> device address space. I've checked and the documentation makes it
> clear that this should be the case (and we really need this to support
> certain insane hw), but the get/put_scatterlist names are a bit
> misleading. Proposal:
>
> - use struct sg_table instead of scatterlist like you've already done
> in you branch. Simply more consistent with the dma api.

yup

> - rename get/put_scatterlist into map/unmap for consistency with all
> the map/unmap dma api functions. The attachement would then serve as
> the abstract cookie to the backing storage, similar to how struct page
> * works as an abstract cookie for dma_map/unmap_page. The only special
> thing is that struct device * parameter because that's already part of
> the attachment.

yup

> - add new wrapper functions dma_buf_map_attachment and
> dma_buf_unmap_attachement to hide all the pointer/vtable-chasing that
> we currently expose to users of this interface.

I thought that was one of the earlier comments on the initial dmabuf
patch, but either way: yup

BR,
-R

> Comments?
>
> Cheers, Daniel
> --
> Daniel Vetter
> daniel.vetter@ffwll.ch - +41 (0) 79 364 57 48 - http://blog.ffwll.ch
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Marek Szyprowski Nov. 28, 2011, 7:47 a.m. UTC | #25
Hello,

I'm sorry for the late reply, I must have missed this mail...

On Tuesday, November 08, 2011 7:43 PM Daniel Vetter wrote:

> On Tue, Nov 08, 2011 at 05:55:17PM +0000, Russell King - ARM Linux wrote:
> > On Tue, Nov 08, 2011 at 06:42:27PM +0100, Daniel Vetter wrote:
> > > Actually I think the importer should get a _mapped_ scatterlist when it
> > > calls get_scatterlist. The simple reason is that for strange stuff like
> > > memory remapped into e.g. omaps TILER doesn't have any sensible notion of
> > > an address in physical memory. For the USB-example I think the right
> > > approach is to attach the usb hci to the dma_buf, after all that is the
> > > device that will read the data and move over the usb bus to the udl
> > > device. Similar for any other device that sits behind a bus that can't do
> > > dma (or it doesn't make sense to do dma).
> > >
> > > Imo if there's a use-case where the client needs to frob the sg_list
> > > before calling dma_map_sg, we have an issue with the dma subsystem in
> > > general.
> >
> > Let's clear something up about the DMA API, which I think is causing some
> > misunderstanding here.  For this purpose, I'm going to talk about
> > dma_map_single(), but the same applies to the scatterlist and _page
> > variants as well.
> >
> > 	dma = dma_map_single(dev, cpuaddr, size, dir);
> >
> > dev := the device _performing_ the DMA operation.  You are quite correct
> >        that in the case of a USB peripheral device, the device is normally
> >        the USB HCI device.
> >
> > dma := dma address to be programmed into 'dev' which corresponds (by some
> >        means) with 'cpuaddr'.  This may not be the physical address due
> >        to bus offset translations or mappings setup in IOMMUs.
> >
> > Therefore, it is wrong to talk about a 'physical address' when talking
> > about the DMA API.
> >
> > We can take this one step further.  Lets say that the USB HCI is not
> > capable of performing memory accesses itself, but it is connected to a
> > separate DMA engine device:
> >
> > 	mem <---> dma engine <---> usb hci <---> usb peripheral
> >
> > (such setups do exist, but despite having such implementations I've never
> > tried to support it.)
> >
> > In this case, the dma engine, in response to control signals from the
> > USB host controller, will generate the appropriate bus address to access
> > memory and transfer the data into the USB HCI device.
> >
> > So, in this case, the struct device to be used for mapping memory for
> > transfers to the usb peripheral is the DMA engine device, not the USB HCI
> > device nor the USB peripheral device.
> 
> Thanks for the clarification. I think this is another reason why
> get_scatterlist should return the sg_list already mapped into the device
> address space - it's more consisten with the other dma apis. Another
> reason to completely hide everything but mapped addresses is crazy stuff
> like this
> 
> 	mem <---> tiling iommu <-+-> gpu
> 	                         |
> 	                         +-> scanout engine
> 	                         |
> 				 +-> mpeg decoder
> 
> where it doesn't really make sense to talk about the memory backing the
> dma buffer because that's smeared all over the place due to tiling. IIRC
> for the case of omap these devices can also access memory through other
> paths and iommut that don't tile (but just remap like a normal iommu)

I really don't get why you want to force the exporter to map the buffer into
clients dma address space. Only the client device might know all the quirks
required to do this correctly. Exporter should only provide a scatter-list 
with the memory that belongs to the exported buffer (might be pinned). How
do you want to solve the following case - the gpu hardware from your diagram
and a simple usb webcam with generic driver. The application would like to
export a buffer from the webcam to scanout engine. How the generic webcam 
driver might know HOW to set up the tiller to create correct mappings for 
the GPU/scanout? IMHO only a GPU driver is capable of doing that assuming
it got just a scatter list from the webcam driver.

Best regards
  
Daniel Vetter Nov. 28, 2011, 10:34 a.m. UTC | #26
On Mon, Nov 28, 2011 at 08:47:31AM +0100, Marek Szyprowski wrote:
> On Tuesday, November 08, 2011 7:43 PM Daniel Vetter wrote:
> > Thanks for the clarification. I think this is another reason why
> > get_scatterlist should return the sg_list already mapped into the device
> > address space - it's more consisten with the other dma apis. Another
> > reason to completely hide everything but mapped addresses is crazy stuff
> > like this
> > 
> > 	mem <---> tiling iommu <-+-> gpu
> > 	                         |
> > 	                         +-> scanout engine
> > 	                         |
> > 				 +-> mpeg decoder
> > 
> > where it doesn't really make sense to talk about the memory backing the
> > dma buffer because that's smeared all over the place due to tiling. IIRC
> > for the case of omap these devices can also access memory through other
> > paths and iommut that don't tile (but just remap like a normal iommu)
> 
> I really don't get why you want to force the exporter to map the buffer into
> clients dma address space. Only the client device might know all the quirks
> required to do this correctly. Exporter should only provide a scatter-list 
> with the memory that belongs to the exported buffer (might be pinned). How
> do you want to solve the following case - the gpu hardware from your diagram
> and a simple usb webcam with generic driver. The application would like to
> export a buffer from the webcam to scanout engine. How the generic webcam 
> driver might know HOW to set up the tiller to create correct mappings for 
> the GPU/scanout? IMHO only a GPU driver is capable of doing that assuming
> it got just a scatter list from the webcam driver.

You're correct that only the gpu knows how to put things into the tiler
(and maybe other devices that have access to it). Let me expand my diagram
so that you're webcam fits into the picture.

 	mem <-+-> tiling iommu <-+-> gpu
 	      |                  |
 	      |                  +-> scanout engine
 	      |                  |
              |                  +-> mpeg decoder
              |                  |                
              |                  |                
              +-> direct dma   <-+
              |                                  
              +-> iommua A <-+-> usb hci                                   
                             |                    
                             +-> other devices                   
                             |                    
                             ...                    

A few notes:
- might not be exactly how omap really looks like
- the devices behind tiler have different device address space windows to
  access the different paths to main memory. No other device can access
  the tiler, iirc.
- your webcam doesn't exist on this because we can't dma from it's memory,
  we can only zero-copy from the memory the usb hci transferred the frame
  to.

Now when when e.g. the scanout engine calls get_scatterlist you only call
dma_map_sg (which does nothing, because there's no iommu that's managed by
the core kernel code for it). The scanout engine will then complain that
your stuff is not contiguous and bail out. Or it is indeed contiguous and
things Just Work.

The much more interesting case is when the mpeg decoder and the gpu share
a buffer (think video on rotating cube or whatever other gui transition
you fancy). Then the omap tiler code can check whether the the device sits
behind the tiler (e.g. with some omap-specific device tree attribute) and
hand out a linear view to a tiled buffer.

In other words, whereever you're currently calling one of the map/unmap
dma api variants, you would call get/put_scatterlist (or better the new
name I'm proposing). So I also don't see your argument about only the
client knows how to map something into address space.

Yours, Daniel
  
Semwal, Sumit Dec. 1, 2011, 5:55 a.m. UTC | #27
Hi Dave, Daniel, Rob,
>
> On Sun, Nov 27, 2011 at 12:29 PM, Rob Clark <robdclark@gmail.com> wrote:
>>
>> On Sat, Nov 26, 2011 at 8:00 AM, Daniel Vetter <daniel@ffwll.ch> wrote:
>> > On Fri, Nov 25, 2011 at 17:28, Dave Airlie <airlied@gmail.com> wrote:
>> >> I've rebuilt my PRIME interface on top of dmabuf to see how it would
>> >> work,
>> >>
>> >> I've got primed gears running again on top, but I expect all my object
>> >> lifetime and memory ownership rules need fixing up (i.e. leaks like a
>> >> sieve).
>> >>
>> >> http://cgit.freedesktop.org/~airlied/linux/log/?h=drm-prime-dmabuf
>> >>
>> >> has the i915/nouveau patches for the kernel to produce the prime
>> >> interface.
>> >
>> > I've noticed that your implementations for get_scatterlist (at least
>> > for the i915 driver) doesn't return the sg table mapped into the
>> > device address space. I've checked and the documentation makes it
>> > clear that this should be the case (and we really need this to support
>> > certain insane hw), but the get/put_scatterlist names are a bit
>> > misleading. Proposal:
>> >
>> > - use struct sg_table instead of scatterlist like you've already done
>> > in you branch. Simply more consistent with the dma api.
>>
>> yup
>>
>> > - rename get/put_scatterlist into map/unmap for consistency with all
>> > the map/unmap dma api functions. The attachement would then serve as
>> > the abstract cookie to the backing storage, similar to how struct page
>> > * works as an abstract cookie for dma_map/unmap_page. The only special
>> > thing is that struct device * parameter because that's already part of
>> > the attachment.
>>
>> yup
>>
>> > - add new wrapper functions dma_buf_map_attachment and
>> > dma_buf_unmap_attachement to hide all the pointer/vtable-chasing that
>> > we currently expose to users of this interface.
>>
>> I thought that was one of the earlier comments on the initial dmabuf
>> patch, but either way: yup
>
Thanks for your comments; I will incorporate all of these in the next
version I'll send out.
>>
>>
>> BR,
>> -R
>
BR,
Sumit.
>>
>>
>> > Comments?
>> >
>> > Cheers, Daniel
>> > --
>> > Daniel Vetter
>> > daniel.vetter@ffwll.ch - +41 (0) 79 364 57 48 - http://blog.ffwll.ch
>> > --
>> > To unsubscribe from this list: send the line "unsubscribe linux-media"
>> > in
>> > the body of a message to majordomo@vger.kernel.org
>> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> >
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  

Patch

diff --git a/drivers/base/Kconfig b/drivers/base/Kconfig
index 21cf46f..07d8095 100644
--- a/drivers/base/Kconfig
+++ b/drivers/base/Kconfig
@@ -174,4 +174,14 @@  config SYS_HYPERVISOR
 
 source "drivers/base/regmap/Kconfig"
 
+config DMA_SHARED_BUFFER
+	bool "Buffer framework to be shared between drivers"
+	default n
+	depends on ANON_INODES
+	help
+	  This option enables the framework for buffer-sharing between
+	  multiple drivers. A buffer is associated with a file using driver
+	  APIs extension; the file's descriptor can then be passed on to other
+	  driver.
+
 endmenu
diff --git a/drivers/base/Makefile b/drivers/base/Makefile
index 99a375a..d0df046 100644
--- a/drivers/base/Makefile
+++ b/drivers/base/Makefile
@@ -8,6 +8,7 @@  obj-$(CONFIG_DEVTMPFS)	+= devtmpfs.o
 obj-y			+= power/
 obj-$(CONFIG_HAS_DMA)	+= dma-mapping.o
 obj-$(CONFIG_HAVE_GENERIC_DMA_COHERENT) += dma-coherent.o
+obj-$(CONFIG_DMA_SHARED_BUFFER) += dma-buf.o
 obj-$(CONFIG_ISA)	+= isa.o
 obj-$(CONFIG_FW_LOADER)	+= firmware_class.o
 obj-$(CONFIG_NUMA)	+= node.o
diff --git a/drivers/base/dma-buf.c b/drivers/base/dma-buf.c
new file mode 100644
index 0000000..58c51a0
--- /dev/null
+++ b/drivers/base/dma-buf.c
@@ -0,0 +1,242 @@ 
+/*
+ * Framework for buffer objects that can be shared across devices/subsystems.
+ *
+ * Copyright(C) 2011 Linaro Limited. All rights reserved.
+ * Author: Sumit Semwal <sumit.semwal@ti.com>
+ *
+ * Many thanks to linaro-mm-sig list, and specially
+ * Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and
+ * Daniel Vetter <daniel@ffwll.ch> for their support in creation and
+ * refining of this idea.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published by
+ * the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/fs.h>
+#include <linux/slab.h>
+#include <linux/dma-buf.h>
+#include <linux/anon_inodes.h>
+
+static inline int is_dma_buf_file(struct file *);
+
+static int dma_buf_mmap(struct file *file, struct vm_area_struct *vma)
+{
+	struct dma_buf *dmabuf;
+
+	if (!is_dma_buf_file(file))
+		return -EINVAL;
+
+	dmabuf = file->private_data;
+
+	if (!dmabuf->ops->mmap)
+		return -EINVAL;
+
+	return dmabuf->ops->mmap(dmabuf, vma);
+}
+
+static int dma_buf_release(struct inode *inode, struct file *file)
+{
+	struct dma_buf *dmabuf;
+
+	if (!is_dma_buf_file(file))
+		return -EINVAL;
+
+	dmabuf = file->private_data;
+
+	dmabuf->ops->release(dmabuf);
+	kfree(dmabuf);
+	return 0;
+}
+
+static const struct file_operations dma_buf_fops = {
+	.mmap		= dma_buf_mmap,
+	.release	= dma_buf_release,
+};
+
+/*
+ * is_dma_buf_file - Check if struct file* is associated with dma_buf
+ */
+static inline int is_dma_buf_file(struct file *file)
+{
+	return file->f_op == &dma_buf_fops;
+}
+
+/**
+ * dma_buf_export - Creates a new dma_buf, and associates an anon file
+ * with this buffer,so it can be exported.
+ * Also connect the allocator specific data and ops to the buffer.
+ *
+ * @priv:	[in]	Attach private data of allocator to this buffer
+ * @ops:	[in]	Attach allocator-defined dma buf ops to the new buffer.
+ * @flags:	[in]	mode flags for the file.
+ *
+ * Returns, on success, a newly created dma_buf object, which wraps the
+ * supplied private data and operations for dma_buf_ops. On failure to
+ * allocate the dma_buf object, it can return NULL.
+ *
+ */
+struct dma_buf *dma_buf_export(void *priv, struct dma_buf_ops *ops,
+				int flags)
+{
+	struct dma_buf *dmabuf;
+	struct file *file;
+
+	BUG_ON(!priv || !ops);
+
+	dmabuf = kzalloc(sizeof(struct dma_buf), GFP_KERNEL);
+	if (dmabuf == NULL)
+		return dmabuf;
+
+	dmabuf->priv = priv;
+	dmabuf->ops = ops;
+
+	file = anon_inode_getfile("dmabuf", &dma_buf_fops, dmabuf, flags);
+
+	dmabuf->file = file;
+
+	mutex_init(&dmabuf->lock);
+	INIT_LIST_HEAD(&dmabuf->attachments);
+
+	return dmabuf;
+}
+EXPORT_SYMBOL(dma_buf_export);
+
+
+/**
+ * dma_buf_fd - returns a file descriptor for the given dma_buf
+ * @dmabuf:	[in]	pointer to dma_buf for which fd is required.
+ *
+ * On success, returns an associated 'fd'. Else, returns error.
+ */
+int dma_buf_fd(struct dma_buf *dmabuf)
+{
+	int error, fd;
+
+	if (!dmabuf->file)
+		return -EINVAL;
+
+	error = get_unused_fd_flags(0);
+	if (error < 0)
+		return error;
+	fd = error;
+
+	fd_install(fd, dmabuf->file);
+
+	return fd;
+}
+EXPORT_SYMBOL(dma_buf_fd);
+
+/**
+ * dma_buf_get - returns the dma_buf structure related to an fd
+ * @fd:	[in]	fd associated with the dma_buf to be returned
+ *
+ * On success, returns the dma_buf structure associated with an fd; uses
+ * file's refcounting done by fget to increase refcount. returns ERR_PTR
+ * otherwise.
+ */
+struct dma_buf *dma_buf_get(int fd)
+{
+	struct file *file;
+
+	file = fget(fd);
+
+	if (!file)
+		return ERR_PTR(-EBADF);
+
+	if (!is_dma_buf_file(file)) {
+		fput(file);
+		return ERR_PTR(-EINVAL);
+	}
+
+	return file->private_data;
+}
+EXPORT_SYMBOL(dma_buf_get);
+
+/**
+ * dma_buf_put - decreases refcount of the buffer
+ * @dmabuf:	[in]	buffer to reduce refcount of
+ *
+ * Uses file's refcounting done implicitly by fput()
+ */
+void dma_buf_put(struct dma_buf *dmabuf)
+{
+	BUG_ON(!dmabuf->file);
+
+	fput(dmabuf->file);
+
+	return;
+}
+EXPORT_SYMBOL(dma_buf_put);
+
+/**
+ * dma_buf_attach - Add the device to dma_buf's attachments list; optionally,
+ * calls attach() of dma_buf_ops to allow device-specific attach functionality
+ * @dmabuf:	[in]	buffer to attach device to.
+ * @dev:	[in]	device to be attached.
+ *
+ * Returns struct dma_buf_attachment * for this attachment; may return NULL.
+ *
+ */
+struct dma_buf_attachment *dma_buf_attach(struct dma_buf *dmabuf,
+						struct device *dev)
+{
+	struct dma_buf_attachment *attach;
+	int ret;
+
+	BUG_ON(!dmabuf || !dev);
+
+	mutex_lock(&dmabuf->lock);
+
+	attach = kzalloc(sizeof(struct dma_buf_attachment), GFP_KERNEL);
+	if (attach == NULL)
+		goto err_alloc;
+
+	attach->dev = dev;
+	if (dmabuf->ops->attach) {
+		ret = dmabuf->ops->attach(dmabuf, dev, attach);
+		if (!ret)
+			goto err_attach;
+	}
+	list_add(&attach->node, &dmabuf->attachments);
+
+err_alloc:
+	mutex_unlock(&dmabuf->lock);
+	return attach;
+err_attach:
+	kfree(attach);
+	mutex_unlock(&dmabuf->lock);
+	return ERR_PTR(ret);
+}
+EXPORT_SYMBOL(dma_buf_attach);
+
+/**
+ * dma_buf_detach - Remove the given attachment from dmabuf's attachments list;
+ * optionally calls detach() of dma_buf_ops for device-specific detach
+ * @dmabuf:	[in]	buffer to detach from.
+ * @attach:	[in]	attachment to be detached; is free'd after this call.
+ *
+ */
+void dma_buf_detach(struct dma_buf *dmabuf, struct dma_buf_attachment *attach)
+{
+	BUG_ON(!dmabuf || !attach);
+
+	mutex_lock(&dmabuf->lock);
+	list_del(&attach->node);
+	if (dmabuf->ops->detach)
+		dmabuf->ops->detach(dmabuf, attach);
+
+	kfree(attach);
+	mutex_unlock(&dmabuf->lock);
+	return;
+}
+EXPORT_SYMBOL(dma_buf_detach);
diff --git a/include/linux/dma-buf.h b/include/linux/dma-buf.h
new file mode 100644
index 0000000..5bdf16a
--- /dev/null
+++ b/include/linux/dma-buf.h
@@ -0,0 +1,162 @@ 
+/*
+ * Header file for dma buffer sharing framework.
+ *
+ * Copyright(C) 2011 Linaro Limited. All rights reserved.
+ * Author: Sumit Semwal <sumit.semwal@ti.com>
+ *
+ * Many thanks to linaro-mm-sig list, and specially
+ * Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and
+ * Daniel Vetter <daniel@ffwll.ch> for their support in creation and
+ * refining of this idea.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published by
+ * the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef __DMA_BUF_H__
+#define __DMA_BUF_H__
+
+#include <linux/file.h>
+#include <linux/err.h>
+#include <linux/device.h>
+#include <linux/scatterlist.h>
+#include <linux/list.h>
+#include <linux/dma-mapping.h>
+
+struct dma_buf;
+
+/**
+ * struct dma_buf_attachment - holds device-buffer attachment data
+ * @dmabuf: buffer for this attachment.
+ * @dev: device attached to the buffer.
+ * @node: list_head to allow manipulation of list of dma_buf_attachment.
+ * @priv: exporter-specific attachment data.
+ */
+struct dma_buf_attachment {
+	struct dma_buf *dmabuf;
+	struct device *dev;
+	struct list_head node;
+	void *priv;
+};
+
+/**
+ * struct dma_buf_ops - operations possible on struct dma_buf
+ * @create: creates a struct dma_buf of a fixed size. Actual allocation
+ *	    does not happen here.
+ * @attach: allows different devices to 'attach' themselves to the given
+ *	    buffer. It might return -EBUSY to signal that backing storage
+ *	    is already allocated and incompatible with the requirements
+ *	    of requesting device. [optional]
+ * @detach: detach a given device from this buffer. [optional]
+ * @get_scatterlist: returns list of scatter pages allocated, increases
+ *		     usecount of the buffer. Requires atleast one attach to be
+ *		     called before. Returned sg list should already be mapped
+ *		     into _device_ address space.
+ * @put_scatterlist: decreases usecount of buffer, might deallocate scatter
+ *		     pages.
+ * @mmap: memory map this buffer - optional.
+ * @release: release this buffer; to be called after the last dma_buf_put.
+ * @sync_sg_for_cpu: sync the sg list for cpu.
+ * @sync_sg_for_device: synch the sg list for device.
+ */
+struct dma_buf_ops {
+	int (*attach)(struct dma_buf *, struct device *,
+			struct dma_buf_attachment *);
+
+	void (*detach)(struct dma_buf *, struct dma_buf_attachment *);
+
+	/* For {get,put}_scatterlist below, any specific buffer attributes
+	 * required should get added to device_dma_parameters accessible
+	 * via dev->dma_params.
+	 */
+	struct scatterlist * (*get_scatterlist)(struct dma_buf_attachment *,
+						enum dma_data_direction,
+						int *nents);
+	void (*put_scatterlist)(struct dma_buf_attachment *,
+						struct scatterlist *,
+						int nents);
+	/* TODO: Add interruptible and interruptible_timeout versions */
+
+	/* allow mmap optionally for devices that need it */
+	int (*mmap)(struct dma_buf *, struct vm_area_struct *);
+	/* after final dma_buf_put() */
+	void (*release)(struct dma_buf *);
+
+	/* allow allocator to take care of cache ops */
+	void (*sync_sg_for_cpu) (struct dma_buf *, struct device *);
+	void (*sync_sg_for_device)(struct dma_buf *, struct device *);
+};
+
+/**
+ * struct dma_buf - shared buffer object
+ * @file: file pointer used for sharing buffers across, and for refcounting.
+ * @attachments: list of dma_buf_attachment that denotes all devices attached.
+ * @ops: dma_buf_ops associated with this buffer object
+ * @priv: user specific private data
+ */
+struct dma_buf {
+	size_t size;
+	struct file *file;
+	struct list_head attachments;
+	const struct dma_buf_ops *ops;
+	/* mutex to serialize list manipulation and other ops */
+	struct mutex lock;
+	void *priv;
+};
+
+#ifdef CONFIG_DMA_SHARED_BUFFER
+struct dma_buf_attachment *dma_buf_attach(struct dma_buf *dmabuf,
+							struct device *dev);
+void dma_buf_detach(struct dma_buf *dmabuf,
+				struct dma_buf_attachment *dmabuf_attach);
+struct dma_buf *dma_buf_export(void *priv, struct dma_buf_ops *ops, int flags);
+int dma_buf_fd(struct dma_buf *dmabuf);
+struct dma_buf *dma_buf_get(int fd);
+void dma_buf_put(struct dma_buf *dmabuf);
+
+#else
+
+static inline struct dma_buf_attachment *dma_buf_attach(struct dma_buf *dmabuf,
+							struct device *dev)
+{
+	return ERR_PTR(-ENODEV);
+}
+
+static inline void dma_buf_detach(struct dma_buf *dmabuf,
+				  struct dma_buf_attachment *dmabuf_attach)
+{
+	return;
+}
+
+static inline struct dma_buf *dma_buf_export(void *priv,
+						struct dma_buf_ops *ops,
+						int flags)
+{
+	return ERR_PTR(-ENODEV);
+}
+
+static inline int dma_buf_fd(struct dma_buf *dmabuf)
+{
+	return -ENODEV;
+}
+
+static inline struct dma_buf *dma_buf_get(int fd)
+{
+	return ERR_PTR(-ENODEV);
+}
+
+static inline void dma_buf_put(struct dma_buf *dmabuf)
+{
+	return;
+}
+#endif /* CONFIG_DMA_SHARED_BUFFER */
+
+#endif /* __DMA_BUF_H__ */