                    DMA Buffer Sharing API Guide
                    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~

                            Sumit Semwal
                <sumit dot semwal at linaro dot org>
                 <sumit dot semwal at ti dot com>

This document serves as a guide to device-driver writers on what is the dma-buf
buffer sharing API, how to use it for exporting and using shared buffers.

Any device driver which wishes to be a part of DMA buffer sharing, can do so as
either the 'exporter' of buffers, or the 'user' of buffers.

Say a driver A wants to use buffers created by driver B, then we call B as the
exporter, and A as buffer-user.

The exporter
- implements and manages operations[1] for the buffer
- allows other users to share the buffer by using dma_buf sharing APIs,
- manages the details of buffer allocation,
- decides about the actual backing storage where this allocation happens,
- takes care of any migration of scatterlist - for all (shared) users of this
   buffer,

The buffer-user
- is one of (many) sharing users of the buffer.
- doesn't need to worry about how the buffer is allocated, or where.
- needs a mechanism to get access to the scatterlist that makes up this buffer
   in memory, mapped into its own address space, so it can access the same area
   of memory.

*IMPORTANT*: [see https://lkml.org/lkml/2011/12/20/211 for more details]
For this first version, A buffer shared using the dma_buf sharing API:
- *may* be exported to user space using "mmap" *ONLY* by exporter, outside of
  this framework.
- with this new iteration of the dma-buf api cpu access from the kernel has been
  enable, see below for the details.

dma-buf operations for device dma only
--------------------------------------

The dma_buf buffer sharing API usage contains the following steps:

1. Exporter announces that it wishes to export a buffer
2. Userspace gets the file descriptor associated with the exported buffer, and
   passes it around to potential buffer-users based on use case
3. Each buffer-user 'connects' itself to the buffer
4. When needed, buffer-user requests access to the buffer from exporter
5. When finished with its use, the buffer-user notifies end-of-DMA to exporter
6. when buffer-user is done using this buffer completely, it 'disconnects'
   itself from the buffer.


1. Exporter's announcement of buffer export

   The buffer exporter announces its wish to export a buffer. In this, it
   connects its own private buffer data, provides implementation for operations
   that can be performed on the exported dma_buf, and flags for the file
   associated with this buffer.

   Interface:
      struct dma_buf *dma_buf_export(void *priv, struct dma_buf_ops *ops,
				     size_t size, int flags)

   If this succeeds, dma_buf_export allocates a dma_buf structure, and returns a
   pointer to the same. It also associates an anonymous file with this buffer,
   so it can be exported. On failure to allocate the dma_buf object, it returns
   NULL.

2. Userspace gets a handle to pass around to potential buffer-users

   Userspace entity requests for a file-descriptor (fd) which is a handle to the
   anonymous file associated with the buffer. It can then share the fd with other
   drivers and/or processes.

   Interface:
      int dma_buf_fd(struct dma_buf *dmabuf)

   This API installs an fd for the anonymous file associated with this buffer;
   returns either 'fd', or error.

3. Each buffer-user 'connects' itself to the buffer

   Each buffer-user now gets a reference to the buffer, using the fd passed to
   it.

   Interface:
      struct dma_buf *dma_buf_get(int fd)

   This API will return a reference to the dma_buf, and increment refcount for
   it.

   After this, the buffer-user needs to attach its device with the buffer, which
   helps the exporter to know of device buffer constraints.

   Interface:
      struct dma_buf_attachment *dma_buf_attach(struct dma_buf *dmabuf,
                                                struct device *dev)

   This API returns reference to an attachment structure, which is then used
   for scatterlist operations. It will optionally call the 'attach' dma_buf
   operation, if provided by the exporter.

   The dma-buf sharing framework does the bookkeeping bits related to managing
   the list of all attachments to a buffer.

Until this stage, the buffer-exporter has the option to choose not to actually
allocate the backing storage for this buffer, but wait for the first buffer-user
to request use of buffer for allocation.


4. When needed, buffer-user requests access to the buffer

   Whenever a buffer-user wants to use the buffer for any DMA, it asks for
   access to the buffer using dma_buf_map_attachment API. At least one attach to
   the buffer must have happened before map_dma_buf can be called.

   Interface:
      struct sg_table * dma_buf_map_attachment(struct dma_buf_attachment *,
                                         enum dma_data_direction);

   This is a wrapper to dma_buf->ops->map_dma_buf operation, which hides the
   "dma_buf->ops->" indirection from the users of this interface.

   In struct dma_buf_ops, map_dma_buf is defined as
      struct sg_table * (*map_dma_buf)(struct dma_buf_attachment *,
                                                enum dma_data_direction);

   It is one of the buffer operations that must be implemented by the exporter.
   It should return the sg_table containing scatterlist for this buffer, mapped
   into caller's address space.

   If this is being called for the first time, the exporter can now choose to
   scan through the list of attachments for this buffer, collate the requirements
   of the attached devices, and choose an appropriate backing storage for the
   buffer.

   Based on enum dma_data_direction, it might be possible to have multiple users
   accessing at the same time (for reading, maybe), or any other kind of sharing
   that the exporter might wish to make available to buffer-users.

   map_dma_buf() operation can return -EINTR if it is interrupted by a signal.


5. When finished, the buffer-user notifies end-of-DMA to exporter

   Once the DMA for the current buffer-user is over, it signals 'end-of-DMA' to
   the exporter using the dma_buf_unmap_attachment API.

   Interface:
      void dma_buf_unmap_attachment(struct dma_buf_attachment *,
                                    struct sg_table *);

   This is a wrapper to dma_buf->ops->unmap_dma_buf() operation, which hides the
   "dma_buf->ops->" indirection from the users of this interface.

   In struct dma_buf_ops, unmap_dma_buf is defined as
      void (*unmap_dma_buf)(struct dma_buf_attachment *, struct sg_table *);

   unmap_dma_buf signifies the end-of-DMA for the attachment provided. Like
   map_dma_buf, this API also must be implemented by the exporter.


6. when buffer-user is done using this buffer, it 'disconnects' itself from the
   buffer.

   After the buffer-user has no more interest in using this buffer, it should
   disconnect itself from the buffer:

   - it first detaches itself from the buffer.

   Interface:
      void dma_buf_detach(struct dma_buf *dmabuf,
                          struct dma_buf_attachment *dmabuf_attach);

   This API removes the attachment from the list in dmabuf, and optionally calls
   dma_buf->ops->detach(), if provided by exporter, for any housekeeping bits.

   - Then, the buffer-user returns the buffer reference to exporter.

   Interface:
     void dma_buf_put(struct dma_buf *dmabuf);

   This API then reduces the refcount for this buffer.

   If, as a result of this call, the refcount becomes 0, the 'release' file
   operation related to this fd is called. It calls the dmabuf->ops->release()
   operation in turn, and frees the memory allocated for dmabuf when exported.

NOTES:
- Importance of attach-detach and {map,unmap}_dma_buf operation pairs
   The attach-detach calls allow the exporter to figure out backing-storage
   constraints for the currently-interested devices. This allows preferential
   allocation, and/or migration of pages across different types of storage
   available, if possible.

   Bracketing of DMA access with {map,unmap}_dma_buf operations is essential
   to allow just-in-time backing of storage, and migration mid-way through a
   use-case.

- Migration of backing storage if needed
   If after
   - at least one map_dma_buf has happened,
   - and the backing storage has been allocated for this buffer,
   another new buffer-user intends to attach itself to this buffer, it might
   be allowed, if possible for the exporter.

   In case it is allowed by the exporter:
    if the new buffer-user has stricter 'backing-storage constraints', and the
    exporter can handle these constraints, the exporter can just stall on the
    map_dma_buf until all outstanding access is completed (as signalled by
    unmap_dma_buf).
    Once all users have finished accessing and have unmapped this buffer, the
    exporter could potentially move the buffer to the stricter backing-storage,
    and then allow further {map,unmap}_dma_buf operations from any buffer-user
    from the migrated backing-storage.

   If the exporter cannot fulfil the backing-storage constraints of the new
   buffer-user device as requested, dma_buf_attach() would return an error to
   denote non-compatibility of the new buffer-sharing request with the current
   buffer.

   If the exporter chooses not to allow an attach() operation once a
   map_dma_buf() API has been called, it simply returns an error.

Kernel cpu access to a dma-buf buffer object
--------------------------------------------

The motivation to allow cpu access from the kernel to a dma-buf object from the
importers side are:
- fallback operations, e.g. if the devices is connected to a usb bus and the
  kernel needs to shuffle the data around first before sending it away.
- full transparency for existing users on the importer side, i.e. userspace
  should not notice the difference between a normal object from that subsystem
  and an imported one backed by a dma-buf. This is really important for drm
  opengl drivers that expect to still use all the existing upload/download
  paths.

Access to a dma_buf from the kernel context involves three steps:

1. Prepare access, which invalidate any necessary caches and make the object
   available for cpu access.
2. Access the object page-by-page with the dma_buf map apis
3. Finish access, which will flush any necessary cpu caches and free reserved
   resources.

1. Prepare access

   Before an importer can access a dma_buf object with the cpu from the kernel
   context, it needs to notify the exporter of the access that is about to
   happen.

   Interface:
      int dma_buf_begin_cpu_access(struct dma_buf *dmabuf,
				   size_t start, size_t len,
				   enum dma_data_direction direction)

   This allows the exporter to ensure that the memory is actually available for
   cpu access - the exporter might need to allocate or swap-in and pin the
   backing storage. The exporter also needs to ensure that cpu access is
   coherent for the given range and access direction. The range and access
   direction can be used by the exporter to optimize the cache flushing, i.e.
   access outside of the range or with a different direction (read instead of
   write) might return stale or even bogus data (e.g. when the exporter needs to
   copy the data to temporary storage).

   This step might fail, e.g. in oom conditions.

2. Accessing the buffer

   To support dma_buf objects residing in highmem cpu access is page-based using
   an api similar to kmap. Accessing a dma_buf is done in aligned chunks of
   PAGE_SIZE size. Before accessing a chunk it needs to be mapped, which returns
   a pointer in kernel virtual address space. Afterwards the chunk needs to be
   unmapped again. There is no limit on how often a given chunk can be mapped
   and unmapped, i.e. the importer does not need to call begin_cpu_access again
   before mapping the same chunk again.

   Interfaces:
      void *dma_buf_kmap(struct dma_buf *, unsigned long);
      void dma_buf_kunmap(struct dma_buf *, unsigned long, void *);

   There are also atomic variants of these interfaces. Like for kmap they
   facilitate non-blocking fast-paths. Neither the importer nor the exporter (in
   the callback) is allowed to block when using these.

   Interfaces:
      void *dma_buf_kmap_atomic(struct dma_buf *, unsigned long);
      void dma_buf_kunmap_atomic(struct dma_buf *, unsigned long, void *);

   For importers all the restrictions of using kmap apply, like the limited
   supply of kmap_atomic slots. Hence an importer shall only hold onto at most 2
   atomic dma_buf kmaps at the same time (in any given process context).

   dma_buf kmap calls outside of the range specified in begin_cpu_access are
   undefined. If the range is not PAGE_SIZE aligned, kmap needs to succeed on
   the partial chunks at the beginning and end but may return stale or bogus
   data outside of the range (in these partial chunks).

   Note that these calls need to always succeed. The exporter needs to complete
   any preparations that might fail in begin_cpu_access.

3. Finish access

   When the importer is done accessing the range specified in begin_cpu_access,
   it needs to announce this to the exporter (to facilitate cache flushing and
   unpinning of any pinned resources). The result of of any dma_buf kmap calls
   after end_cpu_access is undefined.

   Interface:
      void dma_buf_end_cpu_access(struct dma_buf *dma_buf,
				  size_t start, size_t len,
				  enum dma_data_direction dir);


Miscellaneous notes
-------------------

- Any exporters or users of the dma-buf buffer sharing framework must have
  a 'select DMA_SHARED_BUFFER' in their respective Kconfigs.

- In order to avoid fd leaks on exec, the FD_CLOEXEC flag must be set
  on the file descriptor.  This is not just a resource leak, but a
  potential security hole.  It could give the newly exec'd application
  access to buffers, via the leaked fd, to which it should otherwise
  not be permitted access.

  The problem with doing this via a separate fcntl() call, versus doing it
  atomically when the fd is created, is that this is inherently racy in a
  multi-threaded app[3].  The issue is made worse when it is library code
  opening/creating the file descriptor, as the application may not even be
  aware of the fd's.

  To avoid this problem, userspace must have a way to request O_CLOEXEC
  flag be set when the dma-buf fd is created.  So any API provided by
  the exporting driver to create a dmabuf fd must provide a way to let
  userspace control setting of O_CLOEXEC flag passed in to dma_buf_fd().

References:
[1] struct dma_buf_ops in include/linux/dma-buf.h
[2] All interfaces mentioned above defined in include/linux/dma-buf.h
[3] https://lwn.net/Articles/236486/