[PATCH] amdgpu: add amdgpu_find_bo_by_cpu_mapping interface V2

[Christian König]

On 07.12.2015 03:41, Chunming Zhou wrote:
> userspace needs to know if the user memory is from BO or malloc.
> V2: add bo_table_mutex protection.
>
> Change-Id: Ie2dbc13f1c02bc0a996f64f9db83a21da63c1d70
> Signed-off-by: Chunming Zhou <David1.Zhou at amd.com>
> Reviewed-by: Jammy Zhou <Jammy.Zhou at amd.com> (V1)
> Reviewed-by: Christian König <christian.koenig at amd.com> (V1)
The new version is Reviewed-by: Christian König 
<christian.koenig at amd.com> as well.

Thanks,
Christian.

> ---
>   amdgpu/amdgpu.h          | 24 ++++++++++++++++++++++++
>   amdgpu/amdgpu_bo.c       | 46 ++++++++++++++++++++++++++++++++++++++++++++++
>   include/drm/amdgpu_drm.h | 12 ++++++++++++
>   3 files changed, 82 insertions(+)
>
> diff --git a/amdgpu/amdgpu.h b/amdgpu/amdgpu.h
> index baae113..4925056 100644
> --- a/amdgpu/amdgpu.h
> +++ b/amdgpu/amdgpu.h
> @@ -672,6 +672,30 @@ int amdgpu_create_bo_from_user_mem(amdgpu_device_handle dev,
>   				    amdgpu_bo_handle *buf_handle);
>   
>   /**
> + * Validate if the user memory comes from BO
> + *
> + * \param dev - [in] Device handle. See #amdgpu_device_initialize()
> + * \param cpu - [in] CPU address of user allocated memory which we
> + * want to map to GPU address space (make GPU accessible)
> + * (This address must be correctly aligned).
> + * \param size - [in] Size of allocation (must be correctly aligned)
> + * \param buf_handle - [out] Buffer handle for the userptr memory
> + * if the user memory is not from BO, the buf_handle will be NULL.
> + * \param offset_in_bo - [out] offset in this BO for this user memory
> + *
> + *
> + * \return   0 on success\n
> + *          <0 - Negative POSIX Error code
> + *
> +*/
> +int amdgpu_find_bo_by_cpu_mapping(amdgpu_device_handle dev,
> +				  void *cpu,
> +				  uint64_t size,
> +				  amdgpu_bo_handle *buf_handle,
> +				  uint64_t *offset_in_bo);
> +
> +
> +/**
>    * Free previosuly allocated memory
>    *
>    * \param   dev	       - \c [in] Device handle. See #amdgpu_device_initialize()
> diff --git a/amdgpu/amdgpu_bo.c b/amdgpu/amdgpu_bo.c
> index 5a0f4ac..2e794ef 100644
> --- a/amdgpu/amdgpu_bo.c
> +++ b/amdgpu/amdgpu_bo.c
> @@ -528,6 +528,52 @@ int amdgpu_bo_wait_for_idle(amdgpu_bo_handle bo,
>   	}
>   }
>   
> +int amdgpu_find_bo_by_cpu_mapping(amdgpu_device_handle dev,
> +				  void *cpu,
> +				  uint64_t size,
> +				  amdgpu_bo_handle *buf_handle,
> +				  uint64_t *offset_in_bo)
> +{
> +	int r;
> +	struct amdgpu_bo *bo;
> +	struct drm_amdgpu_gem_find_bo args;
> +
> +	pthread_mutex_lock(&dev->bo_table_mutex);
> +	args.addr = (uintptr_t)cpu;
> +	args.size = size;
> +	r = drmCommandWriteRead(dev->fd, DRM_AMDGPU_GEM_FIND_BO,
> +				&args, sizeof(args));
> +	if (r) {
> +		pthread_mutex_unlock(&dev->bo_table_mutex);
> +		return r;
> +	}
> +	if (args.handle == 0) {
> +		pthread_mutex_unlock(&dev->bo_table_mutex);
> +		return -EINVAL;
> +	}
> +	bo = util_hash_table_get(dev->bo_handles,
> +				 (void*)(uintptr_t)args.handle);
> +	if (!bo) {
> +		bo = calloc(1, sizeof(struct amdgpu_bo));
> +		if (!bo) {
> +			pthread_mutex_unlock(&dev->bo_table_mutex);
> +			return -ENOMEM;
> +		}
> +		atomic_set(&bo->refcount, 1);
> +		bo->dev = dev;
> +		bo->alloc_size = size;
> +		bo->handle = args.handle;
> +	} else
> +		atomic_inc(&bo->refcount);
> +
> +	*buf_handle = bo;
> +	*offset_in_bo = args.offset;
> +	pthread_mutex_unlock(&dev->bo_table_mutex);
> +
> +	return r;
> +}
> +
> +
>   int amdgpu_create_bo_from_user_mem(amdgpu_device_handle dev,
>   				    void *cpu,
>   				    uint64_t size,
> diff --git a/include/drm/amdgpu_drm.h b/include/drm/amdgpu_drm.h
> index 050e7fe..e07904c 100644
> --- a/include/drm/amdgpu_drm.h
> +++ b/include/drm/amdgpu_drm.h
> @@ -47,6 +47,7 @@
>   #define DRM_AMDGPU_GEM_OP		0x10
>   #define DRM_AMDGPU_GEM_USERPTR		0x11
>   #define DRM_AMDGPU_WAIT_FENCES		0x12
> +#define DRM_AMDGPU_GEM_FIND_BO          0x13
>   
>   #define DRM_IOCTL_AMDGPU_GEM_CREATE	DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_GEM_CREATE, union drm_amdgpu_gem_create)
>   #define DRM_IOCTL_AMDGPU_GEM_MMAP	DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_GEM_MMAP, union drm_amdgpu_gem_mmap)
> @@ -61,6 +62,7 @@
>   #define DRM_IOCTL_AMDGPU_GEM_OP		DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_GEM_OP, struct drm_amdgpu_gem_op)
>   #define DRM_IOCTL_AMDGPU_GEM_USERPTR	DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_GEM_USERPTR, struct drm_amdgpu_gem_userptr)
>   #define DRM_IOCTL_AMDGPU_WAIT_FENCES	DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_WAIT_FENCES, union drm_amdgpu_wait_fences)
> +#define DRM_IOCTL_AMDGPU_GEM_FIND_BO      DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_GEM_FIND_BO, struct drm_amdgpu_gem_find_bo)
>   
>   #define AMDGPU_GEM_DOMAIN_CPU		0x1
>   #define AMDGPU_GEM_DOMAIN_GTT		0x2
> @@ -201,6 +203,16 @@ struct drm_amdgpu_gem_userptr {
>   	uint32_t		handle;
>   };
>   
> +struct drm_amdgpu_gem_find_bo {
> +       uint64_t                addr;
> +       uint64_t                size;
> +       uint32_t                flags;
> +       /* Resulting GEM handle */
> +       uint32_t                handle;
> +       /* offset in bo */
> +       uint64_t                offset;
> +};
> +
>   /* same meaning as the GB_TILE_MODE and GL_MACRO_TILE_MODE fields */
>   #define AMDGPU_TILING_ARRAY_MODE_SHIFT			0
>   #define AMDGPU_TILING_ARRAY_MODE_MASK			0xf