From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752051AbeFAM62 (ORCPT ); Fri, 1 Jun 2018 08:58:28 -0400 Received: from mail-wr0-f193.google.com ([209.85.128.193]:41885 "EHLO mail-wr0-f193.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751769AbeFAM6V (ORCPT ); Fri, 1 Jun 2018 08:58:21 -0400 X-Google-Smtp-Source: ADUXVKLSUbnZ/EIa3S9LdarfdFb/oEiHNPyUI2LpEHh52PRxGru8jFlJplPMb8hM78yeUW+8aSBiPFwKy26SkBRLzRI= MIME-Version: 1.0 In-Reply-To: <20180531161708.5894-3-michel@daenzer.net> References: <20180531161708.5894-1-michel@daenzer.net> <20180531161708.5894-3-michel@daenzer.net> From: Alex Deucher Date: Fri, 1 Jun 2018 08:58:19 -0400 Message-ID: Subject: Re: [PATCH 3/3] drm/amdgpu: Add documentation for PRIME related code To: =?UTF-8?Q?Michel_D=C3=A4nzer?= Cc: amd-gfx list , linux-doc@vger.kernel.org, LKML , Maling list - DRI developers , Jonathan Corbet Content-Type: text/plain; charset="UTF-8" Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from quoted-printable to 8bit by mail.home.local id w51CwwPD003677 On Thu, May 31, 2018 at 12:17 PM, Michel Dänzer wrote: > From: Michel Dänzer > > Signed-off-by: Michel Dänzer Series is: Reviewed-by: Alex Deucher > --- > Documentation/gpu/amdgpu.rst | 14 +++ > drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c | 119 ++++++++++++++++++++++ > 2 files changed, 133 insertions(+) > > diff --git a/Documentation/gpu/amdgpu.rst b/Documentation/gpu/amdgpu.rst > index 41a14e4aa4ac..f557866f6788 100644 > --- a/Documentation/gpu/amdgpu.rst > +++ b/Documentation/gpu/amdgpu.rst > @@ -4,3 +4,17 @@ > > The drm/amdgpu driver supports all AMD Radeon GPUs based on the Graphics Core > Next (GCN) architecture. > + > +Core Driver Infrastructure > +========================== > + > +This section covers core driver infrastructure. > + > +PRIME Buffer Sharing > +-------------------- > + > +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c > + :doc: PRIME Buffer Sharing > + > +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c > + :internal: > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c > index 4683626b065f..d1f05489595b 100644 > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c > @@ -23,6 +23,14 @@ > * > * Authors: Alex Deucher > */ > + > +/** > + * DOC: PRIME Buffer Sharing > + * > + * The following callback implementations are used for :ref:`sharing GEM buffer > + * objects between different devices via PRIME `. > + */ > + > #include > > #include "amdgpu.h" > @@ -32,6 +40,14 @@ > > static const struct dma_buf_ops amdgpu_dmabuf_ops; > > +/** > + * amdgpu_gem_prime_get_sg_table - &drm_driver.gem_prime_get_sg_table > + * implementation > + * @obj: GEM buffer object > + * > + * Returns: > + * A scatter/gather table for the pinned pages of the buffer object's memory. > + */ > struct sg_table *amdgpu_gem_prime_get_sg_table(struct drm_gem_object *obj) > { > struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); > @@ -40,6 +56,15 @@ struct sg_table *amdgpu_gem_prime_get_sg_table(struct drm_gem_object *obj) > return drm_prime_pages_to_sg(bo->tbo.ttm->pages, npages); > } > > +/** > + * amdgpu_gem_prime_vmap - &dma_buf_ops.vmap implementation > + * @obj: GEM buffer object > + * > + * Sets up an in-kernel virtual mapping of the buffer object's memory. > + * > + * Returns: > + * The virtual address of the mapping or an error pointer. > + */ > void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj) > { > struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); > @@ -53,6 +78,13 @@ void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj) > return bo->dma_buf_vmap.virtual; > } > > +/** > + * amdgpu_gem_prime_vunmap - &dma_buf_ops.vunmap implementation > + * @obj: GEM buffer object > + * @vaddr: virtual address (unused) > + * > + * Tears down the in-kernel virtual mapping of the buffer object's memory. > + */ > void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr) > { > struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); > @@ -60,6 +92,17 @@ void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr) > ttm_bo_kunmap(&bo->dma_buf_vmap); > } > > +/** > + * amdgpu_gem_prime_mmap - &drm_driver.gem_prime_mmap implementation > + * @obj: GEM buffer object > + * @vma: virtual memory area > + * > + * Sets up a userspace mapping of the buffer object's memory in the given > + * virtual memory area. > + * > + * Returns: > + * 0 on success or negative error code. > + */ > int amdgpu_gem_prime_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma) > { > struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); > @@ -94,6 +137,19 @@ int amdgpu_gem_prime_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma > return ret; > } > > +/** > + * amdgpu_gem_prime_import_sg_table - &drm_driver.gem_prime_import_sg_table > + * implementation > + * @dev: DRM device > + * @attach: DMA-buf attachment > + * @sg: Scatter/gather table > + * > + * Import shared DMA buffer memory exported by another device. > + * > + * Returns: > + * A new GEM buffer object of the given DRM device, representing the memory > + * described by the given DMA-buf attachment and scatter/gather table. > + */ > struct drm_gem_object * > amdgpu_gem_prime_import_sg_table(struct drm_device *dev, > struct dma_buf_attachment *attach, > @@ -132,6 +188,19 @@ amdgpu_gem_prime_import_sg_table(struct drm_device *dev, > return ERR_PTR(ret); > } > > +/** > + * amdgpu_gem_map_attach - &dma_buf_ops.attach implementation > + * @dma_buf: shared DMA buffer > + * @target_dev: target device > + * @attach: DMA-buf attachment > + * > + * Makes sure that the shared DMA buffer can be accessed by the target device. > + * For now, simply pins it to the GTT domain, where it should be accessible by > + * all DMA devices. > + * > + * Returns: > + * 0 on success or negative error code. > + */ > static int amdgpu_gem_map_attach(struct dma_buf *dma_buf, > struct device *target_dev, > struct dma_buf_attachment *attach) > @@ -181,6 +250,14 @@ static int amdgpu_gem_map_attach(struct dma_buf *dma_buf, > return r; > } > > +/** > + * amdgpu_gem_map_detach - &dma_buf_ops.detach implementation > + * @dma_buf: shared DMA buffer > + * @attach: DMA-buf attachment > + * > + * This is called when a shared DMA buffer no longer needs to be accessible by > + * the other device. For now, simply unpins the buffer from GTT. > + */ > static void amdgpu_gem_map_detach(struct dma_buf *dma_buf, > struct dma_buf_attachment *attach) > { > @@ -202,6 +279,13 @@ static void amdgpu_gem_map_detach(struct dma_buf *dma_buf, > drm_gem_map_detach(dma_buf, attach); > } > > +/** > + * amdgpu_gem_prime_res_obj - &drm_driver.gem_prime_res_obj implementation > + * @obj: GEM buffer object > + * > + * Returns: > + * The buffer object's reservation object. > + */ > struct reservation_object *amdgpu_gem_prime_res_obj(struct drm_gem_object *obj) > { > struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); > @@ -209,6 +293,18 @@ struct reservation_object *amdgpu_gem_prime_res_obj(struct drm_gem_object *obj) > return bo->tbo.resv; > } > > +/** > + * amdgpu_gem_begin_cpu_access - &dma_buf_ops.begin_cpu_access implementation > + * @dma_buf: shared DMA buffer > + * @direction: direction of DMA transfer > + * > + * This is called before CPU access to the shared DMA buffer's memory. If it's > + * a read access, the buffer is moved to the GTT domain if possible, for optimal > + * CPU read performance. > + * > + * Returns: > + * 0 on success or negative error code. > + */ > static int amdgpu_gem_begin_cpu_access(struct dma_buf *dma_buf, > enum dma_data_direction direction) > { > @@ -253,6 +349,18 @@ static const struct dma_buf_ops amdgpu_dmabuf_ops = { > .vunmap = drm_gem_dmabuf_vunmap, > }; > > +/** > + * amdgpu_gem_prime_export - &drm_driver.gem_prime_export implementation > + * @dev: DRM device > + * @gobj: GEM buffer object > + * @flags: flags like DRM_CLOEXEC and DRM_RDWR > + * > + * The main work is done by the &drm_gem_prime_export helper, which in turn > + * uses &amdgpu_gem_prime_res_obj. > + * > + * Returns: > + * Shared DMA buffer representing the GEM buffer object from the given device. > + */ > struct dma_buf *amdgpu_gem_prime_export(struct drm_device *dev, > struct drm_gem_object *gobj, > int flags) > @@ -273,6 +381,17 @@ struct dma_buf *amdgpu_gem_prime_export(struct drm_device *dev, > return buf; > } > > +/** > + * amdgpu_gem_prime_import - &drm_driver.gem_prime_import implementation > + * @dev: DRM device > + * @dma_buf: Shared DMA buffer > + * > + * The main work is done by the &drm_gem_prime_import helper, which in turn > + * uses &amdgpu_gem_prime_import_sg_table. > + * > + * Returns: > + * GEM buffer object representing the shared DMA buffer for the given device. > + */ > struct drm_gem_object *amdgpu_gem_prime_import(struct drm_device *dev, > struct dma_buf *dma_buf) > { > -- > 2.17.0 > > _______________________________________________ > amd-gfx mailing list > amd-gfx@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/amd-gfx