Re: [PATCH v6 0/7] Add mediate-drm secure flow for SVP

["Jason-JH Lin (林睿祥)" <Jason-JH.Lin@mediatek.com>]

Hi Maxime,

On Fri, 2024-06-21 at 11:00 +0200, mripard@kernel.org wrote:
> Hi,
> 
> On Tue, Jun 11, 2024 at 09:13:03AM GMT, Jason-JH Lin (林睿祥) wrote:
> > Hi Maxime,
> > 

[snip]

> > I have changed user space to use DMA_HEAP_IOCTL_ALLOC to allocate
> > secure buffer, but I still encounter the problem of determining
> > whether
> > the buffer is secure in mediatek-drm driver to add some secure
> > configure for hardware.
> > 
> > 
> > As the comment in [1], dma driver won't provide API for use.
> > [1]: 
> > 
https://patchwork.kernel.org/project/linux-mediatek/patch/20240515112308.10171-3-yong.wu@mediatek.com/#25857255
> > 
> > 
> > So I use name checking at [PATCH v6 3/7] like this currently:
> > 
> > struct drm_gem_object *mtk_gem_prime_import_sg_table(struct
> > drm_device
> > *dev,
> >             struct dma_buf_attachment *attach, struct sg_table *sg)
> > {
> >     struct mtk_gem_obj *mtk_gem;
> > 
> >     /* check if the entries in the sg_table are contiguous */
> >     if (drm_prime_get_contiguous_size(sg) < attach->dmabuf->size) {
> >         DRM_ERROR("sg_table is not contiguous");
> >         return ERR_PTR(-EINVAL);
> >     }
> > 
> >     mtk_gem = mtk_gem_init(dev, attach->dmabuf->size);
> >     if (IS_ERR(mtk_gem))
> >         return ERR_CAST(mtk_gem);
> > 
> > +   mtk_gem->secure = (!strncmp(attach->dmabuf->exp_name,
> > "restricted",
> > 10));
> >     mtk_gem->dma_addr = sg_dma_address(sg->sgl);
> > +   mtk_gem->size = attach->dmabuf->size;
> >     mtk_gem->sg = sg;
> > 
> >     return &mtk_gem->base;
> > }
> > 
> > But I want to change this name checking to the information brought
> > from
> > user space.
> > I tried to use arg->flags to append the secure flag in user space
> > and
> > call drmPrimeHandleToFD() to pass it to DRM driver, but it will be
> > blocked by at the beginning of the drm_prime_handle_to_fd_ioctl().
> 
> I agree with you, it's something to discuss mostly with the dma-buf
> maintainers but it would be better to just set a flag on the dma-buf,
> and use that flag whenever necessary.
> 
> It might be related to the recent work I did to introduce allocation
> flags too:
> 
https://lore.kernel.org/linux-media/20240515-dma-buf-ecc-heap-v1-0-54cbbd049511@kernel.org/
> 

Thanks for your information!
I'll discuss with dma-buf maintainer in series [1].

Regards,
Jason-JH Lin

> Maxime