Re: [PATCH v8 1/6] drm/fourcc: Add modifier definitions for describing Amlogic Video Framebuffer Compression

[Daniel Vetter]

On Thu, Jul 02, 2020 at 05:11:25PM +0200, Neil Armstrong wrote:
> Hi,
> 
> On 02/07/2020 16:15, Daniel Vetter wrote:
> > On Thu, Jul 2, 2020 at 3:34 PM Neil Armstrong <narmstrong@xxxxxxxxxxxx> wrote:
> >>
> >> On 02/07/2020 15:18, Daniel Vetter wrote:
> >>> On Thu, Jul 02, 2020 at 09:23:11AM +0000, Simon Ser wrote:
> >>>> On Thursday, July 2, 2020 9:47 AM, Neil Armstrong <narmstrong@xxxxxxxxxxxx> wrote:
> >>>>
> >>>>> Finally is also adds the Scatter Memory layout, meaning the header contains IOMMU
> >>>>> references to the compressed frames content to optimize memory access
> >>>>> and layout.
> >>>>>
> >>>>> In this mode, only the header memory address is needed, thus the content
> >>>>> memory organization is tied to the current producer execution and cannot
> >>>>> be saved/dumped neither transferrable between Amlogic SoCs supporting this
> >>>>> modifier.
> >>>>
> >>>> Still not sure how to handle this one, since this breaks fundamental
> >>>> assumptions about modifiers.
> >>>
> >>> I wonder whether we should require special allocations for these, and then
> >>> just outright reject mmap on these buffers. mmap on dma-buf isn't a
> >>> required feature.
> >>
> >> Yes, it's the plan to reject mmap on these buffers, but it can't be explained
> >> in the modifiers description and it's a requirement of the producer, not the
> >> consumer.
> > 
> > Hm I think worth to add that as a note to the modifier. Just to make
> > sure. And avoids questions like the one from Simon.
> 
> Something like:
> 
>  /*
>   * Amlogic FBC Scatter Memory layout
>   *
>   * Indicates the header contains IOMMU references to the compressed
>   * frames content to optimize memory access and layout.
>   *
>   * In this mode, only the header memory address is needed, thus the
>   * content memory organization is tied to the current producer
>   * execution and cannot be saved/dumped neither transferrable between
>   * Amlogic SoCs supporting this modifier.
> + *
> + * Due to the nature of the layout, these buffers are not expected to
> + * be accessible by the user-space clients but only accessible by the
> + * hardware producers and consumers.
> + *
> + * The user-space clients should expect a failure while trying to mmap
> + * the DMA-BUF handle returned by the producer.
>   */

lgtm, Acked-by: Daniel Vetter <daniel.vetter@xxxxxxxx>
> 
> Thanks,
> Neil
> 
> > -Daniel
> > 
> >>
> >>>
> >>> That would make sure that userspace cannot look at them.
> >>>
> >>> Also I'm kinda suspecting that there's not unlimited amounts of this magic
> >>> invisible storage available anyway.
> >>> -Daniel
> >>>
> >>
> >> _______________________________________________
> >> dri-devel mailing list
> >> dri-devel@xxxxxxxxxxxxxxxxxxxxx
> >> https://lists.freedesktop.org/mailman/listinfo/dri-devel
> > 
> > 
> > 
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch