[virglrenderer-devel] Potentially attack surface for virgl_renderer_submit_cmd

[Dave Airlie]

On 19 July 2018 at 06:17, Po-Hsien Wang <pwang at chromium.org> wrote:
> Hi Dave,
> Thanks for pointing this out. I'm not aware of those variable and looks into
> the fill_caps path as I think most caller use this to determine its
> behavior.
> Based on briefly examination, it seems those have_ variable is initialized
> in a similar way as fill_caps. I'm wondering why we did this in two
> different code path? Is there a specific reason for this?

No this is definitely an area we could do better in, it just sort of
grew this way,

We could likely consolidate this into a single code path that gathers the
data and then builds the caps from it.

Dave.