Re: [PATCH 5.10 154/167] Revert "drm/i915: Propagate errors on awaiting already signaled fences"

[Greg Kroah-Hartman]

On Tue, Jul 27, 2021 at 11:35:46PM +0200, Pavel Machek wrote:
> Hi!
> 
> According to changelog, this introduces security hole.
> 
> > From: Jason Ekstrand <jason@xxxxxxxxxxxxxx>
> > 
> > commit 3761baae908a7b5012be08d70fa553cc2eb82305 upstream.
> > 
> > This reverts commit 9e31c1fe45d555a948ff66f1f0e3fe1f83ca63f7.  Ever
> > since that commit, we've been having issues where a hang in one
> > client
> 
> Hmm. Sounds like problem I'm seeing in mainline. So... good to know.
> 
> > For backporters: Please note that you _must_ have a backport of
> > https://lore.kernel.org/dri-devel/20210602164149.391653-2-jason@xxxxxxxxxxxxxx/
> > for otherwise backporting just this patch opens up a security bug.
> 
> AFAICT we don't have that c9d9fdbc108af8915d3f497bbdf3898bf8f321b8
> drm/i915: Revert "drm/i915/gem: Asynchronous cmdparser" in 5.10 tree.
> 
> Hmm, and it needs follow up fix:
> 6e0b6528d783b2b87bd9e1bea97cf4dac87540d7 drm/i915: Correct the docs
> for intel_engine_cmd_parser.
> 
> (Someone please double check this).

thanks, let me drop this for now.

Jason, you sent me a single patch to backport, should this be 3 patches
instead?

thanks,

greg k-h