On 08/02/2021 20:55, Dave Airlie wrote:
On Mon, 8 Feb 2021 at 20:53, Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> wrote:
Re-enable secure dispatch for gen6/gen7, primarily to workaround the
command parser and overly zealous command validation on Haswell. For
example this prevents making accurate measurements using a journal for
store results from the GPU without CPU intervention.
There's 31 patches in this series, and I can't find any 00/31 or
justification for any of this work.
I see patches like this which seem to undo work done for security
reasons under CVE patches with no oversight.
Again, the GT team is not doing the right thing here, stop focusing on
individual pieces of Chris's work, push back for high level
architectural reviews and I want them on the list in public.
All I want from the GT team in the next pull request is dma_resv
locking work and restoring the hangcheck timers that seems like a
regression that Chris found acceptable and nobody has pushed back on.
For like the 500th time, if you want DG1 and stuff in the tree, stop
this shit already, real reviewers, high-level architectural reviews,
NAK the bullshit in public on the list.
Since it's mostly been me reviewing the scheduler improvements in this
series, I gather we have met and talked, or that you have at least have
been following me closely enough to conclude I am not a "real" reviewer.
Fair?
Regards,
Tvrtko
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
