Re: [PATCH v7 0/3] arm64/sve: Improve performance when handling SVE access traps

On Wed, Jul 21, 2021 at 03:33:54PM +0100, Dave Martin wrote:

While I think this was a worthwhile experiment, my concern here is that
while the approach taken in this series is reasonable, it doesn't seem
to reduce the amount of code or result in a net simplification.  From my
side I think it's probably best to stick with what we have, until
someone comes up with something that's clearly easier to understand.

I did find it was making it easier to understand some of what was going
on TBH - I forget which specific bits but I found the whole model of
specifying the goal state at a higher level clarified things for me.
It's definitely not saving much in the way of code though and the code
that was already merged to do the zeroing in place gives us most of the
win with dramatically less code, it just doesn't help if we do context
switch.

So, I'd still favour the version based on Julien's code, which is more
of an incremental change to what we already had (and I think was most of
the way there in your post recent version of it).

I prefer Julien's approach too, the requirement to trigger the slow path
on return to userspace doesn't really work with the newer approach
AFAICT.  If this gets resurrected I'll go back to the last _NO_FLUSH
version.