Re: [PATCH v6 3/7] ppc/svp64: support setvl instructions

[Dmitry Selyutin <ghostmansd@gmail.com>]

On Mon, Aug 15, 2022 at 9:18 AM Jan Beulich <jbeulich@suse.com> wrote:
> Out of curiosity - was it intentional to omit a case with the last
> operand non-zero?

Nice catch! This was not intentional, I'll submit the patch to handle
this case. Perhaps I either forgot to add this line, or removed it
somehow...

On Mon, Aug 15, 2022 at 3:59 PM lkcl <luke.leighton@gmail.com> wrote:
> one for Dmitry - i believe these are all auto-generated?

Except for the very first revisions with huge tests, I added the tests
manually. One of the issues for generation is that we don't, in fact,
maintain information about whether some operands should be non-zero.
Another issue was that, until very recently, we had to generate the
instructions manually, not from the parsed fields. In fact, we still
encode these instructions manually even now, but at least now we have
instruments to stop doing it.

For now, I'll simply update the test; in an ideal world, yes, we
should generate these tests.

-- 
Best regards,
Dmitry Selyutin