On 7/8/2026 12:40 AM, Uros Bizjak wrote:
On Wed, Jul 8, 2026 at 3:50 AM <[email protected]> wrote:
From: MITSUNARI Shigeo <[email protected]>
Ping.
v8 was posted on Jun 19 and I have not seen any further comments, so a
gentle ping. For convenience, the current state:
- [1/2] expmed: all five points from the v6 review are addressed, and
v8 fixes the sign-compare bootstrap failure the Linaro tcwg CI hit on
v7 (classic_cost is now unsigned, per Andrew's suggestion). There are
no outstanding review comments that I am aware of.
- [2/2] i386: adds the BMI2 MULX highpart pattern; unchanged since v5.
The two parts touch different areas ([1/2] middle-end/expmed, [2/2]
i386), so they likely need separate acks. Jeff, would [1/2] be OK for
trunk, or is anything else needed? For [2/2] I would appreciate a look
from an x86 maintainer.
I already approved 2/2 a while ago [1].
[1] https://gcc.gnu.org/pipermail/gcc-patches/2026-June/719916.html
Yea, I still need to go through the V8 patch. Given I think the biggest
concerns were generalizing so that we're not so tied to one size I'd
think when I get the time it should be straightforward to work through.
My queue has gotten deep over the last month or so.
jeff
