Re: [PING] [PATCH v2 2/2] i386: Canonicalize PAVG builtins to IFN_AVG_CEIL [PR122715]

Ping for this patch.
This is the x86 part of PR122715 v2, reworked to canonicalize PAVG target builtins to IFN_AVG_CEIL as suggested.
Bootstrapped on x86_64-pc-linux-gnu; the new x86 tests and existing PAVG tests pass.
Is this OK for trunk?
Thanks,
Odysseas
________________________________
From: Odysseas Georgoudis <[email protected]>
Sent: 23 July 2026 02:00
To: Richard Biener <[email protected]>
Cc: [email protected] <[email protected]>; [email protected] <[email protected]>
Subject: [PATCH v2 2/2] i386: Canonicalize PAVG builtins to IFN_AVG_CEIL [PR122715]

  This is the x86 part of v2, reworked to canonicalize PAVG target builtins to IFN_AVG_CEIL as suggested.

  Bootstrapped on x86_64-pc-linux-gnu. The new x86 tests and the existing PAVG tests pass.
________________________________
From: Richard Biener <[email protected]>
Sent: 22 July 2026 14:17
To: Odysseas Georgoudis <[email protected]>
Cc: [email protected] <[email protected]>; [email protected] <[email protected]>
Subject: Re: [PATCH] middle-end, i386: Fold averages with equal operands [PR122715]

On Wed, Jul 15, 2026 at 7:43 AM Odysseas Georgoudis <[email protected]> wrote:
>
> This patch fixes PR middle-end/122715 by folding IFN_AVG_FLOOR (x, x) and IFN_AVG_CEIL (x, x) to x.
>
> x86 PAVG intrinsics are represented by target builtins instead of IFN_AVG_CEIL. The patch therefore also folds calls with equal operands in ix86_gimple_fold_builtin. For masked forms, the fold is applied only when every lane is enabled by the mask.

The match.pd / gimple testcase part is OK.  Please split the x86
target part to a separate
patch (we might want to instead fold the target builtin to the
internal function call?)

Thanks,
Richard.

> Bootstrapped on x86_64-pc-linux-gnu. The new tests and the existing PAVG focused tests pass. The target-independent match.pd fold was also tested using an AArch64 cross-compiler.