Re: LLVM JITLink attempt II (WIP)

From: Thomas Munro <thomas(dot)munro(at)gmail(dot)com>
To: Anthonin Bonnefoy <anthonin(dot)bonnefoy(at)datadoghq(dot)com>
Cc: PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>, Gregory Burd <greg(at)burd(dot)me>, Matheus Alcantara <matheusssilv97(at)gmail(dot)com>
Subject: Re: LLVM JITLink attempt II (WIP)
Date: 2026-01-11 04:14:51
Message-ID: CA+hUKG+5Qnsx5A3QdTT4_g8JV2HN57HkxDHqNM2OCJPKxJJW3g@mail.gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Fri, Jan 9, 2026 at 2:33 AM Anthonin Bonnefoy
<anthonin(dot)bonnefoy(at)datadoghq(dot)com> wrote:
> On Tue, Jan 6, 2026 at 1:09 AM Thomas Munro <thomas(dot)munro(at)gmail(dot)com> wrote:
> > It sounds like we might need to wait for or write a similar
> > function for perf support.
>
> Yeah, there's no C api provided by LLVM yet. I've just opened a
> tentative PR[1] to add it.

Thanks Anthonin! The configure script needs the library for the
"orcdebugging" component (Meson must be adding all components
automatically), and we should check the return value. I also figured
we should set the error report handler as soon as possible, just in
case it wants to tell us something. v4 attached.

jit_profiling_support=1 doesn't fail nicely on a system without perf,
but I don't think it's the fault of your patch. I posted some details
to your PR. Let's see if it can land in time... I think there are
just a couple of days left, so it might just be too late for 22.

I suppose we should only use this for ELF systems initially, given
that debugging support is missing for Macs.

Attachment Content-Type Size
v4-0001-jit-Use-JITLink-for-LLVM-22.patch text/x-patch 5.7 KB

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message zengman 2026-01-11 04:24:22 Re: [PATCH] Fix minor issues in astreamer_zstd.c
Previous Message Tatsuro Yamada 2026-01-11 04:01:33 Re: Inconsistent terminology for -j/--jobs option in documentation