Обсуждение: BUG #19368: f54af9f does not create the correct macro for autotools build

The following bug has been logged on the website:

Bug reference:      19368
Logged by:          Evan Si
Email address:      evsi@amazon.com
PostgreSQL version: 18.1
Operating system:   Amazon Linux 2, Amazon Linux 2023
Description:

https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=f54af9f2679d

The commit f54af9f2679d added a macro HAVE_LIBURING_QUEUE_INIT_MEM that is
checked to see if slowdown from memory mappings can be avoided when using
io_uring.
This HAVE_LIBURING_QUEUE_INIT_MEM is set by Meson build after checking
liburing.
However, the autotools build seems to set a different macro,
HAVE_IO_URING_QUEUE_INIT_MEM, which is never used.

This means when the server is built with autotools, the slowdown when using
io_uring can never be avoided.

PG Bug reporting form <noreply@postgresql.org> writes:
> The commit f54af9f2679d added a macro HAVE_LIBURING_QUEUE_INIT_MEM that is
> checked to see if slowdown from memory mappings can be avoided when using
> io_uring.
> This HAVE_LIBURING_QUEUE_INIT_MEM is set by Meson build after checking
> liburing.
> However, the autotools build seems to set a different macro,
> HAVE_IO_URING_QUEUE_INIT_MEM, which is never used.

> This means when the server is built with autotools, the slowdown when using
> io_uring can never be avoided.

This is indeed busted, but I'd argue that the autoconf build has the
right macro name and meson and the C code are using the wrong one:

  if cc.has_function('io_uring_queue_init_mem',
      dependencies: liburing, args: test_c_args)
    cdata.set('HAVE_LIBURING_QUEUE_INIT_MEM', 1)
  endif

That's clearly not following the "HAVE_function" convention.

            regards, tom lane

On Wed, Dec 31, 2025 at 7:59 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> PG Bug reporting form <noreply@postgresql.org> writes:
> > The commit f54af9f2679d added a macro HAVE_LIBURING_QUEUE_INIT_MEM that is
> > checked to see if slowdown from memory mappings can be avoided when using
> > io_uring.
> > This HAVE_LIBURING_QUEUE_INIT_MEM is set by Meson build after checking
> > liburing.
> > However, the autotools build seems to set a different macro,
> > HAVE_IO_URING_QUEUE_INIT_MEM, which is never used.
>
> > This means when the server is built with autotools, the slowdown when using
> > io_uring can never be avoided.
>
> This is indeed busted, but I'd argue that the autoconf build has the
> right macro name and meson and the C code are using the wrong one:
>
>   if cc.has_function('io_uring_queue_init_mem',
>       dependencies: liburing, args: test_c_args)
>     cdata.set('HAVE_LIBURING_QUEUE_INIT_MEM', 1)
>   endif
>
> That's clearly not following the "HAVE_function" convention.

Agreed. The attached patch should fix the issue.

Regards,

--
Masahiko Sawada
Amazon Web Services: https://aws.amazon.com

Masahiko Sawada <sawada.mshk@gmail.com> writes:
> On Wed, Dec 31, 2025 at 7:59 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> This is indeed busted, but I'd argue that the autoconf build has the
>> right macro name and meson and the C code are using the wrong one:

> Agreed. The attached patch should fix the issue.

You beat me to it ;-).  Patch LGTM.

            regards, tom lane

On Wed, Dec 31, 2025 at 8:52 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> Masahiko Sawada <sawada.mshk@gmail.com> writes:
> > On Wed, Dec 31, 2025 at 7:59 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> >> This is indeed busted, but I'd argue that the autoconf build has the
> >> right macro name and meson and the C code are using the wrong one:
>
> > Agreed. The attached patch should fix the issue.
>
> You beat me to it ;-).  Patch LGTM.

Thank you for reviewing the patch! Pushed.

Regards,

--
Masahiko Sawada
Amazon Web Services: https://aws.amazon.com

Hi,

On 2025-12-31 11:20:17 -0800, Masahiko Sawada wrote:
> On Wed, Dec 31, 2025 at 8:52 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> >
> > Masahiko Sawada <sawada.mshk@gmail.com> writes:
> > > On Wed, Dec 31, 2025 at 7:59 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> > >> This is indeed busted, but I'd argue that the autoconf build has the
> > >> right macro name and meson and the C code are using the wrong one:

Ugh, yea, not sure what I running on that day.

> > > Agreed. The attached patch should fix the issue.
> >
> > You beat me to it ;-).  Patch LGTM.
> 
> Thank you for reviewing the patch! Pushed.

Thanks!

Greetings,

Andres Freund