Date: Fri, 13 Feb 2026 08:03:04 +0000
From: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>
To: Heikki Linnakangas <hlinnaka@iki.fi>
Cc: Andres Freund <andres@anarazel.de>,
	"pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Subject: Re: PGPROC alignment (was Re: pgsql: Separate
 RecoveryConflictReasons from procsignals)
Message-ID: <aY7auF2/xuT58vRI@ip-10-97-1-34.eu-west-3.compute.internal>
References: <aYtMbb28qxqLrhvC@ip-10-97-1-34.eu-west-3.compute.internal>
 <a6159474-a06b-4874-9e4f-f06516995860@iki.fi>
 <aYtcMg5hMdGdhX4q@alap3.anarazel.de>
 <1cb0d7e9-d6dd-4517-a7cd-0ad98e1207f3@iki.fi>
 <aYtsrQ-kmA8nqeiH@alap3.anarazel.de>
 <aYuDz2kvi7haNORb@ip-10-97-1-34.eu-west-3.compute.internal>
 <aYuIrKbh1kW36xbc@alap3.anarazel.de>
 <3dd6f70c-b94d-4428-8e75-74a7136396be@iki.fi>
 <aYwISFVCAJon5ZxT@ip-10-97-1-34.eu-west-3.compute.internal>
 <787be980-0878-4f4a-be01-d042ab5d370e@iki.fi>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <787be980-0878-4f4a-be01-d042ab5d370e@iki.fi>
Archived-At: 
 <https://www.postgresql.org/message-id/aY7auF2/xuT58vRI%40ip-10-97-1-34.eu-west-3.compute.internal>
Precedence: bulk

Hi,

On Wed, Feb 11, 2026 at 12:03:51PM +0200, Heikki Linnakangas wrote:
> On 11/02/2026 06:40, Bertrand Drouvot wrote:
> > A few comments:
> > 
> > 0001:
> > 
> > + * and (b) to make the multiplication / division to convert between PGPROC *
> > + * and ProcNumber be a little cheaper
> > 
> > Is that correct if PGPROC size is not a power of 2?
> 
> You're right, it's not.

Looking more closely at:

"
/* GCC supports aligned and packed */
#if defined(__GNUC__)
#define pg_attribute_aligned(a) __attribute__((aligned(a)))
#define pg_attribute_packed() __attribute__((packed))
#elif defined(_MSC_VER)
/*
 * MSVC supports aligned.
 *
 * Packing is also possible but only by wrapping the entire struct definition
 * which doesn't fit into our current macro declarations.
 */
#define pg_attribute_aligned(a) __declspec(align(a))
#else
/*
 * NB: aligned and packed are not given default definitions because they
 * affect code functionality; they *must* be implemented by the compiler
 * if they are to be used.
 */
#endif
"

and what the patch adds:

+/*
+ * If compiler understands aligned pragma, use it to align the struct at cache
+ * line boundaries.  This is just for performance, to (a) avoid false sharing
+ * and (b) to make the multiplication / division to convert between PGPROC *
+ * and ProcNumber be a little cheaper.
+ */
+#if defined(pg_attribute_aligned)
+                       pg_attribute_aligned(PG_CACHE_LINE_SIZE)
+#endif
+PGPROC;

It means that PGPROC is "acceptable" without padding (on compiler that does not
understand the aligned attribute).

OTOH, looking at:

"
typedef union WALInsertLockPadded
{
    WALInsertLock l;
    char        pad[PG_CACHE_LINE_SIZE];
} WALInsertLockPadded;
"

It seems to mean that WALInsertLockPadded is unacceptable without padding (since
it's not using pg_attribute_aligned()).

That looks ok to see PGPROC as an "acceptable" one, if not, should we use the
union trick?

Regards,

-- 
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com