pg.ddx.io pgsql-docs@postgresql.org mailing list archive
help / color / mirror / Atom feedThe huge_pages parameter doesn't work when shared_memory_type is sysv
4+ messages / 3 participants
[nested] [flat]
* The huge_pages parameter doesn't work when shared_memory_type is sysv
@ 2021-09-27 05:00 PG Doc comments form <noreply@postgresql.org>
2021-10-22 03:59 ` Re: The huge_pages parameter doesn't work when shared_memory_type is sysv Thomas Munro <thomas.munro@gmail.com>
0 siblings, 1 reply; 4+ messages in thread
From: PG Doc comments form @ 2021-09-27 05:00 UTC (permalink / raw)
To: pgsql-docs@lists.postgresql.org; +Cc: exclusion@gmail.com
The following documentation comment has been logged on the website:
Page: https://www.postgresql.org/docs/13/kernel-resources.html
Description:
It seems that the huge_pages parameter is ignored when shared_memory_type is
sysv. The server successfully starts with huge_pages=on, but actually
doesn't use huge pages (relying on /proc/meminfo). I think that this is
worth mentioning in the "Linux Huge Pages" section.
^ permalink raw reply [nested|flat] 4+ messages in thread
* Re: The huge_pages parameter doesn't work when shared_memory_type is sysv
2021-09-27 05:00 The huge_pages parameter doesn't work when shared_memory_type is sysv PG Doc comments form <noreply@postgresql.org>
@ 2021-10-22 03:59 ` Thomas Munro <thomas.munro@gmail.com>
2021-10-22 20:20 ` Re: The huge_pages parameter doesn't work when shared_memory_type is sysv Alexander Lakhin <exclusion@gmail.com>
0 siblings, 1 reply; 4+ messages in thread
From: Thomas Munro @ 2021-10-22 03:59 UTC (permalink / raw)
To: Alexander Lakhin <exclusion@gmail.com>; Pg Docs <pgsql-docs@lists.postgresql.org>
On Mon, Sep 27, 2021 at 9:04 PM PG Doc comments form
<noreply@postgresql.org> wrote:
> The following documentation comment has been logged on the website:
>
> Page: https://www.postgresql.org/docs/13/kernel-resources.html
> Description:
>
> It seems that the huge_pages parameter is ignored when shared_memory_type is
> sysv. The server successfully starts with huge_pages=on, but actually
> doesn't use huge pages (relying on /proc/meminfo). I think that this is
> worth mentioning in the "Linux Huge Pages" section.
Yeah. The thread that added shared_memory_type=sysv had an unfinished
follow-up patch that stalled[1]. The goal of that patch was to add
huge page support for shared_memory_type=sysv on AIX, and it is still
waiting for testing and polishing by an AIX-enabled hacker. That
second patch included exactly such a documentation adjustment, and an
error message too (the idea was to reject that combination on Linux
but allow it on AIX). Here are those bits, extracted.
https://www.postgresql.org/message-id/flat/HE1PR0202MB28126DB4E0B6621CC6A1A91286D90%40HE1PR0202MB281...
Attachments:
0001-Reject-huge_pages-on-if-shared_memory_type-sysv.patch (text/x-patch, 1.9K) download
From c27b1eae1a528f04ab09f60de80ccfa8fbaa7534 Mon Sep 17 00:00:00 2001
From: Thomas Munro <thomas.munro@gmail.com>
Date: Fri, 22 Oct 2021 16:49:22 +1300
Subject: [PATCH] Reject huge_pages=on if shared_memory_type=sysv.
It doesn't work (it could, but hasn't been implemented).
Back-patch to 12, where shared_memory_type arrived.
Reported-by: Alexander Lakhin <exclusion@gmail.com>
Discussion: https://postgr.es/m/163271880203.22789.1125998876173795966@wrigleys.postgresql.org
---
doc/src/sgml/config.sgml | 4 +++-
src/backend/port/sysv_shmem.c | 6 ++++++
2 files changed, 9 insertions(+), 1 deletion(-)
diff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml
index 0bcc6fd322..1c13f995a5 100644
--- a/doc/src/sgml/config.sgml
+++ b/doc/src/sgml/config.sgml
@@ -1664,7 +1664,9 @@ include_dir 'conf.d'
<para>
At present, this setting is supported only on Linux and Windows. The
setting is ignored on other systems when set to
- <literal>try</literal>.
+ <literal>try</literal>. On Linux, it is only supported when
+ <varname>shared_memory_type</varname> is set to <literal>mmap</literal>
+ (the default).
</para>
<para>
diff --git a/src/backend/port/sysv_shmem.c b/src/backend/port/sysv_shmem.c
index cd385c4df6..e7474cee59 100644
--- a/src/backend/port/sysv_shmem.c
+++ b/src/backend/port/sysv_shmem.c
@@ -703,6 +703,12 @@ PGSharedMemoryCreate(Size size,
errmsg("huge pages not supported on this platform")));
#endif
+ /* For now, we don't support huge pages in SysV memory */
+ if (huge_pages == HUGE_PAGES_ON && shared_memory_type != SHMEM_TYPE_MMAP)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("huge pages not supported with the current shared_memory_type setting")));
+
/* Room for a header? */
Assert(size > MAXALIGN(sizeof(PGShmemHeader)));
--
2.30.2
* Re: The huge_pages parameter doesn't work when shared_memory_type is sysv
2021-09-27 05:00 The huge_pages parameter doesn't work when shared_memory_type is sysv PG Doc comments form <noreply@postgresql.org>
2021-10-22 03:59 ` Re: The huge_pages parameter doesn't work when shared_memory_type is sysv Thomas Munro <thomas.munro@gmail.com>
@ 2021-10-22 20:20 ` Alexander Lakhin <exclusion@gmail.com>
2021-10-26 00:19 ` Re: The huge_pages parameter doesn't work when shared_memory_type is sysv Thomas Munro <thomas.munro@gmail.com>
0 siblings, 1 reply; 4+ messages in thread
From: Alexander Lakhin @ 2021-10-22 20:20 UTC (permalink / raw)
To: Thomas Munro <thomas.munro@gmail.com>; Pg Docs <pgsql-docs@lists.postgresql.org>
Hello Thomas,
22.10.2021 06:59, Thomas Munro wrote:
> That
> second patch included exactly such a documentation adjustment, and an
> error message too (the idea was to reject that combination on Linux
> but allow it on AIX). Here are those bits, extracted.
>
> https://www.postgresql.org/message-id/flat/HE1PR0202MB28126DB4E0B6621CC6A1A91286D90%40HE1PR0202MB281...
Thanks for the fix! It works and looks correct to me.
Best regards,
Alexander
^ permalink raw reply [nested|flat] 4+ messages in thread
* Re: The huge_pages parameter doesn't work when shared_memory_type is sysv
2021-09-27 05:00 The huge_pages parameter doesn't work when shared_memory_type is sysv PG Doc comments form <noreply@postgresql.org>
2021-10-22 03:59 ` Re: The huge_pages parameter doesn't work when shared_memory_type is sysv Thomas Munro <thomas.munro@gmail.com>
2021-10-22 20:20 ` Re: The huge_pages parameter doesn't work when shared_memory_type is sysv Alexander Lakhin <exclusion@gmail.com>
@ 2021-10-26 00:19 ` Thomas Munro <thomas.munro@gmail.com>
0 siblings, 0 replies; 4+ messages in thread
From: Thomas Munro @ 2021-10-26 00:19 UTC (permalink / raw)
To: Alexander Lakhin <exclusion@gmail.com>; +Cc: Pg Docs <pgsql-docs@lists.postgresql.org>
On Sat, Oct 23, 2021 at 9:20 AM Alexander Lakhin <exclusion@gmail.com> wrote:
> Thanks for the fix! It works and looks correct to me.
Pushed. Thanks.
^ permalink raw reply [nested|flat] 4+ messages in thread
end of thread, other threads:[~2021-10-26 00:19 UTC | newest]
Thread overview: 4+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2021-09-27 05:00 The huge_pages parameter doesn't work when shared_memory_type is sysv PG Doc comments form <noreply@postgresql.org>
2021-10-22 03:59 ` Thomas Munro <thomas.munro@gmail.com>
2021-10-22 20:20 ` Alexander Lakhin <exclusion@gmail.com>
2021-10-26 00:19 ` Thomas Munro <thomas.munro@gmail.com>
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox