Feedback-ID: ia2694551:Fastmail
Date: Fri, 1 Jul 2022 17:41:05 +0200
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Pg Hackers <pgsql-hackers@lists.postgresql.org>
Subject: EINTR in ftruncate()
Message-ID: <20220701154105.jjfutmngoedgiad3@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="knr42z5qyvhth5as"
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
Archived-At: 
 <https://www.postgresql.org/message-id/20220701154105.jjfutmngoedgiad3%40alvherre.pgsql>
Precedence: bulk


--knr42z5qyvhth5as
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit

Nicola Contu reported two years ago to pgsql-general[1] that they were
having sporadic query failures, because EINTR is reported on some system
call.  I have been told that the problem persists, though it is very
infrequent.  I propose the attached patch.  Kyotaro proposed a slightly
different patch which also protects write(), but I think that's not
necessary.

Thomas M. produced some more obscure theories for other things that
could fail, but I think we should patch this problem first, which seems
the most obvious one, and deal with others if and when they are
reported.

[1] https://www.postgresql.org/message-id/CAMTZZh2V%2B0wJVgSqTVvXUAVMduF57Uxubvvw58%3DkbOae%2B53%2BQQ%40mail.gmail.com

-- 
Álvaro Herrera        Breisgau, Deutschland  —  https://www.EnterpriseDB.com/
"Use it up, wear it out, make it do, or do without"

--knr42z5qyvhth5as
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v2-0001-retry-ftruncate.patch"

From 466ed63b9b399c2914aa44f56f56e044341a77f5 Mon Sep 17 00:00:00 2001
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Fri, 1 Jul 2022 17:16:33 +0200
Subject: [PATCH v2] retry ftruncate

---
 src/backend/storage/ipc/dsm_impl.c | 17 +++++++++++++++--
 1 file changed, 15 insertions(+), 2 deletions(-)

diff --git a/src/backend/storage/ipc/dsm_impl.c b/src/backend/storage/ipc/dsm_impl.c
index 873867e856..4f35e7e3d1 100644
--- a/src/backend/storage/ipc/dsm_impl.c
+++ b/src/backend/storage/ipc/dsm_impl.c
@@ -362,8 +362,21 @@ dsm_impl_posix_resize(int fd, off_t size)
 {
 	int			rc;
 
-	/* Truncate (or extend) the file to the requested size. */
-	rc = ftruncate(fd, size);
+	/*
+	 * Truncate (or extend) the file to the requested size.  We may get
+	 * interrupted.  If so, just retry unless there is an interrupt pending.
+	 * This avoids the possibility of looping forever if another backend is
+	 * repeatedly trying to interrupt us.
+	 */
+	for (;;)
+	{
+		errno = 0;
+		rc = ftruncate(fd, size);
+		if (rc == 0)
+			break;
+		if (errno != EINTR || ProcDiePending || QueryCancelPending)
+			return rc;
+	}
 
 	/*
 	 * On Linux, a shm_open fd is backed by a tmpfs file.  After resizing with
-- 
2.30.2


--knr42z5qyvhth5as--