public inbox for [email protected]
 help / color / mirror / Atom feed
From: Bijan Mottahedeh <[email protected]>
To: [email protected], [email protected]
Subject: [PATCH v2 07/10] test/buffer-share: add interruptible deadlock test
Date: Fri, 22 Jan 2021 14:54:56 -0800	[thread overview]
Message-ID: <[email protected]> (raw)
In-Reply-To: <[email protected]>

Signed-off-by: Bijan Mottahedeh <[email protected]>
---
 test/buffer-share.c | 85 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 84 insertions(+), 1 deletion(-)

diff --git a/test/buffer-share.c b/test/buffer-share.c
index cbfbfb2..2997333 100644
--- a/test/buffer-share.c
+++ b/test/buffer-share.c
@@ -72,6 +72,7 @@ unsigned long shm_sz = SZ_2M;
 
 int no_read;
 int warned;
+int deadlock;
 int private;
 int verbose;
 
@@ -1095,6 +1096,82 @@ test_exit(int expect)
 	exit(ret);
 }
 
+
+/*
+ * Test for deadlock scenario raised by Pavel.
+ * Test hangs but is interruptible.
+ *
+ * [Pavel's comments]
+ *
+ * Ok, now the original io_uring instance will wait until the attached
+ * once get rid of their references. That's a versatile ground to have
+ * in kernel deadlocks.
+ * 
+ * task1: uring1 = create()
+ * task2: uring2 = create()
+ * task1: uring3 = create(share=uring2);
+ * task2: uring4 = create(share=uring1);
+ * 
+ * task1: io_sqe_buffers_unregister(uring1)
+ * task2: io_sqe_buffers_unregister(uring2)
+ * 
+ * If I skimmed through the code right, that should hang unkillably.
+ */
+static int test_deadlock(void)
+{
+	int i, pid, ret, fd0, fd1;
+	struct io_uring rings[4];
+	struct io_uring_params p;
+
+	for (i = 0; i < 2; i++) {
+		memset(&p, 0, sizeof(p));
+		p.flags = IORING_SETUP_SHARE_BUF;
+		ret = io_uring_queue_init_params(1, &rings[i], &p);
+		if (ret) {
+			verror("queue_init share");
+			return ret;
+		}
+	}
+
+	fd0 = rings[0].ring_fd;
+	fd1 = rings[1].ring_fd;
+
+	pid = fork();
+
+	memset(&p, 0, sizeof(p));
+	p.flags = IORING_SETUP_ATTACH_BUF;
+
+	if (pid) {
+		p.wq_fd = fd1;
+		ret = io_uring_queue_init_params(1, &rings[2], &p);
+	} else {
+		p.wq_fd = fd0;
+		ret = io_uring_queue_init_params(1, &rings[3], &p);
+	}
+
+	if (ret) {
+		verror("queue_init attach");
+		return ret;
+	}
+
+	vinfo(V1, "unregister\n");
+
+	if (pid) {
+		close(fd1);
+		ret = io_uring_unregister_buffers(&rings[0]);
+	} else {
+		close(fd0);
+		ret = io_uring_unregister_buffers(&rings[1]);
+	}
+	
+	vinfo(V1, "unregister done\n");
+
+	if (ret)
+		verror("unregister");
+
+	return ret;
+}
+
 /*
  * main()
  * -> shm_create()
@@ -1122,8 +1199,11 @@ main(int argc, char *argv[])
 	int shmid, pid, ret = 0;
 	char c;
 
-	while ((c = getopt(argc, argv, "ps:v:")) != -1)
+	while ((c = getopt(argc, argv, "dps:v:")) != -1)
 		switch (c) {
+		case 'd':
+			deadlock++;
+			break;
 		case 'p':
 			private++;
 			break;
@@ -1146,6 +1226,9 @@ main(int argc, char *argv[])
 	if (!nids || nids > MAXID)
 		exit(1);
 
+	if (deadlock)
+		exit(test_deadlock());
+
 	shmstat = shmstat_create();
 	if (shmstat == MAP_FAILED)
 		exit(1);
-- 
1.8.3.1


  parent reply	other threads:[~2021-01-22 22:58 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-01-22 22:54 [PATCH v2 00/10] liburing: buffer registration enhancements Bijan Mottahedeh
2021-01-22 22:54 ` [PATCH v2 01/10] liburing: support buffer registration updates Bijan Mottahedeh
2021-01-22 22:54 ` [PATCH v2 02/10] liburing: support buffer registration sharing Bijan Mottahedeh
2021-01-22 22:54 ` [PATCH v2 03/10] test/buffer-register: add buffer registration test Bijan Mottahedeh
2021-01-22 22:54 ` [PATCH v2 04/10] test/buffer-update: add buffer registration update test Bijan Mottahedeh
2021-01-22 22:54 ` [PATCH v2 05/10] test/buffer-share: add buffer registration sharing test Bijan Mottahedeh
2021-01-22 22:54 ` [PATCH v2 06/10] test/buffer-share: add private memory option Bijan Mottahedeh
2021-01-22 22:54 ` Bijan Mottahedeh [this message]
2021-01-22 22:54 ` [PATCH v2 08/10] man/io_uring_setup.2: document buffer registration sharing Bijan Mottahedeh
2021-01-22 22:54 ` [PATCH v2 09/10] man/io_uring_register.2: document buffer registration updates Bijan Mottahedeh
2021-01-22 22:54 ` [PATCH v2 10/10] man/io_uring_enter.2: document IORING_OP_BUFFERS_UPDATE Bijan Mottahedeh

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1611356099-60732-8-git-send-email-bijan.mottahedeh@oracle.com \
    [email protected] \
    [email protected] \
    [email protected] \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox