Skip to contentSkip to conversation

inotify07: Add test for kernel crash during event notification

PATCH In review Jan Kara 2 participants 4 messages May 3, 2018 – May 10, 2018 2 messages not in the archive

lore
Share & tools
lore

Jan Kara

author
#0 lore raw

PATCH 1/1 inotify07: Add test for kernel crash during event notification

When a removal of the last mark on inode / vfsmount races with a notification about new event on that object, the kernel can crash. Test for this failure. The problem was fixed by kernel commit d90a10e2444b "fsnotify: Fix fsnotify_mark_connector race".

---
 testcases/kernel/syscalls/inotify/inotify07.c | 104 ++++++++++++++++++++++++++
 1 file changed, 104 insertions(+)
 create mode 100644 testcases/kernel/syscalls/inotify/inotify07.c
testcases/kernel/syscalls/inotify/inotify07.c new file +104 -0
new file mode 100644
index 000000000000..d20a824f0e8d
@@ -0,0 +1,104 @@
+/*
+ * Copyright (c) 2015 SUSE Linux.  All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * Further, this software is distributed without any warranty that it is
+ * free of the rightful claim of any third person regarding infringement
+ * or the like.  Any license provided herein, whether implied or
+ * otherwise, applies only to this software file.  Patent licenses, if
+ * any, provided herein do not apply to combinations of this program with
+ * other software, or any other product whatsoever.
+ *
+ * Started by Jan Kara <[email protected]>
+ *
+ * DESCRIPTION
+ * Test for inotify mark connector destruction race.
+ *
+ * Kernels prior to 4.17 have a race when the last fsnotify mark on the inode
+ * is being deleted while another process reports event happening on that
+ * inode. When the race is hit, the kernel crashes or loops.
+ *
+ * The problem has been fixed by commit:
+ * d90a10e2444b "fsnotify: Fix fsnotify_mark_connector race"
+ */
+
+#include "config.h"
+
+#include <stdio.h>
+#include <unistd.h>
+#include <stdlib.h>
+#include <fcntl.h>
+#include <time.h>
+#include <signal.h>
+#include <sys/time.h>
+#include <sys/wait.h>
+#include <sys/syscall.h>
+
+#include "tst_test.h"
+#include "inotify.h"
+
+#if defined(HAVE_SYS_INOTIFY_H)
+#include <sys/inotify.h>
+
+/* Number of test loops to run the test for */
+#define TEARDOWNS 1000000
+
+static void verify_inotify(void)
+{
+	int inotify_fd, fd;
+	int wd;
+	pid_t pid;
+	int tests;
+	char *name = "stress_fname";
+
+	fd = SAFE_OPEN(name, O_CREAT | O_RDWR, 0600);
+	pid = SAFE_FORK();
+	if (pid == 0) {
+		char buf[64];
+
+		memset(buf, 'a', sizeof(buf));
+		while (1) {
+			SAFE_WRITE(0, fd, buf, sizeof(buf));
+			SAFE_LSEEK(fd, 0, SEEK_SET);
+		}
+		/* Never reached */
+	}
+	SAFE_CLOSE(fd);
+
+	inotify_fd = myinotify_init1(0);
+	if (inotify_fd < 0)
+		tst_brk(TBROK | TERRNO, "inotify_init failed");
+	for (tests = 0; tests < TEARDOWNS; tests++) {
+		wd = myinotify_add_watch(inotify_fd, name, IN_MODIFY);
+		if (wd < 0)
+			tst_brk(TBROK | TERRNO, "inotify_add_watch() failed.");
+		wd = myinotify_rm_watch(inotify_fd, wd);
+		if (wd < 0)
+			tst_brk(TBROK | TERRNO, "inotify_rm_watch() failed.");
+	}
+	SAFE_CLOSE(inotify_fd);
+	/* We survived for given time - test succeeded */
+	tst_res(TPASS, "kernel survived inotify beating");
+
+	/* Kill the child writing to file and wait for it */
+	SAFE_KILL(pid, SIGKILL);
+	SAFE_WAIT(NULL);
+}
+
+static struct tst_test test = {
+	.timeout = 600,
+	.needs_tmpdir = 1,
+	.forks_child = 1,
+	.test_all = verify_inotify,
+};
+
+#else
+	TST_TEST_TCONF("system doesn't have required inotify support");
+#endif
Signature · 10 lines
2.13.6


------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
#1 lore raw

Hi!
This is the old mailing list that has been abandoned a few years ago when sourceforge had frequent outages, the new one is at [email protected].

Signature · 10 lines
Cyril Hrubis
[email protected]

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

A message that is not in the archive (sent off-list or not yet received): CAOQ4uxgQK9HtojVteoQNJfaP5qVX67HxmHoQ-XKLtjaTq4==Lg@mail.gmail.com

#3 lore raw

Hi!

Quoted 7 lines

inotify07 and inotify08 have already been merged to master and FYI already "posted" inotify09:
https://github.com/linux-test-project/ltp/pull/260

BTW, Cyril, do you have a preference between github pull request and posting the patch to the list?
I suppose I should have posted to fsdevel and CC Jan anyway...

I do prefer mailing list over pull requests but I work with both.

Signature · 10 lines
Cyril Hrubis
[email protected]

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

A message that is not in the archive (sent off-list or not yet received): [email protected]

#5 lore raw

Hi!

Quoted 8 lines
Quoted 5 lines
Quoted 3 lines

BTW, Cyril, do you have a preference between github pull request and posting the patch to the list?
I suppose I should have posted to fsdevel and CC Jan anyway...

I do prefer mailing list over pull requests but I work with both.

Cyril, should I rebase on top of that pull request or will you handle the conflict on your own?

I usually handle simple enough collisions myself, no need to rebase and resend just because of that.

Also as we are in the middle of pre-release freeze please be patient as the review will have to wait at least a week or two.

Signature · 10 lines
Cyril Hrubis
[email protected]

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list