Message-Id: <20180522193421.7017-1-christian@brauner.io>
Date:   Tue, 22 May 2018 21:34:21 +0200
From:   Christian Brauner <christianvanbrauner@...il.com>
To:     shuah@...nel.org, keescook@...omium.org, tglx@...utronix.de,
        kstewart@...uxfoundation.org, gregkh@...uxfoundation.org,
        mic@...ikod.net, linux-kernel@...r.kernel.org,
        linux-kselftest@...r.kernel.org
Cc:     ebiederm@...ssion.com, netdev@...r.kernel.org, davem@...emloft.net,
        Christian Brauner <christian@...uner.io>
Subject: [PATCH] selftests: uevent filtering

Recent discussions around uevent filtering (cf. net-next commit [1], [2],
and [3] and discussions in [4], [5], and [6]) have shown that the semantics
around uevent filtering where not well understood.
Now that we have settled - at least for the moment - how uevent filtering
should look like let's add some selftests to ensure we don't regress
anything in the future.
Note, the semantics of uevent filtering are described in detail in my
commit message to [2] so I won't repeat them here.

[1]: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=90d52d4fd82007005125d9a8d2d560a1ca059b9d
[2]: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=a3498436b3a0f8ec289e6847e1de40b4123e1639
[3]: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=26045a7b14bc7a5455e411d820110f66557d6589
[4]: https://lkml.org/lkml/2018/4/4/739
[5]: https://lkml.org/lkml/2018/4/26/767
[6]: https://lkml.org/lkml/2018/4/26/738

Signed-off-by: Christian Brauner <christian@...uner.io>
---
 tools/testing/selftests/uevent/Makefile       |  17 +
 tools/testing/selftests/uevent/config         |   2 +
 .../selftests/uevent/uevent_filtering.c       | 486 ++++++++++++++++++
 3 files changed, 505 insertions(+)
 create mode 100644 tools/testing/selftests/uevent/Makefile
 create mode 100644 tools/testing/selftests/uevent/config
 create mode 100644 tools/testing/selftests/uevent/uevent_filtering.c

diff --git a/tools/testing/selftests/uevent/Makefile b/tools/testing/selftests/uevent/Makefile
new file mode 100644
index 000000000000..f7baa9aa2932
--- /dev/null
+++ b/tools/testing/selftests/uevent/Makefile
@@ -0,0 +1,17 @@
+# SPDX-License-Identifier: GPL-2.0
+all:
+
+include ../lib.mk
+
+.PHONY: all clean
+
+BINARIES := uevent_filtering
+CFLAGS += -Wl,-no-as-needed -Wall
+
+uevent_filtering: uevent_filtering.c ../kselftest.h ../kselftest_harness.h
+	$(CC) $(CFLAGS) $< -o $@
+
+TEST_PROGS += $(BINARIES)
+EXTRA_CLEAN := $(BINARIES)
+
+all: $(BINARIES)
diff --git a/tools/testing/selftests/uevent/config b/tools/testing/selftests/uevent/config
new file mode 100644
index 000000000000..1038f4515be8
--- /dev/null
+++ b/tools/testing/selftests/uevent/config
@@ -0,0 +1,2 @@
+CONFIG_USER_NS=y
+CONFIG_NET=y
diff --git a/tools/testing/selftests/uevent/uevent_filtering.c b/tools/testing/selftests/uevent/uevent_filtering.c
new file mode 100644
index 000000000000..f83391aa42cf
--- /dev/null
+++ b/tools/testing/selftests/uevent/uevent_filtering.c
@@ -0,0 +1,486 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#define _GNU_SOURCE
+#include <errno.h>
+#include <fcntl.h>
+#include <linux/netlink.h>
+#include <signal.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/prctl.h>
+#include <sys/socket.h>
+#include <sched.h>
+#include <sys/eventfd.h>
+#include <sys/stat.h>
+#include <sys/syscall.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "../kselftest.h"
+#include "../kselftest_harness.h"
+
+#define __DEV_FULL "/sys/devices/virtual/mem/full/uevent"
+#define __UEVENT_BUFFER_SIZE (2048 * 2)
+#define __UEVENT_HEADER "add@...vices/virtual/mem/full"
+#define __UEVENT_HEADER_LEN sizeof("add@...vices/virtual/mem/full")
+#define __UEVENT_LISTEN_ALL -1
+
+ssize_t read_nointr(int fd, void *buf, size_t count)
+{
+	ssize_t ret;
+
+again:
+	ret = read(fd, buf, count);
+	if (ret < 0 && errno == EINTR)
+		goto again;
+
+	return ret;
+}
+
+ssize_t write_nointr(int fd, const void *buf, size_t count)
+{
+	ssize_t ret;
+
+again:
+	ret = write(fd, buf, count);
+	if (ret < 0 && errno == EINTR)
+		goto again;
+
+	return ret;
+}
+
+int wait_for_pid(pid_t pid)
+{
+	int status, ret;
+
+again:
+	ret = waitpid(pid, &status, 0);
+	if (ret == -1) {
+		if (errno == EINTR)
+			goto again;
+
+		return -1;
+	}
+
+	if (ret != pid)
+		goto again;
+
+	if (!WIFEXITED(status) || WEXITSTATUS(status) != 0)
+		return -1;
+
+	return 0;
+}
+
+static int uevent_listener(unsigned long post_flags, bool expect_uevent,
+			   int sync_fd)
+{
+	int sk_fd, ret;
+	socklen_t sk_addr_len;
+	int fret = -1, rcv_buf_sz = __UEVENT_BUFFER_SIZE;
+	uint64_t sync_add = 1;
+	struct sockaddr_nl sk_addr = { 0 }, rcv_addr = { 0 };
+	char buf[__UEVENT_BUFFER_SIZE] = { 0 };
+	struct iovec iov = { buf, __UEVENT_BUFFER_SIZE };
+	char control[CMSG_SPACE(sizeof(struct ucred))];
+	struct msghdr hdr = {
+		&rcv_addr, sizeof(rcv_addr), &iov, 1,
+		control,   sizeof(control),  0,
+	};
+
+	sk_fd = socket(AF_NETLINK, SOCK_RAW | SOCK_CLOEXEC,
+		       NETLINK_KOBJECT_UEVENT);
+	if (sk_fd < 0) {
+		fprintf(stderr, "%s - Failed to open uevent socket\n", strerror(errno));
+		return -1;
+	}
+
+	ret = setsockopt(sk_fd, SOL_SOCKET, SO_RCVBUF, &rcv_buf_sz,
+			 sizeof(rcv_buf_sz));
+	if (ret < 0) {
+		fprintf(stderr, "%s - Failed to set socket options\n", strerror(errno));
+		goto on_error;
+	}
+
+	sk_addr.nl_family = AF_NETLINK;
+	sk_addr.nl_groups = __UEVENT_LISTEN_ALL;
+
+	sk_addr_len = sizeof(sk_addr);
+	ret = bind(sk_fd, (struct sockaddr *)&sk_addr, sk_addr_len);
+	if (ret < 0) {
+		fprintf(stderr, "%s - Failed to bind socket\n", strerror(errno));
+		goto on_error;
+	}
+
+	ret = getsockname(sk_fd, (struct sockaddr *)&sk_addr, &sk_addr_len);
+	if (ret < 0) {
+		fprintf(stderr, "%s - Failed to retrieve socket name\n", strerror(errno));
+		goto on_error;
+	}
+
+	if ((size_t)sk_addr_len != sizeof(sk_addr)) {
+		fprintf(stderr, "Invalid socket address size\n");
+		goto on_error;
+	}
+
+	if (post_flags & CLONE_NEWUSER) {
+		ret = unshare(CLONE_NEWUSER);
+		if (ret < 0) {
+			fprintf(stderr,
+				"%s - Failed to unshare user namespace\n",
+				strerror(errno));
+			goto on_error;
+		}
+	}
+
+	if (post_flags & CLONE_NEWNET) {
+		ret = unshare(CLONE_NEWNET);
+		if (ret < 0) {
+			fprintf(stderr,
+				"%s - Failed to unshare network namespace\n",
+				strerror(errno));
+			goto on_error;
+		}
+	}
+
+	ret = write_nointr(sync_fd, &sync_add, sizeof(sync_add));
+	close(sync_fd);
+	if (ret != sizeof(sync_add)) {
+		fprintf(stderr, "Failed to synchronize with parent process\n");
+		goto on_error;
+	}
+
+	fret = 0;
+	for (;;) {
+		ssize_t r;
+
+		r = recvmsg(sk_fd, &hdr, 0);
+		if (r <= 0) {
+			fprintf(stderr, "%s - Failed to receive uevent\n", strerror(errno));
+			ret = -1;
+			break;
+		}
+
+		/* ignore libudev messages */
+		if (memcmp(buf, "libudev", 8) == 0)
+			continue;
+
+		/* ignore uevents we didn't trigger */
+		if (memcmp(buf, __UEVENT_HEADER, __UEVENT_HEADER_LEN) != 0)
+			continue;
+
+		if (!expect_uevent) {
+			fprintf(stderr, "Received unexpected uevent:\n");
+			ret = -1;
+		}
+
+		if (TH_LOG_ENABLED) {
+			/* If logging is enabled dump the received uevent. */
+			(void)write_nointr(STDERR_FILENO, buf, r);
+			(void)write_nointr(STDERR_FILENO, "\n", 1);
+		}
+
+		break;
+	}
+
+on_error:
+	close(sk_fd);
+
+	return fret;
+}
+
+int trigger_uevent(unsigned int times)
+{
+	int fd, ret;
+	unsigned int i;
+
+	fd = open(__DEV_FULL, O_RDWR | O_CLOEXEC);
+	if (fd < 0) {
+		if (errno != ENOENT)
+			return -EINVAL;
+
+		return -1;
+	}
+
+	for (i = 0; i < times; i++) {
+		ret = write_nointr(fd, "add\n", sizeof("add\n") - 1);
+		if (ret < 0) {
+			fprintf(stderr, "Failed to trigger uevent\n");
+			break;
+		}
+	}
+	close(fd);
+
+	return ret;
+}
+
+int set_death_signal(void)
+{
+	int ret;
+	pid_t ppid;
+
+	ret = prctl(PR_SET_PDEATHSIG, SIGKILL, 0, 0, 0);
+
+	/* Check whether we have been orphaned. */
+	ppid = getppid();
+	if (ppid == 1) {
+		pid_t self;
+
+		self = getpid();
+		ret = kill(self, SIGKILL);
+	}
+
+	if (ret < 0)
+		return -1;
+
+	return 0;
+}
+
+static int do_test(unsigned long pre_flags, unsigned long post_flags,
+		   bool expect_uevent, int sync_fd)
+{
+	int ret;
+	uint64_t wait_val;
+	pid_t pid;
+	sigset_t mask;
+	sigset_t orig_mask;
+	struct timespec timeout;
+
+	sigemptyset(&mask);
+	sigaddset(&mask, SIGCHLD);
+
+	ret = sigprocmask(SIG_BLOCK, &mask, &orig_mask);
+	if (ret < 0) {
+		fprintf(stderr, "%s- Failed to block SIGCHLD\n", strerror(errno));
+		return -1;
+	}
+
+	pid = fork();
+	if (pid < 0) {
+		fprintf(stderr, "%s - Failed to fork() new process\n", strerror(errno));
+		return -1;
+	}
+
+	if (pid == 0) {
+		/* Make sure that we go away when our parent dies. */
+		ret = set_death_signal();
+		if (ret < 0) {
+			fprintf(stderr, "Failed to set PR_SET_PDEATHSIG to SIGKILL\n");
+			_exit(EXIT_FAILURE);
+		}
+
+		if (pre_flags & CLONE_NEWUSER) {
+			ret = unshare(CLONE_NEWUSER);
+			if (ret < 0) {
+				fprintf(stderr,
+					"%s - Failed to unshare user namespace\n",
+					strerror(errno));
+				_exit(EXIT_FAILURE);
+			}
+		}
+
+		if (pre_flags & CLONE_NEWNET) {
+			ret = unshare(CLONE_NEWNET);
+			if (ret < 0) {
+				fprintf(stderr,
+					"%s - Failed to unshare network namespace\n",
+					strerror(errno));
+				_exit(EXIT_FAILURE);
+			}
+		}
+
+		if (uevent_listener(post_flags, expect_uevent, sync_fd) < 0)
+			_exit(EXIT_FAILURE);
+
+		_exit(EXIT_SUCCESS);
+	}
+
+	ret = read_nointr(sync_fd, &wait_val, sizeof(wait_val));
+	if (ret != sizeof(wait_val)) {
+		fprintf(stderr, "Failed to synchronize with child process\n");
+		_exit(EXIT_FAILURE);
+	}
+
+	/* Trigger 10 uevents to account for the case where the kernel might
+	 * drop some.
+	 */
+	ret = trigger_uevent(10);
+	if (ret < 0)
+		fprintf(stderr, "Failed triggering uevents\n");
+
+	/* Wait for 2 seconds before considering this failed. This should be
+	 * plenty of time for the kernel to deliver the uevent even under heavy
+	 * load.
+	 */
+	timeout.tv_sec = 2;
+	timeout.tv_nsec = 0;
+
+again:
+	ret = sigtimedwait(&mask, NULL, &timeout);
+	if (ret < 0) {
+		if (errno == EINTR)
+			goto again;
+
+		if (!expect_uevent)
+			ret = kill(pid, SIGTERM); /* success */
+		else
+			ret = kill(pid, SIGUSR1); /* error */
+		if (ret < 0)
+			return -1;
+	}
+
+	ret = wait_for_pid(pid);
+	if (ret < 0)
+		return -1;
+
+	return ret;
+}
+
+static void signal_handler(int sig)
+{
+	if (sig == SIGTERM)
+		_exit(EXIT_SUCCESS);
+
+	_exit(EXIT_FAILURE);
+}
+
+TEST(uevent_filtering)
+{
+	int ret, sync_fd;
+	struct sigaction act;
+
+	if (geteuid()) {
+		TH_LOG("Uevent filtering tests require root privileges. Skipping test");
+		_exit(KSFT_SKIP);
+	}
+
+	ret = access(__DEV_FULL, F_OK);
+	EXPECT_EQ(0, ret) {
+		if (errno == ENOENT) {
+			TH_LOG(__DEV_FULL " does not exist. Skipping test");
+			_exit(KSFT_SKIP);
+		}
+
+		_exit(KSFT_FAIL);
+	}
+
+	act.sa_handler = signal_handler;
+	act.sa_flags = 0;
+	sigemptyset(&act.sa_mask);
+
+	ret = sigaction(SIGTERM, &act, NULL);
+	ASSERT_EQ(0, ret);
+
+	sync_fd = eventfd(0, EFD_CLOEXEC);
+	ASSERT_GE(sync_fd, 0);
+
+	/*
+	 * Setup:
+	 * - Open uevent listening socket in initial network namespace owned by
+	 *   initial user namespace.
+	 * - Trigger uevent in initial network namespace owned by initial user
+	 *   namespace.
+	 * Expected Result:
+	 * - uevent listening socket receives uevent
+	 */
+	ret = do_test(0, 0, true, sync_fd);
+	ASSERT_EQ(0, ret) {
+		goto do_cleanup;
+	}
+
+	/*
+	 * Setup:
+	 * - Open uevent listening socket in non-initial network namespace
+	 *   owned by initial user namespace.
+	 * - Trigger uevent in initial network namespace owned by initial user
+	 *   namespace.
+	 * Expected Result:
+	 * - uevent listening socket receives uevent
+	 */
+	ret = do_test(CLONE_NEWNET, 0, true, sync_fd);
+	ASSERT_EQ(0, ret) {
+		goto do_cleanup;
+	}
+
+	/*
+	 * Setup:
+	 * - unshare user namespace
+	 * - Open uevent listening socket in initial network namespace
+	 *   owned by initial user namespace.
+	 * - Trigger uevent in initial network namespace owned by initial user
+	 *   namespace.
+	 * Expected Result:
+	 * - uevent listening socket receives uevent
+	 */
+	ret = do_test(CLONE_NEWUSER, 0, true, sync_fd);
+	ASSERT_EQ(0, ret) {
+		goto do_cleanup;
+	}
+
+	/*
+	 * Setup:
+	 * - Open uevent listening socket in non-initial network namespace
+	 *   owned by non-initial user namespace.
+	 * - Trigger uevent in initial network namespace owned by initial user
+	 *   namespace.
+	 * Expected Result:
+	 * - uevent listening socket receives no uevent
+	 */
+	ret = do_test(CLONE_NEWUSER | CLONE_NEWNET, 0, false, sync_fd);
+	ASSERT_EQ(0, ret) {
+		goto do_cleanup;
+	}
+
+	/*
+	 * Setup:
+	 * - Open uevent listening socket in initial network namespace
+	 *   owned by initial user namespace.
+	 * - unshare network namespace
+	 * - Trigger uevent in initial network namespace owned by initial user
+	 *   namespace.
+	 * Expected Result:
+	 * - uevent listening socket receives uevent
+	 */
+	ret = do_test(0, CLONE_NEWNET, true, sync_fd);
+	ASSERT_EQ(0, ret) {
+		goto do_cleanup;
+	}
+
+	/*
+	 * Setup:
+	 * - Open uevent listening socket in initial network namespace
+	 *   owned by initial user namespace.
+	 * - unshare user namespace
+	 * - Trigger uevent in initial network namespace owned by initial user
+	 *   namespace.
+	 * Expected Result:
+	 * - uevent listening socket receives uevent
+	 */
+	ret = do_test(0, CLONE_NEWUSER, true, sync_fd);
+	ASSERT_EQ(0, ret) {
+		goto do_cleanup;
+	}
+
+	/*
+	 * Setup:
+	 * - Open uevent listening socket in initial network namespace
+	 *   owned by initial user namespace.
+	 * - unshare user namespace
+	 * - unshare network namespace
+	 * - Trigger uevent in initial network namespace owned by initial user
+	 *   namespace.
+	 * Expected Result:
+	 * - uevent listening socket receives uevent
+	 */
+	ret = do_test(0, CLONE_NEWUSER | CLONE_NEWNET, true, sync_fd);
+	ASSERT_EQ(0, ret) {
+		goto do_cleanup;
+	}
+
+do_cleanup:
+	close(sync_fd);
+}
+
+TEST_HARNESS_MAIN
-- 
2.17.0

Powered by blists - more mailing lists