seccomp: allow TSYNC and USER_NOTIF together

The restriction introduced in 7a0df7fbc1 ("seccomp: Make NEW_LISTENER and
TSYNC flags exclusive") is mostly artificial: there is enough information
in a seccomp user notification to tell which thread triggered a
notification. The reason it was introduced is because TSYNC makes the
syscall return a thread-id on failure, and NEW_LISTENER returns an fd, and
there's no way to distinguish between these two cases (well, I suppose the
caller could check all fds it has, then do the syscall, and if the return
value was an fd that already existed, then it must be a thread id, but
bleh).

Matthew would like to use these two flags together in the Chrome sandbox
which wants to use TSYNC for video drivers and NEW_LISTENER to proxy
syscalls.

So, let's fix this ugliness by adding another flag, TSYNC_ESRCH, which
tells the kernel to just return -ESRCH on a TSYNC error. This way,
NEW_LISTENER (and any subsequent seccomp() commands that want to return
positive values) don't conflict with each other.

Suggested-by: Matthew Denton <mpdenton@google.com>
Signed-off-by: Tycho Andersen <tycho@tycho.ws>
Link: https://lore.kernel.org/r/20200304180517.23867-1-tycho@tycho.ws
Signed-off-by: Kees Cook <keescook@chromium.org>

include/linux/seccomp.h

@@ -7,7 +7,8 @@
 #define SECCOMP_FILTER_FLAG_MASK	(SECCOMP_FILTER_FLAG_TSYNC | \
 					 SECCOMP_FILTER_FLAG_LOG | \
 					 SECCOMP_FILTER_FLAG_SPEC_ALLOW | \
-					 SECCOMP_FILTER_FLAG_NEW_LISTENER)
+					 SECCOMP_FILTER_FLAG_NEW_LISTENER | \
+					 SECCOMP_FILTER_FLAG_TSYNC_ESRCH)
 
 #ifdef CONFIG_SECCOMP
 

include/uapi/linux/seccomp.h

@@ -22,6 +22,7 @@
 #define SECCOMP_FILTER_FLAG_LOG			(1UL << 1)
 #define SECCOMP_FILTER_FLAG_SPEC_ALLOW		(1UL << 2)
 #define SECCOMP_FILTER_FLAG_NEW_LISTENER	(1UL << 3)
+#define SECCOMP_FILTER_FLAG_TSYNC_ESRCH		(1UL << 4)
 
 /*
  * All BPF programs must return a 32-bit value.

kernel/seccomp.c

@@ -528,8 +528,12 @@ static long seccomp_attach_filter(unsigned int flags,
 		int ret;
 
 		ret = seccomp_can_sync_threads();
-		if (ret)
+		if (ret) {
-			return ret;
+			if (flags & SECCOMP_FILTER_FLAG_TSYNC_ESRCH)
+				return -ESRCH;
+			else
+				return ret;
+		}
 	}
 
 	/* Set log flag, if present. */
@@ -1288,10 +1292,12 @@ static long seccomp_set_mode_filter(unsigned int flags,
 	 * In the successful case, NEW_LISTENER returns the new listener fd.
 	 * But in the failure case, TSYNC returns the thread that died. If you
 	 * combine these two flags, there's no way to tell whether something
-	 * succeeded or failed. So, let's disallow this combination.
+	 * succeeded or failed. So, let's disallow this combination if the user
+	 * has not explicitly requested no errors from TSYNC.
 	 */
 	if ((flags & SECCOMP_FILTER_FLAG_TSYNC) &&
-	    (flags & SECCOMP_FILTER_FLAG_NEW_LISTENER))
+	    (flags & SECCOMP_FILTER_FLAG_NEW_LISTENER) &&
+	    ((flags & SECCOMP_FILTER_FLAG_TSYNC_ESRCH) == 0))
 		return -EINVAL;
 
 	/* Prepare the new filter before holding any locks. */

tools/testing/selftests/seccomp/seccomp_bpf.c

@@ -212,6 +212,10 @@ struct seccomp_notif_sizes {
 #define SECCOMP_USER_NOTIF_FLAG_CONTINUE 0x00000001
 #endif
 
+#ifndef SECCOMP_FILTER_FLAG_TSYNC_ESRCH
+#define SECCOMP_FILTER_FLAG_TSYNC_ESRCH (1UL << 4)
+#endif
+
 #ifndef seccomp
 int seccomp(unsigned int op, unsigned int flags, void *args)
 {
@@ -2187,7 +2191,8 @@ TEST(detect_seccomp_filter_flags)
 	unsigned int flags[] = { SECCOMP_FILTER_FLAG_TSYNC,
 				 SECCOMP_FILTER_FLAG_LOG,
 				 SECCOMP_FILTER_FLAG_SPEC_ALLOW,
-				 SECCOMP_FILTER_FLAG_NEW_LISTENER };
+				 SECCOMP_FILTER_FLAG_NEW_LISTENER,
+				 SECCOMP_FILTER_FLAG_TSYNC_ESRCH };
 	unsigned int exclusive[] = {
 				SECCOMP_FILTER_FLAG_TSYNC,
 				SECCOMP_FILTER_FLAG_NEW_LISTENER };
@@ -2645,6 +2650,55 @@ TEST_F(TSYNC, two_siblings_with_one_divergence)
 	EXPECT_EQ(SIBLING_EXIT_UNKILLED, (long)status);
 }
 
+TEST_F(TSYNC, two_siblings_with_one_divergence_no_tid_in_err)
+{
+	long ret, flags;
+	void *status;
+
+	ASSERT_EQ(0, prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0)) {
+		TH_LOG("Kernel does not support PR_SET_NO_NEW_PRIVS!");
+	}
+
+	ret = seccomp(SECCOMP_SET_MODE_FILTER, 0, &self->root_prog);
+	ASSERT_NE(ENOSYS, errno) {
+		TH_LOG("Kernel does not support seccomp syscall!");
+	}
+	ASSERT_EQ(0, ret) {
+		TH_LOG("Kernel does not support SECCOMP_SET_MODE_FILTER!");
+	}
+	self->sibling[0].diverge = 1;
+	tsync_start_sibling(&self->sibling[0]);
+	tsync_start_sibling(&self->sibling[1]);
+
+	while (self->sibling_count < TSYNC_SIBLINGS) {
+		sem_wait(&self->started);
+		self->sibling_count++;
+	}
+
+	flags = SECCOMP_FILTER_FLAG_TSYNC | \
+		SECCOMP_FILTER_FLAG_TSYNC_ESRCH;
+	ret = seccomp(SECCOMP_SET_MODE_FILTER, flags, &self->apply_prog);
+	ASSERT_EQ(ESRCH, errno) {
+		TH_LOG("Did not return ESRCH for diverged sibling.");
+	}
+	ASSERT_EQ(-1, ret) {
+		TH_LOG("Did not fail on diverged sibling.");
+	}
+
+	/* Wake the threads */
+	pthread_mutex_lock(&self->mutex);
+	ASSERT_EQ(0, pthread_cond_broadcast(&self->cond)) {
+		TH_LOG("cond broadcast non-zero");
+	}
+	pthread_mutex_unlock(&self->mutex);
+
+	/* Ensure they are both unkilled. */
+	PTHREAD_JOIN(self->sibling[0].tid, &status);
+	EXPECT_EQ(SIBLING_EXIT_UNKILLED, (long)status);
+	PTHREAD_JOIN(self->sibling[1].tid, &status);
+	EXPECT_EQ(SIBLING_EXIT_UNKILLED, (long)status);
+}
+
 TEST_F(TSYNC, two_siblings_not_under_filter)
 {
 	long ret, sib;
@@ -3196,6 +3250,24 @@ TEST(user_notification_basic)
 	EXPECT_EQ(0, WEXITSTATUS(status));
 }
 
+TEST(user_notification_with_tsync)
+{
+	int ret;
+	unsigned int flags;
+
+	/* these were exclusive */
+	flags = SECCOMP_FILTER_FLAG_NEW_LISTENER |
+		SECCOMP_FILTER_FLAG_TSYNC;
+	ASSERT_EQ(-1, user_trap_syscall(__NR_getppid, flags));
+	ASSERT_EQ(EINVAL, errno);
+
+	/* but now they're not */
+	flags |= SECCOMP_FILTER_FLAG_TSYNC_ESRCH;
+	ret = user_trap_syscall(__NR_getppid, flags);
+	close(ret);
+	ASSERT_LE(0, ret);
+}
+
 TEST(user_notification_kill_in_middle)
 {
 	pid_t pid;