Merge tag 'seccomp-v5.9-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux

Pull seccomp updates from Kees Cook:
 "There are a bunch of clean ups and selftest improvements along with
  two major updates to the SECCOMP_RET_USER_NOTIF filter return:
  EPOLLHUP support to more easily detect the death of a monitored
  process, and being able to inject fds when intercepting syscalls that
  expect an fd-opening side-effect (needed by both container folks and
  Chrome). The latter continued the refactoring of __scm_install_fd()
  started by Christoph, and in the process found and fixed a handful of
  bugs in various callers.

   - Improved selftest coverage, timeouts, and reporting

   - Add EPOLLHUP support for SECCOMP_RET_USER_NOTIF (Christian Brauner)

   - Refactor __scm_install_fd() into __receive_fd() and fix buggy
     callers

   - Introduce 'addfd' command for SECCOMP_RET_USER_NOTIF (Sargun
     Dhillon)"

* tag 'seccomp-v5.9-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux: (30 commits)
  selftests/seccomp: Test SECCOMP_IOCTL_NOTIF_ADDFD
  seccomp: Introduce addfd ioctl to seccomp user notifier
  fs: Expand __receive_fd() to accept existing fd
  pidfd: Replace open-coded receive_fd()
  fs: Add receive_fd() wrapper for __receive_fd()
  fs: Move __scm_install_fd() to __receive_fd()
  net/scm: Regularize compat handling of scm_detach_fds()
  pidfd: Add missing sock updates for pidfd_getfd()
  net/compat: Add missing sock updates for SCM_RIGHTS
  selftests/seccomp: Check ENOSYS under tracing
  selftests/seccomp: Refactor to use fixture variants
  selftests/harness: Clean up kern-doc for fixtures
  seccomp: Use -1 marker for end of mode 1 syscall list
  seccomp: Fix ioctl number for SECCOMP_IOCTL_NOTIF_ID_VALID
  selftests/seccomp: Rename user_trap_syscall() to user_notif_syscall()
  selftests/seccomp: Make kcmp() less required
  seccomp: Use pr_fmt
  selftests/seccomp: Improve calibration loop
  selftests/seccomp: use 90s as timeout
  selftests/seccomp: Expand benchmark to per-filter measurements
  ...

arch/mips/include/asm/seccomp.h

@@ -9,12 +9,12 @@ static inline const int *get_compat_mode1_syscalls(void)
 	static const int syscalls_O32[] = {
 		__NR_O32_Linux + 3, __NR_O32_Linux + 4,
 		__NR_O32_Linux + 1, __NR_O32_Linux + 193,
-		0, /* null terminated */
+		-1, /* negative terminated */
 	};
 	static const int syscalls_N32[] = {
 		__NR_N32_Linux + 0, __NR_N32_Linux + 1,
 		__NR_N32_Linux + 58, __NR_N32_Linux + 211,
-		0, /* null terminated */
+		-1, /* negative terminated */
 	};
 
 	if (IS_ENABLED(CONFIG_MIPS32_O32) && test_thread_flag(TIF_32BIT_REGS))

fs/file.c

@@ -18,6 +18,7 @@
 #include <linux/bitops.h>
 #include <linux/spinlock.h>
 #include <linux/rcupdate.h>
+#include <net/sock.h>
 
 unsigned int sysctl_nr_open __read_mostly = 1024*1024;
 unsigned int sysctl_nr_open_min = BITS_PER_LONG;
@@ -613,6 +614,10 @@ void __fd_install(struct files_struct *files, unsigned int fd,
 	rcu_read_unlock_sched();
 }
 
+/*
+ * This consumes the "file" refcount, so callers should treat it
+ * as if they had called fput(file).
+ */
 void fd_install(unsigned int fd, struct file *file)
 {
 	__fd_install(current->files, fd, file);
@@ -931,6 +936,62 @@ int replace_fd(unsigned fd, struct file *file, unsigned flags)
 	return err;
 }
 
+/**
+ * __receive_fd() - Install received file into file descriptor table
+ *
+ * @fd: fd to install into (if negative, a new fd will be allocated)
+ * @file: struct file that was received from another process
+ * @ufd: __user pointer to write new fd number to
+ * @o_flags: the O_* flags to apply to the new fd entry
+ *
+ * Installs a received file into the file descriptor table, with appropriate
+ * checks and count updates. Optionally writes the fd number to userspace, if
+ * @ufd is non-NULL.
+ *
+ * This helper handles its own reference counting of the incoming
+ * struct file.
+ *
+ * Returns newly install fd or -ve on error.
+ */
+int __receive_fd(int fd, struct file *file, int __user *ufd, unsigned int o_flags)
+{
+	int new_fd;
+	int error;
+
+	error = security_file_receive(file);
+	if (error)
+		return error;
+
+	if (fd < 0) {
+		new_fd = get_unused_fd_flags(o_flags);
+		if (new_fd < 0)
+			return new_fd;
+	} else {
+		new_fd = fd;
+	}
+
+	if (ufd) {
+		error = put_user(new_fd, ufd);
+		if (error) {
+			if (fd < 0)
+				put_unused_fd(new_fd);
+			return error;
+		}
+	}
+
+	if (fd < 0) {
+		fd_install(new_fd, get_file(file));
+	} else {
+		error = replace_fd(new_fd, file, o_flags);
+		if (error)
+			return error;
+	}
+
+	/* Bump the sock usage counts, if any. */
+	__receive_sock(file);
+	return new_fd;
+}
+
 static int ksys_dup3(unsigned int oldfd, unsigned int newfd, int flags)
 {
 	int err = -EBADF;

fs/proc/array.c

@@ -341,6 +341,8 @@ static inline void task_seccomp(struct seq_file *m, struct task_struct *p)
 	seq_put_decimal_ull(m, "NoNewPrivs:\t", task_no_new_privs(p));
 #ifdef CONFIG_SECCOMP
 	seq_put_decimal_ull(m, "\nSeccomp:\t", p->seccomp.mode);
+	seq_put_decimal_ull(m, "\nSeccomp_filters:\t",
+			    atomic_read(&p->seccomp.filter_count));
 #endif
 	seq_puts(m, "\nSpeculation_Store_Bypass:\t");
 	switch (arch_prctl_spec_ctrl_get(p, PR_SPEC_STORE_BYPASS)) {

include/asm-generic/seccomp.h

@@ -33,7 +33,7 @@ static inline const int *get_compat_mode1_syscalls(void)
 	static const int mode1_syscalls_32[] = {
 		__NR_seccomp_read_32, __NR_seccomp_write_32,
 		__NR_seccomp_exit_32, __NR_seccomp_sigreturn_32,
-		0, /* null terminated */
+		-1, /* negative terminated */
 	};
 	return mode1_syscalls_32;
 }

include/linux/file.h

@@ -9,6 +9,7 @@
 #include <linux/compiler.h>
 #include <linux/types.h>
 #include <linux/posix_types.h>
+#include <linux/errno.h>
 
 struct file;
 
@@ -91,6 +92,24 @@ extern void put_unused_fd(unsigned int fd);
 
 extern void fd_install(unsigned int fd, struct file *file);
 
+extern int __receive_fd(int fd, struct file *file, int __user *ufd,
+			unsigned int o_flags);
+static inline int receive_fd_user(struct file *file, int __user *ufd,
+				  unsigned int o_flags)
+{
+	if (ufd == NULL)
+		return -EFAULT;
+	return __receive_fd(-1, file, ufd, o_flags);
+}
+static inline int receive_fd(struct file *file, unsigned int o_flags)
+{
+	return __receive_fd(-1, file, NULL, o_flags);
+}
+static inline int receive_fd_replace(int fd, struct file *file, unsigned int o_flags)
+{
+	return __receive_fd(fd, file, NULL, o_flags);
+}
+
 extern void flush_delayed_fput(void);
 extern void __fput_sync(struct file *);
 

include/linux/seccomp.h

@@ -10,9 +10,14 @@
 					 SECCOMP_FILTER_FLAG_NEW_LISTENER | \
 					 SECCOMP_FILTER_FLAG_TSYNC_ESRCH)
 
+/* sizeof() the first published struct seccomp_notif_addfd */
+#define SECCOMP_NOTIFY_ADDFD_SIZE_VER0 24
+#define SECCOMP_NOTIFY_ADDFD_SIZE_LATEST SECCOMP_NOTIFY_ADDFD_SIZE_VER0
+
 #ifdef CONFIG_SECCOMP
 
 #include <linux/thread_info.h>
+#include <linux/atomic.h>
 #include <asm/seccomp.h>
 
 struct seccomp_filter;
@@ -29,6 +34,7 @@ struct seccomp_filter;
  */
 struct seccomp {
 	int mode;
+	atomic_t filter_count;
 	struct seccomp_filter *filter;
 };
 
@@ -82,10 +88,10 @@ static inline int seccomp_mode(struct seccomp *s)
 #endif /* CONFIG_SECCOMP */
 
 #ifdef CONFIG_SECCOMP_FILTER
-extern void put_seccomp_filter(struct task_struct *tsk);
+extern void seccomp_filter_release(struct task_struct *tsk);
 extern void get_seccomp_filter(struct task_struct *tsk);
 #else  /* CONFIG_SECCOMP_FILTER */
-static inline void put_seccomp_filter(struct task_struct *tsk)
+static inline void seccomp_filter_release(struct task_struct *tsk)
 {
 	return;
 }

include/net/sock.h

@@ -891,6 +891,8 @@ static inline int sk_memalloc_socks(void)
 {
 	return static_branch_unlikely(&memalloc_socks_key);
 }
+
+void __receive_sock(struct file *file);
 #else
 
 static inline int sk_memalloc_socks(void)
@@ -898,6 +900,8 @@ static inline int sk_memalloc_socks(void)
 	return 0;
 }
 
+static inline void __receive_sock(struct file *file)
+{ }
 #endif
 
 static inline gfp_t sk_gfp_mask(const struct sock *sk, gfp_t gfp_mask)

include/uapi/linux/seccomp.h

@@ -113,6 +113,25 @@ struct seccomp_notif_resp {
 	__u32 flags;
 };
 
+/* valid flags for seccomp_notif_addfd */
+#define SECCOMP_ADDFD_FLAG_SETFD	(1UL << 0) /* Specify remote fd */
+
+/**
+ * struct seccomp_notif_addfd
+ * @id: The ID of the seccomp notification
+ * @flags: SECCOMP_ADDFD_FLAG_*
+ * @srcfd: The local fd number
+ * @newfd: Optional remote FD number if SETFD option is set, otherwise 0.
+ * @newfd_flags: The O_* flags the remote FD should have applied
+ */
+struct seccomp_notif_addfd {
+	__u64 id;
+	__u32 flags;
+	__u32 srcfd;
+	__u32 newfd;
+	__u32 newfd_flags;
+};
+
 #define SECCOMP_IOC_MAGIC		'!'
 #define SECCOMP_IO(nr)			_IO(SECCOMP_IOC_MAGIC, nr)
 #define SECCOMP_IOR(nr, type)		_IOR(SECCOMP_IOC_MAGIC, nr, type)
@@ -123,5 +142,9 @@ struct seccomp_notif_resp {
 #define SECCOMP_IOCTL_NOTIF_RECV	SECCOMP_IOWR(0, struct seccomp_notif)
 #define SECCOMP_IOCTL_NOTIF_SEND	SECCOMP_IOWR(1,	\
 						struct seccomp_notif_resp)
-#define SECCOMP_IOCTL_NOTIF_ID_VALID	SECCOMP_IOR(2, __u64)
+#define SECCOMP_IOCTL_NOTIF_ID_VALID	SECCOMP_IOW(2, __u64)
+/* On success, the return value is the remote process's added fd number */
+#define SECCOMP_IOCTL_NOTIF_ADDFD	SECCOMP_IOW(3, \
+						struct seccomp_notif_addfd)
+
 #endif /* _UAPI_LINUX_SECCOMP_H */

init/init_task.c

@@ -204,6 +204,9 @@ struct task_struct init_task
 #ifdef CONFIG_SECURITY
 	.security	= NULL,
 #endif
+#ifdef CONFIG_SECCOMP
+	.seccomp	= { .filter_count = ATOMIC_INIT(0) },
+#endif
 };
 EXPORT_SYMBOL(init_task);
 

kernel/exit.c

@@ -217,6 +217,7 @@ void release_task(struct task_struct *p)
 	}
 
 	write_unlock_irq(&tasklist_lock);
+	seccomp_filter_release(p);
 	proc_flush_pid(thread_pid);
 	put_pid(thread_pid);
 	release_thread(p);

kernel/fork.c

@@ -479,7 +479,6 @@ void free_task(struct task_struct *tsk)
 #endif
 	rt_mutex_debug_task_free(tsk);
 	ftrace_graph_exit_task(tsk);
-	put_seccomp_filter(tsk);
 	arch_release_task_struct(tsk);
 	if (tsk->flags & PF_KTHREAD)
 		free_kthread_struct(tsk);

kernel/pid.c

@@ -42,6 +42,7 @@
 #include <linux/sched/signal.h>
 #include <linux/sched/task.h>
 #include <linux/idr.h>
+#include <net/sock.h>
 
 struct pid init_struct_pid = {
 	.count		= REFCOUNT_INIT(1),
@@ -635,17 +636,8 @@ static int pidfd_getfd(struct pid *pid, int fd)
 	if (IS_ERR(file))
 		return PTR_ERR(file);
 
-	ret = security_file_receive(file);
-	if (ret) {
+	ret = receive_fd(file, O_CLOEXEC);
 	fput(file);
-		return ret;
-	}
-
-	ret = get_unused_fd_flags(O_CLOEXEC);
-	if (ret < 0)
-		fput(file);
-	else
-		fd_install(ret, file);
 
 	return ret;
 }

kernel/seccomp.c

@@ -13,6 +13,7 @@
  * Mode 2 allows user-defined system call filters in the form
  *        of Berkeley Packet Filters/Linux Socket Filters.
  */
+#define pr_fmt(fmt) "seccomp: " fmt
 
 #include <linux/refcount.h>
 #include <linux/audit.h>
@@ -41,6 +42,15 @@
 #include <linux/tracehook.h>
 #include <linux/uaccess.h>
 #include <linux/anon_inodes.h>
+#include <linux/lockdep.h>
+
+/*
+ * When SECCOMP_IOCTL_NOTIF_ID_VALID was first introduced, it had the
+ * wrong direction flag in the ioctl number. This is the broken one,
+ * which the kernel needs to keep supporting until all userspaces stop
+ * using the wrong command number.
+ */
+#define SECCOMP_IOCTL_NOTIF_ID_VALID_WRONG_DIR	SECCOMP_IOR(2, __u64)
 
 enum notify_state {
 	SECCOMP_NOTIFY_INIT,
@@ -77,10 +87,42 @@ struct seccomp_knotif {
 	long val;
 	u32 flags;
 
-	/* Signals when this has entered SECCOMP_NOTIFY_REPLIED */
+	/*
+	 * Signals when this has changed states, such as the listener
+	 * dying, a new seccomp addfd message, or changing to REPLIED
+	 */
 	struct completion ready;
 
 	struct list_head list;
+
+	/* outstanding addfd requests */
+	struct list_head addfd;
+};
+
+/**
+ * struct seccomp_kaddfd - container for seccomp_addfd ioctl messages
+ *
+ * @file: A reference to the file to install in the other task
+ * @fd: The fd number to install it at. If the fd number is -1, it means the
+ *      installing process should allocate the fd as normal.
+ * @flags: The flags for the new file descriptor. At the moment, only O_CLOEXEC
+ *         is allowed.
+ * @ret: The return value of the installing process. It is set to the fd num
+ *       upon success (>= 0).
+ * @completion: Indicates that the installing process has completed fd
+ *              installation, or gone away (either due to successful
+ *              reply, or signal)
+ *
+ */
+struct seccomp_kaddfd {
+	struct file *file;
+	int fd;
+	unsigned int flags;
+
+	/* To only be set on reply */
+	int ret;
+	struct completion completion;
+	struct list_head list;
 };
 
 /**
@@ -94,27 +136,35 @@ struct seccomp_knotif {
  *           filter->notify_lock.
  * @next_id: The id of the next request.
  * @notifications: A list of struct seccomp_knotif elements.
- * @wqh: A wait queue for poll.
  */
 struct notification {
 	struct semaphore request;
 	u64 next_id;
 	struct list_head notifications;
-	wait_queue_head_t wqh;
 };
 
 /**
  * struct seccomp_filter - container for seccomp BPF programs
  *
- * @usage: reference count to manage the object lifetime.
- *         get/put helpers should be used when accessing an instance
- *         outside of a lifetime-guarded section.  In general, this
- *         is only needed for handling filters shared across tasks.
+ * @refs: Reference count to manage the object lifetime.
+ *	  A filter's reference count is incremented for each directly
+ *	  attached task, once for the dependent filter, and if
+ *	  requested for the user notifier. When @refs reaches zero,
+ *	  the filter can be freed.
+ * @users: A filter's @users count is incremented for each directly
+ *         attached task (filter installation, fork(), thread_sync),
+ *	   and once for the dependent filter (tracked in filter->prev).
+ *	   When it reaches zero it indicates that no direct or indirect
+ *	   users of that filter exist. No new tasks can get associated with
+ *	   this filter after reaching 0. The @users count is always smaller
+ *	   or equal to @refs. Hence, reaching 0 for @users does not mean
+ *	   the filter can be freed.
  * @log: true if all actions except for SECCOMP_RET_ALLOW should be logged
  * @prev: points to a previously installed, or inherited, filter
  * @prog: the BPF program to evaluate
  * @notif: the struct that holds all notification related information
  * @notify_lock: A lock for all notification-related accesses.
+ * @wqh: A wait queue for poll if a notifier is in use.
  *
  * seccomp_filter objects are organized in a tree linked via the @prev
  * pointer.  For any task, it appears to be a singly-linked list starting
@@ -124,15 +174,17 @@ struct notification {
  * how namespaces work.
  *
  * seccomp_filter objects should never be modified after being attached
- * to a task_struct (other than @usage).
+ * to a task_struct (other than @refs).
  */
 struct seccomp_filter {
-	refcount_t usage;
+	refcount_t refs;
+	refcount_t users;
 	bool log;
 	struct seccomp_filter *prev;
 	struct bpf_prog *prog;
 	struct notification *notif;
 	struct mutex notify_lock;
+	wait_queue_head_t wqh;
 };
 
 /* Limit any path through the tree to 256KB worth of instructions. */
@@ -366,6 +418,59 @@ static inline pid_t seccomp_can_sync_threads(void)
 	return 0;
 }
 
+static inline void seccomp_filter_free(struct seccomp_filter *filter)
+{
+	if (filter) {
+		bpf_prog_destroy(filter->prog);
+		kfree(filter);
+	}
+}
+
+static void __seccomp_filter_orphan(struct seccomp_filter *orig)
+{
+	while (orig && refcount_dec_and_test(&orig->users)) {
+		if (waitqueue_active(&orig->wqh))
+			wake_up_poll(&orig->wqh, EPOLLHUP);
+		orig = orig->prev;
+	}
+}
+
+static void __put_seccomp_filter(struct seccomp_filter *orig)
+{
+	/* Clean up single-reference branches iteratively. */
+	while (orig && refcount_dec_and_test(&orig->refs)) {
+		struct seccomp_filter *freeme = orig;
+		orig = orig->prev;
+		seccomp_filter_free(freeme);
+	}
+}
+
+static void __seccomp_filter_release(struct seccomp_filter *orig)
+{
+	/* Notify about any unused filters in the task's former filter tree. */
+	__seccomp_filter_orphan(orig);
+	/* Finally drop all references to the task's former tree. */
+	__put_seccomp_filter(orig);
+}
+
+/**
+ * seccomp_filter_release - Detach the task from its filter tree,
+ *			    drop its reference count, and notify
+ *			    about unused filters
+ *
+ * This function should only be called when the task is exiting as
+ * it detaches it from its filter tree. As such, READ_ONCE() and
+ * barriers are not needed here, as would normally be needed.
+ */
+void seccomp_filter_release(struct task_struct *tsk)
+{
+	struct seccomp_filter *orig = tsk->seccomp.filter;
+
+	/* Detach task from its filter tree. */
+	tsk->seccomp.filter = NULL;
+	__seccomp_filter_release(orig);
+}
+
 /**
  * seccomp_sync_threads: sets all threads to use current's filter
  *
@@ -390,14 +495,19 @@ static inline void seccomp_sync_threads(unsigned long flags)
 
 		/* Get a task reference for the new leaf node. */
 		get_seccomp_filter(caller);
+
 		/*
 		 * Drop the task reference to the shared ancestor since
 		 * current's path will hold a reference.  (This also
 		 * allows a put before the assignment.)
 		 */
-		put_seccomp_filter(thread);
+		__seccomp_filter_release(thread->seccomp.filter);
+
+		/* Make our new filter tree visible. */
 		smp_store_release(&thread->seccomp.filter,
 				  caller->seccomp.filter);
+		atomic_set(&thread->seccomp.filter_count,
+			   atomic_read(&thread->seccomp.filter_count));
 
 		/*
 		 * Don't let an unprivileged task work around
@@ -461,7 +571,9 @@ static struct seccomp_filter *seccomp_prepare_filter(struct sock_fprog *fprog)
 		return ERR_PTR(ret);
 	}
 
-	refcount_set(&sfilter->usage, 1);
+	refcount_set(&sfilter->refs, 1);
+	refcount_set(&sfilter->users, 1);
+	init_waitqueue_head(&sfilter->wqh);
 
 	return sfilter;
 }
@@ -544,6 +656,7 @@ static long seccomp_attach_filter(unsigned int flags,
 	 */
 	filter->prev = current->seccomp.filter;
 	current->seccomp.filter = filter;
+	atomic_inc(&current->seccomp.filter_count);
 
 	/* Now that the new filter is in place, synchronize to all threads. */
 	if (flags & SECCOMP_FILTER_FLAG_TSYNC)
@@ -554,7 +667,7 @@ static long seccomp_attach_filter(unsigned int flags,
 
 static void __get_seccomp_filter(struct seccomp_filter *filter)
 {
-	refcount_inc(&filter->usage);
+	refcount_inc(&filter->refs);
 }
 
 /* get_seccomp_filter - increments the reference count of the filter on @tsk */
@@ -564,30 +677,7 @@ void get_seccomp_filter(struct task_struct *tsk)
 	if (!orig)
 		return;
 	__get_seccomp_filter(orig);
-}
-
-static inline void seccomp_filter_free(struct seccomp_filter *filter)
-{
-	if (filter) {
-		bpf_prog_destroy(filter->prog);
-		kfree(filter);
-	}
-}
-
-static void __put_seccomp_filter(struct seccomp_filter *orig)
-{
-	/* Clean up single-reference branches iteratively. */
-	while (orig && refcount_dec_and_test(&orig->usage)) {
-		struct seccomp_filter *freeme = orig;
-		orig = orig->prev;
-		seccomp_filter_free(freeme);
-	}
-}
-
-/* put_seccomp_filter - decrements the ref count of tsk->seccomp.filter */
-void put_seccomp_filter(struct task_struct *tsk)
-{
-	__put_seccomp_filter(tsk->seccomp.filter);
+	refcount_inc(&orig->users);
 }
 
 static void seccomp_init_siginfo(kernel_siginfo_t *info, int syscall, int reason)
@@ -684,20 +774,20 @@ static inline void seccomp_log(unsigned long syscall, long signr, u32 action,
  */
 static const int mode1_syscalls[] = {
 	__NR_seccomp_read, __NR_seccomp_write, __NR_seccomp_exit, __NR_seccomp_sigreturn,
-	0, /* null terminated */
+	-1, /* negative terminated */
 };
 
 static void __secure_computing_strict(int this_syscall)
 {
-	const int *syscall_whitelist = mode1_syscalls;
+	const int *allowed_syscalls = mode1_syscalls;
 #ifdef CONFIG_COMPAT
 	if (in_compat_syscall())
-		syscall_whitelist = get_compat_mode1_syscalls();
+		allowed_syscalls = get_compat_mode1_syscalls();
 #endif
 	do {
-		if (*syscall_whitelist == this_syscall)
+		if (*allowed_syscalls == this_syscall)
 			return;
-	} while (*++syscall_whitelist);
+	} while (*++allowed_syscalls != -1);
 
 #ifdef SECCOMP_DEBUG
 	dump_stack();
@@ -735,6 +825,17 @@ static u64 seccomp_next_notify_id(struct seccomp_filter *filter)
 	return filter->notif->next_id++;
 }
 
+static void seccomp_handle_addfd(struct seccomp_kaddfd *addfd)
+{
+	/*
+	 * Remove the notification, and reset the list pointers, indicating
+	 * that it has been handled.
+	 */
+	list_del_init(&addfd->list);
+	addfd->ret = receive_fd_replace(addfd->fd, addfd->file, addfd->flags);
+	complete(&addfd->completion);
+}
+
 static int seccomp_do_user_notification(int this_syscall,
 					struct seccomp_filter *match,
 					const struct seccomp_data *sd)
@@ -743,6 +844,7 @@ static int seccomp_do_user_notification(int this_syscall,
 	u32 flags = 0;
 	long ret = 0;
 	struct seccomp_knotif n = {};
+	struct seccomp_kaddfd *addfd, *tmp;
 
 	mutex_lock(&match->notify_lock);
 	err = -ENOSYS;
@@ -755,25 +857,43 @@ static int seccomp_do_user_notification(int this_syscall,
 	n.id = seccomp_next_notify_id(match);
 	init_completion(&n.ready);
 	list_add(&n.list, &match->notif->notifications);
+	INIT_LIST_HEAD(&n.addfd);
 
 	up(&match->notif->request);
-	wake_up_poll(&match->notif->wqh, EPOLLIN | EPOLLRDNORM);
+	wake_up_poll(&match->wqh, EPOLLIN | EPOLLRDNORM);
 	mutex_unlock(&match->notify_lock);
 
 	/*
 	 * This is where we wait for a reply from userspace.
 	 */
+wait:
 	err = wait_for_completion_interruptible(&n.ready);
 	mutex_lock(&match->notify_lock);
 	if (err == 0) {
+		/* Check if we were woken up by a addfd message */
+		addfd = list_first_entry_or_null(&n.addfd,
+						 struct seccomp_kaddfd, list);
+		if (addfd && n.state != SECCOMP_NOTIFY_REPLIED) {
+			seccomp_handle_addfd(addfd);
+			mutex_unlock(&match->notify_lock);
+			goto wait;
+		}
 		ret = n.val;
 		err = n.error;
 		flags = n.flags;
 	}
 
+	/* If there were any pending addfd calls, clear them out */
+	list_for_each_entry_safe(addfd, tmp, &n.addfd, list) {
+		/* The process went away before we got a chance to handle it */
+		addfd->ret = -ESRCH;
+		list_del_init(&addfd->list);
+		complete(&addfd->completion);
+	}
+
 	/*
 	 * Note that it's possible the listener died in between the time when
-	 * we were notified of a respons (or a signal) and when we were able to
+	 * we were notified of a response (or a signal) and when we were able to
 	 * re-acquire the lock, so only delete from the list if the
 	 * notification actually exists.
 	 *
@@ -1011,6 +1131,11 @@ static int seccomp_notify_release(struct inode *inode, struct file *file)
 		knotif->error = -ENOSYS;
 		knotif->val = 0;
 
+		/*
+		 * We do not need to wake up any pending addfd messages, as
+		 * the notifier will do that for us, as this just looks
+		 * like a standard reply.
+		 */
 		complete(&knotif->ready);
 	}
 
@@ -1021,6 +1146,23 @@ static int seccomp_notify_release(struct inode *inode, struct file *file)
 	return 0;
 }
 
+/* must be called with notif_lock held */
+static inline struct seccomp_knotif *
+find_notification(struct seccomp_filter *filter, u64 id)
+{
+	struct seccomp_knotif *cur;
+
+	lockdep_assert_held(&filter->notify_lock);
+
+	list_for_each_entry(cur, &filter->notif->notifications, list) {
+		if (cur->id == id)
+			return cur;
+	}
+
+	return NULL;
+}
+
+
 static long seccomp_notify_recv(struct seccomp_filter *filter,
 				void __user *buf)
 {
@@ -1064,7 +1206,7 @@ static long seccomp_notify_recv(struct seccomp_filter *filter,
 	unotif.data = *(knotif->data);
 
 	knotif->state = SECCOMP_NOTIFY_SENT;
-	wake_up_poll(&filter->notif->wqh, EPOLLOUT | EPOLLWRNORM);
+	wake_up_poll(&filter->wqh, EPOLLOUT | EPOLLWRNORM);
 	ret = 0;
 out:
 	mutex_unlock(&filter->notify_lock);
@@ -1078,15 +1220,8 @@ static long seccomp_notify_recv(struct seccomp_filter *filter,
 		 * may have died when we released the lock, so we need to make
 		 * sure it's still around.
 		 */
-		knotif = NULL;
 		mutex_lock(&filter->notify_lock);
-		list_for_each_entry(cur, &filter->notif->notifications, list) {
-			if (cur->id == unotif.id) {
-				knotif = cur;
-				break;
-			}
-		}
-
+		knotif = find_notification(filter, unotif.id);
 		if (knotif) {
 			knotif->state = SECCOMP_NOTIFY_INIT;
 			up(&filter->notif->request);
@@ -1101,7 +1236,7 @@ static long seccomp_notify_send(struct seccomp_filter *filter,
 				void __user *buf)
 {
 	struct seccomp_notif_resp resp = {};
-	struct seccomp_knotif *knotif = NULL, *cur;
+	struct seccomp_knotif *knotif;
 	long ret;
 
 	if (copy_from_user(&resp, buf, sizeof(resp)))
@@ -1118,13 +1253,7 @@ static long seccomp_notify_send(struct seccomp_filter *filter,
 	if (ret < 0)
 		return ret;
 
-	list_for_each_entry(cur, &filter->notif->notifications, list) {
-		if (cur->id == resp.id) {
-			knotif = cur;
-			break;
-		}
-	}
-
+	knotif = find_notification(filter, resp.id);
 	if (!knotif) {
 		ret = -ENOENT;
 		goto out;
@@ -1150,7 +1279,7 @@ static long seccomp_notify_send(struct seccomp_filter *filter,
 static long seccomp_notify_id_valid(struct seccomp_filter *filter,
 				    void __user *buf)
 {
-	struct seccomp_knotif *knotif = NULL;
+	struct seccomp_knotif *knotif;
 	u64 id;
 	long ret;
 
@@ -1161,17 +1290,109 @@ static long seccomp_notify_id_valid(struct seccomp_filter *filter,
 	if (ret < 0)
 		return ret;
 
-	ret = -ENOENT;
-	list_for_each_entry(knotif, &filter->notif->notifications, list) {
-		if (knotif->id == id) {
-			if (knotif->state == SECCOMP_NOTIFY_SENT)
+	knotif = find_notification(filter, id);
+	if (knotif && knotif->state == SECCOMP_NOTIFY_SENT)
 		ret = 0;
+	else
+		ret = -ENOENT;
+
+	mutex_unlock(&filter->notify_lock);
+	return ret;
+}
+
+static long seccomp_notify_addfd(struct seccomp_filter *filter,
+				 struct seccomp_notif_addfd __user *uaddfd,
+				 unsigned int size)
+{
+	struct seccomp_notif_addfd addfd;
+	struct seccomp_knotif *knotif;
+	struct seccomp_kaddfd kaddfd;
+	int ret;
+
+	BUILD_BUG_ON(sizeof(addfd) < SECCOMP_NOTIFY_ADDFD_SIZE_VER0);
+	BUILD_BUG_ON(sizeof(addfd) != SECCOMP_NOTIFY_ADDFD_SIZE_LATEST);
+
+	if (size < SECCOMP_NOTIFY_ADDFD_SIZE_VER0 || size >= PAGE_SIZE)
+		return -EINVAL;
+
+	ret = copy_struct_from_user(&addfd, sizeof(addfd), uaddfd, size);
+	if (ret)
+		return ret;
+
+	if (addfd.newfd_flags & ~O_CLOEXEC)
+		return -EINVAL;
+
+	if (addfd.flags & ~SECCOMP_ADDFD_FLAG_SETFD)
+		return -EINVAL;
+
+	if (addfd.newfd && !(addfd.flags & SECCOMP_ADDFD_FLAG_SETFD))
+		return -EINVAL;
+
+	kaddfd.file = fget(addfd.srcfd);
+	if (!kaddfd.file)
+		return -EBADF;
+
+	kaddfd.flags = addfd.newfd_flags;
+	kaddfd.fd = (addfd.flags & SECCOMP_ADDFD_FLAG_SETFD) ?
+		    addfd.newfd : -1;
+	init_completion(&kaddfd.completion);
+
+	ret = mutex_lock_interruptible(&filter->notify_lock);
+	if (ret < 0)
 		goto out;
+
+	knotif = find_notification(filter, addfd.id);
+	if (!knotif) {
+		ret = -ENOENT;
+		goto out_unlock;
 	}
+
+	/*
+	 * We do not want to allow for FD injection to occur before the
+	 * notification has been picked up by a userspace handler, or after
+	 * the notification has been replied to.
+	 */
+	if (knotif->state != SECCOMP_NOTIFY_SENT) {
+		ret = -EINPROGRESS;
+		goto out_unlock;
 	}
 
-out:
+	list_add(&kaddfd.list, &knotif->addfd);
+	complete(&knotif->ready);
+	mutex_unlock(&filter->notify_lock);
+
+	/* Now we wait for it to be processed or be interrupted */
+	ret = wait_for_completion_interruptible(&kaddfd.completion);
+	if (ret == 0) {
+		/*
+		 * We had a successful completion. The other side has already
+		 * removed us from the addfd queue, and
+		 * wait_for_completion_interruptible has a memory barrier upon
+		 * success that lets us read this value directly without
+		 * locking.
+		 */
+		ret = kaddfd.ret;
+		goto out;
+	}
+
+	mutex_lock(&filter->notify_lock);
+	/*
+	 * Even though we were woken up by a signal and not a successful
+	 * completion, a completion may have happened in the mean time.
+	 *
+	 * We need to check again if the addfd request has been handled,
+	 * and if not, we will remove it from the queue.
+	 */
+	if (list_empty(&kaddfd.list))
+		ret = kaddfd.ret;
+	else
+		list_del(&kaddfd.list);
+
+out_unlock:
 	mutex_unlock(&filter->notify_lock);
+out:
+	fput(kaddfd.file);
+
 	return ret;
 }
 
@@ -1181,13 +1402,22 @@ static long seccomp_notify_ioctl(struct file *file, unsigned int cmd,
 	struct seccomp_filter *filter = file->private_data;
 	void __user *buf = (void __user *)arg;
 
+	/* Fixed-size ioctls */
 	switch (cmd) {
 	case SECCOMP_IOCTL_NOTIF_RECV:
 		return seccomp_notify_recv(filter, buf);
 	case SECCOMP_IOCTL_NOTIF_SEND:
 		return seccomp_notify_send(filter, buf);
+	case SECCOMP_IOCTL_NOTIF_ID_VALID_WRONG_DIR:
 	case SECCOMP_IOCTL_NOTIF_ID_VALID:
 		return seccomp_notify_id_valid(filter, buf);
+	}
+
+	/* Extensible Argument ioctls */
+#define EA_IOCTL(cmd)	((cmd) & ~(IOC_INOUT | IOCSIZE_MASK))
+	switch (EA_IOCTL(cmd)) {
+	case EA_IOCTL(SECCOMP_IOCTL_NOTIF_ADDFD):
+		return seccomp_notify_addfd(filter, buf, _IOC_SIZE(cmd));
 	default:
 		return -EINVAL;
 	}
@@ -1200,7 +1430,7 @@ static __poll_t seccomp_notify_poll(struct file *file,
 	__poll_t ret = 0;
 	struct seccomp_knotif *cur;
 
-	poll_wait(file, &filter->notif->wqh, poll_tab);
+	poll_wait(file, &filter->wqh, poll_tab);
 
 	if (mutex_lock_interruptible(&filter->notify_lock) < 0)
 		return EPOLLERR;
@@ -1216,6 +1446,9 @@ static __poll_t seccomp_notify_poll(struct file *file,
 
 	mutex_unlock(&filter->notify_lock);
 
+	if (refcount_read(&filter->users) == 0)
+		ret |= EPOLLHUP;
+
 	return ret;
 }
 
@@ -1244,7 +1477,6 @@ static struct file *init_listener(struct seccomp_filter *filter)
 	sema_init(&filter->notif->request, 0);
 	filter->notif->next_id = get_random_u64();
 	INIT_LIST_HEAD(&filter->notif->notifications);
-	init_waitqueue_head(&filter->notif->wqh);
 
 	ret = anon_inode_getfile("seccomp notify", &seccomp_notify_ops,
 				 filter, O_RDWR);
@@ -1822,7 +2054,7 @@ static int __init seccomp_sysctl_init(void)
 
 	hdr = register_sysctl_paths(seccomp_sysctl_path, seccomp_sysctl_table);
 	if (!hdr)
-		pr_warn("seccomp: sysctl registration failed\n");
+		pr_warn("sysctl registration failed\n");
 	else
 		kmemleak_not_leak(hdr);
 

net/compat.c

@@ -281,39 +281,31 @@ int put_cmsg_compat(struct msghdr *kmsg, int level, int type, int len, void *dat
 	return 0;
 }
 
-void scm_detach_fds_compat(struct msghdr *kmsg, struct scm_cookie *scm)
+static int scm_max_fds_compat(struct msghdr *msg)
 {
-	struct compat_cmsghdr __user *cm = (struct compat_cmsghdr __user *) kmsg->msg_control;
-	int fdmax = (kmsg->msg_controllen - sizeof(struct compat_cmsghdr)) / sizeof(int);
-	int fdnum = scm->fp->count;
-	struct file **fp = scm->fp->fp;
-	int __user *cmfptr;
-	int err = 0, i;
+	if (msg->msg_controllen <= sizeof(struct compat_cmsghdr))
+		return 0;
+	return (msg->msg_controllen - sizeof(struct compat_cmsghdr)) / sizeof(int);
+}
 
-	if (fdnum < fdmax)
-		fdmax = fdnum;
+void scm_detach_fds_compat(struct msghdr *msg, struct scm_cookie *scm)
+{
+	struct compat_cmsghdr __user *cm =
+		(struct compat_cmsghdr __user *)msg->msg_control;
+	unsigned int o_flags = (msg->msg_flags & MSG_CMSG_CLOEXEC) ? O_CLOEXEC : 0;
+	int fdmax = min_t(int, scm_max_fds_compat(msg), scm->fp->count);
+	int __user *cmsg_data = CMSG_USER_DATA(cm);
+	int err = 0, i;
 
-	for (i = 0, cmfptr = (int __user *) CMSG_COMPAT_DATA(cm); i < fdmax; i++, cmfptr++) {
-		int new_fd;
-		err = security_file_receive(fp[i]);
-		if (err)
-			break;
-		err = get_unused_fd_flags(MSG_CMSG_CLOEXEC & kmsg->msg_flags
-					  ? O_CLOEXEC : 0);
+	for (i = 0; i < fdmax; i++) {
+		err = receive_fd_user(scm->fp->fp[i], cmsg_data + i, o_flags);
 		if (err < 0)
 			break;
-		new_fd = err;
-		err = put_user(new_fd, cmfptr);
-		if (err) {
-			put_unused_fd(new_fd);
-			break;
-		}
-		/* Bump the usage count and install the file. */
-		fd_install(new_fd, get_file(fp[i]));
 	}
 
 	if (i > 0) {
 		int cmlen = CMSG_COMPAT_LEN(i * sizeof(int));
+
 		err = put_user(SOL_SOCKET, &cm->cmsg_level);
 		if (!err)
 			err = put_user(SCM_RIGHTS, &cm->cmsg_type);
@@ -321,16 +313,19 @@ void scm_detach_fds_compat(struct msghdr *kmsg, struct scm_cookie *scm)
 			err = put_user(cmlen, &cm->cmsg_len);
 		if (!err) {
 			cmlen = CMSG_COMPAT_SPACE(i * sizeof(int));
-			kmsg->msg_control += cmlen;
-			kmsg->msg_controllen -= cmlen;
+			if (msg->msg_controllen < cmlen)
+				cmlen = msg->msg_controllen;
+			msg->msg_control += cmlen;
+			msg->msg_controllen -= cmlen;
 		}
 	}
-	if (i < fdnum)
-		kmsg->msg_flags |= MSG_CTRUNC;
+
+	if (i < scm->fp->count || (scm->fp->count && fdmax <= 0))
+		msg->msg_flags |= MSG_CTRUNC;
 
 	/*
-	 * All of the files that fit in the message have had their
-	 * usage counts incremented, so we just free the list.
+	 * All of the files that fit in the message have had their usage counts
+	 * incremented, so we just free the list.
 	 */
 	__scm_destroy(scm);
 }

net/core/scm.c

@@ -280,36 +280,6 @@ void put_cmsg_scm_timestamping(struct msghdr *msg, struct scm_timestamping_inter
 }
 EXPORT_SYMBOL(put_cmsg_scm_timestamping);
 
-static int __scm_install_fd(struct file *file, int __user *ufd, int o_flags)
-{
-	struct socket *sock;
-	int new_fd;
-	int error;
-
-	error = security_file_receive(file);
-	if (error)
-		return error;
-
-	new_fd = get_unused_fd_flags(o_flags);
-	if (new_fd < 0)
-		return new_fd;
-
-	error = put_user(new_fd, ufd);
-	if (error) {
-		put_unused_fd(new_fd);
-		return error;
-	}
-
-	/* Bump the usage count and install the file. */
-	sock = sock_from_file(file, &error);
-	if (sock) {
-		sock_update_netprioidx(&sock->sk->sk_cgrp_data);
-		sock_update_classid(&sock->sk->sk_cgrp_data);
-	}
-	fd_install(new_fd, get_file(file));
-	return 0;
-}
-
 static int scm_max_fds(struct msghdr *msg)
 {
 	if (msg->msg_controllen <= sizeof(struct cmsghdr))
@@ -319,25 +289,25 @@ static int scm_max_fds(struct msghdr *msg)
 
 void scm_detach_fds(struct msghdr *msg, struct scm_cookie *scm)
 {
-	struct cmsghdr __user *cm
-		= (__force struct cmsghdr __user*)msg->msg_control;
-	int o_flags = (msg->msg_flags & MSG_CMSG_CLOEXEC) ? O_CLOEXEC : 0;
+	struct cmsghdr __user *cm =
+		(__force struct cmsghdr __user *)msg->msg_control;
+	unsigned int o_flags = (msg->msg_flags & MSG_CMSG_CLOEXEC) ? O_CLOEXEC : 0;
 	int fdmax = min_t(int, scm_max_fds(msg), scm->fp->count);
 	int __user *cmsg_data = CMSG_USER_DATA(cm);
 	int err = 0, i;
 
+	/* no use for FD passing from kernel space callers */
+	if (WARN_ON_ONCE(!msg->msg_control_is_user))
+		return;
+
 	if (msg->msg_flags & MSG_CMSG_COMPAT) {
 		scm_detach_fds_compat(msg, scm);
 		return;
 	}
 
-	/* no use for FD passing from kernel space callers */
-	if (WARN_ON_ONCE(!msg->msg_control_is_user))
-		return;
-
 	for (i = 0; i < fdmax; i++) {
-		err = __scm_install_fd(scm->fp->fp[i], cmsg_data + i, o_flags);
-		if (err)
+		err = receive_fd_user(scm->fp->fp[i], cmsg_data + i, o_flags);
+		if (err < 0)
 			break;
 	}
 

net/core/sock.c

@@ -2842,6 +2842,27 @@ int sock_no_mmap(struct file *file, struct socket *sock, struct vm_area_struct *
 }
 EXPORT_SYMBOL(sock_no_mmap);
 
+/*
+ * When a file is received (via SCM_RIGHTS, etc), we must bump the
+ * various sock-based usage counts.
+ */
+void __receive_sock(struct file *file)
+{
+	struct socket *sock;
+	int error;
+
+	/*
+	 * The resulting value of "error" is ignored here since we only
+	 * need to take action when the file is a socket and testing
+	 * "sock" for NULL is sufficient.
+	 */
+	sock = sock_from_file(file, &error);
+	if (sock) {
+		sock_update_netprioidx(&sock->sk->sk_cgrp_data);
+		sock_update_classid(&sock->sk->sk_cgrp_data);
+	}
+}
+
 ssize_t sock_no_sendpage(struct socket *sock, struct page *page, int offset, size_t size, int flags)
 {
 	ssize_t res;

tools/testing/selftests/kselftest_harness.h

@@ -195,8 +195,9 @@
  *
  * .. code-block:: c
  *
- *     FIXTURE_DATA(datatype name)
+ *     FIXTURE_DATA(datatype_name)
  *
+ * Almost always, you want just FIXTURE() instead (see below).
  * This call may be used when the type of the fixture data
  * is needed.  In general, this should not be needed unless
  * the *self* is being passed to a helper directly.
@@ -211,7 +212,7 @@
  *
  * .. code-block:: c
  *
- *     FIXTURE(datatype name) {
+ *     FIXTURE(fixture_name) {
  *       type property1;
  *       ...
  *     };
@@ -238,7 +239,7 @@
  *
  * .. code-block:: c
  *
- *     FIXTURE_SETUP(fixture name) { implementation }
+ *     FIXTURE_SETUP(fixture_name) { implementation }
  *
  * Populates the required "setup" function for a fixture.  An instance of the
  * datatype defined with FIXTURE_DATA() will be exposed as *self* for the
@@ -264,7 +265,7 @@
  *
  * .. code-block:: c
  *
- *     FIXTURE_TEARDOWN(fixture name) { implementation }
+ *     FIXTURE_TEARDOWN(fixture_name) { implementation }
  *
  * Populates the required "teardown" function for a fixture.  An instance of the
  * datatype defined with FIXTURE_DATA() will be exposed as *self* for the
@@ -285,7 +286,7 @@
  *
  * .. code-block:: c
  *
- *     FIXTURE_VARIANT(datatype name) {
+ *     FIXTURE_VARIANT(fixture_name) {
  *       type property1;
  *       ...
  *     };
@@ -305,8 +306,8 @@
  *
  * .. code-block:: c
  *
- *     FIXTURE_ADD(datatype name) {
- *       .property1 = val1;
+ *     FIXTURE_VARIANT_ADD(fixture_name, variant_name) {
+ *       .property1 = val1,
  *       ...
  *     };
  *

tools/testing/selftests/seccomp/config

 CONFIG_SECCOMP=y
 CONFIG_SECCOMP_FILTER=y
+CONFIG_USER_NS=y

tools/testing/selftests/seccomp/seccomp_benchmark.c

@@ -18,9 +18,9 @@
 
 unsigned long long timing(clockid_t clk_id, unsigned long long samples)
 {
-	pid_t pid, ret;
-	unsigned long long i;
 	struct timespec start, finish;
+	unsigned long long i;
+	pid_t pid, ret;
 
 	pid = getpid();
 	assert(clock_gettime(clk_id, &start) == 0);
@@ -31,30 +31,43 @@ unsigned long long timing(clockid_t clk_id, unsigned long long samples)
 	assert(clock_gettime(clk_id, &finish) == 0);
 
 	i = finish.tv_sec - start.tv_sec;
-	i *= 1000000000;
+	i *= 1000000000ULL;
 	i += finish.tv_nsec - start.tv_nsec;
 
-	printf("%lu.%09lu - %lu.%09lu = %llu\n",
+	printf("%lu.%09lu - %lu.%09lu = %llu (%.1fs)\n",
 		finish.tv_sec, finish.tv_nsec,
 		start.tv_sec, start.tv_nsec,
-		i);
+		i, (double)i / 1000000000.0);
 
 	return i;
 }
 
 unsigned long long calibrate(void)
 {
-	unsigned long long i;
-
-	printf("Calibrating reasonable sample size...\n");
+	struct timespec start, finish;
+	unsigned long long i, samples, step = 9973;
+	pid_t pid, ret;
+	int seconds = 15;
 
-	for (i = 5; ; i++) {
-		unsigned long long samples = 1 << i;
+	printf("Calibrating sample size for %d seconds worth of syscalls ...\n", seconds);
 
-		/* Find something that takes more than 5 seconds to run. */
-		if (timing(CLOCK_REALTIME, samples) / 1000000000ULL > 5)
-			return samples;
+	samples = 0;
+	pid = getpid();
+	assert(clock_gettime(CLOCK_MONOTONIC, &start) == 0);
+	do {
+		for (i = 0; i < step; i++) {
+			ret = syscall(__NR_getpid);
+			assert(pid == ret);
 		}
+		assert(clock_gettime(CLOCK_MONOTONIC, &finish) == 0);
+
+		samples += step;
+		i = finish.tv_sec - start.tv_sec;
+		i *= 1000000000ULL;
+		i += finish.tv_nsec - start.tv_nsec;
+	} while (i < 1000000000ULL);
+
+	return samples * seconds;
 }
 
 int main(int argc, char *argv[])
@@ -68,32 +81,55 @@ int main(int argc, char *argv[])
 	};
 	long ret;
 	unsigned long long samples;
-	unsigned long long native, filtered;
+	unsigned long long native, filter1, filter2;
+
+	printf("Current BPF sysctl settings:\n");
+	system("sysctl net.core.bpf_jit_enable");
+	system("sysctl net.core.bpf_jit_harden");
 
 	if (argc > 1)
 		samples = strtoull(argv[1], NULL, 0);
 	else
 		samples = calibrate();
 
-	printf("Benchmarking %llu samples...\n", samples);
+	printf("Benchmarking %llu syscalls...\n", samples);
 
+	/* Native call */
 	native = timing(CLOCK_PROCESS_CPUTIME_ID, samples) / samples;
 	printf("getpid native: %llu ns\n", native);
 
 	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
 	assert(ret == 0);
 
+	/* One filter */
 	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &prog);
 	assert(ret == 0);
 
-	filtered = timing(CLOCK_PROCESS_CPUTIME_ID, samples) / samples;
-	printf("getpid RET_ALLOW: %llu ns\n", filtered);
+	filter1 = timing(CLOCK_PROCESS_CPUTIME_ID, samples) / samples;
+	printf("getpid RET_ALLOW 1 filter: %llu ns\n", filter1);
+
+	if (filter1 == native)
+		printf("No overhead measured!? Try running again with more samples.\n");
+
+	/* Two filters */
+	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &prog);
+	assert(ret == 0);
+
+	filter2 = timing(CLOCK_PROCESS_CPUTIME_ID, samples) / samples;
+	printf("getpid RET_ALLOW 2 filters: %llu ns\n", filter2);
+
+	/* Calculations */
+	printf("Estimated total seccomp overhead for 1 filter: %llu ns\n",
+		filter1 - native);
+
+	printf("Estimated total seccomp overhead for 2 filters: %llu ns\n",
+		filter2 - native);
 
-	printf("Estimated seccomp overhead per syscall: %llu ns\n",
-		filtered - native);
+	printf("Estimated seccomp per-filter overhead: %llu ns\n",
+		filter2 - filter1);
 
-	if (filtered == native)
-		printf("Trying running again with more samples.\n");
+	printf("Estimated seccomp entry overhead: %llu ns\n",
+		filter1 - native - (filter2 - filter1));
 
 	return 0;
 }

tools/testing/selftests/seccomp/seccomp_bpf.c

@@ -45,12 +45,19 @@
 #include <sys/socket.h>
 #include <sys/ioctl.h>
 #include <linux/kcmp.h>
+#include <sys/resource.h>
 
 #include <unistd.h>
 #include <sys/syscall.h>
 #include <poll.h>
 
 #include "../kselftest_harness.h"
+#include "../clone3/clone3_selftests.h"
+
+/* Attempt to de-conflict with the selftests tree. */
+#ifndef SKIP
+#define SKIP(s, ...)	XFAIL(s, ##__VA_ARGS__)
+#endif
 
 #ifndef PR_SET_PTRACER
 # define PR_SET_PTRACER 0x59616d61
@@ -167,7 +174,9 @@ struct seccomp_metadata {
 
 #ifndef SECCOMP_FILTER_FLAG_NEW_LISTENER
 #define SECCOMP_FILTER_FLAG_NEW_LISTENER	(1UL << 3)
+#endif
 
+#ifndef SECCOMP_RET_USER_NOTIF
 #define SECCOMP_RET_USER_NOTIF 0x7fc00000U
 
 #define SECCOMP_IOC_MAGIC		'!'
@@ -180,7 +189,7 @@ struct seccomp_metadata {
 #define SECCOMP_IOCTL_NOTIF_RECV	SECCOMP_IOWR(0, struct seccomp_notif)
 #define SECCOMP_IOCTL_NOTIF_SEND	SECCOMP_IOWR(1,	\
 						struct seccomp_notif_resp)
-#define SECCOMP_IOCTL_NOTIF_ID_VALID	SECCOMP_IOR(2, __u64)
+#define SECCOMP_IOCTL_NOTIF_ID_VALID	SECCOMP_IOW(2, __u64)
 
 struct seccomp_notif {
 	__u64 id;
@@ -203,6 +212,39 @@ struct seccomp_notif_sizes {
 };
 #endif
 
+#ifndef SECCOMP_IOCTL_NOTIF_ADDFD
+/* On success, the return value is the remote process's added fd number */
+#define SECCOMP_IOCTL_NOTIF_ADDFD	SECCOMP_IOW(3,	\
+						struct seccomp_notif_addfd)
+
+/* valid flags for seccomp_notif_addfd */
+#define SECCOMP_ADDFD_FLAG_SETFD	(1UL << 0) /* Specify remote fd */
+
+struct seccomp_notif_addfd {
+	__u64 id;
+	__u32 flags;
+	__u32 srcfd;
+	__u32 newfd;
+	__u32 newfd_flags;
+};
+#endif
+
+struct seccomp_notif_addfd_small {
+	__u64 id;
+	char weird[4];
+};
+#define SECCOMP_IOCTL_NOTIF_ADDFD_SMALL	\
+	SECCOMP_IOW(3, struct seccomp_notif_addfd_small)
+
+struct seccomp_notif_addfd_big {
+	union {
+		struct seccomp_notif_addfd addfd;
+		char buf[sizeof(struct seccomp_notif_addfd) + 8];
+	};
+};
+#define SECCOMP_IOCTL_NOTIF_ADDFD_BIG	\
+	SECCOMP_IOWR(3, struct seccomp_notif_addfd_big)
+
 #ifndef PTRACE_EVENTMSG_SYSCALL_ENTRY
 #define PTRACE_EVENTMSG_SYSCALL_ENTRY	1
 #define PTRACE_EVENTMSG_SYSCALL_EXIT	2
@@ -236,6 +278,40 @@ int seccomp(unsigned int op, unsigned int flags, void *args)
 #define SIBLING_EXIT_FAILURE	0xbadface
 #define SIBLING_EXIT_NEWPRIVS	0xbadfeed
 
+static int __filecmp(pid_t pid1, pid_t pid2, int fd1, int fd2)
+{
+#ifdef __NR_kcmp
+	errno = 0;
+	return syscall(__NR_kcmp, pid1, pid2, KCMP_FILE, fd1, fd2);
+#else
+	errno = ENOSYS;
+	return -1;
+#endif
+}
+
+/* Have TH_LOG report actual location filecmp() is used. */
+#define filecmp(pid1, pid2, fd1, fd2)	({		\
+	int _ret;					\
+							\
+	_ret = __filecmp(pid1, pid2, fd1, fd2);		\
+	if (_ret != 0) {				\
+		if (_ret < 0 && errno == ENOSYS) {	\
+			TH_LOG("kcmp() syscall missing (test is less accurate)");\
+			_ret = 0;			\
+		}					\
+	}						\
+	_ret; })
+
+TEST(kcmp)
+{
+	int ret;
+
+	ret = __filecmp(getpid(), getpid(), 1, 1);
+	EXPECT_EQ(ret, 0);
+	if (ret != 0 && errno == ENOSYS)
+		SKIP(return, "Kernel does not support kcmp() (missing CONFIG_CHECKPOINT_RESTORE?)");
+}
+
 TEST(mode_strict_support)
 {
 	long ret;
@@ -1470,6 +1546,7 @@ pid_t setup_trace_fixture(struct __test_metadata *_metadata,
 
 	return tracer_pid;
 }
+
 void teardown_trace_fixture(struct __test_metadata *_metadata,
 			    pid_t tracer)
 {
@@ -1750,7 +1827,7 @@ void change_syscall(struct __test_metadata *_metadata,
 	EXPECT_EQ(0, ret);
 }
 
-void tracer_syscall(struct __test_metadata *_metadata, pid_t tracee,
+void tracer_seccomp(struct __test_metadata *_metadata, pid_t tracee,
 		    int status, void *args)
 {
 	int ret;
@@ -1827,6 +1904,24 @@ FIXTURE(TRACE_syscall) {
 	pid_t tracer, mytid, mypid, parent;
 };
 
+FIXTURE_VARIANT(TRACE_syscall) {
+	/*
+	 * All of the SECCOMP_RET_TRACE behaviors can be tested with either
+	 * SECCOMP_RET_TRACE+PTRACE_CONT or plain ptrace()+PTRACE_SYSCALL.
+	 * This indicates if we should use SECCOMP_RET_TRACE (false), or
+	 * ptrace (true).
+	 */
+	bool use_ptrace;
+};
+
+FIXTURE_VARIANT_ADD(TRACE_syscall, ptrace) {
+	.use_ptrace = true,
+};
+
+FIXTURE_VARIANT_ADD(TRACE_syscall, seccomp) {
+	.use_ptrace = false,
+};
+
 FIXTURE_SETUP(TRACE_syscall)
 {
 	struct sock_filter filter[] = {
@@ -1842,12 +1937,11 @@ FIXTURE_SETUP(TRACE_syscall)
 		BPF_STMT(BPF_RET|BPF_K, SECCOMP_RET_TRACE | 0x1005),
 		BPF_STMT(BPF_RET|BPF_K, SECCOMP_RET_ALLOW),
 	};
-
-	memset(&self->prog, 0, sizeof(self->prog));
-	self->prog.filter = malloc(sizeof(filter));
-	ASSERT_NE(NULL, self->prog.filter);
-	memcpy(self->prog.filter, filter, sizeof(filter));
-	self->prog.len = (unsigned short)ARRAY_SIZE(filter);
+	struct sock_fprog prog = {
+		.len = (unsigned short)ARRAY_SIZE(filter),
+		.filter = filter,
+	};
+	long ret;
 
 	/* Prepare some testable syscall results. */
 	self->mytid = syscall(__NR_gettid);
@@ -1865,60 +1959,48 @@ FIXTURE_SETUP(TRACE_syscall)
 	ASSERT_NE(self->parent, self->mypid);
 
 	/* Launch tracer. */
-	self->tracer = setup_trace_fixture(_metadata, tracer_syscall, NULL,
-					   false);
-}
+	self->tracer = setup_trace_fixture(_metadata,
+					   variant->use_ptrace ? tracer_ptrace
+							       : tracer_seccomp,
+					   NULL, variant->use_ptrace);
 
-FIXTURE_TEARDOWN(TRACE_syscall)
-{
-	teardown_trace_fixture(_metadata, self->tracer);
-	if (self->prog.filter)
-		free(self->prog.filter);
-}
+	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
+	ASSERT_EQ(0, ret);
 
-TEST_F(TRACE_syscall, ptrace_syscall_redirected)
-{
-	/* Swap SECCOMP_RET_TRACE tracer for PTRACE_SYSCALL tracer. */
-	teardown_trace_fixture(_metadata, self->tracer);
-	self->tracer = setup_trace_fixture(_metadata, tracer_ptrace, NULL,
-					   true);
+	if (variant->use_ptrace)
+		return;
 
-	/* Tracer will redirect getpid to getppid. */
-	EXPECT_NE(self->mypid, syscall(__NR_getpid));
+	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &prog, 0, 0);
+	ASSERT_EQ(0, ret);
 }
 
-TEST_F(TRACE_syscall, ptrace_syscall_errno)
+FIXTURE_TEARDOWN(TRACE_syscall)
 {
-	/* Swap SECCOMP_RET_TRACE tracer for PTRACE_SYSCALL tracer. */
 	teardown_trace_fixture(_metadata, self->tracer);
-	self->tracer = setup_trace_fixture(_metadata, tracer_ptrace, NULL,
-					   true);
-
-	/* Tracer should skip the open syscall, resulting in ESRCH. */
-	EXPECT_SYSCALL_RETURN(-ESRCH, syscall(__NR_openat));
 }
 
-TEST_F(TRACE_syscall, ptrace_syscall_faked)
+TEST(negative_ENOSYS)
 {
-	/* Swap SECCOMP_RET_TRACE tracer for PTRACE_SYSCALL tracer. */
-	teardown_trace_fixture(_metadata, self->tracer);
-	self->tracer = setup_trace_fixture(_metadata, tracer_ptrace, NULL,
-					   true);
+	/*
+	 * There should be no difference between an "internal" skip
+	 * and userspace asking for syscall "-1".
+	 */
+	errno = 0;
+	EXPECT_EQ(-1, syscall(-1));
+	EXPECT_EQ(errno, ENOSYS);
+	/* And no difference for "still not valid but not -1". */
+	errno = 0;
+	EXPECT_EQ(-1, syscall(-101));
+	EXPECT_EQ(errno, ENOSYS);
+}
 
-	/* Tracer should skip the gettid syscall, resulting fake pid. */
-	EXPECT_SYSCALL_RETURN(45000, syscall(__NR_gettid));
+TEST_F(TRACE_syscall, negative_ENOSYS)
+{
+	negative_ENOSYS(_metadata);
 }
 
 TEST_F(TRACE_syscall, syscall_allowed)
 {
-	long ret;
-
-	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &self->prog, 0, 0);
-	ASSERT_EQ(0, ret);
-
 	/* getppid works as expected (no changes). */
 	EXPECT_EQ(self->parent, syscall(__NR_getppid));
 	EXPECT_NE(self->mypid, syscall(__NR_getppid));
@@ -1926,14 +2008,6 @@ TEST_F(TRACE_syscall, syscall_allowed)
 
 TEST_F(TRACE_syscall, syscall_redirected)
 {
-	long ret;
-
-	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &self->prog, 0, 0);
-	ASSERT_EQ(0, ret);
-
 	/* getpid has been redirected to getppid as expected. */
 	EXPECT_EQ(self->parent, syscall(__NR_getpid));
 	EXPECT_NE(self->mypid, syscall(__NR_getpid));
@@ -1941,33 +2015,17 @@ TEST_F(TRACE_syscall, syscall_redirected)
 
 TEST_F(TRACE_syscall, syscall_errno)
 {
-	long ret;
-
-	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &self->prog, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	/* openat has been skipped and an errno return. */
+	/* Tracer should skip the open syscall, resulting in ESRCH. */
 	EXPECT_SYSCALL_RETURN(-ESRCH, syscall(__NR_openat));
 }
 
 TEST_F(TRACE_syscall, syscall_faked)
 {
-	long ret;
-
-	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &self->prog, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	/* gettid has been skipped and an altered return value stored. */
+	/* Tracer skips the gettid syscall and store altered return value. */
 	EXPECT_SYSCALL_RETURN(45000, syscall(__NR_gettid));
 }
 
-TEST_F(TRACE_syscall, skip_after_RET_TRACE)
+TEST_F(TRACE_syscall, skip_after)
 {
 	struct sock_filter filter[] = {
 		BPF_STMT(BPF_LD|BPF_W|BPF_ABS,
@@ -1982,14 +2040,7 @@ TEST_F(TRACE_syscall, skip_after_RET_TRACE)
 	};
 	long ret;
 
-	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	/* Install fixture filter. */
-	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &self->prog, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	/* Install "errno on getppid" filter. */
+	/* Install additional "errno on getppid" filter. */
 	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &prog, 0, 0);
 	ASSERT_EQ(0, ret);
 
@@ -1999,69 +2050,7 @@ TEST_F(TRACE_syscall, skip_after_RET_TRACE)
 	EXPECT_EQ(EPERM, errno);
 }
 
-TEST_F_SIGNAL(TRACE_syscall, kill_after_RET_TRACE, SIGSYS)
-{
-	struct sock_filter filter[] = {
-		BPF_STMT(BPF_LD|BPF_W|BPF_ABS,
-			offsetof(struct seccomp_data, nr)),
-		BPF_JUMP(BPF_JMP|BPF_JEQ|BPF_K, __NR_getppid, 0, 1),
-		BPF_STMT(BPF_RET|BPF_K, SECCOMP_RET_KILL),
-		BPF_STMT(BPF_RET|BPF_K, SECCOMP_RET_ALLOW),
-	};
-	struct sock_fprog prog = {
-		.len = (unsigned short)ARRAY_SIZE(filter),
-		.filter = filter,
-	};
-	long ret;
-
-	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	/* Install fixture filter. */
-	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &self->prog, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	/* Install "death on getppid" filter. */
-	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &prog, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	/* Tracer will redirect getpid to getppid, and we should die. */
-	EXPECT_NE(self->mypid, syscall(__NR_getpid));
-}
-
-TEST_F(TRACE_syscall, skip_after_ptrace)
-{
-	struct sock_filter filter[] = {
-		BPF_STMT(BPF_LD|BPF_W|BPF_ABS,
-			offsetof(struct seccomp_data, nr)),
-		BPF_JUMP(BPF_JMP|BPF_JEQ|BPF_K, __NR_getppid, 0, 1),
-		BPF_STMT(BPF_RET|BPF_K, SECCOMP_RET_ERRNO | EPERM),
-		BPF_STMT(BPF_RET|BPF_K, SECCOMP_RET_ALLOW),
-	};
-	struct sock_fprog prog = {
-		.len = (unsigned short)ARRAY_SIZE(filter),
-		.filter = filter,
-	};
-	long ret;
-
-	/* Swap SECCOMP_RET_TRACE tracer for PTRACE_SYSCALL tracer. */
-	teardown_trace_fixture(_metadata, self->tracer);
-	self->tracer = setup_trace_fixture(_metadata, tracer_ptrace, NULL,
-					   true);
-
-	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	/* Install "errno on getppid" filter. */
-	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &prog, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	/* Tracer will redirect getpid to getppid, and we should see EPERM. */
-	EXPECT_EQ(-1, syscall(__NR_getpid));
-	EXPECT_EQ(EPERM, errno);
-}
-
-TEST_F_SIGNAL(TRACE_syscall, kill_after_ptrace, SIGSYS)
+TEST_F_SIGNAL(TRACE_syscall, kill_after, SIGSYS)
 {
 	struct sock_filter filter[] = {
 		BPF_STMT(BPF_LD|BPF_W|BPF_ABS,
@@ -2076,15 +2065,7 @@ TEST_F_SIGNAL(TRACE_syscall, kill_after_ptrace, SIGSYS)
 	};
 	long ret;
 
-	/* Swap SECCOMP_RET_TRACE tracer for PTRACE_SYSCALL tracer. */
-	teardown_trace_fixture(_metadata, self->tracer);
-	self->tracer = setup_trace_fixture(_metadata, tracer_ptrace, NULL,
-					   true);
-
-	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
-	ASSERT_EQ(0, ret);
-
-	/* Install "death on getppid" filter. */
+	/* Install additional "death on getppid" filter. */
 	ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &prog, 0, 0);
 	ASSERT_EQ(0, ret);
 
@@ -3069,7 +3050,7 @@ TEST(get_metadata)
 
 	/* Only real root can get metadata. */
 	if (geteuid()) {
-		XFAIL(return, "get_metadata requires real root");
+		SKIP(return, "get_metadata requires real root");
 		return;
 	}
 
@@ -3112,7 +3093,7 @@ TEST(get_metadata)
 	ret = ptrace(PTRACE_SECCOMP_GET_METADATA, pid, sizeof(md), &md);
 	EXPECT_EQ(sizeof(md), ret) {
 		if (errno == EINVAL)
-			XFAIL(goto skip, "Kernel does not support PTRACE_SECCOMP_GET_METADATA (missing CONFIG_CHECKPOINT_RESTORE?)");
+			SKIP(goto skip, "Kernel does not support PTRACE_SECCOMP_GET_METADATA (missing CONFIG_CHECKPOINT_RESTORE?)");
 	}
 
 	EXPECT_EQ(md.flags, SECCOMP_FILTER_FLAG_LOG);
@@ -3128,7 +3109,7 @@ TEST(get_metadata)
 	ASSERT_EQ(0, kill(pid, SIGKILL));
 }
 
-static int user_trap_syscall(int nr, unsigned int flags)
+static int user_notif_syscall(int nr, unsigned int flags)
 {
 	struct sock_filter filter[] = {
 		BPF_STMT(BPF_LD+BPF_W+BPF_ABS,
@@ -3174,7 +3155,7 @@ TEST(user_notification_basic)
 
 	/* Check that we get -ENOSYS with no listener attached */
 	if (pid == 0) {
-		if (user_trap_syscall(__NR_getppid, 0) < 0)
+		if (user_notif_syscall(__NR_getppid, 0) < 0)
 			exit(1);
 		ret = syscall(__NR_getppid);
 		exit(ret >= 0 || errno != ENOSYS);
@@ -3191,12 +3172,12 @@ TEST(user_notification_basic)
 	EXPECT_EQ(seccomp(SECCOMP_SET_MODE_FILTER, 0, &prog), 0);
 
 	/* Check that the basic notification machinery works */
-	listener = user_trap_syscall(__NR_getppid,
+	listener = user_notif_syscall(__NR_getppid,
 				      SECCOMP_FILTER_FLAG_NEW_LISTENER);
 	ASSERT_GE(listener, 0);
 
 	/* Installing a second listener in the chain should EBUSY */
-	EXPECT_EQ(user_trap_syscall(__NR_getppid,
+	EXPECT_EQ(user_notif_syscall(__NR_getppid,
 				     SECCOMP_FILTER_FLAG_NEW_LISTENER),
 		  -1);
 	EXPECT_EQ(errno, EBUSY);
@@ -3258,15 +3239,20 @@ TEST(user_notification_with_tsync)
 	int ret;
 	unsigned int flags;
 
+	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
+	ASSERT_EQ(0, ret) {
+		TH_LOG("Kernel does not support PR_SET_NO_NEW_PRIVS!");
+	}
+
 	/* these were exclusive */
 	flags = SECCOMP_FILTER_FLAG_NEW_LISTENER |
 		SECCOMP_FILTER_FLAG_TSYNC;
-	ASSERT_EQ(-1, user_trap_syscall(__NR_getppid, flags));
+	ASSERT_EQ(-1, user_notif_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);
+	ret = user_notif_syscall(__NR_getppid, flags);
 	close(ret);
 	ASSERT_LE(0, ret);
 }
@@ -3284,7 +3270,7 @@ TEST(user_notification_kill_in_middle)
 		TH_LOG("Kernel does not support PR_SET_NO_NEW_PRIVS!");
 	}
 
-	listener = user_trap_syscall(__NR_getppid,
+	listener = user_notif_syscall(__NR_getppid,
 				      SECCOMP_FILTER_FLAG_NEW_LISTENER);
 	ASSERT_GE(listener, 0);
 
@@ -3338,7 +3324,7 @@ TEST(user_notification_signal)
 
 	ASSERT_EQ(socketpair(PF_LOCAL, SOCK_SEQPACKET, 0, sk_pair), 0);
 
-	listener = user_trap_syscall(__NR_gettid,
+	listener = user_notif_syscall(__NR_gettid,
 				      SECCOMP_FILTER_FLAG_NEW_LISTENER);
 	ASSERT_GE(listener, 0);
 
@@ -3408,7 +3394,7 @@ TEST(user_notification_closed_listener)
 		TH_LOG("Kernel does not support PR_SET_NO_NEW_PRIVS!");
 	}
 
-	listener = user_trap_syscall(__NR_getppid,
+	listener = user_notif_syscall(__NR_getppid,
 				      SECCOMP_FILTER_FLAG_NEW_LISTENER);
 	ASSERT_GE(listener, 0);
 
@@ -3440,9 +3426,12 @@ TEST(user_notification_child_pid_ns)
 	struct seccomp_notif req = {};
 	struct seccomp_notif_resp resp = {};
 
-	ASSERT_EQ(unshare(CLONE_NEWUSER | CLONE_NEWPID), 0);
+	ASSERT_EQ(unshare(CLONE_NEWUSER | CLONE_NEWPID), 0) {
+		if (errno == EINVAL)
+			SKIP(return, "kernel missing CLONE_NEWUSER support");
+	};
 
-	listener = user_trap_syscall(__NR_getppid,
+	listener = user_notif_syscall(__NR_getppid,
 				      SECCOMP_FILTER_FLAG_NEW_LISTENER);
 	ASSERT_GE(listener, 0);
 
@@ -3482,7 +3471,7 @@ TEST(user_notification_sibling_pid_ns)
 		TH_LOG("Kernel does not support PR_SET_NO_NEW_PRIVS!");
 	}
 
-	listener = user_trap_syscall(__NR_getppid,
+	listener = user_notif_syscall(__NR_getppid,
 				      SECCOMP_FILTER_FLAG_NEW_LISTENER);
 	ASSERT_GE(listener, 0);
 
@@ -3505,7 +3494,10 @@ TEST(user_notification_sibling_pid_ns)
 	}
 
 	/* Create the sibling ns, and sibling in it. */
-	ASSERT_EQ(unshare(CLONE_NEWPID), 0);
+	ASSERT_EQ(unshare(CLONE_NEWPID), 0) {
+		if (errno == EPERM)
+			SKIP(return, "CLONE_NEWPID requires CAP_SYS_ADMIN");
+	}
 	ASSERT_EQ(errno, 0);
 
 	pid2 = fork();
@@ -3547,7 +3539,7 @@ TEST(user_notification_fault_recv)
 
 	ASSERT_EQ(unshare(CLONE_NEWUSER), 0);
 
-	listener = user_trap_syscall(__NR_getppid,
+	listener = user_notif_syscall(__NR_getppid,
 				      SECCOMP_FILTER_FLAG_NEW_LISTENER);
 	ASSERT_GE(listener, 0);
 
@@ -3585,16 +3577,6 @@ TEST(seccomp_get_notif_sizes)
 	EXPECT_EQ(sizes.seccomp_notif_resp, sizeof(struct seccomp_notif_resp));
 }
 
-static int filecmp(pid_t pid1, pid_t pid2, int fd1, int fd2)
-{
-#ifdef __NR_kcmp
-	return syscall(__NR_kcmp, pid1, pid2, KCMP_FILE, fd1, fd2);
-#else
-	errno = ENOSYS;
-	return -1;
-#endif
-}
-
 TEST(user_notification_continue)
 {
 	pid_t pid;
@@ -3609,7 +3591,7 @@ TEST(user_notification_continue)
 		TH_LOG("Kernel does not support PR_SET_NO_NEW_PRIVS!");
 	}
 
-	listener = user_trap_syscall(__NR_dup, SECCOMP_FILTER_FLAG_NEW_LISTENER);
+	listener = user_notif_syscall(__NR_dup, SECCOMP_FILTER_FLAG_NEW_LISTENER);
 	ASSERT_GE(listener, 0);
 
 	pid = fork();
@@ -3619,20 +3601,14 @@ TEST(user_notification_continue)
 		int dup_fd, pipe_fds[2];
 		pid_t self;
 
-		ret = pipe(pipe_fds);
-		if (ret < 0)
-			exit(1);
+		ASSERT_GE(pipe(pipe_fds), 0);
 
 		dup_fd = dup(pipe_fds[0]);
-		if (dup_fd < 0)
-			exit(1);
+		ASSERT_GE(dup_fd, 0);
+		EXPECT_NE(pipe_fds[0], dup_fd);
 
 		self = getpid();
-
-		ret = filecmp(self, self, pipe_fds[0], dup_fd);
-		if (ret)
-			exit(2);
-
+		ASSERT_EQ(filecmp(self, self, pipe_fds[0], dup_fd), 0);
 		exit(0);
 	}
 
@@ -3673,7 +3649,7 @@ TEST(user_notification_continue)
 	resp.val = 0;
 	EXPECT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_SEND, &resp), 0) {
 		if (errno == EINVAL)
-			XFAIL(goto skip, "Kernel does not support SECCOMP_USER_NOTIF_FLAG_CONTINUE");
+			SKIP(goto skip, "Kernel does not support SECCOMP_USER_NOTIF_FLAG_CONTINUE");
 	}
 
 skip:
@@ -3681,15 +3657,342 @@ TEST(user_notification_continue)
 	EXPECT_EQ(true, WIFEXITED(status));
 	EXPECT_EQ(0, WEXITSTATUS(status)) {
 		if (WEXITSTATUS(status) == 2) {
-			XFAIL(return, "Kernel does not support kcmp() syscall");
+			SKIP(return, "Kernel does not support kcmp() syscall");
 			return;
 		}
 	}
 }
 
+TEST(user_notification_filter_empty)
+{
+	pid_t pid;
+	long ret;
+	int status;
+	struct pollfd pollfd;
+	struct clone_args args = {
+		.flags = CLONE_FILES,
+		.exit_signal = SIGCHLD,
+	};
+
+	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
+	ASSERT_EQ(0, ret) {
+		TH_LOG("Kernel does not support PR_SET_NO_NEW_PRIVS!");
+	}
+
+	pid = sys_clone3(&args, sizeof(args));
+	ASSERT_GE(pid, 0);
+
+	if (pid == 0) {
+		int listener;
+
+		listener = user_notif_syscall(__NR_mknod, SECCOMP_FILTER_FLAG_NEW_LISTENER);
+		if (listener < 0)
+			_exit(EXIT_FAILURE);
+
+		if (dup2(listener, 200) != 200)
+			_exit(EXIT_FAILURE);
+
+		close(listener);
+
+		_exit(EXIT_SUCCESS);
+	}
+
+	EXPECT_EQ(waitpid(pid, &status, 0), pid);
+	EXPECT_EQ(true, WIFEXITED(status));
+	EXPECT_EQ(0, WEXITSTATUS(status));
+
+	/*
+	 * The seccomp filter has become unused so we should be notified once
+	 * the kernel gets around to cleaning up task struct.
+	 */
+	pollfd.fd = 200;
+	pollfd.events = POLLHUP;
+
+	EXPECT_GT(poll(&pollfd, 1, 2000), 0);
+	EXPECT_GT((pollfd.revents & POLLHUP) ?: 0, 0);
+}
+
+static void *do_thread(void *data)
+{
+	return NULL;
+}
+
+TEST(user_notification_filter_empty_threaded)
+{
+	pid_t pid;
+	long ret;
+	int status;
+	struct pollfd pollfd;
+	struct clone_args args = {
+		.flags = CLONE_FILES,
+		.exit_signal = SIGCHLD,
+	};
+
+	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
+	ASSERT_EQ(0, ret) {
+		TH_LOG("Kernel does not support PR_SET_NO_NEW_PRIVS!");
+	}
+
+	pid = sys_clone3(&args, sizeof(args));
+	ASSERT_GE(pid, 0);
+
+	if (pid == 0) {
+		pid_t pid1, pid2;
+		int listener, status;
+		pthread_t thread;
+
+		listener = user_notif_syscall(__NR_dup, SECCOMP_FILTER_FLAG_NEW_LISTENER);
+		if (listener < 0)
+			_exit(EXIT_FAILURE);
+
+		if (dup2(listener, 200) != 200)
+			_exit(EXIT_FAILURE);
+
+		close(listener);
+
+		pid1 = fork();
+		if (pid1 < 0)
+			_exit(EXIT_FAILURE);
+
+		if (pid1 == 0)
+			_exit(EXIT_SUCCESS);
+
+		pid2 = fork();
+		if (pid2 < 0)
+			_exit(EXIT_FAILURE);
+
+		if (pid2 == 0)
+			_exit(EXIT_SUCCESS);
+
+		if (pthread_create(&thread, NULL, do_thread, NULL) ||
+		    pthread_join(thread, NULL))
+			_exit(EXIT_FAILURE);
+
+		if (pthread_create(&thread, NULL, do_thread, NULL) ||
+		    pthread_join(thread, NULL))
+			_exit(EXIT_FAILURE);
+
+		if (waitpid(pid1, &status, 0) != pid1 || !WIFEXITED(status) ||
+		    WEXITSTATUS(status))
+			_exit(EXIT_FAILURE);
+
+		if (waitpid(pid2, &status, 0) != pid2 || !WIFEXITED(status) ||
+		    WEXITSTATUS(status))
+			_exit(EXIT_FAILURE);
+
+		exit(EXIT_SUCCESS);
+	}
+
+	EXPECT_EQ(waitpid(pid, &status, 0), pid);
+	EXPECT_EQ(true, WIFEXITED(status));
+	EXPECT_EQ(0, WEXITSTATUS(status));
+
+	/*
+	 * The seccomp filter has become unused so we should be notified once
+	 * the kernel gets around to cleaning up task struct.
+	 */
+	pollfd.fd = 200;
+	pollfd.events = POLLHUP;
+
+	EXPECT_GT(poll(&pollfd, 1, 2000), 0);
+	EXPECT_GT((pollfd.revents & POLLHUP) ?: 0, 0);
+}
+
+TEST(user_notification_addfd)
+{
+	pid_t pid;
+	long ret;
+	int status, listener, memfd, fd;
+	struct seccomp_notif_addfd addfd = {};
+	struct seccomp_notif_addfd_small small = {};
+	struct seccomp_notif_addfd_big big = {};
+	struct seccomp_notif req = {};
+	struct seccomp_notif_resp resp = {};
+	/* 100 ms */
+	struct timespec delay = { .tv_nsec = 100000000 };
+
+	memfd = memfd_create("test", 0);
+	ASSERT_GE(memfd, 0);
+
+	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
+	ASSERT_EQ(0, ret) {
+		TH_LOG("Kernel does not support PR_SET_NO_NEW_PRIVS!");
+	}
+
+	/* Check that the basic notification machinery works */
+	listener = user_notif_syscall(__NR_getppid,
+				      SECCOMP_FILTER_FLAG_NEW_LISTENER);
+	ASSERT_GE(listener, 0);
+
+	pid = fork();
+	ASSERT_GE(pid, 0);
+
+	if (pid == 0) {
+		if (syscall(__NR_getppid) != USER_NOTIF_MAGIC)
+			exit(1);
+		exit(syscall(__NR_getppid) != USER_NOTIF_MAGIC);
+	}
+
+	ASSERT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_RECV, &req), 0);
+
+	addfd.srcfd = memfd;
+	addfd.newfd = 0;
+	addfd.id = req.id;
+	addfd.flags = 0x0;
+
+	/* Verify bad newfd_flags cannot be set */
+	addfd.newfd_flags = ~O_CLOEXEC;
+	EXPECT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD, &addfd), -1);
+	EXPECT_EQ(errno, EINVAL);
+	addfd.newfd_flags = O_CLOEXEC;
+
+	/* Verify bad flags cannot be set */
+	addfd.flags = 0xff;
+	EXPECT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD, &addfd), -1);
+	EXPECT_EQ(errno, EINVAL);
+	addfd.flags = 0;
+
+	/* Verify that remote_fd cannot be set without setting flags */
+	addfd.newfd = 1;
+	EXPECT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD, &addfd), -1);
+	EXPECT_EQ(errno, EINVAL);
+	addfd.newfd = 0;
+
+	/* Verify small size cannot be set */
+	EXPECT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD_SMALL, &small), -1);
+	EXPECT_EQ(errno, EINVAL);
+
+	/* Verify we can't send bits filled in unknown buffer area */
+	memset(&big, 0xAA, sizeof(big));
+	big.addfd = addfd;
+	EXPECT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD_BIG, &big), -1);
+	EXPECT_EQ(errno, E2BIG);
+
+
+	/* Verify we can set an arbitrary remote fd */
+	fd = ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD, &addfd);
+	/*
+	 * The child has fds 0(stdin), 1(stdout), 2(stderr), 3(memfd),
+	 * 4(listener), so the newly allocated fd should be 5.
+	 */
+	EXPECT_EQ(fd, 5);
+	EXPECT_EQ(filecmp(getpid(), pid, memfd, fd), 0);
+
+	/* Verify we can set an arbitrary remote fd with large size */
+	memset(&big, 0x0, sizeof(big));
+	big.addfd = addfd;
+	fd = ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD_BIG, &big);
+	EXPECT_EQ(fd, 6);
+
+	/* Verify we can set a specific remote fd */
+	addfd.newfd = 42;
+	addfd.flags = SECCOMP_ADDFD_FLAG_SETFD;
+	fd = ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD, &addfd);
+	EXPECT_EQ(fd, 42);
+	EXPECT_EQ(filecmp(getpid(), pid, memfd, fd), 0);
+
+	/* Resume syscall */
+	resp.id = req.id;
+	resp.error = 0;
+	resp.val = USER_NOTIF_MAGIC;
+	EXPECT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_SEND, &resp), 0);
+
+	/*
+	 * This sets the ID of the ADD FD to the last request plus 1. The
+	 * notification ID increments 1 per notification.
+	 */
+	addfd.id = req.id + 1;
+
+	/* This spins until the underlying notification is generated */
+	while (ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD, &addfd) != -1 &&
+	       errno != -EINPROGRESS)
+		nanosleep(&delay, NULL);
+
+	memset(&req, 0, sizeof(req));
+	ASSERT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_RECV, &req), 0);
+	ASSERT_EQ(addfd.id, req.id);
+
+	resp.id = req.id;
+	resp.error = 0;
+	resp.val = USER_NOTIF_MAGIC;
+	EXPECT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_SEND, &resp), 0);
+
+	/* Wait for child to finish. */
+	EXPECT_EQ(waitpid(pid, &status, 0), pid);
+	EXPECT_EQ(true, WIFEXITED(status));
+	EXPECT_EQ(0, WEXITSTATUS(status));
+
+	close(memfd);
+}
+
+TEST(user_notification_addfd_rlimit)
+{
+	pid_t pid;
+	long ret;
+	int status, listener, memfd;
+	struct seccomp_notif_addfd addfd = {};
+	struct seccomp_notif req = {};
+	struct seccomp_notif_resp resp = {};
+	const struct rlimit lim = {
+		.rlim_cur	= 0,
+		.rlim_max	= 0,
+	};
+
+	memfd = memfd_create("test", 0);
+	ASSERT_GE(memfd, 0);
+
+	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
+	ASSERT_EQ(0, ret) {
+		TH_LOG("Kernel does not support PR_SET_NO_NEW_PRIVS!");
+	}
+
+	/* Check that the basic notification machinery works */
+	listener = user_notif_syscall(__NR_getppid,
+				      SECCOMP_FILTER_FLAG_NEW_LISTENER);
+	ASSERT_GE(listener, 0);
+
+	pid = fork();
+	ASSERT_GE(pid, 0);
+
+	if (pid == 0)
+		exit(syscall(__NR_getppid) != USER_NOTIF_MAGIC);
+
+
+	ASSERT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_RECV, &req), 0);
+
+	ASSERT_EQ(prlimit(pid, RLIMIT_NOFILE, &lim, NULL), 0);
+
+	addfd.srcfd = memfd;
+	addfd.newfd_flags = O_CLOEXEC;
+	addfd.newfd = 0;
+	addfd.id = req.id;
+	addfd.flags = 0;
+
+	/* Should probably spot check /proc/sys/fs/file-nr */
+	EXPECT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD, &addfd), -1);
+	EXPECT_EQ(errno, EMFILE);
+
+	addfd.newfd = 100;
+	addfd.flags = SECCOMP_ADDFD_FLAG_SETFD;
+	EXPECT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD, &addfd), -1);
+	EXPECT_EQ(errno, EBADF);
+
+	resp.id = req.id;
+	resp.error = 0;
+	resp.val = USER_NOTIF_MAGIC;
+
+	EXPECT_EQ(ioctl(listener, SECCOMP_IOCTL_NOTIF_SEND, &resp), 0);
+
+	/* Wait for child to finish. */
+	EXPECT_EQ(waitpid(pid, &status, 0), pid);
+	EXPECT_EQ(true, WIFEXITED(status));
+	EXPECT_EQ(0, WEXITSTATUS(status));
+
+	close(memfd);
+}
+
 /*
  * TODO:
- * - add microbenchmarks
  * - expand NNP testing
  * - better arch-specific TRACE and TRAP handlers.
  * - endianness checking when appropriate
@@ -3697,7 +4000,6 @@ TEST(user_notification_continue)
  * - arch value testing (x86 modes especially)
  * - verify that FILTER_FLAG_LOG filters generate log messages
  * - verify that RET_LOG generates log messages
- * - ...
  */
 
 TEST_HARNESS_MAIN

tools/testing/selftests/seccomp/settings

+timeout=90