signal/timer/event: signalfd core

This patch series implements the new signalfd() system call.

I took part of the original Linus code (and you know how badly it can be
broken :), and I added even more breakage ;) Signals are fetched from the same
signal queue used by the process, so signalfd will compete with standard
kernel delivery in dequeue_signal().  If you want to reliably fetch signals on
the signalfd file, you need to block them with sigprocmask(SIG_BLOCK).  This
seems to be working fine on my Dual Opteron machine.  I made a quick test
program for it:

http://www.xmailserver.org/signafd-test.c

The signalfd() system call implements signal delivery into a file descriptor
receiver.  The signalfd file descriptor if created with the following API:

int signalfd(int ufd, const sigset_t *mask, size_t masksize);

The "ufd" parameter allows to change an existing signalfd sigmask, w/out going
to close/create cycle (Linus idea).  Use "ufd" == -1 if you want a brand new
signalfd file.

The "mask" allows to specify the signal mask of signals that we are interested
in.  The "masksize" parameter is the size of "mask".

The signalfd fd supports the poll(2) and read(2) system calls.  The poll(2)
will return POLLIN when signals are available to be dequeued.  As a direct
consequence of supporting the Linux poll subsystem, the signalfd fd can use
used together with epoll(2) too.

The read(2) system call will return a "struct signalfd_siginfo" structure in
the userspace supplied buffer.  The return value is the number of bytes copied
in the supplied buffer, or -1 in case of error.  The read(2) call can also
return 0, in case the sighand structure to which the signalfd was attached,
has been orphaned.  The O_NONBLOCK flag is also supported, and read(2) will
return -EAGAIN in case no signal is available.

If the size of the buffer passed to read(2) is lower than sizeof(struct
signalfd_siginfo), -EINVAL is returned.  A read from the signalfd can also
return -ERESTARTSYS in case a signal hits the process.  The format of the
struct signalfd_siginfo is, and the valid fields depends of the (->code &
__SI_MASK) value, in the same way a struct siginfo would:

struct signalfd_siginfo {
	__u32 signo;	/* si_signo */
	__s32 err;	/* si_errno */
	__s32 code;	/* si_code */
	__u32 pid;	/* si_pid */
	__u32 uid;	/* si_uid */
	__s32 fd;	/* si_fd */
	__u32 tid;	/* si_fd */
	__u32 band;	/* si_band */
	__u32 overrun;	/* si_overrun */
	__u32 trapno;	/* si_trapno */
	__s32 status;	/* si_status */
	__s32 svint;	/* si_int */
	__u64 svptr;	/* si_ptr */
	__u64 utime;	/* si_utime */
	__u64 stime;	/* si_stime */
	__u64 addr;	/* si_addr */
};

[akpm@linux-foundation.org: fix signalfd_copyinfo() on i386]
Signed-off-by: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

3 files changed, 359 insertions, 2 deletions

diff --git a/fs/Makefile b/fs/Makefile
index b5cd46a88cb..cd8a57aeac0 100644
--- a/fs/Makefile
+++ b/fs/Makefile
@@ -23,6 +23,7 @@ obj-$(CONFIG_INOTIFY)		+= inotify.o
 obj-$(CONFIG_INOTIFY_USER)	+= inotify_user.o
 obj-$(CONFIG_EPOLL)		+= eventpoll.o
 obj-$(CONFIG_ANON_INODES)	+= anon_inodes.o
+obj-$(CONFIG_SIGNALFD)		+= signalfd.o
 obj-$(CONFIG_COMPAT)		+= compat.o compat_ioctl.o
 
 nfsd-$(CONFIG_NFSD)		:= nfsctl.o
diff --git a/fs/exec.c b/fs/exec.c
index 2255dc72dee..955a8eb66d7 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -50,6 +50,7 @@
 #include <linux/tsacct_kern.h>
 #include <linux/cn_proc.h>
 #include <linux/audit.h>
+#include <linux/signalfd.h>
 
 #include <asm/uaccess.h>
 #include <asm/mmu_context.h>
@@ -582,6 +583,13 @@ static int de_thread(struct task_struct *tsk)
 	int count;
 
 	/*
+	 * Tell all the sighand listeners that this sighand has
+	 * been detached. The signalfd_detach() function grabs the
+	 * sighand lock, if signal listeners are present on the sighand.
+	 */
+	signalfd_detach(tsk);
+
+	/*
 	 * If we don't share sighandlers, then we aren't sharing anything
 	 * and we can just re-use it all.
 	 */
@@ -757,8 +765,7 @@ no_thread_group:
 		spin_unlock(&oldsighand->siglock);
 		write_unlock_irq(&tasklist_lock);
 
-		if (atomic_dec_and_test(&oldsighand->count))
-			kmem_cache_free(sighand_cachep, oldsighand);
+		__cleanup_sighand(oldsighand);
 	}
 
 	BUG_ON(!thread_group_leader(tsk));
diff --git a/fs/signalfd.c b/fs/signalfd.c
new file mode 100644
index 00000000000..7cfeab412b4
--- /dev/null
+++ b/fs/signalfd.c
@@ -0,0 +1,349 @@
+/*
+ *  fs/signalfd.c
+ *
+ *  Copyright (C) 2003  Linus Torvalds
+ *
+ *  Mon Mar 5, 2007: Davide Libenzi <davidel@xmailserver.org>
+ *      Changed ->read() to return a siginfo strcture instead of signal number.
+ *      Fixed locking in ->poll().
+ *      Added sighand-detach notification.
+ *      Added fd re-use in sys_signalfd() syscall.
+ *      Now using anonymous inode source.
+ *      Thanks to Oleg Nesterov for useful code review and suggestions.
+ *      More comments and suggestions from Arnd Bergmann.
+ */
+
+#include <linux/file.h>
+#include <linux/poll.h>
+#include <linux/init.h>
+#include <linux/fs.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/signal.h>
+#include <linux/list.h>
+#include <linux/anon_inodes.h>
+#include <linux/signalfd.h>
+
+struct signalfd_ctx {
+	struct list_head lnk;
+	wait_queue_head_t wqh;
+	sigset_t sigmask;
+	struct task_struct *tsk;
+};
+
+struct signalfd_lockctx {
+	struct task_struct *tsk;
+	unsigned long flags;
+};
+
+/*
+ * Tries to acquire the sighand lock. We do not increment the sighand
+ * use count, and we do not even pin the task struct, so we need to
+ * do it inside an RCU read lock, and we must be prepared for the
+ * ctx->tsk going to NULL (in signalfd_deliver()), and for the sighand
+ * being detached. We return 0 if the sighand has been detached, or
+ * 1 if we were able to pin the sighand lock.
+ */
+static int signalfd_lock(struct signalfd_ctx *ctx, struct signalfd_lockctx *lk)
+{
+	struct sighand_struct *sighand = NULL;
+
+	rcu_read_lock();
+	lk->tsk = rcu_dereference(ctx->tsk);
+	if (likely(lk->tsk != NULL))
+		sighand = lock_task_sighand(lk->tsk, &lk->flags);
+	rcu_read_unlock();
+
+	if (sighand && !ctx->tsk) {
+		unlock_task_sighand(lk->tsk, &lk->flags);
+		sighand = NULL;
+	}
+
+	return sighand != NULL;
+}
+
+static void signalfd_unlock(struct signalfd_lockctx *lk)
+{
+	unlock_task_sighand(lk->tsk, &lk->flags);
+}
+
+/*
+ * This must be called with the sighand lock held.
+ */
+void signalfd_deliver(struct task_struct *tsk, int sig)
+{
+	struct sighand_struct *sighand = tsk->sighand;
+	struct signalfd_ctx *ctx, *tmp;
+
+	BUG_ON(!sig);
+	list_for_each_entry_safe(ctx, tmp, &sighand->signalfd_list, lnk) {
+		/*
+		 * We use a negative signal value as a way to broadcast that the
+		 * sighand has been orphaned, so that we can notify all the
+		 * listeners about this. Remember the ctx->sigmask is inverted,
+		 * so if the user is interested in a signal, that corresponding
+		 * bit will be zero.
+		 */
+		if (sig < 0) {
+			if (ctx->tsk == tsk) {
+				ctx->tsk = NULL;
+				list_del_init(&ctx->lnk);
+				wake_up(&ctx->wqh);
+			}
+		} else {
+			if (!sigismember(&ctx->sigmask, sig))
+				wake_up(&ctx->wqh);
+		}
+	}
+}
+
+static void signalfd_cleanup(struct signalfd_ctx *ctx)
+{
+	struct signalfd_lockctx lk;
+
+	/*
+	 * This is tricky. If the sighand is gone, we do not need to remove
+	 * context from the list, the list itself won't be there anymore.
+	 */
+	if (signalfd_lock(ctx, &lk)) {
+		list_del(&ctx->lnk);
+		signalfd_unlock(&lk);
+	}
+	kfree(ctx);
+}
+
+static int signalfd_release(struct inode *inode, struct file *file)
+{
+	signalfd_cleanup(file->private_data);
+	return 0;
+}
+
+static unsigned int signalfd_poll(struct file *file, poll_table *wait)
+{
+	struct signalfd_ctx *ctx = file->private_data;
+	unsigned int events = 0;
+	struct signalfd_lockctx lk;
+
+	poll_wait(file, &ctx->wqh, wait);
+
+	/*
+	 * Let the caller get a POLLIN in this case, ala socket recv() when
+	 * the peer disconnects.
+	 */
+	if (signalfd_lock(ctx, &lk)) {
+		if (next_signal(&lk.tsk->pending, &ctx->sigmask) > 0 ||
+		    next_signal(&lk.tsk->signal->shared_pending,
+				&ctx->sigmask) > 0)
+			events |= POLLIN;
+		signalfd_unlock(&lk);
+	} else
+		events |= POLLIN;
+
+	return events;
+}
+
+/*
+ * Copied from copy_siginfo_to_user() in kernel/signal.c
+ */
+static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
+			     siginfo_t const *kinfo)
+{
+	long err;
+
+	BUILD_BUG_ON(sizeof(struct signalfd_siginfo) != 128);
+
+	/*
+	 * Unused memebers should be zero ...
+	 */
+	err = __clear_user(uinfo, sizeof(*uinfo));
+
+	/*
+	 * If you change siginfo_t structure, please be sure
+	 * this code is fixed accordingly.
+	 */
+	err |= __put_user(kinfo->si_signo, &uinfo->signo);
+	err |= __put_user(kinfo->si_errno, &uinfo->err);
+	err |= __put_user((short)kinfo->si_code, &uinfo->code);
+	switch (kinfo->si_code & __SI_MASK) {
+	case __SI_KILL:
+		err |= __put_user(kinfo->si_pid, &uinfo->pid);
+		err |= __put_user(kinfo->si_uid, &uinfo->uid);
+		break;
+	case __SI_TIMER:
+		 err |= __put_user(kinfo->si_tid, &uinfo->tid);
+		 err |= __put_user(kinfo->si_overrun, &uinfo->overrun);
+		 err |= __put_user((long)kinfo->si_ptr, &uinfo->svptr);
+		break;
+	case __SI_POLL:
+		err |= __put_user(kinfo->si_band, &uinfo->band);
+		err |= __put_user(kinfo->si_fd, &uinfo->fd);
+		break;
+	case __SI_FAULT:
+		err |= __put_user((long)kinfo->si_addr, &uinfo->addr);
+#ifdef __ARCH_SI_TRAPNO
+		err |= __put_user(kinfo->si_trapno, &uinfo->trapno);
+#endif
+		break;
+	case __SI_CHLD:
+		err |= __put_user(kinfo->si_pid, &uinfo->pid);
+		err |= __put_user(kinfo->si_uid, &uinfo->uid);
+		err |= __put_user(kinfo->si_status, &uinfo->status);
+		err |= __put_user(kinfo->si_utime, &uinfo->utime);
+		err |= __put_user(kinfo->si_stime, &uinfo->stime);
+		break;
+	case __SI_RT: /* This is not generated by the kernel as of now. */
+	case __SI_MESGQ: /* But this is */
+		err |= __put_user(kinfo->si_pid, &uinfo->pid);
+		err |= __put_user(kinfo->si_uid, &uinfo->uid);
+		err |= __put_user((long)kinfo->si_ptr, &uinfo->svptr);
+		break;
+	default: /* this is just in case for now ... */
+		err |= __put_user(kinfo->si_pid, &uinfo->pid);
+		err |= __put_user(kinfo->si_uid, &uinfo->uid);
+		break;
+	}
+
+	return err ? -EFAULT: sizeof(*uinfo);
+}
+
+/*
+ * Returns either the size of a "struct signalfd_siginfo", or zero if the
+ * sighand we are attached to, has been orphaned. The "count" parameter
+ * must be at least the size of a "struct signalfd_siginfo".
+ */
+static ssize_t signalfd_read(struct file *file, char __user *buf, size_t count,
+			     loff_t *ppos)
+{
+	struct signalfd_ctx *ctx = file->private_data;
+	ssize_t res = 0;
+	int locked, signo;
+	siginfo_t info;
+	struct signalfd_lockctx lk;
+	DECLARE_WAITQUEUE(wait, current);
+
+	if (count < sizeof(struct signalfd_siginfo))
+		return -EINVAL;
+	locked = signalfd_lock(ctx, &lk);
+	if (!locked)
+		return 0;
+	res = -EAGAIN;
+	signo = dequeue_signal(lk.tsk, &ctx->sigmask, &info);
+	if (signo == 0 && !(file->f_flags & O_NONBLOCK)) {
+		add_wait_queue(&ctx->wqh, &wait);
+		for (;;) {
+			set_current_state(TASK_INTERRUPTIBLE);
+			signo = dequeue_signal(lk.tsk, &ctx->sigmask, &info);
+			if (signo != 0)
+				break;
+			if (signal_pending(current)) {
+				res = -ERESTARTSYS;
+				break;
+			}
+			signalfd_unlock(&lk);
+			schedule();
+			locked = signalfd_lock(ctx, &lk);
+			if (unlikely(!locked)) {
+				/*
+				 * Let the caller read zero byte, ala socket
+				 * recv() when the peer disconnect. This test
+				 * must be done before doing a dequeue_signal(),
+				 * because if the sighand has been orphaned,
+				 * the dequeue_signal() call is going to crash.
+				 */
+				res = 0;
+				break;
+			}
+		}
+		remove_wait_queue(&ctx->wqh, &wait);
+		__set_current_state(TASK_RUNNING);
+	}
+	if (likely(locked))
+		signalfd_unlock(&lk);
+	if (likely(signo))
+		res = signalfd_copyinfo((struct signalfd_siginfo __user *) buf,
+					&info);
+
+	return res;
+}
+
+static const struct file_operations signalfd_fops = {
+	.release	= signalfd_release,
+	.poll		= signalfd_poll,
+	.read		= signalfd_read,
+};
+
+/*
+ * Create a file descriptor that is associated with our signal
+ * state. We can pass it around to others if we want to, but
+ * it will always be _our_ signal state.
+ */
+asmlinkage long sys_signalfd(int ufd, sigset_t __user *user_mask, size_t sizemask)
+{
+	int error;
+	sigset_t sigmask;
+	struct signalfd_ctx *ctx;
+	struct sighand_struct *sighand;
+	struct file *file;
+	struct inode *inode;
+	struct signalfd_lockctx lk;
+
+	if (sizemask != sizeof(sigset_t) ||
+	    copy_from_user(&sigmask, user_mask, sizeof(sigmask)))
+		return error = -EINVAL;
+	sigdelsetmask(&sigmask, sigmask(SIGKILL) | sigmask(SIGSTOP));
+	signotset(&sigmask);
+
+	if (ufd == -1) {
+		ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
+		if (!ctx)
+			return -ENOMEM;
+
+		init_waitqueue_head(&ctx->wqh);
+		ctx->sigmask = sigmask;
+		ctx->tsk = current;
+
+		sighand = current->sighand;
+		/*
+		 * Add this fd to the list of signal listeners.
+		 */
+		spin_lock_irq(&sighand->siglock);
+		list_add_tail(&ctx->lnk, &sighand->signalfd_list);
+		spin_unlock_irq(&sighand->siglock);
+
+		/*
+		 * When we call this, the initialization must be complete, since
+		 * anon_inode_getfd() will install the fd.
+		 */
+		error = anon_inode_getfd(&ufd, &inode, &file, "[signalfd]",
+					 &signalfd_fops, ctx);
+		if (error)
+			goto err_fdalloc;
+	} else {
+		file = fget(ufd);
+		if (!file)
+			return -EBADF;
+		ctx = file->private_data;
+		if (file->f_op != &signalfd_fops) {
+			fput(file);
+			return -EINVAL;
+		}
+		/*
+		 * We need to be prepared of the fact that the sighand this fd
+		 * is attached to, has been detched. In that case signalfd_lock()
+		 * will return 0, and we'll just skip setting the new mask.
+		 */
+		if (signalfd_lock(ctx, &lk)) {
+			ctx->sigmask = sigmask;
+			signalfd_unlock(&lk);
+		}
+		wake_up(&ctx->wqh);
+		fput(file);
+	}
+
+	return ufd;
+
+err_fdalloc:
+	signalfd_cleanup(ctx);
+	return error;
+}
+