signalfd.c 9.16 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13
/*
 *  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.
Davide Libenzi's avatar
Davide Libenzi committed
14
 *  Sat May 19, 2007: Davi E. M. Arnaut <davi@haxent.com.br>
15
 *      Retrieve multiple signals with one read() call
Davide Libenzi's avatar
Davide Libenzi committed
16 17
 *  Sun Jul 15, 2007: Davide Libenzi <davidel@xmailserver.org>
 *      Attach to the sighand only during read() and poll().
18 19 20 21 22 23 24
 */

#include <linux/file.h>
#include <linux/poll.h>
#include <linux/init.h>
#include <linux/fs.h>
#include <linux/sched.h>
25
#include <linux/slab.h>
26 27 28 29 30
#include <linux/kernel.h>
#include <linux/signal.h>
#include <linux/list.h>
#include <linux/anon_inodes.h>
#include <linux/signalfd.h>
31
#include <linux/syscalls.h>
32
#include <linux/proc_fs.h>
33
#include <linux/compat.h>
34

35 36 37
void signalfd_cleanup(struct sighand_struct *sighand)
{
	wait_queue_head_t *wqh = &sighand->signalfd_wqh;
38 39 40 41 42
	/*
	 * The lockless check can race with remove_wait_queue() in progress,
	 * but in this case its caller should run under rcu_read_lock() and
	 * sighand_cachep is SLAB_DESTROY_BY_RCU, we can safely return.
	 */
43 44 45 46 47 48 49
	if (likely(!waitqueue_active(wqh)))
		return;

	/* wait_queue_t->func(POLLFREE) should do remove_wait_queue() */
	wake_up_poll(wqh, POLLHUP | POLLFREE);
}

50 51 52 53 54 55
struct signalfd_ctx {
	sigset_t sigmask;
};

static int signalfd_release(struct inode *inode, struct file *file)
{
Davide Libenzi's avatar
Davide Libenzi committed
56
	kfree(file->private_data);
57 58 59 60 61 62 63 64
	return 0;
}

static unsigned int signalfd_poll(struct file *file, poll_table *wait)
{
	struct signalfd_ctx *ctx = file->private_data;
	unsigned int events = 0;

Davide Libenzi's avatar
Davide Libenzi committed
65
	poll_wait(file, &current->sighand->signalfd_wqh, wait);
66

Davide Libenzi's avatar
Davide Libenzi committed
67 68 69 70
	spin_lock_irq(&current->sighand->siglock);
	if (next_signal(&current->pending, &ctx->sigmask) ||
	    next_signal(&current->signal->shared_pending,
			&ctx->sigmask))
71
		events |= POLLIN;
Davide Libenzi's avatar
Davide Libenzi committed
72
	spin_unlock_irq(&current->sighand->siglock);
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87

	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);

	/*
88
	 * Unused members should be zero ...
89 90 91 92 93 94 95
	 */
	err = __clear_user(uinfo, sizeof(*uinfo));

	/*
	 * If you change siginfo_t structure, please be sure
	 * this code is fixed accordingly.
	 */
96 97 98
	err |= __put_user(kinfo->si_signo, &uinfo->ssi_signo);
	err |= __put_user(kinfo->si_errno, &uinfo->ssi_errno);
	err |= __put_user((short) kinfo->si_code, &uinfo->ssi_code);
99 100
	switch (kinfo->si_code & __SI_MASK) {
	case __SI_KILL:
101 102
		err |= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
		err |= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
103 104
		break;
	case __SI_TIMER:
105 106 107
		 err |= __put_user(kinfo->si_tid, &uinfo->ssi_tid);
		 err |= __put_user(kinfo->si_overrun, &uinfo->ssi_overrun);
		 err |= __put_user((long) kinfo->si_ptr, &uinfo->ssi_ptr);
108
		 err |= __put_user(kinfo->si_int, &uinfo->ssi_int);
109 110
		break;
	case __SI_POLL:
111 112
		err |= __put_user(kinfo->si_band, &uinfo->ssi_band);
		err |= __put_user(kinfo->si_fd, &uinfo->ssi_fd);
113 114
		break;
	case __SI_FAULT:
115
		err |= __put_user((long) kinfo->si_addr, &uinfo->ssi_addr);
116
#ifdef __ARCH_SI_TRAPNO
117
		err |= __put_user(kinfo->si_trapno, &uinfo->ssi_trapno);
118 119 120 121 122 123
#endif
#ifdef BUS_MCEERR_AO
		/* 
		 * Other callers might not initialize the si_lsb field,
		 * so check explicitly for the right codes here.
		 */
124 125 126
		if (kinfo->si_signo == SIGBUS &&
		    (kinfo->si_code == BUS_MCEERR_AR ||
		     kinfo->si_code == BUS_MCEERR_AO))
127 128
			err |= __put_user((short) kinfo->si_addr_lsb,
					  &uinfo->ssi_addr_lsb);
129 130 131
#endif
		break;
	case __SI_CHLD:
132 133 134 135 136
		err |= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
		err |= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
		err |= __put_user(kinfo->si_status, &uinfo->ssi_status);
		err |= __put_user(kinfo->si_utime, &uinfo->ssi_utime);
		err |= __put_user(kinfo->si_stime, &uinfo->ssi_stime);
137 138 139
		break;
	case __SI_RT: /* This is not generated by the kernel as of now. */
	case __SI_MESGQ: /* But this is */
140 141 142
		err |= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
		err |= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
		err |= __put_user((long) kinfo->si_ptr, &uinfo->ssi_ptr);
143
		err |= __put_user(kinfo->si_int, &uinfo->ssi_int);
144
		break;
145 146 147 148
	default:
		/*
		 * This case catches also the signals queued by sigqueue().
		 */
149 150
		err |= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
		err |= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
151 152
		err |= __put_user((long) kinfo->si_ptr, &uinfo->ssi_ptr);
		err |= __put_user(kinfo->si_int, &uinfo->ssi_int);
153 154 155 156 157 158
		break;
	}

	return err ? -EFAULT: sizeof(*uinfo);
}

159 160 161 162 163 164
static ssize_t signalfd_dequeue(struct signalfd_ctx *ctx, siginfo_t *info,
				int nonblock)
{
	ssize_t ret;
	DECLARE_WAITQUEUE(wait, current);

Davide Libenzi's avatar
Davide Libenzi committed
165 166
	spin_lock_irq(&current->sighand->siglock);
	ret = dequeue_signal(current, &ctx->sigmask, info);
167 168 169 170 171 172
	switch (ret) {
	case 0:
		if (!nonblock)
			break;
		ret = -EAGAIN;
	default:
Davide Libenzi's avatar
Davide Libenzi committed
173
		spin_unlock_irq(&current->sighand->siglock);
174 175 176
		return ret;
	}

Davide Libenzi's avatar
Davide Libenzi committed
177
	add_wait_queue(&current->sighand->signalfd_wqh, &wait);
178 179
	for (;;) {
		set_current_state(TASK_INTERRUPTIBLE);
Davide Libenzi's avatar
Davide Libenzi committed
180
		ret = dequeue_signal(current, &ctx->sigmask, info);
181 182 183 184 185 186
		if (ret != 0)
			break;
		if (signal_pending(current)) {
			ret = -ERESTARTSYS;
			break;
		}
Davide Libenzi's avatar
Davide Libenzi committed
187
		spin_unlock_irq(&current->sighand->siglock);
188
		schedule();
Davide Libenzi's avatar
Davide Libenzi committed
189
		spin_lock_irq(&current->sighand->siglock);
190
	}
Davide Libenzi's avatar
Davide Libenzi committed
191
	spin_unlock_irq(&current->sighand->siglock);
192

Davide Libenzi's avatar
Davide Libenzi committed
193
	remove_wait_queue(&current->sighand->signalfd_wqh, &wait);
194 195 196 197 198
	__set_current_state(TASK_RUNNING);

	return ret;
}

199
/*
Davide Libenzi's avatar
Davide Libenzi committed
200 201 202
 * Returns a multiple of the size of a "struct signalfd_siginfo", or a negative
 * error code. The "count" parameter must be at least the size of a
 * "struct signalfd_siginfo".
203 204 205 206 207
 */
static ssize_t signalfd_read(struct file *file, char __user *buf, size_t count,
			     loff_t *ppos)
{
	struct signalfd_ctx *ctx = file->private_data;
208 209 210
	struct signalfd_siginfo __user *siginfo;
	int nonblock = file->f_flags & O_NONBLOCK;
	ssize_t ret, total = 0;
211 212
	siginfo_t info;

213 214
	count /= sizeof(struct signalfd_siginfo);
	if (!count)
215 216
		return -EINVAL;

217 218 219 220 221 222 223 224 225 226 227 228 229
	siginfo = (struct signalfd_siginfo __user *) buf;
	do {
		ret = signalfd_dequeue(ctx, &info, nonblock);
		if (unlikely(ret <= 0))
			break;
		ret = signalfd_copyinfo(siginfo, &info);
		if (ret < 0)
			break;
		siginfo++;
		total += ret;
		nonblock = 1;
	} while (--count);

Davide Libenzi's avatar
Davide Libenzi committed
230
	return total ? total: ret;
231 232
}

233
#ifdef CONFIG_PROC_FS
234
static void signalfd_show_fdinfo(struct seq_file *m, struct file *f)
235 236 237 238 239 240 241 242 243 244
{
	struct signalfd_ctx *ctx = f->private_data;
	sigset_t sigmask;

	sigmask = ctx->sigmask;
	signotset(&sigmask);
	render_sigset_t(m, "sigmask:\t", &sigmask);
}
#endif

245
static const struct file_operations signalfd_fops = {
246 247 248
#ifdef CONFIG_PROC_FS
	.show_fdinfo	= signalfd_show_fdinfo,
#endif
249 250 251
	.release	= signalfd_release,
	.poll		= signalfd_poll,
	.read		= signalfd_read,
252
	.llseek		= noop_llseek,
253 254
};

255 256
SYSCALL_DEFINE4(signalfd4, int, ufd, sigset_t __user *, user_mask,
		size_t, sizemask, int, flags)
257 258 259 260
{
	sigset_t sigmask;
	struct signalfd_ctx *ctx;

261 262 263 264
	/* Check the SFD_* constants for consistency.  */
	BUILD_BUG_ON(SFD_CLOEXEC != O_CLOEXEC);
	BUILD_BUG_ON(SFD_NONBLOCK != O_NONBLOCK);

265
	if (flags & ~(SFD_CLOEXEC | SFD_NONBLOCK))
266 267
		return -EINVAL;

268 269
	if (sizemask != sizeof(sigset_t) ||
	    copy_from_user(&sigmask, user_mask, sizeof(sigmask)))
Ulrich Drepper's avatar
Ulrich Drepper committed
270
		return -EINVAL;
271 272 273 274 275 276 277 278 279 280 281 282 283 284
	sigdelsetmask(&sigmask, sigmask(SIGKILL) | sigmask(SIGSTOP));
	signotset(&sigmask);

	if (ufd == -1) {
		ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
		if (!ctx)
			return -ENOMEM;

		ctx->sigmask = sigmask;

		/*
		 * When we call this, the initialization must be complete, since
		 * anon_inode_getfd() will install the fd.
		 */
285
		ufd = anon_inode_getfd("[signalfd]", &signalfd_fops, ctx,
286
				       O_RDWR | (flags & (O_CLOEXEC | O_NONBLOCK)));
287 288
		if (ufd < 0)
			kfree(ctx);
289
	} else {
290 291
		struct fd f = fdget(ufd);
		if (!f.file)
292
			return -EBADF;
293 294 295
		ctx = f.file->private_data;
		if (f.file->f_op != &signalfd_fops) {
			fdput(f);
296 297
			return -EINVAL;
		}
Davide Libenzi's avatar
Davide Libenzi committed
298 299 300 301 302
		spin_lock_irq(&current->sighand->siglock);
		ctx->sigmask = sigmask;
		spin_unlock_irq(&current->sighand->siglock);

		wake_up(&current->sighand->signalfd_wqh);
303
		fdput(f);
304 305 306 307
	}

	return ufd;
}
308

309 310
SYSCALL_DEFINE3(signalfd, int, ufd, sigset_t __user *, user_mask,
		size_t, sizemask)
311 312 313
{
	return sys_signalfd4(ufd, user_mask, sizemask, 0);
}
314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343

#ifdef CONFIG_COMPAT
COMPAT_SYSCALL_DEFINE4(signalfd4, int, ufd,
		     const compat_sigset_t __user *,sigmask,
		     compat_size_t, sigsetsize,
		     int, flags)
{
	compat_sigset_t ss32;
	sigset_t tmp;
	sigset_t __user *ksigmask;

	if (sigsetsize != sizeof(compat_sigset_t))
		return -EINVAL;
	if (copy_from_user(&ss32, sigmask, sizeof(ss32)))
		return -EFAULT;
	sigset_from_compat(&tmp, &ss32);
	ksigmask = compat_alloc_user_space(sizeof(sigset_t));
	if (copy_to_user(ksigmask, &tmp, sizeof(sigset_t)))
		return -EFAULT;

	return sys_signalfd4(ufd, ksigmask, sizeof(sigset_t), flags);
}

COMPAT_SYSCALL_DEFINE3(signalfd, int, ufd,
		     const compat_sigset_t __user *,sigmask,
		     compat_size_t, sigsetsize)
{
	return compat_sys_signalfd4(ufd, sigmask, sigsetsize, 0);
}
#endif