xref: /openbsd/sys/kern/sys_pipe.c (revision b9ae17a0)

/*	$OpenBSD: sys_pipe.c,v 1.148 2024/12/30 02:46:00 guenther Exp $	*/

/*
 * Copyright (c) 1996 John S. Dyson
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice immediately at the beginning of the file, without modification,
 *    this list of conditions, and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Absolutely no warranty of function or purpose is made by the author
 *    John S. Dyson.
 * 4. Modifications may be freely made to this file if the above conditions
 *    are met.
 */

/*
 * This file contains a high-performance replacement for the socket-based
 * pipes scheme originally used in FreeBSD/4.4Lite.  It does not support
 * all features of sockets, but does do everything that pipes normally
 * do.
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/fcntl.h>
#include <sys/file.h>
#include <sys/filedesc.h>
#include <sys/pool.h>
#include <sys/ioctl.h>
#include <sys/stat.h>
#include <sys/signalvar.h>
#include <sys/mount.h>
#include <sys/syscallargs.h>
#include <sys/event.h>
#ifdef KTRACE
#include <sys/ktrace.h>
#endif

#include <uvm/uvm_extern.h>

#include <sys/pipe.h>

struct pipe_pair {
	struct pipe pp_wpipe;
	struct pipe pp_rpipe;
	struct rwlock pp_lock;
};

/*
 * interfaces to the outside world
 */
int	pipe_read(struct file *, struct uio *, int);
int	pipe_write(struct file *, struct uio *, int);
int	pipe_close(struct file *, struct proc *);
int	pipe_kqfilter(struct file *fp, struct knote *kn);
int	pipe_ioctl(struct file *, u_long, caddr_t, struct proc *);
int	pipe_stat(struct file *fp, struct stat *ub, struct proc *p);

static const struct fileops pipeops = {
	.fo_read	= pipe_read,
	.fo_write	= pipe_write,
	.fo_ioctl	= pipe_ioctl,
	.fo_kqfilter	= pipe_kqfilter,
	.fo_stat	= pipe_stat,
	.fo_close	= pipe_close
};

void	filt_pipedetach(struct knote *kn);
int	filt_piperead(struct knote *kn, long hint);
int	filt_pipewrite(struct knote *kn, long hint);
int	filt_pipeexcept(struct knote *kn, long hint);
int	filt_pipemodify(struct kevent *kev, struct knote *kn);
int	filt_pipeprocess(struct knote *kn, struct kevent *kev);

const struct filterops pipe_rfiltops = {
	.f_flags	= FILTEROP_ISFD | FILTEROP_MPSAFE,
	.f_attach	= NULL,
	.f_detach	= filt_pipedetach,
	.f_event	= filt_piperead,
	.f_modify	= filt_pipemodify,
	.f_process	= filt_pipeprocess,
};

const struct filterops pipe_wfiltops = {
	.f_flags	= FILTEROP_ISFD | FILTEROP_MPSAFE,
	.f_attach	= NULL,
	.f_detach	= filt_pipedetach,
	.f_event	= filt_pipewrite,
	.f_modify	= filt_pipemodify,
	.f_process	= filt_pipeprocess,
};

const struct filterops pipe_efiltops = {
	.f_flags	= FILTEROP_ISFD | FILTEROP_MPSAFE,
	.f_attach	= NULL,
	.f_detach	= filt_pipedetach,
	.f_event	= filt_pipeexcept,
	.f_modify	= filt_pipemodify,
	.f_process	= filt_pipeprocess,
};

/*
 * Default pipe buffer size(s), this can be kind-of large now because pipe
 * space is pageable.  The pipe code will try to maintain locality of
 * reference for performance reasons, so small amounts of outstanding I/O
 * will not wipe the cache.
 */
#define MINPIPESIZE (PIPE_SIZE/3)

/*
 * Limit the number of "big" pipes
 */
#define LIMITBIGPIPES	32
unsigned int nbigpipe;
static unsigned int amountpipekva;

struct pool pipe_pair_pool;

int	dopipe(struct proc *, int *, int);
void	pipe_wakeup(struct pipe *);

int	pipe_create(struct pipe *);
void	pipe_destroy(struct pipe *);
int	pipe_rundown(struct pipe *);
struct pipe *pipe_peer(struct pipe *);
int	pipe_buffer_realloc(struct pipe *, u_int);
void	pipe_buffer_free(struct pipe *);

int	pipe_iolock(struct pipe *);
void	pipe_iounlock(struct pipe *);
int	pipe_iosleep(struct pipe *, const char *);

struct pipe_pair *pipe_pair_create(void);
void	pipe_pair_destroy(struct pipe_pair *);

/*
 * The pipe system call for the DTYPE_PIPE type of pipes
 */

int
sys_pipe(struct proc *p, void *v, register_t *retval)
{
	struct sys_pipe_args /* {
		syscallarg(int *) fdp;
	} */ *uap = v;

	return (dopipe(p, SCARG(uap, fdp), 0));
}

int
sys_pipe2(struct proc *p, void *v, register_t *retval)
{
	struct sys_pipe2_args /* {
		syscallarg(int *) fdp;
		syscallarg(int) flags;
	} */ *uap = v;

	if (SCARG(uap, flags) & ~(O_CLOEXEC | FNONBLOCK))
		return (EINVAL);

	return (dopipe(p, SCARG(uap, fdp), SCARG(uap, flags)));
}

int
dopipe(struct proc *p, int *ufds, int flags)
{
	struct filedesc *fdp = p->p_fd;
	struct file *rf, *wf;
	struct pipe_pair *pp;
	struct pipe *rpipe, *wpipe = NULL;
	int fds[2], cloexec, error;

	cloexec = (flags & O_CLOEXEC) ? UF_EXCLOSE : 0;

	pp = pipe_pair_create();
	if (pp == NULL)
		return (ENOMEM);
	wpipe = &pp->pp_wpipe;
	rpipe = &pp->pp_rpipe;

	fdplock(fdp);

	error = falloc(p, &rf, &fds[0]);
	if (error != 0)
		goto free2;
	rf->f_flag = FREAD | FWRITE | (flags & FNONBLOCK);
	rf->f_type = DTYPE_PIPE;
	rf->f_data = rpipe;
	rf->f_ops = &pipeops;

	error = falloc(p, &wf, &fds[1]);
	if (error != 0)
		goto free3;
	wf->f_flag = FREAD | FWRITE | (flags & FNONBLOCK);
	wf->f_type = DTYPE_PIPE;
	wf->f_data = wpipe;
	wf->f_ops = &pipeops;

	fdinsert(fdp, fds[0], cloexec, rf);
	fdinsert(fdp, fds[1], cloexec, wf);

	error = copyout(fds, ufds, sizeof(fds));
	if (error == 0) {
		fdpunlock(fdp);
#ifdef KTRACE
		if (KTRPOINT(p, KTR_STRUCT))
			ktrfds(p, fds, 2);
#endif
	} else {
		/* fdrelease() unlocks fdp. */
		fdrelease(p, fds[0]);
		fdplock(fdp);
		fdrelease(p, fds[1]);
	}

	FRELE(rf, p);
	FRELE(wf, p);
	return (error);

free3:
	fdremove(fdp, fds[0]);
	closef(rf, p);
	rpipe = NULL;
free2:
	fdpunlock(fdp);
	pipe_destroy(wpipe);
	pipe_destroy(rpipe);
	return (error);
}

/*
 * Allocate kva for pipe circular buffer, the space is pageable.
 * This routine will 'realloc' the size of a pipe safely, if it fails
 * it will retain the old buffer.
 * If it fails it will return ENOMEM.
 */
int
pipe_buffer_realloc(struct pipe *cpipe, u_int size)
{
	caddr_t buffer;

	/* buffer uninitialized or pipe locked */
	KASSERT((cpipe->pipe_buffer.buffer == NULL) ||
	    (cpipe->pipe_state & PIPE_LOCK));

	/* buffer should be empty */
	KASSERT(cpipe->pipe_buffer.cnt == 0);

	buffer = km_alloc(size, &kv_any, &kp_pageable, &kd_waitok);
	if (buffer == NULL)
		return (ENOMEM);

	/* free old resources if we are resizing */
	pipe_buffer_free(cpipe);

	cpipe->pipe_buffer.buffer = buffer;
	cpipe->pipe_buffer.size = size;
	cpipe->pipe_buffer.in = 0;
	cpipe->pipe_buffer.out = 0;

	atomic_add_int(&amountpipekva, cpipe->pipe_buffer.size);

	return (0);
}

/*
 * initialize and allocate VM and memory for pipe
 */
int
pipe_create(struct pipe *cpipe)
{
	int error;

	error = pipe_buffer_realloc(cpipe, PIPE_SIZE);
	if (error != 0)
		return (error);

	sigio_init(&cpipe->pipe_sigio);

	getnanotime(&cpipe->pipe_ctime);
	cpipe->pipe_atime = cpipe->pipe_ctime;
	cpipe->pipe_mtime = cpipe->pipe_ctime;

	return (0);
}

struct pipe *
pipe_peer(struct pipe *cpipe)
{
	struct pipe *peer;

	rw_assert_anylock(cpipe->pipe_lock);

	peer = cpipe->pipe_peer;
	if (peer == NULL || (peer->pipe_state & PIPE_EOF))
		return (NULL);
	return (peer);
}

/*
 * Lock a pipe for exclusive I/O access.
 */
int
pipe_iolock(struct pipe *cpipe)
{
	int error;

	rw_assert_wrlock(cpipe->pipe_lock);

	while (cpipe->pipe_state & PIPE_LOCK) {
		cpipe->pipe_state |= PIPE_LWANT;
		error = rwsleep_nsec(cpipe, cpipe->pipe_lock, PRIBIO | PCATCH,
		    "pipeiolk", INFSLP);
		if (error)
			return (error);
	}
	cpipe->pipe_state |= PIPE_LOCK;
	return (0);
}

/*
 * Unlock a pipe I/O lock.
 */
void
pipe_iounlock(struct pipe *cpipe)
{
	rw_assert_wrlock(cpipe->pipe_lock);
	KASSERT(cpipe->pipe_state & PIPE_LOCK);

	cpipe->pipe_state &= ~PIPE_LOCK;
	if (cpipe->pipe_state & PIPE_LWANT) {
		cpipe->pipe_state &= ~PIPE_LWANT;
		wakeup(cpipe);
	}
}

/*
 * Unlock the pipe I/O lock and go to sleep. Returns 0 on success and the I/O
 * lock is relocked. Otherwise if a signal was caught, non-zero is returned and
 * the I/O lock is not locked.
 *
 * Any caller must obtain a reference to the pipe by incrementing `pipe_busy'
 * before calling this function in order ensure that the same pipe is not
 * destroyed while sleeping.
 */
int
pipe_iosleep(struct pipe *cpipe, const char *wmesg)
{
	int error;

	pipe_iounlock(cpipe);
	error = rwsleep_nsec(cpipe, cpipe->pipe_lock, PRIBIO | PCATCH, wmesg,
	    INFSLP);
	if (error)
		return (error);
	return (pipe_iolock(cpipe));
}

void
pipe_wakeup(struct pipe *cpipe)
{
	rw_assert_wrlock(cpipe->pipe_lock);

	knote_locked(&cpipe->pipe_klist, 0);

	if (cpipe->pipe_state & PIPE_ASYNC)
		pgsigio(&cpipe->pipe_sigio, SIGIO, 0);
}

int
pipe_read(struct file *fp, struct uio *uio, int fflags)
{
	struct pipe *rpipe = fp->f_data;
	size_t nread = 0, size;
	int error;

	rw_enter_write(rpipe->pipe_lock);
	++rpipe->pipe_busy;
	error = pipe_iolock(rpipe);
	if (error) {
		--rpipe->pipe_busy;
		pipe_rundown(rpipe);
		rw_exit_write(rpipe->pipe_lock);
		return (error);
	}

	while (uio->uio_resid) {
		/* Normal pipe buffer receive. */
		if (rpipe->pipe_buffer.cnt > 0) {
			size = rpipe->pipe_buffer.size - rpipe->pipe_buffer.out;
			if (size > rpipe->pipe_buffer.cnt)
				size = rpipe->pipe_buffer.cnt;
			if (size > uio->uio_resid)
				size = uio->uio_resid;
			rw_exit_write(rpipe->pipe_lock);
			error = uiomove(&rpipe->pipe_buffer.buffer[rpipe->pipe_buffer.out],
					size, uio);
			rw_enter_write(rpipe->pipe_lock);
			if (error) {
				break;
			}
			rpipe->pipe_buffer.out += size;
			if (rpipe->pipe_buffer.out >= rpipe->pipe_buffer.size)
				rpipe->pipe_buffer.out = 0;

			rpipe->pipe_buffer.cnt -= size;
			/*
			 * If there is no more to read in the pipe, reset
			 * its pointers to the beginning.  This improves
			 * cache hit stats.
			 */
			if (rpipe->pipe_buffer.cnt == 0) {
				rpipe->pipe_buffer.in = 0;
				rpipe->pipe_buffer.out = 0;
			}
			nread += size;
		} else {
			/*
			 * detect EOF condition
			 * read returns 0 on EOF, no need to set error
			 */
			if (rpipe->pipe_state & PIPE_EOF)
				break;

			/* If the "write-side" has been blocked, wake it up. */
			if (rpipe->pipe_state & PIPE_WANTW) {
				rpipe->pipe_state &= ~PIPE_WANTW;
				wakeup(rpipe);
			}

			/* Break if some data was read. */
			if (nread > 0)
				break;

			/* Handle non-blocking mode operation. */
			if (fp->f_flag & FNONBLOCK) {
				error = EAGAIN;
				break;
			}

			/* Wait for more data. */
			rpipe->pipe_state |= PIPE_WANTR;
			error = pipe_iosleep(rpipe, "piperd");
			if (error)
				goto unlocked_error;
		}
	}
	pipe_iounlock(rpipe);

	if (error == 0)
		getnanotime(&rpipe->pipe_atime);
unlocked_error:
	--rpipe->pipe_busy;

	if (pipe_rundown(rpipe) == 0 && rpipe->pipe_buffer.cnt < MINPIPESIZE) {
		/* Handle write blocking hysteresis. */
		if (rpipe->pipe_state & PIPE_WANTW) {
			rpipe->pipe_state &= ~PIPE_WANTW;
			wakeup(rpipe);
		}
	}

	if (rpipe->pipe_buffer.size - rpipe->pipe_buffer.cnt >= PIPE_BUF)
		pipe_wakeup(rpipe);

	rw_exit_write(rpipe->pipe_lock);
	return (error);
}

int
pipe_write(struct file *fp, struct uio *uio, int fflags)
{
	struct pipe *rpipe = fp->f_data, *wpipe;
	struct rwlock *lock = rpipe->pipe_lock;
	size_t orig_resid;
	int error;

	rw_enter_write(lock);
	wpipe = pipe_peer(rpipe);

	/* Detect loss of pipe read side, issue SIGPIPE if lost. */
	if (wpipe == NULL) {
		rw_exit_write(lock);
		return (EPIPE);
	}

	++wpipe->pipe_busy;
	error = pipe_iolock(wpipe);
	if (error) {
		--wpipe->pipe_busy;
		pipe_rundown(wpipe);
		rw_exit_write(lock);
		return (error);
	}


	/* If it is advantageous to resize the pipe buffer, do so. */
	if (uio->uio_resid > PIPE_SIZE &&
	    wpipe->pipe_buffer.size <= PIPE_SIZE &&
	    wpipe->pipe_buffer.cnt == 0) {
	    	unsigned int npipe;

		npipe = atomic_inc_int_nv(&nbigpipe);
		if (npipe > LIMITBIGPIPES ||
		    pipe_buffer_realloc(wpipe, BIG_PIPE_SIZE) != 0)
			atomic_dec_int(&nbigpipe);
	}

	orig_resid = uio->uio_resid;

	while (uio->uio_resid) {
		size_t space;

		if (wpipe->pipe_state & PIPE_EOF) {
			error = EPIPE;
			break;
		}

		space = wpipe->pipe_buffer.size - wpipe->pipe_buffer.cnt;

		/* Writes of size <= PIPE_BUF must be atomic. */
		if (space < uio->uio_resid && orig_resid <= PIPE_BUF)
			space = 0;

		if (space > 0) {
			size_t size;	/* Transfer size */
			size_t segsize;	/* first segment to transfer */

			/*
			 * Transfer size is minimum of uio transfer
			 * and free space in pipe buffer.
			 */
			if (space > uio->uio_resid)
				size = uio->uio_resid;
			else
				size = space;
			/*
			 * First segment to transfer is minimum of
			 * transfer size and contiguous space in
			 * pipe buffer.  If first segment to transfer
			 * is less than the transfer size, we've got
			 * a wraparound in the buffer.
			 */
			segsize = wpipe->pipe_buffer.size -
				wpipe->pipe_buffer.in;
			if (segsize > size)
				segsize = size;

			/* Transfer first segment */

			rw_exit_write(lock);
			error = uiomove(&wpipe->pipe_buffer.buffer[wpipe->pipe_buffer.in],
					segsize, uio);
			rw_enter_write(lock);

			if (error == 0 && segsize < size) {
				/*
				 * Transfer remaining part now, to
				 * support atomic writes.  Wraparound
				 * happened.
				 */
#ifdef DIAGNOSTIC
				if (wpipe->pipe_buffer.in + segsize !=
				    wpipe->pipe_buffer.size)
					panic("Expected pipe buffer wraparound disappeared");
#endif

				rw_exit_write(lock);
				error = uiomove(&wpipe->pipe_buffer.buffer[0],
						size - segsize, uio);
				rw_enter_write(lock);
			}
			if (error == 0) {
				wpipe->pipe_buffer.in += size;
				if (wpipe->pipe_buffer.in >=
				    wpipe->pipe_buffer.size) {
#ifdef DIAGNOSTIC
					if (wpipe->pipe_buffer.in != size - segsize + wpipe->pipe_buffer.size)
						panic("Expected wraparound bad");
#endif
					wpipe->pipe_buffer.in = size - segsize;
				}

				wpipe->pipe_buffer.cnt += size;
#ifdef DIAGNOSTIC
				if (wpipe->pipe_buffer.cnt > wpipe->pipe_buffer.size)
					panic("Pipe buffer overflow");
#endif
			}
			if (error)
				break;
		} else {
			/* If the "read-side" has been blocked, wake it up. */
			if (wpipe->pipe_state & PIPE_WANTR) {
				wpipe->pipe_state &= ~PIPE_WANTR;
				wakeup(wpipe);
			}

			/* Don't block on non-blocking I/O. */
			if (fp->f_flag & FNONBLOCK) {
				error = EAGAIN;
				break;
			}

			/*
			 * We have no more space and have something to offer,
			 * wake up select/poll.
			 */
			pipe_wakeup(wpipe);

			wpipe->pipe_state |= PIPE_WANTW;
			error = pipe_iosleep(wpipe, "pipewr");
			if (error)
				goto unlocked_error;

			/*
			 * If read side wants to go away, we just issue a
			 * signal to ourselves.
			 */
			if (wpipe->pipe_state & PIPE_EOF) {
				error = EPIPE;
				break;
			}
		}
	}
	pipe_iounlock(wpipe);

unlocked_error:
	--wpipe->pipe_busy;

	if (pipe_rundown(wpipe) == 0 && wpipe->pipe_buffer.cnt > 0) {
		/*
		 * If we have put any characters in the buffer, we wake up
		 * the reader.
		 */
		if (wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}
	}

	/* Don't return EPIPE if I/O was successful. */
	if (wpipe->pipe_buffer.cnt == 0 &&
	    uio->uio_resid == 0 &&
	    error == EPIPE) {
		error = 0;
	}

	if (error == 0)
		getnanotime(&wpipe->pipe_mtime);
	/* We have something to offer, wake up select/poll. */
	if (wpipe->pipe_buffer.cnt)
		pipe_wakeup(wpipe);

	rw_exit_write(lock);
	return (error);
}

/*
 * we implement a very minimal set of ioctls for compatibility with sockets.
 */
int
pipe_ioctl(struct file *fp, u_long cmd, caddr_t data, struct proc *p)
{
	struct pipe *mpipe = fp->f_data;
	int error = 0;

	switch (cmd) {

	case FIOASYNC:
		rw_enter_write(mpipe->pipe_lock);
		if (*(int *)data) {
			mpipe->pipe_state |= PIPE_ASYNC;
		} else {
			mpipe->pipe_state &= ~PIPE_ASYNC;
		}
		rw_exit_write(mpipe->pipe_lock);
		break;

	case FIONREAD:
		rw_enter_read(mpipe->pipe_lock);
		*(int *)data = mpipe->pipe_buffer.cnt;
		rw_exit_read(mpipe->pipe_lock);
		break;

	case FIOSETOWN:
	case SIOCSPGRP:
	case TIOCSPGRP:
		error = sigio_setown(&mpipe->pipe_sigio, cmd, data);
		break;

	case FIOGETOWN:
	case SIOCGPGRP:
	case TIOCGPGRP:
		sigio_getown(&mpipe->pipe_sigio, cmd, data);
		break;

	default:
		error = ENOTTY;
	}

	return (error);
}

int
pipe_stat(struct file *fp, struct stat *ub, struct proc *p)
{
	struct pipe *pipe = fp->f_data;

	memset(ub, 0, sizeof(*ub));

	rw_enter_read(pipe->pipe_lock);
	ub->st_mode = S_IFIFO;
	ub->st_blksize = pipe->pipe_buffer.size;
	ub->st_size = pipe->pipe_buffer.cnt;
	ub->st_blocks = (ub->st_size + ub->st_blksize - 1) / ub->st_blksize;
	ub->st_atim.tv_sec  = pipe->pipe_atime.tv_sec;
	ub->st_atim.tv_nsec = pipe->pipe_atime.tv_nsec;
	ub->st_mtim.tv_sec  = pipe->pipe_mtime.tv_sec;
	ub->st_mtim.tv_nsec = pipe->pipe_mtime.tv_nsec;
	ub->st_ctim.tv_sec  = pipe->pipe_ctime.tv_sec;
	ub->st_ctim.tv_nsec = pipe->pipe_ctime.tv_nsec;
	ub->st_uid = fp->f_cred->cr_uid;
	ub->st_gid = fp->f_cred->cr_gid;
	rw_exit_read(pipe->pipe_lock);
	/*
	 * Left as 0: st_dev, st_ino, st_nlink, st_rdev, st_flags, st_gen.
	 * XXX (st_dev, st_ino) should be unique.
	 */
	return (0);
}

int
pipe_close(struct file *fp, struct proc *p)
{
	struct pipe *cpipe = fp->f_data;

	fp->f_ops = NULL;
	fp->f_data = NULL;
	pipe_destroy(cpipe);
	return (0);
}

/*
 * Free kva for pipe circular buffer.
 * No pipe lock check as only called from pipe_buffer_realloc() and pipeclose()
 */
void
pipe_buffer_free(struct pipe *cpipe)
{
	u_int size;

	if (cpipe->pipe_buffer.buffer == NULL)
		return;

	size = cpipe->pipe_buffer.size;

	km_free(cpipe->pipe_buffer.buffer, size, &kv_any, &kp_pageable);

	cpipe->pipe_buffer.buffer = NULL;

	atomic_sub_int(&amountpipekva, size);
	if (size > PIPE_SIZE)
		atomic_dec_int(&nbigpipe);
}

/*
 * shutdown the pipe, and free resources.
 */
void
pipe_destroy(struct pipe *cpipe)
{
	struct pipe *ppipe;

	if (cpipe == NULL)
		return;

	rw_enter_write(cpipe->pipe_lock);

	pipe_wakeup(cpipe);
	sigio_free(&cpipe->pipe_sigio);

	/*
	 * If the other side is blocked, wake it up saying that
	 * we want to close it down.
	 */
	cpipe->pipe_state |= PIPE_EOF;
	while (cpipe->pipe_busy) {
		wakeup(cpipe);
		cpipe->pipe_state |= PIPE_WANTD;
		rwsleep_nsec(cpipe, cpipe->pipe_lock, PRIBIO, "pipecl", INFSLP);
	}

	/* Disconnect from peer. */
	if ((ppipe = cpipe->pipe_peer) != NULL) {
		pipe_wakeup(ppipe);

		ppipe->pipe_state |= PIPE_EOF;
		wakeup(ppipe);
		ppipe->pipe_peer = NULL;
	}

	pipe_buffer_free(cpipe);

	rw_exit_write(cpipe->pipe_lock);

	if (ppipe == NULL)
		pipe_pair_destroy(cpipe->pipe_pair);
}

/*
 * Returns non-zero if a rundown is currently ongoing.
 */
int
pipe_rundown(struct pipe *cpipe)
{
	rw_assert_wrlock(cpipe->pipe_lock);

	if (cpipe->pipe_busy > 0 || (cpipe->pipe_state & PIPE_WANTD) == 0)
		return (0);

	/* Only wakeup pipe_destroy() once the pipe is no longer busy. */
	cpipe->pipe_state &= ~(PIPE_WANTD | PIPE_WANTR | PIPE_WANTW);
	wakeup(cpipe);
	return (1);
}

int
pipe_kqfilter(struct file *fp, struct knote *kn)
{
	struct pipe *rpipe = kn->kn_fp->f_data, *wpipe;
	struct rwlock *lock = rpipe->pipe_lock;
	int error = 0;

	rw_enter_write(lock);
	wpipe = pipe_peer(rpipe);

	switch (kn->kn_filter) {
	case EVFILT_READ:
		kn->kn_fop = &pipe_rfiltops;
		kn->kn_hook = rpipe;
		klist_insert_locked(&rpipe->pipe_klist, kn);
		break;
	case EVFILT_WRITE:
		if (wpipe == NULL) {
			/*
			 * The other end of the pipe has been closed.
			 * Since the filter now always indicates a pending
			 * event, attach the knote to the current side
			 * to proceed with the registration.
			 */
			wpipe = rpipe;
		}
		kn->kn_fop = &pipe_wfiltops;
		kn->kn_hook = wpipe;
		klist_insert_locked(&wpipe->pipe_klist, kn);
		break;
	case EVFILT_EXCEPT:
		if (kn->kn_flags & __EV_SELECT) {
			/* Prevent triggering exceptfds. */
			error = EPERM;
			break;
		}
		if ((kn->kn_flags & __EV_POLL) == 0) {
			/* Disallow usage through kevent(2). */
			error = EINVAL;
			break;
		}
		kn->kn_fop = &pipe_efiltops;
		kn->kn_hook = rpipe;
		klist_insert_locked(&rpipe->pipe_klist, kn);
		break;
	default:
		error = EINVAL;
	}

	rw_exit_write(lock);

	return (error);
}

void
filt_pipedetach(struct knote *kn)
{
	struct pipe *cpipe = kn->kn_hook;

	klist_remove(&cpipe->pipe_klist, kn);
}

int
filt_piperead(struct knote *kn, long hint)
{
	struct pipe *rpipe = kn->kn_fp->f_data, *wpipe;

	rw_assert_wrlock(rpipe->pipe_lock);

	wpipe = pipe_peer(rpipe);

	kn->kn_data = rpipe->pipe_buffer.cnt;

	if ((rpipe->pipe_state & PIPE_EOF) || wpipe == NULL) {
		kn->kn_flags |= EV_EOF;
		if (kn->kn_flags & __EV_POLL)
			kn->kn_flags |= __EV_HUP;
		return (1);
	}

	return (kn->kn_data > 0);
}

int
filt_pipewrite(struct knote *kn, long hint)
{
	struct pipe *rpipe = kn->kn_fp->f_data, *wpipe;

	rw_assert_wrlock(rpipe->pipe_lock);

	wpipe = pipe_peer(rpipe);

	if (wpipe == NULL) {
		kn->kn_data = 0;
		kn->kn_flags |= EV_EOF;
		if (kn->kn_flags & __EV_POLL)
			kn->kn_flags |= __EV_HUP;
		return (1);
	}
	kn->kn_data = wpipe->pipe_buffer.size - wpipe->pipe_buffer.cnt;

	return (kn->kn_data >= PIPE_BUF);
}

int
filt_pipeexcept(struct knote *kn, long hint)
{
	struct pipe *rpipe = kn->kn_fp->f_data, *wpipe;
	int active = 0;

	rw_assert_wrlock(rpipe->pipe_lock);

	wpipe = pipe_peer(rpipe);

	if (kn->kn_flags & __EV_POLL) {
		if ((rpipe->pipe_state & PIPE_EOF) || wpipe == NULL) {
			kn->kn_flags |= __EV_HUP;
			active = 1;
		}
	}

	return (active);
}

int
filt_pipemodify(struct kevent *kev, struct knote *kn)
{
	struct pipe *rpipe = kn->kn_fp->f_data;
	int active;

	rw_enter_write(rpipe->pipe_lock);
	active = knote_modify(kev, kn);
	rw_exit_write(rpipe->pipe_lock);

	return (active);
}

int
filt_pipeprocess(struct knote *kn, struct kevent *kev)
{
	struct pipe *rpipe = kn->kn_fp->f_data;
	int active;

	rw_enter_write(rpipe->pipe_lock);
	active = knote_process(kn, kev);
	rw_exit_write(rpipe->pipe_lock);

	return (active);
}

void
pipe_init(void)
{
	pool_init(&pipe_pair_pool, sizeof(struct pipe_pair), 0, IPL_MPFLOOR,
	    PR_WAITOK, "pipepl", NULL);
}

struct pipe_pair *
pipe_pair_create(void)
{
	struct pipe_pair *pp;

	pp = pool_get(&pipe_pair_pool, PR_WAITOK | PR_ZERO);
	pp->pp_wpipe.pipe_pair = pp;
	pp->pp_rpipe.pipe_pair = pp;
	pp->pp_wpipe.pipe_peer = &pp->pp_rpipe;
	pp->pp_rpipe.pipe_peer = &pp->pp_wpipe;
	/*
	 * One lock is used per pipe pair in order to obtain exclusive access to
	 * the pipe pair.
	 */
	rw_init(&pp->pp_lock, "pipelk");
	pp->pp_wpipe.pipe_lock = &pp->pp_lock;
	pp->pp_rpipe.pipe_lock = &pp->pp_lock;

	klist_init_rwlock(&pp->pp_wpipe.pipe_klist, &pp->pp_lock);
	klist_init_rwlock(&pp->pp_rpipe.pipe_klist, &pp->pp_lock);

	if (pipe_create(&pp->pp_wpipe) || pipe_create(&pp->pp_rpipe))
		goto err;
	return (pp);
err:
	pipe_destroy(&pp->pp_wpipe);
	pipe_destroy(&pp->pp_rpipe);
	return (NULL);
}

void
pipe_pair_destroy(struct pipe_pair *pp)
{
	klist_free(&pp->pp_wpipe.pipe_klist);
	klist_free(&pp->pp_rpipe.pipe_klist);
	pool_put(&pipe_pair_pool, pp);
}