freebsd-dev/sys/dev/tcp_log/tcp_log_dev.c
Jonathan T. Looney 2529f56ed3 Add the "TCP Blackbox Recorder" which we discussed at the developer
summits at BSDCan and BSDCam in 2017.

The TCP Blackbox Recorder allows you to capture events on a TCP connection
in a ring buffer. It stores metadata with the event. It optionally stores
the TCP header associated with an event (if the event is associated with a
packet) and also optionally stores information on the sockets.

It supports setting a log ID on a TCP connection and using this to correlate
multiple connections that share a common log ID.

You can log connections in different modes. If you are doing a coordinated
test with a particular connection, you may tell the system to put it in
mode 4 (continuous dump). Or, if you just want to monitor for errors, you
can put it in mode 1 (ring buffer) and dump all the ring buffers associated
with the connection ID when we receive an error signal for that connection
ID. You can set a default mode that will be applied to a particular ratio
of incoming connections. You can also manually set a mode using a socket
option.

This commit includes only basic probes. rrs@ has added quite an abundance
of probes in his TCP development work. He plans to commit those soon.

There are user-space programs which we plan to commit as ports. These read
the data from the log device and output pcapng files, and then let you
analyze the data (and metadata) in the pcapng files.

Reviewed by:	gnn (previous version)
Obtained from:	Netflix, Inc.
Relnotes:	yes
Differential Revision:	https://reviews.freebsd.org/D11085
2018-03-22 09:40:08 +00:00

522 lines
14 KiB
C

/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2016-2017
* Netflix Inc. 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, 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/conf.h>
#include <sys/fcntl.h>
#include <sys/filio.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/module.h>
#include <sys/poll.h>
#include <sys/queue.h>
#include <sys/refcount.h>
#include <sys/mutex.h>
#include <sys/selinfo.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/sysctl.h>
#include <sys/tree.h>
#include <sys/uio.h>
#include <machine/atomic.h>
#include <sys/counter.h>
#include <dev/tcp_log/tcp_log_dev.h>
#ifdef TCPLOG_DEBUG_COUNTERS
extern counter_u64_t tcp_log_que_read;
extern counter_u64_t tcp_log_que_freed;
#endif
static struct cdev *tcp_log_dev;
static struct selinfo tcp_log_sel;
static struct log_queueh tcp_log_dev_queue_head = STAILQ_HEAD_INITIALIZER(tcp_log_dev_queue_head);
static struct log_infoh tcp_log_dev_reader_head = STAILQ_HEAD_INITIALIZER(tcp_log_dev_reader_head);
MALLOC_DEFINE(M_TCPLOGDEV, "tcp_log_dev", "TCP log device data structures");
static int tcp_log_dev_listeners = 0;
static struct mtx tcp_log_dev_queue_lock;
#define TCP_LOG_DEV_QUEUE_LOCK() mtx_lock(&tcp_log_dev_queue_lock)
#define TCP_LOG_DEV_QUEUE_UNLOCK() mtx_unlock(&tcp_log_dev_queue_lock)
#define TCP_LOG_DEV_QUEUE_LOCK_ASSERT() mtx_assert(&tcp_log_dev_queue_lock, MA_OWNED)
#define TCP_LOG_DEV_QUEUE_UNLOCK_ASSERT() mtx_assert(&tcp_log_dev_queue_lock, MA_NOTOWNED)
#define TCP_LOG_DEV_QUEUE_REF(tldq) refcount_acquire(&((tldq)->tldq_refcnt))
#define TCP_LOG_DEV_QUEUE_UNREF(tldq) refcount_release(&((tldq)->tldq_refcnt))
static void tcp_log_dev_clear_refcount(struct tcp_log_dev_queue *entry);
static void tcp_log_dev_clear_cdevpriv(void *data);
static int tcp_log_dev_open(struct cdev *dev __unused, int flags,
int devtype __unused, struct thread *td __unused);
static int tcp_log_dev_write(struct cdev *dev __unused,
struct uio *uio __unused, int flags __unused);
static int tcp_log_dev_read(struct cdev *dev __unused, struct uio *uio,
int flags __unused);
static int tcp_log_dev_ioctl(struct cdev *dev __unused, u_long cmd,
caddr_t data, int fflag __unused, struct thread *td __unused);
static int tcp_log_dev_poll(struct cdev *dev __unused, int events,
struct thread *td);
enum tcp_log_dev_queue_lock_state {
QUEUE_UNLOCKED = 0,
QUEUE_LOCKED,
};
static struct cdevsw tcp_log_cdevsw = {
.d_version = D_VERSION,
.d_read = tcp_log_dev_read,
.d_open = tcp_log_dev_open,
.d_write = tcp_log_dev_write,
.d_poll = tcp_log_dev_poll,
.d_ioctl = tcp_log_dev_ioctl,
#ifdef NOTYET
.d_mmap = tcp_log_dev_mmap,
#endif
.d_name = "tcp_log",
};
static __inline void
tcp_log_dev_queue_validate_lock(int lockstate)
{
#ifdef INVARIANTS
switch (lockstate) {
case QUEUE_LOCKED:
TCP_LOG_DEV_QUEUE_LOCK_ASSERT();
break;
case QUEUE_UNLOCKED:
TCP_LOG_DEV_QUEUE_UNLOCK_ASSERT();
break;
default:
kassert_panic("%s:%d: unknown queue lock state", __func__,
__LINE__);
}
#endif
}
/*
* Clear the refcount. If appropriate, it will remove the entry from the
* queue and call the destructor.
*
* This must be called with the queue lock held.
*/
static void
tcp_log_dev_clear_refcount(struct tcp_log_dev_queue *entry)
{
KASSERT(entry != NULL, ("%s: called with NULL entry", __func__));
TCP_LOG_DEV_QUEUE_LOCK_ASSERT();
if (TCP_LOG_DEV_QUEUE_UNREF(entry)) {
#ifdef TCPLOG_DEBUG_COUNTERS
counter_u64_add(tcp_log_que_freed, 1);
#endif
/* Remove the entry from the queue and call the destructor. */
STAILQ_REMOVE(&tcp_log_dev_queue_head, entry, tcp_log_dev_queue,
tldq_queue);
(*entry->tldq_dtor)(entry);
}
}
static void
tcp_log_dev_clear_cdevpriv(void *data)
{
struct tcp_log_dev_info *priv;
struct tcp_log_dev_queue *entry, *entry_tmp;
priv = (struct tcp_log_dev_info *)data;
if (priv == NULL)
return;
/*
* Lock the queue and drop our references. We hold references to all
* the entries starting with tldi_head (or, if tldi_head == NULL, all
* entries in the queue).
*
* Because we don't want anyone adding addition things to the queue
* while we are doing this, we lock the queue.
*/
TCP_LOG_DEV_QUEUE_LOCK();
if (priv->tldi_head != NULL) {
entry = priv->tldi_head;
STAILQ_FOREACH_FROM_SAFE(entry, &tcp_log_dev_queue_head,
tldq_queue, entry_tmp) {
tcp_log_dev_clear_refcount(entry);
}
}
tcp_log_dev_listeners--;
KASSERT(tcp_log_dev_listeners >= 0,
("%s: tcp_log_dev_listeners is unexpectedly negative", __func__));
STAILQ_REMOVE(&tcp_log_dev_reader_head, priv, tcp_log_dev_info,
tldi_list);
TCP_LOG_DEV_QUEUE_LOCK_ASSERT();
TCP_LOG_DEV_QUEUE_UNLOCK();
free(priv, M_TCPLOGDEV);
}
static int
tcp_log_dev_open(struct cdev *dev __unused, int flags, int devtype __unused,
struct thread *td __unused)
{
struct tcp_log_dev_info *priv;
struct tcp_log_dev_queue *entry;
int rv;
/*
* Ideally, we shouldn't see these because of file system
* permissions.
*/
if (flags & (FWRITE | FEXEC | FAPPEND | O_TRUNC))
return (ENODEV);
/* Allocate space to hold information about where we are. */
priv = malloc(sizeof(struct tcp_log_dev_info), M_TCPLOGDEV,
M_ZERO | M_WAITOK);
/* Stash the private data away. */
rv = devfs_set_cdevpriv((void *)priv, tcp_log_dev_clear_cdevpriv);
if (!rv) {
/*
* Increase the listener count, add this reader to the list, and
* take references on all current queues.
*/
TCP_LOG_DEV_QUEUE_LOCK();
tcp_log_dev_listeners++;
STAILQ_INSERT_HEAD(&tcp_log_dev_reader_head, priv, tldi_list);
priv->tldi_head = STAILQ_FIRST(&tcp_log_dev_queue_head);
if (priv->tldi_head != NULL)
priv->tldi_cur = priv->tldi_head->tldq_buf;
STAILQ_FOREACH(entry, &tcp_log_dev_queue_head, tldq_queue)
TCP_LOG_DEV_QUEUE_REF(entry);
TCP_LOG_DEV_QUEUE_UNLOCK();
} else {
/* Free the entry. */
free(priv, M_TCPLOGDEV);
}
return (rv);
}
static int
tcp_log_dev_write(struct cdev *dev __unused, struct uio *uio __unused,
int flags __unused)
{
return (ENODEV);
}
static __inline void
tcp_log_dev_rotate_bufs(struct tcp_log_dev_info *priv, int *lockstate)
{
struct tcp_log_dev_queue *entry;
KASSERT(priv->tldi_head != NULL,
("%s:%d: priv->tldi_head unexpectedly NULL",
__func__, __LINE__));
KASSERT(priv->tldi_head->tldq_buf == priv->tldi_cur,
("%s:%d: buffer mismatch (%p vs %p)",
__func__, __LINE__, priv->tldi_head->tldq_buf,
priv->tldi_cur));
tcp_log_dev_queue_validate_lock(*lockstate);
if (*lockstate == QUEUE_UNLOCKED) {
TCP_LOG_DEV_QUEUE_LOCK();
*lockstate = QUEUE_LOCKED;
}
entry = priv->tldi_head;
priv->tldi_head = STAILQ_NEXT(entry, tldq_queue);
tcp_log_dev_clear_refcount(entry);
priv->tldi_cur = NULL;
}
static int
tcp_log_dev_read(struct cdev *dev __unused, struct uio *uio, int flags)
{
struct tcp_log_common_header *buf;
struct tcp_log_dev_info *priv;
struct tcp_log_dev_queue *entry;
ssize_t len;
int lockstate, rv;
/* Get our private info. */
rv = devfs_get_cdevpriv((void **)&priv);
if (rv)
return (rv);
lockstate = QUEUE_UNLOCKED;
/* Do we need to get a new buffer? */
while (priv->tldi_cur == NULL ||
priv->tldi_cur->tlch_length <= priv->tldi_off) {
/* Did we somehow forget to rotate? */
KASSERT(priv->tldi_cur == NULL,
("%s:%d: tldi_cur is unexpectedly non-NULL", __func__,
__LINE__));
if (priv->tldi_cur != NULL)
tcp_log_dev_rotate_bufs(priv, &lockstate);
/*
* Before we start looking at tldi_head, we need a lock on the
* queue to make sure tldi_head stays stable.
*/
if (lockstate == QUEUE_UNLOCKED) {
TCP_LOG_DEV_QUEUE_LOCK();
lockstate = QUEUE_LOCKED;
}
/* We need the next buffer. Do we have one? */
if (priv->tldi_head == NULL && (flags & FNONBLOCK)) {
rv = EAGAIN;
goto done;
}
if (priv->tldi_head == NULL) {
/* Sleep and wait for more things we can read. */
rv = mtx_sleep(&tcp_log_dev_listeners,
&tcp_log_dev_queue_lock, PCATCH, "tcplogdev", 0);
if (rv)
goto done;
if (priv->tldi_head == NULL)
continue;
}
/*
* We have an entry to read. We want to try to create a
* buffer, if one doesn't already exist.
*/
entry = priv->tldi_head;
if (entry->tldq_buf == NULL) {
TCP_LOG_DEV_QUEUE_LOCK_ASSERT();
buf = (*entry->tldq_xform)(entry);
if (buf == NULL) {
rv = EBUSY;
goto done;
}
entry->tldq_buf = buf;
}
priv->tldi_cur = entry->tldq_buf;
priv->tldi_off = 0;
}
/* Copy what we can from this buffer to the output buffer. */
if (uio->uio_resid > 0) {
/* Drop locks so we can take page faults. */
if (lockstate == QUEUE_LOCKED)
TCP_LOG_DEV_QUEUE_UNLOCK();
lockstate = QUEUE_UNLOCKED;
KASSERT(priv->tldi_cur != NULL,
("%s: priv->tldi_cur is unexpectedly NULL", __func__));
/* Copy as much as we can to this uio. */
len = priv->tldi_cur->tlch_length - priv->tldi_off;
if (len > uio->uio_resid)
len = uio->uio_resid;
rv = uiomove(((uint8_t *)priv->tldi_cur) + priv->tldi_off,
len, uio);
if (rv != 0)
goto done;
priv->tldi_off += len;
#ifdef TCPLOG_DEBUG_COUNTERS
counter_u64_add(tcp_log_que_read, len);
#endif
}
/* Are we done with this buffer? If so, find the next one. */
if (priv->tldi_off >= priv->tldi_cur->tlch_length) {
KASSERT(priv->tldi_off == priv->tldi_cur->tlch_length,
("%s: offset (%ju) exceeds length (%ju)", __func__,
(uintmax_t)priv->tldi_off,
(uintmax_t)priv->tldi_cur->tlch_length));
tcp_log_dev_rotate_bufs(priv, &lockstate);
}
done:
tcp_log_dev_queue_validate_lock(lockstate);
if (lockstate == QUEUE_LOCKED)
TCP_LOG_DEV_QUEUE_UNLOCK();
return (rv);
}
static int
tcp_log_dev_ioctl(struct cdev *dev __unused, u_long cmd, caddr_t data,
int fflag __unused, struct thread *td __unused)
{
struct tcp_log_dev_info *priv;
int rv;
/* Get our private info. */
rv = devfs_get_cdevpriv((void **)&priv);
if (rv)
return (rv);
/*
* Set things. Here, we are most concerned about the non-blocking I/O
* flag.
*/
rv = 0;
switch (cmd) {
case FIONBIO:
break;
case FIOASYNC:
if (*(int *)data != 0)
rv = EINVAL;
break;
default:
rv = ENOIOCTL;
}
return (rv);
}
static int
tcp_log_dev_poll(struct cdev *dev __unused, int events, struct thread *td)
{
struct tcp_log_dev_info *priv;
int revents;
/*
* Get our private info. If this fails, claim that all events are
* ready. That should prod the user to do something that will
* make the error evident to them.
*/
if (devfs_get_cdevpriv((void **)&priv))
return (events);
revents = 0;
if (events & (POLLIN | POLLRDNORM)) {
/*
* We can (probably) read right now if we are partway through
* a buffer or if we are just about to start a buffer.
* Because we are going to read tldi_head, we should acquire
* a read lock on the queue.
*/
TCP_LOG_DEV_QUEUE_LOCK();
if ((priv->tldi_head != NULL && priv->tldi_cur == NULL) ||
(priv->tldi_cur != NULL &&
priv->tldi_off < priv->tldi_cur->tlch_length))
revents = events & (POLLIN | POLLRDNORM);
else
selrecord(td, &tcp_log_sel);
TCP_LOG_DEV_QUEUE_UNLOCK();
} else {
/*
* It only makes sense to poll for reading. So, again, prod the
* user to do something that will make the error of their ways
* apparent.
*/
revents = events;
}
return (revents);
}
int
tcp_log_dev_add_log(struct tcp_log_dev_queue *entry)
{
struct tcp_log_dev_info *priv;
int rv;
bool wakeup_needed;
KASSERT(entry->tldq_buf != NULL || entry->tldq_xform != NULL,
("%s: Called with both tldq_buf and tldq_xform set to NULL",
__func__));
KASSERT(entry->tldq_dtor != NULL,
("%s: Called with tldq_dtor set to NULL", __func__));
/* Get a lock on the queue. */
TCP_LOG_DEV_QUEUE_LOCK();
/* If no one is listening, tell the caller to free the resources. */
if (tcp_log_dev_listeners == 0) {
rv = ENXIO;
goto done;
}
/* Add this to the end of the tailq. */
STAILQ_INSERT_TAIL(&tcp_log_dev_queue_head, entry, tldq_queue);
/* Add references for all current listeners. */
refcount_init(&entry->tldq_refcnt, tcp_log_dev_listeners);
/*
* If any listener is currently stuck on NULL, that means they are
* waiting. Point their head to this new entry.
*/
wakeup_needed = false;
STAILQ_FOREACH(priv, &tcp_log_dev_reader_head, tldi_list)
if (priv->tldi_head == NULL) {
priv->tldi_head = entry;
wakeup_needed = true;
}
if (wakeup_needed) {
selwakeup(&tcp_log_sel);
wakeup(&tcp_log_dev_listeners);
}
rv = 0;
done:
TCP_LOG_DEV_QUEUE_LOCK_ASSERT();
TCP_LOG_DEV_QUEUE_UNLOCK();
return (rv);
}
static int
tcp_log_dev_modevent(module_t mod __unused, int type, void *data __unused)
{
/* TODO: Support intelligent unloading. */
switch (type) {
case MOD_LOAD:
if (bootverbose)
printf("tcp_log: tcp_log device\n");
memset(&tcp_log_sel, 0, sizeof(tcp_log_sel));
memset(&tcp_log_dev_queue_lock, 0, sizeof(struct mtx));
mtx_init(&tcp_log_dev_queue_lock, "tcp_log dev",
"tcp_log device queues", MTX_DEF);
tcp_log_dev = make_dev_credf(MAKEDEV_ETERNAL_KLD,
&tcp_log_cdevsw, 0, NULL, UID_ROOT, GID_WHEEL, 0400,
"tcp_log");
break;
default:
return (EOPNOTSUPP);
}
return (0);
}
DEV_MODULE(tcp_log_dev, tcp_log_dev_modevent, NULL);
MODULE_VERSION(tcp_log_dev, 1);