qemu/tools/virtiofsd/fuse_virtio.c

1080 lines
32 KiB
C
Raw Normal View History

/*
* virtio-fs glue for FUSE
* Copyright (C) 2018 Red Hat, Inc. and/or its affiliates
*
* Authors:
* Dave Gilbert <dgilbert@redhat.com>
*
* Implements the glue between libfuse and libvhost-user
*
* This program can be distributed under the terms of the GNU LGPLv2.
* See the file COPYING.LIB
*/
#include "qemu/osdep.h"
#include "qemu/iov.h"
#include "qapi/error.h"
#include "fuse_i.h"
#include "standard-headers/linux/fuse.h"
#include "fuse_misc.h"
#include "fuse_opt.h"
#include "fuse_virtio.h"
#include <sys/eventfd.h>
#include <sys/socket.h>
#include <sys/un.h>
#include <grp.h>
#include "libvhost-user.h"
struct fv_VuDev;
struct fv_QueueInfo {
pthread_t thread;
/*
* This lock protects the VuVirtq preventing races between
* fv_queue_thread() and fv_queue_worker().
*/
pthread_mutex_t vq_lock;
struct fv_VuDev *virtio_dev;
/* Our queue index, corresponds to array position */
int qidx;
int kick_fd;
int kill_fd; /* For killing the thread */
};
/* A FUSE request */
typedef struct {
VuVirtqElement elem;
struct fuse_chan ch;
/* Used to complete requests that involve no reply */
bool reply_sent;
} FVRequest;
/*
* We pass the dev element into libvhost-user
* and then use it to get back to the outer
* container for other data.
*/
struct fv_VuDev {
VuDev dev;
struct fuse_session *se;
/*
* Either handle virtqueues or vhost-user protocol messages. Don't do
* both at the same time since that could lead to race conditions if
* virtqueues or memory tables change while another thread is accessing
* them.
*
* The assumptions are:
* 1. fv_queue_thread() reads/writes to virtqueues and only reads VuDev.
* 2. virtio_loop() reads/writes virtqueues and VuDev.
*/
pthread_rwlock_t vu_dispatch_rwlock;
/*
* The following pair of fields are only accessed in the main
* virtio_loop
*/
size_t nqueues;
struct fv_QueueInfo **qi;
};
/* Callback from libvhost-user */
static uint64_t fv_get_features(VuDev *dev)
{
return 1ULL << VIRTIO_F_VERSION_1;
}
/* Callback from libvhost-user */
static void fv_set_features(VuDev *dev, uint64_t features)
{
}
/*
* Callback from libvhost-user if there's a new fd we're supposed to listen
* to, typically a queue kick?
*/
static void fv_set_watch(VuDev *dev, int fd, int condition, vu_watch_cb cb,
void *data)
{
fuse_log(FUSE_LOG_WARNING, "%s: TODO! fd=%d\n", __func__, fd);
}
/*
* Callback from libvhost-user if we're no longer supposed to listen on an fd
*/
static void fv_remove_watch(VuDev *dev, int fd)
{
fuse_log(FUSE_LOG_WARNING, "%s: TODO! fd=%d\n", __func__, fd);
}
/* Callback from libvhost-user to panic */
static void fv_panic(VuDev *dev, const char *err)
{
fuse_log(FUSE_LOG_ERR, "%s: libvhost-user: %s\n", __func__, err);
/* TODO: Allow reconnects?? */
exit(EXIT_FAILURE);
}
/*
* Copy from an iovec into a fuse_buf (memory only)
* Caller must ensure there is space
*/
static size_t copy_from_iov(struct fuse_buf *buf, size_t out_num,
const struct iovec *out_sg,
size_t max)
{
void *dest = buf->mem;
size_t copied = 0;
while (out_num && max) {
size_t onelen = out_sg->iov_len;
onelen = MIN(onelen, max);
memcpy(dest, out_sg->iov_base, onelen);
dest += onelen;
copied += onelen;
out_sg++;
out_num--;
max -= onelen;
}
return copied;
}
/*
* Skip 'skip' bytes in the iov; 'sg_1stindex' is set as
* the index for the 1st iovec to read data from, and
* 'sg_1stskip' is the number of bytes to skip in that entry.
*
* Returns True if there are at least 'skip' bytes in the iovec
*
*/
static bool skip_iov(const struct iovec *sg, size_t sg_size,
size_t skip,
size_t *sg_1stindex, size_t *sg_1stskip)
{
size_t vec;
for (vec = 0; vec < sg_size; vec++) {
if (sg[vec].iov_len > skip) {
*sg_1stskip = skip;
*sg_1stindex = vec;
return true;
}
skip -= sg[vec].iov_len;
}
*sg_1stindex = vec;
*sg_1stskip = 0;
return skip == 0;
}
/*
* Copy from one iov to another, the given number of bytes
* The caller must have checked sizes.
*/
static void copy_iov(struct iovec *src_iov, int src_count,
struct iovec *dst_iov, int dst_count, size_t to_copy)
{
size_t dst_offset = 0;
/* Outer loop copies 'src' elements */
while (to_copy) {
assert(src_count);
size_t src_len = src_iov[0].iov_len;
size_t src_offset = 0;
if (src_len > to_copy) {
src_len = to_copy;
}
/* Inner loop copies contents of one 'src' to maybe multiple dst. */
while (src_len) {
assert(dst_count);
size_t dst_len = dst_iov[0].iov_len - dst_offset;
if (dst_len > src_len) {
dst_len = src_len;
}
memcpy(dst_iov[0].iov_base + dst_offset,
src_iov[0].iov_base + src_offset, dst_len);
src_len -= dst_len;
to_copy -= dst_len;
src_offset += dst_len;
dst_offset += dst_len;
assert(dst_offset <= dst_iov[0].iov_len);
if (dst_offset == dst_iov[0].iov_len) {
dst_offset = 0;
dst_iov++;
dst_count--;
}
}
src_iov++;
src_count--;
}
}
/*
* pthread_rwlock_rdlock() and pthread_rwlock_wrlock can fail if
* a deadlock condition is detected or the current thread already
* owns the lock. They can also fail, like pthread_rwlock_unlock(),
* if the mutex wasn't properly initialized. None of these are ever
* expected to happen.
*/
static void vu_dispatch_rdlock(struct fv_VuDev *vud)
{
int ret = pthread_rwlock_rdlock(&vud->vu_dispatch_rwlock);
assert(ret == 0);
}
static void vu_dispatch_wrlock(struct fv_VuDev *vud)
{
int ret = pthread_rwlock_wrlock(&vud->vu_dispatch_rwlock);
assert(ret == 0);
}
static void vu_dispatch_unlock(struct fv_VuDev *vud)
{
int ret = pthread_rwlock_unlock(&vud->vu_dispatch_rwlock);
assert(ret == 0);
}
static void vq_send_element(struct fv_QueueInfo *qi, VuVirtqElement *elem,
ssize_t len)
{
struct fuse_session *se = qi->virtio_dev->se;
VuDev *dev = &se->virtio_dev->dev;
VuVirtq *q = vu_get_queue(dev, qi->qidx);
vu_dispatch_rdlock(qi->virtio_dev);
pthread_mutex_lock(&qi->vq_lock);
vu_queue_push(dev, q, elem, len);
vu_queue_notify(dev, q);
pthread_mutex_unlock(&qi->vq_lock);
vu_dispatch_unlock(qi->virtio_dev);
}
/*
* Called back by ll whenever it wants to send a reply/message back
* The 1st element of the iov starts with the fuse_out_header
* 'unique'==0 means it's a notify message.
*/
int virtio_send_msg(struct fuse_session *se, struct fuse_chan *ch,
struct iovec *iov, int count)
{
FVRequest *req = container_of(ch, FVRequest, ch);
struct fv_QueueInfo *qi = ch->qi;
VuVirtqElement *elem = &req->elem;
int ret = 0;
assert(count >= 1);
assert(iov[0].iov_len >= sizeof(struct fuse_out_header));
struct fuse_out_header *out = iov[0].iov_base;
/* TODO: Endianness! */
size_t tosend_len = iov_size(iov, count);
/* unique == 0 is notification, which we don't support */
assert(out->unique);
assert(!req->reply_sent);
/* The 'in' part of the elem is to qemu */
unsigned int in_num = elem->in_num;
struct iovec *in_sg = elem->in_sg;
size_t in_len = iov_size(in_sg, in_num);
fuse_log(FUSE_LOG_DEBUG, "%s: elem %d: with %d in desc of length %zd\n",
__func__, elem->index, in_num, in_len);
/*
* The elem should have room for a 'fuse_out_header' (out from fuse)
* plus the data based on the len in the header.
*/
if (in_len < sizeof(struct fuse_out_header)) {
fuse_log(FUSE_LOG_ERR, "%s: elem %d too short for out_header\n",
__func__, elem->index);
ret = -E2BIG;
goto err;
}
if (in_len < tosend_len) {
fuse_log(FUSE_LOG_ERR, "%s: elem %d too small for data len %zd\n",
__func__, elem->index, tosend_len);
ret = -E2BIG;
goto err;
}
copy_iov(iov, count, in_sg, in_num, tosend_len);
vq_send_element(qi, elem, tosend_len);
req->reply_sent = true;
err:
return ret;
}
/*
* Callback from fuse_send_data_iov_* when it's virtio and the buffer
* is a single FD with FUSE_BUF_IS_FD | FUSE_BUF_FD_SEEK
* We need send the iov and then the buffer.
* Return 0 on success
*/
int virtio_send_data_iov(struct fuse_session *se, struct fuse_chan *ch,
struct iovec *iov, int count, struct fuse_bufvec *buf,
size_t len)
{
FVRequest *req = container_of(ch, FVRequest, ch);
struct fv_QueueInfo *qi = ch->qi;
VuVirtqElement *elem = &req->elem;
int ret = 0;
g_autofree struct iovec *in_sg_cpy = NULL;
assert(count >= 1);
assert(iov[0].iov_len >= sizeof(struct fuse_out_header));
struct fuse_out_header *out = iov[0].iov_base;
/* TODO: Endianness! */
size_t iov_len = iov_size(iov, count);
size_t tosend_len = iov_len + len;
out->len = tosend_len;
fuse_log(FUSE_LOG_DEBUG, "%s: count=%d len=%zd iov_len=%zd\n", __func__,
count, len, iov_len);
/* unique == 0 is notification which we don't support */
assert(out->unique);
assert(!req->reply_sent);
/* The 'in' part of the elem is to qemu */
unsigned int in_num = elem->in_num;
struct iovec *in_sg = elem->in_sg;
size_t in_len = iov_size(in_sg, in_num);
fuse_log(FUSE_LOG_DEBUG, "%s: elem %d: with %d in desc of length %zd\n",
__func__, elem->index, in_num, in_len);
/*
* The elem should have room for a 'fuse_out_header' (out from fuse)
* plus the data based on the len in the header.
*/
if (in_len < sizeof(struct fuse_out_header)) {
fuse_log(FUSE_LOG_ERR, "%s: elem %d too short for out_header\n",
__func__, elem->index);
return E2BIG;
}
if (in_len < tosend_len) {
fuse_log(FUSE_LOG_ERR, "%s: elem %d too small for data len %zd\n",
__func__, elem->index, tosend_len);
return E2BIG;
}
/* TODO: Limit to 'len' */
/* First copy the header data from iov->in_sg */
copy_iov(iov, count, in_sg, in_num, iov_len);
/*
* Build a copy of the the in_sg iov so we can skip bits in it,
* including changing the offsets
*/
in_sg_cpy = g_new(struct iovec, in_num);
memcpy(in_sg_cpy, in_sg, sizeof(struct iovec) * in_num);
/* These get updated as we skip */
struct iovec *in_sg_ptr = in_sg_cpy;
unsigned int in_sg_cpy_count = in_num;
/* skip over parts of in_sg that contained the header iov */
iov_discard_front(&in_sg_ptr, &in_sg_cpy_count, iov_len);
do {
fuse_log(FUSE_LOG_DEBUG, "%s: in_sg_cpy_count=%d len remaining=%zd\n",
__func__, in_sg_cpy_count, len);
ret = preadv(buf->buf[0].fd, in_sg_ptr, in_sg_cpy_count,
buf->buf[0].pos);
if (ret == -1) {
ret = errno;
if (ret == EINTR) {
continue;
}
fuse_log(FUSE_LOG_DEBUG, "%s: preadv failed (%m) len=%zd\n",
__func__, len);
return ret;
}
if (!ret) {
/* EOF case? */
fuse_log(FUSE_LOG_DEBUG, "%s: !ret len remaining=%zd\n", __func__,
len);
break;
}
fuse_log(FUSE_LOG_DEBUG, "%s: preadv ret=%d len=%zd\n", __func__,
ret, len);
len -= ret;
/* Short read. Retry reading remaining bytes */
if (len) {
fuse_log(FUSE_LOG_DEBUG, "%s: ret < len\n", __func__);
/* Skip over this much next time around */
iov_discard_front(&in_sg_ptr, &in_sg_cpy_count, ret);
buf->buf[0].pos += ret;
}
} while (len);
/* Need to fix out->len on EOF */
if (len) {
struct fuse_out_header *out_sg = in_sg[0].iov_base;
tosend_len -= len;
out_sg->len = tosend_len;
}
vq_send_element(qi, elem, tosend_len);
req->reply_sent = true;
return 0;
}
static __thread bool clone_fs_called;
/* Process one FVRequest in a thread pool */
static void fv_queue_worker(gpointer data, gpointer user_data)
{
struct fv_QueueInfo *qi = user_data;
struct fuse_session *se = qi->virtio_dev->se;
FVRequest *req = data;
VuVirtqElement *elem = &req->elem;
struct fuse_buf fbuf = {};
bool allocated_bufv = false;
struct fuse_bufvec bufv;
struct fuse_bufvec *pbufv;
struct fuse_in_header inh;
assert(se->bufsize > sizeof(struct fuse_in_header));
if (!clone_fs_called) {
int ret;
/* unshare FS for xattr operation */
ret = unshare(CLONE_FS);
/* should not fail */
assert(ret == 0);
clone_fs_called = true;
}
/*
* An element contains one request and the space to send our response
* They're spread over multiple descriptors in a scatter/gather set
* and we can't trust the guest to keep them still; so copy in/out.
*/
fbuf.mem = g_malloc(se->bufsize);
fuse_mutex_init(&req->ch.lock);
req->ch.fd = -1;
req->ch.qi = qi;
/* The 'out' part of the elem is from qemu */
unsigned int out_num = elem->out_num;
struct iovec *out_sg = elem->out_sg;
size_t out_len = iov_size(out_sg, out_num);
fuse_log(FUSE_LOG_DEBUG,
"%s: elem %d: with %d out desc of length %zd\n",
__func__, elem->index, out_num, out_len);
/*
* The elem should contain a 'fuse_in_header' (in to fuse)
* plus the data based on the len in the header.
*/
if (out_len < sizeof(struct fuse_in_header)) {
fuse_log(FUSE_LOG_ERR, "%s: elem %d too short for in_header\n",
__func__, elem->index);
assert(0); /* TODO */
}
if (out_len > se->bufsize) {
fuse_log(FUSE_LOG_ERR, "%s: elem %d too large for buffer\n", __func__,
elem->index);
assert(0); /* TODO */
}
/* Copy just the fuse_in_header and look at it */
copy_from_iov(&fbuf, out_num, out_sg,
sizeof(struct fuse_in_header));
memcpy(&inh, fbuf.mem, sizeof(struct fuse_in_header));
pbufv = NULL; /* Compiler thinks an unitialised path */
if (inh.opcode == FUSE_WRITE &&
out_len >= (sizeof(struct fuse_in_header) +
sizeof(struct fuse_write_in))) {
/*
* For a write we don't actually need to copy the
* data, we can just do it straight out of guest memory
* but we must still copy the headers in case the guest
* was nasty and changed them while we were using them.
*/
fuse_log(FUSE_LOG_DEBUG, "%s: Write special case\n", __func__);
fbuf.size = copy_from_iov(&fbuf, out_num, out_sg,
sizeof(struct fuse_in_header) +
sizeof(struct fuse_write_in));
/* That copy reread the in_header, make sure we use the original */
memcpy(fbuf.mem, &inh, sizeof(struct fuse_in_header));
/* Allocate the bufv, with space for the rest of the iov */
pbufv = g_try_malloc(sizeof(struct fuse_bufvec) +
sizeof(struct fuse_buf) * out_num);
if (!pbufv) {
fuse_log(FUSE_LOG_ERR, "%s: pbufv malloc failed\n",
__func__);
goto out;
}
allocated_bufv = true;
pbufv->count = 1;
pbufv->buf[0] = fbuf;
size_t iovindex, pbufvindex, iov_bytes_skip;
pbufvindex = 1; /* 2 headers, 1 fusebuf */
if (!skip_iov(out_sg, out_num,
sizeof(struct fuse_in_header) +
sizeof(struct fuse_write_in),
&iovindex, &iov_bytes_skip)) {
fuse_log(FUSE_LOG_ERR, "%s: skip failed\n",
__func__);
goto out;
}
for (; iovindex < out_num; iovindex++, pbufvindex++) {
pbufv->count++;
pbufv->buf[pbufvindex].pos = ~0; /* Dummy */
pbufv->buf[pbufvindex].flags = 0;
pbufv->buf[pbufvindex].mem = out_sg[iovindex].iov_base;
pbufv->buf[pbufvindex].size = out_sg[iovindex].iov_len;
if (iov_bytes_skip) {
pbufv->buf[pbufvindex].mem += iov_bytes_skip;
pbufv->buf[pbufvindex].size -= iov_bytes_skip;
iov_bytes_skip = 0;
}
}
} else {
/* Normal (non fast write) path */
copy_from_iov(&fbuf, out_num, out_sg, se->bufsize);
/* That copy reread the in_header, make sure we use the original */
memcpy(fbuf.mem, &inh, sizeof(struct fuse_in_header));
fbuf.size = out_len;
/* TODO! Endianness of header */
/* TODO: Add checks for fuse_session_exited */
bufv.buf[0] = fbuf;
bufv.count = 1;
pbufv = &bufv;
}
pbufv->idx = 0;
pbufv->off = 0;
fuse_session_process_buf_int(se, pbufv, &req->ch);
out:
if (allocated_bufv) {
g_free(pbufv);
}
/* If the request has no reply, still recycle the virtqueue element */
if (!req->reply_sent) {
fuse_log(FUSE_LOG_DEBUG, "%s: elem %d no reply sent\n", __func__,
elem->index);
vq_send_element(qi, elem, 0);
}
pthread_mutex_destroy(&req->ch.lock);
g_free(fbuf.mem);
free(req);
}
/* Thread function for individual queues, created when a queue is 'started' */
static void *fv_queue_thread(void *opaque)
{
struct fv_QueueInfo *qi = opaque;
struct VuDev *dev = &qi->virtio_dev->dev;
struct VuVirtq *q = vu_get_queue(dev, qi->qidx);
struct fuse_session *se = qi->virtio_dev->se;
GThreadPool *pool = NULL;
GList *req_list = NULL;
if (se->thread_pool_size) {
fuse_log(FUSE_LOG_DEBUG, "%s: Creating thread pool for Queue %d\n",
__func__, qi->qidx);
pool = g_thread_pool_new(fv_queue_worker, qi, se->thread_pool_size,
FALSE, NULL);
if (!pool) {
fuse_log(FUSE_LOG_ERR, "%s: g_thread_pool_new failed\n", __func__);
return NULL;
}
}
fuse_log(FUSE_LOG_INFO, "%s: Start for queue %d kick_fd %d\n", __func__,
qi->qidx, qi->kick_fd);
while (1) {
struct pollfd pf[2];
pf[0].fd = qi->kick_fd;
pf[0].events = POLLIN;
pf[0].revents = 0;
pf[1].fd = qi->kill_fd;
pf[1].events = POLLIN;
pf[1].revents = 0;
fuse_log(FUSE_LOG_DEBUG, "%s: Waiting for Queue %d event\n", __func__,
qi->qidx);
int poll_res = ppoll(pf, 2, NULL, NULL);
if (poll_res == -1) {
if (errno == EINTR) {
fuse_log(FUSE_LOG_INFO, "%s: ppoll interrupted, going around\n",
__func__);
continue;
}
fuse_log(FUSE_LOG_ERR, "fv_queue_thread ppoll: %m\n");
break;
}
assert(poll_res >= 1);
if (pf[0].revents & (POLLERR | POLLHUP | POLLNVAL)) {
fuse_log(FUSE_LOG_ERR, "%s: Unexpected poll revents %x Queue %d\n",
__func__, pf[0].revents, qi->qidx);
break;
}
if (pf[1].revents & (POLLERR | POLLHUP | POLLNVAL)) {
fuse_log(FUSE_LOG_ERR,
"%s: Unexpected poll revents %x Queue %d killfd\n",
__func__, pf[1].revents, qi->qidx);
break;
}
if (pf[1].revents) {
fuse_log(FUSE_LOG_INFO, "%s: kill event on queue %d - quitting\n",
__func__, qi->qidx);
break;
}
assert(pf[0].revents & POLLIN);
fuse_log(FUSE_LOG_DEBUG, "%s: Got queue event on Queue %d\n", __func__,
qi->qidx);
eventfd_t evalue;
if (eventfd_read(qi->kick_fd, &evalue)) {
fuse_log(FUSE_LOG_ERR, "Eventfd_read for queue: %m\n");
break;
}
/* Mutual exclusion with virtio_loop() */
vu_dispatch_rdlock(qi->virtio_dev);
pthread_mutex_lock(&qi->vq_lock);
/* out is from guest, in is too guest */
unsigned int in_bytes, out_bytes;
vu_queue_get_avail_bytes(dev, q, &in_bytes, &out_bytes, ~0, ~0);
fuse_log(FUSE_LOG_DEBUG,
"%s: Queue %d gave evalue: %zx available: in: %u out: %u\n",
__func__, qi->qidx, (size_t)evalue, in_bytes, out_bytes);
while (1) {
FVRequest *req = vu_queue_pop(dev, q, sizeof(FVRequest));
if (!req) {
break;
}
req->reply_sent = false;
if (!se->thread_pool_size) {
req_list = g_list_prepend(req_list, req);
} else {
g_thread_pool_push(pool, req, NULL);
}
}
pthread_mutex_unlock(&qi->vq_lock);
vu_dispatch_unlock(qi->virtio_dev);
/* Process all the requests. */
if (!se->thread_pool_size && req_list != NULL) {
virtiofsd: Reverse req_list before processing it With the thread pool disabled, we add the requests in the queue to a GList, processing by iterating over there afterwards. For adding them, we're using "g_list_prepend()", which is more efficient but causes the requests to be processed in reverse order, breaking the read-ahead and request-merging optimizations in the host for sequential operations. According to the documentation, if you need to process the request in-order, using "g_list_prepend()" and then reversing the list with "g_list_reverse()" is more efficient than using "g_list_append()", so let's do it that way. Testing on a spinning disk (to boost the increase of read-ahead and request-merging) shows a 4x improvement on sequential write fio test: Test: fio --directory=/mnt/virtio-fs --filename=fio-file1 --runtime=20 --iodepth=16 --size=4G --direct=1 --blocksize=4K --ioengine libaio --rw write --name seqwrite-libaio Without "g_list_reverse()": ... Jobs: 1 (f=1): [W(1)][100.0%][w=22.4MiB/s][w=5735 IOPS][eta 00m:00s] seqwrite-libaio: (groupid=0, jobs=1): err= 0: pid=710: Tue Aug 24 12:58:16 2021 write: IOPS=5709, BW=22.3MiB/s (23.4MB/s)(446MiB/20002msec); 0 zone resets ... With "g_list_reverse()": ... Jobs: 1 (f=1): [W(1)][100.0%][w=84.0MiB/s][w=21.5k IOPS][eta 00m:00s] seqwrite-libaio: (groupid=0, jobs=1): err= 0: pid=716: Tue Aug 24 13:00:15 2021 write: IOPS=21.3k, BW=83.1MiB/s (87.2MB/s)(1663MiB/20001msec); 0 zone resets ... Signed-off-by: Sergio Lopez <slp@redhat.com> Message-Id: <20210824131158.39970-1-slp@redhat.com> Reviewed-by: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
2021-08-24 16:11:58 +03:00
req_list = g_list_reverse(req_list);
g_list_foreach(req_list, fv_queue_worker, qi);
g_list_free(req_list);
req_list = NULL;
}
}
if (pool) {
g_thread_pool_free(pool, FALSE, TRUE);
}
return NULL;
}
static void fv_queue_cleanup_thread(struct fv_VuDev *vud, int qidx)
{
int ret;
struct fv_QueueInfo *ourqi;
assert(qidx < vud->nqueues);
ourqi = vud->qi[qidx];
/* Kill the thread */
if (eventfd_write(ourqi->kill_fd, 1)) {
fuse_log(FUSE_LOG_ERR, "Eventfd_write for queue %d: %s\n",
qidx, strerror(errno));
}
ret = pthread_join(ourqi->thread, NULL);
if (ret) {
fuse_log(FUSE_LOG_ERR, "%s: Failed to join thread idx %d err %d\n",
__func__, qidx, ret);
}
pthread_mutex_destroy(&ourqi->vq_lock);
close(ourqi->kill_fd);
ourqi->kick_fd = -1;
g_free(vud->qi[qidx]);
vud->qi[qidx] = NULL;
}
static void stop_all_queues(struct fv_VuDev *vud)
{
for (int i = 0; i < vud->nqueues; i++) {
if (!vud->qi[i]) {
continue;
}
fuse_log(FUSE_LOG_INFO, "%s: Stopping queue %d thread\n", __func__, i);
fv_queue_cleanup_thread(vud, i);
}
}
/* Callback from libvhost-user on start or stop of a queue */
static void fv_queue_set_started(VuDev *dev, int qidx, bool started)
{
struct fv_VuDev *vud = container_of(dev, struct fv_VuDev, dev);
struct fv_QueueInfo *ourqi;
fuse_log(FUSE_LOG_INFO, "%s: qidx=%d started=%d\n", __func__, qidx,
started);
assert(qidx >= 0);
/*
* Ignore additional request queues for now. passthrough_ll.c must be
* audited for thread-safety issues first. It was written with a
* well-behaved client in mind and may not protect against all types of
* races yet.
*/
if (qidx > 1) {
fuse_log(FUSE_LOG_ERR,
"%s: multiple request queues not yet implemented, please only "
"configure 1 request queue\n",
__func__);
exit(EXIT_FAILURE);
}
if (started) {
/* Fire up a thread to watch this queue */
if (qidx >= vud->nqueues) {
vud->qi = g_realloc_n(vud->qi, qidx + 1, sizeof(vud->qi[0]));
memset(vud->qi + vud->nqueues, 0,
sizeof(vud->qi[0]) * (1 + (qidx - vud->nqueues)));
vud->nqueues = qidx + 1;
}
if (!vud->qi[qidx]) {
vud->qi[qidx] = g_new0(struct fv_QueueInfo, 1);
vud->qi[qidx]->virtio_dev = vud;
vud->qi[qidx]->qidx = qidx;
} else {
/* Shouldn't have been started */
assert(vud->qi[qidx]->kick_fd == -1);
}
ourqi = vud->qi[qidx];
ourqi->kick_fd = dev->vq[qidx].kick_fd;
ourqi->kill_fd = eventfd(0, EFD_CLOEXEC | EFD_SEMAPHORE);
assert(ourqi->kill_fd != -1);
pthread_mutex_init(&ourqi->vq_lock, NULL);
if (pthread_create(&ourqi->thread, NULL, fv_queue_thread, ourqi)) {
fuse_log(FUSE_LOG_ERR, "%s: Failed to create thread for queue %d\n",
__func__, qidx);
assert(0);
}
} else {
virtiofsd: Release vu_dispatch_lock when stopping queue QEMU can stop a virtqueue by sending a VHOST_USER_GET_VRING_BASE request to virtiofsd. As with all other vhost-user protocol messages, the thread that runs the main event loop in virtiofsd takes the vu_dispatch lock in write mode. This ensures that no other thread can access virtqueues or memory tables at the same time. In the case of VHOST_USER_GET_VRING_BASE, the main thread basically notifies the queue thread that it should terminate and waits for its termination: main() virtio_loop() vu_dispatch_wrlock() vu_dispatch() vu_process_message() vu_get_vring_base_exec() fv_queue_cleanup_thread() pthread_join() Unfortunately, the queue thread ends up calling virtio_send_msg() at some point, which itself needs to grab the lock: fv_queue_thread() g_list_foreach() fv_queue_worker() fuse_session_process_buf_int() do_release() lo_release() fuse_reply_err() send_reply() send_reply_iov() fuse_send_reply_iov_nofree() fuse_send_msg() virtio_send_msg() vu_dispatch_rdlock() <-- Deadlock ! Simply have the main thread to release the lock before going to sleep and take it back afterwards. A very similar patch was already sent by Vivek Goyal sometime back: https://listman.redhat.com/archives/virtio-fs/2021-January/msg00073.html The only difference here is that this done in fv_queue_set_started() because fv_queue_cleanup_thread() can also be called from virtio_loop() without the lock being held. Signed-off-by: Greg Kurz <groug@kaod.org> Reviewed-by: Vivek Goyal <vgoyal@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Message-Id: <20210312092212.782255-8-groug@kaod.org> Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
2021-03-12 12:22:12 +03:00
/*
* Temporarily drop write-lock taken in virtio_loop() so that
* the queue thread doesn't block in virtio_send_msg().
*/
vu_dispatch_unlock(vud);
fv_queue_cleanup_thread(vud, qidx);
virtiofsd: Release vu_dispatch_lock when stopping queue QEMU can stop a virtqueue by sending a VHOST_USER_GET_VRING_BASE request to virtiofsd. As with all other vhost-user protocol messages, the thread that runs the main event loop in virtiofsd takes the vu_dispatch lock in write mode. This ensures that no other thread can access virtqueues or memory tables at the same time. In the case of VHOST_USER_GET_VRING_BASE, the main thread basically notifies the queue thread that it should terminate and waits for its termination: main() virtio_loop() vu_dispatch_wrlock() vu_dispatch() vu_process_message() vu_get_vring_base_exec() fv_queue_cleanup_thread() pthread_join() Unfortunately, the queue thread ends up calling virtio_send_msg() at some point, which itself needs to grab the lock: fv_queue_thread() g_list_foreach() fv_queue_worker() fuse_session_process_buf_int() do_release() lo_release() fuse_reply_err() send_reply() send_reply_iov() fuse_send_reply_iov_nofree() fuse_send_msg() virtio_send_msg() vu_dispatch_rdlock() <-- Deadlock ! Simply have the main thread to release the lock before going to sleep and take it back afterwards. A very similar patch was already sent by Vivek Goyal sometime back: https://listman.redhat.com/archives/virtio-fs/2021-January/msg00073.html The only difference here is that this done in fv_queue_set_started() because fv_queue_cleanup_thread() can also be called from virtio_loop() without the lock being held. Signed-off-by: Greg Kurz <groug@kaod.org> Reviewed-by: Vivek Goyal <vgoyal@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Message-Id: <20210312092212.782255-8-groug@kaod.org> Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
2021-03-12 12:22:12 +03:00
vu_dispatch_wrlock(vud);
}
}
static bool fv_queue_order(VuDev *dev, int qidx)
{
return false;
}
static const VuDevIface fv_iface = {
.get_features = fv_get_features,
.set_features = fv_set_features,
/* Don't need process message, we've not got any at vhost-user level */
.queue_set_started = fv_queue_set_started,
.queue_is_processed_in_order = fv_queue_order,
};
/*
* Main loop; this mostly deals with events on the vhost-user
* socket itself, and not actual fuse data.
*/
int virtio_loop(struct fuse_session *se)
{
fuse_log(FUSE_LOG_INFO, "%s: Entry\n", __func__);
while (!fuse_session_exited(se)) {
struct pollfd pf[1];
bool ok;
pf[0].fd = se->vu_socketfd;
pf[0].events = POLLIN;
pf[0].revents = 0;
fuse_log(FUSE_LOG_DEBUG, "%s: Waiting for VU event\n", __func__);
int poll_res = ppoll(pf, 1, NULL, NULL);
if (poll_res == -1) {
if (errno == EINTR) {
fuse_log(FUSE_LOG_INFO, "%s: ppoll interrupted, going around\n",
__func__);
continue;
}
fuse_log(FUSE_LOG_ERR, "virtio_loop ppoll: %m\n");
break;
}
assert(poll_res == 1);
if (pf[0].revents & (POLLERR | POLLHUP | POLLNVAL)) {
fuse_log(FUSE_LOG_ERR, "%s: Unexpected poll revents %x\n", __func__,
pf[0].revents);
break;
}
assert(pf[0].revents & POLLIN);
fuse_log(FUSE_LOG_DEBUG, "%s: Got VU event\n", __func__);
/* Mutual exclusion with fv_queue_thread() */
vu_dispatch_wrlock(se->virtio_dev);
ok = vu_dispatch(&se->virtio_dev->dev);
vu_dispatch_unlock(se->virtio_dev);
if (!ok) {
fuse_log(FUSE_LOG_ERR, "%s: vu_dispatch failed\n", __func__);
break;
}
}
/*
* Make sure all fv_queue_thread()s quit on exit, as we're about to
* free virtio dev and fuse session, no one should access them anymore.
*/
stop_all_queues(se->virtio_dev);
fuse_log(FUSE_LOG_INFO, "%s: Exit\n", __func__);
return 0;
}
static void strreplace(char *s, char old, char new)
{
for (; *s; ++s) {
if (*s == old) {
*s = new;
}
}
}
static bool fv_socket_lock(struct fuse_session *se)
{
g_autofree gchar *sk_name = NULL;
g_autofree gchar *pidfile = NULL;
g_autofree gchar *dir = NULL;
Error *local_err = NULL;
dir = qemu_get_local_state_pathname("run/virtiofsd");
if (g_mkdir_with_parents(dir, S_IRWXU) < 0) {
fuse_log(FUSE_LOG_ERR, "%s: Failed to create directory %s: %s\n",
__func__, dir, strerror(errno));
return false;
}
sk_name = g_strdup(se->vu_socket_path);
strreplace(sk_name, '/', '.');
pidfile = g_strdup_printf("%s/%s.pid", dir, sk_name);
if (!qemu_write_pidfile(pidfile, &local_err)) {
error_report_err(local_err);
return false;
}
return true;
}
static int fv_create_listen_socket(struct fuse_session *se)
{
struct sockaddr_un un;
mode_t old_umask;
/* Nothing to do if fd is already initialized */
if (se->vu_listen_fd >= 0) {
return 0;
}
if (strlen(se->vu_socket_path) >= sizeof(un.sun_path)) {
fuse_log(FUSE_LOG_ERR, "Socket path too long\n");
return -1;
}
if (!strlen(se->vu_socket_path)) {
fuse_log(FUSE_LOG_ERR, "Socket path is empty\n");
return -1;
}
/* Check the vu_socket_path is already used */
if (!fv_socket_lock(se)) {
return -1;
}
/*
* Create the Unix socket to communicate with qemu
* based on QEMU's vhost-user-bridge
*/
unlink(se->vu_socket_path);
strcpy(un.sun_path, se->vu_socket_path);
size_t addr_len = sizeof(un);
int listen_sock = socket(AF_UNIX, SOCK_STREAM, 0);
if (listen_sock == -1) {
fuse_log(FUSE_LOG_ERR, "vhost socket creation: %m\n");
return -1;
}
un.sun_family = AF_UNIX;
/*
* Unfortunately bind doesn't let you set the mask on the socket,
* so set umask appropriately and restore it later.
*/
if (se->vu_socket_group) {
old_umask = umask(S_IROTH | S_IWOTH | S_IXOTH);
} else {
old_umask = umask(S_IRGRP | S_IWGRP | S_IXGRP |
S_IROTH | S_IWOTH | S_IXOTH);
}
if (bind(listen_sock, (struct sockaddr *)&un, addr_len) == -1) {
fuse_log(FUSE_LOG_ERR, "vhost socket bind: %m\n");
close(listen_sock);
umask(old_umask);
return -1;
}
if (se->vu_socket_group) {
struct group *g = getgrnam(se->vu_socket_group);
if (g) {
if (chown(se->vu_socket_path, -1, g->gr_gid) == -1) {
fuse_log(FUSE_LOG_WARNING,
"vhost socket failed to set group to %s (%d): %m\n",
se->vu_socket_group, g->gr_gid);
}
} else {
fuse_log(FUSE_LOG_ERR,
"vhost socket: unable to find group '%s'\n",
se->vu_socket_group);
close(listen_sock);
umask(old_umask);
return -1;
}
}
umask(old_umask);
if (listen(listen_sock, 1) == -1) {
fuse_log(FUSE_LOG_ERR, "vhost socket listen: %m\n");
close(listen_sock);
return -1;
}
se->vu_listen_fd = listen_sock;
return 0;
}
int virtio_session_mount(struct fuse_session *se)
{
int ret;
/*
* Test that unshare(CLONE_FS) works. fv_queue_worker() will need it. It's
* an unprivileged system call but some Docker/Moby versions are known to
* reject it via seccomp when CAP_SYS_ADMIN is not given.
*
* Note that the program is single-threaded here so this syscall has no
* visible effect and is safe to make.
*/
ret = unshare(CLONE_FS);
if (ret == -1 && errno == EPERM) {
fuse_log(FUSE_LOG_ERR, "unshare(CLONE_FS) failed with EPERM. If "
"running in a container please check that the container "
"runtime seccomp policy allows unshare.\n");
return -1;
}
ret = fv_create_listen_socket(se);
if (ret < 0) {
return ret;
}
se->fd = -1;
fuse_log(FUSE_LOG_INFO, "%s: Waiting for vhost-user socket connection...\n",
__func__);
int data_sock = accept(se->vu_listen_fd, NULL, NULL);
if (data_sock == -1) {
fuse_log(FUSE_LOG_ERR, "vhost socket accept: %m\n");
close(se->vu_listen_fd);
return -1;
}
close(se->vu_listen_fd);
se->vu_listen_fd = -1;
fuse_log(FUSE_LOG_INFO, "%s: Received vhost-user socket connection\n",
__func__);
/* TODO: Some cleanup/deallocation! */
se->virtio_dev = g_new0(struct fv_VuDev, 1);
se->vu_socketfd = data_sock;
se->virtio_dev->se = se;
pthread_rwlock_init(&se->virtio_dev->vu_dispatch_rwlock, NULL);
if (!vu_init(&se->virtio_dev->dev, 2, se->vu_socketfd, fv_panic, NULL,
fv_set_watch, fv_remove_watch, &fv_iface)) {
fuse_log(FUSE_LOG_ERR, "%s: vu_init failed\n", __func__);
return -1;
}
return 0;
}
void virtio_session_close(struct fuse_session *se)
{
close(se->vu_socketfd);
if (!se->virtio_dev) {
return;
}
g_free(se->virtio_dev->qi);
pthread_rwlock_destroy(&se->virtio_dev->vu_dispatch_rwlock);
g_free(se->virtio_dev);
se->virtio_dev = NULL;
}