numam-spdk/lib/nbd/nbd.c
Jim Harris 0441ffcc05 nbd: do not free spdk_nbd_disk with io outstanding
There is a race condition here, where kernel could have
outstanding I/O to nbd device at the same time we terminate
the nbd application.  In this case, we cannot free the
spdk_nbd_disk since it contains the io structure that
will be referenced when the SPDK poller completes one
of those I/O.

Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I1bde240af904957f4d2bfa358dc673105d266986

Reviewed-on: https://review.gerrithub.io/385927
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
2017-11-10 17:00:58 -05:00

479 lines
10 KiB
C

/*-
* BSD LICENSE
*
* Copyright (c) Intel Corporation.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * 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.
* * Neither the name of Intel Corporation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS 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 COPYRIGHT
* OWNER 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 "spdk/stdinc.h"
#include "spdk/string.h"
#include <linux/nbd.h>
#include "spdk/nbd.h"
#include "spdk/bdev.h"
#include "spdk/endian.h"
#include "spdk/env.h"
#include "spdk/log.h"
#include "spdk/util.h"
#include "spdk/io_channel.h"
struct nbd_io {
enum spdk_bdev_io_type type;
int ref;
void *payload;
/* NOTE: for TRIM, this represents number of bytes to trim. */
uint32_t payload_size;
bool payload_in_progress;
struct nbd_request req;
bool req_in_progress;
struct nbd_reply resp;
bool resp_in_progress;
/*
* Tracks current progress on reading/writing a request,
* response, or payload from the nbd socket.
*/
uint32_t offset;
};
struct spdk_nbd_disk {
struct spdk_bdev *bdev;
struct spdk_bdev_desc *bdev_desc;
struct spdk_io_channel *ch;
int fd;
struct nbd_io io;
uint32_t buf_align;
};
static bool
is_read(enum spdk_bdev_io_type io_type)
{
if (io_type == SPDK_BDEV_IO_TYPE_READ) {
return true;
} else {
return false;
}
}
static bool
is_write(enum spdk_bdev_io_type io_type)
{
switch (io_type) {
case SPDK_BDEV_IO_TYPE_WRITE:
case SPDK_BDEV_IO_TYPE_UNMAP:
return true;
default:
return false;
}
}
static void
_nbd_stop(struct spdk_nbd_disk *nbd)
{
if (nbd->ch) {
spdk_put_io_channel(nbd->ch);
}
if (nbd->bdev_desc) {
spdk_bdev_close(nbd->bdev_desc);
}
if (nbd->fd >= 0) {
close(nbd->fd);
}
free(nbd);
}
void
spdk_nbd_stop(struct spdk_nbd_disk *nbd)
{
if (nbd == NULL) {
return;
}
nbd->io.ref--;
if (nbd->io.ref == 0) {
_nbd_stop(nbd);
}
}
static int64_t
read_from_socket(int fd, void *buf, size_t length)
{
ssize_t bytes_read;
bytes_read = read(fd, buf, length);
if (bytes_read == 0) {
return -EIO;
} else if (bytes_read == -1) {
if (errno != EAGAIN) {
return -errno;
}
return 0;
} else {
return bytes_read;
}
}
static int64_t
write_to_socket(int fd, void *buf, size_t length)
{
ssize_t bytes_written;
bytes_written = write(fd, buf, length);
if (bytes_written == 0) {
return -EIO;
} else if (bytes_written == -1) {
if (errno != EAGAIN) {
return -errno;
}
return 0;
} else {
return bytes_written;
}
}
static void
nbd_io_done(struct spdk_bdev_io *bdev_io, bool success, void *cb_arg)
{
struct nbd_io *io = cb_arg;
if (success) {
io->resp.error = 0;
} else {
to_be32(&io->resp.error, EIO);
}
io->resp_in_progress = true;
if (bdev_io != NULL) {
spdk_bdev_free_io(bdev_io);
}
io->ref--;
if (io->ref == 0) {
_nbd_stop(SPDK_CONTAINEROF(io, struct spdk_nbd_disk, io));
}
}
static void
nbd_submit_bdev_io(struct spdk_bdev *bdev, struct spdk_bdev_desc *desc,
struct spdk_io_channel *ch, struct nbd_io *io)
{
int rc;
io->ref++;
switch (io->type) {
case SPDK_BDEV_IO_TYPE_READ:
rc = spdk_bdev_read(desc, ch, io->payload, from_be64(&io->req.from),
io->payload_size, nbd_io_done, io);
break;
case SPDK_BDEV_IO_TYPE_WRITE:
rc = spdk_bdev_write(desc, ch, io->payload, from_be64(&io->req.from),
io->payload_size, nbd_io_done, io);
break;
case SPDK_BDEV_IO_TYPE_UNMAP:
rc = spdk_bdev_unmap(desc, ch, from_be64(&io->req.from),
io->payload_size, nbd_io_done, io);
break;
case SPDK_BDEV_IO_TYPE_FLUSH:
rc = spdk_bdev_flush(desc, ch, 0, spdk_bdev_get_num_blocks(bdev) * spdk_bdev_get_block_size(bdev),
nbd_io_done, io);
break;
default:
rc = -1;
break;
}
if (rc == -1) {
nbd_io_done(NULL, false, io);
}
}
static int
process_request(struct spdk_nbd_disk *nbd)
{
struct nbd_io *io = &nbd->io;
memcpy(&io->resp.handle, &io->req.handle, sizeof(io->resp.handle));
io->resp.error = 0;
io->offset = 0;
io->payload_size = from_be32(&io->req.len);
spdk_dma_free(io->payload);
io->payload = spdk_dma_malloc(io->payload_size, nbd->buf_align, NULL);
if (io->payload == NULL) {
SPDK_ERRLOG("could not allocate io->payload of size %d\n", io->payload_size);
return -ENOMEM;
}
if (from_be32(&io->req.magic) != NBD_REQUEST_MAGIC) {
SPDK_ERRLOG("invalid request magic\n");
return -EINVAL;
}
switch (from_be32(&io->req.type)) {
case NBD_CMD_READ:
io->type = SPDK_BDEV_IO_TYPE_READ;
nbd_submit_bdev_io(nbd->bdev, nbd->bdev_desc, nbd->ch, io);
break;
case NBD_CMD_WRITE:
io->type = SPDK_BDEV_IO_TYPE_WRITE;
io->payload_in_progress = true;
break;
case NBD_CMD_DISC:
return -ECONNRESET;
#ifdef NBD_FLAG_SEND_FLUSH
case NBD_CMD_FLUSH:
io->type = SPDK_BDEV_IO_TYPE_FLUSH;
nbd_submit_bdev_io(nbd->bdev, nbd->bdev_desc, nbd->ch, io);
break;
#endif
#ifdef NBD_FLAG_SEND_TRIM
case NBD_CMD_TRIM:
io->type = SPDK_BDEV_IO_TYPE_UNMAP;
nbd_submit_bdev_io(nbd->bdev, nbd->bdev_desc, nbd->ch, io);
break;
#endif
}
return 0;
}
int
spdk_nbd_poll(struct spdk_nbd_disk *nbd)
{
struct nbd_io *io = &nbd->io;
int fd = nbd->fd;
int64_t ret;
int rc;
if (io->req_in_progress) {
ret = read_from_socket(fd, (char *)&io->req + io->offset, sizeof(io->req) - io->offset);
if (ret <= 0) {
return ret;
}
io->offset += ret;
if (io->offset == sizeof(io->req)) {
io->req_in_progress = false;
rc = process_request(nbd);
if (rc != 0) {
return rc;
}
}
}
if (io->payload_in_progress && is_write(io->type)) {
ret = read_from_socket(fd, io->payload + io->offset, io->payload_size - io->offset);
if (ret <= 0) {
return ret;
}
io->offset += ret;
if (io->offset == io->payload_size) {
io->payload_in_progress = false;
nbd_submit_bdev_io(nbd->bdev, nbd->bdev_desc, nbd->ch, io);
io->offset = 0;
}
}
if (io->resp_in_progress) {
ret = write_to_socket(fd, (char *)&io->resp + io->offset, sizeof(io->resp) - io->offset);
if (ret <= 0) {
return ret;
}
io->offset += ret;
if (io->offset == sizeof(io->resp)) {
io->resp_in_progress = false;
if (is_read(io->type)) {
io->payload_in_progress = true;
} else {
io->req_in_progress = true;
}
io->offset = 0;
}
}
if (io->payload_in_progress && is_read(io->type)) {
ret = write_to_socket(fd, io->payload + io->offset, io->payload_size - io->offset);
if (ret <= 0) {
return ret;
}
io->offset += ret;
if (io->offset == io->payload_size) {
io->payload_in_progress = false;
io->req_in_progress = true;
io->offset = 0;
}
}
return 0;
}
static void
nbd_start_kernel(int nbd_fd, int *sp)
{
int rc;
char buf[64];
close(sp[0]);
rc = ioctl(nbd_fd, NBD_SET_SOCK, sp[1]);
if (rc == -1) {
spdk_strerror_r(errno, buf, sizeof(buf));
SPDK_ERRLOG("ioctl(NBD_SET_SOCK) failed: %s\n", buf);
exit(-1);
}
#ifdef NBD_FLAG_SEND_TRIM
rc = ioctl(nbd_fd, NBD_SET_FLAGS, NBD_FLAG_SEND_TRIM);
if (rc == -1) {
spdk_strerror_r(errno, buf, sizeof(buf));
SPDK_ERRLOG("ioctl(NBD_SET_FLAGS) failed: %s\n", buf);
exit(-1);
}
#endif
/* This will block in the kernel until the client disconnects. */
ioctl(nbd_fd, NBD_DO_IT);
ioctl(nbd_fd, NBD_CLEAR_QUE);
ioctl(nbd_fd, NBD_CLEAR_SOCK);
exit(0);
}
struct spdk_nbd_disk *
spdk_nbd_start(struct spdk_bdev *bdev, const char *nbd_path)
{
struct spdk_nbd_disk *nbd;
int rc;
int sp[2] = { -1, -1 }, nbd_fd = -1;
char buf[64];
nbd = calloc(1, sizeof(*nbd));
if (nbd == NULL) {
return NULL;
}
nbd->fd = -1;
rc = spdk_bdev_open(bdev, true, NULL, NULL, &nbd->bdev_desc);
if (rc != 0) {
SPDK_ERRLOG("could not open bdev %s, error=%d\n", spdk_bdev_get_name(bdev), rc);
goto err;
}
nbd->io.ref = 1;
nbd->bdev = bdev;
nbd->ch = spdk_bdev_get_io_channel(nbd->bdev_desc);
nbd->buf_align = spdk_max(spdk_bdev_get_buf_align(bdev), 64);
rc = socketpair(AF_UNIX, SOCK_STREAM, 0, sp);
if (rc != 0) {
SPDK_ERRLOG("socketpair failed\n");
goto err;
}
nbd_fd = open(nbd_path, O_RDWR);
if (nbd_fd == -1) {
spdk_strerror_r(errno, buf, sizeof(buf));
SPDK_ERRLOG("open(\"%s\") failed: %s\n", nbd_path, buf);
goto err;
}
rc = ioctl(nbd_fd, NBD_SET_BLKSIZE, spdk_bdev_get_block_size(bdev));
if (rc == -1) {
spdk_strerror_r(errno, buf, sizeof(buf));
SPDK_ERRLOG("ioctl(NBD_SET_BLKSIZE) failed: %s\n", buf);
goto err;
}
rc = ioctl(nbd_fd, NBD_SET_SIZE_BLOCKS, spdk_bdev_get_num_blocks(bdev));
if (rc == -1) {
spdk_strerror_r(errno, buf, sizeof(buf));
SPDK_ERRLOG("ioctl(NBD_SET_SIZE_BLOCKS) failed: %s\n", buf);
goto err;
}
rc = ioctl(nbd_fd, NBD_CLEAR_SOCK);
if (rc == -1) {
spdk_strerror_r(errno, buf, sizeof(buf));
SPDK_ERRLOG("ioctl(NBD_CLEAR_SOCK) failed: %s\n", buf);
goto err;
}
printf("Enabling kernel access to bdev %s via %s\n", spdk_bdev_get_name(bdev), nbd_path);
rc = fork();
switch (rc) {
case 0:
nbd_start_kernel(nbd_fd, sp);
break;
case -1:
spdk_strerror_r(errno, buf, sizeof(buf));
SPDK_ERRLOG("could not fork: %s\n", buf);
goto err;
default:
close(nbd_fd);
break;
}
close(sp[1]);
nbd->fd = sp[0];
fcntl(nbd->fd, F_SETFL, O_NONBLOCK);
to_be32(&nbd->io.resp.magic, NBD_REPLY_MAGIC);
nbd->io.req_in_progress = true;
return nbd;
err:
if (sp[0] >= 0) {
close(sp[0]);
}
if (sp[1] >= 0) {
close(sp[1]);
}
if (nbd_fd >= 0) {
close(nbd_fd);
}
spdk_nbd_stop(nbd);
return NULL;
}