mirror of
https://github.com/qemu/qemu.git
synced 2024-11-23 10:53:37 +08:00
06e0f098d6
The ongoing QEMU multi-queue block layer effort makes it possible for multiple threads to process I/O in parallel. The nbd block driver is not compatible with the multi-queue block layer yet because QIOChannel cannot be used easily from coroutines running in multiple threads. This series changes the QIOChannel API to make that possible. In the current API, calling qio_channel_attach_aio_context() sets the AioContext where qio_channel_yield() installs an fd handler prior to yielding: qio_channel_attach_aio_context(ioc, my_ctx); ... qio_channel_yield(ioc); // my_ctx is used here ... qio_channel_detach_aio_context(ioc); This API design has limitations: reading and writing must be done in the same AioContext and moving between AioContexts involves a cumbersome sequence of API calls that is not suitable for doing on a per-request basis. There is no fundamental reason why a QIOChannel needs to run within the same AioContext every time qio_channel_yield() is called. QIOChannel only uses the AioContext while inside qio_channel_yield(). The rest of the time, QIOChannel is independent of any AioContext. In the new API, qio_channel_yield() queries the AioContext from the current coroutine using qemu_coroutine_get_aio_context(). There is no need to explicitly attach/detach AioContexts anymore and qio_channel_attach_aio_context() and qio_channel_detach_aio_context() are gone. One coroutine can read from the QIOChannel while another coroutine writes from a different AioContext. This API change allows the nbd block driver to use QIOChannel from any thread. It's important to keep in mind that the block driver already synchronizes QIOChannel access and ensures that two coroutines never read simultaneously or write simultaneously. This patch updates all users of qio_channel_attach_aio_context() to the new API. Most conversions are simple, but vhost-user-server requires a new qemu_coroutine_yield() call to quiesce the vu_client_trip() coroutine when not attached to any AioContext. While the API is has become simpler, there is one wart: QIOChannel has a special case for the iohandler AioContext (used for handlers that must not run in nested event loops). I didn't find an elegant way preserve that behavior, so I added a new API called qio_channel_set_follow_coroutine_ctx(ioc, true|false) for opting in to the new AioContext model. By default QIOChannel uses the iohandler AioHandler. Code that formerly called qio_channel_attach_aio_context() now calls qio_channel_set_follow_coroutine_ctx(ioc, true) once after the QIOChannel is created. Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Acked-by: Daniel P. Berrangé <berrange@redhat.com> Message-ID: <20230830224802.493686-5-stefanha@redhat.com> [eblake: also fix migration/rdma.c] Signed-off-by: Eric Blake <eblake@redhat.com>
388 lines
11 KiB
C
388 lines
11 KiB
C
/*
|
|
* QEMU I/O channels external command driver
|
|
*
|
|
* Copyright (c) 2015 Red Hat, Inc.
|
|
*
|
|
* This library is free software; you can redistribute it and/or
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
* License as published by the Free Software Foundation; either
|
|
* version 2.1 of the License, or (at your option) any later version.
|
|
*
|
|
* This library is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
* Lesser General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU Lesser General Public
|
|
* License along with this library; if not, see <http://www.gnu.org/licenses/>.
|
|
*
|
|
*/
|
|
|
|
#include "qemu/osdep.h"
|
|
#include "io/channel-command.h"
|
|
#include "io/channel-util.h"
|
|
#include "io/channel-watch.h"
|
|
#include "qapi/error.h"
|
|
#include "qemu/module.h"
|
|
#include "qemu/sockets.h"
|
|
#include "trace.h"
|
|
|
|
/**
|
|
* qio_channel_command_new_pid:
|
|
* @writefd: the FD connected to the command's stdin
|
|
* @readfd: the FD connected to the command's stdout
|
|
* @pid: the PID/HANDLE of the running child command
|
|
* @errp: pointer to a NULL-initialized error object
|
|
*
|
|
* Create a channel for performing I/O with the
|
|
* previously spawned command identified by @pid.
|
|
* The two file descriptors provide the connection
|
|
* to command's stdio streams, either one or which
|
|
* may be -1 to indicate that stream is not open.
|
|
*
|
|
* The channel will take ownership of the process
|
|
* @pid and will kill it when closing the channel.
|
|
* Similarly it will take responsibility for
|
|
* closing the file descriptors @writefd and @readfd.
|
|
*
|
|
* Returns: the command channel object, or NULL on error
|
|
*/
|
|
static QIOChannelCommand *
|
|
qio_channel_command_new_pid(int writefd,
|
|
int readfd,
|
|
GPid pid)
|
|
{
|
|
QIOChannelCommand *ioc;
|
|
|
|
ioc = QIO_CHANNEL_COMMAND(object_new(TYPE_QIO_CHANNEL_COMMAND));
|
|
|
|
ioc->readfd = readfd;
|
|
ioc->writefd = writefd;
|
|
ioc->pid = pid;
|
|
|
|
trace_qio_channel_command_new_pid(ioc, writefd, readfd,
|
|
#ifdef WIN32
|
|
GetProcessId(pid)
|
|
#else
|
|
pid
|
|
#endif
|
|
);
|
|
return ioc;
|
|
}
|
|
|
|
QIOChannelCommand *
|
|
qio_channel_command_new_spawn(const char *const argv[],
|
|
int flags,
|
|
Error **errp)
|
|
{
|
|
g_autoptr(GError) err = NULL;
|
|
GPid pid = 0;
|
|
GSpawnFlags gflags = G_SPAWN_CLOEXEC_PIPES | G_SPAWN_DO_NOT_REAP_CHILD;
|
|
int stdinfd = -1, stdoutfd = -1;
|
|
|
|
flags = flags & O_ACCMODE;
|
|
gflags |= flags == O_WRONLY ? G_SPAWN_STDOUT_TO_DEV_NULL : 0;
|
|
|
|
if (!g_spawn_async_with_pipes(NULL, (char **)argv, NULL, gflags, NULL, NULL,
|
|
&pid,
|
|
flags == O_RDONLY ? NULL : &stdinfd,
|
|
flags == O_WRONLY ? NULL : &stdoutfd,
|
|
NULL, &err)) {
|
|
error_setg(errp, "%s", err->message);
|
|
return NULL;
|
|
}
|
|
|
|
return qio_channel_command_new_pid(stdinfd, stdoutfd, pid);
|
|
}
|
|
|
|
#ifndef WIN32
|
|
static int qio_channel_command_abort(QIOChannelCommand *ioc,
|
|
Error **errp)
|
|
{
|
|
pid_t ret;
|
|
int status;
|
|
int step = 0;
|
|
|
|
/* See if intermediate process has exited; if not, try a nice
|
|
* SIGTERM followed by a more severe SIGKILL.
|
|
*/
|
|
rewait:
|
|
trace_qio_channel_command_abort(ioc, ioc->pid);
|
|
ret = waitpid(ioc->pid, &status, WNOHANG);
|
|
trace_qio_channel_command_wait(ioc, ioc->pid, ret, status);
|
|
if (ret == (pid_t)-1) {
|
|
if (errno == EINTR) {
|
|
goto rewait;
|
|
} else {
|
|
error_setg_errno(errp, errno,
|
|
"Cannot wait on pid %llu",
|
|
(unsigned long long)ioc->pid);
|
|
return -1;
|
|
}
|
|
} else if (ret == 0) {
|
|
if (step == 0) {
|
|
kill(ioc->pid, SIGTERM);
|
|
} else if (step == 1) {
|
|
kill(ioc->pid, SIGKILL);
|
|
} else {
|
|
error_setg(errp,
|
|
"Process %llu refused to die",
|
|
(unsigned long long)ioc->pid);
|
|
return -1;
|
|
}
|
|
step++;
|
|
usleep(10 * 1000);
|
|
goto rewait;
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
#else
|
|
static int qio_channel_command_abort(QIOChannelCommand *ioc,
|
|
Error **errp)
|
|
{
|
|
DWORD ret;
|
|
|
|
TerminateProcess(ioc->pid, 0);
|
|
ret = WaitForSingleObject(ioc->pid, 1000);
|
|
if (ret != WAIT_OBJECT_0) {
|
|
error_setg(errp,
|
|
"Process %llu refused to die",
|
|
(unsigned long long)GetProcessId(ioc->pid));
|
|
return -1;
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
#endif /* ! WIN32 */
|
|
|
|
|
|
static void qio_channel_command_init(Object *obj)
|
|
{
|
|
QIOChannelCommand *ioc = QIO_CHANNEL_COMMAND(obj);
|
|
ioc->readfd = -1;
|
|
ioc->writefd = -1;
|
|
ioc->pid = 0;
|
|
}
|
|
|
|
static void qio_channel_command_finalize(Object *obj)
|
|
{
|
|
QIOChannelCommand *ioc = QIO_CHANNEL_COMMAND(obj);
|
|
if (ioc->readfd != -1) {
|
|
close(ioc->readfd);
|
|
}
|
|
if (ioc->writefd != -1 &&
|
|
ioc->writefd != ioc->readfd) {
|
|
close(ioc->writefd);
|
|
}
|
|
ioc->writefd = ioc->readfd = -1;
|
|
if (ioc->pid > 0) {
|
|
qio_channel_command_abort(ioc, NULL);
|
|
g_spawn_close_pid(ioc->pid);
|
|
}
|
|
}
|
|
|
|
#ifdef WIN32
|
|
static bool win32_fd_poll(int fd, gushort events)
|
|
{
|
|
GPollFD pfd = { .fd = _get_osfhandle(fd), .events = events };
|
|
int res;
|
|
|
|
do {
|
|
res = g_poll(&pfd, 1, 0);
|
|
} while (res < 0 && errno == EINTR);
|
|
if (res == 0) {
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
#endif
|
|
|
|
static ssize_t qio_channel_command_readv(QIOChannel *ioc,
|
|
const struct iovec *iov,
|
|
size_t niov,
|
|
int **fds,
|
|
size_t *nfds,
|
|
int flags,
|
|
Error **errp)
|
|
{
|
|
QIOChannelCommand *cioc = QIO_CHANNEL_COMMAND(ioc);
|
|
ssize_t ret;
|
|
|
|
#ifdef WIN32
|
|
if (!cioc->blocking && !win32_fd_poll(cioc->readfd, G_IO_IN)) {
|
|
return QIO_CHANNEL_ERR_BLOCK;
|
|
}
|
|
#endif
|
|
|
|
retry:
|
|
ret = readv(cioc->readfd, iov, niov);
|
|
if (ret < 0) {
|
|
if (errno == EAGAIN) {
|
|
return QIO_CHANNEL_ERR_BLOCK;
|
|
}
|
|
if (errno == EINTR) {
|
|
goto retry;
|
|
}
|
|
|
|
error_setg_errno(errp, errno,
|
|
"Unable to read from command");
|
|
return -1;
|
|
}
|
|
|
|
return ret;
|
|
}
|
|
|
|
static ssize_t qio_channel_command_writev(QIOChannel *ioc,
|
|
const struct iovec *iov,
|
|
size_t niov,
|
|
int *fds,
|
|
size_t nfds,
|
|
int flags,
|
|
Error **errp)
|
|
{
|
|
QIOChannelCommand *cioc = QIO_CHANNEL_COMMAND(ioc);
|
|
ssize_t ret;
|
|
|
|
#ifdef WIN32
|
|
if (!cioc->blocking && !win32_fd_poll(cioc->writefd, G_IO_OUT)) {
|
|
return QIO_CHANNEL_ERR_BLOCK;
|
|
}
|
|
#endif
|
|
|
|
retry:
|
|
ret = writev(cioc->writefd, iov, niov);
|
|
if (ret <= 0) {
|
|
if (errno == EAGAIN) {
|
|
return QIO_CHANNEL_ERR_BLOCK;
|
|
}
|
|
if (errno == EINTR) {
|
|
goto retry;
|
|
}
|
|
error_setg_errno(errp, errno, "%s",
|
|
"Unable to write to command");
|
|
return -1;
|
|
}
|
|
return ret;
|
|
}
|
|
|
|
static int qio_channel_command_set_blocking(QIOChannel *ioc,
|
|
bool enabled,
|
|
Error **errp)
|
|
{
|
|
QIOChannelCommand *cioc = QIO_CHANNEL_COMMAND(ioc);
|
|
|
|
#ifdef WIN32
|
|
cioc->blocking = enabled;
|
|
#else
|
|
|
|
if ((cioc->writefd >= 0 && !g_unix_set_fd_nonblocking(cioc->writefd, !enabled, NULL)) ||
|
|
(cioc->readfd >= 0 && !g_unix_set_fd_nonblocking(cioc->readfd, !enabled, NULL))) {
|
|
error_setg_errno(errp, errno, "Failed to set FD nonblocking");
|
|
return -1;
|
|
}
|
|
#endif
|
|
return 0;
|
|
}
|
|
|
|
|
|
static int qio_channel_command_close(QIOChannel *ioc,
|
|
Error **errp)
|
|
{
|
|
QIOChannelCommand *cioc = QIO_CHANNEL_COMMAND(ioc);
|
|
int rv = 0;
|
|
#ifndef WIN32
|
|
pid_t wp;
|
|
#endif
|
|
|
|
/* We close FDs before killing, because that
|
|
* gives a better chance of clean shutdown
|
|
*/
|
|
if (cioc->readfd != -1 &&
|
|
close(cioc->readfd) < 0) {
|
|
rv = -1;
|
|
}
|
|
if (cioc->writefd != -1 &&
|
|
cioc->writefd != cioc->readfd &&
|
|
close(cioc->writefd) < 0) {
|
|
rv = -1;
|
|
}
|
|
cioc->writefd = cioc->readfd = -1;
|
|
|
|
#ifndef WIN32
|
|
do {
|
|
wp = waitpid(cioc->pid, NULL, 0);
|
|
} while (wp == (pid_t)-1 && errno == EINTR);
|
|
if (wp == (pid_t)-1) {
|
|
error_setg_errno(errp, errno, "Failed to wait for pid %llu",
|
|
(unsigned long long)cioc->pid);
|
|
return -1;
|
|
}
|
|
#else
|
|
WaitForSingleObject(cioc->pid, INFINITE);
|
|
#endif
|
|
|
|
if (rv < 0) {
|
|
error_setg_errno(errp, errno, "%s",
|
|
"Unable to close command");
|
|
}
|
|
return rv;
|
|
}
|
|
|
|
|
|
static void qio_channel_command_set_aio_fd_handler(QIOChannel *ioc,
|
|
AioContext *read_ctx,
|
|
IOHandler *io_read,
|
|
AioContext *write_ctx,
|
|
IOHandler *io_write,
|
|
void *opaque)
|
|
{
|
|
QIOChannelCommand *cioc = QIO_CHANNEL_COMMAND(ioc);
|
|
|
|
qio_channel_util_set_aio_fd_handler(cioc->readfd, read_ctx, io_read,
|
|
cioc->writefd, write_ctx, io_write,
|
|
opaque);
|
|
}
|
|
|
|
|
|
static GSource *qio_channel_command_create_watch(QIOChannel *ioc,
|
|
GIOCondition condition)
|
|
{
|
|
QIOChannelCommand *cioc = QIO_CHANNEL_COMMAND(ioc);
|
|
return qio_channel_create_fd_pair_watch(ioc,
|
|
cioc->readfd,
|
|
cioc->writefd,
|
|
condition);
|
|
}
|
|
|
|
|
|
static void qio_channel_command_class_init(ObjectClass *klass,
|
|
void *class_data G_GNUC_UNUSED)
|
|
{
|
|
QIOChannelClass *ioc_klass = QIO_CHANNEL_CLASS(klass);
|
|
|
|
ioc_klass->io_writev = qio_channel_command_writev;
|
|
ioc_klass->io_readv = qio_channel_command_readv;
|
|
ioc_klass->io_set_blocking = qio_channel_command_set_blocking;
|
|
ioc_klass->io_close = qio_channel_command_close;
|
|
ioc_klass->io_create_watch = qio_channel_command_create_watch;
|
|
ioc_klass->io_set_aio_fd_handler = qio_channel_command_set_aio_fd_handler;
|
|
}
|
|
|
|
static const TypeInfo qio_channel_command_info = {
|
|
.parent = TYPE_QIO_CHANNEL,
|
|
.name = TYPE_QIO_CHANNEL_COMMAND,
|
|
.instance_size = sizeof(QIOChannelCommand),
|
|
.instance_init = qio_channel_command_init,
|
|
.instance_finalize = qio_channel_command_finalize,
|
|
.class_init = qio_channel_command_class_init,
|
|
};
|
|
|
|
static void qio_channel_command_register_types(void)
|
|
{
|
|
type_register_static(&qio_channel_command_info);
|
|
}
|
|
|
|
type_init(qio_channel_command_register_types);
|