From: "Daniel P. Berrangé" <berrange@redhat.com>
To: "Richard W.M. Jones" <rjones@redhat.com>
Cc: qemu-devel@nongnu.org, pbonzini@redhat.com
Subject: Re: [PATCH WIP] Implement -run-with exit-with-parent=on
Date: Wed, 1 Oct 2025 19:32:56 +0100 [thread overview]
Message-ID: <aN1z2C__6M8E7I8L@redhat.com> (raw)
In-Reply-To: <20251001174118.2756134-1-rjones@redhat.com>
On Wed, Oct 01, 2025 at 06:40:56PM +0100, Richard W.M. Jones wrote:
> Libguestfs wants to use qemu to run a captive appliance. When the
> program linked to libguestfs exits, we want qemu to be cleaned up.
> Libguestfs goes to great lengths to do this at the moment: it either
> forks a separate process to ensure clean-up is done, or it asks
> libvirt to clean up the qemu process. However this is complicated and
> not totally reliable.
>
> On Linux, FreeBSD and macOS, there are mechanisms to ensure a signal
> or message is delivered to a process when its parent process goes
> away. The qemu test suite even uses this mechanism on Linux (see
> PR_SET_PDEATHSIG in tests/qtest/libqtest.c).
>
> In nbdkit we have long had the concept of running nbdkit captively,
> and we have the nbdkit --exit-with-parent flag to help
> (https://libguestfs.org/nbdkit-captive.1.html#EXIT-WITH-PARENT)
>
> This commit adds the same mechanism that the qemu test suite uses, and
> that nbdkit has long implemented, to allow this feature to be used by
> all qemu users. The syntax is:
>
> qemu -run-with exit-with-parent=on [...]
>
> This is not a feature that most typical users of qemu (for running
> general purpose, long-lived VMs) should use, so it defaults to off.
>
> The exit-with-parent.[ch] files are copied from nbdkit, where they
> have a 3-clause BSD license which is compatible with qemu:
>
> https://gitlab.com/nbdkit/nbdkit/-/tree/master/common/utils?ref_type=heads
>
> Signed-off-by: Richard W.M. Jones <rjones@redhat.com>
> ---
> include/qemu/exit-with-parent.h | 49 +++++++++
> qemu-options.hx | 13 ++-
> system/exit-with-parent.c | 171 ++++++++++++++++++++++++++++++++
> system/meson.build | 1 +
> system/vl.c | 13 +++
> 5 files changed, 245 insertions(+), 2 deletions(-)
>
> diff --git a/include/qemu/exit-with-parent.h b/include/qemu/exit-with-parent.h
> new file mode 100644
> index 0000000000..e907ccb5c9
> --- /dev/null
> +++ b/include/qemu/exit-with-parent.h
> @@ -0,0 +1,49 @@
> +/*
> + * SPDX-License-Identifier: BSD-3-Clause
> + * nbdkit
This 'nbdkit' line can be dropped presumably, unless you want
to refer back to the original source ? Same question for the .c
> + * Copyright Red Hat
> + *
> + * 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 Red Hat 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 RED HAT 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 RED HAT 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.
> + */
IIUC, from the nbdkit history exit-with-parent.{ch} only ever
had 3 contributors, all from Red Hat. So if Red Hat is exclusive
copyright holder, I think you're be justified in removing the
license boilerplate here & below, as a representative of RH.
> +
> +#ifndef NBDKIT_EXIT_WITH_PARENT_H
> +#define NBDKIT_EXIT_WITH_PARENT_H
> +
> +#include <stdbool.h>
This is redundant in QEMU since it is guaranteed by osdep.h
> +
> +/* Test if the feature is available on the platform. */
> +extern bool can_exit_with_parent(void);
> +
> +/*
> + * --exit-with-parent: kill the current process if the parent exits.
> + * This may return -1 on error.
> + */
> +extern int set_exit_with_parent(void);
> +
> +#endif /* NBDKIT_EXIT_WITH_PARENT_H */
> diff --git a/system/exit-with-parent.c b/system/exit-with-parent.c
> new file mode 100644
> index 0000000000..fa97b586f0
> --- /dev/null
> +++ b/system/exit-with-parent.c
> @@ -0,0 +1,171 @@
> +
> +/*
> + * Implement the --exit-with-parent feature on operating systems which
> + * support it.
> + */
> +
> +#include "qemu/osdep.h"
> +#include "qemu/exit-with-parent.h"
> +
> +#ifdef CONFIG_LINUX
> +#include <sys/prctl.h>
> +#endif
> +
> +#ifdef __FreeBSD__
> +#include <sys/procctl.h>
> +#endif
> +
> +#if defined(CONFIG_LINUX) && defined(PR_SET_PDEATHSIG)
> +
> +/* For Linux >= 2.1.57. */
I don't think we need the PR_SET_PDEATHSIG check
since the oldest Linux platform QEMU targets will
always be newer than 2.1.57 and so can assume to
have PR_SET_PDEATHSIG defined.
> +
> +int
> +set_exit_with_parent(void)
> +{
> + return prctl(PR_SET_PDEATHSIG, SIGTERM);
> +}
> +
> +bool
> +can_exit_with_parent(void)
> +{
> + return true;
> +}
> +
> +#elif defined(__FreeBSD__) && defined(PROC_PDEATHSIG_CTL)
> +
> +/* For FreeBSD >= 11.2 */
Also don't need the PROC_PDEATHSIG_CTL check either,
since with current FreeBSD being 14.0, the oldest
QEMU targets is now 13.0
> +
> +#include <sys/procctl.h>
> +
> +int
> +set_exit_with_parent(void)
> +{
> + const int sig = SIGTERM;
> + return procctl(P_PID, 0, PROC_PDEATHSIG_CTL, (void *) &sig);
> +}
> +
> +bool
> +can_exit_with_parent(void)
> +{
> + return true;
> +}
> +
> +#elif defined(__APPLE__)
> +
> +/* For macOS. */
> +
> +#include <unistd.h>
> +#include <errno.h>
These two are guaranteed by osdep.h
> +#include <sys/event.h>
> +#include <pthread.h>
Use "qemu/thread.h" instead
> +
> +static void *
> +exit_with_parent_loop(void *vp)
> +{
> + const pid_t ppid = getppid();
> + int fd;
> + struct kevent kev, res[1];
> + int r;
> +
> + /* Register the kevent to wait for ppid to exit. */
> + fd = kqueue();
> + if (fd == -1) {
> + error_report("exit_with_parent_loop: kqueue: %m");
> + return NULL;
> + }
> + EV_SET(&kev, ppid, EVFILT_PROC, EV_ADD | EV_ENABLE, NOTE_EXIT, 0, NULL);
> + if (kevent(fd, &kev, 1, NULL, 0, NULL) == -1) {
> + error_report("exit_with_parent_loop: kevent: %m");
> + close(fd);
> + return NULL;
> + }
> +
> + /* Wait for the kevent to happen. */
> + r = kevent(fd, 0, 0, res, 1, NULL);
> + if (r == 1 && res[0].ident == ppid) {
> + /* Shut down the whole process when the parent dies. */
> + exit(0);
> + }
> +
> + return NULL;
> +}
> +
> +int
> +set_exit_with_parent(void)
> +{
> + int r;
> + pthread_attr_t attrs;
> + pthread_t exit_with_parent_thread;
> +
> + /*
> + * We have to block waiting for kevent, so that requires that we
> + * start a background thread.
> + */
> + pthread_attr_init(&attrs);
> + pthread_attr_setdetachstate(&attrs, PTHREAD_CREATE_DETACHED);
> + r = pthread_create(&exit_with_parent_thread, NULL,
> + exit_with_parent_loop, NULL);
QemuThread exit_with_parent_thread;
qemu_thread_create(&exit_with_parent_thread,
"exit-parent",
exit_with_parent_loop,
QEMU_THREAD_DETACHED);
mostly just so we get the thread name set.
> + if (r != 0) {
> + errno = r;
> + return -1;
> + }
> +
> + return 0;
> +}
> +
> +bool
> +can_exit_with_parent(void)
> +{
> + return true;
> +}
> +
> +#else /* any platform that doesn't support this function */
> +
> +#include <stdlib.h>
Can be assumed with osdep.h
> +
> +int
> +set_exit_with_parent(void)
> +{
> + abort();
Slightly better to use g_assert_not_reached()
> +}
> +
> +bool
> +can_exit_with_parent(void)
> +{
> + return false;
> +}
> +
> +#endif
> diff --git a/system/vl.c b/system/vl.c
> index 00f3694725..47b6e60b94 100644
> --- a/system/vl.c
> +++ b/system/vl.c
> @@ -53,6 +53,7 @@
> #include "qemu/sockets.h"
> #include "qemu/accel.h"
> #include "qemu/async-teardown.h"
> +#include "qemu/exit-with-parent.h"
> #include "hw/usb.h"
> #include "hw/isa/isa.h"
> #include "hw/scsi/scsi.h"
> @@ -783,6 +784,10 @@ static QemuOptsList qemu_run_with_opts = {
> .name = "chroot",
> .type = QEMU_OPT_STRING,
> },
> + {
> + .name = "exit-with-parent",
> + .type = QEMU_OPT_BOOL,
> + },
> {
> .name = "user",
> .type = QEMU_OPT_STRING,
> @@ -3690,6 +3695,14 @@ void qemu_init(int argc, char **argv)
> if (str) {
> os_set_chroot(str);
> }
> + if (qemu_opt_get_bool(opts, "exit-with-parent", false)) {
> + if (!can_exit_with_parent()) {
> + error_report("exit-with-parent is not available"
> + " on this platform");
> + exit(1);
> + }
> + set_exit_with_parent();
> + }
> str = qemu_opt_get(opts, "user");
> if (str) {
> if (!os_set_runas(str)) {
Broadly looks good to me.
With regards,
Daniel
--
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org -o- https://fstop138.berrange.com :|
|: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
next prev parent reply other threads:[~2025-10-01 18:35 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-10-01 17:40 [PATCH WIP] Implement -run-with exit-with-parent=on Richard W.M. Jones
2025-10-01 18:32 ` Daniel P. Berrangé [this message]
2025-10-01 19:14 ` Richard W.M. Jones
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=aN1z2C__6M8E7I8L@redhat.com \
--to=berrange@redhat.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=rjones@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).