| Message ID | 20200312104142.21259-1-alex.bennee@linaro.org |
|---|---|
| State | Accepted |
| Commit | f6698f2b03b0db76aff7298ed4de1f9a0e22cc26 |
| Headers | show |
| Series | tools/virtiofsd: add support for --socket-group | expand |
On Thu, Mar 12, 2020 at 10:41:42AM +0000, Alex Bennée wrote: > If you like running QEMU as a normal user (very common for TCG runs) > but you have to run virtiofsd as a root user you run into connection > problems. Adding support for an optional --socket-group allows the > users to keep using the command line. If we're going to support this, then I think we need to put it in the vhost-user.rst specification so we standardize across backends. > Signed-off-by: Alex Bennée <alex.bennee@linaro.org> > > --- > v1 > - tweak documentation and commentary > --- > docs/tools/virtiofsd.rst | 4 ++++ > tools/virtiofsd/fuse_i.h | 1 + > tools/virtiofsd/fuse_lowlevel.c | 6 ++++++ > tools/virtiofsd/fuse_virtio.c | 20 ++++++++++++++++++-- > 4 files changed, 29 insertions(+), 2 deletions(-) > > diff --git a/docs/tools/virtiofsd.rst b/docs/tools/virtiofsd.rst > index 378594c422a..5a8246b74f8 100644 > --- a/docs/tools/virtiofsd.rst > +++ b/docs/tools/virtiofsd.rst > @@ -85,6 +85,10 @@ Options > > Listen on vhost-user UNIX domain socket at PATH. > > +.. option:: --socket-group=GROUP > + > + Set the vhost-user UNIX domain socket gid to GROUP. > + > .. option:: --fd=FDNUM > > Accept connections from vhost-user UNIX domain socket file descriptor FDNUM. > diff --git a/tools/virtiofsd/fuse_i.h b/tools/virtiofsd/fuse_i.h > index 1240828208a..492e002181e 100644 > --- a/tools/virtiofsd/fuse_i.h > +++ b/tools/virtiofsd/fuse_i.h > @@ -68,6 +68,7 @@ struct fuse_session { > size_t bufsize; > int error; > char *vu_socket_path; > + char *vu_socket_group; > int vu_listen_fd; > int vu_socketfd; > struct fv_VuDev *virtio_dev; > diff --git a/tools/virtiofsd/fuse_lowlevel.c b/tools/virtiofsd/fuse_lowlevel.c > index 2dd36ec03b6..4d1ba2925d1 100644 > --- a/tools/virtiofsd/fuse_lowlevel.c > +++ b/tools/virtiofsd/fuse_lowlevel.c > @@ -2523,6 +2523,7 @@ static const struct fuse_opt fuse_ll_opts[] = { > LL_OPTION("--debug", debug, 1), > LL_OPTION("allow_root", deny_others, 1), > LL_OPTION("--socket-path=%s", vu_socket_path, 0), > + LL_OPTION("--socket-group=%s", vu_socket_group, 0), > LL_OPTION("--fd=%d", vu_listen_fd, 0), > LL_OPTION("--thread-pool-size=%d", thread_pool_size, 0), > FUSE_OPT_END > @@ -2630,6 +2631,11 @@ struct fuse_session *fuse_session_new(struct fuse_args *args, > "fuse: --socket-path and --fd cannot be given together\n"); > goto out4; > } > + if (se->vu_socket_group && !se->vu_socket_path) { > + fuse_log(FUSE_LOG_ERR, > + "fuse: --socket-group can only be used with --socket-path\n"); > + goto out4; > + } > > se->bufsize = FUSE_MAX_MAX_PAGES * getpagesize() + FUSE_BUFFER_HEADER_SIZE; > > diff --git a/tools/virtiofsd/fuse_virtio.c b/tools/virtiofsd/fuse_virtio.c > index 3b6d16a0417..331f9fc65c5 100644 > --- a/tools/virtiofsd/fuse_virtio.c > +++ b/tools/virtiofsd/fuse_virtio.c > @@ -31,6 +31,8 @@ > #include <sys/socket.h> > #include <sys/types.h> > #include <sys/un.h> > +#include <sys/types.h> > +#include <grp.h> > #include <unistd.h> > > #include "contrib/libvhost-user/libvhost-user.h" > @@ -924,15 +926,29 @@ static int fv_create_listen_socket(struct fuse_session *se) > > /* > * Unfortunately bind doesn't let you set the mask on the socket, > - * so set umask to 077 and restore it later. > + * so set umask appropriately and restore it later. > */ > - old_umask = umask(0077); > + 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)) { > + fuse_log(FUSE_LOG_WARNING, > + "vhost socket failed to set group to %s (%d)\n", > + se->vu_socket_group, g->gr_gid); > + } > + } > + } > umask(old_umask); > > if (listen(listen_sock, 1) == -1) { > -- > 2.20.1 > > 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 :|
On Thu, Mar 12, 2020 at 10:41:42AM +0000, Alex Bennée wrote: > If you like running QEMU as a normal user (very common for TCG runs) > but you have to run virtiofsd as a root user you run into connection > problems. Adding support for an optional --socket-group allows the > users to keep using the command line. > > Signed-off-by: Alex Bennée <alex.bennee@linaro.org> > > --- > v1 > - tweak documentation and commentary > --- > docs/tools/virtiofsd.rst | 4 ++++ > tools/virtiofsd/fuse_i.h | 1 + > tools/virtiofsd/fuse_lowlevel.c | 6 ++++++ > tools/virtiofsd/fuse_virtio.c | 20 ++++++++++++++++++-- > 4 files changed, 29 insertions(+), 2 deletions(-) Dan's suggestion sounds like a good idea to me. Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Hi On Thu, Mar 12, 2020 at 11:49 AM Daniel P. Berrangé <berrange@redhat.com> wrote: > > On Thu, Mar 12, 2020 at 10:41:42AM +0000, Alex Bennée wrote: > > If you like running QEMU as a normal user (very common for TCG runs) > > but you have to run virtiofsd as a root user you run into connection > > problems. Adding support for an optional --socket-group allows the > > users to keep using the command line. > > If we're going to support this, then I think we need to put it in > the vhost-user.rst specification so we standardize across backends. > > Perhaps. Otoh, I wonder if the backend spec should be more limited to arguments/introspection that are used by programs. In this case, I even consider --socket-path to be unnecessary, as a management layer can/should provide a preopened & setup fd directly. What do you think? > > > Signed-off-by: Alex Bennée <alex.bennee@linaro.org> > > > > > > --- > > v1 > > - tweak documentation and commentary > > --- > > docs/tools/virtiofsd.rst | 4 ++++ > > tools/virtiofsd/fuse_i.h | 1 + > > tools/virtiofsd/fuse_lowlevel.c | 6 ++++++ > > tools/virtiofsd/fuse_virtio.c | 20 ++++++++++++++++++-- > > 4 files changed, 29 insertions(+), 2 deletions(-) > > > > diff --git a/docs/tools/virtiofsd.rst b/docs/tools/virtiofsd.rst > > index 378594c422a..5a8246b74f8 100644 > > --- a/docs/tools/virtiofsd.rst > > +++ b/docs/tools/virtiofsd.rst > > @@ -85,6 +85,10 @@ Options > > > > Listen on vhost-user UNIX domain socket at PATH. > > > > +.. option:: --socket-group=GROUP > > + > > + Set the vhost-user UNIX domain socket gid to GROUP. > > + > > .. option:: --fd=FDNUM > > > > Accept connections from vhost-user UNIX domain socket file descriptor FDNUM. > > diff --git a/tools/virtiofsd/fuse_i.h b/tools/virtiofsd/fuse_i.h > > index 1240828208a..492e002181e 100644 > > --- a/tools/virtiofsd/fuse_i.h > > +++ b/tools/virtiofsd/fuse_i.h > > @@ -68,6 +68,7 @@ struct fuse_session { > > size_t bufsize; > > int error; > > char *vu_socket_path; > > + char *vu_socket_group; > > int vu_listen_fd; > > int vu_socketfd; > > struct fv_VuDev *virtio_dev; > > diff --git a/tools/virtiofsd/fuse_lowlevel.c b/tools/virtiofsd/fuse_lowlevel.c > > index 2dd36ec03b6..4d1ba2925d1 100644 > > --- a/tools/virtiofsd/fuse_lowlevel.c > > +++ b/tools/virtiofsd/fuse_lowlevel.c > > @@ -2523,6 +2523,7 @@ static const struct fuse_opt fuse_ll_opts[] = { > > LL_OPTION("--debug", debug, 1), > > LL_OPTION("allow_root", deny_others, 1), > > LL_OPTION("--socket-path=%s", vu_socket_path, 0), > > + LL_OPTION("--socket-group=%s", vu_socket_group, 0), > > LL_OPTION("--fd=%d", vu_listen_fd, 0), > > LL_OPTION("--thread-pool-size=%d", thread_pool_size, 0), > > FUSE_OPT_END > > @@ -2630,6 +2631,11 @@ struct fuse_session *fuse_session_new(struct fuse_args *args, > > "fuse: --socket-path and --fd cannot be given together\n"); > > goto out4; > > } > > + if (se->vu_socket_group && !se->vu_socket_path) { > > + fuse_log(FUSE_LOG_ERR, > > + "fuse: --socket-group can only be used with --socket-path\n"); > > + goto out4; > > + } > > > > se->bufsize = FUSE_MAX_MAX_PAGES * getpagesize() + FUSE_BUFFER_HEADER_SIZE; > > > > diff --git a/tools/virtiofsd/fuse_virtio.c b/tools/virtiofsd/fuse_virtio.c > > index 3b6d16a0417..331f9fc65c5 100644 > > --- a/tools/virtiofsd/fuse_virtio.c > > +++ b/tools/virtiofsd/fuse_virtio.c > > @@ -31,6 +31,8 @@ > > #include <sys/socket.h> > > #include <sys/types.h> > > #include <sys/un.h> > > +#include <sys/types.h> > > +#include <grp.h> > > #include <unistd.h> > > > > #include "contrib/libvhost-user/libvhost-user.h" > > @@ -924,15 +926,29 @@ static int fv_create_listen_socket(struct fuse_session *se) > > > > /* > > * Unfortunately bind doesn't let you set the mask on the socket, > > - * so set umask to 077 and restore it later. > > + * so set umask appropriately and restore it later. > > */ > > - old_umask = umask(0077); > > + 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)) { > > + fuse_log(FUSE_LOG_WARNING, > > + "vhost socket failed to set group to %s (%d)\n", > > + se->vu_socket_group, g->gr_gid); > > + } > > + } > > + } > > umask(old_umask); > > > > if (listen(listen_sock, 1) == -1) { > > -- > > 2.20.1 > > > > > > 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 :| > > -- Marc-André Lureau
On Sat, Mar 14, 2020 at 02:33:25PM +0100, Marc-André Lureau wrote: > Hi > > On Thu, Mar 12, 2020 at 11:49 AM Daniel P. Berrangé <berrange@redhat.com> wrote: > > > > On Thu, Mar 12, 2020 at 10:41:42AM +0000, Alex Bennée wrote: > > > If you like running QEMU as a normal user (very common for TCG runs) > > > but you have to run virtiofsd as a root user you run into connection > > > problems. Adding support for an optional --socket-group allows the > > > users to keep using the command line. > > > > If we're going to support this, then I think we need to put it in > > the vhost-user.rst specification so we standardize across backends. > > > > > > Perhaps. Otoh, I wonder if the backend spec should be more limited to > arguments/introspection that are used by programs. > > In this case, I even consider --socket-path to be unnecessary, as a > management layer can/should provide a preopened & setup fd directly. > > What do you think? I think there's value in standardization even if it is an option targetted at human admins, rather than machine usage. You are right though that something like libvirt would never use --socket-group, or --socket-path. Even admins would benefit if all programs followed the same naming for these. We could document such options as "SHOULD" rather than "MUST" IOW, we don't mandate --socket-group, but if you're going to provide a way to control socket group, this option should be used. 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 :|
On Mon, Mar 16, 2020 at 10:33:31AM +0000, Daniel P. Berrangé wrote: > On Sat, Mar 14, 2020 at 02:33:25PM +0100, Marc-André Lureau wrote: > > Hi > > > > On Thu, Mar 12, 2020 at 11:49 AM Daniel P. Berrangé <berrange@redhat.com> wrote: > > > > > > On Thu, Mar 12, 2020 at 10:41:42AM +0000, Alex Bennée wrote: > > > > If you like running QEMU as a normal user (very common for TCG runs) > > > > but you have to run virtiofsd as a root user you run into connection > > > > problems. Adding support for an optional --socket-group allows the > > > > users to keep using the command line. > > > > > > If we're going to support this, then I think we need to put it in > > > the vhost-user.rst specification so we standardize across backends. > > > > > > > > > > Perhaps. Otoh, I wonder if the backend spec should be more limited to > > arguments/introspection that are used by programs. > > > > In this case, I even consider --socket-path to be unnecessary, as a > > management layer can/should provide a preopened & setup fd directly. > > > > What do you think? > > I think there's value in standardization even if it is an option targetted > at human admins, rather than machine usage. You are right though that > something like libvirt would never use --socket-group, or --socket-path. > Even admins would benefit if all programs followed the same naming for > these. We could document such options as "SHOULD" rather than "MUST" > IOW, we don't mandate --socket-group, but if you're going to provide a > way to control socket group, this option should be used. I agree. It's still useful to have a convention that most vhost-user backend programs follow. Stefan
* Stefan Hajnoczi (stefanha@redhat.com) wrote: > On Mon, Mar 16, 2020 at 10:33:31AM +0000, Daniel P. Berrangé wrote: > > On Sat, Mar 14, 2020 at 02:33:25PM +0100, Marc-André Lureau wrote: > > > Hi > > > > > > On Thu, Mar 12, 2020 at 11:49 AM Daniel P. Berrangé <berrange@redhat.com> wrote: > > > > > > > > On Thu, Mar 12, 2020 at 10:41:42AM +0000, Alex Bennée wrote: > > > > > If you like running QEMU as a normal user (very common for TCG runs) > > > > > but you have to run virtiofsd as a root user you run into connection > > > > > problems. Adding support for an optional --socket-group allows the > > > > > users to keep using the command line. > > > > > > > > If we're going to support this, then I think we need to put it in > > > > the vhost-user.rst specification so we standardize across backends. > > > > > > > > > > > > > > Perhaps. Otoh, I wonder if the backend spec should be more limited to > > > arguments/introspection that are used by programs. > > > > > > In this case, I even consider --socket-path to be unnecessary, as a > > > management layer can/should provide a preopened & setup fd directly. > > > > > > What do you think? > > > > I think there's value in standardization even if it is an option targetted > > at human admins, rather than machine usage. You are right though that > > something like libvirt would never use --socket-group, or --socket-path. > > Even admins would benefit if all programs followed the same naming for > > these. We could document such options as "SHOULD" rather than "MUST" > > IOW, we don't mandate --socket-group, but if you're going to provide a > > way to control socket group, this option should be used. > > I agree. It's still useful to have a convention that most vhost-user > backend programs follow. Alex: Can you add the doc entry that Stefan and Marc-André are asking for; it's probably good they go together. Dave > Stefan -- Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
Dr. David Alan Gilbert <dgilbert@redhat.com> writes: > * Stefan Hajnoczi (stefanha@redhat.com) wrote: >> On Mon, Mar 16, 2020 at 10:33:31AM +0000, Daniel P. Berrangé wrote: >> > On Sat, Mar 14, 2020 at 02:33:25PM +0100, Marc-André Lureau wrote: >> > > Hi >> > > >> > > On Thu, Mar 12, 2020 at 11:49 AM Daniel P. Berrangé <berrange@redhat.com> wrote: >> > > > >> > > > On Thu, Mar 12, 2020 at 10:41:42AM +0000, Alex Bennée wrote: >> > > > > If you like running QEMU as a normal user (very common for TCG runs) >> > > > > but you have to run virtiofsd as a root user you run into connection >> > > > > problems. Adding support for an optional --socket-group allows the >> > > > > users to keep using the command line. >> > > > >> > > > If we're going to support this, then I think we need to put it in >> > > > the vhost-user.rst specification so we standardize across backends. >> > > > >> > > > >> > > >> > > Perhaps. Otoh, I wonder if the backend spec should be more limited to >> > > arguments/introspection that are used by programs. >> > > >> > > In this case, I even consider --socket-path to be unnecessary, as a >> > > management layer can/should provide a preopened & setup fd directly. >> > > >> > > What do you think? >> > >> > I think there's value in standardization even if it is an option targetted >> > at human admins, rather than machine usage. You are right though that >> > something like libvirt would never use --socket-group, or --socket-path. >> > Even admins would benefit if all programs followed the same naming for >> > these. We could document such options as "SHOULD" rather than "MUST" >> > IOW, we don't mandate --socket-group, but if you're going to provide a >> > way to control socket group, this option should be used. >> >> I agree. It's still useful to have a convention that most vhost-user >> backend programs follow. > > Alex: > Can you add the doc entry that Stefan and Marc-André are asking > for; it's probably good they go together. Sure - is docs/interop/vhost-user.rst the master spec for vhost-user daemons? > > Dave > >> Stefan -- Alex Bennée
Hi On Wed, Mar 18, 2020 at 12:21 AM Alex Bennée <alex.bennee@linaro.org> wrote: > > > Dr. David Alan Gilbert <dgilbert@redhat.com> writes: > > > * Stefan Hajnoczi (stefanha@redhat.com) wrote: > >> On Mon, Mar 16, 2020 at 10:33:31AM +0000, Daniel P. Berrangé wrote: > >> > On Sat, Mar 14, 2020 at 02:33:25PM +0100, Marc-André Lureau wrote: > >> > > Hi > >> > > > >> > > On Thu, Mar 12, 2020 at 11:49 AM Daniel P. Berrangé <berrange@redhat.com> wrote: > >> > > > > >> > > > On Thu, Mar 12, 2020 at 10:41:42AM +0000, Alex Bennée wrote: > >> > > > > If you like running QEMU as a normal user (very common for TCG runs) > >> > > > > but you have to run virtiofsd as a root user you run into connection > >> > > > > problems. Adding support for an optional --socket-group allows the > >> > > > > users to keep using the command line. > >> > > > > >> > > > If we're going to support this, then I think we need to put it in > >> > > > the vhost-user.rst specification so we standardize across backends. > >> > > > > >> > > > > >> > > > >> > > Perhaps. Otoh, I wonder if the backend spec should be more limited to > >> > > arguments/introspection that are used by programs. > >> > > > >> > > In this case, I even consider --socket-path to be unnecessary, as a > >> > > management layer can/should provide a preopened & setup fd directly. > >> > > > >> > > What do you think? > >> > > >> > I think there's value in standardization even if it is an option targetted > >> > at human admins, rather than machine usage. You are right though that > >> > something like libvirt would never use --socket-group, or --socket-path. > >> > Even admins would benefit if all programs followed the same naming for > >> > these. We could document such options as "SHOULD" rather than "MUST" > >> > IOW, we don't mandate --socket-group, but if you're going to provide a > >> > way to control socket group, this option should be used. > >> > >> I agree. It's still useful to have a convention that most vhost-user > >> backend programs follow. > > > > Alex: > > Can you add the doc entry that Stefan and Marc-André are asking > > for; it's probably good they go together. > > Sure - is docs/interop/vhost-user.rst the master spec for vhost-user > daemons? So far, yes. But it might make sense to create a standalone vhost-user-daemons.rst. -- Marc-André Lureau
diff --git a/docs/tools/virtiofsd.rst b/docs/tools/virtiofsd.rst index 378594c422a..5a8246b74f8 100644 --- a/docs/tools/virtiofsd.rst +++ b/docs/tools/virtiofsd.rst @@ -85,6 +85,10 @@ Options Listen on vhost-user UNIX domain socket at PATH. +.. option:: --socket-group=GROUP + + Set the vhost-user UNIX domain socket gid to GROUP. + .. option:: --fd=FDNUM Accept connections from vhost-user UNIX domain socket file descriptor FDNUM. diff --git a/tools/virtiofsd/fuse_i.h b/tools/virtiofsd/fuse_i.h index 1240828208a..492e002181e 100644 --- a/tools/virtiofsd/fuse_i.h +++ b/tools/virtiofsd/fuse_i.h @@ -68,6 +68,7 @@ struct fuse_session { size_t bufsize; int error; char *vu_socket_path; + char *vu_socket_group; int vu_listen_fd; int vu_socketfd; struct fv_VuDev *virtio_dev; diff --git a/tools/virtiofsd/fuse_lowlevel.c b/tools/virtiofsd/fuse_lowlevel.c index 2dd36ec03b6..4d1ba2925d1 100644 --- a/tools/virtiofsd/fuse_lowlevel.c +++ b/tools/virtiofsd/fuse_lowlevel.c @@ -2523,6 +2523,7 @@ static const struct fuse_opt fuse_ll_opts[] = { LL_OPTION("--debug", debug, 1), LL_OPTION("allow_root", deny_others, 1), LL_OPTION("--socket-path=%s", vu_socket_path, 0), + LL_OPTION("--socket-group=%s", vu_socket_group, 0), LL_OPTION("--fd=%d", vu_listen_fd, 0), LL_OPTION("--thread-pool-size=%d", thread_pool_size, 0), FUSE_OPT_END @@ -2630,6 +2631,11 @@ struct fuse_session *fuse_session_new(struct fuse_args *args, "fuse: --socket-path and --fd cannot be given together\n"); goto out4; } + if (se->vu_socket_group && !se->vu_socket_path) { + fuse_log(FUSE_LOG_ERR, + "fuse: --socket-group can only be used with --socket-path\n"); + goto out4; + } se->bufsize = FUSE_MAX_MAX_PAGES * getpagesize() + FUSE_BUFFER_HEADER_SIZE; diff --git a/tools/virtiofsd/fuse_virtio.c b/tools/virtiofsd/fuse_virtio.c index 3b6d16a0417..331f9fc65c5 100644 --- a/tools/virtiofsd/fuse_virtio.c +++ b/tools/virtiofsd/fuse_virtio.c @@ -31,6 +31,8 @@ #include <sys/socket.h> #include <sys/types.h> #include <sys/un.h> +#include <sys/types.h> +#include <grp.h> #include <unistd.h> #include "contrib/libvhost-user/libvhost-user.h" @@ -924,15 +926,29 @@ static int fv_create_listen_socket(struct fuse_session *se) /* * Unfortunately bind doesn't let you set the mask on the socket, - * so set umask to 077 and restore it later. + * so set umask appropriately and restore it later. */ - old_umask = umask(0077); + 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)) { + fuse_log(FUSE_LOG_WARNING, + "vhost socket failed to set group to %s (%d)\n", + se->vu_socket_group, g->gr_gid); + } + } + } umask(old_umask); if (listen(listen_sock, 1) == -1) {
If you like running QEMU as a normal user (very common for TCG runs) but you have to run virtiofsd as a root user you run into connection problems. Adding support for an optional --socket-group allows the users to keep using the command line. Signed-off-by: Alex Bennée <alex.bennee@linaro.org> --- v1 - tweak documentation and commentary --- docs/tools/virtiofsd.rst | 4 ++++ tools/virtiofsd/fuse_i.h | 1 + tools/virtiofsd/fuse_lowlevel.c | 6 ++++++ tools/virtiofsd/fuse_virtio.c | 20 ++++++++++++++++++-- 4 files changed, 29 insertions(+), 2 deletions(-) -- 2.20.1