[v2] vhost: fix VDUSE device destruction failure

  From: Maxime Coquelin <maxime.coquelin@redhat.com>

VDUSE_DESTROY_DEVICE ioctl can fail because the device's
chardev is not released despite close syscall having been
called. It happens because the events handler thread is
still polling the file descriptor.

fdset_pipe_notify() is not enough because it does not
ensure the notification has been handled by the event
thread, it just returns once the notification is sent.

To fix this, this patch introduces a synchronization
mechanism based on pthread's condition, so that
fdset_pipe_notify_sync() only returns once the pipe's
read callback has been executed.

Fixes: 51d018fdac4e ("vhost: add VDUSE events handler")
Cc: stable@dpdk.org

Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Signed-off-by: David Marchand <david.marchand@redhat.com>
---
Changes since v1:
- sync'd only when in VDUSE destruction path,
- added explicit init of sync_mutex,

---
 lib/vhost/fd_man.c | 23 +++++++++++++++++++++--
 lib/vhost/fd_man.h |  6 ++++++
 lib/vhost/socket.c |  1 +
 lib/vhost/vduse.c  |  3 ++-
 4 files changed, 30 insertions(+), 3 deletions(-)
  

Le lun. 4 mars 2024, 11:36, David Marchand <david.marchand@redhat.com> a
écrit :

> From: Maxime Coquelin <maxime.coquelin@redhat.com>
>
> VDUSE_DESTROY_DEVICE ioctl can fail because the device's
> chardev is not released despite close syscall having been
> called. It happens because the events handler thread is
> still polling the file descriptor.
>
> fdset_pipe_notify() is not enough because it does not
> ensure the notification has been handled by the event
> thread, it just returns once the notification is sent.
>
> To fix this, this patch introduces a synchronization
> mechanism based on pthread's condition, so that
> fdset_pipe_notify_sync() only returns once the pipe's
> read callback has been executed.
>
> Fixes: 51d018fdac4e ("vhost: add VDUSE events handler")
> Cc: stable@dpdk.org
>
> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
> Signed-off-by: David Marchand <david.marchand@redhat.com>
> ---
> Changes since v1:
> - sync'd only when in VDUSE destruction path,
> - added explicit init of sync_mutex,
>
> ---
>  lib/vhost/fd_man.c | 23 +++++++++++++++++++++--
>  lib/vhost/fd_man.h |  6 ++++++
>  lib/vhost/socket.c |  1 +
>  lib/vhost/vduse.c  |  3 ++-
>  4 files changed, 30 insertions(+), 3 deletions(-)
>

Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

Thanks for improving the patch,
Maxime


> diff --git a/lib/vhost/fd_man.c b/lib/vhost/fd_man.c
> index 79a8d2c006..481e6b900a 100644
> --- a/lib/vhost/fd_man.c
> +++ b/lib/vhost/fd_man.c
> @@ -309,10 +309,11 @@ fdset_event_dispatch(void *arg)
>  }
>
>  static void
> -fdset_pipe_read_cb(int readfd, void *dat __rte_unused,
> +fdset_pipe_read_cb(int readfd, void *dat,
>                    int *remove __rte_unused)
>  {
>         char charbuf[16];
> +       struct fdset *fdset = dat;
>         int r = read(readfd, charbuf, sizeof(charbuf));
>         /*
>          * Just an optimization, we don't care if read() failed
> @@ -320,6 +321,11 @@ fdset_pipe_read_cb(int readfd, void *dat __rte_unused,
>          * compiler happy
>          */
>         RTE_SET_USED(r);
> +
> +       pthread_mutex_lock(&fdset->sync_mutex);
> +       fdset->sync = true;
> +       pthread_cond_broadcast(&fdset->sync_cond);
> +       pthread_mutex_unlock(&fdset->sync_mutex);
>  }
>
>  void
> @@ -342,7 +348,7 @@ fdset_pipe_init(struct fdset *fdset)
>         }
>
>         ret = fdset_add(fdset, fdset->u.readfd,
> -                       fdset_pipe_read_cb, NULL, NULL);
> +                       fdset_pipe_read_cb, NULL, fdset);
>
>         if (ret < 0) {
>                 VHOST_FDMAN_LOG(ERR,
> @@ -366,5 +372,18 @@ fdset_pipe_notify(struct fdset *fdset)
>          * compiler happy
>          */
>         RTE_SET_USED(r);
> +}
> +
> +void
> +fdset_pipe_notify_sync(struct fdset *fdset)
> +{
> +       pthread_mutex_lock(&fdset->sync_mutex);
> +
> +       fdset->sync = false;
> +       fdset_pipe_notify(fdset);
> +
> +       while (!fdset->sync)
> +               pthread_cond_wait(&fdset->sync_cond, &fdset->sync_mutex);
>
> +       pthread_mutex_unlock(&fdset->sync_mutex);
>  }
> diff --git a/lib/vhost/fd_man.h b/lib/vhost/fd_man.h
> index 6315904c8e..7816fb11ac 100644
> --- a/lib/vhost/fd_man.h
> +++ b/lib/vhost/fd_man.h
> @@ -6,6 +6,7 @@
>  #define _FD_MAN_H_
>  #include <pthread.h>
>  #include <poll.h>
> +#include <stdbool.h>
>
>  #define MAX_FDS 1024
>
> @@ -35,6 +36,10 @@ struct fdset {
>                         int writefd;
>                 };
>         } u;
> +
> +       pthread_mutex_t sync_mutex;
> +       pthread_cond_t sync_cond;
> +       bool sync;
>  };
>
>
> @@ -53,5 +58,6 @@ int fdset_pipe_init(struct fdset *fdset);
>  void fdset_pipe_uninit(struct fdset *fdset);
>
>  void fdset_pipe_notify(struct fdset *fdset);
> +void fdset_pipe_notify_sync(struct fdset *fdset);
>
>  #endif
> diff --git a/lib/vhost/socket.c b/lib/vhost/socket.c
> index a2fdac30a4..96b3ab5595 100644
> --- a/lib/vhost/socket.c
> +++ b/lib/vhost/socket.c
> @@ -93,6 +93,7 @@ static struct vhost_user vhost_user = {
>                 .fd = { [0 ... MAX_FDS - 1] = {-1, NULL, NULL, NULL, 0} },
>                 .fd_mutex = PTHREAD_MUTEX_INITIALIZER,
>                 .fd_pooling_mutex = PTHREAD_MUTEX_INITIALIZER,
> +               .sync_mutex = PTHREAD_MUTEX_INITIALIZER,
>                 .num = 0
>         },
>         .vsocket_cnt = 0,
> diff --git a/lib/vhost/vduse.c b/lib/vhost/vduse.c
> index d462428d2c..e0c6991b69 100644
> --- a/lib/vhost/vduse.c
> +++ b/lib/vhost/vduse.c
> @@ -36,6 +36,7 @@ static struct vduse vduse = {
>                 .fd = { [0 ... MAX_FDS - 1] = {-1, NULL, NULL, NULL, 0} },
>                 .fd_mutex = PTHREAD_MUTEX_INITIALIZER,
>                 .fd_pooling_mutex = PTHREAD_MUTEX_INITIALIZER,
> +               .sync_mutex = PTHREAD_MUTEX_INITIALIZER,
>                 .num = 0
>         },
>  };
> @@ -618,7 +619,7 @@ vduse_device_destroy(const char *path)
>         vduse_device_stop(dev);
>
>         fdset_del(&vduse.fdset, dev->vduse_dev_fd);
> -       fdset_pipe_notify(&vduse.fdset);
> +       fdset_pipe_notify_sync(&vduse.fdset);
>
>         if (dev->vduse_dev_fd >= 0) {
>                 close(dev->vduse_dev_fd);
> --
> 2.43.0
>
>
  

On Mon, Mar 4, 2024 at 11:36 AM David Marchand
<david.marchand@redhat.com> wrote:
>
> From: Maxime Coquelin <maxime.coquelin@redhat.com>
>
> VDUSE_DESTROY_DEVICE ioctl can fail because the device's
> chardev is not released despite close syscall having been
> called. It happens because the events handler thread is
> still polling the file descriptor.
>
> fdset_pipe_notify() is not enough because it does not
> ensure the notification has been handled by the event
> thread, it just returns once the notification is sent.
>
> To fix this, this patch introduces a synchronization
> mechanism based on pthread's condition, so that
> fdset_pipe_notify_sync() only returns once the pipe's
> read callback has been executed.
>
> Fixes: 51d018fdac4e ("vhost: add VDUSE events handler")
> Cc: stable@dpdk.org
>
> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
> Signed-off-by: David Marchand <david.marchand@redhat.com>
> ---
> Changes since v1:
> - sync'd only when in VDUSE destruction path,
> - added explicit init of sync_mutex,

Applied, thanks.