[pve-devel] [PATCH qemu] add patch fixing fd leak for vhost

[pve-devel] [PATCH qemu] add patch fixing fd leak for vhost

[Fiona Ebner]

Each pause+resume operation (which is also done as part of taking a VM
snapshot) would increase the number of open file descriptors by the
number of vhost devices (e.g. network devices by default). This could
lead to crashes during backup and surely other issues once the system
limit (default 1024) was reached [0].

[0]: https://forum.proxmox.com/threads/131603/

Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
---
 .../extra/0011-vhost-fix-the-fd-leak.patch    | 29 +++++++++++++++++++
 debian/patches/series                         |  1 +
 2 files changed, 30 insertions(+)
 create mode 100644 debian/patches/extra/0011-vhost-fix-the-fd-leak.patch

diff --git a/debian/patches/extra/0011-vhost-fix-the-fd-leak.patch b/debian/patches/extra/0011-vhost-fix-the-fd-leak.patch
new file mode 100644
index 0000000..31392fb
--- /dev/null
+++ b/debian/patches/extra/0011-vhost-fix-the-fd-leak.patch
@@ -0,0 +1,29 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Li Feng <fengli@smartx.com>
+Date: Mon, 31 Jul 2023 20:10:06 +0800
+Subject: [PATCH] vhost: fix the fd leak
+
+When the vhost-user reconnect to the backend, the notifer should be
+cleanup. Otherwise, the fd resource will be exhausted.
+
+Fixes: f9a09ca3ea ("vhost: add support for configure interrupt")
+
+Signed-off-by: Li Feng <fengli@smartx.com>
+Reviewed-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
+---
+ hw/virtio/vhost.c | 2 ++
+ 1 file changed, 2 insertions(+)
+
+diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
+index a266396576..8e3311781f 100644
+--- a/hw/virtio/vhost.c
++++ b/hw/virtio/vhost.c
+@@ -2034,6 +2034,8 @@ void vhost_dev_stop(struct vhost_dev *hdev, VirtIODevice *vdev, bool vrings)
+     event_notifier_test_and_clear(
+         &hdev->vqs[VHOST_QUEUE_NUM_CONFIG_INR].masked_config_notifier);
+     event_notifier_test_and_clear(&vdev->config_notifier);
++    event_notifier_cleanup(
++        &hdev->vqs[VHOST_QUEUE_NUM_CONFIG_INR].masked_config_notifier);
+ 
+     trace_vhost_dev_stop(hdev, vdev->name, vrings);
+ 
diff --git a/debian/patches/series b/debian/patches/series
index a4dd4c2..74578f6 100644
--- a/debian/patches/series
+++ b/debian/patches/series
@@ -8,6 +8,7 @@ extra/0007-bcm2835_property-disable-reentrancy-detection-for-io.patch
 extra/0008-raven-disable-reentrancy-detection-for-iomem.patch
 extra/0009-apic-disable-reentrancy-detection-for-apic-msi.patch
 extra/0010-migration-block-dirty-bitmap-fix-loading-bitmap-when.patch
+extra/0011-vhost-fix-the-fd-leak.patch
 bitmap-mirror/0001-drive-mirror-add-support-for-sync-bitmap-mode-never.patch
 bitmap-mirror/0002-drive-mirror-add-support-for-conditional-and-always-.patch
 bitmap-mirror/0003-mirror-add-check-for-bitmap-mode-without-bitmap.patch
-- 
2.39.2

[pve-devel] applied: [PATCH qemu] add patch fixing fd leak for vhost

[Thomas Lamprecht]

Am 03/08/2023 um 15:56 schrieb Fiona Ebner:
> Each pause+resume operation (which is also done as part of taking a VM
> snapshot) would increase the number of open file descriptors by the
> number of vhost devices (e.g. network devices by default). This could
> lead to crashes during backup and surely other issues once the system
> limit (default 1024) was reached [0].
> 
> [0]: https://forum.proxmox.com/threads/131603/
> 
> Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
> ---
>  .../extra/0011-vhost-fix-the-fd-leak.patch    | 29 +++++++++++++++++++
>  debian/patches/series                         |  1 +
>  2 files changed, 30 insertions(+)
>  create mode 100644 debian/patches/extra/0011-vhost-fix-the-fd-leak.patch
> 
>

applied, thanks!