qemu-kvm/kvm-vhost-fix-vhost_dev_enable_notifiers-error-case.patch
Miroslav Rezanina b56a1fa35b * Mon Jul 10 2023 Miroslav Rezanina <mrezanin@redhat.com> - 8.0.0-7
- kvm-numa-Validate-cluster-and-NUMA-node-boundary-if-requ.patch [bz#2171363]
- kvm-hw-arm-Validate-cluster-and-NUMA-node-boundary.patch [bz#2171363]
- kvm-hw-arm-virt-Validate-cluster-and-NUMA-node-boundary-.patch [bz#2171363]
- kvm-vhost-fix-vhost_dev_enable_notifiers-error-case.patch [RHEL-330]
- kvm-kvm-reuse-per-vcpu-stats-fd-to-avoid-vcpu-interrupti.patch [bz#2218644]
- kvm-vhost-vdpa-do-not-cleanup-the-vdpa-vhost-net-structu.patch [bz#2128929]
- Resolves: bz#2171363
  ([aarch64] Kernel hits Call trace with irregular CPU-to-NUMA association)
- Resolves: RHEL-330
  ([virtual network][qemu-kvm-8.0.0-rc1]qemu core dump: qemu-kvm: ../softmmu/memory.c:2592: void memory_region_del_eventfd(MemoryRegion *, hwaddr, unsigned int, _Bool, uint64_t, EventNotifier *): Assertion `i != mr->ioeventfd_nb' failed)
- Resolves: bz#2218644
  (query-stats QMP command interrupts vcpus, the Max Latencies could be more than 100us (rhel 9.3.0 clone))
- Resolves: bz#2128929
  ([rhel9.2] hotplug/hotunplug mlx vdpa device to the occupied addr port, then qemu core dump occurs after shutdown guest)
2023-07-10 01:43:27 -04:00

139 lines
5.0 KiB
Diff

From ac54f5f746782da89ab674733af5622e524b58eb Mon Sep 17 00:00:00 2001
From: Laurent Vivier <lvivier@redhat.com>
Date: Fri, 2 Jun 2023 18:27:35 +0200
Subject: [PATCH 4/6] vhost: fix vhost_dev_enable_notifiers() error case
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
RH-Author: Laurent Vivier <lvivier@redhat.com>
RH-MergeRequest: 176: vhost: fix vhost_dev_enable_notifiers() error case
RH-Jira: RHEL-330
RH-Acked-by: MST <mst@redhat.com>
RH-Acked-by: Cindy Lu <lulu@redhat.com>
RH-Acked-by: Eugenio Pérez <eperezma@redhat.com>
RH-Acked-by: Jason Wang <jasowang@redhat.com>
RH-Commit: [1/1] fd30d7501be59f7e5b9d6fc5ed84efcc4037d08e (lvivier/qemu-kvm-centos)
JIRA: https://issues.redhat.com/browse/RHEL-330
in vhost_dev_enable_notifiers(), if virtio_bus_set_host_notifier(true)
fails, we call vhost_dev_disable_notifiers() that executes
virtio_bus_set_host_notifier(false) on all queues, even on queues that
have failed to be initialized.
This triggers a core dump in memory_region_del_eventfd():
virtio_bus_set_host_notifier: unable to init event notifier: Too many open files (-24)
vhost VQ 1 notifier binding failed: 24
.../softmmu/memory.c:2611: memory_region_del_eventfd: Assertion `i != mr->ioeventfd_nb' failed.
Fix the problem by providing to vhost_dev_disable_notifiers() the
number of queues to disable.
Fixes: 8771589b6f81 ("vhost: simplify vhost_dev_enable_notifiers")
Cc: longpeng2@huawei.com
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Message-Id: <20230602162735.3670785-1-lvivier@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 92099aa4e9a3bb6856c290afaf41c76f9e3dd9fd)
---
hw/virtio/vhost.c | 65 ++++++++++++++++++++++++++---------------------
1 file changed, 36 insertions(+), 29 deletions(-)
diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index a266396576..ae0a033e60 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -1545,6 +1545,40 @@ void vhost_dev_cleanup(struct vhost_dev *hdev)
memset(hdev, 0, sizeof(struct vhost_dev));
}
+static void vhost_dev_disable_notifiers_nvqs(struct vhost_dev *hdev,
+ VirtIODevice *vdev,
+ unsigned int nvqs)
+{
+ BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(vdev)));
+ int i, r;
+
+ /*
+ * Batch all the host notifiers in a single transaction to avoid
+ * quadratic time complexity in address_space_update_ioeventfds().
+ */
+ memory_region_transaction_begin();
+
+ for (i = 0; i < nvqs; ++i) {
+ r = virtio_bus_set_host_notifier(VIRTIO_BUS(qbus), hdev->vq_index + i,
+ false);
+ if (r < 0) {
+ error_report("vhost VQ %d notifier cleanup failed: %d", i, -r);
+ }
+ assert(r >= 0);
+ }
+
+ /*
+ * The transaction expects the ioeventfds to be open when it
+ * commits. Do it now, before the cleanup loop.
+ */
+ memory_region_transaction_commit();
+
+ for (i = 0; i < nvqs; ++i) {
+ virtio_bus_cleanup_host_notifier(VIRTIO_BUS(qbus), hdev->vq_index + i);
+ }
+ virtio_device_release_ioeventfd(vdev);
+}
+
/* Stop processing guest IO notifications in qemu.
* Start processing them in vhost in kernel.
*/
@@ -1574,7 +1608,7 @@ int vhost_dev_enable_notifiers(struct vhost_dev *hdev, VirtIODevice *vdev)
if (r < 0) {
error_report("vhost VQ %d notifier binding failed: %d", i, -r);
memory_region_transaction_commit();
- vhost_dev_disable_notifiers(hdev, vdev);
+ vhost_dev_disable_notifiers_nvqs(hdev, vdev, i);
return r;
}
}
@@ -1591,34 +1625,7 @@ int vhost_dev_enable_notifiers(struct vhost_dev *hdev, VirtIODevice *vdev)
*/
void vhost_dev_disable_notifiers(struct vhost_dev *hdev, VirtIODevice *vdev)
{
- BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(vdev)));
- int i, r;
-
- /*
- * Batch all the host notifiers in a single transaction to avoid
- * quadratic time complexity in address_space_update_ioeventfds().
- */
- memory_region_transaction_begin();
-
- for (i = 0; i < hdev->nvqs; ++i) {
- r = virtio_bus_set_host_notifier(VIRTIO_BUS(qbus), hdev->vq_index + i,
- false);
- if (r < 0) {
- error_report("vhost VQ %d notifier cleanup failed: %d", i, -r);
- }
- assert (r >= 0);
- }
-
- /*
- * The transaction expects the ioeventfds to be open when it
- * commits. Do it now, before the cleanup loop.
- */
- memory_region_transaction_commit();
-
- for (i = 0; i < hdev->nvqs; ++i) {
- virtio_bus_cleanup_host_notifier(VIRTIO_BUS(qbus), hdev->vq_index + i);
- }
- virtio_device_release_ioeventfd(vdev);
+ vhost_dev_disable_notifiers_nvqs(hdev, vdev, hdev->nvqs);
}
/* Test and clear event pending status.
--
2.39.3