[PATCH v2 hci-8.0] hostmem: fix live update crash of VMs with bound guest memory #VSTOR-147016

[PATCH v2 hci-8.0] hostmem: fix live update crash of VMs with bound guest memory #VSTOR-147016

[Denis V. Lunev]

From: Denis V. Lunev <den@openvz.org>

A live QEMU update kills a VM whose memory backend has host-nodes when
the emulator cgroup no longer allows those nodes. The old QEMU saves its
state, the new one exits during startup with

  qemu-kvm: cannot bind memory to host NUMA nodes: Invalid argument

and the guest is lost.

Guest RAM survives cpr-exec as the same anonymous mapping, carrying its
pages and its memory policy. host_memory_backend_memory_complete() does
not know that and treats it as fresh memory: it calls mbind() with
MPOL_MF_STRICT | MPOL_MF_MOVE and preallocates again. The exec keeps the
PID and the cgroup, so if cpuset.mems has been changed to nodes outside
host-nodes since the VM started, mbind() returns EINVAL and QEMU exits.
Where it succeeds, it only walks or migrates pages that are already in
place, while the guest is stopped.

Skip the NUMA binding and preallocation for an anonymous backend on
incoming cpr-exec. ram_block_add() refuses to allocate anonymous RAM in
that mode, so such a block is always an inherited one. Backends with an
fd get a new mapping after the exec and keep mbind(), and memory added
after the update is back on the normal path.

Suggested-by: Andrey Drobyshev <andrey.drobyshev@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
---
 backends/hostmem.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/backends/hostmem.c b/backends/hostmem.c
index 35734d6f4d1..bb7f0c4c18f 100644
--- a/backends/hostmem.c
+++ b/backends/hostmem.c
@@ -22,6 +22,7 @@
 #include "qemu/madvise.h"
 #include "qemu/cutils.h"
 #include "hw/qdev-core.h"
+#include "migration/cpr.h"
 
 #ifdef CONFIG_NUMA
 #include <numaif.h>
@@ -363,6 +364,10 @@ host_memory_backend_memory_complete(UserCreatable *uc, Error **errp)
     if (!backend->dump) {
         qemu_madvise(ptr, sz, QEMU_MADV_DONTDUMP);
     }
+    /* Anonymous memory is inherited on cpr-exec with its policy and pages */
+    if (cpr_exec_is_incoming() && memory_region_get_fd(&backend->mr) < 0) {
+        return;
+    }
 #ifdef CONFIG_NUMA
     unsigned long lastbit = find_last_bit(backend->host_nodes, MAX_NODES);
     /* lastbit == MAX_NODES means maxnode = 0 */

base-commit: e8a73d39faf578d58aa2c4b4b024226796c390e6
-- 
2.53.0

Re: [PATCH v2 hci-8.0] hostmem: fix live update crash of VMs with bound guest memory #VSTOR-147016

[Andrey Drobyshev]

LGTM
Ack

On 9/23/26 8:31 PM, Denis V. Lunev wrote:
> From: Denis V. Lunev <den@openvz.org>
> 
> A live QEMU update kills a VM whose memory backend has host-nodes when
> the emulator cgroup no longer allows those nodes. The old QEMU saves its
> state, the new one exits during startup with
> 
>   qemu-kvm: cannot bind memory to host NUMA nodes: Invalid argument
> 
> and the guest is lost.
> 
> Guest RAM survives cpr-exec as the same anonymous mapping, carrying its
> pages and its memory policy. host_memory_backend_memory_complete() does
> not know that and treats it as fresh memory: it calls mbind() with
> MPOL_MF_STRICT | MPOL_MF_MOVE and preallocates again. The exec keeps the
> PID and the cgroup, so if cpuset.mems has been changed to nodes outside
> host-nodes since the VM started, mbind() returns EINVAL and QEMU exits.
> Where it succeeds, it only walks or migrates pages that are already in
> place, while the guest is stopped.
> 
> Skip the NUMA binding and preallocation for an anonymous backend on
> incoming cpr-exec. ram_block_add() refuses to allocate anonymous RAM in
> that mode, so such a block is always an inherited one. Backends with an
> fd get a new mapping after the exec and keep mbind(), and memory added
> after the update is back on the normal path.
> 
> Suggested-by: Andrey Drobyshev <andrey.drobyshev@virtuozzo.com>
> Signed-off-by: Denis V. Lunev <den@openvz.org>
> ---
>  backends/hostmem.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/backends/hostmem.c b/backends/hostmem.c
> index 35734d6f4d1..bb7f0c4c18f 100644
> --- a/backends/hostmem.c
> +++ b/backends/hostmem.c
> @@ -22,6 +22,7 @@
>  #include "qemu/madvise.h"
>  #include "qemu/cutils.h"
>  #include "hw/qdev-core.h"
> +#include "migration/cpr.h"
>  
>  #ifdef CONFIG_NUMA
>  #include <numaif.h>
> @@ -363,6 +364,10 @@ host_memory_backend_memory_complete(UserCreatable *uc, Error **errp)
>      if (!backend->dump) {
>          qemu_madvise(ptr, sz, QEMU_MADV_DONTDUMP);
>      }
> +    /* Anonymous memory is inherited on cpr-exec with its policy and pages */
> +    if (cpr_exec_is_incoming() && memory_region_get_fd(&backend->mr) < 0) {
> +        return;
> +    }
>  #ifdef CONFIG_NUMA
>      unsigned long lastbit = find_last_bit(backend->host_nodes, MAX_NODES);
>      /* lastbit == MAX_NODES means maxnode = 0 */
> 
> base-commit: e8a73d39faf578d58aa2c4b4b024226796c390e6