[pve-devel] [PATCH 24/31] PVE: move snapshot cleanup into bottom half

[Thomas Lamprecht]

On 3/6/20 12:30 PM, Dietmar Maurer wrote:
> From: Wolfgang Bumiller <w.bumiller at proxmox.com>
> 
> as per:
>     (0ceccd858a8d) migration: qemu_savevm_state_cleanup() in cleanup
> 
> may affect held locks and therefore change assumptions made
> by that function!
> 
> Signed-off-by: Wolfgang Bumiller <w.bumiller at proxmox.com>
> Signed-off-by: Thomas Lamprecht <t.lamprecht at proxmox.com>
> ---
>  savevm-async.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/savevm-async.c b/savevm-async.c
> index e4bb0d24b2..10837fc858 100644
> --- a/savevm-async.c
> +++ b/savevm-async.c
> @@ -200,6 +200,8 @@ static void process_savevm_cleanup(void *opaque)
>      int ret;
>      qemu_bh_delete(snap_state.cleanup_bh);
>      snap_state.cleanup_bh = NULL;
> +    qemu_savevm_state_cleanup();
> +
>      qemu_mutex_unlock_iothread();
>      qemu_thread_join(&snap_state.thread);
>      qemu_mutex_lock_iothread();
> @@ -276,7 +278,6 @@ static void *process_savevm_thread(void *opaque)
>                      save_snapshot_error("qemu_savevm_state_iterate error %d", ret);
>                      break;
>              }
> -            qemu_savevm_state_cleanup();
>              DPRINTF("save complete\n");
>              break;
>          }
> 

for the record, this could be squashed into "[PATCH 17/31] PVE: internal
snapshot async" no biggie, but if we already go for a cleanup round..

We can naturally decide to keep this for historical value, but one can go
back in git history still anyway..