Re: [PATCH v2 09/12] plugins: add migration blocker

[Thomas Huth <thuth@redhat.com>]

On 20/06/2024 17.22, Alex Bennée wrote:
> If the plugin in controlling time there is some state that might be
> missing from the plugin tracking it. Migration is unlikely to work in
> this case so lets put a migration blocker in to let the user know if
> they try.
> 
> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
> Suggested-by: "Dr. David Alan Gilbert" <dave@treblig.org>
> ---
>   plugins/api.c | 8 ++++++++
>   1 file changed, 8 insertions(+)
> 
> diff --git a/plugins/api.c b/plugins/api.c
> index 4431a0ea7e..c4239153af 100644
> --- a/plugins/api.c
> +++ b/plugins/api.c
> @@ -47,6 +47,8 @@
>   #include "disas/disas.h"
>   #include "plugin.h"
>   #ifndef CONFIG_USER_ONLY
> +#include "qapi/error.h"
> +#include "migration/blocker.h"
>   #include "exec/ram_addr.h"
>   #include "qemu/plugin-memory.h"
>   #include "hw/boards.h"
> @@ -589,11 +591,17 @@ uint64_t qemu_plugin_u64_sum(qemu_plugin_u64 entry)
>    * Time control
>    */
>   static bool has_control;
> +Error *migration_blocker;
>   
>   const void *qemu_plugin_request_time_control(void)
>   {
>       if (!has_control) {
>           has_control = true;
> +#ifdef CONFIG_SOFTMMU
> +        error_setg(&migration_blocker,
> +                   "TCG plugin time control does not support migration");
> +        migrate_add_blocker(&migration_blocker, NULL);
> +#endif
>           return &has_control;
>       }
>       return NULL;

Reviewed-by: Thomas Huth <thuth@redhat.com>