Re: [Qemu-devel] [PATCH] migrate: Migration aborts abruptly for machine "none"

[Ashijeet Acharya]

On Tue, Jan 31, 2017 at 2:18 AM, Paolo Bonzini <address@hidden> wrote:
>
>
> On 29/01/2017 08:19, Ashijeet Acharya wrote:
>> Also, Paolo has a similar opinion to yours that we should fix it
>> instead of blocking it. He suggests to migrate only the device states
>> and skip all the RAM related stuff. Maybe he can explain it better
>> when he is available.
>
> Yes, there is even support already in ram_migration_cleanup for not
> allocating the bitmap:
>
>     struct BitmapRcu *bitmap = migration_bitmap_rcu;
>     atomic_rcu_set(&migration_bitmap_rcu, NULL);
>     if (bitmap) {
>         memory_global_dirty_log_stop();
>         call_rcu(bitmap, migration_bitmap_free, rcu);
>     }
>
> I think you should just iterate until you find no more misbehaving code.
>  The allocation is one, but ram_find_and_save_block should also return 0
> if QLIST_FIRST_RCU(&ram_list.blocks) is NULL, for example.
>

Okay, I will drop the idea of adding a migration blocker in this case
and proceed as you say.

Ashijeet
> Paolo