Re: [Qemu-devel] [PATCH] migration: Fix multi-thread compression bug

[Juan Quintela]

"Dr. David Alan Gilbert" <address@hidden> wrote:
> * Liang Li (address@hidden) wrote:
>> Recently, a bug related to multiple thread compression feature for
>> live migration is reported. The destination side will be blocked
>> during live migration if there are heavy workload in host and
>> memory intensive workload in guest, this is most likely to happen
>> when there is one decompression thread.
>> 
>> Some parts of the decompression code are incorrect:
>> 1. The main thread receives data from source side will enter a busy
>> loop to wait for a free decompression thread.
>> 2. A lock is needed to protect the decomp_param[idx]->start, because
>> it is checked in the main thread and is updated in the decompression
>> thread.
>> 
>> Fix these two issues by following the code pattern for compression.
>> 
>
> OK, I think that's an improvement - but I have a question.
> Since it's an improvement (and basically now the same as compress):
>
> Reviewed-by: Dr. David Alan Gilbert <address@hidden>
>
> however, my question is:
> What guarantee's that all of the decompression has finished by the time
> the VM is started?  I see in migration/migration.c that we have:
>
>     if (!global_state_received() ||
>         global_state_get_runstate() == RUN_STATE_RUNNING) {
>         if (autostart) {
>             vm_start();
>         } else {
>             runstate_set(RUN_STATE_PAUSED);
>         }
>     } else {
>         runstate_set(global_state_get_runstate());
>     }
>     migrate_decompress_threads_join();

You are right here.  If we don't want to do the join earlier, we need to
do the equivalent of looking that all threads have set param->done as true.
That is what I do on the multiple-fd code.

Later, Juan.