Re: [External] Re: [PATCH v3 13/20] migration/multifd: Prepare to introduce DSA acceleration on the multifd path.

[Peter Xu <peterx@redhat.com>]

On Mon, Jan 22, 2024 at 04:37:09PM -0800, Hao Xiang wrote:
> > > +static void set_normal_pages(MultiFDSendParams *p)
> > > +{
> > > +    for (int i = 0; i < p->pages->num; i++) {
> > > +        p->batch_task->results[i] = false;
> > > +    }
> > > +}
> > Please correct me if I am wrong but set_normal_pages will not be a part of the final patch, right? They are there for testing out the performance against different zero page ration scenarios. If it’s so, can we isolate these parts into a separate patch.
> 
> set_normal_pages is used for performance testing and testing only. It
> won't introduce any "incorrect" behavior and I would love to see it
> being part of the upstream code. But the argument that testing change
> should remain private is always correct. So I am totally OK with
> isolating the parts into a separate patch.

IMHO we can allow that to be production code; as long as the new zeropage
detection parameter can allow user to choose "none" (as I mentioned in
another reply), then it's not a test code only but allow the user to
disable zeropage detections when the user wants.  Thanks,

-- 
Peter Xu