Re: [RFC PATCH 3/3] mm,memory_hotplug: Allocate memmap from the added memory range

From: David Hildenbrand <david@redhat.com>
To: Oscar Salvador <osalvador@suse.de>
Cc: mhocko@kernel.org, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org, vbabka@suse.cz, pasha.tatashin@soleen.com
Subject: Re: [RFC PATCH 3/3] mm,memory_hotplug: Allocate memmap from the added memory range
Date: Fri, 20 Nov 2020 10:31:05 +0100	[thread overview]
Message-ID: <ddcb1d70-2e4b-a185-b749-d4f300feec18@redhat.com> (raw)
In-Reply-To: <20201119104847.GA5281@localhost.localdomain>

>>> -	move_pfn_range_to_zone(zone, pfn, nr_pages, NULL, MIGRATE_ISOLATE);
>>> +	move_pfn_range_to_zone(zone, pfn, nr_pages, nr_vmemmap_pages, NULL,
>>> +			       MIGRATE_ISOLATE);
>>
>> As mentioned, I'd suggest properly initializing the memmap here
>>
>> if (nr_vmemmap_pages) {
>> 	move_pfn_range_to_zone(zone, pfn, nr_vmemmap_pages, NULL,
>> 			       MIGRATE_UNMOVABLE);
>> }
>> move_pfn_range_to_zone(zone, valid_start_pfn, valid_nr_pages, NULL,
> 
> Sure, agreed.
> 
>>> +	if (!support_memmap_on_memory(size))
>>> +		mhp_flags &= ~MEMHP_MEMMAP_ON_MEMORY;
>>
>> Callers (e.g., virtio-mem) might rely on this. We should reject this with
>> -EINVAL and provide a way for callers to test whether this flag is possible.
> 
> Uhm, we might want to make "support_memmap_on_memory" public, and
> callers who might want to it use can check its return value?
> Or do you have something else in mind?

Right, a way for callers to check if it's supported. 
"mhp_supports_memmap_on_memory" or sth. like that.

Thanks!

-- 
Thanks,

David / dhildenb