Re: [RFC PATCH 2/4] mm: unified fault and migrate device page paths

[Alistair Popple <apopple@nvidia.com>]

On Thu, Aug 21, 2025 at 08:10:31AM +0300, Mika Penttilä wrote:
> 
> On 8/21/25 07:30, Balbir Singh wrote:
> 
> > On 8/14/25 17:19, Mika Penttilä wrote:

[...]

> >>  EXPORT_SYMBOL(hmm_range_fault);
> >> diff --git a/mm/migrate_device.c b/mm/migrate_device.c
> >> index e05e14d6eacd..87ddc0353165 100644
> >> --- a/mm/migrate_device.c
> >> +++ b/mm/migrate_device.c
> >> @@ -535,7 +535,18 @@ static void migrate_vma_unmap(struct migrate_vma *migrate)
> >>   */
> >>  int migrate_vma_setup(struct migrate_vma *args)
> >>  {
> >> +	int ret;
> >>  	long nr_pages = (args->end - args->start) >> PAGE_SHIFT;
> >> +	struct hmm_range range = {
> >> +		.notifier = NULL,
> >> +		.start = args->start,
> >> +		.end = args->end,
> >> +		.migrate = args,
> >> +		.hmm_pfns = args->src,
> >> +		.default_flags = HMM_PFN_REQ_MIGRATE,
> >> +		.dev_private_owner = args->pgmap_owner,
> >> +		.migrate = args
> >> +	};
> >>  
> >>  	args->start &= PAGE_MASK;
> >>  	args->end &= PAGE_MASK;
> >> @@ -560,17 +571,19 @@ int migrate_vma_setup(struct migrate_vma *args)
> >>  	args->cpages = 0;
> >>  	args->npages = 0;
> >>  
> >> -	migrate_vma_collect(args);
> >> +	if (args->flags & MIGRATE_VMA_FAULT)
> >> +		range.default_flags |= HMM_PFN_REQ_FAULT;
> >>  
> >> -	if (args->cpages)
> >> -		migrate_vma_unmap(args);
> >> +	ret = hmm_range_fault(&range);
> >> +
> >> +	migrate_hmm_range_setup(&range);
> >>  
> >>  	/*
> >>  	 * At this point pages are locked and unmapped, and thus they have
> >>  	 * stable content and can safely be copied to destination memory that
> >>  	 * is allocated by the drivers.
> >>  	 */
> >> -	return 0;
> >> +	return ret;
> >>  
> >>  }
> >>  EXPORT_SYMBOL(migrate_vma_setup);
> >> @@ -1014,3 +1027,54 @@ int migrate_device_coherent_folio(struct folio *folio)
> >>  		return 0;
> >>  	return -EBUSY;
> >>  }
> >> +
> >> +void migrate_hmm_range_setup(struct hmm_range *range)
> >> +{
> >> +
> >> +	struct migrate_vma *migrate = range->migrate;
> >> +
> >> +	if (!migrate)
> >> +		return;
> >> +
> >> +	migrate->npages = (migrate->end - migrate->start) >> PAGE_SHIFT;
> >> +	migrate->cpages = 0;
> >> +
> >> +	for (unsigned long i = 0; i < migrate->npages; i++) {
> >> +
> >> +		unsigned long pfn = range->hmm_pfns[i];
> >> +
> >> +		/*
> >> +		 *
> >> +		 *  Don't do migration if valid and migrate flags are not both set.
> >> +		 *
> >> +		 */
> >> +		if ((pfn & (HMM_PFN_VALID | HMM_PFN_MIGRATE)) !=
> >> +		    (HMM_PFN_VALID | HMM_PFN_MIGRATE)) {
> >> +			migrate->src[i] = 0;
> >> +			migrate->dst[i] = 0;
> >> +			continue;
> >> +		}
> >> +
> >> +		migrate->cpages++;
> >> +
> >> +		/*
> >> +		 *
> >> +		 * The zero page is encoded in a special way, valid and migrate is
> >> +		 * set, and pfn part is zero. Encode specially for migrate also.
> >> +		 *
> >> +		 */
> >> +		if (pfn == (HMM_PFN_VALID|HMM_PFN_MIGRATE)) {
> >> +			migrate->src[i] = MIGRATE_PFN_MIGRATE;
> >> +			continue;
> >> +		}
> >> +
> >> +		migrate->src[i] = migrate_pfn(page_to_pfn(hmm_pfn_to_page(pfn)))
> >> +			| MIGRATE_PFN_MIGRATE;
> >> +		migrate->src[i] |= (pfn & HMM_PFN_WRITE) ? MIGRATE_PFN_WRITE : 0;
> >> +	}
> >> +
> >> +	if (migrate->cpages)
> >> +		migrate_vma_unmap(migrate);
> >> +
> >> +}
> >> +EXPORT_SYMBOL(migrate_hmm_range_setup);
> >
> > I've not had a chance to test the code, do you have any numbers with the changes
> > to show the advantages of doing both fault and migrate together?
> 
> Not yet, but plan to have some numbers later. 

I don't have any recent numbers, but I profiled this a while ago and having to
walk the page tables multiple times was a significant overhead in our driver
at least.

> >
> > Balbir
> >
> Thanks,
> --Mika
>