linux-kernel - Re: [PATCH v2] mempolicy: Optimize queue_folios_pte

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <56d25bb1-c6ff-4e92-bbab-72c9eee8015a@redhat.com>
Date: Tue, 15 Apr 2025 19:19:42 +0200
From: David Hildenbrand <david@...hat.com>
To: Dev Jain <dev.jain@....com>, akpm@...ux-foundation.org
Cc: ryan.roberts@....com, willy@...radead.org, linux-mm@...ck.org,
 linux-kernel@...r.kernel.org, hughd@...gle.com, vishal.moola@...il.com,
 yang@...amperecomputing.com, ziy@...dia.com
Subject: Re: [PATCH v2] mempolicy: Optimize queue_folios_pte_range by PTE
 batching

On 15.04.25 16:57, Dev Jain wrote:
> After the check for queue_folio_required(), the code only cares about the
> folio in the for loop, i.e the PTEs are redundant. Therefore, optimize
> this loop by skipping over a PTE batch mapping the same folio.
> 
> With a test program migrating pages of the calling process, which includes
> a mapped VMA of size 4GB with pte-mapped large folios of order-9, and
> migrating once back and forth node-0 and node-1, the average execution
> time reduces from 7.5 to 4 seconds, giving an approx 47% speedup.
> 
> v1->v2:
>   - Follow reverse xmas tree declarations
>   - Don't initialize nr
>   - Move folio_pte_batch() immediately after retrieving a normal folio
>   - increment nr_failed in one shot
> 
> Signed-off-by: Dev Jain <dev.jain@....com>
> ---
>   mm/mempolicy.c | 12 ++++++++++--
>   1 file changed, 10 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/mempolicy.c b/mm/mempolicy.c
> index b28a1e6ae096..ca90cdcd3207 100644
> --- a/mm/mempolicy.c
> +++ b/mm/mempolicy.c
> @@ -566,6 +566,7 @@ static void queue_folios_pmd(pmd_t *pmd, struct mm_walk *walk)
>   static int queue_folios_pte_range(pmd_t *pmd, unsigned long addr,
>   			unsigned long end, struct mm_walk *walk)
>   {
> +	const fpb_t fpb_flags = FPB_IGNORE_DIRTY | FPB_IGNORE_SOFT_DIRTY;
>   	struct vm_area_struct *vma = walk->vma;
>   	struct folio *folio;
>   	struct queue_pages *qp = walk->private;
> @@ -573,6 +574,7 @@ static int queue_folios_pte_range(pmd_t *pmd, unsigned long addr,
>   	pte_t *pte, *mapped_pte;
>   	pte_t ptent;
>   	spinlock_t *ptl;
> +	int max_nr, nr;
>   
>   	ptl = pmd_trans_huge_lock(pmd, vma);
>   	if (ptl) {
> @@ -586,7 +588,8 @@ static int queue_folios_pte_range(pmd_t *pmd, unsigned long addr,
>   		walk->action = ACTION_AGAIN;
>   		return 0;
>   	}
> -	for (; addr != end; pte++, addr += PAGE_SIZE) {
> +	for (; addr != end; pte += nr, addr += nr * PAGE_SIZE) {
> +		nr = 1;
>   		ptent = ptep_get(pte);
>   		if (pte_none(ptent))
>   			continue;
> @@ -598,6 +601,11 @@ static int queue_folios_pte_range(pmd_t *pmd, unsigned long addr,
>   		folio = vm_normal_folio(vma, addr, ptent);
>   		if (!folio || folio_is_zone_device(folio))
>   			continue;
> +		if (folio_test_large(folio) &&
> +		   (max_nr = ((end - addr) >> PAGE_SHIFT)) != 1)

That's real nasty :)

Let's simply do at the beginning of the loop:

	max_nr = (end - addr) >> PAGE_SHIFT;
	nr = 1;

Then here

if (folio_test_large(folio) && max_nr != 1)
	nr = ...

The compiler is smart enough to optimize the computation of values where 
really required.

With that

Acked-by: David Hildenbrand <david@...hat.com>

Thanks!

-- 
Cheers,

David / dhildenb