Message-ID: <3734c5a7-6b3f-4af1-ac35-6bd680823be5@linux.dev>
Date: Tue, 23 Sep 2025 20:04:18 +0800
From: Lance Yang <lance.yang@...ux.dev>
To: David Hildenbrand <david@...hat.com>,
 Catalin Marinas <catalin.marinas@....com>
Cc: akpm@...ux-foundation.org, lorenzo.stoakes@...cle.com,
 usamaarif642@...il.com, yuzhao@...gle.com, ziy@...dia.com,
 baolin.wang@...ux.alibaba.com, baohua@...nel.org, voidice@...il.com,
 Liam.Howlett@...cle.com, cerasuolodomenico@...il.com, hannes@...xchg.org,
 kaleshsingh@...gle.com, npache@...hat.com, riel@...riel.com,
 roman.gushchin@...ux.dev, rppt@...nel.org, ryan.roberts@....com,
 dev.jain@....com, ryncsn@...il.com, shakeel.butt@...ux.dev,
 surenb@...gle.com, hughd@...gle.com, willy@...radead.org,
 matthew.brost@...el.com, joshua.hahnjy@...il.com, rakie.kim@...com,
 byungchul@...com, gourry@...rry.net, ying.huang@...ux.alibaba.com,
 apopple@...dia.com, qun-wei.lin@...iatek.com, Andrew.Yang@...iatek.com,
 casper.li@...iatek.com, chinwen.chang@...iatek.com,
 linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
 linux-mediatek@...ts.infradead.org, linux-mm@...ck.org, ioworker0@...il.com,
 stable@...r.kernel.org
Subject: Re: [PATCH 1/1] mm/thp: fix MTE tag mismatch when replacing
 zero-filled subpages



On 2025/9/23 20:00, David Hildenbrand wrote:
> On 23.09.25 13:52, Catalin Marinas wrote:
>> On Mon, Sep 22, 2025 at 07:59:00PM +0200, David Hildenbrand wrote:
>>> On 22.09.25 19:24, Catalin Marinas wrote:
>>>> On Mon, Sep 22, 2025 at 10:14:58AM +0800, Lance Yang wrote:
>>>>> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
>>>>> index 32e0ec2dde36..28d4b02a1aa5 100644
>>>>> --- a/mm/huge_memory.c
>>>>> +++ b/mm/huge_memory.c
>>>>> @@ -4104,29 +4104,20 @@ static unsigned long 
>>>>> deferred_split_count(struct shrinker *shrink,
>>>>>    static bool thp_underused(struct folio *folio)
>>>>>    {
>>>>>        int num_zero_pages = 0, num_filled_pages = 0;
>>>>> -    void *kaddr;
>>>>>        int i;
>>>>>        for (i = 0; i < folio_nr_pages(folio); i++) {
>>>>> -        kaddr = kmap_local_folio(folio, i * PAGE_SIZE);
>>>>> -        if (!memchr_inv(kaddr, 0, PAGE_SIZE)) {
>>>>> -            num_zero_pages++;
>>>>> -            if (num_zero_pages > khugepaged_max_ptes_none) {
>>>>> -                kunmap_local(kaddr);
>>>>> +        if (pages_identical(folio_page(folio, i), ZERO_PAGE(0))) {
>>>>> +            if (++num_zero_pages > khugepaged_max_ptes_none)
>>>>>                    return true;
>>>>
>>>> I wonder what the overhead of doing a memcmp() vs memchr_inv() is. The
>>>> former will need to read from two places. If it's noticeable, it would
>>>> affect architectures that don't have an MTE equivalent.
>>>>
>>>> Alternatively we could introduce something like folio_has_metadata()
>>>> which on arm64 simply checks PG_mte_tagged.
>>>
>>> We discussed something similar in the other thread (I suggested
>>> page_is_mergable()). I'd prefer to use pages_identical() for now, so 
>>> we have
>>> the same logic here and in ksm code.
>>>
>>> (this patch here almost looks like a cleanup :) )
>>>
>>> If this becomes a problem, what we could do is in pages_identical() 
>>> would be
>>> simply doing the memchr_inv() in case is_zero_pfn(). KSM might 
>>> benefit from
>>> that as well when merging with the shared zeropage through
>>> try_to_merge_with_zero_page().
>>
>> Yes, we can always optimise it later.
>>
>> I just realised that on arm64 with MTE we won't get any merging with the
>> zero page even if the user page isn't mapped with PROT_MTE. In
>> cpu_enable_mte() we zero the tags in the zero page and set
>> PG_mte_tagged. The reason is that we want to use the zero page with
>> PROT_MTE mappings (until tag setting causes CoW). Hmm, the arm64
>> memcmp_pages() messed up KSM merging with the zero page even before this
>> patch.
>>
>> The MTE tag setting evolved a bit over time with some locking using PG_*
>> flags to avoid a set_pte_at() race trying to initialise the tags on the
>> same page. We also moved the swap restoring to arch_swap_restore()
>> rather than the set_pte_at() path. So it is safe now to merge with the
>> zero page if the other page isn't tagged. A subsequent set_pte_at()
>> attempting to clear the tags would notice that the zero page is already
>> tagged.
>>
>> We could go a step further and add tag comparison (I had some code
>> around) but I think the quick fix is to just not treat the zero page as
>> tagged.
> 
> I assume any tag changes would result in CoW.
> 
> It would be interesting to know if there are use cases with VMs or other 
> workloads where that could be beneficial with KSM.
> 
>> Not fully tested yet:
>>
>> diff --git a/arch/arm64/kernel/mte.c b/arch/arm64/kernel/mte.c
>> index e5e773844889..72a1dfc54659 100644
>> --- a/arch/arm64/kernel/mte.c
>> +++ b/arch/arm64/kernel/mte.c
>> @@ -73,6 +73,8 @@ int memcmp_pages(struct page *page1, struct page 
>> *page2)
>>   {
>>       char *addr1, *addr2;
>>       int ret;
>> +    bool page1_tagged = page_mte_tagged(page1) && !is_zero_page(page1);
>> +    bool page2_tagged = page_mte_tagged(page2) && !is_zero_page(page2);
>>       addr1 = page_address(page1);
>>       addr2 = page_address(page2);
>> @@ -83,11 +85,10 @@ int memcmp_pages(struct page *page1, struct page 
>> *page2)
>>       /*
>>        * If the page content is identical but at least one of the 
>> pages is
>> -     * tagged, return non-zero to avoid KSM merging. If only one of the
>> -     * pages is tagged, __set_ptes() may zero or change the tags of the
>> -     * other page via mte_sync_tags().
>> +     * tagged, return non-zero to avoid KSM merging. Ignore the zero 
>> page
>> +     * since it is always tagged with the tags cleared.
>>        */
>> -    if (page_mte_tagged(page1) || page_mte_tagged(page2))
>> +    if (page1_tagged || page2_tagged)
>>           return addr1 != addr2;
> 
> That looks reasonable to me.

Yeah, looks good to me as well.

> 
> @Lance as you had a test setup, could you give this a try as well with 
> KSM shared zeropage deduplication enabled whether it now works as 
> expected as well?

Sure. I'll test that and get back to you ;)

> 
> Then, this should likely be an independent fix.
> 
> For KSM you likely have to enable it first through /sys/kernel/mm/ksm/ 
> use_zero_pages.

Got it.

Powered by blists - more mailing lists