Message-ID: <13276FAF-85A1-41DB-8C30-CDA856FA6BF7@nvidia.com>
Date: Tue, 13 Jan 2026 17:02:44 -0500
From: Zi Yan <ziy@...dia.com>
To: Harry Yoo <harry.yoo@...cle.com>
Cc: Jiaqi Yan <jiaqiyan@...gle.com>, jackmanb@...gle.com, hannes@...xchg.org,
 linmiaohe@...wei.com, willy@...radead.org, nao.horiguchi@...il.com,
 david@...hat.com, lorenzo.stoakes@...cle.com, william.roche@...cle.com,
 tony.luck@...el.com, wangkefeng.wang@...wei.com, jane.chu@...cle.com,
 akpm@...ux-foundation.org, osalvador@...e.de, muchun.song@...ux.dev,
 rientjes@...gle.com, duenwen@...gle.com, jthoughton@...gle.com,
 linux-mm@...ck.org, linux-kernel@...r.kernel.org, Liam.Howlett@...cle.com,
 vbabka@...e.cz, rppt@...nel.org, surenb@...gle.com, mhocko@...e.com
Subject: Re: [PATCH v3 2/3] mm/page_alloc: only free healthy pages in
 high-order has_hwpoisoned folio

On 13 Jan 2026, at 0:39, Harry Yoo wrote:

> On Mon, Jan 12, 2026 at 12:49:22AM +0000, Jiaqi Yan wrote:
>> At the end of dissolve_free_hugetlb_folio(), a free HugeTLB folio
>> becomes non-HugeTLB, and it is released to buddy allocator
>> as a high-order folio, e.g. a folio that contains 262144 pages
>> if the folio was a 1G HugeTLB hugepage.
>>
>> This is problematic if the HugeTLB hugepage contained HWPoison
>> subpages. In that case, since buddy allocator does not check
>> HWPoison for non-zero-order folio, the raw HWPoison page can
>> be given out with its buddy page and be re-used by either
>> kernel or userspace.
>>
>> Memory failure recovery (MFR) in kernel does attempt to take
>> raw HWPoison page off buddy allocator after
>> dissolve_free_hugetlb_folio(). However, there is always a time
>> window between dissolve_free_hugetlb_folio() frees a HWPoison
>> high-order folio to buddy allocator and MFR takes HWPoison
>> raw page off buddy allocator.
>
> I wonder if this is something we want to backport to -stable.
>
>> One obvious way to avoid this problem is to add page sanity
>> checks in page allocate or free path. However, it is against
>> the past efforts to reduce sanity check overhead [1,2,3].
>>
>> Introduce free_has_hwpoisoned() to only free the healthy pages
>> and to exclude the HWPoison ones in the high-order folio.
>> The idea is to iterate through the sub-pages of the folio to
>> identify contiguous ranges of healthy pages. Instead of freeing
>> pages one by one, decompose healthy ranges into the largest
>> possible blocks having different orders. Every block meets the
>> requirements to be freed via __free_one_page().
>>
>> free_has_hwpoisoned() has linear time complexity wrt the number
>> of pages in the folio. While the power-of-two decomposition
>> ensures that the number of calls to the buddy allocator is
>> logarithmic for each contiguous healthy range, the mandatory
>> linear scan of pages to identify PageHWPoison() defines the
>> overall time complexity. For a 1G hugepage having several
>> HWPoison pages, free_has_hwpoisoned() takes around 2ms on
>> average.
>>
>> Since free_has_hwpoisoned() has nontrivial overhead, it is
>> wrapped inside free_pages_prepare_has_hwpoisoned() and done
>> only PG_has_hwpoisoned indicates HWPoison page exists and
>> after free_pages_prepare() succeeded.
>>
>> [1] https://lore.kernel.org/linux-mm/1460711275-1130-15-git-send-email-mgorman@techsingularity.net
>> [2] https://lore.kernel.org/linux-mm/1460711275-1130-16-git-send-email-mgorman@techsingularity.net
>> [3] https://lore.kernel.org/all/20230216095131.17336-1-vbabka@suse.cz
>>
>> Signed-off-by: Jiaqi Yan <jiaqiyan@...gle.com>
>>
>> ---
>>  mm/page_alloc.c | 157 +++++++++++++++++++++++++++++++++++++++++++++++-
>>  1 file changed, 154 insertions(+), 3 deletions(-)
>>
>> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
>> index 822e05f1a9646..9393589118604 100644
>> --- a/mm/page_alloc.c
>> +++ b/mm/page_alloc.c
>> @@ -2923,6 +2928,152 @@ static bool free_frozen_page_commit(struct zone *zone,
>>  	return ret;
>>  }
>
> From correctness point of view I think it looks good to me.
> Let's see what the page allocator folks say.
>
> A few nits below.
>
>> +static bool compound_has_hwpoisoned(struct page *page, unsigned int order)
>> +{
>> +	if (order == 0 || !PageCompound(page))
>> +		return false;
>
> nit: since order-0 compound page is not a thing,
> !PageCompound(page) check should cover order == 0 case.
>
>> +	return folio_test_has_hwpoisoned(page_folio(page));
>> +}
>> +
>> +/*
>> + * Do free_has_hwpoisoned() when needed after free_pages_prepare().
>> + * Returns
>> + * - true: free_pages_prepare() is good and caller can proceed freeing.
>> + * - false: caller should not free pages for one of the two reasons:
>> + *   1. free_pages_prepare() failed so it is not safe to proceed freeing.
>> + *   2. this is a compound page having some HWPoison pages, and healthy
>> + *      pages are already safely freed.
>> + */
>> +static bool free_pages_prepare_has_hwpoisoned(struct page *page,
>> +					      unsigned int order,
>> +					      fpi_t fpi_flags)
>
> nit: Hope we'll come up with a better name than
> free_pages_prepare_has_poisoned(), but I don't have any better
> suggestion... :)
>
> And I hope somebody familiar with compaction (as compaction_free() calls
> free_pages_prepare() and ignores its return value) could confirm that
> it is safe to do a compound_has_hwpoisoned() check and, when it returns
> true, call free_has_hwpoisoned() in free_pages_prepare(),
> so that we won't need a separate function to do this.

I wrote the function. compact_control->freepages[] is an array of free
pages isolated from buddy allocator as a temporary free page buffer.
I wonder if compaction_free() can get a HWPoisoned folio or not, since
to get to compaction_free(), the folio needs to be copied to a new folio.
If the memory goes bad, folio_mc_copy() will return -EHWPOISON to abort
the migration, but I do not see any code marking the folio has_hwpoisoned.
When it reaches to compact_free(), there is no way to detect the error.
I think the corrupted page will be put back to buddy and be marked as
has_hwpoison after it is allocated and accessed by an application.

You might need to add has_hwpoisoned code in migration code path to
catch folio_mc_copy() errors.


Best Regards,
Yan, Zi

Powered by blists - more mailing lists