linux-kernel - Re: [PATCH V3 1/2] mm: export folio_expected

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 for Android: free password hash cracker in your pocket

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

Message-ID: <03495e46-234a-4c0d-bec7-f7325789a160@amd.com>
Date: Tue, 22 Apr 2025 16:58:51 +0530
From: Shivank Garg <shivankg@....com>
To: David Hildenbrand <david@...hat.com>, shaggy@...nel.org,
 akpm@...ux-foundation.org
Cc: willy@...radead.org, wangkefeng.wang@...wei.com, jane.chu@...cle.com,
 ziy@...dia.com, donettom@...ux.ibm.com, apopple@...dia.com,
 jfs-discussion@...ts.sourceforge.net, linux-kernel@...r.kernel.org,
 linux-mm@...ck.org, syzbot+8bb6fd945af4e0ad9299@...kaller.appspotmail.com
Subject: Re: [PATCH V3 1/2] mm: export folio_expected_refs for JFS migration
 handler



On 4/17/2025 2:10 PM, David Hildenbrand wrote:
> On 17.04.25 08:06, Shivank Garg wrote:
>> Export folio_expected_refs() to allow filesystem-specific migration
>> handlers like JFS metapage_migrate_folio to properly verify reference
>> counts before migration.
>>
>> Signed-off-by: Shivank Garg <shivankg@....com>
>> ---
>>   include/linux/migrate.h | 1 +
>>   mm/migrate.c            | 3 ++-
>>   2 files changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/include/linux/migrate.h b/include/linux/migrate.h
>> index aaa2114498d6..cb31c5b1eb6a 100644
>> --- a/include/linux/migrate.h
>> +++ b/include/linux/migrate.h
>> @@ -60,6 +60,7 @@ struct movable_operations {
>>   /* Defined in mm/debug.c: */
>>   extern const char *migrate_reason_names[MR_TYPES];
>>   +int folio_expected_refs(struct address_space *mapping, struct folio *folio);
>>   #ifdef CONFIG_MIGRATION
>>     void putback_movable_pages(struct list_head *l);
>> diff --git a/mm/migrate.c b/mm/migrate.c
>> index 6e2488e5dbe4..0f01b8a87dec 100644
>> --- a/mm/migrate.c
>> +++ b/mm/migrate.c
>> @@ -445,7 +445,7 @@ void pmd_migration_entry_wait(struct mm_struct *mm, pmd_t *pmd)
>>   }
>>   #endif
>>   -static int folio_expected_refs(struct address_space *mapping,
>> +int folio_expected_refs(struct address_space *mapping,
>>           struct folio *folio)
>>   {
>>       int refs = 1;
>> @@ -458,6 +458,7 @@ static int folio_expected_refs(struct address_space *mapping,
>>         return refs;
>>   }
>> +EXPORT_SYMBOL_GPL(folio_expected_refs);
>>     /*
>>    * Replace the folio in the mapping.
> 
> Can we make that an inline function instead, and add some documentation?
> 
> This function is only suitable if we know the folio is unmapped from page tables (no references from page table mappings: !folio_mapped()).
> 
> So when exporting this function we either
> 
> a) Need a clearer name
> 
> b) Should generalize it to consider folio_mapcount() etc, and lookup the
>    mapping from the folio.
> 
> For your use case, a) might be easier. Maybe call  it something like "folio_migration_expected_refs" ?
> 

Thank you David for the review. I'll make it a static inline function in the header file
with proper documentation. I don't need to EXPORT it then, since it will be an inline function.

Best Regards,
Shivank