Re: [PATCH v3 08/14] erofs: use folio_attach/detach_private() instead of direct assignment

[David Hildenbrand (Arm)]

On 9/8/26 19:19, Zi Yan wrote:
> On Tue Sep 8, 2026 at 12:13 PM EDT, David Hildenbrand (Arm) wrote:
>> On 9/8/26 04:56, Zi Yan wrote:
>>> erofs_onlinefolio_init/split/end() use folio->private without setting
>>> PG_private or increasing folio refcount and it works. But after PG_private
>>> is replaced by checking folio->private in a future commit, it can break
>>> folio_expected_ref_count(), since the folio has private data without
>>> elevated refcount. Change them to use folio_attach/detach_private().
>>
>> Freezing the folio will properly work I assume, as we'd always freeze with that
>> refcount in mind or remove the reference (detach private) early?
> 
> Yes.
> 
>>
>> I'm thinking of folio_ref_freeze() in __folio_freeze_and_split_unmapped().
>>
>> Or is the ref only very short-lived either way such that we don't care?
> 
> The folios here are locked since they are in the process of
> readahead, which requires all folios locked. So they are not reachable
> from folio split code. In additio, the private is deatched and ref is
> decreased before the folio lock is dropped.

It would be great to summarize this interaction with core-mm (migration/split)
in the patch description.

-- 
Cheers,

David