Re: [PATCH 05/11] mm/sparse-vmemmap: set section order for device DAX

[Muchun Song <muchun.song@linux.dev>]

> On Sep 5, 2026, at 16:20, Qi Zheng <qi.zheng@linux.dev> wrote:
> 
> Hi Muchun,

Hi,

> 
> On 8/31/26 3:53 PM, Muchun Song wrote:
>> Device DAX can use vmemmap optimization only when a full section is
>> populated with a compound-page geometry. Record that geometry in the
>> section order before populating the section, so later vmemmap accounting
>> and population decisions can use the section state directly.
>> Clear the section order when the section becomes empty again. Also reject
>> partial additions to a section that already has optimized vmemmap mappings,
>> because a section cannot safely mix optimized and ordinary vmemmap layouts.
> 
> Perhaps we should explain why they cannot be safely mixed.

Yes, because compound_nr_pages() determines the number of struct pages to
initialize with section as the smallest granularity, they cannot be mixed.

> 
>> Partial additions continue to use ordinary vmemmap population, so they do
>> not save vmemmap memory. Such additions are uncommon, and the lost saving
>> is negligible.
>> Signed-off-by: Muchun Song <songmuchun@bytedance.com>
>> ---
>>  mm/mm_init.c        | 13 ++++---------
>>  mm/sparse-vmemmap.c | 16 ++++++++++++----
>>  2 files changed, 16 insertions(+), 13 deletions(-)
> 
> Also, the issue sashiko reported looks like it was pre-existing, not
> introduced by this patch. So:
> 
> Acked-by: Qi Zheng <qi.zheng@linux.dev>

Thanks.

> 
> Thanks,
> Qi
>