Re: [PATCH v2] hfs: avoid reusing occupied CNIDs after deletion

From: Hengyu Liang

Date: Sat Nov 21 2026 - 05:19:28 EST


Hi Slava,

Sorry for the late reply. I have an important deadline in early
December, so I may be slow to reply until then.

On Mon, 2026-11-16 at 15:21 -0800, Viacheslav Dubeyko wrote:
> You explained a simple case. :) My worry was about the case when we
> could delete some number of items between [16, 20]. For example, we
> delete 17, 18, and 19 at first. And now we delete item with #20. So,
> leaf A should have 16 as a maximum CNID. And we can use 17 as a
> next_id. Correct?

Correct. I ran this case with the patch on a fresh volume. Creating
a, c, e, b, d gives CNIDs 16 to 20 in three leaves:

leaf 1: root folder (2), thread of 2, a (16), b (19) max 19
leaf 2: c (17), d (20), e (18) max 20
leaf 4: threads of 16 to 20 no file or folder

rm c (17), rm e (18): not the newest CNID, next_id stays 21;
leaf maximums 19 and 20 stay
rm b (19): leaf 1's maximum has to be recomputed, the
root keeps 20 (from leaf 2)
rm d (20): leaf 2 is now empty and goes away, the root's
maximum has to be recomputed: leaf 1 -> 16,
root -> 16, next_id = 17

The next new file gets CNID 17. I checked it three ways: with the
values cached before the deletes, with nothing cached yet, and with a
remount between the creates and the deletes. A debug build that
compares every cached value with the records after each insert and
remove found no mismatch.

A larger case: 2000 files created in shuffled name order, so the
CNIDs are spread over the leaves. Deleting CNIDs 1916..2014 first and
then 2015 gives next_id 1916, and the next file gets 1916. Without the
patch the same steps end with "File exists". fsck.hfs reports all
images as OK.

Thanks,
Hengyu