Re: [PATCH] mm: mglru: splice all generations when reparenting a memcg
From: Kairui Song
Date: Wed Nov 11 2026 - 04:03:26 EST
On Wed, Nov 11, 2026 at 8:29 AM Barry Song <baohua@xxxxxxxxxx> wrote:
>
> On Wed, Nov 11, 2026 at 8:09 AM Shakeel Butt <shakeel.butt@xxxxxxxxx> wrote:
> >
> > Noah reported list_del corruption on single-node hosts running 7.2.11,
> > which already has commit bf4ade7dbd76 ("memcg: keep folio's objcg same
> > as its node"). The bad pointer is an MGLRU list head in a freed
> > mem_cgroup_per_node:
> >
> > slab mem_cgroup_per_node start ff31bfa36a3c9e00 pointer offset 528 size 1920
> > list_del corruption. next->prev should be ffacf881ccc70e08, but was ff31bfa36a3ca010. (next=ff31bfa36a3ca010)
> >
> > When a memcg goes offline, __lru_gen_reparent_memcg() moves its MGLRU
> > lists to the parent. It moves only the generations (gens) from min_seq
> > to max_seq, and it puts each list at the same index in the parent.
> >
> > A folio's gen in folio->flags can be ahead of the list it is on. In the
> > parent, the same index can stand for an older gen, so the folio's gen
> > can end up older than its list. Once the parent's min_seq passes that
> > gen, sort_folio() puts the folio on the list for max_seq + 1, where it
> > stays until the next aging.
> >
> > If the parent goes offline before that aging, the folio is not moved to
> > the grandparent. It stays on the parent's list after the parent is
> > freed, and the next list_del() on it touches freed memory. This needs
> > two cgroups on one branch to go offline one after the other, such as a
> > container and then its pod.
> >
> > The parent always has MAX_NR_GENS gens at this point, so every index is
> > valid in it. Move all MAX_NR_GENS lists, not only the gens from min_seq
> > to max_seq.
> >
> > A reproducer that drives these steps through the lru_gen debugfs file
> > gets a KASAN use-after-free on the freed mem_cgroup_per_node without
> > this patch, and no report with it.
> >
> > Fixes: f304652609ea ("mm: vmscan: prepare for reparenting MGLRU folios")
> > Reported-by: Noah Feldt <noah@feldt.systems>
> > Closes: https://lore.kernel.org/all/20261110132853.71228-1-noah@feldt.systems/
> > Signed-off-by: Shakeel Butt <shakeel.butt@xxxxxxxxx>
> > Cc: <stable@xxxxxxxxxxxxxxx>
>
> Thanks very much, Shakeel! I saw Kairui has exactly the same code here:
>
> [PATCH v4 14/17] mm/mglru: reparent folios from all generations
> https://lore.kernel.org/linux-mm/20261016-mglru-fg-v4-14-2106750c0bfa@xxxxxxxxxxx/
Thanks for looking into this!
FG needed that because FG's folio_inc_lru_refs is a generic MGLRU
lockless lazy promotion, it may cause race between aging & lazy
promotion, landing a folio in a stall gen list. This isn't a big
problem, the worst outcome is a hotness inversion, which happens
almost once in a million times.
And right as Shakeel pointed out the problem already exists for
reparenting and lazy promotion, so we can fix it first.
>
> somehow, Shakeel's patch makes us need the change earlier even before
> mglru-fg lands?
Landing this one first is totally fine by me. The code is actually
safer and cleaner, in my opinion, with this patch.