Re: [PATCH] mm: swap: check for xa_zero_entry() on vma in swapoff path

[Charan Teja Kalla <quic_charante@quicinc.com>]

Thanks David, for the reply!!
On 8/8/2025 5:34 PM, David Hildenbrand wrote:
>>         if (mpnt) {
>>             mas_set_range(&vmi.mas, mpnt->vm_start, mpnt->vm_end - 1);
>>             mas_store(&vmi.mas, XA_ZERO_ENTRY);
>>             /* Avoid OOM iterating a broken tree */
>>             set_bit(MMF_OOM_SKIP, &mm->flags);
>>         }
>>         /*
>>          * The mm_struct is going to exit, but the locks will be dropped
>>          * first.  Set the mm_struct as unstable is advisable as it is
>>          * not fully initialised.
>>          */
>>         set_bit(MMF_UNSTABLE, &mm->flags);
>>     }
>>
>> Shouldn't we just remove anything from the tree here that was not copied
>> immediately?
> 
> Another fix would be to just check MMF_UNSTABLE in unuse_mm(). But
> having these MMF_UNSTABLE checks all over the place feels a bit like
> whack-a-mole.
> 
Seems MMF_UNSTABLE is the expectation per the commit,
64c37e134b12("kernel: be more careful about dup_mmap() failures and
uprobe registering"). Excerpt(s) from the commit message:

This patch sets the MMF_OOM_SKIP to avoid the iteration of the vmas on
the oom side (even though this is extremely unlikely to be selected as
an oom victim in the race window), and __sets MMF_UNSTABLE to avoid
other potential users from using a partially initialised mm_struct.

When registering vmas for uprobe, skip the vmas in an mm that is marked
unstable.  Modifying a vma in an unstable mm may cause issues if the mm
isn't fully initialised.__

> Is there anything preventing us from just leaving a proper tree that
> reflects reality in place before we drop the write lock?

When you mean proper tree, is this about the your previous question? --
Shouldn't we just remove anything from the tree here that was not copied
immediately?

Anyway, would request Liam/Lorenzo to comment on this.

Thanks,
Charan