Re: [PATCH] mm: fix pgdat->kswap accessed concurrently

[Kefeng Wang <wangkefeng.wang@huawei.com>]

On 2022/8/23 9:07, Kefeng Wang wrote:
>
> On 2022/8/21 4:59, Andrew Morton wrote:
>> On Sat, 20 Aug 2022 15:33:04 +0800 Muchun Song 
>> <muchun.song@linux.dev> wrote:
>>
>>>
>>>> +    if (IS_ERR(t)) {
>>>>         /* failure at boot is fatal */
>>>>         BUG_ON(system_state < SYSTEM_RUNNING);
>>>>         pr_err("Failed to start kswapd on node %d\n", nid);
>>>> -        pgdat->kswapd = NULL;
>>>> +        WRITE_ONCE(pgdat->kswapd, NULL);
>>>> +    } else {
>>>> +        WRITE_ONCE(pgdat->kswapd, t);
>>>>     }
>>>> }
>>> IIUC, the race is like the followings:
>>>
>>> CPU 0:                    CPU 1:
>>>
>>> kswapd_run()
>>>     pgdat->kswapd = kthread_run()
>>>     if (IS_ERR(pgdat->kswapd))
>>>                     kswapd_is_running
>>>                         // load pgdat->kswapd and it is NOT NULL.
>>>         pgdat->kswapd = NULL
>>>                         task_is_running(pgdat->kswapd); // NULL 
>>> pointer dereference
>>>
>> But don't we still have a bug?  Sure, kswapd_is_running() will no
>> longer deref a null pointer.  But it now runs kswapd_is_running()
>> against a task which has exited - a use-after-free?

The UAF is caused by race between kswapd_stop() and kcompactd(), right?

so  kcompactd() should be stop before kswapd_stop() to avoid the above UAF.

$ git diff
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index fad6d1f2262a..2fd45ccbce45 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1940,8 +1940,8 @@ int __ref offline_pages(unsigned long start_pfn, 
unsigned long nr_pages,

         node_states_clear_node(node, &arg);
         if (arg.status_change_nid >= 0) {
-               kswapd_stop(node);
                 kcompactd_stop(node);
+               kswapd_stop(node);
         }

         writeback_set_ratelimit();

> we could add get/put_task_struct() to avoid the UAF， will update， 
> thanks.

sorry, the task refcount won't fix anything.


> .