Re: [PATCH] mm/hugetlb: fix memory offline failure due to hwpoisoned file hugetlb

[Miaohe Lin <linmiaohe@huawei.com>]

On 2026/3/20 10:43, Jinjiang Tu wrote:
> 
> 在 2026/3/20 10:34, Miaohe Lin 写道:
>> On 2026/3/18 10:07, Jinjiang Tu wrote:
>>> When a file hugetlb folio triggers UCE, me_huge_page() will keep the
>>> hugetlb folio in pagcahe with refcount increased and PG_hwpoison set. Even
>>> after the hugetlb file is deleted, the hugetlb folio is still leaked.
>>>
>>> If we want to offline the memory block that the hwpoisoned hugetlb folio
>>> belongs to, it fails in dissolve_free_hugetlb_folios() due to the
>>> hwpoisoned hugetlb folio isn't free.
>>>
>>> I can reproduce this issue with the following steps in qemu:
>>>   1) echo offline >/sys/devices/system/memory/auto_online_blocks
>>>   2) in qemu monitor:
>>>         object_add memory-backend-ram,id=mem10,size=1G
>>>         device_add pc-dimm,id=dimm1,memdev=mem10,node=2
>>>   3) echo online_movable > /sys/devices/system/node/node2/memory136/state
>>>   4) echo 5 > /sys/devices/system/node/node2/hugepages/hugepages-2048kB/nr_hugepages
>>>   5) run ./hugetlb_file. This process will receive SIGBUS.
>>>   6) remove the hugetlbfs file.
>>>   7) echo offline > /sys/devices/system/node/node2/memory136/state
>>>
>>> hugetlb_file.c:
>>>    fd = open("/dev/hugepages/my_hugepage_file", O_CREAT | O_RDWR, 0755);
>>>    fallocate(fd, 0, 0, HUGEPAGE_SIZE * 2);
>>>    addr = mmap(NULL, HUGEPAGE_SIZE * 2, PROT_READ | PROT_WRITE,
>>>         MAP_SHARED | MAP_HUGETLB, fd, 0);
>>>    memset(addr, 0xaa, HUGEPAGE_SIZE * 2);
>>>    madvise(addr, HUGEPAGE_SIZE, MADV_HWPOISON);
>>>
>>> To fix it, when deleting hugetlb folio from pagecache, mark the hugetlb
>>> folio temporary, and put the refcount increased by memory-failure. After
>>> the hugetlb folio is deleted from pagecache, the refcount is decreased to
>>> zero and the hugetlb folio is dissolved.
>> Thanks for your patch.
>>
>>> Signed-off-by: Jinjiang Tu <tujinjiang@huawei.com>
>>> ---
>>>   fs/hugetlbfs/inode.c | 5 +++++
>>>   1 file changed, 5 insertions(+)
>>>
>>> diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c
>>> index 3f70c47981de..6bebe2e67f3e 100644
>>> --- a/fs/hugetlbfs/inode.c
>>> +++ b/fs/hugetlbfs/inode.c
>>> @@ -603,6 +603,11 @@ static void remove_inode_hugepages(struct inode *inode, loff_t lstart,
>>>                               index, truncate_op))
>>>                   freed++;
>>>   +            if (unlikely(folio_test_hwpoison(folio))) {
>>> +                folio_set_hugetlb_temporary(folio);
>> I think it is not needed to mark the hugetlb folio as temporary because
>> offline_pages() will call dissolve_free_hugetlb_folios().
> 
> Indeed. I was intended to avoid this hugetlb allocated again, but
> dequeue_hugetlb_folio_node_exact() will check PG_hwpoison.
> 
>>
>>> +                folio_put(folio);
>> __get_huge_page_for_hwpoison() will always set hwpoison for hugetlb folio
>> even without page refcnt increased. So this folio_put() might be unexpected.
>> Please see [1] for detail.
> 
> When memory-failure races with migration, __get_huge_page_for_hwpoison()
> will set hwpoison without page refcnt increased. In common case, ref is
> increased before setting hwpoison.

Right. And in those rare cases, VM_BUG_ON_PAGE(page_ref_count(page) == 0, page) will be triggered
in folio_put().

Thanks.
.