[PATCH] mm: fix panic caused by __page_handle_poison()

[PATCH] mm: fix panic caused by __page_handle_poison()

[王贇]

By commit 510d25c92ec4 ("mm/hwpoison: disable pcp for
page_handle_poison()"), __page_handle_poison() was
introduced, and if we mark:

RET_A = dissolve_free_huge_page();
RET_B = take_page_off_buddy();

then __page_handle_poison was supposed to return TRUE When
RET_A == 0 && RET_B == TRUE

But since it failed to take care the case when RET_A is
-EBUSY or -ENOMEM, and just return the ret as a bool which
actually become TRUE, it break the original logical.

The following result is a huge page in freelist but was
referenced as poisoned, and lead into the final panic:

  kernel BUG at mm/internal.h:95!
  invalid opcode: 0000 [#1] SMP PTI
  skip...
  RIP: 0010:set_page_refcounted mm/internal.h:95 [inline]
  RIP: 0010:remove_hugetlb_page+0x23c/0x240 mm/hugetlb.c:1371
  skip...
  Call Trace:
   remove_pool_huge_page+0xe4/0x110 mm/hugetlb.c:1892
   return_unused_surplus_pages+0x8d/0x150 mm/hugetlb.c:2272
   hugetlb_acct_memory.part.91+0x524/0x690 mm/hugetlb.c:4017

This patch replace 'bool' with 'int' to handle RET_A correctly.

Reported-by: Abaci <abaci@linux.alibaba.com>
Signed-off-by: Michael Wang <yun.wang@linux.alibaba.com>
---
 mm/memory-failure.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 470400c..0fff717 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -68,7 +68,7 @@

 static bool __page_handle_poison(struct page *page)
 {
-	bool ret;
+	int ret;

 	zone_pcp_disable(page_zone(page));
 	ret = dissolve_free_huge_page(page);
@@ -76,7 +76,7 @@ static bool __page_handle_poison(struct page *page)
 		ret = take_page_off_buddy(page);
 	zone_pcp_enable(page_zone(page));

-	return ret;
+	return ret > 0;
 }

 static bool page_handle_poison(struct page *page, bool hugepage_or_freepage, bool release)
-- 
1.8.3.1

Re: [PATCH] mm: fix panic caused by __page_handle_poison()

[HORIGUCHI NAOYA(堀口 直也)]

On Mon, Aug 30, 2021 at 06:07:56PM +0800, 王贇 wrote:
> By commit 510d25c92ec4 ("mm/hwpoison: disable pcp for
> page_handle_poison()"), __page_handle_poison() was
> introduced, and if we mark:
> 
> RET_A = dissolve_free_huge_page();
> RET_B = take_page_off_buddy();
> 
> then __page_handle_poison was supposed to return TRUE When
> RET_A == 0 && RET_B == TRUE
> 
> But since it failed to take care the case when RET_A is
> -EBUSY or -ENOMEM, and just return the ret as a bool which
> actually become TRUE, it break the original logical.

s/logical/logic/ ?

> 
> The following result is a huge page in freelist but was
> referenced as poisoned, and lead into the final panic:
> 
>   kernel BUG at mm/internal.h:95!
>   invalid opcode: 0000 [#1] SMP PTI
>   skip...
>   RIP: 0010:set_page_refcounted mm/internal.h:95 [inline]
>   RIP: 0010:remove_hugetlb_page+0x23c/0x240 mm/hugetlb.c:1371
>   skip...
>   Call Trace:
>    remove_pool_huge_page+0xe4/0x110 mm/hugetlb.c:1892
>    return_unused_surplus_pages+0x8d/0x150 mm/hugetlb.c:2272
>    hugetlb_acct_memory.part.91+0x524/0x690 mm/hugetlb.c:4017
> 
> This patch replace 'bool' with 'int' to handle RET_A correctly.
> 
> Reported-by: Abaci <abaci@linux.alibaba.com>
> Signed-off-by: Michael Wang <yun.wang@linux.alibaba.com>

Thank you very much, this fix is totally right.

Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>

Could you add the following tags, too?

Fixes: 510d25c92ec4 ("mm/hwpoison: disable pcp for page_handle_poison()")
Cc: <stable@vger.kernel.org> # 5.14

Thanks,
Naoya Horiguchi

Re: [PATCH] mm: fix panic caused by __page_handle_poison()

[王贇]

On 2021/8/31 下午5:11, HORIGUCHI NAOYA(堀口 直也) wrote:
> On Mon, Aug 30, 2021 at 06:07:56PM +0800, 王贇 wrote:
>> By commit 510d25c92ec4 ("mm/hwpoison: disable pcp for
>> page_handle_poison()"), __page_handle_poison() was
>> introduced, and if we mark:
>>
>> RET_A = dissolve_free_huge_page();
>> RET_B = take_page_off_buddy();
>>
>> then __page_handle_poison was supposed to return TRUE When
>> RET_A == 0 && RET_B == TRUE
>>
>> But since it failed to take care the case when RET_A is
>> -EBUSY or -ENOMEM, and just return the ret as a bool which
>> actually become TRUE, it break the original logical.
> 
> s/logical/logic/ ?

Will Fix it in v2.

> 
>>
>> The following result is a huge page in freelist but was
>> referenced as poisoned, and lead into the final panic:
>>
>>   kernel BUG at mm/internal.h:95!
>>   invalid opcode: 0000 [#1] SMP PTI
>>   skip...
>>   RIP: 0010:set_page_refcounted mm/internal.h:95 [inline]
>>   RIP: 0010:remove_hugetlb_page+0x23c/0x240 mm/hugetlb.c:1371
>>   skip...
>>   Call Trace:
>>    remove_pool_huge_page+0xe4/0x110 mm/hugetlb.c:1892
>>    return_unused_surplus_pages+0x8d/0x150 mm/hugetlb.c:2272
>>    hugetlb_acct_memory.part.91+0x524/0x690 mm/hugetlb.c:4017
>>
>> This patch replace 'bool' with 'int' to handle RET_A correctly.
>>
>> Reported-by: Abaci <abaci@linux.alibaba.com>
>> Signed-off-by: Michael Wang <yun.wang@linux.alibaba.com>
> 
> Thank you very much, this fix is totally right.
> 
> Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
> 
> Could you add the following tags, too?
> 
> Fixes: 510d25c92ec4 ("mm/hwpoison: disable pcp for page_handle_poison()")
> Cc: <stable@vger.kernel.org> # 5.14

Thanks for the review, will add in v2 and send out soon.

Regards,
Michael Wang

> 
> Thanks,
> Naoya Horiguchi
> 

[PATCH v2] mm: fix panic caused by __page_handle_poison()

[王贇]

By commit 510d25c92ec4 ("mm/hwpoison: disable pcp for
page_handle_poison()"), __page_handle_poison() was
introduced, and if we mark:

RET_A = dissolve_free_huge_page();
RET_B = take_page_off_buddy();

then __page_handle_poison was supposed to return TRUE When
RET_A == 0 && RET_B == TRUE

But since it failed to take care the case when RET_A is
-EBUSY or -ENOMEM, and just return the ret as a bool which
actually become TRUE, it break the original logic.

The following result is a huge page in freelist but was
referenced as poisoned, and lead into the final panic:

  kernel BUG at mm/internal.h:95!
  invalid opcode: 0000 [#1] SMP PTI
  skip...
  RIP: 0010:set_page_refcounted mm/internal.h:95 [inline]
  RIP: 0010:remove_hugetlb_page+0x23c/0x240 mm/hugetlb.c:1371
  skip...
  Call Trace:
   remove_pool_huge_page+0xe4/0x110 mm/hugetlb.c:1892
   return_unused_surplus_pages+0x8d/0x150 mm/hugetlb.c:2272
   hugetlb_acct_memory.part.91+0x524/0x690 mm/hugetlb.c:4017

This patch replace 'bool' with 'int' to handle RET_A correctly.

Fixes: 510d25c92ec4 ("mm/hwpoison: disable pcp for page_handle_poison()")
Cc: <stable@vger.kernel.org> # 5.14
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Reported-by: Abaci <abaci@linux.alibaba.com>
Signed-off-by: Michael Wang <yun.wang@linux.alibaba.com>
---
 mm/memory-failure.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 470400c..0fff717 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -68,7 +68,7 @@

 static bool __page_handle_poison(struct page *page)
 {
-	bool ret;
+	int ret;

 	zone_pcp_disable(page_zone(page));
 	ret = dissolve_free_huge_page(page);
@@ -76,7 +76,7 @@ static bool __page_handle_poison(struct page *page)
 		ret = take_page_off_buddy(page);
 	zone_pcp_enable(page_zone(page));

-	return ret;
+	return ret > 0;
 }

 static bool page_handle_poison(struct page *page, bool hugepage_or_freepage, bool release)
-- 
1.8.3.1