linux-kernel - Re: [PATCH] mm: fix apply_to_existing_page

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <297fdd97-d53b-4018-bd15-50235e235561@redhat.com>
Date: Wed, 9 Apr 2025 12:26:09 +0200
From: David Hildenbrand <david@...hat.com>
To: "Kirill A. Shutemov" <kirill@...temov.name>
Cc: "Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>,
 Andrew Morton <akpm@...ux-foundation.org>, Vlastimil Babka <vbabka@...e.cz>,
 linux-mm@...ck.org, linux-kernel@...r.kernel.org,
 Daniel Axtens <dja@...ens.net>
Subject: Re: [PATCH] mm: fix apply_to_existing_page_range()

On 09.04.25 12:23, Kirill A. Shutemov wrote:
> On Wed, Apr 09, 2025 at 11:52:43AM +0200, David Hildenbrand wrote:
>> On 09.04.25 11:40, Kirill A. Shutemov wrote:
>>> In the case of apply_to_existing_page_range(), apply_to_pte_range() is
>>> reached with 'create' set to false. When !create, the loop over the PTE
>>> page table is broken.
>>>
>>> apply_to_pte_range() will only move to the next PTE entry if 'create' is
>>> true or if the current entry is not pte_none().
>>>
>>> This means that the user of apply_to_existing_page_range() will not have
>>> 'fn' called for any entries after the first pte_none() in the PTE page
>>> table.
>>>
>>> Fix the loop logic in apply_to_pte_range().
>>>
>>> Signed-off-by: Kirill A. Shutemov <kirill.shutemov@...ux.intel.com>
>>> Fixes: be1db4753ee6 ("mm/memory.c: add apply_to_existing_page_range() helper")
>>> Cc: Daniel Axtens <dja@...ens.net>
>>> ---
>>>    mm/memory.c | 4 ++--
>>>    1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/mm/memory.c b/mm/memory.c
>>> index fb7b8dc75167..2094564f4dfb 100644
>>> --- a/mm/memory.c
>>> +++ b/mm/memory.c
>>> @@ -2907,11 +2907,11 @@ static int apply_to_pte_range(struct mm_struct *mm, pmd_t *pmd,
>>>    	if (fn) {
>>>    		do {
>>>    			if (create || !pte_none(ptep_get(pte))) {
>>> -				err = fn(pte++, addr, data);
>>> +				err = fn(pte, addr, data);
>>>    				if (err)
>>>    					break;
>>>    			}
>>> -		} while (addr += PAGE_SIZE, addr != end);
>>> +		} while (pte++, addr += PAGE_SIZE, addr != end);
>>>    	}
>>>    	*mask |= PGTBL_PTE_MODIFIED;
>>
>> LGTM. just curious, did you run into any actual issues that are worth
>> describing?
> 
> I stepped on it in my non-upstream code debugging. I am not sure how it
> affects existing users.
> 
>> It should affect apply_to_existing_page_range() users where create==false.
>> There are not many, and likely most PTEs in the range they are passing are
>> all non-none.
> 
> Or we just silently leak memory :P

That's exactly what I am trying to figure out: is there something 
upstream that could actually run into this such that we should CC stable?

-- 
Cheers,

David / dhildenb