linux-kernel - Re: [PATCH] mm: skip reserved page for kmem leak scanning

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]

Date:   Tue, 30 Aug 2022 10:03:21 +0200
From:   David Hildenbrand <david@...hat.com>
To:     Zhaoyang Huang <huangzhaoyang@...il.com>
Cc:     "zhaoyang.huang" <zhaoyang.huang@...soc.com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Catalin Marinas <catalin.marinas@....com>,
        "open list:MEMORY MANAGEMENT" <linux-mm@...ck.org>,
        LKML <linux-kernel@...r.kernel.org>, Ke Wang <ke.wang@...soc.com>
Subject: Re: [PATCH] mm: skip reserved page for kmem leak scanning

On 30.08.22 04:41, Zhaoyang Huang wrote:
> On Mon, Aug 29, 2022 at 8:19 PM David Hildenbrand <david@...hat.com> wrote:
>>
>> On 26.08.22 05:23, Zhaoyang Huang wrote:
>>> On Fri, Aug 26, 2022 at 11:13 AM zhaoyang.huang
>>> <zhaoyang.huang@...soc.com> wrote:
>>>>
>>>> From: Zhaoyang Huang <zhaoyang.huang@...soc.com>
>>>>
>>>> It is no need to scan reserved page, skip it.
>>>>
>>>> Signed-off-by: Zhaoyang Huang <zhaoyang.huang@...soc.com>
>>>> ---
>>>>  mm/kmemleak.c | 2 +-
>>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>>
>>>> diff --git a/mm/kmemleak.c b/mm/kmemleak.c
>>>> index a182f5d..c546250 100644
>>>> --- a/mm/kmemleak.c
>>>> +++ b/mm/kmemleak.c
>>>> @@ -1471,7 +1471,7 @@ static void kmemleak_scan(void)
>>>>                         if (page_zone(page) != zone)
>>>>                                 continue;
>>>>                         /* only scan if page is in use */
>>>> -                       if (page_count(page) == 0)
>>>> +                       if (page_count(page) == 0 || PageReserved(page))
>>> Sorry for previous stupid code by my faint, correct it here
>>
>> Did you even test the initial patch?
>>
>> I wonder why we should consider this change
>>
>> (a) I doubt it's a performance issue. If it is, please provide numbers
>>     before/after.
> For Android-like SOC systems where AP(cpu runs linux) are one of the
> memory consumers which are composed of other processors such as modem,
> isp,wcn etc. The reserved memory occupies a certain number of
> memory(could be 30% of MemTotal) which makes scan reserved pages
> pointless.

But we only scan the memmap (struct page) here and not the actual
memory. Do you have any performance numbers showing that there is even
an observable change?

>> (b) We'll stop scanning early allocations. As the memmap is usually
>>     allocated early during boot ... we'll stop scanning essentially the
>>     whole mmap and that whole loop would be dead code? What am i
>>     missing?
> memmap refers to pages here? If we can surpass these as it exist
> permanently during life period. Besides, I wonder if PageLRU should
> also be skipped?
> -                       if (page_count(page) == 0)
> +                       if (page_count(page) == 0 ||
> PageReserved(page) || PageLRU(page))

I think we need a really good justification to start poking holes into
the memmap scanner. I'm no expert on this code (and under which
circumstances we actually might find referenced objects in a memmap),
though.

But we should be careful with that.

-- 
Thanks,

David / dhildenb