linux-kernel - Re: [PATCH 1/1] mm/page_poison: slightly optimize check_poison

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] [day] [month] [year] [list]

Message-ID: <a8f186c5-78fc-30b6-4019-80325db540c3@huawei.com>
Date: Thu, 22 Aug 2024 09:27:14 +0800
From: "Leizhen (ThunderTown)" <thunder.leizhen@...wei.com>
To: David Hildenbrand <david@...hat.com>, Andrew Morton
	<akpm@...ux-foundation.org>, <linux-mm@...ck.org>,
	<linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/1] mm/page_poison: slightly optimize check_poison_mem()



On 2024/8/21 19:28, David Hildenbrand wrote:
> On 21.08.24 09:53, Zhen Lei wrote:
>> When the debug information needs to be suppressed due to ratelimit,
>> it is unnecessary to determine the end of the corrupted memory.
>>
>> Signed-off-by: Zhen Lei <thunder.leizhen@...wei.com>
>> ---
>>   mm/page_poison.c | 7 ++++---
>>   1 file changed, 4 insertions(+), 3 deletions(-)
>>
>> diff --git a/mm/page_poison.c b/mm/page_poison.c
>> index 3e9037363cf9d85..23fa799214720f1 100644
>> --- a/mm/page_poison.c
>> +++ b/mm/page_poison.c
>> @@ -55,14 +55,15 @@ static void check_poison_mem(struct page *page, unsigned char *mem, size_t bytes
>>       if (!start)
>>           return;
>>   +    if (!__ratelimit(&ratelimit))
>> +        return;
>> +
>>       for (end = mem + bytes - 1; end > start; end--) {
>>           if (*end != PAGE_POISON)
>>               break;
>>       }
>>   -    if (!__ratelimit(&ratelimit))
>> -        return;
>> -    else if (start == end && single_bit_flip(*start, PAGE_POISON))
>> +    if (start == end && single_bit_flip(*start, PAGE_POISON))
>>           pr_err("pagealloc: single bit error\n");
>>       else
>>           pr_err("pagealloc: memory corruption\n");
> 
> This way, you will be ratelimiting on every function call, possibly skipping PAGE_POISON checks even if there was no prior corruption detected?
> 

No, the previous memchr_inv() does the PAGE_POISON check, determine the start of the corrupted memory. !start means no corruption.

	start = memchr_inv(mem, PAGE_POISON, bytes);
	if (!start)
		return;

+	if (!__ratelimit(&ratelimit))
+		return;

	for (end = mem + bytes - 1; end > start; end--) {
		if (*end != PAGE_POISON)
			break;
	}



-- 
Regards,
  Zhen Lei