linux-kernel - Re: [PATCH] padata: Fix possible divide-by-0 panic in padata_mt

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: <986ef2c0-4ed7-4b13-82f2-5f14e30d08e1@redhat.com>
Date: Tue, 13 Aug 2024 14:28:17 -0400
From: Waiman Long <longman@...hat.com>
To: Kamlesh Gurudasani <kamlesh@...com>,
 Steffen Klassert <steffen.klassert@...unet.com>,
 Daniel Jordan <daniel.m.jordan@...cle.com>,
 Andrew Morton <akpm@...ux-foundation.org>
Cc: linux-crypto@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] padata: Fix possible divide-by-0 panic in
 padata_mt_helper()


On 8/11/24 01:44, Kamlesh Gurudasani wrote:
> Waiman Long <longman@...hat.com> writes:
>
>> On 8/10/24 13:44, Kamlesh Gurudasani wrote:
>>> Waiman Long <longman@...hat.com> writes:
>>>
> ...
>>>> diff --git a/kernel/padata.c b/kernel/padata.c
>>>> index 53f4bc912712..0fa6c2895460 100644
>>>> --- a/kernel/padata.c
>>>> +++ b/kernel/padata.c
>>>> @@ -517,6 +517,13 @@ void __init padata_do_multithreaded(struct padata_mt_job *job)
>>>>    	ps.chunk_size = max(ps.chunk_size, job->min_chunk);
>>>>    	ps.chunk_size = roundup(ps.chunk_size, job->align);
>>>>    
>>>> +	/*
>>>> +	 * chunk_size can be 0 if the caller sets min_chunk to 0. So force it
>>>> +	 * to at least 1 to prevent divide-by-0 panic in padata_mt_helper().`
>>>> +	 */
>>> Thanks for the patch and detailed comment.
>>>> +	if (!ps.chunk_size)
>>>> +		ps.chunk_size = 1U;
>>>> +
>>> could it be
>>>           ps.chunk_size = max(ps.chunk_size, 1U);
>>>           
>>> or can be merged with earlier max()
>>>     	ps.chunk_size = max(ps.chunk_size, max(job->min_chunk, 1U));
>>>     	ps.chunk_size = roundup(ps.chunk_size, job->align);
>>>
>>> sits well with how entire file is written and compiler is optimizing
>>> them to same level.
>> I had actually thought about doing that as an alternative. I used the
>> current patch to avoid putting too many max() calls there. I can go this
>> route if you guys prefer this.
> Just curious, what is your reason for avoiding too many max() calls? Both
>          if (!ps.chunk_size)
>          	ps.chunk_size = 1U;
> and
>          ps.chunk_size = max(ps.chunk_size, 1U);
>
> are having same number of instructions [1].
>
> [1] https://godbolt.org/z/ajrK59c67
>
> We can avoid nested max(), though following would make it easier to understand.
>
>     ps.chunk_size = max(ps.chunk_size, 1U);

That will certainly work. My current patch has been merged into the 
Linus tree. You are welcome to post another patch to clean it up if you 
want.

Cheers,
Longman

>
> Cheers,
> Kamlesh
>
>> Cheers,
>> Longman