linux-kernel - Re: [PATCH] mm: fix unexpected changes to {failslab|fail_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: <0fd0c72d-badc-ad75-f0fe-91bc148820f2@bytedance.com>
Date:   Tue, 8 Nov 2022 10:47:58 +0800
From:   Qi Zheng <zhengqi.arch@...edance.com>
To:     Jason Gunthorpe <jgg@...dia.com>
Cc:     dvyukov@...gle.com, willy@...radead.org, akinobu.mita@...il.com,
        akpm@...ux-foundation.org, linux-kernel@...r.kernel.org,
        linux-mm@...ck.org, linux-fsdevel@...r.kernel.org,
        stable@...r.kernel.org
Subject: Re: [PATCH] mm: fix unexpected changes to
 {failslab|fail_page_alloc}.attr



On 2022/11/8 00:26, Jason Gunthorpe wrote:
> On Mon, Nov 07, 2022 at 11:05:42PM +0800, Qi Zheng wrote:
>>
>>
>> On 2022/11/7 20:42, Jason Gunthorpe wrote:
>>> On Mon, Nov 07, 2022 at 11:31:09AM +0800, Qi Zheng wrote:
>>>
>>>> @@ -31,9 +33,9 @@ bool __should_failslab(struct kmem_cache *s, gfp_t gfpflags)
>>>>    		return false;
>>>>    	if (gfpflags & __GFP_NOWARN)
>>>> -		failslab.attr.no_warn = true;
>>>> +		flags |= FAULT_NOWARN;
>>>
>>> You should add a comment here about why this is required, to avoid
>>> deadlocking printk
>>
>> I think this comment should be placed where __GFP_NOWARN is specified
>> instead of here. What do you think? :)
> 
> NOWARN is clear what it does, it is this specifically that is very
> subtle about avoiding deadlock aginst allocations triggered by
> printk/etc code.

Oh, maybe I understand your concern. Some people may think that this
is just a print of fault injection information, not a warning. I'll
add a comment explaining why in some cases there must be no printing.

Thanks,
Qi

> 
> Jason

-- 
Thanks,
Qi