linux-kernel - Re: [PATCH] x86/mpx: fix recursive munmap() corruption

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: <02389b9c-141c-f5b7-756a-516599063766@gmail.com>
Date:   Tue, 3 Nov 2020 21:08:45 +0000
From:   Dmitry Safonov <0x7f454c46@...il.com>
To:     Laurent Dufour <ldufour@...ux.vnet.ibm.com>,
        Christophe Leroy <christophe.leroy@...roup.eu>,
        Michael Ellerman <mpe@...erman.id.au>
Cc:     Dave Hansen <dave.hansen@...el.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Dave Hansen <dave.hansen@...ux.intel.com>, mhocko@...e.com,
        rguenther@...e.de, x86@...nel.org,
        LKML <linux-kernel@...r.kernel.org>, stable@...r.kernel.org,
        luto@...capital.net, linux-mm@...ck.org,
        Andrew Morton <akpm@...ux-foundation.org>,
        linuxppc-dev@...ts.ozlabs.org, vbabka@...e.cz
Subject: Re: [PATCH] x86/mpx: fix recursive munmap() corruption

Hi Laurent, Christophe, Michael, all,

On 11/3/20 5:11 PM, Laurent Dufour wrote:
> Le 23/10/2020 à 14:28, Christophe Leroy a écrit :
[..]
>>>> That seems like it would work for CRIU and make sense in general?
>>>
>>> Sorry for the late answer, yes this would make more sense.
>>>
>>> Here is a patch doing that.
>>>
>>
>> In your patch, the test seems overkill:
>>
>> +    if ((start <= vdso_base && vdso_end <= end) ||  /* 1   */
>> +        (vdso_base <= start && start < vdso_end) || /* 3,4 */
>> +        (vdso_base < end && end <= vdso_end))       /* 2,3 */
>> +        mm->context.vdso_base = mm->context.vdso_end = 0;
>>
>> What about
>>
>>      if (start < vdso_end && vdso_start < end)
>>          mm->context.vdso_base = mm->context.vdso_end = 0;
>>
>> This should cover all cases, or am I missing something ?
>>
>>
>> And do we really need to store vdso_end in the context ?
>> I think it should be possible to re-calculate it: the size of the VDSO
>> should be (&vdso32_end - &vdso32_start) + PAGE_SIZE for 32 bits VDSO,
>> and (&vdso64_end - &vdso64_start) + PAGE_SIZE for the 64 bits VDSO.
> 
> Thanks Christophe for the advise.
> 
> That is covering all the cases, and indeed is similar to the Michael's
> proposal I missed last year.
> 
> I'll send a patch fixing this issue following your proposal.

It's probably not necessary anymore. I've sent patches [1], currently in
akpm, the last one forbids splitting of vm_special_mapping.
So, a user is able munmap() or mremap() vdso as a whole, but not partly.

[1]:
https://lore.kernel.org/linux-mm/20201013013416.390574-1-dima@arista.com/

Thanks,
          Dmitry