[buildd-tools-devel] Bug#797781: Bug#797781: diffoscope does not seem to work with schroot]

[Roger Leigh]

On 06/09/2015 18:28, Aurelien Jarno wrote:
> On 2015-09-02 14:28, Santiago Vila wrote:
>> Hello.
>>
>> I've just reassigned this report to schroot.
>>
>> Below is the reason.
>>
>> Thanks.
>>
>> ----- Forwarded message from Mattia Rizzolo <mattia at mapreri.org> -----
>>
>> Date: Wed, 2 Sep 2015 14:08:42 +0000
>> From: Mattia Rizzolo <mattia at mapreri.org>
>> To: Santiago Vila <sanvila at debian.org>, 797781 at bugs.debian.org
>> Subject: [Reproducible-builds] Bug#797781: Bug#797781: diffoscope does not seem to work with schroot
>> User-Agent: Every email client sucks, this one just sucks less.
>>
>> On Wed, Sep 02, 2015 at 01:41:23PM +0000, Santiago Vila wrote:
>>> Package: diffoscope
>>> Version: 31
>>>
>>> Greetings. I'm running jessie with several chroots created with
>>> schroot. As a normal user, I do this:
>>>
>>> schroot -c sid
>>> diffoscope some.deb someother.deb
>>>
>>> and this is the result:
>>>
>>> CRITICAL /dev/shm is not available or not on a tmpfs. Unable to create semaphore.
>>>
>>> I believe such error is not supposed to happen.
>>
>> Well, quite a lot of stuff requires shm nowadays.
>>
>> Consider that we rb people run diffoscope inside scrhoot, and it just
>> works.  We have
>>
>>      /dev/shm    /dev/shm    none    rw,bind     0   0
>>
>> in /etc/schroot/default/fstab.
>>
>>
>> Personally I'd not consider this a diffoscope bug.
>
> The buildd flavour of the configuration mount a tmpfs in /dev/shm. AFAIK
> this is not done for the default flavour as too options are possible
> there:
> - Bind mount like above. This means sharing the shm directory with the
>    outside world. This might have some security implications, and
>    unwanted consequences.
> - Empty tmpfs like for buildds. This means the processes do not have
>    accesses to shared memory from processes outside of the chroot.
>
> Depending on how the user want to use schroot, one or the other
> configuration should be used. I don't know how we should choose the
> default one.

I would think the main question is this:
- does any process in the build chroot need to share memory with the 
host using POSIX SHM?

I can't think of any valid reason to do so in the context of package 
building, so it should be safe to simply mount an empty tmpfs in the 
chroot in all cases.


Regards,
Roger