Tinderbox chroot change mount to readonly on demand

[Joe Marcus Clarke]

On 6/10/11 6:33 AM, Chris Rees wrote:
> 
> On 10 Jun 2011 09:28, "Joe Marcus Clarke" <marcus at marcuscom.com
> <mailto:marcus at marcuscom.com>> wrote:
>>
>> On 6/9/11 5:02 PM, Chris Rees wrote:
>> > On 8 June 2011 20:36, Chris Rees <crees at marcuscom.com
> <mailto:crees at marcuscom.com>> wrote:
>> >> Hi all,
>> >>
>> >> Done a couple of little changes to the mounting logic [1], as well as
>> >> modify requestMount to auto-add the -u flag to already mounted
>> >> filesystems, this will allow us to toggle the readonly flag.
>> >>
>> >> It looks rather tricky to be playing with something as major as where
>> >> the build is called from, but I think I'm doing it properly...
>> >>
>> >> Does anyone see any major errors in what I've done so far? Progress is
>> >> slow because I'm at work this week, and very busy writing pupil
>> >> reports as well :/
>> >>
>> >> Chris
>> >>
>> >
>> > OK, so I trashed the mounting changes, the complexity and problems with
>> > mounting ${build}/tmp onto ${buildmountpoint}/tmp when ${build} is
> mounted
>> > onto ${buildmountpoint} were just becoming depressing.
>> >
>> > Then I thought... why not just su to nobody for the building phases?
>>
>> Yes, it should be fine as ports should be buildable by non-root users.
>>
>> >
>> > Patch at [1], and net/rsync was tinderboxed with the patch and
> passed [2],
>> > and mail/mailman was tinderboxed with the patch and failed [3] (as it
>> > now should).
>> >
>> > Would anyone care to test this patch with their ports? Any
>> > well-behaved ports should be fine.
>>
>> Out of curiosity, does mailman fail on pointyhat due to its abuse of the
>> file system?
>>
> 
> I'm discussing a similar patch with liaison at the moment for pointy, so
> I guess not.

Cool.  This is exactly what I was getting at.  I would approve this
patch for TB provided the same can be done in pointyhat.

Joe

-- 
PGP Key : http://www.marcuscom.com/pgp.asc