[RFC] Making D-Bus suitable for being run early during boot

David Zeuthen zeuthen at gmail.com
Fri Jul 9 09:16:06 PDT 2010


Hi,

On Fri, Jul 9, 2010 at 12:10 PM, Colin Walters <walters at verbum.org> wrote:
> On Fri, Jul 9, 2010 at 11:58 AM, David Zeuthen <zeuthen at gmail.com> wrote:
>> Hi,
>>
>> On Thu, Jul 8, 2010 at 9:09 PM, Lennart Poettering <mzqohf at 0pointer.de> wrote:
>>> b) The system bus should become an abstract namespace socket by
>>>   default.
>>
>> So what happens when the system bus daemon crashes and some random
>> unprivileged process starts listening on this well-known but abstract
>> address?
>
> If the system bus crashes your OS gets hosed =)

Sure, and I'm the first to agree, but that's beside the point. I mean,
at least part of my current OS keeps running just fine if I kill the
system bus daemon.

>  But in a systemd
> world of course, it's actually systemd listening on the socket.

Ah, OK, I see - very nice, systemd FTW. So, yeah, if we can guarantee
that no unprivileged process can grab the socket in the meantime (if
even for a split-second) then I guess that my objection doesn't
matter.

We still have the other problems to deal with though.

    David


More information about the dbus mailing list