Question about D-Bus and cygrunsrv

[Mike Brady]

Thanks for the comments. I’ll certainly check the security side of things.

What is a bit intriguing is that another application — the Avahi daemon (see http://avahi.org) that provides ZeroConf / Bonjour service discovery services — also uses the D-Bus system bus and it works fine as a cygrunsrv-based daemon. The difference between the two applications with respect to D-Bus is that Shairport Sync uses the high-level GDbus interface that is part of GIO (https://developer.gnome.org/gio/stable/ch01.html), whereas the Avahi daemon uses the dbus library directly.

Mike

> On 26 Feb 2019, at 06:51, L A Walsh <cygwin@tlinx.org> wrote:
> 
> On 2/22/2019 2:16 PM, René Berber wrote:
>> 
>> 
>> My guess is that its a permission problem: D-Bus by default on Windows
>> uses unix sockets (fills up your temp directory with those pesky files :
>> - -)
>> 
>> In Cygwin the temp directory has the sticky permission (t), but the
>> socket file created by the server most likely doesn't have access for
>> anybody else.
>> 
> ---
>    I'm not sure, but I'd think dbus demon would be enabled to
> talk to the system socket
>> Check those permissions to maybe get a clue.
>> 
>> The problem comes from upstream, I'm just not sure because I used to
>> build my own, patched, version of DBUS (for a workaround which is to
>> use tcp, not unix sockets... but the Windows maintainer says "nobody
>> uses tcp sockets on Windows", and then they "improved" security which
>> is why I suspect access permissions).
>> 
> ---
>    Yeah, they are more than a little bit security paranoid.
> 
> I'd like to get the dbus sessions between my desktop & server to talk --
> like when I am running an X-app, and press help, it really would be
> better if it brought up the browser on my desktop instead of one
> on the server via X.
> 
> I had hopes as one of their examples specifies tcp in the protocol
> field, but supposedly the tcp doesn't work and likely won't due to
> security concerns -- mainly because the bus architecture was structured
> around the idea of an internal HW bus -- where normally there isn't
> much in the way of security.  If they allowed it to be open to
> a network, then they think everyone on the internet will have access
> and that dbus will be blaimed for a security problem.
> 
> I did mention that some people use closed nets between machines
> and should be allowed to use dbus to talk between them where desirable,
> but he thinks that most hand-held users wouldn't know how to isolate
> a net.  On top of that, dbus doesn't have the ability to categorize
> if the net is public, private(home) or enterprise(work).  Seemed
> like a pain to get working for what would be a plus for me, but
> not sure about demand, not to mention working with such paranoid
> folk really can be frustrating.
> 
> There's a dbus list where they talk about development and security
> issues @
> 
> https://lists.freedesktop.org/mailman/listinfo/dbus
> 
> best of luck!
> 
> 
> 
> 


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple