2016-06-15 17:00:25

by Don Zickus

Subject: Debugging bluetoothd, dbus and selinux help?

Hi,

I am trying to debug why bluetoothd decides to disconnect from the d-bus on
the host, when the client does an object push to an obex agent on the host.

When I set selinux to permissive mode ('setenforce 0'), things work fine (ie
the object push goes through).

I tried searching /var/log/audit/audit.log, journalctl and dbus-monitor for
clues (on the host) and can't find anything obvious.

Does anyone have any tips on how to find what security policy might be
interfering with my test?

Cheers,
Don

2016-06-17 14:43:19

by Don Zickus

Subject: Re: Debugging bluetoothd, dbus and selinux help?

On Fri, Jun 17, 2016 at 10:32:23AM +0200, Szymon Janc wrote:
> Hi,
>
> On Wednesday 15 June 2016 13:00:25 Don Zickus wrote:
> > Hi,
> >
> > I am trying to debug why bluetoothd decides to disconnect from the d-bus on
> > the host, when the client does an object push to an obex agent on the host.
> >
> > When I set selinux to permissive mode ('setenforce 0'), things work fine (ie
> > the object push goes through).
> >
> > I tried searching /var/log/audit/audit.log, journalctl and dbus-monitor for
> > clues (on the host) and can't find anything obvious.
> >
> > Does anyone have any tips on how to find what security policy might be
> > interfering with my test?
> >
>
> To me it looks like FD passing on D-Bus is being blocked by SELinux.

Hi Szymon,

Thanks for the tip! Is there a way I can verify that. I am scratching my
head for the right log file or monitor probe. Or better yet, the right file
to modify to fix that permission problem..

Cheers,
Don

2016-06-17 08:32:23

by Szymon Janc

Subject: Re: Debugging bluetoothd, dbus and selinux help?

Hi,

On Wednesday 15 June 2016 13:00:25 Don Zickus wrote:
> Hi,
>
> I am trying to debug why bluetoothd decides to disconnect from the d-bus on
> the host, when the client does an object push to an obex agent on the host.
>
> When I set selinux to permissive mode ('setenforce 0'), things work fine (ie
> the object push goes through).
>
> I tried searching /var/log/audit/audit.log, journalctl and dbus-monitor for
> clues (on the host) and can't find anything obvious.
>
> Does anyone have any tips on how to find what security policy might be
> interfering with my test?
>

To me it looks like FD passing on D-Bus is being blocked by SELinux.

--
pozdrawiam
Szymon Janc