Roland,
On Thu, Oct 19, 2006 at 11:10:26PM +0200, Roland Hostettler wrote:
> Attached there are two patches to get the D-Bus daemon working on the
> PXA270.
Thanks a lot. I had added dbus very quickly yesterday, just to make
another program compiling that broke allyesarm.
> First, there are some modifications to the dbus.make file
> ('dbus.make.diff'), particularly the install section is affected. In
> the original, the daemon was not copied to the target root. Also the
> dbus-monitor, dbus-send and dbus-uuidgen in the 'tools/' directory are
> only wrapper scripts. The binaries itself are located in
> 'tools/.libs'. Further I've fixed some typos in the symlinks pointing
> to libdbus.
I'll add this in a few minutes; please make sure that your patches are
p1 style as seen from the ptxdist toplevel directory, so it is a little
bit easier for me to handle them. Please also make sure you add a header
to the patch, like in patches/HEADER. Thanks.
> Second, there's a patch to the source (which is included in a patch for
> the 'ptxdist-trunk/patches/dbus-0.94/generic' directory
> ('patches-dbus-0.94.diff'), so it can be applied automatically using
> 'ptxdist'). This patch modifies the 'system.conf.in' template in the
> 'bus/' source directory. I'm not sure if these modifications are
> appropriate. I'm not sure about my assumptions, but here some thougts
> (I'm not really familiar with D-Bus yet):
>
> * As OSELAS only knows the user root(?), the bus is run as root
Does this have any security implications? Would it make more sense to
add a dbus user? How do other systems handle that?
> * Services dir directive according to [1]
> * The rest seems feasible to me
Who is replacing this @DBUS_SESSION_SOCKET_DIR@ in the configuration?
> I agree with point 1) of [1], that we only need a system bus in an
> embedded system. What do others think about that?
Sounds sane.
> Maybe we could add a configuration option to the ptxdist config.
I'm not going to add the config patch, please rework it in a way that
there is a predefined config (like your one) which can be superseeded by
a project specific one. For example, search for ROOTFS_ETC_INITD_HTTPD
in rules/rootfs.in.
I know that the way we define these options in rootfs.in is not very
elegant and that the config options should really be with the packets
and not in rootfs.in, but it's the way it's currently done. So we will
rework this all at once in the future.
> Known issues:
> * Per-session bus is not tested and unlikely to work
> * Manual creation of directories is nasty
> * Please note: the whole thing is not tested out-and-out, so if you
> find any discrepancies, let us know.
>
> I hope to be able to contribute some more patches for configuration
> (dbus.in) and an init script to get rid of the recreation of the above
> directories soon.
Please do so.
Robert
--
Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de
Pengutronix - Linux Solutions for Science and Industry
Handelsregister: Amtsgericht Hildesheim, HRA 2686
Hannoversche Str. 2, 31134 Hildesheim, Germany
Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9
--
ptxdist mailing list
[email protected]
