polkit fails to authenticate on dbus

Simon McVittie simon.mcvittie at collabora.co.uk
Mon Jan 26 04:32:08 PST 2015


On 26/01/15 00:44, karu.pruun wrote:
> You were right! dbus was fine: in dbus, DragonFly runs through the
> same codepath as FreeBSD, namely, the one defined by the HAVE_CMSGCRED
> macro. (Interestingly DragonFly is mentioned in the code.)

That was based on me doing a few minutes' research with Google while
trying to get some idea of which mechanisms corresponded to which OSs,
and should not be taken as evidence that it ever actually worked :-)

Please consider testing and submitting a patch for dbus that adds
Dragonfly to the same conditionals as __FreeBSD__ (bus/dispatch.c,
dbus/dbus-sysdeps-unix.c and test/dbus-daemon.c). If you do that, any
future regressions in dbus' ability to pass credentials on Dragonfly
will be signalled by a regression test failure (or an #error while
compiling, if detection fails altogether), which should let Dragonfly
maintainers detect and fix problems considerably earlier than by
noticing that dbus has fallen back to the less reliable DBUS_COOKIE_SHA1
authentication mechanism.

    S



More information about the dbus mailing list