l***@jmbreuer.net
2013-07-10 08:24:00 UTC
I've run into an issue while taking care of DVD resume code in XBMC.
A bit of background: XBMC computes a hash over the content of all .IFO
files to use as a Disk Unique ID, to make it possible to identify the
same disk when it is inserted again later and have per-disk persistent
player state / resume bookmarks.
XBMC uses libdvdnav/libdvdread/libdvdcss for DVD access.
I've observed the following: On both of my systems (gentoo on x86_64 and
OpenELEC on x86_64 [different HW]), performing CSS authentication (e.g.
"starting playback") will cause a different hash value to be computed
than before - see https://github.com/xbmc/xbmc/issues/2936
So it seems that performing CSS (my guess: the drive authentication
part) causes different data to be returned from the drive - FOR ME.
The hashes returned "before" and "after" authentication each are stable,
i.e. I always get the same two hashes for a given disc, one before auth,
the other after.
While discussing the issue with upstream, it turns out that another
developer on Windows cannot reproduce the issue - he always gets the
same content/hashes, although the logs show CSS authentication being
performed against his drive as well.
See https://github.com/xbmc/xbmc/pull/2951
Can anyone shed some light on why our systems would behave differently
as they do?
Thanks a lot!
So long,
Joe
A bit of background: XBMC computes a hash over the content of all .IFO
files to use as a Disk Unique ID, to make it possible to identify the
same disk when it is inserted again later and have per-disk persistent
player state / resume bookmarks.
XBMC uses libdvdnav/libdvdread/libdvdcss for DVD access.
I've observed the following: On both of my systems (gentoo on x86_64 and
OpenELEC on x86_64 [different HW]), performing CSS authentication (e.g.
"starting playback") will cause a different hash value to be computed
than before - see https://github.com/xbmc/xbmc/issues/2936
So it seems that performing CSS (my guess: the drive authentication
part) causes different data to be returned from the drive - FOR ME.
The hashes returned "before" and "after" authentication each are stable,
i.e. I always get the same two hashes for a given disc, one before auth,
the other after.
While discussing the issue with upstream, it turns out that another
developer on Windows cannot reproduce the issue - he always gets the
same content/hashes, although the logs show CSS authentication being
performed against his drive as well.
See https://github.com/xbmc/xbmc/pull/2951
Can anyone shed some light on why our systems would behave differently
as they do?
Thanks a lot!
So long,
Joe