Since the latest upgrade of my Bubba 2 to 1.0.13, my Roku Soundbridge cannot connect to the Firefly daap server on my Bubba 2.
I have experienced that setting the rx and tx settings "on" with the ethtool for eth1 does help a little, but even with those settings before the upgrade I was always having problems with tracks jumping after 30 secs or skipping altogether.
I have now built a new version of firefly from source and have eliminated the absence of the bonjour service in the delivered package as a cause, as my built server does have this service.
For those interested, I used the svn-1696 build from the Firefly home page and needed to install (apt-get install xyz) these packages
and I used the following commands to build and run:g++
pkg-config
libavahi-client-dev
libid3tag0-dev
libsqlite3-dev
libogg-dev
libvorbis-dev
libFLAC-dev
the mt-daap.conf file from the build needed these changes, as I didn't do a proper install./configure --enable-avahi --enable-mdns --enable-sqlite3 --enable-oggvorbis --enable-flac --enable-upnp
make
src/mt-daapd -c `pwd`/contrib/mt-daapd.conf -d 9 -f
I also built and ran the same code of firefly on my Suse 11.1 machine with the same processweb_root = /usr/share/mt-daapd/admin-root
db_type = sqlite3
mp3_dir = /home/storage/music
plugin_dir = /home/david/firefly/mt-daapd-svn-1696/src/plugins/.libs
I now have 2 firefly server hosts [bubba 2 and suse11.1] and 2 clients[soundbridge and suse11.1 amarok]
The only connection that is not working is bubba 2 --> soundbridge. It sees the presence of the bubba firefly, but if I try to connect on the soundbridge it just says "connection to server failed". The bubba firefly log (level 9 debug) reports nothing at all.
The soundbridge works immaculately with my Suse11.1 Firefly, and amarok on my Susu11.1 works immaculately with the bubba firefly.
The soundbridge is connected wirelessly and the bubba and Suse11.1 connected to each other via a 100mbps hub.
As far as I can see all bubba settings are now 1.0.13 out-of-the box standard.
[Mediatomb works correctly except that it occasionally jumps the second (and only the second) track in the playlist.]
Is there possibly yet another place in the system where the Freescale CPU buggy checksum instruction is being used? Since the connection request is not arriving with the Firefly server it looks like the problem is with the incoming message from the SB to the Bubba on eth1.
Any suggestions of where to look gratefully received...
It would also be useful to know more about the Freescale bug itself if anyone knows where it is described. The help information about the ethtool settings says that the ethernet hardware is what does the checksum, but this does not seem to fit with the mention from Excito elsewhere in this forum that the bug is in the Freescale CPU - or does the CPU also support the ethernet???