unmerged patches

jolly andreas at eversberg.eu
Sun Apr 8 14:49:27 UTC 2012


hi,

i just collected all my patches together that i would have merged:


1. i added two patches to jolly/battery branch of osmocombb. it will add
a font with symbols and display them at rssi. since i did not receive a
reply about that from christian, i pushed it in this seperate branch so
far. christian, if you find the time, just look at it.


2. there are several patches in my "testing" branch at jolly/ui of
osmocombb. these patches are tested and work quite well. the are not
related to the ui. i think they could me mergend:

commit 66c21b3b7d30db205202d46893032ae2a73992a2
    layer23: Send SIM APDUs via GSMTAP
    Usefull to trace SIM messages together with Um messages.

commit b9ff5044a5fcbd07e6a295d75e14c7c504259914
    layer23: Be sure to close mncc socket on exit of mobile instance

commit e30e351d66f07e6effee768412a9d8f31202b4ed
    layer1: Retry fist power measurement, if it seems to fail
    In some cases (e.g. after a call with TCH) the first power measurement
    after a full reset will always return 0 (-110dbm). In this case the
    measurment is repeated once again.
    This is just a workarround, and it will not fix the actual cause.


3. there are several additions and fixes at jolly/rtpmux branch of openbsc.

commit db3a7dd357bd7cd842a80655e734a1b4afdb6f7c and commit
872c6c002add0a741514aa505df9379a6fcdb955 allow to exchange traffic via
rtp with a given rtp endpoint. the destination can be controlled via
mncc interface. the result is that the traffic is not routed via mncc
interface, but directly exchanged with the given rtp endpoint. lcr
supports that, so traffic between sip and openbsc is directly forwarded
and not routed through lcr and mncc interface.

commit 3d407e7c8e5c4e01dc07a530f910d29fac687809 and commit
f47d13e55888576c9201a3a7fee04fc58f98ff66 will handle bad frames from e1
bts. if a frame is bad, the rtp packet is dropped (if forwarded via
rtp). if the bad frame is received by lcr via mncc interface, lcr will
extrapolate the missing audio by repeating last valid frame with reduced
value. instead of having a distrorted sound, the audio stream will now
be clear, even if some frames are bad.

commit 0193b8a76824cdce9d9da3f7374a928efac6f96c allows dynamic payload
types when forwarding rtp traffic to a given rtp endpoint. (used for
EFR/AMR/HR)

commit ea724c6af9e1a5b6df8d0e5965f357896834d3ce and commit
a22e598c0a93f88433c6a00bd2acdde5c2f496d5 will fix the problems with
delay and loosing audio at nanobts. it uses system clock as a basis to
correct timestamp and sequence number of frames transmitted to the bts.

commit bf14b25358f7ceb021e2397e90c2fb9484245b7b fixes problem with
interruption of traffic, if packet transmission via rtp fails in the
beginning.

the result of all these patches is a reliable audio stream. the call
waiting works, as well as hold/retrieve of calls without audio
interruption or increasing/high delay. (even if database access makes
openbsc stop for some time.) in conjuction with lcr, a sip gateway can
directly exchange audio traffic with openbsc. the codec to be used is
negotiated between SIP gateway and MS. (depending on support and preference)


4. long time ago i extraced the sms protocols "smc" and "smr" (TS 11.11)
from openbsc and added a state machines. they are now part of
libosmocore and are used for sending/receiving sms via osmocombb. i
removed all that code from openbsc and use libosmocore instead. see
jolly/sms for the 4 patches.


i hope that was not too much at a time :)

regards,

andreas

 




More information about the baseband-devel mailing list