Attention is currently required from: fixeria, pespin. Hoernchen has posted comments on this change. ( https://gerrit.osmocom.org/c/osmo-trx/+/30416 )
Change subject: ms-trx support ......................................................................
Patch Set 4:
(37 comments)
File Makefile.am:
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/8557aa21_1a4ffec0 PS3, Line 37: osmocom-bb/src/host/trxcon \
Ack. Most likely he forgot to remove this line.
Done
File Transceiver52M/Makefile.am:
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/ab495219_3ad23e55 PS3, Line 26: AM_CPPFLAGS
'-I' stuff should be in CPPFLAGS only, it makes no sense to have it in CFLAGS/CXXFLAGS.
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/ef318324_7f99c032 PS3, Line 77: LIBOSMOCODING_LIBS
Having LIBOSMOCODING_LIBS in COMMON_LDADD makes all osmo-trx-* binaries depend on libosmocoding, whi […]
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/7dbf1157_40ebb075 PS3, Line 90: noinst_HEADERS
This should be done conditionally too, but not critical I guess.
Done
File Transceiver52M/ms/itrq.h:
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/27a2e661_cd56fde8 PS3, Line 34: /*
weird comment formatting, please make it consistent with the copyright header
Done
File Transceiver52M/ms/l1ctl_server_cb.cpp:
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/1c7497e2_ec2b34dc PS3, Line 65: osmocom_l2
TODO: make path configurable
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/57fa3766_3292a919 PS3, Line 74: return
so the app keeps running if l1ctl_server_alloc() fails?
Done
File Transceiver52M/ms/logging.cpp:
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/a2f6a6eb_8ec2be8e PS3, Line 54: loglevel = LOGL_NOTICE
spacing issues
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/ed44894c_88a9b155 PS3, Line 57: DTRXC
not used by libtrxcon nor needed
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/278cf1d8_349e019f PS3, Line 64: DTRXD
not used by libtrxcon nor needed
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/10a88d01_ee4fa15a PS3, Line 75: .loglevel = LOGL_NOTICE,
spacing issues
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/d69f7024_80c92d1a PS3, Line 76: enabled = 0,
why disabled?
too nuch unbounded log spam impacts timing.
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/b6ec8fb5_6b42d5f4 PS3, Line 82: .loglevel = LOGL_NOTICE,
spacing issues
Done
File Transceiver52M/ms/ms_upper.h:
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/925d5d4e_ee890a3f PS3, Line 1:
ws
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/6b046a98_0c6d47aa PS3, Line 36: trx_if
do you really need/use TRXC/TRXD here?
Done
File Transceiver52M/ms/ms_upper.cpp:
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/cd859f9b_9180321b PS3, Line 61: #include <osmocom/core/gsmtap_util.h>
you're not using GSMTAP API here...
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/499172f8_05f316dc PS3, Line 64: // #include <osmocom/core/application.h>
removeme
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/0d7e67c3_d3a9ea3a PS3, Line 71: trx_if
do you really need/use TRXC/TRXD here?
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/e6956264_e4693cf4 PS3, Line 74: #include <osmocom/bb/l1sched/l1sched.h>
I don't see any l1sched_* API used in this file.
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/50ed90bd_c350e3e7 PS3, Line 177: return trash
can you clarify this a bit further? thrash?
fcch bursts contain "nothing", but the old trxcon needed "something" to tickle the scheduler. There are only 3 burst types, so explictily listing all of those is fine, even if the new scheduler allows skipping those and the function returns early.
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/4b19fa0c_2a6ef870 PS3, Line 184: RSSI = 10
why 10?
random reasonable value, the SCH burst is only demodulated once in the timekeeping part because doing it twice introduces massive latency spikes comapared to NB that are only handled in the upper layer, so there is no opportunity here to measure anything because only the demodded bits are passed to the upper layer in the SCH case.
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/dbf421e4_47d1aabf PS3, Line 223: timingOffset = (int)round(0);
TODO/FIXME?
Done
In practice the alignment does not matter because the lower loop continuously tracks and updates it anyway so the only offset available would be the short term offset until it gets updated and reset to 0.
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/b42d80bc_257ccd2a PS3, Line 238: trx_data_rx_handler
removeme
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/c7baa8e3_a29f5253 PS3, Line 246: trxcon_phyif_handle_clock_ind
removeme
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/a644c068_f9d7149e PS3, Line 304: case trxcon::TRXCON_PHYIF_CMDT_RESET
Better use value-string for that. […]
This mostly exists for debugging the order of actions here and should not really be used for anything else. It's just easier to use than signaltap or other custom markers. In practice for general purpose logging trxcon/the l1ctl handler knows what is going on anyway.
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/067b8844_59e02d0a PS3, Line 352: set_ta
Shouldn't this be done on receipt of TRXCON_PHYIF_CMDT_RESET instead?
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/14fe4b6f_7d7255a9 PS3, Line 357: dbm = -80
please add a FIXME or TODO, so it's clear that the measurement logic is missing
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/70971d36_b612055b PS3, Line 364: gsm_arfcn2freq10
careful, this function may return 0xffff
In practice we mostly want to use different frequencies anyway, but for current testing and usage we just stick to "a" arfcn and that's it. If wrong arfcns are passed that are not part of the gsm ranges everything is broken anyway...
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/dff33073_293e488a PS3, Line 385: assert
OSMO_ASSERT?
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/bd8bb51d_b634bc66 PS3, Line 425: trxcon
it's not trxcon app anymore
but this "is" the global trxcon context used for all trxcon stuff and this is only being used by the "main loop" upper thread doing trxcon things...
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/aef3dc97_b8c24196 PS3, Line 429: 3
pass 0 to avoid confusion, we're not using fn_advance here
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/7945fb2f_e09c12db PS3, Line 430: gsmtap = 0;
gsmtap is a pointer, so = NULL
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/98c4b4de_b3d3dfdb PS3, Line 431: 0x1234
What's this? Why not NULL?
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/2b0108db_523b8342 PS3, Line 433: trxc
I would avoid using 'trxc' here to avoid confusion with the TRXC protocol.
Done
File configure.ac:
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/fc9cdab4_377081c8 PS3, Line 330: test -d osmocom-bb
A proper approach would be making this configurable: --enable-ms or --with-ms.
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/619bda41_d8b301b6 PS3, Line 333: AC_MSG_NOTICE(["Enabling ms-trx..."])
spacing issues
Done
https://gerrit.osmocom.org/c/osmo-trx/+/30416/comment/dfc70a11_6afaee32 PS3, Line 364: doc/manuals/Makefile \
Ack, it's one of my fixups: https://cgit.osmocom. […]
Done
-
Hoernchen