These patches eradicate some of the easier-to-fix compiler warnings I keep
seeing in almost every dev cycle. I'd be glad to have them on master.
One of them is an actual error, ignoring return values.
Neels Hofmeyr (3):
bsc_nat: fail if VTY telnet port cannot be bound, clarify comment
ipaccess_rcvmsg: fix returncode, add partial write warning
gsm340_rx_tpdu: comment-out two unused vars
openbsc/src/ipaccess/ipaccess-proxy.c | 8 +++++++-
openbsc/src/libmsc/gsm_04_11.c | 7 +++++--
openbsc/src/osmo-bsc_nat/bsc_nat.c | 7 +++++--
3 files changed, 17 insertions(+), 5 deletions(-)
--
2.1.4
This time SMPP is included. Here is the abridged complete overview:
Following the previous patch sets for libosmo-abis and libosmocore, this patch
set allows configuring ALL of osmo-nitb's local IP addresses and makes it
possible to run several osmo-nitb processes alongside each other.
(1) Abis/IP (from the libosmo-abis patch)
In the config file, have:
e1_input
ipa bind 10.9.8.7
(2) telnet VTY (prepared by libosmocore patch set)
In the config file, have:
line vty
bind 10.9.8.7
(3) ctrl interface (prepared by libosmocore patch set)
In the config file, have:
ctrl
bind 10.9.8.7
(4) MNCC socket
In addition to the old -m option with a fixed socket path, you may now
supply a cmdline argument with explicit path:
-M /path/to/socket/file
(5) SMPP SMSC
In the config file, have:
smpp
local-tcp 10.9.8.7 2775
Neels Hofmeyr (6):
enable telnet VTY bind address config for various programs
osmo-nitb: add -M to pass specific MNCC socket path
osmo-nitb: cosmetic: rename to rf_ctrl_path, following mncc_sock_path
osmo-nitb: be strict about cmdline args
enable ctrl bind config for various programs
smpp: refactor initialization, add bind address
openbsc/include/openbsc/bsc_nat.h | 3 +-
openbsc/include/openbsc/ctrl.h | 3 +-
openbsc/include/openbsc/gprs_sgsn.h | 3 +-
openbsc/include/openbsc/mncc.h | 2 +-
openbsc/include/openbsc/smpp.h | 4 +-
openbsc/src/gprs/gb_proxy_main.c | 12 +++--
openbsc/src/gprs/gtphub_main.c | 11 ++--
openbsc/src/gprs/sgsn_ctrl.c | 5 +-
openbsc/src/gprs/sgsn_main.c | 42 ++++++++++-----
openbsc/src/libbsc/bsc_ctrl_lookup.c | 6 ++-
openbsc/src/libbsc/bsc_init.c | 6 ++-
openbsc/src/libmsc/mncc_sock.c | 9 ++--
openbsc/src/libmsc/smpp_openbsc.c | 43 +++++++++------
openbsc/src/libmsc/smpp_smsc.c | 93 ++++++++++++++++++++++++---------
openbsc/src/libmsc/smpp_smsc.h | 7 ++-
openbsc/src/libmsc/smpp_vty.c | 75 +++++++++++++++++++++-----
openbsc/src/osmo-bsc/osmo_bsc_main.c | 10 +++-
openbsc/src/osmo-bsc_mgcp/mgcp_main.c | 6 ++-
openbsc/src/osmo-bsc_nat/bsc_nat.c | 22 ++++++--
openbsc/src/osmo-bsc_nat/bsc_nat_ctrl.c | 5 +-
openbsc/src/osmo-nitb/bsc_hack.c | 43 ++++++++++-----
21 files changed, 298 insertions(+), 112 deletions(-)
--
2.1.4
Following the previous patch sets for libosmo-abis and libosmocore, this patch
set allows configuring almost all of the osmo-nitb IP addresses. The goal is to
allow several osmo-nitb processes to run alongside each other.
(1) Abis/IP (from the libosmo-abis patch)
in the config file, have:
e1_input
ipa bind 10.9.8.7
(2) telnet VTY (prepared by libosmocore patch set)
in the config file, have:
line vty
bind 127.0.0.99
(3) ctrl interface (prepared by libosmocore patch set)
in the config file, have:
ctrl
bind 127.0.0.99
(4) MNCC socket
In addition to the old -m option with a fixed socket path, you may now
supply a cmdline argument with explicit path:
-M /path/to/socket/file
(5) still TODO: SMPP SMSC
libsmpp34 is still listening on 0.0.0.0:2775, requires a change in libsmpp34.
Will follow in a subsequent patch.
Neels Hofmeyr (6):
enable telnet VTY bind address config for various programs
osmo-nitb: add -M to pass specific MNCC socket path
osmo-nitb: cosmetic: rename to rf_ctrl_path, following mncc_sock_path
osmo-nitb: be strict about cmdline args
enable ctrl bind config for various programs
bsc_nat: fail if VTY telnet port cannot be bound
openbsc/include/openbsc/bsc_nat.h | 3 ++-
openbsc/include/openbsc/ctrl.h | 3 ++-
openbsc/include/openbsc/gprs_sgsn.h | 3 ++-
openbsc/include/openbsc/mncc.h | 2 +-
openbsc/src/gprs/gb_proxy_main.c | 12 ++++++----
openbsc/src/gprs/gtphub_main.c | 11 ++++++---
openbsc/src/gprs/sgsn_ctrl.c | 5 ++--
openbsc/src/gprs/sgsn_main.c | 42 ++++++++++++++++++++++-----------
openbsc/src/libbsc/bsc_ctrl_lookup.c | 6 +++--
openbsc/src/libbsc/bsc_init.c | 6 ++++-
openbsc/src/libmsc/mncc_sock.c | 9 +++----
openbsc/src/osmo-bsc/osmo_bsc_main.c | 10 +++++++-
openbsc/src/osmo-bsc_mgcp/mgcp_main.c | 6 ++++-
openbsc/src/osmo-bsc_nat/bsc_nat.c | 21 ++++++++++++++---
openbsc/src/osmo-bsc_nat/bsc_nat_ctrl.c | 5 ++--
openbsc/src/osmo-nitb/bsc_hack.c | 39 ++++++++++++++++++++----------
16 files changed, 130 insertions(+), 53 deletions(-)
--
2.1.4
FYI, Rusty's CCAN contains some interesting option parsing code released
under GPLv2+, i.e. compatible to libosmocore:
Usage is explained in
https://github.com/rustyrussell/ccan/blob/master/ccan/opt/_info
It seems rather small and simple, and permits the subsequent addition of
options, i.e. some shared code can register options, and other parts of
the code can register even more options to it (like our libraries, or
bts-specific code in osmo-bts, ...)
I have more pressing things on my todo list than convert this now, but
as there was some discussion regarding gengetopt here recently, I
thought I might point out an alternative.
We might also look into the LGPL 2.1+ CCAN htable
https://github.com/rustyrussell/ccan/blob/master/ccan/htable/_info
as a possible replacement for those areas where our linear llist
iterations should turn out to be problematic.
Regards,
Harald
--
- Harald Welte <laforge(a)gnumonks.org> http://laforge.gnumonks.org/
============================================================================
"Privacy in residential applications is a desirable marketing option."
(ETSI EN 300 175-7 Ch. A6)