OpenBSC

openbsc@lists.osmocom.org

16 participants
4280 discussions

[PATCH 1/9] mgcp/rtp: Refactor mgcp_send to avoid code duplication
by Jacob Erlbeck 16 Jan '14

16 Jan '14

Currently there are two symmetric code paths which are selected by the packet destination (NET or BTS). This patch introduces 3 variables that take the different values and unifies both code paths into one. Sponsored-by: On-Waves ehf --- openbsc/src/libmgcp/mgcp_network.c | 50 ++++++++++++++++-------------------- 1 file changed, 22 insertions(+), 28 deletions(-) diff --git a/openbsc/src/libmgcp/mgcp_network.c b/openbsc/src/libmgcp/mgcp_network.c index 657019e..1c7c3da 100644 --- a/openbsc/src/libmgcp/mgcp_network.c +++ b/openbsc/src/libmgcp/mgcp_network.c @@ -537,6 +537,10 @@ static int mgcp_send(struct mgcp_endpoint *endp, int dest, int is_rtp, struct sockaddr_in *addr, char *buf, int rc) { struct mgcp_trunk_config *tcfg = endp->tcfg; + struct mgcp_rtp_end *rtp_end; + struct mgcp_rtp_state *rtp_state; + int tap_idx; + /* For loop toggle the destination and then dispatch. */ if (tcfg->audio_loop) dest = !dest; @@ -546,35 +550,25 @@ static int mgcp_send(struct mgcp_endpoint *endp, int dest, int is_rtp, dest = !dest; if (dest == MGCP_DEST_NET) { - if (is_rtp) { - mgcp_patch_and_count(endp, &endp->bts_state, - &endp->net_end, - addr, buf, rc); - forward_data(endp->net_end.rtp.fd, - &endp->taps[MGCP_TAP_NET_OUT], buf, rc); - return mgcp_udp_send(endp->net_end.rtp.fd, - &endp->net_end.addr, - endp->net_end.rtp_port, buf, rc); - } else if (!tcfg->omit_rtcp) { - return mgcp_udp_send(endp->net_end.rtcp.fd, - &endp->net_end.addr, - endp->net_end.rtcp_port, buf, rc); - } + rtp_end = &endp->net_end; + rtp_state = &endp->bts_state; + tap_idx = MGCP_TAP_NET_OUT; } else { - if (is_rtp) { - mgcp_patch_and_count(endp, &endp->net_state, - &endp->bts_end, - addr, buf, rc); - forward_data(endp->bts_end.rtp.fd, - &endp->taps[MGCP_TAP_BTS_OUT], buf, rc); - return mgcp_udp_send(endp->bts_end.rtp.fd, - &endp->bts_end.addr, - endp->bts_end.rtp_port, buf, rc); - } else if (!tcfg->omit_rtcp) { - return mgcp_udp_send(endp->bts_end.rtcp.fd, - &endp->bts_end.addr, - endp->bts_end.rtcp_port, buf, rc); - } + rtp_end = &endp->bts_end; + rtp_state = &endp->net_state; + tap_idx = MGCP_TAP_BTS_OUT; + } + + if (is_rtp) { + mgcp_patch_and_count(endp, rtp_state, rtp_end, addr, buf, rc); + forward_data(rtp_end->rtp.fd, &endp->taps[tap_idx], buf, rc); + return mgcp_udp_send(rtp_end->rtp.fd, + &rtp_end->addr, + rtp_end->rtp_port, buf, rc); + } else if (!tcfg->omit_rtcp) { + return mgcp_udp_send(rtp_end->rtcp.fd, + &rtp_end->addr, + rtp_end->rtcp_port, buf, rc); } return 0; -- 1.7.9.5

2 19

[RFC] [ADMIN] Making lists subscriber-only?
by Harald Welte 15 Jan '14

15 Jan '14

Dear all, so far the osmocom.org mailing lists have always been in a 'non-members are manually moderated' mode. This has created a lot of work for manual list moderation, where a lot of the messages caught are simply spam, and only the occasional valid message is being received. I'd like to thank the list moderators for taking care of this. However, in more recent discussions, we were considering to move the lists to a completely closed mode, i.e. postings would automatically be rejected from non-members. The automatic response would contain a description of how to subscribe in 'nomail' mode, i.e. to subscribe in a way to be able to post to the list, while still not receiving any incoming traffic. The latter should be fine for occasional posters who don't want the bulk e-mail that goes with a full/regular subscription. Please provide feedback in case you disagree with that change. Unless there is major opposition, we will likely transition to the 'closed' mode within one month. Thanks, 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)

7 7

[PATCH] Support of multiple RSL connections for ABIS/ipaccess (BTS side)
by Andreas Eversberg 14 Jan '14

14 Jan '14

In order to support multiple TRX, multiple RSL connections can be establised. e1inp_ipa_bts_rsl_connect() requires an additional parameter to set the TRX number. The code was tested with osmobts-trx. The user of e1inp_ipa_bts_rsl_connect() (which is osmo-bts) must be upgraded after applying the patch. src/common/oml.c - rc = e1inp_ipa_bts_rsl_connect(oml_link->ts->line, inet_ntoa(in), port); + rc = e1inp_ipa_bts_rsl_connect(oml_link->ts->line, inet_ntoa(in), port, 0); --- include/osmocom/abis/e1_input.h | 3 ++- src/input/ipaccess.c | 18 +++++++++++------- tests/e1inp_ipa_bts_test.c | 2 +- 3 files changed, 14 insertions(+), 9 deletions(-) diff --git a/include/osmocom/abis/e1_input.h b/include/osmocom/abis/e1_input.h index 0abf0b8..2a9890c 100644 --- a/include/osmocom/abis/e1_input.h +++ b/include/osmocom/abis/e1_input.h @@ -258,7 +258,8 @@ struct subch_mux *e1inp_get_mux(uint8_t e1_nr, uint8_t ts_nr); /* on an IPA BTS, the BTS needs to establish the RSL connection much * later than the OML connection. */ int e1inp_ipa_bts_rsl_connect(struct e1inp_line *line, - const char *rem_addr, uint16_t rem_port); + const char *rem_addr, uint16_t rem_port, + uint8_t trx_id); void e1inp_sign_link_destroy(struct e1inp_sign_link *link); int e1inp_line_update(struct e1inp_line *line); diff --git a/src/input/ipaccess.c b/src/input/ipaccess.c index 9722b2f..20894e4 100644 --- a/src/input/ipaccess.c +++ b/src/input/ipaccess.c @@ -717,7 +717,7 @@ err_line: #define IPA_STRING_MAX 64 static struct msgb * -ipa_bts_id_resp(struct ipaccess_unit *dev, uint8_t *data, int len) +ipa_bts_id_resp(struct ipaccess_unit *dev, uint8_t *data, int len, int trx_nr) { struct msgb *nmsg; char str[IPA_STRING_MAX]; @@ -738,7 +738,7 @@ ipa_bts_id_resp(struct ipaccess_unit *dev, uint8_t *data, int len) switch (data[1]) { case IPAC_IDTAG_UNIT: snprintf(str, sizeof(str), "%u/%u/%u", - dev->site_id, dev->bts_id, dev->trx_id); + dev->site_id, dev->bts_id, trx_nr); break; case IPAC_IDTAG_MACADDR: snprintf(str, sizeof(str), @@ -838,6 +838,7 @@ static int ipaccess_bts_read_cb(struct ipa_client_conn *link, struct msgb *msg) struct e1inp_sign_link *sign_link; uint8_t *data = msgb_l2(msg); int len = msgb_l2len(msg); + int trx_nr = 0; LOGP(DLINP, LOGL_NOTICE, "received ID get\n"); if (!link->line->ops->sign_link_up) { @@ -847,8 +848,10 @@ static int ipaccess_bts_read_cb(struct ipa_client_conn *link, struct msgb *msg) ret = -EINVAL; goto err; } + if (link->ofd->priv_nr >= E1INP_SIGN_RSL) + trx_nr = link->ofd->priv_nr - E1INP_SIGN_RSL; rmsg = ipa_bts_id_resp(link->line->ops->cfg.ipa.dev, - data + 1, len - 1); + data + 1, len - 1, trx_nr); ret = ipaccess_send(link->ofd->fd, rmsg->data, rmsg->len); if (ret != rmsg->len) { @@ -885,7 +888,7 @@ static int ipaccess_bts_read_cb(struct ipa_client_conn *link, struct msgb *msg) } else if (link->port == IPA_TCP_PORT_OML) e1i_ts = &link->line->ts[0]; else if (link->port == IPA_TCP_PORT_RSL) - e1i_ts = &link->line->ts[1]; + e1i_ts = &link->line->ts[link->ofd->priv_nr-1]; OSMO_ASSERT(e1i_ts != NULL); @@ -1016,13 +1019,14 @@ static int ipaccess_line_update(struct e1inp_line *line) } int e1inp_ipa_bts_rsl_connect(struct e1inp_line *line, - const char *rem_addr, uint16_t rem_port) + const char *rem_addr, uint16_t rem_port, + uint8_t trx_id) { struct ipa_client_conn *rsl_link; rsl_link = ipa_client_conn_create(tall_ipa_ctx, - &line->ts[E1INP_SIGN_RSL-1], - E1INP_SIGN_RSL, + &line->ts[E1INP_SIGN_RSL+trx_id-1], + E1INP_SIGN_RSL+trx_id, rem_addr, rem_port, ipaccess_bts_updown_cb, ipaccess_bts_read_cb, diff --git a/tests/e1inp_ipa_bts_test.c b/tests/e1inp_ipa_bts_test.c index 02a4cb3..a43dba3 100644 --- a/tests/e1inp_ipa_bts_test.c +++ b/tests/e1inp_ipa_bts_test.c @@ -70,7 +70,7 @@ sign_link_up(void *unit, struct e1inp_line *line, enum e1inp_sign_type type) /* Now we can send OML messages to the BSC. */ bts_state = BTS_TEST_OML_SIGN_LINK_UP; } - e1inp_ipa_bts_rsl_connect(line, "127.0.0.1", IPA_TCP_PORT_RSL); + e1inp_ipa_bts_rsl_connect(line, "127.0.0.1", IPA_TCP_PORT_RSL, 0); break; case E1INP_SIGN_RSL: LOGP(DBTSTEST, LOGL_NOTICE, "RSL link up request received.\n"); -- 1.8.1.5 --------------080608010203010806040100--

1 0

[PATCH] Support for multiple RSL connections with ABIS/ipaccess (BTS side)
by Andreas Eversberg 14 Jan '14

14 Jan '14

In order to support multiple TRX, multiple RSL connections can be establised. e1inp_ipa_bts_rsl_connect() requires an additional parameter to set the TRX number. The code was successfully tested with osmobts-trx and UmTRX with two transceivers. The user of e1inp_ipa_bts_rsl_connect() and e1inp_line_bind_ops() (which is osmo-bts) must be upgraded after applying the patch. --- include/osmocom/abis/e1_input.h | 3 ++- src/input/ipaccess.c | 16 ++++++++++------ tests/e1inp_ipa_bts_test.c | 2 +- 3 files changed, 13 insertions(+), 8 deletions(-) diff --git a/include/osmocom/abis/e1_input.h b/include/osmocom/abis/e1_input.h index 0abf0b8..2a9890c 100644 --- a/include/osmocom/abis/e1_input.h +++ b/include/osmocom/abis/e1_input.h @@ -258,7 +258,8 @@ struct subch_mux *e1inp_get_mux(uint8_t e1_nr, uint8_t ts_nr); /* on an IPA BTS, the BTS needs to establish the RSL connection much * later than the OML connection. */ int e1inp_ipa_bts_rsl_connect(struct e1inp_line *line, - const char *rem_addr, uint16_t rem_port); + const char *rem_addr, uint16_t rem_port, + uint8_t trx_id); void e1inp_sign_link_destroy(struct e1inp_sign_link *link); int e1inp_line_update(struct e1inp_line *line); diff --git a/src/input/ipaccess.c b/src/input/ipaccess.c index 9722b2f..86ddd3a 100644 --- a/src/input/ipaccess.c +++ b/src/input/ipaccess.c @@ -838,6 +838,7 @@ static int ipaccess_bts_read_cb(struct ipa_client_conn *link, struct msgb *msg) struct e1inp_sign_link *sign_link; uint8_t *data = msgb_l2(msg); int len = msgb_l2len(msg); + struct ipaccess_unit *unit_data; LOGP(DLINP, LOGL_NOTICE, "received ID get\n"); if (!link->line->ops->sign_link_up) { @@ -847,7 +848,9 @@ static int ipaccess_bts_read_cb(struct ipa_client_conn *link, struct msgb *msg) ret = -EINVAL; goto err; } - rmsg = ipa_bts_id_resp(link->line->ops->cfg.ipa.dev, + unit_data = (struct ipaccess_unit *) + link->line->ops->cfg.ipa.dev; + rmsg = ipa_bts_id_resp(&unit_data[link->ofd->priv_nr-1], data + 1, len - 1); ret = ipaccess_send(link->ofd->fd, rmsg->data, rmsg->len); @@ -869,7 +872,7 @@ static int ipaccess_bts_read_cb(struct ipa_client_conn *link, struct msgb *msg) } msgb_free(rmsg); - sign_link = link->line->ops->sign_link_up(msg, + sign_link = link->line->ops->sign_link_up(unit_data, link->line, link->ofd->priv_nr); if (sign_link == NULL) { @@ -885,7 +888,7 @@ static int ipaccess_bts_read_cb(struct ipa_client_conn *link, struct msgb *msg) } else if (link->port == IPA_TCP_PORT_OML) e1i_ts = &link->line->ts[0]; else if (link->port == IPA_TCP_PORT_RSL) - e1i_ts = &link->line->ts[1]; + e1i_ts = &link->line->ts[link->ofd->priv_nr-1]; OSMO_ASSERT(e1i_ts != NULL); @@ -1016,13 +1019,14 @@ static int ipaccess_line_update(struct e1inp_line *line) } int e1inp_ipa_bts_rsl_connect(struct e1inp_line *line, - const char *rem_addr, uint16_t rem_port) + const char *rem_addr, uint16_t rem_port, + uint8_t trx_id) { struct ipa_client_conn *rsl_link; rsl_link = ipa_client_conn_create(tall_ipa_ctx, - &line->ts[E1INP_SIGN_RSL-1], - E1INP_SIGN_RSL, + &line->ts[E1INP_SIGN_RSL+trx_id-1], + E1INP_SIGN_RSL+trx_id, rem_addr, rem_port, ipaccess_bts_updown_cb, ipaccess_bts_read_cb, diff --git a/tests/e1inp_ipa_bts_test.c b/tests/e1inp_ipa_bts_test.c index 02a4cb3..a43dba3 100644 --- a/tests/e1inp_ipa_bts_test.c +++ b/tests/e1inp_ipa_bts_test.c @@ -70,7 +70,7 @@ sign_link_up(void *unit, struct e1inp_line *line, enum e1inp_sign_type type) /* Now we can send OML messages to the BSC. */ bts_state = BTS_TEST_OML_SIGN_LINK_UP; } - e1inp_ipa_bts_rsl_connect(line, "127.0.0.1", IPA_TCP_PORT_RSL); + e1inp_ipa_bts_rsl_connect(line, "127.0.0.1", IPA_TCP_PORT_RSL, 0); break; case E1INP_SIGN_RSL: LOGP(DBTSTEST, LOGL_NOTICE, "RSL link up request received.\n"); -- 1.8.1.5 --------------030007020209040104000503 Content-Type: text/x-diff; name="osmo-bts_Support-for-multiple-RSL-connections.patch" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="osmo-bts_Support-for-multiple-RSL-connections.patch"

1 0

[PATCH] libabis: Enable TCP keepalive by default on all BTS connections
by Jacob Erlbeck 14 Jan '14

14 Jan '14

From: Daniel Willmann <daniel(a)totalueberwachung.de> This way we can find out fast if the connection is broken. Ticket: OW#1060 --- src/input/ipaccess.c | 35 ++++++++++++++++++++++++++++++++++- 1 file changed, 34 insertions(+), 1 deletion(-) diff --git a/src/input/ipaccess.c b/src/input/ipaccess.c index 4a56569..371b21c 100644 --- a/src/input/ipaccess.c +++ b/src/input/ipaccess.c @@ -28,6 +28,7 @@ #include <stdlib.h> #include <stdbool.h> #include <errno.h> +#include <netinet/tcp.h> #include <string.h> #include <time.h> #include <sys/fcntl.h> @@ -50,6 +51,10 @@ static void *tall_ipa_ctx; #define TS1_ALLOC_SIZE 900 +#define DEFAULT_TCP_KEEPALIVE_IDLE_TIMEOUT 30 +#define DEFAULT_TCP_KEEPALIVE_INTERVAL 3 +#define DEFAULT_TCP_KEEPALIVE_RETRY_COUNT 10 + /* * Common propietary IPA messages: * - PONG: in reply to PING. @@ -605,7 +610,7 @@ static int ipaccess_bsc_oml_cb(struct ipa_server_link *link, int fd) { int ret; int idx = 0; - int i; + int i, val; struct e1inp_line *line; struct e1inp_ts *e1i_ts; struct osmo_fd *bfd; @@ -638,6 +643,34 @@ static int ipaccess_bsc_oml_cb(struct ipa_server_link *link, int fd) goto err_line; } + /* Enable TCP keepalive to find out if the connection is gone */ + val = 1; + ret = setsockopt(bfd->fd, SOL_SOCKET, SO_KEEPALIVE, &val, sizeof(val)); + if (ret < 0) + LOGP(DLINP, LOGL_NOTICE, "Failed to set keepalive: %s\n", + strerror(errno)); + else + LOGP(DLINP, LOGL_NOTICE, "Keepalive is set: %i\n", ret); + +#if defined(TCP_KEEPIDLE) && defined(TCP_KEEPINTVL) && defined(TCP_KEEPCNT) + /* The following options are not portable! */ + val = DEFAULT_TCP_KEEPALIVE_IDLE_TIMEOUT; + ret = setsockopt(bfd->fd, IPPROTO_TCP, TCP_KEEPIDLE, &val, sizeof(val)); + if (ret < 0) + LOGP(DLINP, LOGL_NOTICE, "Failed to set keepalive idle time: %s\n", + strerror(errno)); + val = DEFAULT_TCP_KEEPALIVE_INTERVAL; + ret = setsockopt(bfd->fd, IPPROTO_TCP, TCP_KEEPINTVL, &val, sizeof(val)); + if (ret < 0) + LOGP(DLINP, LOGL_NOTICE, "Failed to set keepalive interval: %s\n", + strerror(errno)); + val = DEFAULT_TCP_KEEPALIVE_RETRY_COUNT; + ret = setsockopt(bfd->fd, IPPROTO_TCP, TCP_KEEPCNT, &val, sizeof(val)); + if (ret < 0) + LOGP(DLINP, LOGL_NOTICE, "Failed to set keepalive count: %s\n", + strerror(errno)); +#endif + /* Request ID. FIXME: request LOCATION, HW/SW VErsion, Unit Name, Serno */ ret = ipaccess_send_id_req(bfd->fd); if (ret < 0) { -- 1.7.9.5

3 15

1 OpenBSC moderator request(s) waiting
by openbsc-bounces＠lists.osmocom.org 14 Jan '14

14 Jan '14

The OpenBSC(a)lists.osmocom.org mailing list has 1 request(s) waiting for your consideration at: https://lists.osmocom.org/mailman/admindb/openbsc Please attend to this at your earliest convenience. This notice of pending requests, if any, will be sent out daily. Pending posts: From: edoardo.rispoli(a)gmail.com on Mon Jan 13 23:46:56 2014 Subject: Re: [RFC] [ADMIN] Making lists subscriber-only? Cause: Post by non-member to a members-only list

1 0

OsmoBTS on RAD1
by Kurtis Heimerl 14 Jan '14

14 Jan '14

Hey OpenBSC, Tom, I've been working on getting OsmoBTS running on the recently open-sourced Range Networks RAD1 radio. The code is available here: https://github.com/kheimerl/vbts-openbts/tree/osmotrx. I am stuck now and I was hoping the mailing list might be able to provide some perspective. With the RAD1, the system is beaconing correctly. However, phones are unable to camp. I logged a phone trying to camp on both the RAD1 and a B100 to compare the output and see if anything jumps out. The osmoBTS/osmo-nitb logs are seemingly identical, but the transceiver outputs are different. Both of the transceiver outputs are attached. The only big difference I see is in the "underflows" on the RAD1, which in my experience is a deal-breaker; that's not usually an easy fix. Does anyone have a different perspective on what might be going on in here? Thanks!

2 3

[PATCH] gsm/gsm48ie: Fix range 256 W[i] decoding
by Jacob Erlbeck 14 Jan '14

14 Jan '14

Currently w[14]/w[15] and w[18]/w[19] are swapped in range 256 format decoding in gsm48_decode_freq_list(). This patch fixes this. Sponsored-by: On-Waves ehf --- src/gsm/gsm48_ie.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/src/gsm/gsm48_ie.c b/src/gsm/gsm48_ie.c index 78619b9..2cc0645 100644 --- a/src/gsm/gsm48_ie.c +++ b/src/gsm/gsm48_ie.c @@ -985,17 +985,17 @@ int gsm48_decode_freq_list(struct gsm_sysinfo_freq *f, uint8_t *cd, if (len >= 12) w[13] = r->w13; if (len >= 13) - w[14] = r->w15; + w[14] = (r->w14_hi << 2) | r->w14_lo; if (len >= 13) - w[15] = (r->w14_hi << 2) | r->w14_lo; + w[15] = r->w15; if (len >= 14) w[16] = (r->w16_hi << 3) | r->w16_lo; if (len >= 14) w[17] = r->w17; if (len >= 15) - w[18] = r->w19; + w[18] = (r->w18_hi << 3) | r->w18_lo; if (len >= 15) - w[19] = (r->w18_hi << 3) | r->w18_lo; + w[19] = r->w19; if (len >= 16) w[20] = (r->w20_hi << 3) | r->w20_lo; if (len >= 16) -- 1.7.9.5

1 0

3 OpenBSC moderator request(s) waiting
by openbsc-bounces＠lists.osmocom.org 13 Jan '14

13 Jan '14

The OpenBSC(a)lists.osmocom.org mailing list has 3 request(s) waiting for your consideration at: https://lists.osmocom.org/mailman/admindb/openbsc Please attend to this at your earliest convenience. This notice of pending requests, if any, will be sent out daily. Pending posts: From: kheimerl(a)cs.berkeley.edu on Thu Jan 9 21:27:04 2014 Subject: OsmoBTS on RAD1 Cause: Message body is too big: 5402368 bytes with a limit of 400 KB From: kaber(a)trash.net on Sun Jan 12 20:39:11 2014 Subject: Re: [RFC] [ADMIN] Making lists subscriber-only? Cause: Too many recipients to the message From: holger(a)freyther.de on Sun Jan 12 20:55:56 2014 Subject: Re: [RFC] [ADMIN] Making lists subscriber-only? Cause: Too many recipients to the message

1 0

abis_nm_pchan4chcomb API and return value
by Holger Hans Peter Freyther 13 Jan '14

13 Jan '14

Dear Harald, there is an open coverity issue in osmo-bts (CID 1040768) and it is about mixing enum types. enum abis_nm_chan_comb abis_nm_pchan4chcomb(uint8_t chcomb)

2 1

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

OpenBSC