sr-dev September 2011

sr-dev@lists.kamailio.org

40 participants
298 discussions

by Timo Reimann

Hi all, is there a specific reason we do not set git tags for our releases (e.g., 3.1.x)? While we used to do it for Kamailio 1.x.y, it doesn't seem to be the case for SR 3. Thanks and cheers, --Timo

13 years

dialog functions in reply/failure route

by Jasmin Schnatterbeck

Hello list, in order to deal with NAT-Clients properly I am using dialog-flags to store the information, whether callee or caller is NATed. So if there is an incoming call to a NATed client, the callee-NAT-DLGflag is being set directly after usrloc lookup. But: If the client returns a 3xx response which is processed by the kamailio instance that has just contaced the device, the callee (usually one of our sip-proxies) is NOT NATed anymore - so I need to reset the callee-nat-dlgflag out of 3xx-reply route. And that seems to be a problem. In reply route the dlg functions/flags are not available, and trying to call a normal route with dlg_manage() out of reply(or failure) route results in strange behaviour... I think it is generally a good idea to have access to dlg flags/functions in reply/failure route. So what about the suggestion to identify the dialog, if a reply on a request arrives? Have a nice weekend Jasmin

13 years

kamailio debian/ubuntu support

by Jon Bonilla

Hi all with the release of kamailio 3.1.5, I've done a small change to the debian repositories: Nightly builds of kamailio_3.0 branch have been dropped and the repository has been erased. There is no development in this branch and it has no sense to keep nightly builds for it. Those (few) people using kamailio_3.0-nightly repositories are advised to move their sources.list to the stable 3.0 repository of upgrade to 3.1 or master branches. kamailio 3.0 will continue to be packaged if any new stable version is released for etch, lenny, squeeze and lucid. For 3.1 branch, nightly builds and stable repos will be provided for lenny, lucid and squeeze When the time for 3.2 (master branch at the moment) comes, I'll add Wheezy support. Not sure about Ubuntu distro support. Should we add support to Ubuntu 11.10 or wait until the lts 12.04 is released? Any opinions welcomed cheers, Jon

13 years

git:master: cfg framework: fix the freeing of the replaced strings

by Miklos Tirpak

Module: sip-router Branch: master Commit: 67172188fa23112fa449cf60d790bc84d02fed28 URL: http://git.sip-router.org/cgi-bin/gitweb.cgi/sip-router/?a=commit;h=6717218… Author: Miklos Tirpak <miklos(a)iptel.org> Committer: Miklos Tirpak <miklos(a)iptel.org> Date: Thu Sep 15 17:05:36 2011 +0200 cfg framework: fix the freeing of the replaced strings The replaced strings and the memory block of the replaced group instances cannot be freed when the old configuration block is freed. There might be a child process using an even older configuration that references to the same string value or to the same group instance that is beeing replaced. Hence, as long as there is any child process with an older configuration, the replaced strings cannot be freed. The fix is to link the replaced strings to the per-child process callback list instead of the old cfg block. When the last child process updates its configuration, it also frees the old string values. --- cfg/cfg_ctx.c | 4 ++-- cfg/cfg_struct.c | 33 +++++++++++++++++++++++++-------- cfg/cfg_struct.h | 38 ++++++++++++++++++++++---------------- 3 files changed, 49 insertions(+), 26 deletions(-) diff --git a/cfg/cfg_ctx.c b/cfg/cfg_ctx.c index 304a8c8..e0a18e7 100644 --- a/cfg/cfg_ctx.c +++ b/cfg/cfg_ctx.c @@ -649,7 +649,7 @@ error: if (cfg_shmized) CFG_WRITER_UNLOCK(); if (block) cfg_block_free(block); if (new_array) shm_free(new_array); - if (child_cb) cfg_child_cb_free(child_cb); + if (child_cb) cfg_child_cb_free_list(child_cb); if (replaced) shm_free(replaced); error0: @@ -1235,7 +1235,7 @@ error: error0: CFG_CTX_UNLOCK(ctx); - if (child_cb_first) cfg_child_cb_free(child_cb_first); + if (child_cb_first) cfg_child_cb_free_list(child_cb_first); if (replaced) shm_free(replaced); return -1; diff --git a/cfg/cfg_struct.c b/cfg/cfg_struct.c index e07098a..c2e6520 100644 --- a/cfg/cfg_struct.c +++ b/cfg/cfg_struct.c @@ -401,7 +401,7 @@ void cfg_destroy(void) cfg_free_selects(); if (cfg_child_cb_first) { - if (*cfg_child_cb_first) cfg_child_cb_free(*cfg_child_cb_first); + if (*cfg_child_cb_first) cfg_child_cb_free_list(*cfg_child_cb_first); shm_free(cfg_child_cb_first); cfg_child_cb_first = NULL; } @@ -533,7 +533,7 @@ void cfg_child_destroy(void) if (*cfg_child_cb_first == prev_cb) { /* yes, this process was blocking the deletion */ *cfg_child_cb_first = cfg_child_cb; - shm_free(prev_cb); + cfg_child_cb_free_item(prev_cb); } } else { /* no need to continue, because there is at least @@ -793,6 +793,26 @@ void cfg_install_global(cfg_block_t *block, void **replaced, CFG_REF(block); + if (replaced) { + /* The replaced array is specified, it has to be linked to the child cb structure. + * The last child process processing this structure will free the old strings and the array. */ + if (cb_first) { + cb_first->replaced = replaced; + } else { + /* At least one child cb structure is needed. */ + cb_first = cfg_child_cb_new(NULL, NULL, NULL, 0 /* gname, name, cb, type */); + if (cb_first) { + cb_last = cb_first; + cb_first->replaced = replaced; + } else { + LOG(L_ERR, "ERROR: cfg_install_global(): not enough shm memory\n"); + /* Nothing more can be done here, the replaced strings are still needed, + * they cannot be freed at this moment. + */ + } + } + } + CFG_LOCK(); old_cfg = *cfg_global; @@ -803,11 +823,8 @@ void cfg_install_global(cfg_block_t *block, void **replaced, CFG_UNLOCK(); - if (old_cfg) { - if (replaced) (old_cfg)->replaced = replaced; + if (old_cfg) CFG_UNREF(old_cfg); - } - } /* creates a structure for a per-child process callback */ @@ -854,7 +871,7 @@ cfg_child_cb_t *cfg_child_cb_new(str *gname, str *name, } /* free the memory allocated for a child cb list */ -void cfg_child_cb_free(cfg_child_cb_t *child_cb_first) +void cfg_child_cb_free_list(cfg_child_cb_t *child_cb_first) { cfg_child_cb_t *cb, *cb_next; @@ -863,7 +880,7 @@ void cfg_child_cb_free(cfg_child_cb_t *child_cb_first) cb = cb_next ) { cb_next = cb->next; - shm_free(cb); + cfg_child_cb_free_item(cb); } } diff --git a/cfg/cfg_struct.h b/cfg/cfg_struct.h index dca39a0..505c0b4 100644 --- a/cfg/cfg_struct.h +++ b/cfg/cfg_struct.h @@ -136,10 +136,6 @@ typedef struct _cfg_block { atomic_t refcnt; /*!< reference counter, the block is automatically deleted when it reaches 0 */ - void **replaced; /*!< set of the strings and other memory segments - that must be freed - together with the block. The content depends - on the block that replaces this one */ unsigned char vars[1]; /*!< blob that contains the values */ } cfg_block_t; @@ -159,7 +155,12 @@ typedef struct _cfg_child_cb { * <=0 the cb no longer needs to be executed */ str gname, name; /*!< name of the variable that has changed */ - cfg_on_set_child cb; /*!< callback function that has to be called */ + cfg_on_set_child cb; /*!< callback function that has to be called */ + void **replaced; /*!< set of strings and other memory segments + that must be freed together with this structure. + The content depends on the new config block. + This makes sure that the replaced strings are freed + after all the child processes release the old configuration. */ struct _cfg_child_cb *next; } cfg_child_cb_t; @@ -274,20 +275,23 @@ void cfg_set_group(cfg_group_t *group, /* copy the variables to shm mem */ int cfg_shmize(void); -/* free the memory of a config block */ -static inline void cfg_block_free(cfg_block_t *block) +/* free the memory of a child cb structure */ +static inline void cfg_child_cb_free_item(cfg_child_cb_t *cb) { int i; /* free the changed variables */ - if (block->replaced) { - for (i=0; block->replaced[i]; i++) - shm_free(block->replaced[i]); - shm_free(block->replaced); + if (cb->replaced) { + for (i=0; cb->replaced[i]; i++) + shm_free(cb->replaced[i]); + shm_free(cb->replaced); } - shm_free(block); + shm_free(cb); } +#define cfg_block_free(block) \ + shm_free(block) + /* Move the group handle to the specified group instance pointed by dst_ginst. * src_ginst shall point to the active group instance. * Both parameters can be NULL meaning that the src/dst config is the default, @@ -369,13 +373,15 @@ static inline void cfg_update_local(int no_cbs) /* yes, this process was blocking the deletion */ *cfg_child_cb_first = cfg_child_cb; CFG_UNLOCK(); - shm_free(prev_cb); + cfg_child_cb_free_item(prev_cb); } else { CFG_UNLOCK(); } } - if (atomic_add(&cfg_child_cb->cb_count, -1) >= 0) /* the new value is returned - by atomic_add() */ + if (cfg_child_cb->cb + && (atomic_add(&cfg_child_cb->cb_count, -1) >= 0) /* the new value is returned + by atomic_add() */ + ) /* execute the callback */ cfg_child_cb->cb(&cfg_child_cb->gname, &cfg_child_cb->name); /* else the callback no longer needs to be executed */ @@ -500,7 +506,7 @@ cfg_child_cb_t *cfg_child_cb_new(str *gname, str *name, unsigned int type); /* free the memory allocated for a child cb list */ -void cfg_child_cb_free(cfg_child_cb_t *child_cb_first); +void cfg_child_cb_free_list(cfg_child_cb_t *child_cb_first); /* Allocate memory for a new additional variable * and link it to a configuration group.

13 years

Re: [sr-dev] registrar save() with multiple Contact headers

by Henning Westerholt

On Friday 09 September 2011, Henning Westerholt wrote: > [..] > You can try to apply this patch locally, wait until next week for 3.1.4 or > check out the 3.1 master branch from the repository. small correction, i meant of course 3.1.5, the 3.1.4 was released some month ago. Henning

13 years

multiple values in mtree modes

by Juha Heinanen

daniel, if i have understood correctly, current mtree implementation does not allow more than one value per node (prefix) in the three. it some cases it might make sense to have more that one value. for example, if an mtree is used to map telephone numbers to values, one could have: prefix value 0046 sweden 0046 nordic country 0046 eu member 0046 european country 00375 belarus 00375 european country and when a telephone number, such as 004692345678 or 0037592345678 is matched, all values that the prefix has, would be assigned to an avp. is is ok for you as author of mtree module, if i try to extend it so that a prefix could be associated with more than one value? -- juha

13 years

git:master: Fix typo in acc_cdr

by Sven Knoblich

Module: sip-router Branch: master Commit: 4e5338445645a0f4a51122b9224f801bba4107f3 URL: http://git.sip-router.org/cgi-bin/gitweb.cgi/sip-router/?a=commit;h=4e53384… Author: Sven Knoblich <sven.knoblich(a)1und1.de> Committer: Sven Knoblich <sven.knoblich(a)1und1.de> Date: Thu Sep 15 10:22:07 2011 +0200 Fix typo in acc_cdr - rename dot_adress into dot_address --- modules_k/acc/acc_cdr.c | 12 ++++++------ 1 files changed, 6 insertions(+), 6 deletions(-) diff --git a/modules_k/acc/acc_cdr.c b/modules_k/acc/acc_cdr.c index 9595718..fda436a 100644 --- a/modules_k/acc/acc_cdr.c +++ b/modules_k/acc/acc_cdr.c @@ -199,7 +199,7 @@ static int write_cdr( struct dlg_cell* dialog, /* convert a string into a timeb struct */ static struct timeb time_from_string( str* time_value) { - char* dot_adress = NULL; + char* dot_address = NULL; int dot_position = -1; char zero_terminated_value[TIME_STR_BUFFER_SIZE]; @@ -220,9 +220,9 @@ static struct timeb time_from_string( str* time_value) memcpy( zero_terminated_value, time_value->s, time_value->len); zero_terminated_value[time_value->len] = '\0'; - dot_adress = strchr( zero_terminated_value, time_separator); + dot_address = strchr( zero_terminated_value, time_separator); - if( !dot_adress) + if( !dot_address) { LM_ERR( "failed to find separator('%c') in '%s'!\n", time_separator, @@ -230,17 +230,17 @@ static struct timeb time_from_string( str* time_value) return time_error; } - dot_position = dot_adress-zero_terminated_value + 1; + dot_position = dot_address-zero_terminated_value + 1; if( dot_position >= strlen(zero_terminated_value) || - strchr(dot_adress + 1, time_separator)) + strchr(dot_address + 1, time_separator)) { LM_ERR( "invalid time-string '%s'\n", zero_terminated_value); return time_error; } return (struct timeb) { atoi( zero_terminated_value), - atoi( dot_adress + 1), + atoi( dot_address + 1), 0, 0}; }

13 years

Kamailio v3.1.5 Released

by Daniel-Constantin Mierla

Hello, Kamailio SIP Server v3.1.5 stable release is out. This is a maintenance release of latest stable branch, 3.1, that includes fixes since release of v3.1.4. There is no change to database schema or configuration language structure. Deployments running previous v3.1.x versions are strongly recommended to be upgraded to v3.1.5. For more details about version 3.1.5, visit: http://www.kamailio.org/w/2011/09/kamailio-v3-1-5-released/ Cheers, Daniel -- Daniel-Constantin Mierla -- http://www.asipto.com Kamailio Advanced Training, Oct 10-13, Berlin: http://asipto.com/u/kat http://linkedin.com/in/miconda -- http://twitter.com/miconda

13 years

git:3.1: tm: proper data type for xavp in tm_xdata_t struct

by Daniel-Constantin Mierla

Module: sip-router Branch: 3.1 Commit: ec672ad06125b86af81ba4b056e5853cf1086ece URL: http://git.sip-router.org/cgi-bin/gitweb.cgi/sip-router/?a=commit;h=ec672ad… Author: Daniel-Constantin Mierla <miconda(a)gmail.com> Committer: Daniel-Constantin Mierla <miconda(a)gmail.com> Date: Wed Sep 14 20:59:54 2011 +0200 tm: proper data type for xavp in tm_xdata_t struct - reported by Juha Heinanen - no real effect on branch 3.1 as xavps not enabled, but anyhow safer (cherry picked from commit 260d43bb8bda0e9f0445d7d942dc10e976ff4621) --- modules/tm/h_table.h | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/modules/tm/h_table.h b/modules/tm/h_table.h index ea8acd4..6dbca5d 100644 --- a/modules/tm/h_table.h +++ b/modules/tm/h_table.h @@ -313,7 +313,7 @@ typedef struct tm_xdata struct usr_avp *domain_avps_from; struct usr_avp *domain_avps_to; #ifdef WITH_XAVP - sr_xavp_t **xavps_list; + sr_xavp_t *xavps_list; #endif } tm_xdata_t;

13 years

git:master: tm: proper data type for xavp in tm_xdata_t struct

by Daniel-Constantin Mierla

Module: sip-router Branch: master Commit: 260d43bb8bda0e9f0445d7d942dc10e976ff4621 URL: http://git.sip-router.org/cgi-bin/gitweb.cgi/sip-router/?a=commit;h=260d43b… Author: Daniel-Constantin Mierla <miconda(a)gmail.com> Committer: Daniel-Constantin Mierla <miconda(a)gmail.com> Date: Wed Sep 14 20:59:54 2011 +0200 tm: proper data type for xavp in tm_xdata_t struct - reported by Juha Heinanen - no real effect on branch 3.1 as xavps not enabled, but anyhow safer --- modules/tm/h_table.h | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/modules/tm/h_table.h b/modules/tm/h_table.h index 8b8741b..7ee089e 100644 --- a/modules/tm/h_table.h +++ b/modules/tm/h_table.h @@ -312,7 +312,7 @@ typedef struct tm_xdata struct usr_avp *domain_avps_from; struct usr_avp *domain_avps_to; #ifdef WITH_XAVP - sr_xavp_t **xavps_list; + sr_xavp_t *xavps_list; #endif } tm_xdata_t;

13 years

← Newer
1
...
12
13
14
15
16
17
18
...
30
Older →

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

sr-dev September 2011