- sr-dev - kamailio.org

[kamailio/kamailio] ims_registrar_scscf: add S-CSCF restoration (TS 23.380) (PR #4968)
by Lyubomir Yatanski 08 Oct '26

08 Oct '26

#### Pre-Submission Checklist    - [x] Commit message has the format required by CONTRIBUTING guide - [x] Commits are split per component (core, individual modules, libs, utils, ...) - [x] Each component has a single commit (if not, squash them into one commit) - [x] LLM/AI Coding Assistants were involved in creating the C code or other included content - [x] Code is formatted with `clang-format` using the config file `.clang-format` from source code folder - [x] No commits to README files for modules (changes must be done to docbook files in `doc/` subfolder, the README file is autogenerated) #### Type Of Change - [ ] Small bug fix (non-breaking change which fixes an issue) - [x] New feature (non-breaking change which adds new functionality) - [ ] Breaking change (fix or feature that would change existing functionality) #### Checklist:  - [ ] PR should be backported to stable branches - [x] Tested changes locally - [ ] Related to issue #XXXX (replace XXXX with an open issue number) #### Description  Back up the Contact and Path at the HSS and fetch them back over Cx, as an alternative to the usrloc database: - REGISTRATION/RE_REGISTRATION SARs carry SCSCF-Restoration-Info and Supported-Features (IMSRestorationInd) - SAA restoration info is restored into usrloc for UNREGISTERED_USER and NO_ASSIGNMENT; DIAMETER_ERROR_IN_ASSIGNMENT_TYPE with restoration info is handled as the restoration answer - new assign_server(route, domain, direction, type) for script-initiated SARs, NO_ASSIGNMENT being the originating restoration case New modparam scscf_restoration_info_enabled, default 0: an HSS without IMSRestorationInd rejects the mandatory Supported-Features AVP. With it off, behaviour is unchanged. Compared to db_mode, this needs no SQL backend (usrloc_db.c uses raw MySQL queries) and lets a replacement S-CSCF sharing the Server-Name recover bindings. Only the first binding per IMPI is restored, with placeholder Call-ID/CSeq and the default expiry, until the UE re-registers. You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4968 -- Commit Summary -- * ims_registrar_scscf: add S-CSCF restoration (TS 23.380) -- File Changes -- M src/modules/ims_registrar_scscf/cxdx_avp.c (146) M src/modules/ims_registrar_scscf/cxdx_avp.h (44) M src/modules/ims_registrar_scscf/cxdx_sar.c (147) M src/modules/ims_registrar_scscf/cxdx_sar.h (1) M src/modules/ims_registrar_scscf/doc/ims_registrar_scscf_admin.xml (121) M src/modules/ims_registrar_scscf/ims_registrar_scscf_mod.c (43) M src/modules/ims_registrar_scscf/ims_registrar_scscf_mod.h (1) M src/modules/ims_registrar_scscf/save.c (244) M src/modules/ims_registrar_scscf/save.h (15) M src/modules/ims_registrar_scscf/server_assignment.c (5) M src/modules/ims_usrloc_scscf/ucontact.c (8) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4968.patch https://github.com/kamailio/kamailio/pull/4968.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4968 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4968(a)github.com>

2 5

[kamailio/kamailio] tm, tmx: add pseudo-variables to identify set transaction branches (PR #4987)
by Ben Kaufman 08 Oct '26

08 Oct '26

- expose the names of armed failure, branch, onreply, and branch-failure routes through $T(failure_route), $T(branch_route), $T(onreply_route), and $T(branch_failure_route) - add TM API getters so tmx can resolve the armed route names   #### Pre-Submission Checklist    - [x] Commit message has the format required by CONTRIBUTING guide - [x] Commits are split per component (core, individual modules, libs, utils, ...) - [x] Each component has a single commit (if not, squash them into one commit) - [x] LLM/AI Coding Assistants were involved in creating the C code or other included content - [x] Code is formatted with `clang-format` using the config file `.clang-format` from source code folder - [x] No commits to README files for modules (changes must be done to docbook files in `doc/` subfolder, the README file is autogenerated) #### Type Of Change - [ ] Small bug fix (non-breaking change which fixes an issue) - [x] New feature (non-breaking change which adds new functionality) - [ ] Breaking change (fix or feature that would change existing functionality) #### Checklist:  - [ ] PR should be backported to stable branches - [x] Tested changes locally - [x] Related to issue #4986 #### Description Adds the following pseudo variables, each returning the name of the corresponding transaction route (if set), and null if not set. - `$T(branch_route)` - `$T(failure_route)` - `$T(onreply_route)` - `$T(branch_failure_route)` You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4987 -- Commit Summary -- * tm: expose armed route getters through module API * tmx: add armed route name pseudo-variables -- File Changes -- M src/modules/tm/tm_load.c (4) M src/modules/tm/tm_load.h (6) M src/modules/tmx/doc/tmx_admin.xml (16) M src/modules/tmx/t_var.c (98) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4987.patch https://github.com/kamailio/kamailio/pull/4987.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4987 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4987(a)github.com>

2 1

[kamailio/kamailio] topos: add handle_unmask_miss modparam (PR #4813)
by Stefan Mititelu 08 Oct '26

08 Oct '26

Use P-TPS-Owner header hing, when unmaks miss happens. Always remove all P-TPS-Owner headers upon topos receiving a message, to avoid spoofing.   #### Pre-Submission Checklist    - [X] Commit message has the format required by CONTRIBUTING guide - [X] Commits are split per component (core, individual modules, libs, utils, ...) - [X] Each component has a single commit (if not, squash them into one commit) - [X] LLM/AI Coding Assistants were involved in creating the C code or other included content - [X] Code is formatted with `clang-format` using the config file `.clang-format` from source code folder - [X] No commits to README files for modules (changes must be done to docbook files in `doc/` subfolder, the README file is autogenerated) #### Type Of Change - [ ] Small bug fix (non-breaking change which fixes an issue) - [ ] New feature (non-breaking change which adds new functionality) - [ ] Breaking change (fix or feature that would change existing functionality) #### Checklist:  - [ ] PR should be backported to stable branches - [X] Tested changes locally - [ ] Related to issue #XXXX (replace XXXX with an open issue number) #### Description  When topos used with topos_htable storage, races can happen between: SIP ACK of initial INVITE, roamed to different DMQ cluster node vs KDMQ sync of topos_dialog => sometimes SIP ACK can be faster and topos on the roamed node has no idea where to forward to. Fix this problem by always forwarding to the "owner", that already has the routing info (the one that sent the KDMQ sync for topos_dialog) You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4813 -- Commit Summary -- * topos: add handle_unmask_miss modparam -- File Changes -- M src/modules/topos/doc/topos_admin.xml (52) M src/modules/topos/topos_mod.c (2) M src/modules/topos/tps_msg.c (80) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4813.patch https://github.com/kamailio/kamailio/pull/4813.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4813 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4813(a)github.com>

4 17

git:master:f5561aa3: modules: readme files regenerated - pipelimit ... [skip ci]
by Kamailio Dev 08 Oct '26

08 Oct '26

Module: kamailio Branch: master Commit: f5561aa35928e04e8ddc2f853dba4a040a51ab05 URL: https://github.com/kamailio/kamailio/commit/f5561aa35928e04e8ddc2f853dba4a0… Author: Kamailio Dev <kamailio.dev(a)kamailio.org> Committer: Kamailio Dev <kamailio.dev(a)kamailio.org> Date: 2026-10-08T08:46:09+02:00 modules: readme files regenerated - pipelimit ... [skip ci] --- Modified: src/modules/pipelimit/README --- Diff: https://github.com/kamailio/kamailio/commit/f5561aa35928e04e8ddc2f853dba4a0… Patch: https://github.com/kamailio/kamailio/commit/f5561aa35928e04e8ddc2f853dba4a0… --- diff --git a/src/modules/pipelimit/README b/src/modules/pipelimit/README index 99ce78671f7..26a7b90471d 100644 --- a/src/modules/pipelimit/README +++ b/src/modules/pipelimit/README @@ -160,13 +160,14 @@ Chapter 1. Admin Guide The Random Early Detection Algorithm tries to circumvent the synchronization problem imposed by the tail drop algorithm by measuring - the average load and adapting the drop rate dynamically. When running - with the RED algorithm (enabled by default) Kamailio will return errors - to the Kamailio routing engine every n'th packet trying to evenly - spread the measured load of the last timer interval onto the current - interval. As a negative side effect Kamailio might drop messages - although the limit might not be reached within the interval. Decrease - the timer interval if you encounter this. + the average CPU load and adapting the drop rate dynamically. When + running with the RED algorithm (enabled by default) Kamailio will + return errors to the Kamailio routing engine every n'th packet trying + to evenly spread the measured load of the last timer interval onto the + current interval. As a negative side effect Kamailio might drop + messages although the limit might not be reached within the interval. + Decrease the timer interval if you encounter this. It requires + load_fetch bit 1 set (value 1). Network Algorithm (NETWORK) @@ -174,7 +175,8 @@ Chapter 1. Admin Guide The total amount of bytes waiting to be consumed on all the network interfaces is retrieved once every timer_interval seconds. If the returned amount exceeds the limit specified in the modparam, pl_check() - returns false (negative value). + returns false (negative value). It requires load_fetch bit 2 set (value + 2). Feedback Algorithm (FEEDBACK) @@ -325,14 +327,16 @@ modparam("pipelimit", "timer_mode", 1) 4.9. load_fetch (integer) Control if the module should fetch the load for CPU and network - traffic. If set to 0, no load is fetched; if set to 1, the load is - fetched every timer interval. + traffic. If set to 0, no load is fetched; if bit 1 is set to 1 (value + 1), the CPU load is fetched every timer interval; if bit 2 is set to 1 + (value 2), the network load is fetched every timer interval. It can be + set to value 3 to fetch both CPU and network loads. - Default value is 1. + Default value is 0. Example 1.9. Set load_fetch parameter ... -modparam("pipelimit", "load_fetch", 0) +modparam("pipelimit", "load_fetch", 3) ... 4.10. reply_code (integer)

2 1

git:master:2685d7a9: pipelimit: docs updated for load fetch
by Daniel-Constantin Mierla 08 Oct '26

08 Oct '26

Module: kamailio Branch: master Commit: 2685d7a99f2bbc538e1de5794eb430be9b1e489f URL: https://github.com/kamailio/kamailio/commit/2685d7a99f2bbc538e1de5794eb430b… Author: Daniel-Constantin Mierla <miconda(a)gmail.com> Committer: Daniel-Constantin Mierla <miconda(a)gmail.com> Date: 2026-10-08T08:44:21+02:00 pipelimit: docs updated for load fetch --- Modified: src/modules/pipelimit/doc/pipelimit_admin.xml --- Diff: https://github.com/kamailio/kamailio/commit/2685d7a99f2bbc538e1de5794eb430b… Patch: https://github.com/kamailio/kamailio/commit/2685d7a99f2bbc538e1de5794eb430b… --- diff --git a/src/modules/pipelimit/doc/pipelimit_admin.xml b/src/modules/pipelimit/doc/pipelimit_admin.xml index 0ba56679d28..f8051c965ca 100644 --- a/src/modules/pipelimit/doc/pipelimit_admin.xml +++ b/src/modules/pipelimit/doc/pipelimit_admin.xml @@ -56,13 +56,14 @@ <para> The Random Early Detection Algorithm tries to circumvent the synchronization problem imposed by the tail drop algorithm by measuring - the average load and adapting the drop rate dynamically. When running + the average CPU load and adapting the drop rate dynamically. When running with the RED algorithm (enabled by default) Kamailio will return errors to the Kamailio routing engine every n'th packet trying to evenly spread the measured load of the last timer interval onto the current interval. As a negative side effect Kamailio might drop messages although the limit might not be reached within the interval. Decrease - the timer interval if you encounter this. + the timer interval if you encounter this. It requires load_fetch bit + 1 set (value 1). </para> <para> <emphasis>Network Algorithm (NETWORK)</emphasis> @@ -72,7 +73,8 @@ The total amount of bytes waiting to be consumed on all the network interfaces is retrieved once every timer_interval seconds. If the returned amount exceeds the limit specified in the modparam, pl_check() - returns false (negative value). + returns false (negative value). It requires load_fetch bit + 2 set (value 2). </para> <para> <emphasis>Feedback Algorithm (FEEDBACK)</emphasis> @@ -302,19 +304,21 @@ modparam("pipelimit", "timer_mode", 1) <title><varname>load_fetch</varname> (integer)</title> <para> Control if the module should fetch the load for CPU and network traffic. - If set to 0, no load is fetched; if set to 1, the load is fetched every - timer interval. + If set to 0, no load is fetched; if bit 1 is set to 1 (value 1), the CPU + load is fetched every timer interval; if bit 2 is set to 1 (value 2), + the network load is fetched every timer interval. It can be set to value + 3 to fetch both CPU and network loads. </para> <para> <emphasis> - Default value is 1. + Default value is 0. </emphasis> </para> <example> <title>Set <varname>load_fetch</varname> parameter</title> <programlisting format="linespecific"> ... -modparam("pipelimit", "load_fetch", 0) +modparam("pipelimit", "load_fetch", 3) ... </programlisting> </example>

1 0

git:master:c39c580f: pipelimit: flags to control fetching CPU or NETWORK loads
by Daniel-Constantin Mierla 08 Oct '26

08 Oct '26

Module: kamailio Branch: master Commit: c39c580f86094dcf169ec58ea875ef3bf2e4d291 URL: https://github.com/kamailio/kamailio/commit/c39c580f86094dcf169ec58ea875ef3… Author: Daniel-Constantin Mierla <miconda(a)gmail.com> Committer: Daniel-Constantin Mierla <miconda(a)gmail.com> Date: 2026-10-08T08:42:34+02:00 pipelimit: flags to control fetching CPU or NETWORK loads - fetch_load is now default 0 (no load fetched) --- Modified: src/modules/pipelimit/pipelimit.c Modified: src/modules/pipelimit/pl_ht.h --- Diff: https://github.com/kamailio/kamailio/commit/c39c580f86094dcf169ec58ea875ef3… Patch: https://github.com/kamailio/kamailio/commit/c39c580f86094dcf169ec58ea875ef3… --- diff --git a/src/modules/pipelimit/pipelimit.c b/src/modules/pipelimit/pipelimit.c index 2b4539f48dc..0562252323f 100644 --- a/src/modules/pipelimit/pipelimit.c +++ b/src/modules/pipelimit/pipelimit.c @@ -122,7 +122,7 @@ int pl_clean_unused = 0; /* === */ -static int pl_load_fetch = 1; +static int pl_load_fetch = 0; /** module functions */ static int mod_init(void); @@ -640,16 +640,25 @@ static int pipe_push_direct(pl_pipe_t *pipe) ret = (pipe->counter <= pipe->limit) ? 1 : -1; break; case PIPE_ALGO_RED: - if(pipe->load == 0) + if(pl_load_fetch & PL_LOAD_FETCH_CPU) { + LM_ERR("RED algorithm requires load_fetch bit 1 set\n"); + return -1; + } + if(pipe->load == 0) { ret = 1; - else + } else { ret = (!(pipe->counter % pipe->load)) ? 1 : -1; + } break; case PIPE_ALGO_FEEDBACK: hash_idx = ((unsigned int)pipe->counter) % 100; ret = (hash[hash_idx] < *drop_rate) ? -1 : 1; break; case PIPE_ALGO_NETWORK: + if(pl_load_fetch & PL_LOAD_FETCH_NETWORK) { + LM_ERR("network algorithm requires load_fetch bit 2 set\n"); + return -1; + } ret = -1 * pipe->load; break; default: @@ -823,13 +832,15 @@ static int w_pl_active(sip_msg_t *msg, char *p1, char *p2) static void pl_timer_refresh(void) { - if(pl_load_fetch != 0) { + if(pl_load_fetch & PL_LOAD_FETCH_CPU) { switch(*load_source) { case LOAD_SOURCE_CPU: update_cpu_load(); break; } + } + if(pl_load_fetch & PL_LOAD_FETCH_NETWORK) { *network_load_value = get_total_bytes_waiting(); } diff --git a/src/modules/pipelimit/pl_ht.h b/src/modules/pipelimit/pl_ht.h index cf88c9d7afb..06c7f1ca287 100644 --- a/src/modules/pipelimit/pl_ht.h +++ b/src/modules/pipelimit/pl_ht.h @@ -92,6 +92,9 @@ enum PIPE_ALGO_NETWORK }; +#define PL_LOAD_FETCH_CPU 1 +#define PL_LOAD_FETCH_NETWORK 2 + typedef struct str_map { str str;

1 0

[kamailio/kamailio] Issue bug CPS pipelimit (Issue #4985)
by joseaponted 08 Oct '26

08 Oct '26

joseaponted created an issue (kamailio/kamailio#4985) ### Description pipelimit keeps calling `get_total_bytes_waiting()` (reading `/proc/net/udp` and `/proc/net/tcp`) from the main fast timer even with `load_fetch=0` and `timer_mode=1`. Only TAILDROP pipes are used, so network load is not needed. On a host that also runs RTPengine (tens of thousands of UDP sockets), each read of `/proc/net/udp` takes ~20 s, which blocks the main `timer` process at 100% CPU (almost all %system). As a consequence: - pipelimit counters are reset every ~20 s instead of every `timer_interval` (1 s), so TAILDROP limits effectively apply per ~20 s window. - tm timers are delayed too: `tm.stats` shows `waiting` transactions growing continuously (22k -> 31k in a few minutes) because wait timers don't fire on time. Commit 61d1fbcc8c ("pipelimit: allow to disable fetching cpu and network load") suggests `load_fetch=0` should prevent this. ### Troubleshooting #### Reproduction 1. Run Kamailio with pipelimit on a host with many UDP sockets (e.g. RTPengine on the same host handling thousands of calls). 2. Use only TAILDROP pipes, with: modparam("pipelimit", "timer_interval", 1) modparam("pipelimit", "timer_mode", 1) modparam("pipelimit", "load_fetch", 0) 3. Watch `kamcmd pl.list`: `last_counter` changes every ~20 s, not every 1 s. 4. Check the main timer process: 100% CPU, with `/proc/<pid>/net/udp` open. #### Debugging Data Main timer process (`core.ps` -> "timer") at 100% CPU, ~0% usr / ~100% sys: pidstat -t -u -p <timer_pid> 1 5 |__kamailio 0.00 %usr 100.00 %system Open file descriptor: ls -l /proc/<timer_pid>/fd | grep net 19 -> /proc/<timer_pid>/net/udp perf top -p <timer_pid>: 70.15% [kernel] [k] udp_get_next 16.11% [kernel] [k] udp_get_first 1.92% [kernel] [k] udp_seq_start perf record --call-graph dwarf -p <timer_pid>: udp_get_next udp_seq_start seq_read_iter / seq_read / proc_reg_read / vfs_read / read get_total_bytes_waiting 0x7443089c0066 0x7443089c0a5c timer_main main_loop main Address resolution via /proc/<timer_pid>/maps: 7443089b3000-7443089c8000 r-xp 00003000 ... /usr/lib/x86_64-linux-gnu/kamailio/modules/pipelimit.so offset in file: 0x10066 So the call comes from pipelimit.so, from `timer_main` (main fast timer), despite `timer_mode=1` and `load_fetch=0`. #### Log Messages No errors logged. #### SIP Traffic Not relevant; occurs under normal load (~hundreds of CPS per pipe, thousands of concurrent calls with RTPengine on the same host). ### Possible Solutions - Skip `get_total_bytes_waiting()` entirely when `load_fetch=0`. - Ideally, only fetch network load if at least one NETWORK pipe exists. - Also verify that `timer_mode=1` actually moves the routine off the main timer. Workaround used: removed pipelimit and implemented per-second CPS limiting with htable (`$shtinc` on a key built with `$Ts`). ### Additional Information * **Kamailio Version** - output of `kamailio -v` version: kamailio 6.0.8 (x86_64/linux) flags: USE_TCP, USE_TLS, USE_SCTP, TLS_HOOKS, USE_RAW_SOCKS, DISABLE_NAGLE, USE_MCAST, DNS_IP_HACK, SHM_MMAP, PKG_MALLOC, MEM_JOIN_FREE, Q_MALLOC, F_MALLOC, TLSF_MALLOC, DBG_SR_MEMORY, USE_FUTEX, FAST_LOCK-ADAPTIVE_WAIT, USE_DNS_CACHE, USE_DNS_FAILOVER, USE_NAPTR, USE_DST_BLOCKLIST, HAVE_RESOLV_RES, TLS_PTHREAD_MUTEX_SHARED compiled with gcc 11.4.0 * **Operating System**: Ubuntu 22.04.5 LTS, kernel 6.8.0-138-generic, 32 CPUs RTPengine running on the same host -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/issues/4985 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/issues/4985(a)github.com>

2 1

[kamailio/kamailio] Add read-only variables to expose the currently set transaction branch, failure, onreply, and branch failure routes. (Issue #4986)
by Ben Kaufman 08 Oct '26

08 Oct '26

whosgonna created an issue (kamailio/kamailio#4986) ### Description Add the following read-only pseudo-variables to expose the currently set branch_route, onreply_route, failure_route, and branch_failure_route ### Expected behavior Implement these within the `$T()` variables exported by the TMX module. ``` ## Set transaction routes: t_on_branch("MANAGE_BRANCH"); t_on_failure("MANAGE_FAILURE"); t_on_reply("MANAGE_REPLY"); t_on_branch_failure("MANAGE_BRANCH_FAILURE"); ## Retrieve branch route: xinfo("Branch route is $T(branch_route)\n"); xinfo("Failure route is $T(failure_route)\n"); xinfo("Onreply route is $T(onreply_route)\n"); xinfo("Branch failure route is $T(branch_failure_route)\n"); ``` Will print to the log: > Branch route is MANAGE_BRANCH > Failure route is MANAGE_FAILURE > Onreply route is MANAGE_REPLY > Branch failure route is MANAGE_BRANCH_FAILURE -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/issues/4986 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/issues/4986(a)github.com>

1 0

[kamailio/kamailio] Parsing enhancments (PR #4917)
by David Trihy 07 Oct '26

07 Oct '26

#### Pre-Submission Checklist    - [x] Commit message has the format required by CONTRIBUTING guide - [x] Commits are split per component (core, individual modules, libs, utils, ...) - [x] Each component has a single commit (if not, squash them into one commit) - [x] LLM/AI Coding Assistants were involved in creating the C code or other included content - [x] Code is formatted with `clang-format` using the config file `.clang-format` from source code folder - [x] No commits to README files for modules (changes must be done to docbook files in `doc/` subfolder, the README file is autogenerated) #### Type Of Change - [ ] Small bug fix (non-breaking change which fixes an issue) - [ ] New feature (non-breaking change which adds new functionality) - [x] Breaking change (fix or feature that would change existing functionality) #### Checklist:  - [x] PR should be backported to stable branches - [ ] Tested changes locally - [ ] Related to issue #XXXX (replace XXXX with an open issue number) #### Description This PR is to make the sip parsing more deterministic when in both strict and non strict mode. **Strict mode changes** When in strict parsing mode ie. `sip_parser_mode=1` when receiving a new message the Kamailio core will test for what it calls the transaction quadruple, which is the `To`, `From`, `Call-ID`, `CSeq` and fail to enter the route if the SIP messages does not contain those headers. The parsing of the message is deliberately lazy in nature as opposed to eager parsing but it can leave the core in an odd state when strict parsing is set on. Example invalid message which with strict parsing on will fail to enter the route due to a duplicate quadruple(To in this case) ``` INVITE sip:user@domain.com SIP 2.0 To: <valid nameaddr> To: <valid nameaddr> From: <valid nameaddr> Call-ID: valid-callid CSeq: INVITE 1 X-Custom-Header: value .... Content-Length: 0 ``` Example of a request which will enter the route but result in undefined behaviour ``` INVITE sip:user@domain.com SIP 2.0 To: <valid nameaddr> From: <valid nameaddr> Call-ID: valid-callid CSeq: INVITE 1 X-Custom-Header: value To: <valid nameaddr> .... Content-Length: 0 ``` The reason why this happens is because the second To header is not parsed as the check_required_headers keeps going until it finds an instance of the quadruple and it will enter the route and depending on how you are using the route you could fail at another point where the rest of the headers would be parsed or when construct the SIP reply or request once the route finishes, given the expectation would be that strict parsing would fail straight away but in this case it would not. The change turns the strict parsing mode into an eager parser so it fails early, this shouldn't meaningfully change the behaviour of most configuration but it does in the case outlined above so it technically creates a backwards incompatible change. **Non strict mode changes** When the `sip_parser_mode=0` and there is a duplicate check it will only copy the first valid instance of the quadruple in the reply, the current behaviour is such that when constructing a reply eg. `400 Bad Request` both To headers are copied, this change ensures only one of them is copied which is the first instance of that header during parsing. One other change is when in this mode if there is a message such as this ``` INVITE sip:user@domain.com SIP 2.0 To: <invalid nameaddr> To: <valid nameaddr> From: <valid nameaddr> Call-ID: valid-callid CSeq: INVITE 1 X-Custom-Header: value .... Content-Length: 0 ``` It will still parse the quadruple but will take into account the valid `To` header only which in this case is the second instance of the header but the first instance is tracked as an error so that when calling `sanity_check` in the `sanity` module it can behave correctly if the check for required headers is set in the modparam. The reason for fixing this case is that if the request contains at least one valid instance of the quadruple then it can be used to generate a reply. In the above message case the core would drop the request. However the below example would not drop and then the sanity check would fail at this point and due to how the check required headers works it means we would not be able to reply but we would do the sanity check. ``` INVITE sip:user@domain.com SIP 2.0 To: <valid nameaddr> From: <valid nameaddr> Call-ID: valid-callid CSeq: INVITE 1 X-Custom-Header: value To: <invalid nameaddr> .... Content-Length: 0 ``` This just once again tightens up the behaviour so that is deterministic and the script writer can reasonably infer how the parsing will behave based on their requirements. You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4917 -- Commit Summary -- * core: skip bad headers, copy one SIP quad when sip_parser_mode=0 * pua: pass NULL next to get_hdr_field() * sipcapture: check get_hdr_field() return * sanity: reply 400 on bad or duplicate SIP quad, check errored headers -- File Changes -- M src/core/msg_translator.c (35) M src/core/parser/msg_parser.c (129) M src/core/parser/msg_parser.h (5) M src/core/parser/parse_addr_spec.c (1) M src/core/parser/parse_to.c (1) M src/core/parser/parse_via.c (20) M src/modules/pua/send_subscribe.c (7) M src/modules/sanity/sanity.c (25) M src/modules/sipcapture/sipcapture.c (4) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4917.patch https://github.com/kamailio/kamailio/pull/4917.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4917 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4917(a)github.com>

2 4

git:master:e846c113: modules: readme files regenerated - uuid ... [skip ci]
by Kamailio Dev 07 Oct '26

07 Oct '26

Module: kamailio Branch: master Commit: e846c11384aa86bf1bad208ec3b8dc0b3e50c913 URL: https://github.com/kamailio/kamailio/commit/e846c11384aa86bf1bad208ec3b8dc0… Author: Kamailio Dev <kamailio.dev(a)kamailio.org> Committer: Kamailio Dev <kamailio.dev(a)kamailio.org> Date: 2026-10-07T16:31:08+02:00 modules: readme files regenerated - uuid ... [skip ci] --- Modified: src/modules/uuid/README --- Diff: https://github.com/kamailio/kamailio/commit/e846c11384aa86bf1bad208ec3b8dc0… Patch: https://github.com/kamailio/kamailio/commit/e846c11384aa86bf1bad208ec3b8dc0… --- diff --git a/src/modules/uuid/README b/src/modules/uuid/README index 3034dab7003..eac4faefd74 100644 --- a/src/modules/uuid/README +++ b/src/modules/uuid/README @@ -75,6 +75,10 @@ Chapter 1. Admin Guide If the generation was not safe, the variable returns $null. The function is not available on Mac OS X - the value is the one returned by uuid_generate_time(). + * $uuid(7) - return a UUID version 7 value (RFC 9562). The value is + generated with uuid_generate_time_v7(), which requires libuuid from + util-linux 2.41 or newer. If the module was built against an older + libuuid, the variable returns $null and an error is logged. Example 1.1. $uuid(type) usage ...

1 0