August 2026 - sr-dev - kamailio.org

[kamailio/kamailio] presence: use same logic to calculate header length (PR #4749)
by Victor Seva 09 Sep '26

09 Sep '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] 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 - [x] 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:  - [x] PR should be backported to stable branches - [ ] Tested changes locally - [ ] Related to [2966](https://github.com/kamailio/kamailio/security/code-scanning/2966) #### Description  You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4749 -- Commit Summary -- * presence: use same logic to calculate header length -- File Changes -- M src/modules/presence/notify.c (3) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4749.patchhttps://github.com/kama… -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4749 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4749(a)github.com>

1 3

[kamailio/kamailio] presence: preserve etag uniqueness when marking presentity for removal (PR #4275)
by Alessio Garzi 09 Sep '26

09 Sep '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] 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 - [x] 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: - [x] PR should be backported to stable branches - [x] Tested changes locally - [ ] Related to issue #### Description When the `presence` module attempts to clean up records in the `presentity` table, it replaces the `etag` with a static, hardcoded string: ``` "*#-OFFLINE-#*" ``` However, the `presentity` table enforces a uniqueness constraint on the following combination of fields: ``` MariaDB [kamailio]> show create table presentity; | presentity | CREATE TABLE `presentity` ( ... UNIQUE KEY `presentity_idx` (`username`,`domain`,`event`,`etag`), ``` This approach breaks the uniqueness of etag, which is intended to differentiate records , especially in cases such as parallel calls, or when using kamailio as a forking proxy. Instead of overwriting the etag with the static offline marker, this patch appends the marker to the existing etag value. This ensures the offline flag is retained while preserving the uniqueness of the etag. The deletion query was also updated accordingly: it now identifies offline records using a LIKE '%*#-OFFLINE-#*' clause. Based on testing, the duplicate key errors no longer occur, and the cleanup logic now behaves as expected. This patch improves the reliability of presentity cleanup without requiring schema or index changes. Please consider merging into master. Thanks! You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4275 -- Commit Summary -- * presence: preserve etag uniqueness when marking presentity for removal -- File Changes -- M src/modules/presence/presentity.c (90) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4275.patch https://github.com/kamailio/kamailio/pull/4275.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4275 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4275(a)github.com>

3 15

[kamailio/kamailio] dialog: add DMQ load callbacks option (PR #4885)
by Simone de Blasiis 09 Sep '26

09 Sep '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 - [x] Related to issue #3254 #### Description  ## Objective Address #3254, where active calls replicated through DMQ may not produce a CDR after failover because the new active node has no accounting callbacks attached to the replicated dialogs. This change adds the optional `enable_dmq_load_callbacks` dialog parameter. When enabled, dialogs created from DMQ replication run `DLGCB_LOADED` callbacks and synchronize dialog variables in state updates, allowing modules such as `acc` to restore the callbacks and state required for CDR generation. The parameter defaults to `0`, preserving the dialog module’s existing behavior unless explicitly enabled. A startup warning and documentation describe the known impact on modules that register `DLGCB_LOADED`. You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4885 -- Commit Summary -- * dialog: add DMQ load callbacks option -- File Changes -- M src/modules/dialog/dialog.c (36) M src/modules/dialog/dlg_dmq.c (27) M src/modules/dialog/doc/dialog.xml (5) M src/modules/dialog/doc/dialog_admin.xml (73) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4885.patch https://github.com/kamailio/kamailio/pull/4885.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4885 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4885(a)github.com>

2 4

[kamailio/kamailio] Extended HAproxy portocol support (PR #4746)
by sergey-safarov 09 Sep '26

09 Sep '26

#### Pre-Submission Checklist - [ ] 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] 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:  - [ ] PR should be backported to stable branches - [x] Tested changes locally - [ ] Related to issue #xxxx #### Description In PR extended HAproxy protocol support. For server and client IP addresses used values from HAproxy header. Via headers generated by the values from Haproxy protol. Adjusted `path`, `rr`, `siptrace` module to use IP address values from HAproxy protocol. For PR recommended review by commits. You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4746 -- Commit Summary -- * core: extended haproxy protocol support * core: search connections by haproxy headers * core: set Via header onto haproxy address if haproxy is used * core: socket may not match when haproxy used * rr: reuse Haproxy connection if available * websocket: extendented haproxy protocol support * siptrace: extended haproxy support -- File Changes -- M src/core/msg_translator.c (88) M src/core/parser/msg_parser.c (4) M src/core/parser/msg_parser.h (2) M src/core/receive.c (12) M src/core/receive.h (4) M src/core/tcp_conn.h (1) M src/core/tcp_main.c (193) M src/core/tcp_read.c (2) M src/modules/rr/loose.c (32) M src/modules/siptrace/siptrace.c (251) M src/modules/websocket/ws_frame.c (15) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4746.patchhttps://github.com/kama… -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4746 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4746(a)github.com>

2 8

[kamailio/kamailio] dialog: add 1 new modparam and 2 new cfg functions (PR #4850)
by Stefan Mititelu 09 Sep '26

09 Sep '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  All can be useful in an active-active dialog DMQ setup. You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4850 -- Commit Summary -- * dialog: add enable_dmq_ka_iflags_sync modparam * dialog: add new cfg functions for dialog legs data -- File Changes -- M src/modules/dialog/dialog.c (113) M src/modules/dialog/dlg_dmq.c (19) M src/modules/dialog/doc/dialog_admin.xml (80) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4850.patch https://github.com/kamailio/kamailio/pull/4850.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4850 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4850(a)github.com>

2 2

[kamailio/kamailio] topos_htable: escape the field separator in the stored records (PR #4896)
by Kaixuan Li 09 Sep '26

09 Sep '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 - [x] 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: - [x] PR should be backported to stable branches - [x] Tested changes locally - [ ] Related to issue #XXXX (replace XXXX with an open issue number) #### Description The `topos_htable` records are built by joining the field values with `|` and are parsed back by position with `strsep()`, so a `|` inside a field value taken from a header shifts every following field of the record. A `|` in the Call-ID is enough to trigger it. The stored dialog record ``` 1787806973|probe|PIPE|callid|atpsh-...|btpsh-...|0|SIP/2.0/UDP 127.0.0.1:5080;... ``` has the Call-ID occupying three fields, so on the reply `topos` loads a record where `a_uuid` is something else, does not restore the Via set, and `tps_prepare_msg()` then fails for the outgoing message: ``` ERROR: topos [topos_mod.c:528]: tps_prepare_msg(): mandatory headers missing - via1: (nil) callid: 0x... ``` At that point `tps_msg_sent()` goes to `done:` without touching `obuf`, so the core sends the reply as it arrived, with topology hiding off — the caller gets the callee's real address instead of a masked one. This is specific to the htable backend: the db backend stores each field in its own column and reads it back through `RES_ROWS(...)[0].values[_c]`, so there is no separator to inject there. ##### Changes - escape the separator (and the escape character itself) when building the dialog, branch and initial method branch records - parse the records back with a scanner that skips the escaped separators and unescapes each field in place, instead of `strsep()` - apply the same escaping to the key of the initial method branch record, where a `|` in the Call-ID or in the To-tag could otherwise build the key of a different dialog (`callid=a|b, tag=c` and `callid=a, tag=b|c` gave the same key) Field values without `|` or `\` are stored unchanged, so the record format stays byte-identical to the one written by the previous versions, and the bounds of the existing buffers are still enforced (a value that no longer fits makes the insert fail instead of being silently truncated). `topos_htable_base64=1` was already unaffected, because base64url values contain neither the separator nor the escape character; the new escaping is a no-op there. ##### Testing Three dialogs back to back through one proxy, same UAS, one kamailio instance, everything default except `storage=htable` and `contact_host`. The only variable is the Call-ID: before: ``` clean Call-ID -> MASKED Contact seen by UAC: <sip:atpsh-6a9142b2-3ba5a1-1@127.0.0.1> Call-ID with '|' -> *** LEAKED *** Contact seen by UAC: <sip:bob@127.0.0.1:5070> clean Call-ID (2) -> MASKED Contact seen by UAC: <sip:atpsh-6a9142b2-3ba5a1-3@127.0.0.1> ``` after: ``` clean Call-ID -> MASKED Contact seen by UAC: <sip:atpsh-6a9142c3-3ba789-1@127.0.0.1> Call-ID with '|' -> MASKED Contact seen by UAC: <sip:atpsh-6a9142c3-3ba789-2@127.0.0.1> clean Call-ID (2) -> MASKED Contact seen by UAC: <sip:atpsh-6a9142c3-3ba789-3@127.0.0.1> ``` A full INVITE / 200 / ACK / BYE / 200 flow was also run for ten Call-IDs (clean, single `|`, 23 `|`, backslash, `a\|b|c`, leading / trailing / only separators, trailing backslash), checking that the caller never sees the real UAS address and that the Call-ID is restored on both legs for the in-dialog requests and for the BYE reply. All pass with the patch, and again with `topos_htable_base64=1`; 6 of the 10 leak without it. The record builder and the record scanner were also exercised standalone under ASan/UBSan for round-trip equality, empty fields, separator-only fields, key distinctness and oversized values. You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4896 -- Commit Summary -- * topos_htable: escape the field separator in the stored records -- File Changes -- M src/modules/topos_htable/topos_htable_storage.c (332) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4896.patch https://github.com/kamailio/kamailio/pull/4896.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4896 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4896(a)github.com>

4 11

[kamailio/kamailio] topos: document contact_mode 3 A/B mapping and host xavu params (PR #4858)
by li xuanqun 09 Sep '26

09 Sep '26

Document the previously missing xavu_field_a/b_contact_host module parameters and clarify that contact_mode 3 maps A/B opposite to mode 2 (mode 2: a=downstream/b=upstream; mode 3: a=upstream/b=downstream). Also clarify the get_xavu_host() comment to record this intentionally different mapping. Closes: #4725   #### Pre-Submission Checklist    - [ ] Commit message has the format required by CONTRIBUTING guide - [ ] Commits are split per component (core, individual modules, libs, utils, ...) - [ ] Each component has a single commit (if not, squash them into one commit) - [ ] LLM/AI Coding Assistants were involved in creating the C code or other included content - [ ] Code is formatted with `clang-format` using the config file `.clang-format` from source code folder - [ ] 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 - [ ] Tested changes locally - [ ] Related to issue #XXXX (replace XXXX with an open issue number) #### Description  You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4858 -- Commit Summary -- * topos: document contact_mode 3 A/B mapping and host xavu params -- File Changes -- M src/modules/topos/doc/topos_admin.xml (86) M src/modules/topos/tps_storage.c (8) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4858.patch https://github.com/kamailio/kamailio/pull/4858.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4858 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4858(a)github.com>

3 3

[kamailio/kamailio] ldap: improve parameter parsing in ldap_params_search (PR #4899)
by AlexisHadj 09 Sep '26

09 Sep '26

**The issue:** The ldap module passes the LDAP search filter to vsnprintf() as the format string. This potentially allows an unauthenticated attacker to trigger crashes or information disclosure by supplying format specifiers (such as %n or %s) via SIP-derived data (e.g., Request-URI user parts). **The fix:** Fix a format-string vulnerability in ldap_params_search where the LDAP search filter was passed directly to vsnprintf() as a format string by ensuring that search filters are treated as literal string parameters. You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4899 -- Commit Summary -- * ldap: improve parameter parsing in ldap_params_search -- File Changes -- M src/modules/ldap/ldap_api_fn.c (2) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4899.patch https://github.com/kamailio/kamailio/pull/4899.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4899 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4899(a)github.com>

3 5

[kamailio/kamailio] smsops: fix GSM 7-bit alphabet conversion tables (PR #4883)
by 0xabe-io 09 Sep '26

09 Sep '26

The two conversion tables were not inverses of each other, and neither matched the default alphabet and extension table in 3GPP TS 23.038. Twenty-one code points were wrong. Transcription errors: - gsm7bit_codes mapped 0x2D to 0xAD (SOFT HYPHEN) rather than 0x2D (HYPHEN-MINUS), so a hyphen was corrupted on decode - gsm7bit_ext_codes carried VERTICAL LINE at 0x48, which is not an extension code, leaving 0x40 unmapped even though ascii2gsm7bit_ext_codes escapes '|' to it - decoding one emitted a bare 0xFF - ascii2gsm7bit_codes mapped CURRENCY SIGN to 0x23 (NUMBER SIGN) rather than 0x24 Decode entries that contradicted the encode table: - gsm7bit_codes returned an unaccented ASCII letter for eighteen code points - 0x04 as 'e', 0x5B as 'A', 0x7F as 'a' and so on - while ascii2gsm7bit_codes maps the ISO-8859-1 byte for that same character onto that same code point. The folding was not applied consistently either: 0x01, 0x03, 0x06, 0x07, 0x1C, 0x1D, 0x1E, 0x24, 0x40, 0x5F and 0x60 already returned their ISO-8859-1 bytes. decode(encode(x)) == x now holds for every character both tables cover.   #### 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 - [x] 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: - [x] PR should be backported to stable branches - [x] Tested changes locally - [ ] Related to issue #XXXX (replace XXXX with an open issue number) #### Description `smsops` has two GSM 7-bit conversion tables, one per direction. They are not inverses of each other, and neither matches the default alphabet and extension table in 3GPP TS 23.038. Twenty-one code points are wrong. ##### Three transcription errors | table | index | was | is | | | --- | --- | --- | --- | --- | | `gsm7bit_codes` | `0x2D` | `0xAD` | `0x2D` | `0x2D` is HYPHEN-MINUS; `0xAD` is SOFT HYPHEN | | `gsm7bit_ext_codes` | `0x40` / `0x48` | unmapped / `0x7C` | `0x7C` / unmapped | VERTICAL LINE is extension code `0x40`; `0x48` is not an extension code | | `ascii2gsm7bit_codes` | `0xA4` | `0x23` | `0x24` | `0x24` is CURRENCY SIGN; `0x23` is NUMBER SIGN | The first two lose messages. `gsm_to_ascii()` writes the table value straight into the output with no validation, so a decoded hyphen becomes a bare `0xAD` and a decoded `|` a bare `0xFF`. Neither is valid UTF-8, so a configuration storing `$rpdata` / `$tpdu(payload)` in a UTF-8 column drops the message: ``` db_mysql_submit_query(): driver error on query: Incorrect string value: '\xADsim 3...' for column `smsc`.`messages`.`text` at row 1 (1366) ``` By then the sender has had `202 Accepted` and an RP-ACK built from the same successfully-parsed RP-DATA, so from the handset the SMS was delivered. Nothing reports an error anywhere. The hyphen is the only affected character in printable ASCII, which is presumably how it survived since f82c4775ef, and only senders using `application/vnd.3gpp.sms` reach the decode table at all. Both misplaced entries look like the same kind of slip: in the 8-wide layout the tables originally had, `0x7C` sits at the start of the row *after* the one it belongs to, and the hyphen sits in an otherwise pure identity run (`( ) * + , - . /`). ##### Eighteen decode entries that contradict the encode table `gsm7bit_codes` returns an unaccented ASCII letter for `0x04` `è`, `0x5B` `Ä`, `0x7F` `à` and fifteen others, while `ascii2gsm7bit_codes` maps the ISO-8859-1 byte for that same character onto that same code point — so a character that survives encoding is altered by decoding. This does not look deliberate: the folding is not applied consistently, and `0x01` `£`, `0x03` `¥`, `0x06` `ù`, `0x07` `ì`, `0x1C` `Æ`, `0x1D` `æ`, `0x1E` `ß`, `0x24` `¤`, `0x40` `¡`, `0x5F` `§` and `0x60` `¿` already return their ISO-8859-1 bytes. **This group is separable** — happy to drop it and land only the three above if you would rather take it on its own. One thing to weigh: it makes eighteen more characters emit ISO-8859-1, so a deployment storing decoded text in a UTF-8 column will start seeing the error above for `é`, `ä`, `à`. That collision is pre-existing and not introduced here — `ù` produces it today, unmodified: ``` Incorrect string value: '\xF9' for column `smsc`.`messages`.`text` ``` Whether `smsops` should emit UTF-8 rather than ISO-8859-1 on this path is a larger question and is not addressed here. `ucs2_to_utf8()` on the DCS≠0 path already emits real UTF-8, so `$tpdu(payload)` is two different encodings depending on the coding scheme; given #3043, #3546 and #3690 that looks like the intended direction, but a `unsigned char[128]` table cannot express it and the encode side takes ISO-8859-1 input. Glad to follow up separately. ##### Testing `make modules modules=src/modules/smsops` builds clean, and `clang-format` 18 (the version CI pins) reports the file unchanged. Verified end to end against a live Kamailio IMS deployment (P/I/S-CSCF plus an SMSC using `smsops`), driving each direction separately so a fault in one cannot mask the other: * **decode** — a UE sends `application/vnd.3gpp.sms` carrying each 7-bit code point as a raw septet, bypassing the alphabet entirely, and the byte the SMSC stored is read back. * **encode** — a known byte is placed in the SMSC's queue and the 7-bit value transmitted is read out of the delivered TPDU. All 128 default-alphabet code points and all 10 extension code points, both directions. Before: 20 code points decode to the wrong byte and 1 byte encodes to the wrong code point. After: none. Nothing else about SMS handling changes — `text/plain` delivery, RP-DATA in both directions, TP-OA TON/NPI, off-net routing and store-and-forward all behave identically. You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4883 -- Commit Summary -- * smsops: fix GSM 7-bit alphabet conversion tables -- File Changes -- M src/modules/smsops/smsops_impl.c (20) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4883.patch https://github.com/kamailio/kamailio/pull/4883.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4883 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4883(a)github.com>

4 6

[kamailio/kamailio] permissions: refcount table objects (PR #4875)
by Richard Fuchs 09 Sep '26

09 Sep '26

- similar approach to 1f1cbcb022 (dispatcher) - all existing table objects are encapsulated in higher-level struct which includes a reference count (e.g. `struct addr_list *` is encapsulated in `struct addr_table`, etc) - reference counts are protected by a mutex (one for `struct trusted_table` and a second one for the tables in `address.c`) - access to all table objects is only allowed through a call to the respective get_*() function, which returns a pointer to the table and increases the reference count - every call to get_*() must be paired with a matching call to put_*() to release the reference - the only write operation to all tables is a complete reload, which rebuilds brand new table objects and the swaps out the global pointers - after swapping out the pointers, the reload process waits until all reference counts to the old table objects drops to zero, and only then frees the old tables - with all this, concurrent read access is possible and safe while allowing for concurrent safe reloads - the standalone empty_*() functions have become obsolete and have been integrated into the free_*() functions - functions in address.c generally need access to all 3 tables, so we use the existing `address_tables_group_t` together with matching get/put functions to facilitate this - the subnet table previously used an extra element to keep track of the number of subnets - this count was integrated into the higher-level struct, therefore obsoleting the need for the extra element - the subnet table has a runtime-configurable (but not runtime-mutable) size: we use the empty-array syntax (`[0]`) in the struct together with an adjusted malloc size for that - all get_*() and put_*() functions follow the same schema, so a generic macro is used for the body of these functions - the cleanup function scheduled at regular intervals is not needed any more, therefore `trusted_cleanup_interval` has been removed - the `reload_delta` option is kept, but now defaults to zero (no limit) as all concurrent reloads and read accesses are safe - the `perm_rpc_reload_time` value is protected by an additional mutex, which is needed to eliminate a small race condition between reading and writing the value #### 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) - [ ] 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 - [X] 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 - [X] Related to issue #4809 #### Description Eliminates race conditions between concurrent reloads and read accesses. The previous double-buffering approach is now obsolete and has been replaced with proper reference-counting You can view, comment on, or merge this pull request online at: https://github.com/kamailio/kamailio/pull/4875 -- Commit Summary -- * permissions: refcount table objects -- File Changes -- M src/modules/permissions/address.c (366) M src/modules/permissions/address.h (45) M src/modules/permissions/doc/permissions_admin.xml (40) M src/modules/permissions/hash.c (236) M src/modules/permissions/hash.h (147) M src/modules/permissions/permissions.c (25) M src/modules/permissions/permissions.h (3) M src/modules/permissions/rpc.c (53) M src/modules/permissions/trusted.c (116) M src/modules/permissions/trusted.h (20) -- Patch Links -- https://github.com/kamailio/kamailio/pull/4875.patch https://github.com/kamailio/kamailio/pull/4875.diff -- Reply to this email directly or view it on GitHub: https://github.com/kamailio/kamailio/pull/4875 You are receiving this because you are subscribed to this thread. Message ID: <kamailio/kamailio/pull/4875(a)github.com>

2 2