summaryrefslogtreecommitdiff
path: root/src (follow)
AgeCommit message (Collapse)Author
12 daysReduce the lifetime of cryptographic materialjvoisin
12 daysFix a possible null-deref in sp_stream_wrapper_registerjvoisin
`protocol_name` can be NULL if `zend_parse_parameters_ex` fails (it uses `ZEND_PARSE_PARAMS_QUIET`), but it was then unconditionally passed to `strcasecmp`.
12 daysFix an spprintf undefined behaviourjvoisin
`getenv("REMOTE_ADDR")` can return NULL, and it is passed directly to `spprintf`. While `spprintf` might handle `NULL` gracefully, it's not always the case.
12 daysFix a memory leak in sp_log_disable/sp_log_disable_retjvoisin
php_raw_url_encode returned a new zend_string, but the old arg_value_dup is never released. Also arg_value_dup was reassigned, leaking the initial zend_string_init allocation.
12 daysAddress multiple sign issues in ifilterjvoisin
`sp_is_dangerous_char[(int)*p]` is indexed by `(int)*p`. If char is signed (default on x86), values 0x80–0xFF produce negative indices into the array, causing an out-of-bounds read. The `sp_server_encode` function has the same issue.
12 daysFix a possible null-pointer dereference in cookies encryptionjvoisin
12 daysFix a type confusion on cookie encryptionjvoisin
Cookies can be arrays (session_id[]=x), and we called `Z_STRLEN_P(pDest)` without checking if `Z_TYPE_P(pDest) == IS_STRING`, leading to a type confusion, leaking at least `HashTable->arData`, and the rest of the code is going to corrupt some stuff, leading to a crash. While exploitation can't be ruled out, it looks stupidly harder. The array will be decoded as base64 into another variable, decrypted, and have this value written back to the array. To obtain a controlled read, an attacker would have to bruteforce decryption to find an encrypted value that could be properly decrypted, as we're using authenticated encryption. The next steps are depending on the heap's layout, which should be pretty deterministic/simple as cookie decryption happens at RINIT. But since the heap overflow is happening in the cookies, odds are that they're stored somewhere with other data like POST/GET/… and thus nothing super-duper juicy. While remote heap exploitation in PHP have been done in the past, this primitive looks a tad too limited to yield something powerful enough to pop a shell. Reported-by: Vozec
2026-03-29Add a test for validate idempotence of (un)serializejvoisin
2026-03-28Fix the usage of strlen() which will return a wrong size when serialized ↵W0rty
objects contains null bytes (for example in private fields)
2026-03-25Add a test for Dateinterval::__constructjvoisin
As it has been privately reported that the rule might not be working, so better safe than sorry. Moreover, we didn't have tests for `__construct`
2026-02-23Simplify formattingChristian Göttsche
The members sid_min_length and sid_max_length are of type unsigned long, thus use %lu instead of %zu and a cast.
2026-02-22Constify function pointer tablesChristian Göttsche
2026-02-22Log session ID lengths on failureChristian Göttsche
2026-02-22Harden against snprintf(3) truncationChristian Göttsche
2026-02-22Drop dead codeChristian Göttsche
Dead since almost 5 years with commit ae4ac9f ("properly free memory on shutdown")
2026-02-22Generalize disabled_functions_regexp_multiple.phpt expected outputChristian Göttsche
On uncommon architectures, like s390x, `-2` instead of `-1` might be printed.
2026-02-22Use correct parser for session ID lengthsChristian Göttsche
Can causes issues on uncommon architectures, like s390x.
2026-02-22Define PATH_MAX and update its usagecgzones
2026-02-15tests: add several skip reasonsChristian Göttsche
Co-authored-by: Julien "jvoisin" Voisin <julien.voisin@dustri.org>
2026-01-25Avoid double checking if function is hookedJakub Onderka
2026-01-07Bump the changelogv0.13.0jvoisin
2025-12-05Fix url_encode cookie flag in php_setcookie callAngus Johnston
setcookie() is always URL encoded, urlencode is only turned off for setrawcookie(). Turning it off breaks cookies that have a value containing certain characters (namely spaces) https://github.com/php/php-src/blob/685e99655ae97c667950f7f7d176985958718f56/ext/standard/head.c#L97
2025-10-25Show the logging file is set in phpinfo()jvoisin
2025-10-24feat(log): add the possibility to log to a filelog2filejvoisin
2025-10-24fix(unserialize): don't bail in simulation mode when there's no HMACjvoisin
2025-10-14Add a testjvoisin
2025-10-02fix(log): systematically drop when .drop() is usedjvoisin
When the `php` logging facility is used, the error could have been caught by using `set_error_handler` and whatnot. This commit ensures that if the `.drop()` option is set, we're calling `zend_bailout()` that can't be caught. An attacker could have used this issue to silently perform some recon of the running environment. This isn't considered a vulnerability as an attacker with arbitrary php code execution can simply use the use-after-free of the day to gain arbitrary (native) code execution anyway, after detecting that Snuffleupagus is in use, to take little risks of detection.
2025-10-02Rename a handful of global constantsjvoisin
2025-10-01Fix a cookie-related warning for PHP8.5.0jvoisin
``` ========DIFF======== 001- OK 001+ Fatal error: Uncaught ValueError: setcookie(): "partitioned" option cannot be used without "secure" option in /builddir/build/BUILD/snuffleupagus-1c7598c432551d0c49c2c57f249ccd5ccabce638/src/tests/samesite_cookies.php:2 002+ Stack trace: 003+ #0 /builddir/build/BUILD/snuffleupagus-1c7598c432551d0c49c2c57f249ccd5ccabce638/src/tests/samesite_cookies.php(2): setcookie('super_cookie', 'super_value') 004+ #1 {main} 005+ thrown in /builddir/build/BUILD/snuffleupagus-1c7598c432551d0c49c2c57f249ccd5ccabce638/src/tests/samesite_cookies.php on line 2 ========DONE======== FAIL Cookie samesite [tests/samesite_cookies.phpt] ``` Even though the warning might be spurious, let's fix this properly, by initialising `partitioned` to false, and by setting it only if `secure` is set as well.
2025-09-01Add support for PHP8.5jvoisin
2025-08-19Bump the changelogv0.12.0jvoisin
2025-08-17Fix a NULL-ptr derefjvoisin
``` Program terminated with signal SIGSEGV, Segmentation fault. 20 if (!(func->common.function_name)) { (gdb) info locals func = 0x0 function_name = 0xffb25f6d0190 "SearchByCallback" complete_path_function = 0xffb26c8a0570 "\240\005\212l\262\377" ``` It seems that in some callback shenanigans, there is currently no non-NULL `func` member in execute_data. PHP truly is marvelous. This should close #515
2025-07-15fix: Build PHPPierre Tondereau
2025-06-25Fix debug log statementChristian Göttsche
sp_log_debug() does not take a feature as first argument: src/sp_wrapper.c: In function 'sp_reregister_php_wrapper': src/sp_utils.h:61:53: warning: too many arguments for format [-Wformat-extra-args] 61 | if (sp_debug_stderr > 0) dprintf(sp_debug_stderr, "[snuffleupagus][DEBUG] %s(): " fmt "\n", __FUNCTION__, ##__VA_ARGS__); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ src/sp_wrapper.c:144:3: note: in expansion of macro 'sp_log_debug' 144 | sp_log_debug(LOG_FEATURE, "Stream \"php\" successfully re-registered"); | ^~~~~~~~~~~~
2025-06-25Cast format argument to expected typeChristian Göttsche
Please GCC conversion warning: src/sp_upload_validation.c: In function 'sp_rfc1867_callback': src/sp_utils.h:61:53: warning: format '%lld' expects argument of type 'long long int', but argument 7 has type 'zend_long' {aka 'long int'} [-Wformat=] 61 | if (sp_debug_stderr > 0) dprintf(sp_debug_stderr, "[snuffleupagus][DEBUG] %s(): " fmt "\n", __FUNCTION__, ##__VA_ARGS__); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ src/sp_upload_validation.c:48:7: note: in expansion of macro 'sp_log_debug' 48 | sp_log_debug("Filename: %s\nTmpname: %s\nSize: %zd\nError: %lld\nScript: %s", | ^~~~~~~~~~~~
2025-06-25Do not crash on no cookie hash keyChristian Göttsche
Do not dereference the hash key for cookie encryption if it's NULL: Program terminated with signal SIGSEGV, Segmentation fault. #0 zend_string_equal_content (s1=0x79bdb92170f0, s2=0x0) at /usr/include/php/20240924/Zend/zend_string.h:386 No locals. #1 zend_string_equals (s1=0x79bdb92170f0, s2=0x0) at /usr/include/php/20240924/Zend/zend_string.h:391 No locals. #2 sp_match_value (value=0x0, to_match=0x79bdb92170f0, rx=0x0) at ./src/sp_utils.c:273 No locals. #3 0x00007989377b0709 in sp_lookup_cookie_config (key=0x0) at ./src/sp_cookie_encryption.c:8 config = 0x79bdb92158d0 it = 0x79ae80dabd00 it = <optimized out> config = <optimized out> #4 decrypt_cookie (pDest=0x79893b6787c0, num_args=<optimized out>, args=<optimized out>, hash_key=0x7ffe657c3880) at ./src/sp_cookie_encryption.c:19 cookie = <optimized out> #5 0x000061875aac52df in zend_hash_apply_with_arguments () No symbol table info available. #6 0x00007989377ae74b in zm_activate_snuffleupagus (type=<optimized out>, module_number=<optimized out>) at ./src/snuffleupagus.c:228 config_wrapper = 0x7989377c3490 <snuffleupagus_globals+144> #7 0x000061875aa21710 in zend_activate_modules () No symbol table info available. #8 0x000061875a9a7f18 in php_request_startup () No symbol table info available.
2025-06-16Fix spellingChristian Göttsche
2025-05-13Remove the useless show_old_php_warning featurejvoisin
People are usually well-aware of the outdatedness of the PHP version they're running, which is likely why they're running Snuffleupagus in the first place. This feature shouldn't have been enabled by default, and I fail to see any case where anyone would want to enable it. Moreover, it doesn't take LTS versions from vendors/distributions into account, thus breaking on RHEL/Debian (old)stable.
2025-04-11Minor code simplificationjvoisin
2024-09-06Fix a licensing typojvoisin
Spotted by @kpcyrd
2024-09-06Bump the changelogv0.11.0jvoisin
2024-09-06Skip a test on PHP8.4jvoisin
2024-07-09Fix compilation for PHP84+jvoisin
PHP changed some includes for PHP8.4. This commit also adds php8.4-rc to the CI This fixes #497 and #496 Reported-by: rainerjung
2024-06-18Updated the cached version of the config scannerjvoisin
2024-06-18Fix an UBSAN warningjvoisin
No need to use an ulong, an uint is more than enough for our usecases. This fixes the following warning: ``` src/sp_config.c:207:20: runtime error: store to misaligned address 0x796d1f9a78d4 for type 'u_long', which requires 8 byte alignment 0x796d1f9a78d4: note: pointer points here 00 00 00 00 ff 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ^ #0 0x796d1f92251a in parse_ulong src/sp_config.c:207 #1 0x796d1f9239a9 in sp_process_rule src/sp_config.c:76 #2 0x796d1f92404f in sp_process_config_root src/sp_config.c:31 #3 0x796d1f96770d in sp_config_scan src/sp_config_scanner.c:1482 #4 0x796d1f922cd3 in sp_parse_config src/sp_config.c:62 #5 0x796d1f91e6cf in OnUpdateConfiguration src/snuffleupagus.c:522 #6 0x598aae727646 in zend_register_ini_entries_ex (/usr/bin/php8.2+0x36d646) (BuildId: 5228f916ded87172ddf0f3eca448f1e43874d60f) #7 0x796d1f91cae1 in zm_startup_snuffleupagus src/snuffleupagus.c:121 #8 0x598aae6b70df in zend_startup_module_ex (/usr/bin/php8.2+0x2fd0df) (BuildId: 5228f916ded87172ddf0f3eca448f1e43874d60f) #9 0x598aae6b717f (/usr/bin/php8.2+0x2fd17f) (BuildId: 5228f916ded87172ddf0f3eca448f1e43874d60f) #10 0x598aae6c566a in zend_hash_apply (/usr/bin/php8.2+0x30b66a) (BuildId: 5228f916ded87172ddf0f3eca448f1e43874d60f) #11 0x598aae64c0b5 in php_module_startup (/usr/bin/php8.2+0x2920b5) (BuildId: 5228f916ded87172ddf0f3eca448f1e43874d60f) #12 0x598aae4e2265 (/usr/bin/php8.2+0x128265) (BuildId: 5228f916ded87172ddf0f3eca448f1e43874d60f) #13 0x796d21b4dc89 in __libc_start_call_main ../sysdeps/nptl/libc_start_call_main.h:58 #14 0x796d21b4dd44 in __libc_start_main_impl ../csu/libc-start.c:360 #15 0x598aae4e3550 in _start (/usr/bin/php8.2+0x129550) (BuildId: 5228f916ded87172ddf0f3eca448f1e43874d60f) SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior src/sp_config.c:207:20 in ```
2024-06-14Fix a portability issuejvoisin
This should fix the following compilation issue: ``` /wrkdirs/usr/ports/security/snuffleupagus/work-php83/snuffleupagus-0.10.0/src/sp_utils.c:438:37: error: too few arguments provided to function-like macro invocation 438 | memcpy(mb_name, ZEND_STRL("mb_")); | ^ /usr/include/ssp/string.h:117:9: note: macro 'memcpy' defined here 117 | #define memcpy(dst, src, len) __ssp_bos_check3(memcpy, dst, src, len) ```
2024-06-09Avoid logging success on failureChristian Göttsche
2024-06-09Ensure to inject php-stream filter also on stream_wrapper_restore()Christian Göttsche
Not sure this is needed, but better safe than sorry.
2024-06-09Use more generic expected test result formatChristian Göttsche
2024-06-09Add option to specify the allowed "php" wrapper typesChristian Göttsche
In addition of the current possibility to filter wrappers by their protocol name, also add the option to filter the "php" wrapper by the requested kind. Especially the 'filter' backend can be disabled that way.