]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Arran Cudbard-Bell [Tue, 17 Sep 2019 18:50:29 +0000 (13:50 -0500)]
Don't keep wide_path or wide_name around
Arran Cudbard-Bell [Tue, 17 Sep 2019 18:50:09 +0000 (13:50 -0500)]
Prefer whatever version python-config is, and then python3.
Alan T. DeKok [Tue, 17 Sep 2019 18:27:19 +0000 (14:27 -0400)]
minor simplifications
Alan T. DeKok [Tue, 17 Sep 2019 14:56:26 +0000 (10:56 -0400)]
"filter" is now like "update", too
Alan T. DeKok [Tue, 17 Sep 2019 14:31:25 +0000 (10:31 -0400)]
un-set special flag when we're done with update / map
Alan T. DeKok [Tue, 17 Sep 2019 13:59:24 +0000 (09:59 -0400)]
it helps to stat() the file
Alan T. DeKok [Tue, 17 Sep 2019 13:10:15 +0000 (09:10 -0400)]
make cf_file_include iterative instead of recursive
Alan T. DeKok [Mon, 16 Sep 2019 23:36:05 +0000 (19:36 -0400)]
check max stack only if we're loading the file
Alan T. DeKok [Mon, 16 Sep 2019 22:36:07 +0000 (18:36 -0400)]
rename for clarity
Alan T. DeKok [Mon, 16 Sep 2019 22:33:01 +0000 (18:33 -0400)]
use stack->bufsize instead of talloc_array_length
Alan T. DeKok [Mon, 16 Sep 2019 21:16:39 +0000 (17:16 -0400)]
check max stack depth too
Alan T. DeKok [Mon, 16 Sep 2019 21:11:52 +0000 (17:11 -0400)]
move map/update flags to stack frame
which incidentally also fixes issues with $INCLUDE in the middle
of an "update" section. That may be a bad idea, but there's no
real reason why it should be impossible
Alan T. DeKok [Mon, 16 Sep 2019 20:52:17 +0000 (16:52 -0400)]
merge cf_file_include() with cf_section_read()
so that it's less code, and we don't have two mutually recusive
functions.
Alan T. DeKok [Mon, 16 Sep 2019 20:42:56 +0000 (16:42 -0400)]
move filename strdup and assignment outside of cf_file_include
so that we can look at removing the function entirely, and moving
to an iterative approach
Alan T. DeKok [Mon, 16 Sep 2019 19:27:17 +0000 (15:27 -0400)]
convert cf reader to use file stack
right now it's the naive "use stack struct instead of passing
variables to a function". It still recurses. This is pass one.
Pass two is fixing the reader so that it's one function which
just pushes things into the stack *instead* of recursing.
Alan T. DeKok [Mon, 16 Sep 2019 19:25:06 +0000 (15:25 -0400)]
Revert "CID #145370 - Re-enable commented out code"
This reverts commit
1ceefc9925eb9570a67384ee4935e7f34cc5d3be .
Which breaks control and tls-cache virtual servers
Alan T. DeKok [Mon, 16 Sep 2019 18:52:51 +0000 (14:52 -0400)]
remove unused variable and definitions
Matt Rose [Tue, 17 Sep 2019 00:49:37 +0000 (20:49 -0400)]
various jenkinsfile fixes (#2996)
Alan T. DeKok [Mon, 16 Sep 2019 13:04:38 +0000 (09:04 -0400)]
set User here, too
Alan T. DeKok [Mon, 16 Sep 2019 13:03:20 +0000 (09:03 -0400)]
force user/group to be radiusd
Alan T. DeKok [Mon, 16 Sep 2019 11:41:37 +0000 (07:41 -0400)]
add newly created VPs to the state, not old VPs
Alan T. DeKok [Mon, 16 Sep 2019 11:33:52 +0000 (07:33 -0400)]
remove chown.
It's not needed according to the Systemd documentation
---
RuntimeDirectory=foo/bar baz
the service manager creates /run/foo (if it does not exist), /run/foo/bar, and /run/baz. The directories /run/foo/bar and /run/baz except /run/foo are owned by the user and group specified in User= and Group=, and removed when the service is stopped.
---
Alan T. DeKok [Mon, 16 Sep 2019 00:38:03 +0000 (20:38 -0400)]
don't increment variable if we're not going to use it
i.e. leave things alone on error instead of changing them, and
then returning an error
Arran Cudbard-Bell [Sun, 15 Sep 2019 19:24:28 +0000 (14:24 -0500)]
Don't deref actions when passing to NUM_ELEMENTS
Arran Cudbard-Bell [Sun, 15 Sep 2019 19:19:59 +0000 (14:19 -0500)]
CID #
1453698 - Don't know what this is about... maybe this will help?
Arran Cudbard-Bell [Sun, 15 Sep 2019 19:16:47 +0000 (14:16 -0500)]
CID #
1453690 - Bounds check on result
Arran Cudbard-Bell [Sun, 15 Sep 2019 19:14:33 +0000 (14:14 -0500)]
CID #
1453691 - Check return value of cass_result_column_name
Arran Cudbard-Bell [Sun, 15 Sep 2019 19:13:16 +0000 (14:13 -0500)]
CID #
1453692 - Check stack overrun *AFTER* incrementing ctx->stack_depth
Arran Cudbard-Bell [Sun, 15 Sep 2019 19:10:28 +0000 (14:10 -0500)]
CID #
1453693 - Bounds check on result
Arran Cudbard-Bell [Sun, 15 Sep 2019 19:08:33 +0000 (14:08 -0500)]
CID #
1453694 - Head must be initialised‽ How the hell did clang scan miss this.
Arran Cudbard-Bell [Sun, 15 Sep 2019 19:07:26 +0000 (14:07 -0500)]
CID #
1453695 - Fix TOCTOU on directory permissions
Arran Cudbard-Bell [Sun, 15 Sep 2019 19:04:35 +0000 (14:04 -0500)]
CID #
1453696 - Check return codes when checking timer events
Arran Cudbard-Bell [Sun, 15 Sep 2019 19:02:02 +0000 (14:02 -0500)]
CID #
1453697 - Array length vs index check incorrect
Arran Cudbard-Bell [Sun, 15 Sep 2019 18:59:37 +0000 (13:59 -0500)]
CID #
1453699 - Bad offset in bytesex swap
Arran Cudbard-Bell [Sun, 15 Sep 2019 18:57:09 +0000 (13:57 -0500)]
CID #145370 - Re-enable commented out code
Jorge Pereira [Sat, 14 Sep 2019 18:35:13 +0000 (15:35 -0300)]
rlm_python: Don't leak if got problems to parse the tuple (#2992)
Jorge Pereira [Sat, 14 Sep 2019 18:29:18 +0000 (15:29 -0300)]
rlm_python: Only set the tuple if both parameters are valid (#2991)
Arran Cudbard-Bell [Sat, 14 Sep 2019 16:06:55 +0000 (11:06 -0500)]
Invert test pass/fail calls
Alan T. DeKok [Sat, 14 Sep 2019 15:32:12 +0000 (11:32 -0400)]
define variables outside of 'ifdef openssl'
Alan T. DeKok [Sat, 14 Sep 2019 15:05:51 +0000 (11:05 -0400)]
create 'if' with correct name2
Arran Cudbard-Bell [Sat, 14 Sep 2019 03:17:06 +0000 (22:17 -0500)]
Add log output to some python tests
Arran Cudbard-Bell [Sat, 14 Sep 2019 03:05:39 +0000 (22:05 -0500)]
Reorganise tests, and add explicit test for shared storage
Arran Cudbard-Bell [Sat, 14 Sep 2019 02:57:35 +0000 (21:57 -0500)]
Produce errors if we fail marshalling
Jorge Pereira [Sat, 14 Sep 2019 02:40:18 +0000 (23:40 -0300)]
rlm_python: Fix the mod_populate_vptuple() to return correct 'octets' value (#2989)
Arran Cudbard-Bell [Sat, 14 Sep 2019 01:50:30 +0000 (20:50 -0500)]
Fix radiusd.config in rlm_python
Misunderstanding about how the inittab works, just creates a singleton, which still need to be imported individually into each interpreter. It does not create a module that's *automatically* imported into each interpreter.
Jorge Pereira [Sat, 14 Sep 2019 01:17:30 +0000 (22:17 -0300)]
rlm_python: PyObject functions should return 'Py_RETURN_NONE' instead of 'NULL' (#2988)
Arran Cudbard-Bell [Sat, 14 Sep 2019 00:21:14 +0000 (19:21 -0500)]
Skip auth_cext_compat for python3
Arran Cudbard-Bell [Sat, 14 Sep 2019 00:20:46 +0000 (19:20 -0500)]
Add version number in rlm_python
Arran Cudbard-Bell [Sat, 14 Sep 2019 00:20:13 +0000 (19:20 -0500)]
Fixup the dependency code to cache the default sections for version numbers and features
Jorge Pereira [Fri, 13 Sep 2019 23:38:20 +0000 (20:38 -0300)]
rlm_python: Fix the comments (#2987)
Arran Cudbard-Bell [Fri, 13 Sep 2019 20:24:16 +0000 (15:24 -0500)]
Fix some potentially invalid uses of PyDECREF
Arran Cudbard-Bell [Fri, 13 Sep 2019 20:19:22 +0000 (15:19 -0500)]
Ensure the thread state is the same as the one we set, after calling python land
Arran Cudbard-Bell [Fri, 13 Sep 2019 20:05:30 +0000 (15:05 -0500)]
Typo in comment
Arran Cudbard-Bell [Fri, 13 Sep 2019 19:50:39 +0000 (14:50 -0500)]
Don't need this twice
Arran Cudbard-Bell [Fri, 13 Sep 2019 19:48:33 +0000 (14:48 -0500)]
Log information about the certificate chain at higher debug levels, when making OCSP stapling requests
Arran Cudbard-Bell [Fri, 13 Sep 2019 16:56:29 +0000 (11:56 -0500)]
Fix comment headers
Matt Rose [Fri, 13 Sep 2019 16:46:01 +0000 (12:46 -0400)]
use git commit info to generate release string for packages (#2985)
Matt Rose [Fri, 13 Sep 2019 15:53:10 +0000 (11:53 -0400)]
return empty string when no changes are detected (#2984)
Alan T. DeKok [Fri, 13 Sep 2019 15:25:03 +0000 (11:25 -0400)]
actually fix it <sigh>
Alan T. DeKok [Fri, 13 Sep 2019 15:02:07 +0000 (11:02 -0400)]
fix doxygen
Alan T. DeKok [Fri, 13 Sep 2019 13:55:34 +0000 (09:55 -0400)]
one more "shut the heck up"
Alan T. DeKok [Fri, 13 Sep 2019 13:32:49 +0000 (09:32 -0400)]
shut up stupid analyzer
Jorge Pereira [Fri, 13 Sep 2019 11:52:10 +0000 (08:52 -0300)]
rlm_python: show also the stacktrace (#2979)
It shows the "<file>:<line_num> <func()>
Jorge Pereira [Fri, 13 Sep 2019 11:50:58 +0000 (08:50 -0300)]
rlm_python: the radiusd.py is deprecated (#2982)
Due to currently be loaded as a built-in function.
Jorge Pereira [Fri, 13 Sep 2019 11:50:34 +0000 (08:50 -0300)]
rlm_python: Fix the log() call (#2981)
Arran Cudbard-Bell [Fri, 13 Sep 2019 00:35:08 +0000 (19:35 -0500)]
Start splitting LDAP tutorial content into sections
Arran Cudbard-Bell [Thu, 12 Sep 2019 23:07:52 +0000 (18:07 -0500)]
Formatting fixes
Jorge Pereira [Thu, 12 Sep 2019 23:01:13 +0000 (20:01 -0300)]
rlm_python: Update example.py to be Python3 friendly (#2980)
Of course, keep to be supported by Python 2.x also.
Arran Cudbard-Bell [Thu, 12 Sep 2019 23:00:15 +0000 (18:00 -0500)]
Rcode and formatting fixes
Arran Cudbard-Bell [Thu, 12 Sep 2019 22:56:27 +0000 (17:56 -0500)]
Fix processing instantiate function return codes
Arran Cudbard-Bell [Thu, 12 Sep 2019 22:56:02 +0000 (17:56 -0500)]
We don't care if detach fails
Jorge Pereira [Thu, 12 Sep 2019 22:38:24 +0000 (19:38 -0300)]
rlm_python: The do_python_single() returns rlm_code_t (#2978)
* rlm_python: The do_python_single() returns rlm_code_t
Never return values less than 0.
* rlm_python: Only call 'instantiate' if it was defined
* rlm_python: Just use the correct rlm_rcode_t instead of int
* rlm_python: If the execution failed, just print the errors
It's helpful to see errors in runtime.
Arran Cudbard-Bell [Thu, 12 Sep 2019 21:14:46 +0000 (16:14 -0500)]
Use the correct osname
Matt Rose [Thu, 12 Sep 2019 18:51:25 +0000 (14:51 -0400)]
detect changes to package manifests and pass no-cache flag to docker if found (#2959)
Alan T. DeKok [Thu, 12 Sep 2019 18:31:07 +0000 (14:31 -0400)]
shut up clang scan
Jorge Pereira [Thu, 12 Sep 2019 18:03:44 +0000 (15:03 -0300)]
rlm_python: Load the libpython using correct extension (#2977)
Alan T. DeKok [Thu, 12 Sep 2019 17:21:20 +0000 (13:21 -0400)]
don't "goto exit" where exit is "return NULL"
Alan T. DeKok [Thu, 12 Sep 2019 17:20:55 +0000 (13:20 -0400)]
actually add CC_HINT(nonnull)
Alan T. DeKok [Thu, 12 Sep 2019 16:58:35 +0000 (12:58 -0400)]
add CC_HINT(nonnull) to shut up static analyzer
Alan T. DeKok [Thu, 12 Sep 2019 16:07:28 +0000 (12:07 -0400)]
allow '@', too
Alan T. DeKok [Thu, 12 Sep 2019 15:12:43 +0000 (11:12 -0400)]
fix canonicalize error function
so that it prints the correct thing
Alan T. DeKok [Thu, 12 Sep 2019 14:32:04 +0000 (10:32 -0400)]
remove extraneous comments
Alan T. DeKok [Thu, 12 Sep 2019 14:23:17 +0000 (10:23 -0400)]
more fixes for parser
allow more characters
if we don't use regex, treat '/' as a normal string
Alan T. DeKok [Thu, 12 Sep 2019 13:21:44 +0000 (09:21 -0400)]
remove unused variable
Alan T. DeKok [Wed, 11 Sep 2019 23:20:48 +0000 (19:20 -0400)]
nope, don't un-escape things ever
Alan T. DeKok [Wed, 11 Sep 2019 23:15:23 +0000 (19:15 -0400)]
remove last vestige of getstring() etc.
Alan T. DeKok [Wed, 11 Sep 2019 19:17:35 +0000 (15:17 -0400)]
allow cf_get_token() to do escaping or not
it generally escapes names, etc. But not values
Alan T. DeKok [Wed, 11 Sep 2019 20:41:54 +0000 (16:41 -0400)]
better checks for BARE_WORD strings
which may have ${..} in the middle of them. The old parser
would just accept anything that wasn't whitespace. Which would
cause other side-effects.
The updated parser only allows a limited set of characters. Which
works for all test cases, but which might disallow some corner
cases not tested for.
Arran Cudbard-Bell [Wed, 11 Sep 2019 21:30:15 +0000 (16:30 -0500)]
Fix calling doc test target in travis.yml
Arran Cudbard-Bell [Wed, 11 Sep 2019 20:53:49 +0000 (15:53 -0500)]
Aways Py_DecRef(inst->module)
Arran Cudbard-Bell [Wed, 11 Sep 2019 20:53:34 +0000 (15:53 -0500)]
Fix stupid timestamp issue
Jorge Pereira [Wed, 11 Sep 2019 20:34:21 +0000 (17:34 -0300)]
rlm_python: Fix marked unused complains (#2976)
Matthew Newton [Wed, 11 Sep 2019 20:16:51 +0000 (21:16 +0100)]
crossbuild: sh needs container up
Arran Cudbard-Bell [Wed, 11 Sep 2019 20:07:07 +0000 (15:07 -0500)]
Not this one...
Arran Cudbard-Bell [Wed, 11 Sep 2019 20:01:30 +0000 (15:01 -0500)]
Fix tests target names to be les lisfy
Arran Cudbard-Bell [Wed, 11 Sep 2019 19:44:18 +0000 (14:44 -0500)]
Minor fix
Jorge Pereira [Wed, 11 Sep 2019 19:43:08 +0000 (16:43 -0300)]
rlm_python: Add hack to bypass bug with 'pythoh2.7-config' in the OSX (#2975)
As discussed with Arran. the 'python2.7-config' has a weird bug on the
OSX that if you have CFLAGS="..." declared with whatever content you get
the below behavior.
$ python2.7-config --cflags | grep -o "\-arch [^ ]*"
$ CFLAGS='whatever' python2.7-config --cflags | grep -o "\-arch [^ ]*"
-arch x86_64
-arch i386
$
It results to break the build.
Arran Cudbard-Bell [Wed, 11 Sep 2019 19:35:28 +0000 (14:35 -0500)]
Catch more g and w flags
Alan T. DeKok [Wed, 11 Sep 2019 19:05:45 +0000 (15:05 -0400)]
be more forgiving about break and return
when there are things after them
Alan T. DeKok [Wed, 11 Sep 2019 16:54:22 +0000 (12:54 -0400)]
make conditional parser check for regexes, too
if tmpl_preparse() is told to not require a regex, then it should
treat '/' as any other character