]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Arran Cudbard-Bell [Mon, 7 Oct 2019 00:11:18 +0000 (20:11 -0400)]
Add mini state machine for running auth sections
Alan DeKok [Sun, 6 Oct 2019 19:19:12 +0000 (15:19 -0400)]
Merge pull request #3028 from jpereira/v4/tests2
Fix radmin 'set module'
Jorge Pereira [Sun, 6 Oct 2019 15:07:22 +0000 (12:07 -0300)]
Fix typo, it should be radiud.log
Jorge Pereira [Wed, 2 Oct 2019 16:51:15 +0000 (13:51 -0300)]
Add tests for radmin 'set module'
Jorge Pereira [Wed, 2 Oct 2019 16:50:56 +0000 (13:50 -0300)]
Fix radmin 'set module'
Alan T. DeKok [Sun, 6 Oct 2019 13:55:20 +0000 (09:55 -0400)]
make tests use TEST_BOOTSTRAP function
Alan T. DeKok [Sun, 6 Oct 2019 12:11:50 +0000 (08:11 -0400)]
move test bootstrap code into common function
Alan T. DeKok [Sun, 6 Oct 2019 11:42:04 +0000 (07:42 -0400)]
renaming
* we don't need RADMIN_OUTPUT_DIR, we can just use OUTOUT
* it's better to use ".txt" for text files, even if they are
input files
* similar ".out" instead of ".ret". They're still text, but
we care more to know that they're outputs
* test files don't need numbers. If the tests are ordered,
we should put that information into the tests instead of
in the names
* test files don't need to be in a subdirectory.
Jorge Pereira [Wed, 2 Oct 2019 01:36:24 +0000 (22:36 -0300)]
Add tests for radmin/control-socket
Jorge Pereira [Fri, 4 Oct 2019 15:58:35 +0000 (12:58 -0300)]
Add new 'RADIUSD_SERVICE' make target
More informations can be found in src/tests/radiusd.mk
Alan T. DeKok [Sat, 5 Oct 2019 13:18:13 +0000 (09:18 -0400)]
rename unlang functions for clarity
running a request is now just unlang_interpret(request)
functions taking more arguments have longer names
Alan T. DeKok [Sat, 5 Oct 2019 13:13:39 +0000 (09:13 -0400)]
no longer needed
Alan T. DeKok [Sat, 5 Oct 2019 12:48:00 +0000 (08:48 -0400)]
rename function
we're compiling a section, not getting a subsection from a
parent section
Alan T. DeKok [Sat, 5 Oct 2019 12:47:14 +0000 (08:47 -0400)]
reset when going to next, so that it is always done
Arran Cudbard-Bell [Sat, 5 Oct 2019 05:32:37 +0000 (01:32 -0400)]
Remove server/base.h from io/base.h
Because circular dependencies aren't fun
Arran Cudbard-Bell [Sat, 5 Oct 2019 03:07:52 +0000 (23:07 -0400)]
Allow subrequest sessions to be enabled/disabled and for custom unique_ptr and unique_int values to be specified
This is required for modules pushing subrequests as the instructions usually used to identify sessions are dynamically allocated
Arran Cudbard-Bell [Sat, 5 Oct 2019 01:58:29 +0000 (21:58 -0400)]
Don't assign empty string
Alan T. DeKok [Fri, 4 Oct 2019 14:56:18 +0000 (10:56 -0400)]
fix last commit
Alan T. DeKok [Fri, 4 Oct 2019 14:09:23 +0000 (10:09 -0400)]
make '-i filename' work again :(
Alan T. DeKok [Thu, 3 Oct 2019 02:08:59 +0000 (22:08 -0400)]
start of framework for "local" commands
Alan T. DeKok [Thu, 3 Oct 2019 01:45:48 +0000 (21:45 -0400)]
remove debugging message
Alan Buxey [Wed, 2 Oct 2019 23:48:38 +0000 (00:48 +0100)]
grammar and spelling in sites-available (#3021)
Alan T. DeKok [Wed, 2 Oct 2019 02:19:55 +0000 (22:19 -0400)]
clients can be proto UDP, TCP, or *
Alan T. DeKok [Wed, 2 Oct 2019 02:12:59 +0000 (22:12 -0400)]
!whitespace is also "not zero and not whitespace"
When parsing zero-terminated strings, stop at zero.
Alan T. DeKok [Wed, 2 Oct 2019 02:05:58 +0000 (22:05 -0400)]
if no proto is specified, look for both udp and tcp
Alan T. DeKok [Wed, 2 Oct 2019 02:00:51 +0000 (22:00 -0400)]
fix assertion
Alan T. DeKok [Wed, 2 Oct 2019 00:32:10 +0000 (20:32 -0400)]
hacks to get parallel children resumable
Arran Cudbard-Bell [Tue, 1 Oct 2019 22:02:35 +0000 (17:02 -0500)]
Use table_len when iterating over tables Closes #3017
Jorge Pereira [Tue, 1 Oct 2019 21:54:52 +0000 (18:54 -0300)]
Fix typo (#3018)
Jorge Pereira [Tue, 1 Oct 2019 21:54:30 +0000 (18:54 -0300)]
Fix potential access to dereferenced value (#3016)
Arran Cudbard-Bell [Tue, 1 Oct 2019 19:55:01 +0000 (14:55 -0500)]
Return RLM_MODULE_YIELD from module functions
Alan T. DeKok [Tue, 1 Oct 2019 18:59:26 +0000 (14:59 -0400)]
change check to assertion
Alan T. DeKok [Tue, 1 Oct 2019 18:57:22 +0000 (14:57 -0400)]
add new unlang_subrequest_start()
to start a subrequest
Alan T. DeKok [Tue, 1 Oct 2019 18:53:30 +0000 (14:53 -0400)]
remove "fixup" function
and move repeatable_set() to unlang_module_yield()
Arran Cudbard-Bell [Tue, 1 Oct 2019 18:51:11 +0000 (13:51 -0500)]
Move the store/restore subrequest calls out of rlm_eap
Arran Cudbard-Bell [Tue, 1 Oct 2019 18:19:19 +0000 (13:19 -0500)]
Make asserts less verbose
Arran Cudbard-Bell [Tue, 1 Oct 2019 18:13:03 +0000 (13:13 -0500)]
Be entirely unforgiving about fr_state_store_in_parent and friends being called incorrectly
Arran Cudbard-Bell [Tue, 1 Oct 2019 17:39:58 +0000 (12:39 -0500)]
Ensure the child has its own unique state_ctx, i.e. it's in the same state as when request_data_restore_to_child was called
Jorge Pereira [Tue, 1 Oct 2019 17:13:51 +0000 (14:13 -0300)]
Fix typo (#3015)
Arran Cudbard-Bell [Tue, 1 Oct 2019 17:12:04 +0000 (12:12 -0500)]
Only free the existing state_ctx if we're sure it's not the parent's
Alan T. DeKok [Tue, 1 Oct 2019 17:02:52 +0000 (13:02 -0400)]
doxygen
Alan T. DeKok [Tue, 1 Oct 2019 16:41:23 +0000 (12:41 -0400)]
try to work around things with unlang_interpret_fixup()
which runs things locally via playing with the stack.
Alan T. DeKok [Tue, 1 Oct 2019 14:29:48 +0000 (10:29 -0400)]
fix return code
Alan T. DeKok [Tue, 1 Oct 2019 14:29:26 +0000 (10:29 -0400)]
add missing declaration
Alan T. DeKok [Tue, 1 Oct 2019 13:43:40 +0000 (09:43 -0400)]
check child state_ctx against parent
The EAP code should arguably call request_detach() or something,
but this is a quick and easy fix
Alan T. DeKok [Tue, 1 Oct 2019 13:25:48 +0000 (09:25 -0400)]
rename "persist" as it's confusing. Use and check "detachable"
Alan T. DeKok [Tue, 1 Oct 2019 13:11:52 +0000 (09:11 -0400)]
only parent requests are runnable
Alan T. DeKok [Tue, 1 Oct 2019 12:39:16 +0000 (08:39 -0400)]
more debug messages
Jorge Pereira [Mon, 30 Sep 2019 23:15:25 +0000 (20:15 -0300)]
Don't leak if failed the MS-CHAP password handle
Arran Cudbard-Bell [Mon, 30 Sep 2019 21:02:02 +0000 (16:02 -0500)]
Return correct rcode
Arran Cudbard-Bell [Mon, 30 Sep 2019 21:01:50 +0000 (16:01 -0500)]
Dump packet in radius encoder
Arran Cudbard-Bell [Mon, 30 Sep 2019 20:10:03 +0000 (15:10 -0500)]
Print hex blocks with whitespace
Arran Cudbard-Bell [Mon, 30 Sep 2019 19:47:16 +0000 (14:47 -0500)]
Fix eapmschapv2
Arran Cudbard-Bell [Mon, 30 Sep 2019 19:46:21 +0000 (14:46 -0500)]
Add tracing to mschap_challenge_hash
Arran Cudbard-Bell [Mon, 30 Sep 2019 19:46:01 +0000 (14:46 -0500)]
Fixup rlm_mschap to support EAP-Identity too
Alan T. DeKok [Mon, 30 Sep 2019 19:34:37 +0000 (15:34 -0400)]
use sw->id which is local
Alan T. DeKok [Mon, 30 Sep 2019 19:34:08 +0000 (15:34 -0400)]
use worker->name
instead of function call and thread-local storage
Alan T. DeKok [Mon, 30 Sep 2019 18:39:17 +0000 (14:39 -0400)]
clarify comments and code structure
Alan T. DeKok [Mon, 30 Sep 2019 15:55:52 +0000 (11:55 -0400)]
print "Resuming execution" when we resume
Alan T. DeKok [Mon, 30 Sep 2019 15:30:51 +0000 (11:30 -0400)]
get rid of unlang_group_type
which was always a horrific hack. Instead, just fix the
compile_action_default() function to sneak a peek at the parent.
Then, add "redundant" as a top-level type / keyword. which
lets us print out "redundant" in debug mode instead of "group"
Alan T. DeKok [Mon, 30 Sep 2019 13:19:24 +0000 (09:19 -0400)]
rename func / process to interpret
in preparation for adding compile / parse functions to the
unlang_op_t structure
Arran Cudbard-Bell [Mon, 30 Sep 2019 15:02:36 +0000 (10:02 -0500)]
Split out virtual servers and modules used by them
Arran Cudbard-Bell [Mon, 30 Sep 2019 14:55:00 +0000 (09:55 -0500)]
Fix clan scan issues
Arran Cudbard-Bell [Mon, 30 Sep 2019 14:54:51 +0000 (09:54 -0500)]
Fix comparators to return a number within a limited range
Arran Cudbard-Bell [Mon, 30 Sep 2019 13:11:25 +0000 (08:11 -0500)]
Not needed
Alan T. DeKok [Sun, 29 Sep 2019 22:42:37 +0000 (18:42 -0400)]
remove deterministic load balancing, and revert to random
which means we have less magic knowledge in the interpreter, and
less recursive code. At the cost of being less deterministic
Alan T. DeKok [Sun, 29 Sep 2019 20:17:43 +0000 (16:17 -0400)]
remove "detach" magic from main interpreter loop
detach now "does the right thing", and behaves appropriately.
Alan T. DeKok [Sun, 29 Sep 2019 12:17:03 +0000 (08:17 -0400)]
remove unused parameter to signal()
Alan T. DeKok [Sun, 29 Sep 2019 11:53:21 +0000 (07:53 -0400)]
make UNLANG_TYPE_CALL skip using using resumable frames
and now that nothing uses resumable frames, remove them entirely
from the interpreter
Alan T. DeKok [Sat, 28 Sep 2019 19:23:09 +0000 (15:23 -0400)]
make UNLANG_TYPE_XLAT skip using resumable frames
Alan T. DeKok [Sat, 28 Sep 2019 19:09:38 +0000 (15:09 -0400)]
get rid of unlang_module_resume(), as we no longer need it.
And do some other minor cleanups.
Alan T. DeKok [Sat, 28 Sep 2019 19:08:11 +0000 (15:08 -0400)]
this shouldn't happen
Alan T. DeKok [Sat, 28 Sep 2019 14:06:51 +0000 (10:06 -0400)]
make UNLANG_TYPE_SUBREQUEST skip using resumable frames
and clean up subrequest_push() so that it skips the
unlang_subrequest_state_init() function, and just calls the
unlang_subrequest_process() function directly.
Alan T. DeKok [Sat, 28 Sep 2019 14:22:19 +0000 (10:22 -0400)]
remove "resume" callback as it's no longer used
and restructure the code accordingly.
Alan T. DeKok [Sat, 28 Sep 2019 12:18:42 +0000 (08:18 -0400)]
rename to yielded_set, etc.
and fix a couple of bugs where yielded wasn't cleared when
instructions were being run. And where the current request
wasn't being checked for STOP_PROCESSING
Alan T. DeKok [Sat, 28 Sep 2019 11:44:12 +0000 (07:44 -0400)]
make UNLANG_TYPE_PARALLEL skip using resumable frames
Alan T. DeKok [Fri, 27 Sep 2019 18:56:46 +0000 (14:56 -0400)]
make UNLANG_TYPE_MODULE skip using resumable frames
The basic tests seem to work. So let's hope that the rest works.
Alan T. DeKok [Fri, 27 Sep 2019 17:53:11 +0000 (13:53 -0400)]
more docs and hand-holding.
Spoon-feeding developers is a good idea.
Arran Cudbard-Bell [Fri, 27 Sep 2019 17:54:40 +0000 (12:54 -0500)]
Typo
Arran Cudbard-Bell [Fri, 27 Sep 2019 17:51:58 +0000 (12:51 -0500)]
Only build the support eap-aka-sim library if we have OpenSSL
Alan T. DeKok [Fri, 27 Sep 2019 17:44:23 +0000 (13:44 -0400)]
document and clarify check for missing xlats
Alan T. DeKok [Fri, 27 Sep 2019 17:22:32 +0000 (13:22 -0400)]
more fixes
Alan T. DeKok [Fri, 27 Sep 2019 17:19:21 +0000 (13:19 -0400)]
fix authparams. to authparams->
Alan T. DeKok [Fri, 27 Sep 2019 17:15:33 +0000 (13:15 -0400)]
add "inst" and fix typo
Arran Cudbard-Bell [Fri, 27 Sep 2019 17:41:47 +0000 (12:41 -0500)]
Disable test.doc until it meets the following conditions
- Provides hints on how to fix check failures
- Provides a command to run to re-run checks fo failures
- asciidoctor ignores antora includes and stops producing spurious warnings
- Provides a clear indication of what's being used to generate build products
Arran Cudbard-Bell [Fri, 27 Sep 2019 17:22:53 +0000 (12:22 -0500)]
Return error codes in addition to asserting
Arran Cudbard-Bell [Fri, 27 Sep 2019 16:45:18 +0000 (11:45 -0500)]
Fix EAP-SIM includes
Alan T. DeKok [Fri, 27 Sep 2019 16:40:54 +0000 (12:40 -0400)]
header files are named for the dictionary directory
not the source directory.
Alan T. DeKok [Fri, 27 Sep 2019 15:46:38 +0000 (11:46 -0400)]
doxygen fixes
Alan T. DeKok [Fri, 27 Sep 2019 13:59:59 +0000 (09:59 -0400)]
notes on how it works, and what to fix in the future
Alan T. DeKok [Fri, 27 Sep 2019 13:56:36 +0000 (09:56 -0400)]
set parent to be resumable here, too
Alan T. DeKok [Fri, 27 Sep 2019 13:27:33 +0000 (09:27 -0400)]
clean up unlang_interpret_resumable() and parallel
instead of having magic for parallel in unlang_interpret_resumable(),
parallel just pushes a UNLANG_TYPE_FUNCTION into the childs stack.
The function then runs as the last thing before the child exits.
The function can then notify the parent that the child has exited.
Alan T. DeKok [Fri, 27 Sep 2019 13:10:23 +0000 (09:10 -0400)]
check dl_module->dl
which is NULL if we tried to load a module and failed
Alan T. DeKok [Thu, 26 Sep 2019 23:09:52 +0000 (19:09 -0400)]
parallel children are subrequests
Arran Cudbard-Bell [Thu, 26 Sep 2019 23:04:08 +0000 (18:04 -0500)]
Add support for allocating frame state pools
Arran Cudbard-Bell [Thu, 26 Sep 2019 23:03:52 +0000 (18:03 -0500)]
Preprocessor will optimise the calculations away
Arran Cudbard-Bell [Thu, 26 Sep 2019 22:12:49 +0000 (17:12 -0500)]
s/frame_inst/frame_state/g
Arran Cudbard-Bell [Thu, 26 Sep 2019 22:08:10 +0000 (17:08 -0500)]
Uninvert the timeout delete function calls
Alan T. DeKok [Thu, 26 Sep 2019 21:45:14 +0000 (17:45 -0400)]
the delay module can have xlat OR module timeouts
Alan T. DeKok [Thu, 26 Sep 2019 19:57:11 +0000 (15:57 -0400)]
use frame_inst_size