]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Alan T. DeKok [Tue, 30 Jan 2018 17:18:38 +0000 (12:18 -0500)]
sanity check request->async, too
Alan T. DeKok [Tue, 30 Jan 2018 16:33:12 +0000 (11:33 -0500)]
Revert "disable exec wait and ignore child signals"
This reverts commit
5432dc9359752cea74ed29b249696474e0ddb4b7 .
Now that exec.c tracks 'nowait' children, after the previous
commit.
Alan T. DeKok [Tue, 30 Jan 2018 16:32:24 +0000 (11:32 -0500)]
track "nowait" children we've exec'd.
So that they can be cleaned up appropriately.
Alan T. DeKok [Tue, 30 Jan 2018 15:56:34 +0000 (10:56 -0500)]
show the line number where the tests failed
Alan T. DeKok [Tue, 30 Jan 2018 15:43:30 +0000 (10:43 -0500)]
disable exec wait and ignore child signals
until such time as "wait" is integrated into the event loop.
Alan T. DeKok [Tue, 30 Jan 2018 15:29:36 +0000 (10:29 -0500)]
whitespace
Alan T. DeKok [Tue, 30 Jan 2018 14:29:59 +0000 (09:29 -0500)]
doxygen
Alan T. DeKok [Tue, 30 Jan 2018 14:28:57 +0000 (09:28 -0500)]
remove unused variable
Alan T. DeKok [Tue, 30 Jan 2018 14:08:53 +0000 (09:08 -0500)]
instruction may be NULL?
Arran Cudbard-Bell [Tue, 30 Jan 2018 02:37:08 +0000 (19:37 -0700)]
Install signal handler to destroy EAP sessions
Arran Cudbard-Bell [Tue, 30 Jan 2018 02:04:09 +0000 (19:04 -0700)]
Explicitly free rctx
Else we'd leak memory into the REQUEST *
Arran Cudbard-Bell [Tue, 30 Jan 2018 01:34:56 +0000 (18:34 -0700)]
Don't need this anymore
Arran Cudbard-Bell [Tue, 30 Jan 2018 00:35:19 +0000 (17:35 -0700)]
Try and emulate Linux symbol binding behaviour
Arran Cudbard-Bell [Tue, 30 Jan 2018 00:34:48 +0000 (17:34 -0700)]
Don't bake module.c into unit_test_module
It's provided by libfreeradius-server, and building with it here gives us a chimaera effect, where half the symbols resolve in the local version, and half resolve to libfreeradius-server.dylib
Arran Cudbard-Bell [Tue, 30 Jan 2018 00:23:12 +0000 (17:23 -0700)]
Sanity check for unlang_signal
Arran Cudbard-Bell [Tue, 30 Jan 2018 00:23:00 +0000 (17:23 -0700)]
Resume frame should repeat too
Arran Cudbard-Bell [Tue, 30 Jan 2018 00:22:49 +0000 (17:22 -0700)]
Pass a repeat function for going back up the stack
Arran Cudbard-Bell [Tue, 30 Jan 2018 00:19:20 +0000 (17:19 -0700)]
Case where q is NULL
Arran Cudbard-Bell [Mon, 29 Jan 2018 22:31:38 +0000 (15:31 -0700)]
Switch to two level namespace in macOS to fix OpenSSL/BoringSSL conflicts
Arran Cudbard-Bell [Mon, 29 Jan 2018 22:29:50 +0000 (15:29 -0700)]
Set/unset module names when calling signal and resume functions
Not just cosmetic fixes, we hit asserts if the RINDENTs() and REXDENTS() are done with an inconsistent request->module pointer.
This is because there's a separate indentation counter for when the request is being processed by a module, and for when the request is running outside of a module.
Arran Cudbard-Bell [Mon, 29 Jan 2018 22:28:08 +0000 (15:28 -0700)]
Formatting
Arran Cudbard-Bell [Mon, 29 Jan 2018 05:41:57 +0000 (22:41 -0700)]
Fix low level debug output in xlat_register
Arran Cudbard-Bell [Mon, 29 Jan 2018 05:20:45 +0000 (22:20 -0700)]
Fix ocsp.cnf
So we can generate fake certs again...
Arran Cudbard-Bell [Mon, 29 Jan 2018 04:49:14 +0000 (21:49 -0700)]
Start work on making rlm_eap async
Matthew Newton [Mon, 29 Jan 2018 20:37:33 +0000 (20:37 +0000)]
resolve symbols now, and use dlerror
Alan T. DeKok [Mon, 29 Jan 2018 19:53:08 +0000 (14:53 -0500)]
typo
Alan T. DeKok [Mon, 29 Jan 2018 19:04:31 +0000 (14:04 -0500)]
doxygen
Alan T. DeKok [Mon, 29 Jan 2018 19:01:35 +0000 (14:01 -0500)]
"documentation" is "useful"
Alan T. DeKok [Mon, 29 Jan 2018 18:58:54 +0000 (13:58 -0500)]
hacks for LD_PRELOAD, where we just load it ourselves...
Alan T. DeKok [Mon, 29 Jan 2018 18:49:06 +0000 (13:49 -0500)]
we don't use checkrad in v4
or, at least, in the same way as it was used in v3
Alan T. DeKok [Mon, 29 Jan 2018 18:44:55 +0000 (13:44 -0500)]
parse ENV subsection before loading modules or threads
so that we don't have to set LD_PRELOAD on the command line
for stupid libraries
Alan T. DeKok [Mon, 29 Jan 2018 15:21:12 +0000 (10:21 -0500)]
check for address->client
Arran Cudbard-Bell [Mon, 29 Jan 2018 03:01:36 +0000 (20:01 -0700)]
Allow unlang OPs to push generic functions onto the stack
This allows functions deeper in the C call stack to set up resumption points.
This makes submodules and even internal module functions far easier to integrate, as they don't have to communicate with callers further up the C stack, to get them to push the resumption functions they need.
Arran Cudbard-Bell [Mon, 29 Jan 2018 02:59:55 +0000 (19:59 -0700)]
Move more of the module specific code out of unlang_op.c
There's still a bunch of module specific event handler functions which either need to be made generic or moved into module_unlang.c
Arran Cudbard-Bell [Sun, 28 Jan 2018 23:29:30 +0000 (16:29 -0700)]
Always initialise quote
Arran Cudbard-Bell [Sun, 28 Jan 2018 23:06:14 +0000 (16:06 -0700)]
Fix handling of NULL RHS in update sections
Arran Cudbard-Bell [Sun, 28 Jan 2018 22:53:24 +0000 (15:53 -0700)]
Reset alternate on an additional code path
Arran Cudbard-Bell [Sun, 28 Jan 2018 22:36:30 +0000 (15:36 -0700)]
...an xlat can yield multiple times
Arran Cudbard-Bell [Sun, 28 Jan 2018 22:27:26 +0000 (15:27 -0700)]
Improve xlat evaluation debugging
It should be about equivalent to the synchronous debug output, with the added bonus feature of printing the function fmt string and the expanded arguments.
Arran Cudbard-Bell [Sun, 28 Jan 2018 22:24:45 +0000 (15:24 -0700)]
Add function for copying value_box lists and for checking list taint
Alan T. DeKok [Sun, 28 Jan 2018 19:58:20 +0000 (14:58 -0500)]
more hacks for tmpl_verify
Arran Cudbard-Bell [Sun, 28 Jan 2018 04:35:38 +0000 (21:35 -0700)]
Doxygen
Arran Cudbard-Bell [Sun, 28 Jan 2018 04:12:07 +0000 (21:12 -0700)]
Fix inifinite loop in xlat evaluator where both expansions in an alternate expansion evaluate to NULL
Arran Cudbard-Bell [Sun, 28 Jan 2018 04:11:37 +0000 (21:11 -0700)]
Use ephemeral xlat instantiation in unit tests
Arran Cudbard-Bell [Sat, 27 Jan 2018 15:38:15 +0000 (08:38 -0700)]
Add cancellation function for rest xlats
Arran Cudbard-Bell [Sat, 27 Jan 2018 06:46:19 +0000 (23:46 -0700)]
Extra sanity check
Alan T. DeKok [Sat, 27 Jan 2018 13:16:01 +0000 (08:16 -0500)]
if vp is NULL.... don't dereference it
Arran Cudbard-Bell [Sat, 27 Jan 2018 01:33:25 +0000 (18:33 -0700)]
Make debug message clearer
Arran Cudbard-Bell [Sat, 27 Jan 2018 00:56:11 +0000 (17:56 -0700)]
Don't leave stale REST-HTTP-Status-Code attributes around
Alan T. DeKok [Sat, 27 Jan 2018 00:55:31 +0000 (19:55 -0500)]
talloc_get_type is only in debug builds
Alan T. DeKok [Sat, 27 Jan 2018 00:53:51 +0000 (19:53 -0500)]
doxygen
Arran Cudbard-Bell [Sat, 27 Jan 2018 00:51:36 +0000 (17:51 -0700)]
...and another
Arran Cudbard-Bell [Sat, 27 Jan 2018 00:50:28 +0000 (17:50 -0700)]
Typo
Arran Cudbard-Bell [Sat, 27 Jan 2018 00:41:09 +0000 (17:41 -0700)]
Print out more of the error stack
...and fix EAP AKA/SIM identity parsing so it says exactly why it thinks the permanent ID is invalid
Alan T. DeKok [Sat, 27 Jan 2018 00:28:00 +0000 (19:28 -0500)]
add const
Arran Cudbard-Bell [Sat, 27 Jan 2018 00:01:23 +0000 (17:01 -0700)]
Fix talloc_list_get_type_abort when get_type_abort is disabled
Arran Cudbard-Bell [Sat, 27 Jan 2018 00:01:06 +0000 (17:01 -0700)]
Fix input to map processors to be async
Arran Cudbard-Bell [Fri, 26 Jan 2018 23:29:25 +0000 (16:29 -0700)]
Typo
Arran Cudbard-Bell [Fri, 26 Jan 2018 20:47:04 +0000 (13:47 -0700)]
Don't double alloc the output box in xlat_bin
Arran Cudbard-Bell [Fri, 26 Jan 2018 20:44:08 +0000 (13:44 -0700)]
Formatting
Arran Cudbard-Bell [Fri, 26 Jan 2018 20:43:25 +0000 (13:43 -0700)]
Move private structures into a private header file
Arran Cudbard-Bell [Fri, 26 Jan 2018 20:42:52 +0000 (13:42 -0700)]
Organisation
Arran Cudbard-Bell [Fri, 26 Jan 2018 20:42:41 +0000 (13:42 -0700)]
Check for list taint
Arran Cudbard-Bell [Fri, 26 Jan 2018 20:42:12 +0000 (13:42 -0700)]
Add incr_ref argumet to fr_value_box_copy_shallow
We don't want to increment reference counts when shallow copying globally allocated boxes
Arran Cudbard-Bell [Fri, 26 Jan 2018 20:41:32 +0000 (13:41 -0700)]
Add a value box 'meta' type
Arran Cudbard-Bell [Fri, 26 Jan 2018 20:40:20 +0000 (13:40 -0700)]
Add fr_value_box_alloc_null
To allocate null value boxes which are then populated with another function in the value.c API.
This means you don't have to pass the taint/enumv flags in twice (which looked stupid).
Alan T. DeKok [Fri, 26 Jan 2018 23:13:29 +0000 (18:13 -0500)]
don't re-define macro
Alan T. DeKok [Fri, 26 Jan 2018 19:52:08 +0000 (14:52 -0500)]
update track, too
Arran Cudbard-Bell [Fri, 26 Jan 2018 18:25:50 +0000 (11:25 -0700)]
Don't crash if an async function is used in the legacy xlat functions
Arran Cudbard-Bell [Fri, 26 Jan 2018 18:02:40 +0000 (11:02 -0700)]
Add generic talloc_list_get_type_abort function
...and use it to verify the state of the result xlat eval result list
Alan T. DeKok [Fri, 26 Jan 2018 14:11:47 +0000 (09:11 -0500)]
it's now allowed to delete entries with 0 timestamp
because we may receive packet A (slow). The NAS gives up and
sends packet B, which we reply to. After that, packet A is done.
At that point, the tracking entry has timestamp=0, uses=1, so
we can now delete it.
Alan T. DeKok [Fri, 26 Jan 2018 14:07:42 +0000 (09:07 -0500)]
reference count tracking table entries
Alan T. DeKok [Fri, 26 Jan 2018 14:07:23 +0000 (09:07 -0500)]
if we found an entry in the tree, don't re-insert it...
Alan T. DeKok [Fri, 26 Jan 2018 13:59:09 +0000 (08:59 -0500)]
pass recv_time to tracking entry delete
so that it can do all of the logic necessary for cleanups
Arran Cudbard-Bell [Fri, 26 Jan 2018 03:47:34 +0000 (20:47 -0700)]
Zero out input value box
Else we get a garabage enumv field and that breaks everything else
Arran Cudbard-Bell [Fri, 26 Jan 2018 03:47:14 +0000 (20:47 -0700)]
Don't try and use result after passing it into xlat functions
Arran Cudbard-Bell [Fri, 26 Jan 2018 03:18:44 +0000 (20:18 -0700)]
Compule recv Access-Request in TLS verify virtual server
Arran Cudbard-Bell [Fri, 26 Jan 2018 02:56:35 +0000 (19:56 -0700)]
If no input...
Arran Cudbard-Bell [Fri, 26 Jan 2018 02:47:37 +0000 (19:47 -0700)]
Deal with empty xlat expansions in a sane way
Arran Cudbard-Bell [Fri, 26 Jan 2018 02:14:43 +0000 (19:14 -0700)]
We actually want a pointer to a NULL pointer because the value box code gets excited
Arran Cudbard-Bell [Fri, 26 Jan 2018 02:12:57 +0000 (19:12 -0700)]
Concat input arguments into URL string
Arran Cudbard-Bell [Fri, 26 Jan 2018 02:12:36 +0000 (19:12 -0700)]
Fix empty result in xlat_frame_eval
Arran Cudbard-Bell [Fri, 26 Jan 2018 01:55:01 +0000 (18:55 -0700)]
Both variants?
Arran Cudbard-Bell [Fri, 26 Jan 2018 01:51:05 +0000 (18:51 -0700)]
Pass empty result when calling xlat_frame_eval_repeat internally
Arran Cudbard-Bell [Fri, 26 Jan 2018 01:48:01 +0000 (18:48 -0700)]
Need to pass in **result so xlat_frame_eval and functions can indicate when they've consumed result boxes
Arran Cudbard-Bell [Fri, 26 Jan 2018 00:43:28 +0000 (17:43 -0700)]
Set Accouting-Response code on handled
Alan T. DeKok [Fri, 26 Jan 2018 00:22:07 +0000 (19:22 -0500)]
check map, too
Alan T. DeKok [Thu, 25 Jan 2018 23:07:42 +0000 (18:07 -0500)]
Revert "Target needs to be absolute for the rule to run in the correct order"
This reverts commit
e46e496911ae57ce8843d8434bb257bbbc7996e1 .
which breaks the build on most systems
Alan T. DeKok [Thu, 25 Jan 2018 22:34:03 +0000 (17:34 -0500)]
temporary hack
Alan T. DeKok [Thu, 25 Jan 2018 22:19:21 +0000 (17:19 -0500)]
assert that the pointer is of the right data type
Alan T. DeKok [Thu, 25 Jan 2018 22:19:09 +0000 (17:19 -0500)]
memdup the packet we're injecting
Arran Cudbard-Bell [Thu, 25 Jan 2018 21:15:55 +0000 (14:15 -0700)]
Target needs to be absolute for the rule to run in the correct order
Arran Cudbard-Bell [Thu, 25 Jan 2018 21:05:34 +0000 (14:05 -0700)]
Fixup more tests
Adding [0] in conditions and & before attribute references
Matthew Newton [Thu, 25 Jan 2018 20:57:32 +0000 (20:57 +0000)]
rhel7 libwbclient RPM dependencies
Arran Cudbard-Bell [Thu, 25 Jan 2018 20:27:02 +0000 (13:27 -0700)]
Doxygen
Arran Cudbard-Bell [Thu, 25 Jan 2018 19:50:57 +0000 (12:50 -0700)]
Use generic talloc list free function
Arran Cudbard-Bell [Thu, 25 Jan 2018 19:50:25 +0000 (12:50 -0700)]
Doxygen
Alan T. DeKok [Thu, 25 Jan 2018 19:31:15 +0000 (14:31 -0500)]
clean up pointers on clone
and only free in the master
Alan T. DeKok [Thu, 25 Jan 2018 19:12:04 +0000 (14:12 -0500)]
retransmits start at 1, not at 0. Fixes #2166
The "total packet" counters are maintained separately.
So we need to count retransmits starting from 1, not 0.
Which also means we *print* retransmits starting from 1, not 0
Arran Cudbard-Bell [Thu, 25 Jan 2018 19:10:25 +0000 (12:10 -0700)]
Hack dynamic LHS functionality back in
...I will fix this mess when there's not an impending deadline