]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Alan T. DeKok [Thu, 8 Feb 2018 13:41:12 +0000 (08:41 -0500)]
Revert "make fr_network_socket_delete() thread-safe"
This reverts commit
5dc5d9b97f523607c7829bdeaeed448bac49d8ba .
fr_network_socket_delete() should only be called by the IO
handler, which is already in the network thread
Arran Cudbard-Bell [Wed, 7 Feb 2018 23:12:38 +0000 (23:12 +0000)]
Store current module, and set old_backlog correctly when calling unlang_interpret_synchronous
Arran Cudbard-Bell [Wed, 7 Feb 2018 22:18:41 +0000 (22:18 +0000)]
Doxygen
Arran Cudbard-Bell [Wed, 7 Feb 2018 22:16:42 +0000 (22:16 +0000)]
Only build proto_tls_cache.c if we have OpenSSL
Arran Cudbard-Bell [Wed, 7 Feb 2018 21:32:22 +0000 (21:32 +0000)]
Add stub proto_tls_cache module to compile tls_cache sections
Arran Cudbard-Bell [Wed, 7 Feb 2018 21:06:50 +0000 (21:06 +0000)]
Ignore all subdirectories in src/include
Because manually updating .gitignore every time we create a new library or protocol is going to get very tedious
Alan T. DeKok [Wed, 7 Feb 2018 20:37:08 +0000 (15:37 -0500)]
this shouldn't be commited either
Alan T. DeKok [Wed, 7 Feb 2018 20:35:22 +0000 (15:35 -0500)]
ignore DHCP things
Alan T. DeKok [Wed, 7 Feb 2018 20:34:56 +0000 (15:34 -0500)]
this should NOT be committed
Arran Cudbard-Bell [Wed, 7 Feb 2018 17:50:37 +0000 (17:50 +0000)]
Call/compile TLS Session and OCSP server sections
This is technically async safe, but the calls are still potentially blocking. More work needed to fix that.
There's no advantage to having them actually async capable until the cache module is async capable anyway...
Arran Cudbard-Bell [Wed, 7 Feb 2018 16:12:07 +0000 (16:12 +0000)]
Check permissions last, right before we call the panic action
Alan T. DeKok [Wed, 7 Feb 2018 15:39:35 +0000 (10:39 -0500)]
add assertion that deletion succeeds
Alan T. DeKok [Wed, 7 Feb 2018 15:28:21 +0000 (10:28 -0500)]
clarify debug message
Arran Cudbard-Bell [Wed, 7 Feb 2018 14:23:08 +0000 (14:23 +0000)]
Add the -f back into LN-SF
Arran Cudbard-Bell [Wed, 7 Feb 2018 14:15:38 +0000 (14:15 +0000)]
Doxygen
Arran Cudbard-Bell [Wed, 7 Feb 2018 13:44:28 +0000 (13:44 +0000)]
Update kqueue FD removal assert to print errno
Arran Cudbard-Bell [Wed, 7 Feb 2018 12:47:23 +0000 (12:47 +0000)]
Move all rad_cond_assert to fr_cond_assert
The distinction isn't that important, and they ended up being mixed in the source files anyway.
Arran Cudbard-Bell [Wed, 7 Feb 2018 12:44:43 +0000 (12:44 +0000)]
Boilerplate functions for DHCPv6 (hey it's a start)
Alan T. DeKok [Tue, 6 Feb 2018 20:14:25 +0000 (15:14 -0500)]
make fr_network_socket_delete() thread-safe
Alan T. DeKok [Tue, 6 Feb 2018 20:08:53 +0000 (15:08 -0500)]
call fr_network_socket_delete() before closing the FDs
Alan T. DeKok [Tue, 6 Feb 2018 20:05:52 +0000 (15:05 -0500)]
allow listeners to remove themselves from the network
Alan T. DeKok [Mon, 5 Feb 2018 14:24:21 +0000 (09:24 -0500)]
limit max_outstanding to 1
Alan T. DeKok [Fri, 2 Feb 2018 15:51:35 +0000 (10:51 -0500)]
port #2175 from v3.0.x
Alan T. DeKok [Fri, 2 Feb 2018 14:10:30 +0000 (09:10 -0500)]
clean up error handling
Alan T. DeKok [Fri, 2 Feb 2018 14:06:25 +0000 (09:06 -0500)]
rely on parent function to insert retry timer
Alan T. DeKok [Fri, 2 Feb 2018 14:04:38 +0000 (09:04 -0500)]
allow faster retries for empty files
Alan T. DeKok [Thu, 1 Feb 2018 22:41:38 +0000 (17:41 -0500)]
deal with race conditions on duplicate packets
Untested as of yet. :(
Alan T. DeKok [Thu, 1 Feb 2018 18:48:48 +0000 (13:48 -0500)]
do de-dup detection for accounting packets, too
Alan T. DeKok [Thu, 1 Feb 2018 18:31:46 +0000 (13:31 -0500)]
more sanity checks
Alan T. DeKok [Thu, 1 Feb 2018 18:29:13 +0000 (13:29 -0500)]
make code/id 8-bit values
that's all we need
Alan T. DeKok [Thu, 1 Feb 2018 18:28:34 +0000 (13:28 -0500)]
I hate my life
Alan T. DeKok [Thu, 1 Feb 2018 13:49:49 +0000 (08:49 -0500)]
rely on talloc to clean things up
Alan T. DeKok [Thu, 1 Feb 2018 12:48:08 +0000 (07:48 -0500)]
don't pass bad values to the encoder
i.e. we have layers of checks, just in case one layer breaks
Alan T. DeKok [Thu, 1 Feb 2018 12:45:32 +0000 (07:45 -0500)]
fail accounting == do not respond
Alan T. DeKok [Thu, 1 Feb 2018 12:43:01 +0000 (07:43 -0500)]
move assert to run-time check
Arran Cudbard-Bell [Thu, 1 Feb 2018 03:03:15 +0000 (20:03 -0700)]
Don't leak message buffers
Arran Cudbard-Bell [Thu, 1 Feb 2018 02:56:01 +0000 (19:56 -0700)]
Don't leak xlat_thread_instances for epehemeral xlats
Arran Cudbard-Bell [Thu, 1 Feb 2018 02:14:12 +0000 (19:14 -0700)]
Doxygen
Matthew Newton [Thu, 1 Feb 2018 01:47:07 +0000 (01:47 +0000)]
developer build should have debug symbols in it
Arran Cudbard-Bell [Thu, 1 Feb 2018 01:33:21 +0000 (18:33 -0700)]
Peform adds after deletes because libkqueue/epoll are not smart enough to allow an add and a delete in the same operation even if they're for different filters
Arran Cudbard-Bell [Wed, 31 Jan 2018 23:40:45 +0000 (16:40 -0700)]
Check for both UMTS rcodes
Arran Cudbard-Bell [Wed, 31 Jan 2018 23:15:33 +0000 (16:15 -0700)]
Make that global log
Arran Cudbard-Bell [Wed, 31 Jan 2018 23:11:56 +0000 (16:11 -0700)]
Mark up duplicate requests as warnings
Arran Cudbard-Bell [Wed, 31 Jan 2018 22:36:12 +0000 (15:36 -0700)]
Remember the request that previously thawed the state
Alan T. DeKok [Wed, 31 Jan 2018 22:32:50 +0000 (17:32 -0500)]
doxygen
Alan T. DeKok [Wed, 31 Jan 2018 22:31:30 +0000 (17:31 -0500)]
const issues
Alan T. DeKok [Wed, 31 Jan 2018 21:56:47 +0000 (16:56 -0500)]
be more forgiving here, too
Alan T. DeKok [Wed, 31 Jan 2018 21:52:20 +0000 (16:52 -0500)]
be more forgiving...
Alan T. DeKok [Wed, 31 Jan 2018 21:45:59 +0000 (16:45 -0500)]
more talloc_get_type_abort
Alan T. DeKok [Wed, 31 Jan 2018 21:45:46 +0000 (16:45 -0500)]
convert assert to run-time check
so that it doesn't stomp on memory
Arran Cudbard-Bell [Wed, 31 Jan 2018 20:40:09 +0000 (13:40 -0700)]
when walking up the stack signalling all the things, update their view of the current stack frame
Alan T. DeKok [Wed, 31 Jan 2018 19:55:29 +0000 (14:55 -0500)]
send duplicate replies on duplicate requests
Alan T. DeKok [Wed, 31 Jan 2018 19:50:51 +0000 (14:50 -0500)]
set duplicate timestamp
Alan T. DeKok [Wed, 31 Jan 2018 19:15:38 +0000 (14:15 -0500)]
one more clean up tracking table entry...
Alan T. DeKok [Wed, 31 Jan 2018 18:56:15 +0000 (13:56 -0500)]
cleanups
Arran Cudbard-Bell [Wed, 31 Jan 2018 18:14:54 +0000 (11:14 -0700)]
Only parse the username portion of the identity string in fr_sim_id_type
Arran Cudbard-Bell [Wed, 31 Jan 2018 17:55:08 +0000 (10:55 -0700)]
Fix zero length format string warnings
Arran Cudbard-Bell [Wed, 31 Jan 2018 17:47:47 +0000 (10:47 -0700)]
Make tls-verify-cert work with async calls
Arran Cudbard-Bell [Wed, 31 Jan 2018 17:47:24 +0000 (10:47 -0700)]
Use const free when freeing reply in fr_radius_tracking_entry_reply
Arran Cudbard-Bell [Wed, 31 Jan 2018 17:47:03 +0000 (10:47 -0700)]
Fixup some eap submodule return codes
Arran Cudbard-Bell [Wed, 31 Jan 2018 16:22:55 +0000 (09:22 -0700)]
If entry->reply is already set, free it
Alan T. DeKok [Wed, 31 Jan 2018 16:26:56 +0000 (11:26 -0500)]
track FD filter type
and delete sockets before calling module detach or freeing the
event list, to ensure proper ordering.
Alan T. DeKok [Wed, 31 Jan 2018 16:25:26 +0000 (11:25 -0500)]
assert if FD is closed before it's removed from the event list
and continue deleting things even if something goes wrong.
We can't recover, so we might as well keep going.
Alan T. DeKok [Wed, 31 Jan 2018 15:49:22 +0000 (10:49 -0500)]
parse the limit section, and retransmit.
These aren't *technically* part of the detail "file" reader,
they're part of the worker. But they *are* copied to the worker.
And they are in the default config. So we should parse and use them here.
Alan T. DeKok [Wed, 31 Jan 2018 15:48:55 +0000 (10:48 -0500)]
be more accepting of boundary values
Alan T. DeKok [Wed, 31 Jan 2018 15:40:26 +0000 (10:40 -0500)]
lower the limits on num_messages
which should drastically decrease the memory usage of the detail
file reader
Arran Cudbard-Bell [Wed, 31 Jan 2018 14:13:31 +0000 (07:13 -0700)]
Don't leak memory on zero length expansion
Arran Cudbard-Bell [Wed, 31 Jan 2018 05:41:03 +0000 (22:41 -0700)]
Add basic, outer eap result functions
Arran Cudbard-Bell [Wed, 31 Jan 2018 05:40:22 +0000 (22:40 -0700)]
Remove unused function eap_basic_compose
Arran Cudbard-Bell [Wed, 31 Jan 2018 05:38:14 +0000 (22:38 -0700)]
Formatting
Alan T. DeKok [Wed, 31 Jan 2018 11:36:56 +0000 (06:36 -0500)]
one more assert...
Alan T. DeKok [Wed, 31 Jan 2018 11:36:51 +0000 (06:36 -0500)]
typo
Alan T. DeKok [Tue, 30 Jan 2018 20:07:15 +0000 (15:07 -0500)]
detached children can run forever... mostly
Alan T. DeKok [Tue, 30 Jan 2018 20:04:10 +0000 (15:04 -0500)]
remove unused struct entry
Arran Cudbard-Bell [Wed, 31 Jan 2018 02:27:58 +0000 (19:27 -0700)]
The behaviour of -C and -X should be identical in terms of what's allowed
Arran Cudbard-Bell [Wed, 31 Jan 2018 02:16:59 +0000 (19:16 -0700)]
Pass in structure names, because sizeof(sizeof(rest_xlat_thread_inst_t)) is not what we want
Arran Cudbard-Bell [Wed, 31 Jan 2018 01:20:33 +0000 (18:20 -0700)]
Initialise state tree earlier?
We seem to start processing requests before it's started
Arran Cudbard-Bell [Wed, 31 Jan 2018 01:14:25 +0000 (18:14 -0700)]
Set xlat instance data correctly instead of synthesizing it
Arran Cudbard-Bell [Wed, 31 Jan 2018 00:34:56 +0000 (17:34 -0700)]
Raise state tree exhaustion as an issue in the logs, and don't respond if we can't alloc state entries
Arran Cudbard-Bell [Wed, 31 Jan 2018 00:07:41 +0000 (17:07 -0700)]
Fix build issues
Arran Cudbard-Bell [Tue, 30 Jan 2018 23:55:38 +0000 (16:55 -0700)]
Fix const issues
Arran Cudbard-Bell [Tue, 30 Jan 2018 23:54:39 +0000 (16:54 -0700)]
doxygen
Arran Cudbard-Bell [Tue, 30 Jan 2018 23:46:30 +0000 (16:46 -0700)]
Indent debug
Arran Cudbard-Bell [Tue, 30 Jan 2018 23:46:24 +0000 (16:46 -0700)]
async xlat functions with no arguments can yield too!
Arran Cudbard-Bell [Tue, 30 Jan 2018 21:52:10 +0000 (14:52 -0700)]
Add xlat for rlm_delay
...it delays things.
Arran Cudbard-Bell [Tue, 30 Jan 2018 23:43:03 +0000 (16:43 -0700)]
Add xlat specific infrastructure for timeouts
Arran Cudbard-Bell [Tue, 30 Jan 2018 23:41:21 +0000 (16:41 -0700)]
Pass right arguments to xlat async function
Needs the xlat instance, not the wrapper struct
Arran Cudbard-Bell [Tue, 30 Jan 2018 23:40:48 +0000 (16:40 -0700)]
Add more sanity checks and pass correct values to instantiate functions
Arran Cudbard-Bell [Tue, 30 Jan 2018 21:51:49 +0000 (14:51 -0700)]
Typo
Arran Cudbard-Bell [Tue, 30 Jan 2018 20:29:12 +0000 (13:29 -0700)]
Store the current result in the stack structure so we can get at it
Arran Cudbard-Bell [Tue, 30 Jan 2018 20:28:35 +0000 (13:28 -0700)]
Add function to check for compiled unlang data
Arran Cudbard-Bell [Tue, 30 Jan 2018 20:28:01 +0000 (13:28 -0700)]
s/expanded/expansion/
Arran Cudbard-Bell [Tue, 30 Jan 2018 20:25:47 +0000 (13:25 -0700)]
unlang_push_section should take a top_frame argument
Arran Cudbard-Bell [Tue, 30 Jan 2018 20:08:26 +0000 (13:08 -0700)]
Be better about restoring previous module name, modules can be called inside modules
Arran Cudbard-Bell [Tue, 30 Jan 2018 20:05:35 +0000 (13:05 -0700)]
Set type of talloc chunk when memduping it
Alan T. DeKok [Tue, 30 Jan 2018 19:50:54 +0000 (14:50 -0500)]
detach MUST have a parent request
Alan T. DeKok [Tue, 30 Jan 2018 19:50:45 +0000 (14:50 -0500)]
more assertions and cleanups
Alan T. DeKok [Tue, 30 Jan 2018 17:29:22 +0000 (12:29 -0500)]
more debug3 macros
Alan T. DeKok [Tue, 30 Jan 2018 17:20:20 +0000 (12:20 -0500)]
add more verification of requests
Alan T. DeKok [Tue, 30 Jan 2018 17:18:38 +0000 (12:18 -0500)]
sanity check request->async, too