]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Alan T. DeKok [Wed, 25 Apr 2018 16:52:12 +0000 (12:52 -0400)]
moved master IO handler to src/lib/io
Alan T. DeKok [Wed, 25 Apr 2018 16:52:00 +0000 (12:52 -0400)]
print out what we received
Alan T. DeKok [Wed, 25 Apr 2018 16:28:19 +0000 (12:28 -0400)]
privatize internal structures to io.c
Alan T. DeKok [Wed, 25 Apr 2018 16:26:55 +0000 (12:26 -0400)]
whitespace
Alan T. DeKok [Wed, 25 Apr 2018 16:12:40 +0000 (12:12 -0400)]
move inst->io.FOO to inst->io in io.c
and various other minor cleanups
Alan T. DeKok [Wed, 25 Apr 2018 16:07:01 +0000 (12:07 -0400)]
move "network_get" to public app_io API
Alan T. DeKok [Wed, 25 Apr 2018 16:00:57 +0000 (12:00 -0400)]
move "connection_set" to public app_io API
Alan T. DeKok [Wed, 25 Apr 2018 15:49:34 +0000 (11:49 -0400)]
remove last vestiges of proto_radius knowledge from io.c
Alan T. DeKok [Wed, 25 Apr 2018 15:42:09 +0000 (11:42 -0400)]
add mod_instantiate to master IO handler
and rearrange code in proto_radius to match
Alan T. DeKok [Wed, 25 Apr 2018 15:37:25 +0000 (11:37 -0400)]
rearrangement
Alan T. DeKok [Wed, 25 Apr 2018 15:37:03 +0000 (11:37 -0400)]
make process_by_code smaller
we only need FR_MAX_PACKET_CODE, because that's all we support
Alan T. DeKok [Wed, 25 Apr 2018 15:01:12 +0000 (11:01 -0400)]
add mod_bootstrap to io.c
and move code from proto_radius over there
Alan T. DeKok [Tue, 24 Apr 2018 23:07:59 +0000 (19:07 -0400)]
no need to check process_by_code any more in io.c
Alan T. DeKok [Tue, 24 Apr 2018 23:04:36 +0000 (19:04 -0400)]
move app_io to fr_io_instance_t
Alan T. DeKok [Tue, 24 Apr 2018 22:53:37 +0000 (18:53 -0400)]
use set priority API
Alan T. DeKok [Tue, 24 Apr 2018 22:51:34 +0000 (18:51 -0400)]
add "set priority" API
to abstract that from the protocol agnostic IO handler
Alan T. DeKok [Tue, 24 Apr 2018 19:06:16 +0000 (15:06 -0400)]
remove assertion
Alan T. DeKok [Tue, 24 Apr 2018 18:53:52 +0000 (14:53 -0400)]
whitespace
Alan T. DeKok [Tue, 24 Apr 2018 18:50:39 +0000 (14:50 -0400)]
rename struct entries in preparation for splitting out io.c
Alan T. DeKok [Mon, 23 Apr 2018 20:02:13 +0000 (16:02 -0400)]
rename many proto_radius_foo to fr_io_foo
in preparation for lib-ifying io.c. So that other protocols
can use it, too
Alan T. DeKok [Mon, 23 Apr 2018 19:40:18 +0000 (15:40 -0400)]
abstract names for debug messages
so that io.c is largely protocol agnostic.
Alan T. DeKok [Mon, 23 Apr 2018 18:47:13 +0000 (14:47 -0400)]
only do duplicate tree if we track duplicates
Alan T. DeKok [Mon, 23 Apr 2018 18:44:30 +0000 (14:44 -0400)]
use new dedup function
Alan T. DeKok [Mon, 23 Apr 2018 18:44:04 +0000 (14:44 -0400)]
add packet comparison function
for doing dedup
Arran Cudbard-Bell [Wed, 25 Apr 2018 11:32:17 +0000 (23:32 +1200)]
Autoload dictionary attributes in rlm_pap might close #2220
...and a few other cleanups.
Arran Cudbard-Bell [Wed, 25 Apr 2018 11:26:25 +0000 (23:26 +1200)]
Move Auth-Type registration to bootstrap in rlm_mschap
Arran Cudbard-Bell [Wed, 25 Apr 2018 11:26:01 +0000 (23:26 +1200)]
Use common boilerplate for setting Auth-Type
Arran Cudbard-Bell [Wed, 25 Apr 2018 11:25:37 +0000 (23:25 +1200)]
Fixup dictionary numbers for hashed password attributes so they don't conflict
Arran Cudbard-Bell [Tue, 24 Apr 2018 00:45:54 +0000 (12:45 +1200)]
Rename README to README.md
Alan T. DeKok [Mon, 23 Apr 2018 15:21:47 +0000 (11:21 -0400)]
add "deny" to limited networks
Arran Cudbard-Bell [Sun, 22 Apr 2018 00:39:16 +0000 (12:39 +1200)]
Fix goto labels in auth_wbclient.c
Arran Cudbard-Bell [Sun, 22 Apr 2018 00:20:57 +0000 (12:20 +1200)]
Missing semicolon
Arran Cudbard-Bell [Sat, 21 Apr 2018 22:45:12 +0000 (10:45 +1200)]
Revert "quiet stupid compilers" - Shouldn't be needed
This reverts commit
91a7f2eb41195d27524746409e05beb8ff71c958 .
Arran Cudbard-Bell [Sat, 21 Apr 2018 11:18:57 +0000 (23:18 +1200)]
Always enable ASAN if requested, even if it's not a dev build
Only check for LSAN interface if we're building with ASAN
Arran Cudbard-Bell [Sat, 21 Apr 2018 01:20:51 +0000 (13:20 +1200)]
quiet stupid compilers
Alan T. DeKok [Fri, 20 Apr 2018 19:21:08 +0000 (15:21 -0400)]
Revert "add prototypes"
This reverts commit
1a789072c2c5ab77e74ba08ed53833b064fa7fbc .
GCC complains there's no function prototype. And if you add one,
complains about duplicate definition.
leave it un-prototyped until such time as we figure out WTF is
wrong with GCC
Alan T. DeKok [Fri, 20 Apr 2018 18:39:06 +0000 (14:39 -0400)]
add prototypes
Alan T. DeKok [Fri, 20 Apr 2018 17:52:40 +0000 (13:52 -0400)]
don't hard-code "UDP" as the transport name
Arran Cudbard-Bell [Fri, 20 Apr 2018 09:26:28 +0000 (21:26 +1200)]
Commit lsan.h
Arran Cudbard-Bell [Fri, 20 Apr 2018 08:50:25 +0000 (20:50 +1200)]
Hardcode supressions
Arran Cudbard-Bell [Fri, 20 Apr 2018 07:21:51 +0000 (19:21 +1200)]
Disable LSAN for more of the python initialisation functions as python sucks
Arran Cudbard-Bell [Fri, 20 Apr 2018 07:20:46 +0000 (19:20 +1200)]
Use LSAN top level header file
Arran Cudbard-Bell [Fri, 20 Apr 2018 07:10:23 +0000 (19:10 +1200)]
Fix unused args
Arran Cudbard-Bell [Fri, 20 Apr 2018 06:34:23 +0000 (18:34 +1200)]
destroy more functions
Arran Cudbard-Bell [Fri, 20 Apr 2018 06:21:15 +0000 (18:21 +1200)]
Switch to thread instantiate and destruction functions for rlm_python
Arran Cudbard-Bell [Fri, 20 Apr 2018 04:21:49 +0000 (16:21 +1200)]
Use the fr_strerror() logging macros
Alan T. DeKok [Thu, 19 Apr 2018 23:07:01 +0000 (19:07 -0400)]
xlat expansion errors are debug statements
Alan T. DeKok [Thu, 19 Apr 2018 23:03:55 +0000 (19:03 -0400)]
utility functions call fr_strerror_printf(), not ERROR
Arran Cudbard-Bell [Thu, 19 Apr 2018 07:34:22 +0000 (19:34 +1200)]
Fixup rlm_python to use mod_load and mod_unload
Arran Cudbard-Bell [Thu, 19 Apr 2018 06:53:39 +0000 (18:53 +1200)]
and some more
Arran Cudbard-Bell [Thu, 19 Apr 2018 06:52:14 +0000 (18:52 +1200)]
Fix memory leaks in 3gpp_pseydonym_encrypt
Only happened on error but definitely worth addressing
Alan T. DeKok [Wed, 18 Apr 2018 16:19:35 +0000 (12:19 -0400)]
clean up the client only if there are no packets for it
Alan T. DeKok [Wed, 18 Apr 2018 16:18:19 +0000 (12:18 -0400)]
delete max_request timer if there are no active requests
Alan T. DeKok [Wed, 18 Apr 2018 16:13:00 +0000 (12:13 -0400)]
use the correct structure entry
Alan T. DeKok [Wed, 18 Apr 2018 13:31:25 +0000 (09:31 -0400)]
we may have multiple clients being defined...
Alan T. DeKok [Wed, 18 Apr 2018 13:10:00 +0000 (09:10 -0400)]
free children before parent
Alan T. DeKok [Wed, 18 Apr 2018 13:00:12 +0000 (09:00 -0400)]
move code that increments pending packets
so that it's encapsulated in push/pop functions
Alan T. DeKok [Wed, 18 Apr 2018 12:32:20 +0000 (08:32 -0400)]
un-confuse the compiler. Hopefully.
Alan T. DeKok [Wed, 18 Apr 2018 11:58:39 +0000 (07:58 -0400)]
travis issues
Alan T. DeKok [Wed, 18 Apr 2018 11:55:15 +0000 (07:55 -0400)]
don't rely on "echo" supporting \n
Arran Cudbard-Bell [Wed, 18 Apr 2018 02:18:51 +0000 (14:18 +1200)]
Don't leak the first unknown attribute in an unknown attribute hierarchy
Arran Cudbard-Bell [Wed, 18 Apr 2018 00:26:51 +0000 (12:26 +1200)]
Fix leak in anything that used fr_value_box_from_fmt
Alan T. DeKok [Tue, 17 Apr 2018 20:19:15 +0000 (16:19 -0400)]
doxygen
Alan T. DeKok [Tue, 17 Apr 2018 20:17:45 +0000 (16:17 -0400)]
remove unused variable
Alan T. DeKok [Tue, 17 Apr 2018 20:06:42 +0000 (16:06 -0400)]
remove unused variables
Alan T. DeKok [Tue, 17 Apr 2018 20:05:35 +0000 (16:05 -0400)]
%zd -> %u
Alan T. DeKok [Tue, 17 Apr 2018 20:05:18 +0000 (16:05 -0400)]
doxygen
Alan T. DeKok [Tue, 17 Apr 2018 19:45:06 +0000 (15:45 -0400)]
always one thing I missed..
Alan T. DeKok [Tue, 17 Apr 2018 19:37:42 +0000 (15:37 -0400)]
Rewrite proto_radius to be not insane.
proto_radius_udp now ONLY does UDP reads/writes (and open/close)
it does not track packets, do dedup, etc.
All of the packet trackign has been moved to io.c, in proto_radius.
That function tracks pending packets, does dedup, etc.
The IO is bounced through io.c, which does the heavy lifting.
Then, if it actually needs to read/write packets to the network,
it calls the underlying app_io.
This means that RADIUS over TCP should be not much larger than
RADIUS over UDP. There will need to be some additional work to
get accepted sockets in, but the base work for connections exists.
As a plus, you can now define a dynamic client and mark it as
using connected sockets. Which means that NAT gateways can now
exist as dynamic clients, AND they can have connected sockets
behind them. i.e. each connection through a NAT gateway can
have a different shared secret.
Alan T. DeKok [Tue, 17 Apr 2018 19:30:25 +0000 (15:30 -0400)]
no longer look at client->network.
It's only used for proto_radius, and should never have been here.
Alan T. DeKok [Tue, 17 Apr 2018 19:30:00 +0000 (15:30 -0400)]
use IP address as part of client name for debugging
Alan T. DeKok [Tue, 17 Apr 2018 19:28:22 +0000 (15:28 -0400)]
undefined fields aren't assertions
Alan T. DeKok [Tue, 17 Apr 2018 19:27:50 +0000 (15:27 -0400)]
copy the memory for the new entry
Alan T. DeKok [Tue, 17 Apr 2018 17:49:11 +0000 (13:49 -0400)]
set packet event list AFTER inserting the FD into the event loop
so that the module can pause (or do whatever) to the FD as necessary
Alan T. DeKok [Tue, 17 Apr 2018 17:11:04 +0000 (13:11 -0400)]
tell dynamic client code if the socket is connected
Alan T. DeKok [Mon, 16 Apr 2018 19:53:08 +0000 (15:53 -0400)]
notes on low priority of detail file reading
Alan T. DeKok [Sun, 15 Apr 2018 21:40:54 +0000 (17:40 -0400)]
constify lookup functions
Alan T. DeKok [Fri, 13 Apr 2018 19:55:13 +0000 (15:55 -0400)]
don't pass a "free" function
the stats info is parented from the rbtree, so using a free
function results in a double free.
Alan T. DeKok [Fri, 13 Apr 2018 17:37:50 +0000 (13:37 -0400)]
add provisions for private APIs
Alan T. DeKok [Fri, 13 Apr 2018 13:08:29 +0000 (09:08 -0400)]
add connection tracking to clients.
Alan T. DeKok [Thu, 12 Apr 2018 14:37:41 +0000 (10:37 -0400)]
magic for network write
Alan T. DeKok [Thu, 12 Apr 2018 14:10:40 +0000 (10:10 -0400)]
pass packet_ct to NAK routine
Arran Cudbard-Bell [Mon, 16 Apr 2018 11:20:08 +0000 (23:20 +1200)]
Fixup odd output from format.pl
Probably something about fixing up the DHCPv4 dictionaries, will look when we have proper protocol dictionaries.
Arran Cudbard-Bell [Mon, 16 Apr 2018 11:17:25 +0000 (23:17 +1200)]
Exit cleanly from unit_test_attribute to avoid pages and pages of memory leak output
Arran Cudbard-Bell [Mon, 16 Apr 2018 08:12:28 +0000 (20:12 +1200)]
Use MEM to check allocation
Arran Cudbard-Bell [Mon, 16 Apr 2018 08:09:22 +0000 (20:09 +1200)]
Add IEC dictionary to help out with https://stackoverflow.com/questions/
47052503 /radius-and-iec-62351?rq=1
Arran Cudbard-Bell [Mon, 16 Apr 2018 08:07:03 +0000 (20:07 +1200)]
Format dictionary files
Arran Cudbard-Bell [Fri, 13 Apr 2018 06:41:02 +0000 (12:41 +0600)]
Autoload eap.c attributes
Arran Cudbard-Bell [Fri, 13 Apr 2018 05:13:28 +0000 (11:13 +0600)]
s/network_id/network_name/ to be consistent with RFC 5448
Arran Cudbard-Bell [Thu, 12 Apr 2018 06:49:02 +0000 (12:49 +0600)]
...and a few more
Arran Cudbard-Bell [Thu, 12 Apr 2018 06:37:59 +0000 (12:37 +0600)]
Once more pragma fixup
Arran Cudbard-Bell [Thu, 12 Apr 2018 06:12:33 +0000 (12:12 +0600)]
More sign fixes
Arran Cudbard-Bell [Thu, 12 Apr 2018 06:10:54 +0000 (12:10 +0600)]
Fixupup rlm_mschap Auth-Type enumv and the last few pair_make calls
Arran Cudbard-Bell [Thu, 12 Apr 2018 05:40:00 +0000 (11:40 +0600)]
Typo
Arran Cudbard-Bell [Thu, 12 Apr 2018 05:37:43 +0000 (11:37 +0600)]
Fix sign
Arran Cudbard-Bell [Thu, 12 Apr 2018 05:31:35 +0000 (11:31 +0600)]
Switch to #pragma once
Arran Cudbard-Bell [Thu, 12 Apr 2018 04:20:35 +0000 (10:20 +0600)]
Autoload rlm_mschap attributes
Arran Cudbard-Bell [Thu, 12 Apr 2018 02:51:16 +0000 (08:51 +0600)]
Formatting
Arran Cudbard-Bell [Thu, 12 Apr 2018 02:33:19 +0000 (08:33 +0600)]
Autoload rlm_logintime attrs
Arran Cudbard-Bell [Thu, 12 Apr 2018 02:32:53 +0000 (08:32 +0600)]
Use same logic as logintime