]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Arran Cudbard-Bell [Sun, 15 Mar 2020 21:06:24 +0000 (17:06 -0400)]
pair: Put legacy functions into their own source file
Alan T. DeKok [Sun, 15 Mar 2020 19:25:18 +0000 (15:25 -0400)]
no need to set "dict" for attributes of type "group"
It's now in da->ref->dict
And we don't want to over-ride *our* definition of da->dict, which
points to the parent dictionary for this attribute.
Alan T. DeKok [Sun, 15 Mar 2020 13:04:10 +0000 (09:04 -0400)]
make the fuzzer a bit smarter
so that it finds the protocol name from argv[0]
and it allows -D dictionary_dir
Arran Cudbard-Bell [Sat, 14 Mar 2020 04:07:08 +0000 (00:07 -0400)]
unit_test_attribute: Encoding 0 bytes is an error too
At least in terms of the unit tests
Arran Cudbard-Bell [Sat, 14 Mar 2020 04:04:51 +0000 (00:04 -0400)]
net: Pretty but slow, and the compiler does a poor job optimising.
Arran Cudbard-Bell [Fri, 13 Mar 2020 22:19:09 +0000 (18:19 -0400)]
dict: Fix more const issues, and start creating public/private versions of structures
Arran Cudbard-Bell [Fri, 13 Mar 2020 21:39:56 +0000 (17:39 -0400)]
dict: Remove fr_dict_dict_enum_by_value, Fixes #3321
Arran Cudbard-Bell [Fri, 13 Mar 2020 21:17:29 +0000 (17:17 -0400)]
dict: Fix issues with last commit
Arran Cudbard-Bell [Fri, 13 Mar 2020 20:34:35 +0000 (16:34 -0400)]
dict: Add a dict pointer to every attribute
Arran Cudbard-Bell [Fri, 13 Mar 2020 18:50:16 +0000 (14:50 -0400)]
radict: Log talloc aborts
Arran Cudbard-Bell [Fri, 13 Mar 2020 15:47:59 +0000 (11:47 -0400)]
trunk: Enable destructor _before_ starting the connection
Arran Cudbard-Bell [Fri, 13 Mar 2020 04:28:48 +0000 (00:28 -0400)]
debug: Free the null context so LSAN doesn't flag the memory as leaked
Arran Cudbard-Bell [Fri, 13 Mar 2020 02:00:31 +0000 (22:00 -0400)]
missing: Cast output of builtin_bswap
Arran Cudbard-Bell [Fri, 13 Mar 2020 01:52:52 +0000 (21:52 -0400)]
thread_local: TLS be static
Arran Cudbard-Bell [Fri, 13 Mar 2020 01:38:16 +0000 (21:38 -0400)]
radsmp: Too early!
Arran Cudbard-Bell [Fri, 13 Mar 2020 01:24:09 +0000 (21:24 -0400)]
Formatting
Arran Cudbard-Bell [Fri, 13 Mar 2020 01:10:51 +0000 (21:10 -0400)]
thread_local: Make sure that the thread_local atexit handler is called _last_
Arran Cudbard-Bell [Fri, 13 Mar 2020 00:21:58 +0000 (20:21 -0400)]
global: Explicit free functions for memory that's also freed by thread-local destructors should be removed
Because they're not needed, because we fixed the underlying issue, and now they cause a double free!
Arran Cudbard-Bell [Fri, 13 Mar 2020 00:09:54 +0000 (20:09 -0400)]
configure.ac: Add check for _builtin_czll net: Use that instead of flsll
Arran Cudbard-Bell [Thu, 12 Mar 2020 23:25:38 +0000 (19:25 -0400)]
thread_local: Rework thread local destructors
pthread_exit() no longer needs to be called in the main thread to ensure all the thread-local memory is cleaned up correctly
Arran Cudbard-Bell [Thu, 12 Mar 2020 23:09:39 +0000 (19:09 -0400)]
trunk: Pause connection signals during connection_notify
We need to write to the connection afterwards, so it's bad if it's freed out from under us...
Arran Cudbard-Bell [Thu, 12 Mar 2020 23:08:50 +0000 (19:08 -0400)]
connection: Add ability to pause deferred signals to stop the connection being freed when the caller isn't aware
Only process deferred signals after the highest handler in the stack exits
Arran Cudbard-Bell [Thu, 12 Mar 2020 14:29:33 +0000 (10:29 -0400)]
net: Use mutually acceptable naming scheme
Alan T. DeKok [Thu, 12 Mar 2020 14:20:19 +0000 (10:20 -0400)]
make sure there are trailing CRs
Alan T. DeKok [Thu, 12 Mar 2020 14:15:51 +0000 (10:15 -0400)]
Fix sigsegv in --mode=execute without parameters fixes #3319
and print out a descriptive error message
Arran Cudbard-Bell [Wed, 11 Mar 2020 23:40:36 +0000 (19:40 -0400)]
net: Fix fr_htonllx and fr_ntohll to produce big endian numbers... oops
Alan T. DeKok [Wed, 11 Mar 2020 12:53:08 +0000 (08:53 -0400)]
call pthread_exit, so thread destructors are called on exit
Alan T. DeKok [Wed, 11 Mar 2020 12:46:22 +0000 (08:46 -0400)]
rearrange for simplicity
Alan T. DeKok [Wed, 11 Mar 2020 00:19:56 +0000 (20:19 -0400)]
use retry API
Alan T. DeKok [Wed, 11 Mar 2020 00:11:00 +0000 (20:11 -0400)]
move to retry_config_t
Jorge Pereira [Tue, 10 Mar 2020 23:14:13 +0000 (20:14 -0300)]
Fix typo
Alan T. DeKok [Tue, 10 Mar 2020 19:49:37 +0000 (15:49 -0400)]
typo
Alan T. DeKok [Tue, 10 Mar 2020 19:42:34 +0000 (15:42 -0400)]
always go to the next step
even if this step didn't receive all of the replies, or if
this step didn't reach it's expected PPS.
Then once we reach maximum PPS, go to the draining state, so that
it stops.
Arran Cudbard-Bell [Tue, 10 Mar 2020 18:30:59 +0000 (12:30 -0600)]
If the trunk it doing work when the event loop is exiting that's a pretty major bug
Jorge Pereira [Tue, 10 Mar 2020 17:34:48 +0000 (14:34 -0300)]
rlm_redis: Add missing 'database' parameter (#3317)
Alan T. DeKok [Tue, 10 Mar 2020 17:15:30 +0000 (13:15 -0400)]
stop doing work if the event loop is exiting
and therefore don't crash...
Alan T. DeKok [Tue, 10 Mar 2020 14:17:40 +0000 (10:17 -0400)]
signed / unsigned int issues
Alan T. DeKok [Tue, 10 Mar 2020 14:11:35 +0000 (10:11 -0400)]
tags of 0 can be omitted when printing
Alan T. DeKok [Tue, 10 Mar 2020 14:03:03 +0000 (10:03 -0400)]
add unit tests for Tunnel-Password
now that we have seeded randomness
Alan T. DeKok [Tue, 10 Mar 2020 14:01:00 +0000 (10:01 -0400)]
use random padding, instead of zeros
that way if the salts are the same, the encoded data will still
be all different.
RFC 2868 Section 3.5 recommends that the padding is all zeros.
But the field already contains a length, so it's not necessary.
And for CoA-Request packets, the "encyption" of the password
is done via the request authenticator, which is all zeros.
So we rely only on the salt for security. Which is small, and
therefore terrible.
Changing to random padding makes it harder to figure out what's
going on.
Matthew Newton [Mon, 9 Mar 2020 14:13:50 +0000 (14:13 +0000)]
crossbuild: add info about current image state
Alan T. DeKok [Mon, 9 Mar 2020 12:22:14 +0000 (08:22 -0400)]
use CS_INT for variable, as that's what the function takes
Instead of making the variable "int", and then casting it to
CS_INT when calling the function. That just won't work.
Stephan E [Mon, 9 Mar 2020 11:18:17 +0000 (12:18 +0100)]
fix HUP rate limit for large deltas in system time
The subtraction of the last HUP's time_t from the current time_t is cast into an int.
Depending on what int defaults to this may underflow and make HUPs impossible.
e.g. for a 32 bit signed int, after 69 years, HUPs will be rejected
Matthew Newton [Mon, 9 Mar 2020 10:46:16 +0000 (10:46 +0000)]
keep doxygen happy
Alan T. DeKok [Mon, 9 Mar 2020 02:24:46 +0000 (22:24 -0400)]
remove profile tests if built without profiling
Alan T. DeKok [Sat, 7 Mar 2020 20:39:08 +0000 (15:39 -0500)]
make the tests pass when readline isn't used
Matthew Newton [Sat, 7 Mar 2020 16:53:45 +0000 (16:53 +0000)]
typo: if defined
Matthew Newton [Sat, 7 Mar 2020 16:49:58 +0000 (16:49 +0000)]
fix up specfile for templates broken in
b2f2d1b082
Arran Cudbard-Bell [Fri, 6 Mar 2020 23:51:47 +0000 (17:51 -0600)]
trunk: Expose tconn state too
Arran Cudbard-Bell [Fri, 6 Mar 2020 23:48:16 +0000 (17:48 -0600)]
rlm_radius_udp: Add assert to catch requests timing out in invalid states
Arran Cudbard-Bell [Fri, 6 Mar 2020 23:48:00 +0000 (17:48 -0600)]
trunk: expose treq state
Arran Cudbard-Bell [Fri, 6 Mar 2020 23:42:51 +0000 (17:42 -0600)]
trunk: Document fail_bound
Arran Cudbard-Bell [Fri, 6 Mar 2020 23:32:07 +0000 (17:32 -0600)]
trunk: Add sanity check to prevent internal asserts firing to fr_trunk_requeuest_requeue
Alan T. DeKok [Fri, 6 Mar 2020 23:19:31 +0000 (18:19 -0500)]
clear ID on edge case when requeue on connection which is failed
Arran Cudbard-Bell [Fri, 6 Mar 2020 23:07:08 +0000 (17:07 -0600)]
value: Have fr_value_box_from_network return the correct negative offsets/errors for decode issues
Arran Cudbard-Bell [Fri, 6 Mar 2020 23:05:59 +0000 (17:05 -0600)]
dhcpv6-decode: Return proper negative offsets and errors when decoding
Arran Cudbard-Bell [Fri, 6 Mar 2020 23:04:18 +0000 (17:04 -0600)]
dhcpv6-decode: Don't need to use fr_pair_list_free when we're freeing a single vp
Alan T. DeKok [Fri, 6 Mar 2020 22:23:18 +0000 (17:23 -0500)]
use seeded randomness for tunnel passwords
and move counter to packet_ctx, instead of a "static" variable
Alan T. DeKok [Fri, 6 Mar 2020 22:14:59 +0000 (17:14 -0500)]
pass encoder_ctx to encode tunnel password
Alan T. DeKok [Fri, 6 Mar 2020 21:54:48 +0000 (16:54 -0500)]
remove unused functions
code duplication is waste
Alan T. DeKok [Fri, 6 Mar 2020 21:48:19 +0000 (16:48 -0500)]
add small / fast random number generator
Arran Cudbard-Bell [Fri, 6 Mar 2020 22:23:18 +0000 (16:23 -0600)]
eap/value: Remove duplicate macros
Arran Cudbard-Bell [Fri, 6 Mar 2020 21:48:24 +0000 (15:48 -0600)]
pair: Establish error codes for decoders
Arran Cudbard-Bell [Fri, 6 Mar 2020 21:19:37 +0000 (15:19 -0600)]
net: Missed space
Arran Cudbard-Bell [Fri, 6 Mar 2020 20:54:40 +0000 (14:54 -0600)]
net: Fix function order
Arran Cudbard-Bell [Fri, 6 Mar 2020 20:51:15 +0000 (14:51 -0600)]
net: Use standard names for bytesex conversion functions
Make all bytesex conversion functions use consistent argument names.
Arran Cudbard-Bell [Fri, 6 Mar 2020 16:24:27 +0000 (10:24 -0600)]
Move htonx and ntohx to net.h
Remove radius_packet_t from net.h
Arran Cudbard-Bell [Fri, 6 Mar 2020 13:47:08 +0000 (07:47 -0600)]
Was a two byte typo
Arran Cudbard-Bell [Fri, 6 Mar 2020 03:51:08 +0000 (21:51 -0600)]
Typo
Arran Cudbard-Bell [Thu, 5 Mar 2020 23:55:18 +0000 (17:55 -0600)]
misc: Add htonx and ntohx
For efficient packing and unpacking of network order integers, where the integer is of a variable size.
Arran Cudbard-Bell [Thu, 5 Mar 2020 23:53:26 +0000 (17:53 -0600)]
misc.c: Move DIVIDE_CEIL into misc.h
Arran Cudbard-Bell [Thu, 5 Mar 2020 22:20:39 +0000 (16:20 -0600)]
redis: Cluster coordinator notes
Arran Cudbard-Bell [Thu, 5 Mar 2020 22:00:47 +0000 (16:00 -0600)]
redis/pipeline: Formatting
Arran Cudbard-Bell [Thu, 5 Mar 2020 20:09:12 +0000 (14:09 -0600)]
rlm_radius: Consistent terminology
Arran Cudbard-Bell [Thu, 5 Mar 2020 20:08:01 +0000 (14:08 -0600)]
rlm_radius: Formatting
Arran Cudbard-Bell [Thu, 5 Mar 2020 20:00:27 +0000 (14:00 -0600)]
rlm_radius: Formatting
Arran Cudbard-Bell [Thu, 5 Mar 2020 19:54:38 +0000 (13:54 -0600)]
rlm_radius: Finish manual test plan for rlm_radius
Arran Cudbard-Bell [Thu, 5 Mar 2020 17:24:22 +0000 (11:24 -0600)]
rlm_radius: Use the v3 name for this config item
Alan T. DeKok [Thu, 5 Mar 2020 13:52:15 +0000 (08:52 -0500)]
clear can_retransmit only for status-server
Arran Cudbard-Bell [Thu, 5 Mar 2020 02:40:31 +0000 (20:40 -0600)]
rlm_radius_udp: el is used
Arran Cudbard-Bell [Thu, 5 Mar 2020 00:32:58 +0000 (18:32 -0600)]
trunk: The correct fix here was to mirror the "init" state, not the failed state
Arran Cudbard-Bell [Thu, 5 Mar 2020 00:26:55 +0000 (18:26 -0600)]
trunk: We don't need to do duplicate work on failed
Arran Cudbard-Bell [Thu, 5 Mar 2020 00:26:24 +0000 (18:26 -0600)]
trunk: Implement backlog drain on fail
Arran Cudbard-Bell [Thu, 5 Mar 2020 00:26:00 +0000 (18:26 -0600)]
trunk: Typo
Arran Cudbard-Bell [Wed, 4 Mar 2020 22:43:02 +0000 (16:43 -0600)]
trunk: Initial state should be halted
Arran Cudbard-Bell [Wed, 4 Mar 2020 22:38:38 +0000 (16:38 -0600)]
cf_parse: Use consistent macro names
Arran Cudbard-Bell [Wed, 4 Mar 2020 22:36:13 +0000 (16:36 -0600)]
retransmission is obnoxiously long, and rtx is well known
Arran Cudbard-Bell [Wed, 4 Mar 2020 22:34:23 +0000 (16:34 -0600)]
We use "max" everywhere else
Arran Cudbard-Bell [Wed, 4 Mar 2020 22:30:46 +0000 (16:30 -0600)]
rlm_radius_udp: Add note
Arran Cudbard-Bell [Wed, 4 Mar 2020 22:27:52 +0000 (16:27 -0600)]
rlm_radius_udp: This comment is wrong too
Arran Cudbard-Bell [Wed, 4 Mar 2020 22:25:19 +0000 (16:25 -0600)]
rlm_radius_udp: Remove incorrect call to fr_retry_next
Resending a status-check packet because we want num_to_alive does not count as a retransmission, and should not increase the retransmission interval, and should not be subject to MRD or MRC.
Arran Cudbard-Bell [Wed, 4 Mar 2020 22:20:07 +0000 (16:20 -0600)]
rlm_radius_udp: Add a simpler implementation status checks that run before the connection opens
Arran Cudbard-Bell [Wed, 4 Mar 2020 21:49:54 +0000 (15:49 -0600)]
rlm_radius_udp: Change status check talloc hierarchy to reflect a normal talloc hierarchy for requests/resume ctxs
Arran Cudbard-Bell [Wed, 4 Mar 2020 21:48:14 +0000 (15:48 -0600)]
rlm_radius_udp: Always enforce reply types
Arran Cudbard-Bell [Wed, 4 Mar 2020 21:25:25 +0000 (15:25 -0600)]
trunk: Typo in timestamp comparison
Arran Cudbard-Bell [Wed, 4 Mar 2020 21:24:52 +0000 (15:24 -0600)]
Better debugging on trunk start
Arran Cudbard-Bell [Wed, 4 Mar 2020 21:24:12 +0000 (15:24 -0600)]
trunk: Failed and closed should _NOT_ be excluded from the set of connections
Arran Cudbard-Bell [Wed, 4 Mar 2020 21:23:05 +0000 (15:23 -0600)]
trunk: zero out treq pointers on free
Arran Cudbard-Bell [Wed, 4 Mar 2020 18:19:26 +0000 (12:19 -0600)]
rlm_radius_udp: Free handle using a talloc destructor
It's safer
Arran Cudbard-Bell [Wed, 4 Mar 2020 18:18:05 +0000 (12:18 -0600)]
connection: close connections which hit their connection timeout too
shuffle asserts for is_closed