]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Arran Cudbard-Bell [Mon, 23 Jan 2017 17:32:08 +0000 (17:32 +0000)]
Fix call to value_box_asprint
Arran Cudbard-Bell [Mon, 23 Jan 2017 17:00:07 +0000 (17:00 +0000)]
Add type field to value_box_t
... and modify all the hundres of places which get pair type from vp->da->type, to get it from the boxed value instead.
Add additional validation functions to check for places where the box type isn't set correctly.
References #1883
Closes #1012
Arran Cudbard-Bell [Mon, 23 Jan 2017 16:58:39 +0000 (16:58 +0000)]
Const to the left
Alan T. DeKok [Fri, 20 Jan 2017 23:13:20 +0000 (18:13 -0500)]
Fixed raw VSA test
Alan T. DeKok [Fri, 20 Jan 2017 23:13:08 +0000 (18:13 -0500)]
comment out failing tests until we can fix them
Alan T. DeKok [Fri, 20 Jan 2017 23:04:11 +0000 (18:04 -0500)]
more unknown vendors
Alan T. DeKok [Fri, 20 Jan 2017 23:02:15 +0000 (18:02 -0500)]
more unknown attributes
Alan T. DeKok [Fri, 20 Jan 2017 23:00:40 +0000 (18:00 -0500)]
one more unknown attribute
Alan T. DeKok [Fri, 20 Jan 2017 22:59:06 +0000 (17:59 -0500)]
Using "Attr-X" means that the RHS should be hex
and not ipaddr, or other data types
Alan T. DeKok [Fri, 20 Jan 2017 22:52:01 +0000 (17:52 -0500)]
show what the expansion was
Alan T. DeKok [Fri, 20 Jan 2017 22:51:45 +0000 (17:51 -0500)]
we can only cast LHS of type attr from hex
Alan T. DeKok [Fri, 20 Jan 2017 21:55:01 +0000 (16:55 -0500)]
Don't look at what we parsed, look at the data
Alan T. DeKok [Fri, 20 Jan 2017 21:47:49 +0000 (16:47 -0500)]
look for "is_raw", not "is_unknown".
And look up attribute by parent / attr, not by vendor / num
Alan T. DeKok [Fri, 20 Jan 2017 21:47:05 +0000 (16:47 -0500)]
use "is_raw" for tokenize condition, not "is_unknown"
Alan T. DeKok [Fri, 20 Jan 2017 21:46:23 +0000 (16:46 -0500)]
Add flags.is_raw. Addresses #1883
Attributes which are in the config files are Attr-26 are known,
but SHOULD still be treated as "raw" attributes
Alan T. DeKok [Fri, 20 Jan 2017 18:27:40 +0000 (13:27 -0500)]
check for ATTR_UNDEFINED, too
Alan T. DeKok [Fri, 20 Jan 2017 18:18:11 +0000 (13:18 -0500)]
define unknown attributes in existence checks, too. Addresses #1883
Alan T. DeKok [Fri, 20 Jan 2017 17:48:01 +0000 (12:48 -0500)]
Look up unknown attributes by name, not by number.
And update the ipaddr.raw test to match.
next, update the tests for attribute existence...
Alan T. DeKok [Fri, 20 Jan 2017 17:44:49 +0000 (12:44 -0500)]
use found parent, not old parent.
The found parent may have just been created
Alan T. DeKok [Fri, 20 Jan 2017 17:44:35 +0000 (12:44 -0500)]
set parent and depth when we create the attribute.
Alan T. DeKok [Fri, 20 Jan 2017 17:24:57 +0000 (12:24 -0500)]
use fr_dict_attr_add as the name for errors
Alan T. DeKok [Fri, 20 Jan 2017 17:23:34 +0000 (12:23 -0500)]
use new fr_dict_attr_add_by_name() function. Addresses #1883
Alan T. DeKok [Fri, 20 Jan 2017 17:16:02 +0000 (12:16 -0500)]
Add fr_dict_attr_add_by_name(). Addresses #1883
This function adds an attribute ONLY by name, and not by number.
It can be used to add Attr-26, without polluting Vendor-Specific
for everyone else.
The dictionary code assumes that when two attributes are defined
for the same number, that the second one takes precedence over
the first one. See "dictionary.compat" for examples
Alan T. DeKok [Fri, 20 Jan 2017 17:12:14 +0000 (12:12 -0500)]
fix fr_dict_unknown_add(). Addresses #1883
When an admin uses Attr-26 in the config files, this function
needs to add Attr-26, and then return the newly-defined Attr-26,
and NOT return Vendor-Specific
Matthew Newton [Fri, 20 Jan 2017 16:26:15 +0000 (16:26 +0000)]
remove Kibana 3 dashboard, as it is now obsolete :(
Matthew Newton [Fri, 20 Jan 2017 16:25:48 +0000 (16:25 +0000)]
update elasticsearch/logstash examples so that they work with elastic stack v5
Alan T. DeKok [Fri, 20 Jan 2017 00:52:52 +0000 (19:52 -0500)]
similar test to raw VSA.
We're not sure what the results are until we fix the rest of the
code, but this is a good place-holder.
Alan T. DeKok [Thu, 19 Jan 2017 23:18:56 +0000 (18:18 -0500)]
use fr_dict_non_data_types instead of local version
Alan T. DeKok [Thu, 19 Jan 2017 23:15:39 +0000 (18:15 -0500)]
Revert "don't re-define known attributes"
This reverts commit
ace93f82666940f95a2e04436e8e29f3440d2c73 .
nope...
Alan T. DeKok [Thu, 19 Jan 2017 22:33:39 +0000 (17:33 -0500)]
We SHOULD be able to parse this as a raw attribute.
However, right now the code magically converts the LHS to
PW_TYPE_VENDOR_SPECIFIC. And then gets a run-time error.
Instead, it should leave the LHS alone, and just create a raw
attribute.
Alan T. DeKok [Thu, 19 Jan 2017 22:31:46 +0000 (17:31 -0500)]
Don't allow cast to structural types
Alan T. DeKok [Thu, 19 Jan 2017 22:31:31 +0000 (17:31 -0500)]
don't re-define known attributes
Alan T. DeKok [Thu, 19 Jan 2017 22:05:39 +0000 (17:05 -0500)]
use new non-data type array
Alan T. DeKok [Thu, 19 Jan 2017 22:04:50 +0000 (17:04 -0500)]
add array for structural data types
Alan T. DeKok [Thu, 19 Jan 2017 20:29:44 +0000 (15:29 -0500)]
Don't parse unknown to structural attributes
Alan T. DeKok [Thu, 19 Jan 2017 17:30:43 +0000 (12:30 -0500)]
Vendor-Specific is not allowed
Alan T. DeKok [Thu, 19 Jan 2017 16:02:06 +0000 (11:02 -0500)]
control plane needs a per-originater ring buffer
Messages are owned by the producer. Memory management is done
by the producer. The control-plane just pushes producer
messages into the consumer's atomic queue, and notifies the
consumer's KQ
Alan T. DeKok [Wed, 18 Jan 2017 22:48:54 +0000 (17:48 -0500)]
fix for dumb compiler
Alan T. DeKok [Wed, 18 Jan 2017 21:28:43 +0000 (16:28 -0500)]
check data size in production build, too
Alan T. DeKok [Wed, 18 Jan 2017 21:28:01 +0000 (16:28 -0500)]
enforce max ident
Alan T. DeKok [Wed, 18 Jan 2017 20:50:50 +0000 (15:50 -0500)]
A simple RADIUS server which uses the scheduler.
The receiver needs to be updated to link to a worker, and to
create a message to send to the worker. But at least the basics
work right now.
Alan T. DeKok [Wed, 18 Jan 2017 20:50:10 +0000 (15:50 -0500)]
add a socket to a scheduler
Alan T. DeKok [Wed, 18 Jan 2017 20:49:42 +0000 (15:49 -0500)]
provisions for adding a socket to a receiver
Alan T. DeKok [Wed, 18 Jan 2017 19:45:13 +0000 (14:45 -0500)]
move receiver to callbacks
The receive can now receive additional control-plane messages.
such as "new socket".
Alan T. DeKok [Wed, 18 Jan 2017 19:33:02 +0000 (14:33 -0500)]
move worker to control-plane callbacks
Alan T. DeKok [Wed, 18 Jan 2017 19:31:43 +0000 (14:31 -0500)]
Add function to service control-plane messages
Alan T. DeKok [Wed, 18 Jan 2017 17:37:46 +0000 (12:37 -0500)]
typo. Fixes #1882
Alan T. DeKok [Wed, 18 Jan 2017 18:08:34 +0000 (13:08 -0500)]
make channels take control plane, instead of kq/aq
so taht we can re-use control planes across channels
Alan T. DeKok [Wed, 18 Jan 2017 17:03:06 +0000 (12:03 -0500)]
add preparation for callbacks based on ident
Alan T. DeKok [Tue, 17 Jan 2017 20:55:09 +0000 (15:55 -0500)]
doxygen
Alan T. DeKok [Tue, 17 Jan 2017 20:38:01 +0000 (15:38 -0500)]
allow programmatic change of network / workers
Alan T. DeKok [Tue, 17 Jan 2017 20:35:19 +0000 (15:35 -0500)]
pass fr_log_t to fr_schedule_create
so that we can do debug logging of messages
Alan T. DeKok [Tue, 17 Jan 2017 20:17:58 +0000 (15:17 -0500)]
move more generic log functions to src/lib/log.c
Alan T. DeKok [Tue, 17 Jan 2017 20:05:42 +0000 (15:05 -0500)]
vradlog() -> fr_vlog(), radlog() -> fr_log()
As they are now library functions, not server-side functions
Alan T. DeKok [Tue, 17 Jan 2017 19:58:45 +0000 (14:58 -0500)]
move radlog() and vradlog() to src/lib/log.c
Along with default_log, and some other non-server logging things
Alan T. DeKok [Tue, 17 Jan 2017 17:50:16 +0000 (12:50 -0500)]
added schedule_test
Alan T. DeKok [Tue, 17 Jan 2017 16:24:01 +0000 (11:24 -0500)]
debugging
Alan T. DeKok [Tue, 17 Jan 2017 16:23:06 +0000 (11:23 -0500)]
create the "done_workers" heap
Alan T. DeKok [Tue, 17 Jan 2017 16:22:30 +0000 (11:22 -0500)]
signal the semaphore once started
Alan T. DeKok [Tue, 17 Jan 2017 16:20:31 +0000 (11:20 -0500)]
fr_event_exit() now signals the event loop to exit
which wakes it up if it was sleeping
Alan T. DeKok [Mon, 16 Jan 2017 21:40:45 +0000 (16:40 -0500)]
create and clean up the network thread
Alan T. DeKok [Mon, 16 Jan 2017 21:38:25 +0000 (16:38 -0500)]
add fr_receiver_exit() function
Alan T. DeKok [Mon, 16 Jan 2017 21:11:54 +0000 (16:11 -0500)]
fix clean up logic
Alan T. DeKok [Mon, 16 Jan 2017 21:03:35 +0000 (16:03 -0500)]
catch errors when starting worker threads.
And wait for all worker threads to start before we proceed to the next phase.
Alan T. DeKok [Mon, 16 Jan 2017 20:32:51 +0000 (15:32 -0500)]
rename so that we can have additional radius tests
Alan T. DeKok [Mon, 16 Jan 2017 15:25:49 +0000 (10:25 -0500)]
Add rule to catch BSDMake
Arran Cudbard-Bell [Fri, 13 Jan 2017 21:09:02 +0000 (21:09 +0000)]
Typo
Alan T. DeKok [Fri, 13 Jan 2017 18:49:43 +0000 (13:49 -0500)]
added minimal transport socket structure
Arran Cudbard-Bell [Thu, 12 Jan 2017 21:45:28 +0000 (21:45 +0000)]
Use RLM_SQL_NO_MORE_ROWS in more places
Arran Cudbard-Bell [Thu, 12 Jan 2017 18:13:06 +0000 (18:13 +0000)]
Use a proper rcode for no more rows
# Conflicts:
# src/modules/rlm_sql/drivers/rlm_sql_db2/rlm_sql_db2.c
# src/modules/rlm_sql/drivers/rlm_sql_firebird/rlm_sql_firebird.c
# src/modules/rlm_sql/drivers/rlm_sql_iodbc/rlm_sql_iodbc.c
# src/modules/rlm_sql/drivers/rlm_sql_postgresql/rlm_sql_postgresql.c
# src/modules/rlm_sql/drivers/rlm_sql_sqlite/rlm_sql_sqlite.c
# src/modules/rlm_sql/drivers/rlm_sql_unixodbc/rlm_sql_unixodbc.c
# src/modules/rlm_sql/rlm_sql.c
# src/modules/rlm_sql/sql.c
Matthew Newton [Thu, 12 Jan 2017 16:58:32 +0000 (16:58 +0000)]
rlm_eap: RERROR type debugs so Module-Failure-Message gets set
Alan T. DeKok [Thu, 12 Jan 2017 16:12:54 +0000 (11:12 -0500)]
travis fixes
Alan T. DeKok [Thu, 12 Jan 2017 15:39:26 +0000 (10:39 -0500)]
doxygen
Alan T. DeKok [Thu, 12 Jan 2017 15:38:30 +0000 (10:38 -0500)]
unused variable
Arran Cudbard-Bell [Thu, 12 Jan 2017 15:10:22 +0000 (15:10 +0000)]
Call finish_select_query if we experience an error retrieving the result
Alan T. DeKok [Thu, 12 Jan 2017 15:06:12 +0000 (10:06 -0500)]
whitespace
Alan T. DeKok [Thu, 12 Jan 2017 15:05:55 +0000 (10:05 -0500)]
assertions and better debugging in foreach
Alan T. DeKok [Thu, 12 Jan 2017 15:05:41 +0000 (10:05 -0500)]
don't indent again if we're resuming
Alan T. DeKok [Wed, 11 Jan 2017 21:17:27 +0000 (16:17 -0500)]
move fr_receiver_channel_message() into it's own function
Alan T. DeKok [Wed, 11 Jan 2017 20:38:43 +0000 (15:38 -0500)]
hoist code to handle channel messages into it's own function
Alan T. DeKok [Wed, 11 Jan 2017 19:56:25 +0000 (14:56 -0500)]
add ID to control messages
So that the caller can grab an arbitrary message, and then route
it to the correct handler.
Alan T. DeKok [Wed, 11 Jan 2017 19:41:48 +0000 (14:41 -0500)]
add ID to control channel, so we can separate message types
Alan T. DeKok [Wed, 11 Jan 2017 19:36:56 +0000 (14:36 -0500)]
more progress on the receiver side.
Alan T. DeKok [Wed, 11 Jan 2017 19:36:11 +0000 (14:36 -0500)]
whitespace
Alan T. DeKok [Wed, 11 Jan 2017 15:54:35 +0000 (10:54 -0500)]
typo
Alan T. DeKok [Tue, 10 Jan 2017 15:12:56 +0000 (10:12 -0500)]
use do/while instead of goto
Alan T. DeKok [Tue, 10 Jan 2017 15:11:25 +0000 (10:11 -0500)]
fr_worker_decode_request() is no longer a separate function
it's only called from one place
Alan T. DeKok [Tue, 10 Jan 2017 15:09:24 +0000 (10:09 -0500)]
always NAK messages which can't be decoded
and move fr_worker_nak()
Arran Cudbard-Bell [Tue, 10 Jan 2017 07:09:04 +0000 (02:09 -0500)]
Commit empty example.lua file to fix RPM build
Alan T. DeKok [Mon, 9 Jan 2017 19:59:20 +0000 (14:59 -0500)]
shut up broken compiler warning
Alan T. DeKok [Mon, 9 Jan 2017 19:16:31 +0000 (14:16 -0500)]
check and service the waiting_to_die list
Alan T. DeKok [Mon, 9 Jan 2017 19:00:58 +0000 (14:00 -0500)]
add preliminary code for "wait for child to die"
Still to do: check the "waiting to die" list.
Alan T. DeKok [Mon, 9 Jan 2017 18:44:24 +0000 (13:44 -0500)]
always send ACKs, even for DONE packets
Alan T. DeKok [Wed, 4 Jan 2017 22:23:16 +0000 (17:23 -0500)]
update documentation to match code behavior.
v4 now only has *filtering*, not *enforcement* operators. That is,
if "Session-Timeout = 0", doing:
update request {
Session-Timeout > 86400
}
Will *delete* the non-matching Session-Timeout. It will not *add*
or *update* the Session-Timeout to be 86400.
Which means that in v4, we should probably have a "filter" section,
instead of an "update" section, to make this behavior clearer.
Alan T. DeKok [Wed, 4 Jan 2017 22:14:45 +0000 (17:14 -0500)]
filter code in v4 works differently than v3
Alan T. DeKok [Mon, 2 Jan 2017 16:13:55 +0000 (11:13 -0500)]
update date
Alan T. DeKok [Mon, 2 Jan 2017 15:12:21 +0000 (10:12 -0500)]
document filtering operators < and >
Alan T. DeKok [Mon, 2 Jan 2017 14:11:16 +0000 (09:11 -0500)]
document != as a filtering operator
Alan T. DeKok [Mon, 2 Jan 2017 14:04:20 +0000 (09:04 -0500)]
fix documentation
Arran Cudbard-Bell [Mon, 2 Jan 2017 02:43:01 +0000 (21:43 -0500)]
Check structure types when casting from generic to specialised