]> git.ipfire.org Git - thirdparty/freeradius-server.git/log
thirdparty/freeradius-server.git
6 years agospecify name for pkey constraint and use it in query for cui postgresql 3298/head
Jakub Jirutka [Thu, 20 Feb 2020 11:12:09 +0000 (12:12 +0100)] 
specify name for pkey constraint and use it in query for cui postgresql

6 years agouse real upsert (ON CONFLICT) instead of workaround in postgresql cui
Jakub Jirutka [Thu, 20 Feb 2020 00:30:04 +0000 (01:30 +0100)] 
use real upsert (ON CONFLICT) instead of workaround in postgresql cui

Clause ON CONFLICT performs better than the RULE workaround and it's
guaranted to be atomic (the workaround is not).

It requires PostgreSQL 9.5 or newer; this version was released in 2016
and the previous versions are already unsupported, so it shouldn't be
a problem for anyone using latest FreeRADIUS versions.

There's a problem with backward compatibility because ON CONFLICT
doesn't work when there's a RULE ON INSERT defined on the table. It
results in the following error in radius.log:

    Error: rlm_sql_postgresql: 0A000: FEATURE NOT SUPPORTED

6 years agofix postgresql schema for cui
Jakub Jirutka [Thu, 20 Feb 2020 00:12:18 +0000 (01:12 +0100)] 
fix postgresql schema for cui

I'm getting the following error with the current schema:

    rlm_sql_postgresql: 22001: STRING DATA RIGHT TRUNCATION

Moreover, in PostgreSQL the text type is highly preferred above
varchar. Schema for radacct already uses it, so it seems that schema
for cui is just outdated.

I also remove quotes around `now()`, it was probably some archaism.

6 years agofix formatting in cui postgresql schema to be consistent
Jakub Jirutka [Thu, 20 Feb 2020 09:30:50 +0000 (10:30 +0100)] 
fix formatting in cui postgresql schema to be consistent

6 years agouse usec since current second, not usec since epoch
Alan T. DeKok [Fri, 28 Feb 2020 13:10:33 +0000 (08:10 -0500)] 
use usec since current second, not usec since epoch

6 years agofix crash when no previous regex seen
Matthew Newton [Thu, 27 Feb 2020 13:54:59 +0000 (13:54 +0000)] 
fix crash when no previous regex seen

6 years agodoxygen
Matthew Newton [Thu, 27 Feb 2020 13:54:46 +0000 (13:54 +0000)] 
doxygen

6 years agoRemove vp_cursor_t from rlm_radius_udp.c
Arran Cudbard-Bell [Thu, 27 Feb 2020 22:12:39 +0000 (16:12 -0600)] 
Remove vp_cursor_t from rlm_radius_udp.c

6 years agos/pair_cursor/fr_cursor/ in rlm_isc_dhcp.c
Arran Cudbard-Bell [Thu, 27 Feb 2020 22:10:29 +0000 (16:10 -0600)] 
s/pair_cursor/fr_cursor/ in rlm_isc_dhcp.c

6 years agoMore reasonable default
Arran Cudbard-Bell [Thu, 27 Feb 2020 22:06:18 +0000 (16:06 -0600)] 
More reasonable default

6 years agoAdd more log prefixes. Switch some statements to debug/info.
Arran Cudbard-Bell [Thu, 27 Feb 2020 20:42:03 +0000 (14:42 -0600)] 
Add more log prefixes.  Switch some statements to debug/info.

We really do want to see rechability messages in the main server log.

6 years agoTypo
Arran Cudbard-Bell [Thu, 27 Feb 2020 20:25:22 +0000 (14:25 -0600)] 
Typo

6 years agoRemove duplicate track delete call
Arran Cudbard-Bell [Thu, 27 Feb 2020 20:25:15 +0000 (14:25 -0600)] 
Remove duplicate track delete call

6 years agoAssert treq is still assigned to a connection on timeout
Arran Cudbard-Bell [Thu, 27 Feb 2020 20:23:04 +0000 (14:23 -0600)] 
Assert treq is still assigned to a connection on timeout

6 years agoPass in tconn to status_check_next
Arran Cudbard-Bell [Thu, 27 Feb 2020 20:11:24 +0000 (14:11 -0600)] 
Pass in tconn to status_check_next

The timer is now parented of h, so this should all be ok...

6 years agoSignalling request is complete frees everything parented by treq, so probably want...
Arran Cudbard-Bell [Thu, 27 Feb 2020 19:48:30 +0000 (13:48 -0600)] 
Signalling request is complete frees everything parented by treq, so probably want to do it after we're done messing around with it

6 years agoRemove udp_connection_t;
Arran Cudbard-Bell [Thu, 27 Feb 2020 19:36:28 +0000 (13:36 -0600)] 
Remove udp_connection_t;

6 years agouse more standard names "packet" and "reply"
Alan T. DeKok [Thu, 27 Feb 2020 18:43:00 +0000 (13:43 -0500)] 
use more standard names "packet" and "reply"

6 years agouse more standard name "packet"
Alan T. DeKok [Thu, 27 Feb 2020 16:55:58 +0000 (11:55 -0500)] 
use more standard name "packet"

6 years agouse more standard name "packet" and "reply"
Alan T. DeKok [Thu, 27 Feb 2020 16:46:42 +0000 (11:46 -0500)] 
use more standard name "packet" and "reply"

in preparation for moving to type "group" for lists.

6 years agouse more standard name "packet"
Alan T. DeKok [Thu, 27 Feb 2020 16:46:19 +0000 (11:46 -0500)] 
use more standard name "packet"

in preparation for moving to type "group" for lists.

6 years agouse more standard name "packet" instead of "current"
Alan T. DeKok [Thu, 27 Feb 2020 16:45:18 +0000 (11:45 -0500)] 
use more standard name "packet" instead of "current"

in preparation for moving to type "group" for lists.

6 years agofix crash on invalid attribute
Matthew Newton [Thu, 27 Feb 2020 12:38:50 +0000 (12:38 +0000)] 
fix crash on invalid attribute

6 years agouse path count
Alan T. DeKok [Wed, 26 Feb 2020 21:26:25 +0000 (16:26 -0500)] 
use path count

6 years agoremove "set fd handlers" hack now that status checks are in the trunk
Alan T. DeKok [Wed, 26 Feb 2020 21:24:36 +0000 (16:24 -0500)] 
remove "set fd handlers" hack now that status checks are in the trunk

6 years agoPast tense as it's after sendmmsg
Arran Cudbard-Bell [Wed, 26 Feb 2020 21:31:31 +0000 (16:31 -0500)] 
Past tense as it's after sendmmsg

6 years agoNot needed, already NULL
Arran Cudbard-Bell [Wed, 26 Feb 2020 21:21:42 +0000 (16:21 -0500)] 
Not needed, already NULL

6 years agoUse sendmmsg in rlm_radius_udp.c
Arran Cudbard-Bell [Wed, 26 Feb 2020 20:52:37 +0000 (15:52 -0500)] 
Use sendmmsg in rlm_radius_udp.c

6 years agoFail requests which can't be requeued
Arran Cudbard-Bell [Wed, 26 Feb 2020 20:52:21 +0000 (15:52 -0500)] 
Fail requests which can't be requeued

6 years agoadd support for glob. Fixes #1018
Alan T. DeKok [Wed, 26 Feb 2020 19:51:13 +0000 (14:51 -0500)] 
add support for glob.  Fixes #1018

This needs documentation added, too.

6 years agomove FAILED to timeout function
Alan T. DeKok [Wed, 26 Feb 2020 18:18:05 +0000 (13:18 -0500)] 
move FAILED to timeout function

6 years agodoxygen (xlat functions)
Matthew Newton [Wed, 26 Feb 2020 16:12:43 +0000 (16:12 +0000)] 
doxygen (xlat functions)

6 years agomove status checks to using the trunk API
Alan T. DeKok [Wed, 26 Feb 2020 16:21:50 +0000 (11:21 -0500)] 
move status checks to using the trunk API

6 years agoprioritize status checks
Alan T. DeKok [Tue, 25 Feb 2020 14:19:55 +0000 (09:19 -0500)] 
prioritize status checks

in preparation for making them part of the trunk

6 years agoFunction names and param order
Matthew Newton [Wed, 26 Feb 2020 01:48:03 +0000 (01:48 +0000)] 
Function names and param order

6 years agoxlat doxygen
Matthew Newton [Wed, 26 Feb 2020 01:38:55 +0000 (01:38 +0000)] 
xlat doxygen

6 years agoRe-order xlat_builtin functions, whitespace
Matthew Newton [Sun, 23 Feb 2020 23:16:54 +0000 (23:16 +0000)] 
Re-order xlat_builtin functions, whitespace

6 years agoRemove invalid assert
Arran Cudbard-Bell [Tue, 25 Feb 2020 22:25:31 +0000 (17:25 -0500)] 
Remove invalid assert

6 years agoMove closefrom into missing
Arran Cudbard-Bell [Tue, 25 Feb 2020 22:03:31 +0000 (17:03 -0500)] 
Move closefrom into missing

6 years agoAdd crappy userland sendmmsg implementation
Arran Cudbard-Bell [Tue, 25 Feb 2020 22:03:15 +0000 (17:03 -0500)] 
Add crappy userland sendmmsg implementation

6 years agoExplicitly free rctx on cancel
Arran Cudbard-Bell [Tue, 25 Feb 2020 04:39:29 +0000 (23:39 -0500)] 
Explicitly free rctx on cancel

6 years agoRe-order more functions, and explicitly clean up more things
Arran Cudbard-Bell [Tue, 25 Feb 2020 04:34:48 +0000 (23:34 -0500)] 
Re-order more functions, and explicitly clean up more things

6 years agoRe-order functions
Arran Cudbard-Bell [Tue, 25 Feb 2020 03:27:23 +0000 (22:27 -0500)] 
Re-order functions

6 years agoAllow laziness on the part of the API client, and free any treq children
Arran Cudbard-Bell [Tue, 25 Feb 2020 03:27:15 +0000 (22:27 -0500)] 
Allow laziness on the part of the API client, and free any treq children

6 years agoset treq
Arran Cudbard-Bell [Tue, 25 Feb 2020 02:59:18 +0000 (21:59 -0500)] 
set treq

6 years agoAlways populate preq and rctx
Arran Cudbard-Bell [Tue, 25 Feb 2020 02:52:56 +0000 (21:52 -0500)] 
Always populate preq and rctx

6 years agoMore asserts
Arran Cudbard-Bell [Tue, 25 Feb 2020 02:41:34 +0000 (21:41 -0500)] 
More asserts

6 years agoReturn the treq to the free list
Arran Cudbard-Bell [Tue, 25 Feb 2020 02:37:06 +0000 (21:37 -0500)] 
Return the treq to the free list

6 years agoSimplify populating treq_out
Arran Cudbard-Bell [Tue, 25 Feb 2020 02:30:11 +0000 (21:30 -0500)] 
Simplify populating treq_out

6 years agoHandle isn't available if the connection failed from the init state
Arran Cudbard-Bell [Tue, 25 Feb 2020 01:52:42 +0000 (20:52 -0500)] 
Handle isn't available if the connection failed from the init state

6 years agoAssert that the treq is a member of a trunk
Arran Cudbard-Bell [Tue, 25 Feb 2020 00:17:01 +0000 (19:17 -0500)] 
Assert that the treq is a member of a trunk

6 years agoAlways delete the request timeout
Arran Cudbard-Bell [Mon, 24 Feb 2020 23:40:52 +0000 (18:40 -0500)] 
Always delete the request timeout

6 years agoMore agressive defaults
Arran Cudbard-Bell [Mon, 24 Feb 2020 23:34:43 +0000 (18:34 -0500)] 
More agressive defaults

6 years agoCleanup failed requests...
Arran Cudbard-Bell [Mon, 24 Feb 2020 23:22:55 +0000 (18:22 -0500)] 
Cleanup failed requests...

6 years agoFix whitespace issues
Arran Cudbard-Bell [Mon, 24 Feb 2020 23:00:44 +0000 (18:00 -0500)] 
Fix whitespace issues

6 years agoNot an assert, a useful hint...
Arran Cudbard-Bell [Mon, 24 Feb 2020 22:59:06 +0000 (17:59 -0500)] 
Not an assert, a useful hint...

6 years agoFix assert
Arran Cudbard-Bell [Mon, 24 Feb 2020 22:56:22 +0000 (17:56 -0500)] 
Fix assert

6 years agoOnly apply last_above_target delay if we have at least one active connection
Arran Cudbard-Bell [Mon, 24 Feb 2020 22:56:13 +0000 (17:56 -0500)] 
Only apply last_above_target delay if we have at least one active connection

6 years agoRecord reconnections (not due to expiry) as failures too
Arran Cudbard-Bell [Mon, 24 Feb 2020 22:55:20 +0000 (17:55 -0500)] 
Record reconnections (not due to expiry) as failures too

6 years agoquotes around names...
Arran Cudbard-Bell [Mon, 24 Feb 2020 22:17:40 +0000 (17:17 -0500)] 
quotes around names...

6 years agoOnly start the response timer if we manage to write out a packet
Arran Cudbard-Bell [Mon, 24 Feb 2020 22:11:24 +0000 (17:11 -0500)] 
Only start the response timer if we manage to write out a packet

6 years agoAgressively reconnect connections on write errors
Arran Cudbard-Bell [Mon, 24 Feb 2020 22:02:43 +0000 (17:02 -0500)] 
Agressively reconnect connections on write errors

6 years agoRemove icky O(n) udata hack
Arran Cudbard-Bell [Mon, 24 Feb 2020 21:39:50 +0000 (16:39 -0500)] 
Remove icky O(n) udata hack

6 years agoUse 'noop' callback for unset function pointers. Fixes SEGV when filters are swapped...
Arran Cudbard-Bell [Mon, 24 Feb 2020 21:36:40 +0000 (16:36 -0500)] 
Use 'noop' callback for unset function pointers.  Fixes SEGV when filters are swapped out between corral and service.

6 years agoMinor cleanup in event.c
Arran Cudbard-Bell [Mon, 24 Feb 2020 21:01:57 +0000 (16:01 -0500)] 
Minor cleanup in event.c

6 years agoclean up status checks
Alan T. DeKok [Mon, 24 Feb 2020 21:31:46 +0000 (16:31 -0500)] 
clean up status checks

6 years agoset 'verify=lib' in test dictionary for RADIUS
Alan T. DeKok [Mon, 24 Feb 2020 21:10:00 +0000 (16:10 -0500)] 
set 'verify=lib' in test dictionary for RADIUS

and change check order for loading protocol libraries.
Try to load them, even if they're not required.  And error out
only if they fail to load AND they're required.

This change makes it a bit more forgiving.

6 years agofree module_name on error
Alan T. DeKok [Mon, 24 Feb 2020 21:09:36 +0000 (16:09 -0500)] 
free module_name on error

6 years agoreplicate now takes a different path
Alan T. DeKok [Mon, 24 Feb 2020 15:54:53 +0000 (10:54 -0500)] 
replicate now takes a different path

6 years agoAlways set te->tt
Arran Cudbard-Bell [Mon, 24 Feb 2020 19:25:59 +0000 (14:25 -0500)] 
Always set te->tt

6 years agoNot reading from the socket is an error...
Arran Cudbard-Bell [Mon, 24 Feb 2020 19:25:45 +0000 (14:25 -0500)] 
Not reading from the socket is an error...

6 years agoFix time travel assert
Arran Cudbard-Bell [Mon, 24 Feb 2020 19:25:33 +0000 (14:25 -0500)] 
Fix time travel assert

6 years agomake the input file optional in the load generation module
Arran Cudbard-Bell [Mon, 24 Feb 2020 19:15:24 +0000 (14:15 -0500)] 
make the input file optional in the load generation module

6 years agoPrint better error when we can't find the dictionary validation libraries
Arran Cudbard-Bell [Mon, 24 Feb 2020 19:03:34 +0000 (14:03 -0500)] 
Print better error when we can't find the dictionary validation libraries

6 years agoReduce possibility of screwups
Arran Cudbard-Bell [Mon, 24 Feb 2020 18:49:09 +0000 (13:49 -0500)] 
Reduce possibility of screwups

6 years agoset h->c by rearranging structs and members
Alan T. DeKok [Mon, 24 Feb 2020 15:42:46 +0000 (10:42 -0500)] 
set h->c by rearranging structs and members

6 years agoSignal as sent first in replicate mode
Arran Cudbard-Bell [Mon, 24 Feb 2020 15:33:34 +0000 (10:33 -0500)] 
Signal as sent first in replicate mode

6 years agopass udp_thread_t instead of other ctx
Alan T. DeKok [Mon, 24 Feb 2020 15:21:14 +0000 (10:21 -0500)] 
pass udp_thread_t instead of other ctx

6 years agoset recv_buff = 0 for replication
Alan T. DeKok [Mon, 24 Feb 2020 14:14:46 +0000 (09:14 -0500)] 
set recv_buff = 0 for replication

6 years agofix retransmission logic
Alan T. DeKok [Mon, 24 Feb 2020 14:11:11 +0000 (09:11 -0500)] 
fix retransmission logic

6 years agoremove unnecessary intermediate variables
Alan T. DeKok [Mon, 24 Feb 2020 13:55:02 +0000 (08:55 -0500)] 
remove unnecessary intermediate variables

6 years agodefine RADIUS_AUTH_VECTOR_OFFSET
Alan T. DeKok [Mon, 24 Feb 2020 13:54:39 +0000 (08:54 -0500)] 
define RADIUS_AUTH_VECTOR_OFFSET

6 years agodrain the socket when we don't care about replies
Alan T. DeKok [Mon, 24 Feb 2020 13:44:05 +0000 (08:44 -0500)] 
drain the socket when we don't care about replies

6 years agoif we're replicating, still read packets, but don't write them
Alan T. DeKok [Mon, 24 Feb 2020 12:47:51 +0000 (07:47 -0500)] 
if we're replicating, still read packets, but don't write them

6 years agostatus checks don't need to be Status-Server
Alan T. DeKok [Mon, 24 Feb 2020 12:45:10 +0000 (07:45 -0500)] 
status checks don't need to be Status-Server

6 years agodon't print out signed numbers for hex data
Alan T. DeKok [Sun, 23 Feb 2020 13:57:01 +0000 (14:57 +0100)] 
don't print out signed numbers for hex data

6 years agoDo significantly less work on replicate
Arran Cudbard-Bell [Mon, 24 Feb 2020 03:26:23 +0000 (22:26 -0500)] 
Do significantly less work on replicate

6 years agoMove things not to do with encoding out of encode
Arran Cudbard-Bell [Mon, 24 Feb 2020 02:41:57 +0000 (21:41 -0500)] 
Move things not to do with encoding out of encode

6 years agoFewer magic numbers
Arran Cudbard-Bell [Mon, 24 Feb 2020 02:11:27 +0000 (21:11 -0500)] 
Fewer magic numbers

6 years agoAlways use per-packet send buffers
Arran Cudbard-Bell [Mon, 24 Feb 2020 00:06:24 +0000 (19:06 -0500)] 
Always use per-packet send buffers

Sacrifice memory for ability to call sendmmsg and friends

6 years agoUse saner names for radius tracking tree and structures
Arran Cudbard-Bell [Sun, 23 Feb 2020 23:13:15 +0000 (18:13 -0500)] 
Use saner names for radius tracking tree and structures

6 years agos/request_io_ctx/rctx/
Arran Cudbard-Bell [Sun, 23 Feb 2020 22:54:52 +0000 (17:54 -0500)] 
s/request_io_ctx/rctx/

6 years agoLess ambiguous name
Arran Cudbard-Bell [Sun, 23 Feb 2020 22:47:53 +0000 (17:47 -0500)] 
Less ambiguous name

6 years agoSwitch to single letter for result struct
Arran Cudbard-Bell [Sun, 23 Feb 2020 22:42:29 +0000 (17:42 -0500)] 
Switch to single letter for result struct

6 years agoHave the submodule allocate the rctx
Arran Cudbard-Bell [Sun, 23 Feb 2020 22:20:45 +0000 (17:20 -0500)] 
Have the submodule allocate the rctx

6 years agoPass "cancel" signal through to submodule
Arran Cudbard-Bell [Sun, 23 Feb 2020 22:20:14 +0000 (17:20 -0500)] 
Pass "cancel" signal through to submodule

Remove duplicate check for synchronous behaviour

6 years agoUse the standard signal function signature instead of creating a new one with a sligh...
Arran Cudbard-Bell [Sun, 23 Feb 2020 22:18:29 +0000 (17:18 -0500)] 
Use the standard signal function signature instead of creating a new one with a slightly different argument order

6 years agoReally need a common API for this...
Arran Cudbard-Bell [Sun, 23 Feb 2020 21:25:38 +0000 (16:25 -0500)] 
Really need a common API for this...

6 years agoDon't do extra work when the module API does it for us...
Arran Cudbard-Bell [Sun, 23 Feb 2020 21:24:24 +0000 (16:24 -0500)] 
Don't do extra work when the module API does it for us...

6 years agorlm_radius doesn't need to cache el, it has no timers or I/O events
Arran Cudbard-Bell [Sun, 23 Feb 2020 21:22:45 +0000 (16:22 -0500)] 
rlm_radius doesn't need to cache el, it has no timers or I/O events