]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Bjørn Mork [Mon, 1 Aug 2011 08:57:55 +0000 (10:57 +0200)]
Adding new attributes to the ERX dictionary
This should make it compatible with JUNOSe version 12.1.1
and JUNOS version 11.2.
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Alan T. DeKok [Fri, 29 Jul 2011 12:05:56 +0000 (08:05 -0400)]
Do record_minus to reset buffer, instead of just memcpy
Alan T. DeKok [Thu, 28 Jul 2011 14:25:23 +0000 (10:25 -0400)]
Manual merge of
f74583d2483d0a5f764c452788dcfc33de2bbb4b
Check cert validity
In the process of checking the OCSP response there are only checks for the
correct signed OCSP answer in the function ocsp_check()
(src/modules/rlm_eap/types/rlm_eap_tls/rlm_eap_tls.c:349).
The problem is that the current code does not check the status of the certificate.
For example if a certificate is revoked. Thus, a user with a revoked certificate
is able to bypass the verification.
Alan T. DeKok [Fri, 22 Jul 2011 12:32:00 +0000 (14:32 +0200)]
Added HUP on log rotate
Alan T. DeKok [Sat, 16 Jul 2011 12:01:30 +0000 (08:01 -0400)]
Note URL on how to create various passwords
Phil Mayers [Mon, 25 Jul 2011 14:18:12 +0000 (15:18 +0100)]
support for PCRE
Alan T. DeKok [Wed, 13 Jul 2011 12:50:41 +0000 (14:50 +0200)]
More fixes for DHCP relaying
Alan T. DeKok [Wed, 13 Jul 2011 12:50:26 +0000 (14:50 +0200)]
Allow it to send offers
Alan T. DeKok [Wed, 13 Jul 2011 12:50:08 +0000 (14:50 +0200)]
When in debugging mode, print out VPs from header
Dmitry Borodaenko [Tue, 5 Jul 2011 13:23:06 +0000 (16:23 +0300)]
Configurable file permissions in rlm_linelog
Alan T. DeKok [Sun, 10 Jul 2011 15:30:54 +0000 (17:30 +0200)]
Add connection pool API
Currently unused, but it should be nice...
Arran Cudbard-Bell [Thu, 28 Jul 2011 14:35:30 +0000 (16:35 +0200)]
Replace stale version of oracle configure script with one generated from current version of configure.in (now supports library versions 9, 10, 11 instead of just 10
Arran Cudbard-Bell [Mon, 11 Jul 2011 10:34:39 +0000 (12:34 +0200)]
Add NAI policy for use in post-proxy
Arran Cudbard-Bell [Mon, 11 Jul 2011 08:33:40 +0000 (10:33 +0200)]
Remove the last of the spaces, re-align query column lists, Change post-auth query to use SQL-User-Name
Arran Cudbard-Bell [Sat, 9 Jul 2011 21:07:28 +0000 (23:07 +0200)]
Use config value expansion instead of xlat expansion
Alan T. DeKok [Sat, 9 Jul 2011 14:18:32 +0000 (16:18 +0200)]
Allow config sections "name2" to be expanded at load time
Alan T. DeKok [Sat, 9 Jul 2011 14:03:25 +0000 (16:03 +0200)]
Replace 8 spaces by tabs
Alan T. DeKok [Sat, 9 Jul 2011 14:00:07 +0000 (16:00 +0200)]
Increase size of log buffer
Stephen R. van den Berg [Sat, 9 Jul 2011 10:42:09 +0000 (12:42 +0200)]
Get rid of some absurd warnings
Stephen R. van den Berg [Sat, 9 Jul 2011 10:38:32 +0000 (12:38 +0200)]
Update debian packaging for v3
Alan T. DeKok [Sat, 9 Jul 2011 11:46:24 +0000 (13:46 +0200)]
Fixes for v3 API
Alan T. DeKok [Sat, 9 Jul 2011 11:45:59 +0000 (13:45 +0200)]
Make UDPfromto code build
Stephen R. van den Berg [Sat, 9 Jul 2011 10:40:27 +0000 (12:40 +0200)]
DHCP compile fixes
Alan T. DeKok [Fri, 8 Jul 2011 12:38:54 +0000 (14:38 +0200)]
reset timer on receiving reply from home server
This lets the reject_delay = 1 code work.
Otherwise, the timer for the request will be left at "wait for
reply from home server", which is wrong.
There is likely still a race condition, which will be fixed when
the "request->proxy" pointer is an actual request.
Phil Mayers [Thu, 7 Jul 2011 15:39:11 +0000 (16:39 +0100)]
save all attributes in the Access-Accept when proxying EAP-MSCHAPv2 as plain MSCHAP, and restore on the final Access-Accept
Alan T. DeKok [Fri, 8 Jul 2011 08:52:58 +0000 (10:52 +0200)]
Don't enable Class by default
Arran Cudbard-Bell [Thu, 7 Jul 2011 11:33:48 +0000 (13:33 +0200)]
Fix xlat expansion of values assigned in rlm_attr_filter
Remove comparison that was generating compiler warning
Alan T. DeKok [Thu, 7 Jul 2011 10:51:07 +0000 (12:51 +0200)]
Initialize answer variable
Alan T. DeKok [Tue, 5 Jul 2011 15:54:59 +0000 (17:54 +0200)]
chown if uid or gid is set
Alan T. DeKok [Tue, 5 Jul 2011 13:28:06 +0000 (15:28 +0200)]
Allow idle_timeout to be zero
And abstract jitter command
Alan T. DeKok [Tue, 5 Jul 2011 13:26:57 +0000 (15:26 +0200)]
Fix debugging to use initialized buffer
Alan T. DeKok [Tue, 5 Jul 2011 10:42:19 +0000 (12:42 +0200)]
Don't need original packet when proxying
Alan T. DeKok [Mon, 4 Jul 2011 17:08:12 +0000 (19:08 +0200)]
Get peer id on new socket, not old one
Alan T. DeKok [Mon, 4 Jul 2011 16:09:00 +0000 (18:09 +0200)]
Set ownership of domain socket when starting
Alan T. DeKok [Mon, 4 Jul 2011 16:02:54 +0000 (18:02 +0200)]
Allow root to connect to control socket
Even if the configured "allowed UID" has a different value.
They're root, so they can do anything. We might as well be polite.
Alan T. DeKok [Mon, 4 Jul 2011 15:59:31 +0000 (17:59 +0200)]
Server closing connection returns 0
We should close our end and complain in that case.
Bug found by Brian Candler
Alan T. DeKok [Mon, 4 Jul 2011 16:13:16 +0000 (18:13 +0200)]
Use correct length
Arran Cudbard-Bell [Mon, 4 Jul 2011 08:47:04 +0000 (10:47 +0200)]
Add relax-filter check item to override the relaxed config item on a filter by filter basis
Arran Cudbard-Bell [Sun, 3 Jul 2011 17:10:59 +0000 (19:10 +0200)]
Add 'relaxed' option to rlm_attr_filter, when 'yes' attributes which do not explicitly match any filter rules are still copied.
Arran Cudbard-Bell [Sun, 3 Jul 2011 10:01:34 +0000 (12:01 +0200)]
Convert Class to string before attempting to do a pattern match
Arran Cudbard-Bell [Sun, 3 Jul 2011 09:58:01 +0000 (11:58 +0200)]
NAI policy should always return updated if the User-Name string matched the format
Alan T. DeKok [Sun, 3 Jul 2011 09:07:49 +0000 (11:07 +0200)]
Fix offset bug in %{string:...}
It prints the correct amount with the correct limits, but
to the wrong location
Alan T. DeKok [Thu, 30 Jun 2011 14:01:56 +0000 (16:01 +0200)]
Be less strict about duplicate virtual servers
If they share the same top-level CONF_SECTION, they're duplicates.
Otherwise, the server is reloading it's configuration, so the new
configuration should be allowed to be loaded.
Arran Cudbard-Bell [Fri, 1 Jul 2011 20:27:59 +0000 (22:27 +0200)]
Section should be post-auth not postauth for cui policy
Arran Cudbard-Bell [Fri, 1 Jul 2011 20:27:15 +0000 (22:27 +0200)]
Remove *all* instances of Reply-Message if there is an EAP-Message in the response
Arran Cudbard-Bell [Fri, 1 Jul 2011 13:41:51 +0000 (15:41 +0200)]
Change default table type for radacct and postauth to be INNODB explicitly. They need an engine that supports row level locking.
Arran Cudbard-Bell [Fri, 1 Jul 2011 08:02:23 +0000 (10:02 +0200)]
Fixup more formatting in polcy.conf and add policy to remove Reply-Messages if the reply also contains an EAP-Message
Arran Cudbard-Bell [Thu, 30 Jun 2011 15:44:29 +0000 (17:44 +0200)]
Add NAI (RFC 4282) User-Name splitting policy
Arran Cudbard-Bell [Thu, 30 Jun 2011 15:24:42 +0000 (17:24 +0200)]
Only use the value of class if its in our specific format
Arran Cudbard-Bell [Thu, 30 Jun 2011 14:41:05 +0000 (16:41 +0200)]
Fixup CUI policies to use the new overloading behaviour
Arran Cudbard-Bell [Thu, 30 Jun 2011 12:49:51 +0000 (14:49 +0200)]
Overload the acct_unique module with a better policy - Should fix some problems with non unique Acct-Session-IDs
Fix formatting for other policies, and add additional internal attributes
Rename some existing policies to make a clear distinction between the special <name>.<section>
policies, and ones which had previously used this convention.
Alan T. DeKok [Wed, 29 Jun 2011 09:09:07 +0000 (11:09 +0200)]
Disallow pings for TCP sockets
The existing UDP ping code assumes that any socket can send to any
server. This isn't true for TCP sockets. For them, the pings
have to be specific to each socket.
The RFC 3539 Status-Server watchdog code will fix that.
Alan T. DeKok [Tue, 28 Jun 2011 15:28:00 +0000 (17:28 +0200)]
Handle relayed packets better...
If the request a client packet, we can relay it using
the existing code.
If the request is a server packet, then it MUST be from
the real server, and we MUST be acting as a relay. In that
case, set the giaddr to 0.0.0.0, and forward the packet to the
yiaddr.
And do something with broadcast replies...
Alan T. DeKok [Tue, 28 Jun 2011 13:54:12 +0000 (15:54 +0200)]
Allow DHCP-Opcode and DHCP-Hop-Count to be set from VPs.
This makes it easier to relay && respond to clients
Alan T. DeKok [Tue, 28 Jun 2011 13:38:33 +0000 (15:38 +0200)]
Allow giaddr to be updated when relaying
Alan T. DeKok [Tue, 28 Jun 2011 08:57:11 +0000 (10:57 +0200)]
Complain about TLS only if there's a TLS subsection
Petr Uzel [Mon, 27 Jun 2011 07:21:18 +0000 (09:21 +0200)]
rlm_mschap: silence gcc buffer overflow detection mechanism
Signed-off-by: Petr Uzel <petr.uzel@suse.cz>
Alan T. DeKok [Mon, 27 Jun 2011 12:52:54 +0000 (14:52 +0200)]
Home servers are unique by proto, too.
Check for duplicates, and print errors if there are duplicates
Alan T. DeKok [Mon, 27 Jun 2011 11:04:46 +0000 (13:04 +0200)]
Fix calculation of response authenticator
The Status-Server packet can get an Accounting-Response
packet in return. Since the Status-Server has a random
authentication vector, the response needs to be calculated
using that. We can't use the normal Accounting-Response
calculation.
Oops. No one found this in RFC 5997.
Alan T. DeKok [Mon, 27 Jun 2011 09:19:29 +0000 (11:19 +0200)]
Fix > vs >= bug
Phil Mayers [Fri, 24 Jun 2011 14:46:18 +0000 (15:46 +0100)]
examples of Doxygen on the source base
Alan T. DeKok [Fri, 24 Jun 2011 10:41:17 +0000 (12:41 +0200)]
fclose() frees buffers, too
Alan T. DeKok [Tue, 21 Jun 2011 12:17:19 +0000 (14:17 +0200)]
Simplify TCP socket timers.
All timers are now handled by one function, which makes it easier
to understand.
Alan T. DeKok [Tue, 21 Jun 2011 10:28:33 +0000 (12:28 +0200)]
TCP connections don't ping home servers.
Instead, we should send pings per CONNECTION, not per SERVER.
Alan T. DeKok [Tue, 21 Jun 2011 10:26:34 +0000 (12:26 +0200)]
Program name isn't available here
Will fix it later
Alan T. DeKok [Tue, 21 Jun 2011 09:23:56 +0000 (11:23 +0200)]
If a child process gets a signal to exit, then just exit.
Alan T. DeKok [Tue, 21 Jun 2011 09:22:36 +0000 (11:22 +0200)]
Print out *which* program is causing the delay
Alan T. DeKok [Tue, 21 Jun 2011 07:06:38 +0000 (09:06 +0200)]
Fix > vs >= bug
Alan T. DeKok [Mon, 20 Jun 2011 14:57:14 +0000 (16:57 +0200)]
Don't go too far ahead
if (..){
is OK. The previous code skipped over the curly brace, assuming
that it was there... the code to check for syntax errors assumed
that the curly brace was not skipped over. This change fixes
that conflict
Alan T. DeKok [Sun, 19 Jun 2011 10:02:41 +0000 (12:02 +0200)]
Start of work to remove need for libtool && libltdl
For now, we rely on gcc to do most of the heavy lifting for us.
This hinders portability, but people can always fall back to
libtool.
Alan T. DeKok [Sun, 19 Jun 2011 08:37:44 +0000 (10:37 +0200)]
More hacks to make dlopen() work natively
Alan T. DeKok [Thu, 16 Jun 2011 10:53:37 +0000 (12:53 +0200)]
Be more stringent about unexpected text
Alan T. DeKok [Wed, 15 Jun 2011 09:39:54 +0000 (11:39 +0200)]
Releases don't get replied to
Alan T. DeKok [Wed, 15 Jun 2011 08:49:58 +0000 (10:49 +0200)]
Document "max_outstanding" for home servers
Alan T. DeKok [Wed, 15 Jun 2011 08:34:21 +0000 (10:34 +0200)]
Look inside of virtuals for homes
This allows home servers to be tied to virtuals
Alan T. DeKok [Wed, 15 Jun 2011 08:31:31 +0000 (10:31 +0200)]
Tie virtual servers to home servers
If a home is defined in a virtual, proxying from that
virtual will use globally defined homes, OR homes specific
to that virtual
Alan T. DeKok [Wed, 15 Jun 2011 08:21:51 +0000 (10:21 +0200)]
Track parent virtual server in home server.
This will eventually let home servers be defined for each
virtual server. Right now, it doesn't yet work.
Alan T. DeKok [Wed, 15 Jun 2011 08:00:29 +0000 (10:00 +0200)]
Make home servers unique by source IP, too
Alan T. DeKok [Tue, 14 Jun 2011 06:10:28 +0000 (08:10 +0200)]
Allow policies by section
authorize {
foo
}
will look for first:
policy {
foo.authorize {
...
}
}
and then
policy {
foo {
...
}
}
This allows section-specific overrides for policies and modules.
Alan T. DeKok [Mon, 13 Jun 2011 09:31:47 +0000 (11:31 +0200)]
Allow policies to refer to modules of the same name
policy {
files {
files
...
}
}
Means that you can over-ride the behavior of the "files" module,
and add anything else you need.
Alan T. DeKok [Sun, 12 Jun 2011 04:41:57 +0000 (06:41 +0200)]
New modules && configs for v2.1.11
Alan T. DeKok [Fri, 10 Jun 2011 13:17:45 +0000 (15:17 +0200)]
Convert BEGIN-TLV to tlv.attr
Alan T. DeKok [Fri, 10 Jun 2011 13:16:16 +0000 (15:16 +0200)]
Fix data types
Alan T. DeKok [Fri, 10 Jun 2011 12:52:57 +0000 (14:52 +0200)]
Make "cadir" and "certdir" globals
Alan T. DeKok [Wed, 8 Jun 2011 16:07:35 +0000 (18:07 +0200)]
Remove unnecessary assertion
Alan T. DeKok [Tue, 7 Jun 2011 21:30:31 +0000 (23:30 +0200)]
Fix build issues
Alan T. DeKok [Tue, 7 Jun 2011 16:51:48 +0000 (18:51 +0200)]
Make the certificates on initial install
So that the server doesn't block when it first starts up
Alan T. DeKok [Tue, 7 Jun 2011 16:51:22 +0000 (18:51 +0200)]
Don't look for eap.conf, etc.
they've been moved to the "modules" directory
Alan T. DeKok [Tue, 7 Jun 2011 09:18:44 +0000 (11:18 +0200)]
Fix typo. Closes bug #150
Alan T. DeKok [Sun, 5 Jun 2011 16:15:05 +0000 (18:15 +0200)]
Note existence of "inner-tunnel"
So that people remember to edit it, too
Alan T. DeKok [Fri, 3 Jun 2011 08:27:11 +0000 (10:27 +0200)]
Split request_receive() into two functions.
One is for sockets, and enforces limits, duplicates, etc.
The other allocates the request and inserts it into the queue,
independent of limits. This is called directly by the detail
file reader
Alan T. DeKok [Fri, 3 Jun 2011 08:23:48 +0000 (10:23 +0200)]
Make RADCLIENT always have stats structures
It's simpler.
Alan T. DeKok [Fri, 3 Jun 2011 08:19:58 +0000 (10:19 +0200)]
Removed unused prototypes
Alan T. DeKok [Fri, 3 Jun 2011 08:19:01 +0000 (10:19 +0200)]
Fix GCD code for new process functionality
Alan T. DeKok [Fri, 3 Jun 2011 08:08:58 +0000 (10:08 +0200)]
Updated year
Alan T. DeKok [Tue, 31 May 2011 19:21:23 +0000 (21:21 +0200)]
Added MS-CHAP-V2
Alan T. DeKok [Tue, 31 May 2011 18:39:42 +0000 (20:39 +0200)]
Last set of vp_print fixes
Alan T. DeKok [Tue, 31 May 2011 11:20:12 +0000 (13:20 +0200)]
Move \t into vp_print, just like the last commit
Alan T. DeKok [Tue, 31 May 2011 07:32:20 +0000 (09:32 +0200)]
vp_print should add a "\n" to the end
All callers already do this, so it's best to move that code
into vp_print
Alan T. DeKok [Mon, 30 May 2011 15:14:18 +0000 (17:14 +0200)]
Revert most of the "checked_write" code.
It apparently caused crashes on some machines. This code
reverts (mostly) back to the original code which worked, but
it should also notice when the disk is full, and return FAIL