]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Alan T. DeKok [Sat, 12 Aug 2017 07:56:59 +0000 (09:56 +0200)]
move to debug macros
Alan T. DeKok [Sat, 12 Aug 2017 06:54:46 +0000 (08:54 +0200)]
adde log level to worker
Alan T. DeKok [Sat, 12 Aug 2017 06:48:58 +0000 (08:48 +0200)]
add log level to scheduler
Alan T. DeKok [Fri, 11 Aug 2017 15:54:15 +0000 (17:54 +0200)]
added libjson-c3
# Conflicts:
# debian/control.in
Alan T. DeKok [Fri, 11 Aug 2017 15:09:11 +0000 (17:09 +0200)]
better scan fixes
Alan T. DeKok [Fri, 11 Aug 2017 14:50:30 +0000 (16:50 +0200)]
remove old todo's
Alan T. DeKok [Fri, 11 Aug 2017 14:41:13 +0000 (16:41 +0200)]
scan fixes
Alan T. DeKok [Fri, 11 Aug 2017 14:30:36 +0000 (16:30 +0200)]
more DEBUG3 messages
Alan T. DeKok [Fri, 11 Aug 2017 14:27:41 +0000 (16:27 +0200)]
don't put timestamps into src_dst.
It means that every packet is unique <sigh>
Alan T. DeKok [Fri, 11 Aug 2017 13:48:03 +0000 (15:48 +0200)]
cleaner debug output
Alan T. DeKok [Fri, 11 Aug 2017 13:15:37 +0000 (15:15 +0200)]
allocate status_check rr when we open the socket
not when we initialize the module
Alan T. DeKok [Fri, 11 Aug 2017 13:14:33 +0000 (15:14 +0200)]
print out raw packets we send in debug3
Alan T. DeKok [Fri, 11 Aug 2017 13:14:13 +0000 (15:14 +0200)]
check the correct fields, so we encode the packet properly...
Alan T. DeKok [Fri, 11 Aug 2017 13:13:51 +0000 (15:13 +0200)]
remove idle timeout if we're writing to the socket
Alan T. DeKok [Fri, 11 Aug 2017 13:13:12 +0000 (15:13 +0200)]
added TIMER to debug messages so we know they're timeouts
we should really register names associated with timers, and
have the event list print this...
Alan T. DeKok [Fri, 11 Aug 2017 13:08:13 +0000 (15:08 +0200)]
move status_check to correct place
Alan T. DeKok [Fri, 11 Aug 2017 12:58:05 +0000 (14:58 +0200)]
print raw hex data in debug3
Alan T. DeKok [Fri, 11 Aug 2017 12:56:49 +0000 (14:56 +0200)]
added new fr_radius_print_hex() which prints raw buffers
Alan T. DeKok [Fri, 11 Aug 2017 12:41:48 +0000 (14:41 +0200)]
rename fr_radius_print_hex to fr_radius_packet_print_hex
Alan T. DeKok [Fri, 11 Aug 2017 12:39:05 +0000 (14:39 +0200)]
don't print TACACS+ packets as RADIUS... they're not.
Alan T. DeKok [Fri, 11 Aug 2017 07:58:48 +0000 (09:58 +0200)]
rename ev to idle_ev for clarity
Alan T. DeKok [Thu, 10 Aug 2017 16:33:13 +0000 (18:33 +0200)]
clang scan issues
Alan T. DeKok [Thu, 10 Aug 2017 16:28:16 +0000 (18:28 +0200)]
remove unnecessary assertion
Alan T. DeKok [Thu, 10 Aug 2017 16:02:38 +0000 (18:02 +0200)]
stupid scan fixes... again
Alan T. DeKok [Thu, 10 Aug 2017 15:52:28 +0000 (17:52 +0200)]
update Event-Timestamp in Status-Server messages
Arran Cudbard-Bell [Thu, 10 Aug 2017 13:28:24 +0000 (15:28 +0200)]
Remove duplicate setting of attr
Check there’s a reply packet?
Arran Cudbard-Bell [Thu, 10 Aug 2017 13:04:30 +0000 (15:04 +0200)]
Temporarily mark signal interval as unused
Arran Cudbard-Bell [Thu, 10 Aug 2017 12:03:33 +0000 (14:03 +0200)]
Disable failing test until there’s time to root cause
Alan T. DeKok [Thu, 10 Aug 2017 11:20:27 +0000 (13:20 +0200)]
notes on Acct-Delay-Time and retransmits
Alan T. DeKok [Thu, 10 Aug 2017 11:18:42 +0000 (13:18 +0200)]
Status-Server checks have unique timeout handlers
so we can re-create and re-sign the packets each time.
Alan T. DeKok [Thu, 10 Aug 2017 11:17:57 +0000 (13:17 +0200)]
remove Status-Server checks on connection error
We probably also want to stop all packet retransmissions,
timers, etc. But it's not (yet) clear exactly what to do
Alan T. DeKok [Thu, 10 Aug 2017 09:47:10 +0000 (11:47 +0200)]
lots of little tweaks to get Status-Server checks working
It doesn't yet send the packets, but it does create them,
their contents, and clean up the associated timers correctly.
Alan T. DeKok [Thu, 10 Aug 2017 09:21:02 +0000 (11:21 +0200)]
most recent notes
Alan T. DeKok [Thu, 10 Aug 2017 09:15:53 +0000 (11:15 +0200)]
it doesn't make sense to replicate and have status checks
Alan T. DeKok [Thu, 10 Aug 2017 09:01:56 +0000 (11:01 +0200)]
Status-Server require Message-Authenticator, too
and not Proxy-State
Alan T. DeKok [Thu, 10 Aug 2017 08:42:37 +0000 (10:42 +0200)]
remove retransmit timers when a connection closes
Alan T. DeKok [Thu, 10 Aug 2017 08:41:38 +0000 (10:41 +0200)]
start of Status-Server checks
Alan T. DeKok [Thu, 10 Aug 2017 06:58:09 +0000 (08:58 +0200)]
Start of status_check code.
it parses 'status_check = ...' and does basic sanity checks
Alan T. DeKok [Tue, 8 Aug 2017 18:58:22 +0000 (20:58 +0200)]
comment out "skip signal" for now
it works in debug mode, as we're single threaded...
in daemon mode, skipping signals when there's no "next" packet
means that it just doesn't respond
We should fix the "skip signal" code to skip the signal only
if the other end hasn't ACKed our last signal. Because when
it drains the queue, it notices that the last packet wasn't
ACKed, and sends an ACK
Alan T. DeKok [Tue, 8 Aug 2017 18:56:22 +0000 (20:56 +0200)]
preliminary implementation of zombie_period
Alan T. DeKok [Tue, 8 Aug 2017 18:54:25 +0000 (20:54 +0200)]
reset idle timeouts when we read a packet.
which means (usually) deleting them
Alan T. DeKok [Tue, 8 Aug 2017 18:54:01 +0000 (20:54 +0200)]
sanitize names
Alan T. DeKok [Tue, 8 Aug 2017 18:53:26 +0000 (20:53 +0200)]
add zombie_period
Alan T. DeKok [Tue, 8 Aug 2017 18:38:51 +0000 (20:38 +0200)]
it helps to call write() on retransmit
Alan T. DeKok [Tue, 8 Aug 2017 08:29:12 +0000 (10:29 +0200)]
timer events are events, too
Alan T. DeKok [Mon, 7 Aug 2017 07:40:54 +0000 (09:40 +0200)]
update documentation
Alan T. DeKok [Sun, 6 Aug 2017 16:27:47 +0000 (18:27 +0200)]
check for possible proxy loops
Alan T. DeKok [Sun, 6 Aug 2017 16:26:36 +0000 (18:26 +0200)]
clean up debug messages
Alan T. DeKok [Sun, 6 Aug 2017 16:20:10 +0000 (18:20 +0200)]
move proxy_state random number to main module
Alan T. DeKok [Sun, 6 Aug 2017 16:16:25 +0000 (18:16 +0200)]
print out Proxy-State and Message-Authenticator when proxying
Alan T. DeKok [Sun, 6 Aug 2017 15:49:50 +0000 (17:49 +0200)]
update notes on todo's
Alan T. DeKok [Sun, 6 Aug 2017 15:45:48 +0000 (17:45 +0200)]
print reply packet VPs, and add to request VPs
Alan T. DeKok [Sun, 6 Aug 2017 15:42:29 +0000 (17:42 +0200)]
create and use Session-State-User-Name
so that we can cache the User-Name in Session-State without
affecting the User-Name in the reply.
The default "send Access-Accept" policy adds all of the session-state
attributes to the reply. Which in this case would mean User-Name...
but we don't want that. i.e. we don't want our internal policies
to pollute whatever the user wants to send back
Alan T. DeKok [Sun, 6 Aug 2017 15:28:08 +0000 (17:28 +0200)]
add base fr_radius_decode()
Alan T. DeKok [Sun, 6 Aug 2017 15:27:12 +0000 (17:27 +0200)]
clean up duplicate / different debug messages
Alan T. DeKok [Sun, 6 Aug 2017 14:56:41 +0000 (16:56 +0200)]
allow replies to Status-Server packets
with provisions for future pings...
Alan T. DeKok [Sun, 6 Aug 2017 14:41:10 +0000 (16:41 +0200)]
Allow Protocol-Error packets as all responses
Alan T. DeKok [Sun, 6 Aug 2017 12:42:14 +0000 (14:42 +0200)]
check allowed return codes for replies
Alan T. DeKok [Sun, 6 Aug 2017 12:28:24 +0000 (14:28 +0200)]
set module return code based on packet return code
Alan T. DeKok [Sun, 6 Aug 2017 12:20:52 +0000 (14:20 +0200)]
try to drain the socket when reading packets
Alan T. DeKok [Sun, 6 Aug 2017 12:10:50 +0000 (14:10 +0200)]
set c->pending in fd_active
Alan T. DeKok [Sun, 6 Aug 2017 12:02:34 +0000 (14:02 +0200)]
update "write immediately" code
so that we don't call yeild / resume when replicating packets,
and we don't flip/flop the event list to set it writable,
call write, and then set it not writable
Alan T. DeKok [Sun, 6 Aug 2017 11:15:59 +0000 (13:15 +0200)]
clean up idle timeout logic
on open, set idle timeouts only if there's no queued packets
on fd active, delete any idle timeout
on no more packets, set idle timeout again
Alan T. DeKok [Sun, 6 Aug 2017 10:55:14 +0000 (12:55 +0200)]
DEBUG3 output, and return 1 on "please retransmit"
Alan T. DeKok [Sun, 6 Aug 2017 10:48:42 +0000 (12:48 +0200)]
re-set "was_sleeping" if there's no data read from the channel
i.e. when the network tries to wake up the worker saying "data ready",
but the worker says "I know, I'm waiting for requests to finish"
Alan T. DeKok [Sun, 6 Aug 2017 10:39:41 +0000 (12:39 +0200)]
check for reply:Packet-Type, and enforce it
just before sending the packet, i.e. if the module returned
"handled". If there's no reply:Packet-Type, set it to Acccess-Reject
Alan T. DeKok [Sun, 6 Aug 2017 10:39:28 +0000 (12:39 +0200)]
enforce "do not respond" policy
Alan T. DeKok [Sun, 6 Aug 2017 10:24:21 +0000 (12:24 +0200)]
add Do-Not-Respond packet type
Alan T. DeKok [Sun, 6 Aug 2017 10:21:42 +0000 (12:21 +0200)]
update notes
Alan T. DeKok [Sun, 6 Aug 2017 10:17:42 +0000 (12:17 +0200)]
minor cleanups
Alan T. DeKok [Sun, 6 Aug 2017 10:17:29 +0000 (12:17 +0200)]
set most recent reply to now on connecton open
Alan T. DeKok [Sun, 6 Aug 2017 10:13:43 +0000 (12:13 +0200)]
update time after removing connection from the heap
Alan T. DeKok [Sun, 6 Aug 2017 10:13:32 +0000 (12:13 +0200)]
fixed event list bug
Alan T. DeKok [Sun, 6 Aug 2017 10:10:05 +0000 (12:10 +0200)]
we can't compare seconds / useconds in isolation
they have to be compared as connected times. The erroneous
comparison was causing earlier times to be pushed to later in
the timer heap
Arran Cudbard-Bell [Fri, 4 Aug 2017 19:21:35 +0000 (15:21 -0400)]
Add support for passlib and crypt pbkdf2 variants
Arran Cudbard-Bell [Fri, 4 Aug 2017 13:43:57 +0000 (09:43 -0400)]
Typo
Arran Cudbard-Bell [Fri, 4 Aug 2017 13:27:54 +0000 (09:27 -0400)]
Disable pbkdf2 tests if we don’t have OpenSSL
Arran Cudbard-Bell [Thu, 3 Aug 2017 23:39:15 +0000 (19:39 -0400)]
Ignore with extreme prejudice
Arran Cudbard-Bell [Thu, 3 Aug 2017 22:38:22 +0000 (18:38 -0400)]
Make the build error go away by ignoring it
Arran Cudbard-Bell [Thu, 3 Aug 2017 22:08:04 +0000 (18:08 -0400)]
More used/unused
Arran Cudbard-Bell [Thu, 3 Aug 2017 21:23:28 +0000 (17:23 -0400)]
Moar format specifiers
Arran Cudbard-Bell [Thu, 3 Aug 2017 20:02:45 +0000 (16:02 -0400)]
Fix build issues in rlm_radius
Arran Cudbard-Bell [Thu, 3 Aug 2017 19:56:22 +0000 (15:56 -0400)]
Full stops...
Arran Cudbard-Bell [Thu, 3 Aug 2017 19:55:09 +0000 (15:55 -0400)]
Make it sound less humble braggy
Arran Cudbard-Bell [Thu, 3 Aug 2017 19:46:32 +0000 (15:46 -0400)]
Reflow to 100 columns
Arran Cudbard-Bell [Thu, 3 Aug 2017 18:30:56 +0000 (14:30 -0400)]
Merge pull request #2034 from jimdigriz/winbind-v4.0.x
winbind build (v4.0.x)
Alexander Clouter [Thu, 3 Aug 2017 16:35:33 +0000 (17:35 +0100)]
winbind build
Alan T. DeKok [Wed, 2 Aug 2017 15:01:53 +0000 (17:01 +0200)]
more notes on what to do
Alan T. DeKok [Wed, 2 Aug 2017 15:01:40 +0000 (17:01 +0200)]
set to NULL after freeing it
Alan T. DeKok [Wed, 2 Aug 2017 15:00:29 +0000 (17:00 +0200)]
parent events off of id, not request
because when we're freeing the request, the event will be removed
before rr, and rr points to the event. Instead, rely on the
rlm_radius_link_t to free up the things properly.
Alan T. DeKok [Wed, 2 Aug 2017 14:51:49 +0000 (16:51 +0200)]
reset cleanup timer if there's no more work to do
Alan T. DeKok [Wed, 2 Aug 2017 14:33:49 +0000 (16:33 +0200)]
set "must signal" flag if the worker is sleeping
Alan T. DeKok [Wed, 2 Aug 2017 14:23:36 +0000 (16:23 +0200)]
start of tracking connection state
Alan T. DeKok [Wed, 2 Aug 2017 14:23:21 +0000 (16:23 +0200)]
this is now done
Alan T. DeKok [Wed, 2 Aug 2017 13:58:18 +0000 (15:58 +0200)]
don't malloc / memcpy the packet if we're replicating
Alan T. DeKok [Wed, 2 Aug 2017 13:53:54 +0000 (15:53 +0200)]
manually add Message-Authenticator to proxied packets
Alan T. DeKok [Mon, 31 Jul 2017 15:51:08 +0000 (11:51 -0400)]
do RADIUS fixups
Always add Message-Authenticator to packets.
add CHAP-Challenge if required
Add Proxy-State manually, as a per-connection 32-bit identifier
We probably want to add Message-Authenticator manually, too...
Alan T. DeKok [Mon, 31 Jul 2017 15:04:41 +0000 (11:04 -0400)]
set idle timeout
and free c->conn before other children
and set c->name to something useful
Alan T. DeKok [Sun, 30 Jul 2017 15:36:37 +0000 (11:36 -0400)]
notes on idle_timeout
0 means "no timeout"
Arran Cudbard-Bell [Wed, 2 Aug 2017 12:47:58 +0000 (08:47 -0400)]
Merge pull request #2032 from mjducharme/patch-1
Fix for mikrotik_snmp not working