]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Alan T. DeKok [Tue, 16 Oct 2018 16:59:49 +0000 (12:59 -0400)]
we don't need Post-Auth-Type any more
the reply code has already been set by the time we're running
this module.
Alan T. DeKok [Tue, 16 Oct 2018 16:57:31 +0000 (12:57 -0400)]
simplify the code now that the old code is gone
Alan T. DeKok [Tue, 16 Oct 2018 16:54:07 +0000 (12:54 -0400)]
remove rad_postauth
Alan T. DeKok [Tue, 16 Oct 2018 16:42:22 +0000 (12:42 -0400)]
remove old rad_authenticate code
Alan T. DeKok [Tue, 16 Oct 2018 16:25:40 +0000 (12:25 -0400)]
remove Autz-Type and Autz-Type redo
It's no longer used
Alan T. DeKok [Tue, 16 Oct 2018 14:33:45 +0000 (10:33 -0400)]
remove Connect-Rate comparison registration
It was wrong and unused.
Alan T. DeKok [Tue, 16 Oct 2018 14:32:38 +0000 (10:32 -0400)]
input is request list, not the attribute we need to compare
Alan T. DeKok [Tue, 16 Oct 2018 14:20:23 +0000 (10:20 -0400)]
inout is request list, not one attribute
so if the User-Name field is not at the start of the packet,
rlm_winbind would do bad things.
Alan T. DeKok [Tue, 16 Oct 2018 13:48:03 +0000 (09:48 -0400)]
remove "compare" flag.
It was only used in a few places, and in the server. So the flag
shouldn't be universal across all dictionary attributes.
if the admin defines a comparison for a pre-existing attribute,
well, that's something we can catch via other methods
Alan T. DeKok [Mon, 15 Oct 2018 17:59:06 +0000 (13:59 -0400)]
tests for struct within tlv within struct
we can't do structs within structs right now.. that's also dumb,
as the fields could just be flattened
Alan T. DeKok [Mon, 15 Oct 2018 15:10:00 +0000 (11:10 -0400)]
canonicalize dictionaries using new format
to allow ".1" which means "child of previous attribute"
Alan T. DeKok [Mon, 15 Oct 2018 15:00:59 +0000 (11:00 -0400)]
add encode / decode tests for structs
Alan T. DeKok [Mon, 15 Oct 2018 14:56:31 +0000 (10:56 -0400)]
allow partial OIDs in ATTRIBUTE definitions
in which case the previous attribute is used as the parent.
This makes the dictionaries MUCH simpler. e.g.
ATTRIBUTE Unit-Struct1 241.247 struct
ATTRIBUTE Unit-Struct1-Int1 .1 integer
ATTRIBUTE Unit-Struct1-Int2 .2 integer
ATTRIBUTE Unit-Struct1-Short .3 uint16
ATTRIBUTE Unit-Struct1-String .4 string
Alan T. DeKok [Mon, 15 Oct 2018 13:15:49 +0000 (09:15 -0400)]
added dictionary
Alan T. DeKok [Sun, 14 Oct 2018 13:22:57 +0000 (09:22 -0400)]
return rcode <sigh>
WiMAX has "continued" attributes, which means that the decoder
needs to decode *more* data than exists in the current attribute
Alan T. DeKok [Sun, 14 Oct 2018 12:56:34 +0000 (08:56 -0400)]
add PROJECT_NAME to includedir path. Fixes #2331
Alan T. DeKok [Fri, 12 Oct 2018 19:53:20 +0000 (15:53 -0400)]
decode fields from struct
Arran Cudbard-Bell [Fri, 12 Oct 2018 21:31:10 +0000 (17:31 -0400)]
Merge pull request #2330 from z-eos/master
[proposal] to add LDAP freeradius-radaact.schema
Zeus Panchenko [Fri, 12 Oct 2018 21:27:30 +0000 (00:27 +0300)]
Add files via upload
Alan T. DeKok [Fri, 12 Oct 2018 19:28:13 +0000 (15:28 -0400)]
it's ssize_t
Alan T. DeKok [Fri, 12 Oct 2018 18:16:17 +0000 (14:16 -0400)]
decode TLVs after fixed-length structs, too
Alan T. DeKok [Fri, 12 Oct 2018 18:02:53 +0000 (14:02 -0400)]
return attr_len, not rcode
i.e. even if the called function decodes *less* than the attribute,
we tell the caller that it decoded *all* of the attribute.
This change ensures that if a function partially decodes an attribute,
that the caller will skip *all* of the attribute. And not start
decoding the next attribute in the middle of the partially decoded
one.
Alan T. DeKok [Fri, 12 Oct 2018 14:45:59 +0000 (10:45 -0400)]
encode TLVs as the last field of a struct in RADIUS
Alan T. DeKok [Fri, 12 Oct 2018 14:21:04 +0000 (10:21 -0400)]
clean up the code a bit
Alan T. DeKok [Fri, 12 Oct 2018 13:58:36 +0000 (09:58 -0400)]
clean up the code a bit
Alan T. DeKok [Fri, 12 Oct 2018 13:51:08 +0000 (09:51 -0400)]
create and export "unknown from network" function
Alan T. DeKok [Thu, 11 Oct 2018 12:31:51 +0000 (08:31 -0400)]
use the correct context for moving output pairs. Fixes #2328
Alan T. DeKok [Thu, 11 Oct 2018 19:57:57 +0000 (15:57 -0400)]
move encoe struct to common function
Alan T. DeKok [Thu, 11 Oct 2018 19:53:39 +0000 (15:53 -0400)]
remove tlv_stack from encode_struct
in preparation for making it a generic function.
After it's a generic function, we can re-add protocol-specific
encoding, with sub-TLVs
Alan T. DeKok [Thu, 11 Oct 2018 19:32:52 +0000 (15:32 -0400)]
move struct.[ch] to src/lib/util
which is where most things need it to be
Alan T. DeKok [Thu, 11 Oct 2018 18:38:21 +0000 (14:38 -0400)]
clean it up to do less work
Alan T. DeKok [Thu, 11 Oct 2018 18:21:56 +0000 (14:21 -0400)]
got to the next VP (but not too far) when encoding structs
Alan T. DeKok [Thu, 11 Oct 2018 17:58:09 +0000 (13:58 -0400)]
move struct decoding into it's own function
Alan T. DeKok [Thu, 11 Oct 2018 17:46:05 +0000 (13:46 -0400)]
we don't need decoder_ctx for struct
Alan T. DeKok [Thu, 11 Oct 2018 17:44:24 +0000 (13:44 -0400)]
remove RADIUS knowledge from encode_struct
Alan T. DeKok [Thu, 11 Oct 2018 17:22:20 +0000 (13:22 -0400)]
structs which are variable length have "flags.length == 0"
Alan T. DeKok [Thu, 11 Oct 2018 17:15:49 +0000 (13:15 -0400)]
more sanity checking for flags of struct && their entries
Alan T. DeKok [Thu, 11 Oct 2018 17:08:26 +0000 (13:08 -0400)]
make decode_struct use standard functions
and not RADIUS ones. Struct entries aren't encrypted, they
shouldn't really use RADIUS-specific data types, etc.
Arran Cudbard-Bell [Thu, 11 Oct 2018 16:38:38 +0000 (12:38 -0400)]
Silence signed/unsigned warning
Arran Cudbard-Bell [Thu, 11 Oct 2018 14:26:20 +0000 (10:26 -0400)]
Include the exec header file
Arran Cudbard-Bell [Thu, 11 Oct 2018 14:07:36 +0000 (10:07 -0400)]
Redundant declaration
Arran Cudbard-Bell [Thu, 11 Oct 2018 13:52:35 +0000 (09:52 -0400)]
s/load/onload due to conflicts with stdatomic.h *sigh*
More header splitting
Alan T. DeKok [Wed, 10 Oct 2018 19:46:44 +0000 (15:46 -0400)]
now that we can decode these structs, create them as structs
Alan T. DeKok [Wed, 10 Oct 2018 19:36:07 +0000 (15:36 -0400)]
allow variable-length fields as the last one in a struct
Alan T. DeKok [Wed, 10 Oct 2018 19:32:43 +0000 (15:32 -0400)]
encode structs, too
Instead of disallowing structural types and allowing everything
else (i.e. non-value types), we only allow value types, and
disallow everything else
Alan T. DeKok [Wed, 10 Oct 2018 18:42:23 +0000 (14:42 -0400)]
decode structs where the last entry is variable length
Alan T. DeKok [Wed, 10 Oct 2018 17:43:49 +0000 (13:43 -0400)]
struct children must be numbered consecutively.
We would prefer to not list numbers, but that's life
Alan T. DeKok [Wed, 10 Oct 2018 14:23:21 +0000 (10:23 -0400)]
fix ethernet to integer conversion
we can't really memcpy() an ethernet address to an integer64
attribute, because the address may be in the high bits of the
integer, or in the lowe bits, depending on endian-ness of the
host CPU.
Instead, we manually copy the bytes over, so that we're sure they
end up where we want. Note that this is the same process used
by fr_value_box_cast_to_ethernet(). i.e. copy to the *low* bytes,
and not to the *high* bytes.
In addition, fix the unit tests to match. Casting a string type
to ethernet isn't the right thing to do. With the previous code,
the string type contained an ASCII representation of the ethernet
copied to the high bytes of integer64 and then printed... which
isn't at all anything useful.
The new test case prints ethernet to integer64 type, and then
casts that to ethernet.
If we want to cast from ethernet to string and back, we MUST
just copy the bytes raw / as-is, and *not* convert them to integer
Alan T. DeKok [Wed, 10 Oct 2018 14:17:17 +0000 (10:17 -0400)]
fix to be pedantically correct
the value in vp_ipaddr is in network order, not host order.
While htonl() and ntohl() are similar, using the right names
is better
Alan T. DeKok [Wed, 10 Oct 2018 14:13:36 +0000 (10:13 -0400)]
iuse the correct line number for error messages
unlike v3, v4 interpreter line numbers are for the original policy,
not where they are used
Alan T. DeKok [Wed, 10 Oct 2018 13:03:43 +0000 (09:03 -0400)]
use consistent names
Arran Cudbard-Bell [Wed, 10 Oct 2018 07:07:00 +0000 (03:07 -0400)]
More header splitting
Arran Cudbard-Bell [Tue, 9 Oct 2018 22:15:21 +0000 (15:15 -0700)]
More header splitting
Arran Cudbard-Bell [Tue, 9 Oct 2018 21:57:48 +0000 (14:57 -0700)]
Split paircmp header out
Arran Cudbard-Bell [Tue, 9 Oct 2018 21:07:50 +0000 (14:07 -0700)]
Cherry-pick RHEL sysconfdir fixes from v3.0.x branch
Arran Cudbard-Bell [Tue, 9 Oct 2018 19:45:19 +0000 (12:45 -0700)]
Correct goto label
Arran Cudbard-Bell [Tue, 9 Oct 2018 19:39:06 +0000 (12:39 -0700)]
Minor formatting tweaks
Arran Cudbard-Bell [Tue, 9 Oct 2018 19:38:40 +0000 (12:38 -0700)]
Decode M3UA_MGMT_ERROR codes, and trigger reset on M3UA_ERR_CODE_REFUSED_MANAGEMENT_BLOCKING
Alan T. DeKok [Sun, 7 Oct 2018 13:41:36 +0000 (09:41 -0400)]
generate better error messages
Arran Cudbard-Bell [Fri, 5 Oct 2018 14:29:50 +0000 (21:29 +0700)]
Script to upload log files to AWS S3 bucket
Arran Cudbard-Bell [Fri, 5 Oct 2018 10:31:41 +0000 (17:31 +0700)]
Needed here too
Arran Cudbard-Bell [Fri, 5 Oct 2018 06:47:42 +0000 (13:47 +0700)]
Split regex functions into a separate header
Arran Cudbard-Bell [Fri, 5 Oct 2018 06:24:03 +0000 (13:24 +0700)]
Remove redundant declaration
Arran Cudbard-Bell [Fri, 5 Oct 2018 05:49:45 +0000 (12:49 +0700)]
More header fixes for the server library
Alan T. DeKok [Thu, 4 Oct 2018 20:00:20 +0000 (16:00 -0400)]
this attribute is text, not binary. Fixes #2322
Alan T. DeKok [Thu, 4 Oct 2018 14:36:26 +0000 (10:36 -0400)]
simplify code
the buffer almost always shrinks, so just call the function
Alan T. DeKok [Thu, 4 Oct 2018 14:00:40 +0000 (10:00 -0400)]
shrink the message, and fix compile warning
Arran Cudbard-Bell [Thu, 4 Oct 2018 13:30:06 +0000 (20:30 +0700)]
Minor fixes
Arran Cudbard-Bell [Thu, 4 Oct 2018 12:53:06 +0000 (19:53 +0700)]
It's ok for the ctx to be NULL
Arran Cudbard-Bell [Thu, 4 Oct 2018 12:09:36 +0000 (19:09 +0700)]
No longer used
Arran Cudbard-Bell [Thu, 4 Oct 2018 10:33:19 +0000 (17:33 +0700)]
Fix PCRE2 version output
Arran Cudbard-Bell [Thu, 4 Oct 2018 02:44:21 +0000 (09:44 +0700)]
Don't leak HMAC_CTXs
Arran Cudbard-Bell [Thu, 4 Oct 2018 02:44:03 +0000 (09:44 +0700)]
Redundant declaration
Arran Cudbard-Bell [Thu, 4 Oct 2018 02:37:05 +0000 (09:37 +0700)]
Declare some structures before includes
Arran Cudbard-Bell [Wed, 3 Oct 2018 19:39:10 +0000 (02:39 +0700)]
Replace remaining references to mainconfig with main_config
Arran Cudbard-Bell [Wed, 3 Oct 2018 19:37:10 +0000 (02:37 +0700)]
Fix redundant declarations
Arran Cudbard-Bell [Wed, 3 Oct 2018 19:26:36 +0000 (02:26 +0700)]
Include map.h in LDAP library
Alan T. DeKok [Wed, 3 Oct 2018 18:41:10 +0000 (14:41 -0400)]
finally make thread_instance data for the master IO handler
next step: do it for the rest of the IO modules
Alan T. DeKok [Wed, 3 Oct 2018 18:40:47 +0000 (14:40 -0400)]
added "terminate" command
which stops the server from running
Alan T. DeKok [Wed, 3 Oct 2018 18:37:45 +0000 (14:37 -0400)]
fix error message
Alan T. DeKok [Wed, 3 Oct 2018 18:16:49 +0000 (14:16 -0400)]
add fr_listen_t -> connected
so that the code becomes simpler.
Alan T. DeKok [Wed, 3 Oct 2018 15:47:47 +0000 (11:47 -0400)]
fix indentation
which seems to have been largely random
Alan T. DeKok [Wed, 3 Oct 2018 13:14:09 +0000 (09:14 -0400)]
add fr_network_pre_event()
in order to get the event loop to run through it's work again.
so that the network side can say "hey, there's work to do!"
and then go do it.
Alan T. DeKok [Tue, 2 Oct 2018 17:24:50 +0000 (13:24 -0400)]
bypass the channels if it's the same thread
this gives it about a 5-10% performance boost in single-threaded
mode.
Alan T. DeKok [Tue, 2 Oct 2018 17:20:17 +0000 (13:20 -0400)]
clean up debug3 messages
Alan T. DeKok [Tue, 2 Oct 2018 17:03:52 +0000 (13:03 -0400)]
move callbacks to "end" structures
Alan T. DeKok [Tue, 2 Oct 2018 16:59:58 +0000 (12:59 -0400)]
move worker side to callbacks, too
Alan T. DeKok [Tue, 2 Oct 2018 14:51:54 +0000 (10:51 -0400)]
move fr_channel_recv_reply() to using callbacks
in preparation for bypassing the atomic queues when both ends
of the channel are in the same thread
Alan T. DeKok [Tue, 2 Oct 2018 14:24:48 +0000 (10:24 -0400)]
remember if both ends are in the same thread
Alan T. DeKok [Tue, 2 Oct 2018 14:17:16 +0000 (10:17 -0400)]
rename _master_ to _network_
which is more consistent
Arran Cudbard-Bell [Wed, 3 Oct 2018 14:46:47 +0000 (21:46 +0700)]
First phase of restructuring server library headers
Everything needs to be broken out more as it was with the utility library, expect more commits in the future
Arran Cudbard-Bell [Wed, 3 Oct 2018 05:07:51 +0000 (13:07 +0800)]
Merge pull request #2317 from cipherboy/hmac-openssl-master
HMAC -> OpenSSL against master
Alexander Scheel [Fri, 28 Sep 2018 15:03:52 +0000 (11:03 -0400)]
Replace HMAC-SHA1 implementation with OpenSSL's
If OpenSSL EVP is not found, fallback to internal implementation of
HMAC-SHA1.
Signed-off-by: Alexander Scheel <ascheel@redhat.com>
Alexander Scheel [Fri, 28 Sep 2018 13:54:46 +0000 (09:54 -0400)]
Replace HMAC-MD5 implementation with OpenSSL's
If OpenSSL EVP is not found, fallback to internal implementation of
HMAC-MD5.
Signed-off-by: Alexander Scheel <ascheel@redhat.com>
Alejandro Perez [Tue, 2 Oct 2018 11:54:26 +0000 (13:54 +0200)]
Add Moonshot-OTP-Secret attribute definition
Alan T. DeKok [Tue, 2 Oct 2018 13:48:20 +0000 (09:48 -0400)]
manual merge of commit
b5b27fedba8f
Alan T. DeKok [Mon, 1 Oct 2018 14:09:51 +0000 (10:09 -0400)]
"chunk" is unsigned
Alan T. DeKok [Mon, 1 Oct 2018 13:17:16 +0000 (09:17 -0400)]
reset "chunk" to ensure we find the correct entry
Alan T. DeKok [Mon, 1 Oct 2018 13:11:09 +0000 (09:11 -0400)]
"my_bool" isn't in MySQL 8.0
but we have "bool"
Arran Cudbard-Bell [Mon, 1 Oct 2018 05:26:16 +0000 (13:26 +0800)]
Trim whitespace