]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Nick Porter [Mon, 25 Jan 2021 09:38:23 +0000 (09:38 +0000)]
Add test for less than and greater than to rlm_files
Nick Porter [Mon, 25 Jan 2021 09:18:42 +0000 (09:18 +0000)]
Add test for overwriting attributes in rlm_files
Nick Porter [Mon, 5 Oct 2020 16:28:43 +0000 (17:28 +0100)]
Add test for prefix search in rlm_files
Alan T. DeKok [Mon, 26 Apr 2021 18:15:19 +0000 (14:15 -0400)]
add patricia trie support
Alan T. DeKok [Mon, 26 Apr 2021 18:12:22 +0000 (14:12 -0400)]
pretty much nothing needs users_file.h
Alan T. DeKok [Mon, 26 Apr 2021 18:11:18 +0000 (14:11 -0400)]
move PAIR_LIST_LIST to using fr_value_box
Alan T. DeKok [Mon, 26 Apr 2021 17:33:05 +0000 (13:33 -0400)]
move DEFAULT handling to it's own list
When we have htries for different data types, then we cannot
insert DEFAULT into the same htrie as for integers, as DEFAULT
doesn't parse as an integer.
Alan T. DeKok [Mon, 26 Apr 2021 15:56:35 +0000 (11:56 -0400)]
move rlm_files to using htrie code
does not (yet) handle patricia tries.
Alan T. DeKok [Mon, 26 Apr 2021 14:54:24 +0000 (10:54 -0400)]
move OSX leak suppression to code
Arran Cudbard-Bell [Sat, 24 Apr 2021 02:39:05 +0000 (21:39 -0500)]
...and another
Arran Cudbard-Bell [Sat, 24 Apr 2021 00:12:48 +0000 (19:12 -0500)]
Remove FreeBSD case
Arran Cudbard-Bell [Sat, 24 Apr 2021 00:11:56 +0000 (19:11 -0500)]
Remov errant commas
Arran Cudbard-Bell [Fri, 23 Apr 2021 21:12:34 +0000 (16:12 -0500)]
The CRYPTO_THREAD_lock_new leak isn't OS specific
Arran Cudbard-Bell [Fri, 23 Apr 2021 21:07:04 +0000 (16:07 -0500)]
Suppress minor/startup leaks reported by heaptrack
Arran Cudbard-Bell [Fri, 23 Apr 2021 19:38:39 +0000 (14:38 -0500)]
Accidentally removed...
Arran Cudbard-Bell [Fri, 23 Apr 2021 19:38:28 +0000 (14:38 -0500)]
Fix error
Arran Cudbard-Bell [Fri, 23 Apr 2021 18:52:37 +0000 (13:52 -0500)]
Remove default_eap_type from test configs and update doucmentation
Arran Cudbard-Bell [Fri, 23 Apr 2021 18:22:43 +0000 (13:22 -0500)]
Add support for type_identity to EAP-SIM/AKA/AKA'
Arran Cudbard-Bell [Fri, 23 Apr 2021 18:22:11 +0000 (13:22 -0500)]
Remove cruft from rlm_eap
Arran Cudbard-Bell [Fri, 23 Apr 2021 18:21:19 +0000 (13:21 -0500)]
Add docs for cf_log_*_by_child
Arran Cudbard-Bell [Fri, 23 Apr 2021 18:21:00 +0000 (13:21 -0500)]
Allow submodules to determine if they should be set for a given identity
Arran Cudbard-Bell [Fri, 23 Apr 2021 17:50:56 +0000 (12:50 -0500)]
Make outputs optional for sim/aka id functions, add conversion function to extract the eap type
Arran Cudbard-Bell [Fri, 23 Apr 2021 17:50:16 +0000 (12:50 -0500)]
Remove duplicate/unused header
Alan T. DeKok [Thu, 22 Apr 2021 19:45:37 +0000 (15:45 -0400)]
Move the CSV module to using htries
Alan T. DeKok [Thu, 22 Apr 2021 19:27:03 +0000 (15:27 -0400)]
update for latest behavior
Jorge Pereira [Thu, 22 Apr 2021 17:51:12 +0000 (14:51 -0300)]
Fix unitialized value (#4068)
Arran Cudbard-Bell [Thu, 22 Apr 2021 16:08:22 +0000 (11:08 -0500)]
Index is special
Arran Cudbard-Bell [Thu, 22 Apr 2021 16:05:36 +0000 (11:05 -0500)]
Initialise output
Arran Cudbard-Bell [Thu, 22 Apr 2021 15:47:35 +0000 (10:47 -0500)]
Add test attribute instances
Makes it easy to duplicate a set of test attributes, giving each one a unique name and identifier
Arran Cudbard-Bell [Thu, 22 Apr 2021 15:15:18 +0000 (10:15 -0500)]
Start of changing how list_perf_test operates
Arran Cudbard-Bell [Wed, 21 Apr 2021 22:54:30 +0000 (17:54 -0500)]
Add instance arg to fr_pair_find_by_da
Arran Cudbard-Bell [Wed, 21 Apr 2021 20:14:31 +0000 (15:14 -0500)]
Quiet GCC
Arran Cudbard-Bell [Wed, 21 Apr 2021 18:20:45 +0000 (13:20 -0500)]
Fix random whitespace issues in pair_list_perf_tests.c
Arran Cudbard-Bell [Wed, 21 Apr 2021 16:37:02 +0000 (11:37 -0500)]
cpp indentation
Arran Cudbard-Bell [Wed, 21 Apr 2021 16:34:55 +0000 (11:34 -0500)]
Remove duplicate code from pair tests
Arran Cudbard-Bell [Wed, 21 Apr 2021 16:25:19 +0000 (11:25 -0500)]
Make fr_rb_node do what it says
Nick Porter [Tue, 20 Apr 2021 16:19:08 +0000 (17:19 +0100)]
v4: Add tests for rlm_rest - both module and xlat (#4065)
* Add setup script for openresty instance to test rlm_rest
* Install openresty in CI
* Add module tests for rlm_rest
* Test for %{rest: } xlat
Arran Cudbard-Bell [Tue, 20 Apr 2021 16:17:50 +0000 (11:17 -0500)]
Remove another memcpy
Arran Cudbard-Bell [Tue, 20 Apr 2021 16:17:24 +0000 (11:17 -0500)]
Disable the impressively useless "super-linter"
Arran Cudbard-Bell [Tue, 20 Apr 2021 13:34:30 +0000 (08:34 -0500)]
Just remove the damn assert. The rest of the code doesn't care about integer overflow, it'll just stop the advance.
Alan T. DeKok [Tue, 20 Apr 2021 13:32:54 +0000 (09:32 -0400)]
don't check the results of delete, there may be no data to delete
Alan T. DeKok [Tue, 20 Apr 2021 13:34:12 +0000 (09:34 -0400)]
Revert "Revert "Fix SQL tests checking properly the query results (#4064)""
This reverts commit
da11cc71763c4c21c6dc70d28d923346ad109dbd .
Arran Cudbard-Bell [Tue, 20 Apr 2021 13:27:19 +0000 (08:27 -0500)]
Is the compiler really this stupid?
Arran Cudbard-Bell [Tue, 20 Apr 2021 13:22:14 +0000 (08:22 -0500)]
Revert "add cast to shut up compiler"
This reverts commit
dfeadbdc389881fe541ced8ed7736202930272fd .
Arran Cudbard-Bell [Tue, 20 Apr 2021 13:22:06 +0000 (08:22 -0500)]
Revert "shut UP compiler"
This reverts commit
8fc06a0d0e5b4cdd937b0a7e8726bc41836dc7c3 .
Arran Cudbard-Bell [Tue, 20 Apr 2021 13:21:47 +0000 (08:21 -0500)]
Revert "Fix SQL tests checking properly the query results (#4064)"
This reverts commit
d8626c17363b83d6329f5321ea67c84ca548ed59 .
Alan T. DeKok [Tue, 20 Apr 2021 10:34:57 +0000 (06:34 -0400)]
shut UP compiler
Alan T. DeKok [Tue, 20 Apr 2021 10:29:54 +0000 (06:29 -0400)]
clean up encode_concat()
slen is for signed return values, not for unsigned data lengths,
among other things.
Alan T. DeKok [Tue, 20 Apr 2021 10:11:44 +0000 (06:11 -0400)]
add cast to shut up compiler
Alan T. DeKok [Tue, 20 Apr 2021 09:51:34 +0000 (05:51 -0400)]
return dbuff values, not intermediate variables
Jorge Pereira [Tue, 20 Apr 2021 09:27:58 +0000 (06:27 -0300)]
Fix SQL tests checking properly the query results (#4064)
Use macros, so the tests don't have huge lines of text
Alan T. DeKok [Tue, 20 Apr 2021 09:19:59 +0000 (05:19 -0400)]
Revert "decode VSAs and vendors as nested attributes."
This reverts commit
e906e3d73ff649d16916405165ed7736fb496372 .
Arran Cudbard-Bell [Mon, 19 Apr 2021 22:44:41 +0000 (17:44 -0500)]
s/attr_shift/attr_fragment
...and make it use dbuffs properly
Arran Cudbard-Bell [Mon, 19 Apr 2021 22:41:29 +0000 (17:41 -0500)]
Allow the dbuff copying macros to work on dbuffs or markers
Alan T. DeKok [Mon, 19 Apr 2021 21:18:52 +0000 (17:18 -0400)]
fix up for extended VSAs
Alan T. DeKok [Mon, 19 Apr 2021 21:15:49 +0000 (17:15 -0400)]
update description of function arguments
and note that the function simply doesn't work for extended VSAs
<sigh>
Alan T. DeKok [Mon, 19 Apr 2021 21:02:11 +0000 (17:02 -0400)]
remove unused variable
Arran Cudbard-Bell [Mon, 19 Apr 2021 18:32:51 +0000 (13:32 -0500)]
More doc cleanups
Alan T. DeKok [Mon, 19 Apr 2021 18:06:56 +0000 (14:06 -0400)]
fix up attr_shift() so it isn't completely broken
convert it to just use pointers, and clean up the code so that
it's readable.
And add these magical things called "tests"
Alan T. DeKok [Mon, 19 Apr 2021 12:06:30 +0000 (08:06 -0400)]
let's check for errors, if it's not too hard
Alan T. DeKok [Mon, 19 Apr 2021 12:03:02 +0000 (08:03 -0400)]
The WiMAX header is 9 octets, not 2.
Alan T. DeKok [Mon, 19 Apr 2021 11:17:23 +0000 (07:17 -0400)]
decode VSAs and vendors as nested attributes.
Update the tests accordingly.
Note that WiMAX fails, as it does not correctly decode
"continued" attributes. But it turns out that didn't work
previously, either.
Alan T. DeKok [Mon, 19 Apr 2021 11:10:14 +0000 (07:10 -0400)]
split on total length of attribute, not on encoded value length
Alan T. DeKok [Mon, 19 Apr 2021 10:59:48 +0000 (06:59 -0400)]
fix typo
Alan T. DeKok [Mon, 19 Apr 2021 10:28:36 +0000 (06:28 -0400)]
comments and renaming
use encode_attribute() instead of encode_tlv(). This function
encodes attributes in the standard 1/1 TLV format. It does not
encode an attribute of type TLV.
In contrast, encode_vsa() will encode an attribute of type VSA,
encode_extended() will encode an attribute of type "extended",
etc.
Alan T. DeKok [Sun, 18 Apr 2021 20:26:33 +0000 (16:26 -0400)]
unify encode_rfc_format() and encode_tlv()
there's just no reason for them to be gratuitously different
Alan T. DeKok [Sun, 18 Apr 2021 19:56:09 +0000 (15:56 -0400)]
fix encoding of TLVs
turns out encode_value() should be calling encode_tlv_children(),
and then we can get rid of more if / then / else checks scattered
throughout the code
Alan T. DeKok [Sun, 18 Apr 2021 19:53:40 +0000 (15:53 -0400)]
we can hard-code this
Alan T. DeKok [Sun, 18 Apr 2021 19:46:46 +0000 (15:46 -0400)]
increase "depth" when encoding
so that the encode_foo() functions get passed a DA of the type
which they're encoding
Alan T. DeKok [Sun, 18 Apr 2021 19:41:06 +0000 (15:41 -0400)]
hoist encode_wimax() into encode_vendor()
which then supports Vendor-Specific = { WiMAX = { ... } }
Arran Cudbard-Bell [Mon, 19 Apr 2021 18:24:52 +0000 (13:24 -0500)]
Additional debuff docs and renaming arguments
Arran Cudbard-Bell [Mon, 19 Apr 2021 18:02:31 +0000 (13:02 -0500)]
Add note on fr_dbuff_move
Arran Cudbard-Bell [Mon, 19 Apr 2021 14:00:22 +0000 (09:00 -0500)]
Allow replacement of inline rbnodes
Much more efficient that a normal replace!
Alan T. DeKok [Sun, 18 Apr 2021 16:44:38 +0000 (12:44 -0400)]
clean up encode_extended()
remove unnecessary duplicate variables, extra dbuff copies.
Alan T. DeKok [Sun, 18 Apr 2021 16:35:20 +0000 (12:35 -0400)]
use consistent names
Alan T. DeKok [Sun, 18 Apr 2021 16:35:01 +0000 (12:35 -0400)]
there's no reason to make this signed
Alan T. DeKok [Sun, 18 Apr 2021 16:31:52 +0000 (12:31 -0400)]
clean up and de-dup FR_DBUFF_MAX_NO_ADVANCE()
which was being called multiple times, depending on the call flow.
It now just passes the input buffers through, and any function
which knows it can only encode X bytes will do
work_dbuff = FR_DBUFF_MAX_NO_ADVANCE(dbuff, X)
There may also be issues with dbuffs, as the encode_extended()
was being called _always_ with max 255 bytes. Yet it somehow
managed to encode more than 255 bytes in the tests
Alan T. DeKok [Sun, 18 Apr 2021 16:26:03 +0000 (12:26 -0400)]
use names, not numbers
Alan T. DeKok [Sun, 18 Apr 2021 16:15:44 +0000 (12:15 -0400)]
use different way of setting the length field
Alan T. DeKok [Sun, 18 Apr 2021 16:00:33 +0000 (12:00 -0400)]
only use one byte
Alan T. DeKok [Sun, 18 Apr 2021 15:14:57 +0000 (11:14 -0400)]
remove double quotes
Arran Cudbard-Bell [Sun, 18 Apr 2021 15:56:56 +0000 (10:56 -0500)]
Remove node_ctx from rb_inline functions
Mark tree arg in fr_rb_find as const
Arran Cudbard-Bell [Sun, 18 Apr 2021 15:54:28 +0000 (10:54 -0500)]
Remove fr_pair_find_by_num
Arran Cudbard-Bell [Sun, 18 Apr 2021 15:51:41 +0000 (10:51 -0500)]
Remove fr_pair_list_free_item
Arran Cudbard-Bell [Sun, 18 Apr 2021 01:39:19 +0000 (20:39 -0500)]
Add test for dlist_move
Arran Cudbard-Bell [Sun, 18 Apr 2021 01:39:04 +0000 (20:39 -0500)]
Add fr_dlist_entry_move
Arran Cudbard-Bell [Sun, 18 Apr 2021 01:38:42 +0000 (20:38 -0500)]
Add function for checking if a node is in an rbtree
Alan T. DeKok [Sun, 18 Apr 2021 14:45:03 +0000 (10:45 -0400)]
simplify encode_wimax()
similar to the previous commit.
Alan T. DeKok [Sun, 18 Apr 2021 14:32:25 +0000 (10:32 -0400)]
simplify encode_vendor_attr()
there's no need to have a large number of markers. 2 will do.
and don't call them src/dest. Those names are meaningless.
Alan T. DeKok [Sun, 18 Apr 2021 14:25:24 +0000 (10:25 -0400)]
move vendor-specific encoding to encode_vendor_attr()
which is now more in line with encode_wimax(), which does the
same thing
Alan T. DeKok [Sun, 18 Apr 2021 13:46:56 +0000 (09:46 -0400)]
remove redundant code
Alan T. DeKok [Sun, 18 Apr 2021 13:39:09 +0000 (09:39 -0400)]
rename functions
If they're all named _hdr(), then that's redundant. We know that
VSAs and TLVs are not values.
Use more descriptive names than foo_internal(). All of these
functions are "static", and therefore internal. Instead, give
them names according to their purpose.
Alan T. DeKok [Sat, 17 Apr 2021 18:34:10 +0000 (14:34 -0400)]
ensure that fr_radius_decode_pair_value() always returns OK
it only returns -1 on programmer error, or OOM. Otherwise,
it always returns a VP. i.e. a real one, or a raw one.
Alan T. DeKok [Sun, 18 Apr 2021 13:42:12 +0000 (09:42 -0400)]
add ASAN / LSAN options as appropriate
so that "make test" on OSX doesn't have endless complaints about
leaks in libc
Arran Cudbard-Bell [Sat, 17 Apr 2021 00:32:15 +0000 (19:32 -0500)]
Fix pair_list_perf_test.c
Arran Cudbard-Bell [Sat, 17 Apr 2021 00:06:58 +0000 (19:06 -0500)]
Typo
Arran Cudbard-Bell [Fri, 16 Apr 2021 23:58:15 +0000 (18:58 -0500)]
Type validation... generally a good thing
Arran Cudbard-Bell [Fri, 16 Apr 2021 23:49:04 +0000 (18:49 -0500)]
Use conversion macros where appropriate in the dlist code
Arran Cudbard-Bell [Fri, 16 Apr 2021 23:21:51 +0000 (18:21 -0500)]
Minor issues...
Arran Cudbard-Bell [Fri, 16 Apr 2021 23:19:07 +0000 (18:19 -0500)]
Add "old" argument to fr_*_replace to return the old data
This is to avoid multiple lookups in pair.c