]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Alan T. DeKok [Tue, 18 Jan 2022 16:00:08 +0000 (11:00 -0500)]
typo
Alan T. DeKok [Tue, 18 Jan 2022 15:47:37 +0000 (10:47 -0500)]
move "parse scope" to fr_inet_pton6()
Alan T. DeKok [Mon, 17 Jan 2022 18:43:50 +0000 (13:43 -0500)]
fix xlat_unregister_module() so that it works.
Nothing was setting "mod_inst". The use of it was left over
from before we had the module_ctx_t
Alan T. DeKok [Mon, 17 Jan 2022 13:41:26 +0000 (08:41 -0500)]
add type names / values as enums to Cast-Base
so that the types are in the dictionaries, which is nice.
Alan T. DeKok [Sun, 16 Jan 2022 19:31:03 +0000 (14:31 -0500)]
we don't need cast_base
Alan T. DeKok [Sat, 15 Jan 2022 12:46:45 +0000 (07:46 -0500)]
prefer "allowed" to "terminals"
If we want to parse
&Service-Type = Framed-User
then we don't want to parse it as
(&Service - Type) = (Framed - User)
or
(&Service-Type = Framed) - User
etc.
The tmpl_afrom_attr_substr() calls fr_dict_attr_by_name_substr()
which calls fr_dict_attr_by_name_substr(), which then ignores
terminals. So this change makes parsing the RHS enum name behave
the same as parsing the LHS attribute name.
Alan T. DeKok [Fri, 14 Jan 2022 20:51:01 +0000 (15:51 -0500)]
add fr_dict_enum_by_name_substr()
which parses an enumv by name, but doesn't take a "terminals"
argument. Because the enums are named, and if we see a correct
name, then it's correct, no matter what the terminals are.
Alan T. DeKok [Fri, 14 Jan 2022 21:57:34 +0000 (16:57 -0500)]
return >0 from fr_value_box_from_substr()
mainly for ip addresses and prefixes. And also parse scopes for
DHCPv6. Line 20 of src/tests/unit/protocols/dhcpv6/addresses.txt
says that "scopes are ignored".
However, the only reason that the previous tests succeeded was that
fr_value_box_from_substr() was returning 0, which meant that
fr_value_box_from_str() didn't check that the entire string was
being parsed.
A first-pass fix was in commit
8a1f819e7f7 , but that used an extra
sbuff, instead of just re-using our_in, among other faults.
Alan T. DeKok [Thu, 13 Jan 2022 15:45:57 +0000 (10:45 -0500)]
add more whitespace for sanity
value boxes shouldn't be printed all smashed together
Alan T. DeKok [Fri, 14 Jan 2022 15:08:36 +0000 (10:08 -0500)]
use -f for "don't fork", not "-X" for "kill me with debug output"
Alan T. DeKok [Fri, 14 Jan 2022 17:43:19 +0000 (12:43 -0500)]
let's use name_len
instead of walking off of the end of arrays
Alan T. DeKok [Fri, 14 Jan 2022 21:43:36 +0000 (16:43 -0500)]
remove useless code
Alan T. DeKok [Wed, 12 Jan 2022 21:15:34 +0000 (16:15 -0500)]
typo
Daniel DE ALMEIDA BRAGA [Wed, 12 Jan 2022 14:48:04 +0000 (15:48 +0100)]
Fix side channels in EAP-pwd leaking information on a password-related value
An unnecessary branching on the seed's parity at each iteration may leaks one
bit of information at each iteration. Setting is_odd to BN_is_odd directly
avoids the branch.
Additionaly, OpenSSL's implementation of EC_POINT_set_compressed_coordinates
is not constant-time with respect to the compression type. Since the
compression type is conditioned by save_is_odd, which is related to the password,
we need to avoid the generic OpenSSL call and perform the operations in
constant-time.
Alan T. DeKok [Tue, 11 Jan 2022 18:26:29 +0000 (13:26 -0500)]
allow sending packets directly to the worker
Arran Cudbard-Bell [Sat, 8 Jan 2022 07:00:30 +0000 (23:00 -0800)]
Quiet clang scan
Arran Cudbard-Bell [Sat, 8 Jan 2022 00:29:54 +0000 (16:29 -0800)]
Fix various issues with dcursor that would prevent them from working correctly with stateful iterators
Terry Burton [Thu, 6 Jan 2022 20:21:23 +0000 (20:21 +0000)]
CI FreeBSD: Enable radclient tests
Terry Burton [Thu, 6 Jan 2022 20:17:54 +0000 (20:17 +0000)]
udpfromto: Use plain sendto if our source address is 0.0.0.0 or ::/0
FreeBSD sendmsg returns EINVAL if IP_SENDSRCADDR is INADDR_ANY{,6}.
Terry Burton [Thu, 6 Jan 2022 20:08:23 +0000 (20:08 +0000)]
eapol_test-build.sh: Set permissive CFLAGs for FreeBSD build (#4352)
... instead of patching the source.
Arran Cudbard-Bell [Thu, 6 Jan 2022 15:24:24 +0000 (09:24 -0600)]
Trim default prefix from generated dlist functions
Arran Cudbard-Bell [Thu, 6 Jan 2022 14:33:01 +0000 (08:33 -0600)]
Make tmpl_attr and tmpl_request lists type safe
Arran Cudbard-Bell [Wed, 5 Jan 2022 19:34:41 +0000 (13:34 -0600)]
Formatting
James Jones [Thu, 6 Jan 2022 15:04:11 +0000 (09:04 -0600)]
Make test_enqueue_and_io_speed() work (#4349)
Not advancing test_time_base kept some trunk requests from
advancing from SENT to COMPLETE state.
Alan T. DeKok [Thu, 6 Jan 2022 14:53:17 +0000 (09:53 -0500)]
more FR_PROTO_TRACE
Alan T. DeKok [Thu, 6 Jan 2022 14:52:57 +0000 (09:52 -0500)]
use temporary "end"
and pass the right parent in decode_wimax
Alan T. DeKok [Thu, 6 Jan 2022 14:48:04 +0000 (09:48 -0500)]
check min/max lengths only for leaf data types
Alan T. DeKok [Thu, 6 Jan 2022 00:07:39 +0000 (19:07 -0500)]
pass the correct 'end' pointer
Alan T. DeKok [Wed, 5 Jan 2022 21:07:30 +0000 (16:07 -0500)]
rename regex sub to subst
Terry Burton [Wed, 5 Jan 2022 21:43:17 +0000 (21:43 +0000)]
CI fixes for FreeBSD (#4351)
* CI FreeBSD: Reenable some tests
* CI: Temporary patch for hostapd build
Build currently fails on LLVM with FreeBSD.
* CI FreeBSD: Cache the eapol_test build
Arran Cudbard-Bell [Wed, 5 Jan 2022 18:41:03 +0000 (12:41 -0600)]
Quiet compiler
Arran Cudbard-Bell [Wed, 5 Jan 2022 17:59:25 +0000 (11:59 -0600)]
Don't evaluate _uctx multiple times in fr_atexit_thread_local
Arran Cudbard-Bell [Wed, 5 Jan 2022 17:18:19 +0000 (11:18 -0600)]
Fix libpcre memory leak on exit
Terry Burton [Wed, 5 Jan 2022 16:26:20 +0000 (16:26 +0000)]
pair_tests fr_pair_value_copy: Allocate dst pair before copying into it (#4350)
Otherwise the copy will attempt to clear uninitialised dst data.
Arran Cudbard-Bell [Wed, 5 Jan 2022 02:54:36 +0000 (20:54 -0600)]
Formatting
Arran Cudbard-Bell [Wed, 5 Jan 2022 02:52:49 +0000 (20:52 -0600)]
Helps to use the right condition...
Arran Cudbard-Bell [Wed, 5 Jan 2022 02:11:35 +0000 (20:11 -0600)]
Typo
Arran Cudbard-Bell [Wed, 5 Jan 2022 02:09:15 +0000 (20:09 -0600)]
More, minor, atexit cleanups
Arran Cudbard-Bell [Wed, 5 Jan 2022 01:58:55 +0000 (19:58 -0600)]
Move OpenSSL version handling code to its own source file
Arran Cudbard-Bell [Tue, 4 Jan 2022 21:29:43 +0000 (15:29 -0600)]
Quiet compiler warnings
fr_time_delta_t is a struct now...
Arran Cudbard-Bell [Tue, 4 Jan 2022 21:29:16 +0000 (15:29 -0600)]
Flesh out the atexit debug functions more
Arran Cudbard-Bell [Tue, 4 Jan 2022 20:45:11 +0000 (14:45 -0600)]
Search in the correct destructor list
Arran Cudbard-Bell [Tue, 4 Jan 2022 20:44:54 +0000 (14:44 -0600)]
Use the correct comparison function in fr_event_report
James Jones [Tue, 4 Jan 2022 19:18:46 +0000 (13:18 -0600)]
Fix other instances of backlogs left hanging (#4311)
Arran Cudbard-Bell [Tue, 4 Jan 2022 19:15:43 +0000 (13:15 -0600)]
Ensure all the thread-specific destructors run before the global destructors
Rework the OpenSSL cleanup logic
Arran Cudbard-Bell [Tue, 4 Jan 2022 02:14:18 +0000 (20:14 -0600)]
Just use WITH_TLS....
Arran Cudbard-Bell [Tue, 4 Jan 2022 02:10:30 +0000 (20:10 -0600)]
Use the EVP RC4 functions
Arran Cudbard-Bell [Tue, 4 Jan 2022 02:08:55 +0000 (20:08 -0600)]
Ensure openssl user macros are always defined
This causes deprecated functions to trigger fatal errors when building under macos
Alan T. DeKok [Tue, 4 Jan 2022 00:25:15 +0000 (19:25 -0500)]
initialize packet_ctx.end
Alan T. DeKok [Tue, 4 Jan 2022 00:25:07 +0000 (19:25 -0500)]
remove assert
Arran Cudbard-Bell [Mon, 3 Jan 2022 21:56:45 +0000 (15:56 -0600)]
Fix version function calls in dependency.c
Alan T. DeKok [Mon, 3 Jan 2022 21:09:21 +0000 (16:09 -0500)]
don't pass packet_len to fr_radius_decode_pair_value()
which makes its function signature match the other decode functions
Alan T. DeKok [Mon, 3 Jan 2022 21:02:46 +0000 (16:02 -0500)]
clean up the providers on exit
Alan T. DeKok [Mon, 3 Jan 2022 19:20:34 +0000 (14:20 -0500)]
protocol-specific extensions are variable sized
and use a protocol-specific structure, which can't be shared
across multiple protocols
Alan T. DeKok [Tue, 28 Dec 2021 19:22:13 +0000 (14:22 -0500)]
rearrange and remove unused code
Alan T. DeKok [Tue, 28 Dec 2021 19:11:15 +0000 (14:11 -0500)]
headers to help with protocol encoding / decoding
Alan T. DeKok [Tue, 28 Dec 2021 19:07:34 +0000 (14:07 -0500)]
update documentation
Arran Cudbard-Bell [Mon, 3 Jan 2022 21:24:30 +0000 (15:24 -0600)]
Swap ERR_get_error_line_data for ERR_get_error_all
Arran Cudbard-Bell [Mon, 3 Jan 2022 21:23:50 +0000 (15:23 -0600)]
Set OpenSSL API compat macros centrally
Arran Cudbard-Bell [Mon, 3 Jan 2022 21:23:15 +0000 (15:23 -0600)]
Apparently has no effect...
Arran Cudbard-Bell [Mon, 3 Jan 2022 21:22:57 +0000 (15:22 -0600)]
We use OPENSSL_VERSION_NUMBER everywhere else...
Arran Cudbard-Bell [Mon, 3 Jan 2022 20:18:50 +0000 (14:18 -0600)]
Fix calls to PEM_read_bio_DHparams for OpenSSL 3.0.0
Arran Cudbard-Bell [Mon, 3 Jan 2022 19:50:32 +0000 (13:50 -0600)]
more ltb-project key references
Arran Cudbard-Bell [Mon, 3 Jan 2022 19:42:50 +0000 (13:42 -0600)]
New LTB GPG key url
Arran Cudbard-Bell [Mon, 3 Jan 2022 19:34:22 +0000 (13:34 -0600)]
2022
Arran Cudbard-Bell [Thu, 23 Dec 2021 03:26:38 +0000 (21:26 -0600)]
Don't need this in 1.1.1
Arran Cudbard-Bell [Thu, 23 Dec 2021 03:21:08 +0000 (21:21 -0600)]
Stupid hack still needed in 3.0.1
Arran Cudbard-Bell [Thu, 23 Dec 2021 02:30:42 +0000 (20:30 -0600)]
Yes, OpenSSL now wants the size of the buffer passed into EVP_DigestSignFinal
Arran Cudbard-Bell [Thu, 23 Dec 2021 01:46:43 +0000 (19:46 -0600)]
maybe OpenSSL 3.0.1 started caring about the output buffer length?
Arran Cudbard-Bell [Thu, 23 Dec 2021 01:34:42 +0000 (19:34 -0600)]
Revert "CI: Pass in -Wno-error=deprecated-declarations to LIB_ALT build for now"
This reverts commit
78f814cc8f99ca69f74a6886ec327bd067fa7415 .
Terry Burton [Wed, 22 Dec 2021 17:05:21 +0000 (17:05 +0000)]
CI: Pass in -Wno-error=deprecated-declarations to LIB_ALT build for now
Should be reverted once port to OpenSSL 3 API is complete.
Terry Burton [Wed, 22 Dec 2021 13:03:49 +0000 (13:03 +0000)]
CI: Use OpenSSL 3.0.1 in LIBS_ALT build
Alan T. DeKok [Wed, 22 Dec 2021 17:22:08 +0000 (12:22 -0500)]
let's return the correct value
Alan T. DeKok [Wed, 22 Dec 2021 16:06:44 +0000 (11:06 -0500)]
fix decode_tlvs()
so that it doesn't return partial decoding when one of the
internal TLVs fails to decode
Alan T. DeKok [Wed, 22 Dec 2021 15:51:30 +0000 (10:51 -0500)]
if the option is invalid, return a raw option
so that the caller does not have to do this work
Alan T. DeKok [Wed, 22 Dec 2021 15:50:45 +0000 (10:50 -0500)]
unify two functions, so there's less code
and add assertion. We cannot create raw "root" attributes.
Alan T. DeKok [Wed, 22 Dec 2021 14:31:59 +0000 (09:31 -0500)]
typo
Alan T. DeKok [Tue, 21 Dec 2021 21:48:22 +0000 (16:48 -0500)]
da->parent may be NULL here, too.
we have fr_dict_unknown_attr_afrom_da() and fr_dict_unknown_afrom_da()
which are 90% identical. Perhaps some common code would help.
Alan T. DeKok [Tue, 21 Dec 2021 20:07:24 +0000 (15:07 -0500)]
da->parent is NULL at the root
Alan T. DeKok [Tue, 21 Dec 2021 17:38:09 +0000 (12:38 -0500)]
rearrange to show how to assign octets to structural members
code isn't done, but it's a start of rearrangement
Alan T. DeKok [Mon, 20 Dec 2021 14:13:41 +0000 (09:13 -0500)]
add := test
Also if the RHS is octets, then it should be parsed as the correct
data type. Right now this happens on compilation, but it should
also happen at run-time
Alan T. DeKok [Mon, 20 Dec 2021 14:05:02 +0000 (09:05 -0500)]
use new hack function, so that we support T_OP_EQ and T_OP_SET
Alan T. DeKok [Tue, 21 Dec 2021 16:48:07 +0000 (11:48 -0500)]
let's use typed arguments
Alan T. DeKok [Tue, 21 Dec 2021 15:46:28 +0000 (10:46 -0500)]
split out functions as per other protocols
Alan T. DeKok [Tue, 21 Dec 2021 15:17:58 +0000 (10:17 -0500)]
remove "do_raw", which doesn't make sense
Alan T. DeKok [Mon, 20 Dec 2021 13:59:47 +0000 (08:59 -0500)]
add temporary hack function
which will be removed as soon as we have the better function
Alan T. DeKok [Sun, 19 Dec 2021 19:18:00 +0000 (14:18 -0500)]
this function is only used by the xlat code
so rename, it, and move the function prototype to a place which
makes more sense.
Alan T. DeKok [Sun, 19 Dec 2021 19:14:14 +0000 (14:14 -0500)]
decode_pair test point should take fr_dict_attr_t const *parent
which lets us start decoding part way through the dictionary,
instead of always at the top
Alan T. DeKok [Sun, 19 Dec 2021 13:56:15 +0000 (08:56 -0500)]
remove fr_dict_t from struct trampoline functions
Alan T. DeKok [Sun, 19 Dec 2021 13:44:35 +0000 (08:44 -0500)]
clarify error message
Alan T. DeKok [Sun, 19 Dec 2021 13:23:36 +0000 (08:23 -0500)]
minor rearrangements, and update arrays for new operators
Alan T. DeKok [Sun, 19 Dec 2021 13:18:45 +0000 (08:18 -0500)]
encode / decode STRUCT too
the same as TLV, because we don't have packed-structures
in the internal protocol. And the internal protocol doesn't use
protocol-specific encodings. It uses the internal encodings
Alan T. DeKok [Sun, 19 Dec 2021 00:58:32 +0000 (19:58 -0500)]
no need to pass dict_dns to all of the dns functions
Alan T. DeKok [Sun, 19 Dec 2021 00:56:12 +0000 (19:56 -0500)]
no need to pass dict_dhcpv6 to all of the dhcpv6 functions
Alan T. DeKok [Sat, 18 Dec 2021 15:58:10 +0000 (10:58 -0500)]
no need to pass dict_dhcpv4 to all of the dhcpv4 functions
Alan T. DeKok [Sat, 18 Dec 2021 15:45:17 +0000 (10:45 -0500)]
no need to pass dict_radius to all of the radius functions
Alan T. DeKok [Fri, 17 Dec 2021 20:09:34 +0000 (15:09 -0500)]
more operations for ip addresses and prefixes
Terry Burton [Tue, 21 Dec 2021 03:07:47 +0000 (03:07 +0000)]
CI FreeBSD: Run on push/pull; more VM memory; use rsync, not sshfs (#4333)
Arran Cudbard-Bell [Fri, 17 Dec 2021 18:34:09 +0000 (12:34 -0600)]
Fix dlist definition macros to be safer
Alan T. DeKok [Fri, 17 Dec 2021 14:40:18 +0000 (09:40 -0500)]
-Wunused