]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Matthew Newton [Wed, 23 Nov 2022 20:21:35 +0000 (20:21 +0000)]
remove suse packaging
It's broken and not been seriously updated in nearly 10 years.
Nick Porter [Wed, 23 Nov 2022 13:34:28 +0000 (13:34 +0000)]
Correct CI test
Nick Porter [Wed, 23 Nov 2022 08:24:51 +0000 (08:24 +0000)]
Remove duplicate code
Nick Porter [Wed, 23 Nov 2022 08:21:28 +0000 (08:21 +0000)]
Don't leak fd on failure (CIS #
1517024 )
Alan T. DeKok [Tue, 22 Nov 2022 22:06:36 +0000 (17:06 -0500)]
one more fix for CI
Alan T. DeKok [Tue, 22 Nov 2022 21:11:18 +0000 (16:11 -0500)]
allow for failed xlats to return empty strings, as a special case
&foo := "%{failed xlat}"
will return
&foo == ""
But
&foo := %{failed xlat}
without quotes, will not create &foo.
Matthew Newton [Fri, 18 Nov 2022 22:52:43 +0000 (22:52 +0000)]
More updates to version number handling
Alan T. DeKok [Tue, 22 Nov 2022 15:40:06 +0000 (10:40 -0500)]
pass variable to function
Alan T. DeKok [Tue, 22 Nov 2022 14:00:54 +0000 (09:00 -0500)]
document request.[x] syntax
Alan T. DeKok [Tue, 22 Nov 2022 13:47:43 +0000 (08:47 -0500)]
use list.[*] instead of list[*]
Alan T. DeKok [Tue, 22 Nov 2022 13:41:50 +0000 (08:41 -0500)]
use control.[x]
and change the tests from
(foo != bar)
to
!(foo == bar)
because the first one passes when "foo" doesn't exist. So because
we didn't have control.[x], this test was actually failing for a
while, and no one noticed.
Alan T. DeKok [Tue, 22 Nov 2022 13:37:23 +0000 (08:37 -0500)]
add tmpl_attr_unspec
because otherwise tmpl_attr_tail_da() returns NULL when we have
constructs like %{control.[*]}. And there are just too many
places in the code which do things like:
tmpl_attr_tail_da()->type
All of those are potential crash points if that function returns
NULL.
Instead, we add a canonical "unspec" attribute. It has no name,
numbered zero, is "unknown", and is of FR_TYPE_NULL. This allows
it to pass all of the derefencing code with "it will never match"
Alan T. DeKok [Tue, 22 Nov 2022 12:15:23 +0000 (07:15 -0500)]
don't auto-convert RHS xlats to attribute references
if the caller cares, he can use
&foo := %{eval:%{xlat}}
If we care, we can add
&foo := &%{xlat}
Alan T. DeKok [Mon, 21 Nov 2022 16:15:23 +0000 (11:15 -0500)]
better error messages
show which assignment failed
Alan T. DeKok [Mon, 21 Nov 2022 16:09:01 +0000 (11:09 -0500)]
update docs to match current behavior
Nick Porter [Fri, 11 Nov 2022 13:55:38 +0000 (13:55 +0000)]
Add test for linelog header option
Nick Porter [Fri, 11 Nov 2022 13:54:53 +0000 (13:54 +0000)]
Add header option to linelog
Nick Porter [Wed, 9 Nov 2022 17:52:22 +0000 (17:52 +0000)]
Amend exfile_open() to optionally return the offset in the opened file
Allow for detecting a new / blank file so headers can be added.
Arran Cudbard-Bell [Tue, 22 Nov 2022 02:04:45 +0000 (18:04 -0800)]
Add support for unspecified attributes
Arran Cudbard-Bell [Tue, 22 Nov 2022 02:04:06 +0000 (18:04 -0800)]
Att "last_is_*" functions for tmpls
Arran Cudbard-Bell [Tue, 22 Nov 2022 02:03:37 +0000 (18:03 -0800)]
s/tmpl_attr_count/tmpl_attr_num_elements/
Arran Cudbard-Bell [Mon, 21 Nov 2022 23:32:00 +0000 (15:32 -0800)]
Make it clear what's being returned with tmpl_attr accessors
Arran Cudbard-Bell [Mon, 21 Nov 2022 21:49:25 +0000 (13:49 -0800)]
Ignore depth
Arran Cudbard-Bell [Mon, 21 Nov 2022 21:13:03 +0000 (13:13 -0800)]
Use tmpl accessor macros, and minor reformatting
Arran Cudbard-Bell [Mon, 21 Nov 2022 21:10:43 +0000 (13:10 -0800)]
Fix sign warning in jlibtool
Arran Cudbard-Bell [Mon, 21 Nov 2022 21:09:42 +0000 (13:09 -0800)]
Missing header in tls/session.h
Alan T. DeKok [Sun, 20 Nov 2022 13:02:12 +0000 (08:02 -0500)]
remove "debug attribute lists" from tests
if the tests fail, a developer can add those manually
Alan T. DeKok [Sat, 19 Nov 2022 19:15:34 +0000 (14:15 -0500)]
it's OK for the RHS of an expansion to *not* be an attr ref
We should really do
&foo := &%{...}
for "RHS is an attr ref".
A reasonable compromise is instead
&foo := %{...}
because that's a "bare word". If the user wants a string on the
RHS, he can do
&foo := "%{...}"
and we won't try to parse the RHS as an attribute
Alan T. DeKok [Sat, 19 Nov 2022 14:21:33 +0000 (09:21 -0500)]
partial move to new edit list
because using &control[*] is just weird. And should likely not
be supported.
Alan T. DeKok [Sat, 19 Nov 2022 12:38:53 +0000 (07:38 -0500)]
move assert to before value-box alloc
because structural VPs are very different from structural
value-boxes.
So we should never create a structural value-box in a situation
where the intent is to use VPs
Alan T. DeKok [Fri, 18 Nov 2022 23:09:05 +0000 (18:09 -0500)]
use macros to initialize fields instead of memset
for the odd case where we cast a pre-existing value-box from
a string, AND that string is zero length, the cast silently
succeeds BUT leaves the destination value-box as FR_TYPE_NULL
Which seems bad
Alan T. DeKok [Fri, 18 Nov 2022 19:50:24 +0000 (14:50 -0500)]
future notes for -=
Alan T. DeKok [Fri, 18 Nov 2022 15:00:12 +0000 (10:00 -0500)]
correct order of comparisons
Alan T. DeKok [Fri, 18 Nov 2022 14:38:30 +0000 (09:38 -0500)]
comments
Alan T. DeKok [Fri, 18 Nov 2022 14:11:19 +0000 (09:11 -0500)]
print operators, too
Alan T. DeKok [Fri, 18 Nov 2022 13:50:50 +0000 (08:50 -0500)]
make use-specific variables
so we can do nested loops with fr_pair_list_foreach() macros
Alan T. DeKok [Fri, 18 Nov 2022 13:29:31 +0000 (08:29 -0500)]
give parse error for behavior differences
and give helpful suggestion for how to fix it
Alan T. DeKok [Fri, 18 Nov 2022 13:16:52 +0000 (08:16 -0500)]
compile-time error for things which aren't supported
Alan T. DeKok [Fri, 18 Nov 2022 13:16:32 +0000 (08:16 -0500)]
notes on which tests aren't converted, and why
Alan T. DeKok [Fri, 18 Nov 2022 13:07:41 +0000 (08:07 -0500)]
allow &request = `exec stuff`
garrymar [Mon, 21 Nov 2022 15:04:24 +0000 (16:04 +0100)]
Added Brocade-Cmd attribute sent in command accounting messages (#4808)
Extreme Networks, a vendor that still actively uses the enterprise
number 1588 (Brocade) in their products, added the attribute 8 (Cmd)
that contains a command line when command accounting is configured.
https://documentation.extremenetworks.com/slxos/sw/20xx/20.4.2/
security/GUID-
D8ACA260 -DA62-4128-B443-
44584AD90D96 .shtml
Nick Porter [Mon, 21 Nov 2022 12:14:48 +0000 (12:14 +0000)]
CI: Remove uneeded package
Arran Cudbard-Bell [Sat, 19 Nov 2022 18:46:55 +0000 (10:46 -0800)]
Fix docs to match dictionary
Arran Cudbard-Bell [Sat, 19 Nov 2022 18:39:39 +0000 (10:39 -0800)]
Typo
Arran Cudbard-Bell [Sat, 19 Nov 2022 18:36:32 +0000 (10:36 -0800)]
Bump AC_PREREQ to 2.71 to avoid configure file churn
Matthew Newton [Fri, 18 Nov 2022 22:52:54 +0000 (22:52 +0000)]
need line continuation
Nick Porter [Thu, 17 Nov 2022 19:42:43 +0000 (19:42 +0000)]
Check the return value of ldap_sync init functions
Nick Porter [Thu, 17 Nov 2022 19:46:52 +0000 (19:46 +0000)]
Ensure that cookie storage which fails to send is left on the queue
Nick Porter [Thu, 17 Nov 2022 19:45:41 +0000 (19:45 +0000)]
Add a timer event to retry sending of queued change packets
Nick Porter [Thu, 17 Nov 2022 19:37:42 +0000 (19:37 +0000)]
Use a function to send change packets to the workers
Nick Porter [Thu, 17 Nov 2022 19:34:15 +0000 (19:34 +0000)]
Build pair list within sync_packet_ctx
So that packets which fail to send to the worker can be kept queued with
their pair list
Arran Cudbard-Bell [Fri, 18 Nov 2022 13:13:35 +0000 (07:13 -0600)]
macos doesn't like big version numbers
Alan T. DeKok [Fri, 18 Nov 2022 12:40:18 +0000 (07:40 -0500)]
use API instead of direct access. Fixes #4608
Matthew Newton [Fri, 18 Nov 2022 11:34:25 +0000 (11:34 +0000)]
RPM spec: there is no LTB release for RH9, and default LDAP is good now
Arran Cudbard-Bell [Thu, 17 Nov 2022 23:54:51 +0000 (17:54 -0600)]
srv_cred itself is NULL
Arran Cudbard-Bell [Thu, 17 Nov 2022 23:46:30 +0000 (17:46 -0600)]
Typo
Arran Cudbard-Bell [Thu, 17 Nov 2022 23:41:50 +0000 (17:41 -0600)]
Don't SEGV at DEBUG3 when running SASL
Arran Cudbard-Bell [Thu, 17 Nov 2022 23:21:54 +0000 (17:21 -0600)]
Deal with null SASL fields
Arran Cudbard-Bell [Thu, 17 Nov 2022 22:58:52 +0000 (16:58 -0600)]
RADIUSD_VERSION shouldn't contain a suffix
Arran Cudbard-Bell [Thu, 17 Nov 2022 22:57:09 +0000 (16:57 -0600)]
Look for ldap_interactive_sasl_bind, else we never build with sasl...
Max Khon [Thu, 17 Nov 2022 22:39:54 +0000 (22:39 +0000)]
fr_time_elapsed_fprint(): Fix tab space calculation (#4805)
Without the fix the output sometimes is improperly indented:
time.requests.100ms 28
time.requests.1s 918
time.requests.10s 2408
Max Khon [Thu, 17 Nov 2022 22:38:00 +0000 (22:38 +0000)]
Add support for recv_buff parameter (#4807)
Arran Cudbard-Bell [Thu, 17 Nov 2022 21:34:48 +0000 (15:34 -0600)]
Produce an error if the initial search failed
Alan T. DeKok [Thu, 17 Nov 2022 21:09:35 +0000 (16:09 -0500)]
remove unnecessary assignment
Arran Cudbard-Bell [Thu, 17 Nov 2022 20:52:26 +0000 (14:52 -0600)]
Modules have prefixes other than rlm_ now 🙄
Alan T. DeKok [Thu, 17 Nov 2022 18:07:13 +0000 (13:07 -0500)]
allow attr names in update to have list references
because "update request" just changes the default list, it doesn't
require that list be used.
Alan T. DeKok [Thu, 17 Nov 2022 17:46:33 +0000 (12:46 -0500)]
start of cleanup
Alan T. DeKok [Thu, 17 Nov 2022 17:41:16 +0000 (12:41 -0500)]
clean up and remove intermediate buffers
Alan T. DeKok [Thu, 17 Nov 2022 15:09:18 +0000 (10:09 -0500)]
no parallel for you
Arran Cudbard-Bell [Thu, 17 Nov 2022 19:56:37 +0000 (13:56 -0600)]
Revert "We don't need this explicit linkage"
This reverts commit
99fd88ca8cd750654fdd661772b8d37ca469c1e7 .
Arran Cudbard-Bell [Thu, 17 Nov 2022 19:42:52 +0000 (13:42 -0600)]
We don't need this explicit linkage
Arran Cudbard-Bell [Thu, 17 Nov 2022 19:27:25 +0000 (13:27 -0600)]
More fixes
Arran Cudbard-Bell [Thu, 17 Nov 2022 18:58:27 +0000 (12:58 -0600)]
Remove git marker from changelog
Arran Cudbard-Bell [Thu, 17 Nov 2022 18:23:40 +0000 (12:23 -0600)]
Fix LDAP sync linking
Arran Cudbard-Bell [Thu, 17 Nov 2022 18:19:34 +0000 (12:19 -0600)]
Add the git market in to keep debian happy
Arran Cudbard-Bell [Thu, 17 Nov 2022 18:10:32 +0000 (12:10 -0600)]
Typo
Arran Cudbard-Bell [Thu, 17 Nov 2022 18:10:07 +0000 (12:10 -0600)]
Clever short circuiting tricks don't seem to work everywhere
Arran Cudbard-Bell [Thu, 17 Nov 2022 18:04:52 +0000 (12:04 -0600)]
Quiet ubsan
Arran Cudbard-Bell [Thu, 17 Nov 2022 18:02:48 +0000 (12:02 -0600)]
set email so we don't get prompted
Arran Cudbard-Bell [Thu, 17 Nov 2022 17:55:25 +0000 (11:55 -0600)]
Specify the debian package version using dch
Arran Cudbard-Bell [Thu, 17 Nov 2022 17:39:35 +0000 (11:39 -0600)]
Typo
Arran Cudbard-Bell [Thu, 17 Nov 2022 17:34:49 +0000 (11:34 -0600)]
Don't need to print the version string twice.
Arran Cudbard-Bell [Thu, 17 Nov 2022 17:30:05 +0000 (11:30 -0600)]
Update versioning scheme to use commit depth for incrimental component
Arran Cudbard-Bell [Thu, 17 Nov 2022 16:29:41 +0000 (10:29 -0600)]
Fix return of xlat failure
Arran Cudbard-Bell [Thu, 17 Nov 2022 16:24:54 +0000 (10:24 -0600)]
Simplify release parsing, add +git into git package versions
Arran Cudbard-Bell [Thu, 17 Nov 2022 16:09:29 +0000 (10:09 -0600)]
Produce git versioned .deb files
Arran Cudbard-Bell [Thu, 17 Nov 2022 15:14:20 +0000 (09:14 -0600)]
Add explicit return types for xlat functions
Alan T. DeKok [Thu, 17 Nov 2022 14:34:37 +0000 (09:34 -0500)]
more corner cases, as caught by ASAN
Alan T. DeKok [Thu, 17 Nov 2022 13:56:05 +0000 (08:56 -0500)]
more cleanups
require that attributes always have leading &, and deal with the
consequences
Alan T. DeKok [Wed, 16 Nov 2022 20:14:35 +0000 (15:14 -0500)]
bump up buffer size to shut up compiler
Alan T. DeKok [Wed, 16 Nov 2022 20:13:41 +0000 (15:13 -0500)]
remove unused assignment
Alan T. DeKok [Wed, 16 Nov 2022 18:44:20 +0000 (13:44 -0500)]
run some edits under the new regime
The behavior which isn't yet converted is:
* giving "update list { ... }" and then a *different* list for
an attribute inside of the update section.
The new code doesn't like this, for good reason.
* assigning &request += &control[*]
the new code doesn't expect to see the [*], and therefore
just ignores the assignment.
* creating empty attributes if the RHS of an evaluate fails.
The new code does nothing.
* using array indexes in LHS of assignment for sub-list,
update foo { bar[1] = baz}.
The new code doesn't allow it.
* some compile-time errors are produced at different locations
Alan T. DeKok [Wed, 16 Nov 2022 18:43:55 +0000 (13:43 -0500)]
notes on what needs fixing for new "auto-convert to edit"
Alan T. DeKok [Wed, 16 Nov 2022 18:13:04 +0000 (13:13 -0500)]
more cleanups and fixes for corner cases
Alan T. DeKok [Wed, 16 Nov 2022 18:02:40 +0000 (13:02 -0500)]
forbid array references in sublists
&list := {
&foo[1] = "bar"
}
The edit code does not support this, because it doesn't really
make a lot of sense. If we do need to support it, then the edit
code needs to be updated to support it.
Note that we can already do
&list.foo[1] = "bar"
so this limitation isn't much of an issue. The problem comes when
doing multiple attribute assignments, and those using array indexes
inside of a sublist. What does it mean when we do
&list := {
&foo[0] = "bar"
&foo[1] = "baz"
&foo[0] = "ugh"
}
What the heck is going on here? The real answer is "don't do that"
Alan T. DeKok [Wed, 16 Nov 2022 17:50:54 +0000 (12:50 -0500)]
print attributes in debug, not info level
Alan T. DeKok [Wed, 16 Nov 2022 15:11:25 +0000 (10:11 -0500)]
don't need " \" at EOL any more
Alan T. DeKok [Wed, 16 Nov 2022 15:07:35 +0000 (10:07 -0500)]
use test_fail instead of "fail"
Alan T. DeKok [Wed, 16 Nov 2022 14:43:31 +0000 (09:43 -0500)]
catch more corner cases
update {
&Foo = bar
}
should use "list_def"
Alan T. DeKok [Wed, 16 Nov 2022 13:48:39 +0000 (08:48 -0500)]
this depends on edit, too