]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Arran Cudbard-Bell [Tue, 10 Sep 2019 23:43:34 +0000 (18:43 -0500)]
Quiet clang scan
Arran Cudbard-Bell [Tue, 10 Sep 2019 23:26:18 +0000 (18:26 -0500)]
Strip -Wall when returned from python-config
Fix integer widths in rlm_python.c
Arran Cudbard-Bell [Tue, 10 Sep 2019 22:30:35 +0000 (17:30 -0500)]
Helps to regenerate the configure script
Arran Cudbard-Bell [Tue, 10 Sep 2019 22:22:31 +0000 (17:22 -0500)]
lineno isn't a pointer...
Arran Cudbard-Bell [Tue, 10 Sep 2019 22:07:08 +0000 (17:07 -0500)]
Yet more flag stripping *sigh*
Alan T. DeKok [Tue, 10 Sep 2019 20:29:54 +0000 (16:29 -0400)]
make cf_expand_variables() take lineno instead of *lineno
which was done for who knows what historical reasons
Alan T. DeKok [Tue, 10 Sep 2019 20:13:50 +0000 (16:13 -0400)]
make fr_cond_tokenize() take CONF_SECTION we're parsing
which ends up being easier than the alternative. That way
instead of passing a bunch of things around, we can just pass
a CONF_SECTION
Alan T. DeKok [Tue, 10 Sep 2019 19:55:17 +0000 (15:55 -0400)]
ensure that we always pass a CONF_SECTION
Arran Cudbard-Bell [Tue, 10 Sep 2019 20:09:35 +0000 (15:09 -0500)]
More filtering of ldflags
Arran Cudbard-Bell [Tue, 10 Sep 2019 19:58:45 +0000 (14:58 -0500)]
More python-config stupidity
Arran Cudbard-Bell [Tue, 10 Sep 2019 19:46:13 +0000 (14:46 -0500)]
-Werror args probably don't contain spaces, but they do contain hyphens
Arran Cudbard-Bell [Tue, 10 Sep 2019 19:35:07 +0000 (14:35 -0500)]
Deal with -Werror where it has args
Alan T. DeKok [Tue, 10 Sep 2019 19:19:34 +0000 (15:19 -0400)]
make process_map() call cf_get_token()
which regularizes the parsing of the various things
Alan T. DeKok [Tue, 10 Sep 2019 18:44:06 +0000 (14:44 -0400)]
we only need 4 buffers
rearrange the user of the buffers, and document what they're for.
Alan T. DeKok [Tue, 10 Sep 2019 18:40:14 +0000 (14:40 -0400)]
simplify process_include()
so that it no longer uses the old APIs
Alan T. DeKok [Tue, 10 Sep 2019 18:28:24 +0000 (14:28 -0400)]
allow unlang expansions in name1 for conf section/pairs
which can occasionally be useful. And, is preparation for
removing the rest of the use of gettoken()
Alan T. DeKok [Tue, 10 Sep 2019 18:27:21 +0000 (14:27 -0400)]
use 'this' section, not the top-level one
Alan T. DeKok [Tue, 10 Sep 2019 18:10:54 +0000 (14:10 -0400)]
expand variables in cf_get_token
Alan T. DeKok [Tue, 10 Sep 2019 18:10:28 +0000 (14:10 -0400)]
errors are errors
Arran Cudbard-Bell [Tue, 10 Sep 2019 19:18:25 +0000 (14:18 -0500)]
s/Ssize_t/Size_t/ in rlm_python
Arran Cudbard-Bell [Tue, 10 Sep 2019 19:12:50 +0000 (14:12 -0500)]
Use PyLong_FromSsize_t where appropriate
Arran Cudbard-Bell [Tue, 10 Sep 2019 19:06:19 +0000 (14:06 -0500)]
Switch numeric interchange to PyLong, but use PyNumber to check for compatibility with Python v2 and v3
Arran Cudbard-Bell [Tue, 10 Sep 2019 19:04:41 +0000 (14:04 -0500)]
Just use python-config to configure the python module
Arran Cudbard-Bell [Tue, 10 Sep 2019 19:04:20 +0000 (14:04 -0500)]
Python 3 compatibility fixes
Arran Cudbard-Bell [Tue, 10 Sep 2019 15:06:27 +0000 (10:06 -0500)]
Revert "We don't need to clear repeatable anymore it's done by frame_cleanup"
This reverts commit
bbc8a8bfe272095cc186d1313b98438d19c65d7b .
Alan T. DeKok [Tue, 10 Sep 2019 14:45:54 +0000 (10:45 -0400)]
fix nested %{...}
Alan T. DeKok [Tue, 10 Sep 2019 14:33:56 +0000 (10:33 -0400)]
allow bare %{...} in conditions
which for now are parsed and printed as double-quoted strings.
But it makes some things much easier
Alan T. DeKok [Tue, 10 Sep 2019 14:33:43 +0000 (10:33 -0400)]
decrement before checks
Alan T. DeKok [Tue, 10 Sep 2019 14:19:06 +0000 (10:19 -0400)]
do less memcpy
Alan T. DeKok [Tue, 10 Sep 2019 14:09:54 +0000 (10:09 -0400)]
make cf_expand_variables() take a length of input
which makies other things simpler, such as the conditional
tokenizer
Alan T. DeKok [Tue, 10 Sep 2019 13:54:58 +0000 (09:54 -0400)]
cleanups and create cf_get_token()
which wraps tmpl_preparse(), and eventually cf_expand_variables()
so that the config file parser *doesn't* use gettoken().
Instead, it should use the new APIs which provide for better
error reporting and more functionality
Arran Cudbard-Bell [Tue, 10 Sep 2019 14:29:17 +0000 (09:29 -0500)]
Add test for return after policy
Policy sets return_point on that stack frame, so we need to make sure it's cleared when the next instruction is executed
Arran Cudbard-Bell [Tue, 10 Sep 2019 14:28:29 +0000 (09:28 -0500)]
We don't need to clear repeatable anymore it's done by frame_cleanup
Arran Cudbard-Bell [Tue, 10 Sep 2019 14:27:54 +0000 (09:27 -0500)]
Clear break_point and return_point when moving to the next instruction or popping things
Arran Cudbard-Bell [Tue, 10 Sep 2019 14:27:31 +0000 (09:27 -0500)]
We really do only need to set top_frame when pushing new things onto the stack
Arran Cudbard-Bell [Tue, 10 Sep 2019 14:27:09 +0000 (09:27 -0500)]
Print unwind at the stack level
Arran Cudbard-Bell [Tue, 10 Sep 2019 14:26:47 +0000 (09:26 -0500)]
Set dependencies correctly for return keyword tests
Alan T. DeKok [Tue, 10 Sep 2019 13:29:07 +0000 (09:29 -0400)]
less code == better code
Alan T. DeKok [Tue, 10 Sep 2019 13:29:02 +0000 (09:29 -0400)]
typo
Jorge Pereira [Mon, 9 Sep 2019 23:17:53 +0000 (20:17 -0300)]
Fix missing free() in $INCLUDE
Alan T. DeKok [Tue, 10 Sep 2019 13:01:58 +0000 (09:01 -0400)]
fix list for proper nesting
Alan T. DeKok [Tue, 10 Sep 2019 12:58:55 +0000 (08:58 -0400)]
text was already in the partial include file
Alan T. DeKok [Tue, 10 Sep 2019 12:58:12 +0000 (08:58 -0400)]
typo
Alan T. DeKok [Tue, 10 Sep 2019 12:57:03 +0000 (08:57 -0400)]
use ":" instead of "::" everywhere
Alan T. DeKok [Tue, 10 Sep 2019 12:56:47 +0000 (08:56 -0400)]
should be fixed-width
Alan T. DeKok [Tue, 10 Sep 2019 12:55:17 +0000 (08:55 -0400)]
this file was moved
Alan T. DeKok [Tue, 10 Sep 2019 12:55:07 +0000 (08:55 -0400)]
document changed behavior
Alan T. DeKok [Mon, 9 Sep 2019 19:16:37 +0000 (15:16 -0400)]
use standard macro
Alan T. DeKok [Mon, 9 Sep 2019 19:14:17 +0000 (15:14 -0400)]
unescape bare %{...} too
Matthew Newton [Tue, 10 Sep 2019 09:18:37 +0000 (10:18 +0100)]
doesn't seem to make any difference, even though it is sometimes set
Alan T. DeKok [Mon, 9 Sep 2019 19:04:11 +0000 (15:04 -0400)]
make some functions static
they are old, and now only used in one place
Alan T. DeKok [Mon, 9 Sep 2019 18:58:17 +0000 (14:58 -0400)]
use tmpl_preparse(), which now does most necessary things
Alan T. DeKok [Mon, 9 Sep 2019 16:50:46 +0000 (12:50 -0400)]
tmpl_preparse() is now capable of parsing %{...}
Jorge Pereira [Mon, 9 Sep 2019 17:20:17 +0000 (14:20 -0300)]
Create an initial, empty CRL in DER encoding #2964
Alan T. DeKok [Mon, 9 Sep 2019 15:59:23 +0000 (11:59 -0400)]
be a bit more flexible about 'map'
map { ...} is OK in sigtran as a config section
Alan T. DeKok [Mon, 9 Sep 2019 15:57:34 +0000 (11:57 -0400)]
allow , and ; after words
Alan T. DeKok [Mon, 9 Sep 2019 15:46:11 +0000 (11:46 -0400)]
less code is better code
Alan T. DeKok [Mon, 9 Sep 2019 13:50:56 +0000 (09:50 -0400)]
simplify the parser.
it's clearer to just check for characters directly, instead of
calling functions and then checking for token values
Alan T. DeKok [Mon, 9 Sep 2019 13:41:20 +0000 (09:41 -0400)]
move "expand variables" into add_pair() function
Alan T. DeKok [Mon, 9 Sep 2019 13:19:36 +0000 (09:19 -0400)]
move add_pair() to separate function
and start cleaning up the parser to be simpler and more understandable
Alan T. DeKok [Mon, 9 Sep 2019 12:51:36 +0000 (08:51 -0400)]
remove WITH_CONF_WRITE
unused, untested, likely not to work. If we re-do the parser,
we can make this functionality part of the requirements.
Until then, this code is making refactoring mode difficult
Stefan Winter [Sun, 8 Sep 2019 06:35:54 +0000 (08:35 +0200)]
ue correct CA password when not the default "whatever"
without passin and passout, the CA passphrase is taken from ca.cnf where it is "whatever". In passwords.mk it is also "whatever" so that is usually okay.
However when changing the password in passwords.mk, the CA is still generated with "whatever" but subsequent cert generation commands use the custom password -> failure.
Arran Cudbard-Bell [Sat, 7 Sep 2019 18:20:08 +0000 (13:20 -0500)]
Port unit changes from v3.0.x
Arran Cudbard-Bell [Sat, 7 Sep 2019 17:11:03 +0000 (12:11 -0500)]
More unwind tests
Arran Cudbard-Bell [Sat, 7 Sep 2019 16:46:22 +0000 (11:46 -0500)]
Fix multiple issues in return/unwind
Arran Cudbard-Bell [Sat, 7 Sep 2019 14:46:06 +0000 (09:46 -0500)]
Convert the macros to inline functions
Better type warnings, and slightly cleaner use of "unwind_to_*"
Jorge Pereira [Fri, 6 Sep 2019 17:55:49 +0000 (14:55 -0300)]
jenkins: The 'QA_RPATHS' is already in 'make rpm' target
Jorge Pereira [Fri, 6 Sep 2019 20:12:26 +0000 (17:12 -0300)]
Fix missing free() in 'decode-tacacs' decode
Arran Cudbard-Bell [Sat, 7 Sep 2019 05:28:20 +0000 (00:28 -0500)]
Consistent case
Arran Cudbard-Bell [Sat, 7 Sep 2019 05:19:55 +0000 (00:19 -0500)]
I guess we just can't use an enum for flag fields anymore...
Arran Cudbard-Bell [Sat, 7 Sep 2019 04:53:32 +0000 (23:53 -0500)]
Quiet the compiler?
Arran Cudbard-Bell [Sat, 7 Sep 2019 02:28:57 +0000 (21:28 -0500)]
Different way of managing where to unwind to
Arran Cudbard-Bell [Sat, 7 Sep 2019 00:00:48 +0000 (19:00 -0500)]
Parent the VPs from the frame state for easy cleanup
this possibly wasn't there before, because there used to be a union in the frame to hold instruction specific data
Arran Cudbard-Bell [Fri, 6 Sep 2019 23:18:06 +0000 (18:18 -0500)]
Switch to using a flags field, and set flags in the correct places, i.e. within the foreach and policy keyword functions
Arran Cudbard-Bell [Fri, 6 Sep 2019 17:54:50 +0000 (12:54 -0500)]
Fix stupid hacks in foreach code
Alan T. DeKok [Fri, 6 Sep 2019 17:37:17 +0000 (13:37 -0400)]
add test for `foo = #`, which shouldn't be allowed
Alan T. DeKok [Fri, 6 Sep 2019 17:36:43 +0000 (13:36 -0400)]
move "map" processing to its own function
Alan T. DeKok [Fri, 6 Sep 2019 17:17:55 +0000 (13:17 -0400)]
move "if / elsif" handling to its own function
Alan T. DeKok [Fri, 6 Sep 2019 17:06:14 +0000 (13:06 -0400)]
try to have most goto's going forward
Alan T. DeKok [Fri, 6 Sep 2019 17:04:50 +0000 (13:04 -0400)]
move $TEMPLATE handling to a separate function
Alan T. DeKok [Fri, 6 Sep 2019 16:57:57 +0000 (12:57 -0400)]
move $INCLUDE handling into its own function
Alan T. DeKok [Fri, 6 Sep 2019 15:49:08 +0000 (11:49 -0400)]
remove unused assignment
Alan T. DeKok [Fri, 6 Sep 2019 15:21:13 +0000 (11:21 -0400)]
clean up and clarify parser
foo = # stuff
is now a parse error. Previously it did who knows what. :(
and two "switch" statements have been changed to one, with
a subsequent "expand variables" stage.
Alan T. DeKok [Fri, 6 Sep 2019 14:52:13 +0000 (10:52 -0400)]
use "op_token" for clarity
this doesn't change the code, but it makes it easier to read
Alan T. DeKok [Fri, 6 Sep 2019 14:48:46 +0000 (10:48 -0400)]
rename t1, t2, t3 to something more reasonable
because 20 years of t1 t2 and t3 are hurting my brain
Arran Cudbard-Bell [Fri, 6 Sep 2019 12:44:40 +0000 (07:44 -0500)]
Move unwind to be part of the stack, not part of a frame
Alan T. DeKok [Thu, 5 Sep 2019 23:13:50 +0000 (19:13 -0400)]
use condition as talloc_ctx
to avoid leaking memory. It's leaked only in the tests,
but whatever.
Alan T. DeKok [Thu, 5 Sep 2019 22:24:29 +0000 (18:24 -0400)]
empty strings are OK
with tests
Alan T. DeKok [Thu, 5 Sep 2019 20:07:25 +0000 (16:07 -0400)]
allow $ENV{...} too
Alan T. DeKok [Thu, 5 Sep 2019 20:01:31 +0000 (16:01 -0400)]
filter out sha2 at the make stage
because even if we have `if (0) { ... }` in the unlang policies,
the inner policies are still getting evaluated.
This fix is the right one, but we still need to add tests for
*not* compiling the contents of `if (0)`
Alan T. DeKok [Thu, 5 Sep 2019 19:27:59 +0000 (15:27 -0400)]
nope. that didn't work
Alan T. DeKok [Thu, 5 Sep 2019 19:05:50 +0000 (15:05 -0400)]
typo
Alan T. DeKok [Thu, 5 Sep 2019 18:45:20 +0000 (14:45 -0400)]
show OpenSSL libs for travis
Jorge Pereira [Thu, 5 Sep 2019 17:52:28 +0000 (14:52 -0300)]
password_process_sha3() Fix the talloc_list_free() call (#2956)
It expects reference instead of value.
Arran Cudbard-Bell [Thu, 5 Sep 2019 15:53:26 +0000 (10:53 -0500)]
Extra check for returning inside subrequests
Matthew Newton [Thu, 5 Sep 2019 15:40:00 +0000 (16:40 +0100)]
nope... try more debugging
Terry Burton [Thu, 5 Sep 2019 00:46:14 +0000 (01:46 +0100)]
Debian packaging: Add builddep on asciidoctor and pandoc for man pages
Minimal fix only for the moment...
Add builddep on asciidoc and pandoc so that we can build the man pages
that were previously source files.
Matthew Newton [Thu, 5 Sep 2019 14:31:52 +0000 (15:31 +0100)]
travis is broken
It's been chopping off debug output for months. See if this hack
from https://github.com/travis-ci/travis-ci/issues/8920 helps
Alan T. DeKok [Thu, 5 Sep 2019 14:23:25 +0000 (10:23 -0400)]
Revert "Add openssl-perl as a BuildRequires in the Redhat spec file"
This reverts commit
d7cdb717fc336f68ed8faea7278468c4881c4eae .
revert this for now as docker isn't picking up the package.
Which then lets us see / fix other potential build failures
Alan T. DeKok [Thu, 5 Sep 2019 14:22:52 +0000 (10:22 -0400)]
fix patch for new docs