]>
git.ipfire.org Git - thirdparty/freeradius-server.git/log
Arran Cudbard-Bell [Tue, 2 Jul 2024 23:06:38 +0000 (17:06 -0600)]
Allow the name of the instant client packages to be specified
Arran Cudbard-Bell [Tue, 2 Jul 2024 20:53:29 +0000 (14:53 -0600)]
Use the new macro name
Arran Cudbard-Bell [Tue, 2 Jul 2024 20:16:21 +0000 (14:16 -0600)]
Don't set the oracle version by default
Arran Cudbard-Bell [Tue, 2 Jul 2024 18:58:05 +0000 (12:58 -0600)]
Revert "Take RPMBUILD_FLAGS in builddep too"
This reverts commit
675e4e3f41b326b4561ca89ee80bee444a7e019c .
Arran Cudbard-Bell [Tue, 2 Jul 2024 18:57:48 +0000 (12:57 -0600)]
Revert "Pass in the oracle version explicitly"
This reverts commit
2b1a72f8ffb1d7e526cd7c55c7319332b1733ff9 .
Arran Cudbard-Bell [Tue, 2 Jul 2024 18:53:12 +0000 (12:53 -0600)]
Move definitions to the top of the spec file
Arran Cudbard-Bell [Tue, 2 Jul 2024 18:43:19 +0000 (12:43 -0600)]
Pass in the oracle version explicitly
All the rpmbuild tools seem very broken
Arran Cudbard-Bell [Tue, 2 Jul 2024 18:27:40 +0000 (12:27 -0600)]
Take RPMBUILD_FLAGS in builddep too
Arran Cudbard-Bell [Tue, 2 Jul 2024 18:13:08 +0000 (12:13 -0600)]
rpmbuild doesn't like the separation
Arran Cudbard-Bell [Tue, 2 Jul 2024 18:02:44 +0000 (12:02 -0600)]
Accept --define _oracle_version
Arran Cudbard-Bell [Tue, 2 Jul 2024 17:19:55 +0000 (11:19 -0600)]
Regenerate the _ORACLE_ configure script
Arran Cudbard-Bell [Mon, 1 Jul 2024 23:19:36 +0000 (17:19 -0600)]
Update rlm_sql_oracle configure scripts
Arran Cudbard-Bell [Sun, 30 Jun 2024 22:33:55 +0000 (16:33 -0600)]
Const
Arran Cudbard-Bell [Sun, 30 Jun 2024 20:00:27 +0000 (14:00 -0600)]
WS
Arran Cudbard-Bell [Sun, 30 Jun 2024 19:33:28 +0000 (13:33 -0600)]
Remove invalid assert that was triggered evaluating log messages
Arran Cudbard-Bell [Sat, 29 Jun 2024 03:55:50 +0000 (21:55 -0600)]
Sync require_message_authenticator field names, it's called three different things at different places in the server
Nick Porter [Fri, 28 Jun 2024 11:20:50 +0000 (12:20 +0100)]
Set a capacity for trunk backlog
And allow requests to be allocated / enqueued even if the max
connections and requests per connection limit is reached, up to the
backlog limit.
Helps handle spikes in traffic, particularly where trunks are limited to
one request per connection, and there are server side limitations on the
number of connections that can be made - e.g. SQL servers.
Nick Porter [Fri, 28 Jun 2024 12:46:41 +0000 (13:46 +0100)]
Run docker build on ubuntu-latest
Seems to resolve the issue with the debiansid container build hanging
Nick Porter [Fri, 28 Jun 2024 09:43:12 +0000 (10:43 +0100)]
Always mark trunk requests as complete when freeing query ctx
Due to query_ctx being re-used multiple times when running transactions
(e.g. allocating IPs in sqlippool) it is possible for the current query
to not be submitted, but there still to be a trunk request associated
with the query_ctx.
Arran Cudbard-Bell [Fri, 28 Jun 2024 03:58:34 +0000 (21:58 -0600)]
Typo
Arran Cudbard-Bell [Thu, 27 Jun 2024 17:44:00 +0000 (11:44 -0600)]
Refactor LDAP URI xlat code so that we can call it easily from a future modify and delete xlats
Arran Cudbard-Bell [Thu, 27 Jun 2024 17:43:20 +0000 (11:43 -0600)]
Add API level support for deleting objects
This is just for completeness. We don't currently have any use case for it, but it'll likely be called by a delete xlat in the near future.
Arran Cudbard-Bell [Thu, 27 Jun 2024 17:42:12 +0000 (11:42 -0600)]
Whitespace fixes
Arran Cudbard-Bell [Thu, 27 Jun 2024 16:39:50 +0000 (10:39 -0600)]
Fix struct names
Arran Cudbard-Bell [Thu, 27 Jun 2024 16:19:05 +0000 (10:19 -0600)]
s/memberof/group/g for group checking XLAT to allow redundant xlats over SQL and LDAP
James Jones [Thu, 27 Jun 2024 14:00:32 +0000 (09:00 -0500)]
Do size calculation in appropriate type (CID #
1604603 )
In the error message in lst_expand(), the calculation of the size
has to be done in size_t. The cast of sizeof(void *) to uint32_t
causes it to be done in uint32_t, but that has the possibility of
overflow; note that n_capacity may be set to UINT_MAX, which would
guarantee it. The format conversion is changed to match.
Stephen Blackwell [Wed, 26 Jun 2024 21:14:56 +0000 (15:14 -0600)]
cassandra start query is missing two escape characters
James Jones [Wed, 26 Jun 2024 16:36:09 +0000 (11:36 -0500)]
Only add positive r values to total (CID #
16046001 )
With the addition of r to total in the for loop header, the continue
for EINTR causes a negative value of r to be added to total. That's the
only thing that looks like it could cause an overflow and taint total.
Nick Porter [Wed, 26 Jun 2024 12:22:31 +0000 (13:22 +0100)]
Add open_query to MySQL queries.conf
Particularly in the case of sqlippool use, it has been found that
setting the transaction isolation level to READ COMMITTED is needed to
avoid deadlocks.
Nick Porter [Wed, 26 Jun 2024 12:17:15 +0000 (13:17 +0100)]
Reinstate running open_query for MySQL
Nick Porter [Wed, 26 Jun 2024 12:12:47 +0000 (13:12 +0100)]
Correct enum case
James Jones [Tue, 25 Jun 2024 16:37:32 +0000 (11:37 -0500)]
Guard allocation with MEM() (CID #
1604618 )
Nick Porter [Tue, 25 Jun 2024 14:53:15 +0000 (15:53 +0100)]
Correct docs
Nick Porter [Mon, 24 Jun 2024 18:23:34 +0000 (19:23 +0100)]
sqlippool query fixes
Nick Porter [Mon, 24 Jun 2024 17:35:32 +0000 (18:35 +0100)]
Don't attempt to run zero length queries
Since these come from conf file entries it's all to easy for something
such as
alloc_requested = ''
to be set rather than commenting out the entry.
Nick Porter [Mon, 24 Jun 2024 13:46:06 +0000 (14:46 +0100)]
Add client_port to proto_dhcpv4_udp
Nick Porter [Fri, 21 Jun 2024 16:02:47 +0000 (17:02 +0100)]
Correct DHCP attribute sorting comparison function
Alan T. DeKok [Fri, 21 Jun 2024 16:12:28 +0000 (12:12 -0400)]
add more documentation
Alan T. DeKok [Fri, 21 Jun 2024 16:04:41 +0000 (12:04 -0400)]
correct decode of length + offset fields in structs
Alan T. DeKok [Fri, 21 Jun 2024 16:03:14 +0000 (12:03 -0400)]
the correct offset is 3
Alan T. DeKok [Fri, 21 Jun 2024 16:02:33 +0000 (12:02 -0400)]
the offset here is 3
Alan T. DeKok [Fri, 21 Jun 2024 14:53:44 +0000 (10:53 -0400)]
add tests for simple password auth type
Nick Porter [Fri, 21 Jun 2024 10:45:26 +0000 (11:45 +0100)]
Remove spurious debug message
Matthew Newton [Fri, 21 Jun 2024 09:06:39 +0000 (10:06 +0100)]
Fix error in dictionary documentation
Fixes #5345
Arran Cudbard-Bell [Fri, 21 Jun 2024 00:31:52 +0000 (18:31 -0600)]
More const issues
Arran Cudbard-Bell [Fri, 21 Jun 2024 00:06:36 +0000 (18:06 -0600)]
Fix some missed renames
Arran Cudbard-Bell [Thu, 20 Jun 2024 23:51:01 +0000 (17:51 -0600)]
Const issues
Arran Cudbard-Bell [Thu, 20 Jun 2024 23:36:01 +0000 (17:36 -0600)]
Evaluate all chained method groups when searching for methods or printing methods
Arran Cudbard-Bell [Thu, 20 Jun 2024 23:18:02 +0000 (17:18 -0600)]
Common indent levels in module_rlm.h
Arran Cudbard-Bell [Thu, 20 Jun 2024 23:16:42 +0000 (17:16 -0600)]
Remove unused functions
Arran Cudbard-Bell [Thu, 20 Jun 2024 23:13:27 +0000 (17:13 -0600)]
Initialise name
Arran Cudbard-Bell [Thu, 20 Jun 2024 23:13:17 +0000 (17:13 -0600)]
Correct example xlat calls
Arran Cudbard-Bell [Thu, 20 Jun 2024 21:57:23 +0000 (15:57 -0600)]
Final set of changes for making redundant xlat function registration not suck
We now group module xlats by name (all self-named module expansions are considered equal).
This allows:
redundant {
sql
linelog
}
Which would issue a query against SQL, and then log it to disk.
The instantiate code catches/disallows calls to incompatible module expansions in much the same way as C's _Generic does...
Nick Porter [Thu, 20 Jun 2024 16:09:32 +0000 (17:09 +0100)]
Test attr_filter on reply list
Nick Porter [Thu, 20 Jun 2024 16:08:56 +0000 (17:08 +0100)]
Use correct context for pairs
Nick Porter [Thu, 20 Jun 2024 15:53:26 +0000 (16:53 +0100)]
Add list name based methods to attr_filter
James Jones [Tue, 4 Jun 2024 13:57:21 +0000 (08:57 -0500)]
Python script to add an lldb command
This, once imported with an lldb "script" command, adds a "dd" command
to display certain types of values in an intelligible fashion using the
existing foo_debug() functions not only is it intelligible, it uses a
format people are used to.
The foo_debug() functions take a pointer to the structure. The dd command
will take names of variables that have the structure type as well, and
pass their address.
James Jones [Thu, 13 Jun 2024 21:47:26 +0000 (16:47 -0500)]
Add doc file for lldb
Initially at least it concentrates on extending it with Python scripts,
though it does describe how to adapt the sample command line output
when a test fails to invoke lldb on the program being tested.
Nick Porter [Thu, 20 Jun 2024 09:00:36 +0000 (10:00 +0100)]
Don't use MySQL read / write timeout options
The model for async is different to the sync client. Timeouts are
expected to be handled by the calling application rather than the client
library.
Arran Cudbard-Bell [Wed, 19 Jun 2024 20:39:58 +0000 (14:39 -0600)]
Backticks
Arran Cudbard-Bell [Wed, 19 Jun 2024 20:37:35 +0000 (14:37 -0600)]
Use the correct context
Arran Cudbard-Bell [Wed, 19 Jun 2024 20:11:52 +0000 (14:11 -0600)]
Put the proxy reply in the right list...
Arran Cudbard-Bell [Wed, 19 Jun 2024 20:11:40 +0000 (14:11 -0600)]
Fix confusing name
Alan T. DeKok [Wed, 19 Jun 2024 12:49:17 +0000 (08:49 -0400)]
quiet coverity
We can decode packets if there's no original request, but we
cannot verify the Response Authenticator
Alan T. DeKok [Wed, 19 Jun 2024 12:47:53 +0000 (08:47 -0400)]
Status-Client has no definition and is never used
Alan T. DeKok [Wed, 12 Jun 2024 18:50:33 +0000 (14:50 -0400)]
add connection_timeout to RADIUS client BIO
Alan T. DeKok [Wed, 12 Jun 2024 18:08:30 +0000 (14:08 -0400)]
remove UNUSED
variables are used.
Alan T. DeKok [Wed, 12 Jun 2024 18:07:05 +0000 (14:07 -0400)]
fix compiler warning
Nick Porter [Tue, 18 Jun 2024 15:02:14 +0000 (16:02 +0100)]
Ensure queries are freed before pool connection is released
Arran Cudbard-Bell [Tue, 18 Jun 2024 02:58:37 +0000 (20:58 -0600)]
Fix docs
Arran Cudbard-Bell [Tue, 18 Jun 2024 02:48:12 +0000 (20:48 -0600)]
Fix "RADIUS checksum" mode in radsniff, when dealing with Access-Requests
Nick Porter [Mon, 17 Jun 2024 16:04:18 +0000 (17:04 +0100)]
Add 3GPP2 attributes
Found in 3GPP2 specification document A.S0008-C
Arran Cudbard-Bell [Sun, 16 Jun 2024 04:53:42 +0000 (22:53 -0600)]
Record which xlat registered the module instance
Arran Cudbard-Bell [Sun, 16 Jun 2024 04:39:00 +0000 (22:39 -0600)]
Not supported by GCC
Arran Cudbard-Bell [Sun, 9 Jun 2024 22:46:00 +0000 (18:46 -0400)]
Wrap the module method array in a grouping structure
Nick Porter [Fri, 14 Jun 2024 11:56:54 +0000 (12:56 +0100)]
Fix policy name
There is no preacct section or method any more
Nick Porter [Fri, 14 Jun 2024 08:55:35 +0000 (09:55 +0100)]
Remove dangling symlink
Nick Porter [Tue, 11 Jun 2024 09:42:03 +0000 (10:42 +0100)]
Pacify Coverity (CID #
1598300 )
Nick Porter [Mon, 10 Jun 2024 07:40:28 +0000 (08:40 +0100)]
Satisfy both clang scan and Coverity (CID #
1603335 )
Arran Cudbard-Bell [Sun, 9 Jun 2024 22:02:46 +0000 (18:02 -0400)]
Whitespace
James Jones [Mon, 6 May 2024 17:24:50 +0000 (12:24 -0500)]
Correct uninitialized scalar false positives (CID #
1504037 , #
1503938 )
Coverity doesn't realize that uninitialized char arrays used as
buffer space for sbuffs are in fact set by sbuff functions that
write to the sbuff. (Yes, we tried modeling, but it didn't work.)
The solution is to keep the sbuff FR_SBUFF_OUT() creates around
long enough to use fr_sbuff_start(&sbuff) "instead of" the array
(even though it really is the array); that satisfies coverity.
Arran Cudbard-Bell [Sat, 8 Jun 2024 20:03:47 +0000 (16:03 -0400)]
s/idle/reapable/ in trunk code
Idle is a little ambiguous. Reapable describes more or less the exact behaviour that's implemented... treq finishes, and then hangs around in the trunk waiting for something to collect its status code, and other outputs.
Arran Cudbard-Bell [Sat, 8 Jun 2024 20:01:59 +0000 (16:01 -0400)]
These are attribute numbers...
Arran Cudbard-Bell [Sat, 8 Jun 2024 20:01:40 +0000 (16:01 -0400)]
Fix conflict between pub/private function
Arran Cudbard-Bell [Sat, 8 Jun 2024 19:53:07 +0000 (15:53 -0400)]
Wrap trunk allocations in MEM()
Arran Cudbard-Bell [Sat, 8 Jun 2024 19:51:13 +0000 (15:51 -0400)]
Fix enums
Arran Cudbard-Bell [Sat, 8 Jun 2024 19:47:19 +0000 (15:47 -0400)]
Remove fr_ prefix from trunk and connection
Arran Cudbard-Bell [Sat, 8 Jun 2024 19:40:48 +0000 (15:40 -0400)]
Not needed, we always sort the bindings now
Arran Cudbard-Bell [Fri, 7 Jun 2024 21:00:00 +0000 (17:00 -0400)]
Clear references to markers in fr_sbuff_reset_talloc
Arran Cudbard-Bell [Fri, 7 Jun 2024 01:32:48 +0000 (21:32 -0400)]
Quiet compiler
Arran Cudbard-Bell [Fri, 7 Jun 2024 00:30:52 +0000 (20:30 -0400)]
Use new module method resolution function
This supports static and dynamic modules
Pass in call_env_ctx_t to parsing callbacks instead of the section name, module instance data directly
Arran Cudbard-Bell [Fri, 7 Jun 2024 00:23:14 +0000 (20:23 -0400)]
Method names need to match section names exactly
Arran Cudbard-Bell [Sat, 1 Jun 2024 01:32:41 +0000 (19:32 -0600)]
Use correct type for cf_pair_debug
Arran Cudbard-Bell [Sat, 1 Jun 2024 01:32:27 +0000 (19:32 -0600)]
Fixes to section cmp functions
Arran Cudbard-Bell [Sat, 1 Jun 2024 01:29:04 +0000 (19:29 -0600)]
Move module xlat registrations into the module code
Arran Cudbard-Bell [Tue, 28 May 2024 02:03:13 +0000 (22:03 -0400)]
Split section name match
Alan T. DeKok [Fri, 7 Jun 2024 14:40:10 +0000 (10:40 -0400)]
double-check iterations size
Arran Cudbard-Bell [Fri, 7 Jun 2024 03:02:42 +0000 (23:02 -0400)]
Fixup configure script to look for MariaDB specifically
Nick Porter [Tue, 4 Jun 2024 16:56:58 +0000 (17:56 +0100)]
Don't crash if fr_trunk_request_alloc() returns NULL
This is typically due to configuration limits
Nick Porter [Tue, 4 Jun 2024 07:42:42 +0000 (08:42 +0100)]
Don't close the final connection on a trunk if the last event was a failure
Closing this last connection can lead to no active connections and the
last event being a failure, which then blocks enqueueing of new
requests.