aland [Fri, 1 Dec 2000 18:48:33 +0000 (18:48 +0000)]
oops... pass a pointer, not an int
aland [Fri, 1 Dec 2000 18:41:54 +0000 (18:41 +0000)]
documentation on how to submit patches to the server, and what
sort of coding methods/styles to follow
aland [Fri, 1 Dec 2000 18:11:49 +0000 (18:11 +0000)]
removed the definition of INCLTDL. It's defined properly in the
top-level 'Make.inc'
aland [Thu, 30 Nov 2000 22:00:20 +0000 (22:00 +0000)]
minor updates (alignment, etc), based on warnings from Solaris build
jcarneal [Thu, 30 Nov 2000 21:57:23 +0000 (21:57 +0000)]
Removing some debugging stuff I left in.
jcarneal [Thu, 30 Nov 2000 20:57:00 +0000 (20:57 +0000)]
Added INCLTDL so modules.c will compile on Solaris.
jcarneal [Thu, 30 Nov 2000 20:53:05 +0000 (20:53 +0000)]
Several changes:
1. Cleaned up auth / user find code
2. Added functionality for multiple users with same check items
3. Put all users at end of hash bucket linked list
4. Add preacct and acctusersfile capability from rlm_files
Tried to change the module to rlm_users in the repository, but
'cvs remove' / 'cvs add' and I aren't getting along well.
The functionality now *should* exactly mimic rlm_files.
jcarneal [Thu, 30 Nov 2000 20:06:46 +0000 (20:06 +0000)]
Bugfix. Don't want to overwrite 'i' when updating
tailto after pairdelete.
cmiller [Thu, 30 Nov 2000 18:31:00 +0000 (18:31 +0000)]
Fixed a buglet in the add() code and finished the next() code.
cmiller [Thu, 30 Nov 2000 18:29:38 +0000 (18:29 +0000)]
Added explicit rules for request_{tree,list}.o , for consistency.
hartwick [Thu, 30 Nov 2000 16:56:26 +0000 (16:56 +0000)]
* Remove all references to %y which was removed from the code.
aland [Thu, 30 Nov 2000 15:55:29 +0000 (15:55 +0000)]
massive changes to enable smarter configure script, so that it
auto-discovers SNMP. (most times, I hope...)
Updates to various SNMP header and source files, to make sure it
builds, and use massively paranoid 'const' everywhere.
aland [Thu, 30 Nov 2000 15:49:20 +0000 (15:49 +0000)]
made more variables 'const'
aland [Thu, 30 Nov 2000 15:45:01 +0000 (15:45 +0000)]
added 'const'
aland [Thu, 30 Nov 2000 15:23:25 +0000 (15:23 +0000)]
corrected typo in SNMP data structure
aland [Thu, 30 Nov 2000 14:40:32 +0000 (14:40 +0000)]
Added some more comments.
Removed 'radutmp' from the 'session' section, as it isn't finished
cmiller [Wed, 29 Nov 2000 21:41:30 +0000 (21:41 +0000)]
Added a next() routine to the request_tree schtick.
aland [Wed, 29 Nov 2000 19:23:09 +0000 (19:23 +0000)]
The VP pointed to by 'tailto' may have been deleted in pairdelete().
After calling pairdelete(), update 'tailto' again.
aland [Wed, 29 Nov 2000 17:15:27 +0000 (17:15 +0000)]
If we're NOT using developer C flags, then define -DNDEBUG, so
that the assert's go away.
cmiller [Wed, 29 Nov 2000 14:07:34 +0000 (14:07 +0000)]
Found another possible NULL deref point. Alan's(=efficient) coding
style is slightly different than mine(=pedantic).
cmiller [Wed, 29 Nov 2000 13:57:00 +0000 (13:57 +0000)]
Fixed NULL deref. Sorry, Jeff!
aland [Tue, 28 Nov 2000 20:11:26 +0000 (20:11 +0000)]
corrected minor typo
aland [Tue, 28 Nov 2000 15:37:37 +0000 (15:37 +0000)]
single-threaded ldap code from Weston Bustraan <weston@itdonline.net>
aland [Mon, 27 Nov 2000 23:47:09 +0000 (23:47 +0000)]
initial implementation of mschap, from patch supplied for Cistron.
it builds, but I have no idea if it works or not...
cmiller [Mon, 27 Nov 2000 23:24:16 +0000 (23:24 +0000)]
Removed REQUEST struct's reference to next and prev -- these are now hidden.
cmiller [Mon, 27 Nov 2000 23:21:14 +0000 (23:21 +0000)]
Added my as-yet lame tree storage method. Need to add next() and
find_proxy() methods.
cmiller [Mon, 27 Nov 2000 23:19:40 +0000 (23:19 +0000)]
Generalized REQUEST structure so that core code really doesn't know
anything about the storage.
cmiller [Mon, 27 Nov 2000 23:15:59 +0000 (23:15 +0000)]
rlm_preprocess uses auth_name(), so it can't be static.
aland [Mon, 27 Nov 2000 17:07:42 +0000 (17:07 +0000)]
made one more function 'static' to the only source file that
uses it.
cmiller [Sun, 26 Nov 2000 23:40:22 +0000 (23:40 +0000)]
Completed debian changes. (Built first package.)
autoconf'd on top level. Maybe this should be in the Makefile.
cmiller [Sun, 26 Nov 2000 14:46:29 +0000 (14:46 +0000)]
Fixed misplaced space in CFLAGS generation.
aland [Fri, 24 Nov 2000 22:13:29 +0000 (22:13 +0000)]
added '--with-developer'. For people with CVS directories,
and using GCC, it assumes that they're developers, and adds mondo
paranoid GCC warning flags to CFLAGS.
aland [Fri, 24 Nov 2000 21:24:29 +0000 (21:24 +0000)]
made attrib_nmc compiled in by default. There's no point in
having it #ifdef'd in the source.
aland [Fri, 24 Nov 2000 21:23:41 +0000 (21:23 +0000)]
belated set of updates
aland [Fri, 24 Nov 2000 21:23:17 +0000 (21:23 +0000)]
made USR (attrib_nmc) attributes compiled in by default.
aland [Fri, 24 Nov 2000 21:20:54 +0000 (21:20 +0000)]
removed no longer used build flags from the version string
aland [Fri, 24 Nov 2000 19:23:45 +0000 (19:23 +0000)]
Correct the method fo rsnipping attributes to length.
Instead of dropping them, snip them to the maximum allowed length.
Do this for ALL attributes, based on the pair length + the attribute
header, and NOT on MAX_STRING_LEN.
aland [Fri, 24 Nov 2000 19:14:14 +0000 (19:14 +0000)]
corrected typo with USR attributes
aland [Fri, 24 Nov 2000 19:11:08 +0000 (19:11 +0000)]
rad_send() now packs multiple vendor attributes into one VSA,
if it can. If not, it uses multiple VSA's.
It also checks if the vendor attribute *can* be packed into a VSA.
If not, it silently ignores that attribute.
cmiller [Fri, 24 Nov 2000 19:10:41 +0000 (19:10 +0000)]
Repaired make expansion of $$foo to $${foo}.
aland [Fri, 24 Nov 2000 17:54:21 +0000 (17:54 +0000)]
add MD5 to the list of targets
cmiller [Fri, 24 Nov 2000 14:46:15 +0000 (14:46 +0000)]
Updated stuff needed to build a debian package.
cmiller [Fri, 24 Nov 2000 14:43:36 +0000 (14:43 +0000)]
Added $(R)=root setting.
cmiller [Thu, 23 Nov 2000 19:41:44 +0000 (19:41 +0000)]
Ran automake, sorta.
aland [Thu, 23 Nov 2000 17:57:57 +0000 (17:57 +0000)]
defined new function: AC_CHECK_TYPE_INCLUDE, which is like the
pre-defined AC_CHECK_TYPE, but it also allows you to specify a
list of include files to check.
This function simplifies a number of checks in the 'configure.in'
script
aland [Thu, 23 Nov 2000 16:41:47 +0000 (16:41 +0000)]
added more people to the official list of credits
aland [Thu, 23 Nov 2000 15:30:44 +0000 (15:30 +0000)]
Applied patch from Chad Miller <cmiller@surfsouth.com>
Add $(R) in front of install targets, so that the user can build
a 'tar' image which will untar to the normal root, WITHOUT
touching the root, or installing the module. e.g.
make R=/home/user/fake-root/ install
will install everying to /home/user/fake-root/usr/bin
/home/user/fake-root/etc, etc.
jcarneal [Thu, 23 Nov 2000 00:10:52 +0000 (00:10 +0000)]
Jeff Carneal <jeff@apex.net>
Usercollide bugfix.
aland [Wed, 22 Nov 2000 21:36:20 +0000 (21:36 +0000)]
moved the LTDL_SET_PRELOADED_SYMBOLS() call to *before* the call
to lt_dlinit(). This appears to make a huge difference in getting
ltdl to do what we want.
Note that the libtool documentation doesn't mention that this order
is necessary for libtool to go off it's crack habit...
aland [Wed, 22 Nov 2000 21:12:57 +0000 (21:12 +0000)]
removed un-needed variable
aland [Wed, 22 Nov 2000 21:12:24 +0000 (21:12 +0000)]
include libradius.h at the top of the include file list
aland [Wed, 22 Nov 2000 21:06:09 +0000 (21:06 +0000)]
properly build ONLY the static module, if it's one of the declared
static modules.
Also, don't build ANY dynamic modules if we're not doing shared libs
aland [Wed, 22 Nov 2000 20:22:14 +0000 (20:22 +0000)]
new configure option: --with-static-modules=QUOTED-MODULE-LIST
./configure --with-static-modules="pam ns_mta_md5 sql"
builds all shared libraries, BUT pam, ns_mta_md5, and sql are
statically linked to the server.
If we have --disable-shared, then all of the successfully built
modules are linked to the server at build time, through an in-line
shell script in src/main/Makefile
jcarneal [Wed, 22 Nov 2000 17:34:39 +0000 (17:34 +0000)]
Jeff Carneal <jeff@apex.net>
Minor changes for:
a) hash table statistics logging
b) storing users in the hash bucket forward (not reverse) order
aland [Wed, 22 Nov 2000 17:12:15 +0000 (17:12 +0000)]
define USE_SHARED_LIBS and USE_STATIC_LIBS in the top-level
Make.inc file. Update 'configure.in' to substitute them,
from the 'configure' --enable-shared and --enable-static defaults.
Update the modules rules so that the static/dynamic rules only
apply of the top-level 'USE_FOO_LIBS' is defined to 'yes'.
We still don't know how to pick up libltdl knowledge about the
*ability* to build static/dynamic objects... but that's for later,
I guess.
aland [Wed, 22 Nov 2000 17:06:14 +0000 (17:06 +0000)]
micro-patch from David Kerry <dk@snti.com>
aland [Wed, 22 Nov 2000 17:02:14 +0000 (17:02 +0000)]
added missing #endif
aland [Wed, 22 Nov 2000 16:43:29 +0000 (16:43 +0000)]
if we do NOT have setlinebuf, but we DO have setvbuf, THEN
fake out a setlinebuf by wrapping it in a #define which calls
setvbuf
aland [Wed, 22 Nov 2000 16:42:45 +0000 (16:42 +0000)]
check for setvbuf, too
jcarneal [Wed, 22 Nov 2000 00:50:27 +0000 (00:50 +0000)]
Jeff Carneal <jeff@apex.net>
The new rlm_fastusers. It should be a faster rlm_files now (without
the preacct function, which I will add later).
jcarneal [Wed, 22 Nov 2000 00:24:40 +0000 (00:24 +0000)]
Jeff Carneal <jeff@apex.net>
Usercollide bugfix. Musta deleted the config var before sometime.
jcarneal [Wed, 22 Nov 2000 00:21:12 +0000 (00:21 +0000)]
Jeff Carneal <jeff@apex.net>
The new rlm_fastusers. It should be a faster rlm_files now (without
the preacct function, which I will add later).
jcarneal [Tue, 21 Nov 2000 23:58:23 +0000 (23:58 +0000)]
Jeff Carneal <jeff@apex.net>
Bugfixes. Auth-type counting problem in rad_check_password().
Updated pairfree() calls with '&'
aland [Tue, 21 Nov 2000 21:24:59 +0000 (21:24 +0000)]
removed 'allowed' and 'trusted' attributes from the proxy receive
code. A module will replace this functionality.
aland [Tue, 21 Nov 2000 21:02:00 +0000 (21:02 +0000)]
pull the definition for CC out of the top-level Make.inc file.
bug found by David Kerry <dk@snti.com>
aland [Tue, 21 Nov 2000 20:47:59 +0000 (20:47 +0000)]
more examples of realms
aland [Tue, 21 Nov 2000 20:44:35 +0000 (20:44 +0000)]
minor re-formatting.
allow NULL realms. Bug found by David Kerry <dk@snti.com>
aland [Tue, 21 Nov 2000 20:22:40 +0000 (20:22 +0000)]
radclient now does it's own encryption of the Password and
CHAP-Password attributes. This change is because of the last
bug fix to lib/radius.c
aland [Tue, 21 Nov 2000 20:16:11 +0000 (20:16 +0000)]
when doing a rad_send(), do NOT encode the PW_PASSWORD or the
PW_CHAP_PASSWORD attribute. This is to prevent double encryption.
This also prevents the server from breaking the CHAP-Password
when doing proxying. Bug found by David Kerry <dk@snti.com>
aland [Tue, 21 Nov 2000 20:06:03 +0000 (20:06 +0000)]
patch from David Kerry <dk@snti.com> to properly define variables
aland [Tue, 21 Nov 2000 20:04:23 +0000 (20:04 +0000)]
ALWAYS use our own 'install-sh' program, as we cannot depend
on the system's one.
aland [Tue, 21 Nov 2000 18:31:07 +0000 (18:31 +0000)]
updated dependencies on include files
aland [Tue, 21 Nov 2000 18:30:07 +0000 (18:30 +0000)]
convince the thrice-damned ltldl code to actually open "foo.a",
like it's supposed to, like it claims to, but which it doesn't
do, until I've drop-kicked the fix through it's thick head.
aland [Tue, 21 Nov 2000 16:46:10 +0000 (16:46 +0000)]
use the new 'request->number' to uniquely tag each request
aland [Mon, 20 Nov 2000 21:51:47 +0000 (21:51 +0000)]
<whew>! Finally made the 'request_list' structure 'static'
to request_list.c. This allows us to completely change the
implementation, without breaking any external code.
Removed all references to 'request_list_busy' from radiusd.c
These sort of checks should be in request_list.c. (Not that
they're actually there right now...)
aland [Mon, 20 Nov 2000 21:36:51 +0000 (21:36 +0000)]
removed old code handling stripped-user-name from the 'authorize'
entry. That functionality is now done by rlm_preprocess
aland [Mon, 20 Nov 2000 21:31:22 +0000 (21:31 +0000)]
minor cleanups & touchups, to make the code simpler
aland [Mon, 20 Nov 2000 20:37:26 +0000 (20:37 +0000)]
added comment header with description of GPL.
added 'libradius.h' to top of includes, to fix include file
order.
aland [Mon, 20 Nov 2000 19:43:55 +0000 (19:43 +0000)]
updated arguments to pairfree()
aland [Mon, 20 Nov 2000 18:59:13 +0000 (18:59 +0000)]
changes as per latest discussion on the list
aland [Mon, 20 Nov 2000 17:56:13 +0000 (17:56 +0000)]
added a 'number' element to the REQUEST structure. This is a
unique value per request, incrementing for each new request.
This allows the debug messages to print out 'request NNNN',
which makes it easier for the administrator to correlate
log messages.
aland [Mon, 20 Nov 2000 16:54:05 +0000 (16:54 +0000)]
walk over 255 or 256 ID's
aland [Mon, 20 Nov 2000 16:50:00 +0000 (16:50 +0000)]
updated arguments to pairfree() and rad_free()
aland [Mon, 20 Nov 2000 16:48:55 +0000 (16:48 +0000)]
updated the arguments to pairfree()
aland [Mon, 20 Nov 2000 16:48:21 +0000 (16:48 +0000)]
updated arguments to pairfree()
aland [Mon, 20 Nov 2000 16:35:42 +0000 (16:35 +0000)]
made pairfree() take a VALUE_PAIR**, not a VALUE_PAIR*. This
allows it to set the input pointer to NULL.
aland [Mon, 20 Nov 2000 16:23:00 +0000 (16:23 +0000)]
made a request_free() and rad_free() take a '**' pointer, instead
of a '*' pointer. This allows those functions to set the input
variable to NULL, to explicitely tell the caller it's no longer
valid.
It also minimizes the code, as there was a lot of:
rad_free(foo);
foo = NULL;
jcarneal [Sat, 18 Nov 2000 18:08:30 +0000 (18:08 +0000)]
Jeff Carneal <jeff@apex.net>
Changed lower/nospace code slightly. It's now more configurable, if
slightly less efficient. You can now specify when to lower/nospace
both the user and the pass, rather than always doing both at the
same time.
jcarneal [Sat, 18 Nov 2000 16:44:17 +0000 (16:44 +0000)]
Jeff Carneal <jeff@apex.net>
Bugfig. Need to explicitly set the config item list to NULL
after freeing.
jcarneal [Sat, 18 Nov 2000 16:43:33 +0000 (16:43 +0000)]
Jeff Carneal <jeff@apex.net>
Bugfix. rl_next was returning the same request it was passed.
This hopefully fixes that.
aland [Fri, 17 Nov 2000 21:14:25 +0000 (21:14 +0000)]
new function: rl_num_requests(), which returns the number of
requests in the request list.
After this, there's only one more reference to the request_list
data structure, outside of request_list.c
aland [Fri, 17 Nov 2000 20:37:38 +0000 (20:37 +0000)]
miscellanous cleanups, and do NOT set the timestamp to zero.
Let refresh_request() take care of figuring out which requests
to delete.
aland [Fri, 17 Nov 2000 19:38:53 +0000 (19:38 +0000)]
be a little neater about checking last_request in rad_clean_list
jcarneal [Fri, 17 Nov 2000 19:08:29 +0000 (19:08 +0000)]
Bugfix. pairfree(), not pairlist_free().
aland [Fri, 17 Nov 2000 18:33:37 +0000 (18:33 +0000)]
Modified code to do intermediate processing of the request list,
as posted to the freeradius-devel list.
After spawning a new thread/process, the server now checks a few
old requests for deletion/proxy-retry/etc. This amortizes the
work over each second, and helps to minimize the bursty response
of the server.
aland [Fri, 17 Nov 2000 16:11:39 +0000 (16:11 +0000)]
added rl_next() function, which walks through the request list one
entry at a time.
aland [Fri, 17 Nov 2000 15:33:09 +0000 (15:33 +0000)]
two new functions: radutmp_lock() and radutmp_unlock(), to wrap
the platform-dependent lock functions, so that we have fewer
ifdef's in the code.
Also, when calling rad_check_ts, do unlock/check/lock, so that
we don't block other users from accessing the utmp file.
jcarneal [Fri, 17 Nov 2000 15:21:56 +0000 (15:21 +0000)]
Jeff Carneal <jeff@apex.net>
Bugfix on lower case stuff.
aland [Fri, 17 Nov 2000 15:20:11 +0000 (15:20 +0000)]
update long name with network & mask
aland [Thu, 16 Nov 2000 22:24:51 +0000 (22:24 +0000)]
added new 'component_names' array, which contains the component
names. This is so that we don't have multiple copies of them
all over the place....