summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* Add PRNG improvements to CHANGES.Linus Nordberg2014-02-031-0/+2
|
* Update documentation with info on multithread protection.Linus Nordberg2013-12-201-1/+6
|
* Detect POSIX threads at configure time and use them for OpenSSL.Linus Nordberg2013-12-203-2/+65
|
* Don't use deprecated 'INCLUDES' variable in Makefile.am files.Linus Nordberg2013-12-205-5/+5
|
* Remove unused code.Linus Nordberg2013-12-202-77/+0
|
* Improve initialisation of OpenSSL PRNG.Linus Nordberg2013-12-207-25/+99
| | | | Basic idea taken from Tor.
* Bump version number.Linus Nordberg2013-12-181-1/+1
|
* Distribute new CHANGES file.libradsec-0.0.4Linus Nordberg2013-12-181-1/+1
|
* Update documentation and configure.ac for libradsec-0.0.4.Linus Nordberg2013-12-184-7/+15
|
* Clarify README.Linus Nordberg2013-12-181-3/+4
|
* Add a debug printout when dropping an incoming response message.Linus Nordberg2013-11-181-1/+3
| | | | Also, terminate another debug message with \n.
* Avoid leaking memory when receiving a bad response.Linus Nordberg2013-11-181-11/+14
| | | | | | | | A badly authenticated response message or one that didn't decode or decrypt correctly was never freed. If caller didn't pass pkt_out, any response was leaked as well. As a bonus, the code is now readable too.
* Stop rs_request_send() from leaking RSE_TIMEOUT_CONN and RSE_TIMEOUT_IO.Linus Nordberg2013-11-181-5/+7
| | | | | | | | If sending or receiving time out, pop the error off the stack before continuing the loop. Push a new error, RS_TIMEOUT, before timing out for real. Addresses LIBRADSEC-3.
* Disable/delete events in the timeout callback.Linus Nordberg2013-11-181-0/+15
| | | | | | | | | | | | | Without this, the sequence (1) send (successfully) (2) receive (timeout) (3) resend (successfully) will have the read event from (2) trigger a read in (3) and the response is lost. Addresses LIBRADSEC-3.
* Make sure that configure without --enable-tls works.Linus Nordberg2013-11-151-11/+14
| | | | | Also, --disable-tls, --enable-tls and --enable-tls=no. Closes LIBRADSEC-2.
* Merge branch 'libradsec' of git.nordu.net:radsecproxy into libradsecLinus Nordberg2013-10-031-1/+2
|\ | | | | | | | | | | | | | | # Please enter a commit message to explain why this merge is necessary, # especially if it merges an updated upstream into a topic branch. # # Lines starting with '#' will be ignored, and an empty message aborts # the commit.
| * Enable TLS by default, configure time.Linus Nordberg2013-10-011-1/+2
| |
* | Implement disable_hostname_check config option.Linus Nordberg2013-10-033-0/+6
|/ | | | Patch by Sam Hartman.
* Have the library depend on radsec.sym.Linus Nordberg2013-05-171-0/+1
|
* Merge branch 'libradsec-add-avp-2' into libradsecLinus Nordberg2013-05-175-5/+61
|\ | | | | | | | | Conflicts: lib/Makefile.am
| * Export rs_packet_add_avp() and set library revision.libradsec-add-avp-2Linus Nordberg2013-05-082-1/+2
| | | | | | | | | | Correct library revision is 1:0:1, given that the last "released" library was 0:0:0. (The current 1:0:0 is wrong.)
| * Add rs_packet_add_avp() and use it.Linus Nordberg2013-05-083-4/+59
| | | | | | | | | | | | | | | | | | | | | | | | | | rs_packet_create_authn_request() now uses rs_packet_add_avp() instead of rs_packet_append_avp() which makes it possible to create a authentication packet without knowing the shared secret. Calling rs_packet_add_avp() on a packet is incompatible with using rs_packet_append_avp() on the same packet but since rs_packet_create_authn_request() adds attribute-value pairs for user name and password only if those arguments are supplied, code that doesn't use user name and password (i.e. mech_eap) should still be fine.
* | Don't provide bogus info in an error.Linus Nordberg2013-05-171-3/+1
| |
* | Move PSK example configuration to a separate file.Linus Nordberg2013-05-172-15/+18
| |
* | Fix typo.Linus Nordberg2013-05-151-1/+1
| |
* | Don't crash on reading invalid messages.Linus Nordberg2013-05-154-37/+40
| | | | | | | | | | Also, invoke disconnected callback and close connection in error cases.
* | Whitespace changes in license headers.Linus Nordberg2013-05-0915-15/+15
| |
* | Update copyright years.Linus Nordberg2013-05-0918-31/+35
| |
* | Include stdlib.h everywhere we call (m|c)alloc.Linus Nordberg2013-05-093-2/+5
| |
* | Use malloc+memcpy rather than calloc+strcpy in rs_strdup.Linus Nordberg2013-05-091-5/+10
| | | | | | | | For effiency (but triggered by calloc needing unistd.h on Darwin).
* | Revert "Bump library interface revision."Linus Nordberg2013-05-081-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | So there were two things wrong with that commit. 1. Library interface revision should be bumped only immediately before a public release. 2. Given the changes in the library, it should change to 0:1:0 since the interface didn't change (including not being extended). This reverts commit b9e967b3cde6af41cd0e863e9ff073897019625a. Conflicts: lib/Makefile.am
* | Clarify and reformat comments on how to use Libtool's -version-info.Linus Nordberg2013-05-081-10/+10
| |
* | Pass make distcheck.Linus Nordberg2013-05-085-12/+19
| |
* | Don't include tests in SUBIDRS after all.Linus Nordberg2013-05-081-1/+2
| | | | | | | | | | It requires a running radius server. That's not a nice thing to require for something like distcheck.
* | Initial RPM packagingLinus Nordberg2013-05-083-5/+97
| | | | | | | | | | | | | | | | | | | | | | Adapted from commit 8ff4e9ab2308fc6ee1e9b140d85ba45eff5287ce Author: Sam hartman <hartmans@painless-security.com> Date: Mon Oct 10 15:25:11 2011 +0100 Conflicts: lib/Makefile.am lib/configure.ac
* | Add a few bits to README.Linus Nordberg2013-05-081-8/+23
| |
* | Exit tests with number of failures.Linus Nordberg2013-05-081-2/+4
| | | | | | | | Now 'make check' really fails when a test fails.
* | Add 'tests' to SUBDIRS to make 'make check' work.Linus Nordberg2013-05-081-1/+1
| |
* | Constify the MD5 implementation.Linus Nordberg2013-05-082-4/+4
| |
* | Push an error on the error stack when returning !RSE_OK.Linus Nordberg2013-05-081-1/+1
|/
* Revive RSE_MAX.Linus Nordberg2013-05-081-0/+1
| | | | It's being used after all.
* Add an assert in error handling code.Linus Nordberg2013-05-081-0/+2
|
* Remove an unused error code and unusued RSE_MAX.Linus Nordberg2013-05-082-50/+0
| | | | Also, remove unused file attr.c.
* Follow API change in tests.Linus Nordberg2013-05-081-5/+3
|
* Revert "Add formal argument 'secret' to two public functions."Linus Nordberg2013-05-085-21/+12
| | | | | | | This reverts commit 09d1cff2418a900b587b2113f508984f2417cc11. Conflicts: lib/include/radsec/request.h
* Add and fix RADIUS attributes (4b9e4cb1, e4b6e972).Sam Hartman2013-05-073-5/+12
| | | | | | | | | Fix capitalization in abfab dictionary Update to IETF RADIUS attributes draft-ietf-abfab-gss-eap has been approved; include IANA-issued standard radius attributes for Moonshot. Fix capitalization in abfab dictionary
* Build include before building '.'Linus Nordberg2013-05-071-1/+1
| | | | Patch by Sam Hartman (ff1af013 in moonshot).
* Use CUnit for tests.Linus Nordberg2013-05-067-78/+71
| | | | cgreen didn't seem properly maintained. CUnit seems to be widely used.
* Bump library interface revision.Linus Nordberg2013-05-061-1/+1
| | | | | | | Commit edf4c047 claimed it did this but didn't really do it. Should really have been done as part of bumping the library version (0.0.2.dev in configure.ac).
* Bump version to 0.0.4.dev to keep ahead of moonshot.Linus Nordberg2013-05-062-2/+2
|
generated by cgit v1.1 at 2024-12-27 07:32:30 +0000