Commit graph

268 commits

Author SHA1 Message Date
Aaron Jones
45d05d8882
OpenSSL: Improve error logging in rb_setup_ssl_server() 2016-09-15 20:12:21 +00:00
Aaron Jones
bd8097c459
OpenSSL: Tweak connection shutdown logic 2016-09-15 20:12:21 +00:00
Aaron Jones
485b5b8084
OpenSSL: Import the callback/handshake logic from the MbedTLS backend 2016-09-15 20:12:21 +00:00
Aaron Jones
9114e3a2dc
OpenSSL: Improve rb_setup_ssl_server()
* Move file/data assignments to the top of the function

* Don't attempt to set a hardcoded P-384 ECDH key if we have the new
  SSL_CTX_set1_curves_list() function (OpenSSL 1.0.2+)

* Rename variables consistent with other backends and wrap the function
  arguments.

* Disable OpenSSL's disabling of TLS 1/n-1 record splitting.
  In other words, enable TLS 1/n-1 record splitting.

* Other misc cleanups.
2016-09-15 20:12:21 +00:00
Aaron Jones
47d51fe3ac
OpenSSL: Use C99 __func__ declaration instead of writing function names 2016-09-15 20:12:21 +00:00
Aaron Jones
cc04fbe3f9
OpenSSL: Fix up rb_init_ssl() to use proper define from openssl_ratbox.h 2016-09-15 20:12:21 +00:00
Aaron Jones
62fc0eab03
OpenSSL: Rename error functions consistent with other backends. 2016-09-15 20:12:21 +00:00
Aaron Jones
4e9441a1cb
OpenSSL: Add generic direction enum for negotiation setup.
Also define an SSL_P(x) macro as in other backends and use that to refer
to the client session context.
2016-09-15 20:12:21 +00:00
Aaron Jones
e569720fe1
OpenSSL: Misc code cleanups
Make use of C99 for loop initialisers, declarations with immediate
rvalues, etc.
2016-09-15 20:12:20 +00:00
Aaron Jones
a61e06e1d1
OpenSSL: Add const-correctness to function and variable declarations. 2016-09-15 20:12:20 +00:00
Aaron Jones
1c39c519fe
OpenSSL: Reorder functions in line with the MbedTLS backend.
The diff for this commit will look like I have changed a lot of code;
in reality, nothing was changed, just whole functions moved up or down.
2016-09-15 20:12:20 +00:00
Aaron Jones
1c7d295320
OpenSSL: Move connect structure declaration to top of file 2016-09-15 20:12:20 +00:00
Aaron Jones
8a40573369
OpenSSL: Remove context duplication
OpenSSL is perfectly capable of having a single context that is shared
by both client and server sessions alike; one simply needs to call
SSL_set_accept_state (for server) or SSL_set_connect_state (for client)
before attempting handshaking.
2016-09-15 20:12:20 +00:00
Aaron Jones
2aec9b6d68
OpenSSL: Remove unnecessary handshake info callback 2016-09-15 20:12:20 +00:00
Aaron Jones
1f30c8943b
OpenSSL: Tidy up headers
Move all the header includes to a single header file, rename said file.
2016-09-15 20:12:20 +00:00
Aaron Jones
265dc4e53c
MbedTLS: Final round of const correctness
I'm happy with the state of this backend now.
I don't anticipate making any further changes.
2016-09-15 20:10:43 +00:00
Aaron Jones
f66a6390b0
MbedTLS: More const correctness 2016-09-15 13:24:29 +00:00
Aaron Jones
101c659117
MbedTLS: Cast addrlen rb_ssl_accept_setup to avoid compiler warnings 2016-09-15 13:24:29 +00:00
Aaron Jones
1083d8557b
MbedTLS: Cast return types for reading/writing only on success 2016-09-15 13:24:29 +00:00
Aaron Jones
f0ad82013c
MbedTLS: More const correctness 2016-09-15 13:24:28 +00:00
Aaron Jones
ac62792970
MbedTLS: Correct NULL checks for functions in line with other backends 2016-09-15 13:24:28 +00:00
Aaron Jones
988fedf212
MbedTLS: Move structure cert/key assignment to after cert/key loading 2016-09-15 13:24:28 +00:00
Aaron Jones
bef81a5d45
MbedTLS: Use C99 __func__ declaration instead of writing function names 2016-09-15 13:24:28 +00:00
Aaron Jones
8cd8b24ffb
MbedTLS: Make error string printing prettier. 2016-09-15 13:24:28 +00:00
Aaron Jones
db12df5c16
MbedTLS: Store error codes properly.
OpenSSL uses `unsigned long' type for its error codes, so that's
what (lib)ratbox used to store the error values.

Unfortunately, MbedTLS uses int, and its error codes are negative.
On machines where `int' and `long' are the same size, this could
result in storing a truncated error code.

This patch inverts the sign bit on error codes and then casts them
to unsigned long for storage.

MbedTLS itself (specifically, `mbedtls_strerror()') will function
properly with negative or positive input values. It even converts
negative input values to positive before checking them against the
list of known error codes!

See also: `library/error.c' in the MbedTLS 2.1+ distribution.
2016-09-15 13:24:28 +00:00
Aaron Jones
8668cb9b9d
MbedTLS: Const correctness in rb_ssl_init_fd
We shouldn't ever change this input variable.
Tell the compiler that we won't.
2016-09-15 13:24:28 +00:00
Aaron Jones
46c61dd478
MbedTLS: Set socket send/receive functions after initialising session 2016-09-15 13:24:28 +00:00
Aaron Jones
978c8ae828
MbedTLS: Move memory allocation to the beginning of rb_ssl_init_fd 2016-09-15 13:24:28 +00:00
Aaron Jones
163a4a9d06
MbedTLS: Remove default case in switch for an enum with all values
Having default here doesn't make sense (using something not in that
enum will generate a compile-time warning).
2016-09-15 13:24:28 +00:00
Aaron Jones
5b900411bf
MbedTLS: Rename error printing function
All 3 backends (MbedTLS, OpenSSL, GNUTLS) are going to have the same
function name for returning error strings. This will help to reduce the
diffs between them.
2016-09-15 13:24:28 +00:00
Aaron Jones
295c8f7d37
MbedTLS: Tidy up headers
Move all the header includes to a single header file, rename said file.
2016-09-15 13:24:25 +00:00
Aaron Jones
566f46785f
MbedTLS: Misc backend cleanups
* Add generic direction enum for negotiation setup.

* Rename a rather long wrapper function to a shorter one consistent with
  what it does.

* Rework context setup function.

* Don't check for handshake state before beginning handshaking.

  The old backend began a handshake and then stepped into the callback
  function if it was interrupted; the current one just jumps right into
  it, so there is no need to check if it has previously succeeded,
  because it hasn't been attempted yet.

* Add missing errno assignment to one of the handshake wrappers.

* Don't bother checking if SSL_P(F) is NULL when we already checked if
  F->ssl is NULL -- this should be impossible.

* Don't bother checking if SSL_C(F) is NULL -- this was a no-op.

* Change the socket send and recv functions to not peer into a foreign
  ratbox structure -- use the correct function to get the socket fd.

* Rewrap some lines and function arguments.

Other backends will be brought into line with this backend soon.

This will enable easier maintenance of the backends, by reducing the
diffs between them, which should make different behaviour easier to
spot.
2016-09-10 08:42:04 +00:00
Aaron Jones
2a8ec58c15
MbedTLS: Treat 0 bytes read/written to socket properly
At the moment, if a link quits in just the right (wrong [1]) way,
the quit reason will resemble:

    <-- foo (~bar@baz) has quit (Read error: (-0x0) )

This should resolve that.

[1] Peers should send a close_notify alert before abruptly shutting
    down their socket. This will result in a sane quit message:

    <-- foo (~bar@baz) has quit (Read error: (-0x7880) SSL -
    The peer notified us that the connection is going to be closed)

[ci skip]
2016-09-09 01:47:08 +00:00
William Pitcock
89d4c468b6 Merge pull request #214 from aaronmdjones/release/3.5
Fix up the MbedTLS backend
2016-09-03 10:34:43 -07:00
Aaron Jones
be31ac33d5 MbedTLS: Use correct error code for failed socket writes
This should make writing more efficient.
2016-09-02 00:28:17 +00:00
Aaron Jones
0db0805ed5 MbedTLS: Don't include the sentinel in suites count calculation 2016-09-01 20:57:07 +00:00
Aaron Jones
df51e80717 MbedTLS: Provide default list of configured ciphersuites 2016-09-01 20:47:34 +00:00
Aaron Jones
f92b4d81d4
OpenSSL: Initialise if LibreSSL
LibreSSL's definition of OPENSSL_VERSION_NUMBER bites us in the ass,
*again*.
2016-09-01 19:28:18 +00:00
Aaron Jones
6df12e8169 MbedTLS: Cleaner iteration of ciphersuite list 2016-09-01 18:18:09 +00:00
Aaron Jones
ede25e0a8a MbedTLS: Log success or failure to parse ciphersuite list 2016-08-31 22:03:42 +00:00
Aaron Jones
6f3651f8ec MbedTLS: Remove pointless no-op cast 2016-08-31 18:34:21 +00:00
Aaron Jones
b21ed5c0aa MbedTLS: Ciphersuite configuration fixes 2016-08-31 17:06:51 +00:00
Aaron Jones
42b029d0d6 MbedTLS: Preliminary attempt at ciphersuite configuration 2016-08-31 17:03:02 +00:00
Aaron Jones
865e70f529
Revert "Backport c1fc044c to release/3.5"
This reverts commit c9c2d6ea12.

This commit included some as yet untested and unrelated code by mistake.
2016-08-31 14:19:43 +00:00
Aaron Jones
c9c2d6ea12
Backport c1fc044c to release/3.5 2016-08-31 14:13:45 +00:00
Aaron Jones
531e6323d8 MbedTLS: Explicitly ignore rb_snprintf() return value 2016-08-31 01:01:42 +00:00
Aaron Jones
036419c344 MbedTLS: Misc security improvements
As a client, require all peers (i.e. other IRC servers) to support secure
renegotiation. Break handshakes with servers that don't. We do not
renegotiate our sessions, but this is the most secure option regardless.

As a client, disable TLS Session Tickets. The server side MbedTLS code
does not have any ticket callbacks configured, so an MbedTLS IRC Server
will not issue tickets -- however, others could. Server connections are
not expected to be short-lived enough to benefit from the usage of tickets,
and their issuance harms forward secrecy.
2016-08-31 00:13:56 +00:00
Aaron Jones
19d9c417af MbedTLS: Fix casing on opening comment block 2016-08-30 23:38:25 +00:00
Aaron Jones
f2fbec4510 MbedTLS: More const-correctness 2016-08-30 23:31:47 +00:00
Aaron Jones
f89406ac72 MbedTLS: Misc sizeof prettiness 2016-08-30 23:22:41 +00:00
Aaron Jones
c1007a93d5 MbedTLS: Move more code to appropriate section 2016-08-30 23:16:33 +00:00
Aaron Jones
3ba0923c0e MbedTLS: Move some MbedTLS-specific code to the appropriate section 2016-08-30 23:13:53 +00:00
Aaron Jones
4c9ab80f6b MbedTLS: Major restructuring
I have removed all non-MbedTLS-specific code from this backend and
copied the non-OpenSSL-specific portions of the OpenSSL backend code
over; as it seems to be more reliable.
2016-08-30 22:57:25 +00:00
Aaron Jones
07b6e728b5
OpenSSL: Initialise one context at a time
If initialising the server context fails, but the client one succeeds,
we will not only leak memory, but the error message reported for
initialising the server context might not make sense, because we
initialise the client context after and that could erase or change the
list of queued errors.

This scenario is considered rare. Nevertheless, we now initialise the
client context after *successfully* initialising the server context.
2016-08-30 10:21:46 +00:00
Aaron Jones
f70b6f55f9
TLS Backends: Harmomise the rb_ssl_get_cipher() function
The GNUTLS backend reports the version in use for the client as well
as its ciphersuite -- do the same for the other 2 backends.
2016-08-20 04:08:30 +01:00
Aaron Jones
3288fc4648
GNUTLS: Fixup fingerprint generation across library versions
Also remove some unnecessary variables, e.g. write directy to the
return buffer, in line with the other backends.
2016-08-19 19:05:22 +00:00
Aaron Jones
f15a30a16f
GNUTLS: I need to wake up. Fix more. 2016-08-17 17:37:03 +00:00
Aaron Jones
b24cfd7c50
GNUTLS: Fix typo on previous commit 2016-08-17 17:27:26 +00:00
Aaron Jones
06feeb244d
GNUTLS: Avoid null derefence in constructing ciphersuite 2016-08-17 16:58:40 +00:00
Aaron Jones
0bd2f0b710
openssl: Avoid use-after-free when rehashing fails to load new files
Commit 5c8da48 introduced a fix for issue #186 by freeing the old SSL_CTX
structure before constructing a new one, which could disconnect existing
clients otherwise.

Unfortunately, the freeing is done first, which means that if setting up
a new structure fails for any reason, there will be no usable structures
left, but they are still referenced.

This fix moves the freeing to the end of the function, using intermediate
new variables in the meantime. This problem was discovered while testing
against OpenSSL 1.1.0 RC6.
2016-08-12 13:29:02 +00:00
Aaron Jones
e719e46d27
mbedtls backend: indicate reason for TLS session termination
[ci skip]
2016-06-12 11:32:30 +00:00
Aaron Jones
03e6030ed2
openssl: More LibreSSL compatibility
LibreSSL does not have the new version macros & functions that OpenSSL
1.1.0 implements. This causes a compile-time failure against LibreSSL.

Further, the runtime function for returning the library version returns
the wrong number (the hardcoded constant number SSLEAY_VERSION_NUMBER
aka OPENSSL_VERSION_NUMBER, instead of LIBRESSL_VERSION_NUMBER).

Add more ifdef soup to remedy the situation.
2016-06-01 17:45:36 +00:00
Aaron Jones
82d827469c
openssl: change how we load DH parameters
The code already assumes the presence of fopen(3) and errno, and, by
extension, fclose(3) and strerror(3), so just use those instead of the
BIO wrappers.

Additionally, don't fail to initialise if the DH file does exist but
parsing it fails, as per the pre-existing comment about them being
optional.
2016-05-25 21:53:09 +00:00
William Pitcock
96129f7d4d charybdis 3.5.2. 2016-05-14 17:00:59 -05:00
Aaron Jones
57d3cd1159
[mbedtls] Fix up backend to allow fingerprint generation
See the comments in the newly created file for an explanation.
2016-05-04 09:09:01 +00:00
Aaron Jones
fd5af836b7
[mbedtls] Various fixes and improvements
* Move certificate, key, DH parameters and configuration to heap
  (Documentation states that setting new configuration, e.g.
   during a rehash, is unsupported while connections using that
   configuration are active)

  This is the same approach as the fix for #186

  Refcount these structures so as to not introduce a memory leak

  On rehash, it will use new structures only if there are no
  errors in constructing them

* Add better error-reporting (strings in addition to numbers)
  where possible

* Coalesce several connection memory allocations into one function

* Reduce boilerplate where possible (Charybdis targets C99)

* Support private key being in certificate file, and having no
  DH parameters file

* Correct erroneous closing comment
2016-05-04 02:12:23 +00:00
Aaron Jones
d35caf56cb
[TLS backends] Make version strings more useful and consistent 2016-05-04 00:20:07 +00:00
Aaron Jones
5c8da48264
Backport more TLS backend and ssld fixes & improvements from 3.6
openssl:
 * Don't manually initialise libssl 1.1.0 -- it does this automatically
 * SSL_library_init() should be called first otherwise
 * Move SSL_CTX construction to rb_setup_ssl_server()
 * Test for all required files (certificate & key) before doing anything
 * Free the old CTX before constructing a new one (Fixes #186)
 * Properly abort rb_setup_ssl_server() on CTX construction failures
 * Support ECDHE on more than one curve on OpenSSL 1.0.2 and above
 * Clean up ifdef indentation
 * Fix DH parameters memory leak

mbedtls:
 * Fix certificate fingerprint generation
 * Fix library linking order
 * Fix incorrect printf()-esque argument count
 * Return digest length for fingerprints instead of 1, consistent
   with the other backends

sslproc / ssld:
 * Fingerprint methods have no assocated file descriptors
 * Send TLS information (cipher, fingerprint) before data
 * Use correct header length for fingerprint method

Authored-by: Aaron Jones <aaronmdjones@gmail.com>
Authored-by: William Pitcock <nenolod@dereferenced.org>
Authored-by: Simon Arlott <sa.me.uk>
2016-04-30 21:39:05 +00:00
Valerii Iatsko
bf9e0a6ed5 Fixed compilation w/ gnutls v3 2016-04-02 17:28:37 -05:00
William Pitcock
e253d010ed libratbox: gnutls: add gnutls 3.4 support (closes #123) 2016-01-24 14:52:40 -05:00
William Pitcock
6dcf35b167 libratbox: don't build arc4random support if mbedtls is present. libratbox r29245 2015-12-27 21:21:33 -06:00
William Pitcock
5cc7ba2577 libratbox: fix scoping issue with alloca()'d buffer which could result in undefined behaviour.
this is ported from upstream libratbox r29267
2015-12-27 21:19:17 -06:00
William Pitcock
7233e364cc gnutls: fix typo 2015-12-12 08:19:58 -06:00
William Pitcock
673ec98e71 gnutls: allow priorities to be configured 2015-12-12 08:03:59 -06:00
William Pitcock
c1725bda3c ssl: allow cipher list to be overridden (closes #67) 2015-12-12 07:50:48 -06:00
William Pitcock
5225f83df1 libratbox: import zstring functions 2015-12-11 15:56:33 -06:00
William Pitcock
7c16cc9085 libratbox: implement nossl variants of rb_get_ssl_certfp() and rb_get_ssl_cipher() 2015-12-11 08:36:21 -06:00
William Pitcock
833b2f9cbf libratbox: implement rb_get_ssl_cipher() 2015-12-11 08:32:02 -06:00
Elizabeth Myers
e6bbb41030 Add ability to change CertFP hash.
Presently this only supports SHA1, as the machinery to actually change
the cipher is not hooked up to anything yet.
2015-12-07 01:14:02 -06:00
William Pitcock
677d3cb1a3 mbedtls: implement rb_get_ssl_certfp() 2015-12-05 06:37:04 -06:00
William Pitcock
493897d67c mbedtls: use server certificate for client mode too 2015-12-04 22:42:10 -06:00
William Pitcock
162a91d6ed mbedtls: make client mode work too 2015-12-04 22:41:02 -06:00
William Pitcock
539d912b95 libratbox: fix up mbedtls backend 2015-12-04 00:01:40 -06:00
William Pitcock
fab6f9e80e libratbox: gnutls: we want to pass F, not F->fd to rb_listen() 2015-12-03 22:38:54 -06:00
William Pitcock
cd492e44b5 libratbox: initial cut at an mbedtls implementation, which will replace openssl and gnutls backends in charybdis 3.6
notably, it presently is lacking CertFP support, but everything else is implemented
2015-12-03 22:32:33 -06:00
William Pitcock
6f57a957b0 libratbox: prepare build system for mbed tls backend 2015-12-03 18:37:32 -06:00
Aaron Jones
0b05d1aad5
[libratbox] Remove an unused random function
Nothing in the Charybdis or libratbox code calls rb_get_pseudo_random
and under OpenSSL it uses RAND_pseudo_bytes() which is both dangerous
and removed in OpenSSL version 1.1.0.
2015-10-25 16:04:57 +00:00
Aaron Jones
25f7ee7dd6
Improve the versions of TLS used for server to server linking
When building against current OpenSSL (<= 1.0.2) or old LibreSSL
(< 2.2.2) the server will use TLSv1.0 only when connecting to other
servers.

This patch corrects that.
2015-10-23 16:08:15 +00:00
Aaron Jones
3ae24413ca
Use new TLS method APIs with new LibreSSL
OpenBSD 5.8 includes LibreSSL 2.2.2, which finally brings the API up to
what they claim it is by implementing the new TLS client and server
method APIs. Therefore, in furtherance of commits a4c8c827 and 1a4e224a
we can build with the new APIs if building against (real) OpenSSL 1.1.0
or LibreSSL 2.2.2.

Reported-by: Juuso Lapinlampi <wub@partyvan.eu>
2015-10-23 16:05:33 +00:00
Jilles Tjoelker
df3de4e913 kqueue: Remove unnecessary cast. 2015-09-13 22:59:03 +02:00
William Pitcock
23e6a59007 Merge pull request #95 from jailbird777/master
Spring cleaning redux
2015-07-06 17:13:50 -05:00
Aaron Jones
1a4e224a4e LibreSSL have far advanced OPENSSL_VERSION_NUMBER beyond the
feature set they support (2.0 even!), deliberately breaking
backward compatibility. Therefore, in order to fix a regression
introduced by commit a4c8c827 with regard to LibreSSL's stupidity,
unconditionally use the old TLS API if building against LibreSSL.
2015-06-25 13:57:07 +00:00
Aaron Jones
cb266283f8 libratbox/openssl: Set explicit cipher list for the client context aswell
This is in furtherance of commits 9799bea4 and 1f384464 and addresses
any potential vulnerability to LogJam <https://weakdh.org/>
2015-05-20 16:41:34 +00:00
Aaron Jones
c86f11da1c Fix regression introduced by previous commit
I really shouldn't copy and paste code.
2015-05-20 10:39:04 +00:00
Aaron Jones
a4c8c82703 Tidy up OpenSSL options code, support new version-agnostic client and server APIs 2015-05-20 02:27:59 +00:00
Jail Bird
29c92cf95f Spring cleaning redux:
- Implemented changes suggested by Jilles
- Remove some unused parameters in functions
- Remove some unused ssl procs
- 63-bit time_t support in TS deltas
- const char * vs char * cleanup
- struct alignment (void *) casts
- signed vs unsigned fixes
- bad memset() call
- Bad LT_MAIN in libratbox
- char -> unsigned char casts for isdigit/isspace/etc calls

Thanks Jilles!
2015-04-20 00:55:20 -05:00
Aaron Jones
614502a63c Generate fingerprints for chained certificates with an unknown root 2015-03-24 05:25:38 +00:00
Aaron Jones
d3806d0503 Use X509_digest() instead of memcpy() to obtain cert fingerprint
This will continue to work even if the OpenSSL developers make the
X509* structure opaque, the current approach will not.
2015-03-24 05:22:25 +00:00
Aaron Jones
1f384464fa Update ciphersuite string to prohibit RC4
This is in accordance with RFC 7465
<https://tools.ietf.org/html/rfc7465>

Also correct the key exchange mechanism strings; these should be
prefixed with 'k'.
2015-03-22 06:14:39 +00:00
William Pitcock
e861902fd9 libratbox: remove RB_PRNG_EGD in its entirety (closes #85) 2015-03-01 01:05:14 -06:00