Vincent Hanquez
d787160713
rename connect in client module to initiate.
...
add a deprecated pragma for connect and keep it for compatibility
2010-11-28 10:30:05 +00:00
Vincent Hanquez
d061b37730
remove RandomGen generator and provide a reseed functions to regenerate the RNG.
2010-11-07 10:19:12 +00:00
Vincent Hanquez
a95dd8f45e
Generalize key exchange and use in-house RSA.
...
Remove need for spoon, since RSA will fails gracefully.
Add support for full private key format for fast decryption.
Generalization of key exchange to add future support for DH, etc.
2010-11-04 19:10:00 +00:00
Vincent Hanquez
65942b945f
massive change on the RNG and add support for CryptoRandomGen
...
use an inline AES counter system to generate random data.
2010-11-04 19:05:36 +00:00
Vincent Hanquez
863c09e0d4
put the throwError inside the runGet instead of outside.
...
related to a monad either overlapping report.
2010-11-03 23:06:04 +00:00
Vincent Hanquez
0d342a68f7
use monads-fd instead of mtl
2010-10-24 12:02:45 +01:00
Vincent Hanquez
7a5139c5ce
depends on cryptohash 0.6 and adapt to the new modules location.
2010-10-24 12:02:41 +01:00
Vincent Hanquez
a1469322e4
don't check padding content in SSL3 mode
2010-10-06 09:15:36 +01:00
Vincent Hanquez
f54be74cc4
properly encode RSA structure without extra length when doing SSL3
2010-10-06 09:11:57 +01:00
Vincent Hanquez
cfff801bd9
properly finish SSL3 digest computation.
...
change the cipher structure to contain the hash algorithm
instead of the mac algorithm.
2010-10-06 09:07:48 +01:00
Vincent Hanquez
1bbd893e95
use modify instead of get >>= put (and same for modifyTLS)
2010-10-05 18:48:32 +01:00
Vincent Hanquez
3c2ebe5c08
more generation of SSL block/finished values.
2010-10-05 18:48:28 +01:00
Vincent Hanquez
9c4a3a0223
use the AES rng as the TLSState rng, and generate random bytes on demand
...
client/premaster/secret random bytes are now generated by the TLSstate rng
on demand, simplifying the use of basic routines (connect/listen) and the
renegociation process.
also the AES rng is a CPRNG, compared to system.random PRNG, which
might gives a better warmer fuzzy random feeling.
2010-10-03 11:23:12 +01:00
Vincent Hanquez
6a9296727b
improve the regeneration of client and server rng datas
2010-10-03 11:01:22 +01:00
Vincent Hanquez
9641ca7c83
add prototype to processServerInfo
2010-10-03 10:32:23 +01:00
Vincent Hanquez
383cf4c021
properly handle multiple packet fragments.
...
as a bonus it cleans lots of differents part since the state machine
is inside receiving/sending code
2010-10-02 22:41:00 +01:00
Vincent Hanquez
e189f37a67
new state machine
2010-10-02 22:02:37 +01:00
Vincent Hanquez
5a6ff3abe8
take in account that we can receive multiple handshakes in the same tls fragment.
2010-10-02 10:58:41 +01:00
Vincent Hanquez
8049ad6c6f
add a way to update Digest when we have a handshaket type and the content of the header
2010-10-02 10:54:49 +01:00
Vincent Hanquez
eb3ed06af1
add TLS state machine to track that we receive correct message at the correct type
2010-10-02 10:32:29 +01:00
Vincent Hanquez
d188a180cc
refactorize receiving packet thing
2010-10-02 09:09:46 +01:00
Vincent Hanquez
f033a0d973
reorganize the way we decrypt data to be nicer.
...
as a bonus, finally check if padding is valid.
2010-09-26 20:56:51 +01:00
Vincent Hanquez
6d1e38a337
add partition3 utility
2010-09-26 20:56:47 +01:00
Vincent Hanquez
07bfcd2094
typo
2010-09-26 20:54:14 +01:00
Vincent Hanquez
cd2f8f8ee2
get a util file for some bytestring stuff
2010-09-26 18:51:23 +01:00
Vincent Hanquez
349406b64a
remove data.word
2010-09-26 16:33:39 +01:00
Vincent Hanquez
2f76b2a245
add non finished method to generate finished content for protocol < TLS10
2010-09-26 16:32:28 +01:00
Vincent Hanquez
60de6551e4
add missing signature
2010-09-26 16:07:29 +01:00
Vincent Hanquez
c664f30407
add support for SSL generation of master secret
2010-09-26 16:07:14 +01:00
Vincent Hanquez
fc4c76862c
use specified version when initiating connection.
2010-09-26 15:31:46 +01:00
Vincent Hanquez
938e8db365
remove dead field
2010-09-26 15:31:35 +01:00
Vincent Hanquez
f37b2e3bce
support TLS1.1 explicit block IV
...
despite the fact that it works, it's missing a step at key block set time,
so that we don't use the computed IV, but use a random generated one seeded by
the computed IV.
2010-09-26 14:57:35 +01:00
Vincent Hanquez
b71ea6729c
remove few more unpacking/packing
2010-09-26 10:37:20 +01:00
Vincent Hanquez
8f91009884
use strict bytestring instead of lazy bytestring.
...
the API stays mostly similar except for clientkeyxchg that need a bytes instead of [word8].
remove lots of unnessary packing/unpacking when setting up ciphers.
2010-09-26 10:34:47 +01:00
Vincent Hanquez
c70736cf19
add a cap file to differenciate protocol version capabilities.
...
define 2 capabilities for hello extensions and explicit IV.
use hello extensions checking in decode / encode of clientHello
2010-09-26 08:46:09 +01:00
Vincent Hanquez
2fd8087211
remove the haskell98 dependency and switch to the random package.
2010-09-24 08:30:25 +01:00
Vincent Hanquez
663436a3cb
expose some needed modules and expose TLSClient type
...
(noticed by Michael Snoyman)
2010-09-22 18:31:42 +01:00
Vincent Hanquez
cb850131da
add a server callbacks when receiving Certificates
2010-09-20 08:45:41 +01:00
Vincent Hanquez
2fe1d7e99a
use <$> instead of fmap
2010-09-19 10:49:42 +01:00
Vincent Hanquez
8c20758158
use client callback to callback on certificate verification
2010-09-19 10:42:29 +01:00
Vincent Hanquez
31fac5df44
introduce Client Callbacks structure
2010-09-19 10:42:02 +01:00
Vincent Hanquez
abf299db36
don't try to send an empty string when there's no rsa key to encrypt with.
2010-09-18 11:00:30 +01:00
Vincent Hanquez
b289546721
gives all certificate to processCertificate, so that we can check the chain later.
2010-09-18 11:00:07 +01:00
Vincent Hanquez
f4f4968a82
change clientkeyxchg data to be a specific type
2010-09-13 21:10:25 +01:00
Vincent Hanquez
0b5a0dc548
initial import
2010-09-09 22:47:19 +01:00
