Parent Directory | Revision Log
|Links to HEAD:||(view) (download) (annotate)|
Fix buffer overflow in key-ids and explicitly only compare the correct number of chars
Remove vestiges of v3 key management.
Updated copyright dates, doc issues, misc fixes.
Set keyword expansion for Id on all files that already have $Id$ tags.
Do away with the conditional use of RSA, always using it. Without RSA we aren't much good anyway.
Add a few #ifdef USE_PGP where they belong. -- Hauke
Added missing newline characters.
Made the protocol version a string constant.
We prepend the protocol version string to the software version number in the type 2 capstring. This is necessary to allow existing Mixmaster versions to interoperate with future versions of Mixmaster. Once we have confirmation that the use of the colon in the version string doesn't break any clients or other servers, I'll commit this to -stable and 2.0.4b as well, and modify the spec accordingly.
Whitespace cleanup: for i in *.c; do cp -a $i $i.orig; sed -e 's/^ /<TAB>/' $i.orig > $i; rm $i.orig; done for i in *.c *.h; do cp -a $i $i.orig; sed -e 's/[ <TAB>]*$//' $i.orig > $i; rm $i.orig; done
Only return latest pgp keys to remailer-key requests
Do not recreate the public keys from the private keyring if there still is a public keyring in pgpkey.txt. This means we don't recreate the self signature on binding keys amoung other things. It probably also means that one can replace the keys in pgpkey.txt with keys signed by the adminkey and not have them overriden the next time one does ./mix -[GK] (although I did not test this). We still need to fix keymgt to only return the latest keys - this should be simple once I find out how to get expiration date for a specific key. There's a small problem with the current keymanagement for PGP keys I guess. If the latest key is not the last in the keyring we might screw up. Can you verify this Disastry?
We now handle CRLF end of lines in pubring.mix
Added closing comments for all #ifdef statements. All #endif's, as well as nested braces, should be commented to reference their start. We need to provide comments before every function as well.
Use the key later in the file if they have the same expiry date (weasel aka uid23509)
Mixmaster keys now have creation and expiration date. It is not secured by any crypto voodoo, it's only informational for clients to decide which keys to use should they have more. - on the client side we do not show remailers (and therefore not use them) if their key is expired. - the remailer refuses to decrypt messages to keys that expired one month ago or earlier. - the remailer automatically creates new mixmaster keys if the current one are about to expire or already are expired. - the latest key from secring.mix is written to key.txt. It used to be the first one. Since creation of new mix key appends the key, this seemed sensible.
Fix segfault with no keys
Mixmaster now prompts for user input for the secret key passphrase when mix is started in daemon mode (./mix -D). Mixmaster will try to find either PGP/RSA, PGP/ELG or Mix key that can be decrypted with the provided passphrase. If it finds one, it assumes the passphrase is good. (We may have a problem if the user manages to get his secret key passphrases out of sync. Suggestions on handling this case?) FYI: the passphrase in mix.cfg takes precedence over the passphrase compiled into the binary. If neither of these are present, only then is the user prompted for a keyboard-input passphrase. We still need to make mix retain the user-provided passphrase across SIGHUP signals.
Make sure files are closed
Submitted By: Sami Farin (safari): Included correct header files. Src/main.c had incorrect param to buf_appendf. Added buf_write_sync() (used by Src/rem.c:mix_pool()). Should a mixmaster machine crash, pool file can contain any random data after reboot/journal recovery (at least when using reiserfs). Maybe not relevant, but strrchr should be used instead of strchr when looking for domain part. Src/pgpdata.c:pgp_elgdecrypt() passes uninitialized values to BN_free if BN_CTX_new fails. Src/pgpdata.c:pgp_elgencrypt() passes uninitialized value to BN_free if the third mpi_get(key, i) fails. Src/random.c does not check return value of read(). Src/rndseed.c does not check return value of read().
The -G option now forces creation of new keys even in client mode (Closes: #585176)
We have changed the compile-time option PASSPHRASE to be named COMPILEDPASS. We have changed the configuration file option PASS_PHRASE to be named PASSPHRASE. We have added documentation for the new configuration file option and made changes in the man page to reflect the name change. This was done to avoid confusion due to the similarity in names of the options.
Added the ability to store the key passphrase in the mix.cfg file instead of having it compiled into the binary. This offers a number of advantages over the compiled-in method and still allows for encrypted remote storage of keys. In order to take advantage of the new changes, the Install script will need to be modified. As this is being replaced with a proper autoconf system, we're defering that until later. The changes we're committing replace PASSPHRASE with PASS_PHRASE, which is read from the config. If the compiled-in passphrase, PASSPHRASE, is set, that is used instead. (This is how Mixmaster will function if you use the Install script.)
Mixmaster now does not attempt to generate keys when run in client-only mode (REMAIL set to "n") as there is no reason to do so. Requested by Richard Christman for Quicksilver.
Put "M" in key file (key.txt) if remailer is middleman. (disastry) Remember the flags are: C accepts compressed messages. M will forward messages to another mix, when used as the final hop. Nm supports posting to Usenet throught a mail-to-news gateway. Np supports direct posting to Usenet.
This form allows you to request diffs between any two revisions of this file. For each of the two "sides" of the diff, enter a numeric revision.
|Powered by ViewVC 1.1.5|