[PATCH] enable show-session-key on a truncated encrypted file

[Hank Leininger]

[ Sent to gnupg-devel a couple of days ago but it never went through;
  perhaps -devel is subscriber-only.  Apologies if you eventually see it
  twice. ]

Here is a patch (quick and dirty) to show a session key for an encrypted
file using --show-session-key even if the encrypted file is truncated.

Consider the following scenario:

- There's a big file encrypted to your key on a machine you don't trust
  enough to put your private key on / feed it your passphrase

- You need to have the decrypted version of that file on that machine
  (you do trust it enough to have that)

- You have a slow link to that machine; pulling down, decrypting, and
  pushing the plaintext version back would be painful

Maybe someone knows a better way to do this, but what I did some years
ago with gpg was basically:

  local$ ssh remote head -c1000000 bigfile.pgp > bigfile_fragment.pgp
  local$ gpg --show-session-key -o /dev/null --max-output 1 \
                        bigfile_fragment.pgp 2>&1 | egrep 'session key'
  remote$ gpg -d --override-session-key KEYSTRING bigfile.pgp

This fails with current gnupg without the attached patch.

The key here is the ability to do --override-session-key on a fragment
of a .pgp'ed file.  The current behavior of gnupg is to error out
because of the broken file prior to checking if opt.show_session_key is
set.  This is not "wrong"--but it is not helpful in the above scenario.

The below patch moves up the opt.show_session_key check and prints
the session key if known, even if gnupg is erroring out.  Is there any
reason this is a terrible idea *in the case that* you have already
decided to use --show-session-key / --override-session-key?

Thanks,

Hank Leininger <hl...@korelogic.com>
BE5D FCCA 673B D18B 98A9  3175 896E 3D4A 1B4D C5AC

####

diff -urP gnupg-2.0.17/g10/mainproc.c gnupg-2.0.17-showtrunc/g10/mainproc.c
--- gnupg-2.0.17/g10/mainproc.c 2011-01-09 17:06:16.000000000 -0500
+++ gnupg-2.0.17-showtrunc/g10/mainproc.c       2011-07-17 18:29:30.000000000 
-0400
@@ -561,6 +561,18 @@
     if( !result )
        result = decrypt_data( c, pkt->pkt.encrypted, c->dek );
 
+    /* If told to show the session key, try even on failed operations */
+    if(opt.show_session_key && c->dek != NULL && c->dek->keylen > 0)
+    {
+       int i;
+       char *buf = xmalloc ( c->dek->keylen*2 + 20 );
+       sprintf ( buf, "%d:", c->dek->algo );
+       for(i=0; i < c->dek->keylen; i++ )
+           sprintf(buf+strlen(buf), "%02X", c->dek->key[i] );
+       log_info( "session key: `%s'\n", buf );
+       write_status_text ( STATUS_SESSION_KEY, buf );
+    }
+
     if( result == -1 )
        ;
     else if( !result || (gpg_err_code (result) == GPG_ERR_BAD_SIGNATURE
@@ -572,16 +584,6 @@
            write_status( STATUS_GOODMDC );
        else if(!opt.no_mdc_warn)
            log_info (_("WARNING: message was not integrity protected\n"));
-       if(opt.show_session_key)
-         {
-           int i;
-           char *buf = xmalloc ( c->dek->keylen*2 + 20 );
-           sprintf ( buf, "%d:", c->dek->algo );
-           for(i=0; i < c->dek->keylen; i++ )
-             sprintf(buf+strlen(buf), "%02X", c->dek->key[i] );
-           log_info( "session key: `%s'\n", buf );
-           write_status_text ( STATUS_SESSION_KEY, buf );
-         }
     }
     else if( result == G10ERR_BAD_SIGN ) {
        log_error(_("WARNING: encrypted message has been manipulated!\n"));

signature.asc
Description: Digital signature

_______________________________________________
Gnupg-users mailing list
Gnupg-users@gnupg.org
http://lists.gnupg.org/mailman/listinfo/gnupg-users