* g10/parse-packet.c (parse_key): Check PKTLEN before calling mpi_read
et al.
--
Due to the missing length checks PKTLEN may turn negative. Because
PKTLEN is an unsigned int the malloc in read_rest would try to malloc
a too large number and terminate the process with "error reading rest
of packet: Cannot allocate memory".
Reported-by: Hanno Böck.
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/parse-packet.c (set_packet_list_mode): Set mpi_print_mode.
* g10/misc.c (mpi_print): Do not print an extra leading zero.
--
This was in older versions possible using "--debug 4" but that was
disabled in 2.1 due to a conflict using this values also for
Libgcrypt. Now the values are dumped either with --debug 4 or using
--list-packets along with --verbose.
Because OpenPGP only uses unsigned integers an extra leading zero will
not be printed anymore.
Signed-off-by: Werner Koch <wk@gnupg.org>
* common/host2net.h (buf16_to_ulong, buf16_to_uint): New.
(buf16_to_ushort, buf16_to_u16): New.
(buf32_to_size_t, buf32_to_ulong, buf32_to_uint, buf32_to_u32): New.
--
Commit 91b826a388 was not enough to
avoid all sign extension on shift problems. Hanno Böck found a case
with an invalid read due to this problem. To fix that once and for
all almost all uses of "<< 24" and "<< 8" are changed by this patch to
use an inline function from host2net.h.
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/parse-packet.c (MAX_KEY_PACKET_LENGTH): New.
(MAX_UID_PACKET_LENGTH): New.
(MAX_COMMENT_PACKET_LENGTH): New.
(MAX_ATTR_PACKET_LENGTH): New.
(parse_key): Limit the size of a key packet to 256k.
(parse_user_id): Use macro for the packet size limit.
(parse_attribute): Ditto.
(parse_comment): Ditto.
--
Without that it is possible to force gpg to allocate large amounts of
memory by using a bad encoded MPI. This would be an too easy DoS.
Another way to mitigate would be to change the MPI read function to
allocate memory dynamically while reading the MPI. However, that
complicates and possibly slows down the code. A too large key packet
is in any case a sign for broken data and thus gpg should not use it.
Reported-by: Hanno Böck
GnuPG-bug-id: 1823
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/gpg.h (g10_errstr): Remove macro and change all occurrences by
gpg_strerror.
(G10ERR_): Remove all macros and change all occurrences by their
GPG_ERR_ counterparts.
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/parse-packet.c (can_handle_critical): Check content length
before calling can_handle_critical_notation.
--
The problem was found by Jan Bee and gniibe proposed the used fix.
Thanks.
This bug can't be exploited: Only if the announced length of the
notation is 21 or 32 a memcmp against fixed strings using that length
would be done. The compared data is followed by the actual signature
and thus it is highly likely that not even read of unallocated memory
will happen. Nevertheless such a bug needs to be fixed.
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/parse-packet.c (dump_sig_subpkt): Print regex subpacket
sanitized.
--
We may not use "%s" to print an arbitrary buffer. At least "%.*s"
should have been used. However, it is in general preferable to escape
control characters while printf user data.
Reported-by: Hanno Böck
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/parse-packet.c (parse_attribute_subpkts): Check that the
attribute packet is large enough for the subpacket type.
--
Reported-by: Hanno Böck
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/parse-packet.c (parse_key): Store even unsupported packet
versions.
* g10/keyring.c (keyring_rebuild_cache): Do not copy keys with
versions less than 4.
--
That function, which is implicitly called while checking the keydb, led
to corruption of v3 key packets in the keyring which would later spit
out "packet(6)too short" messages.
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/build-packet.c (do_key): Remove support for building v3 keys.
* g10/parse-packet.c (read_protected_v3_mpi): Remove.
(parse_key): Remove support for v3-keys. Add dedicated warnings for
v3-key packets.
* g10/keyid.c (hash_public_key): Remove v3-key support.
(keyid_from_pk): Ditto.
(fingerprint_from_pk): Ditto.
* g10/options.h (opt): Remove fields force_v3_sigs and force_v4_certs.
* g10/gpg.c (cmd_and_opt_values): Remove oForceV3Sigs, oNoForceV3Sigs,
oForceV4Certs, oNoForceV4Certs.
(opts): Turn --force-v3-sigs, --no-force-v3-sigs, --force-v4-certs,
--no-force-v4-certs int dummy options.
(main): Remove setting of the force_v3_sigs force_v4_certs flags.
* g10/revoke.c (gen_revoke, create_revocation): Always create v4 certs.
* g10/sign.c (hash_uid): Remove support for v3-signatures
(hash_sigversion_to_magic): Ditto.
(only_old_style): Remove this v3-key function.
(write_signature_packets): Remove support for creating v3-signatures.
(sign_file): Ditto.
(sign_symencrypt_file): Ditto.
(clearsign_file): Ditto. Remove code to emit no Hash armor line if
only v3-keys are used.
(make_keysig_packet): Remove arg SIGVERSION and force using
v4-signatures. Change all callers to not pass a value for this arg.
Remove all v3-key related code.
(update_keysig_packet): Remove v3-signature support.
* g10/keyedit.c (sign_uids): Always create v4-signatures.
* g10/textfilter.c (copy_clearsig_text): Remove arg pgp2mode and
change caller.
--
v3 keys are deprecated for about 15 years and due the severe
weaknesses of MD5 it does not make any sense to keep code around to
use these old and broken keys. Users who need to decrypt old messages
should use gpg 1.4 and best re-encrypt them to modern standards.
verification of old (i.e. PGP2) created signatures is thus also not
anymore possible but such signatures have no values anyway - MD5 is
just too broken.
We have also kept support for v3 signatures until now. With the
removal of support for v3 keys it is questionable whether it makes any
sense to keep support for v3-signatures. What we do now is to keep
support for verification of v3-signatures but we force the use of
v4-signatures. The latter makes the --pgp6 and --pgp7 switch a bit
obsolete because those PGP versions require v3-signatures for
messages. These versions of PGP are also really old and not anymore
maintained so they have not received any bug fixes and should not be
used anyway.
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/parse-packet.c (parse_attribute): Avoid xmalloc failure and cap
size of packet.
--
Tavis Ormandy reported a fatal error for attribute packets with a zero
length payload. This is due to a check in Libgcrypt's xmalloc which
rejects a malloc(0) instead of silently allocating 1 byte. The fix is
obvious.
In addition we cap the size of attribute packets similar to what we do
with user id packets. OpenPGP keys are not the proper way to store
movies.
* g10/parse-packet.c (mpi_read): Change error message on overflow.
--
This gets gpg 2.x in sync to what gpg 1.4 does. No need to die for a
broken MPI.
GnuPG-bug-id: 1593
Resolved conflicts:
g10/parse-packet.c - whitespaces fixes.
* agent/cvt-openpgp.c (try_do_unprotect_arg_s): Add field "curve".
(get_keygrip): Add and use arg CURVE.
(convert_secret_key): Ditto.
(convert_transfer_key): Ditto.
(get_npkey_nskey): New.
(prepare_unprotect): Replace gcrypt functions by
get_npkey_nskey. Allow opaque MPIs.
(do_unprotect): Use CURVE instead of parameters.
(convert_from_openpgp_main): Ditto.
(convert_to_openpgp): Simplify.
* g10/import.c (one_mpi_from_pkey): Remove.
(transfer_secret_keys): Rewrite to use the curve instead of the
parameters.
* g10/parse-packet.c (parse_key): Mark protected MPIs with USER1 flag.
* common/openpgp-oid.c (openpgp_curve_to_oid): Allow the use of
"NIST P-256" et al.
* g10/keygen.c (ask_curve): Add arg ALGO.
(generate_keypair): Rewrite the ECC key logic.
* tests/openpgp/ecc.test: Provide the "ecc" passphrase.
* common/sexputil.c (get_pk_algo_from_canon_sexp): Change to return a
string.
* g10/keygen.c (check_keygrip): Adjust for change.
* sm/certreqgen-ui.c (check_keygrip): Likewise.
* agent/pksign.c (do_encode_dsa): Remove bogus map_pk_openpgp_to_gcry.
* g10/misc.c (map_pk_openpgp_to_gcry): Remove.
(openpgp_pk_test_algo): Change to a wrapper for openpgp_pk_test_algo2.
(openpgp_pk_test_algo2): Rewrite.
(openpgp_pk_algo_usage, pubkey_nbits): Add support for EdDSA.
(openpgp_pk_algo_name): Rewrite to remove need for gcry calls.
(pubkey_get_npkey, pubkey_get_nskey): Ditto.
(pubkey_get_nsig, pubkey_get_nenc): Ditto.
* g10/keygen.c(do_create_from_keygrip): Support EdDSA.
(common_gen, gen_ecc, ask_keysize, generate_keypair): Ditto.
* g10/build-packet.c (do_key): Ditto.
* g10/export.c (transfer_format_to_openpgp): Ditto.
* g10/getkey.c (cache_public_key): Ditto.
* g10/import.c (transfer_secret_keys): Ditto.
* g10/keylist.c (list_keyblock_print, list_keyblock_colon): Ditto.
* g10/mainproc.c (proc_pubkey_enc): Ditto.
* g10/parse-packet.c (parse_key): Ditto,
* g10/sign.c (hash_for, sign_file, make_keysig_packet): Ditto.
* g10/keyserver.c (print_keyrec): Use openpgp_pk_algo_name.
* g10/pkglue.c (pk_verify, pk_encrypt, pk_check_secret_key): Use only
OpenPGP algo ids and support EdDSA.
* g10/pubkey-enc.c (get_it): Use only OpenPGP algo ids.
* g10/seskey.c (encode_md_value): Ditto.
--
This patch separates Libgcrypt and OpenPGP public key algorithms ids
and in most cases completely removes the Libgcrypt ones. This is
useful because for Libgcrypt we specify the algorithm in the
S-expressions and the public key ids are not anymore needed.
This patch also adds some support for PUBKEY_ALGO_EDDSA which will
eventually be used instead of merging EdDSA with ECDSA. As of now an
experimental algorithm id is used but the plan is to write an I-D so
that we can get a new id from the IETF. Note that EdDSA (Ed25519)
does not yet work and that more changes are required.
The ECC support is still broken right now. Needs to be fixed.
Signed-off-by: Werner Koch <wk@gnupg.org>
* agent/findkey.c (key_parms_from_sexp): Add algo name "ecc".
(agent_is_dsa_key): Ditto.
(agent_is_eddsa_key): New. Not finished, though.
* agent/pksign.c (do_encode_eddsa): New.
(agent_pksign_do): Use gcry_log_debug functions.
* agent/protect.c (agent_protect): Parse a flags parameter.
* g10/keygen.c (gpg_curve_to_oid): Move to ...
* common/openpgp-oid.c (openpgp_curve_to_oid): here and rename.
(oid_ed25519): New.
(openpgp_oid_is_ed25519): New.
(openpgp_oid_to_curve): New.
* common/t-openpgp-oid.c (test_openpgp_oid_is_ed25519): New.
* g10/build-packet.c (gpg_mpi_write): Write the length header also for
opaque MPIs.
(gpg_mpi_write_nohdr): New.
(do_key): Use gpg_mpi_write_nohdr depending on algorithm.
(do_pubkey_enc): Ditto.
* g10/ecdh.c (pk_ecdh_encrypt_with_shared_point): Use
gpg_mpi_write_nohdr.
* g10/export.c (transfer_format_to_openpgp):
* g10/keygen.c (ecckey_from_sexp): Return the error.
(gen_ecc): Repalce arg NBITS by CURVE.
(read_parameter_file): Add keywords "Key-Curve" and "Subkey-Curve".
(ask_curve): New.
(generate_keypair, generate_subkeypair): Use ask_curve.
(do_generate_keypair): Also pass curve name.
* g10/keylist.c (list_keyblock_print, list_keyblock_colon): Print
curve name.
* g10/parse-packet.c (mpi_read): Remove workaround for
Libcgrypt < 1.5.
(parse_key): Fix ECC case. Print the curve name.
* g10/pkglue.c (mpi_from_sexp): Rename to get_mpi_from_sexp.
(pk_verify, pk_check_secret_key): Add special case for Ed25519.
* g10/seskey.c (encode_md_value): Ditto.
* g10/sign.c (do_sign, hash_for, sign_file): Ditto.
--
Be warned that this code is subject to further changes and that the
format will very likely change before a release. There are also known
bugs and missing code.
Signed-off-by: Werner Koch <wk@gnupg.org>
This helps in the case of an unknown key algorithm with a corrupted
packet which claims a longer packet length. This used to allocate the
announced packet length and then tried to fill it up without detecting
an EOF, thus taking quite some time. IT is easy to fix, thus we do
it. However, there are many other ways to force gpg to use large
amount of resources; thus as before it is strongly suggested that the
sysadm uses ulimit do assign suitable resource limits to the gpg
process. Suggested by Timo Schulz.
We better do this once and for all instead of cluttering all future
commits with diffs of trailing white spaces. In the majority of cases
blank or single lines are affected and thus this change won't disturb
a git blame too much. For future commits the pre-commit scripts
checks that this won't happen again.
Changed order of some conditional to make to put the special case into
the true branch. Indentation changes. Minor other changes to make the
ECC code more similar to the rest of our code.
It builds but many sefltests still fail. Need to fix that before
using it with an ECDH enabled libgcrypt.
[/]
2011-01-21 Werner Koch <wk@g10code.com>
* configure.ac: Need Libgcrypt 1.4.6 due to AESWRAP.
(HAVE_GCRY_PK_ECDH): Add new test.
[agent/]
2011-01-21 Werner Koch <wk@g10code.com>
* cvt-openpgp.c (GCRY_PK_ECDH) [!HAVE_GCRY_PK_ECDH]: New.
[include/]
2011-01-21 Werner Koch <wk@g10code.com>
* cipher.h (GCRY_PK_USAGE_CERT): Remove compatibility macros
because we now require libgcrypt 1.4.6.
(GCRY_PK_ECDH): Add replacement.
The following works:
gpg2 --gen-key (ECC)
gpg2 --list-keys
gpg2 --list-packets ~/.gnupg/pubring.gpg
gpg2 --list-packets <private key from http://sites.google.com/site/brainhub/pgpecckeys>
ECDH doesn't work yet as the code must be re-written to adjust for gpg-agent refactoring.
