* g10/options.h (IMPORT_NO_SECKEY): New.
* g10/keyserver.c (keyserver_spawn, keyserver_import_cert): Set new
flag.
* g10/import.c (import_secret_one): Deny import if flag is set.
--
By modifying a keyserver or a DNS record to send a secret key, an
attacker could trick a user into signing using a different key and
user id. The trust model should protect against such rogue keys but
we better make sure that secret keys are never received from remote
sources.
Suggested-by: Stefan Tomanek
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit e7abed3448)
Resolved conflicts:
g10/import.c
g10/keyserver.c
* g10/gpg.c (oLegacyListMode, opts, main): Add --legacy-list-mode.
* g10/options.h (struct opt): Add field legacy_list_mode.
* g10/keydb.h (PUBKEY_STRING_SIZE): New.
* g10/keyid.c (pubkey_string): New.
* g10/import.c (import_one, import_secret_one): Use pubkey_string.
* g10/keylist.c (print_seckey_info): Ditto.
(print_pubkey_info, print_card_key_info): Ditto.
(list_keyblock_print): Ditto.
* g10/mainproc.c (list_node): Ditto.
* g10/pkclist.c (do_edit_ownertrust, build_pk_list): Ditto.
* g10/keyedit.c (show_key_with_all_names): Ditto. Also change the
format.
(show_basic_key_info): Ditto.
* common/openpgp-oid.c (openpgp_curve_to_oid): Also allow "ed25519".
(openpgp_oid_to_curve): Downcase "ed25519"
--
For ECC it seems to be better to show the name of the curve and not
just the size of the prime field. The curve name does not anymore fit
into the "<size><letter>" descriptor (e.g. "2048R") and a fixed length
format does not work either. Thus the new format uses
"rsa2048" - RSA with 2048 bit
"elg1024" - Elgamal with 1024 bit
"ed25519" - ECC using the curve Ed25519.
"E_1.2.3.4" - ECC using the unsupported curve with OID "1.2.3.4".
unless --legacy-list-mode is given. In does not anymore line up
nicely in columns thus I expect further changes to this new format.
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/options.h (opt): Rename field no_version to emit_version.
* g10/gpg.c (main): Init opt.emit_vesion to 1. Change --emit-version
to bump up opt.emit_version.
* g10/armor.c (armor_filter): Implement different --emit-version
values.
--
GnuPG-bug-id: 1572
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/gpg.c: Include shareddefs.h.
(main): Add option --pinentry-mode.
* g10/options.h (struct opt): Add field pinentry_mode.
* g10/passphrase.c: Include shareddefs.h.
(have_static_passphrase): Take care of loopback pinentry_mode.
(read_passphrase_from_fd): Ditto.
(get_static_passphrase): New.
(passphrase_to_dek_ext): Factor some code out to ...
(emit_status_need_passphrase): new.
* g10/call-agent.c (start_agent): Send the pinentry mode.
(default_inq_cb): Take care of the PASSPHRASE inquiry. Return a
proper error code.
(agent_pksign): Add args keyid, mainkeyid and pubkey_algo.
(agent_pkdecrypt): Ditto.
* g10/pubkey-enc.c (get_it): Pass new args.
* g10/sign.c (do_sign): Pass new args.
* g10/call-agent.c (struct default_inq_parm_s): New. Change all
similar structs to reference this one. Change all users and inquire
callback to use this struct, instead of NULL or some undefined but not
used structs. This change will help to eventually get rid of global
variables.
--
This new features allows to use gpg without a Pinentry. As a
prerequisite the agent must be configured to allow the loopback
pinentry mode (option --allow-loopback-pinentry). For example
gpg2 --pinentry-mode=loopback FILE.gpg
may be used to decrypt FILE.gpg while entering the passphrase on the
tty. If batch is used, --passphrase et al. may be used, if
--command-fd is used, the passphrase may be provided by another
process. Note that there are no try-again prompts in case of a bad
passphrase.
* common/logging.c (log_clock): New.
* g10/gpg.c (set_debug): Print clock debug flag.
* g10/options.h (DBG_CLOCK_VALUE, DBG_CLOCK): New.
--
To actually use log_clock you need to enable the code in
logginc.c:log_check() and link against librt. --debug 4096 may then
be used to enable it at runtime.
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.
* seskey.c (encode_session_key): Debug output of the session key.
* pubkey-enc.c (get_it): Handle card case.
* call-agent.c (agent_scd_pkdecrypt): New.
* pkglue.c (pk_encrypt): Add RSA support.
* g10.c (main): Default to --use-agent.
* keygen.c (show_smartcard): Print info about the public key.
(check_smartcard): Check for existing key here.
(gen_card_key): And not anymore here.
(fpr_is_zero): New.
(generate_keypair): Generate both keys for a card.
(smartcard_change_url): Nw.
* misc.c (checksum_u16_nobug): Removed.
(checksum_u16): Removed the bug emulation.
(checksum_mpi): Ditto.
(checksum_mpi_counted_nbits): Removed and replaced all calls
with checksum_mpi.
* parse-packet.c (read_protected_v3_mpi): New.
(parse_key): Use it here to store it as an opaque MPI.
* seckey-cert.c (do_check): Changed the v3 unprotection to the new
why to store these keys.
(protect_secret_key): Likewise.
* build-packet.c (do_secret_key): And changed the writing.
decryption failed error if a MDC does not verify. Warn if a MDC is not
present (can disable via --no-mdc-warning).
* exec.c (exec_write), g10.c (main), keyserver.c (keyserver_spawn): Use
new DISABLE_KEYSERVER_PATH rather than FIXED_EXEC_PATH.
keyserver.c (keyserver_spawn): If the user does not use "exec-path",
completely replace $PATH with GNUPG_LIBEXECDIR before calling the
keyserver helper. If the user does use "exec-path", append
GNUPG_LIBEXECDIR after the specified path.
delete_inv_parts), keyserver.c (parse_keyserver_options): add new
--import-options option. The only current flag is "allow-local-sigs".
* g10.c (main): Don't disable MDC in pgp7 mode.
* options.h, g10.c (main), keyserver.c (parse_keyserver_options): Remove
old keyserver-option include-attributes now that there is an export-option
for the same thing.
g10.c (main): add new --export-options option. Current flags are
"include-non-rfc", "include-local-sigs", "include-attributes", and
"include-sensitive-revkeys".
* options.h, hkp.c (hkp_export), keyserver.c (parse_keyserver_options,
keyserver_spawn): try passing unknown keyserver options to export options,
and if successful, use them when doing a keyserver --send-key.
* build-packet.c (build_sig_subpkt): We do not generate
SIGSUBPKT_PRIV_VERIFY_CACHE anymore.
designated revoker means "sensitive", not "local". It's exportable under
the right circumstances.
* main.h, options.h, export.c (do_export_stream), g10.c (main), hkp.c
(hkp_export), keyserver.c (keyserver_spawn: Add a flag to skip attribute
packets and their signatures while exporting. This is to accomodate
keyservers (pksd again) that choke on attributes. Use keyserver-option
"include-attributes" to control it. This defaults to ON (i.e. don't
skip).
keyserver_work), hkp.c (hkp_ask_import, hkp_export, hkp_search): Use a
much more strict reading of RFC-2396 for the keyserver URIs. Specifically,
don't try and be smart about checking the value of ":port" so long as it
is all digits, and properly handle opaque data (those scheme specific
parts that do not start with "//").
use the original keyserver URI for cosmetics rather than trying to
recreate it when needed.
* mainproc.c (check_sig_and_print): Properly disregard expired uids.
Make sure that the first uid listed is a real uid and not an attribute
(attributes should only be listed in the "aka" section). When there are
no valid textual userids, try for an invalid textual userid before using
any attribute uid.
