* tools/gpgconf-comp.c (gc_component_kill): Reverse the order.
--
The order matters in a corner case; On a busy machine, there was a
race condition between gpg-agent's running KILLAGENT command and its
accepting incoming request on the socket. If a request by
gpg-connect-agent was accepted, it resulted an error by sudden
shutdown. This change of the order can remove such a race.
Here, we know backend=0 is none.
GnuPG-bug-id: 4577
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* agent/call-pinentry.c (watch_sock_start): Factor out
from do_getpin.
(watch_sock_end): Likewise.
(do_getpin): Use those functions.
(agent_get_confirmation): Likewise.
(popup_message_thread): Likewise.
--
Pinentry's dialog for confirmation should be also closed cleanly, as
well as the dialog for pin-input.
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* scd/app.c (check_application_conflict): Add args to pass a serialno.
* scd/command.c (open_card_with_request): Pass the serialno to
check_application_conflict.
--
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/app.c (compare_card_list_items): New.
(app_send_card_list): Sort the card objects by slot.
--
This is required so that in gpg-card a "list N" command always returns
the expected card. Sorting by slot should be sufficient.
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/app-common.h (struct app_ctx_s): Add func ptr 'reselect'.
* scd/app-piv.c (do_reselect): New.
(app_select_piv): Move AID constant to file scope.
* scd/app-openpgp.c (do_reselect): New.
(app_select_openpgp): Move AID constant to file scope.
* scd/app.c (apptype_from_name): New.
(check_application_conflict): Check against all apps of the card.
Always set current_apptype.
(select_additional_application): New.
(maybe_switch_app): New.
(app_write_learn_status, app_readcert, app_readkey, app_getattr)
(app_setattr, app_sign, app_auth, app_decipher, app_writecert)
(app_writekey, app_genkey, app_change_pin, app_check_pin): Use it here.
(app_do_with_keygrip): Force reselect on success.
(app_new_register): Move setting of CURRENT_APPTYPE to ...
(select_application): here so that it will be set to the requested
card.
* scd/command.c (open_card_with_request): Select additional
application if possible.
--
Noet that we will likely need to rework this even more so to get well
defined semantics for card access.
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/app.c (app_do_with_keygrip): Lock the card.
--
Better safe than sorry.
We should also review the card reference counting to see whether we
better ref the returned card object already here.
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/app.c: Chnage all function dispatcher.
--
This change will allow us to easier integrate an app swithcing logic.
The change should have no user visible effect. The error checking we
do now with the card locked will rarely be asserted. It is the
correct thing to do anyway.
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/app.c (check_conflict): Fold into ...
(check_application_conflict): this and adjust callers. Return a
different error code if it is possible to switch apps.
--
Right now this change does nothing visible.
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/scdaemon.h: Include app-common.h. Remove inclusion of that
header from all other files.
(card_t, app_t): Move typedef to ...
* scd/app-common.h: here. Use them in the defs.
--
In another patch we will need apptype_t in the ctrl object and thus we
need to reorganize things a bit now. Given that most files need
app-common anyway it makes sense to always include it.
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/app-common.h (cardtype_t): New.
(apptype_t): New.
(struct card_ctx_s): Change type of cardtype.
(struct app_ctx_s): Change type of apptype. Adjust all users.
* scd/app.c (struct app_priority_list_s): Add field apptype.
(strcardtype): New. Use as needed.
(strapptype): New. Use as needed.
--
Using strcmp is lame and we can't use a switch to let the compiler
complain about missed cases.
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/app-common.h (struct card_ctx_s): New.
(struct app_ctx_s): Factor card specific fields out to card_ctx_s.
(app_get_slot): New.
* scd/scdaemon.h (card_t): New.
(struct server_control_s): Rename field app_ctx to card_ctx and change
all users.
* scd/app-dinsig.c: Use app_get_slot and adjust for chang in card
related fields.
* scd/app-geldkarte.c: Ditto.
* scd/app-nks.c: Ditto.
* scd/app-openpgp.c: Ditto.
* scd/app-p15.c: Ditto.
* scd/app-sc-hsm.c: Ditto.
* scd/app.c: Lost of changes to adjust for the changed data
structures. Change all callers.
(app_list_lock): Rename to card_list_lock.
(app_top): Remove.
(card_top): New.
(lock_app): Rename to lock_card and change arg type.
(unlock_app): Rename to unlock_card.
(app_dump_state): Print card and app info.
(app_reset): Rename to card_reset.
(app_new_register): Change for the new data structure.
(deallocate_card): Dealloc card and all apps.
(app_ref): Rename to card_ref.
(app_unref): Rename to card_unref.
(app_unref_locked): Rename to card_unref_locked.
(card_get_serialno): New.
* scd/command.c (cmd_pkdecrypt): Actually use the looked up card and
former app object and not the standard one from the context.
--
Although quite large, this is a straightforward change to separate
card/token related data from card application related data. Before
this change there was a one-to-one relation between card and
application and no way to represent several applications on a card.
The new data structure will allow for such a representation.
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/app.c (app_ref): New.
(app_unref): New.
(release_application): Renamed to ...
(app_unref_locked): this and remove arg locked_already. Change
callers to use this or app_ref.
* scd/command.c (open_card_with_request):
(cmd_pksign, cmd_pkauth, cmd_pkdecrypt): Use app_ref and app_unref
instead of accessing the counter directly.
--
This is better in case we need to debug stuff. There is a real change
however: We now lock and unlock the app before changing the reference
count.
The whole app locking business should be reviewed because we pass
pointers along without immediately bumping the refcount.
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/app-openpgp.c (do_with_keygrip): Return a real error code to
avoid misinterpretation of the result. Also fix the case for a too
small buffer.
--
The only real chnage is the case for a too small buffer. That should
in general never happen but if so we now return an error instead of
success.
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/app-openpgp.c (get_public_key): Use gnupg_module_name instead of
just "gpg".
--
There is no bug report regarding this and it would be very unlikely
but we should always use the gpg belonging to our code.
Signed-off-by: Werner Koch <wk@gnupg.org>
--
Please use useful subjects so that there is no need to lookup what a
fix is. A commit fix should be indicated with the keyword
"Fixes-commit: xxxxx"
A new user who sees "policy-file" and searches naively through the
documentation to find it again won't be able to tell what this refers
to, since "policies.txt" doesn't otherwise match the search string
"policy". This gives them a fighting chance at finding the
documentation.
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
* tools/gpgparsemail.c (parse_message): Revert the change.
* tools/rfc822parse.c (transition_to_body): Set ERRNO.
(transition_to_header, insert_header): Likewise.
--
In the comment of rfc822parse_* functions, it explicitly explained
setting ERRNO on error. For parser errors, it may not have
appropriate ERRNO, in such a case, use ENOENT.
Fixes-commit: c13e459ffeffb8c5387c44b3c04bb92b7111a75b
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* sm/certlist.c (gpgsm_add_to_certlist): Add diagnostic and fold two
similar branches.
--
Without this patch gpgsm printed:
gpgsm[23045]: DBG: chan_6 <- RECIPIENT edward.tester@demo.gnupg.com
gpgsm[23045]: certificate is not usable for encryption
gpgsm[23045]: certificate is good
with this patch a
gpgsm[23045]: looking for another certificate
is inserted into the log.
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/gpg.c (g10_exit): Block all signals before calling
emergency_cleanup.
--
There is a race condition here which results crash of the process.
When a signal is delivered in emergency_cleanup, it is called again.
This change fixes the problem.
GnuPG-bug-id: 2747
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* agent/call-pinentry.c (start_pinentry): When TERM is none,
don't send OPTION ttytype to pinentry.
--
GnuPG-bug-id: 4137
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* doc/scdaemon.texi (card-timeout): Remove.
* scd/scdaemon.c (main): Remove oCardTimeout handling.
--
There was the card-timeout option in GnuPG 2.0, but it was never
implemented correctly. The intention of this option was to allow
sharing smartcard among multiple applications, but this didn't work
well as user's expectation (it only worked with DISCONNECT command).
This is because other parts of scdaemon assumes exclusive access. In
GnuPG 2.1, the support of the option was removed, improving
"DISCONNECT" command always works well without this option.
GnuPG-bug-id: 3383
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* agent/command.c (cmd_getinfo): Return GPG_ERR_FALSE as boolean False.
* g13/server.c (cmd_getinfo): Ditto.
* sm/server.c (cmd_getinfo): Ditto.
--
GPG_ERR_FALSE was introduced with libgpg-error 1.21 and we now require
a later version for gnupg 2. Thus we can switch to this more
descriptive code.
Signed-off-by: Werner Koch <wk@gnupg.org>
* agent/command.c (do_one_keyinfo): Add ON_CARD argument to put
A-flag.
(cmd_keyinfo): Call agent_card_keyinfo to offer additional information
if it's on card.
--
This is a modification in gpg-agent, intended for better
enum_secret_keys in gpg frontend.
GnuPG-bug-id: 4244
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* dirmngr/ocsp.c (do_ocsp_request): Remove arg md. Add args r_sigval,
r_produced_at, and r_md. Get the hash algo from the signature and
create the context here.
(check_signature): Allow any hash algo. Print a diagnostic if the
signature does not verify.
--
GnuPG-bug-id: 3966
Signed-off-by: Werner Koch <wk@gnupg.org>
* dirmngr/certcache.c (find_cert_bysubject): Add better debug output
and try to locate by keyid.
--
This chnages was suggested in
GnuPG-bug-id: 4536
but we do not have any test cases for this.
Signed-off-by: Werner Koch <wk@gnupg.org>
* agent/pksign.c (do_encode_md): Use ascii_tolower and avoid
uninitalized TMP in the error case.
--
This is just in case libgcrypt ever returns an algorithm name longer
than 15 bytes.
Signed-off-by: Werner Koch <wk@gnupg.org>
* sm/certlist.c (cert_usage_p): Add arg 'silent' and change all
callers.
(gpgsm_cert_use_sign_p): Add arg 'silent' and pass to cert_usage_p.
Change all callers.
* sm/sign.c (gpgsm_get_default_cert): Set SILENT when calling
gpgsm_cert_use_sign_p
--
GnuPG-bug-id: 4535
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/tofu.c: Removed some translation markups which either make no
sense or are not possble.
--
Error message which are not helpful for the user but indicate a
problem of the installation or the code do not need a translation.
The translator may not understand them correctly and the use support
can't immediately locate the problem because it needs to be reverse
translated.
There is also one case where certain grammar constructs are
assumed (concatenating parts of a sentence at runtime). Better do not
translate that than getting weird sentences.
* common/userids.c (classify_user_id): Do not set the EXACT flag in
the default case.
* g10/export.c (exact_subkey_match_p): Make static,
* g10/delkey.c (do_delete_key): Implement subkey only deleting.
--
GnuPG-bug-id: 4457
* g10/keydb.c (parse_keyblock_image): Treat invalid packet special.
--
This is in particular useful to run --list-keys on a keyring with
corrupted packets. The extra flush is to keep the diagnostic close to
the regular --list-key output.
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/parse-packet.c: Move max packet lengths constants to ...
* g10/packet.h: ... here.
* g10/build-packet.c (do_user_id): Return an error if too data is too
large.
* g10/keygen.c (write_uid): Return an error for too large data.
--
This can lead to keyring corruption becuase we expect that our parser
is abale to parse packts created by us. Test case is
gpg --batch --passphrase 'abc' -v \
--quick-gen-key $(yes 'a'| head -4000|tr -d '\n')
GnuPG-bug-id: 4532
Signed-off-by: Werner Koch <wk@gnupg.org>