* common/asshelp.c (wait_for_sock): instead of checking the socket
every second, we check 10 times in the first second (with exponential
decay).
--
This cuts the wall clock time for the standard test suite roughly by
half.
GnuPG-bug-id: 3490
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
(cherry picked from commit 149041b0b917f4298239fe18b5ebd5ead71584a6)
* common/asshelp.c (wait_for_sock): New function, collecting
codepaths from...
(start_new_gpg_agent) here and...
(start_new_dirmngr) here.
--
This has no functional change, but makes it easier to make this
function more efficient.
GnuPG-bug-id: 3490
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
(cherry picked from commit 0471ff9d3bf8d6b9a359f3c426d70d0935066907)
* g10/getkey.c (getkey_ctx_s): New field found_via_akl.
(get_pubkey_byname): Set it.
(only_expired_enc_subkeys): New.
(get_best_pubkey_byname): Add support to refresh expired keys from the
WKD.
--
A little drawback of that code is that if the WKD has no update for an
expired key each access of the key will trigger a WKD lookup (unless
cached by the dirmngr). To avoid this we need to record the last time
we have checked for an update but that would in turn require that we
update the keyring for each check. We defer this until we have a
better key database which allows for fast updates of meta data.
Testing the code is currently a bit cumbersome because it requires to
update a key in the WKD several times. Eventually we we need a
network emulation layer to provide sample data for the regression
tests.
GnuPG-bug-id: 2917
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit 7f172404bfcf719b9b1af4a182d4803525ebff7c)
* g10/parse-packet.c (parse_symkeyenc): Use iobuf_get_noeof.
(parse_key): Likewise.
--
Cherry picked from master commit:
1b309d9f6199a91caa0ca0b97b92d599e00b736e
When EOF comes at parsing s2k.count, it is possible the value will
be (unsigned long)-1. Then, the result of S2K_DECODE_COUNT will be
undefined. This patch fixes undefined behavior.
Reported-by: Philippe Antoine
GnuPG-bug-id: 4093
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* g10/call-dirmngr.c (MAX_WKD_RESULT_LENGTH): New.
(gpg_dirmngr_wkd_get): Use it.
--
WKD should return only a single key with just one UID. For key
rollover 2 keys may be send. A total of 256 KiB seems to be a
generous limit here.
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit e88f56f1937ac92f6a3b94e50b6db2649ec0be41)
* dirmngr/server.c (proc_wkd_get): Check the returned SRV record names
to mitigate rogue DNS servers.
--
I am not sure wether this really is very useful because the security
relies on a trustworthy DNS system anyway. However, that check is
easy enough to do.
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit ebe727ef596eefebb5eff7d03a98649ffc7ae3ee)
* g10/key-clean.c (clean_key): Rename to clean_all_uids and split
subkey cleaning into ...
(clean_all_subkeys): new. Call that always after the former clean_key
invocations.
--
Note that the clean_all_subkeys function will later be extended.
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit 6c3567196f7e72552f326ce07dccbcce31926e5d)
* g10/mainproc.c (symkey_decrypt_seskey): Check for a valid algorithm.
(proc_symkey_enc): Clear passpharse on error from above function.
--
This does not work reliable as stated in bug 3795 but we can try to
fix ~95% of all cases. The real fix is to use AEAD which will come
with 2.3
GnuPG-bug-id: 3795, 4050
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/getkey.c (get_pubkey_for_sig): New.
(get_pubkeyblock_for_sig): New.
* g10/mainproc.c (issuer_fpr_raw): Give global scope.
(check_sig_and_print): Use get_pubkeyblock_for_sig.
* g10/pkclist.c (check_signatures_trust): Use get_pubkey_for_sig.
* g10/sig-check.c (check_signature2): Ditto.
(check_signature_over_key_or_uid): Ditto.
--
GnuPG-bug-id: 4046
The whole getkey stuff is still a mess with way to much duplication
and missing caching of already fetched data.
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit f7526c7bc754acf68bde0b79c785e875a9365d60)
* po/de.po (decryption forced to fail!): Fix translation.
--
The unmatched %s actually produced a crash on Windows.
GnuPG-Bug-Id: T4053
GnuPG-Bug-Id: T4054
* g10/import.c (read_block): Add special treatment for bad user ids
and comment packets.
--
See
GnuPG-bug-id: 4022
for an example of a bogus user id.
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit 01cd66f9faf1623833e6afac84164de5a136ecff)
* g10/sig-check.c (check_signature_end_simple): Check sign usage.
--
Without this patch the signature verification fails only due to the
missing back signature. This check better explains what went wrong.
GnuPG-bug-id: 4014
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit 214b0077264e35c079e854a8b6374704aea45cd5)
* g10/import.c (get_revocation_reason): New.
(list_standalone_revocation): Extend function.
--
Note that this function extends the "rvs" field signature-class (field
11) with the revocation reason. GPGME does not yet parse this but it
can be expected that the comma delimiter does not break other parsers.
A new field is added to the "rvs" (and in future also the "rev")
record to carry a record specific comment. Hopefully all parsers
meanwhile learned the lesson from other new fields and don't bail out
on more fields than they know about.
This is partial solution to
GnuPG-bug-id: 1173
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit b7cd2c2093ae1b47645be50fa1d431a028187cad)
* dirmngr/dns.c (dns_res_exec): If it's DNS_SO_SOCKS_CONN, don't
iterate to other server, but return the error immediately.
--
Cherry picked from master commit:
1aacd12471935a354cfd85ee1805edc7eb16e6c5
In the function libdns_switch_port_p in dns-stuff.c, this patch
allows to fallback using TOR_PORT2 correctly.
Fixes-commit: bcdbf8b8ebe9d61160e0b007dabe1b6462ffbc93
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* dirmngr/dns.c (LEAVE_SELECTION_OF_PORT_TO_KERNEL): New.
(dns_socket): Don't select ephemeral port in user space.
--
Cherry picked from master commit:
861f1da0731bf29dcb9221c4f22c76b40ec15a78
There is no good reason to bind local port aggressively. It might be
some reason to do so, then, a user can specify it in /etc/resolv.conf
by the second argument of "interface" directive.
At least, it causes a problem on Windows. Binding a specified port in
user space can trigger the Firewall dialog on Windows. Since it can
be considered valid question, it is better not to bind with an
ephemeral port which is selected in user space, by default.
GnuPG-bug-id: 3610
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* dirmngr/dns.c (dns_so_check): When EINVAL, release the association
by connect with AF_UNSPEC and try again. Also try again for
ECONNREFUSED.
(dns_res_exec): Try next nameserver when ECONNREFUSED.
--
Cherry pick from master commit:
bcdbf8b8ebe9d61160e0b007dabe1b6462ffbc93
GnuPG-bug-id: T3374
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* dirmngr/dns.c (dns_resconf_pton): Clear SS.
(dns_resconf_setiface): Clear ->IFACE.
(dns_hints_root, send_query): Clear SS.
--
Cherry pick from master commit:
1c0b6681e4f322b88ac35d1f21c03d3cfc35fc23
POSIX requires clear the structure of struct sockaddr_in6. On macOS,
in some case like bind, it is better to clear even for struct
sockaddr_in.
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* dirmngr/dns-stuff.c (libdns_init): Initialize options.recurse.
--
Cherry pick from master commit:
5b40338f12762cd74238c2d2b3101c33dd2d0ed3
To reproduce an error, run:
./t-dns-stuff --debug --recursive-resolver www.gnupg.org
Then, it returns "No name" error. That's because there was only setup
for root servers, and no setup for recursive query in fact.
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* g10/import.c (import_revoke_cert): Add arg 'options'. Take care of
IMPORT_DRY_RUN.
--
GnuPG-bug-id: 4017
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit 6c9e37b2e541e96e7c65e1679982a614b4863fdb)
* g10/import.c (impex_filter_getval): Add new "usage" property for
drop-subkey filter.
--
For example, this permits extraction of only encryption-capable
subkeys like so:
gpg --export-filter 'drop-subkey=usage !~ e' --export $FPR
GnuPG-Bug-id: 4019
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
(cherry picked from commit 2ddfb5bef920919443309ece9fa2930282bbce85)
* g10/gpg.c (main): Set some list options.
--
The new command --show-keys is commonly used to check the content of a
file with keys. In this case it can be expected that all included
subkeys and uids are of interested, even when they are already expired
or have been revoked.
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit d2bc66f241a66cc95140cbb3a07555f6301290ed)
* g10/mainproc.c (proc_encrypted): Use constant from logging.h
--
Because the log levels are enums I had to change there names in
libgpg-error to avoid clashes. Master uses the new names but 2.2
needs to stick to the old names.
Fixes-commit: 825909e9cd5f344ece6c0b0ea3a9475df1d643de
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/mainproc.c (proc_plaintext): Sanitize verbose output.
--
This fixes a forgotten sanitation of user supplied data in a verbose
mode diagnostic. The mention CVE is about using this to inject
status-fd lines into the stderr output. Other harm good as well be
done. Note that GPGME based applications are not affected because
GPGME does not fold status output into stderr.
CVE-id: CVE-2018-12020
GnuPG-bug-id: 4012
(cherry picked from commit 13f135c7a252cc46cff96e75968d92b6dc8dce1b)
* g10/key-check.c (key_check_all_keysigs): Factor some code out to ...
(remove_duplicate_sigs): new.
(key_check_all_keysigs): Call remove_duplicate_sigs again after
reordering.
--
This is a follupup for commit 26bce2f01d2029ea2b8a8dbbe36118e3c83c5cba
to cleanup the code and to add a second de-duplicate step when needed.
GnuPG-bug-id: 3994
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit 26746fe65d14a00773473c2d0d271406a5105bca)
* g10/packet.h (PKG_siganture): Add field 'help_counter'.
* g10/key-check.c (sig_comparison): Take care of HELP_COUNTER.
(key_check_all_keysigs): De-duplicate on a per-block base.
--
The key_check_all_keysigs first does a detection of duplicate
signature. This is done over all signatures at once. The problem
here is for example:
key
uid_1
sig_uid_1.1
sig_uid_1.2
subkey_1
sig_sub_1.1
subkey_2
sig_sub_2.1
sig_sub_2.2 (duplicate of sig_sub_1.1)
Now the de-duplication deletes the first signature and keeps the
second. That works in most cases for foreign signature on userids but
in the above constellation the code simply removes sig_sub_1.1 so that
subkey_1 has no binding signature anymore. In a later step during
import the missing binding is detected and subkey_1 is removed because
it is not anymore valid. The sig_sub_2.2 will also be removed later
because it does not check out for subkey_2 (that is as expected).
The fix is to let the de-duplication work only on blocks (ie. within
the signatures of a user id or a subkey). This will not detect all
duplicates but that does not harm because later steps will detect and
remove them.
In the above case (with this patch applied) the second phase of
key_check_all_keysigs will reorder key signatures and move the
duplicate sig_sub_2.2 directly after sig_sub_1.1. This duplicates the
signature and for cleanness we should kick the de-duplication process
again. This will be done with a followup patch.
GnuPG-bug-id: 3994
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit 26bce2f01d2029ea2b8a8dbbe36118e3c83c5cba)
* g10/import.c (chk_self_sigs): Print the subkeyid in addition to the
keyid.
(delete_inv_parts): Ditto.
Signed-off-by: Werner Koch <wk@gnupg.org>
(cherry picked from commit 1bc6b5174248ba4d83d648ef6d6f4550540d1f20)
