* g10/armor.c (asctobin): Larger look-up table for fast path.
(initialize): Update 'asctobin' initialization.
(radix64_read): Add fast path for radix64 to binary conversion.
--
This patch adds fast path for radix64 to binary conversion in
armored decryption.
Benchmark results below, tested on Intel Core i7-4790K (turbo off).
Encrypted 2 GiB through pipe to ramfs file using AES128. Decrypt
ramfs file out through pipe to /dev/null.
before patch-set
----------------
gpg process
armor: user time pipe transfer rate
encrypt-aead: 13.8 140 MB/s
decrypt-aead: 30.6 68 MB/s
encrypt-cfb: 17.4 114 MB/s
decrypt-cfb: 32.6 64 MB/s
after (decrypt+iobuf+crc+radix64 opt)
-------------------------------------
gpg process
armor: user time pipe transfer rate
decrypt-aead: 9.8 200 MB/s
decrypt-cfb: 11.9 168 MB/s
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@iki.fi>
* g10/armor.c (bintoasc): Change to read-only.
(initialize): Use const pointer for 'bintoasc'.
(armor_output_buf_as_radix64): New function for faster binary to
radix64 conversion.
(armor_filter): Use new conversion function.
--
This patch adds faster binary to radix64 conversion to speed up
armored encryption.
Benchmark results below, tested on Intel Core i7-4790K (turbo off).
Encrypted 2 GiB through pipe to ramfs file using AES128. Decrypt
ramfs file out through pipe to /dev/null.
before patch-set
----------------
gpg process
armor: user time pipe transfer rate
encrypt-aead: 13.8 140 MB/s
decrypt-aead: 30.6 68 MB/s
encrypt-cfb: 17.4 114 MB/s
decrypt-cfb: 32.6 64 MB/s
after (decrypt+iobuf+crc+radix64 opt)
-------------------------------------
gpg process
armor: user time pipe transfer rate
encrypt-aead: 2.7 523 MB/s
encrypt-cfb: 6.7 264 MB/s
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@iki.fi>
* g10/armor.c (CRCINIT, CRCPOLY, CRCUPDATE, crc_table): Remove.
(new_armor_context): Open libgcrypt CRC24 context.
(release_armor_context): Close CRC24 context.
(initialize): Remove CRC table generation.
(get_afx_crc): New.
(check_input, fake_packet, radix64_read, armor_filter): Update to use
CRC24 context.
* g10/filter.h (armor_filter_context_t): Replace crc intermediate value
with libgcrypt md context pointer.
--
This patch changes armor filter to use optimized CRC24 implementation
from libgcrypt to speed up encryption and decryption.
Benchmark results below, tested on Intel Core i7-4790K (turbo off).
Encrypted 2 GiB through pipe to ramfs file using AES128. Decrypt
ramfs file out through pipe to /dev/null.
before patch-set
----------------
gpg process
armor: user time pipe transfer rate
encrypt-aead: 13.8 140 MB/s
decrypt-aead: 30.6 68 MB/s
encrypt-cfb: 17.4 114 MB/s
decrypt-cfb: 32.6 64 MB/s
after (decrypt+iobuf+crc opt)
-----------------------------
gpg process
armor: user time pipe transfer rate
encrypt-aead: 8.7 211 MB/s
decrypt-aead: 17.6 116 MB/s
encrypt-cfb: 12.6 153 MB/s
decrypt-cfb: 19.6 105 MB/s
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@iki.fi>
* common/iobuf.c (iobuf_read_line): Add fast path for finding '\n'
character in buffer.
--
This patch reduce per byte overhead in iobuf_read_line by avoiding
using iobuf_get when possible and use memchr to find '\n'. This
speeds armored decryption.
Benchmark results below, tested on Intel Core i7-4790K (turbo off).
Encrypted 2 GiB through pipe to ramfs file using AES128. Decrypt
ramfs file out through pipe to /dev/null.
before patch-set
----------------
gpg process
armor: user time pipe transfer rate
encrypt-aead: 13.8 140 MB/s
decrypt-aead: 30.6 68 MB/s
encrypt-cfb: 17.4 114 MB/s
decrypt-cfb: 32.6 64 MB/s
after (decrypt+iobuf opt)
-------------------------
gpg process
armor: user time pipe transfer rate
decrypt-aead: 22.5 92 MB/s
decrypt-cfb: 24.4 85 MB/s
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@iki.fi>
* g10/armor.c (radix64_read): Check EOF with '!afx->buffer_len' instead
of 'c == -1', as 'c' is never set to this value.
--
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@iki.fi>
* g10/decrypt-data.c (fill_buffer): Use iobuf_read instead of iobuf_get
for reading data.
--
This patch reduces iobuf_read per byte processing overhead and speeds
up decryption.
Benchmark results below, tested on Intel Core i7-4790K (turbo off).
Encrypted 2 GiB through pipe to ramfs file using AES128. Decrypt
ramfs file out through pipe to /dev/null.
before patch-set
----------------
gpg process
no-armor: user time pipe transfer rate
encrypt-aead: 1.02 1.0 GB/s
decrypt-aead: 10.8 185 MB/s
encrypt-cfb: 4.8 342 MB/s
decrypt-cfb: 12.7 157 MB/s
gpg process
armor: user time pipe transfer rate
encrypt-aead: 13.8 140 MB/s
decrypt-aead: 30.6 68 MB/s
encrypt-cfb: 17.4 114 MB/s
decrypt-cfb: 32.6 64 MB/s
after (decrypt opt)
-------------------
gpg process
no-armor: user time pipe transfer rate
decrypt-aead: 7.3 263 MB/s
decrypt-cfb: 9.3 211 MB/s
gpg process
armor: user time pipe transfer rate
decrypt-aead: 27.0 77 MB/s
decrypt-cfb: 29.0 72 MB/s
Note: decryption results are much slower than encryption because of
extra SHA1 & RIPEMD160 hashing.
GnuPG-bug-id: 3786
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@iki.fi>
* g10/cipher-aead.c (do_flush): No cast is correct.
* g10/decrypt-data.c (aead_underflow): No cast needed.
Use "%j" for uint64_t for chunklen.
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* tools/gpg-wks.h (opt): Add field with_colons.
* tools/gpg-wks-client.c (oWithColons): New const.
(opts, parse_arguments): Add option --with-colons.
(main): Change aSupported to take several domains in --with-colons
mode.
(command_send): Factor policy getting code out to ...
(get_policy_and_sa): New function.
(command_supported): Make use of new function.
--
In addition to this the --create command now also supports a
submission address only in the policy file. That means the
submission-address file is not anymore required and can be replaced by
the policy file.
Signed-off-by: Werner Koch <wk@gnupg.org>
* build-aux/speedo.mk (speedo_pkg_gpgme_configure): Remove
--disable-w32-qt option.
--
This option is obsolete since GPGME 1.7 (in 2016)
Signed-off-by: Werner Koch <wk@gnupg.org>
* dirmngr/misc.c (host_and_port_from_url): Fix bad port parsing and a
segv for a missing slash after the host name.
--
Reportted-by: Tomas Mraz
GnuPG-bug-id: 4230
Signed-off-by: Werner Koch <wk@gnupg.org>
* tools/gpg-wks-server.c (opts): Add '--directory',
(main): Explain how to set correct permissions.
(command_list_domains): Create an empty policy file and remove the
warning for an empty policy file.
--
Note that a policy file is meanwhile required and thus is is useful to
create it.
Signed-off-by: Werner Koch <wk@gnupg.org>
* kbx/keybox-search-desc.h (fpr): Increase the size.
--
In the function keydb_search_fpr in g10/keydb.c, it is copied using
MAX_FINGERPRINT_LEN. So, more size is required.
Fixes-commit: ecbbafb88d920e713439b6b1b8e1b41a6f8d0e38
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* dirmngr/ks-engine-ldap.c (extract_attributes): Don't use a variabale
out of scope and cleanup the entire pgpKeySize block.
--
GnuPG-bug-id: 4229
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/call-agent.c (learn_status_cb): Parse "bt" flag.
* g10/call-agent.h: New member field "bt".
* g10/card-util.c (uif): Limit its access only when it is supported.
* scd/app-openpgp.c (do_setattr): Allow access to UIF objects only
when there is a button.
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* dirmngr/server.c (proc_wkd_get): Tack the raw local address to the
request.
--
We append the raw non-canonicalized local address part to the hash.
Servers who serve the requests from static files will ignore the
parameters and a test with posteo shows that also services using a
database ignore the parameter. The general idea is that service
providers may use their own canonicalization rules. The problem is
that we currently filter the returned key for the full mail address
and thus we will never see a key if the service did a different
canonicalization than we. So consider this to be an experiment.
Signed-off-by: Werner Koch <wk@gnupg.org>
* agent/command-ssh.c (get_client_info): Turn client_uid into an int.
Fix setting of it in case of a failed getsocketopt.
* agent/command.c (start_command_handler): Fix setting of the pid and
uid for Windows.
--
Fixes-commit: 28aa6890588cc108639951bb4bef03ac17743046
which obviously was only added to master.
Signed-off-by: Werner Koch <wk@gnupg.org>
* tools/rfc822parse.c (HEADER_NAME_CHARS): New. Taken from
mime-maker.c.
(rfc822_valid_header_name_p): New. Based on code from mime-maker.c.
(rfc822_capitalize_header_name): New. Copied from mime-maker.c.
(capitalize_header_name): Remove. Replace calls by new func.
(my_toupper, my_strcasecmp): New.
* tools/mime-maker.c: Include rfc822parse.h.
(HEADER_NAME_CHARS, capitalize_header_name): Remove.
(add_header): Replace check and capitalization by new functions.
--
This is a straightforward change with two minor chnages:
- In rfc822parse.c the capitalization handles MIME-Version special.
- The check in mime-maker bow detects a zero-length name as invalid.
my_toupper and my_strcasecmp are introduced to allow standalone use
of that file.
Signed-off-by: Werner Koch <wk@gnupg.org>
* scd/app-openpgp.c (do_sign): Clear DID_CHV1 after signing.
--
We have a corner case: In "not forced" situation and authenticated,
and it is changed to "forced", card implementaiton can actually accept
signing, but GnuPG requires authentication, because it is "forced".
GnuPG-bug-id: 4177
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* agent/divert-scd.c (getpin_cb): Support --ack option.
--
We are now introducing "acknowledge button" feature to scdaemon,
so that we can support OpenPGPcard User Interaction Flag.
We will (re)use the mechanism of POPUPPINPADPROMPT for this. Perhaps,
we will change the name of POPUPPINPADPROMPT, since it will be no
longer for PINPAD only.
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* g10/tdbio.c (tdbio_update_version_record): Never store a TOFU model.
(create_version_record): Don't init as TOFU.
(tdbio_db_matches_options): Don't indicate a change in case TOFU is
stored in an old trustdb file.
--
This change allows to switch between a tofu and pgp or tofu+pgp trust
model without an auto rebuild of the trustdb. This also requires that
the tofu trust model is requested on the command line. If TOFU will
ever be the default we need to tweak the model detection via TM_AUTO
by also looking into the TOFU data base,
GnuPG-bug-id: 4134
* g10/sig-check.c (check_signature_end_simple):
--
Obviously we should not ignore a back signature here.
Fixes-commit: 214b0077264e35c079e854a8b6374704aea45cd5
GnuPG-bug-id: 4014
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/pubkey-enc.c (get_session_key): Take care of
opt.skip_hidden_recipients.
--
This was lost due to
Fixes-commit: ce2f71760155b71a71418fe145a557c99bd52290
GnuPG-bug-id: 4169
Signed-off-by: Werner Koch <wk@gnupg.org>
* g10/options.h (IMPORT_DROP_UIDS): New.
(EXPORT_DROP_UIDS): New.
* g10/import.c (parse_import_options): Add option "import-drop-uids".
(import_one): Don't bail out with that options and no uids found.
Also remove all uids.
(remove_all_uids): New.
* g10/export.c (parse_export_options): Add option "export-drop-uids".
(do_export_one_keyblock): Implement option.
--
These options are required for experiments with changes to the
keyserver infrastructure.
Signed-off-by: Werner Koch <wk@gnupg.org>
* common/sysutils.c (gnupg_reopen_std): Use fcntl instead of fstat.
--
When gpg was invoked by a Perl web application on FreeBSD, fstat in
gnupg_reopen_std failed with EBADF. Using fcntl, which is considered
lighter than fstat, it works fine. Since uur purpose is to check if
file descriptor is valid or not, lighter operation is better.
Reported-by: Marcin Gryszkalis <mg@fork.pl>
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
* g10/skclist.c (enum_secret_keys): Use SK_LIST instead of pubkey_t.
--
The use of pubkey_t was wrong. The use is just a list of keys, not
with keyblock. With SK_LIST, release_sk_list releases memory by
free_public_key.
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>
