Heiko Schlittermann (HS12-RIPE) [Wed, 19 Aug 2015 16:18:22 +0000 (18:18 +0200)]
Testsuite: Update README about flavours
Heiko Schlittermann (HS12-RIPE) [Wed, 19 Aug 2015 14:43:35 +0000 (16:43 +0200)]
Testsuite: Update debian8 flavour for 5600
Heiko Schlittermann (HS12-RIPE) [Wed, 19 Aug 2015 13:22:41 +0000 (15:22 +0200)]
Fix post-transport-crash.
The crash probably was introduced in
a39bd74d3e94 and
needs 'split_spool_directory=yes' to expose.
Thanks to Wolfgang Breyha, who found the same fix.
Jeremy Harris [Mon, 17 Aug 2015 18:08:52 +0000 (19:08 +0100)]
Docs: add note on spf lookup type, in experimental-spec
Heiko Schlittermann (HS12-RIPE) [Mon, 17 Aug 2015 15:38:47 +0000 (17:38 +0200)]
Docs: Remove trailing whitespace
Heiko Schlittermann (HS12-RIPE) [Mon, 17 Aug 2015 15:37:08 +0000 (17:37 +0200)]
Docs: Clarify white space in $h_<name> expansion
Tony Finch [Thu, 13 Aug 2015 14:16:51 +0000 (15:16 +0100)]
Improve the consistency of logging incoming and outgoing interfaces.
The I= interface field on outgoing lines is now after the H= remote
host field, same as incoming lines. There is a separate outgoing_interface
log selector which allows you to disable the outgoing I= field.
(slight massaging by JH)
Tony Finch [Thu, 13 Aug 2015 14:16:48 +0000 (15:16 +0100)]
Overhaul the debug_selector and log_selector machinery to support variable-length bit vectors. No functional change.
Heiko Schlittermann (HS12-RIPE) [Wed, 12 Aug 2015 22:20:12 +0000 (00:20 +0200)]
Fix timeout adjustment in
c528cec4
Heiko Schlittermann (HS12) [Wed, 12 Aug 2015 22:13:51 +0000 (00:13 +0200)]
Testsuite: Add short help to server
Heiko Schlittermann (HS12-RIPE) [Wed, 12 Aug 2015 21:45:44 +0000 (23:45 +0200)]
Adjust the timeout after interrupted select()
Jeremy Harris [Tue, 11 Aug 2015 21:54:53 +0000 (22:54 +0100)]
GnuTLS: avoid whining about OCSP when not requested by config
Heiko Schlittermann (HS12-RIPE) [Tue, 11 Aug 2015 15:36:29 +0000 (17:36 +0200)]
Fix ESMTP MAIL command option processing
If the address containes spaces, the option processing
was confused.
Heiko Schlittermann (HS12-RIPE) [Tue, 11 Aug 2015 07:13:11 +0000 (09:13 +0200)]
Build: Make test_{os,parse,dbfn,string} work
Heiko Schlittermann (HS12) [Mon, 10 Aug 2015 16:10:26 +0000 (18:10 +0200)]
Testsuite: Add test for MAIL with spaces + options
Heiko Schlittermann (HS12-RIPE) [Sun, 9 Aug 2015 22:39:36 +0000 (00:39 +0200)]
Really re-select() when interrupted.
Heiko Schlittermann (HS12-RIPE) [Sun, 9 Aug 2015 21:29:44 +0000 (23:29 +0200)]
Compiler quietening
Jeremy Harris [Thu, 6 Aug 2015 20:38:33 +0000 (21:38 +0100)]
OpenSSL: fix complile on pre-EC-capable library versions
Jeremy Harris [Wed, 5 Aug 2015 20:19:40 +0000 (21:19 +0100)]
Dup GnuTLS test to OpenSSL
Jeremy Harris [Wed, 5 Aug 2015 15:01:50 +0000 (16:01 +0100)]
Testsuite: Avoid requesting TLS cert-status requests from GnuTLS test utility unless wanted
Jeremy Harris [Sun, 2 Aug 2015 15:09:54 +0000 (16:09 +0100)]
Testsuite: pass GnuTLS despite OCSP whining
Jeremy Harris [Sun, 2 Aug 2015 13:33:56 +0000 (14:33 +0100)]
Docs: add notes on library version limitations on OCSP stapling. Bug 1664
Jeremy Harris [Sun, 2 Aug 2015 12:53:15 +0000 (13:53 +0100)]
GnuTLS: avoid using OCSP on buggy library versions. Bug 1664
Jeremy Harris [Sun, 2 Aug 2015 12:44:31 +0000 (13:44 +0100)]
Testsuite: Add testcase for OCSP-nonaware client, to supporting server. Bug 1664
The logfile here is for (I hope) the passing case, though the fixed GnuTLS library
is not yet available. Also due to the bug, client-gnutls is not usable for the
test; client-openssl must be used - meaning that a GnuTLS-only system cannot run
the testcase:
OCSP-GnuTLS/5650 OCSP stapling, server
** Command 15 ("client-ssl", starting at line 98)
** Return code 127 (expected 0)
Heiko Schlittermann (HS12-RIPE) [Thu, 30 Jul 2015 07:43:51 +0000 (09:43 +0200)]
Docs: Add a note about the maximum spam bar length
Heiko Schlittermann (HS12-RIPE) [Wed, 29 Jul 2015 21:55:09 +0000 (23:55 +0200)]
Docs: Clarify the "any output" phrase
Jeremy Harris [Sun, 26 Jul 2015 17:21:33 +0000 (18:21 +0100)]
Add ref from logging chapter to slow_lookup_log main config option
Jeremy Harris [Sun, 28 Jun 2015 14:14:02 +0000 (15:14 +0100)]
Compiler quietening
Jeremy Harris [Sat, 25 Jul 2015 12:50:43 +0000 (13:50 +0100)]
Docs: add detail on keys for lookups
Jeremy Harris [Sun, 26 Jul 2015 15:54:16 +0000 (16:54 +0100)]
Clean docs for next release
Heiko Schlittermann (HS12-RIPE) [Thu, 23 Jul 2015 21:20:37 +0000 (23:20 +0200)]
Doc: parallel builds (make -j) work
Jeremy Harris [Thu, 16 Jul 2015 15:25:53 +0000 (16:25 +0100)]
Docs: emphasize that the smtp_connection log selector applies to inbound
Jeremy Harris [Wed, 15 Jul 2015 22:45:39 +0000 (23:45 +0100)]
Add check on tls_auth pseudo-command. Bug 1659
Phil Pennock [Sun, 12 Jul 2015 23:28:30 +0000 (23:28 +0000)]
Compile with DISABLE_PRDR set
Andreas Metzler [Sat, 4 Jul 2015 16:07:21 +0000 (18:07 +0200)]
Multiple typo fixes.
Andreas Metzler [Thu, 2 Jul 2015 06:48:58 +0000 (08:48 +0200)]
Bump LOCAL_SCAN_ABI_VERSION.
In 4.86 the size of struct recipient_item changed when EXPERIMENTAL_DSN
was made default. This broke the local scan ABI (rebuild required).
See <http://bugs.debian.org/790616>.
Jeremy Harris [Sat, 27 Jun 2015 15:01:28 +0000 (16:01 +0100)]
Change note
Jeremy Harris [Thu, 25 Jun 2015 16:17:30 +0000 (17:17 +0100)]
Docs: add note on string lists
Jeremy Harris [Sun, 21 Jun 2015 13:36:01 +0000 (14:36 +0100)]
Fix error message for router headers_remove expansion failure
Associated with Bug 1533
Heiko Schlittermann (HS12-RIPE) [Mon, 22 Jun 2015 21:10:13 +0000 (23:10 +0200)]
Doc: Fix typo
Heiko Schlittermann (HS12) [Mon, 22 Jun 2015 20:02:30 +0000 (22:02 +0200)]
Doc: Update dns_trust_aa documentation
Jeremy Harris [Mon, 22 Jun 2015 13:17:07 +0000 (14:17 +0100)]
Docs: mark up with changebars
Jeremy Harris [Mon, 22 Jun 2015 12:21:04 +0000 (13:21 +0100)]
Fix support of $spam_ variables at delivery time. Bug 1647
This change is forced on us by the documentation claiming clearly
the support is there, though the code does not and never has.
The doc change that introduced the claim is
7d9f747b5ef8
Jeremy Harris [Mon, 22 Jun 2015 12:55:12 +0000 (14:55 +0200)]
Testsuite: fix operator precedence in dns_extract_auth_name()
Heiko Schlittermann (HS12) [Mon, 22 Jun 2015 09:44:36 +0000 (11:44 +0200)]
Testsuite: Add a first test for dns_trust_aa
Heiko Schlittermann (HS12) [Mon, 22 Jun 2015 07:57:02 +0000 (09:57 +0200)]
Testsuite: fakens may return AUTHORITY records
If an entry in db.<zone> is prefixed with "AA ", fakens
will put a valid NS record into the AUTHORITY section of the
returned packet. This will be used by dns_trust_aa checks.
Jeremy Harris [Mon, 22 Jun 2015 09:32:01 +0000 (10:32 +0100)]
Before importing a certificate, free any previous one. Bug 1648
Second try
Heiko Schlittermann (HS12) [Sun, 21 Jun 2015 15:06:37 +0000 (17:06 +0200)]
Extract NS/SOA in dns_extract_auth_name() more precisly.
Heiko Schlittermann (HS12) [Sun, 21 Jun 2015 15:03:50 +0000 (17:03 +0200)]
Fix and extend the checks in dns_is_secure()
Jeremy Harris [Sun, 21 Jun 2015 18:07:47 +0000 (19:07 +0100)]
Change note
Jeremy Harris [Sun, 21 Jun 2015 17:17:09 +0000 (18:17 +0100)]
Before importing a certificate, free any previous one. Bug 1648
Because the SSL libraries do not use Exim's heap management
this was a memory-leak in "exim -bp".
Jeremy Harris [Sun, 21 Jun 2015 13:26:16 +0000 (14:26 +0100)]
Docs: add warning on list-sep in headerss_remove
Jeremy Harris [Sat, 20 Jun 2015 16:46:42 +0000 (17:46 +0100)]
Docs typo
Gedalya [Sat, 20 Jun 2015 13:33:14 +0000 (14:33 +0100)]
Fix build script. Bug 1646
Jeremy Harris [Sat, 20 Jun 2015 14:20:54 +0000 (15:20 +0100)]
Add docs and massage coding standards for dns_trust_aa
Heiko Schlittermann (HS12) [Fri, 19 Jun 2015 22:45:00 +0000 (00:45 +0200)]
Add dns_trust_aa
This new global option allows to trust the AA bit for
specific domains the same way we'd trust the AD bit.
Jeremy Harris [Tue, 16 Jun 2015 18:56:28 +0000 (19:56 +0100)]
Docs: clarify notes on .ifdef Bug 1155
Jeremy Harris [Mon, 15 Jun 2015 16:43:43 +0000 (17:43 +0100)]
Docs: clarify notes on Events
Wolfgang Breyha [Mon, 15 Jun 2015 13:52:36 +0000 (14:52 +0100)]
DSN: fix null deref when bounce is due to conn-timeout. Bug 1630
Wolfgang Breyha [Mon, 15 Jun 2015 14:43:43 +0000 (15:43 +0100)]
Clarify that preceding 10ca4f was provided by Wolfgang Breyha
Massaged by JH
Jeremy Harris [Tue, 26 May 2015 15:36:08 +0000 (16:36 +0100)]
Add tls_eccurve main config option. Bug 1397
Patch from Suse, massaged by JH
Jeremy Harris [Fri, 12 Jun 2015 16:19:09 +0000 (17:19 +0100)]
minor tidying
Phil Pennock [Sat, 13 Jun 2015 01:07:05 +0000 (01:07 +0000)]
Doc fix: server_secret expansions should fail
The `cyrusless_sasl` authenticator example failed to explicitly fail if
no result was found from the lookup. Using `server_secret`, we should
_always_ fail instead of expanding to an empty string.
Doc-fix only.
Jeremy Harris [Wed, 10 Jun 2015 20:33:06 +0000 (21:33 +0100)]
Testsuite: additional EC encryptions seen
Jeremy Harris [Wed, 10 Jun 2015 19:37:33 +0000 (20:37 +0100)]
Testsuite: Increase test delays and retry rule times
to allow slow hosts more reliable testing
Jeremy Harris [Tue, 9 Jun 2015 22:00:39 +0000 (23:00 +0100)]
Testsuite: less agressive PID-hiding
Jeremy Harris [Tue, 9 Jun 2015 21:08:49 +0000 (22:08 +0100)]
Testsuite: quietening
Heiko Schlittermann (HS12) [Tue, 9 Jun 2015 20:14:26 +0000 (22:14 +0200)]
Revert "Show the DNSSEC status (ad=) always in -bt/-bv output"
This reverts commit
e7a1b6ff65f1bebbc290f2a4fd7554fde00ae2f6.
It's not production grade, since the wording (ad vs. trusted)
is not final yet.
Jeremy Harris [Tue, 9 Jun 2015 15:46:12 +0000 (16:46 +0100)]
Testsuite: avoid IPv6 to avoid "no route to host" log lines
Jeremy Harris [Tue, 9 Jun 2015 12:02:18 +0000 (13:02 +0100)]
Tighter guard for POLLRDHUP
Jeremy Harris [Mon, 8 Jun 2015 20:48:50 +0000 (21:48 +0100)]
Truncate delay when peer closes connection. Bug 348
This is now possible on Linux, at least.
Jeremy Harris [Sun, 7 Jun 2015 21:07:24 +0000 (22:07 +0100)]
Testsuite: avoid IPv6 interfaces to avoid extra debug stderr lines
Andreas Metzler [Sun, 7 Jun 2015 14:16:35 +0000 (15:16 +0100)]
Content scan: Use ETIMEDOUT not ETIME, as having better portability. Bug 1640
Jeremy Harris [Sat, 6 Jun 2015 20:59:05 +0000 (21:59 +0100)]
PRDR: enable server-side in the default config
Jeremy Harris [Sat, 6 Jun 2015 20:43:29 +0000 (21:43 +0100)]
Logging: add log_selector items in the default config. Bug 1333
Jeremy Harris [Sat, 6 Jun 2015 19:53:21 +0000 (20:53 +0100)]
Doc: Add DKIM info in main sections. Bug 1607
Wolfgang Breyha [Sat, 6 Jun 2015 19:07:04 +0000 (20:07 +0100)]
DSN: fix null deref when bounce is due to conn-timeout. Bug 1630
Jeremy Harris [Sat, 6 Jun 2015 18:35:16 +0000 (19:35 +0100)]
Spamd: add missing initialiser. Rspamd mode was incorrectly sometimes seen.
Reported-by: Andreas Metzler
Jeremy Harris [Fri, 5 Jun 2015 14:30:33 +0000 (15:30 +0100)]
Guard routing against a null-deref. Bug 1639
Phil Pennock [Fri, 5 Jun 2015 04:44:20 +0000 (00:44 -0400)]
release tooling: unbreak website build when not verbose
Phil Pennock [Fri, 5 Jun 2015 03:31:50 +0000 (23:31 -0400)]
release tooling: let make cmd be overriden
Also let tar flag actually take an argument
Phil Pennock [Fri, 5 Jun 2015 02:43:13 +0000 (22:43 -0400)]
Copyright year updates (things touched in 2015)
Update current year in docs and banner copyright in src/src/globals.c
Rest of changes from:
vi $(git whatchanged --since=2015-01-01 | grep '^:100' | sed -n 's/^[^M]*M//p' | sort -u | fgrep -v test/)
Note that there are a lot of changes made because of const propagation;
I opted to include the copyright year updates in that, but we could be
doing a better job with who gets the copyright credit for these changes.
Changes visible with:
git diff $(git rev-list -n1 --before="2015-01-01" master)
Jeremy Harris [Thu, 4 Jun 2015 19:28:25 +0000 (20:28 +0100)]
TLS authenticator
Jeremy Harris [Sun, 31 May 2015 22:04:01 +0000 (23:04 +0100)]
refactor build script
Phil Pennock [Fri, 29 May 2015 19:52:50 +0000 (15:52 -0400)]
Adjust my maintainership status to reflect reality
Phil Pennock [Fri, 29 May 2015 19:46:47 +0000 (15:46 -0400)]
OpenSSL: guard X509_check_host against LibreSSL
LibreSSL's fork does not have this new function; as well as adding a
`LIBRESSL_VERSION_NUMBER` value, that project bumped the OpenSSL version
number in such a way as to conflict with our existing version checks.
* Add a guard.
* Add commentary, suggesting how to avoid getting into twistier knots
with API divergence.
Reported by Jasper Wallace, who provided a slightly different patch.
Fixes bug 1635
Heiko Schlittermann (HS12) [Wed, 27 May 2015 21:41:35 +0000 (23:41 +0200)]
Testsuite: Add $USER to env if missing
Andreas Metzler [Wed, 27 May 2015 12:05:03 +0000 (13:05 +0100)]
Expand docs re. logs dir, and make eximon logs dir match exim's. Bug 1324
Jeremy Harris [Wed, 27 May 2015 11:41:08 +0000 (12:41 +0100)]
Note MAIL commands in -bS batch, to avoid smtp_no_mail logline. Bug 1346
Heiko Schlittermann (HS12) [Tue, 26 May 2015 20:44:23 +0000 (22:44 +0200)]
Fix some typos in EDITME
Phil Pennock [Tue, 26 May 2015 09:48:46 +0000 (10:48 +0100)]
TLS: Enable ECDHE on OpenSSL, just the NIST P-256 curve. Bug 1397
Original by Phil Pennock; tweaked by JH.
Jeremy Harris [Sat, 23 May 2015 20:48:26 +0000 (21:48 +0100)]
New ${env {NAME}} expansion. Bug 1604
Jeremy Harris [Sat, 23 May 2015 17:07:58 +0000 (18:07 +0100)]
Testsuite: move test.again.dns and test.fail.dns handling to fakens
Jeremy Harris [Sat, 23 May 2015 16:45:48 +0000 (17:45 +0100)]
tidying
Jeremy Harris [Fri, 22 May 2015 17:32:04 +0000 (18:32 +0100)]
DANE: do not fail/defer message due to TLSA lookup but dane is only requested
Jeremy Harris [Thu, 21 May 2015 22:22:16 +0000 (23:22 +0100)]
Fix DANE for multiple-MX when all TLSA lookup defer. Bug 1634
Heiko Schlittermann (HS12) [Wed, 20 May 2015 21:08:21 +0000 (23:08 +0200)]
Testsuite: Check debug message if we requested AD but got AA
Heiko Schlittermann (HS12) [Wed, 20 May 2015 21:07:33 +0000 (23:07 +0200)]
Testsuite: Add support for authoritive answer to fakens
Heiko Schlittermann (HS12) [Wed, 13 May 2015 21:50:23 +0000 (23:50 +0200)]
Add DNS debug aid if we requested AD but got AA
If the resolver we ask is authoritive (AA) for some domain,
we never ever get the AD (authentic data) bit in the answer.
Heiko Schlittermann (HS12) [Wed, 13 May 2015 21:50:23 +0000 (23:50 +0200)]
Add DNS debug aid if we requsted AD but got AA
If the resolver we ask is authoritive (AA) for some domain,
we never ever get the AD (authentic data) bit in the answer.