]>
git.ipfire.org Git - thirdparty/dovecot/core.git/log
Timo Sirainen [Wed, 7 Jun 2017 21:35:08 +0000 (00:35 +0300)]
lib-master: Use master_service_import_environment() for preserving environments
Timo Sirainen [Wed, 7 Jun 2017 21:33:32 +0000 (00:33 +0300)]
lib-master: master_service_import_environment() - use a data stack frame
This way the caller doesn't have to do it.
Timo Sirainen [Wed, 7 Jun 2017 21:24:19 +0000 (00:24 +0300)]
master: Move master_set_import_environment() to lib-master
Timo Sirainen [Wed, 7 Jun 2017 21:20:38 +0000 (00:20 +0300)]
master: Append to existing DOVECOT_PRESERVE_ENVS instead of overwriting it
Timo Sirainen [Wed, 7 Jun 2017 21:14:30 +0000 (00:14 +0300)]
master: Move import_environment setting to lib-master's master_service_settings
Timo Sirainen [Sat, 4 Feb 2017 18:45:30 +0000 (20:45 +0200)]
lib-http: Try to reuse memory for host->ips allocation.
If the host already had an IP, most of the time ips_count doesn't change
anymore.
Timo Sirainen [Sat, 10 Jun 2017 07:35:54 +0000 (10:35 +0300)]
lib: Fix test-path-util unit test with some cwd path lengths
The component-component directory might not have been created at all,
in which case the following symlink creations would fail since they were
already created to the same directory in a previous test.
Timo Sirainen [Fri, 9 Jun 2017 15:10:14 +0000 (18:10 +0300)]
mail-crypt-acl: Fix assert-crash due to missing namespace initialization
This was broken by
e031d9aaae59a9f79710dc1138b76b69272615a3 .
Timo Sirainen [Fri, 9 Jun 2017 13:18:56 +0000 (16:18 +0300)]
quota: count backend - Don't ignore mailbox list iteration errors
Aki Tuomi [Fri, 9 Jun 2017 11:08:22 +0000 (14:08 +0300)]
lib: Fix test-path-util.c formatting to conform coding style
Aki Tuomi [Fri, 9 Jun 2017 10:15:35 +0000 (13:15 +0300)]
lib: Fix infinite loop in path-util.c
Infinite loop was reached if the path was longer than
128 bytes, and the 128 boundary was in middle of path
name component.
Timo Sirainen [Fri, 9 Jun 2017 09:07:00 +0000 (12:07 +0300)]
quota: Fix negative quota_warnings with count backend
For example this never triggered:
quota_warning = -messages=100%% quota-warning %u -100
The change to quota_alloc() to update the count_used doesn't seem to
actually fix anything right now, but it makes the code more correct.
Timo Sirainen [Fri, 9 Jun 2017 09:42:51 +0000 (12:42 +0300)]
lib-storage: Fix sort index writing to not leave sort_id=0 gaps
This fixes errors like:
Error: INBOX: Broken sort-f indexes, resetting
Timo Sirainen [Fri, 9 Jun 2017 09:25:36 +0000 (12:25 +0300)]
lib-storage: Add details to "Broken sort-* indexes" error
Aki Tuomi [Fri, 9 Jun 2017 12:22:56 +0000 (15:22 +0300)]
dsync: Display namespace locations on init in debug mode
This simplifies debugging as you can more easily tell
which brain is using which mail location.
Timo Sirainen [Fri, 9 Jun 2017 11:31:15 +0000 (14:31 +0300)]
lib-storage: Fix setting the correct cache record corrupted
It was mixing UIDs and sequences, so a wrong mail could have been set
corrupted or it could have crashed with:
Panic: file mail-index-transaction-update.c: line 1018 (mail_index_update_ext): assertion failed: (seq > 0 && (seq <= mail_index_view_get_messages_count(t->view) || seq <= t->last_new_seq))
Timo Sirainen [Sun, 4 Jun 2017 10:07:48 +0000 (13:07 +0300)]
quota: Remove unused quota_free()
Timo Sirainen [Sun, 4 Jun 2017 10:06:06 +0000 (13:06 +0300)]
quota: Fix quota_vsizes=yes to work with all backends
Josef 'Jeff' Sipek [Wed, 7 Jun 2017 11:37:47 +0000 (14:37 +0300)]
lib-storage: simplify mailbox_get_path_to()
Instead of special casing MAILBOX_LIST_PATH_TYPE_{MAILBOX,INDEX}, we can use
a common helper to remove code duplication.
Timo Sirainen [Tue, 30 May 2017 21:07:46 +0000 (00:07 +0300)]
imapc: Don't send UID FETCH 1:* after SELECT if mailbox is empty
Timo Sirainen [Wed, 7 Jun 2017 16:54:10 +0000 (19:54 +0300)]
lib-storage: test-mail-search-args-simplify - test with initialized args
Timo Sirainen [Wed, 7 Jun 2017 16:53:00 +0000 (19:53 +0300)]
lib-storage: mail_search_args_simplify() - Fix merging already-initialized keywords
It was deinitializing the arg that was kept instead of the one being removed.
Timo Sirainen [Wed, 7 Jun 2017 16:37:46 +0000 (19:37 +0300)]
lib-storage: Fix mail_search_args_simplify() to not deinit too many args
Removing an arg should deinit it, but not its following siblings.
Timo Sirainen [Wed, 7 Jun 2017 16:36:19 +0000 (19:36 +0300)]
lib-storage: Add mail_search_arg_one_deinit()
Timo Sirainen [Wed, 7 Jun 2017 20:47:56 +0000 (23:47 +0300)]
master: Don't include imap-hibernate when counting auth's max client_limit
imap-hibernate doesn't do any auth connections, so it doesn't need it.
It's a bit ugly to add an explicit service name comparison here, but
there didn't really seem to be many other good choices:
* Could have removed protocol=imap from imap-hibernate. I was close to
doing this, but maybe it's better that imap-only processes would have
their protocol set to "imap".
* Could have added a new service type, but seems pretty unnecessary to add
some kind of a "nonauth" just for this. A reverse "needs-auth" might have
been ok though.
Perhaps in future we'll implement proper dependencies across services and
those dependencies could be used to calculate this limit more precisely.
Timo Sirainen [Wed, 7 Jun 2017 07:17:55 +0000 (10:17 +0300)]
lib: Add unit tests for p_strdup*()
Timo Sirainen [Tue, 30 May 2017 15:25:50 +0000 (18:25 +0300)]
lib-storage: Fix crash in mail_get_header_stream() when its previous stream wasn't at EOF
At least this could have happened when indexes were disabled and running:
FETCH 1 (envelope body.peek[header.fields (foo)] bodystructure)
Fixes:
Panic: file index-mail-headers.c: line 198 (index_mail_parse_header_init): assertion failed: (!mail->data.header_parser_initialized)
Timo Sirainen [Wed, 7 Jun 2017 08:35:18 +0000 (11:35 +0300)]
global: Use i_realloc_type() wherever possible
Timo Sirainen [Wed, 7 Jun 2017 08:24:38 +0000 (11:24 +0300)]
lib: Add test-mempool unit test.
It verifies that the p_new() and p_realloc_type() handle overflows
correctly. test-malloc-overflow already does this for the low-level
MALLOC_*() macros, but here we're cheking that the actually used
memory allocation macros also work.
Timo Sirainen [Wed, 7 Jun 2017 08:24:03 +0000 (11:24 +0300)]
lib: Add i_realloc_type() for i_realloc() that checks for overflows
Timo Sirainen [Thu, 8 Jun 2017 08:35:58 +0000 (11:35 +0300)]
imapc: Fix potential crash if initial sync fails.
Happened at least if mailbox had 0 mails and a failing mailbox_sync() was
called twice.
Fixes:
Panic: file imapc-sync.c: line 477 (imapc_sync_index): assertion failed: (mbox->sync_fetch_first_uid == 1)
Timo Sirainen [Wed, 7 Jun 2017 15:10:10 +0000 (18:10 +0300)]
lib-mail: message_address_parse() - Fix fill_missing==TRUE handling
Mainly MISSING_DOMAIN wasn't set in all situations. Also added unit tests.
Timo Sirainen [Wed, 7 Jun 2017 12:33:42 +0000 (15:33 +0300)]
lib-mail: message_address_parse() - Handle invalid standalone phrases better
We'll treat "local-part" (without quotes) as a mailbox even without
@domain, but if it continues with anything or if it's a quoted-string,
we'll treat it as a display-name instead.
This is probably better than just converting everything to display-name,
since there are likely to exist headers like "To: localuser"
Timo Sirainen [Wed, 7 Jun 2017 12:22:22 +0000 (15:22 +0300)]
lib-mail: Improve test-message-address to test invalid addresses in lists
Test in both a group and list of addresses.
Timo Sirainen [Wed, 7 Jun 2017 12:21:10 +0000 (15:21 +0300)]
lib-mail: message_address_parse() - don't stop at <> when parsing address lists
And similarly don't stop at <@domain>
Timo Sirainen [Wed, 7 Jun 2017 12:08:27 +0000 (15:08 +0300)]
lib-mail: message_address_parse() - don't stop at invalid addresses when parsing group
Also the end-group's message_address shouldn't have invalid_syntax=TRUE,
except when we didn't actually find the ending ';'
Timo Sirainen [Wed, 7 Jun 2017 11:57:12 +0000 (14:57 +0300)]
lib-mail: message_address_parse() - Fix reading out-of-bounds read on invalid input
Timo Sirainen [Wed, 7 Jun 2017 10:34:39 +0000 (13:34 +0300)]
lib-mail: message_address_parse() - Remove unnecessary NULL checks
message_address_parse() always initialized rfc822_parser_init() with
last_comment string.
Timo Sirainen [Wed, 7 Jun 2017 10:19:40 +0000 (13:19 +0300)]
lib-mail: test-message-address - add more and better organized tests
Timo Sirainen [Wed, 7 Jun 2017 09:58:21 +0000 (12:58 +0300)]
lib-mail: message_address_parse() - Add INVALID_ROUTE/SYNTAX_ERROR only if fill_missing==TRUE
This allows message_address_write() to write the parsed invalid address in a
way that at least somewhat resembles the original input.
Timo Sirainen [Wed, 7 Jun 2017 09:42:28 +0000 (12:42 +0300)]
lib-mail: message_address_write() - Write empy username as ""
We can't output <@domain> because that conflicts with the <@route:...>
syntax. Also ""@domain seems to be a valid address.
Timo Sirainen [Wed, 7 Jun 2017 09:37:09 +0000 (12:37 +0300)]
lib-mail: message_address_write() - Always write <user@domain> as output
This is the standard format nowadays, so don't convert <user@domain> into
plain user@domain.
Timo Sirainen [Wed, 7 Jun 2017 09:35:35 +0000 (12:35 +0300)]
lib-mail: message_address_write() - Don't write empty <>
Timo Sirainen [Wed, 7 Jun 2017 09:34:17 +0000 (12:34 +0300)]
lib-mail: message_address_write() - Don't add '@' if domain is missing
Timo Sirainen [Wed, 7 Jun 2017 09:33:10 +0000 (12:33 +0300)]
lib-mail: test-message-address - Test also writing invalid addresses
Timo Sirainen [Wed, 7 Jun 2017 09:01:07 +0000 (12:01 +0300)]
lib-mail: test-message-address cleanup
Merge input and output into the same struct to make it clearer.
Josef 'Jeff' Sipek [Mon, 5 Jun 2017 11:21:23 +0000 (14:21 +0300)]
lib: test for guid_128_generate() time handling
Make sure that guids always increase regardless of what is happening to
the time.
Josef 'Jeff' Sipek [Mon, 5 Jun 2017 09:55:56 +0000 (12:55 +0300)]
lib: guid_128_generate() shouldn't generate invalid timestamps
The nsec validity check must take into account the soon-to-be-done increment.
Josef 'Jeff' Sipek [Mon, 5 Jun 2017 09:53:23 +0000 (12:53 +0300)]
lib: guid_128_generate() needs to correctly convert usecs to nsecs
The timestamp in the guid starts off with real time, but "runs ahead" if a lot
of guids are generated. If not many guids are being generated, the timestamp
is fast-forwarded to the current ioloop_timeval. The buggy comparison prevented
fast forwarding when tv_sec matched, but tv_nsec value was > 0 (which
was true most of the time).
Sergey Kitov [Wed, 31 May 2017 11:32:30 +0000 (14:32 +0300)]
quota: assertion for *args != NULL is added in order to avoid warning from static analyzer and *args[0] changed to (*args)[0]
Aki Tuomi [Tue, 6 Jun 2017 10:59:14 +0000 (13:59 +0300)]
dict-cdb: Add simple iteration support
Timo Sirainen [Tue, 6 Jun 2017 14:18:27 +0000 (17:18 +0300)]
lib-index: Fix reset_id handling in mail_cache_get_missing_reason() cache
Broken by
bd897f82a5d9b00ba256b462d2056c2dc7df257c
Timo Sirainen [Fri, 7 Apr 2017 06:19:02 +0000 (09:19 +0300)]
lib: test-utc-mktime: Remove duplicate test
Also fixes a compiler warning when time_t is signed 32bit.
Timo Sirainen [Tue, 6 Jun 2017 06:39:34 +0000 (09:39 +0300)]
lib-storage: Fix assert-crash in SORT caused by earlier changes
Broken by
0a6a527f0c42b5478d80ac53ab357885676fd516
Fixes:
Panic: file index-sort-string.c: line 668 (index_sort_add_ids_range): assertion failed: (str != NULL)
Aki Tuomi [Tue, 6 Jun 2017 09:47:53 +0000 (12:47 +0300)]
lib-oauth2: Make sure fields are always initialized
Aki Tuomi [Tue, 6 Jun 2017 09:44:55 +0000 (12:44 +0300)]
lib-oauth2: Accept empty responses
Timo Sirainen [Mon, 5 Jun 2017 18:53:18 +0000 (21:53 +0300)]
lib-index: Add cache to mail_cache_get_missing_reason()
This avoids excessive CPU usage when it's called in a loop for many mails.
Timo Sirainen [Mon, 5 Jun 2017 18:44:45 +0000 (21:44 +0300)]
lib-index: mail_cache_get_missing_reason() - Minor code cleanup
Timo Sirainen [Mon, 5 Jun 2017 18:09:51 +0000 (21:09 +0300)]
lib-storage: Optimize error handling for SORT
If a lot of mails have failed in a large mailbox, this check reduces the
CPU usage significantly.
Timo Sirainen [Mon, 5 Jun 2017 18:07:32 +0000 (21:07 +0300)]
lib-storage: Sort code cleanup - Change index_sort_get_string() to return bool
None of the callers cared about the difference between "expunged" vs
"error".
Timo Sirainen [Mon, 5 Jun 2017 18:05:07 +0000 (21:05 +0300)]
lib-storage: Sort code cleanup - pass node to index_sort_get_string()
Timo Sirainen [Mon, 5 Jun 2017 17:14:54 +0000 (20:14 +0300)]
lib-storage: Make sure istream-mail doesn't replace an existing storage error
I'm not aware of it actually having broken anything, so this is just for
extra safety.
Timo Sirainen [Mon, 5 Jun 2017 17:12:43 +0000 (20:12 +0300)]
lib-storage: Fix SORT error message when reaching mail_sort_max_read_count
It should have been:
NO [LIMIT] Requested sort would have taken too long
Instead it was just:
NO [SERVERBUG] Mail field not cached
It was broken for CC, FROM, TO, SUBJECT, DISPLAYFROM and DISPLAYTO.
The numeric sorting was working correctly.
Timo Sirainen [Mon, 5 Jun 2017 16:05:58 +0000 (19:05 +0300)]
push-notification: Fix crash in OX driver's deinit
The http_client doesn't always exist.
Aki Tuomi [Mon, 5 Jun 2017 09:14:57 +0000 (12:14 +0300)]
auth: Access always first entry when flushing failures
The code is deleting the first item after accessing
it, and then moving forward. It will eventually go
beyond the array and get NULL ptr and fail.
Instead we need to always get the first item,
since the array deletion is moving the queued items
forward.
Broken by
e18b4e41
Timo Sirainen [Tue, 30 May 2017 13:23:28 +0000 (16:23 +0300)]
lib-auth: Code cleanup - remove unnecessary code
Besides being unnecessary, it was also obsolete since not all the strings
were p_strdup()ed.
Josef 'Jeff' Sipek [Tue, 23 May 2017 11:12:49 +0000 (14:12 +0300)]
pop3: send back auth reply sooner
If possible (i.e., pop3_lock_session=no), we can send back the OK
response earlier.
Josef 'Jeff' Sipek [Tue, 23 May 2017 10:01:48 +0000 (13:01 +0300)]
pop3: initialize namespaces explicitly
(instead of relying on mail user initialization to instantiate them for
us)
Josef 'Jeff' Sipek [Fri, 19 May 2017 07:58:18 +0000 (10:58 +0300)]
pop3: client_create_from_input shouldn't process input
This makes the pop3 code resemble the imap code more. Making the subsequent
changes easier.
Josef 'Jeff' Sipek [Fri, 19 May 2017 07:51:22 +0000 (10:51 +0300)]
pop3: client_create_from_input should return the client struct
This makes the code more similar to imap's, making subsequent changes
easier.
Josef 'Jeff' Sipek [Fri, 19 May 2017 07:46:05 +0000 (10:46 +0300)]
pop3: rename login_client_connected() client arg to login_client
This makes the next commit cleaner.
Josef 'Jeff' Sipek [Fri, 19 May 2017 07:42:03 +0000 (10:42 +0300)]
pop3: move pop3 session locking out of client_create
As a result, we can directly return the client structure (instead of
passing it back via a _r arg).
This makes the pop3 client_create look more like the imap version.
Timo Sirainen [Tue, 30 May 2017 07:37:24 +0000 (10:37 +0300)]
lib-imap-client: Fix test-imapc-client unit test to work on OSX
connect() to port 0 fails in it, so for the "connect failure" test first
open a random free listener port and then close it. Hopefully nothing else
reopens it in the mean time.
Timo Sirainen [Tue, 30 May 2017 07:28:43 +0000 (10:28 +0300)]
lib: test-path-util unit test - small fixes and cleanups
Don't use /tmp, especially because on OSX it's a symlink to /private/tmp/,
which causes these tests to fail.
Timo Sirainen [Wed, 31 May 2017 11:27:16 +0000 (14:27 +0300)]
director: Use longer timeout for receiving user list in handshake
The sender also has a send timeout, so this allows the sender to abort
early and give a nicer error message.
Timo Sirainen [Wed, 31 May 2017 10:42:31 +0000 (13:42 +0300)]
director: Log more connection state information on handshaking errors
Timo Sirainen [Wed, 31 May 2017 10:17:37 +0000 (13:17 +0300)]
director: Use more accurate timestamps for handshake timeout logging
There could be some time between the create, connect and when ME was
received.
Timo Sirainen [Wed, 31 May 2017 10:15:04 +0000 (13:15 +0300)]
director: Fix "sending handshake timed out" error to be actually logged
io was never NULL here, so it was never logged.
Timo Sirainen [Wed, 31 May 2017 10:11:08 +0000 (13:11 +0300)]
director: Fix handshake timeout lengths
They weren't too bad, but also they weren't what the #defines described.
Also added a separate connect() timeout.
Timo Sirainen [Tue, 30 May 2017 18:57:06 +0000 (21:57 +0300)]
quota: Fix imapc backend not to try to find out mail's size.
This removes unnecessary FETCH RFC822.SIZE command sent to server when
expunging mails.
Timo Sirainen [Tue, 30 May 2017 21:00:32 +0000 (00:00 +0300)]
imapc: SEARCH - Don't add parenthesis around OR
There's no need to write "(OR a b)" when "OR a b" works just the same.
Timo Sirainen [Tue, 30 May 2017 20:53:44 +0000 (23:53 +0300)]
lib-storage: mail_search_args_to_imap() - don't add parenthesis around OR
There's no need to write "(OR a b)" when "OR a b" works just the same.
Aki Tuomi [Wed, 31 May 2017 11:39:55 +0000 (14:39 +0300)]
lib: Use unix streams with connection when using unix socket
This makes it easier to send file descriptors over connection
Timo Sirainen [Tue, 30 May 2017 13:57:45 +0000 (16:57 +0300)]
auth: Don't lose all forward_ fields if the first passdb lookup fails.
Aki Tuomi [Mon, 29 May 2017 10:39:18 +0000 (13:39 +0300)]
lib-mail: Add qp encoder
Aki Tuomi [Thu, 25 May 2017 16:03:32 +0000 (19:03 +0300)]
lib-mail: Link test binaries to libmail.la instead of .lo files
Aki Tuomi [Tue, 30 May 2017 11:03:23 +0000 (14:03 +0300)]
mail-filter: tabescape args, this is required by script client 4
Aki Tuomi [Tue, 30 May 2017 11:01:39 +0000 (14:01 +0300)]
welcome: tabescape args, this is required by script client 4
Aki Tuomi [Tue, 30 May 2017 11:01:17 +0000 (14:01 +0300)]
global: Update script client version number
Broken by
b383ed51d75bce0f69f126bc4ff7192985ca30f2
Timo Sirainen [Tue, 30 May 2017 08:30:10 +0000 (11:30 +0300)]
auth: Fix crash on passdb lookup when all passdbs were skipped
Finishes the fix in
614f5b6febaf3c825f9200ab3b62d9d047197b0e
Fixes:
auth: Panic: file auth-request.c: line 2493 (get_log_prefix): assertion failed: (auth_request->mech != NULL)
Timo Sirainen [Tue, 30 May 2017 06:36:49 +0000 (09:36 +0300)]
lib-storage: Fix LAYOUT=fs potentially not ignoring ACLs when requested
fs_get_existence_info_flag() was only being called with LAYOUT=fs and
Maildir combination. This mainly affected that combination when using
mailbox_list_index=yes and trying to list mailboxes with ACL checking
disabled.
Timo Sirainen [Tue, 30 May 2017 06:18:14 +0000 (09:18 +0300)]
lib-storage: Fix ACL in parent mailbox potentially hiding its children.
fs_get_existence_info_flag() was only being called with LAYOUT=fs and
Maildir combination, so only it was broken if the parent mailbox wasn't
visible but its children were.
Timo Sirainen [Tue, 30 May 2017 06:15:09 +0000 (09:15 +0300)]
lib-storage: fs_list_get_mailbox_flags() - Return mailbox files as selectable
Avoids a second stat() later on, which would do the same anyway.
Timo Sirainen [Tue, 30 May 2017 06:12:05 +0000 (09:12 +0300)]
lib-storage: fs_list_get_mailbox_flags() - Don't assume directories always have children.
This makes the directory handling code equivalent for type==UNKNOWN and
type==DIR.
Timo Sirainen [Tue, 30 May 2017 06:10:55 +0000 (09:10 +0300)]
lib-storage: fs_list_get_mailbox_flags() - Add comments
Timo Sirainen [Tue, 30 May 2017 06:09:39 +0000 (09:09 +0300)]
lib-storage: fs_list_get_mailbox_flags() - Reorganize code
No functional changes.
Timo Sirainen [Mon, 29 May 2017 14:58:35 +0000 (17:58 +0300)]
acl: Global ACL file was parsed as if it was local ACL file
This caused some of the ACL handling not work exactly as expected.
Timo Sirainen [Mon, 29 May 2017 14:26:52 +0000 (17:26 +0300)]
lib-storage: Make mailbox_list_iter_init_autocreate() static again
It's no longer needed elsewhere.
Timo Sirainen [Mon, 29 May 2017 14:20:48 +0000 (17:20 +0300)]
lib-storage: Fix mailbox_list_index=yes + ACLs + auto=subscribe mailboxes
If the auto=subscribe mailbox didn't have a lookup ACL, it shouldn't have
been listed. This didn't work properly, because ACL plugin initialized the
autocreate_ctx only for the index iterator, while the autoboxes were
listed using the backend iterator. Fixed by not creating index iterator at
all when doing a passthrough iteration.
Timo Sirainen [Mon, 29 May 2017 13:44:44 +0000 (16:44 +0300)]
lib-storage: Cleanup - Don't use ctx in iter_use_index()
Shrinks the following change.