Commit Graph

1249 Commits

Author SHA1 Message Date
Oswald Buddenhagen
a8b26dc4ac soft-limit peak memory usage
propagating many messages from a fast store (typically maildir or a
local IMAP server) to a slow asynchronous store could cause gigabytes of
data being buffered. avoid this by throttling fetches if the target
context reports memory usage above a configurable limit.

REFMAIL: 9737edb14457c71af4ed156c1be0ae59@mpcjanssen.nl
2015-02-15 18:13:05 +01:00
Oswald Buddenhagen
4b31522fdf complain about excess values supplied to options 2015-02-15 12:48:46 +01:00
Oswald Buddenhagen
d4392c9220 handle clean SSL connection shutdowns
some servers actually bother to close down the SSL connection before
closing the socket.

this fixes the spurious "unhandled SSL error 6" messages.

REFMAIL: 20150120114805.GA17586@leeloo.kyriasis.com
2015-02-15 12:19:47 +01:00
Oswald Buddenhagen
9e15ab4a5a refactor socket EOF handling
handling EOF already at the socket level isn't a very good idea - it
breaks the abstraction, and makes implementing sane semantics hard.
2015-02-15 12:15:46 +01:00
Oswald Buddenhagen
13c742529c fix crash on shutdown of compressed connection
the callback may destroy the socket, so it is very unwise to use it for
buffering the return value. use a temporary instead.
2015-02-15 11:49:58 +01:00
Oswald Buddenhagen
c9b8cefc29 handle clean shutdown of zlib stream
the server can actually close the zlib stream before closing the socket,
so we need to accept it.

we don't do anything beyond that - the actual EOF will be signaled by
the socket, and if the server (erroneously) sends more data, zlib will
tell us about it.

REFMAIL: 1423048708-975-1-git-send-email-alex.bennee@linaro.org
2015-02-15 11:49:58 +01:00
Oswald Buddenhagen
ef70bd4a40 don't try to flush if there is nothing to flush
zlib reports Z_BUF_ERROR when a flush is attempted without any activity
since the previous flush (if any). while this is harmless as such,
discerning the condition from genuine errors would be much harder than
avoiding the pointless flush in the first place.

REFMAIL: eb5681612f17be777bc8d138d31dd6d6@mpcjanssen.nl
2015-02-15 11:49:58 +01:00
Oswald Buddenhagen
f4240761f1 introduce and use pending_wakeup()
so we don't need to peek into internal data structures.
2015-02-15 11:49:52 +01:00
Oswald Buddenhagen
74c78c70b9 deal sensibly with permanent errors
don't retry dead Stores for every Channel.

this also introduces a state for transient errors (specifically, connect
failures), but this is currently unused.
2015-01-17 17:51:20 +01:00
Oswald Buddenhagen
5f265ad7da unify .isyncuidmap.db handling with that of .uidvalidity
that is, open the database on demand when locking and close it again
when unlocking.
2015-01-17 17:51:20 +01:00
Oswald Buddenhagen
4da89af7be refactor maildir_set_uid() and maildir_store_uidval()
the latter now handles both the db and the plaintext file, so the former
can make use of it.
2015-01-17 17:51:20 +01:00
Oswald Buddenhagen
f61efdbb9d rename some *uid*() => *uidval*() to better reflect their function 2015-01-17 17:51:20 +01:00
Oswald Buddenhagen
2eece82276 lock .uidvalidity on demand
a maildir re-scan doesn't need to lock it if it doesn't need to allocate
any new uids.
2015-01-17 17:51:20 +01:00
Oswald Buddenhagen
4aad8c9e04 delay unlocking of .uidvalidity
it's wasteful to re-lock all the time. instead, unlock in the background
after 1-2 seconds of inactivity.
2015-01-17 17:51:20 +01:00
Oswald Buddenhagen
d9a983add6 add support for propagating folder deletions 2015-01-17 17:51:20 +01:00
Oswald Buddenhagen
a7eddc6ede don't make intermediate directories proper maildirs
"phantom" mailboxes waste time on syncing. furthermore, mutt's mailbox
navigator provides no means to enter subfolders of maildirs.
2015-01-17 17:51:20 +01:00
Oswald Buddenhagen
7489ff8613 deal sensibly with incomplete maildir directories
a directory is no mailbox unless it contains a cur/ subdir.
but if that one is present, create new/ and tmp/ if they are missing.

this makes it possible to resume interrupted maildir creations.
2015-01-17 17:51:19 +01:00
Oswald Buddenhagen
5f4e3b285e factor out maildir_clear_tmp() 2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
926788f3ae supplement open_box() with box existence information from list_store()
there is no point in trying to open a non-existing box before trying to
create it.
2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
7b7304b625 split create_box() off from open_box()
this allows us to do something else than creating missing boxes
depending on circumstances. hypothetically, that is.
2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
f1809ddd2b open the mailboxes after loading the sync state
this allows us to react differently to a box'es absence depending on the
state. hypothetically, so far.
2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
f43617cd94 lock sync state lazily
don't try to lock it until we actually read or write it.
the idea is to not fail with SyncState * if we tried to load the state
before selecting a non-existing mailbox. this is ok, because if the
mailbox is missing, we obviously have no sync state pertaining to it,
either.
as a side effect, this allows simplifying an error path.
2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
fb19d644f7 split off open_box() from select_box()
aka prepare_paths() reloaded. we'll need it in a moment.
2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
97a42cd825 factor out {prepare,lock,save,load}_state() 2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
5af1796777 de-duplicate handling of box operation (create & expunge) options
loops work just fine ...
2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
9982e7bf08 make some driver function names more descriptive 2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
00ebf45be2 rename driver::prepare_opts() => prepare_load()
... and move it to the right place in the structure and fix the doc to
not claim that it is called before select().
2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
bac2b00f1b disable use of LITERAL+ for payloads > 100k
when LITERAL+ is used, the server has no chance for early rejection of
messages. this means that the client can upload megabytes for nothing.
so simply don't use LITERAL+ for big messages. of course this adds
server roundtrips, but that's tough luck.

the limit could be arguably higher than 100k (or even configurable).
i set it to ~2 sec with my fairly average DSL line.
2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
6c959c3ee4 fix handling of unsolicited BYE responses
they can come in at any time, after which we must expect the connection
to be closed (and not complain about it).
2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
4f3ef54f3a fix treatment of untagged NO and BAD responses
they aren't possible greeting responses. however, they are warning resp.
error reports from the server, so print them accordingly.
2015-01-11 15:05:29 +01:00
Oswald Buddenhagen
139b90be29 added support for IMAP DEFLATE
initial patch by Jesse Weaver <pianohacker@gmail.com>, but mostly
rewritten by me.
2015-01-11 15:05:28 +01:00
Oswald Buddenhagen
f0b80e7d35 make socket writing buffered
the primary objective is reducing the number of small SSL packets (which
are always padded), but fewer syscalls in the non-SSL case should be
good as well.
2015-01-11 15:05:28 +01:00
Oswald Buddenhagen
efb23ab96a vectorize socket_write()
the objective is making the buffer code aware of the total size of a
write in advance. this commit doesn't take advantage of that yet.
2015-01-11 15:05:28 +01:00
Oswald Buddenhagen
3f629af07e remove support for faking notifications
with the existence of timers, this is now superfluous.
2015-01-11 15:05:28 +01:00
Oswald Buddenhagen
a4e2f1a60d use null timer instead of faking a socket notification 2015-01-11 15:05:28 +01:00
Oswald Buddenhagen
7b76d9ff7e add timers to mainloop
they are called "wakeups", though, as timer_t already exists in time.h.
2015-01-11 15:05:28 +01:00
Oswald Buddenhagen
1fd66195d8 change socket notifier design
instead of keeping the structures in an opaque array (which was a shadow
of the struct pollfd array if poll() was supported), make them directly
addressable.

this has the advantage that notifier-altering operations (mostly
en-/disabling) don't need to look up the structure by file handle each
time.
on the downside, data locality in the main loop is worse.
neither of these have any real effect on performance.

note that the structures are not allocated separately, but embedded into
the the parent structure (like sockets already were).
2015-01-11 15:05:28 +01:00
Oswald Buddenhagen
f68e021b90 factor out socket_open_internal() 2015-01-11 15:05:28 +01:00
Oswald Buddenhagen
42cedc8f81 introduce uchar, ushort & uint typedefs 2015-01-11 15:05:28 +01:00
Oswald Buddenhagen
b730f66f7d Merge branch 'isync_1_1_branch' into HEAD
Conflicts:
	src/socket.c
2015-01-11 14:32:15 +01:00
Oswald Buddenhagen
9a4be0af5f skip merges during ChangeLog generation 2015-01-11 14:32:05 +01:00
Oswald Buddenhagen
2fa75cf159 fix UID assignment with some non-UIDPLUS servers
the seznam.cz IMAP server seems very eager to send UIDNEXT responses
despite not supporting UIDPLUS. this doesn't appear to be a particularly
sensible combination, but it's valid nonetheless.

however, that means that we need to save the UIDNEXT value before we
start storing messages, lest imap_find_new_msgs() will simply overlook
them. we do that outside the driver, in an already present field - this
actually makes the main path more consistent with the journal recovery
path.

analysis by Tomas Tintera <trosos@seznam.cz>.

REFMAIL: 20141220215032.GA10115@kyvadlo.trosos.seznam.cz
2015-01-11 14:29:19 +01:00
Oswald Buddenhagen
958af473a0 fix conditional for early failure in cancel_done() 2015-01-02 12:38:48 +01:00
Oswald Buddenhagen
9eba3d8cd9 don't leave 2nd store in limbo if opening 1st store fails synchronously
we can't leave the store FRESH, as otherwise the error handling code
will assume it is still being opened and will return to the main loop.
depending on the config this would cause an immediate termination or an
indefinite wait.
2015-01-02 11:29:51 +01:00
Oswald Buddenhagen
3db3f4718e remove stray CRLF from AUTHENTICATE continuations
this re-introduces 6741bc94 (just a bit differently), thus effectively
reverting fbfcfea5. i suppose this extra CRLF is needed by a broken
CRAM-MD5 implementation of some server, which is why it was there in the
original implementation as well. however, it breaks more pedantic
non-broken servers. if somebody complains, we'll need to add a much
more sophisticated hack.
2014-12-29 02:30:33 +01:00
Oswald Buddenhagen
518b5630dc cleanup dead cram() prototype 2014-12-29 02:30:33 +01:00
Oswald Buddenhagen
5dfca41422 fix more memcmp() abuse
amends 1217193fb and 4f383a807.
2014-12-29 02:30:01 +01:00
Oswald Buddenhagen
6f7d416bb8 fix acceptance of trusted SSL certs
we should make no assumptions about the layout of OpenSSL's certificate
store, in particular when they are wrong. so copy the interesting part
instead of "deep-linking" into it.

this code is uglier than it should be, but OpenSSL's extensive use of
macros to manage data types would force us to include the ssl headers
into our headers otherwise, which would be even uglier.

REFMAIL: <545442CC.9020400@nodivisions.com>
2014-11-08 14:00:28 +01:00
Oswald Buddenhagen
f377e7b696 introduce FieldDelimiter and InfoDelimiter options
... for windows fs compatibility.

the maildir-specific InfoDelimiter inherits the global FieldDelimiter
(which affects SyncState), based on the assumption that if the sync
state is on a windows FS, the mailboxes certainly will be as well, while
the inverse is not necessarily true (when running on unix, anyway).

REFMAIL: <CA+m_8J1ynqAjHRJagvKt9sb31yz047Q7NH-ODRmHOKyfru8vtA@mail.gmail.com>
2014-10-25 17:42:48 +02:00
Oswald Buddenhagen
85fd5ceb54 move orig_name out of store_t
it's state specific to the synchronizer.
2014-10-25 15:06:50 +02:00