| Age | Commit message (Collapse) | Author |
|---|
|
The logic for calling onItemAdded and onItemRemoved signals when
setting a new list of block items using BlockListImpl::setItems
used to be broken.
This commit fixes and documents the correct signal notification
behavior
Test-Information:
Added a unit test which verifies the notification behavior in case
of added block list items, removed block list items and a complete
change of the block list.
Change-Id: I3061545e25ddfc2d9d1a3c987045a58e5c9230ac
|
|
According to boost doucmentation shared deadline_timers are not
thread-safe. Adding a mutext to protect access to
boost::asio::deadline_timer instance in Swift::BoostTimer.
This fixes a data-race reported by TSAN when running
Swiften/QA/ClientTest/ClientTest.
Test-Information:
Verified that the data-race report is gone with this fix.
Change-Id: I62c8c3a07d6ea16fe6e2d24c879340040406699b
|
|
FileTransferManagerImpl should have a virtual dtor, as the
parent class also has a virtual dtor.
Test-Information:
Unit tests still run.
Change-Id: I72a0757a930254caadfd1463b3f90af6c4829bd0
|
|
Running './scons check_headers=1 will compile each header file
as an object file.
Test-Information:
Tested on OS X 10.9.5 and Elementary OS 0.2.
Change-Id: Ia429ccc6dc0a5bd76c08a2c2e91bf192c6ad2901
|
|
As the correct API usage might not be obvious
Change-Id: I1d4887d7101ce59eb8f3a3dc13d94befe4a58948
|
|
Added Swiften/Base/Concat.h to the exception list for <algorithm>
include, as Concat.h uses std::copy, defined in <algorithm>.
Added Swiften/FileTransfer/S5BTransportSession.h to the exception
list for <boost/bind.hpp> include, as S5BTransportSession.h
declares a template class using boost::bind.
Test-Information:
Tested that each of Swiften's public header files compiles on its
own.
Change-Id: Iac794b6e0f5959cd8261888a67761813b710fe85
|
|
Explicitly disconnect from remaining resources in the
ChainedConnected and ClientSessionStanzaChannel destructors, so
the event loop will not call the signal handler methods on a
freed object.
Test-Information:
Repeating the test case of creating a Swift::Client instance,
connecting it and then deleting it after a random time below one
second.
On Mac OS X 10.9.5 running this test case causes two ASAN
heap-use-after-free errors and with this patch the errors are
gone.
Change-Id: I3e48150c3633f4076ca9172aad9e85ba389df950
|
|
Now Client allocates a FileTransferManager in its ctor and frees
it in its dtor, like all the other resources of the Client class.
This fixes an ASAN crash in the use case where you do a file
transfer, go offline and online again and try to start another
file transfer.
Test-Information:
Tested that the mentioned use case does not crash anymore,
that file transfers still work, and that FileTransferTest still
passes.
Change-Id: Iddbcd8522af7df528bdc2b030fe616ad3f0c4701
|
|
A recent commit introduced resolving of S5B proxy domain names
to their IPv4 and IPv6 addresses. With that a proxy identified
by a JID can have more than one host and we try them in parallel
until the first succeeds.
The old code just handled one host per proxy JID and a failed
IPv6 attempt would override the succeeded connection. The code
uses shared pointers and the succeeded connecting is deallocated
and disconnected when it is replaced with the failing IPv6
connection.
The result is the proxy server complaining that we are not
connected as we try to activate the proxy stream.
This commit changes the the proxy management to handle multiple
connections per proxy JID. Failing connections are removed from
the proxy sessions data structure. With the first succeeding
connections, others are stopped and also removed.
Test-Information:
Tested on Linux (Elementary OS 0.2) with
"Swiften/QA/FileTransferTest/FileTransferTest 4 4", which forces
the use of SOCKS5 bytestream proxy.
Change-Id: If3071c3d058e1040556bb72702bf83f4f5f25334
|
|
Test-Information:
This LSAN reported leak is gone with this fix.
Change-Id: Ib25fdfb7f4a82f62cd3f82ddb99f45618ea72ea9
|
|
Reported by LSAN.
Test-Information:
Detected by running FileTransferTest with LSAN on Linux.
The leak reports for the allocations in FileReadBytestream and
FileWriteBytestream are gone with this fix.
Change-Id: I32711990eca0c9a2a2982837cfac38cb11a28caa
|
|
Adds InBandRegistrationPayloadParserTest, IBBSerializerTest, IsodeIQDelegationSerializerTest.
Adds UserTune Parser and Serializer.
Adds UserLocation Parser and Serializer.
License:
This patch is BSD-licensed, see Documentation/Licenses/BSD-simplified.txt for details.
Test-Information:
All tests passes.
Change-Id: Ida220574c33ca9ee6f2aa8a2f4fba4c68e3fec60
|
|
Test-Information:
Detected with valgrind-3.10.1 on OS X 10.9.5 and verified they are
gone with this fix.
Change-Id: I8cab09efd2227a8d43fa77bd2c315bd9b67435fb
|
|
This is an attempt to fix crash reports with a segmentation fault on
call of onLocalTransportCandidatesGenerated().
Test-Information:
Tested that file-transfer still works.
Change-Id: I415be0d8bc5fa15dcd68d0794ee2478a0d836f27
|
|
This patch uses the GSSAPI authenticator on Windows if the server advertises it
and the client requests it. (The user is not able to request it in the UI yet)
Also sends the manual port to the GSSAPI authenticator to construct the SPN if
a non-default port is used.
Test-information:
Tested on Windows using WIP code. Tested both on TLS & without.
Unit tests pass.
Change-Id: I9a9ad9604fe084d5fb2003b7a91174a9512e2eec
|
|
Test-Information:
Added unit test that checks for the expected behavior.
Change-Id: I4079bdc1182af466eedd2496b9837e024f14acb2
|
|
If the S5B proxy lookup result has no direct IP address listed the
name is resolved using DNS. This change will create a proxy result
per resolved result address and not only the first result address.
Test-Information:
Verified the candidate list contains both entires for a proxy name
that resolves to an A and an AAAA record.
Change-Id: Iec21ff90af981030ff49fb53803d88a59694767c
|
|
ASAN reported heap-use-after-free because if the timer is running
and an instance is freed, the signal of the timer is still emitted and
connected slots are called.
Test-Information:
Tested that the ASAN report wents away with this fix.
Change-Id: Id785737c4c015e5c638e9d9f6419d740d6cf83b9
|
|
Test-Information:
Still builds
Change-Id: I93cd349364093eae320104b5b551c25da47b1d74
|
|
Due to missing signal connection and a copy'n'paste bug, Swift did not
wait for port forwarding/public IP detection before emitting the list
of local candidates. This is fixed now.
The signal is automatically disconnected when the file transfer
is finished and s5bServerPortForwardingUser_ is freed.
Test-Information:
Send a file between two Swift instances and verified log output.
Change-Id: I6530a7ac1cbf6941061bd99aa3f3b0624ebc984c
|
|
Introducing SWIFT_NETWORK_TEST_IPV4 and SWIFT_NETWORK_TEST_IPV6 for
the host addresses used in BoostConnectionTest.cpp.
Passing these variables and ASAN/LSAN_OPTIONS for memory leak
detection when calling the test from SCons.
Test-Information:
Tested on Elementary OS with correct available hosts and unavailable
hosts. In both cases the behavior is the same as before.
Change-Id: I24128c766642c5a2ee273f7c11624525ae33e270
|
|
Test-Information:
Tested on Elementary OS with a LLVM/Clang 3.7.0 build. With this patch
the reports for the fixed locations are gone.
Change-Id: I0260fc85ad662335a69ace331f51ebe2f864ef97
|
|
This patch adds an option 'disconnectOnCardRemoval' to system-settings.xml which
when set to false allows the user's session to stay connected if the smartcard
is removed. The default value of this option is true if it is not specified.
Test-information:
Tested on Windows using NIST smartcards.
Tested true and false values set for this option in the file and also when
option is not specified (true).
Unit tests pass.
Change-Id: I7e421b4153ff7d3000f41999add20d339076c96e
|
|
Test-information:
Tested on Windows using WIP code.
Unit tests pass.
Change-Id: I766294e57dc6374830b865f3e57b07b67e7d2fe2
|
|
When the card is reset in shared mode (which is the mode we use), an application
trying to access certain commands will be returned the value SCARD_W_RESET_CARD.
When this occurs SCardReconnect() must be called. This wasn't done before so
this patch fixes it.
Also provides more logging for SCard function returns.
Test-information:
Tested on Windows using NIST smart cards.
Before the fix, the Card Removed Error was seen often even after the initial
connection was established. After the fix, the reconnect attempt is done so the
error is not seen.
Also verified that if a card is removed deliberately, then the user is logged
out.
Change-Id: I94748ab9ff944a79de655646e1e06a8b61776f4b
|
|
Test-information:
Tested on Windows using WIP GSSAPI code.
Unit tests pass.
Change-Id: I21f8f637480a21a014ec172431dd8d4a01a11620
|
|
In RFC 6120, there are only 2 places where "=" is allowed - in the initial
client auth and in the final server success response.
While testing challenge response exchanges in Kerberos, it was seen that Swift
was adding an '=' in other empty responses. This patch fixes it by sending an
empty response instead of an '='.
Test-information:
Tested on Windows using a WIP GSSAPI authentication exchange with M-Link.
Change-Id: I1f82bddbd1380361cbe43e45a2804156249582ae
|
|
Test-information:
Tested on Windows using WIP GSSAPI code.
Unit tests pass.
Change-Id: If872863d6a8b5a164f8ebec4f88e9939b4e73c62
|
|
This patch creates and passes on an optional boost::system::error_code variable
which contains more error info which can be displayed in case of an error.
Test-information:
Tested using WIP code on Windows.
Change-Id: I285b8aec5e9c00d3a8e0d8cc0d5e7b4c5d94c099
|
|
Test-information:
Verified with M-Link.
Unit tests pass.
Change-Id: Ic675c8d7cd70e01be61c51c0280e1d7208b364ba
|
|
Clang raises a warning in non-experimental builds because the full
type of HistoryStorage is not known and it tries to delete an
instance (even though a null ptr) in the dtor.
Test-Information:
Tested on OS X 10.9.5 with Clang (Apple LLVM version 6.0).
Change-Id: I8a06890967427af21e11e077e9ac59b67bdb810c
|
|
Template classes cannot have SWIFTEN_API annotations as with annotation
code for instantiations of the template are expected to be in the DLL
which cannot be guaranteed for any user type. With the complete
implementation in available in the header it is not needed because
3rdParty Swiften users can instantiate an implementation as needed.
This also conditionally includes SQLiteHistoryStorage.h conditionally
in MemoryStorages.cpp, as otherwise the linker will expect an
implementation of SQLiteHistoryStorage in the DLL. However, it is only
built into the DLL if experimental features are turned on.
Test-Information:
Tested with experimental=on/off and swiften_dll=on/off.
Change-Id: Ieec85675c167ec34cffd4745ac854a5949fb2037
|
|
Test-Information:
Send a file from one Swift to another. Previously the progress bar did
not update in the receiving client. Now it does.
Change-Id: Ie710a4c1cfc389d6fcfb93de96250a80787883d3
|
|
Some servers have very restrictive TLS stacks that respond badly
to a bug in the SChannel TLS implementation, meaning that TLS
has to be limited to 1.0.
Add ClientOptions.tlsOptions. This is a method of passing options into
the TLS stack. It's currently
only used for the TLS 1.0 workaround in SChannel, but we might reasonably
expose other options in the future, such as limiting cypher suites.
Disables use of SSLv3 for SChannel
Also updates the coding style in SchannelContext a bit.
Test-Information:
Compiles on both OS X and Windows(SChannel). OS X doesn't show the new
option. Windows shows it, and remembers it between logins. Not tested
against a server requiring 1.0 only, but a previous hack with the
same approach was tested.
Change-Id: I1e7854d43811fd173f21f98d4dc3915fc7a4b322
|
|
Test-Information:
Tested build on Windows 8 with VS 2014 and ran unit tests.
Change-Id: I3d8096df4801be6901f22564e36eecba0e7310c4
|
|
Use scoped_connection to prevent missing disconnection from signals.
Stop and free S5BServer when stopping SOCKS5BytestreamServerManager.
Test-Information:
Tried sending a file to myself multiple times and it did not crash.
Change-Id: If32075d8e9c243cab254776b924248227520e030
|
|
Test-Information:
Verified that SCons still runs on OS X.
Change-Id: I7e9b97f90ee5581a691a959b6f2c999d93e0be53
|
|
Change-Id: Ieeb9636b40bfd43b9b79dede71eefc288e6014d2
|
|
released early
This patch also disconnects possible slots on stop() from the signals.
Test-Infromation:
This fixes a crash that occured during testing with
ConcurrentFileTransferTest.
Change-Id: I2b443f40ac3359c8142fcffc084647312b9a1e76
|
|
This also adds missing resets pointers to NULL after they are deleted.
Test-Information:
FileTransferTest crashed in some scenarios before; now it always runs
successful.
Change-Id: Ic63fd51eeb46e708221a04dc912e8bf2f1d4f9fb
|
|
running
Add a check on whether we have a running SOCKS5 bytestream server
before trying to access it to close unneeded S5B connections to it
after we decided on a candidate to use.
Test-Information:
Without this FileTransferTest crashes in the configuration that does
not allow direct or assisted S5B connections. This patch fixes this
issue.
Change-Id: Ifcf59f82755b9620e7f5fa5841e45815a41ac442
|
|
This patchs adds management of the SOCKS5 bytestream server and port
forwarding setup to Swiften. The first file-transfer, regardless of
incoming or outgoing transfer, will start and initialize the S5B server
and if configured try to set up port forwarding. The last file-transfer
to finish will will close the server and remove the port forwarding.
Test-Information:
Tested with upcoming ConcurrentFileTransferTest.cpp and running multiple
S5B file-transfers in parallel.
Change-Id: Idd09d3d0ef9498fc9435b0aee81f2b1061783342
|
|
This modifies the feature detection in the ChatController to try to use
the common features of all available resources feature detection if no
full JID has been bound to the chat yet.
Test-Information:
Tested with two Swift instances. Tested
a) the initial chat start case and,
b) the offline/online.
In case a) Swift used to initally show a yellow warning about no support
for message receipts. This warning is gone now.
In case b), after a user gone offline and online again in a running chat,
Swift used to show a warning about missing support for message receipts.
This warning is gone now.
Change-Id: I7a769fde8d14847b180503aeaa58280c572d81b3
|
|
To enable native Android logging you need to pass
"-DSWIFT_ANDROID_LOGGING" to the "cxxflags" and "-llog" to the
"linkflags".
Test-Information:
Tested on OS X 10.9.5, with NDK API level 14 on an emulated Android 4.0
ARM instance.
Tested that if enabled, Swift log messages appear in the Android
system log via "adb logcat" instead of appearing on stdout.
Change-Id: If89c870f0b664a5e65ccc1fa8a244826d2875ec1
|
|
Removed dead debugging code and unnecessary includes as well.
Test-Information:
Tested with Swiften/QA/ClientTest, adjusted to have a manual hostname
set in the ClientOptions.
Tested on OS X 10.9.5, with NDK API level 14 on an emulated Android 4.0
ARM instance.
Without this patch the connector timed out during name lookup trying
to resolve the IP set in the ClientOptions::manualHostname. With this
patch it skips lookup and connects successfully.
Tested and verified unchanged behavior with this patch and without the
manual address override set.
Change-Id: I737327b2e66c9da78a1963e754bcf201b7d40626
|
|
Test-Information:
Compiles without the warning, unit tests pass.
Change-Id: I31bce47e4d6595b4ef405a87b56cf71db004e6a0
|
|
IncomingJingleFileTransfer::terminate already sets the finshed state
for the file-transfer.
Test-Information:
Verified in the logs that onFinished signal is only called once.
Change-Id: Ib2084e1487ded2a921f61b237cd036e1ae4cf992
|
|
Test-Information:
The FileTransfer::onFinished was not called when the file transfer finished.
Verified with log output that it is called correctly now.
Change-Id: I05f2a749ca3f0aece08cae9503a212cf0327f98a
|
|
Previously all IQ handlers were final. The IQ handles were processed
in last-added-first order and if the IQ payload matched the handler
payload, the IQ would only be passed to that handler.
This conflicts with our IBBReceiveSession. When running multiple
concurrent file-transfers using multiple IBBReceiveSessions there are
multiple IQ handlers for IBB content; one for every transfer.
This commit allows a Responder to be set as non-final. In this case
unhandled IQs will not be responded with an error but returned to
the IQRouter so it can pass it to the next possible IQ handler.
Test-Information:
Tested with ConcurrentFileTransferTest with runs multiple IBB transfers
in parallel.
Change-Id: I8237e234cbe5c110deaa8c3d6ba303b65fd53d00
|
|
Test-Information:
Builds.
Change-Id: Ica4b1bca4ff71f103ee5ebad707bc0f95595ad37
|
Swift