Global static variables do not work, if more than one instance
of an RDP client is running in the same process space.
Removed the varaibles where possible and replaced them with
thread local storage where necessary.
Function pa_stream_trigger has to be called under lock to avoid
following crashes on asserts:
Assertion 'e->mainloop->n_enabled_defer_events > 0' failed at pulse/mainloop.c:257, function mainloop_defer_enable(). Aborting.
Assertion '!e->next' failed at pulsecore/queue.c:104, function pa_queue_pop(). Aborting.
Assertion 'q->front' failed at pulsecore/queue.c:81, function pa_queue_push(). Aborting.
Not use sizeof on packet memory structure for stream length check/estimating and use number directly.
Most of the packet size could be obviously calculated/verified from code and comments in same function.
Macro is defined for those were referenced multiple times.
1. Fix style for channels/rdpgfx/server/rdpgfx_main.c according to scripts/format_code.sh
2. Refactor and introduce APIs to allow rdpgfx server channel running in existing thread (similar to rdpsnd)
1. Fix stream leak in rdpgfx
2. Make src data const in zgfx. Harden zgfx to be independent to byte order
3. Fix written bytes return value in channel write
4. Add check for return value in shadow_client.c
5. Add gfx callback to send surface command with frame marker pdu.
6. Check remain length for recv subroutine
7. Fix compile errors
When no shared libs are built is required to export all generated
archive files otherwise programs linking against
freerdp-client/freerdp-server might miss symbols.
Fixes#3350
Add all missing dependencies found with --no-undefined. Since
dependencies aren't exported anymore (if not required) it is no necessary
to explicitly list all required libraries.
If a target is linked against libraries with cmake
(target_link_libraries) and the libraries are not marked as PRIVATE
they are "exported" and in case a other target is linked against this
target it is also linked against *all* (not private) libraries.
Without declaring private libraries PRIVATE a lot of over linking
(linking against unneeded libraries) was done.
PR #3173 fixed argument parsing. However it breaks the usage without specifying subsystem: "/microphone"
In particular, aFreerdp microphone redirect is broken as it exactly uses "/microphone"
This PR fixes the regression.
Fix: We should not call CommandLineParseArgumentsA if argc is only 1 (same case as rdpsnd_main.c)
- Added missing ConvertFromUnicode checks
- If ConvertToUnicode allocates memory, guarantee the null termination
similar to ConvertFromUnicode's implementation
- Fixed some TestUnicodeConversion.c CTest return values
- Added some CTests for ConvertFromUnicode and ConvertToUnicode
- Misc code and protocol hardening fixes in the surrounding code regions
that have been touched
After #3097, the java side pass command line argument to JNI for freerdp settings. However there's several issues need to be fixed:
1. The argument /sound should be appended if freerdp is required to play sound at local device
2. The option value for "audio-mode" is not correct.
It should match the definition in client/common/cmdline.c
/* Audio Mode */
define AUDIO_MODE_REDIRECT 0 /* Bring to this computer */
define AUDIO_MODE_PLAY_ON_SERVER 1 /* Leave at remote computer */
define AUDIO_MODE_NONE 2 /* Do not play */
3. Uncomment support for WAVE_FORMAT_PCM in audin. I tested on my android phone and Nokia N1 tablet. It works on both device
Starting with cmake 2.8.10 FreeRDP exports a cmake find module. With 2.8.12
the PRIVATE/PUBLIC keywords were introduced in cmake. When building with
2.8.11 it is not possible to mark link dependencies as private and
therefore they need to be exported.
With this commit the "exported" components (usable with pkg-config and
cmake find module package)
* winpr - winpr library and headers
* freerdp - core library and headers
* freerdp-client - client specific library
* freerdp-server - server specific library
* rdtk - rdtk headers and library
To allow the installation of multiple different version (different major
number) the include files were moved into the respective sub folder:
freerdp -> freerdp{MAJOR}/freerdp (currently freerdp2/freerdp/)
winpr -> winpr{MAJOR}/winpr (currently winrp1/winpr/)
rdtk -> rdpk{MAJOR}/rdtk (currently rdtk0/rdtk/
The generated pkg-config and cmake find modules now also include the major
version number. Currently the following pkg-config are generated and
installed.
* winpr1
* freerdp2
* freerdp-server2
* freerdp-client2
* rdtk0
As cmake is able to handle multiple versions out of the box the
following can be used to find a specific module:
find_package(WinPR)
find_package(FreeRDP)
find_package(FreeRDP-Server)
find_package(FreeRDP-Client)
find_package(RdTk)
As cmake doesn't automatically resolve dependencies for packages it is
necessary to manually include the requirements. For example if
FreeRDP-Client is required WinPR and FreeRDP need to be included
(find_package) as well.
This commit also fixes the installation when STATIC_CHANNELS are built.
WITH STATIC_CHANNELS all channels are linked into libfreerdp-client, for
this all channels are generated as linker archive and linked together in
the final step. Before the intermediate linker archives were, although
not required and useful, installed. Same applies for server side
channels.
build-config.h should contain configure/compile time settings that are
relevant for projects that use FreeRDP.
For example the compiled in plugin search paths.
- Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec
- Use autoaudiosink - let gstreamer choose best audio playback plugin
- Catch when autosinks add known elements so that we can manipulate properties on them
- Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available
- Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec
- Name audio/video gstreamer elements better for easier debugging
- Update gstreamer pipeline and element properties to handle playback better
- Detect when valid timestamps are available for buffer from server and try to account for when they are not valid
- Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time
- Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time
- Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf
- Remove ack function from gstreamer, instead rely on ack thread to handle acks
- Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display
- Add check to see if a decoder is available for telling the server the client various media types
- Add in support for M4S2 and WMA1 media types
- Fix flush message handling, they are for individual streams and not the entire presentation
- Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos.
- Fix issue with geometry info being ignored when resent for new streams within existing presentation
- Fixed volume level initialization issue when a stream is stopped and restarted
- Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt
to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
* oss: fix function signature
* autodetection - if only /sound is given all enabled
plugins are tried in order and the first successful loaded is used.
- this restores the previous behavior
* alsa/pulse fix command line parsing - no parameters shouldn't be
treated as error
The rdpContext gets an event which will
get set if an error occoured in a channel.
If a thread or a void callback has to report an
error it will get signaled by this system.
winsock.h pulls in a lot of defines and dependencies that are not
required and partially unwanted in winpr's core (for parts that are not
related to network). In order to get rid of this dependency and have an
independent defines for extended winpr functions the WINPR_FD_* defines
are used internally (and for exposed functions). Where required, like in
WSAEventSelect, the FD_* is mapped to WINPR_FD_*.
This patch adds some checks for the TSMF channel:
* checks malloc / calloc / strdup / _strdup return value
* checks for available bytew before reading in a stream
This patch adds checks for the sound channel and plugins:
* checks for protocol bytes;
* checks for malloc/calloc;
* be more strict when interacting with sound plugin
This patch contains:
* checks for malloc return value + treat callers;
* modified malloc() + ZeroMemory() to calloc();
* misc fixes of micro errors seen during the code audit:
** some invalid checks in gcc.c, also there were some possible
integer overflow. This is interesting because at the end the data are parsed
and freed directly, so it's a vulnerability in some kind of dead code (at least
useless);
** fixed usage of GetComputerNameExA with just one call, when 2 were used
in misc places. According to MSDN GetComputerNameA() is supposed to return
an error when called with NULL;
** there were a bug in the command line parsing of shadow;
** in freerdp_dynamic_channel_collection_add() the size of array was multiplied
by 4 instead of 2 on resize
The function with the same name is defined in multiple libraries
(libfreerdp-shadow.so, libfreerdp-server.so), which might confuse
the dynamic linker at runtime, binding a wrong version of the function
and thus causing segmentation faults.
Since remdesk_virtual_channel_write()s aren't used outside the files
they are defined in, we can declare them static to make them invisible
to the linker.
