1)
Added missing checks for CreateEvent which also required the
following related changes:
- changed freerdp_context_new API to BOOL
- changed freerdp_peer_context_new API to BOOL
- changed pRdpClientNew callback to BOOL
- changed pContextNew callback to BOOL
- changed psPeerAccepted callback to BOOL
- changed psPeerContextNew callback to BOOL
2)
Fixed lots of missing alloc and error checks in the
changed code's neighbourhood.
3)
Check freerdp_client_codecs_prepare result to avoid segfaults
caused by using non-initialized codecs.
4)
Fixed deadlocks in x11 caused by missing xf_unlock_x11() calls
in some error handlers
5)
Some fixes in thread pool:
- DEFAULT_POOL assignment did not match TP_POOL definition
- don't free the pool pointer if it points to the static DEFAULT_POOL
- added error handling and cleanup in InitializeThreadpool
* top level GDI functions return 0 on error and != 0 otherwise but the
low level functions (16bpp.c, 8bpp.c 32bpp.c) which are called did it
exactly the other way around. Those were adapted.
* change gdi_InvalidateRegion to BOOL and check calls where appropriate
* integrate comments from pull request
Now using nCount as in and out argument.
When called, set nCount to the number of available handles.
This value is checked and an error returned, if not enough
handles are available.
* Though not frequent, it's possible to get TsProxySetupReceivePipe
data of stublength 4 that is actual data. This happens when
header->common.call_id == rpc->PipeCallId &&
!(header->common.pfc_flags & PFC_LAST_FRAG).
This should address GW disconnects that manifest as SSL read errors.
Change the return type of Stream_Ensure*Capacity from void to BOOL to be
able to detect realloc problems easily. Otherwise the only way to detect
this was to check if the capacity after the call was >= the required
size.
In case Stream_Ensure*Capacity fails the old memory is still available
and need to freed outside.
This commit also adds checks to most calls of Stream_Ensure*Capacity to
check if the call was successful.
The X.224 Connection Request PDU might contain an optional cookie or
routing token before the optional RDP Negotiation Request (rdpNegReq).
If present, both of these fields must be terminated by a 0x0D0A
two-byte sequence. It seems that until now FreeRDP has incorrectly
assumed that a token or cookie must always be present.
If that was not the case, FreeRDP was searching for 0x0D0A until it
arrived at the end of the stream which prevented the remaining data
(RDP Negotiation Request, RDP Correlation Info) from being parsed.
Small cleanup of passing around decorations flag.
Limit PercentScreen to single monitor vs. entire desktop. IMO - this is better behavior in a multimonitor environment.
Handle fullscreen windows better:
1. Ensure that size hints are set to allow resizing before setting a window to fullscreen as some window managers do not behave properly.
2. Handle fullscreen toggles without destroying and recreating window.
3. Use NET_WM_STATE_FULLSCREEN Extended Window Manager Hint for fullscreen functionality
4. Use the NET_WM_FULLSCREEN_MONITORS Extended Window Manager Hint when appropriate
5. When a single monitor fullscreen is requested - use the current monitor(as determined from mouse location)
6. Handle cases where there is no local monitor at coordinate 0,0. The Windows server expect there to be a monitor at this location, so we maintain offset if necessary between our local primary monitor and the server side primary monitor located at 0,0.
Situation: we have fragmented TPKT PDU without two last bytes
(or one last byte - for fast-path) in network stack.
First call to transport_read_pdu() works normally, read
available bytes and exit with status 0 - no whole PDU readed.
Before second call this missed bytes arrive.
Optionally with next PDU.
In second call header parsing code unconditionally read this
two bytes(one byte) despite this is not header bytes.
And increase stream position, so stream now contains whole PDU.
This cause (pduLength - Stream_GetPosition(s)) calculation to be 0.
So transport_read_layer_bytes()-->transport_read_layer() return 0
and transport_read_pdu() exits with "not enough data is available"
status.
If next PDU isn't available next calls to transport_read_pdu()
give same result.
If next PDU arrive - (pduLength - Stream_GetPosition(s)) will be
less than 0. Stream position will grow, grow and grow on each call.
And transport_read_pdu() never signals that PDU is readed.
Caught on Android FreeRDP client with high RDP traffic (several MBytes/s).
Keepalive settings are usually (depending on the implementation) only
used if the TCP connection is idle.
If the network is interrupted/disconnected/... click or keyboard input
generates outgoing traffic therefore the connection isn't idle
anymore and keepalives might not be used causing the connection to
stay open and the client to stall.
Linux 2.6.36 added a TCP_USER_TIMEOUT TCP socket option that lets a
program specify the maximum time transmitted data may remain
unacknowledged before TCP will close the corresponding connection with
ETIMEDOUT.
Setting TCP_USER_TIMEOUT allows us to detect a network problem (like
cable disconnect) even if the connection isn't idle.
refreshRectSupport and suppressOutputSupport of the General
Capability Set (MS-RDPBCGR 2.2.7.1.1) are server-only flags
that indicate whether the Refresh Rect or Suppress Output
PDUs are supported by the server.
Therefore in rdp_read_general_capability_set() we must only
change the respective settings if we are not in server mode.
WTSStartRemoteControlSession doesn't allow to specify additional flags
therefore add a new extended version WTSStartRemoteControlSessionEx
with an additional "flags" parameter.
The following flags are defined:
REMOTECONTROL_FLAG_DISABLE_KEYBOARD - disable keyboard input
REMOTECONTROL_FLAG_DISABLE_MOUSE - disable mouse input
REMOTECONTROL_FLAG_DISABLE_INPUT - disable input (keyboard and mouse)
This patch fixes a bug with mstsc connecting to a RDP security only FreeRDP server.
It seems like the mstsc shipped with Windows Seven considers packets after the nego_failure
packet as an error. So after trying to do TLS, depending on the timing, mstsc can print an
error message instead of retrying to connect with RDP security. With this patch, we
don't send the MCS disconnect message when the negociation has failed.
Before this patch, RDP security was (wrongly) the fallback when negociating a
security protocol between the client and the server. For example when a client
was claiming TLS-only when connecting to a FreeRDP based-server with RDP security only,
the result of the negociation was that the server started to do RDP security.
The expected behaviour is to send a nego failure packet with error code
SSL_NOT_ALLOWED_BY_SERVER. This patch fixes this.
We also try to handle all cases of failed negociation and return the corresponding
error code.
Commit 0357a38e31 modified the function
fastpath_send_update_pdu() to check if the desired update is possible
by checking the payload size against the computed maxLength and the
clients's advertised max request size.
If the check failed that commit added a workaround which simply
copied the payload to a slow path updade.
This workaround is totally flawed and causes protocol errors:
- the fast path update code is not checked and required data format
conversions are missing
- depending on the fast path update code rdp_send_data_pdu() would
have to be called with differend data pdu type values but the
workaround always uses DATA_PDU_TYPE_UPDATE
- the workaround does not check if the total size would exceed
the maximum possible size for a slow path update
The check if a fast path output is actually possible with the
passed parameters is basically a good idea.
However, if that check fails it would only indicate an error in
the server implementation who must not generate updates that
exceed the client's max request size.
Even though a slow-path conversion would be possible there is
much more involved than simply copying the payload stream.
In addition it is highly doubtful if there is a benefit at all.
Even the oldest rdesktop and windows ce clients do support fast
path and although some lack the multi-fragment update capability
we cannot really send larger updates using slow-path outputs.
For the reasons elucidated above, I have removed the workaround
but kept a modified version of the check if a fast-path output
is possible at all.
Commit 0357a38e31 has added some code
without any effect.
That commit added code to rdp_read_capability_sets() to check if
CAPSET_TYPE_MULTI_FRAGMENT_UPDATE was not received which caused
settings->MultifragMaxRequestSize to be set to 0.
- this was done in the wrong place because we do these kind
of checks in rdp_recv_confirm_active() by consulting the
variable settings->ReceivedCapabilities[]
- the code had no effect at all because MultifragMaxRequestSize gets
set to FASTPATH_FRAGMENT_SAFE_SIZE in rdp_recv_confirm_active()
if the CAPSET_TYPE_MULTI_FRAGMENT_UPDATE was not received.
Extend rdp_pointer with function SetPosition. Can then be used by
clients support setting pointer by server which might be used in
shadowing scenarios.
When a client disconnects from a server and its channel structures are removed, the global hash g_OpenHandles should not be destroyed. Only freed channels must be removed from the hash.
