Pierre Ossman [Mon, 18 May 2020 16:53:43 +0000 (18:53 +0200)]
Support calling methods from timers
We can't safely use the normal timers in base classes as we cannot
guarantee that subclasses will call the base class' handleTimeout()
properly if the subclass overrides it.
Lukas Vrabec [Wed, 6 May 2020 08:55:29 +0000 (10:55 +0200)]
[SELinux] Allow vnc_session_t type execute itself
vncsession-start is running in SELinux vnc_session_t domain because of
"SELinuxContext=system_u:system_r:vnc_session_t:s0" option in systemd
vncserver@.service unit file. vncsession-start executing binary
vncsession with SELinux label/type vnc_session_t. This access was not
allowed in vncsession policy.
Pierre Ossman [Mon, 20 Apr 2020 06:55:13 +0000 (08:55 +0200)]
Handle empty changes for every operation
It seems like many of the X11 operations can end up with no pixels
actually changing. So instead of discovering and adding workarounds for
each individually we'll just check very region added if it's empty.
Alex Tanskanen [Tue, 10 Mar 2020 12:33:01 +0000 (13:33 +0100)]
Throw GAIException() for getaddrinfo errors
Created a new subclass of Exception called GAIException() that will
handle error messages from getaddrinfo() instead of letting Exception()
handle it. GAIException() will make use of gai_strerror() to map the
error code to text. On Windows, gai_strerrorW() must be used if the text
is encoded with UTF-8.
Alex Tanskanen [Wed, 4 Mar 2020 09:27:02 +0000 (10:27 +0100)]
Make system error messeges in Windows 10 use UTF-8
The previous error messages did not support Unicode characters. This
commit will use UTF-8 encoding to be able to display error messages in
every language.
Pierre Ossman [Thu, 19 Jul 2018 14:04:23 +0000 (16:04 +0200)]
Add SELinux policy file for vncsession
Running as a service on a SELinux system requires rules so we can
transition to our own context. We also need the proper permissions
to start new user sessions.
Pierre Ossman [Mon, 16 Jul 2018 13:58:06 +0000 (15:58 +0200)]
Start sessions via PAM
This sets up a more correct session as there are key tasks that
need to be performed by PAM. E.g. systemd will allocate cgroups
and start base services.
In order to easily handle this as a system service the mapping of
displays is now done via a configuration file.
Pierre Ossman [Wed, 23 Jan 2019 14:29:41 +0000 (15:29 +0100)]
Make vncserver always run in the foreground
We need to be started as a system service for things to work correctly
anyway, so delegate the work of starting and stopping things to the
system service manager (e.g. systemd).
Linus Heckemann [Sat, 1 Feb 2020 10:08:26 +0000 (11:08 +0100)]
xserver: add no-op input thread init function
This allows Xvnc to build with xorg-server 1.20.7, which requires OS
layers to implement a ddxInputThreadInit function when configured with
--enable-input-thread (the default).
Pierre Ossman [Fri, 3 Jan 2020 12:10:20 +0000 (13:10 +0100)]
Clean up initialization of DIBSectionBuffer
We had an unintentional conflict with PixelBuffer::setSize() here.
But we can simplify this further as this initialization is only used
by the subclass DeviceFrameBuffer, and only once.
Pierre Ossman [Mon, 30 Dec 2019 09:50:52 +0000 (10:50 +0100)]
Provide correct dimensions for XShm setup
Since 53f913a we initialize the underlying PixelBuffer with 0x0
dimensions, which means we need to keep more explicit track of what
we are trying to allocate in the setup methods.
Samuel Mannehed [Fri, 20 Dec 2019 15:06:48 +0000 (16:06 +0100)]
Use the correct minimum value for compression lvl
There was even some confusion in the RFB protocol regarding this, but
the zlib implementation confirms that accepted values for compression
level is 0-9.
Pierre Ossman [Mon, 9 Dec 2019 15:58:07 +0000 (16:58 +0100)]
Check the correct stream if there is more data pending
The input stream might no longer be the raw socket, so we need to
query what's currently active. That wrapping stream might have its
own buffering and may have more data even if the socket is drained.
Pierre Ossman [Mon, 9 Dec 2019 15:57:04 +0000 (16:57 +0100)]
Handle pending data in TLS buffers
There might be more bytes left in the current TLS record, even if
there is nothing on the underlying stream. Make sure we properly
return this when we aren't being requested to block.
Pierre Ossman [Mon, 9 Dec 2019 12:54:17 +0000 (13:54 +0100)]
Don't background the main session command
When used with -fg we expect the startup script to remain running until
the session is over. This will not happen if the session command is put
in the background using &.
Pierre Ossman [Wed, 2 Oct 2019 14:06:08 +0000 (16:06 +0200)]
Handle pixel formats with odd shift values
Our fast paths assume that each channel fits in to a separate byte.
That means the shift needs to be a multiple of 8. Start actually
checking this so that a client cannot trip us up and possibly cause
incorrect code exection.
Issue found by Pavel Cheremushkin from Kaspersky Lab.
Pierre Ossman [Tue, 24 Sep 2019 07:41:07 +0000 (09:41 +0200)]
Be defensive about overflows in stream objects
We use a lot of lengths given to us over the network, so be more
paranoid about them causing an overflow as otherwise an attacker
might trick us in to overwriting other memory.
This primarily affects the client which often gets lengths from the
server, but there are also some scenarios where the server might
theoretically be vulnerable.
Issue found by Pavel Cheremushkin from Kaspersky Lab.
Pierre Ossman [Tue, 10 Sep 2019 14:07:50 +0000 (16:07 +0200)]
Add sanity checks for PixelFormat shift values
Otherwise we might be tricked in to reading and writing things at
incorrect offsets for pixels which ultimately could result in an
attacker writing things to the stack or heap and executing things
they shouldn't.
This only affects the server as the client never uses the pixel
format suggested by th server.
Issue found by Pavel Cheremushkin from Kaspersky Lab.
Pierre Ossman [Tue, 10 Sep 2019 13:21:03 +0000 (15:21 +0200)]
Restrict PixelBuffer dimensions to safe values
We do a lot of calculations based on pixel coordinates and we need
to make sure they do not overflow. Restrict the maximum dimensions
we support rather than try to switch over all calculations to use
64 bit integers.
This prevents attackers from from injecting code by specifying a
huge framebuffer size and relying on the values overflowing to
access invalid areas of the heap.
This primarily affects the client which gets both the screen
dimensions and the pixel contents from the remote side. But the
server might also be affected as a client can adjust the screen
dimensions, as can applications inside the session.
Issue found by Pavel Cheremushkin from Kaspersky Lab.
Pierre Ossman [Tue, 10 Sep 2019 13:18:30 +0000 (15:18 +0200)]
Encapsulate PixelBuffer internal details
Don't allow subclasses to just override dimensions or buffer details
directly and instead force them to go via methods. This allows us
to do sanity checks on the new values and catch bugs and attacks.