Freerdp: ERRINFO_INVALID_INPUT_PDU_MOUSE and exit on middle mouse button click

[ddombrowsky]

Describe the bug

When connected to any client, if I use the middle mouse button, the program forcibly disconnects from the remote host.

[10:43:01:915] [8258:8259] [INFO][com.freerdp.core] - ERRINFO_INVALID_INPUT_PDU_MOUSE (0x000010D0):(a) A Slow-Path Mouse Event (section 2.2.8.1.1.3.1.1.3) or Extended Mouse Event (section 2.2.8.1.1.3.1.1.4) has been received with an invalid pointerFlags field.
(b) A Fast-Path Mouse Event (section 2.2.8.1.2.2.3) or Fast-Path Extended Mouse Event (section 2.2.8.1.2.2.4) has been received with an invalid pointerFlags field.
[10:43:01:917] [8258:8259] [ERROR][com.freerdp.core.transport] - BIO_read returned a system error 104: Connection reset by peer
[10:43:01:917] [8258:8259] [ERROR][com.freerdp.core] - freerdp_check_fds() failed - 0
[10:43:01:917] [8258:8259] [ERROR][com.freerdp.client.x11] - Failed to check FreeRDP file descriptor

I can live without it, but it would be nice to have a work-around that doesn't close the connection.

To Reproduce
Steps to reproduce the behavior:

1. open a connection from linux to a local windows VM.
2. click the middle mouse button

Expected behavior
Either nothing should happen, or the middle mouse button should be passed through to the remote host.

** Application details

• Version of FreeRDP
  This is FreeRDP version 2.0.0-dev (git n/a)

• Command line used

xfreerdp /size:1724x960 /v:davek-win /sec:rdp /u:davek -grab-keyboard

• output of /buildconfig

This is FreeRDP version 2.0.0-dev (git n/a)
Build configuration: BUILD_TESTING=OFF BUILTIN_CHANNELS=ON HAVE_AIO_H=1 HAVE_EVENTFD_H=1 HAVE_EXECINFO_H=1 HAVE_FCNTL_H=1 HAVE_INTTYPES_H=1 HAVE_JOURNALD_H=TRUE HAVE_MATH_C99_LONG_DOUBLE=1 HAVE_POLL_H=1 HAVE_PTHREAD_MUTEX_TIMEDLOCK=ON HAVE_PTHREAD_MUTEX_TIMEDLOCK_LIB=1 HAVE_PTHREAD_MUTEX_TIMEDLOCK_SYMBOL= HAVE_SYSLOG_H=1 HAVE_SYS_FILIO_H= HAVE_SYS_MODEM_H= HAVE_SYS_SELECT_H=1 HAVE_SYS_SOCKIO_H= HAVE_SYS_STRTIO_H= HAVE_TIMERFD_H=1 HAVE_TM_GMTOFF=1 HAVE_UNISTD_H=1 HAVE_XI_TOUCH_CLASS=1 WITH_ALSA=ON WITH_CCACHE=ON WITH_CHANNELS=ON WITH_CLIENT=ON WITH_CLIENT_AVAILABLE=1 WITH_CLIENT_CHANNELS=ON WITH_CLIENT_CHANNELS_AVAILABLE=1 WITH_CLIENT_COMMON=ON WITH_CLIENT_INTERFACE=OFF WITH_CUPS=ON WITH_DEBUG_ALL=OFF WITH_DEBUG_CAPABILITIES=OFF WITH_DEBUG_CERTIFICATE=OFF WITH_DEBUG_CHANNELS=OFF WITH_DEBUG_CLIPRDR=OFF WITH_DEBUG_DVC=OFF WITH_DEBUG_KBD=OFF WITH_DEBUG_LICENSE=OFF WITH_DEBUG_MUTEX=OFF WITH_DEBUG_NEGO=OFF WITH_DEBUG_NLA=OFF WITH_DEBUG_NTLM=OFF WITH_DEBUG_RAIL=OFF WITH_DEBUG_RDP=OFF WITH_DEBUG_RDPDR=OFF WITH_DEBUG_RDPEI=OFF WITH_DEBUG_REDIR=OFF WITH_DEBUG_RFX=OFF WITH_DEBUG_RINGBUFFER=OFF WITH_DEBUG_SCARD=OFF WITH_DEBUG_SND=OFF WITH_DEBUG_SVC=OFF WITH_DEBUG_SYMBOLS=OFF WITH_DEBUG_THREADS=OFF WITH_DEBUG_TIMEZONE=OFF WITH_DEBUG_TRANSPORT=OFF WITH_DEBUG_TSG=OFF WITH_DEBUG_TSMF=OFF WITH_DEBUG_WND=OFF WITH_DEBUG_X11=OFF WITH_DEBUG_X11_CLIPRDR=OFF WITH_DEBUG_X11_LOCAL_MOVESIZE=OFF WITH_DEBUG_XV=OFF WITH_DIRECTFB=OFF WITH_EVENTFD_READ_WRITE=1 WITH_FFMPEG=OFF WITH_GFX_H264=OFF WITH_GPROF=OFF WITH_GSM=OFF WITH_GSTREAMER_0_10=OFF WITH_GSTREAMER_1_0=ON WITH_IPP=OFF WITH_JPEG=ON WITH_KRB5=OFF WITH_LIBRARY_VERSIONING=ON WITH_LIBSYSTEMD=ON WITH_MACAUDIO=OFF WITH_MACAUDIO=OFF WITH_MACAUDIO_AVAILABLE=0 WITH_MANPAGES=ON WITH_MBEDTLS=OFF WITH_OPENH264=OFF WITH_OPENSLES=OFF WITH_OPENSSL=ON WITH_OSS=ON WITH_PAM=ON WITH_PCSC=ON WITH_PROFILER=OFF WITH_PULSE=ON WITH_SAMPLE=OFF WITH_SANITIZE_ADDRESS=OFF WITH_SANITIZE_ADDRESS_AVAILABLE=1 WITH_SANITIZE_LEAK=OFF WITH_SANITIZE_LEAK_AVAILABLE=1 WITH_SERVER=ON WITH_SERVER_CHANNELS=ON WITH_SERVER_INTERFACE=ON WITH_SMARTCARD_INSPECT=OFF WITH_SSE2=ON WITH_THIRD_PARTY=OFF WITH_VALGRIND_MEMCHECK=OFF WITH_VALGRIND_MEMCHECK_AVAILABLE=1 WITH_WAYLAND=ON WITH_X11=ON WITH_X264=OFF WITH_XCURSOR=ON WITH_XDAMAGE=ON WITH_XEXT=ON WITH_XFIXES=ON WITH_XI=ON WITH_XINERAMA=ON WITH_XKBFILE=ON WITH_XRANDR=ON WITH_XRENDER=ON WITH_XSHM=ON WITH_XTEST=ON WITH_XV=ON WITH_ZLIB=ON
Build type:          RELWITHDEBINFO
CFLAGS:              -g -O2 -fdebug-prefix-map=/build/freerdp2-jffv_t/freerdp2-2.0.0~git20170725.1.1648deb+dfsg1=. -fstack-protector-strong -Wformat -Werror=format-security -Wdate-time -D_FORTIFY_SOURCE=2 -Wdate-time -D_FORTIFY_SOURCE=2 -fPIC -Wall -Wno-unused-result -Wno-unused-but-set-variable -Wno-deprecated-declarations -fvisibility=hidden -Wimplicit-function-declaration -Wredundant-decls -g -msse2
Compiler:            GNU, 7.3.0
Target architecture: x64

• OS version connecting to
  windows 7

Desktop (please complete the following information):

• OS: ubuntu 18.04

[akallabeth]

Could you check if it works with current nightlies?
Working with these daily (and remmina uses them too) so this seems to be already resolved.

[bmiklautz]

https://ci.freerdp.com/job/freerdp-nightly-binaries/ you find the nightly builds for testing there.

[ddombrowsky]

I will check.

[ddombrowsky]

nightly builds will not run for me, as installed by apt.

/opt/freerdp-nightly/bin/xfreerdp --help
==20631==ASan runtime does not come first in initial library list; you should either link runtime to your application or manually preload it with LD_PRELOAD.

I tried fudging LD_LIBRARY_PATH with /opt/freerdp-nightly/lib/ but that didn't work. I'll try the LD_PRELOAD var.

[ddombrowsky]

After specifying the environment, it fails with the same error on middle click:

LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libasan.so.4 /opt/freerdp-nightly/bin/xfreerdp /size:1724x960 /v:davek-win /sec:rdp /u:davek -grab-keyboard
[06:37:42:187] [20880:20881] [INFO][com.freerdp.client.common.cmdline] - loading channelEx cliprdr
[06:37:42:189] [20880:20881] [ERROR][com.freerdp.client.x11] - Mouse physical button 1 is mapped to logical button 0
[06:37:42:290] [20880:20881] [INFO][com.freerdp.gdi] - Local framebuffer format  PIXEL_FORMAT_BGRX32
[06:37:42:290] [20880:20881] [INFO][com.freerdp.gdi] - Remote framebuffer format PIXEL_FORMAT_RGB16
[06:37:42:307] [20880:20881] [INFO][com.winpr.clipboard] - initialized POSIX local file subsystem
[06:37:47:304] [20880:20881] [INFO][com.freerdp.core] - ERRINFO_INVALID_INPUT_PDU_MOUSE (0x000010D0):(a) A Slow-Path Mouse Event (section 2.2.8.1.1.3.1.1.3) or Extended Mouse Event (section 2.2.8.1.1.3.1.1.4) has been received with an invalid pointerFlags field.
(b) A Fast-Path Mouse Event (section 2.2.8.1.2.2.3) or Fast-Path Extended Mouse Event (section 2.2.8.1.2.2.4) has been received with an invalid pointerFlags field.
[06:37:47:304] [20880:20881] [ERROR][com.freerdp.core.transport] - BIO_read returned a system error 104: Connection reset by peer

This is FreeRDP version 2.0.0-dev5 (fc12b9336)

[akallabeth]

@ddombrowsky Ok, the line [06:37:42:189] [20880:20881] [ERROR][com.freerdp.client.x11] - Mouse physical button 1 is mapped to logical button 0 is interesting.
You have some local mouse button mapping magic activated?

[ddombrowsky]

@akallabeth I use a logi marble mouse, with a defined button mapping, yes. I use xinput to map the left "page down" button to the middle button.

xinput set-button-map "Logitech USB Trackball" 1 0 3 4 5 6 7 2 9

[akallabeth]

@ddombrowsky Are you sure a button id of 0 is allowed? This seems to cause the problems.

[ddombrowsky]

@akallabeth yes. From the manpage:

      --set-button-map device map_button_1 [map_button_2 [...]]
               Change the button mapping of device. The buttons are specified in physical order (starting with button 1) and are mapped to the logical button provided. 0 disables a
               button. The default button mapping for a device is 1 2 3 4 5 6 etc.

I also tried mapping it to "8" and that fails with a similar error message.

[bmiklautz]

@ddombrowsky would you test with xfreerdp without the mapping? That could help to pin down the problem.

[ddombrowsky]

@bmiklautz Yes, if I change the xinput mapping to default, xfreerdp does not disconnect. The button defaults to the "back" button and works as designed (which, in my opinion is a bug not a feature, but that's my complaint to logitech). Without the mapping I have no middle mouse button, so I can't test that.

[akallabeth]

@ddombrowsky I've checked and the mouse button remapping code in xfreerdp was not really working for all butttons.
Extended that and ignoring all unmapped buttons now.

[bmiklautz]

@ddombrowsky @akallabeth 's fix is already merged and should be in tomorrows nightly build. Would nice if you can verify if your problem is fixes with it and close the issue.

[akallabeth]

Fix was merged and is working here.
Reopen if still not fixed.

[ddombrowsky]

I can confirm this fixed. The button now doesn't do anything, but that's better than a disconnect.

Thank you all for your help.

[akallabeth]

@ddombrowsky thx for the feedback.
The button you mean your second one set to 0? Isn't that what it is supposed to do as 0 stands for deactivated?

[ddombrowsky]

@akallabeth The logi marble mouse has no "middle" mouse button. You need to use software to decide which button to use (if any) as the middle one. On windows, there's a separate hook in the driver (I think). On linux, which I use, it's an xinput setting which tells X11 to return the code for button "2" when button 8 is clicked. I'm not sure why that code isn't being sent through to xfreerdp when it works in all other programs.
A very specific use case, I understand. It really affects only the universe of people using this specific mouse on linux, which is probably just me :)

[ddombrowsky]

@akallabeth Sure enough, right after I posted that, I tried +unmap-buttons which does exactly what I was looking for. Now that the disconnect is fixed, that flag works and everything is right with the world!