Do not keep bus, address, vid and pid of the USB device
in SpiceUsbDeviceInfo structure. Getters for these properties
can easily obtain them from respective backend device.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>

No need to add USE_POLKIT block on channel-usbredir code

Signed-off-by: Victor Toso <victortoso@redhat.com>

As a step toward possibility to present emulated USB devices
to the guest, we remove the knowledge about libusb and
usbredirhost (which depends on libusb) from all the modules
and concentrate it in one (usb backend) which presents
abstract USB objects and internal API to all other modules.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>

In this patch series, spice_usbutil_libusb_strerror() becomes useless
in favor of libusb_error_name(). Instead of making the helper function
to call the native one, drop the usage of the helper function
entirely.

Signed-off-by: Victor Toso <victortoso@redhat.com>

libusb has libusb_error_name procedure that returns name
for any error that libusb may return, so we do not need
to analyze error values by ourselves.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>

There is no need to direct interaction with polkit in
channel-usbredir. Let's keep all the access control related code to
usb-acl-helper APIs.

This patch moves the g_task_run_in_thread() to
spice_usb_acl_helper_open_acl_async() to keep code behaving similarly
to what we had before this patch.

Signed-off-by: Victor Toso <victortoso@redhat.com>

Explanation on this mutex in commit 4eea5f3a in 2016-03-24 by Kirill
Moizik <kmoizik@redhat.com>. This is missing in the polkit code path
and its the major difference with _open_device_async_cb() callback.

Next patch will remove _open_device_async_cb() to keep code path for
access control mandatory there so add this lock/unlock now with some
context on why.

Signed-off-by: Victor Toso <victortoso@redhat.com>

This will allow us to simplify channel-usbredir code, which contains a
few #ifdef USE_POLKIT blocks while the code itself is only calling
the usb-acl-helper APIs.

Let's keep the API working even when polkit is disabled.

This is a preparation patch, after this patch, the channel-usbredir.c
still only calls usb-acl-helper API's when polkit is enabled. The next
patch adds the missing bits so the API keeps working without Polkit.

Signed-off-by: Victor Toso <victortoso@redhat.com>

Create helper exec_usb_acl_helper_bin() to handle the execution of
spice-client-glib-usb-acl-helper binary. This bin is only compiled in
spice-gtk if Polkit is enabled.

This is a preparation patch for enabling the build of usb-acl-helper.c
when Polkit is disabled.

Signed-off-by: Victor Toso <victortoso@redhat.com>

Victor Toso authored May 07, 2019 and Frediano Ziglio committed May 07, 2019

Not really interesting to keep tracking what is inside or not
USE_USBREDIR on channel-usbredir itself. When usbredir is disabled,
just dummy functions are needed with spice-client.h include.

After this patch, on channel-usbredir.c, we have a single #ifdef
USE_USBREDIR and the #else block is defined in the end of the file.

This patch also adds #ifdef USE_USBREDIR for channel-usbredir-priv.h
as none those functions are defined when usbredir is disabled.

Signed-off-by: Victor Toso <victortoso@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>

Otherwise we'll get

    g_type_add_instance_private: assertion 'private_size > 0' failed

Signed-off-by: Victor Toso <victortoso@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>

Headers macros (HAVE_*_H) are guarding some include in the code.

Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Victor Toso <victortoso@redhat.com>

Jeremy White authored Apr 30, 2019 and Frediano Ziglio committed May 01, 2019

This mitigates a fairly rare problem we see with our kiosk mode clients.
That is, normally if something goes wrong with a client connection
(e.g. the session is killed, or the server is restarted ), the kiosk will
exit on disconnect, and we get a chance to retry the connection, or
present the user with a 'server down' style message.

But in the case of a serious network problem or a server hard power
cycle (i.e. no TCP FIN packets can flow), our end user behavior is not
ideal - the kiosk appears to hang solid, requiring a power cycle.

That's because we've got the stock keepalive timeouts, or about 2 hours
and 11 minutes, before the client sees the disconnect.

This change will cause the client to recognize the server has vanished
without a TCP FIN after 75 seconds.

See this thread:
  https://lists.freedesktop.org/archives/spice-devel/2017-March/036553.html

As well as this bug:
  https://bugzilla.redhat.com/show_bug.cgi?id=1436589



Signed-off-by: Jeremy White <jwhite@codeweavers.com>

Roman Bogorodskiy authored Apr 27, 2019 and Frediano Ziglio committed Apr 27, 2019

Currently, spice-gtk uses PKG_CHECK_MODULES(SSL, openssl) to detect
openssl CFLAGS and LIBS. And spice-common defines and uses
SPICE_CHECK_OPENSSL macro which calls
PKG_CHECK_MODULES(OPENSSL, openssl).

This means that in order to override openssl CFLAGS or LIBS a user will
have to set both OPENSSL_(CFLAGS|LIBS) and SSL_(CFLAGS|LIBS).

To make this more consistent, make spice-gtk use the
SPICE_CHECK_OPENSSL macro from spice-common.

Signed-off-by: Roman Bogorodskiy <bogorodskiy@gmail.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>

Clean all git modules, not only the main one.

Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Christophe de Dinechin <dinechin@redhat.com>

"devices" is translated to "dispositivi" (plural) not "dispositivo"
(singular).
In other occurrences "Some devices" is correctly translated to
"Qualche dispositivo".

Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Snir Sheriber <ssheribe@redhat.com>

build-aux/meson/check-spice-common was removed by commit
57645fac
("build-sys: remove meson/check-spice-common")

Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Eduardo Lima (Etrunko) <etrunko@redhat.com>

Uri Lublin authored Apr 23, 2019 and Frediano Ziglio committed Apr 24, 2019

Signed-off-by: Uri Lublin <uril@redhat.com>

Uri Lublin authored Apr 23, 2019 and Frediano Ziglio committed Apr 24, 2019

Fix out-of-tree builds.

Uri Lublin (1):
      codegen Makefile: add common/ to --include client_marshallers.h

David Jaša authored Apr 18, 2019 and Frediano Ziglio committed Apr 22, 2019

Signed-off-by: David Jaša <djasa@redhat.com>
Acked-by: Jakub Janku <jjanku@redhat.com>

Jakub Janků authored Apr 18, 2019 and Frediano Ziglio committed Apr 18, 2019

Translation strings for spice options are already present
in the respective PO files.

g_option_group_set_translation_domain() call needs to be made,
so that the strings marked with N_() actually get translated at runtime.

Signed-off-by: Jakub Janků <jjanku@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>

Jakub Janků authored Apr 17, 2019 and Frediano Ziglio committed Apr 18, 2019

_() should not be used in conjunction with ngettext(),
otherwise it's impossible to properly translate the message into
languages that have multiple plural forms, such as Czech.

Additionally, spice-gtk must use dngettext instead of ngettext,
otherwise the translation strings are searched within a wrong domain.

Fix ngettext usage in usb-device-widget.c and
update the PO files accordingly.

Signed-off-by: Jakub Janků <jjanku@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>

Spice gtk-session would try to synchronize the modifiers state whenever
the keymap changes, but doing so is inherently racy.

While the there is a keyboard grab in effect, all key events are
forwarded to the guest, hence all modifiers key press get processed by
the kernel on the guest.

Trying to synchronize the modifiers will generate additional key press/
release events which will result in the opposite effect and effectively
desynchronize the modifier states in the guest.

Synchronizing modifiers from the host should therefore be limited to
focus change, as actual press/release events might have occurred without
the guest knowing. Otherwise, no need to synchronize the modifiers.

Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Acked-by: Victor Toso <victortoso@redhat.com>

This brings in the following changes:

Christophe Fergeau (8):
      canvas_base: Fix variable shadowing warning
      backtrace: Add missing include
      lz: Don't try to print uninitialized variable
      build: Add missing G_GNUC_PRINTF annotations
      build: Update verify.h to latest version
      test-marshallers: Fix header guard
      quic: Fix QUIC_VERSION definition
      log: Let gcc know about the logging macros which abort

Eduardo Lima (Etrunko) (5):
      meson: Use underscorify() instead of split()
      meson: Bump libcacard requirement to 2.5.1
      meson: switch smartcard option to auto feature
      meson: Bump requirement to 0.48
      meson: Make targets list store dictionaries instead of lists

Frediano Ziglio (31):
      Remove obsolete Autoconf macros
      log: remove deprecated SPICE_ABORT_LEVEL support
      log: remove deprecated SPICE_DEBUG_LEVEL support
      Obsolete Glib cleanup
      codegen: Document ptr_array attribute
      codegen: Use a better type for pointer converted to integer
      codegen: Reduce indentation
      codegen: Fix c_type result for TypeAlias
      codegen: Check wrong attribute
      codegen: Add a test for attribute combination
      mem: Fix compile error if alignment-checks option is used
      log: Remove useless includes
      proto: Remove obsolete TunnelChannel
      protocol: Add a dummy TunnelChannel
      messages: Remove fields not used by the protocol
      codegen: Remove support for --ptrsize
      codegen: Rename --prefix parameter to --suffix
      ci: Remove dependencies from copr build
      build: Remove snd_codec.h from EXTRA_DIST
      test-ssl-verify: Improve subject_to_x509_name coverage
      meson: Remove some useless checks
      codegen: Factor out a function to write output file
      codegen: Generate headers while generating code
      codegen: Allows to generate C declarations automatically
      Allow to generate C declarations for spice.proto
      Generate automatically most C message declarations
      recorder: Update
      build: Add new check for recorder library
      proto: Fix typo in message name
      meson: Fix dependency of generated files
      meson: Do not build generated files twice

Marc-André Lureau (1):
      docs: add spice URI scheme

Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Eduardo Lima (Etrunko) <etrunko@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 27, 2019

Remove unused code related to GUdevDevice.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 27, 2019

Unify SpiceUsbDeviceInfo for Linux and Windows by using
persistent libusb_device. Probably there was some problem
doing it long time ago with winusb, but at least with
existing libusb and UsbDk the the libusb device can be
accessed as long as it is referenced.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 27, 2019

Change the signal to report libusb_device instead of GUdevDevice
object. This simplifies the callback procedure in usb-dev-manager
and avoids additional reenumeration immediately after reenumeration
that already done by win-usb-dev.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

When getting a new list of USB devices after a WM_DEVICECHANGE event,
libusb_device which were already present before the event may be
represented by a different instance than the one we got in the new list.

Since other layers of spice-gtk may be using that older instance of
libusb_device, this commit makes sure that we always keep the older
instance in GUdevClient::udev_list.

At the moment, this should not be making any difference, but will make
things more consistent later on.

Based on a patch from Yuri Benditovich.

Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 25, 2019

Change internal device list to maintain libusb devices
instead of GUdevDevice objects. Create temporary
GUdevDevice object only for indication to usb-dev-manager.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 21, 2019

When processing list of USB devices, avoid allocating memory
for devices which later will be skipped. Use existing libusb
struct first to check whether the device shall be excluded.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 21, 2019

Changing signal definition from (boxed-boxed) to (pointer,int).
There is no need for additional referencing of GUdevDevice
object before signal callback. Second parameter (action) is
FALSE for device removal and TRUE for device addition.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 21, 2019

Discard the optimization of initial device enumeration.
Just after connection to the 'uevent' signal, request to report all the
devices one by one as if they are inserted.
Further commits will remove device enumeration in usb-dev-manager
completely.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 21, 2019

Remove unused g_udev_device_get_sysfs_attr.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 21, 2019

If when calling handle_dev_change, the new device has the same
bus:address as an existing device cached in GUdevClient::udev_list, then
an "uevent" signal will not be emitted to notify about the new device.
For example, when due to some reason the 'redirecting' property is set
for a long time and during this time one device is unplugged and then
replugged.
This commit makes the device comparison more strict by checking not only
bus:addr, but also vid:pid.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 21, 2019

In initial device enumeration, hotplug notification can be
called twice with the same libusb device. For details, see
http://libusb.sourceforge.net/api-1.0/group__libusb__hotplug.html#ga00e0c69ddf1fb1b6774dc918192e8dc7


This commit filters out devices that are already present in the list.
It also removes the identical call in spice_usb_device_manager_add_udev,
which adds devices under Windows.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 21, 2019

Do not create our own libusb context in usb-device-manager.
Reuse existing context created by win-usb-dev instead,
this will later allow usb-device-manager to use libusb
devices enumerated in win-usb-dev.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 21, 2019

Removing unused parameter from GUdevClient constructor.
This code is used only in Windows builds.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

Yuri Benditovich authored Mar 19, 2019 and Christophe Fergeau committed Mar 21, 2019

Replacing USE_GUDEV with G_OS_WIN32 anywhere. GUDEV simulation
is used only in Windows build.

Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>

spice-server does not deal properly with multiple smartcard readers,
only the first one will be working. Add a warning when this happens to
make it easier to diagnose such issues.

Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Acked-by: Marc-André Lureau <marcandre.lureau@gmail.com>