Mir

benchmarks/frame-uniformity/touch_producing_server.cpp (+0/-1)
include/platform/mir/input/platform.h (+2/-2)
include/test/mir_test_framework/executable_path.h (+1/-0)
src/include/server/mir/default_server_configuration.h (+1/-11)
src/include/server/mir/input/input_probe.h (+48/-0)
src/platform/graphics/platform_probe.cpp (+5/-1)
src/platforms/evdev/platform_factory.cpp (+1/-1)
src/platforms/mesa/server/x11/input/input.cpp (+1/-1)
src/server/input/CMakeLists.txt (+1/-0)
src/server/input/default_configuration.cpp (+16/-39)
src/server/input/input_probe.cpp (+104/-0)
src/server/symbols.map (+1/-1)
tests/mir_test_framework/CMakeLists.txt (+1/-0)
tests/mir_test_framework/executable_path.cpp (+10/-0)
tests/mir_test_framework/fake_input_device_impl.cpp (+1/-1)
tests/mir_test_framework/fake_input_server_configuration.cpp (+0/-1)
tests/mir_test_framework/input_testing_server_options.cpp (+0/-1)
tests/mir_test_framework/stub_input.cpp (+2/-2)
tests/mir_test_framework/stub_input_platform.cpp (+1/-1)
tests/umock-acceptance-tests/test_libinput.cpp (+8/-14)
tests/unit-tests/input/CMakeLists.txt (+1/-0)
tests/unit-tests/input/test_input_platform_probing.cpp (+161/-0)

To merge this branch:

bzr merge lp:~andreas-pokorny/mir/load-all-supported-input-platforms

Related bugs:

Bug #1479463: Mir touchpad does not support relative mouse events	High	Fix Released
Bug #1517133: [regression] Mouse speed in Unity8 is much lower than elsewhere (to the point of useless)	High	Fix Released

Link a bug report

Reviewer	Review Type	Date Requested	Status
PS Jenkins bot (community)	continuous-integration	2015-10-14	Approve on 2015-11-25
Cemil Azizoglu (community)			Approve on 2015-11-20
Alan Griffiths		2015-10-14	Abstain on 2015-11-20
Daniel van Vugt		2015-10-14	Needs Fixing on 2015-10-28
Chris Halse Rogers		2015-10-14	Approve on 2015-10-19
Review via email: mp+274421@code.launchpad.net

This proposal supersedes a proposal from 2015-10-05.

Commit message

Add automatic searching and probing for input platforms

With this specifying the input platform via --platform-input-lib is no longer necessary, unless the stub input platform that supports attaching fake devices is required

Description of the change

The most notable change of this MP will make the libinput and x11 platform work without additional command line parameter.

This MP adds probing for input platforms. mirserver will now search inside the default platform-path for loadable modules that fulfil the input platform ABI. Then only those modules that claim to be better than 'dummy' will be used. Hence input-stub.so will not be loaded by default, but requires loading through the platform-input-lib parameter.

Next step: remove input reader and event hub..

WRT to testing this MP - or the evdev input platform in general: pick silo-56 for phone images. This will still have problems with the clock (which is fixed by lp:~andreas-pokorny/mir/use-realtime-clock-for-input-events). Xenial already has the right libinput package.

PS Jenkins bot (ps-jenkins) wrote on 2015-10-05: Posted in a previous version of this proposal

FAILED: Continuous integration, rev:2992
http://jenkins.qa.ubuntu.com/job/mir-ci/5149/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-vivid-i386-build/4287
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-clang-vivid-amd64-build/3194
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-vivid-touch/4229/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-wily-touch/187/console
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-wily-amd64-ci/1303
        deb: http://jenkins.qa.ubuntu.com/job/mir-wily-amd64-ci/1303/artifact/work/output/*zip*/output.zip
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-wily-i386-ci/187/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/4230/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-wily-armhf/188/console

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5149/rebuild

review: Needs Fixing (continuous-integration)

PS Jenkins bot (ps-jenkins) wrote on 2015-10-05: Posted in a previous version of this proposal

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5154/rebuild

review: Needs Fixing (continuous-integration)

PS Jenkins bot (ps-jenkins) wrote on 2015-10-06: Posted in a previous version of this proposal

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5158/rebuild

review: Needs Fixing (continuous-integration)

Chris Halse Rogers (raof) wrote on 2015-10-08: Posted in a previous version of this proposal

+ mir::log_info("Found input driver: %s", desc->name);

Could you also log driver version please?

+ return std::move(platforms);

Pessimisation: this breaks return value optimisation. When the Wily clang builds work again this will be an error.

+ EXPECT_THAT(platforms.size(), Eq(2)); // we cannot disable the default evdev platform with only umock dev is in place
+ EXPECT_THAT(platforms[0].get(), AnyOf(OfType<mi::evdev::Platform>(), OfType<mi::X::XInputPlatform>()));
+ EXPECT_THAT(platforms[1].get(), AnyOf(OfType<mi::evdev::Platform>(), OfType<mi::X::XInputPlatform>()));

Seems like you could replace OfType with OfPointerType and then use

EXPECT_THAT(platforms, UnorderedElementsAre(OfPointerType<mi::evdev::Platform>(), OfPointerType<mi::X::XInputPlatform>());

Which I think is both more natural and also is a slightly stronger test.

========================
Non-review comment:
+ mtf::UdevEnvironment env; // TODO replace with mock_libevdev und mock_libudev when the new input reading stack is the only in use.

Heads up: I think that doing this will provide negative value. Also, und→and :)

review: Needs Fixing

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-08: Posted in a previous version of this proposal

Fixed all of the above.

> ========================
> Non-review comment:
> + mtf::UdevEnvironment env; // TODO replace with mock_libevdev und
> mock_libudev when the new input reading stack is the only in use.
>
> Heads up: I think that doing this will provide negative value. Also, und→and
> :)

I want to simulate the cases no devices found, devices found and some or all can be opened, devices found and none can be opened. I cannot do that with umockdev...

PS Jenkins bot (ps-jenkins) wrote on 2015-10-08: Posted in a previous version of this proposal

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5205/rebuild

review: Needs Fixing (continuous-integration)

Chris Halse Rogers (raof) wrote on 2015-10-08: Posted in a previous version of this proposal

On Thu, Oct 8, 2015 at 5:56 PM, Andreas Pokorny
<email address hidden> wrote:
> Fixed all of the above.
>
>> ========================
>> Non-review comment:
>> + mtf::UdevEnvironment env; // TODO replace with mock_libevdev
>> und
>> mock_libudev when the new input reading stack is the only in use.
>>
>> Heads up: I think that doing this will provide negative value.
>> Also, und→and
>> :)
>
> I want to simulate the cases no devices found, devices found and some
> or all can be opened, devices found and none can be opened. I cannot
> do that with umockdev...

You can do all except the last, and the last would be a simple patch to
umockdev.

PS Jenkins bot (ps-jenkins) wrote on 2015-10-08: Posted in a previous version of this proposal

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5206/rebuild

review: Needs Fixing (continuous-integration)

Chris Halse Rogers (raof) wrote on 2015-10-08: Posted in a previous version of this proposal

LGTM

review: Approve

Alan Griffiths (alan-griffiths) wrote on 2015-10-09: Posted in a previous version of this proposal

Looks like infrastructure glitch:

inflating: package_archive/archive/work/output/libmirclient-debug-extension1_0.17.0bzr2998pkg0wily4303+autopilot0_armhf.deb
error: closed
[timestamp] Start PPAs and local Archive : 2015-10-08T10:09:40+0000
rm: cannot remove 'ppa_sources': No such file or directory

~~~~

(Non blocking) looks like we need to bump the server ABI

Alan Griffiths (alan-griffiths) wrote on 2015-10-09: Posted in a previous version of this proposal

"The most notable change of this MP will make the libinput and x11 platform work without additional command line parameter."

Am I expecting too much?

$ sudo DISPLAY= bin/mir_demo_server --vt 1 --arw-file&
$ bin/mir_demo_server --file mir-socket --host-socket /tmp/mir_socket
[Switch to vt 1]
[Switch to vt 7]

ERROR: /home/alan/display_server/mir/src/platforms/mesa/server/common/display_helpers.cpp(256): Throw in function int mir::graphics::mesa::helpers::DRMHelper::open_drm_device(const std::shared_ptr<mir::udev::Context>&)
Dynamic exception type: boost::exception_detail::clone_impl<boost::exception_detail::error_info_injector<std::runtime_error> >
std::exception::what: Error opening DRM device
13, "Permission denied"

It looks as though the nested server is selecting the wrong drivers. (It should "agree" with the host.)

NB I can workaround like this:

$ DISPLAY= bin/mir_demo_server --file mir-socket --host-socket /tmp/mir_socket --launch bin/mir_demo_client_animated_cursor

review: Needs Information

Alan Griffiths (alan-griffiths) wrote on 2015-10-09: Posted in a previous version of this proposal

OK, I don't think the auto-configuration is quite there yet. But this is a step forwards.

I guess there will be an ABI break needed for 0.18.

review: Approve

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-09: Posted in a previous version of this proposal

Oh .. I wasnt aware there is a mismatch between the x11 graphics and input probing.
Soooo.. a host-socket setting should win over a DISPLAY env, shouidnt it?

That should be a simple fix in graphics.cpp.

PS Jenkins bot (ps-jenkins) wrote on 2015-10-09: Posted in a previous version of this proposal

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5220/rebuild

review: Approve (continuous-integration)

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-09: Posted in a previous version of this proposal

> Oh .. I wasnt aware there is a mismatch between the x11 graphics and input
> probing.
> Soooo.. a host-socket setting should win over a DISPLAY env, shouidnt it?
>
> That should be a simple fix in graphics.cpp.

Nested.. not as simple as I thought..

Daniel van Vugt (vanvugt) wrote on 2015-10-12: Posted in a previous version of this proposal

Landed and then reverted:
* Modifiers don't work any more (can't move/resize windows using Alt)
* Default mouse speed too high
* A little too many log messages come from this.

Only the first issue is a blocker.

review: Needs Fixing

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-12: Posted in a previous version of this proposal

> Landed and then reverted:
> * Modifiers don't work any more (can't move/resize windows using Alt)

Yes modifiers are per device right now .. which isnt what we want within the dispatcher. Working on a proper abstraction/fix for that.

> * Default mouse speed too high

Yay! So mouse acceleration is working. Configuration branches are on the way..

> * A little too many log messages come from this.
>
> Only the first issue is a blocker.

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-14: Posted in a previous version of this proposal

> Landed and then reverted:
> * Modifiers don't work any more (can't move/resize windows using Alt)
> * Default mouse speed too high
> * A little too many log messages come from this.
>
> Only the first issue is a blocker.

This should fix the issue..
https://code.launchpad.net/~andreas-pokorny/mir/combine-modifier-state-accross-devices/+merge/274372

PS Jenkins bot (ps-jenkins) wrote on 2015-10-14:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5258/rebuild

review: Needs Fixing (continuous-integration)

Chris Halse Rogers (raof) wrote on 2015-10-19:

I'm still happy with this.

review: Approve

Alan Griffiths (alan-griffiths) wrote on 2015-10-21:

Nits:

+#include "mir/module_deleter.h"

#include <memory>
#include <string>
+#include <vector>

The other changes to src/include/server/mir/default_server_configuration.h don't justify these additions.

~~~~

- virtual std::shared_ptr<input::Platform> the_input_platform();
     virtual std::shared_ptr<dispatch::MultiplexingDispatchable> the_input_reading_multiplexer();
     virtual std::shared_ptr<input::InputDeviceRegistry> the_input_device_registry();
     virtual std::shared_ptr<input::InputDeviceHub> the_input_device_hub();
@@ -384,8 +385,7 @@
     CachedPtr<input::CompositeEventFilter> composite_event_filter;
     CachedPtr<input::InputManager> input_manager;
     CachedPtr<input::SurfaceInputDispatcher> surface_input_dispatcher;
- CachedPtr<input::DefaultInputDeviceHub> default_input_device_hub; // currently not used by default
- CachedPtr<input::Platform> input_platform; // currently not used by default
+ CachedPtr<input::DefaultInputDeviceHub> default_input_device_hub;

You can also lose the declaration of Platform on line 135.

~~~~~

Probably not something to address in this MP but...

+#ifdef MIR_BUILD_PLATFORM_MESA_X11
+ ON_CALL(mock_x11, XOpenDisplay(_)).WillByDefault(Return(nullptr));
+#endif

Should the test setup really depend on the *build* platform?

review: Needs Fixing

PS Jenkins bot (ps-jenkins) wrote on 2015-10-22:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5371/rebuild

review: Needs Fixing (continuous-integration)

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-23:

This is currently blocked by:
https://code.launchpad.net/~andreas-pokorny/mir/move-cookie-test-to-acceptance-tests/+merge/275502

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-26:

^ No longer blocking.. brandon made a similar MP and that landed today.

PS Jenkins bot (ps-jenkins) wrote on 2015-10-26:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5400/rebuild

review: Needs Fixing (continuous-integration)

PS Jenkins bot (ps-jenkins) wrote on 2015-10-26:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5405/rebuild

review: Needs Fixing (continuous-integration)

PS Jenkins bot (ps-jenkins) wrote on 2015-10-27:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5415/rebuild

review: Needs Fixing (continuous-integration)

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-27:

^
demo server and acceptances tests crash...

Daniel van Vugt (vanvugt) wrote on 2015-10-28:

OK, I've invested time in testing and lots needs fixing unfortunately. At least three are show-stoppers...

(1) Mouse now moves too fast. Even my slowest mouse.

(2) "evdev-input" log messages on startup report nothing useful. Can we remove or change them to report actual device descriptions?

(3) Input resampling has been lost (can only verify on desktop due to #5). It was introduced on 2014-08-29 but we only just fixed Unity8 this week to utilize it and achieve halved latency. It's taken 14 months to come to fruition this week, so please try to avoid regressing on it.
Test: env MIR_CLIENT_INPUT_RATE=59 mir_demo_client_fingerpaint
Verify only one square painted per frame.

(4) Input events are strangely erratic and no longer smooth (even when not resampled), compared to lp:mir. Compare 'mir_demo_client_fingerpaint' using this branch with lp:mir and you will see a lot of jitter when using this branch. Weirdly, MIR_CLIENT_INPUT_RATE=59 does affect the result but only in that it makes things even more erratic, and not limited to 59Hz either (#3).

(5) Servers crash on startup on mako (vivid rc-proposed):
[1446022093.989552] <WARNING> mircommon: Failed to load module: bin/../lib/server-modules/input-evdev.so.4 (error was:libinput.so.10: cannot open shared object file: No such file or directory)
[1446022093.989948] mircommon: Loading module: bin/../lib/server-modules/graphics-dummy.so
[1446022094.000600] mircommon: Loading module: bin/../lib/server-modules/input-stub.so
ERROR: /home/dan/bzr/mir/tmp.all/src/server/input/default_configuration.cpp(328): Throw in function mir::DefaultServerConfiguration::the_input_manager()::<lambda()>
Dynamic exception type: N5boost16exception_detail10clone_implINS0_19error_info_injectorISt13runtime_errorEEEE
std::exception::what: No input platforms found

review: Needs Fixing

Daniel van Vugt (vanvugt) wrote on 2015-10-28:

(5) Fixed by installing package 'libinput10' but ...

(6) mir_demo_client_fingerpaint on mako paints very small, indicating the touch_major/touch_minor values are wrong using this branch.

Daniel van Vugt (vanvugt) wrote on 2015-10-28:

(7) Multi-touch gestures no longer work (on mako). Try moving a window in mir_proving_server using three fingers and it mostly does nothing, or has a mind of its own. If you run 'mir_demo_client_target' you can see what the problem is with multiple touches: some go missing mid-gesture (flickering targets).

Daniel van Vugt (vanvugt) wrote on 2015-10-29:

Although (1) sounds unimportant and not like a blocker, if you fix (1) and remove the default acceleration that libinput introduces, then that might actually fix (4) as well.

Alan Griffiths (alan-griffiths) wrote on 2015-10-29:

My issue above fixed. Looks like enough issues remain to be dealt with for me to wait before looking again.

review: Abstain

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-29:

(1) This is the default setting of libinput - 1.1.0 will change that i.e use differnt profiles depending on the mouse or device type. I found the default setting just good..

(2)

(3) There is a known problem with libinput-0.21 in the overlay ppas. Patches are w

(4)

(5) better than hanging the system

(6) they are reported like that from the kernel.. in general to small.. but d

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-29:

oops hit reply to soon..

(3) (4) require a libinput patch that has not yet landed in archive or those ppas. This might be the reason for (7) too.

(6) proper touch screen drivers should report resolution values to allow scaling values to actual units. All touch screen drives that we use only report a resolution of zero. That invalid setting is handled inside libinput and it returns the plain value reported by the kernel. Those values as you have noticed are too small. To come up with proper values we need to have a device dependent scaling. I dont see

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-29:

.. any other client than finger paint use those values at the moment. I do hope that since they are now exposed via libinput that the other display servers and toolkits start to use those. Within the Qt event feeder they are mapped to a rectangular region.. which isnt really accurate either.

Andreas Pokorny (andreas-pokorny) wrote on 2015-10-30:

just checked mako again: (3) (4) and (7) is the same cause - and needs the current libinput patch-set as in:

https://bugs.launchpad.net/ubuntu/+source/libinput/+bug/1504168

Daniel van Vugt (vanvugt) wrote on 2015-11-02:

(3) and (4) were discovered with a mouse on desktop. So any touchscreen fixes will not help (3) and (4). It appears likely that fixing (1), while not a major issue itself, will probably fix (3) and (4). Because (1) is the only reason I can think of for libinput to tinker with the mouse event stream.

Andreas Pokorny (andreas-pokorny) wrote on 2015-11-03:

(2) I just improved logs with lp:~andreas-pokorny/mir/better-input-device-related-logs
We should really consider having something like a severity filter..

Ok I misunderstood (3) and (4) so what I just tried on lp:mir was:

A: Resampling with input-evdev:
export MIR_CLIENT_INPUT_RATE=59
mir_demo_server --platform-input-lib ../lib/server-modules/input-evdev.so.4 --launch-client=./mir_demo_client_fingerpaint

B: Resampling with android input
export MIR_CLIENT_INPUT_RATE=59
mir_demo_server --launch-client=./mir_demo_client_fingerpaint

C: No resampling with input-evdev
export MIR_CLIENT_INPUT_RATE=0
mir_demo_server --platform-input-lib ../lib/server-modules/input-evdev.so.4 --launch-client=./mir_demo_client_fingerpaint

D: No resampling with android input
export MIR_CLIENT_INPUT_RATE=0
mir_demo_server --launch-client=./mir_demo_client_fingerpaint

using a cheap USB MS mouse on top of libinput-1.1.0 with my patches

A and B look nearly identical - mouse moves slightly faster than with android input but there are ugly gaps/jumps .. in both cases every 5 or 10 seconds in the trace drawn by fingerpaint.

C and D look nearly identical - mouse moves slightly faster than with android input but there in both cases the trace displayed is smooth and sticks close to the cursor. Additionally there is no odd jump in both the data sets. As opposed to the resampled cases the plotted squares have a similar distance and give a good indication of mouse movement.

I repeated A and B a few times and I was able to reproduce the odd jumps for A and B sometimes. It seems that the resampling is very sensitive to scheduling. My conclusion from that:
we should make sure that we only use resampling when we have a horrible input device (driver) or when the client is unable to process the user input fast enough. Additionally we should spend the next 14 month with verifying that the algorithm yields proper results.

PS: I did mess with android input in the past and found the bug that disabled the pointer acceleration:
https://code.launchpad.net/~andreas-pokorny/mir/activate-pointer-acceleration
Not sure if it still works..

Daniel van Vugt (vanvugt) wrote on 2015-11-04:

AFAIK, resampling uses the original event timestamps which in android-input are accurate because they come directly from the kernel. So if those timestamps are being faked in libinput then that might explain the irregularity and sensitivity to scheduling. Better check that libinput (and Mir itself) is not inserting new timestamps. Our input reports also rely on those timestamps being the original event time from the kernel.

It's also worth noting my mouse setting is 1000Hz (kernel param usbhid.mousepoll=1). You might need that to notice some issues.

Andreas Pokorny (andreas-pokorny) wrote on 2015-11-04:

> AFAIK, resampling uses the original event timestamps which in android-input
> are accurate because they come directly from the kernel. So if those
> timestamps are being faked in libinput then that might explain the
> irregularity and sensitivity to scheduling. Better check that libinput (and
> Mir itself) is not inserting new timestamps. Our input reports also rely on
> those timestamps being the original event time from the kernel.
>
> It's also worth noting my mouse setting is 1000Hz (kernel param
> usbhid.mousepoll=1). You might need that to notice some issues.

Well whatever you observe I cannot observe that. I see negative effects of resampling with and without libinput involved.

PS Jenkins bot (ps-jenkins) wrote on 2015-11-04:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5499/rebuild

review: Approve (continuous-integration)

Andreas Pokorny (andreas-pokorny) wrote on 2015-11-05:

There is now a silo for vivid overlay and this newer libinput package is about to land in xenial:
https://launchpad.net/~ci-train-ppa-service/+archive/ubuntu/landing-056/+packages

With that the known touchscreen troubles are gone - mouse is still fast though (as actually requested by many people)

PS Jenkins bot (ps-jenkins) wrote on 2015-11-09:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5523/rebuild

review: Needs Fixing (continuous-integration)

PS Jenkins bot (ps-jenkins) wrote on 2015-11-09:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5532/rebuild

review: Approve (continuous-integration)

Alan Griffiths (alan-griffiths) wrote on 2015-11-13:

-if(MIR_TEST_PLATFORM STREQUAL "mesa-kms" OR MIR_TEST_PLATFORM STREQUAL "mesa-x11")
+if(MIR_BUILD_PLATFORM_MESA_KMS OR MIR_BUILD_PLATFORM_MESA_X11)
...
+#ifdef MIR_BUILD_PLATFORM_MESA_X11
+ ON_CALL(mock_x11, XOpenDisplay(_)).WillByDefault(Return(nullptr));
+#endif

Should the test setup really depend on the *build* platform? (Or is this poor naming - should MIR_BUILD_PLATFORM_* really be MIR_TARGET_PLATFORM_*)

review: Needs Information

Andreas Pokorny (andreas-pokorny) wrote on 2015-11-13:

> -if(MIR_TEST_PLATFORM STREQUAL "mesa-kms" OR MIR_TEST_PLATFORM STREQUAL
> "mesa-x11")
> +if(MIR_BUILD_PLATFORM_MESA_KMS OR MIR_BUILD_PLATFORM_MESA_X11)
> ...
> +#ifdef MIR_BUILD_PLATFORM_MESA_X11
> + ON_CALL(mock_x11, XOpenDisplay(_)).WillByDefault(Return(nullptr));
> +#endif
>
> Should the test setup really depend on the *build* platform? (Or is this poor
> naming - should MIR_BUILD_PLATFORM_* really be MIR_TARGET_PLATFORM_*)

Yeah rather poor naming. MIR_BUILD_PLATFORM indicates that we build for platform X11 .. to be honest the way autoconf names target, host and build still confuses me regularly.

Good that you mention this. This is one of the things I wanted to have another look.

Alan Griffiths (alan-griffiths) wrote on 2015-11-13:

Right. I'll consider it preexisting.

review: Approve

Alan Griffiths (alan-griffiths) wrote on 2015-11-13:

Nit:

+struct StubEmergencyCleanupRegistry : mir::EmergencyCleanupRegistry
+{
+ void add(mir::EmergencyCleanupHandler const&) override {}
+};

This needs to implement the other overload.

review: Needs Fixing

Alan Griffiths (alan-griffiths) wrote on 2015-11-13:

Also, in the following scenario:

$ sudo bin/mir_demo_server --window-manager system-compositor --display-config sidebyside --vt 1 --arw-file
$ bin/mir_demo_server --host /tmp/mir_socket --display-config clone --window-manager tiling
$ bin/mir_demo_client_egltriangle

I find that Alt+middle-button+drag seems intermittently to not work (neither resizing nor moving the cursor).

review: Needs Fixing

PS Jenkins bot (ps-jenkins) wrote on 2015-11-16:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5557/rebuild

review: Needs Fixing (continuous-integration)

PS Jenkins bot (ps-jenkins) wrote on 2015-11-17:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5563/rebuild

review: Needs Fixing (continuous-integration)

Alan Griffiths (alan-griffiths) wrote on 2015-11-18:

A bit more detail on the problems I see:

I'm using a lenovo with a trackpoint.

The trackpoint controls the cursor unless the middle "mouse" button is pressed. E.g. Alt+left button drags the triangle, With Alt on its own or with Alt+right it just move the cursor.

Just pressing the middle button (with or without Alt) and the cursor no longer responds to the trackpoint.

It might be diagnostic that the left and right buttons do work, but the middle button doesn't - is something misidentifying it as a two button mouse?

Plugging in a spare 3-button mouse does work.

Cemil Azizoglu (cemil-azizoglu) wrote on 2015-11-18:

613 -if(MIR_TEST_PLATFORM STREQUAL "mesa-kms" OR MIR_TEST_PLATFORM STREQUAL "mesa-x11")
614 +if(MIR_BUILD_PLATFORM_MESA_KMS OR MIR_BUILD_PLATFORM_MESA_X11)

Why is this needed? If the test is being built for Android only why should mesa related stuff be linked in?

review: Needs Information

Andreas Pokorny (andreas-pokorny) wrote on 2015-11-20:

> A bit more detail on the problems I see:
>
> I'm using a lenovo with a trackpoint.
>
> The trackpoint controls the cursor unless the middle "mouse" button is
> pressed. E.g. Alt+left button drags the triangle, With Alt on its own or with
> Alt+right it just move the cursor.
>
> Just pressing the middle button (with or without Alt) and the cursor no longer
> responds to the trackpoint.
>
> It might be diagnostic that the left and right buttons do work, but the middle
> button doesn't - is something misidentifying it as a two button mouse?
>
> Plugging in a spare 3-button mouse does work.

Not quite. I diagnosed that with your evemu record and libinput-debug-events - which is a tool provided by libinput but is not part of the ubuntu libinput package so far.

So by default - it configures your device to use the middle button as scroll-button - so trackpoint motion is converted to scroll axis events while middle button is pressed. You should be able to deactivate that behaviour with the lp:~andreas-pokorny/mir/example-configure-input-devices MP - since it overwrite those settings.

Alan Griffiths (alan-griffiths) wrote on 2015-11-20:

> So by default - it configures your device to use the middle button as scroll-
> button - so trackpoint motion is converted to scroll axis events while middle
> button is pressed. You should be able to deactivate that behaviour with the lp
> :~andreas-pokorny/mir/example-configure-input-devices MP - since it overwrite
> those settings.

I tried adding that, but (after a few merge conflicts - I'm a little way off trunk because of the landings backlog) the server exited because if couldn't find an input platform.

I'll try again later. Meanwhile I won't block on a problem that's being worked on elsewhere.

review: Abstain

Andreas Pokorny (andreas-pokorny) wrote on 2015-11-20:

I thought we do not build the unit tests for

> 613 -if(MIR_TEST_PLATFORM STREQUAL "mesa-kms" OR MIR_TEST_PLATFORM
> STREQUAL "mesa-x11")
> 614 +if(MIR_BUILD_PLATFORM_MESA_KMS OR MIR_BUILD_PLATFORM_MESA_X11)
>
> Why is this needed? If the test is being built for Android only why should
> mesa related stuff be linked in?

Back then I got linker errors whithout that depending on the ordering of the MIR_PLATFORM variable. This is no longer the case..

In general we should find a way to avoid the machinery of MIR_TEST_PLATFORM as far as possible.

PS Jenkins bot (ps-jenkins) wrote on 2015-11-20:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5606/rebuild

review: Needs Fixing (continuous-integration)

PS Jenkins bot (ps-jenkins) wrote on 2015-11-20:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5608/rebuild

review: Needs Fixing (continuous-integration)

Cemil Azizoglu (cemil-azizoglu) wrote on 2015-11-20:

review: Approve

PS Jenkins bot (ps-jenkins) wrote on 2015-11-20:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5609/rebuild

review: Approve (continuous-integration)

PS Jenkins bot (ps-jenkins) wrote on 2015-11-20:

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5614/rebuild

review: Needs Fixing (continuous-integration)

PS Jenkins bot (ps-jenkins) wrote on 2015-11-21:

review: Needs Fixing (continuous-integration)

PS Jenkins bot (ps-jenkins) wrote on 2015-11-21:

review: Needs Fixing (continuous-integration)

PS Jenkins bot (ps-jenkins) on 2015-11-25:

review: Approve (continuous-integration)

Preview Diff

[H/L] Next/Prev Comment, [J/K] Next/Prev File, [N/P] Next/Prev Hunk

Subscribers

People subscribed via source and target branches

to all changes:

Andreas Pokorny

Emanuele Antonio Faraone

Gerry Boland

Mir development team

 === modified file 'benchmarks/frame-uniformity/touch_producing_server.cpp'
 --- benchmarks/frame-uniformity/touch_producing_server.cpp	2015-10-05 02:18:18 +0000
 +++ benchmarks/frame-uniformity/touch_producing_server.cpp	2015-11-20 15:25:54 +0000
@@ -26,7 +26,6 @@
  #include <functional>
  namespace mi = mir::input;
--namespace mia = mi::android;
  namespace mg = mir::graphics;
  namespace mis = mi::synthesis;
  namespace geom = mir::geometry;
 === modified file 'include/platform/mir/input/platform.h'
 --- include/platform/mir/input/platform.h	2015-10-12 08:08:58 +0000
 +++ include/platform/mir/input/platform.h	2015-11-20 15:25:54 +0000
@@ -91,7 +91,7 @@
  };
  typedef mir::UniqueModulePtr<Platform>(*CreatePlatform)(
--    std::shared_ptr<options::Option> const& options,
++    options::Option const& options,
      std::shared_ptr<EmergencyCleanupRegistry> const& emergency_cleanup_registry,
      std::shared_ptr<InputDeviceRegistry> const& input_device_registry,
      std::shared_ptr<InputReport> const& report);
@@ -123,7 +123,7 @@
   * \ingroup platform_enablement
   */
  mir::UniqueModulePtr<mir::input::Platform> create_input_platform(
--    std::shared_ptr<mir::options::Option> const& options,
++    mir::options::Option const& options,
      std::shared_ptr<mir::EmergencyCleanupRegistry> const& emergency_cleanup_registry,
      std::shared_ptr<mir::input::InputDeviceRegistry> const& input_device_registry,
      std::shared_ptr<mir::input::InputReport> const& report);
 === modified file 'include/test/mir_test_framework/executable_path.h'
 --- include/test/mir_test_framework/executable_path.h	2015-10-20 12:46:15 +0000
 +++ include/test/mir_test_framework/executable_path.h	2015-11-20 15:25:54 +0000
@@ -26,6 +26,7 @@
  std::string executable_path();
  std::string library_path();
++std::string server_platform_path();
  std::string udev_recordings_path();
  std::string server_platform(std::string const& name);
  std::string server_input_platform(std::string const& name);
 === modified file 'src/include/server/mir/default_server_configuration.h'
 --- src/include/server/mir/default_server_configuration.h	2015-11-19 04:54:42 +0000
 +++ src/include/server/mir/default_server_configuration.h	2015-11-20 15:25:54 +0000
@@ -132,7 +132,6 @@
  class Scene;
  class InputManager;
  class SurfaceInputDispatcher;
--class Platform;
  class InputDeviceRegistry;
  class InputDeviceHub;
  class DefaultInputDeviceHub;
@@ -144,15 +143,8 @@
  class InputRegion;
  class InputSender;
  class InputSendObserver;
--class NestedInputRelay;
--class EventHandler;
  class CursorImages;
  class LegacyInputDispatchable;
--namespace android
--{
--class InputRegistrar;
--class InputThread;
--}
+ }
  namespace logging
@@ -331,7 +323,6 @@
      virtual std::shared_ptr<droidinput::InputListenerInterface> the_input_translator();
      // new input reading related parts:
--    virtual std::shared_ptr<input::Platform> the_input_platform();
      virtual std::shared_ptr<dispatch::MultiplexingDispatchable> the_input_reading_multiplexer();
      virtual std::shared_ptr<input::InputDeviceRegistry> the_input_device_registry();
      virtual std::shared_ptr<input::InputDeviceHub> the_input_device_hub();
@@ -390,8 +381,7 @@
      CachedPtr<input::CompositeEventFilter> composite_event_filter;
      CachedPtr<input::InputManager>    input_manager;
      CachedPtr<input::SurfaceInputDispatcher>    surface_input_dispatcher;
--    CachedPtr<input::DefaultInputDeviceHub>    default_input_device_hub; // currently not used by default
--    CachedPtr<input::Platform>    input_platform; // currently not used by default
++    CachedPtr<input::DefaultInputDeviceHub>    default_input_device_hub;
      CachedPtr<dispatch::MultiplexingDispatchable> input_reading_multiplexer;
      CachedPtr<input::InputDispatcher> input_dispatcher;
      CachedPtr<input::InputSender>     input_sender;
 === added file 'src/include/server/mir/input/input_probe.h'
 --- src/include/server/mir/input/input_probe.h	1970-01-01 00:00:00 +0000
 +++ src/include/server/mir/input/input_probe.h	2015-11-20 15:25:54 +0000
@@ -0,0 +1,48 @@
++/*
++ * Copyright © 2015 Canonical Ltd.
++ *
++ * This program is free software: you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 3,
++ * as published by the Free Software Foundation.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
++ *
++ * Authored by: Andreas Pokorny <andreas.pokorny@canonical.com>
++ */
++
++#ifndef MIR_INPUT_PROBE_H_
++#define MIR_INPUT_PROBE_H_
++
++#include <vector>
++#include "mir/module_deleter.h"
++
++namespace mir
++{
++namespace options
++{
++class Option;
++}
++class EmergencyCleanupRegistry;
++class SharedLibraryProberReport;
++
++namespace input
++{
++class InputReport;
++class Platform;
++class InputDeviceRegistry;
++
++std::vector<mir::UniqueModulePtr<Platform>> probe_input_platforms(
++    options::Option const& options, std::shared_ptr<EmergencyCleanupRegistry> const& emergency_cleanup,
++    std::shared_ptr<InputDeviceRegistry> const& device_registry, std::shared_ptr<InputReport> const& input_report,
++    SharedLibraryProberReport & prober_report);
++
++}
++}
++
++#endif
 === modified file 'src/platform/graphics/platform_probe.cpp'
 --- src/platform/graphics/platform_probe.cpp	2015-10-12 08:08:58 +0000
 +++ src/platform/graphics/platform_probe.cpp	2015-11-20 15:25:54 +0000
@@ -46,7 +46,11 @@
                  "describe_graphics_module",
                  MIR_SERVER_GRAPHICS_PLATFORM_VERSION);
              auto desc = describe();
--            mir::log_info("Found graphics driver: %s", desc->name);
++            mir::log_info("Found graphics driver: %s, (version: %d.%d.%d)",
++                          desc->name,
++                          desc->major_version,
++                          desc->minor_version,
++                          desc->micro_version);
+         }
          catch (std::runtime_error const&)
+         {
 === modified file 'src/platforms/evdev/platform_factory.cpp'
 --- src/platforms/evdev/platform_factory.cpp	2015-10-20 12:46:15 +0000
 +++ src/platforms/evdev/platform_factory.cpp	2015-11-20 15:25:54 +0000
@@ -67,7 +67,7 @@
+ }
  mir::UniqueModulePtr<mi::Platform> create_input_platform(
--    std::shared_ptr<mo::Option> const& /*options*/,
++    mo::Option const& /*options*/,
      std::shared_ptr<mir::EmergencyCleanupRegistry> const& /*emergency_cleanup_registry*/,
      std::shared_ptr<mi::InputDeviceRegistry> const& input_device_registry,
      std::shared_ptr<mi::InputReport> const& report)
 === modified file 'src/platforms/mesa/server/x11/input/input.cpp'
 --- src/platforms/mesa/server/x11/input/input.cpp	2015-10-12 08:08:58 +0000
 +++ src/platforms/mesa/server/x11/input/input.cpp	2015-11-20 15:25:54 +0000
@@ -28,7 +28,7 @@
  extern mx::X11Resources x11_resources;
  mir::UniqueModulePtr<mi::Platform> create_input_platform(
--    std::shared_ptr<mo::Option> const& /*options*/,
++    mo::Option const& /*options*/,
      std::shared_ptr<mir::EmergencyCleanupRegistry> const& /*emergency_cleanup_registry*/,
      std::shared_ptr<mi::InputDeviceRegistry> const& input_device_registry,
      std::shared_ptr<mi::InputReport> const& /*report*/)
 === modified file 'src/server/input/CMakeLists.txt'
 --- src/server/input/CMakeLists.txt	2015-11-04 16:30:42 +0000
 +++ src/server/input/CMakeLists.txt	2015-11-20 15:25:54 +0000
@@ -14,6 +14,7 @@
    display_input_region.cpp
    event_filter_chain_dispatcher.cpp
    input_modifier_utils.cpp
++  input_probe.cpp
    key_repeat_dispatcher.cpp
    null_input_channel_factory.cpp
    null_input_dispatcher.cpp
 === modified file 'src/server/input/default_configuration.cpp'
 --- src/server/input/default_configuration.cpp	2015-10-13 17:42:07 +0000
 +++ src/server/input/default_configuration.cpp	2015-11-20 15:25:54 +0000
@@ -39,16 +39,16 @@
  #include "surface_input_dispatcher.h"
  #include "mir/input/touch_visualizer.h"
++#include "mir/input/input_probe.h"
  #include "mir/input/platform.h"
  #include "mir/options/configuration.h"
  #include "mir/options/option.h"
  #include "mir/dispatch/multiplexing_dispatchable.h"
  #include "mir/compositor/scene.h"
  #include "mir/emergency_cleanup.h"
--#include "mir/report/legacy_input_report.h"
  #include "mir/main_loop.h"
--#include "mir/shared_library.h"
  #include "mir/glib_main_loop.h"
++#include "mir/log.h"
  #include "mir/dispatch/action_queue.h"
  #include "mir_toolkit/cursors.h"
@@ -290,26 +290,6 @@
          });
+ }
--std::shared_ptr<mi::Platform>
--mir::DefaultServerConfiguration::the_input_platform()
--{
--    return input_platform(
--        [this]() -> std::shared_ptr<mi::Platform>
--        {
--            auto options = the_options();
--
--            if (!options->is_set(options::platform_input_lib))
--                return nullptr;
--
--            auto lib = std::make_shared<mir::SharedLibrary>(
--                options->get<std::string>(options::platform_input_lib));
--            auto create = lib->load_function<mi::CreatePlatform>(
--                "create_input_platform",
--                MIR_SERVER_INPUT_PLATFORM_VERSION);
--            return create(the_options(), the_emergency_cleanup(), the_input_device_registry(), the_input_report());
--        });
--}
--
  namespace
+ {
  class NullLegacyInputDispatchable : public mi::LegacyInputDispatchable
@@ -341,23 +321,20 @@
+             }
              else
+             {
--                if (options->get<std::string>(options::legacy_input_report_opt) == options::log_opt_value)
--                    mr::legacy_input::initialize(the_logger());
--
--                if (auto platform = the_input_platform())
--                {
--                    auto const ret = std::make_shared<mi::DefaultInputManager>(
--                        the_input_reading_multiplexer(),
--                        std::make_shared<NullLegacyInputDispatchable>());
--
--                    ret->add_platform(platform);
--                    return ret;
--                }
--                else
--                {
--                    return std::make_shared<mi::DefaultInputManager>(
--                        the_input_reading_multiplexer(), the_legacy_input_dispatchable());
--                }
++                auto platforms = probe_input_platforms(*options, the_emergency_cleanup(), the_input_device_registry(),
++                                                       the_input_report(), *the_shared_library_prober_report());
++
++                if (platforms.empty())
++                    BOOST_THROW_EXCEPTION(std::runtime_error("No input platforms found"));
++
++                auto const ret = std::make_shared<mi::DefaultInputManager>(
++                    the_input_reading_multiplexer(),
++                    std::make_shared<NullLegacyInputDispatchable>());
++
++                for (auto & platform : platforms)
++                    ret->add_platform(std::move(platform));
++
++                return ret;
+             }
+         }
      );
 === added file 'src/server/input/input_probe.cpp'
 --- src/server/input/input_probe.cpp	1970-01-01 00:00:00 +0000
 +++ src/server/input/input_probe.cpp	2015-11-20 15:25:54 +0000
@@ -0,0 +1,104 @@
++/*
++ * Copyright © 2015 Canonical Ltd.
++ *
++ * This program is free software: you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 3,
++ * as published by the Free Software Foundation.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
++ *
++ * Authored by: Andreas Pokorny <andreas.pokorny@canonical.com>
++ */
++
++#include "mir/input/input_probe.h"
++#include "mir/input/platform.h"
++
++#include "mir/options/configuration.h"
++#include "mir/options/option.h"
++
++#include "mir/shared_library_prober.h"
++#include "mir/shared_library.h"
++#include "mir/log.h"
++#include "mir/module_deleter.h"
++
++#include <stdexcept>
++
++namespace mi = mir::input;
++namespace mo = mir::options;
++
++namespace
++{
++auto probe_input_platform(mir::SharedLibrary const& lib, mir::options::Option const& options)
++{
++    auto probe = lib.load_function<mi::ProbePlatform>("probe_input_platform", MIR_SERVER_INPUT_PLATFORM_VERSION);
++
++    return probe(options);
++}
++
++void describe_input_platform(mir::SharedLibrary const& lib)
++{
++    auto describe =
++        lib.load_function<mi::DescribeModule>("describe_input_module", MIR_SERVER_INPUT_PLATFORM_VERSION);
++    auto desc = describe();
++    mir::log_info("Selected input driver: %s (version: %d.%d.%d)", desc->name, desc->major_version, desc->minor_version,
++                  desc->micro_version);
++}
++
++mir::UniqueModulePtr<mi::Platform> create_input_platform(
++    mir::SharedLibrary const& lib, mir::options::Option const& options,
++    std::shared_ptr<mir::EmergencyCleanupRegistry> const& cleanup_registry,
++    std::shared_ptr<mi::InputDeviceRegistry> const& registry, std::shared_ptr<mi::InputReport> const& report)
++{
++
++    auto create = lib.load_function<mi::CreatePlatform>("create_input_platform", MIR_SERVER_INPUT_PLATFORM_VERSION);
++
++    return create(options, cleanup_registry, registry, report);
++}
++}
++
++std::vector<mir::UniqueModulePtr<mi::Platform>> mi::probe_input_platforms(
++    mo::Option const& options, std::shared_ptr<EmergencyCleanupRegistry> const& emergency_cleanup,
++    std::shared_ptr<mi::InputDeviceRegistry> const& device_registry, std::shared_ptr<mi::InputReport> const& input_report,
++    mir::SharedLibraryProberReport& prober_report)
++{
++    std::vector<UniqueModulePtr<Platform>> platforms;
++
++    if (options.is_set(mo::platform_input_lib))
++    {
++        mir::SharedLibrary lib(options.get<std::string>(mo::platform_input_lib));
++
++        platforms.emplace_back(create_input_platform(lib, options, emergency_cleanup, device_registry, input_report));
++
++        describe_input_platform(lib);
++    }
++    else
++    {
++        auto const& path = options.get<std::string>(mo::platform_path);
++        auto platforms_libs = mir::libraries_for_path(path, prober_report);
++
++        for (auto const& platform_lib : platforms_libs)
++        {
++            try
++            {
++                if (probe_input_platform(*platform_lib, options) > mi::PlatformPriority::dummy)
++                {
++                    platforms.emplace_back(
++                        create_input_platform(*platform_lib, options, emergency_cleanup, device_registry, input_report));
++
++                    describe_input_platform(*platform_lib);
++                }
++            }
++            catch (std::runtime_error const&)
++            {
++            }
++        }
++    }
++
++    return platforms;
++}
 === modified file 'src/server/symbols.map'
 --- src/server/symbols.map	2015-11-19 06:43:10 +0000
 +++ src/server/symbols.map	2015-11-20 15:25:54 +0000
@@ -83,6 +83,7 @@
      mir::input::TouchVisualizer::operator*;
      mir::input::TouchVisualizer::?TouchVisualizer*;
      mir::input::TouchVisualizer::TouchVisualizer*;
++    mir::input::probe_input_platforms*;
      mir::LockableCallback::?LockableCallback*;
      mir::LockableCallback::LockableCallback*;
      mir::LockableCallback::operator*;
@@ -687,7 +688,6 @@
      mir::DefaultServerConfiguration::the_input_dispatcher*;
      mir::DefaultServerConfiguration::the_new_input_dispatcher*;
      mir::DefaultServerConfiguration::the_input_manager*;
--    mir::DefaultServerConfiguration::the_input_platform*;
      mir::DefaultServerConfiguration::the_input_reader*;
      mir::DefaultServerConfiguration::the_input_reader_policy*;
      mir::DefaultServerConfiguration::the_input_reading_multiplexer*;
 === renamed file 'tests/mir_test_framework/stub_input_platform.h' => 'tests/include/mir_test_framework/stub_input_platform.h'
 === modified file 'tests/mir_test_framework/CMakeLists.txt'
 --- tests/mir_test_framework/CMakeLists.txt	2015-11-04 15:42:43 +0000
 +++ tests/mir_test_framework/CMakeLists.txt	2015-11-20 15:25:54 +0000
@@ -38,6 +38,7 @@
    process.cpp
    server_runner.cpp
    socket_detect_server.cpp
++  stub_input_platform.cpp
    stub_client_platform_factory.cpp
    stub_server_platform_factory.cpp
    stub_session.cpp
 === modified file 'tests/mir_test_framework/executable_path.cpp'
 --- tests/mir_test_framework/executable_path.cpp	2015-10-20 12:46:15 +0000
 +++ tests/mir_test_framework/executable_path.cpp	2015-11-20 15:25:54 +0000
@@ -45,6 +45,16 @@
      return executable_path() + "/../lib";
+ }
++std::string mir_test_framework::server_platform_path()
++{
++    for (auto const& option : {library_path() + "/server-modules/",
++                               library_path() + "/server-platform/",
++                               std::string(MIR_SERVER_PLATFORM_PATH) + '/'})
++        if (boost::filesystem::exists(option))
++            return option;
++    BOOST_THROW_EXCEPTION(std::runtime_error("Failed to find server platform in standard search locations"));
++}
++
  std::string mir_test_framework::udev_recordings_path()
+ {
      std::string run_path     = executable_path() + "/udev_recordings";
 === modified file 'tests/mir_test_framework/fake_input_device_impl.cpp'
 --- tests/mir_test_framework/fake_input_device_impl.cpp	2015-11-19 04:54:42 +0000
 +++ tests/mir_test_framework/fake_input_device_impl.cpp	2015-11-20 15:25:54 +0000
@@ -17,7 +17,7 @@
   */
  #include "fake_input_device_impl.h"
--#include "stub_input_platform.h"
++#include "mir_test_framework/stub_input_platform.h"
  #include "mir/input/input_device.h"
  #include "mir/input/input_device_info.h"
 === modified file 'tests/mir_test_framework/fake_input_server_configuration.cpp'
 --- tests/mir_test_framework/fake_input_server_configuration.cpp	2015-08-26 08:44:24 +0000
 +++ tests/mir_test_framework/fake_input_server_configuration.cpp	2015-11-20 15:25:54 +0000
@@ -23,7 +23,6 @@
  namespace mi = mir::input;
  namespace ms = mir::scene;
  namespace msh = mir::shell;
--namespace mia = mir::input::android;
  mtf::FakeInputServerConfiguration::FakeInputServerConfiguration()
+ {
 === modified file 'tests/mir_test_framework/input_testing_server_options.cpp'
 --- tests/mir_test_framework/input_testing_server_options.cpp	2015-06-25 03:00:08 +0000
 +++ tests/mir_test_framework/input_testing_server_options.cpp	2015-11-20 15:25:54 +0000
@@ -38,7 +38,6 @@
  namespace mg = mir::graphics;
  namespace mi = mir::input;
  namespace ms = mir::shell;
--namespace mia = mi::android;
  namespace geom = mir::geometry;
  mtf::InputTestingServerConfiguration::InputTestingServerConfiguration()
 === modified file 'tests/mir_test_framework/stub_input.cpp'
 --- tests/mir_test_framework/stub_input.cpp	2015-10-12 08:08:58 +0000
 +++ tests/mir_test_framework/stub_input.cpp	2015-11-20 15:25:54 +0000
@@ -16,7 +16,7 @@
   * Authored by: Andreas Pokorny <andreas.pokorny@canonical.com>
   */
--#include "stub_input_platform.h"
++#include "mir_test_framework/stub_input_platform.h"
  #include "fake_input_device_impl.h"
  #include "mir/module_properties.h"
@@ -25,7 +25,7 @@
  namespace mi = mir::input;
  mir::UniqueModulePtr<mi::Platform> create_input_platform(
--    std::shared_ptr<mo::Option> const& /*options*/,
++    mo::Option const& /*options*/,
      std::shared_ptr<mir::EmergencyCleanupRegistry> const& /*emergency_cleanup_registry*/,
      std::shared_ptr<mi::InputDeviceRegistry> const& input_device_registry,
      std::shared_ptr<mi::InputReport> const& /*report*/)
 === modified file 'tests/mir_test_framework/stub_input_platform.cpp'
 --- tests/mir_test_framework/stub_input_platform.cpp	2015-10-12 08:08:58 +0000
 +++ tests/mir_test_framework/stub_input_platform.cpp	2015-11-20 15:25:54 +0000
@@ -16,7 +16,7 @@
   * Authored by: Andreas Pokorny <andreas.pokorny@canonical.com>
   */
--#include "stub_input_platform.h"
++#include "mir_test_framework/stub_input_platform.h"
  #include "mir/input/input_device_registry.h"
  #include "mir/dispatch/action_queue.h"
 === modified file 'tests/umock-acceptance-tests/test_libinput.cpp'
 --- tests/umock-acceptance-tests/test_libinput.cpp	2015-10-12 08:08:58 +0000
 +++ tests/umock-acceptance-tests/test_libinput.cpp	2015-11-20 15:25:54 +0000
@@ -17,8 +17,8 @@
   */
  #include "mir/input/platform.h"
++#include "mir/test/doubles/mock_option.h"
  #include "mir/shared_library.h"
--#include "mir/options/option.h"
  #include "mir_test_framework/udev_environment.h"
  #include "mir_test_framework/executable_path.h"
@@ -27,20 +27,13 @@
  #include <gtest/gtest.h>
  namespace mtf = mir_test_framework;
++namespace mtd = mir::test::doubles;
  namespace mo = mir::options;
++
  using namespace ::testing;
  namespace
+ {
--struct MockOption : mo::Option
--{
--    MOCK_CONST_METHOD1(is_set, bool(char const*));
--    MOCK_CONST_METHOD2(get, bool(char const*, bool));
--    MOCK_CONST_METHOD2(get, int(char const*, int));
--    MOCK_CONST_METHOD2(get, std::string(char const* name, char const*));
--    MOCK_CONST_METHOD1(get, boost::any const&(char const*));
--};
--
  auto get_libinput_platform()
+ {
      auto path = mtf::server_input_platform("input-evdev");
@@ -54,7 +47,7 @@
  TEST(LibInput, DISABLED_probes_as_unsupported_without_device_access)
+ {
--    NiceMock<MockOption> options;
++    NiceMock<mtd::MockOption> options;
      // dumb assumption - nobody runs this test cases as root..
      // or allows accessing evdev input devices from non privileged users.
@@ -67,7 +60,7 @@
+ {
      mtf::UdevEnvironment env;
      env.add_standard_device("laptop-keyboard");
--    NiceMock<MockOption> options;
++    NiceMock<mtd::MockOption> options;
      auto library = get_libinput_platform();
      auto probe_fun = library->load_function<mir::input::ProbePlatform>(probe_input_platform_symbol);
@@ -78,7 +71,8 @@
+ {
      mtf::UdevEnvironment env;
      env.add_standard_device("laptop-keyboard");
--    NiceMock<MockOption> options;
++    NiceMock<mtd::MockOption> options;
++
      ON_CALL(options,is_set(StrEq(host_socket_opt)))
          .WillByDefault(Return(true));
      ON_CALL(options,get(StrEq(host_socket_opt), Matcher<char const*>(_)))
@@ -92,7 +86,7 @@
  TEST(LibInput, probes_as_supported_when_umock_dev_available_or_before_input_devices_are_available)
+ {
      mtf::UdevEnvironment env;
--    NiceMock<MockOption> options;
++    NiceMock<mtd::MockOption> options;
      auto library = get_libinput_platform();
      auto probe_fun = library->load_function<mir::input::ProbePlatform>(probe_input_platform_symbol);
 === modified file 'tests/unit-tests/input/CMakeLists.txt'
 --- tests/unit-tests/input/CMakeLists.txt	2015-11-04 19:47:19 +0000
 +++ tests/unit-tests/input/CMakeLists.txt	2015-11-20 15:25:54 +0000
@@ -8,6 +8,7 @@
    ${CMAKE_CURRENT_SOURCE_DIR}/test_xcursor_loader.cpp
    ${CMAKE_CURRENT_SOURCE_DIR}/test_touchspot_controller.cpp
    ${CMAKE_CURRENT_SOURCE_DIR}/test_input_event.cpp
++  ${CMAKE_CURRENT_SOURCE_DIR}/test_input_platform_probing.cpp
    ${CMAKE_CURRENT_SOURCE_DIR}/test_event_builders.cpp
    ${CMAKE_CURRENT_SOURCE_DIR}/test_default_device.cpp
    ${CMAKE_CURRENT_SOURCE_DIR}/test_default_input_device_hub.cpp
 === added file 'tests/unit-tests/input/test_input_platform_probing.cpp'
 --- tests/unit-tests/input/test_input_platform_probing.cpp	1970-01-01 00:00:00 +0000
 +++ tests/unit-tests/input/test_input_platform_probing.cpp	2015-11-20 15:25:54 +0000
@@ -0,0 +1,161 @@
++/*
++ * Copyright © 2015 Canonical Ltd.
++ *
++ * This program is free software: you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License version 3 as
++ * published by the Free Software Foundation.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
++ *
++ * Authored by: Andreas Pokorny <andreas.pokorny@canonical.com>
++ */
++
++#include <gtest/gtest.h>
++
++#include "mir/input/input_probe.h"
++
++#include "mir/input/platform.h"
++#include "src/server/report/null_report_factory.h"
++
++#include "mir/options/option.h"
++#include "mir/emergency_cleanup_registry.h"
++
++#include "mir_test_framework/udev_environment.h"
++#include "mir_test_framework/executable_path.h"
++#include "mir_test_framework/stub_input_platform.h"
++#include "mir/test/doubles/mock_x11.h"
++#include "mir/test/doubles/mock_libinput.h"
++#include "mir/test/doubles/mock_option.h"
++#include "mir/test/doubles/mock_input_device_registry.h"
++#include "src/platforms/evdev/platform.h"
++#include "src/platforms/mesa/server/x11/input/input_platform.h"
++#include "mir/test/fake_shared.h"
++
++namespace mt = mir::test;
++namespace mtd = mt::doubles;
++namespace mi = mir::input;
++namespace mr = mir::report;
++namespace mtf = mir_test_framework;
++
++using namespace ::testing;
++
++namespace
++{
++struct StubEmergencyCleanupRegistry : mir::EmergencyCleanupRegistry
++{
++    void add(mir::EmergencyCleanupHandler const&) override {}
++    void add(mir::ModuleEmergencyCleanupHandler) override {}
++};
++char const platform_input_lib[] = "platform-input-lib";
++char const platform_path[] = "platform-path";
++
++struct InputPlatformProbe : ::testing::Test
++{
++    InputPlatformProbe()
++    {
++        ON_CALL(mock_options, is_set(StrEq(platform_input_lib))).WillByDefault(Return(false));
++        ON_CALL(mock_options, is_set(StrEq(platform_path))).WillByDefault(Return(true));
++        ON_CALL(mock_options, get(StrEq(platform_path),Matcher<char const*>(_)))
++            .WillByDefault(Return(platform_path_value));
++        ON_CALL(mock_options, get(StrEq(platform_path)))
++            .WillByDefault(Invoke(
++                    [this](char const*) -> boost::any const&
++                    {
++                        return platform_path_value_as_any;
++                    }));
++        ON_CALL(mock_options, get(StrEq(platform_input_lib)))
++            .WillByDefault(Invoke(
++                    [this](char const*) -> boost::any const&
++                    {
++                        return platform_input_lib_value_as_any;
++                    }));
++    }
++
++    void disable_x11()
++    {
++#ifdef MIR_BUILD_PLATFORM_MESA_X11
++        ON_CALL(mock_x11, XOpenDisplay(_)).WillByDefault(Return(nullptr));
++#endif
++    }
++
++    // replace with with mocks for udev and evdev to simulate root or non-root
++    // access on evdev devices, and enable the disabled test case(s) below.
++    mtf::UdevEnvironment env;
++
++#ifdef MIR_BUILD_PLATFORM_MESA_X11
++    NiceMock<mtd::MockX11> mock_x11;
++#endif
++    NiceMock<mtd::MockLibInput> mock_libinput;
++    NiceMock<mtd::MockOption> mock_options;
++    mtd::MockInputDeviceRegistry mock_registry;
++    StubEmergencyCleanupRegistry stub_emergency;
++    std::shared_ptr<mir::SharedLibraryProberReport> stub_prober_report{mr::null_shared_library_prober_report()};
++    std::string platform_path_value{mtf::server_platform_path()};
++    boost::any platform_path_value_as_any{platform_path_value};
++    std::string platform_input_lib_value;
++    boost::any platform_input_lib_value_as_any;
++};
++
++template <typename Expected>
++struct OfPtrTypeMatcher
++{
++    template <typename T>
++    bool MatchAndExplain(T&& p, MatchResultListener* /* listener */) const
++    {
++        return dynamic_cast<Expected*>(&*p) != nullptr;
++    }
++    void DescribeTo(::std::ostream* os) const
++    {
++        *os << "is a or derived from a " << typeid(Expected).name();
++    }
++    void DescribeNegationTo(::std::ostream* os) const
++    {
++        *os << "is not or not derived from " << typeid(Expected).name();
++    }
++};
++
++template<typename Dest>
++inline auto OfPtrType()
++{
++    return MakePolymorphicMatcher(OfPtrTypeMatcher<Dest>());
++}
++}
++
++
++TEST_F(InputPlatformProbe, stub_platform_not_picked_up_by_default)
++{
++    disable_x11();
++    auto platforms =
++        mi::probe_input_platforms(mock_options, mt::fake_shared(stub_emergency), mt::fake_shared(mock_registry),
++                                  mr::null_input_report(), *stub_prober_report);
++
++    EXPECT_THAT(platforms, ElementsAre(OfPtrType<mi::evdev::Platform>()));
++}
++
++#ifdef MIR_BUILD_PLATFORM_MESA_X11
++TEST_F(InputPlatformProbe, x11_platform_found_and_used_when_display_connection_works)
++{
++    auto platforms =
++        mi::probe_input_platforms(mock_options, mt::fake_shared(stub_emergency), mt::fake_shared(mock_registry),
++                                  mr::null_input_report(), *stub_prober_report);
++
++    EXPECT_THAT(platforms, UnorderedElementsAre(OfPtrType<mi::evdev::Platform>(), OfPtrType<mi::X::XInputPlatform>()));
++}
++#endif
++
++TEST_F(InputPlatformProbe, allows_forcing_stub_input_platform)
++{
++    ON_CALL(mock_options, is_set(StrEq(platform_input_lib))).WillByDefault(Return(true));
++    platform_input_lib_value = mtf::server_input_platform("input-stub.so");
++    platform_input_lib_value_as_any = platform_input_lib_value;
++    auto platforms =
++        mi::probe_input_platforms(mock_options, mt::fake_shared(stub_emergency), mt::fake_shared(mock_registry),
++                                  mr::null_input_report(), *stub_prober_report);
++    EXPECT_THAT(platforms, ElementsAre(OfPtrType<mtf::StubInputPlatform>()));
++}