Mir sometimes crashes after Bluetooth reconnect

[Víctor R. Ruiz]

Test case.
- In System Settings, pair a Bluetooth keyboard and mouse.
- Wait some minutes so keyboard and mouse suspend.
- Reboot the device.
- Start typing on the password prompt and move the mouse.

Expected result.
- User can log in.

Actual result.
- Unity8 crashes.

It's not an accurate test case but has happened at least three times in similar scenarios which involve waking up the keyboard (or mouse?).

current build number: 180
device name: frieza
channel: ubuntu-touch/rc-proposed/bq-aquaris-pd.en

With silo 37.

[Ken VanDine (ken-vandine)]

I've witnessed this once. I had the OSK showing to type something in an app (forget which). I decided to switch to the bluetooth keyboard, so I turned on the bluetooth keyboard. Once it connected, unity8 crashed.

[Christopher Townsend (townsend)]

I just reflashed my frieza to build 180 and witnessed a Unity 8 crash when my BT mouse try connecting. The system did not reboot however.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Michał Sawicz (saviq)]

Can someone please give up a corresponding .crash file?

[Michał Sawicz (saviq)]

Unfortunately that one's broken, a unity8 .crash file needs to be multiple MBs, not 300kB.

[Víctor R. Ruiz (vrruiz)]

It's the only data it gathered :(

[Víctor R. Ruiz (vrruiz)]

I saved this other crash from yesterday. I think it happened in similar conditions.

[Michał Sawicz (saviq)]

Much better!

I'll try and get symbols, but in the mean time:

 #0 0xf4302920 in ?? () from /usr/lib/arm-linux-gnueabihf/libmirserver.so.41
 No symbol table info available.
 #1 0xf430376a in ?? () from /usr/lib/arm-linux-gnueabihf/libmirserver.so.41
 No symbol table info available.
 #2 0xf4317e66 in ?? () from /usr/lib/arm-linux-gnueabihf/libmirserver.so.41
 No symbol table info available.
 #3 0xf421b488 in ?? () from /usr/lib/arm-linux-gnueabihf/libmirclient.so.9
 No symbol table info available.
 #4 0xf3ffb9d4 in ?? () from /usr/lib/arm-linux-gnueabihf/libmircommon.so.6
 No symbol table info available.

This suggests it's a mir issue, and it roughly coincides with when latest mir was released.

[Michał Sawicz (saviq)]

Yup, this looks like a Mir crash IMO:

#0 0xffffffff in mir::graphics::nested::InputPlatform::InputDevice::emit_event(MirInputEvent const*, mir::geometry::Rectangle const&) (this=0xe7f53350, event=event@entry=0xd61f8cb8, frame=...)
    at /build/mir-V80GEw/mir-0.24.0+15.04.20160815.3/src/server/graphics/nested/input_platform.cpp:131
        pointer_event = <optimized out>
        new_event =
                {_M_t = {<std::_Tuple_impl> = {<std::_Tuple_impl> = {<std::_Tuple_impl> = {<No data fields>}, <std::_Head_base> = {_M_head_impl = 0xe9ad4bfc}, <No data fields>}, <std::_Head_base> = {_M_head_impl = 0x4252d328}, <No data fields>}, <No data fields>}}
        x = <optimized out>
        y = <optimized out>
        type = mir_input_event_type_pointer
#1 0xffffffff in std::_Function_handler::_M_invoke(std::_Any_data const&, MirEvent const&, mir::geometry::Rectangle const&) (area=..., event=..., __closure=0xe7f0cd20)
    at /build/mir-V80GEw/mir-0.24.0+15.04.20160815.3/src/server/graphics/nested/input_platform.cpp:225
        input_ev = 0xd61f8cb8
        id = 6
        event_type = <optimized out>
        this = 0xf1bef72c
#2 0xffffffff in std::_Function_handler::_M_invoke(std::_Any_data const&, MirEvent const&, mir::geometry::Rectangle const&) (__functor=..., __args#0=..., __args#1=...) at /usr/include/c++/4.9/functional:2039
#3 0xffffffff in mir::graphics::nested::detail::DisplayBuffer::event_thunk(MirSurface*, MirEvent const*, void*) (__args#1=..., __args#0=..., this=<optimized out>) at /usr/include/c++/4.9/functional:2439
        event = @0xd61f8cb8: {type_ = mir_event_type_motion}
        this = <optimized out>
#4 0xffffffff in mir::graphics::nested::detail::DisplayBuffer::event_thunk(MirSurface*, MirEvent const*, void*) (source_frame=..., event=..., this=<optimized out>)
    at /build/mir-V80GEw/mir-0.24.0+15.04.20160815.3/src/server/graphics/nested/mir_client_host_connection.cpp:419
        event = @0xd61f8cb8: {type_ = mir_event_type_motion}
        this = <optimized out>
#5 0xffffffff in mir::graphics::nested::detail::DisplayBuffer::event_thunk(MirSurface*, MirEvent const*, void*) (event=..., this=<optimized out>)
    at /build/mir-V80GEw/mir-0.24.0+15.04.20160815.3/src/server/graphics/nested/display_buffer.cpp:115
        event = @0xd61f8cb8: {type_ = mir_event_type_motion}
        this = <optimized out>

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[slash (slash-tux)]

Gents,

I've exactly this problem when the second device got paired.
Where do i find the logfile you need ?

Mako last rc-proposed

[frunu (frundengar)]

Hi,
i have the sme problem since ota13 in stable on mako.
If u need some log i'll do my best.
Actually my Speedlink Athera XE puts the phone to the boot screen nie of ten times on reconnect.

Greets,
frunu

[pedaliero2012 (patrik-simon)]

Hi,
unity crashes When I try to log on
Especially when the mouse is in motion and the login screen is not yet visible

rc-proposed
Bq M10 FHD
Logitech M535 Bluetooth Mouse + Logitech K480 Bluetooth Keyboard