Devices are only added from a single thread but they may be destroyed
concurrently so we need to guard the list against race conditions.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
And name the callbacks a bit more consistently.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
The &IG_00 suffix is now automatically added by winexinput.sys.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
In addition, and before WINEBUS\WINE_COMP_HID, so that winexinput.sys
will match first as soon as it is introduced.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
And match it in winehid.sys instead of individual bus ids.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Instead of a separately allocated device list entry pointer. And link
the device object into the device list directly.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
And use an opaque struct unix_device as private data.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Otherwise the device will still be linked in the device list and any
thread iterating it to find a matching device will access stale pointer.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
And try parsing it from the hid or input and usb parent nodes, but not
beyond.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
The comparison result was wrong, and the device syspath always differs
anyway between two different subsystems. We need to compare the parent
device syspath.
The input subsystem devices also needs to be deduplicated between eventX
devices and jsX devices.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This was inconsistently forcing the first report byte to the report id.
The report ids are stripped when they aren't used, in the packet
hidclass.sys provides, but hidraw API wants them all the time, so we
need to use an intermediate buffer.
Also ioctl length argument have a lower size capacity, so we have to
check for it as well.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Fixing invalid axis range when they differ between two axis of the same
usage page.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Or we may have a race condition where a pending IRP may be completed
after we left the CS but before we read its status.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Breaking out of the switch will leave it already.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>