Merge remote-tracking branch 'whot/for-keith'

I've looked at these patches, but I can't say I've actually
reviewed them...

Merge remote-tracking branch 'airlied/for-keithp'

glx/dri2: initialise api to avoid indirect rendering failing randomly

Running glxinfo under indirect rendering would randomly fail against the
intel driver, as it would create a context with no attribs, and then the
api value would be passed to the driver uninitialised.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Keith Packard <keithp@keithp.com>

Merge branch 'stack-smash-on-touchpoint' into for-keith

Merge branch 'touch-selection-conflict-fixes' into for-keith

Xi: fix touch event selction conflicts (#57301)

There are limits on which client may select for touch events on a given
window, with restrictions being that no two clients can select on the same
device, but narrower selections are allowed, i.e. if one client has
XIAllDevices, a second client may still select for device X.

The current code had a dependency on which client selected first and which
device, resulting in inconsistencies when selecting for events. Fix that,
responding with the right errors regardless of who selected what first.

X.Org Bug 57301 <http://bugs.freedesktop.org/show_bug.cgi?id=57301>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

dix: split xi2_mask_isset into a per-device function

For touch selection conflicts, we need to check not only if the mask is set
for the device, but if it is set for only that specific device (regardless
of XIAll*Devices)

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

dix: add FIXME, TouchRemovePointerGrab does nothing

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Xi: if a TouchEnd appears on a actively grabbing client, always accept

Once the TouchEnd appears on the device, the touch is done. If the client
still has a pointer grab, accept it to avoid clients with TouchOwnership
selections to wait indefinitely for the actual touch event.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

dix: when deactivating pointer-only grabs, don't emulate TouchEnd events

A client with a pointer grab on a touch device must reject the touch when
detactivating the grab while the touch is active. However, such a rejecting
must not trigger a ButtonRelease event to be emulated and sent to the
client.
Set the grabbing listener's state to HAS_END, so we simply skip delivery to
that client.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

dix: only reject active grabs on ungrab and do it before actually ungrabbing

An active grab ungrabbing is the same as rejecting the grab, since the
client is no longer interested in those events. So reject any touch grab,
but do so before actually deactivating since we're interested in the
TouchEnd for the current grabbing client.

A passive grab otoh is _not_ like rejecting a grab, since it deactivates
automatically when the touch ends.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

dix: don't call ProcessInputEvents() when accepting/rejecting touches

TouchListenerAcceptReject may be called during normal event processing, but
ProcessInputEvents is not reentrant and calling it here smashes the event
queue.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Xi: fix typo "mechansims" → "mechanisms"

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Revert "xf86: Fix non-PCI configuration-less setups"

This reverts commit 76d9c62eb2be2010a19bf36285012d086cdd180b.

This breaks multi-GPU setups here, so lets drop it for now.

Signed-off-by: Dave Airlie <airlied@redhat.com>

dri1: fix dri1 startup since 459c6da0f907ba33d733c7e62a116184ba2f14e5

This commit regresses dri1 since it moves the drmSetServerInfo from being
called at module load time to extension init time. However DRIScreenInit
relies on this being called before it gets control.

This patches moves the call into DRIScreenInit and seems to work here.

Signed-off-by: Dave Airlie <airlied@redhat.com>

xfixes: Fix minor number in QueryVersion

Due to a typo the major version number was passed as minor version to
version_compare().

Regression-from: ffd4874798ba54f86acac75779a15b4babeaa5f3

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

xkb: only post a XTest release if the XTest device has the button down

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

xkb: always post XTest button up when the physical button released (#28808)

Regression introduced by commit 2decff6393a44b56d80d53570718f95354fde454
  xkb: ProcesssPointerEvent must work on the VCP if it gets the VCP

XTest buttons must be released when a physical button is released. This was
fixed in 14327858391ebe929b806efb53ad79e789361883, but
2decff6393a44b56d80d53570718f95354fde454 changed a condition that this code
didn't get triggered anymore.

"dev" for pointer events is now always the VCP which doesn't have a xkbi
struct. So move this condition out and always trigger the XTest released for
button events.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Tested-by: Frank Roscher <Frank-Roscher@gmx.net>

Merge remote-tracking branch 'whot/for-keith'

randr: call RRProviderInit in the proper place.

No idea where this got lost across development cycles, but its
definitely missing.

Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=57448

Signed-off-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Keith Packard <keithp@keithp.com>

Merge branch 'high-resolution-touch-devices' into for-keith

hw/dmx: add update_desktop_dimensions() call

The Xdmx server did not update the desktop dimensions when computing screen
origins.

Signed-off-by: Sybren van Elderen <sowmestno@msn.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

dix: use pixman for fp1616 conversions

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Søren Sandmann <ssp@redhat.com>>

xkb: fill in keycode and event type for slow keys enablement

eventType is set for the type that triggered a XkbControlsNotify event.
Technically, SlowKeys is triggered by a timer which doesn't have a matching
core event type. So we used to use 0 here.

Practically, the timer is triggered by a key press + hold and cancelled when
the key is released before the timeout expires. So we might as well set
KeyPress (keycode) in the ControlsNotify to give clients a chance to differ
between timer-triggered SlowKeys and client-triggered ones.

This is a chance in behaviour, though I suspect with little impact.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Daniel Stone <daniel@fooishbar.org>

When resetting device idle time, reset XIAll(Master)Devices too (#56649)

When the screen saver is forcibly deactivated, the idle time counter is
reset for all devices but not for the fake XIAllDevices and
XIAllMasterDevices. XScreenSaverQueryInfo uses XIAlldevices to fill the
"idle" field, thus returning the wrong value.

Regression introduced in
commit 6aef209ebc2e54f5465da505a780f7b4cc273ee0
Author: Peter Hutterer <peter.hutterer@who-t.net>
Date:   Mon Mar 12 13:51:02 2012 +1000

    Change lastDeviceIdleTime to be per-device

X.Org Bug 56649 <http://bugs.freedesktop.org/show_bug.cgi?id=56649>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Tested-by: Giacomo Perale <ghepeu@virgilio.it>
Reviewed-by: Keith Packard <keithp@keithp.com>

cursor: Revise edge cases for the pointer moving towards barriers

Since barriers block the invisible line between pixels, that means
that we need to explicitly check the boundaries, or else we'll have
a potential off-by-one error. This fixes issues when trying to move
down or right across a barrier and having the pointer visibly bounce.

Signed-off-by: Jasper St. Pierre <jstpierre@mecheye.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Merge remote-tracking branch 'jturney/mingw-patches'

hw/xwin: Fix for MinGW-w64 DirectDraw headers

mingw-w64 headers handle NONAMELESSUNION earlier than mingw.org's, so it must be
defined before including any headers.  It also provides a ddraw.h, so use it.

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Keith Packard <keithp@keithp.com>

dix/dispatch.c, os/utils.c: Disable smart scheduler on WIN32

setitimer() and SIGALRM aren't available on WIN32, so smart scheduler
code cannot be built.  Provide only stubs for smart scheduler timer
code, and disable smart scheduler by default.

Signed-off-by: Ryan Pavlik <rpavlik@iastate.edu>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Tested-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Keith Packard <keithp@keithp.com>

os/osinit.c: no getpgrp() and setpgrp() on WIN32

Signed-off-by: Ryan Pavlik <rpavlik@iastate.edu>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Tested-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Keith Packard <keithp@keithp.com>

hw/xwin: use raise() instead of kill()

MinGW doesn't have kill(), so use raise() instead

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Keith Packard <keithp@keithp.com>

os/utils.c: Fix compilation of OsBlockSIGIO when SIGIO isn't defined

Fix compilation of OsBlockSIGIO with -Werror=return-type when SIGIO isn't
defined.

/jhbuild/checkout/xorg/xserver/os/utils.c: In function 'OsBlockSIGIO':
/jhbuild/checkout/xorg/xserver/os/utils.c:1248:1: error: control reaches end of non-void function [-Wreturn-type]

v2: Shuffle around to avoid writing unreachable code

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Tested-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Keith Packard <keithp@keithp.com>

os/osinit.c, os/utils.c: Exclude sigaction code when building for MinGW

MinGW doesn't have sigaction, so this patch is needed for building.

No attempt is made to actually install the fatal error signal handler, as MinGW
will simply terminate the process rather than deliver a fatal signal.

Also avoid using strsignal

Signed-off-by: Ryan Pavlik <rpavlik@iastate.edu>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Tested-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Keith Packard <keithp@keithp.com>

xf86: select a fake output for headless servers

Following commit 37d956e3ac9513b74078882dff489f9b0a7a5a28
Author: Keith Packard <keithp@keithp.com>
Date:   Mon Sep 10 11:14:20 2012 +1000

    xf86: fix compat output selection for no output GPUs

headless servers can no longer startup as we no longer select a compat
output for the fake framebuffer.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56343
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>

xf86: Fix non-PCI configuration-less setups

For non-PCI video devices, such as those found on many ARM embedded
systems, the X server currently requires the BusID option to specify the
full path to the DRM device's sysfs node in order to properly match it
against the probed platform devices.

In order to allow X to start up properly if either the BusID option was
omitted or no configuration is present at all, the first video device is
used by default.

Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>

Merge remote-tracking branch 'daniels/master'

hw/xwin: Don't spam MotionNotify events when the mouse hasn't moved

Don't spam MotionNotify events when the mouse hasn't moved, when polling the
mouse position outside any X window

(Test with 'xev -root' after mouse polling has started.)

Signed-off-by: Colin Harrison <colin.harrison@virgin.net>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>

hw/xwin: Don't log all fbConfigs and GL/WGL extensions

Don't log all fbConfigs and GL/WGL extensions, unless verbose logging
is requested
Log the number of pixelFormats which gave rise to the fbConfigs

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Use pre-computed extent of damage for creating GDI region

When -clipupdates option is specified, use the pre-computed extent of damage,
rather than computing a GDI region which combines all the individual boxes in
the damage.

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Ensure we own the clipboard before checking the format of it's contents

Ensure we own the clipboard before checking the format of it's contents, this
prevents the contents from changing underneath us.

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Disable minimize button on window with skip-taskbar state

If a window has had its taskbar button removed, disable its minimize
button to prevent it becoming lost

Signed-off-by: Colin Harrison <colin.harrison@virgin.net>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>

hw/xwin: Process _NET_WM_STATE_SKIP_TASKBAR hint in multiwindow mode.

Set WS_EX_TOOLWINDOW style to hide window from Alt-Tab switcher

Use ITaskBarList interface to ensure that the taskbar notices if the window has
changed it's style in a way which affects if the taskbar shows it or not

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Add a simple interface to the ITaskbarList COM interface

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Enable RANDR resize by default

Also fix erroneous use of '--resize' not '-resize', and document '-noresize' in man page

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Enable AIGLX by default

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Restore non-latching modifier key state when an X window gains focus

In multiwindow mode, the state of the modifier keys was lost when a window is
created (or raised) and focus moved to that window.

For example: In window A Ctrl + some key opens a window B, then in window B Ctrl
+ some other key triggers the next action. However after the opening of window B
the Ctrl key has to be released and pressed again. If the user keeps the Ctrl
key held down when the window B is opened, the next key press X will be
interpreted as X and not as Ctrl+X.

Extended the function winRestoreModeKeyStates in winkeybd.c to consider not only
the latching modifier keys but also the modifiers Ctrl, Shift, Alt/AltGr by
using the Windows function GetAsyncKeyState.

A combined Ctrl+AltGr modifier state cannot be restored correctly, as Windows
always fakes a Ctrl-L when AltGr is pressed.

Signed-off-by: Oliver Schmidt <oschmidt-mailinglists@gmx.de>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Give our logical xor operator a more logical name

Also, rather than a comment about why we need a logical operator, let's have a
comment about what we are doing to the keyboard state...

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Add correct taskbar grouping of X windows on Windows 7

I created a small patch for XWin that adds correct grouping of taskbar icons
when 'Always combine, hide labels' is set in the taskbar properties. It uses the
new taskbar APIs introduced in Windows 7 to set an application id for each
window. The id is based on the X11 class hints.

v2: Add file to _SOURCES to fix distcheck

v3 : Fix compilation with mingw-w64 w32api headers
Include propkey.h, propsys.h rather than defining necessary stuff ourselves

v4: Fix up names taskbar->propertystore, AppID->AppUserModelID, etc.
Link directly with ole32 for PropVariantClear(), prototyping it if neccessary.

v5: Put winSetAppUserModelID()-related code in a separate file.
Drop a superfluous assign to hr of ignored HRESULT of SetValue()

Signed-off-by: Tobias Häußler <tobias.haeussler@gmx.de>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Fix function name in log message

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Allow XScreenSaverSuspend() to effect Windows screen saver or monitor power off

Reset the idle timer in the block handler if screenSaverSuspended is set.

This isn't quite a complete solution.  We should also set the block timeout to
something less than the idle timer timeout to ensure we will reset the idle
timer before it times out if we are blocking.

The idle timer timeout is presumably the first one to expire of the screen saver
activation or monitor low power or power down timeout, depending on
configuration.

Unfortunately this is probably not straightforward to do. Whilst
SystemParametersInfo(SPI_GETSCREENSAVETIMEOUT) is portable, apparently
SPI_GETLOWPOWERTIMEOUT and SPI_GETPOWEROFFTIMEOUT are not supported by Windows
versions 6.0 or later, and the interface for discovering equivalent value is
complex.

This doesn't matter in the case where a media player or similar application is
the one making the XScreenSaverSuspend() requests, as it will be continuously
drawing, causing the X server to become unblocked often.

In the case where slide show presentation application or similar is the one
making the XScreenSaverSuspend() request, this might be a problem.

Since "Powerpoint is Evil" [1], I think we'll leave it like this till someone
complains :-)

[1] http://www.edwardtufte.com/tufte/books_pp

v2: conditionalize on SCREENSAVER

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Avoid logging an extra blank line if BUILDERSTRING is empty

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Report which drawing engines are being enabled when checking for support

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Show any fatal error message

Report the fatal error message in the dialog we pop up, rather than just
referring the user to the logfile.

v2: Do this a better way since the "Pass the FatalError message to
OsVendorFatalError" patch has landed, and OsVendorFatalError() now gets passed
the fatal error message

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

Ephyr: Find the right host screen when embedded

When Xephyr is embedded into another application (e.g. in xoo), the
input events may come from the parent Window instead of our own. So make
sure we find the host screen in that case as well instead of crashing.

Signed-off-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>

vfb: Initialize the GLX extension again.

This should fix a regression in the 1.13.0 release: commit
5f5bbbe543f65c48ecbb5cce80116a86ca3fbe86 removed a code path used by Xvfb
and made it use the default one when initializing extensions. However, this
meant the GLX extension was not initialized anymore since it is not part of
the `staticExtensions' array.

Since it is not possible to just add it to that array after commit
aad428b8e21c77397c623b78706eb64b1fea77c9, adopt an approach similar to
xwin's and xquartz's and initialize the extension from vfb's `InitOutput'.

Signed-off-by: Raphael Kubo da Costa <raphael.kubo.da.costa@intel.com>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>

DMX: Add DMX and GLX extensions

Unfortunately this also got lost in the extmod fallout, leaving the DMX
server not exposing the DMX or GLX extensions.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

Constify extensions in LoadExtension users

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

Constify argument to LoadExtension

Since we never modify it.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

dix: Save touchpoint last coordinates before transform. #49347

DDXTouchPointInfoRec.valuators used to store axis values after transform.
This resulted in Coordinate Transformation Matrix
being applied multiple times to the last coordinates,
in the case when only pressure changes in the last touch event.

Changed DDXTouchPointInfoRec.valuators to store values before transform.

Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=49347

Signed-off-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Simplify GetTouchEvents

With only one callee left, we are free to assume that
!(flags & TOUCH_CLIENT_ID)

Signed-off-by: Thomas Jaeger <ThJaeger@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Don't use GetTouchEvents in EmitTouchEnd

As before GetTouchEvents causes unwanted side effects.  Add a new
function GetDixTouchEnd, which generates a touch event from the touch
point.  We fill in the event's screen coordinates from the MD's current
sprite position.

Signed-off-by: Thomas Jaeger <ThJaeger@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Don't use GetTouchEvents when replaying events

GetTouchEvents has plenty of side effects such as moving the pointer or
updating the master device, which we don't want to happen when
replaying.  The only reason for calling it was to generate a DCCE event,
but GetTouchEvents doesn't even do that right (we might need a DCCE
event even when replaying a master event, or clients could interpret
valuator data incorrectly).

This discussion is moot at the moment anyway, since DeliverTouchEvents
doesn't appear to deliver DCCE events.

Signed-off-by: Thomas Jaeger <ThJaeger@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
[Added call to processInputProc instead of direct call to DeliverTouchEvents]

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Update the MD's position when a touch event is received

Signed-off-by: Thomas Jaeger <ThJaeger@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Merge remote-tracking branch 'whot/for-keith'

linux: Prefer ioctl(KDSKBMUTE, 1) over ioctl(KDSKBMODE, K_OFF)

K_OFF is a slightly broken interface, since if some other process
(cough, systemd) sets the console state to K_UNICODE then it undoes
K_OFF, and now Alt-F2 will switch terminals instead of summoning the
Gnome "run command" dialog.

KDSKBMUTE separates the "don't enqueue events" logic from the keymap, so
doesn't have this problem.  Try it first, then continue falling back to
older methods.

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=859485
Tested-by: Josh Boyer <jwboyer@redhat.com>
Signed-off-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

xfixes: let PointerBarrierClient->device_ids point to allocated memory

"Fixed" up Jasper's original patch [1] to use pointers instead of
zero-length arrays but didn't actually set the pointer to anywhere.

[1] 9a536820f6174befb22431bab375ef1af0a1ec29

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-By: Jasper St. Pierre <jstpierre@mecheye.net>

Xi: Fix modifier swapping in XIPassiveGrabDevice

XIPassiveGrabDevice uses a list of uint32_t as modifier sets.

The ModifierInfo struct represents the current modifier states and is
therefor used in XIQueryPointer and various events.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

remove init_event

The function is identical to init_device_event from inpututils.c with
the first two arguments swapped.

Signed-off-by: Thomas Jaeger <ThJaeger@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

XKB: Remove component listing support

No-one uses this - not xkbcomp, not GNOME, not KDE.  The preferred way
to deal with component listing (which gives you RMLVO rather than
KcCGST) is to use the XML files on the client side.

Indeed, a couple of hours after making this commit, it emerged that all
*.dir files built with xkbcomp 1.1.1 (released two years ago) and later
have been catastrophically broken and nearly empty.  So I think that's
reasonable proof that no-one uses them.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

input: drop FP1616 macro

The double_to_f1616() functions do the same thing, and they're tested.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>

XQuartz: Opt-in to NSSupportsAutomaticGraphicsSwitching

http://xquartz.macosforge.org/trac/ticket/668

Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>

Merge remote-tracking branch 'whot/for-keith'

Xi: Set modifier mask on touch events

Button mask should be out-of-band with the emulated
pointer events as touch devices don't truly have
"buttons". Even though, it's handy to have the modifier
mask from the paired keyboard on touch events.

Signed-off-by: Carlos Garnacho <carlosg@gnome.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Xi: Update the device after delivering the emulated pointer event(#56558)

Ensure emulated pointer events contain the state that applies before the
event was processed, so the device state must be updated after delivering
such emulated events.

Co-authored-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Carlos Garnacho <carlosg@gnome.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

cursor: Fix up implementation for per-device barriers

Support multiple mast devices being specified as a parameter to the
barrier. This should implement all parts of the XFixes specification,
minus the existing non-specification detail where specifying no devices
is the same as specifying XIAllDevices.

Signed-off-by: Jasper St. Pierre <jstpierre@mecheye.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

cursor: Clean up barrier finding code a bit

The current code manually unrolls the loop, by finding a barrier,
clamping it, and then repeaing the code. Since we want to modify
both passes of the loop to support device-specific barriers, make
it into an actual loop.

Signed-off-by: Jasper St. Pierre <jstpierre@mecheye.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

cursor: Clean up pointer barrier creation code a tiny bit

This will make it much simpler when we add more error paths to the code
that constructs pointer barrier clients.

Signed-off-by: Jasper St. Pierre <jstpierre@mecheye.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

cursor: CreatePointerBarrier has a variable request length

Support this, and swap the devices in the SProc as well. Don't
actually do anything with the devices just yet -- specifying any
should still emit a BadImplementation.

Based on a patch by Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Jasper St. Pierre <jstpierre@mecheye.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

cursor: Fix a minor unused variable warning

Signed-off-by: Jasper St. Pierre <jstpierre@mecheye.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

dix: don't allow disabling XTest devices

Disabling a XTest device followed by an XTest API call crashes the server.
This could be fixed elsewhere but disabled devices must not send events
anyway. The use-case for disabled XTest devices is somewhat limited, so
simply disallow disabling the devices.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>

Apply partial matches for option (#25873)

Rules which match star (*) and option, like one below, should be applied

layout[2] option    = symbols
*         misc:typo = +typo(base)

This is port of patch from #19563 (https://bugs.freedesktop.org/19563)
because here we have own copy of maprules.c

X.Org Bug 25873 <http://bugs.freedesktop.org/show_bug.cgi?id=25873>

Signed-off-by: Alexey Ten (Lynn) <alexeyten@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

os: Add libnettle as a choice of SHA1 implementation

libnettle is smaller than libgcrypt, currently being released more
frequently, and has replaced the latter in gnutls-3.x (which is used
by TigerVNC, so they can avoid pulling in two crypto libraries
simultaneously).

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Julien Cristau <jcristau@debian.org>

Xi: fix fprint format warning

exevents.c: In function 'ProcessTouchEvent':
exevents.c:1601:20: warning: too many arguments for format

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

Xext: fix shadow warnings

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

Xext: fix redundant redeclaration warnings

panoramiX.c:595:13: warning: redundant redeclaration of 'CreateConnectionBlock'
../include/dix.h:167:23: note: previous declaration of 'CreateConnectionBlock' was here
xres.c:193:13: warning: redundant redeclaration of 'ResExtensionInit'
../include/extinit.h:109:13: note: previous declaration of 'ResExtensionInit'
xtest.c:60:12: warning: redundant redeclaration of 'DeviceValuator'
../Xi/exglobals.h:61:12: note: previous declaration of 'DeviceValuator' was here

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

Xext: fix unused variable warnings in xf86bigfont.c

These were rendered unused by commit 2c7c520cfe0df30f4bc3adba59d9c62582823bf8.

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

xkb: fix shadow warnings

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

render: fix shadow warnings

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

randr: export more provider property symbols

These were added as part of commit 66d92afeaeed9f4a19267d95a1f81b9bf27162a5
but never declared or exported.  Fixes warnings:

rrproviderproperty.c:255:1: warning: no previous prototype for 'RRPostProviderPendingProperties'
rrproviderproperty.c:327:1: warning: no previous prototype for 'RRConfigureProviderProperty'

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

mi: fix shadow warnings

mibitblt.c: In function 'miGetImage':
mibitblt.c:617:20: warning: declaration of 'pt' shadows a previous local
mibitblt.c:609:17: warning: shadowed declaration is here
mispans.c: In function 'miFillUniqueSpanGroup':
mispans.c:456:33: warning: declaration of 'i' shadows a previous local
mispans.c:382:9: warning: shadowed declaration is here
mispans.c:488:17: warning: declaration of 'i' shadows a previous local
mispans.c:382:9: warning: shadowed declaration is here

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

fb: fix shadow warnings

fbblt.c: In function 'fbBlt':
fbblt.c:76:16: warning: declaration of 'src' shadows a previous local
fbblt.c:52:13: warning: shadowed declaration is here
fbblt.c:77:16: warning: declaration of 'dst' shadows a previous local
fbblt.c:52:19: warning: shadowed declaration is here
fbbltone.c: In function 'fbBltPlane':
fbbltone.c:742:13: warning: declaration of 'w' shadows a previous local
fbbltone.c:725:9: warning: shadowed declaration is here

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

dix: fix shadow warnings

dispatch.c: In function 'ProcCopyArea':
dispatch.c:1608:5: warning: declaration of 'rc' shadows a previous local
dispatch.c:1604:9: warning: shadowed declaration is here
dispatch.c: In function 'ProcCopyPlane':
dispatch.c:1647:5: warning: declaration of 'rc' shadows a previous local
dispatch.c:1643:9: warning: shadowed declaration is here
events.c: In function 'GetClientsForDelivery':
events.c:2030:68: warning: declaration of 'clients' shadows a global declaration
../include/dix.h:124:28: warning: shadowed declaration is here
events.c: In function 'DeliverEventToWindowMask':
events.c:2113:19: warning: declaration of 'clients' shadows a global declaration
../include/dix.h:124:28: warning: shadowed declaration is here
events.c: In function 'EventSuppressForWindow':
events.c:4420:12: warning: declaration of 'free' shadows a global declaration

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

dix: fix redundant redeclaration warnings in dixfont

These functions are already declared in <X11/fonts/fontproto.h>.
Redeclaring them just for _X_EXPORT causes tons of warnings throughout
xserver, but they need to be declared somewhere to be picked up by
sdksyms.sh.  Doing so in a private header limits the warnings to
sdksyms.c; fixing those as well would require changes to fontsproto.

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

xfree86: os-support: fix old-style function definition warnings

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

macros: clarify documentation

Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

Merge remote-tracking branch 'jturney/mingw-w64-w32api'

hw/xwin: Wrap 'Status' when including ddraw.h

Status is #defined as an alias for a type in xkbsrv.h, which conflicts with it's
use as a parameter name in rpcdce.h

This fixes compilation with MinGW-w64 w32api headers

(The MinGW-w64 w32api headers actually provide a ddraw.h, so this fix tries to
do things in a way which will still work if our private copy of ddraw.h is
eventually removed)

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Remove pointless winFinishCreateWindowsWindowDDNL()

Remove pointless winFinishCreateWindowsWindowDDNL() and the useless DirectDraw
surface pointers in the window privates

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

hw/xwin: Link with libdxguid rather than defining the DirectX GUIDs ourselves

Decruftify by linking with libdxguid rather than defining the DirectX GUIDs
ourselves. This fixes compilation with MinGW-w64 w32api headers, which do not
provide GUID_SECT (which only ever did anything for gcc prior to 2.95, anyhow)

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>

Merge remote-tracking branch 'whot/for-keith'