xserver 1.7.6

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

Revert "XQuartz: Explicitly pass a bellProc to make XBell() work again."

I'm not quite sure why this was necessary, but DDXRingBell is being called
from CoreKeyboardBell, so we don't need a separate bellProc which would
result in multiple rings.

This reverts commit 7e7e7935eeb86b5990664a662471800cfd2e7ce0.

Conflicts:

hw/xquartz/quartzKeyboard.c

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

XQuartz: Use an empty xkb keymap by default

Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
(cherry picked from commit 5172253bae3b9867118c6717434e73c173acd5e9)

configure: restore SHA1_LIB for XSERVER_SYS_LIBS

Accidentally removed in ff5fb43a4b38c707a, caused by
a botched merge when cherry-picking 178da6534fe7ab4b99fb.

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

xserver 1.7.5.902

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

Merge branch 'server-1.7-nominations' into server-1.7-branch

Xext: Link to external libraries when necessary.

Although the DDX should be linked to the necessary libraries, we may
also need to pull them in directly to the module to ensure the symbols
are resolved at runtime. Should fix this bug with XSELINUX:

/usr/bin/X: symbol lookup error:
/usr/lib64/xorg/modules/extensions/libextmod.so: undefined symbol:
is_selinux_enabled

-v2: use _LIBADD instead of _LIBS; remove SELINUX_LIBS from
     XSERVER_SYS_LIBS as it should only be needed in extmod.

Signed-off-by: Dan Nicholson <dbn.lists@gmail.com>
Signed-off-by: Eamon Walsh <ewalsh@tycho.nsa.gov>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 178da6534fe7ab4b99fb87925c04e6f963c88583)

Conflicts:

configure.ac

Record: Avoid duplicates from replaying frozen events

Reintroduce a check which used to be there in the old
ProcessKeyboardEvent/ProcessPointerEvent codepath, which avoids us
recording events subject to a grab twice: once when it's first processed
in EnqueueEvent, and then again when it's thawed and being replayed.

This required a tiny amount of code motion to expose syncEvents.

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>
(cherry picked from commit a2ea8c2f2cc53607d57d60f41c879380ea70cd02)

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

dri2: return FALSE not BadValue.

With 1.7 if you try and send a DRI2Connect with this value set
to anything but 0, the server will segfault due to the strlen
in the function this returns to, since BadValue != FALSE.

Fixed via other work in 1.8.

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

Revert "dix: Use DeliverGrabbedEvent for implicit passive grabs (#25400)"

Reported-by: Florian Mickerl <florian@mickler.org>
"this regresses my desktop. fluxbox is not able to move the windows around
anymore. also popup of context-menue does not work. (well sometimes it does,
but then the fluxbox-keyboard-shortcuts do not work)"

This reverts commit 351d11fd466fae731df860a6ff4cf50ad565f43f.

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

xserver 1.7.5.901

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

Allow for missing or disabled compat_output

When the compat output is missing (I don't think this is actually
possible), or is disabled (and hence has no crtc), we would like to
avoid dereferencing NULL pointers. This patch creates inline functions
to extract the current compat output, crtc or associated RandR crtc
structure, carefully checking for NULL pointers everywhere.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit de86a3a3448f0a55c1cd99aee9ea80070a589877)

configure: move SELINUX_LIBS to XSERVER_SYS_LIBS

All ddxs require linking against selinux if enabled.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Dan Nicholson <dbn.lists@gmail.com>
Reviewed-by: Eamon Walsh <ewalsh@tycho.nsa.gov>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 1c2abec479cfbac9aeadc121af0825d368317658)

Conflicts:

configure.ac

configure: move libselinux requirement up to the common section.

Let's have all version-specific requirements in one block.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Dan Nicholson <dbn.lists@gmail.com>
Reviewed-by: Eamon Walsh <ewalsh@tycho.nsa.gov>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit cab99b947c66a42dbf8a906bdb02e23aa7cab519)

Conflicts:

configure.ac

dix: Use DeliverGrabbedEvent for implicit passive grabs (#25400)

A client requesting a GrabModeSync button grab, owner-events true, with only
the ButtonRelease mask set would never receive the press event even if the
grab window had the ButtonPress mask set.

The protocol requires that if owner-events is true, then the delivery mask
is the combination of the grab mask + the window event mask.

DeliverGrabbedEvents does this already for us, checking first the delivery
based on owner_events and then based on the grab mask. AFAICT, the device
cannot enter the states FREEZE_BOTH_NEXT_EVENT or FREEZE_NEXT_EVENT that
would be handled by DGE in any possible path here.

Bonus point - CheckPassiveGrabsOnWindows suddenly becomes a lot lesss
complicated.

X.Org Bug 25400 <http://bugs.freedesktop.org/show_bug.cgi?id=25400>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit cf72b5437d2d620521279077a29c5df6d0fbb576)

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

Solaris xf86OSRingBell() off-by-one error in filling iov[] array

When generating sound buffers for /dev/audio bells, insert waveform
for beep *or* silence, but not both, so we don't write one entry past
the end of the iov buffer when the final bit of soundwave ends up in
the final entry allocated in the iov array.

Fixes OpenSolaris bug 6894890:
 http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6894890

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>

dix: remove now-erroneous comment about frozen slave devices.

A direct grab on a slave device through XI2 detaches it, regardless of
whether the grab is sync or async. So this comment doesn't apply to XI2
anyway.

For XI1, aside from your life being miserable already, it doesn't matter as
XI1 does not have a concept of attachment. You can freeze a device and if
you don't freeze _all_ other devices at the same time, the master device can
still happily send events to the client.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Adam Jackson <ajax@redhat.com>
(cherry picked from commit c0d941913199ed251fceab62fcf65437f6f9bd31)

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

dix: try to ring the bell even if the current device doesn't have one. (#24503)

Evdev devices do not have the bell proc set, but XTEST devices do. By
exiting early, the bell only rings if the last keyboard used was the XTEST
keyboard and hence the bell proc is still set on the master but not if an
evdev keyboard was used last.

The better approach here is to try to ring the bell on all devices attached
to this master device in case one or more actually do produce an audible
sound. That's also XKB's behaviour if XkbUseCoreKbd is specified as device
identifier.

X.Org Bug 24503 <http://bugs.freedesktop.org/show_bug.cgi?id=24503>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Adam Jackson <ajax@redhat.com>
(cherry picked from commit 758f6971750ed507e64eee817d720a77181439f2)

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

xselinux: Bump extension minor version.

Changes introduced in this version:
- 3 window-related requests now handle pixmaps also.

Signed-off-by: Eamon Walsh <ewalsh@tycho.nsa.gov>
Reviewed-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit db4f676f25c6d8e58263d5151942be730592d444)

xselinux: Rename window-related requests that now support pixmaps.

Renamed requests:
SetWindowCreateContext -> SetDrawableCreateContext
GetWindowCreateContext -> GetDrawableCreateContext
GetWindowContext -> GetDrawableContext

Signed-off-by: Eamon Walsh <ewalsh@tycho.nsa.gov>
Reviewed-by: Keith Packard <keithp@keithp.com>

xselinux: Allow GetWindowContext to be used for pixmaps as well.

Signed-off-by: Eamon Walsh <ewalsh@tycho.nsa.gov>
Reviewed-by: Keith Packard <keithp@keithp.com>

xselinux: Remove reference counting calls for SID objects.

Starting with libselinux 2.0.86, SID objects are no longer
reference counted and the sidput() and sidget() calls are no-ops.

Signed-off-by: Eamon Walsh <ewalsh@tycho.nsa.gov>
Reviewed-by: Keith Packard <keithp@keithp.com>

libselinux now has a pkgconfig file.  Use it.

Also remove HAVE_NETLINK_AVC_ACQUIRE_FD tests, because we now
require a version of libselinux that has it.

Signed-off-by: Eamon Walsh <ewalsh@tycho.nsa.gov>
Reviewed-by: Keith Packard <keithp@keithp.com>

Xi: assume BadMode for non-BadMatch errors returned from SetDeviceMode.

The XI protocol spec only allows for two errors on the SetDeviceMode
requests: BadMatch or BadMode. BadMode however is a dynamically assigned
extension error and the driver doesn't have access to the actual error
number. Hence, if a SetDeviceMode driver returns an error other than
BadMatch, assume BadMode.

The two exceptions are BadAlloc and BadImplementations, pass these on to the
client (any request is allowed to return either of those).

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Dan Nicholson <dbn.lists@gmail.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit e40ba5798c367bec584f6437b23d2c5f801fd013)

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

Re-enable RECORD extension.

RECORD was disabled during the switch to internal events. This patch
modifies the record callback to work with internal events instead of
xEvents. The InternalEvents are converted to core/Xi events as needed.

Since record is a loadable extension, the EventTo* calls must be externed.

Signed-off-by: Chris Dekter <cdekter@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 122fc0e7a0712df4ec2bd9ca6773f90bcd2095cf)

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

xfree86: vgaarb: remove useless debug

This is RAC's remnant. Any sane person would use a more wise method of
debugging instead.

X.Org Bug 26074 <http://bugs.freedesktop.org/show_bug.cgi?id=26074>

Signed-off-by: Tiago Vignatti <tiago.vignatti@nokia.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 5e81078cf56aabbf6551d96d312c7840ba9370ae)

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

dix: don't update the slave coordinates from the VCK.

A keyboard event from a device with both valuators and keys will be posted
through the VCK. In this case, do not update the slave device coordinates
from the VCK - they're always 0/0. Leave them as-is, for the next pointer
event will continue where it left.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 6f265d55a61f9be323583b8acacae783be72bda9)

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

XQuartz: Fix a possible buffer overrun in quartzAudio

Also dropped deprecated API while there

Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
(cherry picked from commit 4e8bf12b13690afa6d9fee0e339d3819ef16fb3f)

XQuartz: clang static analysis fixes

Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
(cherry picked from commit 4127e8e5be5c366eb752bbb8d7f0ff5b519e641b)

xserver 1.7.5

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

dix: restore lastDeviceEventTime update in dixSaveScreens

This was removed in 6b5978dcf1f7ac3ecc2f22df06f7000f360e2066 (Do not
reset lastDeviceEventTime when we do dixSaveScreens), but caused a
regression for XResetScreenSaver.  Add the lastDeviceEventTime update
back, but restrict it to that case.

X.Org bug#25855 <http://bugs.freedesktop.org/25855>

Reported-by: Lubos Lunak <l.lunak@suse.cz>
Tested-by: Lubos Lunak <l.lunak@suse.cz>
Signed-off-by: Julien Cristau <jcristau@debian.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 001ce71dc11287dc94cc2fbc5d35677c046e6c04)

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

Don't double-swap the RandR PropertyNotify event

The event is already swapped in randr.c/SRROutputPropertyNotifyEvent, so
it should not be swapped here.

X.Org Bugzilla #26511: http://bugs.freedesktop.org/show_bug.cgi?id=26511

Tested-by: Leonardo Chiquitto <leonardo@ngdn.org>
Acked-by: Adam Jackson <ajax at redhat.com>
Reviewed-by: Julien Cristau <jcristau at debian.org>
Signed-off-by: Peter Harris <pharris@opentext.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 97b03037f4d99fcebc7603011f41c3aff9871ce2)

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

Xi: reset the sli pointers after copying device classes. (#25640)

If the indicator flags have the XkbSLI_IsDefault bit set, the indicator map
and names aren't their own bit of memory but rather point into the
device->key->xkbInfo->desc structure. XkbCopySrvLedInfo knows about this and
leaves the pointers alone.

When copying the classes from the slave to the master, these pointers are
copied and still point to the dev->key class of the slave device. If the
slave device is removed, the memory becomes invalid and a call to modify
this data (e.g. XkbSetIndicators) may cause a deadlock.

The copying of dev->key relies on dev->kbdfeed to be already set up. Hence
the pointers need to be reset once _both_ kbdfeed and key have been copied
into the master device.

X.Org Bug 25640 <http://bugs.freedesktop.org/show_bug.cgi?id=25640>
Fedora Bug 540584 <https://bugzilla.redhat.com/show_bug.cgi?id=540584>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 48f7298657f91843db36566b8d66d6c4c18dbd4c)

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

RENDER: Fix gradient and solid fill pictures with Xinerama, and misc cleanup

If these aren't wrapped, then procs that are wrapped (such as
RenderChangePicture) will fail in Xinerama when they see the resource
type of a picture created through one of these interfaces is PictureType
and not XRT_PICTURE like those allocated via RenderCreatePicture.

Signed-off-by: Robert Morell <rmorell@nvidia.com>
Reviewed-by: Aaron Plattner <aplattner@nvidia.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 4d575b0559817258f7a0ce6c4d2d0f9e7e5bba63)

xserver 1.7.4.902

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

Allow driver to call DeleteInputDeviceRequest during UnInit

When the input driver (like xf86-input-wacom) removes it's devices
during a call to UnInit, the CloseDownDevices() cannot handle it. The
"next" variable can become a pointer to freed memory.

The patch introduces order-independent device freeing mechanism by
remembering the already freed device ids. The devices can reorder any
time during freeing. No device will be double-freed - if the removing
failed for any reason; some implementations of DeleteInputDeviceRequest
don't free the devices already.

Signed-off-by: Oldřich Jedlička <oldium.pro@seznam.cz>
Reviewed-by: Simon Thum <simon.thum@gmx.de>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 08b22c7faf97217ea4d497eec6624fc3dd916d9b)

xkb: make ctrl+alt+keypad + / ctrl+alt+keypad - work again (#25743)

Video mode switching via keypad keys did not work

Signed-off-by: Horst Wente <horst.wente@acm.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit b91cec26de367f75388e620772747b08eee34294)

Fix typo in updateSlaveDeviceCoords

The index [0] for the second valuator looks bogus; fix it.

Signed-off-by: Oldřich Jedlička <oldium.pro@seznam.cz>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 993e78d6c433f65bd8a87890a6bb6da480b2cdc4)

Render: Fix request size verification

RenderSetPictureClipRectangles and the Xinerama version of
RenderChangePicture were using the wrong structure types for request
size verification.

Signed-off-by: Robert Morell <rmorell@nvidia.com>
Reviewed-by: Aaron Plattner <aplattner@nvidia.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 8d6b1838330625b758ebc55bb0df1b425c5cefaf)

xfree86: remove man page reference inexistent acceleration profile

not implemented in the X.org server 1.7

Signed-off-by: Simon Thum <simon.thum@gmx.de>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

XQuartz: Attatch a stub display when CoreGraphics reports no displays.

This is half of the required changes to address the "stuck mouse pointer"
bug that occurs when X11 launches while the displays are asleep.  The
remainder of the fix is part of libXplugin which needs to be updated to
send XP_EVENT_DISPLAY_CHANGED when the display wakes up.  If you don't
have a recent enough libXplugin (expected in 2.5.0_beta2 or later), you
can cause this event to be sent by changing your display resolution (or
you could just start X11.app with your screens awake).

Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
(cherry picked from commit 52456c602c3cdd7d5eac677889a18fad37dfb7ae)

Avoid segfaults in XF86VidMode GammaRamp functions if randr_crtc is NULL

Fixes crash when xscreensaver tries to use GammaRamp calls to fade out
http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6915712

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>

Solaris: Avoid switching to inactive VT's

Fix for OpenSolaris bug 6876992: "[vconsole] Ctrl+Alt+F12 switchs to blank
console screen with hotkeys property turned-off"
http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6876992

Xorg needs to do sanity test for the VT it is commanded to switch to.
If the VT is not opened by any process, discard the switching request.

The changes also contain the fix for some flaws discovered when
getting the new gdm to run.

Signed-off-by: Aaron Zang <Aaron.Zang@Sun.COM>
Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>

Fix source pictures getting random transforms after 2d6a8f668342a5190cdf43b5.

*xoff and *yoff were uninitialized for source-only pictures.x

Signed-off-by: Pierre-Loup A. Griffais <pgriffais@nvidia.com>
Reviewed-by: Aaron Plattner <aplattner@nvidia.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit a6bd5d2e482a5aa84acb3d4932e2a166d8670ef1)

x86emu: Respect the LEA 67h address size prefix.

Signed-off-by: Christian Zander <chzander@nvidia.com>
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Tested-by: Tiago Vignatti <tiago.vignatti@nokia.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit f57bc0ede8e018c7e264b917927c42a018cd1d5a)

xserver 1.7.4.901

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

XQuartz: Update copyright in bundle for 2010

Signed-off-by: Jeremy Huddleston <jeremyhu@freedesktop.org>
(cherry picked from commit 6bde306f7f6b9bbabeaa8bb910ea549be906cd8b)

XQuartz: Setup the modifier map in the quartz thread

This avoids possible doing it twice which could result in incorrect
keycodes for alt due to our loss of information about its side.

Signed-off-by: Jeremy Huddleston <jeremyhu@freedesktop.org>
(cherry picked from commit 6008cc116493cb2825ad0bda0b407b7aefabb3f4)

os: state effect of -a and -t options more precisely

Signed-off-by: Simon Thum <simon.thum@gmx.de>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 14e4e4a294e648e0bdcb70c34748e1b81c5bb64f)

macros: use PKG_CONFIG variable rather than executable name

User can defined alternate location for pkg-config.
Once option in place, all instances of pkg-config must be converted.

Acked-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Dan Nicholson <dbn.lists@gmail.com>
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 6a2a57832709798d99d19ff4c587e22f3b3c150d)

configure: use backticks rather than $() for commands

This patch to xserver configure.ac is to increase code portability to
non POSIX system by using backticks rather than $() for command
substitution for BUILD_DATE and BUILD_TIME.

Reviewed-by: Rémi Cardona <remi@gentoo.org>
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 6313d2da6c6910827d68cf31fe00b46a34c5bfc7)

xfree86: document pointer acceleration in xorg.conf.man

Signed-off-by: Simon Thum <simon.thum@gmx.de>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Fernando Carrijo <fcarrijo@yahoo.com.br>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 0722c287a4c8a6cdedca9756192547bfcf77ade5)

Conflicts:

hw/xfree86/doc/man/xorg.conf.man.pre

doc: actually document SendDragEvents

Signed-off-by: Simon Thum <simon.thum@gmx.de>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 14039b5a7982fbf8130501bb00766176a4e9bccb)

Conflicts:

hw/xfree86/doc/man/xorg.conf.man.pre

xserver 1.7.4

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

Do not define members of include/eventstr.h:EventType enum conditionally.

Main problem is that EventType enumeration members can be different in
module and in server, which obviously causes problems.

Signed-off-by: Adam Tkac <atkac@redhat.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 8d53d84485fdce8ea9686e6f300a69f7ddebd467)

CloseDevice: call XkbRemoveResourceClient before freeing key class struct

XkbRemoveResourceClient() returns immediately if dev->key is NULL.
CloseDevice calls XkbRemoveResourceClient until it removes all resources.

If we free dev->key and NULL it before XkbRemoveResourceClient, then
infinite loop ensues, and the server appears to hang on exit or crash.

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 13c8bd3fde3b0831921e59f84936022a16379d63)

xserver 1.7.3.902

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

xkb: don't assign garbage value to led_return.

As the comment for the function states, led_return is undefined if map is
NULL. We might as well skip writing to it then.

Found by clang.

Reported-by: Tomas Carnecky <tom@dbservice.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Jamey Sharp <jamey@minilop.net>
(cherry picked from commit 12fb31815db9de9c01f2d4155a2b74531777c0bf)

xfree86: belately init RandR12 if xinerama fails. (#24627)

On Fri, Dec 11, 2009 at 10:19:01AM -0800, Keith Packard wrote:
> On Wed, 9 Dec 2009 11:55:14 +1000, Peter Hutterer <peter.hutterer@who-t.net> wrote:
> > On Tue, Dec 08, 2009 at 05:24:06PM -0800, Aaron Plattner wrote:
> > > On Tue, Dec 08, 2009 at 03:52:27PM -0800, Peter Hutterer wrote:
> > > > Xorg +xinerama crashes immediately due to whacky dependency between Xinerama
> > > > and RandR12. The latter doesn't initialize if Xinerama is enabled, but if
> > > > only one screen is found, Xinerama is disabled again and RandR12 tries to
> > > > access data it never initialized.
>
> I'd sure like to have RandR get enabled when xinerama doesn't; is there
> an easy way of making that happen here? Perhaps having the RandR12 code
> disable Xinerama when only one screen is found? Or some other kludge?

you know the dependency better than I do so any hints are apreciated.
afaict, the screenInfo.numScreens (the check used by Xinerama) isn't
necessarily initialized at this point so we can't use the same check.
The following seems to work though:

From 670b3ebdb7312a6433a8f093d0820785db2aea20 Mon Sep 17 00:00:00 2001
From: Peter Hutterer <peter.hutterer@who-t.net>
Date: Mon, 14 Dec 2009 11:00:58 +1000
Subject: [PATCH] xfree86: if only one screen was found, disable Xinerama (#24627)

Xorg +xinerama crashes immediately due to whacky dependency between Xinerama
and RandR12. The latter doesn't initialize if Xinerama is enabled, but if
only one screen is found, Xinerama is disabled again and RandR12 tries to
access data it never initialized.

Dependency chain is:
- ProcessCommandLine sets noPanoramiXExtension to FALSE
- xf86RandR12Init() is a noop
- PanoramiXExtensionInit sets noPanoramiXExtension to TRUE
- xf86RandR12CreateScreenResources tries to use the devPrivates key it never
  initialized.

This hack checks if there's only one screen at the time RandR12 is
initialized. If so, we expect Xinerama to fail anyhow so we disable it
ourselves and proceed as planned.

X.Org Bug 24627 <http://bugs.freedesktop.org/show_bug.cgi?id=24627>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit faca1bc582e374d32ee9d63d10e072fbef4940a3)

xfree86: remove HistorySize from the xorg.conf man page.

This option isn't parsed by anything anymore.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 190610e0c62170a27ab3e40c6c6210a583ae1ad4)

xfree86: reword InputDevice man sections, deprecate CorePointer/CoreKeyboard

Reshuffle and reword - InputDevice sections are only necessary if
hotplugging is disabled. Put more emphasis on hotplugging and less on HAL
since we'll switch backends eventually.

CorePointer, CoreKeyboard, and AlwaysCore should be listed as deprecated
since they don't do what they used to since 1.4. These days, only
SendCoreEvents matters and it's enabled for any driver calling
xf86ProcessCommonOptions (== every driver).
It only controls the startup behavior too, so document this.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Alan Coopersmith <alan.coopersmith@sun.com>
(cherry picked from commit 094c6b9f97a9f92e5a0ef3cf5be24c09ed4d6063)

xfree86: update man page for special keys handling.

SpecialKeyHandling was removed from the kbd driver with version 1.4.0. Since
this is the only version that will build against server 1.7+ it's not
reasonable to mention it in the man page. Reword, point to XKB instead and
make clear that some key combinations _may_ not be available in any given
config.

Reported-by: Derek Fawcus
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Alan Coopersmith <alan.coopersmith@sun.com>
(cherry picked from commit 801bc8075aee664bd4c6b6ff842ab737c143a1de)

xfree86: DontZap has been disabled for a while now, say so in the man page.

1.7 always shipped with DontZap disabled, it's just the default keymaps that
may not include the symbol to trigger it.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Alan Coopersmith <alan.coopersmith@sun.com>
(cherry picked from commit 753310837cd3812882d8de67f063bb61813db675)

test/xi2: fix maximum max_keycode (bug#25492)

The number of keycodes needs to be lower than 0xFFFD so that the length
field of xXIKeyInfo doesn't overflow.

Signed-off-by: Julien Cristau <jcristau@debian.org>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit b44c9be244cee286835855483a69c69e80b095c0)

Xfake: Nuke -Wl,-undefined=InitExtensions from LDFLAGS

This is not portable, and accodring to Paulo should not be required any more.

Signed-off-by: Jeremy Huddleston <jeremyhu@freedesktop.org>
(cherry picked from commit d50319550458f8127298cf8672b47f914b74ab1f)

dtrace: Add Xserver-dtrace.h to CLEANFILES

Signed-off-by: Jeremy Huddleston <jeremyhu@freedesktop.org>
Reviewed-by: Alan Coopersmith <alan.coopersmith@sun.com>
(cherry picked from commit ed7d2527190ec56aed38db19d9e958abe88441b9)

XQuartz: pbproxy: Fix building of standalone xpbproxy executable

Signed-off-by: Jeremy Huddleston <jeremyhu@freedesktop.org>
(cherry picked from commit 97a6454ea57587db490873fee8ff0f899882972d)

Change default xkb model from pc104 to pc105

PC105 is a more useful default for non-American keyboard users,
not harmful for American PC101/PC104 keyboard users.

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 1df4bd6011e110dcf0649b15bfffd4ab9e6961d6)

Add platform compatibility defines for Sun Studio compilers

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 48749cc21b074ee8e68c3854bd3977ec7c408225)

Convert checks for PC98 support from platform #ifdefs to configure flag

Default remains the same - on for most OS'es on i386 (except Solaris),
off for everyone else.   Can be manually toggled via --enable-pc98 or
--disable-pc98.

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit b63912ed4c69fedd1bea92274d6cae0429a79677)

fb: backport fb changes from master for src window operations.

This rolls up the following commits and also keeps the image_from_pict
API compatiblity. It introduces a new image_from_pict_18 API that is
used by the server leaving the old API alone.

I've rolled this up as I don't want to introduce ABI breaks in bisection.

a72c65e9176c51de95db2fdbf4c5d946a4911695 fb: Adjust transform or composite coordinates for pixman operations
bd567061c8b84b268d9bbb01bc4d8981feefb862 Split fbGetDrawable into fbGetDrawablePixmap and fbGetPixmapBitsData
61335052972a78d67c0ba74f902273b34c63a198 Revert "Fix clipping when windows are used as sources"
071b3c1810d9f2602173acc8578caac20e0b771e Revert "Use IncludeInferiors when copying windows before compositing."
8e640d6b131d2865a9725d8997023865b0ef3d69 Revert "Reserve space for two GC values in copy_drawable()."

Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add freetype & type1 to the LoadModules ignore list

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
Acked-by: Adam Jackson <ajax@redhat.com>

Update Sun license notices to current X.Org standard form

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
Acked-by: Peter Hutterer <peter.hutterer@who-t.net>

COPYING: Update license preamble

Update the comment about "a modification is currently under discussion"
since there hasn't been any such discussion for some time and this is
the consensus agreed upon preferred format.

Also throw in a pointer to the OSI discussion page about the MIT license.

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

Fix builds with --with-int10=stub

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
Acked-by: Adam Jackson <ajax@redhat.com>

Merge branch 'server-1.7-branch' into server-1.7-nominations

Xi: reset device properties to NULL after deleting them. (#25374)

Commit 0e6cee853d8e5bef3274e632ef034d37f14674a9 introduced cleanup code to
remove the accel properties when switching schemes. The same code is
triggered by the default closedown code but only after unconditionally
removing all device properties (as part of the cleanup). The properties,
although deleted never got reset to NULL.

X.Org Bug 25374 <http://bugs.freedesktop.org/show_bug.cgi?id=25374>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Tested-by: Andy Furniss <lists@andyfurniss.entadsl.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
(cherry picked from commit ea14a418214ba4c58910f5650d8b25b3fc3a7099)

Fix possible NULL dereference in XkbFlushLedEvents()

Through some code paths it is possible that NULL is being passed in the
'ed' parameter to XkbFlushLedEvents(). Make sure we don't pass it along
to bzero().

Signed-off-by: Tomas Carnecky <tom@dbservice.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 886140787852c660978a9bb3abc7cc2702cbaed1)

xserver 1.7.3.901

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

Merge branch 'server-1.7-branch' into server-1.7-nominations

xserver 1.7.3

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

configure.ac: error while checking for XDMXCONFIG_DEP

Introduced in commit 9998105a387e0294054502331a56e1e020cd93e4
The replacement third parameters to PKG_CHECK_MODULES([DMXMODULES]
was not quoted.

Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
Reviewed-by: Dan Nicholson <dbn.lists@gmail.com>
Tested-by: Julien Cristau <jcristau@debian.org>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 91c1bd78f7240c92702828f8e5a6b6ce944b9e36)
(cherry picked from commit b8623569a3acff31cb18a9f6046e60d0637cc0e0)

Restore ABI by reintroducing the "is_src" paramter to image_from_pict.

Restores old ABI after 196aff9b18381d700fb28027186cce6e68ad587c.

Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Tested-by: Julien Cristau <jcristau@debian.org>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 155e61a9f0429bf28ce493c0fe7a2d076cb7e137)

[PATCH] Xv: Fix AdjustFrame when driver implements ReputImage.

Finally fixes fd.o #4653, filed more than 4 years ago.

Patch can be happily applied to all modular Xorg versions.

Signed-off-by: Luc Verhaegen <libv@skynet.be>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 0e084d8c71e697a5cf5d5d7c749455ae14bd6eb7)

dix: fix memory leak, free event list on shutdown. (#25028)

X.Org Bug 25028 <http://bugs.freedesktop.org/show_bug.cgi?id=25028>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit c20c8897272427cb3f755a3e28e80a9ad46f08a1)

dix: remove core devices when shutting down. (#25028)

NewInputDeviceRequest (and RemoveDevice) have checks in place to not allow
removal of the VCP/VCK. When shutting down, they need to be cleaned up
nonetheless to free the memory associated.

X.Org Bug 25028 <http://bugs.freedesktop.org/show_bug.cgi?id=25028>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 66bb8c6fbdfc0fc0d971aac4ec6f949bb9288c1b)

Set the source and deviceid for key repeat events (#24785)

X.Org Bug 24785 <http://bugs.freedesktop.org/show_bug.cgi?id=24785>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Adam Jackson <ajax@redhat.com>
(cherry picked from commit b584c224a888c9e7f92d7e49021f74232a727c7f)

xfree86: tell users to disable AutoAddDevices, not AllowEmptyInput.

Technically, disabling AEI is the right suggestion. AEI off forces the
server to init the built-in defaults for input devices (or pick the first
one from the config file). At the same time, hotplugging is still available
with AEI off.

Unfortunatly, in the vast majority of cases users want to simply disable
hotplugging or have a working server while the local HAL configuration is
broken or missing. Disabling AEI will lead to duplicate events, triple
keystrokes, etc. once the configuration works again.
It's not actually required to remove AEI once hotplugging works again,
though it will in many cases lead to a setup that appears broken.

Asking users to disable AutoAddDevices instead means those users disable
hotplugging, can then fix the HAL setup and they _must_ remove the config
line again to test if hotplugging works again. Which doesn't leave them with
a broken config once everything is working nice and dandy. Less bugreports,
everybody wins.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Dan Nicholson <dbn.lists@gmail.com>
Acked-by: Daniel Stone <daniel@fooishbar.org>
Acked-by: Rémi Cardona <remi@gentoo.org>
Acked-by: James Cloos <cloos@jhcloos.com>
(cherry picked from commit 761ae22f880bd79550ccf93d321b8a28b3755956)

configure.ac: error while checking for XDMXCONFIG_DEP

Introduced in commit 9998105a387e0294054502331a56e1e020cd93e4
The replacement third parameters to PKG_CHECK_MODULES([DMXMODULES]
was not quoted.

Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
Reviewed-by: Dan Nicholson <dbn.lists@gmail.com>
Tested-by: Julien Cristau <jcristau@debian.org>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 91c1bd78f7240c92702828f8e5a6b6ce944b9e36)

EXA: Don't defragment offscreen memory at allocation time.

Fixes http://bugs.freedesktop.org/show_bug.cgi?id=24300 .

Offscreen memory allocation can occur from various places, and apparently
doing defragmentation from at least some of them can confuse some driver
acceleration code.

There's still the regular background defragmentation in the WakeupHandler,
which should manage to keep fragmentation at a reasonable level.

Signed-off-by: Michel Dänzer <daenzer@vmware.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 8754db77d8169e5ea506a963cebee1a651bcf094)

EXA: fix mixed-pixmaps issue where we could do accel with pixmap mapped

The problem occurs in the following situation:

1. Create Pixmap
2. Software fallback on pixmap: pExaPixmap->pDamage created
3. Accel to pixmap: driver pixmap created, exaPixmapIsOffscreen() now true
4. Software fallback on pixmap
4.1: PrepareAccess() from driver succeeds, so no DFS to system memory copy
4.2: Software rendering...
4.3: (pExaPixmap->pDamage && exaPixmapIsOffscreen()) is true, so we try
     to (incorrectly) UTS the data back to the driver pixmap, when we
     should just call the driver's FinishAccess() hook.
4.4: fail

The patch adds an additional check on pExaPixmap->offscreen, which should
(for the cases exaPixmapIsOffscreen() will return true) only be FALSE
when we haven't used PrepareAccess() on the pixmap.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Acked-By: Michel Dänzer <michel@daenzer.net>
[ Michel Dänzer: This is a 'backport' of commit
  647b79f87a9891225678dc6fc2fbda3bdef8fa9d from master - actually this patch
  was submitted first but didn't apply to master anymore, so technically the
  other change is a 'forward port' of this one ]

Restore ABI by reintroducing the "is_src" paramter to image_from_pict.

Restores old ABI after 196aff9b18381d700fb28027186cce6e68ad587c.

Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Tested-by: Julien Cristau <jcristau@debian.org>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

xserver 1.7.2

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

Bug #25136: Revert "Fix clipping when windows are used as sources"

That change causes lib(w)fb to make accelerated driver calls after the driver
has entered a software fallback.  Most drivers don't expect this, which leads to
corruption or crashes.  A change to make this code do the copy in software is
unacceptably slow.

This reverts commit e9aa61e9f0d663d5b34a397b943b4d1df44e873d.

Conflicts:

configure.ac
fb/fbpict.c

Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Xi: when deleting all properties, reset property handler to NULL.

Trying to unregister property handlers during the device closure process
leads to invalid memory accesses.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 1b127ab8429616adf9ec31ba4d8bdd9af6e104a9)

Move xdmxconfig modules into DMX conditionals (#25102)

xdmxconfig requires additional modules not checked for if Xdmx build is set
to auto (the default). This may lead to build errors if the Xdmx modules are
installed, but not the extra ones required for xdmxconfig.

X.Org Bug 25102 <http://bugs.freedesktop.org/show_bug.cgi?id=25102>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Remi Cardona <remi@gentoo.org>
Acked-by: Julien Cristau <jcristau@debian.org>
(cherry picked from commit 8806375ed72a3cd465fe0a49ead079a334accd6b)

Xext: Fix a memory leak on GE client disconnect.

Add a call to dixRequestPrivate to inform dixFreePrivates that memory
allocated in GEClientCallback should be released when client
disconnects. Otherwise there is a leak of sizeof(GEClientInfoRec) for
every client connect/disconnect.

Also remove the explicit allocation and let GEGetClient /
dixLookupPrivate do it. This makes GEClientCallback similar to the
other extension callbacks.

Signed-off-by: Rami Ylimaki <ext-rami.ylimaki@nokia.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit eb967ca36cfe0409972ac987a74d498cd6f6bafb)

Xi: don't crash when deleting invalid device properties.

Deleting a property that was not set on a device leads to a null-pointer
reference. The protocol allows deleting those properties - it has to be a
noop.

Reproducible:
xinput --set-prop "My device" --type=int --format=8 "my property" 1
xinput --delete-prop "My other device" "my property"

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit a30e739a144912a68adcaa9f426d600c6ecbd529)

xfree86: Edid quirk for Philips LCD LP154W01

v1->v2: Make one condition case for one quirk instead of merging them
together. This is based on the Keithp's suggestion.

Move the EDID quirk for Philips LCD LP154W01 as the panel reports the vertical
size in cm.

https://bugs.freedesktop.org/show_bug.cgi?id=24482

Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 19f7c15e2008dab3c46ba3e14dfa353d01c74f72)

XQuartz: Allow better compatability with older versions of xinit

If we are id="org.x" and the launchd socket is ":0", we will claim
the socket to match the old behavior before we prefixed the
socket name with our id.

Signed-off-by: Jeremy Huddleston <jeremyhu@freedesktop.org>
(cherry picked from commit 4677b5a80025b50ba2a3e953fd487a549586ae9f)