[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

xorg-server: Changes to 'refs/tags/xorg-server-2_1.6.99.903-1'



Tag 'xorg-server-2_1.6.99.903-1' created by Julien Cristau <jcristau@debian.org> at 2009-09-29 15:51 +0000

Tagging upload of xorg-server 2:1.6.99.903-1 to experimental.

Changes since xorg-server-2_1.6.4-1:
Aaron Plattner (7):
      Don't enable XAA offscreen pixmaps with Option "XaaOffscreenPixmaps" "no".
      Query the DDX in ProcRRGetScreenInfo.
      Export CompositeRegisterAlternateVisuals.
      Bug #22804: Reject out of bounds XGetImage requests
      Fix dist.
      Damage: Add wrappable hooks for damage create, destroy, register, & unregister.
      Damage: Add devPrivates to DamageRec

Aaron Zang (1):
      Add new VT support for OpenSolaris & future Solaris releases

Adam Jackson (134):
      randr: Don't send output property events on server exit
      Default to x86emu even on i386 linux
      randr: Add [GS]etOutputPrimary
      randr: Mangle compat Xinerama reply based on primary output
      randr: Mangle GetScreenResources sort order based on primary output
      randr: use primary output for RRFirstOutput()
      randr: clear primaryOutput when the output is deleted
      xsync: remove cast abuse.
      xsync: Use a local header for server API definitions
      xsync: ANSI cleanups
      xsync: Fix wakeup storm in idletime counter.
      xsync: build fix
      Fix syncsrv.h guard define
      config: Add vboxvideo and tga to the magic driver list
      dmx: Fix calloc macro confusion.
      xv: remove useless XVCALL macro
      XAA: Disable offscreen pixmaps by default.
      Move CreateUnclippedWinSize to mbuf
      Remove CreateUnclippedWinSize from window.h
      Don't include fontmod.h
      Remove xorgcfg 'Options' list.
      Remove a bunch of useless casts.
      Xephyr: Add some rudimentary Xnest command line compat
      XKB: Remove -kb and +kb from -help text and man page.
      loader: Simplify handle allocation and refcounting.
      loader: code motion.
      loader: Make a comment slightly less stupid
      loader: Simplify loader magic
      loader: Remove useless call to LoaderGetOS
      loader: Remove a vestige of archive support
      loader: Remove icache flushes on alpha.
      Bus: make the resource list static.
      Bus: Remove unused xf86ExtractTypeFromList
      Bus: Delete some obfuscatory macros
      Bus: Remove xf86SetAccessFuncs() and related machinery
      Revert "Bus: Remove xf86SetAccessFuncs() and related machinery"
      RANDR: Validate entire mode list for interlace and doublescan
      EDID: Hack for 1366x768 in standard timing descriptors
      Input: Don't re-block SIGIO in xf86SigioReadInput()
      Linux: Remove dead USE_DEV_FB #ifdefs
      PCI: Remove pciBusAddrToHostAddr and associated nonsense
      kdrive: Nuke dead AGP and VGA code. (#19921)
      RANDR: Fail softly on GetPanning if the screen can't do it.
      selinux: Don't bother relabeling resources that are being destroyed
      DDC: Redo extended device probe slightly.
      randr: Fix thinko in xf86TargetPreferred
      Unexport xf86SetPriority
      selinux: Add support for avc_acquire_netlink_fd()
      misprite: In SourceValidate, check that it's a Window first
      misprite: Fix cast abuse
      misprite: Fix a typo
      misprite: RECT_IN_REGION -> miRectIn
      misprite: Do window check first for Get{Image,Spans} too
      misprite: Fix software cursor on multiple ScreenRecs
      misprite: Remove private header, fold into misprite.c
      config: Fall back to vesa for Intel Poulsbo.
      selinux: Only activate if policy says to be an object manager
      APM: Use general handlers, not input handlers.
      Document which bits of ClientRec are currently unused
      dix: Simplify InitClient()
      APM: Don't carp if not available.
      input: Remove xf86ReloadInputDevs hack
      kdrive: s/KdSaveString/strdup/g
      DPMS: Code motion.
      DPMS: Fix cast abuse
      DPMS: Unexport all the various setup variables.
      DPMS: Simplify some macro silliness.
      DPMS: Align the default timeouts with the default screensaver timeout.
      DPMS: Remove unused DPMSGet()
      DPMS: Simplify command line parsing
      DPMS: Remove the defaultDPMS* variables
      os: Remove the useless -x option
      DPMS: Re-export the various DPMS variables.
      Revert accidental Makefile change from previous commit
      Remove some OS/2 leftovers.
      os: signal handlers return void.
      config: s/xf86configStrdup/strdup/
      config: Remove useless xf86conf{{m,c,re}alloc},free} macros
      xdmcp: Don't crash on X -query with more than 255 IP addresses. (#20675)
      randr: Accept gamma set requests from XF86VidMode clients too
      Update several of my and/or Red Hat's licenses to standard form.
      COPYING: SGI FreeB 1.1 -> SGI FreeB 2.0
      EDID: Be more cautious about finding vendor blocks.
      EDID: Add modes from Established Timings III descriptor to mode pool
      xfree86: Remove unused (and useless) xf86{Add,Delete}ModuleInfo
      xfree86: Remove loader symbol list ABI stubs
      vfb: Fix depth setup.
      doc: Remove mention of Kerberos replay cache from Xserver.1
      Remove some libXfont leftovers
      Typo fix
      s/MIN/min/, s/MAX/max/ (#2968)
      cvt: Allow multiple-of-60Hz refresh rates for reduced blanking.
      xephyr: Add -title option.
      kdrive: Remove a lie about PCMCIA support.
      kdrive: undef PSEUDO8
      vfb: Re-enable 30bpp support
      randr: fix typo in swapped dispatch
      EDID: Carp about 1.4 monitors with no preferred refresh rate
      pci: Dump vendor/devices ids in the printed device list
      EDID: Fix timing class names to match the spec
      config: drop i810.
      Since font modules are dead, don't mention them in xorg-server.pc
      dbe: Adapt to new headers
      ddc: Update a comment.
      ddc: Code motion to eliminate forward decls
      ddc: Yet more code motion
      ddc: Give DisplayID a place to hang its hat
      ddc: mv xf86DDC.c ddc.c
      ddc: Skeleton for xf86DoDisplayID()
      ddc: Don't try to publish a root window property for DisplayID
      ddc: Refactor root window property code
      ddc: s/xf86DDCMonitorSet/xf86EdidMonitorSet/
      displayid: Implement mode decoding.
      dbe: Fix indentation
      xfree86: Remove TargetRefresh option
      xfree86: Remove some #if 0
      xfree86: Fix some misleading comments
      randr: Un-duplicate the reduced blanking check.
      fbdevhw: simplify some #if 1
      fbdevhw: Remove a #if 0
      fbdevhw: Remove pointless OS check, this never gets built on non-Linux
      loader: Remove useless TestFree() macro
      loader: remove dead LoaderCheckUnresolved
      loader: Remove a useless check.
      xfree86: Remove xf86SetPriority
      randr: Add Option "Primary" to Monitor sections
      s/xf86EnableOutputs/xf86CollectEnabledOutputs/ for clarity
      linux: hand-roll a backtrace printer instead of using backtrace_symbols
      linux: Yet more malloc() avoidance for backtrace()
      xfree86: dump /proc/cmdline in the log on Linux
      composite: Move screen init before extension registration.
      EDID: Print 1152x864 in established timings, not x870
      EDID: Vendor detailed blocks aren't worth X_WARNING about
      randr: Fix crtcs using set_mode_major()

Adam Tkac (1):
      xfree86: don't try to UnInit virtual devices in DIDR. (#20087)

Alan Coopersmith (59):
      Remove hack to only load font modules if magic strings appear in font path
      Update See Also lists in Xorg & xorg.conf man pages
      Correct warning for unknown GlxVisuals option in conf file
      More man page updates for 1.6 release for Xorg, xorg.conf & exa man pages.
      Remove unused HandleSpecialKeys config option
      Remove unused config flags from FlagValues & FlagOptions
      Don't need to check uid/euid for every commandline argument
      Add atKeynames.h to libdmxinput_a_SOURCES so it's included in tarballs
      Update sample xorg.conf file
      Make sure _X_EXPORT is defined in edid.h
      Fix sdksyms.sh to work with Solaris/Sun compiler builds
      Fix linking of Xorg with dtrace probes on Solaris
      Remove powerpc load/store functions from export list on other platforms
      Remove unused CLOG_DATE (leftover from pre-git ChangeLog)
      Remove obsolete XF86_DATE & xf86Date.h
      Fix XKB default configure options to use the right variables
      Update xorg-server.h.in & Xnest to use new XKB default #defines
      x11-input.fdi: Add options needed to handle adding USB devices on Solaris
      Constify atom name strings
      Correct error message if specified config file is not found
      Check for and report errors writing xorg.conf.new from Xorg -configure
      Make RgbPath keyword in xorg.conf a non-fatal error
      Obsolete InputDevices keyword in xorg.conf Files section
      Add Extensions section to xorg.conf man page
      Remove #ifdef macII code left over from ancient A/UX 3.0 support
      Don't leak canonical module name and patterns if module is built-in
      Don't leak default font path when appending built-ins
      Lift fatal signal handlers from DDX'es up to a common DIX implementation
      Use RTLD_DI_SETSIGNAL to catch runtime dynamic loader errors and clean up
      Fix byte swapping of XF86VidMode{Get,Set}GammaRamp
      Add casts to generated entries in sdksyms.c to silence type mismatch warnings
      Add SIGIO/SIGPOLL support for Solaris
      Solaris: use <sys/agpgart.h> instead of stale copy in Xorg sources
      XkbSetNamedIndicator should ignore SD's without LED's
      Fix a couple off-by-one array boundary checks.
      Don't printf NULL pointers on HAL connection error
      Resync COPYING file with notices in code base as of xorg-server-1.6.1
      COPYING file updates for git master changes since 1.6 branch
      Correct some Sun license notices to Sun's standard X11 license format
      Clarify use of and need for mffs vs. ffs
      Xephyr & Xserver man page fixes
      Fix int10 module build after typo in commit 2638e9899
      Remove hardcoded gcc -Wall option from configure.ac
      Fix blddir != srcdir builds of solaris-*.il files
      Ansify function arguments in VTsw_noop.c
      Correct modifier map built when ProcSetModifierMapping is called
      Remove support for Solaris x86 releases older than Solaris 8
      Sun bug 6872917: Xorg not querying /dev/fb when no xorg.conf exists
      compiler.h: include <string.h> before using memmove()
      xfree86: Change default font path to match configure.ac's --default-font-path.
      Update xorg.conf man page & sample for changes in Xorg 1.7
      Fix module path in xorg.conf man page
      Remove stray semi-colon in protocol-common.c
      test_convert_XIRawEvent: Don't write more data than the struct holds
      Fix num_masks/length overflow test for XiSelectEvents
      Change xf86dristr.h includes to use xf86driproto.h instead
      Add configuration option for use of SIGIO handlers for input events
      Check for -wrap support in the linker
      Fix typo in Xephyr man page

Alan Hourihane (3):
      bump master to 1.6.99.1 (now the 1.6 branch is created)
      glx: fix retval checks when failures occur for drawable creation.
      dri2: support glXWaitGL & glXWaitX by copying fake front to front and

Ander Conselvan de Oliveira (2):
      Honour Option "DPMS" "off" on xorg.conf
      xfree86: Remove device from inputInfo.devices if ActivateDevice failed.

Ben Byer (1):
      Add (ok, fix) support for DTrace under OS X

Ben Gamari (1):
      Xi: Remove redundant and incorrect butmap range check

Ben Skeggs (2):
      exa: driver pixmaps enabled if either CreatePixmap or CreatePixmap2 present
      quirk: use first detailed timing as preferred for PEA prod 9003 (rh#492359)

Benjamin Close (7):
      Don't use gnu specific extensions to awk when builing symbols
      XNest: Fix the build caused by __XKBDEFRULES__ -> XKB_DFLT_RULES in 23862ede59a9ce11a06ec5151bde460fb836c603
      dix: Remove includes which are not actually used directly
      xext: Use proto header rather than the Xext include file, this prevents userspace being pulled in causing issues
      dix: Change AllocMaster into AllocDevicePair, allow creation of SDs too.
      input: propagate XTst events through virtual slave devices.
      xfree86: correctly define barriers for FreeBSD amd64

Benjamin Defnet (2):
      hw/xf86/modes: Set crtc mode/rotation/transform before calling set_mode_major
      randr: fix operation order so that rotation+transform works

Bernhard Rosenkraenzer (1):
      Xext: fix MultiBuffer compilation error with TryClientEvents. (#18835)

Carl Worth (1):
      configure: Fix broken quoting for initialization of REQUIRED_MODULES

Chris Ball (2):
      Build fix for xf86EdidModes.c
      Build fix, remove export of pciBusAddrToHostAddr

Christian Beier (1):
      Xi: fix typo in WarpDevicePointer handling.

Colin Guthrie (1):
      Fix compilation with -Werror=format-security

Colin Harrison (34):
      xsync: Prototype fix.
      Xming, Cygwin/X: Fix crash in function winMultiWindowGetTransientFor() (#11147)
      Xming, Cygwin/X: Tidy up warnings, headers and unused variables (#11132)
      Xming,Cygwin/X: Fix crashes when using the DirectDraw '-refresh rate-in-Hz' option in -fullscreen (#11128)
      Xming: Fix a couple of warnings
      Xming: Reduce MOUSE_POLLING_INTERVAL
      Xming: Correct the way display and screen number is reported in window titles and tooltips
      Xming: Correctly parent XA_WM_TRANSIENT_FOR windows in -multiwindow mode when a windows window is created.
      Xming: Only allow WM_MOUSEWHEEL messages to act on the client area of a focused window.
      Xming: Notify X when the keyboard focus is lost to a pure Windows window in -multiwindow mode.
      Xming: Add styles keyword and attributes.
      Xming: Ensure we dont try to print a null value when displaying xkb configuration
      Xming: Update .rc file to tidy About and Exit dialogs
      Xming: Add Colin Harrison to copyright and authors list for clipboard integration files
      Xming: Select events correctly in clipboard integration code
      Xming: Improve clipboard integration startup in XDMCP mode
      Xming: Fix prototype for winProcessXEventsTimeout() in clipboard integration code
      Xming: Fix a memory leak in clipboard integration code
      Xming: Check clipboard selection ownership after taking it
      Xming: Update keyboard layouts mapping table
      xkb: set bell_func in InitKeyboardDeviceStruct.
      Xming: Slightly adjust the formatting of the logged command line
      Xming: Prevent the mouse wheel from stalling when another window is minimized.
      Xming: Cache atom lookups in clipboard integration code
      Xming: Avoid log spam if a windows application won't provide CF_UNICODETEXT clipboard format
      Xming: Various tidy ups in winClipboardFlushXEvents()
      Xming: Fix some log spam
      Xming: Simplify logic for ownership release of native clipboard
      Xming: Apply window style hints in -multiwindow mode
      Cygwin/X: Update Get/SetWindowLong() to Get/SetWindowLongPtr() everywhere
      Cygwin/X: use GWLP_WNDPROC, GWLP_USERDATA with Get/SetWindowLongPtr
      Xming: Use GetClassLongPtr() instead of superseded GetClassLong()
      Xming: Use RegisterClassEx() instead of superseded RegisterClass()
      Xming: Add NET_WM_ICON to native icon conversion

Cooper Yuan (1):
      Fix typo in xf86PickCrtcs()

Dan Nicholson (6):
      dmx: Require newer libXi for XQueryInputVersion
      Don't reuse PKG_CHECK_MODULES identifiers
      Kill off --with-mesa-source completely
      Ignore generated dmx manpages
      xkb: Use cached XKB keymap when rules haven't changed
      xfree86: Link libselinux with Xorg system libraries

Daniel Stone (54):
      XKB: Also copy keyboard feedback when copying the keymap
      XFree86: Fix build with DGA disabled
      XFree86: fbdevhw: Add helper function to get fd
      XFree86: Xv: Add ClipNotify helper
      RandR: Only export Xinerama symbols when building Xinerama support
      XKB: Allow build-time configuration of XKB defaults
      XKB: Move XkbCopyKeymap definition to xkbsrv.h
      XkbCopyKeymap: inputInfo.keyboard is not a special case
      XKB: Only Xi events are processed
      XKB: Remove unnecessary prototype
      XFree86: Input: Fail NIDR if we can't enable devices
      XKB: Trying to copy to the same keymap is not fatal
      XKB: Be more verbose about XkbCopyKeymap failure in debug mode
      XKB: Add XkbRMLVOSet
      XKB: Remove lock actions disabling
      Xephyr: Disable state inheritance hack
      XKB: Make XKB mandatory
      Input: Overhaul keyboard initialisation process
      Input: Remove state from KeyClassRec
      Input: Ignore modifiers in core input processing
      Input: Remove modifierKeyMap
      Input: Remove modifierMap from core
      XKB: Remove unsupported Xi operation flags
      XKB: Remove 'extra' functionality from rules parsing
      XKB: Remove support for setting combined keymaps
      XKB: Remove descriptions from maprules
      XKB: Explicitly decode action data
      XKB: Add a hell of a lot more FIXMEs
      XKB: Sanitise vmods in actions
      XKB: Sanitise pointer actions
      XKB: Sanitise ctrls action
      XKB: Sanitise vmods for redirected keys
      XKB: Simplify keymap writing a bit
      Xi: Introduce XIShouldNotify
      Input: Clean up keymap change notifications
      Input: Centralise pointer map changing
      Input: Remove core keysyms from KeyClassRec
      XKB: Remove unused DDX functions
      Input: Remove unused CoreProcess{Keyboard,Pointer}Event
      XKB: Remove unused XkbProcessOtherEvent
      XKB: Fix logic error
      Input: Don't allow DDX to generate repeat events
      Input: Add postdown to ButtonClassRec
      Input: Use previous state in valuator events
      Input: Clarify valuator FatalError messages
      KDrive: Warning fixes and cleanups
      OS: Fix compile warnings
      KDrive: Xephyr: DRI: Warning fixes
      KDrive: Warning fixes
      Xi: Fix harmless ButtonPress/ButtonRelease confusion
      Input: Mark Xi input events as critical
      dix: improve code flow in TryClientEvents, better debugging messages.
      input: allow for detectable autorepeat.
      fbdevhw: Test for graphics:fb%d as well as graphics/fb%d

Dave Airlie (24):
      randr12: looking up these bits if randr isn't initialised is bad.
      exa: avoid offscreen pixmap swapped out flag for driver pixmaps
      exa: add CreatePixmap2 hook for driver pixmaps.
      exa: missed exa.h change
      exa: add missing exa.h header include
      crtc/gamma: check xf86_config is valid before using it.
      input: move inputstr.h to where its needed.
      glx: fix open-coded linked list removal function
      xfree86: move didLock assignment down to where the function pointer is valid.
      fbdev: make entity fail if PCI claimed already.
      GLX: note the implicit flushes with ReadPixels in indirect contexts.
      GLX: make function static.
      dix/resource: fix use after free in resource code with DRI
      pci: add support for pci is boot vga call.
      xserver: remove RAC/resource handling code.
      sbus: fixup for rac removal
      ddx: fix xf86Config.a generation
      parser: make libxf86config_internal.la not installed.
      exa: fix CreatePixmap2 to be useful for tiling.
      exa: clarify createpixmap2 new pitch return
      xf86 ddx: add vga arbiter support.
      vgaarb: if arb init fails, make sure locking doesn't occur.
      vgaarb: protect fini as well just in case
      xserver: SIGIO option handling was incorrect.

David Jander (2):
      kdrive: Fix segfault in tslib support
      [kdrive] Fix rotation of pointer

David Marx (1):
      Solaris: Make sure non-inline versions of asm routines end with ret statements

David Miller (1):
      mi: ignore DGA events in ChangeDeviceID

David Nusinow (1):
      xorg.conf (5) refer to mousedrv (4). Debian #394058

Eamon Walsh (37):
      Add xace headers to the SDK when enabled and export the XaceHooks symbol
      xace: Export wrappers around two Xtrans functions used by modules.
      xselinux: Use xace Xtrans wrappers instead of the now-inaccessible wrapees.
      Fix 2 const warnings.
      Correct access mode in call to dixLookupWindow() within RRSelectInput.
      xselinux: Don't require incoming context strings to be null-terminated.
      config: fix crash caused by strdup(NULL)
      xselinux: Don't BadAlloc in List* requests if there are no items to list.
      security: Revert behavior of extension access for compatibility.
      security: Fix a crash caused by wrong ordering of format arguments.
      security: Grant untrusted windows remove access on all windows.
      kdrive: fix Xvfb build with separate libmain.
      Fix most remaining deprecated resource lookups.
      xselinux: Relax ownership restriction on SetSelectionUseContext.
      xace: Fix a bad device access hook call.
      dix/events.c: Use wClient where appropriate.
      glx: Register names for the GLX resource types.
      xace: fix a bad send access hook call.
      xselinux: Move the security class mapping to the header file.
      xselinux: Add new device permissions for XI2.
      Xi: fix 2 memory leaks.
      Xi: fix up access modes for calls to dixLookupDevice().
      Xi: check for Use permission on the device in SetClientPointer().
      Xi: check for GetAttr permission when listing or querying devices.
      dix: add a new DixAccess bit, "DixPostAccess".
      xselinux: ignore property hook calls with the new Post access mode bit set.
      xace: add a new hook for checking property content after it has been set.
      dix/property.c: use memcpy where appropriate.
      xselinux: Add more new device permissions for XI2.
      xselinux: Allow per-client device create contexts.
      Add XI2 requests to protocol.txt
      xace: fix access mode in dixLookupWindow call within ProcUngrabKey.
      xace: fix access mode in dixLookupWindow within several RandR calls.
      Add DRI2 requests to protocol.txt
      xace: fix up access modes in dixLookupDrawable calls from dri2.
      Correct outdated e-mail address in "Author" statements.
      Fix "warning: left-hand operand of comma expression has no effect"

Eric Anholt (14):
      Fix GLX after 180bad84774493d48f2793a6281d825560944863.
      Warning fix (GL likes to call strings GLubyte * instead of char *).
      Warning fix: Remove dead glXDisp{,Swap}_DrawArraysEXT definitions.
      Move the apple fat binary hacks back to a header file, and make it apple-only.
      randr: Avoid re-querying the configuration on everything but GetScreenResources.
      modes: Protect xf86_crtc_supports_gamma() from non-RandR 1.2 drivers.
      glx: Don't match fbconfigs to visuals with mismatched channel masks.
      glx: Replace broken GLX visual setup with a fixed "all" mode.
      xinerama: Put the proto version in the code instead using proto headers.
      Move contributed m4 to a subdir so we can more easily update contributions.
      Move VENDOR_* defines from AC_SUBST to a header to avoid angering shave.
      Add shave so that we can see the steaming piles of warnings generated.
      Fix unused var warning from pci cleanups.
      dri2: Enable GLX_SGI_make_current_read when the DRI driver supports it.

Eric Paris (1):
      This patch changes all places in the X code to use _raw functions.  The

Federico Mena Quintero (2):
      randr: bug #21554 - re-probe outputs when coming back from laptop unsuspend
      dix/randr: Add missing fields to SRR*NotifyEvent()

Felix Kuehling (1):
      Export symbols needed by the RandR implementation in fglrx

Francis Giraldeau (1):
      Don't log audit messages when -audit 0 specified

Fredrik Höglund (2):
      Render: Add support for the PDF blend mode operators.
      Require renderproto >= 0.11 and pixman-1 >= 0.15.14 for the new blend modes.

Gaetan Nadon (1):
      dix: xserver "make dist" fails due to eventconvert.h (#27825)

George Staplin (12):
      XQuartz: GL: Make various changes to makeFormat, so that it works better.  Now glxgears looks normal with the old libGL.
      XQuartz: GL: Make many more fbconfigs and visuals available for the 1.5 branch.
      XQuartz: GL: Set the __GLXconfig renderType to GLX_RGBA_BIT.
      XQuartz: GL: Make indirect.c build and work in the 1.6 branch.
      XQuartz: xpr: Cleanup some of the code and possibly fix part of the GLX Pixmap problem.
      XQuartz: GL: indirect.c changes to fix the build with newer OpenGL frameworks.
      XQuartz: xpr: The dri.c code for pixmaps was wrong in several ways.  They weren't
      XQuartz: Add support for GLXPixmaps to the AppleDRI.
      XQuartz: Add driWrap.h that I missed in the last commit.
      XQuartz: Fix the new ProcAppleDRIDestroyPixmap code REQUEST_SIZE_MATCH.
      XQuartz: GL: Fix a bug with an uninitialized GLX data structure.
      XQuartz: GL: Change from xalloc to xcalloc to potentially avoid

Havoc Pennington (1):
      Set bg pixmap of composite overlay window to None (#20912)

Ian Romanick (15):
      GLX: Changes resulting from changes to Mesa generator scripts / data
      Enable pbuffers
      glx: Add comments around some extension string weirdness
      glx: Inialize best_score before calculating visual scores
      Allow GLX sources to build against Mesa 7.4 sources
      Use a #define instead of a magic number
      DRI2: Add fake front-buffer to request list for windows
      DRI2: Do not send the real front buffer of a window to the client
      DRI2: Synchronize the contents of the real and fake front-buffers
      DRI2: Don't leave empty entries in private->buffers
      DRI2: Add missing front-buffer flush callback.
      DRI2: Send the version the code actually supports
      DRI2: Add interface for drivers to query DRI2 extension version
      DRI2: Implement protocol for DRI2GetBuffersWithFormat
      DRI2: Force allocation of real-front buffer for non-windows as well

Jason Vas Dias (1):
      Fix build with --enable-debug.

Jay Cotton (1):
      Sun bug 6618220: Xorg server core dump in xf86RandRModeRefresh(NULL)

Jeremy Huddleston (167):
      XQuartz: Fix an uninitialized keyboard_type on Tiger
      XQuartz: Fix Czech keyboard dead-acute
      XQuartz: Fix dead-acute on Greek keyboards
      XQuarz: Setup our PATH and PWD earlier, so our initial client benefits from it as well...
      XQuartz: Avoid some warning messages being spewed to system.log by AppKit
      XQuartz: Added option to enable/disable test extensions
      XQuartz: pbproxy: Simplify linking
      Removed unised option from configure.ac for launchd
      XQuartz: Corrected name/command labels in the customization widget
      XQuartz: darwinPointer reports the actual pixel position now rather than a relative position
      XQuartz: Removed some debug spew
      XQuartz: unsetenv(DISPLAY) if we're not org.x.X11
      XQuartz: unset DISPLAY if we didn't get a launchd socket handoff
      XQuartz: Avoid using login /bin/sh blech.  Just use a bash script to start the app, so it will inherit the right environment
      XQuartz: Make debugging output for invalid depths a bit more detailed
      XQuartz: fixed make dist
      XQuartz: Fix path to executable
      mi: Reuse memory in mieqProcessInputEvents rather than making excessive calls to calloc()
      XQuartz: Tiger fix, don't call Xplugin code in the Appkit thread if Xplugin isn't threadsafe.
      XQuartz: Updated menu item ordering for better HIG compliance
      XQuartz: Name the startup shell script X11 for better compatability
      rootless: Make expose_1 static
      XQuartz: Get rid of white rectangle bug
      XQuartz: Changed X11.sh to allow use of a ~/.x11run as requested by users of alternate shells
      XQuartz: update quoting in case X11.app is moved to a directory with a space.
      XQuartz: pbproxy: Release display notification lock when not needed to avoid deadlock
      XQuartz: Run applications via '/bin/sh -c ...' to support users who expect shell parsing
      XQuartz: Update our "screens" when we toggle rootless rather than when we toggle fullscreen
      XQuartz: Don't use keycode 0 to determine !swallow since our most common key to swallow is actual keycode 0 (a)
      XQuartz: Use depth=24 instead of FatalError if we can't figure out our depth
      Updated .gitignore
      XQuartz: Re-enable rlAccel
      XQuartz: Reposition windows when we enter fullscreen to ensure our root window
      XQuartz: Try harder to get the user's login environment
      XQuartz: pbproxy: We explicitly need libX11 for pbproxy
      XQuartz: Updated man page fullscreen_hotkeys fullscreen_menu
      XQuartz: Workaround OSX VNC server bug for modifier key state
      XQuartz: Better avoid stuck keys on context switches
      XQuartz: Honor system key repeat rate
      XQuartz: Make sure to reset the saved key state when deactivating X11.app
      XQuartz: Don't use NX_SECONDARYFNMASK, NX_NUMERICPADMASK, NX_HELPMASK
      XQuartz: pbproxy: Push dpy init and CFRunLoop hook setup into the pbproxy thread to avoid possible deadlock
      XQuartz: Copyright Update... happy new year
      XQuartz: Use AvailabilityMacros.h instead of availability.h for Tiger
      XQuartz: Fat binary buildfixes
      XQuartz: Update padding in appledristr.  This shouldn't change how the struct ends up in memory, but should make it more obvious for human eyes
      XQuartz: Bundle version 2.4.0
      Apple: Don't use DRI2
      XQuartz: misc 1.6 updates (still --disable-glx)
      XQuartz: Add locking to make mieq thread safe on OSX
      XQuartz: Comment explaining the 0x10 flag for [e data2]
      XQuartz: Only call DarwinUpdateModKeys when needed
      mieq: Avoid possible race condition whereby one thread might call mieqEnqueue before InitAndStartDevices finishes
      XQuartz: Don't need explicit Activate/EnableDevice in InitInput
      XQuartz: Fix builddir != srcdir issues and undef _XSERVER64 where appropriate on fat binary compilation
      XQuartz: mieq: Wait for the server to finish initializing before letting other threads mieqEnqueue
      XQuartz: SnowLeopard: Help system book name changed in 10.6
      XQuartz: GLX: OpenGL.framework on Tiger doesn't have glBlitFramebufferEXT
      XQuartz: GLX: Fix accidental exclusion of glBlitFramebufferEXT on Leopard
      XQuartz: Accept bundle version 2.1.6 to work with our startup method
      XQuartz: Fix copyright info in Info.plist to be prettier when viewed in Finder
      XQuartz: Don't need GlxSetVisualConfig any more
      configure.ac: Support version strings like W.X.Y.Z-XXXXXX
      XQuartz: quartzKeyboard compile fixes for recent XKB Changes
      XQuartz: DefineInitialRootWindow is gone
      XQuartz: Comment out the background pointer interaction that seems to be causing CPU spinning on some configurations
      XQuartz: Re-enable support for capslock
      XQuartz: Don't erase the mode_switch+keysym if it's the same as shift+keysym
      XQuartz: Implement SendPSN for AppleWM
      XQuartz: Fix mouse tracking for quake, et. al. in wine
      XQuartz: Don't report mouse events while X11 is not the foreground application
      XQuartz: Pad xAppleDRINotifyEvent to 32bytes to match sizeof(xEvent)
      XQuartz: Return BadRequest when SendPSN isn't implemented rather than success
      XQuartz: Still send mouse events while X11 is in the background if we have test extensions enabled
      XQuartz: Send a MotionNotify event for the mouse cursor when activating X11.app
      XQuartz: Use updated Xplugin API to send overide-redirect windows to the current space when they're ordered in
      XQuartz: Revert most of the previous override redirect patch
      XQuartz: Send MotionNotify before button presses when X11 is in the background
      XQuartz: In rooted mode, make sure we start in the hidden state.
      XQuartz: Properly set the menu bar and hotkey state when changing rootless mode.
      XQuartz: Properly set the window level for the root window
      XQuartz: Fix window levels for rooted mode to allow showing the menu bar.
      XQuartz: Update window levels when changing rootless state
      XQuartz: xprSetWindowLevel updated to store the level requested by the WM
      XQuartz: Re-enable Fn as an option for 3button mouse simulation.
      XQuartz: Make sure the Fn doesn't trigger unneccessary calls to DarwinUpdateModKeys()
      XQuartz: Only set MotionNotify on activation if it is updated.
      XQuartz: Re-enable background window checking
      XQuartz: Solve the tablet 100% CPU bug
      XQuartz: Use correct values for ProximityIn and ProximityOut
      XQuartz: Make sure the bgMouseLocationUpdated state is consistent
      XQuartz: Localization update
      XQuartz: pbproxy standalone: added missing variable declarations.
      XQuartz: More localization updates
      XQuartz: English localization update for HIG compliance
      Rootless: Use miPaintWindow since PaintWindowBackground is dead
      XQuartz: Fix alpha to be 1 (screenshots bug, etc)
      GLX: Purge some glxint.h usage in glxcmds.c
      GLX: Purge glxint.h usage
      Rootless: Despite its name, we still want to allow rootless to mess with the root window
      XQuartz: Default to forcing the dialog box before quitting.
      Rootless: Use serverClient instead of NullClient
      XQuartz: AIGLX dispatch table cleanup
      GLX: Make sure the types match for ALIAS in indirect_reqsize.c
      XQuartz: More localization updates
      XQuartz: More localization updates
      XQuartz: 64bit fix for screen origin in AppleWM
      XQuartz: Don't leave zombied processes at startup
      XQuartz: ProcAppleWMAttachTransient to play nice with the new Dock in SL
      XQuartz: Cleanup keymap locking, fix a possible synchro bug
      XQuartz: Allow more than 3 OSX displays
      XQuartz: Cleanup getGlCapabilities to avoid hardcoding the number of displays
      XQuartz: Localization updates
      XQuartz: Initial support for automatic updates through Sparkle
      XQuartz: Added a "Check for X11 Updates..." menu item.
      XQuartz: Set can_quit to true during a Sparkle-initiated relaunch.
      XQuartz: Change handling of Windows menu to workaround a bug triggered by mixing Sparkle and X11 windows
      XQuartz: Rever the "Set can_quit to true during a Sparkle-initiated relaunch." change
      XQuartz: Copy the keyboard map to the core keyboard
      XQuartz: Use the master device in DarwinSendDDXEvent to avoid duplicate events.
      XQuartz: xpr: Added missing include for RootlessHideAllWindows
      XQuartz: Bump the reported version to X11R7.4
      mieq: Protect from pDev=NULL in mieqEnqueue and mieqProcessInputEvents
      XQuartz: Use pDev=NULL for DarwinSendDDXEvent
      Revert "XQuartz: Copy the keyboard map to the core keyboard"
      XQuartz: Use CopyKeyClass to copy the keymap to the virtual core keyboard.
      XQuartz: Overhaul setting up visuals
      XQuartz: Avoid namespace collission for BOOL in Sparkle
      XQuartz: Avoid a possible spinlock in applicationWillTerminate
      XQuartz: Define DDXRingBell
      XQuartz: no DirectColor
      XQuartz: Cleanup the bitmask setting for GLX visuals.
      XQuartz: Dead code removal for StaticColor visual
      XQuartz: Unify how we set our bitmasks for visuals
      XQuartz: Purge redundant QuartzBell
      XQuartz: Don't use location delta for tablets since NSEvent does not give a precise delta.
      XQuartz: Use mouseLocation rather than locationInWindow when setting lastpt
      XQuartz: Only save lastpt on mouse/tablet events
      XQuartz: AIGLX: Provide empty __glXAquaDrawableResize to avoid crashing in DoMakeCurrent
      CheckWindowOptionalNeed: Ensure w->optional is set to avoid SEGFAULT
      XQuartz: Use applewmproto 1.4 updated headers.
      XQuartz: Check NSINTEGER_DEFINED to make sure the NSInteger and NSUInteger types are defined.
      XQuartz: GL: Explicitly set GL_EXT symbols to 0 if they are not in OpenGL.framework to prevent X11's glext.h from setting them to 1.
      XQuartz: GL: Unset GL_EXT_gpu_program_parameters for Tiger/ppc
      XQuartz: GLX: Drawable does not contain resize anymore.
      Xext: Require newer versions of xcmiscproto, bigreqsproto, and xf86bigfontproto for new *proto.h header file names
      XQuartz: Use --with-launchd-id-prefix for consistency with xinit
      XQuartz: Check the DISPLAY environment variable to see if the socket at startup is ours.  If not, ignore it and fork/exec startx.
      XQuartz: Fix a strcpy/strcmp typo
      XQuartz: pbproxy: 64bit fixes: Properly process an array of Atoms
      XQuartz: Fix "warning: function declaration isn’t a prototype" noise
      XQuartz: launchd: Fallback on :0 if prefix:0 gives an error for the socket name
      XQuartz: pbproxy: Remove debugging XBell()
      XQuartz: GLX capabilities: Allow 16bit accumulation buffers
      XQuartz: Use internal xshm header for new xextproto
      Xi: CopyKeyClass is not static for XQuartz
      XQuartz: Nuke duplicate locks that make painful headaches
      XQuartz: Fix a brain-o array indexing problem
      XQuartz: Add pressure/tilt property labels
      XQuartz: Stop checking version numbers of the bundle because CFBundleGetVersionNumber is gimpish
      XQuartz: Transition from xEvent based mieq to InternalEvent
      XQuartz: Force a keymap resync on the first keypress to workaround XKB mucking with our keymap.
      XQuartz: Fix inverse map from mode_switch to alt
      XQuartz: Nuke TSM
      XQuartz: Fix a bunch of compilation warnings about style
      XQuartz: Fix QuartzSetCursor to match the expected prototype.
      Rootless: Correct border rendering on parent-relative windows

Jerome Glisse (2):
      DRI2: update DRI2 private drawable width & height according to X drawable
      Xext: fix a typo for bigreqsproto.h header file

Jesse Adkins (1):
      xserver doesn't stop all connections to localhost

Jesse Barnes (2):
      Don't prepare outputs & crtcs if set_mode_major is present
      xfree86/linux: don't change VT perms unless we're running as root

Jim Huang (1):
      kdrive: Move a close() to the right place so we don't close(-1) normally.

Joe Krahn (4):
      Cygwin/X: Fix for mis-aligned icon data creates bad background masks (#4491)
      Cygwin/X: Consolidate dialog initialization in winInitDialog()
      Cygwin/X: Update icons directly, rather than modifying the window's class
      Cygwin/X: Change to a single native window class for all X windows

Joel Bosveld (1):
      Correct fcntl(F_SETOWN) error check in non-Solaris sigio code

John Hein (1):
      Actually require bigfontproto when it's enabled.  Bug #13710.

Jon TURNEY (40):
      Cygwin/X: Update the native icon to one based on the X.Org logo
      Cygwin/X: build machinery fixes
      Cygwin/X: update for MPX cursor API changes
      Cygwin/X: update for MPX device changes
      Cygwin/X: update for changes in mieq API
      Cygwin/X: enqueue a pointer motion event on mouse movement
      Cygwin/X: update for changes to shadow framebuffer
      Cygwin/X: update to use standard DPMS stubs
      Cygwin/X: Initialize native HWND atom when built !XWIN_MULTIWINDOWEXTWM
      Cygwin/X: Fix some remaining build issues
      Cygwin/X: Add -static back to linker flags
      Cygwin/X: Add *.exe pattern to .gitignore
      Cygwin/X: Tidy up an unused #define
      GLX: Avoid a crash when we have an uninitialized GL context
      Cygwin/X: should also use GetTickCount(), just like Xming
      Cygwin/X: Fix compilation for mandatory XKB
      Cygwin/X: Fix -Wold-style-definition warnings
      Cygwin/X: Fix several prototypes to return HICON
      cygwin/X: Remove an unused variable to fix a warning
      Cygwin/X: Place prototypes for winCheckKeyPressed() and winFixShiftKeys() in a header file
      Cygwin/X: Place prototype for winSelectIcons() in a header file
      Cygwin/X: Fix keyboard layout mapping for Latin American keyboards
      Remove long-gone '-co' option from Xserver man page
      Remove references to rgb.txt from files section of Xserver and Xorg man pages
      Cygwin/X: link with libmain.a to provide main()
      Cygwin/X: Correctly allow for the native window frame width in ValidateSizing()
      Cygwin/X: Remove unused TimeSinceLastInputEvent()
      Cygwin/X: Tidy up a bodge to avoid collision between X header and native Win32 API definitions of ATOM
      Cygwin/X: Remove obsolete ChangeLog files
      Cygwin/X: Add a needed inputstr.h
      Cygwin/X: Fix permuted args to InitPointerDeviceStruct()
      Cygwin/X: Remove an obsolete mention of xf86Config
      Cygwin/X: more warnings fixes
      Cygwin/X: Improve mouse tracking for moving/resizing undecorated windows
      Cygwin/X: Allow pointer warping to work in rootless modes
      Cygwin/X: Avoid a visual glitch on window move in rootless modes
      Cygwin/X: Fix multiwindow extwm mode to build again
      Cygwin/X: Window positioning improvements for multiwindow mode
      Cygwin/X: winInitMultiWindowClass() should be static
      Cygwin/X: Only try to build rootless extension if multiwindow extwm mode is being built

Julien Cristau (48):
      Add missing include
      Typo fix
      randr: add swapped dispatch for RR[GS]etCrtcTransform
      Move RELEASE_DATE closer to AC_INIT so it's more likely to be updated
      xfree86/linux: fix log flood on acpid open error
      Xext: ANSI cleanups
      xsync: make SyncAlarmCounterDestroyed static
      Xi: fix missing declaration of XkbSetRulesDflts
      config: ANSI cleanups
      dix: ANSI cleanups
      dix: fix cast from pointer to integer
      dix: don't mix declarations and code
      Xvfb: ANSI cleanups
      xfree86: ANSI cleanups
      xfree86: don't mix declarations and code
      xfree86: use %zu to format size_t arguments
      xfree86: linuxPci.c needs a prototype for xf86AccResFromOS
      xnest: ANSI cleanups
      xnest: only define ddxBeforeReset if needed
      mi: ANSI cleanups
      mi: don't mix declarations and code
      xkb: ANSI cleanup
      randr: RRSetPrimaryOutput can be static
      os: ANSI cleanups
      os: backtrace() returns int, not size_t
      os: don't mix declarations and code
      require randrproto 1.2.99.3
      xfree86: ModeDebug is a boolean, not a string
      xfree86: Remove unused DEFAULT_UNRESOLVED and DEFAULT_BEST_REFRESH macros
      Add RandR 1.3 requests to protocol.txt
      Add XI 1.5 event and requests to protocol.txt
      Bug#21324: Add quirk for Iiyama Vision Master 450
      xfree86: Add two __FreeBSD_kernel__ checks
      xfree86: add edid quirk for Samsung Syncmaster 2333HD
      randr: fix server crash in RRGetScreenInfo
      configure.ac: drop dependency on fontenc
      Merge commit 'xorg-server-1.6.4' into upstream-experimental
      Merge branch 'upstream-experimental' into debian-experimental
      Bump changelogs
      Update build deps
      Update patch stack
      Cleanup some libtool cruft in debian/rules clean.
      Bump serverminver, videoabiver and inputabiver.
      Add bug closers
      Bump debhelper compat level to 5
      Actually install the Xephyr manpage
      Bump Standards-Version to 3.8.3.
      Prepare changelog for upload

Keith Packard (36):
      Add server support for RRGetScreenResourcesCurrent
      Clean up rotation data when crtc is turned off
      When disabling SIGIO tracking, use SIG_IGN instead of SIG_DFL.
      Use scrn->virtualX/virtualY in xf86CrtcFitsScreen. Fix bug 19017.
      Patch brown-bag fix for bug 19017 (scrn->virtualX/virtualY 0 before PreInit)
      Make cvt complain about invalid arguments more often.
      Ignore EDID-supplied monitor physical sizes for core screen size
      Make crtc_notify wrap/unwrap code do nothing unless mode code is inuse
      Make RandR CRTC info report panning area instead of just crtc area
      RandR SetOutputPrimary should work with or without a crtc connected
      RandR crtcs not fetched correctly when primary output is set.
      Handle matrix computation overflow in RRTransformCompute
      Damage re-used shadow scanout buffer using new transforms.
      Handle the combination of panning and crtc transforms
      Add XkbDir to Files config file section
      Eliminate the shadow clear on transform change
      RandR rotations and reflections offset by one pixel
      Pre-clip panning coordinates to keep crtc within panning region
      Oops. 62fc98c had crtc offsets applied backwards (fix from server-1.6-branch)
      Replace dixLookupResource by dixLookupResourceBy{Type,Class}
      Make RANDR 'set' timestamps follow client specified time. Bug 21987.
      dri2: Preserve compatibility with 1.6 DRI2 API/ABI
      Remove old DRI2 buffer alloc/free interfaces
      xf86_reload_cursors: fix cursor position to eliminate jumping after mode set
      Perform rotation redisplay before calling driver block handler (which may flush rendering)
      Ensure that rotation updates happen frequently
      Eliminate bogus event resizing.
      probe_devices_from_device_sections: don't leak list of devices
      xf86Xinput.c: get DIX event queue pointer once at InitInput time
      Free libpciaccess iterator in xf86PciProbe
      LoaderSortExtensions: free graph nodes
      xf8CrtcSetModeTransform: free adjusted_mode name too
      Free randr crtc and output pointer arrays
      RRDestroyOutputProperty: Free randr property valid values
      xfree86/modes: Remove all framebuffer support from DGA
      Fix sporadic segfault on resume due to accidentally freeing cursor.

Kevin E Martin (6):
      dmx: Make description match default option.
      Include <X11/extensions/dmx.h> and remove _DMX_SERVER_ define.
      dmx: disable Get/SetWindowPixmap, we don't support composite/redirected windows
      dmx: only free the default pixmaps that we actually allocated.
      dmx: reshuffle linker order to avoid errors when MITSHM is undefined.
      dmx: undefine MITSHM, move undefs to miinitext.c.

Kim Woelders (1):
      dix: Fix handling of do_not_propagate_mask window attribute.

Kristian Høgsberg (12):
      Avoid dereferencing NULL pScreen in xf86CrtcSetModeTransform().
      Drop unused DRI2 vblank infrastructure.
      Bump dri2proto requirement to 1.99.3.
      Test for DRI2 extension in dri_internal.h and only enable AIGLX DRI2 if found.
      Support setTexBuffer2 in AIGLX.
      Make GLX context lookup use dixLookupResourceByType()
      Don't stomp on dixLookupDrawable() return value in DoCreateGLXPixmap().
      Add validGlxDrawable() and use dixLookupResourceByType().
      Convert remaining GLX LookupIDByType() calls
      glx: Fix drawable private leak on destroy
      xfixes: Fix a couple of resource lookups
      composite: Fix resource lookups

Lee Leahu (1):
      seg fault when initializing DMX screens

Maarten Maathuis (62):
      randr: Avoid needlessly creating a shadow framebuffer.
      randr: Improve per-crtc gamma support.
      randr: some improvements, fixes and crtc abi bump
      Forgot to fix a comment.
      randr: check for malloc failure
      randr: add some notes about the stuff that changed in driver ABI 2
      randr: Consider panned crtc's when calculating xinerama screen sizes.
      randr/xfree86: Fix a one off error in the panning calculations.
      exa: a few cleanups
      exa: A more correct fix.
      xfree86/randr: Avoid a crash when xf86CrtcConfigPrivateIndex is -1.
      randr/xfree86: Fix initial gamma computation.
      exa: preparing as source and finishing access as mask is a bad idea
      exa: Allow drivers to set non-NULL devPrivate.ptr for !offscreen pixmaps.
      XKB: ifdef XKB is dead, yet it was still present in a few places.
      exa: Remove one of the many calls directly into the fb layer.
      exa: kill of exaImageGlyphBlt
      exa: add GC private
      exa: properly wrap GC functions
      exa: use proper wrapping in exa.c
      exa: wrap the remainder of exa_unaccel.c
      exa: create ExaCheckGetImage
      fb: move some code to mi
      exa: don't use fbCopyNtoN
      exa: Calling exaMarkSync after UTS is the drivers responsibility.
      mi: clip exposures to pGC->clientClip.
      mi: kill a few warnings
      exa: All fallbacks should have a GC, remove some code.
      exa: fix exaValidateGC.
      dix: always NULL pGC->tile.pixmap in Create{Scratch}GC.
      fb: add fbDoCopy and fbCopyRegion compatibility wrappers.
      exa: fix performance regression from 736b6fbd2c941b6276066cd1503523edebe7bf3d
      exa: reintroduce src rect optimisation, with a slightly higher threshold.
      exa: the extent of the valid region is probably much larger than that of the pending damage.
      exa: exaPixmapDirty should use official damage functions.
      Revert "exa: the extent of the valid region is probably much larger than that of the pending damage."
      exa: minor glyphs cleanup.
      exa: fixup exaAssertNotDirty.
      exa: whitespace
      exa: fixup aux indices and ensure that the indices are used as they should be.
      exa: simplify exaPixmapIsOffscreen
      exa: avoid a potential Prepare/FinishAccess inbalance.
      exa: increase/rework safety checks in Prepare/FinishAccess.
      exa: fix unwrapping of ModifyPixmapHeader upon CloseScreen.
      exa: remove a few pExaPixmap checks.
      exa: fix a serious issue in exaChangeWindowAttributes (and some more related things)
      exa: allow exaModifyPixmapHeader to set sys_ptr for EXA_HANDLES_PIXMAPS
      exa: check if the pixmap is pinned for unsupported AUX indices.
      exa: round of fb_pitch to the next byte
      exa: Accept scratch pixmaps with offscreen memory as such.
      xkb: plug a memory leak in XkbCopySrvLedInfo (#20756)
      exa: implement UTS based upload through CopyArea
      exa: Split out some classic and driver allocated pixmap code into seperate files
      exa: A simple 3rd backend implementation.
      exa: Use damage to optimise away useless copies.
      exa: implement exaMoveInPixmap for "mixed"
      exa: delay malloc for "mixed"
      exa: fix a potential loophole in "mixed"
      exa: one can never be too careful
      exa: Fix the broken upload fallback for "mixed"
      exa: minor cleanup
      exa: more safety

Magnus Kessler (3):
      Xvfb: Remove unused function GetLK201Mappings in InitInput.c
      .gitignore: ignore sdksyms.dep
      Xvfb: add missing include for new xkbsrv.h

Manuel Bouyer (2):
      netbsd: Fix alpha ev6 support.
      netbsd: Force the use of ev56 instructions for register access on ev56.

Martin-Éric Racine (7):
      Updated the PCI Vendor and Device IDs for all Geode variants.
      Removed unnecessary curly braces for Geode LX PCI ID.
      Further explained what the Geode PCI ID is all about as comments.
      Formatted recent Geode PCI ID changes to fit a 80-column coding style.
      Added comment with the PCI device ID of Geode SCx200 series.
      Use lowercase PCI vendor ID for NSC.
      xserver: added more Geode oddities to xf86AutoConfig.c

Matt Turner (19):
      Remove extra / from Loading module line
      Remove unused barrier macros on alpha.
      Fix mem_barrier() on Alpha
      Fix build on Alpha
      Move memory barrier macros into common section
      define barrier macros as nops if not otherwise defined
      Add x86 barrier macros
      Use sfence and mfence instructions on amd64
      Replace 8 nops with proper sync instruction on mips
      Remove unused ppc_flush_icache function
      Remove unused arm_flush_cache function
      Move unaligned access functions to common section
      Clean up unaligned access functions on alpha
      Simplify unaligned access code
      Remove unused ia64_flush_cache function
      Remove unused BUSmemcpy.c
      Remove unused Delay.c
      Make sys.c use compiler.h unaligned access functions
      alpha: kill xf86SlowBCopyToBus and xf86SlowBCopyFromBus

Matthias Hopf (23):
      randr: Weird enough, crtc->version was never set upon creation. Fix that.
      randr: Crtc interface update for panning support.
      randr: Panning support
      randr: Protocol bits for panning support
      randr: Nuke config-timestamp for panning
      randr: Rename pan() to set_origin(), and xf86CrtcPan() to xf86CrtcSetOrigin()
      randr: Don't change panning parameters if verification fails.
      randr: Rework panning area verification
      randr: Allow panning to be disabled per axis
      randr: Update SProcRandrVector for panning
      randr: Add monitor option "Panning" for initial panning configuration
      randr: Fix error message for bad panning config
      randr: Update initial screen size if panning information is present
      randr: Oops, miscalculated panning rectangle's coordinates
      randr: Fix initial panning border copy
      randr: Update EDID_ATOM_NAME to reflect RandR 1.3 name change
      randr: Nuke broken set_origin shortcut
      randr: Setting gamma: inverse logic looks more sane
      Revert "randr: Setting gamma: inverse logic looks more sane"
      Unclaim PCI slot if driver probing fails.
      Work around sporadic segfault on resume with intel/KMS due to cursor->bits == NULL.


Reply to: