Subject: [ANNOUNCE] xorg-server

Keith Packard keithp at
Tue Nov 25 23:39:57 PST 2008

Ok, here's the first beta release on the 1.6 branch. If you want
anything in 1.6 after this, please post them to the release page:

List the commit id of the patch from the master branch; all patches
going to 1.6 must already be on master before I will pull them across.

Aaron Plattner (17):
      Get modes code building with old X servers again.
      Include pciaccess in the xorg-server.pc Requires line.
      Xephyr: Handle depth 30 and reject higher depths rather than crashing.
      Bug #14692: Allow drivers to have a say in Xinerama visual consolidation.
      Short-circuit PanoramiXTranslateVisualID after verifying that the visual actually exists first.
      CreateColormap returns Success on success, not TRUE.
      Work around the DIX losing mmWidth/mmHeight for RandR 1.1 DDXen.
      Add LoaderShouldIgnoreABI to allow drivers that roll their own ABI checks (i.e. nvidia) to perform the check before ScreenInit.
      Try nvidia before nv under the assumption that people who install it usually want to *use* it, too.
      Make shmint.h part of the SDK.
      Add an xf86PrintBacktrace wrapper around xorg_backtrace to aid driver debugging.
      Bump ABI major versions for the TryClientExceptions change from commit 883811c.
      Conditionalize Composite-based backing store on pScreen->backingStoreSupport.
      Revert "Remove unused server state change callbackery" for now.
      Revert "Unexport (and unimplement) xf86SetAccessFuncs" for now.
      Revert "xfree86: xf86{Enable, Disable}InputHandler can be static."
      Restore xf86{Enable, Disable}GeneralHandler.

Adam Jackson (245):
      1.5 has branched, start 1.5.99.x.
      xf86DDCMonitorSet: Honor the DisplaySize from the config file.
      Fix Motif menu drawing in Xnest.
      RANDR 1.2: Inherit PreferredMode from the global configuration, if any.
      Remove all mention of the vga driver from the config logic.
      Make xf86InitialConfiguration slightly smarter.
      Fix distcheck.
      Fix segfault when a monitor exists but has no modes.
      Size xf86DefaultModes explicitly.
      Bug #14927: Fix the math for xf86NumDefaultModes.
      Add the "amd" driver to magic driver selection.
      RANDR 1.2: Don't report a square resolution to RANDR 1.1 clients.
      RANDR 1.2: Fix the RANDR 1.1 screen size estimation to approach reality.
      RANDR 1.2: Fix initial mode aspect ratio match in a corner case.
      Fix a stray use of ALLOCATE_LOCAL.
      Bug #13962: Re-arm the DPMS timer when re-enabling DPMS.
      Bug #11508: Fix build without XV.
      Bug #11510: Fix build without RECORD.
      Fix that last commit.
      Start E-EDID support in the DDC module.
      Refactor DDC2 code to allow for proper segmented addressing.
      Add E-EDID segment selection.
      Compile fix.
      Fix PCI config space cycles from int10 emulator.
      Add doltcompile to .gitignore
      So long, and thanks for all the cfb.
      Stop building mfb/afb/xf1bpp by default.
      Death to Extended Visual Information.
      Death to APPGROUP.
      Finish deleting EVI
      Remove appgroup mentions from
      Death to TOG-CUP.
      Remove fbpseudocolor
      Bug #13104: Don't let XAA glyph pixmaps anywhere near video memory.
      Bug #13104: XAA: Adapt to glyph storage changes.
      Bug #13104: Remove broken XAA a1 glyph fast path.
      X n'est pas une print API.
      Fix fbdevhw initialization for PCI drivers.
      Restore compression of duplicate log file entries.
      Remove global argc/argv variables.
      Simplify default font path setup.
      Move defaultDisplayClass to xdmcp-only visibility.
      The great kdrive purge.
      Fix hal shutdown crash.
      Re-add sync range inference from legacy setup to RANDR 1.2.
      Redo RANDR compatibility output selection.
      Fix initial mode selection even harder.
      xf86SetDesiredModes(): Skip disabled CRTCs first thing.
      RANDR 1.1 compat: remove senseless comparison against the virtual size.
      Fix reduced-blanking mode filtering in RANDR 1.2.
      Death to afb.
      Remove xgl.
      Death to mfb.
      ... and stay dead.
      Fix unterminated endif.
      Bug #12414: Create full-fledged pixmaps in fb24_32ReformatTile().
      Implement bswap in x86emu.
      Remove some default modes that really shouldn't be in the default set.
      Bug #15586: (Correctly) refuse to redirect the root window.
      Remove xorgconfig.  Use X -configure instead.
      Bug #11842: Fix emulation of int1A PCI BIOS services.
      Check for __amd64__, not __x86_64__.
      Also remove xorgconfig from
      XTrap delenda est.
      Bump ABI minor numbers for the devPrivate ABI functions.
      Actually fetch all blocks of EEDID if asked to.
      Hexdump the whole EDID block if extensions present.
      Do EEDID in RANDR 1.2.
      XInput ABI is 2.1, not 3.1.
      Declare xf86InterpretEEDID() properly.
      Don't try to load DRI2 if built without it.
      Auf wiedersehen, KAA.
      Fix GLX in Xvfb and kdrive.
      Bug #7300: Move xf86RAC to hw/xfree86/common
      Update the Allow*Grabs documentation for xf86misc removal.
      Remove the numVideoScreens xprintism.
      Remove font module support from xorgcfg.
      Remove loadable font renderer support.
      Build fixes
      Remove XF86Misc code from xorgcfg.
      Drop hw/xfree86/os-support/lynxos/
      Nuke some ancient code for commandline-challenged OSes.
      Drop the FONTCACHE extension.
      EDID: Add quirk to clamp max pixel clock to single DVI link speed.
      EDID: Correct DDCEstablishedModes to conform to DMT.
      EDID: For standard timing blocks, prefer DMT timings if they exist.
      EDID: Various reduced blanking fixes.
      Build fix.
      Unifdef QNX.
      Remove some broken debugging scaffolding.
      Assume __GNUC__ >= 3.
      Unifdef VXWORKS.
      Unifdef sgi.
      Unifdef DGUX.
      Unifdef AIX.
      Unifdef __osf__
      Unifdef hpux.
      Shape extension is built-in and mandatory.
      Bigreqs are built-in and mandatory.
      Allow extension closedown hook to be null.
      Remove all empty extension reset hooks, replace with NULL.
      XC-MISC is built-in and mandatory.
      Sync is built-in and mandatory.
      Remove some more MEMBUG garbage.
      Document more of the OS and library assumptions.
      Unifdef XIDLE.
      Handle XGI cards in autoconfig.
      Silence the "No matching Device section" warning in some harmless cases.
      Erk.  Actually check all the BARs, not just the first.
      Remove xf86Version.h and related API.
      Sledgehammer off the DRI2 build system.
      Build fix.
      Eviscerate save-under support.
      Drop dead PowerPC PCI code.
      Remove PciProbeType and associated weirdness.
      Remove vestigial MAX_PCI_DEVICES.
      Remove sparcPci.c
      Remove (unused) pciMaxBusNum.
      int10: Fix int1A for %ax == 0xB101.
      Remove MAX_PCI_{BUSES,DOMAINS} and some associated #if 0.
      Remove unused EstimateSizesAggressively option.
      Remove the remainder of grab deactivation and closedown.
      Remove unused xf86Info.sharedMonitor
      Remove unused xf86Info.mouseLocal
      Remove unused xf86Info.pMouse
      Remove unused -showunresolved option
      Remove unused -bestRefresh option.
      Remove unused -co option.
      Refactor misprite a bit.
      The ->Identify hook is not mandatory.
      Simplify calling xf86EnableIO() a bit.
      Simplify driver setup.
      Remove redundant check.
      Centralize declaration of ConnectionInfo.
      Suppress cursor display until the first XDefineCursor() request.
      EDID: Publish the whole block in the RANDR property if we've got it.
      EDID: Publish the whole block on the root window if we've got it.
      EDID: Build fix.
      Reimplement ShmPutImage.
      MIT-SHM pixmaps, if they exist, are ZPixmap.
      Build fix.
      Fix strict-aliasing silliness in XAA.
      Update to SGI FreeB 2.0.
      Revert accidental edid changes from previous commit.
      Default to -br.  Add -retro option for the nostalgic.
      Upgrade GLX Public License 1.0 to FreeB 2.0.
      Rename SaveSetMap() to SaveSetShouldMap().
      Fix un-suppressing the initial cursor.
      Don't inherit XAUTHORITY from the environment.
      Fix bad preprocessor line
      Unifdef ISC
      Remove some stale IPv6 debugging We don't look in root's ~ anymore, don't say we do.
      Remove an utterly, utterly ancient changelog.
      RAC: AccessEnable() and AccessDisable() are mandatory.
      RAC: NO_SEPARATE_*_FROM_* are useless.
      Unexport (and unimplement) xf86SetAccessFuncs.
      Unexport xf86EnterServerState
      Internalise some resource list API.
      Remove the old implementation of RemoveOverlaps
      Typo fix.
      xf86RegisterResources: Always print any failed registrations.
      Remove unused server state change callbackery.
      Unexport xf86ChkConflict
      Unexport xf86SetEntityFuncs
      Unexport xf86DeallocateResourcesForEntity
      Unexport xf86NoSharedResources
      Unexport MakeClientGrab{Imp,P}ervious
      Drop the never-called SkippedRequestCallback list.
      const cleanup
      Remove the Must_have_memory hack.
      Remove unused pciNumBuses
      loader: sort the mi symbol list
      Remove a bunch of unused API from the mi export list.
      mi{Create,Destroy}GCOps are unused, nuke 'em.
      miFindMaxBand is dead code
      staticize mi{G,S}etScreenPixmap
      Remove unused wrapping for visual initialization.
      Delete unused miModifyBanking()
      Fix compile for miSetScreenPixmap prototype.
      Dangling use of miDestroyGCOps
      xalloc+bzero -> xcalloc
      xalloc+memset(0) -> xcalloc
      Loader: Unexport a handful of consumerless symbols from dixsym.
      Build fix.
      Reinstate cursorScreenDevPriv
      int10: Remove useless check.
      int10: Don't warn when scanning for devices we don't have.
      int10: Fix a nasty memory leak.
      -retro enables zapping
      Document zap/-retro interaction more.
      I hate you xlib.  Please die in a fire.
      Fix more xcalloc stupidity.
      EDID: Catch monitors that encode aspect ratio for physical size.
      Revert "Don't leak EDID_block"
      Move xorg_backtrace() up to the OS level so we can call it from DIX.
      mieq: Backtrace when the queue overflows.
      kdrive: Hardcode a different wrong value for number of mouse buttons
      Add backtrace definitions to
      EDID: Catch even more cases of encoding aspect as size.
      int10: Degrade the checksum warning to X_INFO.
      exa: Add exaDrawableIsOffscreen() to the driver API.
      Bug #18159: Spell "anisotropic" correctly
      Fix previous cherry pick for API changes
      Change some #define foo() -> static void foo()
      Argh, macros.
      Add xf86MonitorIsHDMI() helper.
      kdrive: Bye bye Xvesa
      linux: Drain the console fd of data when using evdev for keyboards
      xf86TargetExact should try harder if there's only one monitor attached.
      Remove xf86GetResourcesImplicitly
      PCI: Always build domain support.
      PCI: Remove non-pciaccess path for x86.
      PCI: Remove config mechanism details.
      PCI: Simplify OS PCI function registration a bit.
      PCI: Remove unused ia64 platform code.
      Bus: remove the "reducer"
      Bus: Remove the notion of estimated resources.
      Bus: Remove unused RemoveOverlaps
      Bus: Don't try to find an ISA bus just for fun.
      Bus: remove useless isaConvertRange2Host
      PCI: Unexport xf86scanpci
      Bus: remove special handling for init-only resources
      Remove unused MIN macro
      Bus: Trust the kernel when registering driver resources
      Bus: Simplify a failure case (that pretty much never happens)
      Bus: Don't pretend to care about IRQs, DMA, or PCI config space
      Bus: Remove yet more unused overlap processing.
      Bus: Remove ISA support.
      Remove the remnants of Jensen support
      Remove xf86{En,Dis}ableInterrupts entirely
      Remove xf86IODelay
      Remove dead FreeModuleDesc.
      Warning fix.
      Code motion: subsume xf86DoProbe.c into xf86Init.c
      Remove unused XAAAvailableOptions
      Warning fix.
      Warning fix
      Warning fix
      Warning fix
      Warning fix
      Warning fix
      Dead code cleanup over DBE.

Adam Tkac (1):
      Fix Xvfb input when building against current X sources.

Alan Coopersmith (32):
      Make sure SIOGLIFCONF buffer is properly aligned for socket structures
      Xevie cleanups, byteswapping fixes & request length check fixes
      Make Xevie private symbol names less generic
      Update ac_define_dir macro in acinclude.m4 to 2008-04-12 version
      Bug 14247: If pkg-config can't find openssl.pc, just link with -lcrypto
      Allow using libmd instead of libcrypto for SHA1 hashing in render/glyph.c
      Return a valid X error when stuck in font alias loop
      Check for strcasestr and workaround it on systems without it
      When XKB fails to open rules file, log the file name, not the NULL file pointer
      Restore return type on xf86SetScrnInfoModes
      Don't log null device name in XkbDDXLoadKeymapByNames
      xf86GetOptValFreq wants a pointer to double, but maxPixClock is an int
      Update bsd & solaris descriptions in README.OS-lib
      Improved driver selection when autoconfiguring driver without xorg.conf
      Remove README.font, since it's just an old text copy of xorg-docs/sgml/fonts
      Remove more Lynx leftovers
      Make xstrcasestr prototype return value match the implementation
      Convert Solaris #ifdef's for <sys/kd.h> to AC_CHECK_HEADERS check
      Remove unnecessary #ifdef __SOL8__'s
      Allow non-gcc compilers to build xf86Xinput.c
      Refactor Solaris mapping code to work with standard shared/vidmem.c
      Non-Linux OS'es should default to kbd driver, not now-dead keyboard driver
      Use OsSignal in Popen/Pclose to avoid SysV signal() stupidity
      When HAL returns a NULL property, print "(null)" instead of a NULL pointer
      Use bsd_pci stub routines on Solaris too, since we use pciaccess for real work
      Always use server-<display>.xkm to avoid races when multiple servers start
      Xephyr man page typo fixes.
      Solaris: Make KDSETMODE failure non fatal, and retry it on interrupts
      Simplify filename generation code for Xorg -configure
      Fix typo in 5bf2c88d2317230b95b2904cb975167d03ee13a2
      Remove duplication from code paths in XkbDDXCompileKeymapByNames
      Fix const-mismatch warnings for DisplayModePtr's

Alan Hourihane (4):
      Set driverPriv immediately on CreatePixmap.
      Bump inputproto to for XI_PROP_ENABLED
      require inputproto
      re-enable DRI2

Alex Deucher (3):
      Revert "Fix RandR 1.2 driver interface conversion of two colour cursors to ARGB"
      Add missing newline when printing edid info
      Only add default modes if EDID supports continuous-frequency

Ander Conselvan de Oliveira (1):
      Fix incorrect test regarding keyboard map.

Bart Trojanowski (1):
      Bug #14332: Fix PCI access cycles from x86emu.

Ben Byer (39):
      prevent "fake mouse clicks" from generating spurious extra events
      xquartz: copy in new stubs from Xi/stubs.c to replace our old
      Remove calls to InitValuatorAxisStruct -- these are now handled in dix by
      add debug statements so we can see if/when our Xinput stubs are getting called.
      Add code to track 5 valuators for pointing device, in preparation
      Begin to move all of our Xquartz DDX-specific event handlers
      add logging of current thread ID to DEBUG_LOG macro
      add prototype for DarwinEventHandler
      gut darwinEQEnqueue, and make it just call mieqEnqueue (for the moment)
      remove vestigal DarwinEQPointerPost etc
      nuke DarwinEventQueue
      just a bit of juggling headers around -- we're preparing
      shovelling code around ...
      moved and renamed QuartzMessageServerThread to
      formatting cleanup for X11Application.m (no code changes)
      The AppKit thread should not be calling directly into the X server
      add missing dix-config include
      kludge: miEqEnqueue wants a device, even if we're passing custom messages,
      hack to Xquartz to prevent xmodmap from wiping out our valid modmap, per daniels
      darwinKeyboard: refactor slightly so that we're not cutting and pasting code from dix, kthx
      enable keyboard map debugging -- it's going to x11-debug.txt, anyway ... so no harm
      add support for horizontal scrolling (buttons 6 and 7)
      delete debugging spew
      formatting cleanup
      oops, missed a spot
      removed Xquartz debugging code that leaked into master.  Our Bad.
      Added XKB support for Xquartz
      random flailing
      Fix for pointer-offset issue when using a multi-display environment on
      Nuke a call to deprecated LookupClient, and hopefully prevent a
      add missing DARWIN_GLX_LIBS in
      xquartz: remove vestigial aglGlx.c
      xquartz: clean up linker line for main binary
      glx / xquartz:  we still need glxcontentmodes.[ch] kthx
      xquartz: build fixes for GLX/DRI2 fallout (note: not guaranteed)
      xquartz: remove unused params from DarwinEQInit
      xquartz: created darwinTablet, to represent tablet events
      Xquartz: experimental embedding of Sparkle.framework
      fix compilation on systems that don't have

Ben Gamari (1):
      dix: free FocusPrivates when deleting a window. #16493

Benjamin Close (1):
      dix: Handle the case where a device cursor was removed correctly

Benjamin Tissoires (1):
      kdrive: AddInputClient takes a ClientPtr.

Bernhard R. Link (1):
      Bug#16501: autodetection of sbus graphic cards

Brad Smith (3):
      BSD's also have /dev/urandom.
      OpenBSD doesn't define __x86_64__, simplify tests.
      Fix the configure tests for swap16/__swap16 macros on OpenBSD.

Brian Rogers (1):
      DGA: pass nevents to UpdateDeviceState. #15936

Carl Worth (1):
      Use AC_CHECK_LIB before PKG_CHECK_MODULES for openssl

Chris Ball (1):
      Remove KdOffscreenMarkUsed() as all its callers are gone.

Dan Nicholson (3):
      Fix up paths for GL/glx -> glx renaming
      Actually enable building GLX without Mesa source
      DRI: Ensure we have the DRI headers from Mesa

Daniel Drake (2):
      Don't abort if swrast library is not present
      Don't try to destroy a NULL pixmap.

Daniel Stone (55):
      XKB: Fix initial map setting on startup
      mi: More meaningful assert crashes
      XkbCopyKeymap: Fix broken indentation
      gitignore: Add two more bits Do the dolt
      Revert "Optimize dixLookupPrivate for repeated lookups of the same private."
      GL: Make errors non-fatal
      Build: Ensure xf86DefModeSet.c ends in an empty line
      XKB: Actually explain keymap failures
      Revert "GL: Make errors non-fatal" Add GLX_SYS_LIBS for Xvfb and Xnest
      DIX: Add strncasecmp from FreeBSD, make strcasecmp args const
      OS/KDrive/XFree86: Sanitise colour initialisation
      XFree86: Remove trailing whitespace
      XFree86: Clean up init a tiny bit (no code changes)
      XFree86: Delete empty file & function
      XFree86: Remove useless debugging code
      XFree86: Delete OSMouse code
      XFree86: Remove mysticism from
      config: Don't attempt to use D-Bus when not strictly necessary
      strcasecmp: Actually use the right license
      dix: Actually build str(n)casecmp if we don't have it Fix SHA1 handling
      DIX: Add strcasestr from FreeBSD
      HAL: Remove grotesque open-coded strcasestr
      XFree86: Remove XFree86-Misc extension
      Make --enable-debug usable again
      XFree86: Remove remnants of XFree86-Misc
      KDrive: Delete rafts of unused code
      XFree86: Restore AllowMouseOpenFail usage text
      Dead code removal
      Everyone has urandom Fix release date
      Remove dead code, useless #defines, et al Remove non-existant directory
      dix: Remove insane BC hacks
      Revert "Try nvidia before nv under the assumption that people who install it usually want to *use* it, too." Pass --force to autoreconf
      Remove kbd_mode
      Remove ioport
      Remove xorgcfg
      EXA: Remove unnecessary #includes
      Update COPYING for SHA1 code
      Input: Make DontZap the default
      Update docs to reflect new DontZap default
      XFree86: Fix DontZap interaction with -retro
      XFree86: Simplify DontZap/-retro interaction
      Xi: XINPUT has been mandatory for ages
      A bit of whitespace cleanup in include directives.
      Remove XEvIE
      XKB: Move XKBrules.h to the server too
      XKB: Fix thinko, causing warning (erroneously fixed in 5544c51447)
      XKB: Tiny cleanups to _XkbLookupAnyDevice
      DIX: Remove unnecessary prototype
      XKB: Remove pointless macros

Dave Airlie (23):
      quirk: fix LPL monitors properly.
      quirk: add quirk for ACER EDID
      quirk: add quirk for ACR 640x350 default mode is wrong
      glcore: zero fbconfigs before filling them in.
      glcore: make visualRating GLX_NONE - note GLX_NONE is not == 0
      glx: silly nitpick...
      GLcore: make googleearth not crash the server on sw-rendering.
      glx: test for valid read and draw privates before using them
      xf86edid: fix typo in debug output
      pci: don't do this pci stuff when we don't have hw access
      cursor: don't dereference NULL pointer is devPrivates not yet set.
      int10: add pci_device_enable support on Linux
      dbe: fix DoS reported by iDefense.
      modes: make aspect choosing work on single output case.
      modes: check the crtc is valid before using its desired modes.
      dri2: bump libdrm numbers as 2.3.1 is released without ttm interface
      modes: fix initial xorg.conf mode selection.
      exa: drop cw.h inclusion
      exa: make sure pixmap devPrivate.ptr is NULL at create time
      exa: don't call composite routines with no buffer.
      fixup another calloc build issue
      EXA: avoid copy operations if no boxes in use
      exa: avoid doing prepare/done without intervening copies in exaFillRegionTiled

Dave Miller (1):
      glx: fix crash in driGetConfigAttribIndex

David Nolden (1):
      compilation-fix in debug mode

David Nusinow (1):
      Bug #10016: Implement WM_CLASS hints in Xephyr.

Dodji Seketeli (3):
      [Xephyr/DRI] correctly route motion events targeted at GL drawable
      [Xephyr] Fix #15839
      [Xephyr] various X86_64 build cleanups

Donnie Berkholz (2):
      Xephyr: Distribute ephyrdriext.h in tarballs.
      Xephyr: Build fix: Port across XF86dri.c changes from Mesa.

Doug Chapman (1):
      Bug #14091: Fix build (and runtime) on ia64.

Drew Parsons (3):
      Define XPRINT in XPRINT_CFLAGS (
      Create dix/ for Xprint-specific build of
      Disable D-BUS from Xprint.

Eamon Walsh (40):
      dix: Convert selection list to a linked list.
      XACE: Check the return value of the selection create hook call.
      xselinux: Implement polyinstantiation support and related protocol.
      Apply __glXDisp_GetVisualConfigs message patch
      XSELinux: Correctly handle some permission bits that are used more than once.
      XSELinux: Do a check for whether background "None" is allowed.
      XSELinux: Add xorg.conf option for permissive/enforcing/disabled.
      XSELinux: Add a request to get a client's context from a resource ID.
      Make devPrivates lookup functions ABI instead of static inlines.
      Fix "warning: implicit declaration of function ‘miPointerGetScreen’".
      Fix "warning: assignment from incompatible pointer type".
      Fix "warning: pointer targets in assignment differ in signedness".
      Fix "warning: passing argument 1 of 'pixman_fill' from incompatible pointer type".
      Fix "warning: unused variable ‘pDev’, `pPointer'".
      Fix "warning: no previous prototype for ‘XevieExtensionInit’".
      Fix "warning: label ‘done’ defined but not used".
      Fix multiple warnings in os/xdmauth.c.
      Fix "Warning: no previous prototype for `SecureRPCInit'".
      Fix "warning: no previous prototype for ‘miRROutputGetProperty’".
      Fix "warning: ‘rc’ may be used uninitialized in this function".
      Fix "warning: unused variable `tsize'".
      Fix "warning: unused variable `s'".
      Fix "warning: the address of ‘newTypes’ will always evaluate as ‘true’".
      SELinux: Bring server-side name registry up to date.
      SELinux: Add an extension alias under the OS-agnostic "Flask" name.
      Fix "warning: passing argument 1 of `xf86addNewLoadDirective' from incompatible pointer type".
      Fix "warning: passing argument 1 of `modeIsPresent' from incompatible pointer type".
      Fix "warning: assignment discards qualifiers from pointer target type".
      Fix "warning: suggest parentheses around && within ||".
      Fix a leak in the code that parses the protocol names.
      Fix a leak of window devPrivates in window.c/CrushTree().
      Don't need to request space for Xnest pixmap private.
      Remove unused GetGlyphPrivatesForScreen.
      Add an array of integers for use as per-screen cursor private keys.
      Attempt getpeercon() on remote sockets as well as local ones.
      Array-index based devPrivates implementation.
      xselinux: send more specific message types to libaudit.
      xace: Fix an incorrect call to the RECEIVE_ACCESS hook.
      xselinux: don't pass a NULL key string to selabel_lookup().
      xselinux: use "raw context" variants of getpeercon() and getcon().

Egbert Eich (3):
      xkb: use the correct device instead of an uninitialised "dev". #15614
      DDX/RandR: When positioning outputs assume width/height = 0 if no modes exist.
      int10: Do an mprotect(..,PROT_EXEC) on shmat()ed memory ranges.

Eric Anholt (8):
      Fix clock_gettime presence detect on FreeBSD.
      dolt: add FreeBSD support (this and ppc have been submitted upstream).
      Remove NDEBUG (assert() disable) define from the X Server.
      Fix dolt to error out on compile error, and not print errors on race to mkdir.
      Update dolt from upstream, fixing fallback to libtool.
      Nuke the MIT-SUNDRY-NONSTANDARD extension.
      Add necessary include for COMPILEDDISPLAYCLASS.
      Fix distcheck.

Fernando Carrijo (2):
      input: Remove useless conditional.
      [dix] Remove a duplicate statement.

Fredrik Höglund (3):
      EXA: Improve the algorithm used for tracking offscreen pixmap use.
      EXA: Optimize the eviction scanning loop in exaOffscreenAlloc.
      Fix the tile offset in miPaintWindow for ParentRelative windows.

George Peter Staplin (32):
      Add image/jpeg type, and rename reclaim_clipboard to claim_clipboard.
      Rename reclaim_clipboard to claim_clipboard.
      Add an image_jpeg Atom.
      Rename reclaim_clipboard to claim_clipboard.
      XQuartz: pbproxy: return the TARGETS list as a 32-bit list to fix a bug that was in the original.
      XQuartz: pbproxy: The greedy CLIPBOARD handling now works for text.
      XQuartz: pbproxy: Use an NSBitmapImageRep to convert an "image/png" selection
      XQuartz: pbproxy: pbproxy now responds to selection request events.
      XQuartz: pbproxy: Add COMPOUND_TEXT handling.  Do misc. cleanups with testing
      XQuartz: pbproxy: Add a missing include of Xutil.h.  Another header
      XQuartz: pbproxy: Make the proxy handle copy request races, and PRIMARY
      XQuartz: pbproxy: Fix a bug that occured when a PICT format was available.
      XQuartz: pbproxy: Add an X error handler that returns 0.
      XQuartz: pbproxy: Add a comment to pbproxy.h about how the DB()
      XQuartz: pbproxy: Fix NSObject memory leaks by properly using the
      XQuartz: pbproxy: Fix 3 paths that could leak memory.
      XQuartz: pbproxy: Fix a bug that rxvt brought out in STRING copying.  Now I can copy and paste to/from rxvt correctly.
      XQuartz: pbproxy: Fix another STRING bug.
      XQuartz: pbproxy: Make the signal handler safer by using _exit, and only exit.
      XQuartz: pbproxy: Possibly fix a memory leak by using an [NSApp run] loop,
      XQuartz: pbproxy: Fix 2 bugs with the preferences.
      miext/rootless: Fix the usage of DeleteProperty so that this builds again.
      XQuartz: Remove the enable_stereo default.  We now use feature detection to enable stereo visuals.
      XQuartz: pbproxy: pbproxy.h had invalid macro usage with 2 #else statements, so it wasn't building.  Fix the build.
      XQuartz: GL: Add feature detection of color sizes, and accum sizes.
      XQuartz: GL: Fix several error handling mistakes in handleRenderDescriptions, that were leftover before the final code evolved.
      XQuartz: pbproxy: Add code to handle PICT conversion to PNG and JPEG.
      XQuartz: pbproxy: Work around bugs in some apps that aren't ICCCM compliant.
      XQuartz: pbproxy: Fix a bug that prevented nedit CLIPBOARD from being proxied.
      XQuartz: GL: Mark visuals that are slower than others, due to lack of acceleration.
      XQuartz: GL: Work around problems with really deep visuals.
      XQuartz: GL: Add capability detection for depth buffers, and multisampling.

George Sapountzis (22):
      glcore: prepare for dynamic loading
      glcore: build from mesa, dlopen from xorg
      glcore: drop GLcore (build system)
      glcore: drop GLcore (files)
      drop EVI remnants
      drop TOGCUP remnants
      xephyr: Makefile cleanup 1
      xephyr: Makefile cleanup 2
      xephyr: libGL is client lib
      xephyr: sync with mesa
      xephyr: XEPHYR_DRI is identical to XF86DRI
      Add loader for the swrast dri driver.
      drop xprint remnants: xpstubs
      drop xprint remnants: InitGlobals
      glx: drop drm.h, sareaHandle from swrast loader
      kdrive: need not export symbols for glx anymore
      glx: drop stray glcore.h include
      glx: missing swrast is fatal
      glx: fix forgotten swrast -> SWRast
      glx: fix memory corruption with r5g6b5
      glx: drop unused cleargc
      clean .gitignore

George Staplin (10):
      XQuartz: GL: Provide code for getting the capabilities of the underlying system's CGL.
      XQuartz: GL: Add the new C code to the
      XQuartz: GL: Add the proper license information to the new files, and add more commentary to setVisualConfigs().
      XQuartz: GL: setVisualConfigs() copied the behavior of indirect.c which is wrong, with regard to the handling of xcalloc failure.
      XQuartz: Restore the logic of AppleWMDoReorderWindow to the same as the 1.4 branch.
      XQuartz: Disable 2 calls to miSetVisualTypesAndMasks that seem to be unnecessary, and possibly wrong (with regard to masks and bits per RGB).
      XQuartz: GL: Handle the alpha differently when the alpha is equal to GLCAPS_COLOR_BUF_INVALID_VALUE.
      XQuartz: GL: remove glcontextmodes.* from the  It has been replaced.
      XQuartz: GL: Add a branch to prevent a NULL DrawablePtr structure access.
      XQuartz: GL: Remove the inclusion of glcontextmodes.h.

Goneri Le Bouder (1):
      xfree86: don't crash in AutoConfig if the primary device is not pci

Hans de Goede (1):
      Prefer glxvisuals with stencil buffer for default visuals

Hasso Tepper (2): DragonFly BSD support Fix monotonic test harder

Hong Liu (2):
      Bug #15160: quirk Proview AY765C
      Keep rotation wrappers in place while transforms are in use

Ian Romanick (3):
      Bug #15169: Make the server build again on Alpha.
      VBO: Regenerate files based on recent changes to gl_API.xml
      GLX: Make sure GLX_SGI_make_current_read is enabled when possible

Ivaylo Boyadzhiev (1):
      Fix a longstanding XAA CopyPlane bug.

James Cloos (8):
      Fix some documentation typos
      Clean up whitespace
      Prevent the -wm command line option from causing a SEGV
      Fix compile failure
      Fix LookupColor
      Optimize commit 331cc3f0799a54910a99484264f76569beeee55a
      Remove some null statements.
      Don’t recommend fixing HAL if HAL support is disabled.

Jeremy Huddleston (266):
      Rootless: Removed safeAlphaXXXX() in favor of using fb/pixman
      XQuartz: Initial framework for dealing with spaces on OS-X
      Xquartz: Added separate preference tab for quartz-wm settings
      XQuartz: Disable 256 color option and fullscreen option
      XQuartz: Sanity-check the removal of the 256-color option
      XQuartz: Fixed names of enums to be more consistent
      XQuartz: Fixed missing close-paren in preference pane text.
      turns out we weren't actually using these files. oops
      continue with gutting darwinEvents.c
      XQuartz: Change reporting of space change to debug log rather than stderr
      XQuartz: Fix issue where clicking on an X11 window might send that event to an X11 window in another space.
      Xquartz: Added applicationShouldHandleReopen:hasVisibleWindows to handle dock icon clicking
      XQuartz: Removed a call to RootlessReorderWindow from the Carbon thread
      Xquartz: Don't need to link against rlAccel since we don't use it
      XQuartz: Don't use composite.
      XQuartz: Don't enable rootless accelerated functionality... crashy...
      merged darwinKeyboard.[ch] into quartzKeyboard
      XQuartz: Use strerror(errno)... cause I like text more than grepping header files
      XQuartz: Include version info for CrashReporter
      XQuartz: Added framework for asserting which thread we're in.
      XQuartz: Fixed some missing prototypes
      XQuartz: A little more debugging output from threadSafety
      XQuartz: Use a mutex to ensure we only have one thread calling mieqEnqueue at a time.
      XQuartz: Moved some rootless-specific cruft into xpr
      XQuartz: Forgot to commit xprEvent.[hc] ...
      XQuartz: Handled sanitization of namespace better
      XQuartz: Removed unneccessary include
      XQuartz: More sanitization of the namespace
      Last of the spam... I promise...
      Rootless: Kill off rlAccel
      Removed XWin DDX
      XQuartz: Cleaned up some casting to get rid of compiler warnings
      XQuartz: Cleanup turning off COMPOSITE
      Revert "Removed XWin DDX"
      .gitignore: Added Xquartz stuff
      XQuartz: Fixed quit dialog to be more conforming with HIG.
      Fixed dixLookupClient to work with client=NullClient as it did in the 1.3 branch
      Make rootless use dixLookupClient rather than deprecated LookupClient.
      XQuartz: Fixed cmd-tab to bring all windows forward.
      XQuartz: Make sure QuartzAudioInit() gets run.
      XQuartz: No need to include indirect.c a second time
      Moved Apple GL bits into our DDX subtree
      XQuartz: Compile in missing glx source that we still need
      Added missing HAVE_LAUNCHD to
      darwin gots /dev/urandom, too yo!
      XQuartz: Cleaned up multi-monitor support.
      XQuartz: More multi-monitor work... reverted Ben's workaround (worked for side-by-side only) and added more debugging.
      XQuartz: Added some pseudoramiX debug traces
      XQuartz: Don't subtract the titlebar off of the pointer_y
      XQuartz: Cleaned up startup and thread creation a tad.
      Xquartz: Added missing to EXTRA_DIST
      XQuartz: Added missing to EXTRA_DIST
      XQuartz: Updated Localizable.strings
      XQuartz: More startup / threading house cleaning.
      XQuartz: Added thread debugging to xprFrame.c
      XQuartz: Disabled DPMS extension
      XQuartz: xprIsX11Window can be called from the Appkit thread (see X11Application.m)
      XQuartz: Unset CFProcessPath... blech
      XQuartz: Fix to tablet-event handling code; we now scale
      Dix: Cleaned up sanity checking in Get{Pointer,Keyboard}Events
      XQuartz: Moved some relevant stuff from darwin.h to darwinEvents.h
      XQuartz: Adjust the screen origin offset properly for multimonitor setups
      XQuartz: Avoid a possible crash at startup due to unfavorable context switching.
      XQuartz: Fix mouse input offsets earlier since GetPointerEvents does not like negative (x,y) values
      Rootless: mi doesn't let us resize root, so we need to do it ourselves...
      XQuartz: Move server bits into bundle and setup stub in /usr/X11/bin/Xquartz in prep for startup rewrite
      Fixed up dist
      XQuartz: Added uncommitted files
      XQuartz: Fixed typo
      Set CSRG_BASED on OSX
      XQuartz: Fixed some issue in our bundle creation
      XQuartz: Set bundle version to 2.3.0
      Updated .gitignore for new Xquartz layout
      XQuartz: Reorganized some of the build system in prep for the Mach IPC startup work.
      XQuartz: Starting to work on the new Mach IPC startup stuff for better launchd, ApplicationServices, and Dock support
      XQuartz: More startup work... listen if we're the actual server
      XQuartz: Added some version checking protection so we don't trigger an infinite exec loop with new /usr/X11/bin/Xquartz and older
      Don't need the fink-friendly printf in the DDX anymore.
      XQuartz: Cleaned up the about box.
      Added missing to EXTRA_DIST
      Xquartz: More work on the new Mach startup
      XQuartz: More work on the Mach-IPC startup path
      Xquartz: Xquartz stub now can send Mach IPC to start the server.
      XQuartz: Made DarwinSendDDXEvent a little more robust to context switching diring server init.
      XQuartz: Disable xkb since it doesn't work after getting xkeyboard-config installed
      XQuartz: Added functionality to add a file descriptor to the connection list after the server is already running.
      XQuartz: Ok, pass XQUARTZ_USE_XKB since it breaks worse without it... but we have issues when we have the keyboard configs installed... need to figure out what to do there...
      XQuartz: More work towards Mach-IPC startup... started working on FD passing
      New XQuartz icon thanks to Simone Karin Lehmann
      XQuartz: Cleaned up ListenOnOpenFD...
      XQuartz: Fixed dropped code in the !XKB blocks, disable XKB support until we figure out a solution for xkeyboard-config
      XQuartz: First stab at SCM_RIGHTS passing the $DISPLAY launchd fd from the stub to server
      XQuartz: Fixed a few issues with fd passing... still not working =(
      XQuartz: Move the launchd display grabbing into mach_startup
      XQuartz: Don't need launchd checking in anymore since its functionality has been removed from libxtrans and into the mach startup code.
      XQuartz: Add the launchd fd to AllSockets as well
      XQuartz: A 2 second delay for launchd socket connections should be sufficient.
      XQuartz: Switched over to new startup path for testing.  Cleaned it up a bit.  Server still crashes when using the icon to launch =/
      XQuartz: Fork for trigger
      XQuartz: Don't forget to destroy the mutex and cond after we're done with them
      XQuartz: use a condition variable to signal when darwinEvents is ready rather than polling
      XQuartz: Removed async debugging sleep
      XQuartz: Updated icon with more rsolutions and made it more dark-background friendly
      XQuartz: Fixed the "laggy" startup under the new model.  This was caused by xinit blocking expecting SIGUSR1 and our stub never sending it.
      XQuartz: Removed code path for old startup
      XQuartz: Under the new startup model, we no longer need to do the foreground/background dance.
      Xquartz: Removed include directive for removed header
      XQuartz: Set noPanoramixExtension earlier to avoid a possible race.
      Set machine dependent defaults for ppc64
      XQuartz: Remove deprecated keyboard code.
      XQuartz: Some fd handoff cleanup.
      XQuartz: Use CFEqual to compare keyboards
      XQuartz: Fixed first-client-can't-connect bug
      XQuartz: Nuke Sparkle.
      XQuartz: Fixed proper processing of tablet button presses
      XQUartz: UI Cleanup.  Removed done/cancel buttons from Applications->Customize
      XQuartz: Fixed some errors / typos in the preferences pane.
      XQuartz: More fd handoff cleanup.
      XQuartz: Added code and made comments more helpful for debugging first-client-auth bug.
      XQuartz: Now properly disable xauth checking on launchd socket and mostly fix the xinitrc / launchd race condition
      glcontextmodes.[hc] were not added with the removal of the meas symlinks patch.  Copied from mesa head 2008.08.06.
      XQuartz: Re-added deprecated code fallback failsafe for keyboard layout on Leopard with some debugging spew.
      XQuartz: Disable the Xquartz AIGLX for now since it doesn't even compile any more...
      Removed rogue #include
      Transport/Apple: Add TRANS_NOXAUTH to incoming connections when the listener is flagged as such
      XQuartz: Reverted "Control" text to be accurately "Command" in the input prefs.
      XQuartz: Made 3-button mouse simulation a little more consistent.
      XQuartz: Fixed a stuck modifier key bug.
      XQuartz: Updated bundle version to 2.3.1
      XQuartz: More input fixes
      XQuartz: Added appkit_modifiers defaults option which users can set to 'rcommand' to get access to the input menu with right command
      XQuartz: Just cleaned up formatting of event processing code... no "meat" changed...
      XQuartz: Added window_item_modifiers defaults item (and option to localization) to change the modifier keys used for the windows menu.
      XQuartz: Support windowItemModMask=0 or -1 to disable window shortcuts.
      XQuartz: Added more explanation to debug keylayout spew to cut down on report-spam.
      Xquartz: Added missing ev_type for NSTabletPoint event.
      XQuartz: Cleaned up mouse event translation a bit more.
      XQuartz: Fixed bogus args to debug ErrorF
      XQuartz: Made a note to come back to the xinitrc race condition in the 1.5 branch when we move up to it.
      XQuartz: Use new Xplugin API to determine if a keypress event corresponds to an OS-X hotkey that we should dispatch to Appkit.
      XQuartz: Fixed needed xplugin version. ><
      XQuartz: Always use TIS for the keyboard layout seed since KB* aren't thread safe.
      XQuartz: xp_is_symbolic_hotkey_event catches the input menu, but not our mainMenu, so we still need to do that the old way.
      XQuartz: Added debugging output to the crash log to help track down two crashes, since people don't often report their system.log spew.
      XQuartz: Placate automake which doesn't like escaped newlines even in comments...
      Removed dead glcontextmodes
      XQuartz: Don't even build with GLX support since swrast now depends on libdrm
      XQuartz: Fixed missing symbol in quartzKeyboard debugging
      XQuartz: Fixed editing of Applications menu
      XQuartz: Cleaned up our sleeping before adding the launchd socket.
      XQuartz: Removed dead code (unused option definitions)
      XQuartz: Updated man page
      mieqProcessInputEvents: Check custom event handlers first.
      XQuartz: Removed "Done" button from Applications Customization.  Now properly save on exit.
      XQuartz: Updated Tablet code to include ranges for valuators and added cursor and erasor
      XQuartz: Added pasteboard proxy code stripped out of quartz-wm.
      mieq: Fix my broken cherry pick.
      XQuartz: Fixed a crash in RootlessNativeWindowMoved
      XQuartz: Remove decls of crashtracer strings that we don't use anymore in here
      XQuartz: Fixed proximity logic errors for tablets.
      XQuartz: Dead code removal
      XQuartz: Don't warp the pointer on server start for XQuartz
      XQuartz: Changed around fd handoff model to be more robust
      XQuartz: Dead code removal
      XQuartz: 256 color support
      XQuartz: Dead code removal
      mieq: Made custom event handlers safer for nevents >1
      XQuartz: pbproxy: First round of updates to pbproxy from George.
      XQuartz: pbproxy: Added global variables to customize how pbproxy behaves
      XQuartz: Don't enable 8bit visuals in the TrueColor server, since it's not working yet
      XQuartz: Added a preferences UI for pbproxy options.
      XQuartz: "Disable" the text when syncing is disabled.
      XQuartz: 64bit crash fix ... don't pass pointers through xe.
      XQuartz: pbproxy: Removed stray SPI include.
      XQuartz: Fixed threading issue with TSM.
      Added dep on applewmproto 1.1.0
      XQuartz: Added a --enable-integrated-xpbproxy configure option for building xpbproxy as an app or as a thread.
      XQuartz: Send AppleWMReloadPreferences where appropriate.
      Xquartz: xpbproxy: Split out app-specific stuff to app-main.m in prep for making this into a thread.
      XQuartz: Actually send the pointer event from the pen even if it happes to actually have 0 pressure and 0 tilt.
      XQuartz: Added some localizable strings that were missing.
      XQuartz: Removed resolved comment about localization.
      XQuartz: Using absolute ranges for pointer location to increase resolution and better support tablets.
      XQuartz: Workaround for initial pressure/tilt being sent as 0 with motion during the proximity event
      XQuartz: Use "pointer" and "pen" for device names to please GDK.
      Partially reverts commit 505c216d030a8eb6e9fa757ecebed9597a850c88.
      XQuartz: GLX - enable_stereo defaults option should work now
      XQuartz: xpbproxy: Support some of the preference toggles from, cleaned up CLIPBOARD_MANAGER atom management.
      XQuartz: pbcopy: Re-enabled some code I commented out for testing
      XQuartz: Bumped bundle version to 2.3.2 for beta1
      Depend on xtrans >= 1.2.2 for TRANS_NOXAUTH
      Xquartz: Fix parallel builds
      XQuartz: Call setVisualConfigs in InitOutput, and only when GLXEXT is defined.
      XQuartz: Fixed EXTRA_DIST for new GL files.
      XQuartz: Some motion made towards supporting fullscreen.
      XQuartz: implemented primary-on-grab and fixed clipboard-to-pasteboard
      XQuartz: Fixed window levels.
      XQuartz: Require applewmproto-1.1.1
      XQuartz: Fixed missing key-equiv for Hide-Others.
      XQuartz: More progress towards fullscreen...
      XQuartz: Fixed follow-keyboard-layout
      XQuartz: Cleaned up keyboard init and map reload.
      XQuartz: More fullscreen fixes
      XQuartz: Only send mouse events while in the background if we're over a visible window.
      XQuartz: pbproxy: SIGHUP reloads preferences.
      XQuartz: Default dpi to 96 instead of 75
      XQuartz: Added missing semicolons to Localizable.strings
      XQuartz: Honor the Spaces preference for "When switching to an application, switch to a space with open windows for the application"
      Rootless: Properly typed expose_1
      XQuartz: pbproxy: Fix compilation on case-sensitive file systems...
      miext/rootless: Silence some warnings by using proper C prototypes.
      XQuartz: Do check_xinitrc() later to avoid possible deadlock.
      Xquartz: pbproxy: Some debugging about unhandled mime types... this will spam up quick, but will be useful for a while...
      XQuartz: pbproxy: fixed blocking
      XQuartz: Pass along SIGINT and SIGTERM from the stub to
      XQuartz: Purge the old XQUARTZ_EXPORTS_LAUNCHD_FD option
      XQuartz: if 0 out a block to avoid thread-unsafe lockups
      XQuartz: Fixed Xplugin thread safety issues, so rea-enable background window check
      XQuartz: Don't map the menu bar to the screen area.
      XQuartz: Silence pointer debug spam
      Rootless: Force initialization of static variable before use.
      XQuartz: displayScreenBounds uses quartzEnableRootless rather than !quartzHasRoot
      Rootless: LP64 related casting fixes from Bob Murphy
      XQuartz: LP64 related casting fixes from Bob Murphy
      XQuartz: Add option to allow access to the menu in fullscreen mode
      XQuartz: version string updated for 2.3.2_beta3
      XQuartz: Replace deprecated API usage thanks to Robery Murphy
      XQuartz: Make LS parse our version properly... it didn't like our CFBundleVersion
      XQuartz: CFBundleShortVersionString needs to conform to X.Y.Z as well... sigh
      XQuartz: Added more debugging to handoff spew
      XQuartz: Fixed XBell() when not using system alert.
      os/connection: TRANS_NOXAUTH bit comparison brainfart fix.
      XQuartz: Made X11Application.m a little more tidy.
      Xquartz: Force sending mouse clicks to AppKit if we're over the menu bar
      XQuartz: Initial work on readding GLX to the 1.5 server
      XQuartz: Actually add glcontextmodes.[hc]
      XQuartz: Integrated xpbproxy into the server.
      XQuartz: Don't use TIS on Tiger...
      mi: Only increment tail (push) when the event data is actually in the queue
      mi: Push screen-saver/DPMS handling to after the EQ pop operation.
      XQuartz: pbproxy: Remove thread-main.m
      XQuartz: pbproxy: nuke RCS Id, update header license/copyright
      XQuartz: pbproxy: Push the pbproxy Xevent processing into its own thread
      XQuartz: Resurrect the old audio path for Tiger
      XQuartz: More Tiger ifdefs
      XQuartz: pbproxy: Added some typedefs for types not available on Tiger
      XQuartz: spewCallStack is noop on Tiger now...
      XQuartz: Update applewm deps
      XQuartz: Don't use LS to find on Tiger.
      XQuartz: GLX: sync up indirect.c to match George's updates in the xorg-server-1.5-apple branch
      XQuartz: Updated some code to use newer server API
      XQuartz: Fixed --disable-glx
      XQuartz: More 1.6 server API updates
      XQuartz: More Tiger cleanup: bootstrap_strerror
      XQuartz: Dead code removal
      XQuartz: Disable some error spew on Tiger (where it wouldn't be an error)
      XQuartz: More dead code removal
      XQuartz: Don't hardcode values of org.x.X11 for the preferences domain
      XQuartz: Dead code removal
      XQuartz: Use the environment to pass the bundle's prefs domain on to xinit/quartz-wm for Tiger or no-launchd-LEOPARD
      XQuartz: Add fallback for xpbproxy's display for Tiger or no-launchd-Leopard
      XQuartz: Removed hardcoded org.x.X11 from MachIPC as well
      XQuartz: Force X11Controller to reset a broken DISPLAY envvar.
      XQuartz: Simplify the xquartz_resetenv_display path

Jeremy Uejio (2):
      Sun bug #6685465: Xephyr uses wrong or bad colortable in 8-bit mode
      Refix Sun bug #6685465: Xephyr uses wrong or bad colortable in 8-bit mode

Jerome Glisse (1):
      dri2: fail at context creation if driver fail to create it's context

Jesse Barnes (6):
      Add cscope files to .gitignore
      Allow RandR get output property to call into drivers
      Fail CRTC configuration if !vtSema
      Make xf86SetDesiredModes aware of current output configuration
      Cleanup logic in xf86PrepareOutputs
      Add cscope files to .gitignore

Jesse Ruffin (1):
      Bug #16302: Fix GLX drawable refcounting in DMX.

Jie Luo (4):
      Fix color lookup.
      Remove freetype from the list of autoloaded modules.
      Remove duplicate load of freetype font renderer.
      glx: avoid possible NULLptr deref, fix #16884

John Tapsell (2):
      fbdevhw: Remove pixclock check.
      Render: Use built-in SHA1 library

Jordan Crouse (1):
      xf86: Change AutoConfig driver for PCI ID 022:2091 to 'geode'

Julien Cristau (41):
      Programs in hw/dmx/examples/ want client-side xkb headers.
      exa: use xf86ReturnOptValBool instead of xf86IsOptionSet
      xaa: use xf86ReturnOptValBool instead of xf86IsOptionSet
      Fix the clock_gettime check for glibc-based non-Linux systems
      Fix composite on !darwin
      Minor xorg.conf manpage fixups
      Don't set DRI2=yes if we can't find dri2proto or libdrm
      autoconfig: don't call closedir() when opendir() failed
      kdrive: allow disabling Composite
      xfree86: fix initial output positions with Above or LeftOf and rotation
      xfree86: fix build on GNU/kFreeBSD
      Add xkbstr.h and xkbsrv.h to sdk_HEADERS
      mi: Fix typo in comments about deprecated functions
      kdrive: fix building with --disable-xv (bug#16488)
      x86emu: fix build, fallout from 9719354
      Export xkbfile.h in the SDK
      Bug #16674: Make sure RANDR reports refresh as 0 if pixel clock is 0.
      XF86BigFont is disabled by default
      distcheck fixes
      Don't put xf86Build.h in the tarball (bug#9277)
      Drop some more configure-generated files from the tarball
      xfree86: use xorg.conf input devices if there is no ServerLayout
      xfree86: don't output a spurious newline to the log when loading a module
      Add swapped dispatch for randr 1.2 requests
      Xevie: swap replies as necessary
      exa: disable shared pixmaps
      Don't use our version of ffs() if the system has one
      Bug#5851: Document the "Log" option in xorg.conf(5)
      Xext: Fix compiler warnings
      Nuke unused variables
      xfree86: Fix compiler warnings (implicit declarations)
      xaa: Fix compiler warning (implicit declaration)
      Add prototypes for strlcpy/strlcat
      Nuke prototype for nonexistent function
      Ansify declaration of FontToXError()
      Nuke unused miBankNewSerialNumber
      xfree86: Silence compiler warnings about function prototypes
      xfree86: fix compiler warnings in DoModalias()
      x86emu: fix sparse warnings
      mi: Fix infinite loop on regen when is missing
      xfree86: xf86SetDepthBpp needs to respect the driver's depth24flags

Kazuhiro Inaoka (1):
      Add M32R architecture support (bug #10020)

Keith Packard (49):
      XkbCopyKeymap was mangling doodads and overlays
      Empty the borderClip of manual redirect windows. (bug 15823)
      Rework composite overlay window code to fix several resource management bugs.
      Xi: avoid double button mapping.
      Make button down state a bitmask. Master buttons track union of slave buttons
      dix: reset potential lastSlaves when disabling an SD
      Drop a reference to user mode after create
      Switching from Automatic to Manual redirect needs to unmap/remap
      When resizing a window with redirected descendents, don't expose them
      Revert "Update COPYING for SHA1 code"
      Revert "Render: Use built-in SHA1 library"
      RANDR: Delivery output property events.
      Wrap AddTraps in exa and damage.
      Xinput: use floats for ConstantDeceleration and AdaptiveDeceleration
      Notify DRI when crtc regions change
      Add projective transforms to RandR DIX/DDX API.
      Export a bunch of matrix operations from render.
      Randr now depends on Render for matrices
      Add funcs to convert between protocol and pixman matrices
      Wire up RandR CRTC transform protocol, bump server to RandR 1.3
      [render] Split out filter finding from filter setting.
      [RANDR] Support filters in CRTC transforms.
      Use transform when computing scanout size of modes
      Add kernel size to Render filters.
      Pass filter kernel size through transforms
      Initialize and cleanup new filter fields in xf86Crtc.
      Handle filter widths in xf86Rotate
      Correct bilinear filter kernel size (should be 2x2)
      Actually use filter kernel size to expand transform redisplay box
      Adjust transformed cursor position to account for hotspot
      Add matrix inversion function (uses doubles)
      Compute matrix inversion instead of using wire version in RRCrtcTransformSet
      When converting from double to fixed, round carefully.
      Clear shadow pixmaps before using them.
      Report whether transforms are support from driver through extension to client
      Eliminate inverse matrix from randr transform protocol
      Handle RandR transform matrices in floating point.
      Allow drivers to set crtc transforms.
      Avoid overflow in PictureTransformPoint. Fix PictureTransformIsIdentity.
      Create rrtransform.[ch]. Add RRTransform argument to RRCrtcNotify.
      rrtransform needs randrstr to get RANDR_INTERFACE defines
      Fix rotated/reflected cursor positions.
      Be careful about copying transforms around; they have allocated memory.
      Export rrtransform.h as needed by randrstr.h
      Update RandR global transform when driver notifies of transform change.
      [randr] don't try to compute crtc transform when no mode is set.
      Handle transform failure when computing shadow damage area.
      Move matrix operations from X server to pixman 0.13.2
      New version of dolt

Kim Woelders (1):
      xkb: fix use of uninitialized variable.

Kristian Høgsberg (35):
      Fix DRI2 texture target for GLX_EXT_texture_from_pixmap.
      DRI2: Return event buffer head index in DRI2CreateDrawable.
      GLX: Track changes to DRI_TEX_BUFFER extension.
      DRI2: Add DRI2AuthConnection().
      Make WriteToClient take a const void * like any decent IO write function.
      Silence REGION_INIT() warning.
      Implement DRI2 direct rendering and update AIGLX to DRI interface changes.
      dri2: Unbreak glcore visual setup.
      Add @XORG_CFLAGS@ to satisfy xf86* includes.
      Send the GLX_EXT_texture_from_pixmap attributes to the client.
      Pick up dri2proto from the standard proto header include path.
      Only autoload RECORD if it was enabled.
      Convert __DRIconfigs after we've made sure createNewScreen succeeded.
      Make DRI2 a serverlayout/serverflags option.
      Stop symlinking glcontextmodes.[ch] from mesa.
      Drop glcore GLX provider.
      Use LD_EXPORT_SYMBOLS_FLAG when linking Xnest.
      Get out of the mesa symlinking business.
      Move the windows GL files to the xwin DDX.
      Move GL/glx on level up now that it's the only thing left under GL.
      Clean up remaining afb references.
      Un-K&R shm extension.
      Fix embarrasing GLXPixmap leak.
      Need to unref pixmaps backing pbuffers too.
      Drop unnecessary linked list of contexts from GLXDrawable.
      Drop the glx resize hook and stop chaining PositionWindow.
      Remove SPARC muldiv code.  Was used by ELF loader, which is no more.
      Ugh, remove trailing backslash.
      Drop xf86CommonSpecialKey() and atKeynames.h
      Clean up unused definitions from glx headers.
      Add atKeynames.h in hw/dmx/input and drop evil cross-ddx #include.
      Fix driGetConfigAttribIndex unaligned access to GLboolean.
      DRI2: Drop sarea use, implement server side swap buffers.
      dri2: Update to latest protocol draft.
      CopySubBuffer expects GL style coordinates.

Luc Verhaegen (6):
      DGA: Fix ProcXF86DGASetViewPort for missing support in driver.
      DGA: Mash together xf86dga.c and xf86dga2.c.
      DGA: Track client state even when using old style DGA.
      GLX: fix build when dri2 is not available.
      GLX: fix build when dri2 _is_ available.
      XAA PixmapOps: Sync before accessing unwrapped callbacks.

Maarten Maathuis (18):
      exa: fix assert logic thinko from 361a9eb953aaa38f8ebc057185de29e50f9eef26
      exa: remove useless cache{X,Y}off from UploadToScreen() arguments.
      exa_glyphs: remove useless offset
      exa: move destination damage for internal calls to a special function
      exa: report damage manually for exa{Trapezoids,Triangles} when needed
      exa: some minor cleanup
      exa: fix thinko from 988725f32e082aee9392a71464125157a83d1e67
      {damage,exa}: sanitise damage
      damage: internal functions start with a non-capital letter
      damage: DamageReportRawRegion should set pDamage->damage
      damage: initial attempt at a damage marker mechanism
      damage: choose less ambiguous function names
      xf86Crtc: Set mode to NULL, so nothing bad happens if crtc == NULL.
      exa: remove some excessive whitespace
      exa: remove "direct" case for exa{Trapezoids,Triangles}
      pixmap: Add some notes to devKind and devPrivate.
      dri: don't set the dixPrivate key to NULL, as this is a staticly set variable.
      exa: restore {x,y}{Src,Dst} to their original values when !done

Magnus Vigerlöf (5):
      dix: Ensure Proximity events don't kill the server.
      Make all conversion handling in GPE.
      dix: Correct values in the device axisVal are calculated in GPE
      dix: Correct clipAxis so it can handle devices with value ranges properly
      dix: Cleanup of GetPointerEvents

Mark Kettenis (1):
      OpenBSD support for libpciaccess.

Mathieu Bérard (7):
      Drop a bunch of #ifdef Lynx.
      Fix AddExtension now that CloseDownProc can be NULL.
      The smart scheduler is not optional.
      Remove ancient unused inline hack.
      Remove some unused assembly code and assyntax.h
      Remove dead {bsd,lnx}Resource.c
      Remove xaaTEGlyphBlt.S

Matt Turner (1):
      Check for __x86_64__ when we check for __amd64__

Matthias Hopf (1):
      Return randr interface version in xf86CrtcScreenInit()

Matthieu Herrb (23):
      test for the presence of pci_system_init_dev_mem() before calling it.
      CVE-2008-1377 - RECORD and Security extensions memory corruption
      CVE-2008-1379 - MIT-SHM arbitrary memory read
      CVE-2008-2360 - RENDER Extension heap buffer overflow
      CVE-2008-2361 - RENDER Extension crash
      CVE-2008-2362 - RENDER Extension memory corruption
      sprintf() -> snprintf()
      Let code build with gcc 2.95
      Remove RCS Id.
      preserve errno around the SIGIO handler
      build fix
      Fix ipc check on OpenBSD
      Various messages clean-ups.
      Move string comparaison functions to from dix/ to os/.
      remove OpenBSD/amd64 specific mtrr API.
      add libXinerama to the list of dependencies.
      Move strcasecmp(), strcasencmp() and strcasestr() prototypes to os.h Typo (the the).
      Close well known connections in ServerAbort()
      Support for BSD apm
      Add a man page for Xephyr, with information from the README file.
      Remove the old libusbhid copy from os-support/bsd.
      missing action in if statement.

Michael Vogt (1):
      Do not send VisibilityNotify events when MapUnmapEvents are disabled

Michel Dänzer (36):
      Fix RandR 1.2 driver interface conversion of two colour cursors to ARGB
      Fix RandR 1.2 driver interface conversion of two colour cursors to ARGB
      Fix off-by-one error in ProcXResQueryClients().
      dolt works on powerpc Linux.
      Optimize dixLookupPrivate for repeated lookups of the same private.
      EXA: Offscreen memory eviction improvements.
      EXA: Glyph cache upload tweaks.
      EXA: Accumulate glyphs whenever possible, for full benefits of the glyph cache.
      EXA: Try to accelerate non-antialiased text via the glyph cache as well.
      EXA: Fall back in CompositeRects if the driver can't accelerate Composite.
      EXA: Specify region used for source pixmap migration in exaCopyNtoN.
      EXA: Don't suggest exaCopyDirty be inlined.
      EXA: Don't migrate twice in exaImageGlyphBlt.
      EXA: Only record damage generated by rendering operations.
      EXA: Fix exaGetPixmapFirstPixel() crash if the driver has a CreatePixmap hook.
      AIGLX/DRI1: Switch to server context for calling pScreen->GetImage.
      dix: fix a crash when the client has a different byteorder.
      Fix crack.
      AIGLX/DRI1: Pay more attention to return value from DRIGetDrawableInfo().
      GLX: Unreference drawables bound to the old context, not the new one.
      EXA: Replace open coded CopyArea logic with GC op call.
      EXA: Optimize GXcopy tiled fills.
      EXA: Simplify exaFillRegionTiled() control flow.
      EXA: Fix exponential growth logic for GXcopy tiled fills.
      EXA: Do still return FALSE if the driver PrepareCopy hook failed...
      Remove commented out lines that make automake 1.9 puke.
      EXA: Make sure damage tracking code is inactive if the driver manages pixmaps.
      EXA: Don't use exaGlyphs if the driver doesn't provide a PrepareComposite hook.
      EXA: Inline Prepare/FinishAccessWindow into only caller, ChangeWindowAttributes.
      EXA: Avoid excessive syncing in PutImage
      AIGLX: Allow 2D driver to prevent zero-copy texturing of a pixmap.
      AIGLX: Don't truncate offset returned by driver texOffsetStart hook on 32 bit.
      AIGLX: Reinstate call to driver texOffsetFinish hook.
      Fix typos which caused exaCompositeRects() to use an incorrect damage region.
      Set libdrm flags correctly even when only DRI2 is enabled.
      DRI2: Don't crash in DRI2ClipNotify when DRI2DrawablePtr is NULL.

Mikhail Gusarov (1):
      scrnsaver is optional at build time.

Nathaniel McCallum (1):
      Xorg: add -modalias option

Owen Taylor (7):
      EXA: Use a single large glyph cache pixmap
      EXA: Add exaCompositeRects()
      EXA: Fix overlapping glyphs in glyph cache
      EXA: Clean up debug messages
      EXA: Use UploadToScreen() for uploads to glyph cache
      Bug #15369: Don't reduce Over to Src for transform + RepeatNone.
      Change 'remap' to 'map' in saveset functions/macros

Owen W. Taylor (2):
      EXA: Fix TS origin computation when implementing RenderComposite with tiling.
      Don't remap windows for SaveSetUnmap

Paul Bender (1):
      Bug #15665: Fix building without Composite.

Paulo Cesar Pereira de Andrade (5):
      Fix an off by one read error in drmSIGIOHandler().
      xorgcfg: Fix a crash if xorg.conf doesn't have a Files section.
      Correct a NULL pointer deference
      mi: minor build fix when compiling with debug enabled.
      gl: include assert.h if we're compiling with DEBUG.

Paulo Ricardo Zanoni (3):
      Enable event delivery for multiple heads.
      Allow value None for windows when setting ClientPointer.
      ProcX{Change|Query}WindowAccess: change device list from char* to XID*.

Peter Harris (1):
      Fix panoramiX request and reply swapping

Peter Hutterer (677):
      o fix minor error in comment for GetPointerEvents()
      removed unused variables drag and local from xf86PostMotionEvent
      dix: cleaned up GetPointerEvents() to avoid code duplication.
      dix:    added isMPdev field to _DeviceIntRec, is used in GetPointerEvents()
      mieq:  EQ processing handles MP devices
      dix:    moved isMPdev field to end of _DeviceIntRec structure
      mi:     moved core pointer information from miSpriteScreenRec into a
      mi:     added MPX to miSprite functions.
      mi:     added MPX to miSpriteReportDamage
      mi:     added device ID to miSpriteCursorFuncRec functions
      mi:	constrain MPX cursors to screen
      dix:    moved sprite from static to be pointer of type SpritePtr
      dix:	WindowsRestructured() calls CheckMotion() on all devices
      mi:     added DevToSprite() conversion function
      dix	GetSpriteWindow() adjusted for MPX
      dix:	DeliverEventsToWindow adjusted for MPX
      mi:	closing memory leak, miPointer is freed in miPointerCloseScreen
      mi:     free allocated memory in miDCCloseScreen and in miDCInitialize in case
      mi:     mipointer.c cleanup, reducing code duplication, using MPHasCursor()
      mi:     malloc checks for misprite and midispcur, code cleanup
      mi:     Bugfix: FreeCursor ignores negative refcounts.
      dix:	missing parameter to SetCursorPosition in CheckMotion ifndef MPX
      mpx:    Infrastructure for MPX extension, GetExtensionVersion request works.
      mpx:    SelectEvents and GetEventBase requests.
      Xi:     adding MPX checks to ProcessOtherEvents
      Ironing some glitches caused by the merge
      dix:    Moving SpriteRec into DeviceIntRec
      mi:     removing MPX ifdefs
      mi:     removing MPX ifdefs
      Cleaning up ifdef MPX from remaining files, removing it from
      xfree86: Changing "IsMPDevice" to "SharedPointer" option. Devices will default
      xfree86:  fix xf86ActivateDevice(): only set to MP device if not a keyboard
      removing MPX extension files
      xfree86: fix xf86ProcessCommonOptions() core and shared pointer assignment
      Xi:     Adding ChangeDeviceCursor request
      xfree86:  fix XI86_SHARED_POINTER flag setting
      Xi:     add missing call for SProcXChangeDeviceCursor
      mi:	remove core pointer special handling. No event has core pointer as
      mi:     Quick fix to stop segfault on non-core keyboards.
      o fix minor error in comment for GetPointerEvents()
      mi:     Fix cursor rendering issues.
      Xi:	Adding ChangePointerKeyboardPairing request
      xkb:	Daniel's patch to stop evdev keyboard segfaults.
      dix:    If POINTER_CORE_ONLY is set, GetPointerEvents() only creates a core event
      dix:    Clean up GetPointerEvents. If pDev == inputInfo.pointer, assume that
      dix:    fix typo in condition.
      xfree86: mouse and keyboard hotplugging support. New mouse devices will get
      Xi, dix: Only one client at a time can change the pointer-keyboard pairing,
      Xi: 	Add access control request handling.
      xfree86: Don't initialize sprites for devices that don't own a sprite (stops
      Update device valuators on XTest motion event, otherwise the cursor jumps
      dix:    Remove InitSprite, work can be done by InitializeSprite.
      dix:    Add ClientPointer to client, used for picking which pointer to use in
      dix, Xext, Xtrap, Xi: replace inputInfo.pointer with PickPointer where
      o fix minor error in comment for GetPointerEvents()
      dix:    Add GuessFreePointerDevice(). Runs through device list and tries to
      Xi:    Don't deactivate core grabs from non-core button/key releases.
      dix:    Each device needs to differ between a core grab and an XI grab,
      dix:    Check if client has a grab on any other device before mouse/kbd
      Xi:     Add SetClientPointer request handling.
      dix:    restore commit b3b2a6a0d43d1724e04d69588f8a55c3270e5523 that for some
      dix:    Get the state from the paired device and the real device to fill the
      dix:    fix: don't skip first item in list when seeking for a device cursor.
      kdrive: update kdrive to new mi API with additional device argument. Includes
      dix:    fix: don't return when device is same as argument.
      Xi:     move IsPointerDevice into dix.
      o fix minor error in comment for GetPointerEvents()
      dix:    Move sprite initalization/pairing from ActivateDevice to
      Xi:     add GetClientPointer and GetPointerPairing requests.
      dix:    Increase allocation size for core keyboard keymap to avoid buffer
      Clean up DeviceIntPtr, prepare for Server 1.4 ABI. Move sprite-related
      dix:    Pick keyboard instead of using virtual keyboard in Grab/UngrabKey.
      dix:    Try to get closer to the correct refcnt for cursors.
      dix:    Fix wrong cursor refcount.
      Xi:     fix ProcXListInputDevices to include the virtual core devices.
      o fix minor error in comment for GetPointerEvents()
      Make GrabDevice differ between core grabs and device grabs.
      Revert "o fix minor error in comment for GetPointerEvents()"
      Delete all grabs from all devices when deleting a window.
      Unrealize cursor only once, not per device.
      Allow for multiple animated cursors.
      Synthesize WarpPointer events for the actual device rather than the VCP.
      Fix: pick new ClientPointer when device is closed.
      Adding some comments to devices.c.
      Don't explicitly init sprite for VCP, EnableDevice() will take care of that.
      Close down virtual core devices when closing all devices.
      Alloc sprite memory in devices' devPrivates, allow undisplaying cursors.
      Remove workaround for a NULL core pointer in ProcessInputEvents.
      Change core enter/leave semantics for multiple pointers.
      Use paired kbd instead of VCK for EnterLeaveEvents, sanitize variable naming.
      Deliver FocusIn/FocusOut events for all devices.
      Change enter/leave semantics for events with detail Notify{Nonlinear}Virtual.
      Change FocusIn/Out semantics to match Enter/Leave semantics.
      Change dbus 'listDevices' call to not require an argument.
      config: Return errors as negative numbers, device ids as positive numbers.
      Check for NULL devices in CloseDevice().
      Fix cursor rendering for multi-head.
      Allow events to grabWindows event if the device is not grabbed.
      Add GenericEvent extension to Xext.
      Check VCP's and VCK's ids when searching for a new device id.
      Check and re-set paired devices when initializing sprites.
      Send PointerKeyboardPairingChangedNotifyEvent using XGE.
      Correct enterleave semaphore value for root window.
      Start devices after windows are initialized.
      Adjust EQ to support long events.
      Add missing files for ProcXiSelectEvent.
      Add handling for FakeDeviceEvent request.
      Cleaning up a bit.
      Add selected RamDac functions to xfree86LookupTab.
      Add ExtendedGrabDevice handling.
      Remove double-grab option. We only want one grab per device.
      Add ungrab support to ExtendedGrabDeviceRequest handling.
      Revert "Allow events to grabWindows event if the device is not grabbed."
      Init device axis with -1,-1 minimum values.
      Clean up, correct some comments.
      Squish the flags into the upper bits of the mode field for EnterLeave events.
      Remove unused include "gestr.h"
      Change GrabDevice's parameter name back to "coreGrab".
      Fix up comments for event.c
      Add a deviceMask to the GrabRec and don't interfere with passiveGrabs.
      Zero deviceMask and genericMask when creating a grab via CreateGrab().
      Only decrement buttonsDown when the button count is greater than 0.
      Add implicitGrab field to GrabInfoRec.
      Add a few comments to ProcXGrabDevice.
      Set the detail field for DeviceKeyEvents to the keycode.
      Save processInputProc before wrapping it and restore it later, instead of
      Revert "Save processInputProc before wrapping it and restore it later, instead of"
      Save processInputProc before wrapping it and restore it later, instead of
      Quickfix to stop core key events from doubling up.
      Init genericMasks to 0, make sure list is NULL terminated.
      mieqEnqueue: use modulo for queue tail wrapping.
      Reset grab->next to NULL when we're creating a grab.
      Add 'evfill' field to GEExtensions.
      Use the same struct for generic event masks throughout the code.
      Call CheckMotion for all core events.
      DeliverGrabbedEvent: stop segfault when gemask == NULL.
      Change CheckMotion to ignore non-pointer events but acknowledge XI events.
      Install geext.h, otherwise drivers won't build.
      DeliverEventsToWindow: ensure that genericMask has a defined value.
      mieqEnqueue: Don't try to update the time for GenericEvents.
      dix: check for core event to determine if grab is a core grab (CreateGrab).
      dix: get the paired keyboard for a passive grab (ProcGrabButton).
      dix: Allow flexible devices for passive core grabs.
      Fix typo from last commit. Oh well.
      dix: ProcUngrabKeyboard: make sure grab is initialized.
      dix: ProcChangeActivePointerGrab: make sure variable is initialised.
      dix: Only check device events for possible ACLs.
      randr: RRPointerScreenConfigured needs to move all pointers.
      xkb: Store the action filters per device in the XkbSrvInfoRec.
      Revert "Input: Fix stuck modifiers (bug #11683)"
      dix: don't change the device struct while processing core events.
      dix: close virtual core devices after other devices.
      xkb: enable XI event processing for xkb.
      dix: add XI event support to FixKeyState.
      xfree86: wrap keyboard devices for XKB.
      dix: Enabling devices must not overwrite existing sprites/pairing.
      dix: don't compress motion events from different devices (EventEnqueue)
      dix: Make EnqueueEvent aware of GenericEvents.
      dix: fix comment for DeliverDeviceEvents.
      dix: don't try to access "time" field for GenericEvents in PlayReleasedEvent.
      dix: fix "possible use of uninitialized variable" warning.
      dix: fix compiler warnings about unused variables.
      include: remove register keywords.
      dix: cosmetic change. Remove "pointer" variable, might as well use pDev.
      dix: add comments to ComputeFreezes() and syncEvents struct.
      dix: add comments to PlayReleasedEvents()
      Wrap core event handling through ProcessOtherEvents.
      Xi: unify ErrorFs. Prefix all with [Xi].
      dix: GetPointerEvents: remove unused variable compiler warning.
      dix: unify ErrorFs. prepend all with [dix].
      xkb: unify ErrorFs. Prefix all with [xkb].
      xkb: XkbFilterEvents: Remove unused variable compiler warning.
      xfree86: pass in the XKB event processing proc instead of the XI one.
      Xi: fix the wrapper code for processInputProc wrapping.
      xkb: Unwrap properly in ProcessPointerEvent.
      xkb: xkbHandleActions: let wrapping take care of event delivery.
      dix: GetMaximumEventsNum() needs to return 3 + MAX_VALUATOR_EVENTS.
      dix: take paired keyboard instead of VCK (CoreProcessPointerEvents)
      dix: GetPairedKeyboard() always returns a valid keyboard (VCK if necessary).
      Remove generation of core events, long live XI!
      Xi, xfree86: Remove leftovers of the XI wrapper code.
      xfree86: update all pointers when calling xf86SwitchMode
      xfree86: xf86DGA: some cleanup to use the correct devices instead of VCP/VCK.
      xfree86: fix DGA to use the correct devices.
      mi: switch core keyboard on XI events, not only on core events.
      dix: Ignore devices when deleting passive core grabs.
      dix: change Enter/Leave semaphore handling to accommodate for NotifyGrab.
      dix: send NotifyGrab/NotifyUngrab focus events regardless of semaphore state.
      dix: ignore passive grab if the client already has a grab on the device.
      xkb: remove some warning comments.
      Xext: add missing geint.h to (un-breaks make dist)
      dix: When pairing, only check for access right if client is set.
      dix: when disabling a device, make sure all paired devices are re-paired.
      dix: change a few inputInfo.keyboard over to PickKeyboard()
      include: add "master" field to DeviceIntRec.
      dix: Set core devices to ProcessOtherEvents, set sendCore and XKB.
      dix: add AttachDevice, needed to attach a slave device to a master device.
      include: add "isMaster" field to DeviceIntRec.
      dix: fix up Activate/Enable/Disable device.
      dix: Set bidirectional pairing, don't allow pairing with already paired devs.
      dix: remove pairing/attachment from InitAndStartDevices.
      dix: Remove GetPairedPointer, GetPairedKeyboard in favour of GetPairedDevice.
      dix: don't check VCK and VCP separately when looking for device ids.
      dix: check for isMaster before calling DeviceCursorCleanup.
      dix: don't call CloseDevice on VCP/VCK separately.
      dix: don't allow slave devices as ClientPointer
      dix: adjust PickPointer and PickKeyboard to get the first master device.
      include: remove unused GetPairedPointer/Keyboard declarations.
      dix: GetPairedDevice: return dev paired with master for slave devices.
      Fix up detritus from removing GetPairedPointer/Keyboard.
      dix: don't treat VCK/VCP separately in DevHasCursor, IsPointer/KeyboardDevice.
      xfree86: remove XI86_SHARED_POINTER flag.
      dix: set the device's sprite when using AttachDevice
      include: add "lastSlave" field to DeviceIntRec.
      dix: GetPointerEvents: get state from master device before applying valuators.
      dix: fix detritus from adding lastSlave field.
      dix: CheckPassiveGrabOnWindow: only get paired device for non-keyboards.
      mi: don't exclude inputInfo.pointer from rendering.
      xfree86: NIDR: don't call PairDevices explicitly.
      dix: add AllocMasterDevice for creation of new master devices.
      dix: Make InitCoreDevices use AllocMasterDevice.
      mi: fix some macros to allow multiple cursors for master devices.
      Xi: don't send core events for slave devices.
      mi: call processInputProc for master devices after slave event processing.
      mi: don't call UndisplayCursor for non-sprite-owners.
      Xi: return all master devices as type IsXPointer/Keyboard when listing devs.
      Xi: set master device's id in ListDevices Reply.
      Xi: remove ChangePointerKeyboardPairing in favour of ChangeDeviceHierarchy.
      Xi: advertise as XInput v2 capable
      Xi: allow VCP/VCK to be OpenDevice'd as well.
      dix: Create new sprite for floating devices.
      dix: check for core passive grabs on inferiors when replaying events.
      dix: Don't interfere grabs the interfering device is of different type.
      mi: duplicate event before processing it, so master has original values
      Xi: don't call CheckMotion for slave devices.
      Xi: return BadDevice for slave devices on ProcXSetClientPointer
      mi: change the device id to the master devices' when duplicating the event.
      dix, mi: stop segfaults when a floating device emits events.
      dix: calloc root window's Generic Event mask, stops segfaults.
      Xi: notify the clients when the device hierarchy has been changed.
      dix: grabbing an attached SD sets it floating for the duration of the grab.
      Xext: add simple macro to easily check a generic event's type.
      Xi, dix: Add ability to change MD classes + send event when doing so.
      dix: When the last slave is removed, set master to the original classes.
      mi: avoid SIGABRT by setting master_event to NULL.
      dix: allow grab modifier device to be NULL.
      dix: don't unconditionally update valuators during key events.
      xkb: disable xkb key repeats (temporarily)
      dix: Send MappingNotify when keyboard maps change.
      dix: Attach spriteInfoRec to same memory block as the DeviceIntRec.
      dix: remove trailing/wrong whitespaces from devices.c and events.c
      Xi: remove trailing whitespaces.
      Xi: remove RegisterPairingClient handling. Deprecated with device hierarchy.
      Xi: Move updating the device state from POE into separate function.
      dix: Return Success from ProcBell if no BellProc is set.
      Xi: toggle the public.devPrivate as well when switching device classes.
      Xi: When switching MD classes, make a deep copy instead of pointer flip.
      mi: don't call SwitchCoreKeyboard, we switch during event processing now.
      dix: Make sure core MDs copy their initial classes before they are used.
      Xi: free XkbSrvLedInfos as well when freeing an MD's device classes.
      Revert "xkb: disable xkb key repeats (temporarily)"
      Xi: reset xkb-stuff to NULL after copying (DeepCopyDeviceClasses)
      dix: Add FreeDeviceClass and FreeFeedbackClass for centralised xfree.
      dix: SetKeySymMap should alloc the map if dst doesn't have one already.
      Xi: fix up sloppy class copying causing segfaults.
      Xi: Only alloc modifierKeyMap if <slave device>->maxKeysPerModifier > 0
      dix: Fix up class restoring when last SD disconnects.
      dix: explicitly float all attached SDs before closing down devices.
      Xi: Deep-copy full list of FeedbackClasses.
      dix: Free both current classes and original classes when closing an MD.
      dix: reset MD's devPrivate classes to NULL before device initialisation.
      Xi: add missing XI_DeviceClassesChangedMask to XI filters array.
      dix: getevents.c remove trailing whitespaces.
      Xext: Scruffy the janitor don't like no "register" keywords.
      xtest: switch an inputInfo.pointer over to PickPointer.
      Xi: allow clients to specify pure client id in SetClientPointer.
      DeviceIntRec: move lastx/lasty from valuator into DeviceIntRec.
      Xi: allocate motion history for master device.
      dix: Add special treatment of NotifyUngrab for enter/leave events.
      dix: fix typo
      mi: only call UpdateSpriteForScreen if the screen actually changed.
      Xi: set DeviceXXXState's length fields to the correct size of the struct.
      xkb: swap a LookupKeyboardDevice over in favour of GetPairedDevice
      dix: only freeze the paired MD on a grab, not all other devices.
      dix: avoid activating the VCP/VCK twice.
      dix: Remove RegisterPairingClient and UnregisterPairingClient.
      ephyr: miPointerWarpCursor needs the device parameter.
      dix: remove #ifdef XINPUT.
      dix: comments, whitespaces, copyright fixes.
      dix: change the filters to be per-device.
      dix: update comments about inputInfo.pointer.
      mi: change infamous "Tossed event ..." error for something more explanatory.
      mi: unify last two ErrorF's in mieq.c
      mi: unify ErrorFs, prepend [mi]
      dix: when getting a bogus event, at least print out the type.
      dix: set the correct number of valuators in valuator events.
      include: never overwrite realInputProc with enqueueInputProc. Bug #13511
      dix: free ClassesRec when a BadAlloc occurs during CoreKeyboardProc.
      Xi: switch to using "rc" (instead of "err") as returncode.
      xfixes: switch a few inputInfo.pointer to PickPointer
      Xi: ProcXSetClientPointer should directly check for the client from the id.
      dix: DoChangeKeyboardControl shouldn't be using inputInfo.keyboard.
      dix: don't free device's devPrivates manually, dixFreePrivates does it.
      Xi: remove file for the obsolete ChangePointerKeyboardPairing request handling.
      dix: init GrabPrivateKey to correct value.
      dix: don't try to confine cursor on screen restructure if there is no cursor.
      Xi: keep a counter of buttons down to avoid duplicate press/release events.
      dix: remove obsolete comment.
      dix: Emulate core events within CheckDeviceGrabs and ComputeFreezes.
      dix: Process an input event as a single event, instead of two separate ones.
      Xi: remove some leftovers from ProcessOtherEvents.
      xkb: don't do core key repeats, XI only is the path to light.
      dix: print out an error when core type can't be converted.
      Xi: Check for DestroyAccess when trying to delete a master device.
      dix: remove inputInfo.pointer reference in TryClientEvents.
      dix: remove inputInfo.pointer refernce in EventSelectForWindow
      include: mark CoreProcessPointerEvent, CoreProcessKeyboardEvent as deprecated.
      dix: free the genericMasks when ungrabbing.
      dix: remove some superfluous lines in ProcChangeActivePointerGrab.
      dix: if a client has a core grab, return the grab device on QueryPointer.
      Xi: add XACE hooks for device creation (ChangeDeviceHierarchy)
      Xext: allow per-device GE masks.
      Xi: Change window access right check to RecieveAccess in XiSelectEvent.
      Xext: add GEDeviceMaskIsSet (GE mask checking per device)
      Xi: if deviceid is > 127 assume GE mask is for all devices (XiSelectEvent).
      config: add a debug message, fix a whitespace error.
      config: only shutdown libhal if the connection is valid.
      config: check connection != NULL before getting dbus' dispatch status.
      config: don't reset connection info on disconnect.
      xkb: when copying sections, make sure num_rows is set too.
      Xi: finish XKB initialisation before copying the keymap.
      xkb: when copying the keymap, make sure the structs default to 0/NULL.
      dix: Don't generate RawDeviceEvents anymore.
      mi: remove RawDeviceEvent's ID changing.
      Xi: Remove files for GrabAccessControl.
      Xi: remove raw device events.
      Xi: remove FakeDeviceData calls. And one leftover from GrabAccessControl.
      Xi: QueryDevicePointer doesn't have a shared field anymore.
      dix: set num_events to 1 by default. (no RawDeviceEvents anymore)
      Xi: even if ChangeDeviceHierarchy fails, we may need to send an event.
      dix: add support for WindowAccessAllowAll in device access checking.
      Xi: make SizeDeviceInfo public and re-use from CreateClassesChangedEvent.
      Move input event list initialisation and storage from DDX to DIX.
      dix: add InputEventListLen and SetMinimumEventSize
      dix: change GetEventList to return length of list and set parameter in place.
      dix: Ensure enough memory for ClassesChangedEvent for a new device.
      dix: set evlen to the size of the reallocated memory.
      Xi: remove ungrab handling of ExtendedUngrabDevice request.
      Xi: remove GetPairedPointer handling.
      Xi: MDs return the paired device's ID in attached field of ListInputDevices.
      dix: before copying the classes the first time, set the VCK's classes to NULL.
      Xi: add deviceid to QueryDevicePointer request.
      Xi: remove false memset from DeepCopyDeviceClasses.
      Xi: don't free device classes before copying them.
      Xi: only alloc memory for feedback classes if it doesn't already exist.
      Xi: remove duplicate copying of kbdfeed and ptrfeed.
      Xi: ALLOC_COPY_CLASS_IF should only alloc if to->field doesn't exist.
      Xi: realloc to->valuator instead of allocing it blindly.
      Xi: add comments for DeepCopyDeviceClasses, ChangeMasterDeviceClasses.
      dix: remove debug error message about XI->core type conversion.
      dix: sprite may be NULL, don't dereference it then.
      xkb: don't overwrite CtrlProc in the second run of XkbFinishDeviceInit.
      xkb: don't overwrite CtrlProc in the second run of XkbFinishDeviceInit.
      Xi: check if source has a key class before copying.
      Xi: plug memory leak, free previous motion history before allocating new.
      Xi: If device "to" has a class but "from" doesn't, free the class in "to".
      Xi: When attaching, check for ptr -> ptr and keybd -> keybd.
      dix: When floating, set sprite to NULL before calling InitializeSprite.
      Xi: some extra checks for validity of kbd and mouse.
      xfree86: Sanity check before retrieving the paired device.
      dix: Dont deliver grabbed pointer events to a focus window.
      dix: Extend IsKeyboardDevice() to not include pointer devices.
      Xi: dont copy FocusClassRec if the master already has one.
      Xi: only DeliverFocusedEvents if the event is not a pointer event.
      dix: Rework Enter/Leave semaphore system.
      dix: don't free MDs classes on init.
      dix: Call DeleteInputDeviceRequest from CloseDownDevices (#14418)
      xkb: Add XkbCopySrvLedInfo, deep-copies a XkbSrvLedInfoRec.
      Xi: copy the XkbSrvLedInfo too when copying device classes.
      Xi: modifierKeyMap needs to be set to NULL when copying classes.
      Xi: copy feedback classes first, in some cases xkb relies on kbdfeed.
      xkb: two fixes to avoid server crashes.
      Xi: fix up modifierKeyMap copying.
      dix: Ignore focus for passive grabs if the event is a pointer event.
      Xi: Fix pointer handling in KeyClassRec copy.
      Xi: copy the over from the source.
      Xi: store unused classes in devPrivates.
      Xi: store feedback classes in devProviates system as well.
      Xi: fix up button count.
      dix: float attached devices _before_ disabling the master.
      dix: free the unused device classes when closing a device.
      Xi: when copying button classes, copy xkb_acts as well.
      xkb: mixing up src and dst in a memcpy is suboptimal.
      Xext: xtest.c: death to tabs+spaces indendation, remove #define XINPUT
      Xext: Let XTestFakeInput update the sprite for XI events.
      dix: remove coreOnly check. Core pointer must generate XI events now.
      xnest: re-enable XKB support. #10015
      Xext: Update sprite on fake input only for MDs and floating SDs.
      mi: fix typo in comment.
      Xext: route event through master if required (XTestFakeInput)
      dix: always send FocusIn events if mode is NotifyUngrab.
      xfree86: pass correct value to mieqEnqueue (merge detritus I guess)
      Xext: remove trailing whitespaces in geext.{c|h}
      Xext: re-do the XGE versioning handling.
      Xext: add a few lines of comments to XGE.
      Xext: XGE: change function definition to have return value on separate line.
      xfree86: Don't free master devices.
      xfree86: don't free the config-file related information in DIDR. #15645
      dix: NULL out WindowTable after freeing all the windows.
      Xi: don't attempt to send to a NULL window.
      Xi: remove unused #define DISPATCH
      Xi: handle requests through a dispatch handler.
      Xi: fix two compiler warnings
      dix: remove pairingClient definition.
      dix: remove "unused variable" compiler warning.
      dix: add mi.h include to stop compiler warning
      dix: remove obsolete InitWindowPrivates().
      Xi: remove leftover chpkpair.h file.
      Standardise copyright headers for all new files created as part of MPX.
      dix: let PickPointer() return devices that are core-grabbed by the client.
      dix: remove misleading (stale) comment
      dix: don't search for grabbed devices, PickPointer does it for us now.
      Xi: remove some superflous code
      Xi: remove RemoveOtherCoreGrabs()
      Remove all traces of #ifdef XINPUT and the matching bits from the
      Xi: add versioning support.
      Xi: disable XI for clients < XI 2.0
      Xi: if a pre-XI2 client tries to list the devices, pretend we don't have any.
      Revert "Xi: disable XI for clients < XI 2.0"
      Xi: whoops, stray ! caused a bit of memory mess.
      Xi: Only return VCP, VCK and floating SDs to Xi 1.x clients.
      mi: guard against NULL-pointer dereference.
      dix: if alloc of a master keyboard fails, remove the master pointer.
      dix: store all valuators as they come from the device in the valuator events.
      Xi: apply ValuatorEvent data to the device's valuators.
      Xi: change UpdateDeviceState to be _X_EXPORT.
      Xi: add IEventBase global variable. Stores event base for X Input events.
      xfree86: DGA needs to call into Xi.
      mi: call the custom event handlers for both MD and SD.
      include: add declaration for CopyGetMasterEvent to shut up compiler warning.
      Xi: stop excessive use of _X_EXPORT.
      dix: fix typo in clipAxis.
      input: replace -1 as default axis limit with NO_AXIS_LIMIT define.
      dix: fix typo from last commit.
      dix: resize EQ to minimum size to avoid reallocs during SIGIO.
      dix: continue to check for invalid valuator ranges.
      xfree86: don't try to repaint the cursor before cursors exist.
      xfree86: re-enable hardware cursor (for a single cursor)
      xfree86: switch between to SW cursors if more than 1 sprite is available.
      xfree86: call SetCursor for the VCP when switching between SW/HW rendering.
      config: deprecate (and ignore) the use of input.xkb.whatever options.
      Revert "config: deprecate (and ignore) the use of input.xkb.whatever options."
      config: remove trailing whitespaces.
      config: override xkb_{r,m,l,v} with Xkb{r,m,l,v} if the latter is set.
      config: remove trailing whitespaces.
      config: override xkb_{r,m,l,v} with Xkb{r,m,l,v} if the latter is set.
      render: don't dereference cursor if cursor is NULL.
      mi: don't try to display a cursor for devices w/o cursors.
      Remove UndisplayCursor API.
      Xi: Fix up ProcIVector, got out of sync with the protocol.
      xfree86: remove Ubuntu-ism.
      Xext: core button events have rootX/Y unset too.
      Xext: emulate XI events from core events when faking input.
      mi: protect against NULL-pointer dereference.
      dix: don't use serverClient in AddInputDevice, use the one supplied instead.
      dix: InitAndStartDevices doesn't need a window parameter.
      dmx: fix build.
      dix: don't undisplay cursor if we don't own a sprite.
      Xi: assemble button/modifier state before updating the device. #15934
      Xext: fix typo in condition.
      mi: remove miPointerUpdate() -> replaced by miPointerUpdateSprite().
      xfree86: always render first cursor in HW, all others in SW.
      mi: don't call from miPointerMoved into the spriteFuncs for anything but VCP.
      mi: remove miPointerUpdate() -> replaced by miPointerUpdateSprite().
      Xext: fix typo in condition.
      fb: shut up two compiler warnings.
      mi: sync declaration and definition of mieqResizeEvents.
      dix: shut up two compiler warnings.
      xkb: remove superfluous checks in if statement.
      mi: shut up compiler warning (return w/o a value)
      Replace UniSA with full Uni name in Author affiliation.
      xkb: fix crash caused by uninitialised variable.
      xfree86: suspend signals while removing a device.
      mi: handle sprite even for floating slave devices.
      dix: when floating a device, create a new cursor sprite.
      dix: fill valuators with the correct values depending on the device mode (GPE)
      dix: coreEvents doesn't have meaning here, use master instead.
      input: instead of lastx/y, use a last.valuators[] array on the device.
      dix: more device scaling insanity.
      Revert "dix: Correct values in the device axisVal are calculated in GPE"
      dix: don't allow relative data in deviceValuators.
      Xi: update device valuators based on the event.
      dix: update valuators > 2 as well.
      Xi: remove deprecated comment.
      Xi: reset motion history when swapping device classes.
      dmx: remove some #if 0'd code.
      Remove GetMotionProc from ValuatorClassRec.
      dix: Allocate MD's motion history to be large enough for potential valuators
      Ensure the motion history is merged for master devices.
      If core motion history is required, scale back to screen coords and INT16.
      dix: store the absolute values in the motion history.
      xfree86: Bump XINPUT ABI to 3.0.
      xfree86: fix offset error in EDID parsing. #15964
      mi: ansify.
      Require x11proto 7.0.13, xextproto 7.0.4 and inputproto
      Xext: only free events in case of core motion events.
      dix: null out dummyDev, otherwise Xephyr dereferences random pointers.
      mi: reduce noisyness. DebugF is enough if device ID can't be changed.
      Xext: use GPE/GKE from XTestFakeInput #16145
      dix: Remove superfluous comment.
      dix: ProcQueryPointer doesn't need special grab handling, PickPointer does it
      mi: Remove #ifdef XINPUT, kdrive otherwise won't get events.
      kdrive: fix up callers to KdQueueEvent, need to pass xEvent, not EventListPtr
      dix: don't read valuators past numAxes.
      config: init dev to NULL to shut up a valgrind warning.
      xkb: delete default rules when devices are closed.
      Revert "xfree86: suspend signals while removing a device."
      xfree86: suspend signals while removing a device (corrected version).
      kdrive: don't post motion event if there was no motion. #16179
      dix: Fix build with --disable-xinerama #16204
      xkb: reset xkb_cached_map on CloseDownDevices.
      dix: set dst->mapWidth when allocating a new map.
      dmx: some more build fixes.
      dmx: fix a segfault caused by GC devPrivates never being initalised.
      dmx: claim we support XI 2.
      dmx: learn about the existence of IsXExtensionPointer and IsXExtensionKeyboard
      dmx: for now, don't acknowledge SDs.
      dmx: add stubs for DeviceCursorInitialize, DeviceCursorCleanup
      dmx: don't dereference a nullpointer.
      dmx: fix false memory allocation.
      dmx: add an XFlush.
      dmx: don't free event list after use.
      dmx: remove a ifndef XINPUT, XINPUT is always defined now.
      mi: protect against possible NULL-pointer dereference.
      ephyr: some whitespace changes.
      mi: remove leftover old license comment.
      xnest: call miDCInitialize rather than miPointerInitialize.
      xnest: switch to using EventList rather than xEvents.
      xnest: fix up sprite funcs to stop compiler warnings.
      xfree86: Fix up xf86ScaleAxis
      mi: protect mipointrst.h against multiple includes.
      xnest: wrap the xnest cursor sprite funcs around the mi funcs.
      dix: Call RealizeCursor during InitializeSprite.
      Xext: init firstValuator to zero, otherwise core XTest events may get lost.
      kdrive: fix wrong use of &, should be | instead.
      randr: remove superfluous check against uninitialised variable.
      mi: remove trailing whitespaces from misprite.c
      Xi: don't drop the grab status in XExtendedGrabDevice().
      input: fix up usage of button->down, used to be a bitmask, is now an array.
      dix: don't attempt to send Enter/Leave events if we have a keyboard.
      xkb: remove unused variable is_core.
      xfree86: fix up leftover allocation of xf86Events.
      dix: InitEventList does not need to be _X_EXPORT.
      dix: free memory associated with the EventList. #16507
      dix: don't dereference dev->key if it doesn't exist.
      input: for non-device events (e.g. DevicePresence) use MAX_DEVICES as id.
      xfree86: move declaration of configured_device to start of function.
      xfree86: handle missing Screen lines in the ServerLayout #16301
      xfree86: AllowEmptyInput is now enabled by default if hotplugging is enabled.
      dix: free GenericMask allocations when destroying window.
      Xext: store the GenericMasks in the resource system.
      Xi: add support for input device properties.
      Revert "Xi: add support for input device properties."
      Xi: add support for input device properties.
      Xi: pack the property stuff into a struct, make handlers a linked list.
      Xi: GetDeviceProperty reply includes deviceid.
      Xi: remove ChangeDeviceControl for CoreCtl.
      Xi: protect against NULL handlers, don't try to dereference.
      Xi: expose Enable/DisableDevice through XI_PROP_ENABLED property.
      xfree86: append, not prepent, new input devices to xf86InputDevs.
      dix: shift the duplicate button mapping check to ProcSetPointerMapping.
      xfree86: plug memory leak, free driver's private data when deleting the device.
      dix: fix up enter/leave system once again.
      xkb: don't send core events on SlowKeys.
      xfree86: if AllowEmptyInput is on, warn the user that we rely on HAL now.
      xfree86: warn some more about potential missing input devices.
      xkb: use PickPointer/PickKeyboard in _XkbLookupAnyDevice.
      xkb: remove superfluous inputInfo.keyboard treatment.
      config: note that HAL options must be strings.
      Xi: ChangeDeviceControl presence events should set the appropriate devchange.
      config: add parsing for input.x11_options.XkbOptions. #16874
      config: support type strlist for XkbOptions property.
      config: protect against potential out-of-bounds indexing.
      xkb: break up XkbCopyKeymap into bite-sized chunks.
      xkb: ProcXkbBell should work on all attached SDs.
      xkb: ProcXkbSetMap should work on all attached SDs.
      xkb: ProcXkbSetCompatMap should work on all attached SDs.
      xkb: ProcXkbSetIndicatorMap should work on all attached SDs.
      xkb: ProcXkbSetNamedIndicator should work on all attached SDs.
      xkb: ProcXkbSetNames should work on all attached SDs.
      xkb: ProcXkbSetGeometry should work on all attached SDs.
      xkb: ProcXkbSetDeviceInfo should work on all attached SDs.
      dix: SetModifierMapping should only apply to the ClientPointer and it's SDs.
      dix: remove obsolete FIXME comment.
      xkb: actually initialise sli before using it.
      dix: don't try to create events if we don't have a screen. #16898
      dix: remove superfluous check.
      Xi: return the state from the correct keyboard in QueryDevicePointer.
      Xi: byte-swap device property requests.
      Backport device properties to XI 1.5 instead of XI 2.0
      dix: stick a warning in about loss of subpixel precision.
      Require inputproto
      Xi: swap devices property replies.
      Xi: don't include .c files.
      Xi: remove broken SDeviceEnterNotifyEvent, replace with ...Leave...
      dix: calloc, not malloc, ValuatorClassRec.
      Xi: don't include client-side headers.
      xkb: fix core keyboard map generation. #14373
      xkb: squash canonical types into explicit ones on core reconstruction.
      config: print error code if NIDR fails.
      xfixes: realize the invisible cursor before displaying it.
      Xi: remove configure/query device property calls.
      Xi: allow Set/GetProperties to return a status, and honour this status code.
      Xi: add "deletable" flag to properties, add DeleteProperty handler.
      Push server-known properties into xserver-properties.h.
      Require inputproto
      xfree86: don't let DGA steal key events outside of [9,255].
      dix: don't lose subpixel precision when scaling.
      dix: fix axis scaling.
      xfree86: cache dev->isMaster before freeing the device.
      xfree86: export NewInputDeviceRequest/DeleteInputDeviceRequest
      dix: silence "unused variable" compiler warning.
      kdrive: lastx/y needs to be set to the event coordinates #17728
      kdrive: fix dixflags so they're always defined before enqueuing events #17734
      Xi: don't memcpy the KeyClassRec from SD to MD. #16167
      dix: remove duplicate code in ReleaseActiveGrabs
      Xi: check all handlers before applying property changes.
      xfree86: if AllowEmptyInput is true, enable RAW mode on the console.
      dmx: change remainder of manual EQ allocation to GetEventList().
      dmx: avoid hilarious #define Xcalloc xcalloc game.
      xfree86: AllowEmptyInput is true by default - update the xf86Info defaults.
      dix: init remaining three xkb fields to zero in a new client.
      dix: fix endianess issue in AddInputDevice. #18111
      dix: don't allow VCP/VCK be disabled through properties.
      dix: fix two more endian issues, correct initial "enabled" value. #18111
      config: don't add duplicate devices through HAL.
      xfree86: If AEI is on, disable "kbd" and "mouse" devices.
      xfree86: fix xf86ScaleAxis once again.
      xkb: ProcXkbSetCompatMap should do dry-runs, then normal runs.
      Move MAX_DEVICES to misc.h, rename to MAXDEVICES for consistency.
      dix: PairDevices() can be static, remove from input.
      dix: SwitchCorePointer is long gone, bury it.
      dix: NextFreePointerDevice() can be static.
      dix: remove unused GuessFreePointerDevice().
      include: remove unused declaration for RegisterPairingClient().
      dix: FreeDeviceClass() and friends can be static.
      xkb: remove unused label "out", clean up program flow.
      xnest: I'm sure xfree works fine, even if we're 64 bit.
      dix: clean up GetPointerEvents and GetKeyboardValuatorEvents.
      xkb: when faking mouse button events, fake them on the correct devices.
      Purge device-based WindowAccess code.
      xkb: extract the correct device in XkbFilterEvents.
      mi: clean up mieqProcessInputEvents, copy all events before processing.
      dix: remove superfluous variable devgrab
      dix: formatting fix.
      dix: remove duplicate line in EnterLeaveEvent.
      dix: don't store enter/leave and focus semaphores in a devPrivate.
      xfree86: without CONFIG_HAL, Auto{Add|Enable}Devices and AEI is false.
      gitignore: ignore .patch files
      dix: memset DeviceVelocityPtr to zero.
      dix: fix false comment.
      dix: Enable core devices in InitCoreDevices already.
      xfree86: Split the working code of NIDR into new xf86NewInputDevice.
      DGA: silence compiler warning.
      dix: move Enter-Leave related functions into new enterleave.c
      dix: remove unused EnterLeaveSemaphoresIsset.
      dix: split enter/leave event handling into core and device handling.
      dix: Add EnterWindow, LeaveWindow, HasPointer auxiliary functions.
      dix: add FirstPointerChild, FirstPointerAncestor auxiliary functions.
      dix: updated enter/leave core event model.
      include: remove ENTER_LEAVE_SEMAPHORE macros.
      Revert "dix: Enable core devices in InitCoreDevices already."
      dix: Enable core devices in InitCoreDevices already.

Pierre Willenbrock (1):
      RANDR: Fix output property event delivery.

Pierre-Loup A. Griffais (1):
      Don't return BadAlloc when trying to set a PictureFilter with no parameters when a filter with parameters was previously set.

Roland Bär (1):
      Bug #11857: Avoid running off the end of a MAXCLIENTS-sized array.

Roland Scheidegger (2):
      glx: copy msaa visual capabilities
      dri2: fix compilation (remove no longer existing include file)

Ross Burton (7):
      Fix build when XKB is disabled.
      Update to tslib 1.0
      Add a default touchscreen path
      Add support for AVR32
      Enable the epson kdrive driver
      Add mediumraw support to the linux kdrive driver
      Reuse the existing framebuffer mode in kdrive/fbdev

Rémi Cardona (3):
      Force LC_ALL=C when building xf86DefModeSet.c
      drop unused HAVE_LIBDRM_2_2
      xfree86: don't reset Auto(Add|Enable)Devices, use defaults from xf86Globals

Sascha Hlusiak (3):
      Fix getValuatorEvents to compute number of valuators correctly.
      Support to pass arbitrary options via HAL hotplugging
      Remove stale code

Simon Thum (18):
      input: change dxremaind, dyremaind to floats for better acceleration #9156
      Add support for multiple pointer acceleration schemes. #8583
      dix: add legal statements to ptrvelo.{c|h}
      dix: rename classic accel _scheme_ to lightweight to avoid confusion with
      dix: introduce defines for accel profile numbers
      dix: improve the driver interface to predictable pointer acceleration
      dix: optimize precision in device velocity estimation
      dix: preparation change: make runtime exchange of filter chain safe
      xfree86: perform pointer accel scheme selection by name, not number
      some trivial changes regarding C compat, indentation, etc.
      dix: use average of pointer accel profile
      dix: export driver-side functions for acceleration
      dix: simplify velocity tracking filters
      xf86: don't replicate dix defaults for pointer acceleration
      dix: simplified debugging helper for pointer acceleration
      dix: remove misleading comment in ptrvelo.c
      dix: fix non-initialization of pointer accel in master devices.
      dix: fix typos in comments, one formatting fix.

Stefan Dirsch (1):
      Added '-showopts' option to print available driver options (#5564).

Søren Sandmann (1):
      Give priority to clients that are receiving expose and configure events

Thomas Bodzar (1): typo (the the).

Thomas Jaeger (1):
      XKB: Fix processInputProc wrapping

Thorvald Natvig (1):
      Xevie: always initialize rep.length (bug#17394)

Tiago Vignatti (24):
      Remove redundancy.
      Restructure and organize the code.
      kdrive: changes ephyr API according mpx merge.
      Remove unused xf86DeviceCursorUndisplay() function.
      Remove more unused things.
      CopyGetMasterEvent() can be static.
      No need for inputPending field in xf86Info.
      Welcome to hotplug era.
      xfree86: kill INHERIT_LOCK_STATE macro.
      Set --enable-builtin-fonts as default.
      Remove inexistent function headers.
      Nuke libc wrapper remaining stuffs.
      Update README.OS-lib to reflect reality.
      Remove deprecated function miPointerPosition().
      Fix typo in the last commit.
      Remove unused function and variables.
      xfree86: fix wrong function name message errors.
      xfree86: remove xf86{Enable, Disable}GeneralHandler and others unused.
      xfree86: xf86{Enable, Disable}InputHandler can be static.
      Everyone knows that the loader is present. Silence it.
      Remove unused declarations.
      mi: remove deprecated and unused functions.
      mi: functions here are carrying 'time' without need.

Tilman Sauerbeck (7):
      Fixed for autoconf 2.62.
      EXA: Teach exaCompositeFallbackPictDesc() about x8r8g8b8.
      EXA: Update pixmaps' accel_blocked field in ModifyPixmapHeader.
      EXA: Avoid some fallbacks in exaCopyNtoN.
      Removed fbpseudocolor.h from sdk_HEADERS.
      EXA: Update sys_pitch/fb_pitch in exaModifyPixmapHeader.
      EXA: Set pixmap->accel_blocked on the screen pixmap, too.

Tomas Carnecky (5):
      OutputPropertyNotifyMask is a valid thing to select for.
      Cosmetic cleanup to Xvfb/Xdmx configure check output.
      Redefine WindowTable as a fixed array
      Redefine clients as a fixed array
      Prepare for array-index based devPrivates.

Xavier Bestel (1):
      Cosmetic fix to EDID decoding.

Xiang, Haihao (1):
      dri2: fix crasher if DRI2Connect fails

Zhenyu Wang (4):
      randr: fix user mode create initial reference count
      Revert "randr: fix user mode create initial reference count"
      Check nextEnabledOutput()'s return in bestModeForAspect()
      Don't leak EDID_block

root (1):
      Revert "XInput ABI is 2.1, not 3.1."

git tag: xorg-server-
MD5: c1b7f11700a6ac7d3af51343c8ebf8a6  xorg-server-
SHA1: 68b23c8f4be03382b0a05813b25a161b412336ed  xorg-server-
MD5: 37ad31c700c04de488e9fc4f5afb0fac  xorg-server-
SHA1: 455d310450afab72b5947051b95a5fc70a1baae4  xorg-server-

keith.packard at
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : 

More information about the xorg-announce mailing list