Well, it's been a while since we've done a snapshot, so why not? The highlights are fixing 44 Coverity bugs and 30 Bugzilla bugs, a few crash fixes (Render and appgroup), and pruning of a whole bunch of dead code, including all the old loaders. EXA has seen significant performance action, and GNU/kFreeBSD and DragonFly systems should both be a lot happier. ACPI and APM support has also been fixed. CVS tag: xserver-1_1_99_2 http://xorg.freedesktop.org/archive/development/individual/xserver/xorg-server-1.1.99.2.tar.gz: MD5: b01e06843362a4c6f740b7a42c67656c xorg-server-1.1.99.2.tar.gz SHA1: fcaa4a885ec5554a0214ca61188180a787735c68 xorg-server-1.1.99.2.tar.gz http://xorg.freedesktop.org/archive/development/individual/xserver/xorg-server-1.1.99.2.tar.bz2: MD5: 7fe0adb05ad854d68600b3fa8abf5153 xorg-server-1.1.99.2.tar.bz2 SHA1: b0443e26849ff95d5135c04b500ef4608620d356 xorg-server-1.1.99.2.tar.bz2 Cheers, Daniel Full ChangeLog since 1.1.99.1: Add support for kFreeBSD systems. (Robert Millan, Aurelien Jarno) Completely remove relocation pointer tables. Bug #6956: Fix crash when removing session leader before its children. (Rich Coe) Bug #6619: Fix disappearing hardware cursor. (Colin Harrison) Bug #5089: Die, libbitmap, die! Add bitmap to the ignored module list. Die, XTESTEXT1, die! Get rid of horrendous, somewhat misguided, DDX binary-compatibility glue. Ignore requests to load deprecated GLcore and speedo modules (properly, this time). Kill the a.out, ELF and COFF loaders with extreme prejudice. Good riddance. There's more where this came from, too. Simplify the 'unsupported' list. Remove LBX. Goodbye. Bug #6583: Only parse /proc/bus/pci/devices once. (Bill Nottingham) Powerpc machines also need ioperm_noop.c Properly document the DPMS, SyncOnGreen, and TargetRefresh options. Minor #include cleanups. Remove -xkbmap argument. check buffersize before reading next char, fixes the "mouse stuck at left" bug (bugzilla #3113) for BSD systems using wscons. (Dale Rahn). RGB color database and XErrorDB install to "share" not "lib" by default (by app/rgb and libX11). (TODO: They are customizable, so maybe cpprules.in should be extended.) Don't do fixup_video_driver_list if no drivers. (I had signal 11 and core dump when drivers not installed yet.) (TODO: fix fixup_video_driver_list to handle NULL argument.) Don't destroy a pixmap twice on server exit (bugzilla #4247). update to build against Mesa CVS HEAD (Carlos Eduardo Rodrigues Di?genes) NetBSD also has curses.h instead of ncurses.h. (TODO: Maybe instead of checking for SCO, UnixWare, Sun SVR4 and NetBSD, the curses.h versus ncurses.h should be tested in configure.) On NetBSD, add -li386 for i386 and -lx86_64 for x86_64 to SYS_LIBS. (x86_64 on NetBSD not tested.) Remove tolower() which was missed in the _XkbStrCaseCmp/strcmp changes. Add AC_SUBST([SOLARIS_ASM_CFLAGS]) that I forgot when splitting them out of the XORG_CFLAGS. Bug #6924: Restore the ABI for DrawableRec and ColormapRec to the state they were in prior to the fix for #6438. Based on a patch from Andy Ritger. Update to 2006-05-19 snapshot Remove entries now present in pci.ids Regenerate from updated pci.ids & extrapci.ids Remove dead DPMS timer functions from the installed headers. (Fredrik H?glund) Bug #4139: Fix a BAR remapping bug that could lead to IERR and system hang. (Egbert Eich) Bug #6377: Ignore disabled BARs, and allow matching BARs aligned to less than 16 bytes. (Felix K?hling, ATI) Bug #5877: Avoid burning CPU when acpid dies. Require acpid to be running for ACPI support on Linux. Minor errno handling fixes. (Valery Inozemtsev, Adam Jackson) Bug #6827: Fix texel fetch in fbFetchTransformed to avoid crashes. Still not 100% correct, but better than 7.0. (Radek Doulik, Matthias Hopf) Make DBE support conditional. (Enrico Wiegelt) Make Xv symbols conditional. (Enrico Wiegelt) Bug #5209: Fix APM/ACPI support, again. (Michel D?nzer, Valery Inozemtsev) Don't use AM_CONDITIONAL inside shell conditionals. Bugzilla #6916. previous item and update its next pointer when deleting from the linked list. Bug #5796: Wrong fastpath selection for repeating sources. Bug #4320: Improved XAA Composite fastpath. Bug #3561: Crash fix in the Record extension. (Paul Anderson) Bug #6867: Yet another Render crash fix. (Michel D?nzer) Bug #5209: Re-enable building APM and ACPI support. (Michel D?nzer) Make Xephyr build on Solaris: Add #include <sys/file.h> on Solaris for FNONBLOCK/FASYNC definitions Move agp.c & agp.h to KDRIVE_HW_SOURCES since they're not needed for Xephyr-only builds Add -lrt to XEPHYR_LIBS if needed to get nanosleep(). Redact a few mentions of speedo font support. Typo in ALLLOCATE_LOCAL() arguments, causing mis-computation of the buffer.Bugzilla #6642. Fix Solaris build with Sun compilers to work when exa is built before hw/xfree86/os-support/solaris (as it is by default now). Use min() [defined in include/misc.h] instead of MIN() [not defined in any Xorg header]. Remove NEED_LINEHELPER BC cruft for pre-R6 DDXes. Remove stray mfb/cfb references. (exaPolySegment): Add Polylines and PolyPoint acceleration as well. This is primarily to clean up fallback debugging output, so I can focus on more imporant cases. Performance is comparable but without hardware stalls, and passes Xlib9. Correct some bugs causing performance issues in the "Smart" scheme. (exaImageGlyphBlt), (exaFillRegionSolid), (exaFillRegionTiled): (ExaCheckSetSpans), (ExaCheckPutImage), (ExaCheckCopyArea), (ExaCheckCopyPlane), (ExaCheckPolyPoint), (ExaCheckPolylines), (ExaCheckPolySegment), (ExaCheckPolyArc), (ExaCheckPolyFillRect), (ExaCheckImageGlyphBlt), (ExaCheckPolyGlyphBlt), (ExaCheckPushPixels), (ExaCheckGetImage), (ExaCheckGetSpans), (ExaCheckSaveAreas), (ExaCheckRestoreAreas), (ExaCheckPaintWindow), (ExaCheckComposite): Improve EXA fallback debugging output to include the locations of pixmaps. This is being used in tracking down recent compositing performance regressions. Add trivial PolySegment acceleration for 0-width horizontal/vertical lines, which still happen somewhat frequently and were cluttering up my fallback debugging output. x11perf says it's a major performance win in those cases (though probably irrelevant), and it passes Xlib9. In drawing glyphs, shortcut our way to exaComposite instead of going through the whole CompositePicture stack and doing things like computing damage over again. This is a sizeable win for text drawing with a compmgr. Also avoid calling down into the server for dealing with the scratch pixmap when we are able to do UploadToScreen successfully and never need it. (exaGlyphs): Add a helper for the Component Alpha Over case, which breaks the operation down into an OutReverse and an Add. Turn off the fallback to software glyphs when component alpha, now that we expect all (new) drivers to be able to support it. Also, make Xephyr fall back in the CA Over case to exercise this code. This speeds up my rgb24text and ls -lR in gnome-terminal by a factor of 5. (HandlePciBios): (linuxPciHandleBIOS): Bug #6751: Use the Linux PCI ROM interface on Linux properly The old domain code, still modified BARs not a good idea, Just talk to sysfs, if 0 read, fallback to old methods. (xf86StdAccResFromOS): Bug #6750: This patch detects Intel bridges that are transparent but aren't reported as such. From the Linux kernel fixups. This patch also removes the reserved BIOS area from the area to allocate resources in. (exaMoveOutPixmap): Improve the migration debugging output. Fix a bug in the intersection computation that could concievably cause incorrect results to be returned (but would probably usually be over-conservative). Bug #4668: Check if the lists of glyphs don't have any intersecting glyphs, and if they all have a maskFormat matching the format of the actual glyphs If so, we can avoid the temporary pixmap for accumulating glyphs, which reduces the number of operations done, and makes it easier on the migration system. This fixes some significant performance issues, particularly with subpixel antialiasing. Note that it does increase the amount of damage computation which is done, so is not always a win with a compositing manager running. Add an option to verify at the point of migration that pixmaps which aren't marked dirty are in fact not dirty. This will hopefully help catch issues like the previous commit. Leave it on in fakexa. Don't forget to mark the drawable as dirty in exaPutImage(). Fixes corruption with drivers that have UTS. (Michel D?nzer) Remove another reference to cfb16. strlcpy() doesn't exist on Linux, so use the implementation in os/. Don't access free memory after unloading a module. Bugzilla #4168. When reading the kernel keyboard mapping (readKernelMapping in os-support/linux/lnx_KbdMap.c) we overrun the usefully-named global array 'map', scribbling on other random static variables elsewhere. This is fixed by changing the size of at2lnx. (David Woodhouse). Bug #5169 (fbStore_x4a4), (storeProcForPicture): Add support for x4a4 format (depth 4 at 8bpp). Bug #6325. (exaDoMigration): Add a new migration scheme, called "Smart" for lack of a better name. This one behaves somewhat between Greedy and Always. It moves in if we can accelerate, unless the destination is clean and shouldn't be kept in framebuffer according to the score, in which case we migrate out (and force-migrate anything where migration is free). This should help fix lack of acceleration for drivers without UTS since removing exaAsyncPixmapGCOps, and has removed one performance trap with Radeon I'd noticed. It is the new default. (exaGetPixmapFirstPixel): Fix exaGetPixmapFirstPixel to migrate as unaccelerated. Also adds a bit of fallback debugging info to PolyGlyphBlt. Update all prefixes in config tools, so they mostly work again. Also, make RgbPath commented out when using xorgconfig. Start using /dev/input/mice as the default mouse location on Linux. Wrap a couple more SDK headers in if XORG, as per Dave Airlie's commit on 2006-01-18. Fix kdrive build by linking in libexa before KDRIVE_LIBS. Coverity #804: Another leak on OOM path. Coverity #806: Another memory leak on OOM path. Coverity #847, #848, #849: Three more memory leaks. Coverity #1003, #1004: Two more useless null checks. Coverity #1005: Avoid a null deref. Coverity #1007: Fix a silly null check. CVT means Coordinated Video Timing instead of Common. Bug #6580: Don't install xf86drm.h, that's libdrm's job. Coverity #323, #445, #446, #447: Fix potential NULL dereferences. Coverity #169: Fix potential fgets() into NULL (?!?). Coverity #324: Fix potential NULL dereference. (Alan Coopersmith) Coverity #340: Fix potential NULL dereference. Clean up proliferation of 'register int n' in loops of ProcXkbGetNames. Bump the ABI versions. Due to Glyph privates and the XV update below, the video driver ABI needs to be bumped to 1.0. The rest of the ABI minor versions were bumped to include the LoaderGetABIVersion function. (xf86XVReputVideo), (xf86XVReputImage), (xf86XVPutStill), (xf86XVGetStill), (xf86XVPutImage): Add a DrawblePtr argument to the XV hooks. This allows drivers to determine that the target window is redirected and draw to the appropriate place. FreeResource instead of deleteCompOverlayClient() Remove libc wrapper types from Xisb interfaces. Coverity #1216: Fix double-close of file on error. Coverity #987: Avoid potential NULL dereference. Coverity #844, #845, #846: Fix memory leaks. Coverity #826: Fix potential memory leak. Coverity #333, #334 - eliminate unncessary test for always true condition in fbEvenStipple. Coverity #488: Avoid smashing an array on malformed config files. Coverity #769: Fix a potential memory leak for systems that allocate on malloc(0) Coverity #838: Plug two more memory leaks. Coverity #837: Fix another another memory leak. Coverity #836: Fix another memory leak. Coverity #835: Plug memory leak in extension section parsing. Coverity #812: Fix parser memory leak. Coverity #818: Avoid memory leak on error path. Coverity #985: Avoid segfault on malloc failure. Coverity #1037: Sanity check idx before use. Remove cfb16, no longer used. Put the screensaver extension back in the Xext module. Move the screenSaverSuspended variable to DIX globals. Restore the old link order for the Xorg and Xdmx binaries. Include fbmmx.h when USE_MMX is defined. Fixes build problem on x86-64 resulting from fbHaveMMX being a macro instead of a function on that platform. Bug #5729: Convert xf8_16bpp to fb. chips(4) users please test. Bug #5300: Fix missing spaces in the Build OS line in the log. (Egmont Koblinger) Bug #4806: Dump the raw EDID contents in hex to the log file for better debugging. (Philip Prindeville) Bug #2142: Make font path logging more readable. (Eduard Fuchs) Bug #4766: Convert all Xprint drivers to fb. Bug #5478: More use of fbSOlidFillmmx. (Jim Huang) Bug #6346: Build fix when using gcc -mno-sse. (Jonathan Adamczewski) Bug #1358: Make ISO_Prev_Group cycle/wrap as ISO_Next_Group does. Coverity #38: Dead branch elimination. Coverity #75: Dead variable elimination. Coverity #82: Dead variable elimination. Coverity #271: Fix an unbelievably boneheaded NULL chase. Coverity #616: Fix a rare memory leak. Coverity #833: Fix a rather nasty memory leak. Coverity #983: Move some risky debugging code inside #ifdef DEBUG. Coverity #986: Prevent a NULL chase. Coverity #992: Prevent a NULL chase. -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 191 bytes Desc: Digital signature Url : http://lists.freedesktop.org/archives/xorg-announce/attachments/20060603/1e6333ef/attachment.pgp