CVS log for xc/programs/Xserver/hw/xfree86/drivers/ati/ativersion.h
Request diff between arbitrary revisions
51. Change atimisc to not disable the MMIO area at the tail end of the linear aperture if it is found to be enabled on entry. Fixes a hang on Solaris/sparc upon /dev/console or /dev/wscons output while the server is running (Marc La France).
37. Radeon driver changes: - Merge in various changes from X.Org, via OpenBSD's xenocara repository. Among other things, this adds support for newer Radeon's. - Don't use vgaHW if the video mode on server entry is found to be an accelerator (i.e. non-VGA) video mode (Marc La France). - Avoid hard-failed master aborts while attempting to read video BIOS (Marc La France).
25. Make some of ATIProbe()'s verbose messages less misleading (Marc La France). 24. Fix reversed memset() arguments in atimisc LUT handling (David Krause).
Annual copyright update.
7. Speed up Mach64 block transfers on AMD64s (Marc La France).
111. Various PCI-related changes (Marc La France): - Don't ignore any BARs or ROM pointer a PCI-to-PCI bridge might have. - Don't ignore any BAR a PCI-to-CardBus bridge might have. - Destructively size PCI bases when the approriate decode enable is not set. - By default, destructively size PCI bases whose modifiable bits are all zero. Optionally also do the same in the all-ones case (dangerous). Both behavious are controlled by the XF86Config Server section "EstimateSizesAggressively" option. - When destructively sizing ROMs, do not set the decode enable bit. - Do not destructively size a base more than once. - Do not ask the OS for a base's size if the base was destructively sized. - When retrieving base sizes from the OS, filter out those that are not a power of 2, or are less than 4. - For SunOS and OpenBSD on SPARCs, add a means of extracting base sizes from PROM information. As this information is incomplete, additional changes were needed to handle devices with a mix of estimated and non-estimated resources. - Fix page alignment issues in the mapping of PCI configuration spaces for SunOS and OpenBSD on SPARCs. - Rework the scanpci utility to print type 2 PCI headers, resource size information, and, optionally, a hexadecimal dump of PCI configuration space.
Annual copyright update. Happy New Year!
45. Allow a module to export its own ModuleData structure, and use this to fix atimisc's module version checki. Problem reported by Mike Boyer (Marc La France). 44. Fix typo that caused the loader to incorrectly complain about an undefined reference in atimisc (Marc La France).
19. Replace atimisc's transfers into, and out of, VGA memory with byte-wise copies. This avoids (and stays well away from) 64-bit accesses because some VGA aperture implementations are not designed to reliably implement them (Marc La France).
1. Fix ATIProbe() buglet that can prevent detection of non-PCI adapters on OS'es that provide only estimates of PCI resource sizes (Marc La France).
Fix formatting and resync list of exported symbols.
224. TinyX build fixes, including one for Linux systems that don't provide <sys/io.h>, e.g. libc5 among others (Marc La France). 223. Improve accuracy of atimisc's calculations that are based on a 29.5 MHz reference clock (Marc La France).
208. atimisc changes: another DSP calculation tweak; work-around for screen-to-screen copy corruption on second-generation integrated controllers; make memory range prints faster (Marc La France).
193. Fix one cause of video memory corruption when using XAA with second-generation integrated Mach64 controllers. (I'm still chasing another one.) (Marc La France). 192. Chrontel 8398 related changes to the Mach64 support, including a bug fix in clock selection (Marc La France).
The annual update of my copyright
130. Add loader infrastructure to implement inter-module version checks and use it as part of a rework of VBE information blocks (Marc La France).
Bump patch number
71. Fix for GENENA register handling in ATIProbe() (Marc La France).
69. Change atimisc driver on big-endian systems to allow depth 24 even if a big-endian aperture is not available, by swapping RGB masks (Marc La France).
65. Restore accuracy of atimisc's clock probe despite ceratin I/O errors (Marc La France).
64. ATIProbe() and atimisc changes: - Port to common layer's domain scheme; - Replace AVOID_CPIO and AVOID_NON_PCI rchitecture-specific #define's with run-time checks and implement new common layer infrastructure to allow this to happen; - Replace endianness compile-time checks with run-time ones; - Deal with certain spurious I/O errors; - No longer consider IOBase when matching adapters to device sections This re-instates support for all ATI adapters on all architectures supported by the common layer (Marc La France). 63. Make sparse I/O masks more portable throughout the server (Marc La France).
312. Fix another buglet in atimisc's decoding of a panel's mode on server entry (Marc La France).
Update my copyright. It's gotta be midnight somewhere by now...
251. Fix minor bug in Linux/Alpha's xf86GetPciSizeFromOS() (Marc La France). 250. Cause the Linux port to warn when MAX_PCI_BUSES and/or MAX_PCI_DEVICES need to be increased (Marc La France). 249. Deal with Linux 2.6's PCI bus naming (Marc La France). 248. Resync with Martin Mares's 2004.11.29 pci.ids snapshot and add to it an entry for Sun's Schizo+ (Marc La France). 247. Re-instate DGA support in atimisc driver on UltraSPARC's (Marc La France). 246. Add support for the more recent UltraSPARC implementations to the Linux port (Marc La France). 245. Fix DGA for multi-domain architectures (except perhaps for Linux/Alpha, see comment inserted into axpPci.c). PCI only for now (Marc La France) Note these changes require `make World`.
232. Rationalise driver interface for DGA's OpenFramebuffer(); fix siliconmotion driver's entry for it; bump driver ABI version (not strictly necessary as this is only a source API change) (Marc La France).
178. Fix atimisc driver for unusually small page sizes (Marc La France).
134. Set subpixel order to horizontal RGB in atimisc panel support (Adam J. Richter, Marc La France). 133. Fix off-by-one error in the interpration and programming of VGA sync pulses. This has only been done for atimisc, and should also be for vgahw and callers, but the complexity of doing so right now far exceeds my ability to test such a change (Marc La France). 132. Some RENDER cleanup in drivers (Marc La France).
25. Small tweak to fixed-clock line matching in the atimisc driver (Marc La France).
820. ati & atimisc module changes (Marc La France) - Allow builds with ATIAvoidNonPCI && !ATIAvoidCPIO; - Make handling of unknown Radeon's saner; - Add private option to re-enable BIOS display switches; - Deal with ati & atimisc module version skews; - Fix private option handling;
2004 update to my copyright.
Fix bug introduced by #583.
583. Fix atimisc's decoding of the mode on input for large panels (Marc La France).
327. Add Mach64 I2C framework. Derived from GATOS project (Marc La France).
Correct stupid typo and bump version.
Credit, potential SEGV and version bump
136. Fix bug that prevented fbman from using the last partial scanline of a Mach64 framebuffer (Marc La France). 135. Make ATI Mach64 FIFO cache integrity testing optional (Marc La France). 134. Export ATI Mach64 hardware overlay as an XVideo adaptor (derived from GATOS project, Egbert Eich, Marc La France). 133. Reorganise ATI Mach64 support into separate source files (Marc La France). 132. Refine atimisc's decoding of the panel mode on server entry in an attempt to reduce the effect of atyfb bugs (Marc La France). 131. Make Rage128 and Radeon XVideo available even when 2D acceleration is disabled (Marc La France). 130. There is no longer any need to require hardware cursors during Rage128 and Radeon XVideo displays (Marc La France). 129. Initialise v4l's XVideo adaptors last (Marc La France). 128. Reduce cut&paste effects by adding more helpers to Xv (derived from #5645, Björn Augustsson, Marc La France). 127. Centralise a region comparison primitive into 'mi' and use it instead of local definitions throughout the server (Marc La France). 126. DPMSExtension & XvExtension driver cleanups (Marc La France).
708. Fix for when firmware/BIOS initialisation does not enable the extra command FIFO entries available on second-generation integrated Mach64 variants (Marc La France).
Update my copyright
658. Fix atimisc panel support bug that occurs when the mode on server entry is a VGA mode with large horizontal and/or vertical blanking pulses (Marc La France).
Bump driver version
467. Ignore (with messages) any HorizSync and VertRefresh specifications for panels when using ATI LT, LTPro, XL and Mobility M1 (Marc La France). + some version number tweaks.
405. Fix long-standing colour map bug in ATI driver (Marc La France).
107. Rework ati, atimisc, r128 and radeon module versioning (Marc La France). 106. Code clean up for the original Rage LT (Marc La France). 105. An attempt at disabling display switches while the server is running on an LT Pro or Mobility M1 (Marc La France). 104. Make atimisc, r128 and radeon modules register the symbols they reference at module load time, rather than PreInit() (Marc La France). 103. Fix rounding error in Mach64 DSP calculation (Marc La France).
39. Fix IBM-reported bug that could cause PCI bus lockouts with second generation integrated Mach64 controllers (Marc La France). 38. Rework ATI driver's registration of symbols with the loader (Marc La France). 37. In the ATI driver, add option "CrtDisplay" to force CRT display in a dual display situation (Marc La France). 36. In the ATI driver, rename option "CrtScreen" to "NoPanelDisplay", keeping "CrtScreen" as a private option for compatibility (Marc La France). 35. Document X_NOT_IMPLEMENTED (Marc La France).
Prep work for future changes.
Resync with 188.8.131.52
Update my copyright notice. Drop OldChipID() for R128's and Radeon's. Miscellaneous ATI driver cleanups.
407. Fix for threaded libraries (Marc La France). 406. Finish removal of SuperProbe (Marc La France). 405. A rather large number of warning fixes throughout (Marc La France). 404. Fix bug in HTML install script (Marc La France). 403. Missing ident lines for some XFree86-modified files (Marc La France). 402. Add default half-width doublescanned modes (Marc La France). 401. Mark all driver-registered resources with ResBus (Maarc La France). 400. Fix DPMS-related build problem (Marc La France). 399. Log a message just before calling each ChipProbe() during '-probe' processing (Marc La France). 398. Temporarily disable ISA probing on SPARCs and PowerPCs (Marc La France). 397. Add PCI IDs for Sun hardware (Marc La France). 396. Fix memory leak in resource relocation (Marc La France). 395. Do not relocate resources that only conflict with disabled non-video PCI devices or disabled PCI ROMs (Marc La France). 394. Re-organise SBUS code (Marc La France). 393. Add as-yet-unused definitions for PCI resource types other than I/O and memory (Marc La France). 392. Add doc for Solaris, but don't format it yet (Marc La France). 391. Normalise driver names (Marc La France). 390. For SPARCs, disable DGA support in ATI driver (Marc La France). 389. Clean up some debugging messages (Marc La France). 388. Fix newport driver for when a /proc fs isn't mounted (Marc La France). 387. Fix DAC handling bugs in s3 driver (Marc La France). 386. Fix resource registration bug for PCI Tseng's (Marc La France). 385. Add aperture driver for Solaris (not yet used) (Marc La France). 384. Rework scanpci to fix problems that prevented it from completely displaying non-PCI bridges and Simba bridges (Marc La France). 383. Fix build problems in some input drivers (Marc La France). 382. Fix int10 compile problem for SPARCs and PowerPCs (Marc La France). 381. Ensure master aborts on secondary buses complete normally during PCI scans (Marc La France). 380. Some memory mapping and Solaris cleanups (Marc La France).
258. Fix banked 24bpp displayWidth calculation (Marc La France). 257. Fix accel initialisation bugs in ATI driver (Marc La France). 256. Add entry to initialise framebuffer manager by area (allowing for a possible extra partial scanline) and make it print all largest areas (Marc La France).
217. Experimental change to default behaviour of ATI panel support to override the horizontal and vertical porches of all modes to that found in the mode on entry (Marc La France). 216. Fix detection of a panel's mode on entry in the ATI driver (Marc La France). 215. Fix for Option "CRTScreen" in ATI driver (Marc La France). + build fixes for recent changes.
168. Fix 16bpp colour maps on Mach64's (Marc La France). 167. Fix Mach64 DPMS bug (Marc La France). 166. Fix bug that prevented hardware cursors on Mach64 CT's (Marc La France). 165. Document what a ChipAdjustFrame() function should to to prevent moving of hardware cursors during mode switches (Marc La France). 164. -Wimplicit warning cleanup [`make World` >required<] (Marc La France). 163. Ensure installed headers can find the headers they, in turn, #include (Marc La France). 162. Make imake pass its -W* flags to cpp (Marc La France). 161. More accomodations for newer GCC's (Mike A. Harris).
145. Fix typos in radeon and wacom drivers. 144. Don't disable the secondary CRTC on ATI Rage XL adapters (it doesn't exist) (Marc La France). 143. Speed up ix86 block moves in ATI driver (Marc La France). 142. Temporarily disable hardware cursor support for ATI GX & CX adapters (Marc La France). 141. Make composite sync the default in the ATI driver (David S. Miller). 140. Rework stretching of small DFP modes in ATI driver (Marc La France). 139. Add an interesting PCI ID (Marc La France). 138. Allow ChipValidMode() functions to pre-calculate a mode's horizontal sync and vertical refresh rates (Marc La France). 137. Another fix for mode switch and cursor interactions (Marc La France). 136. Temporary fix for endianness problems in ATI driver (Marc La France). 135. Add XGetIMValues() prototype to Xlib.h (Marc La France). 134. Use a more reasonable numbering for recent binutils versions on Linux (Marc La France). 133. Fix dependencies for building drivers in a static server (Marc La France). + build & warning fixes.
101. Fix bogus handling of BUS_CNTL Mach64 register (Marc La France).
29. ATI DGA changes (Patrick Chase and Marc La France): - Add accelerated DGA primitives for Mach64 variants; - Add accelerated DGA transparent blits for Rage128 and Radeon; - Fix setting of DGA_CONCURRENT_ACCESS; - Take XaaNo* options into consideration. 28. Simplify DGAShutdown() (Marc La France). 27. Fix colourmap bug when switching out of a DGA mode (Marc La France). 26. Make it possible, but not required, for a driver to initialise DGA after XAA so that it may factor in any XaaNo* options in determining what DGA drawing primitives to support (Marc La France). 25. Prevent mode switches while DGA modes are set (Marc La France).
551. More build & warning fixes (Marc La France). 550. Fix bug that caused hardware cursors to be temporarily moved during mode switches (Marc La France). 549. Optimise HARDWARE_CURSOR_AND_SOURCE_WITH_MASK case (Marc La France). 548. Move xf86CursorScreenRec definition into xf86CursorPriv.h (Marc La France). 547. Fix BIOS retrievals in MGA driver (Marc La France). 546. Fix ATIProbe() for newer Rage128 and Radeon chips (Marc La France). 545. Add temporary workaround in ATI driver for interrupts that occur on PowerPC's upon PCI master-aborts (Marc La France).
489. Hardware cursor documentation update (Marc La France). 488. XVideo extension support in ATI driver for generic (e.g. v4l) adapters (based on a patch by Ignacio Garcia Etxebarria). 487. Fix ATI driver bug that caused the hardware cursor to move after mode switches (Marc La France). 486. Fix off-by-one error in screen positioning after a mode switch (Marc La France). 485. Backout part of 479 and implement a proper fix in the ATI driver (Marc La France).
New PCI id's.
399. Implement two new driver-provided options in the common cursor code and use them to work around hardware bug that causes flickering Mach64 hardware cursors (Marc La France).
Update formatted docs; Bump driver version.
317. glxinfo needs libGLU (Marc La France). 316. Rage128 fix to Cards database (Marc La France). 315. Minor fix to PCI resource overlap handling (Marc La France). 314. Loader code simplification and IA-64 cache flushes (Marc La France). 313. Workaround in the ATI driver for troublesome interaction between loader and compiler optimisation (Marc La France). 312. Warning fixes for `gcc -fno-builtin`, which appears to be the default on some systems (Marc La France). + more of #301, tags, warnings and build fixes.
303. XvMC build fix (Marc La France). 302. Fix recently introduced bug in xf86cfg (Marc La France). 301. Re-instate warning fixes clobbered by DRI merge (Marc La France). 300. Ignore disabled non-video PCI ROMs (Marc La France). 299. Hardware cursor implementation for Mach64 variants (William Blew, Ani Joshi, Marc La France). 298. Some endianness fixes in the ATI driver (incomplete) (Marc La France). 297. Fix MMIO ordering problems for Mach64 acceleration (Marc La France). 296. Fix ATI driver behaviour when it finds the LCD panel has been disabled on entry (Marc La France). 295. Fix ECP clock for GATOS (Vladimir Dergachev). 294. Undo part of CHANGELOG 36 (Marc La France).
Bump driver version.
1279. Disable RENDER extension support in the ATI(misc) driver when mibank and/or shadowfb is used (Marc La France). 1278. Remove cfb & friends from the ATI driver (Marc La France). 1277. Some additional ATI PCI IDs (Marc La France). 1276. Make compilation of drivers produce the same object whether or not DPMS and DGA extensions are being built. Partial changes to do the same for RENDER (Marc La France). 1275. Ensure `make depend` goes through the same set of subdirectories as `make all` does. Allows MakefileAdditions to contain depend rule(s) (Marc La France). 1274. Move $(EXTRA_DEFINES) to the end of ALLDEFINES. Ditto for $(CXXEXTRA_DEFINES) and CXXDEFINES. Allows easier overrides at the subdirectory level (Marc La France). 1273. Add a pre-clean phase to `make World` if xc/xmakefile already exists (Marc La France). + update formatted docs + build & warning fixes
Update my copyright and email address.
1177. Fix recognition of valid depth/fbbpp combinations in ATI driver (Marc La France). 1176. Fix 1bpp in ATI driver (Marc La France).
Update credits and version number
Merge Rage128 and Radeon drivers into the ATI driver.
832. Split ATI driver into two modules in preparation for merging in the r128 and radeon drivers (Marc La France).
751. Fix int10 so that it rejects BIOS's that are not aligned on a 512-byte boundary (Marc La France). 750. Fix int10 for ISA adapters on Alpha's (Egbert Eich). 749. Fix for ATI ISA adapters on Alpha's or with -configure (Marc La France). 748. Fix stupid clock probe bug in ATI driver (Marc La France). 747. Reorganise ATI driver in preparation for futuure changes (Marc La France). 746. Formatting changes to hsync/vrefresh messages (Marc La France). 745. Fix spurious free() when using an ISA adapter (Marc La France).
721. PCI chip ID updates to ATI driver (Marc La France). 720. Fix i810 driver for -probe and -configure (Marc La France). 719. Change message when default modes are deleted (Marc La France). 718. Fix Xinerama byte swapping bug (Marc La France). 717. IA-64 and Alpha fixes for pswrap, Mesa, DRI, Xpm, libX11, Xt, Xaw, Xmu, dps, Type1 fonts, cfb24, most output drivers, ELF loader, ramdac module, xf4bpp and xf86cfg (Marc La France). 716. Improve IA-64 support by removing a plethora of 32-bit'isms (Marc La France). 715. Default HasLinuxDoc to NO (Marc La France). + ident lines and warning fixes.
571. Fix input event queueuing for 64-bit platforms (Marc La France). 570. IA-64 changes resulting from a source cross-check with Alpha support for 64-bit dependencies (Marc La France). 569. Fix in ATI driver for DPMS support on panels (Marc La France).
561. Fix ATI driver bug that prevented the disabling of int10, ddc and vbe through options (Marc La France).
Corrections to previous changes.
523. ATI driver changes (Marc La France): - Make ATIProbe() and ATIPreInit() optionally more verbose. - Fix compile problem on Alpha's. - Fix Mach64 hardware clipping bug. - Support transparency during Mach64 screen-to-screen copies. - Disable int10 interface in MMIO-only driver version. - Fix determination of MMIO base address for Mach64 GX, CX, CT, ET, VT and GT. - "Wake" up ATI VGA's before probing for them. - Other cosmetic changes.
492. ATI driver changes (Marc La France): - Fix resource relocation bug. - An attempt at fixing problems reported with Chrontel 8398's. - Workaround for Mobility BIOS bug that affected the driver's ability to restore the mode on server entry. - Fix SEGV that occurs with ATI adapter BIOS'es that do not support some flavour of VBE. - Allow the driver to tolerate the absence of an initialised adapter BIOS image and introduce XF86Config options for the information the driver would otherwise require from it. - Recognise more cases where a DFP panel might be in use. - Fix minor bug that occurred on xf86SetDepthBpp() failures. - Implement a compilation option that produces an MMIO-only version of the driver. Intended for non-Intel architectures. See the driver's Imakefile for details. + post-Egbert fixes to trident driver.
417. Make ATI driver tolerate relocation of conflicting resources (Marc La France). 416. Make Mach64 FIFO handling available to GATOS (Marc La France). 415. Fix bug in Mach64 scissor handling (Marc La France). 414. Implement Mark Vojkovich's suggestions in the ATI driver (LSB ordering of monochrome data and host transfer burst modes) (Marc La France). 413. More fixes to ATI DSP register calculation (Marc La France). 412. ATI DGA support fixes for VGA Wonder capable adapters and the setting of the DGA_CONCURRENT_ACCESS flag (Marc La France). 411. Change imake to generate symbols for the GCC version used to compile it (Marc La France).
327. xdm warning fix (Marc La France). 326. Make x11perf dies gracefully when display cannot be opened (Marc La France). 325. Fix rendition driver compile glitch (Marc La France). 324. #ifdef out glitch in glint driver (Marc La France). 323. Preliminary DGA support in ATI driver (Marc La France). 322. Minor update to DESIGN and Status documents (Marc La France). 321. Fix DGA's reporting of vertical refresh rates to clients (Marc La France). 320. Make driver-provided Sync(), SetViewport() and GetViewport() DGA callbacks optional (Marc La France). 319. Fix XKB compile glitch (Marc La France). 318. Make imstt driver compile (Marc La France).
Compile fixits Fix, -configure, yet again, for multiscreen with identical adapters. EDID formatting changes. 123. Fix mode setting bug in ATI driver (Marc La France). 122. Add int10, DDC & VBE support to ATI driver (Marc La France). 121. Allow driver to supply DDC module to VBE (Marc La France).
32. ATI driver changes: Implement Mach64 engine restriction on virtual resolution; Add DPMS support; More DSP calculation fixes; Fix text mode restoration bug; Fix Mach64 screen blanking; Don't probe for previously claimed VGA compatibles; Fix device section matching for Mach64 controllers made by UMC foundry; Miscellaneous cleanups (Marc La France).
Fix ATI panel clock problem with newer GCC's, and other more minor problems.
Add Mach64 screen-to-screen copy acceleration.
3811. A rework of Kevin Martin's Mach64 acceleration. Only solid fills for now (Marc La France). 3810. ATI Mobility fixes (Marc La France). 3809. A new clock ordering has been reported for Mach32's. Treated as a different clock chip for now (Marc La France). 3808. Build fixups (Marc La France).
3272. Mode validation changes (Marc La France). - Remove assumption that modes of the same name are also of the same size. - Even if Display subsection mode names include duplicates, no two names will ever match the same mode. - If the supplied mode names do not yield a usable mode (including the case where no names are supplied), scan the mode pool until a usable mode is found. This is a step towards allowing successful server runs with XF86Config's that specify no mode information whatsoever. - Implement minHeight and maxHeight checks. - Call the driver's ValidMode() for a final OK just before accepting a mode. - Free mode pool on xf86PruneDriverModes() calls. - Documentation updates. - Minor fixes to C&T and GLINT drivers. 3271. ATI driver updates (Marc La France). - Disallow interlaced modes when scanline pitch exceeds hardware limits. - LCD support no longer depends on the mode on server entry. - Fix bug introduced by CHANGELOG entry 3245. - Re-lock adapters on ATIPreInit() failures. - Make default sync polarities consistent.
3188. Turn MGA soft reset stuff back on but modified so that it doesn't poll for fifos in case the PLLs aren't powered up (#3254, Mark Vojkovich). 3187. Xterm patch #115 (#3253, Thomas Dickey). 3186. Comment out some of the CacheBlit tiling code since the logic is not quite correct and it can get stuck in an endless loop in some cases (#3250, Mark Vojkovich). 3185. Fix tseng 24 bpp mode (#3249, Egbert Eich). 3184. Fix et4000W32p font corruption with ColorExpandScanline in modes > 8bpp (#3249, Egbert Eich). 3183. Add Solaris 8 support (#3241, 3247, David Holland). 3182. Fix some build problems in the Type1 code that show up on Solaris 8 (#3241, David Holland). 3181. Add an ISO8859-1 version of the "micro" font (#3246, Mark Montague). 3180. Minor DDC code improvement: xf86PrintEDID() now returns pointer to monitor structure (#3245, Egbert Eich). 3179. Problem with RAC wrapper (hopefully) fixed (#3245, Egbert Eich). 3178. Neomagic ScreenToScreenCopy bug fix (#3245, Egbert Eich). 3177. Initial INT10 module (Linux-only so far) (#3245, 3248, 3251, Egbert Eich). 3176. Update the DRM Linux kernel module to match that in the Linux 2.3.18 tree, and update libdrm and the DRI routines that use libdrm to reflect changes in the Linux kernel (#3244, Rik Faith). 3175. Fix a WarpPointer bug in Xinerama (#3243, 3252, Mark Vojkovich). 3174. Fix a seg fault when changing DGA mode (#3242, Loic Grenie). 3173. Xterm patch #114 (#3240, Thomas Dickey). 3172. Fix an XAA GC validation problem with Polygons for some fillStyles in some situations (#3239, Mark Vojkovich). 3171. Fix an incorrect method for preallocating a color on the default map (#3237, Mark Vojkovich). 3170. Fix a bad assumption in the XAA NonTE text rendering code (#3236, Mark Vojkovich). 3169. An attempt to autodetect the G400 Max and set the max pixel clock appropriately (#3235, Mark Vojkovich). 3168. Fix shadowfb problems when drawing is done while switched out (#3234, Marc La France). 3167. Attempt to autodetect SDRAM in the MGA driver for the primary head (#3233, Mark Vojkovich). 3166. Make Window->Pixmap copies in Xinerama less broken (#3232, Mark Vojkovich). 3165. Fix a bug in the Xinerama ShmGetImage support added in the last alpha (#3230, Mark Vojkovich). 3164. New cursor warping code. This implements the new way of specifying screen positions in the ServerLayout config file section (#3229, Mark Vojkovich). 3163. Add a xf86CheckMTRR function which returns TRUE when changing the MTRR setting is supported and enabled (#3227, Alan Hourihane). 3162. Resync SuperProbe and scanpci with the 3.3.x version (#3226, Marc La France). 3161. Make 8+24 work on Matrox 2064w (Millennium I) with >4MB (#3224, Mark Montague). 3160. Make xdpyinfo -ext XFree86-VidModeExtension print the current mode line (#3223, Andrew Aitchison). 3159. Minor fix to Xnest Imakefile (#3222, Marc La France). 3158. Make the wrapping or replacement of ScreenRec region code function pointers reliable when NEED_SCREEN_REGIONS is #define'd (#3221, Marc La France). 3157. Remove ScreenRec region code function pointers (except BitmapToRegion) when the server is compiled with NEED_SCREEN_REGIONS #undef'ined (#3221, Marc La France). 3156. Fix memory leak during resource broker initialisation (#3220, Marc La France). 3155. During broker initialisation, use both system and non-system PCI resources to "trim" OS resources. Among other things, this allows the maximum memory limit to be bumped back up to 2G (#3220, Marc La France). 3154. Temporarily allow a driver to register resources that only conflict with resources whose size the common layer has only estimated (#3220, Marc La France). 3153. ATI driver updates, including: - Add option to display on the CRT interface whether or not a digital flat panel is detected. - Add support for ShadowFB (adapted from Henrik Harmsen). - Always read BIOS from legacy location for adapters found enabled on server entry. - Ensure the primary CRTC is always used for the CRT interface. - Fix determination of panel dimensions with older BIOS's. - Stop driver when panel dimensions cannot be determined. (#3219, Marc La France). 3152. Thread-related updates to 'ico' (#3217, Carlos A M dos Santos). 3151. Fix a build problem on 64-bit platforms introduced in 3.9.16a (#3216, Matt Grossman).
2970. Modify DDC for neomagic (#3139, Egbert Eich). 2969. Add DDC support for the S3 ViRGE (DDC2 untested) (#3139, Egbert Eich). 2968. Fix DDC2 support for MGA and C&T drivers (#3139, Egbert Eich). 2967. Add a call to xf86SetPciVideo(NULL,NONE) after each probe (#3139, Egbert Eich). 2966. Modified registration of the OS access windows (#3139, Egbert Eich). 2965. Some Xaw and xedit fixes (#3138, Paulo Cesar Pereira de Andrade). 2964. Next round of the font ansification and warning removal (#3135, Matthieu Herrb). 2963. Fix ATI 18800-x detection problem (#3133, Marc La France). 2962. Check for limitations on vertical blending in panel support for the ATI driver (#3133, Marc La France). 2961. Update ATI driver to complain about bad BIOS data (#3133, Marc La France). 2960. Man page corrections for the MGA driver (#3132, Mark Vojkovich). 2959. Limit PCI BIOS apertures to 16MB as per the PCI 2.1 spec (#3131, Marc La France). 2958. Assume a system has up to 1G of physical memory installed, rather than 2G (#3130, Marc La France). 2957. Fix a bug in the handling for buggy VGA CRTCs (#3129, Marc La France). 2956. Fix to build the mga driver with the driver SDK (#3128, David Bateman). 2955. Xterm patch #113 (#3125, Thomas Dickey). 2954. Port the XAA cfb copyplane functions, adding them to the cfb code (#3124, 3126, 3127, David Bateman).
XFree86 3.9.15a (1 August 1999) 2934. Reduce memory consumption of the bitmap font backends when using sparse fonts, which is important when using Unicode-encoded fonts (#3092, Juliusz Chroboczek). 2933. Fix Option register initialisation in multi-head for MGA 2064/2164 (#3089, Andrew Aitchison). 2931. Add SuperProbe for NetBSD 1.4 wscons (#3083, Matthias Drochner). 2930. Fix button and modifier key state problems in DGA (#3082, Mark Vojkovich). 2929. Avoid a possible segsegmentation fault in "xsm" which, for example, occurs local hostname can't be resolved (#3073, Matthias Scheler). 2928. ATI driver updates, including: - Add preliminary panel support. - Rework RAC interface. - PCI VGA's not on primary bus are now detected. - Fix bug when an adapter's VGA has been disabled. - Fix gamma support. (#3069, Marc La France) 2927. Fix 8514/A pre-defined resource and pre-define VGA sparse I/O resources (#3069, 3081, Marc La France). 2926. Fix a bug in the registration of background device BIOS'es (#3069, Marc La France). 2925. Invert the meaning of ResMinimised bit into ResEstimated (#3069, Marc La France). 2924. Allow drivers to print resource lists, and allow them to enable/disable adapters before creating bus slots for them (#3069, Marc La France). 2923. Resync SuperProbe and scanpci with 3.3.x (#3069, Marc La France). 2922. Add some missing man pages for the DGA and VidMode extensions (#3064, Joe Moss). 2921. Some changes to the DGA and colormap layers to allow gamma changes during DGA mode (#3060, Mark Vojkovich). 2920. Fix XIM segv error (#3033, 3051, Manabe Takashi). 2919. Add a client state callback to the DGA extension so that the server side can be cleaned up when a DGA client exits (#3048, 3049, Mark Vojkovich). 2918. Update the FreeBSD/mouse documentation (#3047, Kazutaka Yokota). 2917. Fix an undocumented blitter constraint on the G200 (#3046, Mark Vojkovich). 2916. Fix the MGA driver sot hat it can support changing depth on the fly in DGA mode. All depths are now offered to DGA clients regardless of what the screen's main depth is (#3045, Mark Vojkovich).
2835. Fix for GetBlock() (#2974, Egbert Eich). 2834. Remove the test for SHAREDIOCLASS when creating a pciAccessInfoRec for a device (#2974, Egbert Eich). 2833. Modified BIOS ROM handling, doing validation only on demand (#2974, Egbert Eich). 2832. Add a ChangeGamma function to the ScrnInfoRec, which allows drivers to optionally provide a function to do this instead of the default one provided by the colormap layer (#2972, Mark Vojkovich). 2831. Sis driver memclock fixes (#2971, Bart Oldeman, Juanjo Santamarta). 2830. Add snprintf() and vsnprintf() to libc emulation (#2970, Marc La France). 2829. Increase array size in PciInfo data (#2970, Marc La France). 2828. Add a LoaderRefSymbols() function (#2970, Marc La France). 2827. When matching fixed clocks, prefer non-CLKDIV2 modes over CLKDIV2 modes (#2970, Marc La France). 2826. Reinstate xf86GetPciConfigInfo() and some warning fixes (#2970, Marc La France). 2825. Build libXext and libX11 client-side libraries when also building GLX, despite BuildServersOnly (#2970, Marc La France). 2824. Update ATI doc (#2970, Marc La France). 2823. ATI driver from 184.108.40.206 (rac interface is preliminary, no LCD panel or Rage 128 support yet) (#2970, Marc La France). 2822. Fix Ukrainian_GHE_with_upturn double definition error (#2969, Peter Novodvorsky).
#1605-1616builds static servers okVGA16 and SVGA run fineSVGA dies with an undefined CloseSreen member of pScreen when exitingthe last client
patches #1041,1044-1047,1050-1058,1061-1065,1069,1072; I didn't include the contrib patches and the patches from Louis Kruger, as well as the TOG stuff; this is not build tested, yet
add all the new files and changes 769+821