NEWS revision 13496ba1
1Snapshot 2.99.917 (2014-12-21)
2==============================
33 months drifted by whilst I looked elsewhere for bugs.. The highlight of
4bugs fixed here are a couple of workarounds required for Broadwell and
5making sure that the rasterisation code is symmetric under inversions.
6However, as a couple of crashers slipped through into 2.99.916 (though
7not actual regressions in 2.99.916 per se) and 3 months have passed, we
8should make one more snapshot before an imminent release.
9
10 * Beware of recomputing the clear hint in the middle of MI recursion
11   https://bugs.freedesktop.org/show_bug.cgi?id=77074
12
13 * Fix crash from rendering an empty Glyph string under PRIME as the
14   secondary driver
15
16 * Restore application of default monitor options to the first output,
17   a regression in 2.99.915 and the MST support
18   https://bugs.gentoo.org/show_bug.cgi?id=522500
19
20 * Finally fix ZaphodHeads blocking on waiting for flip completion events.
21   Before kernel 3.19, O_NONBLOCK support is broken and so we must avoid
22   reading if we are not expecting an event.
23
24 * Backwards compatibilty fix for fake triple buffering with PRIME and
25   Xorg-1.15
26   https://bugs.freedesktop.org/show_bug.cgi?id=85144#c12
27
28 * Fix a rendering issue with output rotation and software fallbacks.
29   https://bugs.freedesktop.org/show_bug.cgi?id=84653
30
31 * Enable MST discovery even without udev support
32   https://bugs.freedesktop.org/show_bug.cgi?id=84718
33
34 * Fix TearFree operation after the GPU is wedged
35   https://bugs.freedesktop.org/show_bug.cgi?id=85058
36
37 * Fix projective sampling on gen6+.
38
39 * Fix rendering and corruption with gen8.
40   https://bugs.freedesktop.org/show_bug.cgi?id=84958
41   https://bugs.freedesktop.org/show_bug.cgi?id=83207
42   https://bugs.freedesktop.org/show_bug.cgi?id=79053
43   https://bugs.freedesktop.org/show_bug.cgi?id=81583
44
45 * Fix crash when using Xinerama.
46   https://bugs.freedesktop.org/show_bug.cgi?id=87207
47
48 * Fix rendering of right-to-left or bottom-to-top PolySegments in UXA
49
50
51Snapshot 2.99.916 (2014-09-08)
52==============================
53Quick update for MST in UXA - we need to hook up the RandR outputs for
54dynamicaly added connectors.
55
56
57Snapshot 2.99.915 (2014-09-08)
58==============================
59A significant change to UXA to enable MST and to keep it working on recent
60kernels needs some soaking before a major release, and the usual
61plethora of bugfixes. One other feature is the support for hardware
62rotations on very recent kernels.
63
64 * Handle rotated scanouts that are all clear correctly. The region to
65   update was not being computed correctly, overdrawing the wrong CRTC.
66   https://bugs.freedesktop.org/show_bug.cgi?id=81820
67
68 * Reset cursor images after rotation and size changes
69   https://bugs.freedesktop.org/show_bug.cgi?id=81886
70   https://bugs.freedesktop.org/show_bug.cgi?id=82273
71   https://bugs.freedesktop.org/show_bug.cgi?id=82337
72
73 * Handle stale DRI2 buffers and Client errors more gracefully
74
75 * Fallback if we fail to render a glyph onto a too-large surface, e.g.
76   direct rendering of glyphs onto extended desktops with gen2/gen3.
77   https://bugs.archlinux.org/task/40949
78
79 * Further work to enable tiled rendering onto large surfaces in
80   severely aperture and memory constrained devices (e.g. gen2/gen3)
81
82 * Honour the Primary option from xorg.conf and mark that as the RandR
83   primary.
84   https://bugs.freedesktop.org/show_bug.cgi?id=82193
85   https://bugs.freedesktop.org/show_bug.cgi?id=82205
86
87 * Fix another Client-Window DRI2 close race
88   https://bugs.freedesktop.org/show_bug.cgi?id=82979
89
90 * Fix incorrect discarding of GPU damage when copying over the TearFree
91   scanout - which caused a flicker in rapidly updated elements (like
92   simple video panes).
93   https://bugs.freedesktop.org/show_bug.cgi?id=81973
94
95
96Snapshot 2.99.914 (2014-07-23)
97==============================
98And a brown paper bag to hide the rebuilding from the tarball with
99'autoreconf -fi' error that arose from not distributing the libobj/
100directory.
101
102
103Snapshot 2.99.913 (2014-07-23)
104==============================
105This should be it... A few fixes from testing the new code, we should be
106ready for the final release. However, we do have one standout feature in
107this snapshot, we now officially recognise HD Graphics 5300/5500/5600,
108Iris Graphics 6100 and Iris Pro Graphics 6200/P6300 (formerly known as
109Broadwell).
110
111 * Check the window actually covers the CRTC before doing a single CRTC flip,
112   and then restore the right framebuffer after completing CRTC flips.
113   Otherwise we would detect an error and disable an output under TearFree
114   Regression in 2.99.912
115   https://bugs.freedesktop.org/show_bug.cgi?id=80191
116
117 * Fix framebuffer creation on kernels older than 3.11
118   Regression in 2.99.912
119
120 * Check that the damage still exists after implicit reduction
121   Regression in 2.99.912
122   https://bugs.freedesktop.org/show_bug.cgi?id=77436
123
124 * Fix direction flags for fallback composited CopyAreas which
125   caused scrolling corruption in a few configurations
126   Regression from 2.20.0
127   https://bugs.freedesktop.org/show_bug.cgi?id=79843
128
129 * Do not throw away damage if there is no redundant copy
130   https://bugs.freedesktop.org/show_bug.cgi?id=79992
131
132 * Check clipping on PolyRect before discarding the clipped damage
133   Regression from 2.99.903
134   https://bugs.freedesktop.org/show_bug.cgi?id=79992
135
136 * Fix hints for GLXPixmapa, as these are never swapped and so
137   miss invalidating the hints on SwapBuffers with the result
138   that they are often presumed blank
139   Regression in 2.99.912
140   https://bugs.freedesktop.org/show_bug.cgi?id=79999
141
142 * Fix incoherent choice of source bo when constructing 8x8 tiles,
143   incorrect pattern origin when extracting, and then fix the
144   alignment of colour patterns for BLT operations
145   https://bugs.freedesktop.org/show_bug.cgi?id=80033
146
147 * Disable blending with the render engine on snoopable buffers
148   https://bugs.freedesktop.org/show_bug.cgi?id=80253
149
150 * Restore throttling to prevent client lag under heavy GPU load
151   Regression from 2.21.10
152   https://bugs.freedesktop.org/show_bug.cgi?id=77436
153
154 * Use ClientGone for notifications on shared DRI2 windows to prevent
155   rare crashes due to use-after-free of the swap requests
156   https://bugs.freedesktop.org/show_bug.cgi?id=80157
157
158 * Ensure the mmaped CPU bo is idle before migrating damage
159   https://bugs.freedesktop.org/show_bug.cgi?id=80560
160
161 * Fix incorrect clipping by the render engine for large DRI2 windows
162
163 * Ensure that the aperture tiling fallbacks are bounded
164
165 * Validate parameter to xf86-video-intel-backlight-helper more carefully
166   (CVE-2014-4910)
167
168 * Fix slaved scanouts for reverse optimus, though rotated slaves will
169   require further patches to Xorg.
170   https://bugs.freedesktop.org/show_bug.cgi?id=81383
171
172 * Fix build without Composite extension.
173
174 * Fix build without gettline().
175
176 * UXA: Allocate and resize frontbuffer consistently to pass sanity checks
177   https://bugs.freedesktop.org/show_bug.cgi?id=80088
178
179 * UXA: Report cached backlight value when the output is off (like sna)
180   https://bugzilla.redhat.com/show_bug.cgi?id=1032978
181
182 * UXA: Mark outputs as off before the kernel does (like sna)
183   This will prevent the internal panel from starting up blank in some
184   multi-monitor configurations
185   https://bugzilla.redhat.com/show_bug.cgi?id=1103806
186
187Note that the DRI2 exchange mechanism introduced in 2.99.912 exposes bugs
188in some compositors, at least kwin and comptom, which discard DRI2 buffer
189invalidates rather than resourcing their texture. For example,
190https://bugs.kde.org/show_bug.cgi?id=336589
191
192Note that the improved triple buffering introduced in DRI2 requires a patch
193to Xorg (now released upstream) to prevent crashes with DRI_PRIME.
194https://bugs.freedesktop.org/show_bug.cgi?id=80001
195
196Note that DRI3/Present require tracking the relevant development trees for
197mesa and the xserver as they are very much still under early testing. Also
198be aware that Mesa provides no support for explicit fencing so Damage
199tracking between compositors and clients is unserialised.
200
201
202Snapshot 2.99.912 (2014-06-10)
203==============================
204A final round of features. We have everything from support for variable
205cursor sizes, support for the DRI3 and Present extensions, improved DRI2
206support, support for Xserver 1.16, userptr from kernel 3.16, and
207precursory support for DP multistream transport,
208
209 * Avoid discarding dirty pixels when promoting a migration to cover the
210   whole pixmap.
211   Regression in 2.99.911
212   https://bugs.freedesktop.org/show_bug.cgi?id=77063
213   https://bugs.freedesktop.org/show_bug.cgi?id=77178
214
215 * Avoid overextending degenerate lines (and consequentially accessing
216   pixels outside of our damaged area).
217   https://bugs.freedesktop.org/show_bug.cgi?id=77074
218
219 * Fix subpixel glyph rendering on gen2 devices (830-865 chipsets)
220   Regression in 2.99.911
221   https://bugs.freedesktop.org/show_bug.cgi?id=77201
222
223 * Share the global pixman glyph cache between ZaphodHeads
224   https://bugs.freedesktop.org/show_bug.cgi?id=54707
225
226 * Light up all connected outputs, even if their status is unknown, on
227   takeover from fbcon. This prevents loss of display after a resume
228   on recent kernels, for example.
229   https://bugs.freedesktop.org/show_bug.cgi?id=77768
230
231 * Show the video overlay (when supported by the hardware) across all
232   outputs.
233   https://bugs.freedesktop.org/show_bug.cgi?id=77802
234
235 * Do not discard damage when performing "BLT" spans inplace with the CPU.
236   Regression from 2.20.10
237
238 * Avoid discarding IO buffers too early during their preparation for a
239   new batch
240   https://bugs.freedesktop.org/show_bug.cgi?id=79238
241
242 * Fix fallback handling for displaying large scaled framebuffers (that
243   are too large to be scaled by the GPU in a single pass)
244   https://bugs.freedesktop.org/show_bug.cgi?id=79320
245
246 * Listen to external modifications of backlight value and propagate
247   the notifications to RandR clients. This should make the GUI report
248   ACPI keypresses to change the backlight correctly.
249   https://bugs.freedesktop.org/show_bug.cgi?id=79699
250
251 * UXA: fix pageflips with 3 heads.
252
253 * UXA: do not report a BadMatch error for DRI2GetMsc - as clients are
254   often unprepared and die when they get the unexpected error.
255
256
257Snapshot 2.99.911 (2014-03-19)
258==============================
259Hans de Geode has been working on making the Xserver work without
260privileges under the supervision of systemd/logind. This necessitated a
261few new features for us: server fds (where we are passed which fd to use
262to talk to our device by the Xserver who may in turn receive it from
263logind or other host) and a small backlight helper so that we can continue
264to provide a RandR backlight property when running without root privileges.
265
266 * Flush when changing blend modes on Ironlake, or else single glyphs
267   are sometimes rendered incorrectly
268   Regression from 2.20.15
269   https://bugs.freedesktop.org/show_bug.cgi?id=74882
270
271 * Fix pixmap offsets for pixman fallbacks onto Composite redirected
272   windows
273   https://bugs.freedesktop.org/show_bug.cgi?id=73811
274
275 * Fix blending onto 8-bit destinations, typically used for generating
276   masks in complex Render operations, on gen2
277   https://bugs.freedesktop.org/show_bug.cgi?id=75818
278
279 * Handle failure to create DRI bo more carefully. For example on gen3,
280   the DDX supports pixmaps that are much, much larger than OpenGL can
281   use and do not support the tiling modes that we request for OpenGL.
282   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1289049
283
284 * Fix a bookkeeping bug with proxy buffers that are marked active but
285   not actually inserted into a request (so they end up permanently
286   active and confuse everybody).
287   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1289923
288
289 * Actually turn off displays with DPMS off for UXA.
290   Regression from 2.99.903, but requires
291
292   kernel commit c9976dcf55c8aaa7037427b239f15e5acfc01a3a
293   Author: Chris Wilson <chris@chris-wilson.co.uk>
294   Date:   Sun Sep 29 19:15:07 2013 +0100
295
296       drm/i915: Only apply DPMS to the encoder if enabled
297
298   instead for correct behaviour on Haswell.
299   https://code.google.com/p/chromium/issues/detail?id=341135
300
301
302Snapshot 2.99.910 (2014-02-10)
303==============================
304Another latent bug exposed by recent changes merit another snapshot for
305final testing.
306
307  * Only discard damage when overwriting the dirty CPU bo, instead
308    of discarding damage that will be shown!
309
310  * Reset operation state when switching between glyph caches.
311    https://bugs.freedesktop.org/show_bug.cgi?id=74494
312
313  * Fully reinitialise pixmaps allocated from the freed cache. Fixes
314    a potential issue (crash or misrendering) when using some compositors.
315    https://bugs.freedesktop.org/show_bug.cgi?id=74550
316
317  * Do not expose the TexturedVideo adaptor in UXA when it is disabled
318    either due to a hung GPU or explicitly disabled by the user.
319
320  * Restore the pipe stall when changing CC state on gen6, otherwise
321    the GPU may not flush intermediate results from all EU resulting
322    in render corruption (usually the occasional black box).
323    Regression from 2.99.906
324    https://bugs.freedesktop.org/show_bug.cgi?id=7237
325
326
327Snapshot 2.99.909 (2014-02-01)
328==============================
329Pass the brown paper bag. Ridiculously stupid bug in last minute coding.
330
331  * Add Xv support using glamor acceleration in addition to adaptors
332    provided by the UXA backend.
333
334  * Fix overeager discarding of CPU damage
335    Regression in 2.99.908 :(
336    https://bugs.freedesktop.org/show_bug.cgi?id=74327
337
338
339Snapshot 2.99.908 (2014-01-31)
340==============================
341A couple of regressions dashed the hopes that .907 was to be the final
342release candidate, so time to start the cycle again after applying a few
343more bugfixes.
344
345  * Fix invalid pageflipping of GLXPixmaps by UXA.
346    https://bugs.freedesktop.org/show_bug.cgi?id=73282
347
348  * Consistently treat DPMS Suspend/Standy as Off, otherwise we may
349    incorrectly restore the backlight in UXA.
350
351  * Fix disabling the backlight in UXA when querying the value whilst off
352    https://bugs.freedesktop.org/show_bug.cgi?id=73181.
353
354  * Invalidate between every operation if rendering into the source or mask.
355    Fixes regression in 2.99.907 on Ivybridge (seen in KDE)
356    https://bugs.freedesktop.org/show_bug.cgi?id=73208
357
358  * Check for available batch buffer state before restoring state on gen4,
359    otherwise we overwrite surface state with commands and cause a GPU hang.
360    https://bugs.freedesktop.org/show_bug.cgi?id=73348
361
362  * Prevent an invalid free by TearFree
363    https://bugs.freedesktop.org/show_bug.cgi?id=73469
364
365  * Fix confusion in TearFree when it tried to pageflip to a stale bo
366    https://bugs.freedesktop.org/show_bug.cgi?id=70905
367
368  * Initialise TearFree contents with the current frontbuffer upon creation,
369    otherwise old content may be visible for significant fractions of a
370    second after resume or DPMS on
371    https://bugs.freedesktop.org/show_bug.cgi?id=73842
372
373  * Nullify the old pointer into a CPU mmapping of the bo when promoting it
374    on the GPU. Eventually that old pointer becomes invalid and will cause
375    the Xserver to crash.
376    Fixes regression in 2.99.906
377    https://bugs.freedesktop.org/show_bug.cgi?id=73351
378
379  * Restore the serialNumber on the GC correctly after falling back to
380    software. This was seen to cause corruption with Wine.
381    https://bugs.freedesktop.org/show_bug.cgi?id=73856
382
383  * Do not ignore CPU damage on a pixmap unless the Composite operation
384    will completely overwrite it. This was observed in xfce4.
385    https://bugs.freedesktop.org/show_bug.cgi?id=69528
386
387  * Skip fake outputs during initial probe as they can cause a NULL pointer
388    dereference.
389    https://bugs.freedesktop.org/show_bug.cgi?id=73981
390
391  * Fix rendering of dashed zero-width lines onto 24bit buffers
392    https://bugzilla.redhat.com/show_bug.cgi?id=1059152
393
394  * Cap the freed Pixmap cache and reuse it more often
395    Fixes a regression in 2.99.907 that caused the memory used by X to
396    grow until it was cleaned up during server regreneration
397    https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1272338
398
399
400Snapshot 2.99.907 (2013-12-30)
401==============================
402The highlight here is that things seem to be quietening down on the bug
403reporting front. All is not quiet as you will see below, and maybe it is
404just the holiday season that is approaching*, but there has been a
405remarkable falloff in bug reports. Lets hope this trend holds and we can
406make a stable release shortly!
407
408* This was intended to be sent a couple of weeks ago and the holiday season
409has been very quiet....
410
411Oh, and acceleration support for Intel's next generation of
412integrated processor graphics has landed, codenamed Broadwell.
413
414 * Fix potential X server infinite recursion (crash) from a gen2 bug fix
415   Regression in 2.99.906
416   https://bugs.freedesktop.org/show_bug.cgi?id=71605
417
418 * Workaround a missing pipeline flush within Ivybridge, that would
419   leave black rectangles randomly over the output
420   https://bugs.freedesktop.org/show_bug.cgi?id=68410
421
422 * Fix tiled fills. gen2-3 and gen4+ had two different bugs that
423   both broke rendering with small 8x8 patterns in some circumstances
424   https://bugs.freedesktop.org/show_bug.cgi?id=71260
425
426 * Fix reads from a cropped video image using a packed pixel format.
427
428 * Another clear the clear hint after DRI2 SwapBuffers, like the bug fixed
429   in 2.99.903. Failure to clear the hint would cause read backs of the
430   frontbuffer (Xvnc) to be blank.
431   https://bugs.freedesktop.org/show_bug.cgi?id=72194
432
433 * Disable VSync on Baytrail
434   https://bugs.freedesktop.org/show_bug.cgi?id=69869
435
436 * Handle partial uploads with TearFree correctly
437   https://bugs.freedesktop.org/show_bug.cgi?id=72343
438   https://bugs.freedesktop.org/show_bug.cgi?id=72430
439
440 * Avoid recusing through DRI event handlers whilst processing TearFree,
441   leading to a double free (and memory corruption)
442   https://bugs.freedesktop.org/show_bug.cgi?id=72690
443
444 * Hide the gen4 render corruption by crippling the GPU
445   https://bugs.freedesktop.org/show_bug.cgi?id=55500
446
447
448Snapshot 2.99.906 (2013-11-13)
449==============================
450Several stability fixes required after the recent tweaking of the core
451mechanics to handle the updated TearFree and attempting to make static
452analyzers happy.
453
454 * Fix damage handling when rendering to a partially damaged GPU surface.
455   Regression in 2.99.905
456   https://bugs.freedesktop.org/show_bug.cgi?id=70527
457
458 * Use asprintf() instead of sprintf()
459   Regression in 2.99.905
460   https://bugs.freedesktop.org/show_bug.cgi?id=70835
461
462 * Improve accounting for fence overallocation on older gen2/3, and
463   improve the tiling mechanism to fit into the same aperture constraints
464   https://bugs.freedesktop.org/show_bug.cgi?id=70924
465
466 * Add an extra GPU flush on Sandybridge to fix some rare font corruption
467
468 * Rasterise lines through all clip boxes
469   https://bugs.freedesktop.org/show_bug.cgi?id=70802
470
471 * Fix regression from stricter handling of failures to move a GC to the GPU
472   Regression in 2.99.905
473   https://bugs.freedesktop.org/show_bug.cgi?id=71415
474
475 * Fix various fail along the memcpy_xor paths, including inadequate error
476   handling and integer overflow
477   https://bugs.freedesktop.org/show_bug.cgi?id=70527
478
479 * Fix outside-of-target stipple uploads
480   https://bugs.launchpad.net/bugs/1247785
481
482 * Fix clip detection for long glyphs
483   Incomplete bug fix (causing a regression) in 2.99.905
484   https://bugs.freedesktop.org/show_bug.cgi?id=70527
485
486 * Fix VSync for the render engine (Xv) on Haswell
487   https://bugs.freedesktop.org/show_bug.cgi?id=70527
488
489
490Snapshot 2.99.905 (2013-10-23)
491==============================
492The highlight for this snapshot is the extension of TearFree to support
493transformed outputs, along with some polishing to eliminate its impact
494upon input and output latency. As always, thanks to everyone who have
495been testing, reporting bugs and helping to improve the stability before
496release.
497
498 * Prevent a crash when starting with a user specified mode or position
499
500 * Prevent some crashes in UXA after allocation failure
501
502 * Stop marking the user's preferred backlight value as 0 if the
503   backlight property is queried whilst the connector is disabled
504   https://bugs.freedesktop.org/show_bug.cgi?id=70406
505
506 * Pad GETCONNECTOR ioctl for compatability between 32/64-bit userspace
507   and kernel
508
509 * Handle long glyph runs correctly
510   https://bugs.freedesktop.org/show_bug.cgi?id=70541
511
512 * Fix clipping of stippled rectangles against clip regions
513   https://bugs.freedesktop.org/show_bug.cgi?id=67865
514
515 * Support TearFree rendering of rotated outputs
516   https://bugs.freedesktop.org/show_bug.cgi?id=22969
517
518
519Snapshot 2.99.904 (2013-10-09)
520==============================
521There is one more feature planned to be completed for 3.0, so time for a
522snapshot beforehand to push out the bug fixes from the last week.
523
524 * Fix video output using sprites when changing the image size
525
526 * Apply more restrictive tile constaints for 915g class devices
527   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1232546
528
529 * Ensure all overlapping rectangles are drawn for XRenderFillRectangles
530   https://bugs.freedesktop.org/show_bug.cgi?id=66313
531
532 * Fix trapezoid clipping against the left-edge
533   https://bugs.freedesktop.org/show_bug.cgi?id=69469
534
535 * Prevent discarding active upload buffers, causing glitches in chromium
536   https://bugs.freedesktop.org/show_bug.cgi?id=66990
537
538 * Prevent specifying a negative timeout to select if the BlockHandler
539   takes too long to update the display
540
541 * Promote the Ironlake pipecontrol to be a full pipeline flush to
542   prevent render cache corruption
543   https://bugs.freedesktop.org/show_bug.cgi?id=51422
544
545 * Never pass an invalid trapezoid to pixman
546   https://bugs.launchpad.net/ubuntu/+source/xorg-server/+bug/1197921
547
548 * Prevent out-of-bounds access by overassigning work amongst threads
549   https://bugs.freedesktop.org/show_bug.cgi?id=70204
550
551 * Make sure the current mode is always listed amongst the output modes
552   https://bugs.freedesktop.org/show_bug.cgi?id=70132
553
554 * Build fixes for 1.14.99.2
555
556
557Snapshot 2.99.903 (2013-09-28)
558==============================
559Lots more stabilization work, not yet peaceful enough to christen 3.0. We
560have everything ranging from build fixes for systems like Suse Linux
561Enterprise Desktop that like to backport hardware enablement to ancient
562packages, to fixes for hardware enabling, and some more performance tuning.
563
564 * Fix VSync on Haswell.
565   https://bugs.freedesktop.org/show_bug.cgi?id=69119
566
567 * Disable Y-tiling on gen4 - it too frequently leads to instability.
568   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1222203
569
570 * Disable same EDID detection based on property id - the kernel likes
571   to reuse ids leading to a failure to detect the right modes on a monitor
572   change.
573
574 * Avoid issuing multiple DPMS requests to the same encoder (alised to
575   multiple connectors) to avoid upsetting Haswell and leaving the
576   screens blank.
577
578 * Honour the user preferrence for the initial mode, even if they are
579   being silly
580   https://bugzilla.novell.com/show_bug.cgi?id=841696
581
582 * Clear the clear hint when apply DRI updates - to prevent some
583   screenshots from GL windows being left blank.
584   https://bugs.freedesktop.org/show_bug.cgi?id=69730
585
586 * Prevent a NULL dereference from trying to undo an non-existent buffer
587   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1228677
588
589 * Handle out-of-memory conditions far more gracefully. If the system is
590   hard against the memory wall, then the kernel will start issuing
591   SIGBUS even for CPU mmaps. Untrapped these will cause X to die.
592   https://bugs.freedesktop.org/show_bug.cgi?id=67889
593
594
595Snapshot 2.99.902 (2013-09-07)
596==============================
597We do not condone or support Canonical in the course of action they have
598chosen, and will not carry XMir patches upstream.
599-The Management
600
601Snapshot 2.99.901 (2013-09-04)
602==============================
603What's this? A pre-release snapshot of an upcoming 3.0 major release?
604What could be the big new feature about to land? SNA by default!
605Not only that, we also have a preview of the XMir integration patches.
606
607Whilst we consider the SNA acceleration method to be ready to take over
608from UXA (the current acceleration method), it still remains possible to
609switch over to the old method with the following xorg.conf snippet:
610  Section "Device"
611    Identifier "Device0"
612    Driver "intel"
613    Option "AccelMethod" "UXA"
614  EndSection
615
616The promise of SNA is that it improves the user experience by providing a
617faster, more fluid, desktop, that is both more power efficient and less
618prone to crashes than UXA. If it still fails to meet your expections,
619please file bugs!
620
621
622Release 2.21.15 (2013-08-21)
623============================
624Some build fixes for the BSDs and alternate compiler, and conflicting
625configure options and a critcial fix for some PRIME setups.
626
627 * Don't fail to configure if DRI1 is not available and the user asks
628   for both DRI and KMS-only
629   [Regression from 2.21.14]
630
631 * Lots of miscellaneous fixes for older gcc, other compilers and BSD.
632
633 * Initial framework support for hosted X.
634
635 * Improve transition from fbcon to X with multiple outputs and extended
636   desktops.
637
638 * Ensure the framebuffer exists before checking it against the CRTC
639   constraints.
640   https://bugs.launchpad.net/ubuntu/+source/xorg-server/+bug/121234
641   [Regression from 2.21.10]
642
643 * Add RGB overlay support for Ironlake and later.
644
645 * Don't release the overlay buffer too early, i.e. before the client
646   disables the Xv image.
647
648 * Tweak the ring switching logic to reduce use of semaphores on
649   Sandybridge and later - a delicate balancing act between trying to
650   use the faster engine for the task and preventing concurrent use of
651   multiple rings.
652
653
654Release 2.21.14 (2013-08-04)
655============================
656A few updates for the top-end Haswell systems (notably GT3 and GT3e
657systems), in particular, setting the appropriate thread counts for the
658larger GPU and handling a few scaling issues with the emedded DRAM. More
659important than the performance tuning though is a fix for a critical
660issue encountered by a few people on gen4/gen5 systems.
661
662 * Initial performance tuning for HSW:GT3e
663
664 * Fix a crash with Planetary Annihilation
665
666 * Disable triple buffering for compositors
667
668 * Reserve space in the exec buffer array for deferred VBOs.
669   https://bugs.freedesktop.org/show_bug.cgi?id=67504
670
671
672Release 2.21.13 (2013-07-27)
673============================
674A minor release to repair the build for non-Linux systems, and to undo a
675few more regressions.
676
677 * Avoid potential memory corruption with allocations of very small
678   depth 1 bitmaps.
679
680 * Fix source clipping whilst computing copy extents and exposures.
681   [Regression from 2.21.12]
682   https://bugs.freedesktop.org/show_bug.cgi?id=66970
683
684 * Use /proc/cpuinfo if cpuid4 is not available (old hardware or old
685   gcc), and repair the build on systems without cpuid.
686   [Regression from 2.21.12]
687
688 * Fix performance regression on Ironlake from inadvertently flushing
689   after every operation.
690   [Regression from 2.21.11]
691   https://bugs.freedesktop.org/show_bug.cgi?id=67157
692
693 * Fix conflict handling when probing initial KMS configuration for
694   ZaphodHeads.
695   [Regression from 2.21.11]
696   https://bugs.freedesktop.org/show_bug.cgi?id=67176
697
698 * Stop being overly restrictive and rejecting stale DRI2 buffers.
699   (DRI2 is inherently racy in that the client may have completed and
700   submitted rendering to buffers that are now invalid on the server.
701   This race is magnified by bugs within the Xserver where it forgets
702   to notify the DRI2 clients of certain invalidation events.)
703   https://bugs.freedesktop.org/show_bug.cgi?id=67210
704
705 * Handle a failure to use a GTT mmap for a pixmap upload and try
706   an alternative method before giving up.
707
708
709Release 2.21.12 (2013-07-14)
710============================
711In this release, we clear up the teething troubles from preserving the KMS
712configuration, notably external connections on Haswell and plugging in new
713outputs after startup were broken. Besides these regression fixes, there
714are a couple of fixes for some long standing issues, such as incorrect
715rendering on gen2, an infinite loop with very, very large pixmaps and a
716slight improvement to the tempermental gen4.
717
718 * Allow untiled scanouts again (required for large extended desktops on
719   gen2 and gen3).
720   [Regression from 2.21.11]
721
722 * Use the correct count of the number of dirty damage boxes for the quick
723   check on whether the existing damage contains the requested area.
724   The danger is that we may get a false result and skip migration and so
725   cause pixmap corruption (in the unlikely event that the application
726   frequently causes fallbacks).
727   [Regression from 2.21.11]
728   https://bugs.freedesktop.org/show_bug.cgi?id=66430
729
730 * Fix initial connection probing for multi-function encoders, such as the
731   external connections on Haswell.
732   [Regression from 2.21.11, initial connection probing]
733   https://bugs.freedesktop.org/show_bug.cgi?id=66488
734
735 * Fix gen2 rendercopy into a8 surfaces, for example, glyph uploads into
736   the glyph cache.
737
738 * Fix detection of user overrides for initial connection configuration.
739   The code used the xorg-server-1.15 values, having missed the
740   introduction of ZoomModes into that release.
741   [Regression from 2.21.11, initial connection probing]
742
743 * Always initialise the gamma ramp, even on unconnected CRTCs.
744   https://bugs.freedesktop.org/show_bug.cgi?id=66563
745   [Regression from 2.21.11, initial connection probing]
746
747 * Some more tuning of the gen4 vertex corruption workaround. The root
748   cause behind the GPU using incorrect texture coordinates is still
749   not solved, but by reducing the maximum number or rectangles in
750   flight through the GPU we reduce the likelihood of corruption.
751
752 * Fix compilation with gcc-4.5
753   [Regression from 2.21.11]
754
755 * Avoid integer overflow when performing tiled uploads and operations
756   on very large (>28k pixels wide or tall pixmaps)
757   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1200766
758
759
760Release 2.21.11 (2013-06-30)
761============================
762An eventful week. What started with a regression with some builds of
763firefox on some machines lead ultimately to the discovery of an older
764kernel bug. Aside from the work to fix the image bug and a few other
765older bugs that were reported and resolved this week, there is also a
766(hopefully) subtle change to the initial configuration of displays. In
767the absence of user overrides in xorg.conf, the DDX will try to preserve
768the same display configuration as used by the kernel, which hopefully
769will be the same configuration as setup by the BIOS. The result should
770be a boot sequence that does not resize at all (aka fastboot) - until
771the display manager takes over and loads a completely different
772configuration!
773
774 * Add reference counting of drmMaster for ZaphodHeads
775   https://bugs.freedesktop.org/show_bug.cgi?id=66041
776
777 * Add a GPU flush before changing blend modes on Ironlake
778   https://bugs.freedesktop.org/show_bug.cgi?id=51422
779
780 * Fix occasional missing images for inplace uploads
781   [regression from 2.21.10]
782   https://bugs.freedesktop.org/show_bug.cgi?id=66059
783
784 * Add missing utility files to the tarball and remove a few unused ones
785
786 * Initialise PolyPoint operand state before calling miWideDash
787   https://bugs.freedesktop.org/show_bug.cgi?id=66104
788
789 * Fix redirection handling for rendering into large surfaces
790   https://bugs.freedesktop.org/show_bug.cgi?id=66168
791   https://bugs.freedesktop.org/show_bug.cgi?id=66249
792
793 * Fix compilation of UXA with xorg-xserver < 1.10
794   [regression from 2.20.0]
795
796 * Fix consideration of gradients for deciding when to migrate render
797   operations
798   [performance regression from 2.21.10, the bug itself is older]
799   https://bugs.freedesktop.org/show_bug.cgi?id=66297
800
801Also fixed this week was:
802
803commit 22fd5ca947b58901927d100d2b1aa0f1672b3435
804Author: Chris Wilson <chris@chris-wilson.co.uk>
805Date:   Fri Jun 28 16:54:08 2013 +0100
806
807    drm/i915: Only clear write-domains after a successful wait-seqno
808
809which affects kernels 3.7 - 3.10, coming to a stable kernel near you soon.
810
811
812Release 2.21.10 (2013-06-22)
813============================
814Fixes missing support for Xv (with the textured video adaptor) on
815Haswell, and an old bug with wide monitors on various generation. Along
816with a few other regression fixes and performance tweaks for various
817corner cases.
818
819 * Do not lose track of fast pageflips across mode changes
820   [regression from 2.20.8]
821
822 * Fix listing of Visuals for Xv
823   [regression from 2.21.8]
824   https://bugs.freedesktop.org/show_bug.cgi?id=65479
825
826 * Improve coherency of concurrent CPU accesses to a pixmap
827   https://bugs.freedesktop.org/show_bug.cgi?id=61628
828
829 * Set sampler swizzling for textured video on Haswell
830   https://bugs.freedesktop.org/show_bug.cgi?id=65699
831
832 * Apply scanout stride limits
833   https://bugs.freedesktop.org/show_bug.cgi?id=65099
834
835 * Undo the self-copy for cloned pixmaps for loimpress's animations
836   [regression from 2.21.7]
837   https://bugs.freedesktop.org/show_bug.cgi?id=65665
838
839
840Release 2.21.9 (2013-06-06)
841===========================
842Consolidating the copy-on-write support, hopefully cleaning up the last of
843the regressions.
844
845 * Restore vsync on textured videos.
846   [regression from 2.21.8]
847   https://bugs.freedesktop.org/show_bug.cgi?id=65048
848
849 * Fix incorrect ordering of possible_clones with certain outputs, which
850   can lead to attempting to incorrectly clone 2 outputs and failing to
851   light them up.
852   [regression from 2.20.10]
853
854 * Fix performance regression from not promoting large fills to the GPU
855   [regression from 2.21.7]
856
857 * Undo the pixmap clone before performing a DRI2CopyRegion
858   [regression from 2.21.7]
859   https://bugs.freedesktop.org/show_bug.cgi?id=65250
860
861
862Release 2.21.8 (2013-05-27)
863===========================
864A quick release to cleanup a few regressions from the introduction of
865copy-on-write support, notably hitting wine applications and a memory
866leak for firefox.
867
868 * Only mark a PolyFillRect operation as replacing if it is unclipped
869   https://bugs.freedesktop.org/show_bug.cgi?id=64841
870
871 * Prevent potential NULL dereference of damage when checking COW support
872
873 * Fix invalidation of clone after dirtying the pixmap via the CPU
874
875 * Prevent discarding an operation before requiring it for a fallback
876
877 * Fix memory leak from replacing the clone under certain circumstances
878   https://bugs.freedesktop.org/show_bug.cgi?id=64978
879
880
881Release 2.21.7 (2013-05-21)
882===========================
883A couple of weeks turned into a month and a couple of weeks... Amidst
884the usual bug fixes, we have added the complete set of Haswell PCI IDs -
885hopefully future proofing ourselves against being surprised by new
886products. We can also now use the correct term for the top of the range
887Haswell variants, GT3.
888
889 * Fix several assertion failures hit by Jiri Slaby.
890
891 * Allow XvMC to also target overlay/sprite planes.
892
893 * Throw in a paranoid MI_FLUSH between BLT and RENDER operations on
894   Ironlake.
895   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1168066
896
897 * Prevent reuse of old framebuffers after a resize.
898   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1157678
899
900 * Fix compilation with --enable-valgrind and no --enable-debug
901
902 * Improve partial migration of render sources.
903
904 * Fix origin of trapezoids.
905   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1178020
906
907 * Introduce copy-on-write support for cloning pixmaps. The ultimate
908   goal here is to efficiently support the TearFree mode of operation,
909   but this provides immediate benefits with firefox - most importantly
910   because of the inefficient way it now implements scrolling.
911
912
913Release 2.21.6 (2013-04-06)
914===========================
915A surprising highlight of this release is a little refresh to the KMS
916support for OpenBSD. OpenBSD now has its own KMS implementation which is
917mostly compatible with the interface in Linux, with one or two tweaks
918supplied by Mark Kettenis. This release continues to cleanup behaviour
919for Haswell.
920
921 * Workaround a failure by the xserver to invalidate DRI buffers
922   following a pixmap change for XComposite redirection.
923   https://bugs.freedesktop.org/show_bug.cgi?id=62614
924
925 * Fix computation of clip extents for stippling
926   https://bugs.freedesktop.org/show_bug.cgi?id=62618
927
928 * Support KMS on OpenBSD, by Mark Kettenis
929
930 * Clean up sockets upon CloseScreen (making ourselves better behaved
931   for muxed setups).
932
933 * Fix the tests for AVX/AVX2 support in CPUID and remember to check for
934   OS support as well.
935
936 * Report a monotonic UST value for undisplayed drawables rather than 0
937   by Daniel Kurtz
938
939 * Fix video playback on gen4 through a complex clip (more gen4 GPU woes)
940   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1162046
941
942
943Release 2.21.5 (2013-03-21)
944===========================
945Haswell reintroduces a command to load the scanline window from the
946command stream and so requires its own specialised wait-for-vsync routine
947- failure to do so was then causing hangs when trying to do tearfree video
948or use a compositor.
949
950 * Prevent buffer leak if a non-fullscreen Window is closed with multiple
951   pending swap events.
952
953 * Fix offset transformation for fallback gradient paths.
954   https://bugs.freedesktop.org/show_bug.cgi?id=62198
955
956 * Prevent Glamor from crashing if misconfigured.
957   Thanks to Michel Dänzer.
958
959 * Prevent UXA from crashing if torn down during PreInit.
960   Thanks to Aaron Plattner.
961
962 * Prevent miscompilation with different functional units having different
963   compiler flags. Some functions were expected to be inlined and so
964   recompiled with the current target. However, some compilers were
965   choosing to emit subroutine calls instead without noticing that the
966   ABI was different between the caller and callee - causing corruption.
967   https://bugs.freedesktop.org/show_bug.cgi?id=62198
968
969 * Fix rendering of CompositeTriFan with recent Xorg.
970
971 * Apply the video src-offset fix highlighted in the last release!
972   A typo prevented the fix from working for gen4+.
973   https://bugs.freedesktop.org/show_bug.cgi?id=62343
974
975 * Fix rendering of multiple glyphs to very large destination surfaces
976   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1156387
977
978 * Fix scanline waits for Haswell
979   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1156679
980
981
982Release 2.21.4 (2013-03-11)
983===========================
984More bugs, more fixes, more releases. A minor new feature being introduced
985is the runtime detection of CPU instructions sets along with specialised
986paths to take advantage of the available CPU.
987
988 * Honour LinearFramebuffer for clumsy PowerXpress integration
989
990 * Disable read-read optimisations of mappings whilst the root
991   cause of a corruption issue remains elusive.
992   https://bugs.freedesktop.org/show_bug.cgi?id=61628
993
994 * Disable 8-bpp framebuffers in UXA as the regressions therein
995   remain unfixed.
996
997 * Restart vertex checks after lock contention
998   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1124576
999
1000 * Handle Screen Pixmap recreation whilst exported via PRIME.
1001
1002 * Correct application of scale factors to video source offsets.
1003   https://bugs.freedesktop.org/show_bug.cgi?id=61610
1004
1005 * Chain up CloseScreen so that resources are actually freed across regen
1006   https://bugs.freedesktop.org/show_bug.cgi?id=56608
1007
1008 * Fix Haswell CRW PCI-IDs
1009
1010 * Handle batch submission failure during DRI copies
1011   https://bugs.freedesktop.org/show_bug.cgi?id=61708
1012
1013 * Probe for kernel support of requested Screen depth
1014   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1135403
1015
1016 * Correct GPU limits for early gen2 and gen3 architectures and prevent
1017   an infinite recursion for particular image sizes.
1018
1019
1020Release 2.21.3 (2013-02-20)
1021===========================
1022A few minor bugfixes, another point release.
1023
1024 * Fix tracking of DRI pixmaps and their backing bo across reparenting. If
1025   we tried to execute a SwapBuffers after a Window was reparented, but
1026   before the DRI client has updated its references, then we would end up
1027   manipulating an exported pixmap without a flush flag set. In the worst
1028   case, this would culminate in a segfault in the driver.
1029   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1127497
1030
1031 * Restore the gen4 workarounds for flickering rendering - a few cases still
1032   remain, as the root cause persists.
1033   https://bugs.freedesktop.org/show_bug.cgi?id=60402
1034
1035 * Double check that the device has KMS enabled before claiming. This allows
1036   X to gracefully fallback to VESA/fbdev rather than bailing out.
1037   https://bugs.freedesktop.org/show_bug.cgi?id=60987
1038
1039 * Fix the UXA render programs for projective transforms on Ivybridge.
1040
1041
1042Release 2.21.2 (2013-02-10)
1043===========================
1044Pass the brown paper bags, I need half a dozen or so. That seemingly
1045innocuous build fix with xorg-1.13 happened to have the little side-effect
1046of breaking glyph rendering with xorg-1.12 and older on 64-bit machines.
1047
1048
1049Release 2.21.1 (2013-02-10)
1050===========================
1051A fix for a potential GPU hang on 945gm (GMA950) and earlier chipsets,
1052along with backporting SNA to the packages found in stable distributions
1053like Debian 6.0 (Squeeze).
1054
1055 * Cleanup compilation warnings from deblint, thanks to Paul Menzel
1056
1057 * Minor build improvements by Damien Lespiau.
1058
1059 * Disable generating span geometry for non-rectilinear spans on gen4
1060   in order to work around and prevent one class of render corruption.
1061
1062 * Prevent cache thrashing and severe performance degradation on LLC
1063   machines for streaming texture updates. However, note the effect was
1064   only observed on just one particular laptop.
1065
1066 * Fix alignment of subsurface proxies for old chipsets.
1067   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1120108
1068
1069 * Repair build against Xserver-1.6 and contemporary packages.
1070
1071
1072Release 2.21.0 (2013-02-01)
1073===========================
1074A few new features:
1075
1076 * Enable render acceleration for Haswell GT1/GT2.
1077
1078 * Enable multi-threaded rasterisation of trapezoids and fallback composition
1079
1080 * Utilise a new kernel interface (v3.9) for processing relocations
1081
1082along with a few older features from the 2.20.x series:
1083
1084 * PRIME support for hotplug GPUs and hybrid systems
1085
1086 * Support for IvyBridge GT1 machines, aka HD2500 graphics.
1087
1088 * Stable 830gm/845g support, at last!
1089
1090As usual we have a large number of bug fixes since the last release:
1091
1092 * Prevent a stray relocation being left after a buffer is removed from
1093   a batch, leading to GPU hangs.
1094
1095 * Make the driver more robust against its own failures to submit batches
1096   by falling back to software rendering.
1097
1098 * Fix emission of scanline waits for secondary pipes on gen6/7. Otherwise
1099   you may encounter GPU hangs in MI_WAIT_FOR_EVENT.
1100
1101 * Fix a missing corner pixel when drawing rectangles with PolyLines
1102   https://bugs.freedesktop.org/show_bug.cgi?id=55484
1103
1104 * Don't try to use Y-tiling colour buffers with mesa/i915c as mesa
1105   doesn't support them and wil fallback to software rendering
1106
1107 * Ensure that any cached mmaps are invalidated for a SwapBuffers
1108   https://bugs.freedesktop.org/show_bug.cgi?id=60042
1109
1110 * Correctly handle the composition of rotated displays too large for the
1111   3D pipeline
1112   https://bugs.freedesktop.org/show_bug.cgi?id=60124
1113
1114 * Fix the computation of the planar video frame size
1115   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1104180
1116
1117
1118Release 2.20.19 (2013-01-20)
1119============================
1120A quick release as the last broke USB DisplayLink slave outputs badly. The
1121performance of those displays was unusable due to an inadvertent change that
1122caused us to flush the entire scanout over the USB for every drawing
1123operation.
1124
1125 * Implement the GNOME Build API. A couple of minor changes to make
1126   integrators and distributors lives a little easier, or at least more
1127   consistent.
1128
1129 * Correctly offset inplace trapezoids for subwindows, such as the GTK+
1130   close button after it has a background image uploaded.
1131
1132 * Explicitly prevent ring-switching for synchronized rendering to
1133   scanouts (for vsync).
1134
1135 * Clip dirty region to slave pixmaps (otherwise UDL is nigh unusuable)
1136   https://bugs.freedesktop.org/show_bug.cgi?id=59539
1137
1138
1139Release 2.20.18 (2013-01-16)
1140============================
1141A bunch of miscellaneous fixes for assertion failures and various
1142performance regressions when mixing new methods for offloads, along with
1143a couple of improvements for rendering with gen4.
1144
1145 * Remove use of packed unnormalized texture coordinates on gen4/5 as
1146   these GPUs do not support unnormalized coordinates in the sampler.
1147
1148 * Remove dependency upon x86 asm for cross-building to unsupported
1149   architectures.
1150   https://bugs.gentoo.org/show_bug.cgi?id=448570
1151
1152 * Apply damage around PRIME updates in the correct order.
1153
1154 * Correctly read the initial backlight level for when the user
1155   overrides UXA's choice of backlight controller.
1156
1157 * Throttle UXA and prevent it queuing work much faster than the GPU can
1158   complete it. This realised itself in impossible performance figures and
1159   the entire display freezing for several seconds whlist the GPU caught
1160   up. One side effect is that it also caused the DDX to consume more
1161   memory than was required as it could not recycle buffers quick
1162   enough, and in some cases this produces a marked improvement in
1163   performance. Also note on gen2/3 this requires a new libdrm [2.4.41]
1164   in order to prevent a bug causing the DDX to fallback to swrast.
1165
1166Release 2.20.17 (2012-12-26)
1167============================
1168A minor update to prepare for co-operating with the kernel over managing
1169stability on 830gm/845g. On this pair of chipsets, the kernel will perform
1170an extra copy of the batchbuffer into reserved memory, which prevents them
1171from randomly dying. However, that extra copy does have a noticeable
1172impact upon throughput, so we also have a mechanism for userspace to
1173opt-out of the kernel workaround and take responsibility for ensuring its
1174batches are coherent.
1175
1176 * Build fixes against xorg-1.14
1177   https://bugs.freedesktop.org/show_bug.cgi?id=58552
1178   https://bugs.freedesktop.org/show_bug.cgi?id=58406
1179
1180 * Fixed the origin of cropped (textured) video windows (Xv and XvMC)
1181   https://bugs.freedesktop.org/show_bug.cgi?id=23033
1182
1183 * Fix potential corruption when using images larger than ~1GiB
1184
1185
1186Release 2.20.16 (2012-12-15)
1187============================
1188Rejoice! We have found a trick to make 830gm/845g stable at long last.
1189Ever since the switch to GEM and dynamic video memory, those early
1190second generation chipsets have been plagued by instability. The lack of
1191flushing cachelines from the CPU to GMCH was eventually solved by using
1192an undocmented bit, but 830/845 were still hanging under memory pressure.
1193These deaths were all due to garbage finding its way into the command
1194streamer, and they go away if we take a leaf out of the original driver
1195and never reuse those pages for anything else. So for the first time
1196ever, I have been able to complete running the test suite on an 845g,
1197even whilst thrashing the page and buffer caches!
1198
1199 * Run the SF stage as single-threaded on gen4 to workaround a few issues
1200   https://bugs.freedesktop.org/show_bug.cgi?id=57410
1201
1202 * Keep the scanout SURFACE_STATE separate to avoid overriding its
1203   memory access control on gen6/7 (i.e. writes to the scanout need to
1204   be kept out of the render cache)
1205
1206 * Tune batch flushing after an operation to an exported surface under a
1207   compositor.
1208
1209 * Make sure the source is on the CPU for inplace composition of trapezoids
1210   using the CPU
1211   https://bugs.freedesktop.org/show_bug.cgi?id=56825
1212
1213 * Immediately flush in the block hander after a split batch to reduce
1214   latency between the two halves of an operation.
1215   https://bugs.freedesktop.org/show_bug.cgi?id=51718
1216
1217 * Install a fallback config if we fail to install the desired config
1218   at VT switch (i.e. booting, after resume with 3 incompatible pipes on
1219   Ivybridge)
1220
1221 * Pin batches to avoid CS incoherence on 830/845
1222   https://bugs.freedesktop.org/show_bug.cgi?id=26345
1223
1224
1225Release 2.20.15 (2012-12-03)
1226============================
1227And lo, enabling more of the common acceleration paths for gen4 revealed
1228another lurking bug - something is wrong with how we prepare Y-tiling
1229surfaces for rendering. For the time being, we can surreptiously disable
1230them for gen4 and avoid hitting GPU hangs.
1231
1232 * Avoid clobbering the render state after failing to convert the
1233   operation to use the blitter.
1234   https://bugs.freedesktop.org/show_bug.cgi?id=57601
1235
1236 * Disable shadow tracking upon server regeneration, and so fix a crash
1237   if you restart the server whilst a RandR transform (e.g. rotation) is
1238   in effect.
1239   https://bugs.freedesktop.org/show_bug.cgi?id=52255
1240   https://bugs.freedesktop.org/show_bug.cgi?id=56608
1241
1242
1243Release 2.20.14 (2012-11-26)
1244============================
1245The highlight of this release is gen4, from 965g to gm45. Quite an old
1246bug surfaced in the shader assembly, sparking a chance to review a few
1247design choices within that backend and experiment on fresh ways to
1248workaround the remaining issues.
1249
1250 * Avoid using inplace XOR'ed uploads for very large buffers
1251   https://bugs.freedesktop.org/show_bug.cgi?id=57031
1252
1253 * Fix the gen4/5 opacity shader
1254   https://bugs.freedesktop.org/show_bug.cgi?id=57054
1255
1256 * Queue a pending vblank request after flip completion
1257   https://bugs.freedesktop.org/show_bug.cgi?id=56423
1258
1259 * Avoid migrating an uninitialised pixmap for use as a render source
1260   https://bugs.freedesktop.org/show_bug.cgi?id=47597
1261
1262 * Improve handing of texture fallbacks for 830/845.
1263   https://bugs.freedesktop.org/show_bug.cgi?id=57392
1264
1265
1266Release 2.20.13 (2012-11-11)
1267============================
1268Nothing but bug fixes. Many thanks to everyone who took the time to
1269report their issues, and for their help in improving the driver.
1270
1271 * Sanity check the platform probe points to our expected i915 device
1272   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1069031
1273
1274 * Prevent 16-bit overflow for computing the sample area to upload of
1275   sources for render operations
1276   https://bugs.freedesktop.org/show_bug.cgi?id=56324
1277
1278 * Clamp the drawable box for migration to prevent 16-bit overflow
1279   https://bugs.freedesktop.org/show_bug.cgi?id=56591
1280
1281 * Disable RandR hotplug events if Xinerama is enabled and thereby prevent
1282   a crash upon hotplug
1283   https://bugs.freedesktop.org/show_bug.cgi?id=55260
1284
1285 * Call ValidatePicture before attempting to flatten the alphamaps
1286   https://bugs.freedesktop.org/show_bug.cgi?id=56367
1287
1288 * Clip the trapezoid correctly if it ends on the boundary pixel
1289   https://bugs.freedesktop.org/show_bug.cgi?id=56395
1290
1291 * Make sure the pipeline choice is propagated to the scanline wait
1292   across a batch flush
1293   https://bugs.freedesktop.org/show_bug.cgi?id=47597
1294
1295 * Set the valid drawable box when choosing placement of BLT composite ops
1296   https://bugs.freedesktop.org/show_bug.cgi?id=47597
1297
1298 * Prevent use-after-free when promoting a partial-GPU bo to a full-GPU bo
1299   https://bugs.freedesktop.org/show_bug.cgi?id=56591
1300
1301 * gen4 opacity spans require the per-rectangle workaround
1302   https://bugs.freedesktop.org/show_bug.cgi?id=55500
1303
1304 * Prevent use of invalid damage pointers when redirecting rendering
1305   https://bugs.freedesktop.org/show_bug.cgi?id=56785
1306
1307
1308Release 2.20.12 (2012-10-20)
1309============================
1310More bug reports, more bug fixes! Perhaps the headline feature is
1311that with a secure batches, coming to a 3.8 kernel near you, we may
1312finally have the ability to perform updates to the scanout synchronized
1313to the refresh rate on later SandyBridge and IvyBridge chipsets. It comes
1314at quite a power cost as we need to keep the GPU out of its power saving
1315modes, but it should allow legacy vsync to function at last. But this
1316should allow us to address a longstanding issue with tearing on
1317SandyBridge+.
1318
1319 * Fix component-alpha rendering on IvyBridge, for example subpixel
1320   antialiased glyphs.
1321   https://bugs.freedesktop.org/show_bug.cgi?id=56037
1322
1323 * Flush before some "pipelined" state changes on gen4. The evidence is
1324   that the same flushes as required on gen5+ are also required for gen4.
1325   https://bugs.freedesktop.org/show_bug.cgi?id=55627
1326
1327 * Prevent a potential crash when forcing a stall on a busy CPU bo
1328   https://bugs.freedesktop.org/show_bug.cgi?id=56180
1329
1330[Release 2.20.11 contained a typo causing UXA to fail immediately.]
1331
1332Release 2.20.10 (2012-10-14)
1333============================
1334The last couple of weeks have been fairly retrospective, a dive into
1335prehistory tidying up the earlier generations which lay languishing as
1336the core progressed and lead to a number of annoying core bugs being
1337fixed.
1338
1339 * Release DRM master earlier during shutdown so switching between
1340   multiple X servers works automatically.
1341   https://bugs.freedesktop.org/show_bug.cgi?id=55446
1342
1343 * Suppress error propagation from DRI2GetMSC and behave as if the pipe
1344   was simply off to avoid unexpected errors in the clients
1345   https://bugs.freedesktop.org/show_bug.cgi?id=55395
1346
1347 * A few fixes to i8xx batch emission, ensuring that the GPU is always
1348   in a valid state.
1349   https://bugs.freedesktop.org/show_bug.cgi?id=55455
1350
1351 * Prevent a use-after-free during UXA shutdown due to inspecting
1352   the glamor flags after the glamor interface had been freed.
1353
1354 * Prevent a crash combining TearFree and rotations.
1355   https://bugs.freedesktop.org/show_bug.cgi?id=55527
1356
1357 * Correct a missing damage upload along PutImage after using the CPU bo
1358   as a source for the GPU.
1359   https://bugs.freedesktop.org/show_bug.cgi?id=55508
1360
1361 * Fix compilation for older glibc without O_CLOEXEC
1362   https://bugs.freedesktop.org/show_bug.cgi?id=55577
1363
1364 * Fix out-of-tree builds failing to recompile the gen4 assemblies
1365   https://bugs.freedesktop.org/show_bug.cgi?id=55645
1366
1367 * Fix non-standard build host configuration handling for intel-gen4asm
1368   https://bugs.freedesktop.org/show_bug.cgi?id=55646
1369
1370 * Fix a potential batch buffer overflow when replacing the last BLT fill
1371   operation with a copy
1372   https://bugs.freedesktop.org/show_bug.cgi?id=55700
1373
1374 * Flush the render pipeline more frequently on Ironlake as not all
1375   pipelined state changes are.
1376   https://bugs.freedesktop.org/show_bug.cgi?id=51422
1377
1378 * Detect when we need to read the destination for the background raster
1379   op during fallbacks.
1380   https://bugs.freedesktop.org/show_bug.cgi?id=55810
1381
1382 * Avoid a potential deference of an invalid CPU mmap after doing an
1383   inplace tiled upload.
1384   https://bugs.freedesktop.org/show_bug.cgi?id=55812
1385
1386 * Prevent sign extension when packing the upload data for CopyPlane
1387   https://bugs.freedesktop.org/show_bug.cgi?id=55823
1388
1389 * Fix some render corruption with a UDL slave output and pageflipping
1390
1391
1392Release 2.20.9 (2012-09-29)
1393===========================
1394And so it came to pass that a critical bug was uncovered in UXA. The
1395kernel does not like to pageflip when the pipe is off, yet due to the
1396delayed nature of a pageflip and the relaxed checking performed by UXA,
1397we could request a pageflip after turning off the display (DPMS). The
1398kernel rejected that pageflip and the error handling path failed to
1399restore sanity, and when the screen came back it was stuck on the image
1400seen before it went to sleep. (Note that there are also some related
1401kernel bugs, but this update should prevent the most conspicious of the
1402freezes.) Many thanks to Timo Aaltonen for his efforts in tracking down
1403the issue.
1404
1405In other news:
1406
1407  * Prepare for xorg-1.14, the api is being tweaked again.
1408
1409  * Handle early FreeScreen in UXA.
1410    https://bugs.freedesktop.org/show_bug.cgi?id=55346
1411
1412  * Reenable XvMC support
1413
1414  * Do not replace the GPU bo when uploading into the shadow/CPU copy
1415    https://bugs.freedesktop.org/show_bug.cgi?id=54978
1416
1417  * Fix use of an uninitialised GC when drawing glyphs to a depth=1 pixmap
1418
1419
1420Release 2.20.8 (2012-09-16)
1421===========================
1422Another new small feature, another new release. And a few more bugs
1423fixed as well! But what is this new feature, do I hear you ask? Why, it
1424is nothing less than enabling the ValleyView SDV! The lucky person to
1425have their hands on one will now be able to enjoy X in full TechniColor.
1426
1427For the rest of us, a few more bugs were fixed with interesting
1428combinations of software and rendering patterns.
1429
1430 * Add an extra layer of defence against trying to use a non-GEM device
1431   with UXA. This should already be taken care of with the new probe,
1432   but the extra sanity check already existed in the code but was doing
1433   nothing.
1434   https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/962892
1435
1436 * Fix computation of valid CRTCs bitmask for ZaphodHead
1437   Some systems can only handle certain outputs on certain pipes
1438   (Screens), and this information was not being propagated through to X
1439   and randr correctly, breaking valid configurations and not detecting
1440   invalid configurations correctly.
1441
1442 * Disable global glyph caching with ZaphodHeads.
1443   The glyph privates need to be Screen private, but at the moment are
1444   global leading to conflicts and invalid rendering with multiple heads.
1445   https://bugs.freedesktop.org/show_bug.cgi?id=54707
1446
1447 * Prevent direct read back of unmappable buffers
1448   https://bugs.freedesktop.org/show_bug.cgi?id=54808
1449
1450 * Tile large uploads whilst replacing the alpha channel
1451   https://bugs.freedesktop.org/show_bug.cgi?id=54808
1452
1453 * Correct the source offset when converting a RENDER composite
1454   operation into a BLT composite operation.
1455   https://bugs.freedesktop.org/show_bug.cgi?id=54868
1456
1457 * Correct a minor typo in flattening alphamaps which caused the
1458   replacement pixmap to often end up with a height of zero.
1459   https://bugs.gentoo.org/show_bug.cgi?id=434860
1460
1461 * Don't discard the CPU damage if only part of is being replaced.
1462   This is most evident when combining software renders like Opera with
1463   accelerated rendering.
1464   https://bugs.freedesktop.org/show_bug.cgi?id=54937
1465
1466Release 2.20.7 (2012-09-08)
1467===========================
1468Continuing the flurry of releases, this week saw the release of
1469xorg-1.13 bringing with it the first steps towards hotpluggable gpu
1470support and the ability to offload DRI rendering onto other GPUs. In
1471light of that advance, this release includes the support code by Dave
1472Airlie to integrate PRIME into xf86-video-intel and make those new
1473features available.
1474
1475Aside from landing PRIME, a few other bugs were fixed since 2.20.6:
1476
1477 * Fix framebuffer leak on server regeneration
1478
1479 * Fix texture cache flushing on IvyBridge and Kwin with plastique
1480   https://bugs.freedesktop.org/show_bug.cgi?id=54488
1481
1482 * Redirect large solid fills on SandyBridge+. By large I mean greater
1483   than 8192 or 16384 pixels on SandyBridge and IvyBridge respectively.
1484   https://bugs.freedesktop.org/show_bug.cgi?id=54134
1485
1486 * Fix up backlight option handling in the manpage and uxa.
1487   https://bugs.freedesktop.org/show_bug.cgi?id=54397
1488
1489 * Unbreak ZaphodHeads.
1490   https://bugs.freedesktop.org/show_bug.cgi?id=52438
1491
1492Release 2.20.6 (2012-09-02)
1493===========================
1494A serious bug that caused a crash on SandyBridge and IvyBridge when
1495mixing CPU and GPU operations on the same buffer, and an annoyance from
1496bad scheduling of windowed swapbuffer updates causing low framerates and
1497jitter. Plus the usual smattering of assertion fixes and a long standing
1498issue with incoherent page access to a streaming buffer.
1499
1500 * Low frame rates in Blobby Valley when "fullscreen"
1501   https://bugs.freedesktop.org/show_bug.cgi?id=54274
1502
1503 * Incoherent concurrent access with the CPU and GPU
1504   https://bugs.freedesktop.org/show_bug.cgi?id=51422
1505   https://bugs.freedesktop.org/show_bug.cgi?id=52299
1506
1507 * Add Option "Backlight" to override automatic selection of the
1508   backlight interface.
1509
1510 * Avoid overwriting the composite operation info when testing if we
1511   can transfer the operation to the BLT.
1512
1513Release 2.20.5 (2012-08-26)
1514===========================
1515Another silly bug found, another small bugfix release. The goal was for
1516the driver to bind to all Intel devices supported by the kernel.
1517Unfortunately we were too successful and started claiming Pouslbo,
1518Medfield and Cedarview devices which are still encumbered by propietary
1519IP and not supported by this driver.
1520
1521Bugs fixed since 2.20.4:
1522
1523 * Only bind to Intel devices using the i915 kernel module
1524
1525 * Regression in the bitmap-to-region code, e.g. icewm window buttons
1526   https://bugs.freedesktop.org/show_bug.cgi?id=53699
1527
1528Release 2.20.4 (2012-08-18)
1529===========================
1530Continuing the small bugfix releases, the only real feature is initial
1531enabling for Haswell for the purpose of rendering verification and
1532validation - by no means is it complete!
1533
1534Bugs fixed since 2.20.3:
1535
1536  * Some potential errors along failure paths found by a static analyser
1537  with the help of Zdenek Kablac.
1538
1539  * Eliminate zero-sized rectangles from PolyFillRectangles as the code
1540  assumes that they did not exist and so caused corruption.
1541
1542  * Remove the UXA warning for failing to tile the front buffer if it is
1543  disallowed by hardware, and so expected.
1544
1545  * Fix the validation of the XV pipe parameter.
1546
1547  * Fix 8x8 tiled pattern fills
1548  https://bugs.freedesktop.org/show_bug.cgi?id=53353
1549
1550  * Fix compile failure when using --with-builderstring
1551
1552  * Restore w/a flush for gen4 fill/copy/video, fortunately rare
1553  operations as at least for fill/copy we prefer to use the BLT.
1554  https://bugs.freedesktop.org/show_bug.cgi?id=53119
1555
1556  * Restore preferred use of the RENDER ring for SNB+ DRI copies.
1557
1558Release 2.20.3 (2012-08-04)
1559===========================
1560Just a minor bugfix for gen4 chipsets (965gm, gm45 and friends) that
1561crept into 2.20.2. As an added bonus, the pessimistic workaround for a
1562GPU hang on gen4 has been relaxed and the shaders have been overhauled
1563which should pave the way to eliminating the last of the uncommon CPU
1564operations, along with immediately realising a small perforamnce
1565improvement.
1566
1567Bugs fixed since 2.20.2:
1568
1569 * Update DPMS bookkeeping after modeset
1570   https://bugs.freedesktop.org/show_bug.cgi?id=52142
1571
1572 * Avoid overlapping gpu/cpu damage after ignoring cpu damage in the
1573   consideration of placement for the operation.
1574
1575 * Enable acceleration by default on 830gm/845g. The GMCH on this pair
1576   of chipsets is notoriously incoherent, so the GPU is almost certainly
1577   going to hang at some point, though unlikely to hang the system and
1578   should automatically disable acceleration (and thence behave
1579   identically as if the acceleration was disabled from the start).
1580   Option "NoAccel" can be used to disable all 2D acceleration and
1581   Option "DRI" can be used to disable all 3D acceleration.
1582   https://bugs.freedesktop.org/show_bug.cgi?id=52624
1583
1584 * Fix vertex bookkeeping for gen4 that was causing corruption in the
1585   command stream.
1586
1587Release 2.20.2 (2012-07-27)
1588===========================
1589For the last 9 months, since 2.16.901, we have been shipping a driver that
1590does not work on IvyBridge GT1 systems (HD2500 graphics); we were telling
1591the GPU to use an invalid number of threads for the pixel shader and this
1592in turned caused the GPU to hang.
1593
1594Also fixed since the last release just a few days ago:
1595
1596* Support for the gmux backlight controller on Apple laptops
1597  https://bugs.freedesktop.org/show_bug.cgi?id=52423
1598
1599* Fix X -configure not to list this driver as matching any Intel device,
1600  just the VGA class devices will do!
1601
1602* A crash in SNA when repeatedly switching xrandr rotations
1603
1604* Corruption in SNA observed in kwin on IvyBridge
1605  https://bugs.freedesktop.org/show_bug.cgi?id=52473
1606
1607Release 2.20.1 (2012-07-22)
1608===========================
1609A week in, grab the brown paper bags, for it is time to reveal a couple
1610of critical bugs that spoilt the 2.20.0 release.
1611
1612Firstly we have the restoration of DRI for i810. I am sure that the
1613solitary user will be overjoyed in a couple of years when a new xserver
1614is forced upon him. That enjoyment will be short-lived when as no actual
1615acceleration remains, not even shadow, for the chipset.
1616
1617Perhaps a little more wildly felt, I hope!, will be that the SNA
1618fallbacks were broken on 64-bit machines if they required clipping. One
1619little misplaced cast of a pointer, and the screen is filled with
1620corruption.
1621
1622Among the other tweaks this week:
1623
1624* A bug affecting gen4 handling of trapezoids was fixed, and CPU
1625  overhead reduced.
1626  https://bugs.freedesktop.org/show_bug.cgi?id=52158
1627
1628* A fix for a bug causing corruption of a DRI2 unredirected client
1629  window that was resized whilst under a compositor.
1630
1631* Support for snoopable buffers on non-LLC architectures, coming to
1632  a future kernel. The aim to accelerate transfers between the CPU
1633  and the GPU, in particular to dramatically improve readback
1634  performance, and to further minimise clflushes.
1635
1636* Improvement to the composite performance on GT2 SandyBridge and
1637  IvyBridge devices, in particular the render copy is significantly
1638  improved.
1639
1640* Improved handling for when acceleration is disabled, including
1641  permitting DRI2 to remain supported even if the X server believes
1642  the GPU wedged.
1643
1644* Shadow support was dropped from UXA as it was neither complete nor
1645  correct, use SNA instead.
1646
1647Release 2.20.0 (2012-07-15)
1648===========================
1649First the big news, a new acceleration method that aims to be faster and
1650consume far less CPU than UXA is now available for selection at runtime.
1651This snazzy new architecture can be selected through use of
1652  Option "AccelMethod" "sna"
1653in your xorg.conf. Whilst it has been under development for some time, it
1654has not yet had the same degree of widespread testing of UXA, so tread
1655lightly. Try it and if you spot anything that can be improved, please do
1656report a bug.
1657
1658Otherwise we have the usual smattering of bug fixes for UXA:
1659
1660 * Use a white source whilst adding glyphs to the glyph mask
1661   (This fixes blank glyphs if using a font that mixes ARGB and A glyphs.)
1662
1663 * Avoid fallbacks for glyph-to-dst in e.g. gnome-terminal
1664   https://bugs.freedesktop.org/show_bug.cgi?id=50508
1665
1666 * Force unused outputs off when VT switching
1667   https://bugs.freedesktop.org/show_bug.cgi?id=50772
1668
1669 * Copy the fbcon across to the Screen pixmap at startup.
1670   (This patch has been kicking around in the distributions for years.)
1671
1672 * Many missed malloc failures checks and forgotten frees found by a static
1673   analyzer. Thanks Zdenek Kabelac!
1674
1675 * Leak of the back buffer when terminating an application after pageflipping
1676   https://bugs.freedesktop.org/show_bug.cgi?id=50670
1677
1678 * Double check that the pipe is on before emitting a WAIT_ON_EVENT.
1679   In conjunction with an uptodate kernel, this should eliminate any
1680   hangs when changing resolutions or adding/removing displays.
1681   https://bugs.freedesktop.org/show_bug.cgi?id=50668
1682
1683 * Update to new Xorg APIs. Future proofing for the next generation of
1684   hotplug Xorg display servers.
1685
1686Many thanks to everyone who has reported a bug and otherwise helped to
1687improve the driver.
1688
1689Release 2.19.0 (2012-04-29)
1690===========================
1691More stability fixes for UXA and support for another variant of IvyBridge.
1692Given the severity of the stability fixes, I strongly recommend everybody
1693to upgrade to 2.19.0.
1694
1695* Prevent waiting on scanlines whilst not in control of the VT and therefore
1696  whilst referencing foreign CRTC configurations.
1697
1698* Pixmap (and bo leak) during fallback glyph composition
1699
1700* Remove broken acceleration for rendering glyphs directly upon the
1701  destination pixmap, exposed by cairo-1.12.0 (and coincidentally fix
1702  another Pixmap leak upon fallback handling).
1703
1704* Add support for Ivy Bridge GT2 Server chipset [PCI id 0x016a]
1705
1706* Remove broken damage flushing with CompositeRectangles
1707  https://bugs.freedesktop.org/show_bug.cgi?id=32547
1708
1709* Fix crash upon server start with multiple monitors
1710  https://bugs.freedesktop.org/show_bug.cgi?id=47395
1711
1712* Fix composition issues resulting from overly aggressive Pixmap reuse
1713  https://bugs.freedesktop.org/show_bug.cgi?id=47345
1714
1715Release 2.18.0 (2012-02-24)
1716===========================
1717Time passes, a few more bugs have crept out of the woodwork that are a
1718compelling reason to update.
1719
1720Bugs fixed in this release (compared to 2.17.0)
1721-----------------------------------------------
1722
1723* Limit maximum object size so that all of the source, mask and
1724  destination can be mapped into the aperture simultaneously by basing the
1725  limit on the mappable aperture size rather than the size of the total
1726  GATT.
1727
1728* Incorrect clipping of polygons
1729  https://bugs.freedesktop.org/show_bug.cgi?id=43649
1730  Regression from 2.15.901
1731
1732* Limit number of VMA cached to avoid hitting the per-process VMA limit
1733  There still is a residual bug in that we seem to have so many objects
1734  floating around in the first place and that still leads to exhaustion
1735  of system limits.
1736  https://bugs.freedesktop.org/show_bug.cgi?id=43075
1737  https://bugs.freedesktop.org/show_bug.cgi?id=40066
1738
1739* Latency in processing user-input during continuous rendering
1740
1741Release 2.17.0 (2011-11-16)
1742==============================
1743A few months have passed, and we have accumulated a surprising number of
1744bug fixes. Oops! We would strongly encourage everyone to upgrade.
1745
1746Bugs fixed in this snapshot (compared to 2.16.902)
1747--------------------------------------------------
1748
1749* Video clobbering composite batch state
1750  http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=635953
1751
1752Snapshot 2.16.902 (2011-11-13)
1753==============================
1754This is the second release candidate in preparation for the upcoming
17552.17.0 release. We will appreciate any feedback we can get from
1756testing of this snapshot, and thanks to everyone who tested and reported
1757upon issues in the last snapshot.
1758
1759A very old bug demonstrating that nobody has ever used alpha maps, and a
1760couple of more recent mistakes were found.
1761
1762Bugs fixed in this snapshot (compared to 2.16.901)
1763--------------------------------------------------
1764
1765 * Incorrect reuse of surface bindings within a batch for multiple formats
1766   https://bugs.freedesktop.org/show_bug.cgi?id=40926
1767
1768 * Nothing was rendered for text with procedural sources
1769   https://bugs.freedesktop.org/show_bug.cgi?id=31819
1770
1771 * Handle fallbacks involving alpha maps
1772
1773Snapshot 2.16.901 (2011-10-30)
1774==============================
1775This is the first release candidate in preparation for the upcoming
17762.17.0 release. We will appreciate any feedback we can get from
1777testing of this snapshot.
1778
1779The basis of this release are to push out a couple of critical stability
1780fixes for SandyBridge and IvyBridge. There is very little else to see
1781here. Those 200+ other commits do not exist...
1782
1783Bugs fixed in this snapshot (compared to 2.16.0)
1784--------------------------------------------------
1785
1786 * Workaround blitter hang on SandyBridge and IvyBridge
1787   https://bugzilla.kernel.org/show_bug.cgi?id=27892
1788   https://bugs.freedesktop.org/show_bug.cgi
1789
1790 * Workaround pipe control issues on SandyBridge
1791
1792 * Use correct maximum PS thread count on IvyBridge
1793
1794 * Protect against failed pixmap allocation for XV
1795   https://bugs.freedesktop.org/show_bug.cgi?id=40439
1796
1797Release 2.16.0 (2011-08-09)
1798==============================
1799A new quarter, a new release! The key feature of this release, looking past
1800the bug fixes, is the enabling of IvyBridge acceleration. We have also fixed
1801many bugs and graphical glitches and would encourage everyone to upgrade.
1802
1803Bugs fixed in this snapshot (compared to 2.15.901)
1804--------------------------------------------------
1805
1806* Build fix for xserver-1.7.7
1807
1808Snapshot 2.15.901 (2011-07-30)
1809==============================
1810This is the first release candidate in preparation for the upcoming
18112.16.0 release. We will appreciate any feedback we can get from
1812testing of this snapshot.
1813
1814The highlight of this snapshot is the full enabling of IvyBridge with
1815acceleration for Render and Xv, along with handling of the shared render
1816buffer allocations required for Mesa. Also of note is that deep-color
1817support is enabled (for all chipsets), which allows you to drive your
181830-bit monitor at its native colour depth.
1819
1820Bugs fixed in this snapshot (compared to 2.15.0)
1821--------------------------------------------------
1822
1823* Misuse of the Resource database causing crashes after DRI clients close
1824  https://bugs.freedesktop.org/show_bug.cgi?id=37700
1825
1826* Crash on large strings
1827  https://bugs.freedesktop.org/show_bug.cgi?id=36860
1828
1829* Incorrect rendering for some core drawing operations
1830  http://bugs.freedesktop.org/show_bug.cgi?id=28768
1831  http://bugs.freedesktop.org/show_bug.cgi?id=28798
1832  http://bugs.freedesktop.org/show_bug.cgi?id=28908
1833  http://bugs.freedesktop.org/show_bug.cgi?id=29401
1834
1835* Crash in Xv due to insufficient checks on batch space
1836  https://bugs.freedesktop.org/show_bug.cgi?id=36319
1837
1838Release 2.15.0 (2011-04-14)
1839==============================
1840We are pleased to announce this major release of the xf86-video-intel
1841driver, roughly on schedule at 3 months since 2.14.0. With the many bug
1842fixes in this release, we encourage everyone to upgrade to 2.15.
1843
1844The priority for this quarter has been simply to be unexciting and stabilise
1845the driver further, seeking to capitalise upon the improvements elsewhere
1846in the stack.
1847
1848Bugs fixed in this snapshot (compared to 2.14.903)
1849--------------------------------------------------
1850
1851* Turn off relaxed fencing by default for older chipsets
1852  This was continuing to destabilize those system, so for the release
1853  we disabled the feature. If you wish to help us debug this, you can
1854  re-enable the optimisation with Option "RelaxedFencing" "True".
1855  Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=36147
1856
1857* Build fix for xserver-1.7.7
1858
1859* KDE glitches on SNB
1860  [Technically fixed in the previous snapshot, but I'm really pleased
1861   that this got fixed in time for the release!]
1862  https://bugs.freedesktop.org/show_bug.cgi?id=35808
1863
1864
1865Snapshot 2.14.903 (2011-04-11)
1866==============================
1867This is the third release candidate in preparation for the upcoming
18682.15.0 release. We will appreciate any feedback we can get from
1869testing of this snapshot.
1870
1871There was a bit of churn since 2.14.902 as a potential fix for a
1872performance regression was tried but had to reverted when it was found to
1873cause glitches running Compiz on SandyBridge. Otherwise, there were just a
1874couple of fixes for building against old xservers and running on an
1875obscure chipset.
1876
1877Bugs fixed in this snapshot (compared to 2.14.902)
1878--------------------------------------------------
1879
1880* Prevent issuing an invalid scanline wait command
1881  https://bugs.freedesktop.org/show_bug.cgi?id=35576
1882
1883* The 946GZ in not a 945, but a 965.
1884  https://bugs.freedesktop.org/show_bug.cgi?id=35854
1885
1886* Fix tile sizes for gen2 (finally).
1887
1888* Allow building of recent dri2 changes against old xservers.
1889
1890
1891Snapshot 2.14.902 (2011-03-29)
1892==============================
1893This is the second release candidate in preparation for the upcoming
18942.15.0 release. We will appreciate any feedback we can get from
1895testing of this snapshot.
1896
1897As befits testing of release candidates, no major regression was found and
1898a couple more bugs have been fixed.
1899
1900Bugs fixed in this snapshot (compared to 2.14.901)
1901--------------------------------------------------
1902
1903* Clients disappearing with pending swaps
1904
1905* Incorrect clipping of Xv output on i915 across extended desktops
1906  https://bugs.freedesktop.org/show_bug.cgi?id=35346
1907
1908* Introduction of a LinearFramebuffer option. (Defaults to tiled for
1909  performance and power saving.)
1910
1911
1912Snapshot 2.14.901 (2011-03-02)
1913==============================
1914This is the first release candidate in preparation for the upcoming
19152.15.0 release. We will appreciate any feedback we can get from
1916testing of this snapshot.
1917
1918Still no further along my grandiose plans to improve Render performance,
1919aside from the performance tuning lower in the stack, instead we have had
1920a steady stream of bug fixes.
1921
1922Bugs fixed in this snapshot (compared to 2.14.0)
1923------------------------------------------------
1924
1925* Green pixels within partially off-screen video playback
1926  https://bugs.freedesktop.org/show_bug.cgi?id=24767
1927
1928* Defer creation of the glyph cache to generation startup
1929  https://bugs.freedesktop.org/show_bug.cgi?id=33412
1930
1931* Incorrect maximum addresses for video decoder state
1932  https://bugs.freedesktop.org/show_bug.cgi?id=34017
1933
1934* Failure to handle oversized temporary surfaces
1935  https://bugs.freedesktop.org/show_bug.cgi?id=34399
1936
1937* Relaxed tiling corruption on gen2
1938
1939* Crash when destroying a foreign DRI drawable
1940  https://bugs.freedesktop.org/show_bug.cgi?id=34787
1941
1942
1943Release 2.14.0 (2011-01-07)
1944===========================
1945We are pleased to announce this major release of the xf86-video-intel
1946driver, on schedule at 3 months since 2.13.0. With the many bug fixes
1947in this release, we encourage everyone to upgrade to 2.14.
1948
1949The most notable feature of this release is the inclusion of acceleration
1950support, both RENDER and XVIDEO, for Sandy Bridge; note that this requires
1951a 2.6.37 kernel. In improving the acceleration code paths to support Sandy
1952Bridge, we were also able to improve the performance of the general i965+
1953RENDER acceleration. For example on i3, glyph (aa10 and rgb10) performance
1954is about 3x faster, as is performing small copies between windows, with the
1955improvement becoming negligible once you hit around 100 pixels square.
1956
1957
1958Snapshot 2.13.903 (2010-01-04)
1959==============================
1960This is the second release candidate in preparation for the upcoming
19612.14.0 release, and if no showstoppers turn up in wider testing will
1962become the next release. Besides the usual bug fixes, we've included
1963patches by Mario Kleiner to improve vblank handing which both improve the
1964timing accuracy and remove a few potential races -- provided you also
1965have the corresponding fixes in the kernel.
1966
1967Bug fixes
1968---------
1969
1970 * Crash on hot-unpluging with an active fullscreen GL application, e.g. a
1971   compositing WM or video player
1972   https://bugs.freedesktop.org/show_bug.cgi?id=32770
1973
1974 * Freeze upon maximising client-side software renderers, e.g. flash.
1975   https://bugs.freedesktop.org/show_bug.cgi?id=31367
1976
1977 * Fix confusion over G35 generation
1978   https://bugs.freedesktop.org/show_bug.cgi?id=32478
1979
1980
1981Snapshot 2.13.902 (2010-12-10)
1982==============================
1983This is the first release candidate in preparation for the upcoming
19842.14.0 release. Sandybridge support should be now be ready for use,
1985and we will appreciate any feedback we can get from this snapshot.
1986Improving the code base for Sandybridge should also have had knock
1987on effects for earlier chips as well, in particular Ironlake and to
1988a lesser extent i965.
1989
1990New requirements compared to 2.13
1991---------------------------------
1992 * Librdrm >= 2.4.23
1993
1994Bug fixes
1995---------
1996
1997 * Crash after failing to allocate memory
1998   https://bugs.freedesktop.org/show_bug.cgi?id=31487
1999
2000 * EFAULT during pwrite
2001   https://bugs.freedesktop.org/show_bug.cgi?id=29752
2002
2003 * Mark outputs as on and restore backlight after mode-set
2004
2005 * A hang with i965+ with newer kernels.
2006
2007
2008Snapshot 2.13.901 (2010-11-05)
2009==============================
2010This is an intermediate snapshot of ongoing driver development. The
2011primary purpose of this snapshot is to capture some recent
2012improvements, (particularly in Sandybridge support), for further
2013testing.
2014
2015Release 2.13.0 (2010-09-30)
2016===========================
2017We are pleased to announce this major release of the xf86-video-intel
2018driver, on schedule at 3 months since 2.12.0. With the many bug fixes
2019in this release, we encourage everyone using 2.12 to upgrade to 2.13.
2020
2021[This release is functionally identical to the earlier 2.12.902
2022release candidate.]
2023
2024New requirements compared to 2.12
2025---------------------------------
2026 * Librdrm >= 2.4.22
2027
2028Bug fixes
2029---------
2030 * Attempt to fix infinite MI_WAIT_FOR_EVENT while watching video
2031   https://bugs.freedesktop.org/show_bug.cgi?id=28964
2032
2033   [Various new checks have been added to the video code here, but the
2034   bug fix hasn't yet been verified by the original reporter.]
2035
2036 * Fix buffer-object leak
2037   https://bugs.freedesktop.org/show_bug.cgi?id=26946
2038
2039 * Fix memory leak on server reset
2040
2041 * Fix crash due to unchecked pixmap allocation
2042   https://bugs.freedesktop.org/show_bug.cgi?id=29187
2043
2044 * Fix for video artifacts when using dualscreen
2045   https://bugs.freedesktop.org/show_bug.cgi?id=29213
2046
2047 * Fix for incorrect characters in gnome-terminal when using compiz
2048   https://bugs.freedesktop.org/show_bug.cgi?id=28438
2049
2050 * Fix for hanging, full-screen applications, (flash, compiz, etc.)
2051   https://bugs.freedesktop.org/show_bug.cgi?id=29584
2052
2053 * Fix selection of backlight device on multi-GPU systems
2054   https://bugs.freedesktop.org/show_bug.cgi?id=29273
2055
2056 * Fix to avoid crash with extremely large glyphs
2057   https://bugs.freedesktop.org/show_bug.cgi?id=29430
2058
2059 * Fix for eDP panels incorrectly being given only a single, valid mode
2060   https://bugs.freedesktop.org/show_bug.cgi?id=30069
2061
2062 * Fix GPU hang involving clipped SRC copies
2063   https://bugs.freedesktop.org/show_bug.cgi?id=30120
2064
2065 * Fix to compile for 1.6 series X server.
2066
2067 * Fix to retry framebuffer allocation after an initial failure.
2068
2069 * Fix to disable dri2 after fallbacks are forced on.
2070
2071Snapshot 2.12.902 (2010-09-28)
2072==============================
2073This is the second release candidate in preparation for the upcoming
20742.13.0 release. The comes exactly one week after 2.12.902 and includes
2075only a handful of changes.
2076
2077One significant change is that the drive now requires libdrm 2.4.22
2078(or newer), as several people reported compilation failures of
2079xf86-video-intel 2.12.901 with libdrm 2.4.21.
2080
2081Other changes include a fix to compile for the 1.6 series X server, a
2082fix to retry framebuffer allocation after an initial failure, and a
2083fix to disable dri2 after fallbacks are forced on.
2084
2085Snapshot 2.12.901 (2010-09-21)
2086==============================
2087This is the first release candidate in preparation for the upcoming
20882.13.0 release. We will appreciate any feedback we can get from
2089testing of this snapshot.
2090
2091Bugs fixed in this snapshot (compared to 2.12.0)
2092------------------------------------------------
2093 * Attempt to fix infinite MI_WAIT_FOR_EVENT while watching video
2094   https://bugs.freedesktop.org/show_bug.cgi?id=28964
2095
2096   [Various new checks have been added to the video code here, but the
2097   bug fix hasn't yet been verified by the original reporter.]
2098
2099 * Fix buffer-object leak
2100   https://bugs.freedesktop.org/show_bug.cgi?id=26946
2101
2102 * Fix memory leak on server reset
2103
2104 * Fix crash due to unchecked pixmap allocation
2105   https://bugs.freedesktop.org/show_bug.cgi?id=29187
2106
2107 * Fix for video artifacts when using dualscreen
2108   https://bugs.freedesktop.org/show_bug.cgi?id=29213
2109
2110 * Fix for incorrect characters in gnome-terminal when using compiz
2111   https://bugs.freedesktop.org/show_bug.cgi?id=28438
2112
2113 * Fix for hanging, full-screen applications, (flash, compiz, etc.)
2114   https://bugs.freedesktop.org/show_bug.cgi?id=29584
2115
2116 * Fix selection of backlight device on multi-GPU systems
2117   https://bugs.freedesktop.org/show_bug.cgi?id=29273
2118
2119 * Fix to avoid crash with extremely large glyphs
2120   https://bugs.freedesktop.org/show_bug.cgi?id=29430
2121
2122 * Fix for eDP panels incorrectly being given only a single, valid mode
2123   https://bugs.freedesktop.org/show_bug.cgi?id=30069
2124
2125 * Fix GPU hang involving clipped SRC copies
2126   https://bugs.freedesktop.org/show_bug.cgi?id=30120
2127
2128Release 2.12.0 (2010-06-24)
2129===========================
2130We are pleased to announce this major release of the xf86-video-intel
21312D driver. It contains many correctness and performance improvements
2132compared to the 2.11 releases. We encourage everyone using 2.11 to
2133upgrade to 2.12.
2134
2135[This release contains several bug fixes since the earlier 2.11.901
2136release candidate---see below for details.]
2137
2138New requirements compared to 2.11
2139---------------------------------
2140 * Libdrm >= 2.4.21
2141
2142Major performance improvements
2143------------------------------
2144There have been many performance improvements targeting the
2145915G/945G/Pineview graphics chipsets (thank to Chris Wilson!). Areas
2146improved include:
2147
2148 * Major improvements to glyph rendering
2149
2150 * Avoid software fallback when using extended desktops, (for example,
2151   video playback)
2152
2153 * Better performance when using large windows, (for example with
2154   firefox scrolling)
2155
2156 * Avoiding migration ping-pong with gigantic drawing
2157
2158The following speedups (measured with cairo-perf-trace/cairo-traces)
2159capture some of the improvements. These were measured on a Pineview
2160system with X server 1.8.1. This compares how this 2.12 release
2161improves performance compared to 2.11.0:
2162
2163	Trace			Speedup
2164	-----			-------
2165	ocitysmap		5.41x speedup
2166	firefox-talos-gfx       2.41x speedup
2167	gnome-terminal-vim      2.15x speedup
2168	poppler                 1.86x speedup
2169	firefox-planet-gnome	1.43x speedup
2170	midori-zoomed		1.39x speedup
2171	swfdec-giant-steps      1.36x speedup
2172	xfce4-terminal-a1	1.17x speedup
2173	evolution		1.09x speedup
2174	swfdec-youtube		1.06x speedup
2175
2176Other improvements
2177------------------
2178Daniel Vetter improved Xvmc and overlay, adding GEM support.
2179
2180Bug fixes
2181---------
2182There have been many conformance improvements targeting the
2183915G/945G/Pineview graphics chipsets. The rendercheck test suite
2184passes once again and there has been a significant reduction in the
2185number of errors detected by the cairo test suite (notably center
2186sampling and EXTEND_NONE behaviors are fixed).
2187
2188Other notable bug fixes include:
2189
2190 * Fix for bug 28446 Garbled fonts with Mathematica
2191   https://bugs.freedesktop.org/show_bug.cgi?id=28446
2192
2193 * Fixes for page-flipping and other DRI2-handling bugs, (Thanks Jesse
2194   Barnes, and Kristian Høgsberg)
2195
2196 * Avoid corruption when using extended desktops, (for example, video
2197   playback)
2198
2199(The bug fixes below are new since the 2.11.901 release candidate)
2200
2201 * Fix corrupted output when screen rotated
2202   https://bugs.freedesktop.org/show_bug.cgi?id=28461
2203
2204 * Fix corrupted rendering in KDE due to missing flush.
2205
2206 * Fixes i830_uxa_put_image for the following issues:
2207
2208      Bug 28569 - [i965] IGN's flash-based video player crashes X
2209      https://bugs.freedesktop.org/show_bug.cgi?id=28569
2210    
2211      Bug 28573 - [i965] Fullscreen flash and windowed SDL games fail to
2212                  update the screen
2213      https://bugs.freedesktop.org/show_bug.cgi?id=28573
2214
2215 * Fix visual corruption of scrollbar in Chromium
2216
2217 * Fix tiling limits to resolve this issue:
2218
2219      Bug 28497 - Graphics corruption after opening a specific website
2220      https://bugs.freedesktop.org/show_bug.cgi?id=28497
2221
2222Snapshot 2.11.901 (2010-06-14)
2223==============================
2224This is the first release candidate in preparation for the upcoming
22252.12.0 release. We will appreciate any feedback we can get from
2226testing of this snapshot to improve the 2.12.0 release.
2227
2228New requirements compared to 2.10
2229---------------------------------
2230 * Libdrm >= 2.4.19
2231
2232Major performance improvements
2233------------------------------
2234There have been many performance improvements targeting the
2235915G/945G/Pineview graphics chipsets (thank to Chris Wilson!). Areas
2236improved include:
2237
2238 * Major improvements to glyph rendering
2239
2240 * Avoid software fallback when using extended desktops, (for example,
2241   video playback)
2242
2243 * Better performance when using large windows, (for example with
2244   firefox scrolling)
2245
2246 * Avoiding migration ping-pong with gigantic drawing
2247
2248The following speedups (measured with cairo-perf-trace/cairo-traces)
2249capture some of the improvements. These were measured on a Pineview
2250system with a very recent X server, (from git just before 1.8.1). This
2251compares how this 2.12 release candidate improves performance compared
2252to 2.11.0:
2253
2254	Trace			Speedup
2255	-----			-------
2256	ocitysmap		5.41x speedup
2257	firefox-talos-gfx       2.41x speedup
2258	gnome-terminal-vim      2.15x speedup
2259	poppler                 1.86x speedup
2260	firefox-planet-gnome	1.43x speedup
2261	midori-zoomed		1.39x speedup
2262	swfdec-giant-steps      1.36x speedup
2263	xfce4-terminal-a1	1.17x speedup
2264	evolution		1.09x speedup
2265	swfdec-youtube		1.06x speedup
2266
2267Other improvements
2268------------------
2269Daniel Vetter improved Xvmc and overlay, adding GEM support.
2270
2271Bug fixes
2272---------
2273There have been many conformance improvements targeting the
2274915G/945G/Pineview graphics chipsets. The rendercheck test suite
2275passes once again and there has been a significant reduction in the
2276number of errors detected by the cairo test suite (notably center
2277sampling and EXTEND_NONE behaviors are fixed).
2278
2279Other notable bug fixes include:
2280
2281 * Fix for bug 28446 Garbled fonts with Mathematica
2282   https://bugs.freedesktop.org/show_bug.cgi?id=28446
2283
2284 * Fixes for page-flipping and other DRI2-handling bugs, (Thanks Jesse
2285   Barnes, and Kristian Høgsberg)
2286
2287 * Avoid corruption when using extended desktops, (for example, video
2288   playback)
2289
2290Release 2.11.0 (2010-03-29)
2291===========================
2292New in 2.11: DRI2 and page flipping
2293-----------------------------------
2294The most significant new feature of this release is support for new
2295DRI2 APIs, allowing page flipping to occur for swaps that are
2296full-screen and not rotated.
2297
2298Performance improvements in 2.11 compared to 2.10
2299-------------------------------------------------
2300Dramatically improved performance of large pixmaps on
2301memory-constrained hardware, (such as 945), by using tiling.
2302Note that this fix is new since the most recent release
2303candidate for this release (2.10.903).
2304
2305    https://bugs.freedesktop.org/show_bug.cgi?id=25375
2306
2307Elimination of software fallback with alpha-only pixmap (a fix which
2308eliminated a full second from the Moblin boot time):
2309
2310    https://bugs.freedesktop.org/show_bug.cgi?id=26189
2311
2312Notable bug fixes in 2.11 compared to 2.10
2313------------------------------------------
2314Fix for undesired black borders on some images, (caued several
2315problems with firefox):
2316
2317    http://bugs.freedesktop.org/show_bug.cgi?id=17933
2318
2319And other fixes to provide more conformance with existing
2320software-rendering, (as measures by the cairo test suite), such as the
2321sampling location for nearest-neighbor sampling.
2322
2323Other changes
2324-------------
2325Eric Anholt and Daniel Vetter both removed large piles of old and
2326useless code now that the driver requires kernel modesetting
2327(KMS). This continues to reduce the mainteance burden of the driver,
2328making it easier to isolate and fix bugs.
2329
2330Snapshot 2.10.903 (2010-03-22)
2331==============================
2332Some notable bug fixes
2333----------------------
2334* Fix unpredictable results in page-flipping code due to access of an
2335  uninitialized variable.
2336
2337* Two fixes that address a number of long-outstanding failures in the
2338  cairo test suite:
2339
2340    Remove coord-adjust for nearest centre-sampling.
2341
2342    Fill alpha on xrgb images. Closes:
2343
2344	Bug 17933 - x8r8g8b8 doesn't sample alpha=0 outside surface bounds
2345	http://bugs.freedesktop.org/show_bug.cgi?id=17933
2346
2347* Fix a regression introduced in 2.10.902:
2348
2349    Fatal server error:
2350    i915_emit_composite_setup: ADVANCE_BATCH:
2351    under-used allocation 100/104
2352
2353	Bug 21723 [i915 bisected] rendering makes X crash
2354	http://bugs.freedesktop.org/show_bug.cgi?id=27123
2355
2356Other minor fixes (avoid trying to print an error message with a freed
2357string, set a variable to NULL after freeing), and documentation
2358updates are also included.
2359
2360Snapshot 2.10.902 (2010-03-15)
2361==============================
2362Some significant bug fixes
2363--------------------------
2364Fix video color problems due to YUV plane ordering issue. This fixed a
2365regression that showed up recently (only in these release candidates).
2366
2367Fix recent regression leading to corruption under compiz:
2368
2369  https://bugs.freedesktop.org/show_bug.cgi?id=26814
2370
2371Further DRI2 improvements/fixes from Mario Kleiner and Jesse Barnes,
2372(including support for SwapBuffers request with divisor != 0).
2373
2374Fix build against X server 1.6 branch.
2375
2376Avoid flooding logs after an error occurs.
2377
2378And dramatic code cleanups
2379--------------------------
2380Eric Anholt removed several cases of old, useless code, (conditions
2381that could not occur now that we require KMS, etc.).
2382
2383Daniel Vetter eliminated piles of code by modernizing the XvMC
2384memory-management and completely removing the old i830_memory
2385allocator (i830_allocate_memory, i830_free_memory), etc.
2386
2387Snapshot 2.10.901 (2010-02-26)
2388==============================
2389The most significant new feature of this release is support for new
2390DRI2 APIs, allowing page flipping to occur for swaps that are
2391full-screen and not rotated.
2392
2393Some significant bug fixes
2394--------------------------
2395* Avoid fallback due to failing to extract pixel value from an
2396  alpha-only solid, (this fix eliminated 1 second from Mobilin boot
2397  time).
2398
2399  https://bugs.freedesktop.org/show_bug.cgi?id=26189
2400
2401* Handle full-height blits without tearing in CopyRegion
2402
2403  https://bugs.freedesktop.org/show_bug.cgi?id=22475
2404
2405Several other fixes and cleanups are included as well, (including the
2406removal of stale utility code that is now maintained in the separate
2407intel-gpu-tools module).
2408
2409Release 2.10.0 (2010-01-04)
2410===========================
2411New requirements of 2.10 compared to 2.9
2412----------------------------------------
2413 * Linux kernel with kernel-modesetting (KMS)
2414
2415   All user-modesetting code has now been removed from the driver.
2416   We highly recommend kernel version 2.6.32 or later.
2417
2418 * Libdrm >= 2.4.16
2419
2420New features in 2.10 compared to 2.9
2421------------------------------------
2422 * New driver debugging options available in xorg.conf
2423   (DebugFlushBatches, DebugFlushCaches, DebugWait). See "man intel"
2424   for more details.
2425
2426 * Video overlay support with KMS. This currently requires Linux
2427   2.6.33, but a backport to 2.6.32 is available here:
2428
2429   http://gitorious.org/daniel-s-linux-stuff/linux-kernel/commits/intel-kms-overlay-for-2.6.32
2430
2431Notable fixes in 2.10 compared to 2.9
2432-----------------------------------
2433 * Fix crash when XRenderComposite is called with a -1 value for
2434   width/height, (crash at login when using compiz).
2435
2436   http://bugs.freedesktop.org/show_bug.cgi?id=24724
2437
2438 * Fix to support larger extended desktops, (such as a 2560x1600 plus a
2439   1920x1200 monitor).
2440
2441
2442 * Fix glyph corruption due to insufficient cache flushing:
2443
2444   https://bugs.freedesktop.org/show_bug.cgi?id=24315
2445
2446Several other correctness and performance fixes are also included.
2447
2448Note: The driver code of the 2.10.0 release is identical to that of
2449the 2.9.99.901 release candidate. The only changes in 2.10.0 compared
2450to 2.9.99.902 are a minor change to the configure script, and some
2451small updates to the documentation.
2452
2453Snapshot 2.9.99.902 (2009-12-10)
2454================================
2455This is the second release candidate in preparation for the upcoming
24562.10.0 release. We will appreciate any feedback we can get from
2457testing of this snapshot to improve the 2.10.0 release.
2458
2459Compared to the previous snapshot, this snapshot now correctly
2460verifies that libdrm 2.4.16 (or newer) is available. The previous
2461snapshot also required libdrm 2.4.16 for correct functionality, but
2462neglected to notify the user of this at configure, compile, or install
2463time and would instead simply fail to run.
2464
2465Snapshot 2.9.99.901 (2009-11-30)
2466================================
2467This is the first release candidate in preparation for the upcoming
24682.10.0 release. We will appreciate any feedback we can get from
2469testing of this snapshot to improve the 2.10.0 release.
2470
2471With this snapshot, the xf86-video-intel driver requires kernel
2472modesetting (KMS) as all of the user modesetting (UMS) support has
2473been removed from the driver.
2474
2475Release 2.9.0 (2009-09-28)
2476==========================
2477We are pleased to announce the 2.9.0 release of the xf86-video-intel
2478driver. (If you have been following along at home, this release is
2479identical to the 2.8.99.902 release candidate other than the version
2480number).
2481
2482New features in 2.9 compared to 2.8
2483-----------------------------------
2484* Support for the B43 chipset.
2485
2486Major fixes in 2.9.0 compared to 2.8.0
2487--------------------------------------
2488* Multiple fixes to make the driver stable for 8xx chipsets, (855GM,
2489  865G, etc.). The 2.8 driver series was extremely unstable with many
2490  of these chipsets.
2491
2492  https://bugs.freedesktop.org/show_bug.cgi?id=22904 (and many duplicates)
2493  https://bugs.freedesktop.org/show_bug.cgi?id=22947
2494
2495* Add support for BACKLIGHT property when using kernel modesetting
2496  (KMS). This allows backlight adjustment with programs such as
2497  "xbacklight -set <percentage>" or "xrandr --set BACKLIGHT <value>".
2498
2499  https://bugs.freedesktop.org/show_bug.cgi?id=20963
2500
2501* Fix so that "xrandr --scale" works when using KMS.
2502
2503  https://bugs.freedesktop.org/show_bug.cgi?id=22893
2504
2505* Fix segfaults of X server when logging out.
2506
2507  https://bugs.freedesktop.org/show_bug.cgi?id=20516
2508
2509* Avoid falling back to software for 1-bit alpha-only masks.
2510
2511  https://bugs.freedesktop.org/show_bug.cgi?id=23184
2512
2513* Fix pixel sampling position for 8xx chipsets, (fixes some cairo-test
2514  suite failures).
2515
2516Snapshot 2.8.99.901 (2009-09-09)
2517================================
2518This is the first release candidate in preparation for the upcoming
25192.9.0 release. We will appreciate any feedback we can get from testing
2520of this snapshot to improve the 2.9.0 release.
2521
2522See the commit log for changes in this release, and developers, please
2523provide me summaries for any major features in this release that you
2524think should appear in NEWS and in the 2.9.0 release notes.
2525
2526Note: We are still investigating failures with 865G chipsets, (these
2527failures existed in the 2.8 series as well). We are hoping to fix
2528these failures before the 2.9.0 release.
2529
2530Release 2.8.0 (2009-07-20)
2531==========================
2532We are pleased to present this major release of the xf86-video-intel
2533driver. This release is very similar to the 2.7.99.902 release
2534candidate but includes a couple of additional bug fixes, (for bugs
2535#22760 and @22483).
2536
2537Compared to the 2.7 releases this driver contains several new features
2538and many bug fixes. The driver now depends on X server 1.6 or later,
2539and eliminates several obsolete code paths, (XAA and EXA removed in
2540favor of UXA, DRI1 support eliminated). The driver certainly will work
2541best with an i915 module from a recent kernel (2.6.31) and with
2542kernel-modesetting (KMS, specified by loading i915 with the option
2543"modeset=1").
2544
2545New features in 2.8.0 compared to 2.7
2546-------------------------------------
2547* Remove XAA/EXA/DRI1 and NoAccel, and only support UXA/DRI2 now
2548  (requiring xserver 1.6).
2549
2550* VLD support added into XvMC for Mpeg2, which decreases CPU usage
2551  when playing video.
2552
2553Major fixes in 2.8.0 compared to 2.7
2554------------------------------------
2555* Fix broken front-buffer rendering
2556  https://bugs.freedesktop.org/show_bug.cgi?id=19174
2557
2558* Fix disabling of XvMC disabled
2559  https://bugs.freedesktop.org/show_bug.cgi?id=20790
2560
2561* Fix broken VT switch on some machines
2562  https://bugs.freedesktop.org/show_bug.cgi?id=19578
2563
2564* Improve performance of trapezoid rendering for UXA
2565  https://bugs.freedesktop.org/show_bug.cgi?id=21376
2566
2567* Vblank sync'd GL buffer swap (fixing video tearing under composite)
2568  https://bugs.freedesktop.org/show_bug.cgi?id=20664
2569
2570* Fix G41 DRM support
2571  https://bugs.freedesktop.org/show_bug.cgi?id=21095
2572
2573* Fix crash on some 8xx
2574  https://bugs.freedesktop.org/show_bug.cgi?id=18974
2575
2576Various fixes for GPU hangs, and more...
2577
2578Snapshot 2.7.99.902 (2009-07-13)
2579--------------------------------
2580This is the first release candidate in preparation for the upcoming
25812.8.0 release. Most major and critical bugs should be fixed, but some
2582minor bugs may still be present. We will appreciate any feedback we
2583can get from testing of this snapshot to improve the 2.8.0 release.
2584
2585Snapshot 2.7.99.901 (2009-06-10)
2586--------------------------------
2587This is the first release candidate in preparation for the upcoming
25882.8.0 release. Most critical bugs should be fixed, but some major
2589bugs may still be present. We will appreciate any feedback we can
2590get from testing of this snapshot to improve the 2.8.0 release.
2591
2592Some of the major fixes in this snapshot include:
2593
2594	* Several fixes for intermittent GPU hangs/crashes, (some of
2595	  which already appeared in the 2.7.1 release)
2596
2597	* Fix to avoid X crash when viewing large images in browser
2598
2599	* Fix X server failure when running old (non-GEM) kernel
2600
2601	* Fixes for SDVO LVDS mode detection
2602
2603	* Fix major performance regression of trapezoid rendering
2604          compared to XAA/EXA
2605
2606	* New support for tear-free video when using KMS
2607
2608	* New support for tear-free DRI2 CopyRegion
2609
2610Snapshot 2.7.99.1 (2009-04-28)
2611------------------------------
2612This is a development snapshot very early in the process toward
2613developing 2.8. There have been some big changes to the code, and
2614we're anxious to get feedback on these changes as early as possible.
2615
2616Here is a summary of the biggest changes:
2617
2618  * Driver now depends on X server 1.6 or later
2619
2620  * Eliminate XAA and EXA support (in favor of UXA)
2621
2622  * Eliminate DRI1 support
2623
2624  * Fixes for running without DRI at all
2625
2626These code removals represent a deletion of a substantial amount of
2627code, (and hopefully piles of bugs), as well as reduce the maintenance
2628effort going forward as the number of combinatorial configurations for
2629the driver are greatly reduced. This means that users are much more
2630likely to be running code that has actually been tested, and it will
2631be much easy for developers to replicate bugs that users experience.
2632
2633One of the things that would be most useful in testing this release is
2634to revisit any outstanding bugs that you have previously reported. If
2635the buggy behavior is gone, (or the bug is no longer relevant---such
2636as a bug that's specific to XAA only), please feel free to indicate so
2637in bugzilla or even just close the bug.
2638
2639If you confirm that the bug is still present, please indicate so in
2640the bug report. (I was going to ask that you select a 1.7.99 version,
2641but it looks like bugzilla only has versions for products not
2642compoenents, while we use a "xorg" product and a "driver/intel"
2643component.) We definitely want to make any such confirmed bugs a
2644priority, so it would be nice to have a consistent mechanism to search
2645for these bugs. Suggestions are welcome on the best approach.
2646
2647Thanks in advance for any testing or feedback on this snapshot.
2648
2649Release 2.7.0 (2009-04-15)
2650--------------------------
2651Compared to the 2.6 series, 2.7.0 has a large number of bug fixes, but
2652also a few significant features, such as:
2653
2654        SDVO-TV support, available on ADD2 card (bug#9992) and
2655        D945GCLF2 board (bug#17776).
2656
2657        Basic SDVO-LVDS support
2658
2659        XV video display without tearing
2660	[Though this isn't working for all users yet, see
2661	https://bugs.freedesktop.org/show_bug.cgi?id=21076 ]
2662
2663	Various fixes for UXA, DRI2, and Kernel modesetting.
2664
2665We encourage users to use kernel modesetting and UXA acceleration with
2666this release, which should give the best performance and
2667robustness. When KMS is available, UXA is the default acceleration
2668used by the driver, (EXA is the default otherwise).
2669
2670Known issue:
2671
2672	Some Linux kernel versions (such as 2.6.29) are known to have
2673	broken PAT code that causes recent versions of this driver to
2674	fail, (which can manifest as the X server simply not
2675	starting). This can be verified by adding the "nopat" option
2676	to the kernel command-line and seeing the failure go away. We
2677	hope that newer kernels in the 2.6.29.x as well as 2.6.30 and
2678	above will have working PAT code.
2679
2680Some of the most notable bugs fixed in 2.7.0 include:
2681
2682	[GM45 965GM] bad htotal causes panel startup failure
2683	https://bugs.freedesktop.org/show_bug.cgi?id=17292
2684
2685	[xrandr TV] need TV output property control
2686	https://bugs.freedesktop.org/show_bug.cgi?id=12763
2687
2688	[TV] "xrandr --set TV_FORMAT" gets BadMatch error
2689	https://bugs.freedesktop.org/show_bug.cgi?id=16566
2690
2691	[945 tiling] Low performance due to no A17 workaround
2692	https://bugs.freedesktop.org/show_bug.cgi?id=16835
2693
2694	[TV]Flicker when launching applications in the 2.4-branch
2695	https://bugs.freedesktop.org/show_bug.cgi?id=17405
2696
2697	[945GM FBC] FBC causes underruns & flicker
2698	https://bugs.freedesktop.org/show_bug.cgi?id=18651
2699
2700	[xv] Textured video suffers from tearing
2701	https://bugs.freedesktop.org/show_bug.cgi?id=19635
2702
2703	[G45] Random hangs with UXA
2704	https://bugs.freedesktop.org/show_bug.cgi?id=19734
2705
2706	[945GM] Any 3D app is slow in resolution higher than 800x600
2707	with UXA+DRI2, due to tiling
2708	https://bugs.freedesktop.org/show_bug.cgi?id=19738
2709
2710	[i915 UXA,EXA] rotation messes display with tiling on
2711	https://bugs.freedesktop.org/show_bug.cgi?id=20265
2712
2713	[G45] DRI2/UXA gives solid white instead of transparency
2714	https://bugs.freedesktop.org/show_bug.cgi?id=20321
2715
2716	LVDS output not detected
2717	https://bugs.freedesktop.org/show_bug.cgi?id=20517
2718
2719	xf86-video-intel-2.6.3: Xv crashes X server
2720	https://bugs.freedesktop.org/show_bug.cgi?id=20525
2721
2722	[G965 non-GEM] systray in KDE 4 completely broken
2723	https://bugs.freedesktop.org/show_bug.cgi?id=20527
2724
2725	[SDVO-TV]the desktop is teared in four sections on the screen
2726	https://bugs.freedesktop.org/show_bug.cgi?id=20550
2727
2728	Intel video driver 2.6.3 crashes with XVideo
2729	https://bugs.freedesktop.org/show_bug.cgi?id=20563
2730
2731	[855GM] Xv crash with non-KMS
2732	https://bugs.freedesktop.org/show_bug.cgi?id=20585
2733
2734	2.6.99.902 breaks native 1680x1050 mode on TMDS -- EDID miss
2735	https://bugs.freedesktop.org/show_bug.cgi?id=20594
2736
2737	[945GM TV] 2.6.99.902 sets a too high CONTRAST-value
2738	https://bugs.freedesktop.org/show_bug.cgi?id=20670
2739
2740	[915GM] fail to detect LVDS with new VBT code
2741	https://bugs.freedesktop.org/show_bug.cgi?id=20752
2742
2743	[regression i965]tiled output when start X
2744	https://bugs.freedesktop.org/show_bug.cgi?id=20803
2745
2746	2.6.99.902: LVDS wrongly detected as disconnected
2747	https://bugs.freedesktop.org/show_bug.cgi?id=20826
2748
2749	vt switching fails and crashes X
2750	https://bugs.freedesktop.org/show_bug.cgi?id=20863
2751
2752	[HDMI] The screen will flicker when some application runs on
2753	G45-64
2754	https://bugs.freedesktop.org/show_bug.cgi?id=20875
2755
2756	TexturedVideo is offsetted wrongly on dualhead
2757	https://bugs.freedesktop.org/show_bug.cgi?id=20980
2758
2759	[EXA] xvideo hang X
2760	https://bugs.freedesktop.org/show_bug.cgi?id=21027
2761
2762	[EXA] x11perf performance regression
2763	https://bugs.freedesktop.org/show_bug.cgi?id=21029
2764
2765	And many others...
2766