diff options
Diffstat (limited to 'NEWS')
| -rw-r--r-- | NEWS | 290 |
1 files changed, 290 insertions, 0 deletions
@@ -1,3 +1,293 @@ +Release 2.21.2 (2013-02-10) +=========================== +Pass the brown paper bags, I need half a dozen or so. That seemingly +innocuous build fix to xorg-1.13 happned to have the little side-effect +of breaking glyph rendering on xorg-1.12 and older on 64-bit machines. + + +Release 2.21.1 (2013-02-10) +=========================== +A fix for a potential GPU hang on 945gm (GMA3100) and earlier chipsets, +along with backporting SNA to the packages found in stable distributions +like Debian 6.0 (Squeeze). + + * Cleanup compilation warnings from deblint, thanks to Paul Menzel + + * Minor build improvements by Damien Lespiau. + + * Disable generating span geometry for non-rectilinear spans on gen4 + in order to work around and prevent one class of render corruption. + + * Prevent cache thrashing and severe performance degradation on LLC + machines for streaming texture updates. However, note the effect was + only observed on just one particular laptop. + + * Fix alignment of subsurface proxies for old chipsets. + https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1120108 + + * Repair build against Xorg-1.6 and contemporary packages. + + +Release 2.21.0 (2013-02-01) +=========================== +A few new features: + + * Enable render acceleration for Haswell GT1/GT2. + + * Enable multi-threaded rasterisation of trapezoids and fallback composition + + * Utilise a new kernel interface (v3.9) for processing relocations + +along with a few older features from the 2.20.x series: + + * PRIME support for hotplug GPUs and hybrid systems + + * Support for IvyBridge GT1 machines, aka HD2500 graphics. + + * Stable 830gm/845g support, at last! + +As usual we have a large number of bug fixes since the last release: + + * Prevent a stray relocation being left after a buffer is removed from + a batch, leading to GPU hangs. + + * Make the driver more robust against its own failures to submit batches + by falling back to software rendering. + + * Fix emission of scanline waits for secondary pipes on gen6/7. Otherwise + you may encounter GPU hangs in MI_WAIT_FOR_EVENT. + + * Fix a missing corner pixel when drawing rectangles with PolyLines + https://bugs.freedesktop.org/show_bug.cgi?id=55484 + + * Don't try to use Y-tiling colour buffers with mesa/i915c as mesa + doesn't support them and wil fallback to software rendering + + * Ensure that any cached mmaps are invalidated for a SwapBuffers + https://bugs.freedesktop.org/show_bug.cgi?id=60042 + + * Correctly handle the composition of rotated displays too large for the + 3D pipeline + https://bugs.freedesktop.org/show_bug.cgi?id=60124 + + * Fix the computation of the planar video frame size + https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1104180 + + +Release 2.20.19 (2013-01-20) +============================ +A quick release as the last broke USB DisplayLink slave outputs badly. The +performance of those displays was unusable due to an inadvertent change that +caused us to flush the entire scanout over the USB for every drawing +operation. + + * Implement the GNOME Build API. A couple of minor changes to make + integrators and distributors lives a little easier, or at least more + consistent. + + * Correctly offset inplace trapezoids for subwindows, such as the GTK+ + close button after it has a background image uploaded. + + * Explicitly prevent ring-switching for synchronized rendering to + scanouts (for vsync). + + * Clip dirty region to slave pixmaps (otherwise UDL is nigh unusuable) + https://bugs.freedesktop.org/show_bug.cgi?id=59539 + + +Release 2.20.18 (2013-01-16) +============================ +A bunch of miscellaneous fixes for assertion failures and various +performance regressions when mixing new methods for offloads, along with +a couple of improvements for rendering with gen4. + + * Remove use of packed unnormalized texture coordinates on gen4/5 as + these GPUs do not support unnormalized coordinates in the sampler. + + * Remove dependency upon x86 asm for cross-building to unsupported + architectures. + https://bugs.gentoo.org/show_bug.cgi?id=448570 + + * Apply damage around PRIME updates in the correct order. + + * Correctly read the initial backlight level for when the user + overrides UXA's choice of backlight controller. + + * Throttle UXA and prevent it queuing work much faster than the GPU can + complete it. This realised itself in impossible performance figures and + the entire display freezing for several seconds whlist the GPU caught + up. One side effect is that it also caused the DDX to consume more + memory than was required as it could not recycle buffers quick + enough, and in some cases this produces a marked improvement in + performance. Also note on gen2/3 this requires a new libdrm [2.4.41] + in order to prevent a bug causing the DDX to fallback to swrast. + +Release 2.20.17 (2012-12-26) +============================ +A minor update to prepare for co-operating with the kernel over managing +stability on 830gm/845g. On this pair of chipsets, the kernel will perform +an extra copy of the batchbuffer into reserved memory, which prevents them +from randomly dying. However, that extra copy does have a noticeable +impact upon throughput, so we also have a mechanism for userspace to +opt-out of the kernel workaround and take responsibility for ensuring its +batches are coherent. + + * Build fixes against xorg-1.14 + https://bugs.freedesktop.org/show_bug.cgi?id=58552 + https://bugs.freedesktop.org/show_bug.cgi?id=58406 + + * Fixed the origin of cropped (textured) video windows (Xv and XvMC) + https://bugs.freedesktop.org/show_bug.cgi?id=23033 + + * Fix potential corruption when using images larger than ~1GiB + + +Release 2.20.16 (2012-12-15) +============================ +Rejoice! We have found a trick to make 830gm/845g stable at long last. +Ever since the switch to GEM and dynamic video memory, those early +second generation chipsets have been plagued by instability. The lack of +flushing cachelines from the CPU to GMCH was eventually solved by using +an undocmented bit, but 830/845 were still hanging under memory pressure. +These deaths were all due to garbage finding its way into the command +streamer, and they go away if we take a leaf out of the original driver +and never reuse those pages for anything else. So for the first time +ever, I have been able to complete running the test suite on an 845g, +even whilst thrashing the page and buffer caches! + + * Run the SF stage as single-threaded on gen4 to workaround a few issues + https://bugs.freedesktop.org/show_bug.cgi?id=57410 + + * Keep the scanout SURFACE_STATE separate to avoid overriding its + memory access control on gen6/7 (i.e. writes to the scanout need to + be kept out of the render cache) + + * Tune batch flushing after an operation to an exported surface under a + compositor. + + * Make sure the source is on the CPU for inplace composition of trapezoids + using the CPU + https://bugs.freedesktop.org/show_bug.cgi?id=56825 + + * Immediately flush in the block hander after a split batch to reduce + latency between the two halves of an operation. + https://bugs.freedesktop.org/show_bug.cgi?id=51718 + + * Install a fallback config if we fail to install the desired config + at VT switch (i.e. booting, after resume with 3 incompatible pipes on + Ivybridge) + + * Pin batches to avoid CS incoherence on 830/845 + https://bugs.freedesktop.org/show_bug.cgi?id=26345 + + +Release 2.20.15 (2012-12-03) +============================ +And lo, enabling more of the common acceleration paths for gen4 revealed +another lurking bug - something is wrong with how we prepare Y-tiling +surfaces for rendering. For the time being, we can surreptiously disable +them for gen4 and avoid hitting GPU hangs. + + * Avoid clobbering the render state after failing to convert the + operation to use the blitter. + https://bugs.freedesktop.org/show_bug.cgi?id=57601 + + * Disable shadow tracking upon server regeneration, and so fix a crash + if you restart the server whilst a RandR transform (e.g. rotation) is + in effect. + https://bugs.freedesktop.org/show_bug.cgi?id=52255 + https://bugs.freedesktop.org/show_bug.cgi?id=56608 + + +Release 2.20.14 (2012-11-26) +============================ +The highlight of this release is gen4, from 965g to gm45. Quite an old +bug surfaced in the shader assembly, sparking a chance to review a few +design choices within that backend and experiment on fresh ways to +workaround the remaining issues. + + * Avoid using inplace XOR'ed uploads for very large buffers + https://bugs.freedesktop.org/show_bug.cgi?id=57031 + + * Fix the gen4/5 opacity shader + https://bugs.freedesktop.org/show_bug.cgi?id=57054 + + * Queue a pending vblank request after flip completion + https://bugs.freedesktop.org/show_bug.cgi?id=56423 + + * Avoid migrating an uninitialised pixmap for use as a render source + https://bugs.freedesktop.org/show_bug.cgi?id=47597 + + * Improve handing of texture fallbacks for 830/845. + https://bugs.freedesktop.org/show_bug.cgi?id=57392 + + +Release 2.20.13 (2012-11-11) +============================ +Nothing but bug fixes. Many thanks to everyone who took the time to +report their issues, and for their help in improving the driver. + + * Sanity check the platform probe points to our expected i915 device + https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1069031 + + * Prevent 16-bit overflow for computing the sample area to upload of + sources for render operations + https://bugs.freedesktop.org/show_bug.cgi?id=56324 + + * Clamp the drawable box for migration to prevent 16-bit overflow + https://bugs.freedesktop.org/show_bug.cgi?id=56591 + + * Disable RandR hotplug events if Xinerama is enabled and thereby prevent + a crash upon hotplug + https://bugs.freedesktop.org/show_bug.cgi?id=55260 + + * Call ValidatePicture before attempting to flatten the alphamaps + https://bugs.freedesktop.org/show_bug.cgi?id=56367 + + * Clip the trapezoid correctly if it ends on the boundary pixel + https://bugs.freedesktop.org/show_bug.cgi?id=56395 + + * Make sure the pipeline choice is propagated to the scanline wait + across a batch flush + https://bugs.freedesktop.org/show_bug.cgi?id=47597 + + * Set the valid drawable box when choosing placement of BLT composite ops + https://bugs.freedesktop.org/show_bug.cgi?id=47597 + + * Prevent use-after-free when promoting a partial-GPU bo to a full-GPU bo + https://bugs.freedesktop.org/show_bug.cgi?id=56591 + + * gen4 opacity spans require the per-rectangle workaround + https://bugs.freedesktop.org/show_bug.cgi?id=55500 + + * Prevent use of invalid damage pointers when redirecting rendering + https://bugs.freedesktop.org/show_bug.cgi?id=56785 + + +Release 2.20.12 (2012-10-20) +============================ +More bug reports, more bug fixes! Perhaps the headline feature is +that with a secure batches, coming to a 3.8 kernel near you, we may +finally have the ability to perform updates to the scanout synchronized +to the refresh rate on later SandyBridge and IvyBridge chipsets. It comes +at quite a power cost as we need to keep the GPU out of its power saving +modes, but it should allow legacy vsync to function at last. But this +should allow us to address a longstanding issue with tearing on +SandyBridge+. + + * Fix component-alpha rendering on IvyBridge, for example subpixel + antialiased glyphs. + https://bugs.freedesktop.org/show_bug.cgi?id=56037 + + * Flush before some "pipelined" state changes on gen4. The evidence is + that the same flushes as required on gen5+ are also required for gen4. + https://bugs.freedesktop.org/show_bug.cgi?id=55627 + + * Prevent a potential crash when forcing a stall on a busy CPU bo + https://bugs.freedesktop.org/show_bug.cgi?id=56180 + +[Release 2.20.11 contained a typo causing UXA to fail immediately.] + Release 2.20.10 (2012-10-14) ============================ The last couple of weeks have been fairly retrospective, a dive into |