Intel driver updates: 2.5.99.2 and 2.6.0, plus a little about DRI2, Compiz and GEM


Judging from the most used searches and posts  (thank you wordpress stats), interest in the intel Xorg. driver is extremely high.  While it's easy to see that's true on the X.org mailing list, looking at my own blog stats was the first time I realized other end users were having issues getting the driver up and running too.

For those who are still having problems, you may want to try one of the recent updates to the driver.  The latest pre-release version is now 2.5.99.2, which came out January 2nd.  I've posted the updates included in that release below.

The latest 2.6.0 release was made on January 15th.  I've posted the changes for that release below as well.

As far as I can tell the 2.5.x series updates the existing driver architecture with new fixes while the 2.6.0 series are intended for use with the next release of X.org to utilize the GEM architecture in the kernel.  I'm not completely positive of this description of the differences, however.  If anyone has a clarification, please post a comment here.

An interesting post I found on the xorg mailing from Gordon Jin regarding some of the recent 2.5.xx releases:

  • 2.5.99.1 adds some new features like hdmi audio, XvMC for Mpeg2 on i965.
  • 2.5.99.2 are mostly bug fixes.

Known issues are listed at http://intellinuxgraphics.org/community_testing.html

These new features are all supported on i915 and later. We don't guarantee the stability on i8xx.
You can test DRI2 by setting
UXA in xorg.conf, then play with composite window manager.
GEM is [enabled] if you use 2.6.28 kernel.
KMS is not covered in this release package. It needs 2.6.29-ish kernel and not stable yet.

So it looks like the driver automatically knows whether or not to use GEM, which means you can grab the latest driver from git and compile for a pre-2.6.28 kernel and it should work (or at least I'm guessing that's the case).  DRI2 is supposed to address many of the issues related to use of the compiz eye candy problems many people are experiencing.  Apparently (and this comes from two different posts from  Adam K Kirchhoff on the xorg mailing list):

This is a limitation in the direct rendering infrastructure that DRI2 is set to address. You will just need to wait for DRI2 to be completed.

and

This effects all video card drivers that use Xorg's direct rendering infrastructure, including the open source and closed source drivers for ATI video cards. nvidia does not use that same infrastructure.

This means you should probably not be using compiz if you're using DRI for any driver (including the intel driver) except the nvidia driver.  I don't use compiz for anything (don't need the eye candy) so I can't comment on how accurate this statement is, but I can say at least one comment on my blog said that disabling compiz fixed problems with white screens when using the intel driver.


2.5.99.2 – January 8th, 2009

Date: Thu, 8 Jan 2009 16:04:55 +0800


Subject: [ANNOUNCE] xf86-video-intel 2.5.99.2

Dave Airlie (5):
modeset: fix xf86CrtcRotate API change across server versions
modeset: transformPresent is a new API member
exa: fix map gtt call to use current API
intel: don't call enter/leave VT for KMS enabled systems
intel/kms: disable overlay when it needs physical address

Eric Anholt (9):
uxa: Correctly prepare/finishaccess of stipple in ValidateGC (and only it)
Remove old mergedfb includes and defines, which bother spatch.
Rely on libdrm 2.4.3 and stop checking for xf86drmMode.h.
Add PCI write posting to LeaveVT path when we're about to wait on write results.
uxa: Do a hack to use the aperture mapping instead of bo_map in sw fallbacks.
Don't touch the pipestat regs for detecting FIFO underrun. The kernel owns them.
Disable DRI2 buffer tiling on non-965, as those need fence regs for 2D blits.
Fix pin leakage with EXA GTT-mapping shortcut, and crash with UXA on KMS.
Always enable KMS if server's new enough, and remove option.

Henry unbongo (1):
Add support for SDVO LVDS.

Keith Packard (3):
Handle drm_bo_map failure in 965 video and composite paths.
uxa: handle uxa_prepare_access failure
FatalError on batchbuffer map failure

Kristian H=F8gsberg (1):
Quiet some KMS warnings.

Ma Ling (1):
SDVO: reset pixel repeat in avi frame

Peter Alfredsen (1):
xvmc: fix up needed libs

Robert Noland (1):
Fix drmOpen for non-linux 965 XvMC

Zhenyu Wang (4):
Don't count vertex buffer in second aperture size check
Bug #18004: Add Aopen 915GM LVDS quirk
UXA: Fallback to dri_bo_map() if pin failed
Bump version 2.5.99.2

git tag: xf86-video-intel-2.5.99.2

http://xorg.freedesktop.org/archive/individual/driver/xf86-video-intel-2.5.99.2.tar.bz2
MD5: 3894f1a9024480fd5d26fe0774f3f405  xf86-video-intel-2.5.99.2.tar.bz2
SHA1: 8b99ff57e6b35278c112831fb839859371da0ac2  xf86-video-intel-2.5.99.2.tar.bz2

http://xorg.freedesktop.org/archive/individual/driver/xf86-video-intel-2.5.99.2.tar.gz
MD5: c12c462271e7edab247b0072c69a945f  xf86-video-intel-2.5.99.2.tar.gz
SHA1: bc933ef3ed97044f9e374177239ae09caafa6167  xf86-video-intel-2.5.99.2.tar.gz


2.6.0 – January 15th, 2009

Date: Thu, 15 Jan 2009 14:21:48 +0800


Subject: [ANNOUNCE] xf86-video-intel 2.6.0

Here's xf86-video-intel 2.6.0 release. Full changelog against 2.5.1 is below.
We had DRI2 and 965 XvMC branch merged, and other bunch of fixes. We also
have basic support for SDVO LVDS from last rc.

This'll be included in Intel 2008 Q4 release, and Gordon Jin will update
on that build soon.

Thanks all contributors! Please test and report issue you'd found to us!

ChangeLog from 2.5.1:

Adam Jackson (1):
Quirk: No LVDS on dell Studio Hybrid

Alan Coopersmith (1):
Remove xorgconfig & xorgcfg from See Also list in man page

Bernhard Rosenkraenzer (1):
UXA: Declare glyph cache picture as component-alpha when necessary.

Bryce Harrington (2):
PipeA quirk for Quanta/W251U (launchpad bug #244242)
Pipe-A quirk for HP 2730p (bug #18852)

Carl Worth (13):
Ignore intel_gtt binary
Rename gen4_state_t to gen4_static_state_t
Rename gen4_state_init to gen4_static_state_init
Rename state_base_offset to static_state_offset in gen4_static_state_init
Use consistent idiom for obtaining static_state
Use buffer object for vertex buffer (in new gen4_dynamic_state)
965: Move composite setup to new _emit_batch_header_for_composite
Rename gen4_dynamic_state to gen4_vertex_buffer
Unreference the vertex_buffer_bo in gen4_render_state_cleanup
Use buffer objects for binding table and surface-state objects.
i965: Add batch_flush_notify hook to create new vertex-buffer bo
Don't smash fixed_mode if skip_panel_detect is set.
Set vertex_buffer_bo to NULL after unreference.

Dave Airlie (7):
Default kernel mode setting to off, add configure flag to enable
uxa: don't call composite routines with no buffer.
modeset: fix xf86CrtcRotate API change across server versions
modeset: transformPresent is a new API member
exa: fix map gtt call to use current API
intel: don't call enter/leave VT for KMS enabled systems
intel/kms: disable overlay when it needs physical address

Eric Anholt (35):
Don't set up sarea or drm mappings in DRI2 mode.
UXA: Re-enable non-965 render.
DRI2: Emit the MI_FLUSH before flushing batch in swapbuffers.
DRI2: Move pixmap pitch alignment for use with depth to pixmap create.
Fix build failures that should have been in the previous merge commit.
Remove the CheckDevices timer.
Make I830FALLBACK debugging a runtime instead of compile-time option.
i965: Support render acceleration with pixmaps in BOs.
Remove DRI_MM defines which are always true now.
UXA: Add support for tiled front/back/depth by cutting over to the GTT map.
Re-enable composite accel on 965 with UXA.
Enable tiling for DRI2 back/depth buffers.
Move debug code for I965DisplayVideoTextured to separate functions.
Move I965DisplayVideoTextured surface/sampler setup to separate functions.
Move I965DisplayVideoTextured unit state setup to separate functions.
Move i965 video cc state to BOs.
Move i965 video vs/sf state to BOs.
Stop allocating unused scratch space for i965 video.
Move i965 video wm and sampler state to BOs.
Move remaining i965 video programs to BOs.
Move i965 video vertex data to BOs.
Move i965 video surface state and binding table to BOs.
Emit proper relocations to pixmaps in BOs in i965 video.
Remove the extra memory allocation for 965 video state now that it's all in BOs.
uxa: Add in EnableDisableFBAccess handling like examodule.c did.
uxa: Reject solid/copy to under-8bpp destinations.
uxa: Correctly prepare/finishaccess of stipple in ValidateGC (and only it)
Remove old mergedfb includes and defines, which bother spatch.
Rely on libdrm 2.4.3 and stop checking for xf86drmMode.h.
Add PCI write posting to LeaveVT path when we're about to wait on write results.
uxa: Do a hack to use the aperture mapping instead of bo_map in sw fallbacks.
Don't touch the pipestat regs for detecting FIFO underrun. The kernel owns them.
Disable DRI2 buffer tiling on non-965, as those need fence regs for 2D blits.
Fix pin leakage with EXA GTT-mapping shortcut, and crash with UXA on KMS.
Always enable KMS if server's new enough, and remove option.

Henry unbongo (1):
Add support for SDVO LVDS.

Ian Romanick (1):
Prevent redefinitions of CARD8 and friends.  Fixes build.

Jesse Barnes (3):
Don't modify render standby if kernel mode setting is active
Default to FULL_ASPECT panel fitting
Make sure DRM library paths are included

Julien Cristau (1):
Don't unconditionally define DRI2

Keith Packard (8):
Fix mis-merge of DRI2 changes related to pI830->directRenderingType
Use long crt hotplug activation time on GM45.
Add RandR 1.3 panning support by supporting the crtc set_origin function
Handle drm_bo_map failure in 965 video and composite paths.
uxa: handle uxa_prepare_access failure
FatalError on batchbuffer map failure
Fix LVDS EDID to match all possible default modes
Assign rotation memory dri_bo to rotation pixmap.

Kristian H=F8gsberg (7):
Add DRI2 support.
Fix broken test for DRI1 in DRI2 conversion.
Update to DRI2 changes.
Fix KMS compilation.
Simplify crtc preinit a bit.
Make sure DRI/DRI2 can initialize properly with KMS.
Quiet some KMS warnings.

Ma Ling (4):
enable Intel G35 SDVO HDMI audio output
SDVO: reset pixel repeat in avi frame
set continuous-frequency flag in get modes function
Disable VGA plane reliably

Maxim Levitsky (1):
Add an option to make the overlay be the first XV adaptor.

Owain G. Ainsworth (2):
use ifdef __linux__ where needed.
Fix ioctl type.

Paulo Cesar Pereira de Andrade (2):
Export libIntelXvMC and libI80XvMC symbols.

Peter Alfredsen (1):
xvmc: fix up needed libs

Robert Lowery (1):
TV: add support to set TV margins in xorg.conf

Robert Noland (1):
Fix drmOpen for non-linux 965 XvMC

Wu Fengguang (3):
introduce i830_hdmi_priv.has_hdmi_sink
enable Intel G45 integrated HDMI audio output
refresh batch_bo reference after intel_batch_flush()

Zhenyu Wang (30):
Make IS_GM45 into IS_G4X define
SDVO: fix wrong order of sdvo version's major/minor
SDVO: add HDMI audio encrypt change bit for GetInterruptEventSource command
SDVO: fix sdvo tv format and sdtv resolution request/reply definition
SDVO: add GetScaledHDTVResolutionSupport command
SDVO: add command for set monitor power state
SDVO: fix more command definition errors
TV: white space cleanup
TV: fix default contrast and saturation modifier
TV: save serveral TV_CTL register fields in mode set
TV: fix timing parameters for PAL, 480p, 1080i
TV: subcarrier fix for NTSC and PAL
quirk LVDS on Asus Eee box
TV: fix contrast and saturation for 915/945G
Disable XvMC on DRI2 for now
Fix maxium backlight level restored issue
Fix TV compiling warning
Fix DRI2 compiling warning
Remove Cappuccino SlimPRO SP625F 855GM LVDS quirk
bug #17395: Quirk CRT for sony VGC-LT71DB
Try to always probe SDVOC on 965G/965GM
Fix directRenderingType check
Don't warn on ring enabled in GEM
Bump version 2.5.99.1
Don't count vertex buffer in second aperture size check
Bug #18004: Add Aopen 915GM LVDS quirk
UXA: Fallback to dri_bo_map() if pin failed
Bump version 2.5.99.2
Revert “Assign rotation memory dri_bo to rotation pixmap.”
Bump version 2.6.0

Zou Nan hai (13):
965 xvmc, current only frame based DCT_type support
[965-xvmc] some fixes for earlier 965 chips
Fix from He Shuang(shuang.he@intel.com),
[965-xvmc] support field prediction and dct type, optimize in GPU
[965-xvmc] add missing g4a file
[965-xvmc] half-pixel bug fix
[965-xvmc] some shader optimize
[965-xvmc] There is some issue with uv half pixel, remove it till fix
[965-xvmc] optimize, pack some payload
[965-xvmc] IGD support
[965-xvmc] 965q indirect media data workaround
[965-xvmc] fix dual prime kernel, flush issue on G4x
Update binary versions of the dual-prime kernels

root (1):
[i965-xvmc] bypass copy when put image with xvmc

git tag: xf86-video-intel-2.6.0

http://xorg.freedesktop.org/archive/individual/driver/xf86-video-intel-2.6.0.tar.bz2
MD5: 70f23db0f5178f63b4d688e6fc4a43de  xf86-video-intel-2.6.0.tar.bz2
SHA1: bc1040a2a2f399b0d141d3addd4735a13c1bacb4  xf86-video-intel-2.6.0.tar.bz2

http://xorg.freedesktop.org/archive/individual/driver/xf86-video-intel-2.6.0.tar.gz
MD5: 6c3071b6655fa0acfda1ae46eceb5039  xf86-video-intel-2.6.0.tar.gz
SHA1: 8545dafd077a715bd3c7a83a5b15a2fb07b71a49  xf86-video-intel-2.6.0.tar.gz

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.