On Wed, 2008-01-16 at 11:47 +1030, Daniel O'Connor wrote: > On Wed, 16 Jan 2008, Eric Anholt wrote: > > I took a quick review of the docs again and I think it's almost > > complete. The gtt_size detection is broken, though. We should have > > gtt_size (the amount of graphics stolen memory used for storing the > > GTT data) equal to zero on this hardware, instead of using the G965 > > path. However, this also requires current xf86-video-intel, which > > fixes the mistaken assumption we originally had for xf86-video-intel > > and linux agp that the gtt stolen memory was stolen out of graphics > > stolen memory. > > OK.. Not sure what GTT is though :)
Think page tables on your CPU, but for the graphics device's virtual addressing inside of the AGP aperture. > > Updating just the AGP driver for fixing the bad stolen memory > > assumption means that the 2d driver tries to bind memory lower than > > the AGP driver will let you, and your server won't start. The > > "proper" fix would be for us to remember the stolen memory address > in > > the GTT and let you bind over them, but restore them on unbind. > Some > > day we would love to release this memory to the general page > > allocator and get rid of this stolen memory disaster. > > > > To test, install xf86-video-intel from ports, uncomment those PCI > > IDs, and start X. If things are working, then it probably means > > you're good to go and we can enable it by default. If you're > getting > > the "agp0: trying to bind into stolen memory" complaint, you need > > current 2D from > > git://anongit.freedesktop.org/git/xorg/driver/xf86-video-intel and > we > > should cherry-pick the relevant patch into our ports. > With the ports xf86-video-intel I get.. > (==) intel(0): Backing store disabled > (==) intel(0): Silken mouse enabled > (II) intel(0): Initializing HW Cursor > (II) intel(0): xf86BindGARTMemory: bind key 1 at 0x006ff000 (pgoffset 1791) > (WW) intel(0): xf86BindGARTMemory: binding of gart memory with key 1 > at offset 0x6ff000 failed (Invalid argument) > > Fatal server error: > Couldn't bind memory for front buffer > > I tried to build your git repo of the driver but I am getting.. > gmake[3]: Entering directory `/local0/tmp/xf86-video-intel/src' > /usr/local/bin/bash ../libtool --tag=CC --mode=compile gcc -DHAVE_CONFIG_H > -I. -I.. -Wall -Wpointer-arith -Wstrict-prototypes -Wmissing-prototypes > -Wmissing-declarations -Wnested-externs -fno-strict-aliasing > -D_THREAD_SAFE -I/usr/local/include/xorg -I/usr/local/include/pixman-1 > -I/usr/local/include -I/usr/local/include -I/usr/local/include/drm > -I/usr/local/include/X11/dri -DI830_XV -DI830_USE_XAA -DI830_USE_EXA -g -O2 > -MT i810_driver.lo -MD -MP -MF .deps/i810_driver.Tpo -c -o i810_driver.lo > i810_driver.c > gcc -DHAVE_CONFIG_H -I. -I.. -Wall -Wpointer-arith -Wstrict-prototypes > -Wmissing-prototypes -Wmissing-declarations -Wnested-externs > -fno-strict-aliasing -D_THREAD_SAFE -I/usr/local/include/xorg > -I/usr/local/include/pixman-1 -I/usr/local/include -I/usr/local/include > -I/usr/local/include/drm -I/usr/local/include/X11/dri -DI830_XV > -DI830_USE_XAA -DI830_USE_EXA -g -O2 -MT i810_driver.lo -MD -MP -MF > .deps/i810_driver.Tpo -c i810_driver.c -fPIC -DPIC -o .libs/i810_driver.o > i810_driver.c:456: error: `PACKAGE_VERSION_MAJOR' undeclared here (not in a > function) > i810_driver.c:456: error: initializer element is not constant > i810_driver.c:456: error: (near initialization for > `intelVersRec.majorversion') > i810_driver.c:456: error: `PACKAGE_VERSION_MINOR' undeclared here (not in a > function) > i810_driver.c:456: error: initializer element is not constant > i810_driver.c:456: error: (near initialization for > `intelVersRec.minorversion') > i810_driver.c:456: error: `PACKAGE_VERSION_PATCHLEVEL' undeclared here (not > in a function) > ... Your xorg-macros package is stale. > I modified config.h to add those #defines and the man Makefile (configure is > fun!) and installed but now I get.. > (==) intel(0): Write-combining range (0xa0000,0x10000) was already clear > (II) EXA(0): Offscreen pixmap area of 30720000 bytes > (II) EXA(0): Driver registered support for the following operations: > (II) Solid > (II) Copy > (II) Composite (RENDER acceleration) > (==) intel(0): Backing store disabled > (==) intel(0): Silken mouse enabled > (II) intel(0): Initializing HW Cursor > (II) intel(0): xf86BindGARTMemory: bind key 8 at 0x006ff000 (pgoffset 1791) > (WW) intel(0): xf86BindGARTMemory: binding of gart memory with key 8 > at offset 0x6ff000 failed (Invalid argument) > > Fatal server error: > Couldn't bind memory for front buffer That's almost surely as a result of the stolen memory issue. > > To be sure that everything's working, also enable DRI by installing > > updated DRM from git://anongit.freedesktop.org/git/mesa/drm (bsd-core > > directory, make all install). It may complain if you have witness > > enabled, since I haven't brought over the witness fix from -current > > yet. > > OK. > I really only want the 2D driver so DPMS works and the screen saver can > turn the monitor off (although faster 2D would be good as well) DPMS should work with the VESA driver even, not to discourage you from the right solution. -- Eric Anholt [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED]
signature.asc
Description: This is a digitally signed message part
