[regression] agp/intel: Sandybridge doesn't require GMCH enabling

[regression] agp/intel: Sandybridge doesn't require GMCH enabling

[Alexey Fisher]

after commit e380f60b22eddec7825224b8d788572c82b63161
intel driver filed to load, so system fallback to vesa driver.

here is the interisting part of the dmesg:
[    2.559921] agpgart-intel 0000:00:00.0: detected 32768K stolen memory
[    2.559957] pci 0000:00:02.0: failed to enable the GTT: GMCH_CTRL=350
[    2.559988] agpgart-intel 0000:00:00.0: error configuring host
chipset
[    2.560069] agpgart-intel 0000:00:00.0: agp_backend_initialize()
failed
[    2.560077] agpgart-intel: probe of 0000:00:00.0 failed with error
-22

[    2.560252] [drm] Initialized drm 1.1.0 20060810
[    2.560272] [drm:i915_init] *ERROR* drm/i915 can't work without
intel_agp module!



last working version 2.6.36-07538-gc584fe4

my hardware, Intel DG45ID board, 
intel_stepping 
Vendor: 0x8086, Device: 0x2e22, Revision: 0x03 (A3)
-- 
Regards,
        Alexey

Re: [regression] agp/intel: Sandybridge doesnt require GMCH enabling

[Chris Wilson]

On Sun, 31 Oct 2010 11:29:18 +0100, Alexey Fisher <bug-track@fisher-privat.net> wrote:
> after commit e380f60b22eddec7825224b8d788572c82b63161
> intel driver filed to load, so system fallback to vesa driver.

*grin*

I'm looking right now at how many systems we were poking bits in
registers that were undefined.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre