Merge branch 'drm-intel-fixes' of git://people.freedesktop.org/~danvet/drm-intel...

Daniel writes: (this pull is the one with the bad patch dropped)
First pile of fixes for 3.6 already, and I'm afraid it's a bit larger than
what I'd wish for. But I've moved all the feature-y stuff to -next, so
this really is all -fixes. Most of it is handling fallout from the hw
context stuff, discovered now that mesa git has started using them for
real. Otherwise all just small fixes:
- unbreak modeset=0 on gen6+ (regressed in next)
- const mismatch fix for ->mode_fixup
- simplify overly clever lvds modeset code (current code can totally
  confuse backlights, resulting in broken panels until a full power draw
  restores them).
- fix some fallout from the flushing_list disabling (regression only
  introduced in -next)
- DP link train improvements (this also kills the last 3.2 dp regression
  afaik)
- bugfix for the new ddc VGA detection on newer platforms
- minor backlight fixes (one of them a -next regression)
- only enable the required PM interrupts (to avoid waking up the cpu
  unnecessarily)
- some really minor bits (workaround clarification, make coverty happy,
  hsw init fix)
* 'drm-intel-fixes' of git://people.freedesktop.org/~danvet/drm-intel: (23 commits)
  drm/i915: unbreak lastclose for failed driver init
  drm/i915: Set the context before setting up regs for the context.
  drm/i915: constify mode in crtc_mode_fixup
  drm/i915/lvds: ditch ->prepare special case
  drm/i915: dereferencing an error pointer
  drm/i915: fix invalid reference handling of the default ctx obj
  drm/i915: Add -EIO to the list of known errors for __wait_seqno
  drm/i915: Flush the context object from the CPU caches upon switching
  drm/i915: Make the lock for pageflips interruptible
  drm/i915: don't forget the PCH backlight registers
  drm/i915: Insert a flush between batches if the breadcrumb was dropped
  drm/i915: missing error case in init status page
  drm/i915: mask tiled bit when updating ILK sprites
  drm/i915: try to train DP even harder
  drm/i915: kill intel_ddc_probe
  drm/i915: check whether we actually received an edid in detect_ddc
  drm/i915: fix up PCH backlight #define mixup
  drm/i915: Add comments to explain the BSD tail write workaround
  drm/i915: Disable the BLT on pre-production SNB hardware
  drm/i915: initialize power wells in modeset_init_hw
  ...

diff --cc drivers/gpu/drm/i915/i915_irq.c
Simple merge

diff --cc drivers/gpu/drm/i915/intel_display.c
index bc25a24,e9e476e..f615976
--- 1/drivers/gpu/drm/i915/intel_display.c
--- 2/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@@ -7172,8 -7175,26 +7175,13 @@@ static void i915_disable_vga(struct drm
        POSTING_READ(vga_reg);
  }
  
 -static void ivb_pch_pwm_override(struct drm_device *dev)
 -{
 -      struct drm_i915_private *dev_priv = dev->dev_private;
 -
 -      /*
 -       * IVB has CPU eDP backlight regs too, set things up to let the
 -       * PCH regs control the backlight
 -       */
 -      I915_WRITE(BLC_PWM_CPU_CTL2, BLM_PWM_ENABLE);
 -      I915_WRITE(BLC_PWM_CPU_CTL, 0);
 -      I915_WRITE(BLC_PWM_PCH_CTL1, BLM_PCH_PWM_ENABLE | BLM_PCH_OVERRIDE_ENABLE);
 -}
 -
  void intel_modeset_init_hw(struct drm_device *dev)
  {
+       /* We attempt to init the necessary power wells early in the initialization
+        * time, so the subsystems that expect power to be enabled can work.
+        */
+       intel_init_power_wells(dev);
+ 
        intel_prepare_ddi(dev);
  
        intel_init_clock_gating(dev);