Bug 67122

Summary: [ivb regression] "[drm:__gen6_gt_force_wake_mt_get] *ERROR* Timed out waiting for forcewake old ack to clear." is back on boot
Product: DRI Reporter: Chris Wilson <chris>
Component: DRM/IntelAssignee: Daniel Vetter <daniel>
Status: CLOSED FIXED QA Contact: Intel GFX Bugs mailing list <intel-gfx-bugs>
Severity: normal    
Priority: medium    
Version: unspecified   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
Clear forcewake before use
none
fix up gt init sequence ordering none

Description Chris Wilson 2013-07-20 16:14:03 UTC
Fallout from

commit 7dcd2677ea912573d9ed4bcd629b0023b2d11505
Author: Konstantin Khlebnikov <khlebnikov@openvz.org>
Date:   Wed Jul 17 10:22:58 2013 +0400

    drm/i915: fix long-standing SNB regression in power consumption after resume
Comment 1 Chris Wilson 2013-07-20 16:22:31 UTC
Created attachment 82733 [details] [review]
Clear forcewake before use
Comment 2 Daniel Vetter 2013-07-21 10:55:22 UTC
Oops. Can you please rebase on top of -fixes, add all the relevant cc's and submit the patch to intel-gfx? I'd like to send my -fixes pile to Dave early next week.

Looking at it a bit more I think on -fixes we only need


diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 5c0663f..ed32c43 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -1593,8 +1593,8 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
        intel_detect_pch(dev);
 
        intel_irq_init(dev);
-       intel_gt_init(dev);
        intel_gt_reset(dev);
+       intel_gt_init(dev);
 
        /* Try to make sure MCHBAR is enabled before poking at it */
        intel_setup_mchbar(dev);

Since that reordering seems to have been an unintentional part of the offending patch.
Comment 3 Chris Wilson 2013-07-21 11:03:38 UTC
Hmm, I thought it was intentional, but reset() doesn't look like it needs init(). Ugh, I'm really disliking this ordering and naming, which is why I felt like adding it to early_sanitize was a good step forward. I think I would like to move the write(RC6_CTL, 0); there as well for clarity.

It's your patch now. :)
Comment 4 Daniel Vetter 2013-07-21 11:15:45 UTC
Created attachment 82772 [details] [review]
fix up gt init sequence ordering
Comment 5 Chris Wilson 2013-07-27 20:41:09 UTC
commit 181d1b9e31c668259d3798c521672afb8edd355c
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Sun Jul 21 13:16:24 2013 +0200

    drm/i915: fix up gt init sequence fallout

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.