Skip to contentSkip to conversation

drm/i915: Fix KMS regression on Sandybridge/CPT

PATCH Applied Zhenyu Wang 2 participants 8 messages Oct 28, 2010 – Nov 4, 2010 1 message not in the archive

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/i915: Fix KMS regression on Sandybridge/CPT

We should enable FDI normal training on Sandybridge/CPT system as well. Also restore back some original register posting read that got removed. LVDS/VGA/HDMI seems back to life but DP still fails.

---
 drivers/gpu/drm/i915/i915_drv.h      |    1 +
 drivers/gpu/drm/i915/intel_display.c |   80 +++++++++++++++++++++++-----------
 2 files changed, 55 insertions(+), 26 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +1 -0
index 5948e05..228c571 100644
@@ -1322,6 +1322,7 @@ static inline void i915_write(struct drm_i915_private *dev_priv, u32 reg,
 
 #define INTEL_PCH_TYPE(dev) (((struct drm_i915_private *)(dev)->dev_private)->pch_type)
 #define HAS_PCH_CPT(dev) (INTEL_PCH_TYPE(dev) == PCH_CPT)
+#define HAS_PCH_IBX(dev) (INTEL_PCH_TYPE(dev) == PCH_IBX)
 
 #define PRIMARY_RINGBUFFER_SIZE         (128*1024)
 
drivers/gpu/drm/i915/intel_display.c +54 -26

152 lines not on this page · raw message

Signature · 1 line
1.7.1
#1 lore raw
Quoted 4 lines · Zhenyu Wang

We should enable FDI normal training on Sandybridge/CPT system as well. Also restore back some original register posting read that got removed. LVDS/VGA/HDMI seems back to life but DP still fails.

The POSTING_READs you added are no-ops since they are all followed by a read. The transconf should have the bpc in the register at this point or else we should have bugged much earlier. The break-exec-wait condition is still documented for gen6, augmented with an additional break-sempahore-wait and obviously per-ring, but this is unrelated to the changelog.

I've shrunk the patch to just the FDI portion, pushed to staging for review. Let's try to keep patches as minimal as possible and addressing a single issue. (I'm as guilty of violating that as anyone.) Mostly so that we have a clear history of why/how the code works (and I don't break it later).
-Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
#2 lore raw
Quoted 9 lines · Chris Wilson

The POSTING_READs you added are no-ops since they are all followed by a read. The transconf should have the bpc in the register at this point or else we should have bugged much earlier. The break-exec-wait condition is still documented for gen6, augmented with an additional break-sempahore-wait and obviously per-ring, but this is unrelated to the changelog.

I've shrunk the patch to just the FDI portion, pushed to staging for review.

Current drm-intel-staging still fails unfortunately..

Quoted 4 lines

Let's try to keep patches as minimal as possible and addressing a single issue. (I'm as guilty of violating that as anyone.) Mostly so that we have a clear history of why/how the code works (and I don't break it later).

yeah, will do that, thanks for remind.

Signature · 3 lines
Open Source Technology Center, Intel ltd.

$gpg --keyserver wwwkeys.pgp.net --recv-keys 4D781827

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#3 lore raw
Quoted 5 lines · Zhenyu Wang
Quoted 2 lines · Chris Wilson

I've shrunk the patch to just the FDI portion, pushed to staging for review.

Current drm-intel-staging still fails unfortunately..

I'm very interested to know which was the magic bit then. :)

Did we lose the transcoder bpc setting? Is an extra READ masking a missing delay? Is the break-scanline-wait fouling gen6?
-Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
#4 lore raw
Quoted 11 lines · Chris Wilson
Quoted 5 lines · Zhenyu Wang
Quoted 2 lines · Chris Wilson

I've shrunk the patch to just the FDI portion, pushed to staging for review.

Current drm-intel-staging still fails unfortunately..

I'm very interested to know which was the magic bit then. :)

Did we lose the transcoder bpc setting? Is an extra READ masking a missing delay? Is the break-scanline-wait fouling gen6?

I don't think transcoder bpc setting should matter, but sorry that I'm short of time to track down which one extra read made the difference, my sandybridge laptop normally refuse to boot on first time.. ;)

The code operation is same as what we have in .36 kernel, so could you restore behavior back first?

thanks.

Signature · 3 lines
Open Source Technology Center, Intel ltd.

$gpg --keyserver wwwkeys.pgp.net --recv-keys 4D781827

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#5 lore raw
Quoted 6 lines · Zhenyu Wang

I don't think transcoder bpc setting should matter, but sorry that I'm short of time to track down which one extra read made the difference, my sandybridge laptop normally refuse to boot on first time.. ;)

The code operation is same as what we have in .36 kernel, so could you restore behavior back first?

The original patch was unacceptable since it did more than it claimed to in its changelog. Just disabling the Ironlake workaround and restoring the FDI normal train on crtc disable is insufficient. I remain dubious that adding the POSTING_READs is sufficient without at least some explanation and some testing.

I could break the original patch down into the 5 separate patches (of which it sounds like we only need 3) and I would still be waiting on someone to test them...
-Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre

A message that is not in the archive (sent off-list or not yet received): [email protected]

#7 lore raw
Quoted 16 lines · Zhenyu Wang
Quoted 5 lines · Chris Wilson
Quoted 4 lines · Zhenyu Wang

I don't think transcoder bpc setting should matter, but sorry that I'm short of time to track down which one extra read made the difference, my sandybridge laptop normally refuse to boot on first time.. ;)

Quoted 9 lines
Quoted 2 lines

The code operation is same as what we have in .36 kernel, so could you restore behavior back first?

The original patch was unacceptable since it did more than it claimed to in its changelog. Just disabling the Ironlake workaround and restoring the FDI normal train on crtc disable is insufficient. I remain dubious that adding the POSTING_READs is sufficient without at least some explanation and some testing.

Quoted 8 lines

Chris, I try to retest your changed version on drm-intel-staging. It looks things work fine now with that. ;)

Sorry that I'm not quite sure why my last test failed...please pick that one to -fixes, so QA team could pick it up for validation.

Thanks.

oops, this was rejected, now with correct subscribed mail address to the list..

Signature · 3 lines
Open Source Technology Center, Intel ltd.

$gpg --keyserver wwwkeys.pgp.net --recv-keys 4D781827

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

Applied
#8 lore raw
In reply to a message not in the archive
Quoted 5 lines · Zhenyu Wang

Chris, I try to retest your changed version on drm-intel-staging. It looks things work fine now with that. ;)

Sorry that I'm not quite sure why my last test failed...please pick that one to -fixes, so QA team could pick it up for validation.

Applied to -fixes, thanks for following up.
-Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre