Bug Team Status

Bug Team Status

[Jani Nikula]

The weekly bug team status for the past week.

== Daniel

Mostly looked at the ILK bug and stiched together a revert of the
shrinker changes. Still no idea about the underlying cause. Details in
the commit [1].

Otherwise the usual bug scrubbing activities with 1-2 failed patch
attempts.

[1] http://cgit.freedesktop.org/~danvet/drm-intel/commit/?id=93927ca52a55c23e0a6a305e7e9082e8411ac9fa

== Ville

Working on bug https://bugs.freedesktop.org/show_bug.cgi?id=46800

Reading CEA-861 and HDMI specs to get to the bottom of how the RGB
quantization range stuff should be handled.

Fixed the current code for ILK-IVB.

Made a prototype patch for fixing it on HSW+ using the pipe CSC block.
Tried to check the power consuptiom impact of the pipe CSC method on my
ILK laptop, didn't see any real difference in ACPI power numbers. Asked
Ouping to make some measurements too, no results yet.

Started to code up support for parsing the QS bit from EDID CEA block,
and based on that setting the Q bits in the AVI infoframe to inform the
display about the currently used quantization range. Unfortunately my
test TV doesn't support this stuff, so it's untested as of now.

Currently working on an "Automatic" mode for the "Broadcast RGB"
property, which would follow the guidelines set forth by CEA-861.
Hopefully it will result in the driver picking the correct quantization
range in most cases w/o user intervention.

I'll try to post the infoframe and automatic property patches still
today.

== Jani

Went through all the open bugs, getting an overview of it all. While at
it, flipped the NEEDINFO state as necessary; roughly a third of our bugs
are pending info from the reporters. About half of the bugs seem to be
KMS related. Particularly in the backlight department we may have a
bunch of dupes, but we prefer to keep them separate to be sure any fix
handles them all.

Debugged a number of bugs, requesting info as necessary, but failed at
cracking them. There was one better FORCEWAKE posting read related patch
attempt that seemingly fixed the issue, but Ben wants us to get to the
bottom of it [2], and rightly so.

[2] https://bugzilla.kernel.org/show_bug.cgi?id=52411

Bug Team Status

[Paulo Zanoni]

Weekly bug team status.

This week's team: Imre Deak, Jesse Barnes and Paulo Zanoni.

- Pointed some more people at the gpu reset patch for RC6 issues,
apparently the people seeing hangs aren't having luck with it.
- Work on gathering more documentation to solve some specific bugs
- Revisited Haswell bugs
- Investigated and tried to reproduce bugs about modes that don't work
reported on freedesktop.org. As part of this debugging, we concluded
that:
    o We need to report more display errors.  We have patches for this
which should land the mailing list soon.
    o We could do some adjustments on the watermarks code for some gens.
    o Our mode set sequence on SNB needs some adjustments.
- Also investigated some hotplug issues.

Thanks,
Paulo

-- 
Paulo Zanoni

Bug Team Status

[Rodrigo Vivi]

Many Closed GPU Hung Bugs (at fd.o):

Bug 55984 - Chris fixed this bug.
Bug 50619 - "sounds like the original issue was a combination of
busted hw/bios, and we're robbed of any chances to do further debug"
Daniel
Bug 59951 - "For 845g, you need to be either using SNA from
xf86-video-intel-2.20.19 (or later) or kernel-3.8 (or later)." Chris
Bug 54964 - I confirmed this bymyself last week. Got same cpu stepping
here and verified both WAs fixed this bug:  drm/i915: Implement
WaSetupGtModeTdRowDispatch && drm/i915: Implement
WaDisableHiZPlanesWhenMSAAEnabled
Bug 50545: Also fixed by Daniel's patch: drm/i915: Implement
WaDisableHiZPlanesWhenMSAAEnabled

Rodrigo still have to playing with his old (and new) VGA modes bugs.

fd.o resolved bug list: http://tinyurl.com/b8ukswa
fd.o open bugs order by update date: http://tinyurl.com/afxbovg

Thanks,
--
Rodrigo Vivi
Blog: http://blog.vivi.eng.br

Bug team status

[Antti Koskipää]

Another bug team week gone...

Antti:

Went through bugzilla, checking that work is still being done, general
stuff.

Mika:

Fixed https://bugs.freedesktop.org/show_bug.cgi?id=58230

Tried to verify a fix (drm/i915: disable shrinker lock
stealing for create_mmap_offset) for a problem i reported
but didn't manage to get ./gem_tiled_swapping to break anymore on
top of clean of clean drm-intel-fixes.

Rodrigo:

Worked on https://bugs.freedesktop.org/show_bug.cgi?id=51146
and https://bugs.freedesktop.org/show_bug.cgi?id=54964 and may have
found a VGA workaround for the latter one.


-- 
- Antti

Bug Team Status

[Damien Lespiau]

Another week, another bug team status:

== Paulo Zanoni

 Investigated and fixed a lot of VGA Haswell bugs with some general
bugzilla bookkeeping:
  - Fix our FDI link "retrain" code (if the first training attempt
failed, the subsequent ones would always fail)
  - Reject modes we can't set (we only have 2 FDI lanes)
  - Make more modes fit on 2 FDI lanes
  - Fix VGA after suspend/resume (init the PCH refclk)
  - Fix VGA black screens on CRW machines (set FDI RX polarity reversal bit)

None of these patches is upstream yet (need review), but they were
already tested by QA and confirmed that they will enable us to fix the
current top-priority Haswell bug.

== Damien Lespiau

Looked at some fdo bugs:

- [Bug 55901] [IVB/HSW]I-G-T/testdisplay -p fails with modes 1400x1050
and 1366x768 (fixed and verified)
- [Bug 50249] [VLV] verify LLC cache settings (closed)
- [Bug 55372] HDMI unplug detection after gen4 should check HPD before
EDID: Added the investigation so far, noone has found the equivalent
on the HPD live status bits, the closest that I could find was
SHOTPLUG_CTL and does not seem very useful
- Spend some time discovering the GTI (GT Interface) modules, learning
about the MMIO fifo and forcewake bits.

-- 
Damien