[Bug 76103] [915GM regression] black screen on boot Linux >= 3.14-rc1

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=76103

--- Comment #3 from Ville Syrjala <ville.syrjala at linux.intel.com> ---
Right, so previously we were failing to enable self refresh since we did the
INSTPM write incorrectly, but now that we do enable it, something fails. I
don't see any underrun reports in the logs though.

Can you run something like gem_exec_nop from intel-gpu-tool after a boot where
the display went blank, but before starting X? That should cause the GPU to
generate some interrupts which would trick the underrun report out if it was
there.

Also can you attach a dmesg log which includes the part where you start X and
the display comes back to life?

Chris, I suppose you could be right about the bit. Maybe it's not really safe
to poke it while the pipes/planes are active. But if we change that then we
probably need to change the watermark code to always compute the SR watermark
regardless of how many pipes are active. Not quite sure what we should stick
there though, probably the worst case value for all pipes. Also if we compute
the SR watermark and it comes out invalid, we should still have a way to
disable SR entirely, even with just one active pipe. I don't see any other bit
for doing that. So for that reason I'm a bit sceptical about this idea.

I did spot a few potential issues with the way we calculate the SR watermark.
I'll attach a few patches for those.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20140313/77c76a14/attachment.html>