<html> <head> <base href="https://bugs.freedesktop.org/" /> </head> <body> <div> <a class="bz_bug_link bz_status_NEW " title="NEW --- - [ILK]igt/kms_render/gpu-blit randomly causes system hang" href="https://bugs.freedesktop.org/show_bug.cgi?id=67243#c24">Comment # 24</a> on <a class="bz_bug_link bz_status_NEW " title="NEW --- - [ILK]igt/kms_render/gpu-blit randomly causes system hang" href="https://bugs.freedesktop.org/show_bug.cgi?id=67243">bug 67243</a> from <a class="email" href="mailto:imre.deak@gmail.com" title="Imre Deak <imre.deak@gmail.com>"> Imre Deak</a> <pre>(In reply to <a href="show_bug.cgi?id=67243#c23">comment #23</a>) > Created <a href="attachment.cgi?id=85305" name="attach_85305" title="dmesg with patch comment 21">attachment 85305</a> <a href="attachment.cgi?id=85305&action=edit" title="dmesg with patch comment 21">[details]</a> > dmesg with patch <a href="show_bug.cgi?id=67243#c21">comment 21</a> Ok, thanks a lot for testing these. I didn't get much closer to the root cause but at least we eliminated some possible causes. To summarize, a loop as in <a href="show_bug.cgi?id=67243#c17">comment 17</a> is enough to trigger the problem and the hang seems to happen in intel_disable_plane(). The wrong ilk disable sequence I mentioned is only a red-herring, since the port gets only disabled in encoder->post_disable(), so we can forget about that. One more pattern I noticed in your dmesg is regular fifo underflows, that happen consistently with certain modes, resulting in lower watermark thresholds. And the hang happened in all cases with the same threshold values. Atm, no idea why you get those, I don't get any underflows on my ilk even with your timings.</pre> </div> <hr> You are receiving this mail because: <ul> <li>You are the QA Contact for the bug.</li> </ul> </body> </html>