<html>
<head>
<base href="https://bugs.freedesktop.org/">
</head>
<body>
<p>
<div>
<b><a class="bz_bug_link
bz_status_NEW "
title="NEW - [CI] igt@kms_flip@*-vs-vblank-race* - Failed assertion: end - start > 0.9 * mode_frame_time(o) && end - start < 2.1 * mode_frame_time(o)"
href="https://bugs.freedesktop.org/show_bug.cgi?id=103060#c25">Comment # 25</a>
on <a class="bz_bug_link
bz_status_NEW "
title="NEW - [CI] igt@kms_flip@*-vs-vblank-race* - Failed assertion: end - start > 0.9 * mode_frame_time(o) && end - start < 2.1 * mode_frame_time(o)"
href="https://bugs.freedesktop.org/show_bug.cgi?id=103060">bug 103060</a>
from <span class="vcard"><a class="email" href="mailto:stanislav.lisovskiy@intel.com" title="Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>"> <span class="fn">Stanislav Lisovskiy</span></a>
</span></b>
<pre>Looked into test case wf_blank-ts-check-interruptible results in the same run
and found one interesting correlation: in case when we have those failures we
also have that one:
Using monotonic timestamps
Starting subtest: wf_vblank-ts-check-interruptible
Beginning wf_vblank-ts-check-interruptible on pipe A, connector eDP-1
3840x2160 60 3840 3888 3920 4000 2160 2163 2168 2222 0xa 0x48 533250
Expected frametime: 16668us; measured 17955.9us +- 5.512us accuracy 0.09%
vblank interval differs from modeline! expected 16667.6us, measured 17956us +-
5.512us, difference 1288.3us (233.7 sigma)
wf_vblank-ts-check-interruptible on pipe A, connector eDP-1: PASSED
...
Beginning wf_vblank-ts-check-interruptible on pipe C, connector eDP-1
3840x2160 60 3840 3888 3920 4000 2160 2163 2168 2222 0xa 0x48 533250
Expected frametime: 16668us; measured 17955.9us +- 5.702us accuracy 0.10%
vblank interval differs from modeline! expected 16667.6us, measured 17956us +-
5.702us, difference 1288.3us (225.9 sigma)
In normal case two vblanks take ~33297 which approximately twice as expected
frame time, however there seems to be cases when the actual vblank interval
gets different: 17956us will give us value rather close or exceeding boundary
condition in assert. So basically this is not about some overhead in
vblank_wait or other code path, but it is really the vblanking interval itself
which is getting different from expected one.
I wonder why we don't consider this an error in
wf_vblank-ts-check-interruptible as it clearly leads to this fail as a
consequence. We should either then not consider this an issue here either then
and compare to the actual frame time or wf_vblank-ts-check-interruptible should
also fail. Any thoughts?</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are on the CC list for the bug.</li>
<li>You are the QA Contact for the bug.</li>
<li>You are the assignee for the bug.</li>
</ul>
</body>
</html>