<div dir="ltr"><div dir="ltr">On Mon, May 11, 2020 at 9:55 PM Seth Forshee <<a href="mailto:seth.forshee@canonical.com">seth.forshee@canonical.com</a>> wrote:<br></div><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">On Mon, May 11, 2020 at 08:49:49AM -0500, Seth Forshee wrote:<br>
> On Mon, May 11, 2020 at 05:00:45PM +0800, <a href="mailto:koba.ko@canonical.com" target="_blank">koba.ko@canonical.com</a> wrote:<br>
> > From: Koba Ko <<a href="mailto:koba.ko@canonical.com" target="_blank">koba.ko@canonical.com</a>><br>
> > <br>
> > BugLink: <a href="https://bugs.launchpad.net/bugs/1876844" rel="noreferrer" target="_blank">https://bugs.launchpad.net/bugs/1876844</a><br>
> > <br>
> > [Impact]<br>
> > Process of s3 resume take 7s and it's too long which is over 5s.<br>
> > <br>
> > [Fix]<br>
> > Reference link_active_reporting field of PCI device to determine whether delay is taken or not.<br>
> > Actual delay would be determined by taking maximum between 100ms and pci device's d3cold delay.<br>
> > <br>
> > [test]<br>
> > With the patch, the total time s3 resume is 3874.298 ms which is smaller than 5s.<br>
> > <br>
> > [Regression Potential]<br>
> > Medium. This patch only changes the rules to determine whether delay is taken or not.<br>
> > Reference the right field(link_active_reporting), but not speed of pci device.<br>
> > <br>
> > v2: correct BugLink typo only<br>
> <br>
> Does this problem also exist in the focal 5.4 kernel? Users of OEM<br>
> kernels in bionic will be moved to the focal GA kernel on upgrade to<br>
> 20.04, so if the problem exists there and we fix it in the osp1 kernel<br>
> users would see a regrssion. I would expect that any sauce patches<br>
> submitted for bionic OEM kernels at this point should either also be<br>
> submitted for focal, or else come with an explanation of why the patch<br>
> is not required for focal.<br>
<br>
And now I see patches for the 5.6 OEM kernel, which tells me the issue<br>
almost certainly does exist in the focal 5.4 kernel.<br></blockquote><div>Hi Seth,</div><div>As per KaiHeng's comment, the patch is not low risk of regression.</div><div>Also this is a performance patch not a functional patch.</div><div>After discussion with AceLan and KaiHeng, I sent a patch only to OEM.</div><div>because we could ask our QA team to verify it first.</div><br class="gmail-Apple-interchange-newline"><div> Koba Ko</div></div></div>