On Mon, Aug 05, 2019 at 08:58:33PM +0800, Kai-Heng Feng wrote: > Hi Greg, > > at 17:22, Kai-Heng Feng <kai.heng.feng@xxxxxxxxxxxxx> wrote: > > > at 22:17, Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote: > > > > > > I agree with Kai-Heng, this seems like a fairly light-weight solution > > > to a reasonable problem. > > > > Thanks for your review. > > > > > As to the issue of how much it will slow down system shutdowns, I have > > > no idea. Probably not very much, unless somebody has an unusually > > > large number of USB devices plugged in, but only testing can give a > > > real answer. > > > > In addition to that, only USB2 devices that enable LPM will slow down > > shutdown process. > > Right now only internally connected USB2 devices enable LPM, so the > > numbers are even lower. > > > > > I suppose we could add an HCD flag for host controllers which require > > > this workaround. Either way, it's probably not a very big deal. > > > > IMO this is not necessary. Only xHCI that reports hw_lpm_support will be > > affected. At least for PC, this only became true after Whiskey Lake. > > > > Kai-Heng > > > > > Alan Stern > > This patch is included in Ubuntu’s kernel for a while now, and there’s no > regression report so far. > Please consider merge this patch. I do not see a patch here at all, sorry. Please resend it. greg k-h
