On Thu, May 08, 2014 at 10:07:25AM -0400, Alan Stern wrote: > On Thu, 8 May 2014, Hans de Goede wrote: > > > Hi, > > > > On 05/08/2014 12:00 AM, Maxime Ripard wrote: > > > On Wed, May 07, 2014 at 10:25:55AM -0400, Alan Stern wrote: > > >> On Tue, 6 May 2014, Maxime Ripard wrote: > > >> > > >>> From: Boris BREZILLON <boris.brezillon@xxxxxxxxxxxxxxxxxx> > > >>> > > >>> On the Allwinner's A31 SoC the reset line connected to the EHCI IP has to > > >>> be deasserted for the EHCI block to be usable. > > >>> > > >>> Add support for an optional reset controller that will be deasserted on > > >>> power off and asserted on power on. > > >>> > > >>> Signed-off-by: Boris BREZILLON <boris.brezillon@xxxxxxxxxxxxxxxxxx> > > >>> Signed-off-by: Maxime Ripard <maxime.ripard@xxxxxxxxxxxxxxxxxx> > > >> > > >> Is this really a _reset_ line? That is, when you assert the reset > > >> line, does it actually reset the EHCI controller, or does it merely > > >> leave the controller in a partially powered-down state? > > > > > > It actually resets the whole controller. > > > > > >> The difference is important. During suspend, the controller is > > >> supposed to remember the state of the port connections as well as other > > >> settings. If it doesn't, the controller and all attached USB devices > > >> will have to be reinitialized every time the controller resumes, which > > >> will increase the latency. > > > > > > So you're saying that we should move this to the probe then? > > > > Yes. > > That's right. The controller should not be reset during suspend, if > you can possibly avoid it. > > There isn't any real benefit to asserting the reset signal during > suspend, is there? I mean, it won't use any less power, right? I don't think it will. I'll update the patches and send a new version. Thanks! Maxime -- Maxime Ripard, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com
Attachment:
signature.asc
Description: Digital signature
