Firefox/4.0b8pre SeaMonkey/2.1b2pre   
   hCOk0C06T8el_hb^CT1A^BL@[ndJ77LFlBibjhTn7`6bEDVdbX7`]3JPdRa   
   From: Lars Erdmann    
      
   Rich Walsh wrote:   
   > On Mon, 24 Jan 2011 22:53:01 UTC, "Doug Bissett" wrote:   
   >> On Tue, 7 Dec 2010 07:11:40 UTC, Lars Erdmann wrote:   
   >>   
   >>> Are you saying that on your system you have NO UHCI even though you   
   >>> have EHCI ? In other words: EHCI "builds on top of" OHCI and not on UHCI ?   
   >>   
   >> FWIW, my new Lenovo ThinkPad T510 has only two EHCI controllers. there   
   >> are no (ZERO) UHCI, or OHCI controllers. I can use USB 1.0, and 1.1   
   >> devices, with no problem, and it doesn't matter if I use the old IBM   
   >> EHCI driver, or your latest one. It still works.   
   >   
   > When I read this I just had to try it myself.  If only it worked....   
   >   
   > With all my USBUHCD drivers rem'd-out and just USBEHCD active, my   
   > USB keyboard and mouse were dead.  However, USBEHCD did disable   
   > legacy mode when I had it enabled - and it still took 70+ seconds.   
   > It's further proof that USBEHCD causes the delay that I experience.   
   >   
   > A question, Lars...  Does the EHCI controller actually participate   
   > in the support of legacy mode or just in disabling it?   
      
   I am not a USB expert but looking at the specs, I'd say that during legacy   
   operation the EHCI controller is also in the loop and the corresponding SMM   
   code   
   active (if the MOBO manufacturer does support legacy mode + it's activated in   
   BIOS).   
   This becomes apparent when as in your case you have a USB 2.0 mouse.   
    From a bandwidth point of view, mice and keyboards would be happy with   
   USB 1.1 but it looks like USB 1.1 becomes more and more deprecated   
   (not surprising) so we obviously have to deal with it.   
   EHCI has a bunch of nifty settings to trigger an SMI (and therefore execute   
   SMM code) when this and that event happens. Basically all the events that   
   would lead to an ordinary interrupt (and therefore to be handled by the   
   OS driver) when legacy mode is disabled plus a few more (2 additional events   
   that will allow the SMM code to be called when someone else mucks around with   
   vital PCI config space settings).   
   The legacy settings are all manipulated via 2 registers in PCI config space.   
      
      
   This new "I am a EHCI controller but I can also handle USB 1.1" thing is   
   interesting.   
   Let's hope for the best it's completely dealt with in HW ...   
      
      
   Lars   
      
   --- Internet Rex 2.31   
    * Origin: Arcor (1:261/20.999)