> HI Alan et al,
>
> I believe that I have solved the problem (at least for USB 1.x) while
> using the 2.6.12.3 kernel. In summary, I changed two things at once
> (yeah, I know I shouldn't do that):
> 1) I added a pci-usb controller with a different chipset
> 2) I compiled/installed the ehci-hcd kernel module
Compiling and installing the ehci-hcd module shouldn't make any
difference. But if you rebuilt the entire kernel instead of just that one
module, then maybe something else could have changed.
> My USB hardware on the dual Opteron server:
>
> # lspci -v | grep -i usb
>
> #add on pci-usb controller
> 0000:00:08.0 USB Controller: OPTi Inc. 82C861 (rev 10) (prog-if 10 [OHCI])
It's a little unusual to come across a PCI USB card that includes an OHCI
controller without an EHCI controller.
> # these are from the motherboard chipset
> 0000:00:10.0 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB
> 1.1 Controller (rev 81) (prog-if 00 [UHCI])
> 0000:00:10.1 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB
> 1.1 Controller (rev 81) (prog-if 00 [UHCI])
> 0000:00:10.2 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB
> 1.1 Controller (rev 81) (prog-if 00 [UHCI])
> 0000:00:10.4 USB Controller: VIA Technologies, Inc. USB 2.0 (rev 86)
> (prog-if 20 [EHCI])
>
> Previously, I had neither compiled the ehci-hcd.ko module nor had I
> enabled USB 2.0 on my motherboard (the motherboard allows one to select
> USB 1.0 or USB 2.0). Since the iAudio G3 supposedly supports USB 2.0, I
> thought I would try enabling USB 2.0. I compiled and loaded ehci-hcd
> and now I can mount and access the iAudio G3 reliably on the UHCI and
> OCHI ports.
>
> Does it make sense that adding the ehci-hcd module should make this
> suddenly work even in USB 1.x mode?
No, if by "USB 1.x mode" you mean that the USB 2.0 option in the BIOS
remains turned off. But if that option is turned on and ehci-hcd is
loaded then there's no way to force the motherboard ports to run at full
(USB 1.1) speed -- other than plugging in a full-speed device. If
ehci-hcd isn't loaded, then the only differences from your earlier
environment are caused by:
Whatever changes result from loading ehci-hcd at boot and then
unloading it later (probably nothing important);
Whatever other changes might have occurred as a result of
rebuilding the entire kernel (only if you did rebuild the
entire kernel, and even then there shouldn't be any such
changes, although you never know);
Whatever changes you might have made to the BIOS.
The BIOS could easily have been the entire reason for your problems all
along. I wouldn't be surprised if you found that the earlier OS version,
without ehci-hcd compiled, also started working correctly once that
setting was changed. And I wouldn't be surprised if you found that
putting the BIOS setting back caused the problems on the motherboard ports
to reappear, even with ehci-hcd loaded.
> I'm pleasantly surprised that the speed of access is *much* faster than
> on the laptop (on the order of 6x faster on writes). My CPU goes to
> 100% while accessing the G3, so it makes sense to me that a much faster
> processor (1.8GHz Opteron vs. 1.0 GHz Athlon) would help. Would you agree?
No. At high (USB 2.0) speed, I/O transfers generally limited by the
device, not by the CPU or the USB connection. Even though your CPU usage
shows up as 100%, it's not at all clear that the CPU is doing anything
useful at those times.
> Now, I can't reliably access the G3 (or my Epson scanner) when it is
> plugged into the EHCI port, but that is the subject of a future thread.
What do you mean by "the EHCI port"? Remember, on a high-speed system
each port is generally connected to two types of controller: the EHCI
controller and a companion full/low-speed UHCI or OHCI controller. That's
how your motherboard is set up.
Alan Stern
-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
linux-usb-devel@lists.sourceforge.net
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
