tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: uhci on non-x86



On Tue, 13 May 2008, Michael Lorenz wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

On May 13, 2008, at 21:02, Brian Buhrow wrote:

        Hello.  It certainly is true that on my 4.0-stable systems, and on my
3.x based systems, files written to umass devices over the ehci(4) driver
produces garbage.  Ohci and uhci connected devices work fine.
(This is on I386 based systems.)  My point here is that I think the ehci(4)
driver is broken everywhere, including on I386 systems.

Who made your ohci/ehci chip? I had one made by ALi which never worked reliably while others, made by NEC, work just fine.

  FWIW, I can at least confirm that..

ehci0 at pci0 dev 11 function 1: NVIDIA nForce4 USB2 Host Controller (rev. 0xa4)

on NetBSD 4.0 (GENERIC.MP) is utterly, utterly broken. The motherboard in that box is an ASUS A8N32-SLI Deluxe. Connecting any kind of umass device invariably leads to complete system deadlock -- sometimes on the first attempt at IO, sometimes after seeming to work for a few seconds. I've seen everything from the system slowly dying as unrelated processes get stuck in a deadlocked io subsystem one after the other -- to immediate, complete hangs (i.e. including the ps/2 keyboard). The wireless mouse and graphire tablet sitting on ohci work flawless.

The devices (a usb2 card reader and usb memory sticks) all work fine under other OSen, and in other computers, so it's not a hardware/media thing.

/ali


Home | Main Index | Thread Index | Old Index