solved (Re: USB ether fails: watchdog timeout)

[Hiroharu Tamaru]

Upgrade to FreeBSD-4.8-RELEASE or cvsup to -stable.
It's already merged to -stable.

http://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/current-stable.html#STABLE

Otherwise, you would need to figure out how to make use of the cvsweb
for patches.

http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/dev/usb/ohci.c

Hiroharu Tamaru

At Wed, 30 Apr 2003 13:49:15 +0100 (WETDST),
VERFAILLIE Eric wrote:

> On Wed, 30 Apr 2003, VERFAILLIE Eric wrote:
> 
> > hello
> > I'new in bsd and i'v got some problems with my ethernet card rtk0, it's
> > has the same irq (11) that ohci0
> > I tried to recompile GENERIC but even if comment the lines with ohci0
> > ,I always have got the same message :'watchdog timeout'
> > 
> > in the message of Hiroharu Tamaru i,ve seen that they is a file
> > dev/usb/ohci.c
> > in order to fixe the problem, where can i get it
> > 
> >    from [Hiroharu Tamaru]
> > 
> > For the sake of record:
> > 
> >  With some discussions in bsd-usb ML, the issue is fixed in rev 1.118
> >  of dev/usb/ohci.c committed by shiba at .  It is waiting to be MFC'ed.
> > 
> >  At Tue, 25 Feb 2003 03:36:43 +0900,
> >  Hiroharu Tamaru wrote:
> >  > 
> >  > Hi list,
> >  > 
> >  > I have an "I/O DATA USB ETTXS" USB ether NIC (aue, Pegasus
> >  > II) which fails to work with Toshiba Libretto M3 (notebook)
> >  > on FreeBSD 4.2--5.0 (at least). 
> >  > 
> >  > From the reasons described below, I am suspecting if it were
> >  > to do with the OHCI controllor on Libretto M3, which is NEC
> >  > uPD 9210.
> >  > 
> >  > ohci0: <NEC uPD 9210 USB controller> mem 0xffaff000-0xffafffff irq 11
> > at 
> >  > device 11.0 on pci0
> >  > usb0: OHCI version 1.0
> >  > usb0: <NEC uPD 9210 USB controller> on ohci0
> >  > usb0: USB revision 1.0
> >  > uhub0: NEC OHCI root hub, class 9/0, rev 1.00/1.00, addr 1
> >  > uhub0: 2 ports with 2 removable, self powered
> >  > 
> >  > The only part that's not working seems to be the transmit
> >  > part.  aue driver attaches and ifconfig succeeds, and tcpdump
> >  > even captures the packets on the wire.  But once instructed
> >  > to send a packet (like ping or dhclient), 
> >  > 
> >  > aue0: watchdog timeout
> >  > 
> >  > occurs (but receive side continues normally) and nothing is
> >  > sent.  I added printfs to see the usb status code that's
> >  > acquired inside aue_watchdog after the call to
> >  > usbd_get_xfer_status, it turned out to be USB_NOT_STARTED.
> >  > 'ifconfig aue0'ing after this timeout shows OACTIVE flag on,
> >  > and it never drops until after 'ifconfig aue0 down' is
> >  > issued.
> >  > 
> >  > And that was about everything I could try in the absence of
> >  > any knowledge for USB programming.
> >  > Could someone suggest snything to try with this?
> >  > 
> >  > TIA.
> >  > 
> >  > 
> >  > PS:
> >  > This NIC works fine on a 5.0-RELEASE desktop box with a UHCI
> >  > host controller: 
> >  > 
> >  > uhci0: <Intel 82801BA/BAM (ICH2) USB controller USB-A>
> >  >  port 0xd000-0xd01f irq 11 at device 31.2 on pci0
> >  > 
> >  > and also on a 4.7-STABLE desktop box with:
> >  > 
> >  > uhci0: <Intel 82801AB (ICH0) USB controller>
> >  >  port 0xd000-0xd01f irq 10 at device 31.2 on pci0
> >  > 
> >  > USB mice (ums) and USB floppy (umass,da0) work fine with
> >  > this Libretto, so the usb subsystem itself is not totally
> >  > broken.
> >  > 
> >  > One other thing I might mention is that this Libretto
> >  > freezes when I try to use pccard and usb together, unless I
> >  > route the pccard (ToPic97) interrups via isa bus by:
> >  > hw.pcic.intr_path="1"
> >  > hw.pcic.irq="0"
> >  > 
> >  > May be it is so because both pcic and ohci grabs irq 11 if
> >  > the pcic is pci routed where as in polling and isa routing
> >  > mode, ed0 that's plugged into the PCCard slot grabs irq3 and
> >  > thus there's no conflict.  BIOS tells me that pci devices
> >  > all grab irq11 and is not configurable to anything else.
> >  > Most of the internal devices were disabled inside the BIOS
> >  > to make things simple during the test.
> >  > 
> >  > FWIW, booting 5.0-RELEASE on this machine fails to find ata
> >  > ad0 disk at the very last step just before the login prompt,
> >  > unless acpi is disabled.
> >  > -- 
> >  > Hiroharu Tamaru.