Benjamin Herrenschmidt wrote:
Booting test4 with a keyspan PDA gives:


Aug 24 20:49:57 gaston kernel: ohci-hcd 0001:01:19.0: OHCI Host Controller

Here's where I want the pci device descriptions bak ... this is an Apple OHCI, yes?


Aug 24 20:49:57 gaston kernel: ohci-hcd 0001:01:19.0: irq 28, pci mem f53b0000
Aug 24 20:49:57 gaston kernel: ohci-hcd 0001:01:19.0: new USB bus registered, assigned 
bus number 2
Aug 24 20:49:57 gaston kernel: bus usb: add device usb2
Aug 24 20:49:57 gaston kernel: bound device 'usb2' to driver 'usb'
Aug 24 20:49:57 gaston kernel: PM: Adding info for usb:usb2
Aug 24 20:49:57 gaston kernel: bus usb: add device 2-0:0
Aug 24 20:49:57 gaston kernel: hub 2-0:0: USB hub found
Aug 24 20:49:57 gaston kernel: hub 2-0:0: 2 ports detected
Aug 24 20:49:57 gaston kernel: bound device '2-0:0' to driver 'hub'
Aug 24 20:49:57 gaston kernel: PM: Adding info for usb:2-0:0
Aug 24 20:49:57 gaston kernel: bound device '0001:01:19.0' to driver 'ohci-hcd'
Aug 24 20:49:57 gaston kernel: hub 1-0:0: debounce: port 1: delay 100ms stable 4 
status 0x101
Aug 24 20:49:57 gaston kernel: hub 1-0:0: new USB device on port 1, assigned address 2
Aug 24 20:49:57 gaston kernel: usb 1-1: control timeout on ep0out

Was it actually getting the IRQs? That last patch of yours had a problem where it never passed IRQs along...



Aug 24 20:49:57 gaston kernel: drivers/usb/core/hcd.c:506: spin_lock(drivers/usb/core/hcd.c:c03cc258) already locked by drivers/usb/core/hcd.c/1182

Then, later on, trying to remove the driver typically, I get a badness in ohci-hcd:344
and the machine typically hangs on reboot.

... badness there would happen if the SF interrupt never arrived, and the controller was still allegedly running. Maybe not that particular badness, but trouble nonetheless.


I suspect it would be a Good Thing if the HCD glue tracked whether or not IRQs had been received since the device initialized, and then used that information to help diagnose what's likely the most common problem folk have getting USB going: IRQs aren't usable.

Seems to me that certain cleanup paths could just emit a KERN_ERR message
if they're doing things like unlinking an urb and no IRQ ever arrived.
Letting folk know for sure they've got IRQ problems is the first step to
getting them fixed.  (Usually by turning off ACPI or similar.)  We've
seen the problem with all of the HCDs.

- Dave




Ben.








-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to