Re: [Question]About port ownership

On Tue, 18 Sep 2007, [ks_c_5601-1987] ±è¹ÎÇü wrote:

> Hi everyone,
> 
> I am a beginner in USB driver.
> I have some questions.
> I don't know whether it is proper or not, to ask you in this mailing list.
> But I need help
> 
> My test environment :
> Kernel Version - 2.6.12
> USB controller - Broadcom (EHCI/OHCI)
> Cable Length - about 3.x m
> (Note. The following phenomenon doesn't occur with NEC HC and Kernel 2.6.12)
> 
> Sequence of reproduction & Questions:
> 1. Connect USB device (Transcend 1GB thumb-drive).
> 2. Then, USB EHCI driver detects new device. The device is registered as
> high speed device and SCSI driver probe the device.
> 3. Remove the USB device.
>    EHCI port status register doesn't report the disconnection of the device.
>    Q. Is it Broadcom chip's problem? 

Yes.

>        Or could cable's electric characteristics affect the detection of
> disconnection?

I doubt it.  But you can always try using a different cable.

>    Port reset resulted from USB_STOR_TRANSPORT_ERROR in
> usb_stor_invoke_transport causes EHCI driver to change a port ownership.
>    Q. According to the EHCI specification, port ownership is returned
> immediately to the EHCI controller when device is disconnected during a
> companion HC has ownership. In this situation, device was already
> disconnected. Is changing ownership to companion HC a correct operation
> although device was already disconnected? If it is, then who is responsible
> for returning port ownership to the EHCI controller?

It is not correct.  The EHCI controller should keep port ownership and 
mark the port as disconnected.

> 4. Wait for about 20 ~ 30 seconds
> 5. Connect the USB device again.
> 6. USB OHCI driver detects new device. New device is registered as full
> speed device. 
>   Q. When connecting new device with enabling all debug prints, port
> ownership is return to the EHCI controller and new device is detected by
> EHCI driver. Who does change port ownership? 

The EHCI controller is in charge of the port ownership switch.

> But when debug prints are disabled, companion HC still has port ownership
> and OHCI driver detects new device. How does debug prints affect changing
> port ownership?

They do not affect it directly.  But they can affect the timing of 
various operations, and the timing might affect the controller's 
operation.

> Additional question.
> Q. When USB driver set PORT_RESET bit, it also clear PORT_ENABLE bit.
>    Who does enable the port after reset? Does HC enable the port
> automatically? Or does driver enable the port?

The controller enables the port automatically.

Alan Stern


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
[email protected]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel