David Brownell wrote:

> > When I first connect the ARCHOS Jukebox Recorder 20 on USB and then
> > afterwards start the usb support
> > (either manually or by rchotplug start), everything works fine: I can
> > mount the drive, read and write it properly.
> >
> > When I start the usb support first (either manually or by "rchotplug
> > start"), it does not (!) work.
>
> Do the startup problems happen when you use USB 1.1 speeds,
> only using an OHCI or UHCI host controller driver (not ehci-hcd)?
>

yes,  same problem. I used the usb-ohci before (its a NEC chip on the
controller).
result: khubd and scsi_eh_1 go to DW state and never leave this until
reboot

when it works (connect the drive before starting the driver) I can notice
the speed difference.


>
> Looks to me like the device takes enough time to initialize
> after enumeration that it confuses the USB storage code.
> Plus, I notice that the Archos page
>
>     http://www.archos.com/us/products/product_500277.html
>
> doesn't say the device has been USB 2.0 qualified, and does
> not use the USB 2.0 logo.  It's not listed on the www.usb.org
> page of products that are compliant.  That's suggestive ... it'd
> be worth contacting Archos to ask what gives.  (USB 2.0 has
> a compliance program for good reasons ... :)
>

The logo is on the drive...?

>
> > I have got the impression that in these cases the first read to the
> > manufacturer infos is failing and somehow the processes become stuck.
>
> Sounds somewhat right to me, I've seen that failure mode.
> So far nothing seems to suggest it's trouble in EHCI.
>

Just want to mention: even if the drive itself misbehaves, the driver
should not become stuck forever in "D " or DW" state.
In my situation, I have to re-boot to restart usb. Killing the processes
doesn't work.

>
> > Any ideas what I should try to overcome this blocked system?
> > try a newer patch?
>
> You might try this on a more recent kernel (2.4.19-pre5?), and try
> enabling mass storage debug to get more info.  Matt recently posted
> some rework of that exception handling code, and it's also possible
> this device needs some storage quirk handling before it behaves.
>
> - Dave

Thanks, I am just about to try 2.4.19-pre5 (downloading this currently).
Do I need to apply additional patches to this?
I will report on 2.4.19-pre5.

Thanks for helping
        Christian

--
Christian Mahr (privat)                  [EMAIL PROTECTED]




_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to