What's happening is the error-handling thread is firing, which is
apparently sending everything to hell. This used to work, but lord knows
what's changed in the SCSI layers without my looking.
And yes, if you poll with TUR, you'll lose the unit attention that gives
the disk-change notification. That should happen with all removable media
devices, not just usb-storage.
Matt
On Wed, Jun 13, 2001 at 05:41:35PM +0200, Sancho Dauskardt wrote:
>
> >
> >At least 2.4.5-ac6 still generates this kernel hang, and can be made
> >to do it for full speed devices too.
> >
> >How to reproduce with usb-storage and a USB 1.1 device and any
> >host controller (well, at least all combinations I've tried):
> >
> > - add to usb-storage/transport.c, about line 1145,
> > a call to wait_ms(10000) ... after reading CSW
> > [ smaller values might work too ]
> >
> > - rebuild, modprobe, etc
> >
> > - connect a bulk-only usb-storage device
> >
> > - watch the blinking lights stop ... :)
> >
> >Anyone have a clue why that gives a hard lockup? Seems
> >likely that'll the same root cause for the high speed version.
>
> Since i haven't had the time to verify this properly - just a guess:
>
> what happens if usb-storage takes longer for a SCSI command, that the
> specified SCSI timeout ???
>
> My scenario: i've got a little app that permanently does TEST_UNIT_READY
> calls to a usb-storage device.
> This device needs around 16 sec's if report OK for some cards (oh yes, 64MB
> SSFDC). Now if i send the TEST_UNIT_READY command via sg interface an
> specify a timout of, say, 5 sec - bang.
>
> Matt ?
>
> - sda
>
> p.s. / o.t. if i poll TEST_UNIT_READY i' loose proper medium-change
> recognition within the VFS.. Is this normal ?
>
>
> _______________________________________________
> [EMAIL PROTECTED]
> To unsubscribe, use the last form field at:
> http://lists.sourceforge.net/lists/listinfo/linux-usb-devel
--
Matthew Dharm Home: [EMAIL PROTECTED]
Maintainer, Linux USB Mass Storage Driver
G: Let me guess, you started on the 'net with AOL, right?
C: WOW! d00d! U r leet!
-- Greg and Customer
User Friendly, 2/12/1999
PGP signature
