Problem with 6.2-RC1 da0/umass and Mitsumi FDD

[David E. Cross]

I have a mitsumi USB FDD as follows under 6.2-RC1 (the same problem
existed under 6.1-p10)

umass0: MITSUMI MITSUMI USB FDD, rev 1.10/1.00, addr 2
umass0: UFI over CBI with CCI (using CBI); quirks = 0x0000
umass0:6:0:-1: Attached to scbus6
da0 at umass-sim0 bus 0 target 0 lun 0
da0: <MITSUMI USB FDD     070M 3.01> Removable Direct Access SCSI-0
device 
da0: 1.000MB/s transfers


(as you can see I have already enabled USB_DEBUG)

Reading from the device works, although it is SLOW.  4K/sec slow.
Writing to the device does NOT work, it just hangs indefinitely (but the
machine stays responsive, until I unplug the device at which point it
panics.. but that's for another email)

Kernel messages from writing are:
umass0: CBI reset failed, TIMEOUT
umass0: CBI bulk-in stall clear failed, TIMEOUT
umass0: CBI bulk-out stall clear failed, TIMEOUT
umass0: CBI reset failed, TIMEOUT
umass0: CBI bulk-in stall clear failed, TIMEOUT
umass0: CBI bulk-out stall clear failed, TIMEOUT
umass0: CBI reset failed, TIMEOUT
umass0: CBI bulk-in stall clear failed, TIMEOUT
umass0: CBI bulk-out stall clear failed, TIMEOUT
umass0: CBI reset failed, TIMEOUT
umass0: CBI bulk-in stall clear failed, TIMEOUT
umass0: CBI bulk-out stall clear failed, TIMEOUT
umass0: CBI reset failed, TIMEOUT


What I have tried (to date) to resolve this issue:
  1) USB quirk to force the protocol to USB_CBI (instead of USB_CBI_I)
  2) SCSI DA quirk to set NO_SYNC (a previous kernel message which I
don't have a copy of made mention that there was an unsupported UFI call
of 0x35  (verbage of that is wrong, but "unsupported" "UFI" and "0x35"
is right) followed by a sync failure.  Also in the mailing list I
noticed some other similar problems with other devices were cured by
that.

Anyway, I am looking for some guidance before I beat against this some
more.

-- 
David E. Cross