Timeouts during initial Mode Sense commands

[Denver Hull]

Hans Petter Selasky wrote:
> On 2019-12-21 00:46, Denver Hull wrote:
>> How hard would it be to change things to use 0x1a instead of 0x5a 
>> temporarily?
>
> There is a tool called usbtest in /usr/src/tools/tools/usbtest which 
> can exercise the SCSI commands for mass storage devices.
>
> --HPS
>
Very nice, thanks.  My tools require an actual SCSI device node so they 
aren't much help when there isn't one.  The usbtest host mode tests seem 
to work ok with the device in question, but not the mass storage tests:
 >30
Attaching to: ugen2.12: <Adafruit Industries LLC CircuitPlayground 
Express> at usbus2 @ iface 2
Resetting device ...
Testing SCSI commands ...
ERROR: CBW reception: 4
ERROR: CBW reception: 4
ERROR: CBW reception: 4
Cannot read disk capacity (0 / 4)
ERROR: CBW reception: 4
Cannot read disk capacity (1 / 4)
ERROR: CBW reception: 4
Cannot read disk capacity (2 / 4)
ERROR: CBW reception: 4
Cannot read disk capacity (3 / 4)
[0.2.4] - Mass Storage Test Parameters:

However, I did dust off one of my old tools and tried using it to send 
both 6 byte and 10 byte mode sense commands from Linux.  The 6 byte 
commands always worked, but not the 10.  They resulted in lots of 
"connection reset by peer" and "broken pipe" status messages, along with 
port reset commands.  I saved wireshark/usbmon traces from both, and 
have attached them.  I think they're small enough to get through this time.

If the problem with these devices really is that they can't respond 
properly to 10 byte mode sense commands (and that's how it's beginning 
to look), then what?

Thanks,

Denver