I will later tonight. For now, I'm using 3.18.x (for another reason), which doesn't have this commit.

My Linus-latest tree has a "fix" that imposed a hard limit of 32767, which worked for me then.

On January 5, 2015 12:07:56 PM PST, Alan Stern [off-list ref] wrote:

On Mon, 5 Jan 2015, Christoph Hellwig wrote:

quoted

On Tue, Dec 30, 2014 at 08:36:34AM -0800, Kenneth R. Crudup wrote:

quoted

OP here. FWIW, this is what I get when running that command on the

SCSI

quoted

quoted

generic device that corresponds to the USB-3 (non-UAS) disk[1] that

had the

quoted

quoted

issue:

So it looks like this one actually provides sane values, but we don't
we never even look at EVPD pages for usb devices due to the wrong

SCSI

quoted

level?

According to James Bottomley, it doesn't matter much what the EVPD
pages say.  The limitation is imposed by the USB _bridge_, whereas the
EVPD data indicates what the _drive_ is capable of.  So we can't rely
on that data anyway.  (Although if the EVPD data indicates a limit 
smaller than the bridge can handle, then we'd need to pay attention to 
it.)

The patch I posted sets a general limit of 32 MB for USB drives that 
don't have a quirk flag for a smaller limit.

Kenneth, have you tried that patch?  Does it fix your problem?

Alan Stern

-- 
Sent from my 2014 Galaxy Note 10.1