[Bug 208668] 11.0-CURRENT since around December 2015 can be remarkably troublesome with USB (for PersonaCrypt)

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=208668

            Bug ID: 208668
           Summary: 11.0-CURRENT since around December 2015 can be
                    remarkably troublesome with USB (for PersonaCrypt)
           Product: Base System
           Version: 11.0-CURRENT
          Hardware: amd64
                OS: Any
            Status: New
          Severity: Affects Only Me
          Priority: ---
         Component: usb
          Assignee: freebsd-usb at FreeBSD.org
          Reporter: grahamperrin at gmail.com
                CC: freebsd-amd64 at FreeBSD.org
                CC: freebsd-amd64 at FreeBSD.org

Created attachment 169141
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=169141&action=edit
A diagnostic file, produced by 'PC-BSD bug report tool'; includes dmesg and
other information

USB 3.1 Gen 1 (USB 3.0) Kingston DataTraveler G4 
http://web.archive.org/web/20151107114009/http://www.kingston.com/en/usb/personal_business#DTIG4 
32GB USB flash drive for PersonaCrypt in PC-BSD 11.0-CURRENT. 

If required:
https://www.ixsystems.com/blog/pc-bsd-version-1012-now-available-for-the-security-conscious-desktop-user/
includes an outline of PersonaCrypt. 

With PC-BSD 11.0-CURRENTNOV2015 and less, the Kingston device is *entirely
reliable* for PersonaCrypt with a circa 2007 Ergo Vista 621 notebook, with 4 GB
memory, where the interface is probably limited to USB 2.0. 

With the following distributions, the same combination of hardware *ceases to
work very quickly* (typically within ten minutes of GELI attachment): 

* PC-BSD 11.0-CURRENTDEC2015
* PC-BSD 11.0-CURRENTJAN2016
* PC-BSD 11.0-CURRENTFEB2016
* PC-BSD 11.0-CURRENTMAR2016

For a few weeks the Kingston device was good for PersonaCrypt with PC-BSD
11.0-CURRENTMAR2016 on a more modern notebook -- an early 2011 Apple
MacBookPro8,2 with 4 GB memory, https://support.apple.com/kb/SP620 – again, USB
2.0. 

Today with PC-BSD 11.0-CURRENTAPRIL2016 before attempting to use the Kingston
device, I gave that Mac notebook an unusually 'hard push' (overnight, with
synth at ttyv1) – hard enough to use around 2 GB of 4 GB available swap. Then,
as I attempted to log in with PersonaCrypt, the problem recurred. 

An extract: 

----

usb_pc_common_mem_cb: Page offset was not preserved
(da0:umass-sim0:0:0:0): READ(10). CDB: 28 00 00 00 00 a8 00 00 80 00 
(da0:umass-sim0:0:0:0): CAM status: CCB request completed with an error
(da0:umass-sim0:0:0:0): Retrying command
usb_pc_common_mem_cb: Page offset was not preserved
(da0:umass-sim0:0:0:0): READ(10). CDB: 28 00 00 00 00 a8 00 00 80 00 
(da0:umass-sim0:0:0:0): CAM status: CCB request completed with an error
(da0:umass-sim0:0:0:0): Retrying command
usb_pc_common_mem_cb: Page offset was not preserved
(da0:umass-sim0:0:0:0): READ(10). CDB: 28 00 00 00 00 a8 00 00 80 00 
(da0:umass-sim0:0:0:0): CAM status: CCB request completed with an error
(da0:umass-sim0:0:0:0): Retrying command
usb_pc_common_mem_cb: Page offset was not preserved
(da0:umass-sim0:0:0:0): READ(10). CDB: 28 00 00 00 00 a8 00 00 80 00 
(da0:umass-sim0:0:0:0): CAM status: CCB request completed with an error
(da0:umass-sim0:0:0:0): Retrying command
usb_pc_common_mem_cb: Page offset was not preserved
(da0:umass-sim0:0:0:0): READ(10). CDB: 28 00 00 00 00 a8 00 00 80 00 
(da0:umass-sim0:0:0:0): CAM status: CCB request completed with an error
(da0:umass-sim0:0:0:0): Error 5, Retries exhausted

----

I have photographs of today's incident, but I can not attach them at this time. 

I understand that users of USB devices may be cautioned against using USB 2.0
for ZFS, but the difference between NOV2015 and DEC2015 was so extreme that a
bug report seems appropriate. 

----

Historically (2010)
-------------------

https://wiki.freebsd.org/USB noted:

    Q: I see the following error and USB stops working 
       "Page offset was not preserved!"
    A: Update your sources, this issue is fixed.

Recently (2015-11-24)
---------------------

svn commit: r291251 - stable/9/sys/dev/usb
https://lists.freebsd.org/pipermail/svn-src-stable-9/2015-November/008097.html

> MFC r290326:
> Relax the BUS_DMA_KEEP_PG_OFFSET requirement to 
> allow optimising allocation of DMA bounce buffers.

Modified:

    stable/9/sys/dev/usb/usb_busdma.c

A wild guess: I wonder whether the problem(s) affecting my use of a USB device
are an unintended consequence of that allowance for optimisation.

-- 
You are receiving this mail because:
You are on the CC list for the bug.