Re: Polling Place conditions, reframing the issue

From: Karl Auerbach <karl_at_cavebear_dot_com>
Date: Fri May 14 2004 - 02:27:39 CDT

On Thu, 13 May 2004, Ed Kennedy wrote:

> ...Thumb drives are non-volatile aren't they?

I'm a big fan of this kind of USB based flash memory. It is very robust
and, if two of 'em are used, then they can be written in an A-B-A-B...
manner so that there is never a time when both are in the process of being
written at the same time. (The Linux kernel is not necessarily our friend
in that regard - I have watched many megabytes of data stored in memory
and only flushed to the USB flash filesystem every now and then - so
explicit cache flushing would be necessary in our code.)

The Linux USB system is still somewhat shakey - it doesn't seem solid yet
with respect to hot removals/re-insertions of USB devices. I'm not sure
whether PCMCIA based flash might be more stable with the current crop of
kernels. (I'm using 2.6.6 on some of my boxes and it's harder to get newer
than that.) But by the time we get something deployed I'm pretty sure
that Linux USB stability will be improved.

Most flash memory is formatted VFAT. It seems to work well. I've tried
ext3 on a flash stick/thumb drive (as well as a three thumb-drive wireless
ISCSI mounted RAID 5 [yes, really!]) I don't know whether it would be
useful to use ext3 rather than VFAT - I'm not sure the journaling buys us
anything in this application.

In order to protect against thumb drive swapping (intentional or
accidental) it would be useful to pre-label the flash drives with unique
codes and to prepare each one pre-election with a file containing some
kind of digital signature of that unique that had better be there at the
end of the day (and which is validated at the end of the day.) We might
also wire up the device-removal handlers on Linux to set up a loud scream
if a thumb drive is removed.

Personally I find the thumb drive more attractive than the writable
multi-session CD-ROM except for the implicit validation that comes with
having everything from boot-code to final data on the same medium.


= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
Received on Mon May 31 23:17:42 2004

This archive was generated by hypermail 2.1.8 : Mon May 31 2004 - 23:18:16 CDT