[Date Prev][Date Next] [Chronological] [Thread] [Top]

RE: Multiple uploads don't verify checksums



Its just a counters thing.  Reset election will clear any corrupt counters
that were uploaded.

Ken

> -----Original Message-----
> From: owner-bugtrack@gesn.com [mailto:owner-bugtrack@gesn.com]On Behalf
> Of Steve Knecht
> Sent: Friday, February 25, 2000 2:16 PM
> To: bugtrack@gesn.com; support@gesn.com
> Subject: Re: Multiple uploads don't verify checksums
>
>
> Guy, Marin bypassed the data checksums on this without my knowledge
> repeatedly into the database that we are using for the election. (The one
> with Voter Reg totals input).  Will RESET ELECTION take away any corrupted
> counters and therefore wipe out any of these situations described below??
> Or do I need to backtrack to a db prior to the uploads and
> reinput Voter Reg
> totals (not a good alternative).
> -----Original Message-----
> From: Guy Lancaster <glanca@gesn.com>
> To: Bugs <bugtrack@gesn.com>
> Date: Thursday, February 24, 2000 10:27 AM
> Subject: Multiple uploads don't verify checksums
>
>
> >  When uploading multiple memory cards during the same
> >connection, the Accu-Vote does not verify on successive
> >memory cards that the checksums match the data.  This is
> >both a blessing and a curse.  Steve Knecht discovered this
> >in dealing with memory card corruptions and was able to
> >upload memory cards that were coming up as "PLEASE
> >RE-INSERT...".  However the checksums SHOULD be verified
> >first and therefore this is a bug.
> >
> >  In fact, all that is checked is that the memory card is
> >not write-protected and that its status is not blank and not
> >audited (i.e. the election finished).  Thus you could upload
> >corrupted counters and if the precinct data isn't reading
> >properly (connection problems???) then the AV could crash or
> >upload meaningless data.  And if there is a connection
> >problem, the AV could write the upload audit log entry to
> >the wrong location and corrupt a previously okay memory
> >card.  In short, not checking means taking unknown risks
> >with the data.
> >
> >  This is my fault (I wrote the code) and I'll "fix" it for
> >the next release (1.94y?) but until then be warned, if you
> >have a memory card that is failing the checks, bypassing the
> >checks using this method is dangerous.
> >
> >  And of course we'd all prefer that these memory card
> >corruptions just went away.  Those and computer crashes.
> >And crime.  And lousy weather.  And...  :-)
> >
> >              Guy
> >
> >