| University of Bielefeld - Faculty of technology | |
|---|---|
|
Networks and distributed Systems
Research group of Prof. Peter B. Ladkin, Ph.D. |
|
| Back to Abstracts of References and Incidents | Back to Root |
| This page was copied from: http://catless.ncl.ac.uk/Risks/11.82.html | |
ACM Committee on Computers and Public Policy, Peter G. Neumann, moderator
Risks of open anonymous ftp
Magellan spacecraft performance; followup
Lauda Air Boeing 767 Aircraft crash
RISKS of posting humor to the net
Digital Fingerprints in California
CPSR Review of FBI Net Surveillance
Computers and Academic Freedom Groups Now at EFF.ORG
I have discovered some risks of having an open anonymous FTP
environment on your machine. By 'open', I mean that there are one or more
writable directories available for people using anonymous FTP into which files
can be uploaded. This arrangement is popular on several of our professors'
workstations, as it allows them to exchange files and papers with colleagues at
other sites. It is particularly nice for dvi or binary files, which can't be
sent directly via e-mail. Occasionally, we have seen a file called
MAKE_MONEY_EASY or something similar advertising some get-rich-quick pyramid
scheme that someone uploads. This hasn't been a problem in the past, but it has
become one when escalated by an order of magnitude or two.
Some background first: the USENET newsgroup alt.sex.pictures is used to
distribute what are usually called X- or R-rated pictures. These pictures are
uuencoded files, usually in the GIF format, sometimes in others. Because of the
size of the pictures, a individual pictures is broken down into parts, usually
3 or 4, but sometimes up to 20. A fully assembled and decoded picture is
sometimes as small as 25 or 30K, but is typically 150-250K and may be even
larger. Monthly posted reports of USENET traffic flow show that this newsgroup
is consistently among the leaders in terms of quantity of megabytes of traffic.
Many sites refuse to carry or forward this newsgroup. For some, it is
simply a question of traffic; they don't want to double or triple their phone
and modems costs to transmit the group. For others, it is a matter of the
material itself. More than one group has been forced to shut down the group
under orders from superiors, often under political pressure.
To make up for the lack of availability of the newsgroup, and also for
the limited bandwidth it provides even if available, seekers of the GIF files
look for ftp sites. A good site can have many pictures available, and the ftper
can grab many megabytes of files at once. Several of these have been found in
the past, but what typically happens is that someone will announce the location
of such a site, at which point the poor machine is swamped with anonymous ftp
sessions and traffic, forcing the administrators of the machine to turn off
anonymous ftp. As a consequence, the source of ftp sites is few.
So, another tack must be taken and it is this: find a site that has a
writable anonymous ftp directory, create gif directories, and ftp away.
Requests and questions are communicated by creating files whose name is the
request/question itself, like this:
-rw-r--r-- 1 ftp 1 May 23 18:57 more-asian-gifs-PLEASE
or
1 -rw-rw-rw- 1 ftp 1 May 23 14:21 00-Otherwise-this-site-wo
uld-be-shut-down-soon
1 -rw-rw-rw- 1 ftp 1 May 23 14:21 00+Please-do-you-file-tra
nsfer-at-non-prime-time.---7pm-6am.eastern-time
Announcement of new sites is handled in a similar fashion:
-rw-r--r-- 1 ftp 1 May 22 00:16 I-saw-somebody-suggesting-t
o-switch-to-xxxxx.yyy.zzz
People upload and download to their hearts conten<t. Everyone is
happy....except for perhaps the unknowing users of the machine and the network
it is on.
This has been happening on my machines the last few weeks. When we
first discovered it several weeks ago, I quickly hacked logging into the FTP
server and saw connections from Canada, Australia, England, Israel, France,
Italy, Switzerland, Denmark, Germany, Austria, Sweden, Norway, Finland and of
course from the US, all within a single 24 hour period. I had to shut it down
at the time, but I started watching my other hosts, and sure enough, it started
again. I began watching intently for a couple of days and was able to determine
who some of the people who were doing this were. I watched one person create
and 'seed' two new sites by uploading a small collection of files to others to
grab. Within hours, these sites had been advertised to others and were being
used.
This activity get clog a machine and a network. I have seen people
upload 10MB of pictures to my machines at once, and then have seen anywhere
from 5 to 8 people grabbing them at the same time. People write scripts to log
in every 15 minutes and check for new files automatically; sometimes these
continued to run even after I had shut down the site.
One way to protect yourself is to make a directory writable, but not
readable. This way, people can upload files, but fetching them is more
difficult, since you can't know what the names are. However, I think that that
would be of limited use in this case, since people can put a list of what
they've uploaded to a machine on another site. Then, others can use that list
to retrieve what they want from the repository. The scheme also curtails the
legitimate use of a writable area.
I see several risks. First, your site can become severely clogged,
often without your having any idea of what is going on. This makes it harder to
fix; we found out about it because we *were* having network problems and just
happened to notice a lof of network activity to a particular machine. A quick
'ps' soon pointed out the problem. However, I believe that many sites are less
sophisticated and will be less able to pick up on this.
Second, this activity affects legitimate use of the anonymous ftp
facility. Sharing and convenience suffer because of it.
Third, shutting it down and then confronting the abusers in a public
forum (newsgroup alt.sex.pictures.d) can attract attention to your site and may
spur people into more diligent efforts to 'pay you back' for spoiling their
fun. While I received some supportive responses, I also received a few telling
me what I could do with myself. Curiously enough, I also got a few requests for
the addresses of the other sites I had discovered being victimized.
Check your systems, you might be suffering from network parasitic
terrorism.
Pete
The Magellan craft doing radar mapping Venus had several failures early in the
mission, including one 32-hour outage, as noted in some previous Risks
postings.
This update on subsequent performance is extracted from a 2 page article in
appeared in Aviation Week of 20 May:
The Magellan spacecraft successfully completed the first cycle of Venus
mapping on May 15, producing more data than required despite occasional
spacecraft glitches...
A second cycle of mapping was started May 16 to cover most of the surface
that was not mapped on the first cycle, with emphasis on south polar regions
that have not been seen before.
Ambitious plans for future use of the spacecraft are being formulated,
including using aerobraking to circularize the orbit...
Magellan's basic mission was to radar map at least 70% of Venus' surface and
83.7% had been covered as of May 15. The Martin Marietta spacecraft has done
a ``great job'' and the Magellan radar, built by Huges has been ``flawless,
beautiful,'' said A. J. Spear, Magellan project manager at JPL.
Nearly all major newspapers in Austria had the main headline on Monday along
the following topic:
Computer failure causes Boeing 767 airplane crash
Niki Lauda, the owner of Lauda Air (and former grand prix champion) gave the
following explanation in a televised press conference in Vienna on Sunday at
1:00 p.m. (I watched it.):
Both the voice recorder and the data recorder have been recovered after last
Sunday's crash of the 767 near Bangkok. The data recorder was unreadable
because it did not withstand the crash. After an analysis of the voice
recorder the following sequence of events has been established:
Sunday, May 26, 23:05 Bangkok time: The plane, coming from Hongkong took off
from Bangkok airport. Everything was normal, the plane has climbed to 7000 m
and was still climbing further.
23:16 An advisory warning light (lowest of three criticality degrees) started
to blink intermittently. The copilot referred to the checklist which read
(about): Another failure can cause the deployment of the thrust reversal. Expect
normal operation of thrust reversal after landing. No immediate action to be
taken.
23:18 The voice of copilot Turner: "It deployed" (i.e. the thrust reversal was
activated while the plane was still climbing with high engine power). A few
moments later one can hear an acoustic warning signal on the tape indicating
that the forces on the plane are critical. Two seconds later the voice
recorder stops, because the plane crashed.
I have been asked by newspapers to comment on the suspected computer failure
but do not have any further information. Do you have access to any further
information about the B 767, in particular:
* Do they use single or multiversion software?
* How is it possible that a dangerous state, which has been indicated,
does not require any action? (Consider this happened in the first
fifteen minutes of a 10 hour flight!)
Looking forward to some more information
Hermann --Tel 43 1 588018180--FAX 43 1 569149 -- ++++++ PLEASE NOTE THE CHANGE
IN THE E-MAIL ADDRESS ++++++
[An anonymous commenter sent in this: "Speculation is now increasing that
the thrust reversal deployment, while the system was under computer control,
was part of the problem. I just spoke to my Airline Pilot Association friend
who does all the interviews after such events, and he tells me that there is
a mechanical system that is supposed to prevent the reversers from deploying
until the plane is on the ground, but he says that they do break down and
such a failure would allow a computer malfunction to deploy the reversers.
He also says that contrary to what Boeing is saying, it can be VERY hard to
avoid having a plane tear apart if the reversers deploy while the plane is in
a high power situation (e.g. climbing--as was this plane). Note that the
767-300 is NOT a "fly by wire" plane in the sense the new Airbuses are, but
that the engines are normally under computer control with no direct
mechanical connection with the pilots." There is some redundancy in the
following messages, and in the preceding ones, but I think that the subject
is potentially too complex for me to try to do an accurate and careful
differential reading and analysis, in the absence of more definitude. PGN]
No, thrust reversal in mid-flight wouldn't rip off the wing. It would have some
potentially unpleasant control consequences, by introducing a very large yaw
moment, but reverse thrust is far less efficient than forward thrust. Also, I
understand that as part of the flight certification of an airplane, it must be
able to fly with one engine at full reverse thrust. I don't know what might
happen if the thrust reverser activated at low altitude, where there isn't a
lot of time to recover, but the plane should hold together until it hits the
ground.
Extracted from this morning's _Washington_Post_ (3 June 91) p. A11:
"What happened in the plane is that the thrust reverser, for whatever
reason, was deployed in the air," Lauda told a news conference upon
returning to Vienna from Washington, where U.S. authorities are examining
equipment retrieved from the wreck. His conclusions were echoed in a
statement by Austrian Transport Minister Rudolf Streicher, who said a
computer error may have activated the thrust reverser.
[...]
However, a spokesman for the Seattle-based Boeing Co. said a 767
should be able to continue flying even if the situation described by
Lauda occurred. The Federal Aviation Administration will not certify
any jet aircraft to fly unless it passes an in-flight test in which
a thrust reverser is deployed at full power.
[...]
Some U.S. sources close to the investigation expressed irritation
at the flow of statements from both Thailand and Austria as the
investigation of the crash continues. In any crash, they noted, an
agency can determine a final cause only after months of painstaking
investigation.
The sources, who asked not to be identified, did not rule out the
possibility that the thrust reverser may have malfunctioned. However,
Boeing spokeswoman Elizabeth Reese said that owners of the more than
350 767's now flying had never reported any thrust reverser problems.
[...]
To receive FAA certification, each type of aircraft produced [in]
the United States must have a thrust reverser deployed at full power
in flight. The pilot must be able to control the plane using
normal procedures.
Joe Morris
Re: In-flight engine reversal as reported in _The Times_ (London), Monday
June 3rd 1991.
> [...] If the diagnosis were confirmed, the accident would
> be unprecedented, Herr Lauda said.
A crash of a 767 is unprecedented. In-flight thrust reversal leading to an
accident is not -- there is a long history of accidents from both intentional
and unintentional in-flight reversing, for both jet and propeller driven
aircraft.
>rmoonen@hvlpa.att.com (Ralph 'Hairy' Moonen) writes;
>And certainly, wouldn't a mid-air reversal of thrust just rip
>off the wing, leaving the plane to plummet down totally out of control?
No, it would not. The engine would depart the aircraft well before that
level of stress could be reached. Application of significant amounts of
reverse thrust would cause a severe controllability problem in and of itself
though. If one engine were providing full foward thrust and the other
significant reverse thrust, the result would be a very large yawing moment.
The crew would likely notice this very quickly. It is possible that the
reversers deployed as part of a engine failure that was already in progress.
More data will be required before this can be ascertained.
Steve Philipson
According to the Wall Street Journal, 6/3/91, In order to receive federal
(US) certification, the Boeing had to *demonstrate* that the 767 could
fly with one thrust reverser deployed (emphasis added). I take this to
mean that they had to actually fly the plane this way.
The same report also said that Boeing engineers knew of one other time
the thrust reverser deployed in mid-air, but that plane landed without
incident, and the situation may not have been entirely analogous.
Jeremy Grodberg jgro@lia.com
Uh, the line
for(;P("\n"),R-;P("|"))for(e=C;e-;P("_"+(*u++/8)%2))P("| "+(*u/4)%2);
does *not* compile cleanly on my system (Sparc). The problem is with the
operators "R-" and "e-". Change them to "R--" and "e--" and it will compile
successfully, but drop core when it runs. Then initialize u, C and R to
something reasonable (like "the address of an array of 100 ints initialized to
0', "10" and "10", respectively), and define P() to be printf(), and you'll get
output that looks something like this:
| _| _| _| _| _| _| _| _| _| _|
| _| _| _| _| _| _| _| _| _| _|
| _| _| _| _| _| _| _| _| _| _|
| _| _| _| _| _| _| _| _| _| _|
| _| _| _| _| _| _| _| _| _| _|
| _| _| _| _| _| _| _| _| _| _|
| _| _| _| _| _| _| _| _| _| _|
| _| _| _| _| _| _| _| _| _| _|
| _| _| _| _| _| _| _| _| _| _|
| _| _| _| _| _| _| _| _| _| _|
Make the initial values of the array random, and you get something that
looks like a maze:
_ _ _ | _| _| _| |
_| | | | _ _ _ _ |
| _| _| _| _| | |
| | _ _ _ _ | _| _| _| |
_| | | | _ _ _|
_ | _| _| _| _|
| | | | _ _ _ _ | _| _|
| _| _| | | | _|
_ _ _ | _| _| _| |
_| | | | _ _ _ _ |
Cute, yes. Useful? You tell me...
Phil
I recently applied for a California driver's license, and was surprised to
learn that the fingerprinting required for a license (right thumbprint) was now
done by a digital scanner instead of with paper and ink pad. The RISK is
obvious -- sometime down the road, when pattern matching of fingerprints has
been more or less totally automated, the State of California will have a
database ready to go without the hassle of scanning paper fingerprints in.
It's my understanding that current matching technology is too labor- and
computer-intensive to perform regularly on anything larger than a database of
known felons, but with advances in computer power, matching against the whole
population may be possible. Anybody know more about the California database,
or how viable thumbprint matching may be? Would one expect many false matches
using just a thumbprint? How many other states require fingerprints for
driver's licenses, and does any other use digital scanners?
I suppose it's possible that the California DMV doesn't retain the digital data
-- but I doubt it. I'm less certain but fairly sure that the "mugshot" is also
taken with a video system. I could imagine it would be awfully tempting for
law enforcement agencies to combine those two databases.
Mike Caplinger, MSSS/Caltech Mars Observer Camera mc@moc.jpl.nasa.gov
I 'd like to add a bit of relevant information to the discussion. Computer
Professionals for Social Responsibility (CPSR) is currently litigating a FOIA
lawsuit against the FBI seeking information on the Bureau's policies and
practices with regard to computer bulletin boards. A couple of months ago, we
received a heavily censored copy of a 1985 internal FBI legal opinion entitled
"Acquisition of Information from Electronic Bulletin Boards."
Although couched in terms of a prohibition, the opinion does *not* establish an
across-the-board prohibition on monitoring and/or surveillance of computer
bulletin boards. All that the opinion prohibits is a "comprehensive"
monitoring program. Bulletin boards may be monitored, so long as Fourth
Amendment standards are satisfied, i.e., where there is a "reasonable
expectation of privacy," a warrant must be obtained. The opinion might not be
using precise language when it refers to "bulletin boards," since most are
public and do not generally involve an expectation of privacy. As I read the
opinion, it permits warrantless monitoring of public bulletin boards on a
case-by-case basis.
CPSR believes that such monitoring, even of public bulletin boards, is
inappropriate. There is an undeniable "chill" placed on the free exchange of
opinions when participants need to worry if the discussion is being monitored
by government agents. The history of the FBI demonstrates that individuals
expressing views deemed to be unpopular or "subversive" became the subjects of
official scrutiny and extensive record-keeping. While some might argue that
bulletin board discussions are completely open and that participants should
expect them to be monitored, such a concession seriously erodes First Amendment
values. How would we feel if we knew that every political meeting or community
gathering we attended was monitored and recorded by government agents? Isn't
that the sort of governmental conduct we so strongly condemned when it was
practiced by the old communist regimes in Eastern Europe?
While it is unclear whether we will be able to learn anything about the
implementation of the FBI's policy through our lawsuit, the legal opinion
described above certainly raises questions that should be pursued. I would be
glad to keep interested folks posted on developments, though I'd prefer to do
so through private e-mail, i.e., "with an expectation of privacy." Send me a
note if you'd like to be kept informed about the litigation.
David Sobel, CPSR Legal Counsel cdp!dsobel@labrea.stanford.edu
["... with an expectation of privacy" is subtle, in that apart from
"privacy enhanced e-mail", e-mail goes over unencrypted local and
global networks, is handled by some decidely unsecure systems, and
is typically forwarded iteratively to other people. "... with some
hope of privacy" might be more accurate! But I imagine David will
get some requests from a few government RISKS contributors, who
might like to know what "developments" are turning up... PGN]
CAF discusses such questions as : How should general principles of academic
freedom (such as freedom of expression, freedom to read, due process, and
privacy) be applied to university computers and networks? How are these
principles actually being applied? How can the principles of academic freedom
as applied to computers and networks be defended?
The EFF has given the discussion a home on the eff.org machine. As of
April 23, less than two week after its creation, the list has 230 members in
four countries.
There are three versions of the mailing list:
comp-academic-freedom-talk
- you'll received dozens of e-mail notes every day.
comp-academic-freedom-batch
- about once a day, you'll receive a compilation of the day's notes.
comp-academic-freedom-news
- about once a week you'll receive a compilation of the best
notes of the week. (I play the editor for this one).
To join a version of the list, send mail to listserv@eff.org.
Include the line "add <name-of-version>". (Other commands are "delete
<name-of-version>" and "help").
In any case, after you join the list you can send e-mail to the
LIST BY addressing it to caf-talk@eff.org.
These mailing lists are also available as the USENET alt groups
'alt.comp.acad-freedom.talk' and 'alt.comp.acad-freedom.news'.
| This page was copied from: | http://catless.ncl.ac.uk/Risks/11.82.html |
| COPY! | |
| COPY! | |
|
by Michael Blume |