Worm storms

[Julian V. Noble]

Dear C Mavens,

Anyone here getting hosts of spam with nefarious attachments,
purporting to be from M$ or its lackeys, into your mailbox?

I neglected to spoof my header, and since Hurricane Isabel
I have gotten well over 10K such messages.
--
Julian V. Noble
Professor Emeritus of Physics
jv*@lessspamfor mother.virginia .edu
^^^^^^^^^^^^^^^ ^^^
http://galileo.phys.virginia.edu/~jvn/

"Science knows only one commandment: contribute to science."
-- Bertolt Brecht, "Galileo".

[Zygmunt Krynicki]

On Wed, 24 Sep 2003 19:22:05 +0300, Ian Tuomi wrote:

Julian V. Noble wrote:

Dear C Mavens,

Anyone here getting hosts of spam with nefarious attachments,
purporting to be from M$ or its lackeys, into your mailbox?

I neglected to spoof my header, and since Hurricane Isabel
I have gotten well over 10K such messages.

I got suprised one day as it turned out that I had ~200 messagess waiting
for me. The bad thing is that I have *slow* connection and those messages
were simply killing my system. I had 100+ of sendmails hanging around and
waiting forever for the mail to arrive.

To be topical: what is the keyword "restricted " for, how old is it? I've
noticed a couple of people giving little hints that it's for telling the
programmer/compiler it's illegal to pass the same thing more than once. I
dont know if I got it correctly or is it just my imagination working.
Anyway what is the reason for such a construct? The olny example I could
think of was something like memcpy - memove (it's a little slopy, I know
it's not exactly the same).
Fell free to correct me

Zygmunt

[Malcolm]

"Zygmunt Krynicki" <zyga@_CUT_2zyg a.MEdyndns._OUT _org> wrote in message

To be topical: what is the keyword "restricted " for, how old is it? I've
noticed a couple of people giving little hints that it's for telling the
programmer/compiler it's illegal to pass the same thing more than once.

Say we've got the following function

int mean(int *data, int N, int *err)
{
/* add up the data, if you get an overflow then set err */
}

The problem comes when err points to one of the data elements pointed to by
data. This is obviously pathological from the point of view of a human
programmer who knows the intent of the function, but to the compiler it is
legal C.
The need to handle pointer aliasing may make it difficult to optimise the
function. For instance, if integers are four bytes but the architecture
allows 8 bytes to be read from memory at one cycle, the compiler cannot take
advantage of this because of the possibility that a write to *err has
invalidated the second data item.

[Malcolm]

"Zygmunt Krynicki" <zyga@_CUT_2zyg a.MEdyndns._OUT _org> wrote in message

To be topical: what is the keyword "restricted " for, how old is it? I've
noticed a couple of people giving little hints that it's for telling the
programmer/compiler it's illegal to pass the same thing more than once.

Say we've got the following function

int mean(int *data, int N, int *err)
{
/* add up the data, if you get an overflow then set err */
}

The problem comes when err points to one of the data elements pointed to by
data. This is obviously pathological from the point of view of a human
programmer who knows the intent of the function, but to the compiler it is
legal C.
The need to handle pointer aliasing may make it difficult to optimise the
function. For instance, if integers are four bytes but the architecture
allows 8 bytes to be read from memory at one cycle, the compiler cannot take
advantage of this because of the possibility that a write to *err has
invalidated the second data item.

[Bertrand Mollinier Toublet]

Julian V. Noble wrote:

Dear C Mavens,

Anyone here getting hosts of spam with nefarious attachments,
purporting to be from M$ or its lackeys, into your mailbox?

I neglected to spoof my header, and since Hurricane Isabel
I have gotten well over 10K such messages.

I do get emails from Swen infected users, to my one and only public
email address, probably collected from c.l.c when I was posting without
mangling it. For some reason, though, I do not get any unmanageable
amount :-/ Maybe 50 emails tops since last friday...

--
Bertrand Mollinier Toublet
"In regard to Ducatis vs. women, it has been said: 'One is a sexy thing
that you've just got to ride, even if it breaks down a lot, costs a lot
of money, and will probably try to kill you'. However, nowadays I can't
seem to remember which one is which." -- Peer Landa

[Bertrand Mollinier Toublet]

Julian V. Noble wrote:

Dear C Mavens,

Anyone here getting hosts of spam with nefarious attachments,
purporting to be from M$ or its lackeys, into your mailbox?

I neglected to spoof my header, and since Hurricane Isabel
I have gotten well over 10K such messages.

I do get emails from Swen infected users, to my one and only public
email address, probably collected from c.l.c when I was posting without
mangling it. For some reason, though, I do not get any unmanageable
amount :-/ Maybe 50 emails tops since last friday...

--
Bertrand Mollinier Toublet
"In regard to Ducatis vs. women, it has been said: 'One is a sexy thing
that you've just got to ride, even if it breaks down a lot, costs a lot
of money, and will probably try to kill you'. However, nowadays I can't
seem to remember which one is which." -- Peer Landa

[Simon Biber]

"Julian V. Noble" <jvn-at-virginia.edu> wrote:

Dear C Mavens,

Anyone here getting hosts of spam with nefarious attachments,
purporting to be from M$ or its lackeys, into your mailbox?

I neglected to spoof my header, and since Hurricane Isabel
I have gotten well over 10K such messages.

Since 2003/9/18 I have received about 4000 copies of the worm
Swen.A. That's about 600 megabytes added to my monthly quota :(

I think a lot of people on comp.lang.c are affected according to a bounce message I received:

---
The file (part0004:q8349 94.exe) attached to mail (with subject: Current Net Critical Pack) sent by
sales.dep-at-xnet.ro to jens.toerring-at-physik.fu-berlin.de, 80bluesky-at-gmx.at,
calum.bulk-at-ntlworld.com, jacob.navia-at-jacob.remcomp.f r, thomas.pfaff-at-tiscali.no,
nicole0169-at-citiz.net, christian.bau-at-cbau.freeserve. co.uk, sbiber-at-optushome.com.a u,
foo.foo-at-gmx.net, debashis_kolkat a-at-rediffmail.com, nimel-at-passagen.se, a.litowka-at-gmx.de,
gah-at-ugcs.caltech.ed u, gin-at-binky.homeunix. org, dagwyn-at-null.net, mambuhl-at-earthlink.net,
mason_verger-at-skincare.com, lawrence.jones-at-eds.com, klachemin-at-home.com,
pyf-at-mail.zjitc.net, nzanella-at-cs.mun.ca, francischeng-at-hong-kong.crosswinds .net,
jcook-at-strobedata.com, emonk-at-slingshot.co.nz .no.uce, pushkar-at-erc.msstate.edu ,
lfw-at-airmail.net, binary-at-eton.powernet.c o.uk, airia-at-acay.com.au, chris-at-sonnack.com,
kst-at-cts.com, derkgwen-at-hotpop.com, dontmail-at-address.co.uk.i nvalid, mkwahler-at-mkwahler.net,
os2guy-at-pc-rosenau.de, richmond-at-ev1.net, horpner-at-yahoo.com, nglen702-at-netscape.net,
stewart.brodie-at-ntlworld.com, ayeameen-at-yahoo.com, parinioa-at-hotmail.com,
malcolm-at-55bank.freeserv e.co.uk, joewwright-at-earthlink.net, m_donaghy50-at-hotmail.com,
robertvazan-at-privateweb.sk, kevin.bracey-at-tematic.com, dan.pop-at-cern.ch, thadsmith-at-acm.org,
nethlek-at-tokyo.com, koster_thomas-at-yahoo.com.sg, ajo-at-andrew.cmu.edu,
first.last-at-company.com, aurer-at-axis.com, palaste-at-cc.helsinki.fi, eric.sosman-at-sun.com,
msgregoryz-at-earthlink.net, kers-at-hpl.hp.com, d99alu-at-efd.lth.se, cmccormick-at-mailsnare.net,
chrisval-at-bigpond.com.au, kuyper-at-saicmodis.com, deliberately-at-made.invalid,
ak+usenet-at-freeshell.org, irrwahn-at-freenet.de, xal-at-abowers.combase .com,
s030768-at-student.dtu.dk, pfiland-at-mindspring.com, scs-at-eskimo.com, noizetogo-at-direct.ca,
glenhallick-at-sprint.ca, cdvanos-at-telus.net, n36170-at-hotmail.com, me-at-here.com,
danmc-at-shaw.ca, magpie-at-shinythings.com , keimdf-at-softek-net.com is infected with virus:
Win32/Swen.A-at-mm.
---

(@ replaced with -at- in this message to try to prevent this
email list from being picked up by spambots.)

--
Simon.

[Simon Biber]

"Julian V. Noble" <jvn-at-virginia.edu> wrote:

Dear C Mavens,

Anyone here getting hosts of spam with nefarious attachments,
purporting to be from M$ or its lackeys, into your mailbox?

I neglected to spoof my header, and since Hurricane Isabel
I have gotten well over 10K such messages.

Since 2003/9/18 I have received about 4000 copies of the worm
Swen.A. That's about 600 megabytes added to my monthly quota :(

I think a lot of people on comp.lang.c are affected according to a bounce message I received:

---
The file (part0004:q8349 94.exe) attached to mail (with subject: Current Net Critical Pack) sent by
sales.dep-at-xnet.ro to jens.toerring-at-physik.fu-berlin.de, 80bluesky-at-gmx.at,
calum.bulk-at-ntlworld.com, jacob.navia-at-jacob.remcomp.f r, thomas.pfaff-at-tiscali.no,
nicole0169-at-citiz.net, christian.bau-at-cbau.freeserve. co.uk, sbiber-at-optushome.com.a u,
foo.foo-at-gmx.net, debashis_kolkat a-at-rediffmail.com, nimel-at-passagen.se, a.litowka-at-gmx.de,
gah-at-ugcs.caltech.ed u, gin-at-binky.homeunix. org, dagwyn-at-null.net, mambuhl-at-earthlink.net,
mason_verger-at-skincare.com, lawrence.jones-at-eds.com, klachemin-at-home.com,
pyf-at-mail.zjitc.net, nzanella-at-cs.mun.ca, francischeng-at-hong-kong.crosswinds .net,
jcook-at-strobedata.com, emonk-at-slingshot.co.nz .no.uce, pushkar-at-erc.msstate.edu ,
lfw-at-airmail.net, binary-at-eton.powernet.c o.uk, airia-at-acay.com.au, chris-at-sonnack.com,
kst-at-cts.com, derkgwen-at-hotpop.com, dontmail-at-address.co.uk.i nvalid, mkwahler-at-mkwahler.net,
os2guy-at-pc-rosenau.de, richmond-at-ev1.net, horpner-at-yahoo.com, nglen702-at-netscape.net,
stewart.brodie-at-ntlworld.com, ayeameen-at-yahoo.com, parinioa-at-hotmail.com,
malcolm-at-55bank.freeserv e.co.uk, joewwright-at-earthlink.net, m_donaghy50-at-hotmail.com,
robertvazan-at-privateweb.sk, kevin.bracey-at-tematic.com, dan.pop-at-cern.ch, thadsmith-at-acm.org,
nethlek-at-tokyo.com, koster_thomas-at-yahoo.com.sg, ajo-at-andrew.cmu.edu,
first.last-at-company.com, aurer-at-axis.com, palaste-at-cc.helsinki.fi, eric.sosman-at-sun.com,
msgregoryz-at-earthlink.net, kers-at-hpl.hp.com, d99alu-at-efd.lth.se, cmccormick-at-mailsnare.net,
chrisval-at-bigpond.com.au, kuyper-at-saicmodis.com, deliberately-at-made.invalid,
ak+usenet-at-freeshell.org, irrwahn-at-freenet.de, xal-at-abowers.combase .com,
s030768-at-student.dtu.dk, pfiland-at-mindspring.com, scs-at-eskimo.com, noizetogo-at-direct.ca,
glenhallick-at-sprint.ca, cdvanos-at-telus.net, n36170-at-hotmail.com, me-at-here.com,
danmc-at-shaw.ca, magpie-at-shinythings.com , keimdf-at-softek-net.com is infected with virus:
Win32/Swen.A-at-mm.
---

(@ replaced with -at- in this message to try to prevent this
email list from being picked up by spambots.)

--
Simon.

[Christian Bau]

In article
<pan.2003.09.24 .21.21.18.47767 9@_CUT_2zyga.ME dyndns._OUT_org >,
"Zygmunt Krynicki" <zyga@_CUT_2zyg a.MEdyndns._OUT _org> wrote:

On Wed, 24 Sep 2003 19:22:05 +0300, Ian Tuomi wrote:

Julian V. Noble wrote:

Dear C Mavens,

Anyone here getting hosts of spam with nefarious attachments,
purporting to be from M$ or its lackeys, into your mailbox?

I neglected to spoof my header, and since Hurricane Isabel
I have gotten well over 10K such messages.

I got suprised one day as it turned out that I had ~200 messagess waiting
for me. The bad thing is that I have *slow* connection and those messages
were simply killing my system. I had 100+ of sendmails hanging around and
waiting forever for the mail to arrive.

Recommendation: Use Mozilla Firebird. It lets you choose "don't download
messages over xx Kilobyte", so it downloads only about 1KB of each of
these messages and then you can delete them.
To be topical: what is the keyword "restricted " for, how old is it? I've
noticed a couple of people giving little hints that it's for telling the
programmer/compiler it's illegal to pass the same thing more than once.
It is there since C99. There are two uses:

1. If you use a pointer like "int * restrict p", then it is undefined
behavior if you modify an object through an expression that is derived
from the value of p, and access it through a different pointer; and it
is also undefined behavior if you access an object through an expression
that is derived from the value of p, and access it modify it through a
different pointer.

This is important for an optimising compiler. Example:

int *restrict p;
int *q;

int x = *q, y;
*p = 2;
y = *q;

The compiler can assume that y == x because the assignment to *p cannot
change *q (if it did you would have violated the first rule).

2. If you use a pointer like "const int * restrict p", then it is
undefined behavior if you modify an object that is accessed through an
expression that is derived from the value of p. In other words, *p
cannot be modified as long as the pointer p exists. Usually, if you have
a const* pointer then the object pointed to can still be modified by
other means, or by casting the const-ness away. Not if it is a const
*restrict pointer.
I dont know if I got it correctly or is it just my imagination working.
Anyway what is the reason for such a construct? The olny example I could
think of was something like memcpy - memove (it's a little slopy, I know
it's not exactly the same).

[Christian Bau]

In article
<pan.2003.09.24 .21.21.18.47767 9@_CUT_2zyga.ME dyndns._OUT_org >,
"Zygmunt Krynicki" <zyga@_CUT_2zyg a.MEdyndns._OUT _org> wrote:

On Wed, 24 Sep 2003 19:22:05 +0300, Ian Tuomi wrote:

Julian V. Noble wrote:

Dear C Mavens,

Anyone here getting hosts of spam with nefarious attachments,
purporting to be from M$ or its lackeys, into your mailbox?

I neglected to spoof my header, and since Hurricane Isabel
I have gotten well over 10K such messages.

I got suprised one day as it turned out that I had ~200 messagess waiting
for me. The bad thing is that I have *slow* connection and those messages
were simply killing my system. I had 100+ of sendmails hanging around and
waiting forever for the mail to arrive.

Recommendation: Use Mozilla Firebird. It lets you choose "don't download
messages over xx Kilobyte", so it downloads only about 1KB of each of
these messages and then you can delete them.
To be topical: what is the keyword "restricted " for, how old is it? I've
noticed a couple of people giving little hints that it's for telling the
programmer/compiler it's illegal to pass the same thing more than once.
It is there since C99. There are two uses:

1. If you use a pointer like "int * restrict p", then it is undefined
behavior if you modify an object through an expression that is derived
from the value of p, and access it through a different pointer; and it
is also undefined behavior if you access an object through an expression
that is derived from the value of p, and access it modify it through a
different pointer.

This is important for an optimising compiler. Example:

int *restrict p;
int *q;

int x = *q, y;
*p = 2;
y = *q;

The compiler can assume that y == x because the assignment to *p cannot
change *q (if it did you would have violated the first rule).

2. If you use a pointer like "const int * restrict p", then it is
undefined behavior if you modify an object that is accessed through an
expression that is derived from the value of p. In other words, *p
cannot be modified as long as the pointer p exists. Usually, if you have
a const* pointer then the object pointed to can still be modified by
other means, or by casting the const-ness away. Not if it is a const
*restrict pointer.
I dont know if I got it correctly or is it just my imagination working.
Anyway what is the reason for such a construct? The olny example I could
think of was something like memcpy - memove (it's a little slopy, I know
it's not exactly the same).

[Irrwahn Grausewitz]

"Simon Biber" <sb****@optusho me.com.au> wrote:

"Julian V. Noble" <jvn-at-virginia.edu> wrote:

Dear C Mavens,

Anyone here getting hosts of spam with nefarious attachments,
purporting to be from M$ or its lackeys, into your mailbox?

I neglected to spoof my header, and since Hurricane Isabel
I have gotten well over 10K such messages.

Since 2003/9/18 I have received about 4000 copies of the worm
Swen.A. That's about 600 megabytes added to my monthly quota :(

I think a lot of people on comp.lang.c are affected according to a bounce message I received:

<who-is-who in c.l.c snipped>

Just what I thought. I had to re-route the traffic to the address I
used when posting here to /dev/null, after receiving about forty virus-
or bounce-messages per hour. The new alias redirects to a working
spam-free account (after removing the capitals).

Irrwahn
(currently using his old 14.4K Hayes Optima on a flaky phone line)
--
Close your eyes and press escape three times.