You might be able to achieve what you want with chattr -A or chattr +A. (Probably in combination with -R). Eg chattr -R +A/usr/ should turn off atime updates on all of/usr
I think it only works to turn off atime updates so you would need to mount with atime and then turn it off on the folders where you don't need it.
(Warning the above is untested: but the LKML suggests it works as does the chattr man page)
the/usr directory is one of the more important ones if you do any compiling as each compile accesses large numbers of files in/usr/include/
So in the period you mention hard disk capacity increased 1000 fold. Hard disk throughput increased 10-20 times. Hard disk access time (remember you need to add half a disk rotation to the seek time) is about 4 times faster. As far I can see 1990 is before the 486 so i am guesing a high end processor was something like a 33MHz 386. Taking a high end processor now as a 3Ghz PIV thats suggests about a 100 times increas in speed of processor.
So hard disk capacity has comfortably outpaced processor speed but processors have outpaced disk throughput and even more so disk latency.
Is it clearly impossible for a computer to generate an AI complete problem? A priori the computer could start with the solution and then work out the question, which may be computationally feasible, whilst working out the answer is not (without intelligence).
And how many local exploits does the unix system have? The distinction between local and root here is not significant. (For example why is there no official 2.4 kernel release fixing the local->root ptrace vulnerability? In part because there are so many other local->root vulnerabilities).
I dont know how many local->root exploits windows has but i doubt that on either system it's very difficult.
One difficulty is the user configuration files. If the attacker changes things like root's bashrc that may not be noticed by such tools as it is probably not deemed a system file. So what you say could help but still leaves large holes. The same is even more difficult to check if it is a user's bashrc as they may be modified moderately frequently (making them harder to check against a known good file).
Pynnonen (the guy who found the exploit) has posted a new message to Bugtraq. If the servers reply is crafted correctly it can cause the program to be downloaded executed with *no* dialogs. See the posting for more details. Still no exploit given though.
sorry I didnt make that clear. If you follow the above advice you should not get any filesystem corruption. The last line is what to do if you have already got a corrupt filesystem!
Breakage happens when you umount filesystem (_any_ local filesystem, be
it ext2, reiserfs, whatever) that still has dirty inodes.
As a workaround - sync before umount (and don't boot
unpatched 2.4.15/2.4.15-pre9 again, obviously).
IOW, if you are running 2.4.15 - build a patched kernel, install it and
do the following:
* switch to single-user
* sync
* umount everything non-busy
* remount the rest read-only
* turn the thing off
* boot with patched kernel or with anything before 2.4.15-pre9
The filesystem corruption can be fixed by a forced fsck. (The fsck must be forced since the filesystem is marked clean.)
It will probably be at least a week until 2.5 really gets moving. Linus on the lkml
...when I release 2.4.15, I'll at
the same time release a 2.5.0 that is identical except for version number
(that makes synchronization easier later on). And I'll probably _not_
start accepting all the big waiting patches immediately, I'd rather wait
for at least a week or two to see that there aren't any other issues.
It's much easier doing some of the IO patches in particular knowing that
the base you start out from is stable.
The GPL is still necessary even with no copyright. In the absence of copyright I could take your source code modify and distribute the binary without the source.
I cannot charge for the binary (at least anyone else can distribute it for free).
I use this method and it works nicely for binaries/paths. But it does not work well for shared libraries because ldconfig does not follow symbolic links. Currently I have to add each directory to/etc/ld.so.conf. Is there a good way round this?
But first you have to decode it. The articles says that decoders will only be available in HDTV sets. So until someone cracks the encryption compression will be difficult.
Whilst life on earth is something like 4 billion years old isn't cellular life much younger at something like one billion years old? To my mind this makes the likelihood of these bacteria not being from space much lower.
Re:The most interesting & missing parts of the sto
on
Chip News To Crunch On
·
· Score: 1
This is probably an error. A slightly older story on the same site quotes final quarter of 2001 for first samples with production in the first quarter 2002. The AMD 64 bit FAQ agrees.
Quite right. Universities in the UK currently pay for their trans-atlantic traffic (approx 2p a MB). Somebody in the US tried to hack into one of the machines in my university and were sending in excess of 5GB of ICMP packets a day. That can get expensive! In this case I believe that the university didnt get charged but with wider metering the potential would be there.
Another difficulty (though much less important) is that you cant show a mirrored surface. You cant even show reflections of parts of the image itself unless you know where the viewer is. Any thoughts on other limitations?
Yes... but how does the other party know the key. I think that it is sent using public key encryption at the start. Eg SSH uses RSA to send a IDEA or DES key when the connection starts. (They switch since RSA is computationally much more expensive than IDEA or DES).
So if you can break the public key at the start the whole session is open.
I know its cheap enough for us to use it. But what of the large web servers? Presumably they buy hardware which can cope with the needs of serving (and given slashdotting presumably not a lot more). Wouldn't encryption add very significantly to their costs?
Correct me if I am wrong. I don't know the relative computational cost. This is just a guess. Anyone with more info?
In my experience NT in a virtual machine is fine whereas 98 was painfully slow (on a PIII 500 with 128MB). So a performance comparison between 98 on VMWare and Win4Lin is not as useful as one between NT on VMWare and Win4Lin (assuming you just want to run windows apps).
Anyone know how other OS compare in the virtual machine?
I agree. How about an encoder which asks the listener to sit some tests when first used so that it can optimise the psychoacoustic model for that listener?
I agree that unix is more secure but it does not seem to be set up to deal with hostile processes. Suppose that you are running X. Then it (any hostile process) can read off all your keystrokes and so find passwords PGP encoding etc. It can also edit *your* startup scripts (eg.cshrc) to make sure that it gets run whenever you login. This may take a time but if it doesn't do anything nasty until it has gathered its information and spread then that does not matter, in fact it may be an advantage as there is less panic about it. Of course this all assumes that someone chose to execute the hostile script....
Slashdot Mirror
You might be able to achieve what you want with chattr -A or chattr +A. (Probably in combination with -R). Eg chattr -R +A /usr/ should turn off atime updates on all of /usr
/usr directory is one of the more important ones if you do any compiling as each compile accesses large numbers of files in /usr/include/
I think it only works to turn off atime updates so you would need to mount with atime and then turn it off on the folders where you don't need it.
(Warning the above is untested: but the LKML suggests it works as does the chattr man page)
the
K
Writer of I love you virus sues for copyright infringement.
"People just kept distributing copies of my IP" the author claimed earlier today.
So in the period you mention hard disk capacity increased 1000 fold. Hard disk throughput increased 10-20 times. Hard disk access time (remember you need to add half a disk rotation to the seek time) is about 4 times faster. As far I can see 1990 is before the 486 so i am guesing a high end processor was something like a 33MHz 386. Taking a high end processor now as a 3Ghz PIV thats suggests about a 100 times increas in speed of processor.
So hard disk capacity has comfortably outpaced processor speed but processors have outpaced disk throughput and even more so disk latency.
Is it clearly impossible for a computer to generate an AI complete problem? A priori the computer could start with the solution and then work out the question, which may be computationally feasible, whilst working out the answer is not (without intelligence).
And how many local exploits does the unix system have? The distinction between local and root here is not significant. (For example why is there no official 2.4 kernel release fixing the local->root ptrace vulnerability? In part because there are so many other local->root vulnerabilities).
I dont know how many local->root exploits windows has but i doubt that on either system it's very difficult.
One difficulty is the user configuration files. If the attacker changes things like root's bashrc that may not be noticed by such tools as it is probably not deemed a system file. So what you say could help but still leaves large holes. The same is even more difficult to check if it is a user's bashrc as they may be modified moderately frequently (making them harder to check against a known good file).
Pynnonen (the guy who found the exploit) has posted a new message to Bugtraq. If the servers reply is crafted correctly it can cause the program to be downloaded executed with *no* dialogs. See the posting for more details. Still no exploit given though.
-K
sorry I didnt make that clear. If you follow the above advice you should not get any filesystem corruption. The last line is what to do if you have already got a corrupt filesystem!
The GPL is still necessary even with no copyright. In the absence of copyright I could take your source code modify and distribute the binary without the source.
I cannot charge for the binary (at least anyone else can distribute it for free).
This is very different from the GPL.
K
I use this method and it works nicely for binaries/paths. But it does not work well for shared libraries because ldconfig does not follow symbolic links. Currently I have to add each directory to /etc/ld.so.conf. Is there a good way round this?
Or am I just being stupid?
-K
But first you have to decode it. The articles says that decoders will only be available in HDTV sets. So until someone cracks the encryption compression will be difficult.
Whilst life on earth is something like 4 billion years old isn't cellular life much younger at something like one billion years old? To my mind this makes the likelihood of these bacteria not being from space much lower.
This is probably an error. A slightly older story on the same site quotes final quarter of 2001 for first samples with production in the first quarter 2002. The AMD 64 bit FAQ agrees.
Quite right. Universities in the UK currently pay for their trans-atlantic traffic (approx 2p a MB). Somebody in the US tried to hack into one of the machines in my university and were sending in excess of 5GB of ICMP packets a day. That can get expensive! In this case I believe that the university didnt get charged but with wider metering the potential would be there.
Another difficulty (though much less important) is that you cant show a mirrored surface. You cant even show reflections of parts of the image itself unless you know where the viewer is. Any thoughts on other limitations?
So if you can break the public key at the start the whole session is open.
Correct me if I am wrong. I don't know the relative computational cost. This is just a guess. Anyone with more info?
In my experience NT in a virtual machine is fine whereas 98 was painfully slow (on a PIII 500 with 128MB). So a performance comparison between 98 on VMWare and Win4Lin is not as useful as one between NT on VMWare and Win4Lin (assuming you just want to run windows apps). Anyone know how other OS compare in the virtual machine?
I agree. How about an encoder which asks the listener to sit some tests when first used so that it can optimise the psychoacoustic model for that listener?
I agree that unix is more secure but it does not seem to be set up to deal with hostile processes. Suppose that you are running X. Then it (any hostile process) can read off all your keystrokes and so find passwords PGP encoding etc. It can also edit *your* startup scripts (eg .cshrc) to make sure that it gets run whenever you login. This may take a time but if it doesn't do anything nasty until it has gathered its information and spread then that does not matter, in fact it may be an advantage as there is less panic about it. Of course this all assumes that someone chose to execute the hostile script....