To think I've thrown out four linux servers and a Sun E450 because I'd forgotten the root passwd on all of them, and thought I'd never be able to get in!!
Some exploit! It's been in the HOWTO's since year dot, yet still nobody has fixed this "vulnerability".
Ah, matey I think, for your own good you should RTFA. After you do, you'll be enlightened and may even realise your system is vulnerable! Anyway, for a sneak peek, the exercise is not finding out your window title, quite the opposite. It's about exploiting your terminal to change the title, then use another exploit to copy the title to your command buffer. What I'm doing here might be called Karma whoring, so I'll leave the rest to your imagination and again encourage you to read up.
Oh, and I don't believe this has been said today yet: All your 25th line are belong to us!
Hm, an IDE cable, are you sure? I've connected IDE cables backward numerous times. IDE cables do not carry power (like SCSI), only low-level signals.
Same here, IDE and floppy cables, I hooked 'em up backwards as a matter of course when I started working on x86 hardware, worst I ever got was refusal to boot, always remedied by putting the cable on the other way.
That's not to say we're bloody lucky we weren't punished for what is not a trivial mistake. I think it's possible to cock up the AT motherboard power connector, which would almost certainly cook everything connected.
Oh yeah, wish I'd mentioned that, I heard on nz.comp of someone plugging in a parallel port ZIP drive into a Mac. Now we all know that Mac's don't have parallel ports, and neither device survived (so I'm told). Years ago at a security company I programmed for, the assistant manager re-set up the Amiga 500 + GVP Hard Drive, and plugged the parallel printer into the GVP scsi, not the parport. I don't know if said printer ever got powered up, but both devices were fine.
Needless to say, the rest of the thread chastised Apple for using DB25 for their scsi ports, not HD50. They're not the only ones guilty of that though, but it's unfortunate that (since there aren't many connectors that are interchangeable) SCSI and parallel (sooooo disimilar) for some time shared the same connector.
It only depends on the capabilities of the human, and what he uses to read it. You wouldn't use more to read/proc/kcore, you'd use a hex browser. You could use more or vi to edit XML, but an XML editor will make this even easier.
Ever seen what happens with a SCSI ribbon back to front!? Most of the time it's impossible because the plugs have a little notch preventing that. Of course, with the myriad of people making cables, some end up with the notches on the wrong side, etc, or without any at all.
My friend did this, and the amount of smoke from just one wire on the ribbon was amazing (to me anyway, he didn't seem to take it the right way).
It was like someone took a knife and sliced the ribbon all the way down, making two parallel ribbons. You ask: was everything OK afterwards? Yes! The scsi card and all devices were fine. Did his scsi card have a fuse on the terminator power wire? Obviously not. The event was locally named the "Lonergan SCSI terminator power-wire-fire". Boy I can't believe how tough some of this hardware is. The only HD I've every blown was because I let the onboard controller touch the case chassis. Spark! I've got cables the wrong way round, forgotten to plug fans in, hotplugged stuff I shouldn't... I can't kill anything!!
Ah, that's good. And thanks, I didn't know about that difference between LVM and RAID. Still, even losing one filesystem at a time is Russian roulette. (Uh oh,/invites Soviet Russia joke).
Add all this to the fact that, on the site there are no photos of the project complete with case screws attached. I'm assuming the machine is still running naked with guts poking out everywhere. His cat might knock that ugly tower of drives over! One drive overheating might cause another drive to overheat! Ugh, the more I think about it the more I shudder. Again, I'm glad I dont' need 1.8TB. If/when I do, I won't do it this way.
Gnome (1.?) (redhat 6.0) -> Gnome 1.4 (liked it a little) -> Gnome 2.0 ->/usr/X11R6/bin/twm
Now [in Gnome2] even the (dismal) file requester takes about five seconds to draw up. I think Gnome is going the wrong way. It's only a good desktop for those that like to fire up a number of apps / windows, and then sit there looking at it / fire screenshots around the globe.
Apart from adding a bit of spiff, all I can find are features removed. I heard someone type 'gnome 2.2...now you can have transparency in the panel'. Well, 1.4 had that, 2.0 didn't.
I started using Gnome about three months after I started using Linux, about five years ago. I've been hoping it hasn't, but all I've seen (apart from a glimmer of hope in 1.4) is Gnome go to shit. I think I'm going to get the 1.4 source, maintain it myself and keep using that.
Not trolling, I'm genuinely disappointed that Gnome has given Linux a bad name, not because Gnome is Linux, but Joe Public who's used to a Windows system thinks that, he can't draw an abstraction between a desktop environment and an OS, and I think that's why a lot of people think Linux sucks. I also think adoption of Gnome by Sun is a bad move. Sure CDE sucks, but it runs.
Exactly. When you think about it, if he doesn't have any redundancy, he's seventeen times more likely to suffer a disk failure than anyone else with one drive.
I can't begin to think how you'd come back up after losing a drive in a concatenated R^HAID. Whoops, no R if it's not redundant eh?
I'm actually quite glad I'm not sitting on 1.8TB of data, and I don't intend to in the near future.
If he does mirror the drives, I wonder if his mobo will be the bottleneck..?
Let's not forget that most OS's that run an X window system also go a long way to prevent stuff like this happening. I like your idea, however, and can assure you that the next release of Outlook will have this vulnerability.
*sigh* don't get me started about Xinerama. Or about GNOME on xinerama anyway. You see by "improved xinerama support", GNOME mean "now EVERY window pops up on the border between both monitors".
Apart from GNOME, some window managers behaved better than others, Enlightenment was pretty well behaved. But you know what? The ONLY combo that hasn't pissed me off completeley, was CDE, yes CDE running on Solaris8 with XFree86, Xinerama, and (sorry) Win98. Of the three OS's my PC can boot, Linux was actually the one that made me take the extra monitor off the desk, and replace it with an actual fishtank.
I can't wait until someone shouts the GNOME developers a second video card, because I don't otherwise mind using it. I just wish I could find the "Stupid-window-placement stochastic prediction engine? y/n" switch.
Exactly, you're right. Which just goes to show, we need to stop the comparison between the timesharing multi-user operating system and the internal combustion engine.
Analogies fall flat on their face everytime you liken a computer to a car.
Internal combustion engine: Converts chemical potential energy to kinetic energy. It will use the fuel at a finite rate, until the fuel is exhausted.
Timesharing multi-user operating system: Continuously and repetitively runs "operating system" code and "user" code. This code will never deplete, save an equipment faulure.
If anything, a computer is more similar to a washing machine.
There are tremendous differences between the Linux kernel and the Solaris kernel. Same with any BSD kernel vs. Linux vs. Solaris.
Are you saying that there wouldn't be any difference between a Solaris kernel and s SCO kernel? Or an IRIX kernel, or even a Solaris 2.x v SunOS 4.x kernel?? After all, these systems are UNIX in their own right.
Yes I know there is more to being UNIX than just having the fork() system call, but what other basket would you put Linux in at the time it came out?
Since almost any source destined for a UNIX system could also compile on a Linux system, why not? Linux was designed with the intent of being a "unix-like" system, and it is at least that. The fact is, to be allowed to call your product a UNIX costs $500,000 a year in the first instance. Then your system gets vetted to make sure it fits the standard.
The only thing I've seen common on UNIX systems the the arrangement of bin usr dev etc directories in the root and plenty of things diverging from there. As for where everything is put, how things work, command line opts, nothing is the same.
I know the poster has already said he's disconnected his speakers and the problem still persists, but the cdrom audio -> snd card cable picked up noise from everything in my box, gfx redraws, hdd access of course, you can even hear the tray motor when the cdrom opens.
So I have my CD mixer all the way down for the most of the time I don't use it. Next cdrom must have spdif out.
BTW poster, memcpy()'s won't do it - they only copy within main memory. Graphics operations will cause data to pass from main memory to gfx card. This IMO will be more likely to recreate the problem. Tried a different graphics card?
Actually, this 'hole' is worse the one in Windows. Windows config data is stored in the registry, which is binary and so is much harder to manually edit than the plain-text files in/etc/ on a Linux box.
True, but do you really think the registry is going to be one of the targets of a hac^H^H^Hintruder anyway?
And if they did want it, I'm sure reverse engineering the registry is much easier than the (common among crackers) skill to find, and craft exploits (architecture specific, ASM coding req'd) against buffer overflows.
I'm sure after you've build a registry decoder, you'll get just as much info from it as you get from/etc.
Step 7 - go on web, find OEM passwords for BIOS
Step 8 - game over, boot floppy/CD.
Even if a default password doesn't get them through the BIOS, you can open the case. Then either muck around with the BIOS jumbers, or for the impatient, slip the HDD into your cargo pants, and head home.
Yes the GRUB password prevents someone from booting another image / device. Even though I have a BIOS password set, I don't expect it to get in anybodys way should they want access to my machine, and I'm not in between them and it.
That's patently untrue
No he's right! All these people are doing are installing [Linux|OSX] on servers just hoping nobody's going to spend the time h4x0ring them.
And for some reason, they just get left alone! Yes, that's why Linux is so lean! They just don't put in any code for checking things like passwords, buffers etc. because nobody even tries to hack into any OS if it isn't windows...
God forbid any h4x0rs read the Linux source, lest they find all the/* FIXME - we probably should compare the password entered with the hash in/etc/shadow, but nobody reads this stuff anyway */
Yes, Linux affords security only through obscurity. Anybody reading the source code could find 10 security holes in as many minutes eh?
No way at all you could do that. That board would fry if the engine was running. You can't use a DC regulator, because the battery voltage would need to be at least 1.25 v above the 12 we need, and we can't always guarantee that.
Furthermore, the 5V supply will generate a hell of a lot of heat. You won't be able to use any old 7805 for it, because you'll be getting at least 5A through it.
Your only option would be an inverter and then a normal power supply, or a dedicated 12v switched mode PSU. When all is said in done, a laptop that can be charged in car is usually a much better option, whether you're just play MP3's, when you may as well get a player anyway, or needing actual computing. A desktop computer in a car is usually a very ugly site.
I believe you should have called me a "Karma Whore", yes, that's the term for someone posting something actually factual or useful here. No worries though.
Slashdot Mirror
To think I've thrown out four linux servers and a Sun E450 because I'd forgotten the root passwd on all of them, and thought I'd never be able to get in!!
Some exploit! It's been in the HOWTO's since year dot, yet still nobody has fixed this "vulnerability".
Oh, and I don't believe this has been said today yet: All your 25th line are belong to us!
Same here, IDE and floppy cables, I hooked 'em up backwards as a matter of course when I started working on x86 hardware, worst I ever got was refusal to boot, always remedied by putting the cable on the other way.
That's not to say we're bloody lucky we weren't punished for what is not a trivial mistake. I think it's possible to cock up the AT motherboard power connector, which would almost certainly cook everything connected.
Needless to say, the rest of the thread chastised Apple for using DB25 for their scsi ports, not HD50. They're not the only ones guilty of that though, but it's unfortunate that (since there aren't many connectors that are interchangeable) SCSI and parallel (sooooo disimilar) for some time shared the same connector.
It only depends on the capabilities of the human, and what he uses to read it. You wouldn't use more to read /proc/kcore, you'd use a hex browser. You could use more or vi to edit XML, but an XML editor will make this even easier.
My friend did this, and the amount of smoke from just one wire on the ribbon was amazing (to me anyway, he didn't seem to take it the right way).
It was like someone took a knife and sliced the ribbon all the way down, making two parallel ribbons. You ask: was everything OK afterwards? Yes! The scsi card and all devices were fine. Did his scsi card have a fuse on the terminator power wire? Obviously not. The event was locally named the "Lonergan SCSI terminator power-wire-fire". Boy I can't believe how tough some of this hardware is. The only HD I've every blown was because I let the onboard controller touch the case chassis. Spark! I've got cables the wrong way round, forgotten to plug fans in, hotplugged stuff I shouldn't... I can't kill anything!!
Add all this to the fact that, on the site there are no photos of the project complete with case screws attached. I'm assuming the machine is still running naked with guts poking out everywhere. His cat might knock that ugly tower of drives over! One drive overheating might cause another drive to overheat! Ugh, the more I think about it the more I shudder. Again, I'm glad I dont' need 1.8TB. If/when I do, I won't do it this way.
Now [in Gnome2] even the (dismal) file requester takes about five seconds to draw up. I think Gnome is going the wrong way. It's only a good desktop for those that like to fire up a number of apps / windows, and then sit there looking at it / fire screenshots around the globe.
Apart from adding a bit of spiff, all I can find are features removed. I heard someone type 'gnome 2.2...now you can have transparency in the panel'. Well, 1.4 had that, 2.0 didn't.
I started using Gnome about three months after I started using Linux, about five years ago. I've been hoping it hasn't, but all I've seen (apart from a glimmer of hope in 1.4) is Gnome go to shit. I think I'm going to get the 1.4 source, maintain it myself and keep using that.
Not trolling, I'm genuinely disappointed that Gnome has given Linux a bad name, not because Gnome is Linux, but Joe Public who's used to a Windows system thinks that, he can't draw an abstraction between a desktop environment and an OS, and I think that's why a lot of people think Linux sucks. I also think adoption of Gnome by Sun is a bad move. Sure CDE sucks, but it runs.
I can't begin to think how you'd come back up after losing a drive in a concatenated R^HAID. Whoops, no R if it's not redundant eh?
I'm actually quite glad I'm not sitting on 1.8TB of data, and I don't intend to in the near future.
If he does mirror the drives, I wonder if his mobo will be the bottleneck..?
Apart from GNOME, some window managers behaved better than others, Enlightenment was pretty well behaved. But you know what? The ONLY combo that hasn't pissed me off completeley, was CDE, yes CDE running on Solaris8 with XFree86, Xinerama, and (sorry) Win98. Of the three OS's my PC can boot, Linux was actually the one that made me take the extra monitor off the desk, and replace it with an actual fishtank.
I can't wait until someone shouts the GNOME developers a second video card, because I don't otherwise mind using it. I just wish I could find the "Stupid-window-placement stochastic prediction engine? y/n" switch.
I would tend to think that Symptom == "root server responses too slow" Disease == "too much unnecessary traffic"
Analogies fall flat on their face everytime you liken a computer to a car.
Internal combustion engine: Converts chemical potential energy to kinetic energy. It will use the fuel at a finite rate, until the fuel is exhausted.
Timesharing multi-user operating system: Continuously and repetitively runs "operating system" code and "user" code. This code will never deplete, save an equipment faulure.
If anything, a computer is more similar to a washing machine.
#ifdef TROLL
Anyone on Slashdot use these?
#endif
Are you saying that there wouldn't be any difference between a Solaris kernel and s SCO kernel? Or an IRIX kernel, or even a Solaris 2.x v SunOS 4.x kernel?? After all, these systems are UNIX in their own right.
Yes I know there is more to being UNIX than just having the fork() system call, but what other basket would you put Linux in at the time it came out?
Since almost any source destined for a UNIX system could also compile on a Linux system, why not? Linux was designed with the intent of being a "unix-like" system, and it is at least that. The fact is, to be allowed to call your product a UNIX costs $500,000 a year in the first instance. Then your system gets vetted to make sure it fits the standard.
The only thing I've seen common on UNIX systems the the arrangement of bin usr dev etc directories in the root and plenty of things diverging from there. As for where everything is put, how things work, command line opts, nothing is the same.
I've got ones that fart, crackle and squeal. Everybody! Degauss on 3..2..1..
So I have my CD mixer all the way down for the most of the time I don't use it. Next cdrom must have spdif out.
BTW poster, memcpy()'s won't do it - they only copy within main memory. Graphics operations will cause data to pass from main memory to gfx card. This IMO will be more likely to recreate the problem. Tried a different graphics card?
I think in future I will use a sarcasm tag, not because I think you're daft, but just like you say, hard to tell in here sometimes...
Sorry, I suppose I really should have used a -sarcasm- tag
True, but do you really think the registry is going to be one of the targets of a hac^H^H^Hintruder anyway?
And if they did want it, I'm sure reverse engineering the registry is much easier than the (common among crackers) skill to find, and craft exploits (architecture specific, ASM coding req'd) against buffer overflows.
I'm sure after you've build a registry decoder, you'll get just as much info from it as you get from /etc.
Sheesh, that's unusual.
Even if a default password doesn't get them through the BIOS, you can open the case. Then either muck around with the BIOS jumbers, or for the impatient, slip the HDD into your cargo pants, and head home.
Yes the GRUB password prevents someone from booting another image / device. Even though I have a BIOS password set, I don't expect it to get in anybodys way should they want access to my machine, and I'm not in between them and it.
No he's right! All these people are doing are installing [Linux|OSX] on servers just hoping nobody's going to spend the time h4x0ring them.
And for some reason, they just get left alone! Yes, that's why Linux is so lean! They just don't put in any code for checking things like passwords, buffers etc. because nobody even tries to hack into any OS if it isn't windows...
God forbid any h4x0rs read the Linux source, lest they find all the /* FIXME - we probably should compare the password entered with the hash in /etc/shadow, but nobody reads this stuff anyway */
Yes, Linux affords security only through obscurity. Anybody reading the source code could find 10 security holes in as many minutes eh?
An OpenPROM on a Sun Sparc, with password protection would be a lot harder to get through, though still not infallible.
You could do this way _any_ os, not just OpenBSD, of course, and I'm sure you knew that.
Even if the computer had no floppy, no cdrom, you could still find ways of circumventing authentication if you had access to the machine.
Furthermore, the 5V supply will generate a hell of a lot of heat. You won't be able to use any old 7805 for it, because you'll be getting at least 5A through it.
Your only option would be an inverter and then a normal power supply, or a dedicated 12v switched mode PSU. When all is said in done, a laptop that can be charged in car is usually a much better option, whether you're just play MP3's, when you may as well get a player anyway, or needing actual computing. A desktop computer in a car is usually a very ugly site.
I believe you should have called me a "Karma Whore", yes, that's the term for someone posting something actually factual or useful here. No worries though.