Check out this hypothetical- the browser was just an URL bar, and then you could add plugins / extensions / whatnot. Into this reasonably open environment, inject different ways of doing pretty much everything- favorites, start pages, content blocking and modification, tabs, etc.
You or I would tinker with this and it would be pretty much perfect, until we have to go to another machine. Then we are reliant on either moving these preferences about via some online doohicky (login to "the browser", really logging into the mothership and pulling down what it needs to look like you are used to), or setting them up from scratch each time.
Each addon could easily be abandoned, with the best version of displaying tabs or whatever ceasing functionality six months after you have finally figured out all the details of it.
Finally, every addon you have to get increases the risk of it being a bad addon- someone needs to vet that, and that's a huge deal.
I'm fine with many of the things Firefox (and others) have decided is default UI- bookmarks, tabs, etc. Like you, I've got no time for things like pocket- no interest in that at all. But my point is, it's not obvious what is core functionality and what is not, and every addon that you have to remember and pull down is a hassle for you at some point in the future- it ceases to be a product and becomes a system you have to integrate. That's perfect for power and customization, but you know that it's a burden many would opt out of if given the chance.
Yes, of course it is. It's been found that way in court, and what the fuck ELSE would it be? Wait, don't answer that: every other possibility is terrifying.
> The weight of a GPU chip and a couple of extra VRAM chips isnâ(TM)t going to break anyoneâ(TM)s back.
I disagree. Along with a GPU comes cooling requirements, and those definitely add both weight and size.
An external GPU would let you:
1)- Easily replace a wonky GPU, which is a gamble right now. "Al, lemme try your graphicbox. Ok, see, mine must be bad. I'll buy a new one." 2)- Easily upgrade a GPU, which is *almost impossible* now. "Huh, the new card just went on sale. It's been a couple years and I want to upgrade, but I don't want to get a whole new laptop. But this is just, buy the new one and drop it in." 3)- Possibly actually repair a GPU. "Ok, I know it's bad, but I can see the problem, and I don't have to be inside the machine I do my work on in order to fix it." 4)- Remove the need for expensive integration. Laptops are almost impossible to build and ones that actually work well with graphics cards are rare. 5)- Offload heavy and loud cooling options to an entirely different case. 6)- Prevent GPU heat from ever touching your HDD, SDD, RAM, and CPU. 7)- Pack in a fan that is large, slow, and quiet, versus slimline high RPM ones. 8)- Use a goddamned DESKTOP GPU. Mobile GPUs have the same names as desktop GPUs, usually with an "m" somewhere in there, but it's an absolute joke- in many cases, a third of the processing. It's simply a much more constrained system. 9)- Game on systems made by companies that don't have to support "gaming systems", such as Apple. Right now, if a company wants to support gaming, that's a whole market choice. If instead you have a standard you can hook up to any compatible box, you have no problems.
It will increase consumer choice, instantly double to quadruple the amount of graphics power brought to bear, lighten your laptop, shrink your laptop, make your laptop last much longer, and let you repair and upgrade it.
It's such a good idea I mostly doubt it will happen.
This country is called "America" by almost everyone here. Our patriotic songs call it that too, as do our politicians and journalists. We're Americans, we live in America, and you don't get to name us or our nation.
If you are looking to reduce greenhouse emissions, you need to address root causes, such as "cars make X CO2, could make less", "giant freighters free monumental amounts of carbon as they take raw materials from the US to China and finished products back to the US", and "cities are oppressive for many, who live in suburbs, creating a vast amount of useless travel each day".
If your philosophy is saying "more people will live better lives- WHAT A PROBLEM!", then just ditch that philosophy. It sucks.
> Where is the morality in creating a system upgrade that would defeat us from ever preventing this?
Same morality that doesn't require the government cut off our hands and feet at birth. Tools should function as tools, not be agents of a state. Remember, if everything is backdoored for state use (to prevent terrorism, or whatever song and dance they have to sing to make it so you own and control nothing for fear that you too are a terrorist), then we are just rolling the dice every year and hoping we continue to roll up governments that aren't willing to tighten that noose. Once you can't be trusted to control your data, everything else will follow soon enough, and you are betting on the government being continuously correct for, say, four hundred years. What are the odds of that? Can it even be calculated?
> The real issue here is that they can apparently update the firmware of the Secure Enclave
Correct.
The 5c doesn't have the same security as the 6 and 6s. However, Apple has implied that some attack on those could even be mounted- just not this attack.
I'm sure it will be fixed later.
> consider that the secure storage in the current Nexus phones and many other devices does not have this flaw
I'm not convinced of that.
Hypothetical: Before this happened, pretend someone asked that question here. Wouldn't security folks and Apple fanboys have told you it was impossible? Spoiler: it isn't really hypothetical. If you look back a few days, there was a slashdot story about this, from before all the details were out. The only reason ANYONE suspects that the 6 and 6s are vulnerable to an attack of this class (having their secure enclave somehow worked around) is that Apple implied that something like that could happen when they posted their letter. No one knows how that would work- we just assume that something like that could exist.
So, is the Nexus immune? Maybe. But on the other hand, the FBI isn't trying to unlock a fucking Nexus, so we don't have them trying to All Writs Act Google. So we won't know for sure.
I do think that this will be fixed in the next ios rev, and if Google's stuff is anything but solid, it will also be addressed in their next wave of stuff.
Absolutely correct. You cannot, within the field of computer science, ever make a 4 digit PIN secure. This is why they step outside computer science to use hardware tricks to hopefully prevent the brute force attack in the first place.
> You could use a key derived from a PBKDF2 hash with such a high number of rounds that it requires 6 hours to unlock the device, after you typed in the correct passcode.
Ok, but then I just try it 10,000 times, on a machine that is faster than the little tiny crypto processor on there. Or if I own a farm of them, I can try all 10,000 at the same time.
Rest assured it is ABSOLUTELY impossible. If the PBKDF2 iterations on correct entry will unlock within your lifetime, then it can be brute forced within hours, days, or at worst months. And that's if you are willing to wait like 50 years to unlock your phone.
It doesn't matter how many repetitions you do of your hash function either- if the ios one takes 10 minutes to check a passcode, then specialized hardware could do that same attempt in a few seconds, and you could buy 10,000 copies of the hardware if you needed to.
You can't solve it with computer science- it can't be done. Hence the reliance on hardware and the hope that the hardware is not breakable. Or you use a real passphrase and then you are safe from this attack.
"The class key is protected with the hardware UID and, for some classes, the user's passcode"
Yes, it's part of the crypto system. Yes, choosing a real one makes it secure and choosing a shitty one means you are trusting other parts to prevent brute force.
> The 4 digit passcode isn't the cryptographic key, it's something that the user can use to identify himself to the phone a limited number of times
No, actually, the passphrase- which can be a 4 digit code, a 6 digit code, or an arbitrary length passphrase that is crypto secure- is combined, inside the secure enclave, with other data. It's not an if/then check, it's actually crypto happening there. The whole point of the hardware drama is to prevent that from being tried multiple times. It's not a user identification code, it's actually crypto- and obviously a 4 digit passcode is shit for that, hence the hardware and software trying hard to prevent multiple attempts.
Well, Apple has been trying to address this with every hardware rev. Later ones have enforcement in the secure enclave. This attack would absolutely not work on the 6 and 6s, but Apple has implied that there's still some way even there- I bet you see that patched in a later version.
LUKS doesn't force you to not use a 4 digit PIN. Does LUKS suck? Veracrypt will let you use a 1 digit passphrase too. Does Veracrypt suck?
You just implicitly shit on the best crypto engines in the world in the process of finding something about Apple to hate. Yuk yuk yuk, good joke.
It's not "getting past the lockscreen". That implies this is a software control. There is a master key that is encrypted by a combination of the user passphrase and some hardware specific stuff. That master key is used to unwrap all the file specific AES-128 XTS keys (or possibly AES-256 CBC keys- I'm not 100% sure which is used on that version of hardware).
So to reiterate, this is NOT a software guard, or "getting past the lockscreen". If you forced your way past the lockscreen, you couldn't access any of the data, which is meaningless. What the FBI needs is to get around the logic that wipes the key, and on this older hardware that's still possible (and possibly on the newer hardware as well), thus allowing many tries. Once many tries are enabled, you are relying on the crypto itself, which, like any crypto, is total shit if you just have a 4 digit fucking PIN.
> crypto isn't worth a god damn thing if all you have to do is load custom firmware
Crypto is NEVER worth anything if you use a 4 digit PIN. Apple has a combination of hardware and firmware to attempt to prevent the 4 digit PIN from being brute forced, and that's what this is all about. The later versions have even more enforcement in hardware (though Apple has implied that there exists some manner of attack against that).
The thing is, this isn't a backdoor. This is a hardware level of security that you aren't obligated to use (you can absolutely make a secure passphrase on ios, and the government can't brute force that). If this stuff gets broken, then your security level falls to (roughly) that of LUKS and Truecrypt and such. But those are still pretty fucking secure- just not if you use a 4 digit PIN.
Again, the current system is secure if you trust the math only by using a crypto secure passphrase. If you don't, you MUST be trusting the hardware or software to guard against the brute force- really the wimp force, because 10k trials is nothing.
You can set up an alphanumeric passphrase of massive length under settings. Then you are secure against brute force no matter what, same as if you used it as your Veracrypt passphrase.
> My point is that if Apple can push such a software update to an existing phone without the user unlocking the device first, then iOS cryptography is broken already.
You should look a bit more into it.
First, if we are talking CRYPTO, lets be real: a 4 digit passcode is triival to brute force. I don't care WHAT you use- Twofish/AES/Serpent in Veracrypt, I will absolutely break your 4 digit passcode in moments. Because it's a fucking FOUR DIGIT PASSCODE.
So, how does Apple try to secure this? The only way it can- with hardware. The crypto is 128 bit AES, so they aren't trying to attack that. Later versions of the iphone have secure hardware implement this sort of logic. The version in question actually IS less secure- it has software that does the task of the wiping. Apple is refusing to build and cryptographically sign software that will do it.
There's no cryptographic way to secure a 4 digit passcode, or a 6 digit passcode. It's physically impossible. Hence the use of hardware. If you have a serious crypto passphrase on your iphone- and you absolutely can- then the only way in is through the crypto, either the AES or the PBKDF2. It's not as strong as AES 256 XTS (because it is AES 128 XTS), but it is still considered unbreakable.
So don't talk shit about their crypto if their crypto isn't even up for debate. This is about a software workaround possible on an older model to brute force requests into the hardware that is expected to defend a 4 digit passcode against repeated attempts. The crypto isn't even in the conversation.
Dude, how many times a day do you WANT to enter your 40+ character passphrase, with caps, lowers, numbers, and specials?
You are correct that having a hardware setup to allow for 4 or 6 digit PINs is not as ideal as that, but realistically nobody would use that security model.
Yes. You get full encryption with a 4 digit PIN. The key is stored in (supposedly) tamperproof hardware, and it is setup to blank the key after 10 failed attempts.
> correct firewall settings...and let me guess, you have these settings, but there's no room to include them in the margin?
I get that Windows users want everything to be fine, just more random configuration scripts, external firewall settings, services to remove from the command line, KBs to blacklist. But this level of configuration is really confusing. On Linux you get all this for free, and you never have to leave a GUI... Or set anything in the first place.
Windows is kill. I hate it too, but it's absolutely true.
Slashdot Mirror
I disagree. Not entirely, but enough to comment.
Check out this hypothetical- the browser was just an URL bar, and then you could add plugins / extensions / whatnot. Into this reasonably open environment, inject different ways of doing pretty much everything- favorites, start pages, content blocking and modification, tabs, etc.
You or I would tinker with this and it would be pretty much perfect, until we have to go to another machine. Then we are reliant on either moving these preferences about via some online doohicky (login to "the browser", really logging into the mothership and pulling down what it needs to look like you are used to), or setting them up from scratch each time.
Each addon could easily be abandoned, with the best version of displaying tabs or whatever ceasing functionality six months after you have finally figured out all the details of it.
Finally, every addon you have to get increases the risk of it being a bad addon- someone needs to vet that, and that's a huge deal.
I'm fine with many of the things Firefox (and others) have decided is default UI- bookmarks, tabs, etc. Like you, I've got no time for things like pocket- no interest in that at all. But my point is, it's not obvious what is core functionality and what is not, and every addon that you have to remember and pull down is a hassle for you at some point in the future- it ceases to be a product and becomes a system you have to integrate. That's perfect for power and customization, but you know that it's a burden many would opt out of if given the chance.
Who fucking cares about 99% of PC users. Gamers care about this shit, and gamers are nerds, and this is news for nerds.
> But, code is not speech.
Yes, of course it is. It's been found that way in court, and what the fuck ELSE would it be? Wait, don't answer that: every other possibility is terrifying.
> The weight of a GPU chip and a couple of extra VRAM chips isnâ(TM)t going to break anyoneâ(TM)s back.
I disagree. Along with a GPU comes cooling requirements, and those definitely add both weight and size.
An external GPU would let you:
1)- Easily replace a wonky GPU, which is a gamble right now. "Al, lemme try your graphicbox. Ok, see, mine must be bad. I'll buy a new one."
2)- Easily upgrade a GPU, which is *almost impossible* now. "Huh, the new card just went on sale. It's been a couple years and I want to upgrade, but I don't want to get a whole new laptop. But this is just, buy the new one and drop it in."
3)- Possibly actually repair a GPU. "Ok, I know it's bad, but I can see the problem, and I don't have to be inside the machine I do my work on in order to fix it."
4)- Remove the need for expensive integration. Laptops are almost impossible to build and ones that actually work well with graphics cards are rare.
5)- Offload heavy and loud cooling options to an entirely different case.
6)- Prevent GPU heat from ever touching your HDD, SDD, RAM, and CPU.
7)- Pack in a fan that is large, slow, and quiet, versus slimline high RPM ones.
8)- Use a goddamned DESKTOP GPU. Mobile GPUs have the same names as desktop GPUs, usually with an "m" somewhere in there, but it's an absolute joke- in many cases, a third of the processing. It's simply a much more constrained system.
9)- Game on systems made by companies that don't have to support "gaming systems", such as Apple. Right now, if a company wants to support gaming, that's a whole market choice. If instead you have a standard you can hook up to any compatible box, you have no problems.
It will increase consumer choice, instantly double to quadruple the amount of graphics power brought to bear, lighten your laptop, shrink your laptop, make your laptop last much longer, and let you repair and upgrade it.
It's such a good idea I mostly doubt it will happen.
>> Quake 3 was the best FPS
> LOL
No, that's the best moba.
This country is called "America" by almost everyone here. Our patriotic songs call it that too, as do our politicians and journalists. We're Americans, we live in America, and you don't get to name us or our nation.
If you are looking to reduce greenhouse emissions, you need to address root causes, such as "cars make X CO2, could make less", "giant freighters free monumental amounts of carbon as they take raw materials from the US to China and finished products back to the US", and "cities are oppressive for many, who live in suburbs, creating a vast amount of useless travel each day".
If your philosophy is saying "more people will live better lives- WHAT A PROBLEM!", then just ditch that philosophy. It sucks.
Exactly. It "disrupts the relationship" in the same way mosquito netting "disrupts the relationship" between mosquitoes and mammal flesh.
> I'll close the tab
hahaha you'll TRY to close the tab- maybe it'll listen
> Where is the morality in creating a system upgrade that would defeat us from ever preventing this?
Same morality that doesn't require the government cut off our hands and feet at birth. Tools should function as tools, not be agents of a state. Remember, if everything is backdoored for state use (to prevent terrorism, or whatever song and dance they have to sing to make it so you own and control nothing for fear that you too are a terrorist), then we are just rolling the dice every year and hoping we continue to roll up governments that aren't willing to tighten that noose. Once you can't be trusted to control your data, everything else will follow soon enough, and you are betting on the government being continuously correct for, say, four hundred years. What are the odds of that? Can it even be calculated?
If the user has enabled the feature and leaves his phone around monkeys for two hours, then yes.
Then he goes home and restores it from his encrypted backup.
> The real issue here is that they can apparently update the firmware of the Secure Enclave
Correct.
The 5c doesn't have the same security as the 6 and 6s. However, Apple has implied that some attack on those could even be mounted- just not this attack.
I'm sure it will be fixed later.
> consider that the secure storage in the current Nexus phones and many other devices does not have this flaw
I'm not convinced of that.
Hypothetical: Before this happened, pretend someone asked that question here. Wouldn't security folks and Apple fanboys have told you it was impossible?
Spoiler: it isn't really hypothetical. If you look back a few days, there was a slashdot story about this, from before all the details were out. The only reason ANYONE suspects that the 6 and 6s are vulnerable to an attack of this class (having their secure enclave somehow worked around) is that Apple implied that something like that could happen when they posted their letter. No one knows how that would work- we just assume that something like that could exist.
So, is the Nexus immune? Maybe. But on the other hand, the FBI isn't trying to unlock a fucking Nexus, so we don't have them trying to All Writs Act Google. So we won't know for sure.
I do think that this will be fixed in the next ios rev, and if Google's stuff is anything but solid, it will also be addressed in their next wave of stuff.
Absolutely correct. You cannot, within the field of computer science, ever make a 4 digit PIN secure. This is why they step outside computer science to use hardware tricks to hopefully prevent the brute force attack in the first place.
> You could use a key derived from a PBKDF2 hash with such a high number of rounds that it requires 6 hours to unlock the device, after you typed in the correct passcode.
Ok, but then I just try it 10,000 times, on a machine that is faster than the little tiny crypto processor on there. Or if I own a farm of them, I can try all 10,000 at the same time.
Rest assured it is ABSOLUTELY impossible. If the PBKDF2 iterations on correct entry will unlock within your lifetime, then it can be brute forced within hours, days, or at worst months. And that's if you are willing to wait like 50 years to unlock your phone.
> Store hash.
Do 10,000 attempts. Trivial.
Or do 1 attempt 10,000 times.
It doesn't matter how many repetitions you do of your hash function either- if the ios one takes 10 minutes to check a passcode, then specialized hardware could do that same attempt in a few seconds, and you could buy 10,000 copies of the hardware if you needed to.
You can't solve it with computer science- it can't be done. Hence the reliance on hardware and the hope that the hardware is not breakable. Or you use a real passphrase and then you are safe from this attack.
> the pin you use to access the phone is the password or key used for encrypting the data
Good fucking grief.
https://www.apple.com/business...
"The class key is protected with the hardware UID and, for some classes, the user's passcode"
Yes, it's part of the crypto system. Yes, choosing a real one makes it secure and choosing a shitty one means you are trusting other parts to prevent brute force.
> Key management is an essential part of a cryptosystem
Good thing this has nothing to do with key management.
https://en.wikipedia.org/wiki/...
> The 4 digit passcode isn't the cryptographic key, it's something that the user can use to identify himself to the phone a limited number of times
No, actually, the passphrase- which can be a 4 digit code, a 6 digit code, or an arbitrary length passphrase that is crypto secure- is combined, inside the secure enclave, with other data. It's not an if/then check, it's actually crypto happening there. The whole point of the hardware drama is to prevent that from being tried multiple times. It's not a user identification code, it's actually crypto- and obviously a 4 digit passcode is shit for that, hence the hardware and software trying hard to prevent multiple attempts.
Well, Apple has been trying to address this with every hardware rev. Later ones have enforcement in the secure enclave. This attack would absolutely not work on the 6 and 6s, but Apple has implied that there's still some way even there- I bet you see that patched in a later version.
LUKS doesn't force you to not use a 4 digit PIN. Does LUKS suck? Veracrypt will let you use a 1 digit passphrase too. Does Veracrypt suck?
You just implicitly shit on the best crypto engines in the world in the process of finding something about Apple to hate. Yuk yuk yuk, good joke.
It's not "getting past the lockscreen". That implies this is a software control. There is a master key that is encrypted by a combination of the user passphrase and some hardware specific stuff. That master key is used to unwrap all the file specific AES-128 XTS keys (or possibly AES-256 CBC keys- I'm not 100% sure which is used on that version of hardware).
https://www.apple.com/business...
So to reiterate, this is NOT a software guard, or "getting past the lockscreen". If you forced your way past the lockscreen, you couldn't access any of the data, which is meaningless. What the FBI needs is to get around the logic that wipes the key, and on this older hardware that's still possible (and possibly on the newer hardware as well), thus allowing many tries. Once many tries are enabled, you are relying on the crypto itself, which, like any crypto, is total shit if you just have a 4 digit fucking PIN.
> crypto isn't worth a god damn thing if all you have to do is load custom firmware
Crypto is NEVER worth anything if you use a 4 digit PIN. Apple has a combination of hardware and firmware to attempt to prevent the 4 digit PIN from being brute forced, and that's what this is all about. The later versions have even more enforcement in hardware (though Apple has implied that there exists some manner of attack against that).
The thing is, this isn't a backdoor. This is a hardware level of security that you aren't obligated to use (you can absolutely make a secure passphrase on ios, and the government can't brute force that). If this stuff gets broken, then your security level falls to (roughly) that of LUKS and Truecrypt and such. But those are still pretty fucking secure- just not if you use a 4 digit PIN.
Again, the current system is secure if you trust the math only by using a crypto secure passphrase. If you don't, you MUST be trusting the hardware or software to guard against the brute force- really the wimp force, because 10k trials is nothing.
You can set up an alphanumeric passphrase of massive length under settings. Then you are secure against brute force no matter what, same as if you used it as your Veracrypt passphrase.
> My point is that if Apple can push such a software update to an existing phone without the user unlocking the device first, then iOS cryptography is broken already.
You should look a bit more into it.
First, if we are talking CRYPTO, lets be real: a 4 digit passcode is triival to brute force. I don't care WHAT you use- Twofish/AES/Serpent in Veracrypt, I will absolutely break your 4 digit passcode in moments. Because it's a fucking FOUR DIGIT PASSCODE.
So, how does Apple try to secure this? The only way it can- with hardware. The crypto is 128 bit AES, so they aren't trying to attack that. Later versions of the iphone have secure hardware implement this sort of logic. The version in question actually IS less secure- it has software that does the task of the wiping. Apple is refusing to build and cryptographically sign software that will do it.
There's no cryptographic way to secure a 4 digit passcode, or a 6 digit passcode. It's physically impossible. Hence the use of hardware. If you have a serious crypto passphrase on your iphone- and you absolutely can- then the only way in is through the crypto, either the AES or the PBKDF2. It's not as strong as AES 256 XTS (because it is AES 128 XTS), but it is still considered unbreakable.
So don't talk shit about their crypto if their crypto isn't even up for debate. This is about a software workaround possible on an older model to brute force requests into the hardware that is expected to defend a 4 digit passcode against repeated attempts. The crypto isn't even in the conversation.
Dude, how many times a day do you WANT to enter your 40+ character passphrase, with caps, lowers, numbers, and specials?
You are correct that having a hardware setup to allow for 4 or 6 digit PINs is not as ideal as that, but realistically nobody would use that security model.
Yes. You get full encryption with a 4 digit PIN. The key is stored in (supposedly) tamperproof hardware, and it is setup to blank the key after 10 failed attempts.
> correct firewall settings ...and let me guess, you have these settings, but there's no room to include them in the margin?
I get that Windows users want everything to be fine, just more random configuration scripts, external firewall settings, services to remove from the command line, KBs to blacklist. But this level of configuration is really confusing. On Linux you get all this for free, and you never have to leave a GUI... Or set anything in the first place.
Windows is kill. I hate it too, but it's absolutely true.