Gadgetory


All Cool Mind-blowing Gadgets You Love in One Place

Myths About Meltdown & Spectre: Expert Interviews

2018-01-29
1995 is the year that the first vulnerable processors to the meltdown and specter attacks were created it's also the year that the discoverer or one of them of those attacks was born fast forward to today January 2018 and we have industry luminaries like linus torvalds not not not that linus no disrespect saying things like the intel patch for meltdown are quote complete and utter garbage probably while waving his middle finger around as he often does before that this video is brought to you by the gamers nexus at patreon and our patreon backers do you want to help us out directly you can go to patreon.com/scishow and check with the GN team or you can support us at $5 or higher and get access to behind the scenes videos as we release them once or twice a month learn more at the link in the description below it's been nearly a month since major news that broke on the meltdown and spectre exploits and the tech press along with all of the users in the space are still swarming like an upturned anthill trying to figure out what exactly is going on Intel is one of those ants they've released a couple of patches now some of which have actually been detrimental and they've had to recall them and AMD arm Apple pretty much everyone has put out some kind of patch Microsoft all of them because everyone at some level is affected by meltdown inspector basically anyone who is using out of order processing on their CPU which would include phones tablets laptops desktops servers anything you can think of the most part smartwatches as well I suppose so we wanted to come back to that topic and see where we are today now that it's been a month what's been going on what's the current state of things and in doing so we realized this is a pretty complex topic let's bring in some experts so we reached out to several of the researchers who were initially credited with the discovery of meltdown inspector and we greatly appreciate their time and responding to our questions to clarify how these exploits affect you and to what extent they have been addressed or resolved and where they are today so we'll be also addressing some misconceptions on specter and meltdown we basically said to the researchers here's your opportunity have you seen any commentary online that you find either mislead or misguided or inversely have you seen anything that's been left out of the discussion so we've got all that today along with a bit of a timeline for a preview of this we have a great quote from cyber s technology who stated to us with respect to Spectre it became obvious that CPU software aka micro code is not being developed with the same level of quality as the hardware at least at Intel so were you talking about that more as well we should start with some terminology here there's a common phrase that is used pertaining to meltdown and Spector and that phrase is a side channel attack which I'm going to read a definition from a brief by the European Union Agency for information security this brief says that a side channel attack is quote an exploit observable and measurable with computational side effects to extract infer otherwise unavailable secret information and data side channel attacks are well known to be used against cryptographic operations while addressing this term we should also address the idea of speculative execution this is something we discussed in our first meltdown inspector explanation video and it's worth coming back to because in speaking with the researchers we've learned that one of the biggest misconceptions that's been going around with all this is the idea of that speculative execution itself is the exploit or is the bad thing in all facts and reality speculative execution is what allows us with modern processors to achieve at the level of performance that we can achieve basically everyone uses it it inspector effects everything basically going back to 1995 so this is a big deal and speculative execution is in all those things because it's an accelerator it allow us CPUs to more or less step ahead of their current queue they can think ahead by looking at what's going on now what has gone in the past in this situation what do we think is going to happen next and then the CPU will try and predict or preempt as the key word is that next task it's possible that in doing so that next task never occurs it might be a wasted preemption step however the upside is significant and losing a couple of cycles every now and then because you predicted something incorrectly as the CPU is far less of a hit to performance than the gains because of how accurate it often is so speculative execution itself is actually a good thing Spectre is the specifically named attack that is sort of branded and leverages speculative execution as an entry or an attack vector to work its way into the data stream more or less which is where you get your exposure in terms of security vulnerabilities so difference there between spectre and speculative execution this has been around for more than 20 years and is why we see the attacks so wide reaching the most comprehensive hub of information on meltdown in Spectre is the website hosted by gratz University of Technology in Austria this is home to one of the research teams that discovered and reported the issues to Intel and that's one of because there are no fewer than three other teams acknowledged by grads in their independently discovered and reported vulnerabilities over the past few months we've assembled a rough timeline of events with the aid of some research from other outlets that you can find on our website linked in the description below in the article of this but for purposes of the video let's focus on the email exchanges with these researchers who are obviously experts in the field and some of the best people to talk to you right now about meltdown and Spectre first off we spoke with anders fogh of GData GData is one of the first if not the first firm to publicly talk about and draw attention to potential abuse of speculative execution as an attack vector they also have one of the more detailed and early blog posts about these attack vectors we further spoke with cyber s technology CTO Verner house and architect Thomas pressure we also spoke as Michael Schwartz of gratz University who discovered meltdown apparently separately from these others so multiple researchers at once working independently of one another coming to similar or the same conclusions and in our separate emails with them they all pretty much agree on what's going on the meltdown attack comm website that's widely sourced is from the team that Michael Schwartz is on so we've got his answers as well the questions and answers will be truncated for video you want to see the full questions and answers again article below we'll have them without any cuts at all we first asked whether Spector the exploit with the least public focus requires physical access to execute we read some early reports and user comments on videos that suggested Spector needed physical i/o access and all three research teams replied to tell us that no neither meltdown nor Spector required physical access to quote anders fogh of GData they are purely software that just utilized how the hardware works referring to the exploits house and pressure of cyber is explained to us that meltdown inspector only need code to execute on the victim machine but noted that physical access is not required Schwartz of Graz University reminded us that spectra can be mounted from JavaScript thus in the browser and therefore requires no physical access so in short no an attacker doesn't need physical access to your computer for spectre to work which is one of the main misconceptions that has been around regarding inspector the next question we asked whether Spector could be executed remotely so this would be for example via browser this is important because if it can execute through JavaScript JavaScript is everywhere basically every website you visit uses some amount of JavaScript for the most part especially anything with an ad network which is 90 plus percent of the Internet so if it can be executed via browsers or hijacked ad networks it's a concern meltdown inspector have both been proven to work with JavaScript and a compromised website or a malicious website could therefore be used as an attack vector or a compromised ad network which is actually somewhat common so we wanted to ask about that GData tells us that that browse vendors have been made of aware of the attack and that many of them have already released updates so if your browser has recently received an annoying icon that says hey it's time to update it's time to update just you're gonna lose your tabs it's okay you'll get them back it's probably a good idea to update though it'll protect you later GData also noted that kpti or kernel page table isolation will work well against meltdown specifically from a browser and suggested using kpti and updating browsers where relevant for example with linux as your OS and g data also emphasized the following further current versions of Spectre and meltdown are members of the class of timing attacks other timing attacks have been shown to be launched a ball over the network even without JavaScript thus it is theoretically possible that we'll see such attacks in the future my personal opinion is that this is more likely to be an academic exercise than a real-world vector regarding the same question the group at Cypress replied and they stated similarly the good news is that browser vendors have already rolled out patches addressing the vulnerabilities and continued to state that an attacker needs good quality timing information in order to be able to detect traces of a misled speculative execution and caches so it sounds like this has been somewhat addressed with the browser updates now this is still a concern as we'll see going through all of this information but the immediate severity of it has been at least somewhat mitigated with the browser patches we haven't yet gotten to hardware level issues though and we'll get there as far as common misconceptions we asked the researchers what online commentary they'd seen that seems misguided or seem to misunderstand the core story here G data noted that speculative execution is used as an attack vector but in and of itself is not an attack something we clarified earlier G data also further emphasized that quote Specter style attacks can be launched without speculative execution but would usually require the presence of a classic software bug even in the absence of software bugs and speculative execution we can do scary things with side channels see the earlier side-channel definition for that house and pressure also highlighted that spectate of execution again as a standalone feature is not a bad thing nor is using cash both are required for modern-day performance that we see out of our CPUs doesn't matter if it's AMD in teller arm they all need those advantages to get the performance we see today they also need to essentially reaaargh attack for the future but we'll see where they go with that the Cypress team noted that quote requests for the replacement of all processors currently in use are unduly exaggerated and further stated having a vulnerable CPU does not imply that the system it can be readily compromised which is also a good bit of level-headedness injected it into the conversation so out of all this the good news from what the researchers tell us is that meltdown has very obvious changes that can be made and are being worked on at least for future architectures whether or not we can get them in today the other bit of good news is that these things can be made to resolve the problem at a hardware level and nip it in the bud so to speak and the researchers believe this will lead to future processors not being vulnerable to the same types of attacks we're seeing today and to be clear meltdown inspector don't really have any proven real-world attacks yet we know that it's possible to do them the researchers have all released their own proofs of concept or discussion conceptually of what can happen but whether or not specter meltdown have been used at a wide scale is either unknown or it just hasn't happened we're also told that the linux kpti and other patches are recommended as an effective protection against meltdown Spectre currently has stopgap solutions proposed and will likely require more testing and research to close up and then finally on the more negative side of things Schwartz highlighted for us that the specter and meltdown attacks can bypass sandboxing or virtualization so if you have JavaScript code from a browser or from a browser application that is usually contained within an isolated sandbox it is capable of breaking out of that sandbox and attacking the hosts this is the nature of the spectra and meltdown attacks so it does that by reaching outside of the sandbox and then if you have stuff going on in the rest of the machine you potentially grant code execution to untrusted code that's running through the browser or wherever its source is from so that's a bit of a concern but again this is under the attempt of being addressed with current patches to at least mitigate browser concerns via browser updates and then Microsoft has its own updates Linux has its kernel page table isolation so everyone's working on a solution as for potentially overlooked aspects of meltdown inspector GData noted that meltdown is quote surely the more dangerous one fortunately it is also the easier problem to fix spectre in my opinion will in variations be with us for a very long time but it'll probably lose importance with time as we figure out better approximations for a solution sigh Buress expressed disappointment and how little information has been revealed about low-level details for meltdown and noted a quote for example initially I postulated that presence in l1 would be required but now I am fairly confident that any memory location can be read the memory subsystem is rather opaque and it would be nice to have some insight this is probably irrelevant from a security perspective but it could help writing more efficient code they further stated with respect to Spector it became obvious that CPU architecture aka micro code is not being developed the same level of quality as the hardware at least at Intel what I find irritating is that enabling three new MSR capabilities related to branch prediction can trigger reboots ok I'm oversimplifying a bit they concluded with the public is left wondering about what is really going on which certainly does not increase my confidence in a solution so this is specifically referring to Intel's patches that they push recently where were some systems like Haswell and broad wealth systems were experiencing reboots or other instability in the environment and those patches at this point have been more or less recalled Intel has actually requested that their partners stop pushing them through official channels so that's where we stand within tiles patches for the time being as for wider reaching impact we queried GData about the extent to which AMD is affected because we know that intel is affected by meltdown we know that everyone's affected by Spectre but what is the impact of meltdown on AMD specifically so GData noted that AMD appears to be unaffected by meltdown and called meltdown the more severe of the two issues GData also noted that handy is affected by spectre and as our mobile devices arm and so forth but they further speculated that there likely isn't much difference in severity between Intel and AMD with regard to spectre specifically spectre isn't as bad as meltdown from what we've been told but can still fully expose higher privilege system memory and quote shouldn't be taken lightly as for responses to meltdown inspector coming to public the vendors had about two hundred days to respond to the attacks before researchers went public with their information thus far most vendors have shipped patches of various functional states some causing issues on specific CPUs but overall G data believes that the defenders did well in their reactions and emphasized the complexity of these decades-old issues some of the proposed changes constitute some of the biggest overall changes to operating systems in the last decade and they were pulled off in about six months which GData says is impressive and says I think there's room for improvement both for software fixes I'm actively researching this and hardware fixes in terms of security benefit and performance costs further stating as for intel's patches they certainly are not good solutions but it's probably all they can do it'll take time for CPUs to fully pass all these issues the only real solution is waiting for new architectures that all fully resolved things at least what we know of today and new things will arise as always CPUs take years to develop looking at long timelines 18 months for some of the recent CPUs and that gets even worse as you introduce new architectures or depart from existing architectures or rework existing designs to quote the GData group again they seem to think that the microcode updates we've seen from Intel are likely to be features meant for development being exposed to operating system vendors and used for something they were never meant to do so this is looking like at this point it's a matter of the hardware vendors and the OS or software vendors working together to try and piecemeal together a solution for the time being which is why you're gonna see some instability and some of those patches or some nebulous performance depend on what type of applications you use now speaking of performance we also asked the researchers about this and it seems the general belief is that performance impacts will not be significant or potentially even noticeable for the average consumer which seems reasonable but also noted that the high-end users which our interpretation here based on their response would be xoc type users anyone fighting for clocks and for higher scores to rank higher and competitive overclocking or anything like that those types of users may feel a bit of pain because they're obviously concerned that to some extent about security depending on what kind of environment you're in and then even more so about absolute speed so that's something to think about but in terms of overall performance penalty the biggest hit is going to come two specific types of the enterprise and data center applications which we're not covering today because that's not really our focus or core competency however from a consumer standpoint nothing to worry about too much then update all your browsers and everything else as soon as you can and then from a gaming standpoint you might lose some performance it just it's gonna depend on which CPU you're on and how it has been patched it's to deal with these issues so that's largely going to be a CPU dependent thin and it will depend a bit on the games as well so potentially some downside there though from what they're telling us but the big focus is on updating browsers so recap here there's no real evidence that these have been used in the wild to attack anyone yet and everyone's working together to try and fix it manufacturers have a lot more to worry about right now than consumers they're the ones you have to push out the patches you just download them obviously you want to get some that are stable though so you don't need to worry too much immediately but everyone is affected by specter Intel and the arm basically anything modern and Intel is primarily affected by meltdown with no real evidence that AMD has significant concerns regarding meltdown specifically software patches remain an effective stopgap solution until there are firm hardware solutions in the future looking a couple years out as we move to new architectures at which point the research to believe that this issue should be largely resolved for gamers and overclockers there might be a bit of a performance cost for the added security but Intel did have a 200 day grace period where they were alerted to meltdown and the public announcement came later along with some Paget so they're working on it but we'll see what happens as it's all ironed out what is embarrassing is that Intel has had a number of patches so far that I've proven just generally unstable if you have a has well abroad well system you might experience rebooting you should probably roll back your patches as for the answer of where are we now we're somewhere between the problem being revealed and a stable solution we're not quite there yet but we're also not left hopeless the progress being made is admirable thus far again these are major operating system overhauls in some cases especially with linux and they've been pulled off in a matter of six months as some of the biggest changes in a decade from what we've been to hold so that's a pretty massive achievement but there's still more work to be done so for consumers what you should do right now is update things as the updates roll out and if you experience instability contact whoever pushed the update and hopefully it'll get resolved with your reports but that's it for this one you can check the full article links in the description below if you want the full interviews without any truncation going on and as always subscribe for more you go to patreon.com/scishow sexist to help us out directly we're gonna store it gamers nexus net to pick of a shirt like this one I'll see you all next time
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.