New LT1 for 2014 6.2l alum block
11-10-2012, 10:42 PM
#221
Teching In
Join Date: Aug 2010
Posts: 10
Likes: 0
Received 0 Likes
on
0 Posts
In order to control the various electro-mechanical devices, you need hard real-time support in the operating system (aka - RTOS), like VxWorks or RtLinux.
If you are running a hypervisor (i.e. virtualization), there are very few hard real-time enabled solutions out and available. To be sure, this will be a type 1 hypervisor (i.e. it runs right on the metal).
Given these two constraints, you would probably running something like Wind River (http://www.windriver.com/products/hypervisor/), Green Hills (http://www.ghs.com/) or LynuxWorks (http://www.lynuxworks.com/).
Now comes the part that does not add up. If I am hosting multiple VMs on the same physical box, I would allow them to communicate via a virtualized network socket. This virtualized networks socket is typically nothing more than a bit of shared memory that the hypervisor owns and is very fast.
Within the box, employing encryption between the VMs is pointless - shared memory is only accessible to the VMs that the hypervisor allows access to. As well, the processor only has a finite amount of processing power (tempered by the SWAP-C trade). Encryption and other PKI uses precious CPU cycles to protect something that is otherwise protected.
So, either you are wasting CPU cycles with the encryption (thus driving up the cost, size, weight, power) and more than likely would need to add a liquid cooling loop to the processor or you are only using encryption that is going off-box and that encryption is not used all the time.
OK, so lets say that the encryption is on the edge, calling back to GM or when an external device is connected. The problem is that I can change the non-volatile persistence and effectively "own" the box.
The XBox360 is a prime example of this. Microsoft put in very sophisticated PKI protecting the application, but at the end of the day, once the box was opened, a way was found around the system (ref: http://www.ubergizmo.com/2011/09/xbo...nently-hacked/). The challenge this has is that this is an embedded real-time system, with real-world bad stuff happening when timing is off, where the XBox is just a gaming system and if it needed to it could shed frames per second to keep the game alive.
But wait, you say. We put on a Trusted Platform Module (TPM) on all of our boards and ensure a secure boot (see: http://www.intel.com/technology/adva...omm/322287.pdf). The problem with this is that a smart person could pull off the different VMs, load those instances on a changed hypervisor (i.e. one that ignores secure boot) and mod the box until the cows come home. In the process, it will be pretty easy to disable / defeat the "phone home" back to GM.
As well, keep in mind the capabilities of very leading edge of processors and constraints on the various embedded systems (like start-up timeline). Starting a hypervisor takes time and most want to simply jump into their car, twist the key and go. You only have a few microseconds and running a VM takes a whole bunch more than that given today's leading edge processors. Thus, I have my doubts any automotive manufacturer is running VMs in something controlling the fuel delivery, transmission and other embedded hard real-time systems.
The other point is that, with any system there are multiple attack vectors that can be employed. For example, did GM secure it's connection to the TPMS sensors? See: http://arstechnica.com/security/2010...-tyre-sensors/.
Bottom line is that nothing is 100% secure and given time / money any system can be cracked.
11-11-2012, 04:11 AM
#224
So why is it that every other GM vehicle with DI can be tuned but you guys insist on not being able to tune the lt1 because of safety? How is it any more dangerous then the rest?
11-11-2012, 07:04 PM
#225
TECH Resident
iTrader: (1)
Join Date: Jul 2012
Location: Miami, FL
Posts: 965
Likes: 0
Received 0 Likes
on
0 Posts
Can't believe I just went through all of this. I agree with 01ssreda4, I literally feel dumber after reading big gunz posts and trying to comprehend how one can be such a ******* tool. It's even more mind blowing that theres still a few people actually taking him seriously...
11-12-2012, 09:55 PM
#228
This is what they said about the 2012+ LML Duramax. It's already been done and they are tuned.
To say an ECM can't be cracked and figured out is flat out stupid, where there's a will there's a way. Especially when companies will make money on it.
Edit: To add from reading your prior posts about "Safety" because of fuel pressure. Do you know what duramax diesels run for fuel rail pressure? About 10-15 TIMES more than this motor. We've been tuning duramaxs since 2001.
My biggest tune holds about 22000 PSI rail pressure at wide open throttle.
Also another thing, why would this vehicle be in "contact" with GM? Is that only when its in for service, or all the time the car has a connection to GM? I think that'd be the first thing I'd remove just like the onstar system.
To say an ECM can't be cracked and figured out is flat out stupid, where there's a will there's a way. Especially when companies will make money on it.
Edit: To add from reading your prior posts about "Safety" because of fuel pressure. Do you know what duramax diesels run for fuel rail pressure? About 10-15 TIMES more than this motor. We've been tuning duramaxs since 2001.
My biggest tune holds about 22000 PSI rail pressure at wide open throttle.
Also another thing, why would this vehicle be in "contact" with GM? Is that only when its in for service, or all the time the car has a connection to GM? I think that'd be the first thing I'd remove just like the onstar system.
But you're right. They also said this back when the bosch PCM first showed up in the 06 Duramax.
Mr Gunnz seems to have avoided my comments, and I get that...
I wonder how long until we can get our hands on one of these magic boxes
11-20-2012, 11:06 AM
#230
TECH Fanatic
iTrader: (2)
Join Date: Jan 2006
Location: Indianapolis Indiana
Posts: 1,299
Likes: 0
Received 0 Likes
on
0 Posts
To be fair, I dont believe that any of the tuners available for the new dmax pcm actually read the stock file out...yet
But you're right. They also said this back when the bosch PCM first showed up in the 06 Duramax.
Mr Gunnz seems to have avoided my comments, and I get that...
I wonder how long until we can get our hands on one of these magic boxes
But you're right. They also said this back when the bosch PCM first showed up in the 06 Duramax.
Mr Gunnz seems to have avoided my comments, and I get that...
I wonder how long until we can get our hands on one of these magic boxes
11-21-2012, 03:58 PM
#231
TECH Fanatic
Join Date: Feb 2009
Location: Saudi Arabia
Posts: 1,020
Likes: 0
Received 0 Likes
on
0 Posts
GM wants to dominate the performance market , thats all
buy our cars , parts , tuning products .
give it a year or two , EFI live/HPtuners will crack that ECM
its not about safty , its all about the money.
buy our cars , parts , tuning products .
give it a year or two , EFI live/HPtuners will crack that ECM
its not about safty , its all about the money.
11-21-2012, 05:03 PM
#232
11 Second Club
iTrader: (13)
Join Date: Oct 2003
Location: apoopka, fl
Posts: 1,414
Likes: 0
Received 0 Likes
on
0 Posts
This is simple, look at the JB4 for my direct injection turbo BMW. Basically a computer inline with the stocker. The ecu puts out x, the jb4 converts that to y, and before it gets back to the ecu its converted back to x.
Not stressing tuning one bit.
AND if worse comes to worse, I have a carburetor and a mother f%&^*ng welder to make an intake mani with.
Not stressing tuning one bit.
AND if worse comes to worse, I have a carburetor and a mother f%&^*ng welder to make an intake mani with.
11-21-2012, 09:47 PM
#235
A bit of thinking here, but this is an embedded system. The constraints that are put on the embedded system is the trade between capability and size / weight / power / cost (SWAP-C).
In order to control the various electro-mechanical devices, you need hard real-time support in the operating system (aka - RTOS), like VxWorks or RtLinux.
If you are running a hypervisor (i.e. virtualization), there are very few hard real-time enabled solutions out and available. To be sure, this will be a type 1 hypervisor (i.e. it runs right on the metal).
Given these two constraints, you would probably running something like Wind River (http://www.windriver.com/products/hypervisor/), Green Hills (http://www.ghs.com/) or LynuxWorks (http://www.lynuxworks.com/).
Now comes the part that does not add up. If I am hosting multiple VMs on the same physical box, I would allow them to communicate via a virtualized network socket. This virtualized networks socket is typically nothing more than a bit of shared memory that the hypervisor owns and is very fast.
Within the box, employing encryption between the VMs is pointless - shared memory is only accessible to the VMs that the hypervisor allows access to. As well, the processor only has a finite amount of processing power (tempered by the SWAP-C trade). Encryption and other PKI uses precious CPU cycles to protect something that is otherwise protected.
So, either you are wasting CPU cycles with the encryption (thus driving up the cost, size, weight, power) and more than likely would need to add a liquid cooling loop to the processor or you are only using encryption that is going off-box and that encryption is not used all the time.
OK, so lets say that the encryption is on the edge, calling back to GM or when an external device is connected. The problem is that I can change the non-volatile persistence and effectively "own" the box.
The XBox360 is a prime example of this. Microsoft put in very sophisticated PKI protecting the application, but at the end of the day, once the box was opened, a way was found around the system (ref: http://www.ubergizmo.com/2011/09/xbo...nently-hacked/). The challenge this has is that this is an embedded real-time system, with real-world bad stuff happening when timing is off, where the XBox is just a gaming system and if it needed to it could shed frames per second to keep the game alive.
But wait, you say. We put on a Trusted Platform Module (TPM) on all of our boards and ensure a secure boot (see: http://www.intel.com/technology/adva...omm/322287.pdf). The problem with this is that a smart person could pull off the different VMs, load those instances on a changed hypervisor (i.e. one that ignores secure boot) and mod the box until the cows come home. In the process, it will be pretty easy to disable / defeat the "phone home" back to GM.
As well, keep in mind the capabilities of very leading edge of processors and constraints on the various embedded systems (like start-up timeline). Starting a hypervisor takes time and most want to simply jump into their car, twist the key and go. You only have a few microseconds and running a VM takes a whole bunch more than that given today's leading edge processors. Thus, I have my doubts any automotive manufacturer is running VMs in something controlling the fuel delivery, transmission and other embedded hard real-time systems.
The other point is that, with any system there are multiple attack vectors that can be employed. For example, did GM secure it's connection to the TPMS sensors? See: http://arstechnica.com/security/2010...-tyre-sensors/.
Bottom line is that nothing is 100% secure and given time / money any system can be cracked.
In order to control the various electro-mechanical devices, you need hard real-time support in the operating system (aka - RTOS), like VxWorks or RtLinux.
If you are running a hypervisor (i.e. virtualization), there are very few hard real-time enabled solutions out and available. To be sure, this will be a type 1 hypervisor (i.e. it runs right on the metal).
Given these two constraints, you would probably running something like Wind River (http://www.windriver.com/products/hypervisor/), Green Hills (http://www.ghs.com/) or LynuxWorks (http://www.lynuxworks.com/).
Now comes the part that does not add up. If I am hosting multiple VMs on the same physical box, I would allow them to communicate via a virtualized network socket. This virtualized networks socket is typically nothing more than a bit of shared memory that the hypervisor owns and is very fast.
Within the box, employing encryption between the VMs is pointless - shared memory is only accessible to the VMs that the hypervisor allows access to. As well, the processor only has a finite amount of processing power (tempered by the SWAP-C trade). Encryption and other PKI uses precious CPU cycles to protect something that is otherwise protected.
So, either you are wasting CPU cycles with the encryption (thus driving up the cost, size, weight, power) and more than likely would need to add a liquid cooling loop to the processor or you are only using encryption that is going off-box and that encryption is not used all the time.
OK, so lets say that the encryption is on the edge, calling back to GM or when an external device is connected. The problem is that I can change the non-volatile persistence and effectively "own" the box.
The XBox360 is a prime example of this. Microsoft put in very sophisticated PKI protecting the application, but at the end of the day, once the box was opened, a way was found around the system (ref: http://www.ubergizmo.com/2011/09/xbo...nently-hacked/). The challenge this has is that this is an embedded real-time system, with real-world bad stuff happening when timing is off, where the XBox is just a gaming system and if it needed to it could shed frames per second to keep the game alive.
But wait, you say. We put on a Trusted Platform Module (TPM) on all of our boards and ensure a secure boot (see: http://www.intel.com/technology/adva...omm/322287.pdf). The problem with this is that a smart person could pull off the different VMs, load those instances on a changed hypervisor (i.e. one that ignores secure boot) and mod the box until the cows come home. In the process, it will be pretty easy to disable / defeat the "phone home" back to GM.
As well, keep in mind the capabilities of very leading edge of processors and constraints on the various embedded systems (like start-up timeline). Starting a hypervisor takes time and most want to simply jump into their car, twist the key and go. You only have a few microseconds and running a VM takes a whole bunch more than that given today's leading edge processors. Thus, I have my doubts any automotive manufacturer is running VMs in something controlling the fuel delivery, transmission and other embedded hard real-time systems.
The other point is that, with any system there are multiple attack vectors that can be employed. For example, did GM secure it's connection to the TPMS sensors? See: http://arstechnica.com/security/2010...-tyre-sensors/.
Bottom line is that nothing is 100% secure and given time / money any system can be cracked.
Yup. Youre just ignorant to really think otherwise.
11-21-2012, 11:47 PM
#236
TECH Addict
iTrader: (2)
Being that you are familiar with TLS with your computer background. Then you would know the ECM has been converted into a inactive virtual client regarding the lack of the digital encrypted keys. The ECM VMs on a single hardware platform without additional network and server resources. There will be no spoofing or packet sniffing of any sort on a TLS connection with the virtual client Renders the fuel time inactive...... the assurance is in place along with many other fail safes. TAMPER PROOF.. This is the interest of consumer safety.
Bigg Gunz
Bigg Gunz
safety MY ***... its about money. always has, always will.
Tuning has come a long way in the past decade for LS vehicles and GM has caught on that theres a HUGE market for it and they want a slice. If they can prevent people form cracking into the computer, thats more money in GMs pocket pure and simple. Safetys just the side effect since theres a rediculous amount of PSI now for what ever reason.
Dont get me wrong im all for moving forward (although calling it an LT1 seems like a step back) and the scinece behind this new engine is amazing and sweet. can wait for some of its technical benefits to trickle down to the LS motors.
but harping on the safety thing just makes it sound like "big Brother" cant trust us so is putting training wheels on us... its actually quite offensive as a car enthusiast
it money, pure and simple.
11-23-2012, 11:57 PM
#240