Difference between revisions of "SHA-1 Image Segment Overflow"

From The iPhone Wiki
Jump to: navigation, search
m (Compatibility: Consistency please.)
m
 
(5 intermediate revisions by 4 users not shown)
Line 1: Line 1:
'''SHA-1 Image Segment Overflow''' or '''SHAtter''' was an exploit that allowed unsigned code execution from a flaw in the bootrom. It was never used in a public jailbreak due to the exploit used in [[limera1n]] being released first. SHAtter was patched in the [[S5L8940|A5]] devices and therefore, never released.
+
'''SHA-1 Image Segment Overflow''' or '''SHAtter''' was an exploit that allowed unsigned code execution from a flaw in the [[bootrom]]. It was never used in a public jailbreak because the [[limera1n]] exploit was released first, and more devices were vulnerable to it. SHAtter was patched in the [[S5L8940|A5]] bootrom and therefore, never officially released.
  +
  +
An implementation of SHAtter is available in [[ipwndfu]].
   
 
== Compatibility ==
 
== Compatibility ==
 
SHAtter only works with [[S5L8930|A4]] devices:
 
SHAtter only works with [[S5L8930|A4]] devices:
* [[k48ap|iPad]]
+
* [[K48AP|iPad]]
 
* [[iPhone 4]] (all models)
 
* [[iPhone 4]] (all models)
* [[n81ap|iPod touch 4G]]
+
* [[N81AP|iPod touch (4th generation)]]
* [[k66ap|Apple TV 2G]]
+
* [[K66AP|Apple TV (2nd generation)]]
   
 
== Credit ==
 
== Credit ==
* '''vulnerability''': [[User:posixninja|posixninja]] (7 May 2010), also discovered independently by [[User:geohot|geohot]]
+
* '''vulnerability''': [[User:posixninja|posixninja]] ({{date|2010|05}}), also discovered independently by [[User:geohot|geohot]]
 
* '''research''': [[User:posixninja|posixninja]], [[User:pod2g|pod2g]], also [[User:MuscleNerd|MuscleNerd]]
 
* '''research''': [[User:posixninja|posixninja]], [[User:pod2g|pod2g]], also [[User:MuscleNerd|MuscleNerd]]
 
* '''exploit''': [[User:Pod2g|pod2g]]
 
* '''exploit''': [[User:Pod2g|pod2g]]
   
 
==Background Info==
 
==Background Info==
In April 2010 [[User:pod2g|pod2g]] wrote a USB [[wikipedia:Fuzz testing|fuzzer]] and tested every single [[USB control messages|USB control message]] possible on his [[N72ap|iPod touch 2G]].
+
In {{date|2010|04}} [[User:pod2g|pod2g]] wrote a USB [[wikipedia:Fuzz testing|fuzzer]] (see also: [[fuzzing]]) and tested every single [[USB control messages|USB control message]] possible on his [[N72AP|iPod touch (2nd generation)]].
 
The [[wikipedia:Fuzz testing|fuzzer]] found 2 vulnerabilities:
 
The [[wikipedia:Fuzz testing|fuzzer]] found 2 vulnerabilities:
 
* a heap overflow caused by [[usb_control_msg(0xA1, 1) Exploit‎|usb_control_msg(0xA1, 1)]]
 
* a heap overflow caused by [[usb_control_msg(0xA1, 1) Exploit‎|usb_control_msg(0xA1, 1)]]
 
* a way to dump the bootrom using USB descriptors request
 
* a way to dump the bootrom using USB descriptors request
   
The team tested these two vulnerabilities on newer devices ([[N88ap|iPhone 3GS]], [[N18ap|iPod touch 3G]], [[K48ap|iPad]]) and both were already fixed by Apple.
+
The team tested these two vulnerabilities on newer devices ([[N88AP|iPhone 3GS]], [[N18AP|iPod touch (3rd generation)]], [[K48AP|iPad]]) and both were already fixed by Apple.
 
[[User:posixninja|posixninja]] continued the [[wikipedia:Fuzz testing|fuzzing]] on these devices and found that with a particular sequence of [[USB control messages|USB messages]] it was possible to dump the [[wikipedia:.bss|BSS]]+Heap+Stack (on new gens only).
 
Having a memory dump is really helpful to make exploits and it was also the first time we had this kind of dump. (Previous bootrom exploits like the [[0x24000 Segment Overflow]] were done blind!)
 
   
  +
[[User:posixninja|posixninja]] continued [[wikipedia:Fuzz testing|fuzzing]] on these devices, and found that with a particular sequence of [[USB control messages]], it was possible for the first time to dump the [[wikipedia:.bss|BSS]]+Heap+Stack (on newer devices only).
Also, his first attempts to dump the memory resulted in rebooting the device. Interesting! We'll see after that this reboot is the base of the [[SHA-1 Image Segment Overflow|SHAtter]] exploit.
 
   
Research began to figure out why the device would reboot. [[User:posixninja|posixninja]] found the reason and proposed different ideas to exploit this. He also reversed tons of assembly code of the bootrom in this period, giving a support discussion to the team. We're not talking about days, but months of work. So, major props to [[User:posixninja|posixninja]]: [[SHA-1 Image Segment Overflow|SHAtter]] would not have been possible without the clever vulnerability he found and the research he did on the bootrom.
+
Interestingly, his first attempts to dump the memory resulted in rebooting the device. Research began to figure out why the device would reboot. [[User:posixninja|posixninja]] found the reason and proposed different ideas to exploit this. He also reversed tons of assembly code of the bootrom in this period, giving a support discussion to the team. This took months of work.
   
In the meanwhile, [[User:pod2g|pod2g]] helped on the USB reversing side and found a way to have more control over the size of the USB packets sent. The finer-grained control of the packet sizes is the key of [[SHA-1 Image Segment Overflow|SHAtter]].
+
In the meanwhile, [[User:pod2g|pod2g]] helped on the USB reversing side and found a way to have more control over the size of the USB packets sent. The finer-grained control of the packet sizes is the key of SHAtter.
   
 
[[User:posixninja|posixninja]] and [[User:pod2g|pod2g]] worked on exploiting the vulnerability for days. Every attempt was a failure because the idea to attack the stack and bypass the [[IMG3 File Format|IMG3]] control routines was just impossible. It took them weeks to understand why they failed and why they couldn't exploit it this way.
 
[[User:posixninja|posixninja]] and [[User:pod2g|pod2g]] worked on exploiting the vulnerability for days. Every attempt was a failure because the idea to attack the stack and bypass the [[IMG3 File Format|IMG3]] control routines was just impossible. It took them weeks to understand why they failed and why they couldn't exploit it this way.
   
They both gave up in July and focused on other subjects.
+
They both gave up in {{date|2010|07}} and focused on other subjects.
   
 
== Vulnerability ==
 
== Vulnerability ==
 
Explanation by [[p0sixninja]] at [[MyGreatFest]]:
 
Explanation by [[p0sixninja]] at [[MyGreatFest]]:
   
It tricked the bootrom to think the size of the image uploading was larger then what it actually was. Then when it would try to load the image, it would see that it was wrong. Then it would try to wipe out the entire image with all zeros and go past it and start wiping out bootrom.
+
:It tricked the bootrom to think the size of the image uploading was larger then what it actually was. Then when it would try to load the image, it would see that it was wrong. Then it would try to wipe out the entire image with all zeroes and go past it and start wiping out bootrom.
   
Exploitation was done by overwriting SHA-1 registers to zeros so then when it went to check images it would copy part of image into memory address zero (where the bootrom is). It would take the image uploaded and copy it over top of the bootrom (which turns out to be writable over the data portion).
+
:Exploitation was done by overwriting SHA-1 registers to zeroes so then when it went to check images it would copy part of image into memory address zero (where the bootrom is). It would take the image uploaded and copy it over top of the bootrom (which turns out to be writable over the data portion).
   
 
== References ==
 
== References ==

Latest revision as of 15:14, 25 October 2021

SHA-1 Image Segment Overflow or SHAtter was an exploit that allowed unsigned code execution from a flaw in the bootrom. It was never used in a public jailbreak because the limera1n exploit was released first, and more devices were vulnerable to it. SHAtter was patched in the A5 bootrom and therefore, never officially released.

An implementation of SHAtter is available in ipwndfu.

Compatibility

SHAtter only works with A4 devices:

Credit

Background Info

In April 2010 pod2g wrote a USB fuzzer (see also: fuzzing) and tested every single USB control message possible on his iPod touch (2nd generation). The fuzzer found 2 vulnerabilities:

The team tested these two vulnerabilities on newer devices (iPhone 3GS, iPod touch (3rd generation), iPad) and both were already fixed by Apple.

posixninja continued fuzzing on these devices, and found that with a particular sequence of USB control messages, it was possible for the first time to dump the BSS+Heap+Stack (on newer devices only).

Interestingly, his first attempts to dump the memory resulted in rebooting the device. Research began to figure out why the device would reboot. posixninja found the reason and proposed different ideas to exploit this. He also reversed tons of assembly code of the bootrom in this period, giving a support discussion to the team. This took months of work.

In the meanwhile, pod2g helped on the USB reversing side and found a way to have more control over the size of the USB packets sent. The finer-grained control of the packet sizes is the key of SHAtter.

posixninja and pod2g worked on exploiting the vulnerability for days. Every attempt was a failure because the idea to attack the stack and bypass the IMG3 control routines was just impossible. It took them weeks to understand why they failed and why they couldn't exploit it this way.

They both gave up in July 2010 and focused on other subjects.

Vulnerability

Explanation by p0sixninja at MyGreatFest:

It tricked the bootrom to think the size of the image uploading was larger then what it actually was. Then when it would try to load the image, it would see that it was wrong. Then it would try to wipe out the entire image with all zeroes and go past it and start wiping out bootrom.
Exploitation was done by overwriting SHA-1 registers to zeroes so then when it went to check images it would copy part of image into memory address zero (where the bootrom is). It would take the image uploaded and copy it over top of the bootrom (which turns out to be writable over the data portion).

References