Vulnerability Note VU#283803

Integrated GPUs may allow side-channel and rowhammer attacks using WebGL ("Glitch")

Original Release date: 03 May 2018 | Last revised: 03 May 2018

Overview

Some platforms with integrated GPUs, such as smartphones, may allow both side-channel and rowhammer attacks via WebGL, which may allow a remote attacker to compromise the browser on an affected platform. An attack technique that leverages these vulnerabilities is called "GLitch."

Description

An academic paper describes an attack called "GLitch," which leverages two different techniques to achieve a compromise of a web browser using WebGL. The attack is only feasible on platforms where the CPU and GPU share the same memory, such as a smartphone or similar device. The two components of the attack are:

  1. A Side-channel attack to determine physical memory layout
  2. A Rowhammer attack to flip the value of one or more bits in physical memory

The side-channel attack

The precise timing capabilities provided by WebGL can allow an attacker to determine the difference between cached DRAM accesses and uncached DRAM accesses. This can allow an attacker to determine contiguous areas of physical DRAM memory. Knowledge of contiguous memory regions are used in a number of microarchitectural attacks, such as rowhammer.

The rowhammer attack

The rowhammer attack targets the design of DRAM memory. On a system where the DRAM is insufficiently refreshed, targeted operations on a row of DRAM memory may be able to influence the memory values on neighboring rows. Protections against the rowhammer attack include the use of ECC DRAM, as well as increased refresh rates. The LPDDR4 mobile memory standard also has optional hardware support for target row refresh, which can mitigate the rowhammer attack.

Combining the attacks with WebGL

The GLitch attack leverages both a side-channel attack to determine contiguous memory, as well as rowhammer. With the knowledge of contiguous memory, an attacker may be able to determine relative physical addresses. This knowledge of relative physical addresses can let the attacker know what memory locations to target with the rowhammer attack. The use of WebGL with precise timers is important in the GLitch attack for these reasons:
  • Precise WebGL timers allow a side-channel to leak memory addresses.
  • GPU capabilities exposed via WebGL allow for fast double-sided DRAM access, enabling the rowhammer attack.
The impact of combining both the side-channel attack and rowhammer attack has been demonstrated to bypass the Firefox sandbox on the Android platform.

GLitch success rates in testing

It is important to realize that the GLitch attack has only successfully been demonstrated on the Nexus 5 phone, which was released in 2013. The Nexus 5 phone received its last software security update in October, 2015, and is therefore an already unsafe device to use. Several other phones released in 2013 were tested, but were not able to successfully be attacked with the GLitch attack. Success rates on phones newer than 2013 models were not provided. Non-Android devices were not tested as well.

Impact

Upon visiting a malicious or compromised website with a vulnerable device, an attacker may be able to bypass security features provided by the web browser.

Solution

Apply an update

Google Chrome and Mozilla Firefox have released updates which disable high precision timers in the browser.
Other browsers do not appear to be affected.

Vendor Information (Learn More)

VendorStatusDate NotifiedDate Updated
GoogleAffected16 Mar 201803 May 2018
MozillaAffected16 Mar 201803 May 2018
MicrosoftNot Affected16 Mar 201825 Apr 2018
AMDUnknown16 Mar 201816 Mar 2018
AppleUnknown16 Mar 201816 Mar 2018
ArmUnknown-26 Apr 2018
BlackBerryUnknown16 Mar 201816 Mar 2018
Brave SoftwareUnknown16 Mar 201816 Mar 2018
BroadcomUnknown16 Mar 201816 Mar 2018
IBM, INC.Unknown26 Apr 201826 Apr 2018
Imagination TechnologiesUnknown16 Mar 201816 Mar 2018
IntelUnknown16 Mar 201816 Mar 2018
NVIDIAUnknown16 Mar 201816 Mar 2018
OperaUnknown16 Mar 201816 Mar 2018
QUALCOMM IncorporatedUnknown16 Mar 201816 Mar 2018
If you are a vendor and your product is affected, let us know.View More »

CVSS Metrics (Learn More)

Group Score Vector
Base 4.0 AV:N/AC:H/Au:N/C:P/I:P/A:N
Temporal 3.6 E:F/RL:W/RC:C
Environmental 2.7 CDP:ND/TD:M/CR:ND/IR:ND/AR:ND

References

Credit

This issue was reported by Pietro Frigo, Cristiano Giuffrida, Herbert Bos, and Kaveh Razavi of the Vrije Universiteit Amsterdam.

This document was written by Will Dormann and Trent Novelly.

Other Information

  • CVE IDs: CVE-2018-10229
  • Date Public: 03 May 2018
  • Date First Published: 03 May 2018
  • Date Last Updated: 03 May 2018
  • Document Revision: 45

Feedback

If you have feedback, comments, or additional information about this vulnerability, please send us email.