Researchers have revealed a new technique that might be used to bypass existing hardware mitigations in modern processors from Intel, AMD, and Arm CPUs and stage speculative execution attacks like Spektre to expose sensitive data from host memory.
Spectre attacks are aimed to disrupt the isolation between different applications by using an optimization technique known as speculative execution in CPU hardware implementations to mislead programmes into accessing arbitrary memory regions and leaking their secrets.
While chipmakers have included software and hardware defences such as Retpoline and safeguards such as Enhanced Indirect Branch Restricted Speculation (eIBRS) and Arm CSV2, the latest technique demonstrated by VUSec researchers seek to circumvent all of these measures.
Branch History Injection (BHI or Spectre-BHB) is a new variant of Spectre-V2 attacks (tracked as CVE-2017-5715) that circumvent both eIBRS and CSV2, according to the researchers, and exposes arbitrary kernel memory on modern Intel CPUs.
"The hardware mitigations do prevent the unprivileged attacker from injecting predictor entries for the kernel," the researchers explained,
"However, the predictor relies on a global history to select the target entries to speculatively execute. And the attacker can poison this history from userland to force the kernel to mispredict to more 'interesting' kernel targets (i.e., gadgets) that leak data," the Systems and Network Security Group at Vrije Universiteit Amsterdam added.
To put it another way, malicious code can use the CPU Branch History Buffer (BHBshared )'s branch history to affect mispredicted branches within the victim's hardware context, leading to speculative execution that can subsequently be used to infer information that would otherwise be inaccessible.
All Intel and Arm processors that were previously vulnerable to Spectre-V2, as well as a number of AMD chipsets, are now vulnerable to Spectre-BHB, forcing the three firms to release software upgrades to address the problem.
Customers should also disable the unprivileged extended Berkeley Packet Filters (eBPF) in Linux, enable both eIBRS and Supervisor-Mode Execution Prevention (SMEP), and apply LFENCE to particularly identified gadgets that are discovered to be susceptible, according to Intel.
The researchers stated, "The [Intel eIBRS and Arm CSV2] mitigations work as intended, but the residual attack surface is much more significant than vendors originally assumed. Nevertheless, finding exploitable gadgets is harder than before since the attacker can't directly inject predictor targets across privilege boundaries. That is, the kernel won't speculatively jump to arbitrary attacker-provided targets, but will only speculatively execute valid code snippets it already executed in the past."
