On-demand time blurring to support side-channel defense

Side-channel attacks are a serious threat to multi-tenant public clouds. Past work showed how secret information in one virtual machine (VM) can be leaked to another, co-resident VM using timing side channels. Recent defenses against timing side channels focus on reducing the degree of resource shar...

Full description

Saved in:
Bibliographic Details
Main Authors: LIU, Weijie, GAO, Debin, REITER, Michael K.
Format: text
Language:English
Published: Institutional Knowledge at Singapore Management University 2017
Subjects:
Online Access:https://ink.library.smu.edu.sg/sis_research/4024
https://ink.library.smu.edu.sg/context/sis_research/article/5026/viewcontent/esorics17.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Singapore Management University
Language: English
Description
Summary:Side-channel attacks are a serious threat to multi-tenant public clouds. Past work showed how secret information in one virtual machine (VM) can be leaked to another, co-resident VM using timing side channels. Recent defenses against timing side channels focus on reducing the degree of resource sharing. However, such defenses necessarily limit the flexibility with which resources are shared. In this paper, we propose a technique that dynamically adjusts the granularity of platform time sources, to interfere with timing side-channel attacks. Our proposed technique supposes an interface by which a VM can request the temporary coarsening of platform time sources as seen by all VMs on the platform, which the hypervisor can effect since it virtualizes accesses to those timers. We show that the VM-Function (VMFUNC) mechanism provides a low-overhead such interface, thereby enabling applications to adjust timer granularity with minimal overhead. We present a proof-of-concept implementation using a Xen hypervisor running Linux-based VMs on a cloud server using commodity Intel processors and supporting adjustment of the timestamp-counter (TSC) granularity. We evaluate our implementation and show that our scheme mitigates timing side-channel attacks, while introducing negligible performance penalties.