BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:ASPLOS @ EPFL Special seminar bundle
DTSTART:20140218T131500
DTEND:20140218T160000
DTSTAMP:20260407T064529Z
UID:7a2c11fc1d6e0c0f4cce795f10836fbeff7736e3ecd4ca3389f62f47
CATEGORIES:Conferences - Seminars
DESCRIPTION:Radu Banabic\, Stefan Bucur\, Stanko Novakovic \nEPFL has thr
 ee papers this year at ASPLOS\, the premier interdisciplinary conference o
 n Architectural Support for Programming Languages and Operating Systems. 
  We will have a special seminar of back-to-back presentations at EPFL ahea
 d of the conference.\nThere will be a break between the talks\, so it is p
 ossible to join-in or leave in the middle.1:15 Radu Banabic\, George Cande
 a\, and Rachid Guerraoui. \nFinding Trojan Message Vulnerabilities in Dis
 tributed Systems.\nhttp://dslab.epfl.ch/pubs/Achilles.pdf\nTrojan messages
  are messages that seem correct to the re- ceiver but cannot be generated 
 by any correct sender. Such messages constitute major vulnerability points
  of a dis- tributed system—they constitute ideal targets for a malicious
  actor and facilitate failure propagation across nodes. We describe Achill
 es\, a tool that searches for Trojan messages in a distributed system. Ach
 illes uses dynamic white-box analysis on the distributed system binaries i
 n order to infer the predicate that defines messages parsed by receiver no
 des and generated by sender nodes\, respectively\, and then com- putes Tro
 jan messages as the difference between the two. \nWe apply Achilles on im
 plementations of real distributed systems: FSP\, a file transfer applicati
 on\, and PBFT\, a Byzantine-fault-tolerant state machine replication libra
 ry. Achilles discovered a new bug in FSP and rediscovered a previously kno
 wn vulnerability in PBFT. In our evaluation we demonstrate that our approa
 ch can perform orders of magnitude better than general approaches based on
  regular fuzzing and symbolic execution.2:15 Stefan Bucur\, Johannes Kinde
 r\, and George Candea.\nPrototyping Symbolic Execution Engines for Interpr
 eted Languages.\nhttp://infoscience.epfl.ch/record/195687\nSymbolic execut
 ion is being successfully used to automat- ically test statically compiled
  code [4\, 7\, 9\, 15]. However\, increasingly more systems and applicatio
 ns are written in dynamic interpreted languages like Python. Building a ne
 w symbolic execution engine is a monumental effort\, and so is keeping it 
 up-to-date as the target language evolves. Furthermore\, ambiguous languag
 e specifications lead to their implementation in a symbolic execution engi
 ne potentially differing from the production interpreter in subtle ways.\n
 We address these challenges by flipping the problem and using the interpre
 ter itself as a specification of the language semantics. We present a reci
 pe and tool (called CHEF) for turning a vanilla interpreter into a sound a
 nd complete sym- bolic execution engine. CHEF symbolically executes the ta
 rget program by symbolically executing the interpreter’s binary while ex
 ploiting inferred knowledge about the pro- gram’s high-level structure.\
 nUsing CHEF\, we developed a symbolic execution engine for Python in 5 per
 son-days and one for Lua in 3 person-days. They offer complete and faithfu
 l coverage of language features in a way that keeps up with future languag
 e versions at near-zero cost. CHEF-produced engines are up to 1000× more 
 performant than if directly executing the interpreter symbolically without
  CHEF.3:15 Stanko Novakovic\; Alexandros Daglis\; Edouard Bugnion\; Babak 
 Falsafi\; Boris Grot.\nScale-Out NUMA\nhttp://infoscience.epfl.ch/record/1
 95238\nEmerging datacenter applications operate on vast datasets that are 
 kept in DRAM to minimize latency. The large number of servers needed to ac
 commodate this massive memory footprint requires frequent server-to-server
  communication in applications such as key-value stores and graph-based ap
 plications that rely on large irregular data structures. The fine-grained 
 nature of the accesses is a poor match to commodity networking technologie
 s\, including RDMA\, which incur delays of 10-1000x over local DRAM operat
 ions. We introduce Scale-Out NUMA (soNUMA) – an architecture\, programmi
 ng model\, and communication protocol for low-latency\, distributed in-mem
 ory processing. soNUMA layers an RDMA-inspired programming model directly 
 on top of a NUMA memory fabric via a stateless messaging protocol. To faci
 litate interactions between the application\, OS\, and the fabric\, soNUMA
  relies on the remote memory controller – a new architecturally-exposed 
 hardware block integrated into the node’s local coherence hierarchy. Our
  results based on cycle-accurate full-system simulation show that soNUMA p
 erforms remote reads at latencies that are within 4x of local DRAM\, can f
 ully utilize the available memory bandwidth\, and can issue up to 10M remo
 te memory operations per second per core.
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR