Microsoft Technology Licensing, LLC

Massachusetts Institute of Technology

Clock Synchronization

Miguel Castro

Manuel Costa

Richendra Khanna

Antony Rowstron

Dushyanth Narayanan

Aleksandar Dragojevic

Anders Tungeland Gjerdrum

Alexander Shamis

Stanko Novakovic

Matthew Renzelmann

Jon Crowcroft

Georgios Chatzopoulos

Marcus Peinado

Jean-Philippe Martin

Periklis Akritidis

Christopher Lee Mullins

Lintao Zhang

Richard John Black

Lidong Zhou

Antony Ian Taylor Rowstron

Austin N Donnelly

Robert Standefer, Iii

Barbara Liskov

Paul Barham

Tim Harris

Matthew Caesar

In various examples, there is provided methods performed by nodes in a cluster of nodes to establish a master clock at a new master node following a reconfiguration of the nodes included in the cluster, whereby the master clock is provided by an old master node prior to the reconfiguration, and synchronize a local clock of slave nodes to clock of the new master node. The new master node sends a message to the slave nodes instructing them to disable their respective local clocks, receives acknowledgements that the local clocks have been disabled, waits until a time at which all leases have expired for any nodes removed from the cluster, sets the value of its clock to be greater than a maximum value that could have been provided by the old master node at the time the leases expired and indicates to the other nodes to re-enable their local clocks.

Clock synchronization

In various examples, there is provided methods performed by nodes in a cluster of nodes for performing transactions comprising one or more read operations and/or one or more write operations. The node comprises a local clock which is synchronized with a master clock and maintains a measure of uncertainty indicating current minimum and maximum values of the master clock. The method to perform transactions involving read operations generates a read timestamp representing a point in time which is earlier than a current minimum value of the master clock. The method then reads the objects and determines, for each of them, whether a timestamp associated with that object is later than the read timestamp. If so, an error handling procedure is performed for that object.

Performing transactions in distributed transactional memory systems

Technologies are described herein for document collaboration discovery. A collaboration system enables users to collaboratively author documents. The collaboration system receives edits to a document in real or near real time, and indexes the edits in a search index. The collaboration system can also receive and index metadata associated with the document. The collaboration system can also receive a search query from a user and perform a search of the search index. If the document is identified by the search, the user can request to be admitted as an active editor of the document. The user can also request to join a real-time messaging session with other active editors of the document. The active editors can be notified of the search terms that led the user to the document, and indicate whether the user is to be admitted to the document as an active editor or the real-time messaging session.

Document collaboration discovery

The majority of such software attacks exploit software vulnerabilities or flaws to write data to unintended locations. For example, control-data attacks exploit buffer overflows or other vulnerabilities to overwrite a return address in the stack, a function pointer, or some other piece of control data. Non-control-data attacks exploit similar vulnerabilities to overwrite security critical data without subverting the intended control flow in the program. We describe a method for securing software against both control-data and non-control-data attacks. A static analysis is carried out to determine data flow information for a software program. Data-flow tracking instructions are formed in order to track data flow during execution or emulation of that software. Also, checking instructions are formed to check the tracked data flow against the static analysis results and thereby identify potential attacks or errors. Optional optimisations are described to reduce the resulting additional overheads.

Securing software by enforcing data flow integrity

Methods of detecting memory errors using write integrity testing are described. In an embodiment, additional analysis is performed when a program is compiled. This analysis identifies a set of objects which can be written by each instruction in the program. Additional code is then inserted into the program so that, at runtime, the program checks before performing a write instruction that the particular object being written is one of the set of objects that it is allowed to write. The inserted code causes an exception to be raised if this check fails and allows the write to proceed if the check is successful. In a further embodiment, code may also be inserted to perform checks before indirect control-flow transfer instructions, to ensure that those instructions cannot transfer control to locations different from those intended.

Detecting memory errors using write integrity testing

Software fault isolation methods using byte-granularity memory protection are described. In an embodiment, untrusted drivers or other extensions to a software system are run in a separate domain from the host portion of the software system, but share the same address space as the host portion. Calls between domains are mediated using an interposition library and access control data is maintained for substantially each byte of relevant virtual address space. Instrumentation added to the untrusted extension at compile-time, before load-time, or at runtime and added by the interposition library enforces the isolation between domains, for example by adding access right checks before any writes or indirect calls and by redirecting function calls to call wrappers in the interposition library. The instrumentation also updates the access control data to grant and revoke access rights on a fine granularity according to the semantics of the operation being invoked.

Software fault isolation using byte-granularity memory protection

Methods and architectures for automatic filter generation are described. In an embodiment, these filters are generated in order to block inputs which would otherwise disrupt the normal functioning of a program. An initial set of filter conditions is generated by analyzing the path of a program from a point at which a bad input is received to the point at which the malfunctioning of the program is detected and creating conditions on an input which ensure that this path is followed. Having generated the initial set of filter conditions, the set is made less specific by determining which instructions do not influence whether the point of detection of the attack is reached and removing the filter conditions which correspond to these instructions.

Automatic filter generation and generalization

Methods and apparatus for generating error reports with enhanced privacy are described. In an embodiment the error is triggered by an input to a software program. An error report is generated by identifying conditions on an input to the program which ensure that, for any input which satisfies the conditions, the software program will follow the same execution path such that the error can be reproduced. The error report may include these conditions or may include a new input generated using the conditions.

Privacy enhanced error reports

To reduce the dependency of overlay networks on underlay networks to route messages, a virtual ring routing architecture may be formed that leverages the design of the overlay network to achieve their desirable scaling and robustness properties but also reduce the dependency on any underlay network to setup and maintain connectivity. More particularly, each node may have a single, fixed, location independent node identifier, to organize the nodes into a virtual ring. The connectivity between nodes through the actual network topology may be formed by a plurality of nodes in the virtual ring by maintaining connectivity to those nodes identified as virtual neighbor nodes within the virtual ring. The path segments defining communication connections between virtual neighbor nodes may be used to route messages between any pair of nodes in the network and may reduce route discovery overhead, reduce delay in transmission, and reduce or eliminate flooding to setup or maintain the path segments.

Network routing

A containment system may include a protection system which may protect the computing device from future attacks. For example, a patch may be automatically generated which resolves a detected vulnerability in a program. IN another example, a filter may be automatically generated which filters actions and/or messages which take advantage of a detected vulnerability in a program.

Filter generation

A containment system may include generating and/or sending an alert as the basis for safely sharing knowledge about detected worms. An alert may contain information that proves that a given program has a vulnerability. The alert may be self-certifying such that its authenticity may be independently verified by a computing system.

Self-certifying alert

One aspect of the invention is a vulnerability detection mechanism that can detect a large class of attacks through dynamic dataflow analysis. Another aspect of the invention includes self-certifying alerts as the basis for safely sharing knowledge about worms. Another aspect of the invention is a resilient and self-organizing protocol to propagate alerts to all non-infected nodes in a timely fashion, even when under active attack during a worm outbreak. Another aspect of the invention is a system architecture that enables a large number of mutually untrusting computers to collaborate in the task of stopping a previously unknown worm, even when the worm is spreading rapidly and exploiting unknown vulnerabilities in popular software packages.

Containment of worms

A new approach for asynchronous state-machine replication in a fault-tolerant system offers both integrity and high availability in the presence of Byzantine faults. The approach also improves the security of previous systems by recovering replicas proactively without necessarily identifying that they have failed or been attacked. This proactive recovery limits the time extent of a particular fault by regularly recovering replicas. In this way, the system works correctly even when all the replicas fail multiple times over the lifetime of the system, provided that less than &frac13; of the replicas are all faulty within a window of vulnerability. The approach also features an efficient implementation of message authentication that prevents an attacker from impersonating a replicated node that was faulty after that node recovers

Growing community of inventors

Cambridge, United Kingdom

Miguel Castro