Internet Draft James Pinkerton
draft-ietf-rddp-security-04.txtdraft-ietf-rddp-security-05.txt Microsoft Corporation Category: Standards Track Ellen Deleganes Expires: February, 2005 Intel Corporation Sara Bitan Microsoft Corporation August 2004 DDP/RDMAP Security 1 Status of this Memo This document is an Internet-Draft and is in full conformance with all provisions of Section 10 of RFC2026. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. 2 Abstract This document analyzes security issues around implementation and use of the Direct Data Placement Protocol(DDP) and Remote Direct Memory Access Protocol (RDMAP). It first defines an architectural model for an RDMA Network Interface Card (RNIC), which can implement DDP or RDMAP and DDP. The document reviews various attacks against the resources defined in the architectural model and the countermeasures that can be used to protect the system. Attacks are grouped into spoofing, tampering, information disclosure, denial of service, and elevation of privilege. Finally, the document concludes with a summary of security services for DDP and RDMAP, such as IPsec. J. Pinkerton, et al. Expires February 2005 1 Table of Contents 1 Status of this Memo.........................................1 2 Abstract....................................................1 2.1 Revision History............................................3 2.1.1 Changes from -02-04 to -03-05 version............................3 2.1.2 Changes from -03 to -04 version............................4 2.1.3 Changes from -02 to -03 version............................4 2.1.4 Changes from the -01 to the -02 version....................5 18.104.22.168.5 Changes from the -00 to -01 version........................5 3 Introduction................................................6Introduction................................................7 4 Architectural Model.........................................8Model.........................................9 4.1 Components..................................................9Components.................................................10 4.2 Resources..................................................11 4.2.1 Stream Context Memory.....................................11 4.2.2 Data Buffers..............................................11 4.2.3 Page Translation Tables...................................11Tables...................................12 4.2.4 STag Namespace............................................12 4.2.5 Completion Queues.........................................12 4.2.6 Asynchronous Event Queue..................................12Queue..................................13 4.2.7 RDMA Read Request Queue...................................13 4.2.8 RNIC Interactions.........................................13 22.214.171.124 Privileged Control Interface Semantics................13 126.96.36.199 Non-Privileged Data Interface Semantics...............13Semantics...............14 188.8.131.52 Privileged Data Interface Semantics...................14 4.2.9 Initialization of RNIC Data Structures for Data Transfer..14 4.2.10 RNIC Data Transfer Interactions.........................15Interactions.........................16 5 Trust and Resource Sharing.................................17 6 Attacker Capabilities......................................18 7 Attacks and Countermeasures................................19 7.1 Tools for Countermeasures..................................19 7.1.1 Protection Domain (PD)....................................19 7.1.2 Limiting STag Scope.......................................20 7.1.3 Access Rights.............................................21 7.1.4 Limiting the Scope of the Completion Queue................21 7.1.5 Limiting the Scope of an Error............................21 7.2 Spoofing...................................................21 7.2.1 Impersonation.............................................22 7.2.2 Stream Hijacking..........................................22 7.2.3 Man in the Middle Attack..................................22 7.2.4 Using an STag on a Different Stream.......................23 7.3 Tampering..................................................24 7.3.1 Buffer Overrun - RDMA Write or Read Response..............24 7.3.2 Modifying a Buffer After Indication.......................25 7.3.3 Multiple STags to access the same buffer..................25 7.3.4 Network based modification of buffer content..............25 7.4 Information Disclosure.....................................26 7.4.1 Probing memory outside of the buffer bounds...............26 7.4.2 Using RDMA Read to Access Stale Data......................26 7.4.3 Accessing a Buffer After the Transfer.....................26 7.4.4 Accessing Unintended Data With a Valid STag...............26 7.4.5 RDMA Read into an RDMA Write Buffer.......................27 7.4.6 Using Multiple STags Which Alias to the Same Buffer.......27 7.4.7 Remote Node Loading Firmware onto the RNIC................28 7.4.8 Controlling Access to PTT & STag Mapping..................28 7.4.9 Network based eavesdropping...............................28 7.5 Denial of Service (DOS)....................................29 7.5.1 RNIC Resource Consumption.................................29 7.5.2 Resource Consumption By Active Applications...............30ULPs.......................30 184.108.40.206 Multiple Streams Sharing Receive Buffers..............30 220.127.116.11 Local PeerULP Attacking a Shared CQ......................31CQ.......................31 18.104.22.168 Remote Peer Attacking a Shared CQ.....................32 22.214.171.124 Attacking the RDMA Read Request Queue.................35 7.5.3 Resource Consumption by Idle Applications.................36ULPs.........................36 7.5.4 Exercise of non-optimal code paths........................36 7.5.5 Remote Invalidate an STag Shared on Multiple Streams......37 7.6 Elevation of Privilege.....................................37 8 Security Services for RDMA and DDP.........................38 8.1 Introduction to Security Options...........................38 8.1.1 Introduction to IPsec.....................................38 8.1.2 Introduction to SSL Limitations on RDMAP..................40 8.1.3 ApplicationsULPs Which Provide Security.......................40Security...............................40 8.2 Requirements for IPsec Encapsulation of DDP................40DDP................41 9 Security considerations....................................42 10 References.................................................43 10.1 Normative References......................................43 10.2 Informative References....................................43 11 Appendix A: ImplementingULP Issues for RDDP Client/Server Protocols...........44Protocols....44 12 Appendix B: Summary Tableof Attacks.......................48RNIC and ULP Implementation Requirements.....................................................48 13 Appendix C: Partial Trust Taxonomy.........................50 14 AuthorÆsAuthor’s Addresses.........................................52 15 Acknowledgments............................................53 16 Full Copyright Statement...................................54 Table of Figures Figure 1 - RDMA Security Model....................................9Model...................................10 2.1 Revision History 2.1.1 Changes from -04 to -05 version * Small modifications to normative statements per phone call review. * 4.1 - Moved MUST statement from Privileged Resource Manager to section 5. Also added additional normative statements around resource sharing and assumptions of who trusts whom. * 7.2.4 - changed last paragraph SHOULD to should. * 7.4.4 - changed last paragraph MUST to SHOULD. * 126.96.36.199 - clarified it is the ULP at issue, and removed reference to Protection Domain - key issue is whether they share partial mutual trust. * 188.8.131.52 - remove MUST statement at the end of the 3rd paragraph - it was replaced with a more general MUST in section <TBD>. Also changed the cap on the number of outstanding RDMA Read Requests at the sender to a SHOULD (from MUST). * 8.1 - first paragraph after enumerated list. Change MAY to may. It is a ULP issue. * Removed "application" from the document and replaced it with "ULP". In some cases also changed "Local Peer" to ULP to clarify what the text meant. 2.1.2 Changes from -03 to -04 version * Removed "issues" section because all issues have been resolved. * Completed section "Applications"ULPs Which Provide Security" by providing a cross reference to channel bindings. * Substantial rewrite of Section 11 Appendix A: ImplementingULP Issues for RDDP Client/Server Protocols. Retargeted it to focus on server application requirements, rather than RNIC requirements. * Changed "IPSec" to "IPsec" everywhere to match the RFC. * Added new ULP requirement in section 184.108.40.206 Attacking the RDMA Read Request Queue. * Reviesed Sectio 12 Appendix B: Summary of RNIC and ULP Implementation Requirements slightly to add one ULP requirement and one RNIC requirement which is stated in the document but was missed in this summary. 220.127.116.11.3 Changes from -02 to -03 version * ID changed from Informational to Standards Track. This caused previous RECOMMENDATIONS to be categorized into the categories of MUST, SHOULD, MAY, RECOMMENDED, and in one case, "recommended". * Completed Appendix B: Summary of Attacks to provide a summary of implementation requirements for applications using RDDP and for RNICs in Appendix B: Summary of Attacks. * Modified intro to better explain when concept of Partial Mutual Trust is useful. * Misc minor changes from Tom Talpey's extensive review, including: * Send Queue/Receive Queue formally defined/used. * RI is gone, now use RNIC interface, RNIC, and Remote Invalidate. * Clarified attackers capabilities. * In many cases replaced "session" with "Stream". * Added definitions for equation variables in section 18.104.22.168. * Changed section 8.2 to normative xref to IPS Security, plus comment on the value of end-to-end IPsec. * Added clarifying example on STag invalidation (e.g. One- Shot STag discussion). * Added clarifying text on why SSL is a bad idea. * Normative statement on mitigation for Shared RQ. 22.214.171.124.4 Changes from the -01 to the -02 version Minimal - some typos, deleted some text previously marked for deletion. 126.96.36.199.5 Changes from the -00 to -01 version * Added two pages to the architectural model to describe the Asynchronous Event Queue, and the types of interactions that can occur between the RNIC and the modules above it. * Addressed Mike Krauses comments submitted on 12/8/2003 * Moved "Trust Models" from the body of the document to an appendix. Removed references to it throughout the document (including use of "partial trust". Document now assumes Remote Peer is untrusted. Thus the key issue is whether local resources are shared, and what the resource is. * Misc cleanup throughout the document. * The Summary of Attacks at the end of the document is now an Appendix. It also now provides a summary. Cleared change bars because became unreadable. Also shortened section names for attacks to fit in table. * Added a new concept of "Partial Mutual Trust" between a collection of Streams to better characterize a set of attacks in a client/server environment. * Filled in Security Services for RDMA and DDP section (almost all is new, except IPsec overview). * Globally tried to change "connection" to "Stream". In some cases it can be either a connection or stream. 3 Introduction RDMA enables new levels of flexibility when communicating between two parties compared to current conventional networking practice (e.g. a stream-based model or datagram model). This flexibility brings new security issues that must be carefully understood when designing application protocolsUpper Layer Protocols (ULPs) utilizing RDMA and when implementing RDMA-aware NICs (RNICs). Note that for the purposes of this security analysis, an RNIC may implement RDMAP and DDP, or just DDP. Also, a ULP may be an application or it may be a middleware library. The specification first develops an architectural model that is relevant for the security analysis - it details components, resources, and system properties that may be attacked in Section 4. It then defines what resources a ULP may share locally across Streams and what resources the ULP may share with the Remote Peer across Streams in Section 5. Intentional sharing of resources between multiple Streams may imply some level of trust between the Streams. However, some types of resource sharing have unmitigated security attacks which would mandate not sharing a specific type of resource unless there is some level of trust between the Streams sharing resources. Partial Mutual Trust is defined to address this concept: Partial Mutual Trust - a collection of RDMAP/DDP Streams, which represent the local and remote end points of the Stream, are willing to assume that the Streams from the collection will not perform malicious attacks against any of the Streams in the collection. ApplicationsULPs have explicit control of which collection of endpoints is in the collection through tools discussed in Section 7.1 Tools for Countermeasures on page 19. An untrusted peer relationship is appropriate when an applicationa ULP wishes to ensure that it will be robust and uncompromised even in the face of a deliberate attack by its peer. For example, a single applicationULP that concurrently supports multiple unrelated Streams (e.g. a server) would presumably treat each of its peers as an untrusted peer. For a collection of Streams which share Partial Mutual Trust, the assumption is that any Stream not in the collection is untrusted. For the untrusted peer, a brief list of capabilities is enumerated in Section 6. The rest of the specification is focused on analyzing attacks. First, the tools for mitigating attacks are listed (Section 7.1), and then a series of attacks on components, resources, or system properties is enumerated in the rest of Section 7. For each attack, possible countermeasures are reviewed. If all recommended mitigations are in place the implemented usage models, the RDMAP/DDP protocol can be shown to not expose any new security vulnerabilities. ApplicationsULPs within a host are divided into two categories - Privileged and Non-Privileged. Both applicationULP types can send and receive data and request resources. The key differences between the two are: The Privileged ApplicationULP is trusted by the system to not maliciously attack the operating environment, but it is not trusted to optimize resource allocation globally. For example, the Privileged ApplicationULP could be a kernel application,ULP, thus the kernel presumably has in some way vetted the applicationULP before allowing it to execute. A Non-Privileged ApplicationÆsULP's capabilities are a logical sub-set of the Privileged ApplicationÆs.ULP's. It is assumed by the local system that a Non-Privileged ApplicationULP is untrusted. All Non-Privileged ApplicationULP interactions with the RNIC Engine that could affect other applicationsULPs need to be done through a trusted intermediary that can verify the Non-Privileged ApplicationULP requests. 4 Architectural Model This section describes an RDMA architectural reference model that is used as security issues are examined. It introduces the components of the model, the resources that can be attacked, the types of interactions possible between components and resources, and the system properties which must be preserved. Figure 1 shows the components comprising the architecture and the interfaces where potential security attacks could be launched. External attacks can be injected into the system from an applicationa ULP that sits above the RNIC Interface or from the network. The intent here is to describe high level components and capabilities which affect threat analysis, and not focus on specific implementation options. Also note that the architectural model is an abstraction, and an actual implementation may choose to subdivide its components along different boundary lines than defined here. For example, the Privileged Resource Manager may be partially or completely encapsulated in the Privileged Application.ULP. Regardless, it is expected that the security analysis of the potential threats and countermeasures still apply. +-------------+ | Privileged | | Resource | Admin<-+>| Manager | AppULP Control Interface | | |<------+-------------------+ | +-------------+ | | | ^ v v | | +-------------+ +-----------------+ |---------------->| Privileged | | Non-Privileged | | | ApplicationULP | | ApplicationULP | | +-------------+ +-----------------+ | ^ ^ |Privileged |Privileged |Non-Privileged |Control |Data |Data |Interface |Interface |Interface RNIC | | | Interface v v v ================================================================= +--------------------------------------+ | | | RNIC Engine | <-- Firmware | | +--------------------------------------+ ^ | v Internet Figure 1 - RDMA Security Model 4.1 Components The components shown in Figure 1 - RDMA Security Model are: * RNIC Engine (RNIC) - the component that implements the RDMA protocol and/or DDP protocol. * Privileged Resource Manager - the component responsible for managing and allocating resources associated with the RNIC Engine. The Resource Manager does not send or receive data. Note that whether the Resource Manager is an independent component, part of the RNIC, or part of the applicationULP is implementation dependent. If a specific implementation does not wish to address security issues resolved by the Resource Manager, there may in fact be no resource manager at all. * Privileged ApplicationULP - See Section 3 Introduction for a definition of Privileged Application.ULP. The local host infrastructure can enable the Privileged ApplicationULP to map a data buffer directly from the RNIC Engine to the host through the RNIC Interface, but it does not allow the Privileged ApplicationULP to directly consume RNIC Engine resources. * Non-Privileged ApplicationULP - See Section 3 Introduction for a definition of Non-Privileged Application. All Non- Privileged Application interactions with the RNIC Engine that could affect other applications MUST be done using the Privileged Resource Manager as a proxy.ULP. A design goal of the DDP and RDMAP protocols is to allow, under constrained conditions, Non-Privileged applicationsULP to send and receive data directly to/from the RDMA Engine without Privileged Resource Manager intervention - while ensuring that the host remains secure. Thus, one of the primary goals of this paper is to analyze this usage model for the enforcement that is required in the RNIC Engine to ensure the system remains secure. The host interfaces that could be exercised include: * Privileged Control Interface - A Privileged Resource Manager uses the RNIC Interface to allocate and manage RNIC Engine resources, control the state within the RNIC Engine, and monitor various events from the RNIC Engine. It also uses this interface to act as a proxy for some operations that a Non-Privileged ApplicationULP may require (after performing appropriate countermeasures). * ApplicationULP Control Interface - An applicationULP uses this interface to the Privileged Resource Manager to allocate RNIC Engine resources. The Privileged Resource Manager implements countermeasures to ensure that if the Non- Privileged ApplicationNon-Privileged ULP launches an attack it can prevent the attack from affecting other applications.ULPs. * Non-Privileged Data Transfer Interface - A Non-Privileged ApplicationULP uses this interface to initiate and to check the status of data transfer operations. * Privileged Data Transfer Interface - A superset of the functionality provided by the Non-Privileged Data Transfer Interface. The applicationULP is allowed to directly manipulate RNIC Engine mapping resources to map an STag to an applicationa ULP data buffer. * Figure 1 also shows the ability to load new firmware in the RNIC Engine. Not all RNICs will support this, but it is shown for completeness and is also reviewed under potential attacks. If Internet control messages, such as ICMP, ARP, RIPv4, etc. are processed by the RNIC Engine, the threat analyses for those protocols is also applicable, but outside the scope of this paper. 4.2 Resources This section describes the primary resources in the RNIC Engine that could be affected if under attack. For RDMAP, all of the defined resources apply. For DDP, all of the resources except the RDMA Read Queue apply. 4.2.1 Stream Context Memory The state information for each Stream is maintained in memory, which could be located in a number of places - on the NIC, inside RAM attached to the NIC, in host memory, or in any combination of the three, depending on the implementation. Stream Context Memory includes state associated with Data Buffers. For Tagged Buffers, this includes how STag names, Data Buffers, and Page Translation Tables inter-relate. It also includes the list of Untagged Data Buffers posted for reception of Untagged Messages (commonly called the Receive Queue), and a list of operations to perform to send data (commonly called the Send Queue). 4.2.2 Data Buffers There are two different ways to expose a data buffer; a buffer can be exposed for receiving RDMAP Send Type Messages (a.k.a. DDP Untagged Messages) on DDP Queue zero or the buffer can be exposed for remote access through STags (a.k.a. DDP Tagged Messages). This distinction is important because the attacks and the countermeasures used to protect against the attack are different depending on the method for exposing the buffer to the network. For the purposes of the security discussion, a single logical Data Buffer is exposed with a single STag. Actual implementations may support scatter/gather capabilities to enable multiple physical data buffers to be accessed with a single STag, but from a threat analysis perspective it is assumed that a single STag enables access to a single logical Data Buffer. In any event, it is the responsibility of the RNIC to ensure that no STag can be created that exposes memory that the consumer had no authority to expose. 4.2.3 Page Translation Tables Page Translation Tables are the structures used by the RNIC to be able to access applicationULP memory for data transfer operations. Even though these structures are called "Page" Translation Tables, they may not reference a page at all - conceptually they are used to map an applicationa ULP address space representation of a buffer to the physical addresses that are used by the RNIC Engine to move data. If on a specific system a mapping is not used, then a subset of the attacks examined may be appropriate. Note that the Page Translation Table may or may not be a shared resource. 4.2.4 STag Namespace The DDP specification defines a 32-bit namespace for the STag. Implementations may vary in terms of the actual number of STags that are supported. In any case, this is a bounded resource that can come under attack. Depending upon STag namespace allocation algorithms, the actual name space to attack may be significantly less than 2^32. 4.2.5 Completion Queues Completion Queues are used in this specification to conceptually represent how the RNIC Engine notifies the ApplicationULP about the completion of the transmission of data, or the completion of the reception of data through the Data Transfer Interface. Because there could be many transmissions or receptions in flight at any one time, completions are modeled as a queue rather than a single event. An implementation may also use the Completion Queue to notify the applicationULP of other activities, for example, the completion of a mapping of an STag to a specific applicationULP buffer. Completion Queues may be shared by a group of Streams, or may be designated to handle a specific Stream's traffic. Some implementations may allow this queue to be manipulated directly by both Non-Privileged and Privileged applications.ULPs. 4.2.6 Asynchronous Event Queue The Asynchronous Event Queue is a queue from the RNIC to the Privileged Resource Manager of bounded size. It is used by the RNIC to notify the host of various events which might require management action, including protocol violations, Stream state changes, local operation errors, low water marks on receive queues, and possibly other events. The Asynchronous Event Queue is a resource that can be attacked because Remote or Local Peers can cause events to occur which have the potential of overflowing the queue. Note that an implementation is at liberty to implement the functions of the Asynchronous Event Queue in a variety of ways, including multiple queues or even simple callbacks. All vulnerabilities identified are intended to apply regardless of the implementation of the Asynchronous Event Queue. For example, a callback function is simply a very short queue. 4.2.7 RDMA Read Request Queue The RDMA Read Request Queue is the memory that holds state information for one or more RDMA Read Request Messages that have arrived, but for which the RDMA Read Response Messages have not yet been completely sent. Because potentially more than one RDMA Read Request can be outstanding at one time, the memory is modeled as a queue of bounded size. 4.2.8 RNIC Interactions With RNIC resources and interfaces defined, it is now possible to examine the interactions supported by the generic RNIC functional interfaces through each of the 3 interfaces - Privileged Control Interface, Privileged Data Interface, and Non-Privileged Data Interface. 188.8.131.52 Privileged Control Interface Semantics Generically, the Privileged Control Interface controls the RNICÆsRNIC’s allocation, deallocation, and initialization of RNIC global resources. This includes allocation and deallocation of Stream Context Memory, Page Translation Tables, STag names, Completion Queues, RDMA Read Request Queues, and Asynchronous Event Queues. The Privileged Control Interface is also typically used for managing Non-Privileged ApplicationULP resources for the Non- Privileged ApplicationNon-Privileged ULP (and possibly for the Privileged ApplicationULP as well). This includes initialization and removal of Page Translation Table resources, and managing RNIC events (possibly managing all events for the Asynchronous Event Queue). 184.108.40.206 Non-Privileged Data Interface Semantics The Non-Privileged Data Interface enables data transfer (transmit and receive) but does not allow initialization of the Page Translation Table resources. However, once the Page Translation Table resources have been initialized, the interface may enable a specific STag mapping to be enabled and disabled by directly communicating with the RNIC, or create an STag mapping for a buffer that has been previously initialized in the RNIC. For RDMAP, transmitting data means sending RDMAP Send Type Messages, RDMA Read Requests, and RDMA Writes. For data reception, for RDMAP it can receive Send Type Messages into buffers that have been posted on the Receive Queue or Shared Receive Queue. It can also receive RDMA Write and RDMA Read Response Messages into buffers that have previously been exposed for external write access through advertisement of an STag. For DDP, transmitting data means sending DDP Tagged or Untagged Messages. For data reception, for DDP it can receive Untagged Messages into buffers that have been posted on the Receive Queue or Shared Receive Queue. It can also receive Tagged DDP Messages into buffers that have previously been exposed for external write access through advertisement of an STag. Completion of data transmission or reception generally entails informing the applicationULP of the completed work by placing completion information on the Completion Queue. 220.127.116.11 Privileged Data Interface Semantics The Privileged Data Interface semantics are a superset of the Non-Privileged Data Transfer semantics. The interface can do everything defined in the prior section, as well as create/destroy buffer to STag mappings directly. This generally entails initialization or clearing of Page Translation Table state in the RNIC. 4.2.9 Initialization of RNIC Data Structures for Data Transfer Initialization of the mapping between an STag and a Data Buffer can be viewed in the abstract as two separate opertions: a. Initialization of the allocated Page Translation Table entries with the location of the Data Buffer, and b. Initialization of a mapping from an allocated STag name to a set of Page Translation Table entry(s) or partial- entries. Note that an implementation may not have a Page Translation Table (i.e. it may support a direct mapping between an STag and a Data Buffer). In this case threats and mitigations associated with the Page Translation Table are not relevant. Initialization of the contents of the Page Translation Table can be done by either the Privileged ApplicationULP or by the Privileged Resource Manager as a proxy for the Non-Privileged Application.ULP. By definition the Non-Privileged ApplicationULP is not trusted to directly manipulate the Page Translation Table. In general the concern is that the Non-Privileged applicationULP may try to maliciously initialize the Page Translation Table to access a buffer for which it does not have permission. The exact resource allocation algorithm for the Page Translation Table is outside the scope of this specification. It may be allocated for a specific Data Buffer, or be allocated as a pooled resource to be consumed by potentially multiple Data Buffers, or be managed in some other way. This paper attempts to abstract implementation dependent issues, and focus on higher level security issues such as resource starvation and sharing of resources between Streams. The next issue is how an STag name is associated with a Data Buffer. For the case of an Untagged Data Buffer, there is no wire visible mapping between an STag and the Data Buffer. Note that there may, in fact, be an STag which represents the buffer. However, because the STag by definition is not visible on the wire, this is a local host specific issue which should be analyzed in the context of local host implementation specific security analysis, and thus is outside the scope of this paper. For a Tagged Data Buffer, either the Privileged Application,ULP, the Non-Privileged Application,Non- Privileged ULP, or the Privileged Resource Manager acting on behalf of the Non-Privileged Resource Manager may initialize a mapping from an STag to a Page Translation Table, or may have the ability to simply enable/disable an existing STag to Page Translation Table mapping. There may also be multiple STag names which map to a specific group of Page Translation Table entries (or sub-entries). Specific security issues with this level of flexibility are examined in Section 7.3.3 Multiple STags to access the same buffer on page 25. There are a variety of implementation options for initialization of Page Translation Table entries and mapping an STag to a group of Page Translation Table entries which have security repercussions. This includes support for separation of Mapping an STag verses mapping a set of Page Translation Table entries, and support for ApplicationsULPs directly manipulating STag to Page Translation Table entry mappings (verses requiring access through the Privileged Resource Manager). 4.2.10 RNIC Data Transfer Interactions RNIC Data Transfer operations can be subdivided into send operations and receive operations. For send operations, there is typically a queue that enables the ApplicationULP to post multiple operations to send data (referred to as the Send Queue). Depending upon the implementation, Data Buffers used in the operations may or may not have Page Translation Table entries associated with them, and may or may not have STags associated with them. Because this is a local host specific implementation issue rather than a protocol issue, the security analysis of threats and mitigations is left to the host implementation. Receive operations are different for Tagged Data Buffers verses Untagged Data Buffers. If more than one Untagged Data Buffer can be posted by the Application,ULP, the DDP specification requires that they be consumed in sequential order. Thus the most general implementation is that there is a sequential queue of receive Untagged Data Buffers (Receive Queue). Some implementations may also support sharing of the sequential queue between multiple Streams. In this case defining "sequential" becomes non-trivial - in general the buffers for a single stream are consumed from the queue in the order that they were placed on the queue, but there is no order guarantee between streams. For receive Tagged Data Buffers, at some time prior to data transfer, the mapping of the STag to specific Page Translation Table entries (if present) and the mapping from the Page Translation Table entries to the Data Buffer must have been initialized (see the prior section for interaction details). 5 Trust and Resource Sharing It is assumed that in general the Local and Remote Peer are untrusted, and thus attacks by either should have mitigations in place. A separate, but related issue is resource sharing between multiple streams. If local resources are not shared, the resources are dedicated on a per Stream basis. Resources are defined in Section 4.2 -Resources on page 10. The advantage of not sharing resources between Streams is that it reduces the types of attacks that are possible. The disadvantage is that applicationsULPs might run out of resources. It is assumed in this paper that the component that implements the mechanism to control sharing of the RNIC Engine resources is the Privileged Resource Manager. The RNIC Engine exposes its resources through the RNIC Interface to the Privileged Resource Manager. All Privileged and Non-Privileged applicationsULPs request resources from the Resource Manager.Manager (note that by definition both the Non- Privileged and the Privileged application might try to greedily consume resources, thus creating a potential Denial of Service (DOS) attack. The Resource Manager implements resource management policies to ensure fair access to resources. The Resource Manager should be designed to take into account security attacks detailed in this specification. Note that for some systems the Privileged Resource Manager may be implemented within the Privileged Application.ULP. All Non-Privileged ULP interactions with the RNIC Engine that could affect other ULPs MUST be done using the Privileged Resource Manager as a proxy. All ULP resource allocation requests for scarce resources MUST also be done using a Privileged Resource Manager. The sharing of resources across Streams should be under the control of the application,ULP, both in terms of the trust model the applicationULP wishes to operate under, as well as the level of resource sharing the applicationULP wishes to give Local Peer processes. For more discussion on types of trust models which combine partial trust and sharing of resources, see Appendix C: Partial Trust Taxonomy on page 50. The Privileged Resource Manager MUST NOT assume different ULPs share Partial Mutual Trust unless there is a mechanism to ensure that the ULPs do indeed share partial mutual trust. This can be done in several ways, including explicit notification from the ULP. 6 Attacker Capabilities An attackerÆsattacker’s capabilities delimit the types of attacks that attacker is able to launch. RDMAP and DDP require that the initial LLP Stream (and connection) be set up prior to transferring RDMAP/DDP Messages. Attackers with send only capabilities must first guess the current LLP Stream parameters before they can attack RNIC resources (e.g. TCP sequence number). Attackers with both send and receive capabilities have presumably setup a valid LLP Stream, and thus have a wider ability to attack RNIC resources. 7 Attacks and Countermeasures This section describes the attacks that are possible against the RDMA system defined in Figure 1 - RDMA Security Model and the RNIC Engine resources defined in Section 4.2. The analysis includes a detailed description of each attack, what is being attacked, and a description of the countermeasures that can be taken to thwart the attack. Note that connection setup and teardown is presumed to be done in stream mode (i.e. no RDMA encapsulation of the payload), so there are no new attacks related to connection setup/teardown beyond what is already present in the LLP (e.g. TCP or SCTP). Note, however, that RDMAP/DDP parameters may be exchanged in stream mode, and if they are corrupted by an attacker unintended consequences will result. Therefore, any existing mitigations for LLP Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, or Elevation of Privilege continues to apply (and is out of scope of this document). Thus the analysis in this section focuses on attacks that are present regardless of the LLP Stream type. The attacks are classified into five categories: Spoofing, Tampering, Information Disclosure, Denail of Service (DoS) attacks, and Elevation of Privileges. Tampering is any modification of the legitimate traffic (machine internal or network). Spoofing attack is a special case of tempering; where the attacker falsifies an identity of the Remote Peer (identity can be an IP address, machine name, ULP level identity etc.). 7.1 Tools for Countermeasures The tools described in this section are the primary mechanisms that can be used to provide countermeasures to potential attacks. 7.1.1 Protection Domain (PD) Protection Domains are associated with two of the resources of concern, Stream Context Memory and STags associated with Page Translation Table entries and data buffers. Protection Domains are used mainly to ensure that an STag can only be used to access the associated data buffer through Streams in the same Protection Domain as that STag. If an implementation chooses to not share resources between Streams, it is recommended that each Stream be associated with its own, unique Protection Domain. If an implementation chooses to allow resource sharing, it is recommended that Protection Domain be limited to the number of Streams that have Partial Mutual Trust. Note that an applicationa ULP (either Privileged or Non-Privileged) can potentially have multiple Protection Domains. This could be used, for example, to ensure that multiple clients of a server do not have the ability to corrupt each other. The server would allocate a Protection Domain per client to ensure that resources covered by the Protection Domain could not be used by another (untrusted) client. 7.1.2 Limiting STag Scope The key to protecting a local data buffer is to limit the scope of its STag to the level appropriate for the Streams which share Partial Mutual Trust. The scope of the STag can be measured in multiple ways. * Number of Connections and/or Streams on which the STag is valid. One way to limit the scope of the STag is to limit the connections and/or Streams that are allowed to use the STag. As noted in the previous section, use of Protection Domains appropriately can limit the scope of the STag. The analysis presented in this document assumes two mechanisms for limiting the scope of Streams for which the STag is valid: * Protection Domain scope. The STag is valid if used on any Stream within a specific Protection Domain, and is invalid if used on any Stream that is not a member of the Protection Domain. * Single Stream scope. The STag is valid on a single Stream, regardless of what the Stream association is to a Protection Domain. If used on any other Stream, it is invalid. * Limit the time an STag is valid. By Invalidating an Advertised STag (e.g., revoking remote access to the buffers described by an STag when done with the transfer), an entire class of attacks can be eliminated. * Limit the buffer the STag can reference. Limiting the scope of an STag access to *just* the intended applicationULP buffers to be exposed is critical to prevent certain forms of attacks. * Allocating and/or advertising STag numbers in an unpredictable way. If STags are allocated/advertised using an algorithm which makes it hard for the attacker to guess which STag(s) are currently in use, it makes it more difficult for an attacker to guess the correct value. As stated in the RDMAP specification [RDMAP], an invalid STag will cause the RDMAP Stream to be terminated. For the case of [DDP], at a minimum it must signal an error to the ULP, and commonly this will cause the DDP stream to be terminated. 7.1.3 Access Rights Access Rights associated with a specific Advertised STag or RDMAP/DDP Stream provide another mechanism for applicationsULPs to limit the attack capabilities of the Remote Peer. The Local Peer can control whether a data buffer is exposed for local only, or local and remote access, and assign specific access privileges (read, write, read and write) on a per stream basis. For DDP, when an STag is advertised, the Remote Peer is presumably given write access rights to the data (otherwise there was not much point to the advertisement). For RDMAP, when an applicationa ULP advertises an STag, it can enable write-only, read- only,read-only, or both write and read access rights. Similarly, some applicationsULPs may wish to provide a single buffer with different access rights on a per-Stream or per-Stream basis. For example, some Streams may have read-only access, some may have remote read and write access, while on other Streams only the Local Peer is allowed access. 7.1.4 Limiting the Scope of the Completion Queue Completions associated with sending and receiving data, or setting up buffers for sending and receiving data, could be accumulated in a shared Completion Queue for a group of RDMAP/DDP Streams, or a specific RDMAP/DDP Stream could have a dedicated Completion Queue. Limiting Completion Queue association to one, or a small number of RDMAP/DDP Streams can prevent several forms of Denial of Service attacks. 7.1.5 Limiting the Scope of an Error To prevent a variety of attacks, it is important that an RDMAP/DDP implementation be robust in the face of errors. If an error on a specific Stream can cause other unrelated Streams to fail, then a broad class of attacks are enabled against the implementation. For example, an error on a specific RDMAP stream should not cause the RNIC to stop processing incoming packets, or corrupt a receive queue for an unrelated stream. 7.2 Spoofing Spoofing attacks can be launched by the Remote Peer, or by a network based attacker. A network based spoofing attack applies to all Remote Peers. Because the RDMAP Stream requires an LLP Stream in the ESTABLISHED state, certain types of traditional forms of wire attacks do not apply -- an end-to-end handshake must have occurred to establish the RDMAP Stream. So, the only form of spoofing that applies is one when a remote node can both send and receive packets. Yet even with this limitation the Stream is still exposed to the following spoofing attacks. 7.2.1 Impersonation A network based attacker can impersonate a legal RDMA/DDP peer (by spoofing a legal IP address), and establish an RDMA/DDP Stream with the victim. End to end authentication (i.e. IPsec, SSL or ULP authentication) provides protection against this attack. For additional information see Section 8, Security Services for RDMA and DDP, on page 38. 7.2.2 Stream Hijacking Stream hijacking happens when a network based attacker follows the Stream establishment phase, and waits until the authentication phase (if such a phase exists) is completed successfully. He can then spoof the IP address and re-direct the Stream from the victim to its own machine. For example, an attacker can wait until an iSCSI authentication is completed successfully, and hijack the iSCSI Stream. The best protection against this form of attack is end-to-end integrity protection and authentication, such as IPsec (see Section 8, Security Services for RDMA and DDP, on page 38), to prevent spoofing. Another option is to provide physical security. Discussion of physical security is out of scope for this document. Because the connection and/or Stream itself is established by the LLP, some LLPs are more difficult to hijack than others. Please see the relevant LLP documentation on security issues around connection and/or Stream hijacking. 7.2.3 Man in the Middle Attack If a network based attacker has the ability to delete, inject replay, or modify packets which will still be accepted by the LLP (e.g., TCP sequence number is correct) then the Stream can be exposed to a man in the middle attack. One style of attack is for the man-in-the-middle to send Tagged Messages (either RDMAP or DDP). If it can discover a buffer that has been exposed for STag enabled access, then the man-in-the-middle can use an RDMA Read operation to read the contents of the associated data buffer, perform an RDMA Write Operation to modify the contents of the associated data buffer, or invalidate the STag to disable further access to the buffer. The only countermeasure for this form of attack is to either secure the RDMAP/DDP Stream (i.e. integrity protect) or attempt to provide physical security to prevent man- in-the-middle type attacks. The best protection against this form of attack is end-to-end integrity protection and authentication, such as IPsec (see Section 8 Security Services for RDMA and DDP on page 38), to prevent spoofing or tampering. If Stream or session level authentication and integrity protection are not used, then a man- in-the-middle attack can occur, enabling spoofing and tampering. Because the connection/Stream itself is established by the LLP, some LLPs are more exposed to man-in-the-middle attack then others. Please see the relevant LLP documentation on security issues around connection and/or Stream hijacking. Another approach is to restrict access to only the local subnet/link, and provide some mechanism to limit access, such as physical security or 802.1.x. This model is an extremely limited deployment scenario, and will not be further examined here. 7.2.4 Using an STag on a Different Stream One style of attack from the Remote Peer is for it to attempt to use STag values that it is not authorized to use. Note that if the Remote Peer sends an invalid STag to the Local Peer, per the DDP and RDMAP specifications, the Stream must be torn down. Thus the threat exists if a STag has been enabled for Remote Access on one Stream and a Remote Peer is able to use it on an unrelated Stream. If the attack is successful, the attacker could potentially be able to perform either RDMA Read Operations to read the contents of the associated data buffer, perform RDMA Write Operations to modify the contents of the associated data buffer, or to Invalidate the STag to disable further access to the buffer. An attempt by a Remote Peer to access a buffer with an STag on a different Stream in the same Protection Domain may or may not be an attack depending on whether resource sharing is intended (i.e. whether the Streams shared Partial Mutual Trust or not). For some applicationsULPs using an STag on multiple Streams within the same Protection Domain could be desired behavior. For other applicationsULPs attempting to use an STag on a different Stream could be considered to be an attack. Since this varies by application, an applicationULP, a ULP typically would need to be able to control the scope of the STag. In the case where an implementation does not share resources between Streams (including STags), this attack can be defeated by assigning each Stream to a different Protection Domain. Before allowing remote access to the buffer, the Protection Domain of the Stream where the access attempt was made is matched against the Protection Domain of the STag. If the Protection Domains do not match, access to the buffer is denied, an error is generated, and the RDMAP Stream associated with the attacking Stream should be terminated. For implementations that share resources between multiple Streams, it may not be practical to separate each Stream into its own Protection Domain. In this case, the applicationULP can still limit the scope of any of the STags to a single Stream (if it is enabling it for remote access). If the STag scope has been limited to a single Stream, any attempt to use that STag on a different Stream will result in an error, and the RDMA Stream should be terminated. Thus for implementations that do not share STags between Streams, each Stream MUST either be in a separate Protection Domain or the scope of an STag MUST be limited to a single Stream. An additional issue may be unintended sharing of STags (i.e. a bug in the application)ULP) or a bug in the Remote Peer which causes an off-by-oneoff- by-one STag to be used. For additional protection, an implementation SHOULDshould allocate STags in such a fashion that it is difficult to predict the next allocated STag number. Allocation methods which deterministically allocate the next STag should be avoided (e.g. a method which always starts with STag equal to one and monotonically increases it for each new allocation, or a method which always uses the same STag for each operation). 7.3 Tampering A Remote Peer or a network based attacker can attempt to tamper with the contents of data buffers on a Local Peer that have been enabled for remote write access. The types of tampering attacks that are possible are outlined in the sections that follow. 7.3.1 Buffer Overrun - RDMA Write or Read Response This attack is an attempt by the Remote Peer to perform an RDMA Write or RDMA Read Response to memory outside of the valid length range of the data buffer enabled for remote write access. This attack can occur even when no resources are shared across Streams. This issue can also arise if the applicationULP has a bug. The countermeasure for this type of attack must be in the RNIC implementation, using the STag. When the Local Peer specifies to the RNIC the base address and the number of bytes in the buffer that it wishes to make accessible, the RNIC must ensure that the base and bounds check are applied to any access to the buffer referenced by the STag before the STag is enabled for access. When an RDMA data transfer operation (which includes an STag) arrives on a Stream, a base and bounds byte granularity access check must be performed to ensure the operation accesses only memory locations within the buffer described by that STag. Thus an RNIC implementation MUST ensure that a Remote Peer is not able to access memory outside of the buffer specified when the STag was enabled for remote access. 7.3.2 Modifying a Buffer After Indication This attack can occur if a Remote Peer attempts to modify the contents of an STag referenced buffer by performing an RDMA Write or an RDMA Read Response after the Remote Peer has indicated to the Local Peer that the STag data buffer contents are ready for use. This attack can occur even when no resources are shared across Streams. Note that a bug in a Remote Peer, or network based tampering, could also result in this problem. For example, assume the STag referenced buffer contains ULP control information as well as ULP payload, and the ULP sequence of operation is to first validate the control information and then perform operations on the control information. If the Remote Peer can perform an additional RDMA Write or RDMA Read Response (thus changing the buffer) after the validity checks have been completed but before the control data is operated on, the Remote Peer could force the ULP down operational paths that were never intended. The Local Peer can protect itself from this type of attack by revoking remote access when the original data transfer has completed and before it validates the contents of the buffer. The Local Peer can either do this by explicitly revoking remote access rights for the STag when the Remote Peer indicates the operation has completed, or by checking to make sure the Remote Peer Invalidated the STag through the RDMAP Invalidate capability, and if it did not, the Local Peer then explicitly revokes the STag remote access rights. The Local Peer SHOULD follow the above procedure to protect the buffer before it validates the contents of the buffer (or uses the buffer in any way). 7.3.3 Multiple STags to access the same buffer See section 7.4.6 Using Multiple STags Which Alias to the Same Buffer on page 27 for this analysis. 7.3.4 Network based modification of buffer content This is actually a man in the middle attack - but only on the content of the buffer, as opposed to the man in the middle attack presented above, where both the signaling and content can be modified. See Section 7.2.3 Man in the Middle Attack on page 22. 7.4 Information Disclosure The main potential source for information disclosure is through a local buffer that has been enabled for remote access. If the buffer can be probed by a Remote Peer on another Stream, then there is potential for information disclosure. The potential attacks that could result in unintended information disclosure and countermeasures are detailed in the following sections. 7.4.1 Probing memory outside of the buffer bounds This is essentially the same attack as described in Section 7.3.1, except an RDMA Read Request is used to mount the attack. The same countermeasure applies. 7.4.2 Using RDMA Read to Access Stale Data If a buffer is being used for a combination of reads and writes (either remote or local), and is exposed to the Remote Peer with at least remote read access rights, the Remote Peer may be able to examine the contents of the buffer before they are initialized with the correct data. In this situation, whatever contents were present in the buffer before the buffer is initialized can be viewed by the Remote Peer, if the Remote Peer performs an RDMA Read. Because of this, the Local Peer SHOULD ensure that no stale data is contained in the buffer before remote read access rights are granted (this can be done by zeroing the contents of the memory, for example). 7.4.3 Accessing a Buffer After the Transfer If the Remote Peer has remote read access to a buffer, and by some mechanism tells the Local Peer that the transfer has been completed, but the Local Peer does not disable remote access to the buffer before modifying the data, it is possible for the Remote Peer to retrieve the new data. This is similar to the attack defined in Section 7.3.2 Modifying a Buffer After Indication on page 25. The same countermeasures apply. In addition, the Local Peer SHOULD grant remote read access rights only for the amount of time needed to retrieve the data. 7.4.4 Accessing Unintended Data With a Valid STag If the Local PeerULP enables remote access to a buffer using an STag that references the entire buffer, but intends only a portion of the buffer to be accessed, it is possible for the Remote Peer to access the other parts of the buffer anyway. To prevent this attack, the Local Peer MUSTULP SHOULD set the base and bounds of the buffer when the STag is initialized to expose only the data to be retrieved. 7.4.5 RDMA Read into an RDMA Write Buffer One form of disclosure can occur if the access rights on the buffer enabled remote read, when only remote write access was intended. If the buffer contained applicationULP data, or data from a transfer on an unrelated Stream, the Remote Peer could retrieve the data through an RDMA Read operation. Note that an RNIC implementation is not required to support STags that have both read and write access. The most obvious countermeasure for this attack is to not grant remote read access if the buffer is intended to be write-only. Then the Remote Peer would not be able to retrieve data associated with the buffer. An attempt to do so would result in an error and the RDMAP Stream associated with the Stream would be terminated. Thus if an applicationa ULP only intends a buffer to be exposed for remote write access, it MUST set the access rights to the buffer to only enable remote write access. 7.4.6 Using Multiple STags Which Alias to the Same Buffer Multiple STags which alias to the same buffer at the same time can result in unintentional information disclosure if the STags are used by different, mutually untrusted, Remote Peers. This model applies specifically to client/server communication, where the server is communicating with multiple clients, each of which do not mutually trust each other. If only read access is enabled, then the Local Peer has complete control over information disclosure. Thus a server which intended to expose the same data (i.e. buffer) to multiple clients by using multiple STags to the same buffer creates no new security issues beyond what has already been described in this document. Note that if the server did not intend to expose the same data to the clients, it should use separate buffers for each client (and separate STags). When one STag has remote read access enabled and a different STag has remote write access enabled to the same buffer, it is possible for one Remote Peer to view the contents that have been written by another Remote Peer. If both STags have remote write access enabled and the two Remote Peers do not mutually trust each other, it is possible for one Remote Peer to overwrite the contents that have been written by the other Remote Peer. Thus multiple Remote Peers which do not share Partial Mutual Trust MUST NOT be granted write access to the same buffer through different STags. A buffer should be exposed to only one untrusted Remote Peer at a time to ensure that no information disclosure or information tampering occurs between peers. 7.4.7 Remote Node Loading Firmware onto the RNIC If the Remote Peer can cause firmware to be loaded onto the RNIC, there is an opportunity for information disclosure. See Elevation of Privilege in Section 7.6 for this analysis. 7.4.8 Controlling Access to PTT & STag Mapping If a Non-Privileged applicationULP is able to directly manipulate the RNIC Page Translation Tables (which translate from an STag to a host address), it is possible that the Non-Privileged applicationULP could point the Page Translation Table at an unrelated applicationÆsULP’s buffers and thereby be able to gain access to information in the unrelated application.ULP. As discussed in Section 4 Architectural Model on page 8,9, introduction of a Privileged Resource Manager to arbitrate the mapping requests is an effective countermeasure. This enables the Privileged Resource Manager to ensure an applicationa ULP can only initialize the Page Translation Table (PTT)to point to its own buffers. Thus if Non-Privileged applicationsULPs are supported, the Privileged Resource Manager MUST verify that the Non-Privileged applicationULP has the right to access a specific Data Buffer before allowing an STag for which the applicationULP has access rights to be associated with a specific Data Buffer. This can be done when the Page Translation Table is initialized to access the Data Buffer or when the STag is initialized to point to a group of Page Translation Table entries, or both. 7.4.9 Network based eavesdropping An attacker that is able to eavesdrop on the network can read the content of all read and write access to the peerÆspeer’s buffers. To prevent information disclosure, the read/written data must be encrypted. See also Section 7.2.3 Man in the Middle Attack on page 22. The encryption can be done either by the ULP, or by a protocol that provides security services to the LLP (e.g. IPsec or SSL). Refer to section 8 for discussion of security services for DDP/RDMA. 7.5 Denial of Service (DOS) A DOS attack is one of the primary security risks of RDMAP. This is because RNIC resources are valuable and scarce, and many applicationULP environments require communication with untrusted Remote Peers. If the remote applicationULP can be authenticated or encrypted, clearly, the DOS profile can be reduced. For the purposes of this analysis, it is assumed that the RNIC must be able to operate in untrusted environments, which are open to DOS style attacks. Denial of service attacks against RNIC resources are not the typical unknown party spraying packets at a random host (such as a TCP SYN attack). Because the connection/Stream must be fully established, the attacker must be able to both send and receive messages over that connection/Stream, or be able to guess a valid packet on an existing RDMAP Stream. This section outlines the potential attacks and the countermeasures available for dealing with each attack. 7.5.1 RNIC Resource Consumption This section covers attacks that fall into the general category of a Local Peer attempting to unfairly allocate scarce (i.e. bounded) RNIC resources. The Local Peer may be attempting to allocate resources on its own behalf, or on behalf of a Remote Peer. Resources that fall into this category include: Protection Domains, Stream Context Memory, Translation and Protection Tables, and STag namespace. These can be attacks by currently active Local Peers or ones that allocated resources earlier, but are now idle. This type of attack can occur regardless of whether resources are shared across Streams. The allocation of all scarce resources MUST be placed under the control of a Privileged Resource Manager. This allows the Privileged Resource Manager to: * prevent a Local Peer from allocating more than its fair share of resources. * detect if a Remote Peer is attempting to launch a DOS attack by attempting to create an excessive number of Streams (with associated resources) and take corrective action (such as refusing the request or applying network layer filters against the Remote Peer). This analysis assumes that the Resource Manager is responsible for handing out Protection Domains, and RNIC implementations will provide enough Protection Domains to allow the Resource Manager to be able to assign a unique Protection Domain for each unrelated, untrusted Local Peer (for a bounded, reasonable number of Local Peers). This analysis further assumes that the Resource Manager implements policies to ensure that untrusted Local Peers are not able to consume all of the Protection Domains through a DOS attack. Note that Protection Domain consumption cannot result from a DOS attack launched by a Remote Peer, unless a Local Peer is acting on the Remote PeerÆsPeer’s behalf. 7.5.2 Resource Consumption By Active ApplicationsULPs This section describes DOS attacks from Local and Remote Peers that are actively exchanging messages. Attacks on each RDMA NIC resource are examined and specific countermeasures are identified. Note that attacks on Stream Context Memory, Page Translation Tables, and STag namespace are covered in Section 7.5.1 RNIC Resource Consumption, so are not included here. 18.104.22.168 Multiple Streams Sharing Receive Buffers The Remote Peer can attempt to consume more than its fair share of receive data buffers (Untagged DDP buffers or for RDMAP buffers consumed with Send Type Messages) if receive buffers are shared across multiple Streams. If resources are not shared across multiple Streams, then this attack is not possible because the Remote Peer will not be able to consume more buffers than were allocated to the Stream. The worst case scenario is that the Remote Peer can consume more receive buffers than the Local Peer allowed, resulting in no buffers to be available, which could cause the Remote PeerÆsPeer’s Stream to the Local Peer to be torn down, and all allocated resources to be released. If local receive data buffers are shared among multiple Streams, then the Remote Peer can attempt to consume more than its fair share of the receive buffers, causing a different Stream to be short of receive buffers, thus possibly causing the other Stream to be torn down. For example, if the Remote Peer sent enough one byte Untagged Messages, they might be able to consume all local shared receive queue resources with little effort on their part. One method the Local Peer could use is to recognize that a Remote Peer is attempting to use more than its fair share of resources and terminate the Stream (causing the allocated resources to be released). However, if the Local Peer is sufficiently slow, it may be possible for the Remote Peer to still mount a denial of service attack. One countermeasure that can protect against this attack is implementing a low-water notification. The low-water notification alerts the applicationULP if the number of buffers in the receive queue is less than a threshold. If all of the following conditions are true, then the Local Peer can size the amount of local receive buffers posted on the receive queue to ensure a DOS attack can be stopped. * a low-water notification is enabled, and * the Local Peer is able to bound the amount of time that it takes to replenish receive buffers, and * the Local Peer maintains statistics to determine which Remote Peer is consuming buffers. The above conditions enable the low-water notification to arrive before resources are depleted and thus the Local Peer can take corrective action (e.g., terminate the Stream of the attacking Remote Peer). A different, but similar attack is if the Remote Peer sends a significant number of out-of-order packets and the RNIC has the ability to use the applicationULP buffer as a reassembly buffer. In this case the Remote Peer can consume a significant number of applicationULP buffers, but never send enough data to enable the applicationULP buffer to be completed to the application.ULP. An effective countermeasure is to create a high-water notification which alerts the applicationULP if there is more than a specified number of receive buffers "in process" (partially consumed, but not completed). The notification is generated when more than the specified number of buffers are in process simultaneously on a specific Stream (i.e., packets have started to arrive for the buffer, but the buffer has not yet been delivered to the ULP). A different countermeasure is for the RNIC Engine to provide the capability to limit the Remote PeerÆsPeer’s ability to consume receive buffers on a per Stream basis. Unfortunately this requires a large amount of state to be tracked in each RNIC on a per Stream basis. Thus, if an RNIC Engine provides the ability to share receive buffers across multiple Streams, the combination of the RNIC Engine and the Privileged Resource Manager MUST be able to detect if the Remote Peer is attempting to consume more than its fair share of resources so that the Local Peer can apply countermeasures to detect and prevent the attack. 22.214.171.124 Local PeerULP Attacking a Shared CQ DOS attacks against a Shared Completion Queue (CQ) can be caused by either the Local PeerPeer's ULP or the Remote Peer if either attempts to cause more completions than its fair share of the number of entries, thus potentially starving another unrelated StreamULP such that no Completion Queue entries are available. A Completion Queue entry can potentially be consumed by a completion from the Send Queue or a completion from the Receive Queue completion.Queue. In the former, the attacker is the Local Peer.Peer's ULP. In the later, the attacker is the Remote Peer. A form of attack can occur where the Local PeersPeer ULPs can consume resources on the CQ. A Local Peer ULP that is slow to free resources on the CQ by not reaping the completion status quickly enough could stall all other Local PeersPeer ULPs attempting to use that CQ. One of two countermeasures can be used to avoid this kind of attack. The first is to only share a CQ between StreamsULPs that share Partial Mutual Trust (i.e. Streams within the same Protection Domain).Trust. The other is to use a trusted Local Peer to act as a third party to free resources on the CQ and place the status in intermediate storage until the untrusted Local PeerULP reaps the status information. For these reasons, an RNIC MUST NOT enable sharing a CQ across StreamsULPs that belong to different Protection Domains. Additionally, an application SHOULD NOT share a CQ between Streams whichdo not share Partial Mutual Trust.partial mutual trust. 126.96.36.199 Remote Peer Attacking a Shared CQ For an overview of the Shared CQ attack model, see Section 188.8.131.52. <TBD: add text that says if not shared there is no security threat). If you get a CQ overflow it MUST NOT affect any resource outside the scope of the current Stream.> The Remote Peer can attack a CQ by consuming more than its fair share of CQ entries by using one of the following methods: * The ULP protocol allows the Remote Peer to reserve a specified number of CQ entries, possibly leaving insufficient entries for other Streams that are sharing the CQ. * If the Remote Peer or Local Peer (or both) can attack the CQ by overwhelming the CQ with completions, then completion processing on other Streams sharing that Completion Queue can be affected (e.g. the Completion Queue overflows and stops functioning). The first method of attack can be avoided if the ULP does not allow a Remote Peer to reserve CQ entries or there is a trusted intermediary such as a Privileged Resource Manager. Unfortunately it is often unrealistic to not allow a Remote Peer to reserve CQ entries - particularly if the number of completion entries is dependent on other ULP negotiated parameters, such as the amount of buffering required by the ULP. Thus an implementation MUST implement a Privileged Resource Manager to control the allocation of CQ entries. See Section 4.1 Components on page 910 for a definition of Privileged Resource Manager. One way that a Local or Remote Peer can attempt to overwhelm a CQ with completions is by sending minimum length RDMAP/DDP Messages to cause as many completions (receive completions for the Remote Peer, send completions for the Local Peer) per second as possible. If it is the Remote Peer attacking, and we assume that the Local Peer does not run out of receive buffers (if they do, then this is a different attack, documented in Section 184.108.40.206 Multiple Streams Sharing Receive Buffers on page 30), then it might be possible for the Remote Peer to consume more than its fair share of Completion Queue entries. Depending upon the CQ implementation, this could either cause the CQ to overflow (if it is not large enough to handle all of the completions generated) or for another Stream to not be able to generate CQ entries (if the RNIC had flow control on generation of CQ entries into the CQ). In either case, the CQ will stop functioning correctly and any Streams expecting completions on the CQ will stop functioning. This attack can occur regardless of whether all of the Streams associated with the CQ are in the same Protection Domain or are in different Protection Domains - the key issue is that the number of Completion Queue entries is less than the number of all outstanding operations that can cause a completion. The Local Peer can protect itself from this type of attack using either of the following methods: * Size the CQ to the appropriate level, as specified below (note that if the CQ currently exists, and it needs to be resized, resizing the CQ can fail, so the CQ resize should be done before sizing the Send Queue and Receive Queue on the Stream), OR * Grant fewer resources than the Remote Peer requested (not supplying the number of Receive Data Buffers requested). The proper sizing of the CQ is dependent on whether the Local Peer will post as many resources to the various queues as the size of the queue enables or not. If the Local Peer can be trusted to post a number of resources that is smaller than the size of the specific resourceÆsresource’s queue, then a correctly sized CQ means that the CQ is large enough to hold completion status for all of the outstanding Data Buffers (both send and receive buffers), or: CQ_MIN_SIZE = SUM(MaxPostedOnEachRQ) + SUM(MaxPostedOnEachSRQ) + SUM(MaxPostedOnEachSQ) Where: MaxPostedOnEachRQ = the maximum number of requests which can cause a completion that will be posted on a specific Receive Queue. MaxPostedOnEachSRQ = the maximum number of requests which can cause a completion that will be posted on a specific Shared Receive Queue. MaxPostedOnEachSQ = the maximum number of requests which can cause a completion that will be posted on a specific Send Queue. If the local peer must be able to completely fill the queues, or can not be trusted to observe a limit smaller than the queues, then the CQ must be sized to accommodate the maximum number of operations that it is possible to post at any one time. Thus the equation becomes: CQ_MIN_SIZE = SUM(SizeOfEachRQ) + SUM(SizeOfEachSRQ) + SUM(SizeOfEachSQ) Where: SizeOfEachRQ = the maximum number of requests which can cause a completion that can ever be posted on a specific Receive Queue. SizeOfEachSRQ = the maximum number of requests which can cause a completion that can ever be posted on a specific Shared Receive Queue. SizeOfEachSQ = the maximum number of requests which can cause a completion that can ever be posted on a specific Send Queue. Where MaxPosted*OnEach*Q and SizeOfEach*Q varies on a per Stream or per Shared Receive Queue basis. The Local PeerIf the ULP is sharing a CQ across multiple streams which do not share partial mutual trust, then the ULP MUST implement a mechanism to ensure that the Completion Queue can not overflow. Note that it is possible to share CQs even if the Remote Peers accessing the CQs are untrusted if either of the above two formulas are implemented. If the Local PeerULP can be trusted to not post more than MaxPostedOnEachRQ, MaxPostedOnEachSRQ, and MaxPostedOnEachSQ, then the first formula applies. If the Local PeerULP can not be trusted to obey the limit, then the second formula applies. 220.127.116.11 Attacking the RDMA Read Request Queue If RDMA Read Request Queue resources are pooled across multiple Streams, one attack is if the Local Peer attempts to unfairly allocate RDMA Read Request Queue resources for its Streams. For example, the Local Peer attempts to allocate all available resources on a specific RDMA Read Request Queue for its Streams, thereby denying the resource to applicationsULPs sharing the RDMA Read Request Queue. The same type of argument applies even if the RDMA Read Request is not shared - but a Local Peer attempts to allocate all of the RNICs resource when the queue is created. Thus access to interfaces that allocate RDMA Read Request Queue entries MUST be restricted to a trusted Local Peer, such as a Privileged Resource Manager. The Privileged Resource Manager SHOULD prevent a Local Peer from allocating more than its fair share of resources. Another form of attack is if the Remote Peer sends more RDMA Read Requests than the depth of the RDMA Read Request Queue at the Local Peer. If the RDMA Read Request Queue is a shared resource, this could corrupt the queue. If the queue is not shared, then the worst case is that the current Stream is disabled. One approach to solving the shared RDMA Read Request Queue would be to create thresholds, similar to those described in Section 18.104.22.168 Multiple Streams Sharing Receive Buffers on page 30. A simpler approach is to not share RDMA Read Request Queue resources amoung Streams or enforce hard limits of consumption per Stream. Thus RDMA Read Request Queue resource consumption MUST be controlled by the Privileged Resource Manager such that RDMAP/DDP Streams which do not share Partial Mutual Trust do not share RDMA Read Request Queue resources. A ULP SHOULD indicate to the Privileged Resource Manager when allocating a RDMA Read Request Queue whether or not it shares partial mutual trust with any other Stream(s).If the issue is a bug in the Remote PeerÆsPeer’s implementation, and not a malicious attack, the issue can be solved by requiring the Remote PeerÆsPeer’s RNIC to throttle RDMA Read Requests. By properly configuring the Stream at the Remote Peer through a trusted agent, the RNIC can be made to not transmit RDMA Read Requests that exceed the depth of the RDMA Read Request Queue at the Local Peer. If the Stream is correctly configured, and if the Remote Peer submits more requests than the Local PeerÆsPeer’s RDMA Read Request Queue can handle, the requests would be queued at the Remote PeerÆsPeer’s RNIC until previous requests complete. If the Remote PeerÆsPeer’s Stream is not configured correctly, the RDMAP Stream is terminated when more RDMA Read Requests arrive at the Local Peer than the Local Peer can handle (assuming the prior paragraphÆsparagraph’s recommendation is implemented). Thus an RNIC implementation MUSTSHOULD provide a mechanism to cap the number of outstanding RDMA Read Requests. The configuration of this limit is outside the scope of this specification. 7.5.3 Resource Consumption by Idle ApplicationsULPs The simplest form of a DOS attack given a fixed amount of resources is for the Remote Peer to create a RDMAP Stream to a Local Peer, and request dedicated resources then do no actual work. This allows the Remote Peer to be very light weight (i.e. only negotiate resources, but do no data transfer) and consumes a disproportionate amount of resources in the server. A general countermeasure for this style of attack is to monitor active RDMAP Streams and if resources are getting low, reap the resources from RDMAP Streams that are not transferring data and possibly terminate the Stream. This would presumably be under administrative control. Refer to Section 7.5.1 for the analysis and countermeasures for this style of attack on the following RNIC resources: Stream Context Memory, Page Translation Tables and STag namespace. Note that some RNIC resources are not at risk of this type of attack from a Remote Peer because an attack requires the Remote Peer to send messages in order to consume the resource. Receive Data Buffers, Completion Queue, and RDMA Read Request Queue resources are examples. These resources are, however, at risk from a Local Peer that attempts to allocate resources, then goes idle. This could also be created if the ULP negotiates the resource levels with the Remote Peer, which causes the Local Peer to consume resources, however the Remote Peer never sends data to consume them. The general countermeasure described in this section can be used to free resources allocated by an idle Local Peer. 7.5.4 Exercise of non-optimal code paths Another form of DOS attack is to attempt to exercise data paths that can consume a disproportionate amount of resources. An example might be if error cases are handled on a "slow path" (consuming either host or RNIC computational resources), and an attacker generates excessive numbers of errors in an attempt to consume these resources. Note that for most RDMAP or DDP errors, the attacking Stream will simply be torn down. Thus for this form of attack to be effective, the Remote Peer needs to exercise data paths which do not cause the Stream to be torn down. If an RNIC implementation contains "slow paths" which do not result in the tear down of the Stream, it is recommended that an implementation provide the ability to detect the above condition and allow an administrator to act, including potentially administratively tearing down the RDMAP Stream associated with the Stream exercising data paths consuming a disproportionate amount of resources. 7.5.5 Remote Invalidate an STag Shared on Multiple Streams If a Local Peer has enabled an STag for remote access, the Remote Peer could attempt to remote invalidate the STag by using the RDMAP Send with Invalidate or Send with SE and Invalidate Message. If the STag is only valid on the current Stream, then the only side effect is that the Remote Peer can no longer use the STag; thus there are no security issues. If the STag is valid across multiple Streams, then the Remote Peer can prevent other Streams from using that STag by using the remote invalidate functionality. Thus if RDDP Streams do not share Partial Mutual Trust (i.e. the Remote Peer may attempt to invalidate the STag prematurely), the applicationULP MUST NOT allowenable an STag towhich would be valid across multiple Streams. 7.6 Elevation of Privilege The RDMAP/DDP Security Architecture explicitly differentiates between three levels of privilege - Non-Privileged, Privileged, and the Privileged Resource Manager. If a Non-Privileged ApplicationULP is able to elevate its privilege level to a Privileged Application,ULP, then mapping a physical address list to an STag can provide local and remote access to any physical address location on the node. If a Privileged Mode ApplicationULP is able to promote itself to be a Resource Manager, then it is possible for it to perform denial of service type attacks where substantial amounts of local resources could be consumed. In general, elevation of privilege is a local implementation specific issue and thus outside the scope of this specification. <TBD: make more general and include authentication etc> There is one issue worth noting, however. If the RNIC implementation, by some insecure mechanism (or implementation defect), can enable a Remote Peer or un-trusted Local Peer to load firmware into the RNIC Engine, it is possible to use the RNIC to attack the host. Thus, an implementation MUST NOT enable firmware to be loaded on the RNIC Engine directly from a Remote Peer, unless the Remote Peer is properly authenticated (by a mechanism outside the scope of this specification. The mechanism presumably entails authenticating that the remote applicationULP has the right to perform the update), and the update is done via a secure protocol, such as IPsec (See Section 8 Security Services for RDMA and DDP on page 38). Further, an implementation MUST NOT allow a Non-Privileged Local Peer to update firmware in the RNIC Engine. 8 Security Services for RDMA and DDP RDMA and DDP are used to control, read and write data buffers over IP networks. Therefore, the control and the data packets of these protocols are vulnerable to the spoofing, tampering and information disclosure attacks listed in Section 7. Generally speaking, Stream confidentiality protects against eavesdropping. Stream and/or session authentication and integrity protection is a counter measurement against various spoofing and tampering attacks. The effectiveness of authentication and integrity against a specific attack, depend on whether the authentication is machine level authentication (as the one provided by IPsec and SSL), or ULP authentication. 8.1 Introduction to Security Options The following security services can be applied to an RDMAP/DDP Stream: 1. Session confidentiality - protects against eavesdropping (section 7.4.9). 2. Per-packet data source authentication - protects against the following spoofing attacks: network based impersonation (section 7.2.1), Stream hijacking (section 7.2.2), and man in the middle (section 7.2.3). 3. Per-packet integrity - protects against tampering done by network based modification of buffer content (section 7.3.4) 4. Packet sequencing - protects against replay attacks, which is a special case of the above tampering attack. If an RDMAP/DDP Stream may be subject to impersonation attacks, or Stream hijacking attacks, it is recommended that the Stream be authenticated, integrity protected, and protected from replay attacks; it MAYmay use confidentiality protection to protect from eavesdropping (in case the RDMAP/DDP Stream traverses a public network). Both IPsec and SSL are capable of providing the above security services for IP and TCP traffic respectively. ULP protocols are able to provide only part of the above security services. The next sections describe the different security options. 8.1.1 Introduction to IPsec IPsec is a protocol suite which is used to secure communication at the network layer between two peers. The IPsec protocol suite is specified within the IP Security Architecture [RFC2401], IKE [RFC2409], IPsec Authentication Header (AH) [RFC2402] and IPsec Encapsulating Security Payload (ESP) [RFC2406] documents. IKE is the key management protocol while AH and ESP are used to protect IP traffic. An IPsec SA is a one-way security association, uniquely identified by the 3-tuple: Security Parameter Index (SPI), protocol (ESP/AH) and destination IP address. The parameters for an IPsec security association are typically established by a key management protocol. These include the encapsulation mode, encapsulation type, session keys and SPI values. IKE is a two phase negotiation protocol based on the modular exchange of messages defined by ISAKMP [RFC2408],and the IP Security Domain of Interpretation (DOI) [RFC2407]. IKE has two phases, and accomplishes the following functions: 1. Protected cipher suite and options negotiation - using keyed MACs and encryption and anti-replay mechanisms. 2. Master key generation - via Diffie-Hellman calculations. 3. Authentication of end-points (usually machine level authentication). 4. IPsec SA management (selector negotiation, options negotiation, create, delete, and rekeying). Items 1 through 3 are accomplished in IKE Phase 1, while item 4 is handled in IKE Phase 2. IKE phase 1 defines four authentication methods; three of them require both sides to have certified signature or encryption public keys; the forth require the side to exchange out-of-band a secret random string - called pre-shared-secret (PSS). An IKE Phase 2 negotiation is performed to establish both an inbound and an outbound IPsec SA. The traffic to be protected by an IPsec SA is determined by a selector which has been proposed by the IKE initiator and accepted by the IKE Responder. The IPsec SA selector can be a "filter" or traffic classifier, defined as the 5-tuple: <Source IP address, Destination IP address, transport protocol (e.g. UDP/SCTP/TCP), Source port, Destination port>. The successful establishment of a IKE Phase-2 SA results in the creation of two uni-directional IPsec SAs fully qualified by the tuple <Protocol (ESP/AH), destination address, SPI>. The session keys for each IPsec SA are derived from a master key, typically via a MODP Diffie-Hellman computation. Rekeying of an existing IPsec SA pair is accomplished by creating two new IPsec SAs, making them active, and then optionally deleting the older IPsec SA pair. Typically the new outbound SA is used immediately, and the old inbound SA is left active to receive packets for some locally defined time, perhaps 30 seconds or 1 minute. Optionally, rekeying can use Diffie-Helman for keying material generation. 8.1.2 Introduction to SSL Limitations on RDMAP SSL and TLS [RFC 2246] provide Stream authentication, integrity and confidentiality for TCP based applications.ULPs. SSL supports one- wayone-way (server only) or mutual certificates based authentication. There are at least two limitations that make SSL underneath RDMAP less appropriate then IPsec for DDP/RDMA security: 1. The maximum length supported by the TLS record layer protocol is 2^14 bytes - longer packets must be fragmented (as a comparison, the maximal length of an IPsec packet is determined by the maximum length of an IP packet). 2. SSL is a connection oriented protocol. If a stream cipher or block cipher in CBC mode is used for bulk encryption, then a packet can be decrypted only after all the packets preceding it have already arrived. If SSL is used to protect DDP/RDMA traffic, then SSL must gather all out-of-order packets before RDMAP/DDP can place them into the ULP buffer, which might cause a significant decrease in its efficiency. If SSL is layered on top of RDMAP or DDP, SSL does not protect the RDMAP and/or DDP headers. Thus a man-in-the-middle attack can still occur by modifying the RDMAP/DDP header to incorrectly place the data into the wrong buffer, thus effectively corrupting the data stream. 8.1.3 ApplicationsULPs Which Provide Security ApplicationsULPs which provide integrated security but wish to leverage lower-layer protocol security should be aware of security concerns around correlating a specific channelÆschannel’s security mechanisms to the authentication performed by the application.ULP. See [NFSv4CHANNEL] for additional information on a promising approach called "channel binding". From [NFSv4CHANNEL]: The"The concept of channel bindings allows applications to prove that the end-points of two secure channels at different network layers are the same by binding authentication at one channel to the session protection at the other channel. The use of channel bindings allows applications to delegate session protection to lower layers, which may significantly improve performance for some applications.applications." 8.2 Requirements for IPsec Encapsulation of DDP The IP Storage working group has spent significant time and effort to define the normative IPsec requirements for IP Storage [RFC3723]. Portions of that specification are applicable to a wide variety of protocols, including the RDDP protocol suite. In order to not replicate this effort, an RNIC implementation MUST follow the requirements defined in RFC3723 Section 2.3 and Section 5, including the associated normative references for those sections. Additionally, since IPsec acceleration hardware may only be able to handle a limited number of active IKE Phase 2 SAs, Phase 2 delete messages may be sent for idle SAs, as a means of keeping the number of active Phase 2 SAs to a minimum. The receipt of an IKE Phase 2 delete message MUST NOT be interpreted as a reason for tearing down an DDP/RDMA Stream. Rather, it is preferable to leave the Stream up, and if additional traffic is sent on it, to bring up another IKE Phase 2 SA to protect it. This avoids the potential for continually bringing Streams up and down. Note that there are serious security issues if IPsec is not implemented end-to-end. For example, if IPsec is implemented as a tunnel in the middle of the network, any hosts between the peer and the IPsec tunneling device can freely attack the unprotected Stream. 9 Security considerations This entire specification is focused on security considerations. 10 References 10.1 Normative References [RFC2828] Shirley, R., "Internet Security Glossary", FYI 36, RFC 2828, May 2000. [DDP] Shah, H., J. Pinkerton, R.Recio, and P. Culley, "Direct Data Placement over Reliable Transports", Internet-Draft draft-ietf-rddp-ddp-01.txt, February 2003. [RDMAP] Recio, R., P. Culley, D. Garcia, J. Hilland, "An RDMA Protocol Specification", Internet-Draft draft-ietf-rddp- rdmap-01.txt, February 2003. [RFC3723] Aboba B., et al, "Securing Block Storage Protocols over IP", Internet draft (work in progress), RFC3723, April 2004. [SCTP] R. Stewart et al., "Stream Control Transmission Protocol", RFC 2960, October 2000. [TCP] Postel, J., "Transmission Control Protocol - DARPA Internet Program Protocol Specification", RFC 793, September 1981. 10.2 Informative References [IPv6-Trust] Nikander, P., J.Kempf, E. Nordmark, "IPv6 Neighbor Discovery Trust Models and threats", Internet-Draft draft- ietf-send-psreq-01.txt, January 2003. [NFSv4CHANNEL] Williams, N., "On the Use of Channel Bindings to Secure Channels", Internet-Draft draft-ietf-nfsv4-channel- bindings-02.txt, July 2004. 11 Appendix A: ImplementingULP Issues for RDDP Client/Server Protocols This section is a normative appendix to the specification that is focused on client/server applicationULP implementation requirements to ensure a secure server implementation. The prior sections outlined specific attacks and their countermeasures. This section summarizes the attacks and countermeasures that have been defined in the prior section which are applicable to creation of a secure applicationULP (e.g. application) server. An applicationA ULP server is defined as an applicationa ULP which must be able to communicate with many clients which do not trust each other and ensure that each client can not attack another client through server interactions. Further, the server may wish to use multiple Streams to communicate with a specific client, and those Streams may share mutual trust. Note that this section assumes a compliant RNIC and Privileged Resource Manager implementation - thus it focuses specifically on applicationULP server (i.e. ULP)(e.g. application) implementation issues. All of the prior section's details on attacks and countermeasures apply to the server. In some cases normative SHOULD statements for the ULP in the main body of this document are made MUST statements for the ULP because the operating conditions can be refined to make the motives for a SHOULD inapplicable. If a prior SHOULD is changed to a MUST in this section, it is explicitly noted. The following list summarizes the relevent attacks that clients can mount on the shared server, by re-stating the previous normative statements to be client/server specific: * Spoofing * Sections 7.2.1 to 7.2.3. For protection against many forms of spoofing attacks, enable IPsec. * Section 7.2.4 Using an STag on a Different Stream on page 23. To ensure that one client can not access another client's data via use of the other client's STag, the server applicationULP MUST either scope an STag to a single Stream or use a Protection Domain per client. If a single client has multiple streams that share Partial Mutual Trust, then the STag can be shared between the associated Streams by using a single Protection Domain amoung the associated Streams (see section 8.1.3 Applications Which Provide Security on page 40for additional issues). To prevent unintended sharing of STags within the associated Streams, a server applicationULP SHOULD use STags in such a fashion that it is difficult to predict the next allocated STag number. * Tampering * 7.3.2 Modifying a Buffer After Indication on page 25. Before the server applicationULP operates on a buffer that was written using an RDMA Write or RDMA Read, the server MUST ensure the the buffer can no longer be modified by invalidating the STag for remote access (note that this is stronger than the SHOULD in section 7.3.2). This can either be done explicitly by revoking remote access rights for the STag when the client indicates the operation has completed, or by checking to make sure the client Invalidated the STag through the RDMAP Invalidate capability, and if it did not, the Local Peer then explicitly revokes the STag remote access rights. * Information Disclosure * 7.4.2 Using RDMA Read to Access Stale Data on page 26. A server applicationULP MUST (this is stronger than the SHOULD in section 7.4.2) ensure that no stale data is contained in a buffer before remote read access rights are granted to a client (this can be done by zeroing the contents of the memory, for example). * 7.4.3 Accessing a Buffer After the Transfer on page 26. This mitigation is already covered by section 7.3.2 (above). * 7.4.4 Accessing Unintended Data With a Valid STag on page 26. The applicationULP server MUST set the base and bounds of the buffer when the STag is initialized to expose only the data to be retrieved. * 7.4.5 RDMA Read into an RDMA Write Buffer on page 27. If a server only intends a buffer to be exposed for remote write access, it MUST set the access rights to the buffer to only enable remote write access. * 7.4.6 Using Multiple STags Which Alias to the Same Buffer on page 27. The requirement in section 7.2.4 (above) mitigates this attack. A buffer is exposed to only one client at a time to ensure that no information disclosure or information tampering occurs between peers. * 7.4.9 Network based eavesdropping on page 28. Enable IPsec if this threat is a concern. * Denial of Service * 22.214.171.124 Multiple Streams Sharing Receive Buffers on page 30. ApplicationULP memory footprint size can be important for some server applications.ULPs. If a server applicationULP is expecting significant network traffic from multiple clients, using a receive buffer queue per Stream where there is a large number of Streams can consume substantial amounts of memory. Thus a receive queue that can be shared by multiple Streams is attractive. However, because of the attacks outlined in this section, sharing a single receive queue between multiple clients MUST ONLYonly be done if the mechanism to ensure one client can't consume too many receive buffers is enabled. For multiple Streams within a single client applicationULP (which presumably shared partial mutual trust) this added overhead does not have to be enabled. * 126.96.36.199 Local PeerULP Attacking a Shared CQ on page 31. <TBD>The normative mitigations were * RNIC MUST NOT enable sharing a CQ across Streams that belong to different Protection Domains. * An applicationA ULP SHOULD NOT share a CQ between Streams which do not share Partial Mutual Trust. Because the attack is a local server applicationULP attacking another server application,ULP, * 188.8.131.52 Remote Peer Attacking a Shared CQ on page 32. There are two mitigations specified in this section - one requires a worst-case size of the CQ, and can be implemented entirely within the Privileged Resource Manager. The second approach requires cooperation with the local applicationULP server (to not post too many buffers), and enables a smaller CQ to be used. In some server environments, partial trust of the server applicationULP (but not the clients) is acceptable, thus the smaller CQ fully mitigates the remote attacker. In other environments, the local server applicationULP could also contain untrusted elements which can attack the local machine (or have bugs). In those environments, the worst-case size of the CQ must be used. * 184.108.40.206 The section requires a Privileged Resource Manager to not enable sharing of RDMA Read Request Queues across multiple Streams that do not share partial mutual trust. However, because the server applicationULP knows best which of its Streams share partial mutual trust, this requirement can be reflected back to the application.ULP. The ULP (i.e. server) requirement is that it MUST NOT request RDMA Read Request Queues to be shared between applicationsULPs which do not have partial mutual trust. * 7.5.5 Remote Invalidate an STag Shared on Multiple Streams on page 37. This mitigation is already covered by section 7.3.2 (above). 12 Appendix B: Summary of RNIC and ULP Implementation Requirements Below is a summary of implementation requirements for the RNIC: * 7.3.1 Buffer Overrun - RDMA Write or Read Response * 7.4.8 Controlling Access to PTT & STag Mapping * 7.5.1 RNIC Resource Consumption * 220.127.116.11 Multiple Streams Sharing Receive Buffers * 18.104.22.168 Local PeerULP Attacking a Shared CQ * 22.214.171.124 Remote Peer Attacking a Shared CQ * 126.96.36.199 Attacking the RDMA Read Request Queue * 7.5.4 Exercise of non-optimal code paths * 7.6 Elevation of Privilege * 8.2 Requirements for IPsec Encapsulation of DDP Below is a summary of implementation requirements for the applicationULP above the RNIC: * 7.2.4 Using an STag on a Different Stream * 7.3.2 Modifying a Buffer After Indication * 7.4.2 Using RDMA Read to Access Stale Data * 7.4.3 Accessing a Buffer After the Transfer * 7.4.4 Accessing Unintended Data With a Valid STag * 7.4.5 RDMA Read into an RDMA Write Buffer * 7.4.6 Using Multiple STags Which Alias to the Same Buffer * 7.4.9 Network based eavesdropping * 188.8.131.52 Local PeerULP Attacking a Shared CQ * 7.5.5 Remote Invalidate an STag Shared on Multiple Streams 13 Appendix C: Partial Trust Taxonomy Partial Trust is defined as when one party is willing to assume that another party will refrain from a specific attack or set of attacks, the parties are said to be in a state of Partial Trust. Note that the partially trusted peer may attempt a different set of attacks. This may be appropriate for many applicationsULPs where any adverse effects of the betrayal is easily confined and does not place other clients or applicationsULPs at risk. The Trust Models described in this section have three primary distinguishing characteristics. The Trust Model refers to a Local Peer and Remote Peer, which are the local and remote applicationULP instances communicating via RDMA/DDP. * Local Resource Sharing (yes/no) - When local resources are shared, they are shared across a grouping of RDMAP/DDP Streams. If local resources are not shared, the resources are dedicated on a per Stream basis. Resources are defined in Section 4.2 - Resources on page 11. The advantage of not sharing resources between Streams is that it reduces the types of attacks that are possible. The disadvantage is that applicationsULPs might run out of resources. * Local Partial Trust (yes/no) - Local Partial Trust is determined based on whether the local grouping of RDMAP/DDP Streams (which typically equates to one applicationULP or group of applications)ULPs) mutually trust each other to not perform a specific set of attacks. * Remote Partial Trust (yes/no) - The Remote Partial Trust level is determined based on whether the Local Peer of a specific RDMAP/DDP Stream partially trusts the Remote Peer of the Stream (see the definition of Partial Trust in Section 3 Introduction). Not all of the combinations of the trust characteristics are expected to be used by applications.ULPs. This paper specifically analyzes five applicationULP Trust Models that are expected to be in common use. The Trust Models are as follows: * NS-NT - Non-Shared Local Resources, no Local Trust, no Remote Trust - typically a server applicationULP that wants to run in the safest mode possible. All attack mitigations are in place to ensure robust operation. * NS-RT - Non-Shared Local Resources, no Local Trust, Remote Partial Trust - typically a peer-to-peer application,ULP, which has, by some method outside of the scope of this specification, authenticated the Remote Peer. Note that unless some form of key based authentication is used on a per RDMA/DDP Stream basis, it may not be possible be possible for man-in-the-middle attacks to occur. See section 8, Security Services for RDMA and DDP on page 38. * S-NT - Shared Local Resources, no Local Trust, no Remote Trust - typically a server applicationULP that runs in an untrusted environment where the amount of resources required is either too large or too dynamic to dedicate for each RDMAP/DDP Stream. * S-LT - Shared Local Resources, Local Partial Trust, no Remote Trust - typically an application,a ULP, which provides a session layer and uses multiple Streams, to provide additional throughput or fail-over capabilities. All of the Streams within the local applicationULP partially trust each other, but do not trust the Remote Peer. This trust model may be appropriate for embedded environments. * S-T - Shared Local Resources, Local Partial Trust, Remote Partial Trust - typically a distributed application, such as a distributed database application or a High Performance Computer (HPC) application, which is intended to run on a cluster. Due to extreme resource and performance requirements, the application typically authenticates with all of its peers and then runs in a highly trusted environment. The application peers are all in a single application fault domain and depend on one another to be well-behaved when accessing data structures. If a trusted Remote Peer has an implementation defect that results in poor behavior, the entire application could be corrupted. Models NS-NT and S-NT above are typical for Internet networking - neither Local Peers nor the Remote Peer is trusted. Sometimes optimizations can be done that enable sharing of Page Translation Tables across multiple Local Peers, thus Model S-LT can be advantageous. Model S-T is typically used when resource scaling across a large parallel applicationULP makes it infeasible to use any other model. Resource scaling issues can either be due to performance around scaling or because there simply are not enough resources. Model NS-RT is probably the least likely model to be used, but is presented for completeness. 14 AuthorÆsAuthor’s Addresses James Pinkerton Microsoft Corporation One Microsoft Way Redmond, WA. 98052 USA Phone: +1 (425) 705-5442 Email: email@example.com Ellen Deleganes Intel Corporation MS JF5-355 2111 NE 25th Ave. Hillsboro, OR 97124 USA Phone: +1 (503) 712-4173 Email: firstname.lastname@example.org Sara Bitan Microsoft Corporation Email: email@example.com 15 Acknowledgments Allyn Romanow Cisco Systems 170 W Tasman Drive San Jose, CA 95134 USA Phone: +1 408 525 8836 Email: firstname.lastname@example.org Catherine Meadows Naval Research Laboratory Code 5543 Washington, DC 20375 Email: email@example.com Patricia Thaler Agilent Technologies, Inc. 1101 Creekside Ridge Drive, #100 M/S-RG10 Roseville, CA 95678 Phone: +1-916-788-5662 email: firstname.lastname@example.org James Livingston NEC Solutions (America), Inc. 7525 166th Ave. N.E., Suite D210 Redmond, WA 98052-7811 Phone: +1 (425) 897-2033 Email: email@example.com John Carrier Adaptec, Inc. 691 S. Milpitas Blvd. Milpitas, CA 95035 USA Phone: +1 (360) 378-8526 Email: firstname.lastname@example.org Caitlin Bestler Email: email@example.com Bernard Aboba Microsoft Corporation One Microsoft Way Redmond, WA. 98052 USA Phone: +1 (425) 706-6606 Email: firstname.lastname@example.org 16 Full Copyright Statement Copyright (C) The Internet Society (2001). All Rights Reserved. This document and translations of it may be copied and furnished to others, and derivative works that comment on or otherwise explain it or assist in its implementation may be prepared, copied, published and distributed, in whole or in part, without restriction of any kind, provided that the above copyright notice and this paragraph are included on all such copies and derivative works. However, this document itself may not be modified in any way, such as by removing the copyright notice or references to the Internet Society or other Internet organizations, except as needed for the purpose of developing Internet standards in which case the procedures for copyrights defined in the Internet Standards process must be followed, or as required to translate it into languages other than English. The limited permissions granted above are perpetual and will not be revoked by the Internet Society or its successors or assigns. This document and the information contained herein is provided on an "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Funding for the RFC Editor function is currently provided by the Internet Society.