The present invention is in the field of virtualizing intelligent network interface circuitry (NIC) connectable to a host computer executing multiple functions or multiple Operating System (OS) instances simultaneously and, in particular, relates to the virtualization of the operation of single NIC as viewed by the multiple functions or multiple OS instances.
In some sense, network interface circuitry to interface a host computer, executing multiple OS's, to a network is known but somewhat limited. For example, for ingress communication (network to host), a Media Access Controller (MAC) within the NIC can be set to accept all layer-2 network packets that arrive at the NIC from the network (promiscuous mode operation) and a network protocol stack within a virtual operating system (VOS), such as provided by Vmware or Xen, on a host computer can then de-multiplex the layer-2 network packets arriving at the host to separate network processing stacks within multiple guest operating systems (guest-OS's) executing on top of the VOS. In other words, the VOS protocol stack forwards unicast packets destined for one or more guest-OS's, and also multicast and broadcast packets, but drops unicast packets that are not destined for any of the guest-OS's. For egress communication, the guest-OS's accomplish sending packets by copying from memory associated with a particular guest-OS to memory associated with the VOS generally, and the VOS then operates to send the packets.
With the recent advances in networking speed, it has become possible to foresee a “unified wire” approach to connectivity. From a network connectivity perspective, it is typical to have in one computer system different NICs for different applications. For example, it is usual to need an Ethernet NIC for general network and Internet connectivity, a Fibre Channel NIC for storage connectivity and a specialized high speed interconnect NIC for high performance computing connectivity. In particular, the Ethernet technology has recently reached speeds high enough to enable the support of these different applications simultaneously over one “multi-function” NIC. The benefits of unifying the “wire” are numerous, including but not limited to the single NIC and single network cost savings as well as the simplified management. However, in-order to preserve the useful characteristics and match the performance of specialized storage and high performance computing networks, it is desirable to enhance the capabilities of the unified wire, for example, to support the storage protocols iSCSI and the Network File System (NFS), and Remote Direct Memory Access (RDMA). In addition, it is needed to provide control on the resource usage of each application type within the unified wire.
For both the egress and ingress direction, the NIC can optionally offload the processing of some network protocols. For example, the protocol offload processing may be according to a Transmission Control Protocol (TCP), whereas communication across the network (layer-2) may be via high-speed Ethernet, such as 10 Gbps Ethernet.
The present invention is in the field of virtualization of network interface circuitry (e.g., as embodied within network interface cards and/or network interface controllers) configurable to connect a network and a host computer having multiple functions or OS's (referred to as guest-functions or guest-OS's) executing simultaneously. In particular, in one aspect, the present invention relates to the virtualization of the operation of the NIC, such that the NIC can be used simultaneously by multiple such guests in such a manner that memory references due to the network traffic originating from and destined to a particular guest are segregated from other network traffic. In one aspect, the Media Access Controller (MAC) within the NIC operates such that the MAC only accepts layer-2 network packets with payload destined to one of the different guests, and the MAC is not required to operate in promiscuous mode, accepting all incoming packets in order to be able to implement the virtualization features. In another aspect, the NIC is equipped with traffic management which provides control on the resources used by each guest.
We now discuss a shortcoming with the approach described in the Background for handling layer-2 network traffic destined for multiple guest-functions or multiple instances of guest-OS's operating on a host computer executing on top of a virtualization layer (also called virtual operating system (VOS) when multiple instances of guest-OS's are present). The “guest function” corresponds to a particular protocol stack instance. This can for example refer to the TCP/IP networking stack, and the SCSI storage stack used or usable from within the same instance of a guest-OS. In this case the TCP/IP networking stack would represent one “guest function” and the SCSI storage stack another “guest function”. This is just one example of the usage of multiple “guest functions” within a particular instance of a “guest OS.” In a case where there is only one protocol stack used or usable from within an instance of a guest-OS, then the instance of the guest-OS is considered to have a single guest-function associated with it, which is the instance of the guest-OS itself. In addition, it should be noted that some or all of the instances of guest-OS's may be separate instances of the same OS, such as, for example, Windows or Linux.
In particular, the approach described in the Background places a computational load on the virtualization layer or VOS to demultiplex the network traffic, and this load can be quite substantial for high-speed network interfaces. Another shortcoming is the overhead associated with copying data from memory associated with the virtualization layer or VOS to memory associated with the destination guest-functions or instances of guest-OS's in all cases for ingress packets, and copying data from memory associated with a particular origin guest-OS instance to memory associated with the VOS in the egress direction.
It is thus desirable to reduce the computational load and overhead on the virtualization layer or VOS associated with processing ingress packets destined to multiple guest functions or instances of guest OS's. It is also desirable to protect against each guest-function or guest-OS instance corrupting data structures for packet ingress associated with the other guest-functions or guest-OS's. Conventionally, data structures for direct memory access (DMA) such as command queues, free-list pages, and completion queues are shared for network traffic destined to the multiple guest-functions and instances of guest-OS's, and memory corruption by one guest-OS instance thus necessarily affects the other guest-functions and instances of guest-OS's. In the remainder of this discussion, we will consider the example where multiple instances of OS's are running atop a VOS. A similar discussion applies to the case where multiple functions are supported atop a virtualization layer or a combination of multiple functions and multiple instances of OS's are in use.
Specifically, referring now to
In accordance with one aspect described here, the promiscuous mode of the MAC (described in the Background) is not used. In accordance with another aspect, memory references resulting from network traffic originating from, and destined to, each guest-OS instance is segregated from memory references resulting from network traffic from/to the other instances of guest OS's. Having the NIC not be in promiscuous mode has an advantage of decreasing the ingress packet traffic (between the NIC and the host) and thereby minimizing or eliminating the overhead of the host processing packets not destined for any of the instances of guest-OS's. Memory reference safety may be accomplished, for example, by segregating the DMA command queues, gather-lists, free-lists, and completion queues used by each guest-OS instance.
Copying and delivery of multicast and broadcast network traffic to the instances of guest OS's can optionally be handled or assisted by the NIC, so that all the processing of network traffic is offloaded to the NIC. In addition, direct data placement (DDP) can be accomplished safely by having the VOS validate the creation of the DDP regions by the instances of guest-OS's. As mentioned above, each of the instances of guest-OS's uses a separate data structure for DMA transfers (e.g., referred to here as command queue 200, gather-list 210, free list 220, direct-data region 230, and response queue 240).
We first discuss the egress (host to network) data path. Referring specifically to
For the ingress (network to host) data path, the NIC 150 either directly writes payload data into the direct data 230 region for the destination guest-OS instance, or writes payload data to the scatter-list 220 for the guest-OS instance indicated in one or more entries of the scatter-list 220 for the guest-OS instance. In the direct data 230 case, placement information (i.e., mapping) for the guest-OS is previously communicated to the NIC 150, for example as part of setting up a remote DMA transfer or to enable direct data placement of socket API read( ) response data. In the free list case, the location of the scatter-list pages for the guest-OS instance has been previously communicated to the NIC 150, e.g., via DMA reads. Finally, the response queue 240 for the guest-OS instance is typically also written for ingress transfers for the guest-OS instance to indicate that a particular entry has been placed in either a free-list 220 entry or entries, or that payload data has been directly placed in the direct data 230 region for the guest-OS instance.
In some examples, the NIC 150 is configured to offload protocol processing for some protocols. For example, the NIC 150 may have capability to offload processing for the TCP protocol. In another example, as previously noted, the NIC 150 may have capability to offload protocols at layers above TCP, such as iSCSI and NFS.
As a result of the virtualization capabilities of the NIC 150, the memory references due to egress network traffic originating from a particular guest-OS instance are segregated in one or more command queues 200, and one or more gather-list 210 data structures on the host computer may also be segregated within the NIC. In addition, segregated response queues 240 may be used. In some examples, the egress traffic is segregated within the NIC by mapping packets from a particular guest-OS instance to particular queues within an egress packet scheduler within the NIC. However, in some examples, the egress packet traffic corresponding to different instances of guest-OS's is interleaved through the NIC.
Furthermore, the virtualization configuration of the NIC 150 is such that memory references due to ingress network traffic arriving at the NIC from the network are segregated within the NIC 150, and are steered to separate segregated scatter-lists 220, direct data region 230, and response queue 240, corresponding to a particular guest-OS instance 110120130.
In sum, the just-described mode of mode of operation is referred to by us as multiple guest-functions or instances of guest-OS's running securely on top of a virtualized NIC. For example, multiple instances of Windows or Linux may be running on top of a VOS such as provided by VMware or Xen. In another example, iSCSI, RDMA and TCP/IP sockets may be running on top of a virtualization layer. Each instance of the guest-function and guest-OS broadly sees itself running on its own (virtual) NIC, while in effect it is running along with other functions and instances of guest OS's on top of a single virtualized NIC. The operation is referred to as “secure” because each guest-function and guest-OS instance has its own data structures in the host memory and one guest-function or guest-OS instance therefore will not corrupt the memory associated with the operation of the other guest-function and guest-OS instances. Because the different guest-functions and instances of guest-OS's use different data structures, a particular guest function or guest-OS instance does not have access permission to the physical memory in use by the other guest functions and instances of guest-OS's.
Having generally described virtualization of a NIC, we now describe three specific approaches to achieving virtualization of a NIC.
In accordance with the first approach, the virtual NIC's are operated in what is referred to as “promiscuous mode,” such that all ingress Ethernet packets are forwarded to the virtualization layer or VOS. For example, an incoming packet with a destination MAC Address that does not match any of the local MAC addresses will not be dropped. Instead, the incoming packet will reach the virtualization layer or VOS. The virtualization layer or VOS may then make a decision to drop packets that have destination MAC addresses not corresponding to any of the guest-functions and instances of guest-OS's and also drop packets with a matching destination MAC address but a non-matching destination IP address. The virtualization layer or VOS also processes broadcast and multicast packets. Note that the guest functions and instances of guest-OS's may or may not utilize the same local MAC address.
In contrast to the first approach, the second approach employs destination address filtering in the NIC, such that only those packets having an address corresponding to one of the guest-OS's are accepted by the NIC and forwarded to the VOS. The multicast packets are typically also filtered, and a packet that is not destined for any of the instances of guest-OS's is dropped. The broadcast packets are typically delivered to all guest functions and instances of guest-OS's. Note that the destination address used in this method may be the MAC address or the IP address. Therefore, the virtualization layer or VOS matches the local MAC or Internet Protocol address (LIP) for the guest functions or instances of guest-OS's in this approach and processes broadcast and multicast packets.
The third approach also employs destination address filters, but uses an indirection-map to map the matching packets to a guest-function or guest-OS instance via an index (sometimes referred to in this description as the “guest_idx” and/or “steering index”). For example, the MAC address of an ingress Ethernet packet is matched to a guest_idx using a lookup table. If the packet encapsulates an IP packet, the destination IP address in the IP packet is then compared to the IP address (one or more) associated with the particular guest-OS instance having the particular MAC address index (guest_idx steering index). Packets that do not have a valid (guest_idx, LIP) tuple are dropped in the NIC.
In this third approach, filtering is achieved by parsing the Ethernet packets and applying a filtering rule associated with the guest_idx (that is part of the tuple that indicates if the packet should be accepted or dropped). In a NIC that implements protocol offload, this approach accommodates offloaded protocols by adding the guest_idx to the tuple that typically includes the 4-tuple (LIP,LP,FIP,FP) for either the TCP or UDP protocols. As an example, the resulting 5-tuple would be used as an identifier of a particular TCP connection or UDP stream.
A step in the virtualization with this third approach employs the guest_idx steering index to steer the incoming packet directly, without the intervention of the virtualization layer or VOS, to the appropriate destination guest-function or guest-OS instance. For this purpose, the NIC uses the guest_idx steering index, optionally along with other fields of the packet or other pertinent information or criteria, to select a queue triplet (e.g., send-queue/response-queue/completion-queues for RDMA/iWARP traffic, or command-queue/free-list-queue/response-queue for other traffic) corresponding to the appropriate guest-function or guest-OS instance. The selection step may further involve the determination of a particular CPU number on which the received packet is to be processed. Using a target CPU number, the packet processing load may be distributed among potentially multiple CPUs on the system.
Two options with this third approach are now described to process multicast and broadcast packets. The distribution of these packets can either be performed by the NIC that receives the multicast and broadcast traffic or, alternately, these packets can be sent to the virtualization layer or VOS for processing. With the latter alternative, the packets may be mapped to one or more specific command-queue/free-list-queue/response-queue triplets, which facilitates processing of the packets by the virtualization layer or VOS. A benefit of the first option is that broadcast and multicast packets may be processed without involvement of the virtualization layer or VOS, but this is at the cost of multiple copies of the packet being provided over the host (e.g., PCI) bus. A decision as to which option to utilize may be based, for example, on a particular context and usage scenario.
In one example for processing multicast Ethernet packets, the Ethernet destination address is provided as input to a hashing function, the result of which is used to determine if the packet is to be accepted or dropped by the NIC. The result of the multicast address hash is compared to the allowed hash values, which are provided to the NIC by the virtualization layer or VOS on behalf of the guest functions or instances of guest-OS's. This results in a partial (imperfect) filtering of undesired multicast traffic, since the filtering criterion is not exact. In other words, the exact MAC address matching criteria is replaced by a hash of the multicast, and if a match to a hash table is produced within the Ethernet media access control block of the NIC, the packet is accepted. The packet may then be multicast/broadcast to the different guest-functions and guest-OS's by an on-chip module that has stored a table with information about which guest-function or guest-OS subscribes to which multicast. The NIC can, based on this determination, either multicast the packet to the appropriate guest functions or instances of guest-OS's, or deliver the packet to the virtualization layer or VOS that performs the multicast/broadcast, possibly based on an indication of this determination provided by the NIC along with the packet.
A guest function or OS may be configured to operate its virtual NIC(s) in promiscuous mode, while other virtual NIC's are independently configurable. Promiscuous mode operation may be useful in certain circumstances, such as for troubleshooting network problems. It is therefore useful for the NIC to support the promiscuous mode operation on a per-virtual NIC basis. In this case, the NIC can relax the filtering rule used for the particular guest-OS instance that is configured to operate the virtual NIC in this mode. As a result, the particular guest-function or guest-OS instance is provided with copies of all the packets received, even those that are determined to be destined to other guest-functions and instances of guest-OS's, regardless of the destination MAC address.
The MAC address is 48-bits wide for Ethernet, and the size of the index is at least the logarithm of the maximum number of supported addresses in the filtering database. For example if 65536, 1024, 256, or 8 addresses are the maximum number of supported addresses in the database, the size of the index is at least 16 bits wide, 10 bits, 8 bits, and 3 bits wide, respectively. In some examples, for efficiency reasons, the steering index is used within the processing path rather than the MAC address itself. The steering index, described below, is used within the processing path to steer a packet 54 to the correct DMA scatter-list, and response queue. If the packet is indicated as being multicast, then it may be handled as discussed above (according to a hash table).
Turning now to
It is noted that the arbiter 102 is a feature of the particular flow processor architecture of the
The protocol processing block 107 includes a lookup block 108 that locates the state for an offloaded protocol such as TCP and that is additionally used to filter packets (e.g., obtain filtering rules) that, for example, do not have a destination IP address that is consistent with the guest_idx steering index. A packet is identified by the header or headers that the packet contains. As an example, the headers for Ethernet packets contain at least a protocol stack layer-2 Ethernet packet, and when the Ethernet packet encapsulates an IP packet, a packet will also contain a layer-3 IP header, and when the IP header encapsulates a layer-4 TCP (or UDP) protocol, it will also contain a TCP (UDP) header. For a TCP packet, a 4-tuple including a source and destination IP address and a source and destination port numbers is said to uniquely identify a point-to-point connection that uses the protocol.
For offloaded connections, the lookup minimally considers the 4-tuple information and the Ethernet address lookup index guest_idx steering index corresponding to the destination Ethernet MAC address. In addition, the lookup may consider information about the Virtual Local Area Network (VLAN) to which the packet belongs, when VLAN's are being used, and perhaps the NIC port on which the packet arrived. In some examples, the lookup block 108 operates to match the protocol header to an internal identification (“tid,” used by the interface device and the host) corresponding to a particular protocol or filtering rule Control Block (CB). In the
The lookup block 108 then provides the tid, received from the TCAM 110, to connection manager circuitry 112 that manages the CB connection state and attributes. In the
In particular, for offloaded connections, the connection manager 112 provides the tid to the CB 114, and the CB 114 provides the current connection state and attributes for the connection (i.e., the connection to which the tid corresponds) to the connection manager 112. Based on the current connection state and attributes provided from the CB 114, the connection manager 112 determines that the connection state corresponds to an offloaded connection, how to appropriately modify the connection state and provides, to the payload command manager 116, an indication of the modification to the connection state. Based on the indication of the modification, the payload command manager 116 issues one or more appropriate payload commands to the payload manager block 118 to cause payload data to be forwarded to the host (via the form packet block 120) or to create Rx modulation events to schedule delivery of data to the host. The packet formed by the form packet block 120 contains the guest_idx steering index. The CB also contains the guest_idx value to use when sending packets to the host.
For filtering rules, the CB includes an indication of whether a packet is to be forwarded to the host or is to be dropped. In the context of the virtualization functionality discussed above, a dropped packet might, for example, correspond to a packet with a matching MAC address but with an IP address that is not consistent with the particular MAC address.
For offloaded connections, the connection manager 112 writes the modified connection state and attributes back into the CB 114. The read, modify and write of the connection state and attributes is done in an atomic operation. Here, “atomic” refers to the property that a read of the CB always returns the most recent state of the particular CB, even though the pipeline might be processing multiple messages simultaneously, that are associated with the same CB. The connection manager 112 provides an appropriate packet header for data transmission to a form packet block 120. Meanwhile, the payload manager block 118 provides the corresponding payload to the form packet block 120 (as discussed above, based on payload commands from the payload command manager 116). The form packet block 120 combines the packet header and corresponding payload into a packet for transmission to the host computer.
In the
Turning now to
An example of processing steps to utilize the
At step 506, the local address index is mapped to the guest_idx (steering index), which indicates the guest OS instance to which the DA is mapped. At step 508, the steering index is included as part of the tuple used to look up the tid (as described with reference to
Before continuing on with describing
A result is that traffic “belonging” to one guest OS instance can be distributed to different CPUs and queues for load balancing. For example, for the eight guest-OS instance example, one entry for each map 540a and 540b could be configured to map to the same 5th tuple value such that each guest-OS instance is allocated one local MAC address in the filtering database on each interface. The 5th-tuple returned for each interface could then map to the same free-list index, and the same response queue index. The ingress messages destined for a particular guest-OS would, in this manner, map to the same free-list and response queue, and the ingress messages destined for different instances of guest-OS's would map to different free-lists and response queues.
Another resulting observation is that different instances of guest-OS's can share the same local source MAC address and the steering of the traffic to the different instances of guest-OS's be based on other information, such as all or part of the 4-tuple.
In summary, 5-tuple virtualization has an advantage of allowing the NIC to be run with MAC address matching that enables Layer-2 demultiplexing, and also enabling the switching of ingress unicast packets on the NIC (other than broadcast and multicast packets, which would typically require additional processing on the NIC or in the host). In addition, the use of VLAN tags (to partition a LAN into many virtual LAN's) is naturally supported by this scheme, by including the VLAN tag as part of a 6-tuple that is used to determine the index.
We now return to
At step 510, it is determined if the packet is an offloaded connect request packet, such as a TCP SYN packet. If so, then at step 512 it is determined if the 6-tuple (step 508) is already in use. If so, then a RST error message may be sent back to the peer that sent the connection request message. If the 6-tuple is not already in use, then a TCB entry is initialized at step 514. In addition, at step 516, the steering index is used to look up a free-list and response queue for use with the connection. At step 518, a CPL command is provided to the host indicating that the connection was setup and indicating the parameters of the connection.
If, at step 510, it is determined that there was no hit in the TCAM for offload processing (step 508) for the 6-tuple, then the layer-2 packet is a tunnel packet, which is to be provided to the host without substantial protocol processing by the network interface circuitry. In this case, at step 520, the steering index is used to look up a scatter-list and response-queue for the connection, and a control protocol message is sent to the host at step 522 indicating receipt of the “tunnel” (non-offloaded) packet.
If it is determined that there was a hit in the TCAM for offload processing for the 6-tuple, then at step 524 it is confirmed that the steering index is contained in the TCB entry indirectly indicated by the 6-tuple. If the steering index is not contained in the TCB entry, then the packet is dropped. Otherwise, the packet is processed according to the offloaded protocol (TCP, in the illustrated case) and the data is communicated as appropriate to the host computer, according to the principles discussed above.
With regard to egress packets, as discussed above, egress virtualization is achieved by using different command queues and response queues for the different guest-OS's, and gather-lists use physical memory locations that can not be accessed by (that are not mapped for) the other guest-OS's.
Internally to the NIC, traffic belonging to different guest functions or guest OS's may be discriminated and treated differently according to traffic management. Traffic management in both the receive and transmit direct is described in several pending patent applications, each of which is incorporated herein by reference in its entirety: U.S. patent application Ser. No. 11/217,661, filed Aug. 31, 2005 and entitled “Protocol offload transmit traffic management”; U.S. patent application Ser. No. 11/282,933, filed Nov. 18, 2005 and entitled “A method for UDP Transmit Protocol Offload Processing with Traffic Management”; and U.S. patent application Ser. No. 11/313,003, filed Dec. 19, 2005 and entitled “A Method for Traffic Scheduling in Intelligent Network Interface Circuitry.”
This application is a continuation of U.S. patent application Ser. No. 11/330,898 (now U.S. Pat. No. 7,660,306), filed Jan. 12, 2006, and entitled “VIRTUALIZING THE OPERATION OF INTELLIGENT NETWORK INTERFACE CIRCUITRY,” which is incorporated herein by reference in its entirety for all purposes.
Number | Name | Date | Kind |
---|---|---|---|
4445116 | Grow | Apr 1984 | A |
4533996 | Hartung et al. | Aug 1985 | A |
5497476 | Oldfield et al. | Mar 1996 | A |
5778189 | Kimura et al. | Jul 1998 | A |
6087581 | Emmer et al. | Jul 2000 | A |
6226680 | Boucher et al. | May 2001 | B1 |
6240094 | Schneider | May 2001 | B1 |
6247060 | Boucher et al. | Jun 2001 | B1 |
6334153 | Boucher et al. | Dec 2001 | B2 |
6389479 | Boucher et al. | May 2002 | B1 |
6393487 | Boucher et al. | May 2002 | B2 |
6397316 | Fesas, Jr. | May 2002 | B2 |
6401177 | Koike | Jun 2002 | B1 |
6427171 | Craft et al. | Jul 2002 | B1 |
6427173 | Boucher et al. | Jul 2002 | B1 |
6434620 | Boucher et al. | Aug 2002 | B1 |
6470415 | Starr et al. | Oct 2002 | B1 |
6510164 | Ramaswamy et al. | Jan 2003 | B1 |
6564267 | Lindsay | May 2003 | B1 |
6591302 | Boucher et al. | Jul 2003 | B2 |
6594268 | Aukia et al. | Jul 2003 | B1 |
6625671 | Collette et al. | Sep 2003 | B1 |
6658480 | Boucher et al. | Dec 2003 | B2 |
6681244 | Cross et al. | Jan 2004 | B1 |
6687758 | Craft et al. | Feb 2004 | B2 |
6697868 | Craft et al. | Feb 2004 | B2 |
6701372 | Yano et al. | Mar 2004 | B2 |
6708223 | Wang et al. | Mar 2004 | B1 |
6708232 | Obara | Mar 2004 | B2 |
6717946 | Hariguchi et al. | Apr 2004 | B1 |
6751665 | Philbrick et al. | Jun 2004 | B2 |
6757245 | Kuusinen et al. | Jun 2004 | B1 |
6757746 | Boucher et al. | Jun 2004 | B2 |
6792502 | Pandya et al. | Sep 2004 | B1 |
6798743 | Ma et al. | Sep 2004 | B1 |
6807581 | Starr et al. | Oct 2004 | B1 |
6813652 | Stadler et al. | Nov 2004 | B2 |
6862648 | Yatziv | Mar 2005 | B2 |
6925055 | Erimli et al. | Aug 2005 | B1 |
6938092 | Burns | Aug 2005 | B2 |
6941386 | Craft et al. | Sep 2005 | B2 |
6965941 | Boucher et al. | Nov 2005 | B2 |
6996070 | Starr et al. | Feb 2006 | B2 |
7031267 | Krumel | Apr 2006 | B2 |
7042898 | Blightman et al. | May 2006 | B2 |
7076568 | Philbrick et al. | Jul 2006 | B2 |
7089326 | Boucher et al. | Aug 2006 | B2 |
7093099 | Bodas et al. | Aug 2006 | B2 |
7114096 | Freimuth et al. | Sep 2006 | B2 |
7124205 | Craft et al. | Oct 2006 | B2 |
7133902 | Saha et al. | Nov 2006 | B2 |
7133914 | Holbrook | Nov 2006 | B1 |
7133940 | Blightman et al. | Nov 2006 | B2 |
7164656 | Foster et al. | Jan 2007 | B2 |
7167926 | Boucher et al. | Jan 2007 | B1 |
7167927 | Philbrick et al. | Jan 2007 | B2 |
7174393 | Boucher et al. | Feb 2007 | B2 |
7185266 | Blightman et al. | Feb 2007 | B2 |
7191241 | Boucher et al. | Mar 2007 | B2 |
7191318 | Tripathy et al. | Mar 2007 | B2 |
7239642 | Chinn et al. | Jul 2007 | B1 |
7254637 | Pinkerton et al. | Aug 2007 | B2 |
7260631 | Johnson et al. | Aug 2007 | B1 |
7284047 | Barham et al. | Oct 2007 | B2 |
7313623 | Elzur et al. | Dec 2007 | B2 |
7376147 | Seto et al. | May 2008 | B2 |
7408906 | Griswold et al. | Aug 2008 | B2 |
7453892 | Buskirk et al. | Nov 2008 | B2 |
7474670 | Nowshadi | Jan 2009 | B2 |
7493427 | Freimuth et al. | Feb 2009 | B2 |
7533176 | Freimuth et al. | May 2009 | B2 |
7583596 | Frink | Sep 2009 | B1 |
7596634 | Mittal et al. | Sep 2009 | B2 |
7609696 | Guygyi et al. | Oct 2009 | B2 |
20010010046 | Muyres et al. | Jul 2001 | A1 |
20010021949 | Blightman et al. | Sep 2001 | A1 |
20020039366 | Sano | Apr 2002 | A1 |
20020101848 | Lee et al. | Aug 2002 | A1 |
20020188753 | Tang et al. | Dec 2002 | A1 |
20020191622 | Zdan | Dec 2002 | A1 |
20030018516 | Ayala et al. | Jan 2003 | A1 |
20030035436 | Denecheau et al. | Feb 2003 | A1 |
20030046330 | Hayes | Mar 2003 | A1 |
20030048751 | Han et al. | Mar 2003 | A1 |
20030158906 | Hayes | Aug 2003 | A1 |
20030200284 | Philbrick et al. | Oct 2003 | A1 |
20030204631 | Pinkerton et al. | Oct 2003 | A1 |
20040003094 | See | Jan 2004 | A1 |
20040019689 | Fan | Jan 2004 | A1 |
20040028069 | Tindal et al. | Feb 2004 | A1 |
20040030745 | Boucher et al. | Feb 2004 | A1 |
20040042487 | Ossman | Mar 2004 | A1 |
20040054813 | Boucher et al. | Mar 2004 | A1 |
20040062245 | Sharp et al. | Apr 2004 | A1 |
20040062246 | Boucher et al. | Apr 2004 | A1 |
20040064578 | Boucher et al. | Apr 2004 | A1 |
20040064590 | Starr et al. | Apr 2004 | A1 |
20040073703 | Boucher et al. | Apr 2004 | A1 |
20040078480 | Boucher et al. | Apr 2004 | A1 |
20040088262 | Boucher et al. | May 2004 | A1 |
20040100952 | Boucher et al. | May 2004 | A1 |
20040111535 | Boucher et al. | Jun 2004 | A1 |
20040117509 | Craft et al. | Jun 2004 | A1 |
20040158640 | Philbrick et al. | Aug 2004 | A1 |
20040165592 | Chen et al. | Aug 2004 | A1 |
20040190533 | Modi et al. | Sep 2004 | A1 |
20040199808 | Freimuth et al. | Oct 2004 | A1 |
20040213235 | Marshall et al. | Oct 2004 | A1 |
20040240435 | Craft et al. | Dec 2004 | A1 |
20050071490 | Craft et al. | Mar 2005 | A1 |
20050083935 | Kounavis et al. | Apr 2005 | A1 |
20050120037 | Maruyama et al. | Jun 2005 | A1 |
20050125195 | Brendel | Jun 2005 | A1 |
20050135378 | Rabie et al. | Jun 2005 | A1 |
20050135396 | McDaniel et al. | Jun 2005 | A1 |
20050135412 | Fan | Jun 2005 | A1 |
20050147126 | Qiu et al. | Jul 2005 | A1 |
20050190787 | Kuik et al. | Sep 2005 | A1 |
20050216597 | Shah et al. | Sep 2005 | A1 |
20050259644 | Huitema et al. | Nov 2005 | A1 |
20050259678 | Gaur | Nov 2005 | A1 |
20050289246 | Easton et al. | Dec 2005 | A1 |
20060031524 | Freimuth | Feb 2006 | A1 |
20060039413 | Nakajima et al. | Feb 2006 | A1 |
20060075119 | Hussain | Apr 2006 | A1 |
20060080733 | Khosmood et al. | Apr 2006 | A1 |
20060133267 | Alex et al. | Jun 2006 | A1 |
20060168649 | Venkat et al. | Jul 2006 | A1 |
20060206300 | Garg et al. | Sep 2006 | A1 |
20060209693 | Davari et al. | Sep 2006 | A1 |
20060221832 | Muller et al. | Oct 2006 | A1 |
20060221946 | Shalev et al. | Oct 2006 | A1 |
20060235977 | Wunderlich et al. | Oct 2006 | A1 |
20060265517 | Hashimoto et al. | Nov 2006 | A1 |
20060281451 | Zur | Dec 2006 | A1 |
20070011358 | Wiegert et al. | Jan 2007 | A1 |
20070064737 | Williams | Mar 2007 | A1 |
20070070901 | Aloni et al. | Mar 2007 | A1 |
20070083638 | Pinkerton et al. | Apr 2007 | A1 |
20070110436 | Bennett | May 2007 | A1 |
20070143848 | Kraemer et al. | Jun 2007 | A1 |
20070162572 | Aloni et al. | Jul 2007 | A1 |
20070201474 | Isobe | Aug 2007 | A1 |
20080002731 | Tripathy et al. | Jan 2008 | A1 |
20080016511 | Hyder et al. | Jan 2008 | A1 |
20080043750 | Keels et al. | Feb 2008 | A1 |
20080232386 | Gorti et al. | Sep 2008 | A1 |
20080273532 | Bar-David et al. | Nov 2008 | A1 |
20090172301 | Ebersole et al. | Jul 2009 | A1 |
20100023626 | Hussain et al. | Jan 2010 | A1 |
Number | Date | Country | |
---|---|---|---|
Parent | 11330898 | Jan 2006 | US |
Child | 12645324 | US |