US20060280193A1 - Method and apparatus for performing packet classification for policy-based packet routing - Google Patents
Method and apparatus for performing packet classification for policy-based packet routing Download PDFInfo
- Publication number
- US20060280193A1 US20060280193A1 US11/466,395 US46639506A US2006280193A1 US 20060280193 A1 US20060280193 A1 US 20060280193A1 US 46639506 A US46639506 A US 46639506A US 2006280193 A1 US2006280193 A1 US 2006280193A1
- Authority
- US
- United States
- Prior art keywords
- priority
- policy
- logic
- memory
- stored
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
- 238000000034 method Methods 0.000 title abstract description 56
- 230000015654 memory Effects 0.000 abstract description 287
- 230000006870 function Effects 0.000 description 48
- 230000008569 process Effects 0.000 description 31
- 238000002955 isolation Methods 0.000 description 26
- 230000004044 response Effects 0.000 description 23
- 210000004027 cell Anatomy 0.000 description 19
- 230000001174 ascending effect Effects 0.000 description 16
- 230000005055 memory storage Effects 0.000 description 15
- 235000019800 disodium phosphate Nutrition 0.000 description 14
- 238000001914 filtration Methods 0.000 description 13
- 238000007726 management method Methods 0.000 description 12
- 230000000295 complement effect Effects 0.000 description 10
- 238000003491 array Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 5
- 238000007599 discharging Methods 0.000 description 4
- 238000003780 insertion Methods 0.000 description 4
- 230000037431 insertion Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 101100537666 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) TOS1 gene Proteins 0.000 description 3
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 2
- 230000000873 masking effect Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000006855 networking Effects 0.000 description 2
- 210000000352 storage cell Anatomy 0.000 description 2
- 239000000758 substrate Substances 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03K—PULSE TECHNIQUE
- H03K3/00—Circuits for generating electric pulses; Monostable, bistable or multistable circuits
- H03K3/02—Generators characterised by the type of circuit or by the means used for producing pulses
- H03K3/353—Generators characterised by the type of circuit or by the means used for producing pulses by the use, as active elements, of field-effect transistors with internal or external positive feedback
- H03K3/356—Bistable circuits
- H03K3/356104—Bistable circuits using complementary field-effect transistors
- H03K3/356113—Bistable circuits using complementary field-effect transistors using additional transistors in the input circuit
- H03K3/35613—Bistable circuits using complementary field-effect transistors using additional transistors in the input circuit the input circuit having a differential configuration
- H03K3/356139—Bistable circuits using complementary field-effect transistors using additional transistors in the input circuit the input circuit having a differential configuration with synchronous operation
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C15/00—Digital stores in which information comprising one or more characteristic parts is written into the store and in which information is read-out by searching for one or more of these characteristic parts, i.e. associative or content-addressed stores
- G11C15/04—Digital stores in which information comprising one or more characteristic parts is written into the store and in which information is read-out by searching for one or more of these characteristic parts, i.e. associative or content-addressed stores using semiconductor elements
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C8/00—Arrangements for selecting an address in a digital store
- G11C8/04—Arrangements for selecting an address in a digital store using a sequential addressing device, e.g. shift register, counter
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03K—PULSE TECHNIQUE
- H03K23/00—Pulse counters comprising counting chains; Frequency dividers comprising counting chains
- H03K23/40—Gating or clocking signals applied to all stages, i.e. synchronous counters
- H03K23/50—Gating or clocking signals applied to all stages, i.e. synchronous counters using bi-stable regenerative trigger circuits
- H03K23/56—Reversible counters
Definitions
- the present invention relates generally to performing packet classification for policy-based packet routing.
- Routers are devices that direct traffic in a network.
- Each router in the network has a route table that typically includes routing information to enable incoming packets to be forwarded or routed to their destination in the network. Some routers forward packets based only on the destination address indicated in the packet. Other, more complex, routers forward or route packets based on policies defined, for example, by a network administrator. The latter routing schemes are commonly referred to as policy-based routing.
- Policy-based routing can enable packets to be forwarded or routed in a network based on any number of criteria, including the source of the packet, the destination of the packet, the cost of forwarding the packet through different routes or paths in the network, or the bandwidth available on different routes in the network. Policy-based routing can also be used to provide a certain Quality of Service (QOS) or Type of Service (TOS) to differentiated traffic in the network. For example, one or more of the various fields (e.g., the TOS bits) in the header of an Internet Protocol (IP) packet can be used by policy-based routers to forward IP packets in a network.
- QOS Quality of Service
- TOS Type of Service
- Each policy-based router implements a policy through the use of route maps that define how to forward the packet in the network.
- Each route map statement or policy statement contains one or more match clauses and a set clause.
- the match clauses are a series of conditions that are used to determine if an incoming packet satisfies a particular policy. If all of the match clauses of a policy statement are satisfied, the set clause specifies how the router should forward the packet in the network. If one of the match clauses of a particular policy statement is not satisfied, then the policy-based router investigates subsequent policy statements.
- FIG. 1 shows exemplary processed policy information 100 of an incoming packet to a policy-based router.
- Policy information 100 includes several policy fields 102 including a destination address (DA) for the packet, a source address (SA) of the packet, protocol type (PTCL) such as those defined by for an IP packet header, TOS, and COST.
- Policy information 100 may be received by a policy-based router that implements a policy such as policy 200 shown in FIG. 2 .
- Policy 200 includes three separate policy statements 201 through 203 . If policy information 100 satisfies the match clause (i.e., the “if” clause) of one of the policy statements, the set clause (i.e., the “then” clause) of that policy statement determines routing information for the packet in the network. For example, if the destination address of the incoming packet is DA 1 , the source address is SA 1 , and the TOS field of the packet is TOS 1 , then routing information RI 2 should be selected.
- match clause i.e., the “
- a policy-based router can use a content addressable memory (CAM)-based system to implement a filtering or classification function to determine whether an incoming packet matches a policy statement.
- FIG. 3 shows one example of a system 300 that implements policy-based routing using a ternary CAM 302 .
- the policy statements or policy words 201 - 203 are stored in separate rows in ternary CAM array 304 .
- a ternary CAM array is one that is able to mask entries in a CAM array on a bit-by-bit basis.
- Ternary CAM array 304 has rows of CAM cells 305 for storing policy field information, and corresponding rows of mask cells 310 for storing mask data.
- Routing information RI 0 -RI 2 is typically stored in an external memory 308 at addresses corresponding to those at which the respective policy words 201 - 203 are stored in ternary CAM array 304 .
- Each policy field that corresponds to a match clause for a given policy statement is unmasked by having its corresponding mask bits set, for example, to a logic zero.
- each policy field that does not have a match clause for a given policy statement is masked by having its corresponding mask bits set, for example, to a logic one.
- a policy-based router When an incoming packet is received by a policy-based router, it is processed to determine the policy field information.
- the processed policy field information is provided to system 300 as policy search key 307 .
- the corresponding match line ML 0 -ML N will be asserted and provided to priority encoder 306 .
- priority encoder 306 In response to the match lines, priority encoder 306 outputs the address of the highest priority matching entry in CAM array 304 to HPM bus 312 . If there is more than one matching policy statement in CAM array 304 , priority encoder 306 determines that the highest priority matching policy statement is the one stored at the lowest logical address of CAM array 304 . For example, as shown in FIG.
- CAM array 304 if CAM array 304 is loaded with policy statement 203 at address zero (i.e., the lowest logical address), statement 202 at address one, and statement 201 at address 2 , and a policy search key of DA 1 , SA 1 , PTCL 1 , TOS 1 , COST 1 is provided to CAM array 304 , then each of policy statements 201 - 203 is identified as a match on match lines ML 0 -ML 2 , respectively.
- Priority encoder 306 outputs address zero on the HPM bus to select route information RI 2 from address zero in memory 308 .
- priority encoder 306 determines the highest priority matching location based on predetermined logical address assignments, policy statements 201 - 203 are preordered or prioritized such that higher priority policy statements are stored in lower logical addresses of CAM array 304 than lower priority policy statements.
- a policy statement has a higher priority than another policy statement when the route information for the first policy statement is to be selected over the second policy statement even though both policy statements may match the policy search key (e.g., with masking).
- the prioritizing of the policy statements is typically performed by table management hardware and/or software, which adds overhead to the router.
- FIG. 4 shows an example in which policy statement 201 is incorrectly prioritized such that it, rather than policy statement 203 , is loaded into the lowest logical address.
- priority encoder 306 still provides address zero on HPM bus 312 . This causes, however, route information RI 0 to be selected in memory 308 rather than the desired route information RI 2 associated with higher priority policy statement 201 . It would be desirable to load policy statements into a CAM array without having to preorder the statements according to their respective priorities.
- CAM 302 may cause an incorrect route to be selected from memory 308 . For example, FIG.
- a new policy statement 204 is loaded into the next free address (i.e., address three) of CAM array 304 without reprioritizing and reloading CAM array 304 .
- Route information RI 3 corresponding to policy statement 204 , is loaded into a corresponding address in memory 308 . If policy statement 204 has a priority greater than that of 201 , 202 , or 203 , then a search with policy search key 307 will not correctly identify policy statement 204 during a compare operation on CAM array 304 . Policy statement 203 , rather, will be identified as the higher priority statement, and route information RI 2 will be selected. It would be desirable to be able to add a new policy statement to a policy without reloading the CAM array storing the policy statements.
- the digital signal processor includes a policy statement table for storing policy statements.
- the policy statement table may be stored in a ternary CAM array that stores the policy statements and associated mask data.
- Each policy statement has associated with it a priority number that indicates the priority of the policy statement relative to other policy statements.
- the priority numbers are separately stored in a priority index table.
- the priority index table includes priority logic that determines the most significant priority number from among the policy statements that match an incoming packet during a classification or filter operation.
- the priority logic also identifies the location in the priority index table of the most significant priority number.
- the most significant priority number may be the priority number with the lowest or highest numerical value.
- the identified location in the priority index table can be used to access associated route information or other information stored in a route memory array located external to, or on the DSP.
- the DSP may include an encoder to encode the identified location in the priority index table into an address for the route memory.
- the DSP configuration obviates preloading the policy statements in the policy statement table in a predetermined order. Instead, the priority logic determines the most significant priority number from among matching policy statements regardless of the order in which the policy statements are stored in the table. This can reduce the hardware and/or software needed for table management of the table, and can increase the performance of a router incorporating the DSP.
- new policy statements can be added at any location in the policy statement table, and associated priority numbers loaded into corresponding locations in the priority index table. If a new policy statement has a priority that is greater than or equal to a priority of a policy statement already stored in the policy statement table, the priority number of the previously stored policy statement may be updated to accommodate the new policy statement. Similarly, when a policy statement is removed (i.e., invalidated or overwritten) from the policy statement table, the priority numbers of the previously stored lower priority policy statements may be updated.
- the updating functions can be performed by the priority logic in the priority index table, or by inequality circuits in the priority index table.
- the updating functions can be performed without the need to physically reorder the policy statements in the policy statement table, or to physically reorder the priority numbers in the priority index table. This also can reduce the hardware and/or software needed for table management of the policy statement table, and can increase the performance of a router incorporating the DSP.
- the DSP can also be used in other non-networking applications.
- the DSP can be used to process if-then-else functions in other applications.
- FIG. 1 is an example of policy information
- FIG. 2 is an example of a policy having policy statements
- FIG. 3 is an example of storing a policy in a conventional ternary CAM
- FIG. 4 is an example of storing a particular policy in a conventional ternary CAM and comparing a search key with the policy
- FIG. 5 is another example of storing a particular policy in a conventional ternary CAM and comparing a search key with the policy
- FIG. 6 is one embodiment of a classification system for a policy-based router
- FIG. 7 is one embodiment of performing the classification operation for the system of FIG. 6 ;
- FIG. 8 is one example of performing the classification operation on a particular policy stored in the CAM array of FIG. 6 ;
- FIG. 9 is one embodiment of reading or writing a policy statement based on a priority number stored in the priority memory of FIG. 6 ;
- FIG. 10 is one embodiment of the priority index table of FIG. 6 ;
- FIG. 11 is another embodiment of the priority index table of FIG. 6 ;
- FIG. 12A is one embodiment of two rows of the priority index table of FIG. 11 ;
- FIG. 12B is another embodiment of two rows of the priority index table of FIG. 11 ;
- FIG. 13 is one example of determining the most significant priority number stored in the rows of the priority index table
- FIG. 14 is one embodiment of a compare circuit and an isolation circuit for the priority logic element of FIG. 11 ;
- FIG. 15 is another embodiment of a compare circuit and an isolation circuit for the priority logic element of FIG. 11 ;
- FIG. 16 is another example of determining the most significant priority number stored in the rows of the priority index table
- FIG. 17 is another embodiment of the priority index table of FIG. 6 including inequality circuits
- FIG. 18 is one example of a policy stored in the digital signal processor of FIG. 17 ;
- FIG. 19 is one example of loading a new policy statement into the digital signal processor of FIG. 18 ;
- FIG. 20 is another example of a policy stored in the digital signal processor of FIG. 17 ;
- FIG. 21 is one example of loading a new policy statement into the digital signal processor of FIG. 20 ;
- FIG. 22 is one embodiment of the rows of the priority index table of FIG. 17 , in which the priority memory locations are configured as counters;
- FIG. 23 is one embodiment of an inequality circuit and counter of FIG. 22 ;
- FIG. 24 is one example of determining whether a new number is greater than or equal to priority numbers stored in the priority memory of FIG. 17 ;
- FIG. 25A is one embodiment of a compare circuit of FIG. 24 ;
- FIG. 25B is another embodiment of a compare circuit of FIG. 24 ;
- FIG. 26 is one example of determining whether a new number is less than or equal to priority numbers stored in the priority memory of FIG. 17 ;
- FIG. 27A is one embodiment of a compare circuit of FIG. 26 ;
- FIG. 27B is another embodiment of a compare circuit of FIG. 26 ;
- FIG. 28 is one embodiment of a process of deleting or invalidating policy statements and their associated priority numbers from the digital signal processor of FIG. 17 ;
- FIG. 29 is one embodiment of priority index table of FIG. 17 ;
- FIG. 30 is another embodiment of priority index table of FIG. 17 ;
- FIG. 31 is one embodiment of an inequality circuit element, a priority logic element and a memory element of FIG. 17 ;
- FIG. 32 is another embodiment of an inequality circuit element, a priority logic element and a memory element of FIG. 17 ;
- FIG. 33 is another embodiment of the priority index table of FIG. 6 including a decrement circuit and storing a policy
- FIG. 34 is one example of loading a new policy statement into the digital signal processor of FIG. 33 ;
- FIG. 35 is one embodiment of a row of the priority index table in FIG. 33 ;
- FIG. 36 is one example, of determining whether a new number is greater than or equal to priority numbers stored in the priority memory of FIG. 33 ;
- FIG. 37 is one embodiment of compare circuit and an isolation circuit for a priority logic element of FIG. 35 ;
- FIG. 38 is one example, of determining whether a new number is less than or equal to priority numbers stored in the priority memory of FIG. 33 ;
- FIG. 39 is a block diagram of one embodiment of depth cascading digital signal processors
- FIG. 40 is a block diagram of one embodiment of a digital signal processor including cascade logic
- FIG. 41 is a block diagram of one embodiment of the cascade logic of FIG. 39 ;
- FIG. 42 is one example of processing Internet protocol addresses based on a classless inter domain routing scheme in the digital signal processor of FIG. 6 .
- circuit elements or blocks may be shown as buses or as single signal lines.
- buses may alternatively be single signal lines, and each of the single signal lines may alternatively be buses.
- prefix symbol “/” or the suffix “B” attached to signal names indicates that the signal is an active low signal.
- Each of the active low signals may be changed to active high signals as generally known in the art.
- the digital signal processor includes a policy statement table for storing policy statements.
- the policy statement table may be stored in a ternary CAM array that stores the policy statements and associated mask data.
- Each policy statement has associated with it a priority number that indicates the priority of the policy statement relative to other policy statements.
- the priority numbers are separately stored in a priority index table.
- the priority index table includes priority logic that determines the most significant priority number from among the policy statements that match an incoming packet during a classification or filter operation.
- the priority logic also identifies the location in the priority index table of the most significant priority number.
- the most significant priority number may be the priority number with the lowest or highest numerical value.
- the identified location in the priority index table can be used to access associated route information or other information stored in a route memory array located external to, or on the DSP.
- the DSP may include an encoder to encode the identified location in the priority index table into an address for the route memory.
- the DSP configuration obviates preloading the policy statements in the policy statement table in a predetermined order. Instead, the priority logic determines the most significant priority number from among matching policy statements regardless of the order in which the policy statements are stored in the table. This can reduce the hardware and/or software needed for table management of the table, and can increase the performance of a router incorporating the DSP.
- new policy statements can be added at any location in the policy statement table, and associated priority numbers loaded into corresponding locations in the priority index table. If a new policy statement has a priority that is greater than or equal to a priority of a policy statement already stored in the policy statement table, the priority number of the previously stored policy statement may be updated to accommodate the new policy statement. Similarly, when a policy statement is removed (i.e., invalidated or overwritten) from the policy statement table, the priority numbers of the previously stored lower priority policy statements may be updated.
- the updating functions can be performed by the priority logic in the priority index table, or by inequality circuits in the priority index table.
- the updating functions can be performed without the need to physically reorder the policy statements in the policy statement table, or to physically reorder the priority numbers in the priority index table. This also can reduce the hardware and/or software needed for table management of the policy statement table, and can increase the performance of a router incorporating the DSP.
- the DSP can also be used in other non-networking applications.
- the DSP can be used to process if-then-else functions in other applications.
- FIG. 6 shows one embodiment of a classification or filtering system 600 for a policy-based router.
- System 600 includes DSP 602 and route memory 614 .
- DSP 602 includes policy statement table 604 , priority index table 606 and encoder 612 , and may be an integrated circuit component formed on a single semiconductor substrate.
- policy statement table 604 is stored in a ternary CAM array that stores policy statements or policy words.
- Ternary CAM array 604 has rows of CAM cells 605 for storing policy field information PF 1 -PFX, where X is any number.
- Each policy field PF 1 -PFX can include any policy information including DA, SA, PTCL, TOS, and COST, or any other type of policy field to assist in the classification or filtering of the policy statement to provide a certain Quality of Service (QoS), Class of Service (CoS), and the like.
- Each policy field may include any number of bits. Additional information associated with each policy field may be stored in one or more additional binary or ternary CAM cells or other types of memory cells disposed in each row of CAM 604 .
- Ternary CAM 604 also has rows of mask cells 607 for storing mask data M 1 -MX corresponding to each row of policy fields 605 .
- Global masks may be used to mask entire columns in CAM array 604 as generally known in the art.
- CAM array 604 may be any other type of CAM including a binary CAM, or any other type of memory to store policy statements to be compared with processed policy information of an incoming packet.
- Ternary CAM array 604 can be any ternary CAM array that is capable of locally masking each entry on a bit-by-bit basis.
- Each policy field that corresponds to a match clause for a given policy statement will be unmasked by having its corresponding mask bits set, for example, to a logic zero (or, alternatively, a logic one).
- each policy field that does not have a match clause for a given policy statement will be masked by having its corresponding mask bits set, for example, to a logic one (or, alternatively, a logic zero).
- DSP 602 also supports rule and route aggregation. That is, DSP 602 supports ranges of addresses or policy field information.
- Each policy statement loaded into ternary CAM array 604 has associated with it a priority number P 0 -P Z and route information RI 0 -RI N-1 .
- the priority number indicates the priority of the policy statement relative to other policy statements in a given policy.
- the policy numbers may be assigned by a user of DSP 602 including, for example, a network administrator or the router itself.
- the priority numbers P 0 -P Z are separately stored at locations 608 0 - 608 N-1 , respectively, of priority memory 608 of priority index table 606 .
- Route information RI 0 -RI N-1 for the particular policy statements are stored at locations 614 0 - 614 N-1 , respectively, in route memory 614 .
- the route information may include, for example, forwarding or next hop information, authentication information, QOS, TOS, time to live information or other packet filtering and classification information for an incoming packet to the router incorporating system 600 .
- a policy statement, its priority number, and its route information are each stored at the corresponding addresses in each of their respective memory arrays.
- Priority memory 608 and route memory 614 may each be any type of memory array including volatile, non-volatile, random access memory (RAM), and/or read only access memory (ROM).
- priority memory 608 comprises a CAM array.
- the number of addressable locations N in priority memory 608 may be greater than, less than, or equal to Z.
- priority memory 608 may be 20 bits wide to accommodate up to 2 20 or 1 Meg (i.e., 1,048,576) priority numbers.
- the priority numbers may be assigned in ascending priority order such that zero is the highest priority number and 2 n ⁇ 1 is the lowest priority number. Alternatively, the priority numbers may be assigned in descending priority order such that 2 n ⁇ 1 is the highest priority number and zero is the lowest priority number.
- Each priority number may be assigned so as to identify the priority of each policy statement relative to other policy statements.
- the priority numbers may be assigned consecutively. For example, the highest priority policy statement can be assigned the highest priority number (e.g., zero or 2 n ⁇ 1), the next lower priority policy statement can be assigned the next lower priority number (e.g., one or 2 n ⁇ 2), and so forth.
- gaps may be left in the priority number assignments to allow for the addition of future priority numbers associated with new policy statements.
- Priority index table 606 also includes priority logic 610 that compares the priority numbers with each other for all corresponding policy statements that match an incoming packet.
- Priority logic 610 identifies the most significant priority number PNUM in memory 608 from among the compared priority numbers, and further identifies the location of PNUM in priority memory 608 .
- PNUM has the lowest numerical value when the priority numbers are assigned is ascending priority order, and PNUM has the highest numerical value when the priority numbers are assigned in descending priority order.
- Priority logic 610 may also output PNUM from DSP 602 .
- the identified location of PNUM in the priority memory is provided on internal address lines IAD 0 -IAD N-1 to encoder 612 .
- one of IAD 0 -IAD N-1 is asserted to indicate the location in priority memory 608 of PNUM from among the compared priority numbers. This location also corresponds to the location of the highest priority matching policy statement in ternary CAM array 604 .
- Encoder 612 may be any encoding logic that takes the information on address lines IAD 0 -IAD N-1 and generates an encoded address.
- encoder 612 is a ROM.
- route memory 614 may also be included within DSP 602 .
- encoder 612 may be omitted and route memory 614 may be accessed directly by internal address lines IAD 0 -IAD N-1 .
- FIG. 7 summarizes the classification or filtering function 700 (i.e., search or compare operation) performed by DSP 602 for an incoming packet according to a policy stored in ternary CAM 604 .
- An incoming packet received by a policy-based router incorporating system 600 is initially processed to determine the policy field information.
- the policy field information is provided to system 600 as policy search key 609 .
- the policy fields of policy search key 609 are compared with the policy statements stored in ternary CAM array 604 . For each policy statement that matches the policy search key, the corresponding match line ML 0 -ML N-1 is asserted. If no match is found, then the process stops at step 704 .
- priority logic 610 determines PNUM and identifies its location in priority memory 608 .
- the identified location is provided on internal address lines IAD 0 -IAD N-1 to encoder 612 .
- encoder 612 determines the address of the identified location in priority index table 606 . This encoded address is also the logical address of the highest priority matching policy statement in ternary CAM array 604 .
- Encoder 612 outputs the encoded address to HPM bus 616 .
- the encoded address can then be used at step 710 to access the corresponding route information in memory 614 . Steps 708 and/or 710 may be omitted when encoder 612 is removed from DSP 602 , and priority logic 610 may directly access the route information in memory 614 .
- IAD 0 -IAD N-1 are provided to CAM array 604 to access the highest priority matching policy statement, which may then be read from DSP 602 .
- HPM bus 616 may be provided to CAM array 604 (e.g., through a decoder) to access the highest priority matching policy statement.
- FIG. 8 shows one example in which policy statements 201 , 202 , and 203 from FIG. 2 are stored at locations 604 0 , 604 1 , and 604 2 (i.e., addresses 0, 1, and 2), respectively, of ternary CAM array 604 .
- the corresponding priority numbers 2, 1, and 0 are stored at locations 608 0 , 608 1 , and 608 2 , respectively, in memory 608 .
- the corresponding route information RI 0 , RI 1 , and RI 2 are stored at locations 614 0 , 614 1 , and 614 2 , respectively of route memory 614 .
- the policy statements and priority numbers are written into their respective memories using conventional write circuits, counters, and/or address decoders, etc. (not shown).
- the priority numbers have been assigned in ascending priority order such that policy statement 203 is identified as the highest priority policy statement by being assigned priority number 0, the lowest numerical value; policy statement 201 is identified as the having the lowest priority policy statement by being assigned priority number 2, the highest numerical value; and, policy statement 202 is identified as having a priority greater than that of policy statement 201 , but less than that of policy statement 203 , by being assigned priority number 1.
- the priority numbers may be assigned in descending priority order such that policy statement 201 is identified as the highest priority policy statement by being assigned priority number 2, the highest numerical value; policy statement 203 is identified as having the lowest priority policy statement by being assigned priority number 0, the lowest numerical value; and, policy statement 202 is identified as having a priority greater than that of policy statement 201 , but less than that of policy statement 203 , by being assigned priority number 1.
- the process of determining the route information for policy search key 609 is illustrated with the aid of FIG. 7 .
- the policy fields of policy search key 609 are compared with the policy statements stored in ternary CAM array 604 .
- CAM array 604 determines that each of the policy statements 201 - 203 , as masked by their respective mask data, matches policy search key 609 . In response, each of match lines ML 0 -ML 2 is asserted.
- priority logic 610 compares, with each other, priority numbers 0, 1, and 2 associated with matching policy statements 203 , 202 , and 201 respectively. Priority logic 610 determines that priority number 0 is the most significant priority number, asserts IAD 2 , and de-asserts the other internal address lines. Encoder 612 encodes the internal address information, at step 708 , and generates an external address of 2 on HPM bus 616 . The external address can be used to access route information RI 2 stored at address two in route memory 614 .
- DSP 602 is able to identify the highest priority matching policy statement stored in ternary CAM array 604 regardless of where the policy statements are stored in CAM array 604 .
- the process illustrated in FIG. 7 identifies the location in priority memory 608 of the most significant priority number from among the compared priority numbers. Once this location is identified, the priority number stored at the identified location can be read out from DSP 602 by a read circuit (not shown), or a new priority number can be written into that location by a write circuit (not shown).
- the user of the DSP e.g., a network administrator or the policy-based router itself
- priority numbers can be updated by the user for already stored policy statements. This provides the user with flexibility in the control and management of the policy statements stored in DSP 602 .
- priority memory 608 may be a CAM.
- each priority number may be separately compared by a comparison circuit with an externally applied priority number.
- FIG. 9 shows one embodiment of a process 900 for reading or writing a policy statement based on a priority number already stored in the priority memory.
- an external priority number is compared with the priority numbers stored in priority memory 608 .
- the valid priority numbers stored in memory 608 may be indicated by one or more validity bits (not shown) stored in CAM memory 604 or memory 608 . If there is no match, or more than one match (i.e., a multiple match), then the process stops at step 904 .
- a priority encoder circuit can be coupled to priority index table 606 to select one of the multiple matches for step 906 .
- IAD 0 -IAD N-1 may be the match lines associated with each priority number storage location.
- encoder 612 determines the address of the matching location in priority memory 608 and provides this address on HPM bus 616 . This address is also the address of the corresponding policy statement in CAM array 604 .
- the encoded address can then be provided to CAM array 604 (e.g., through a decoder), and the policy statement stored at that address either read from DSP 602 , or overwritten.
- IAD 0 -IAD N-1 may be provided directly to CAM array 604 to access the desired policy statement and the process may omit step 908 . Additionally, a new priority number can be written to priority memory 608 at the address determined in step 910 .
- priority index table 606 stores priority numbers in priority memory 608
- priority logic 610 determines the most significant priority number among those associated with policy statements that match policy search key 609 .
- FIG. 10 shows DSP 1000 that includes priority index table 1001 that is one embodiment of priority index table 606 .
- CAM array 1002 is one embodiment of priority memory 608
- priority logic 1004 is one embodiment of priority logic 610 .
- DSP 1000 performs the filtering function of FIG. 7 as follows.
- a policy search key is compared with the policy fields stored in CAM array 604 and match lines ML 0 -ML N-1 are asserted for the matching locations.
- the asserted match lines access the associated priority numbers stored in CAM array 1002 .
- ML 0 -ML N-1 are coupled to corresponding word lines of the rows of CAM cells in CAM array 1002 .
- CAM array 1002 provides the selected priority numbers to priority logic 1004 over bus 1006 to determine which priority number is the most significant priority number.
- the priority numbers may be provided one at a time over bus 1006 , or in parallel.
- Priority logic 1004 includes one or more compare circuits to compare the priority numbers and determine the most significant priority number.
- priority logic 1004 includes a compare circuit and a register. The first priority number provided on bus 1006 is stored in the register and compared with the second priority number. Each subsequent priority number provided over bus 1006 is then compared and the result (either greater than or less than) is then stored in the register. This process continues until no more priority numbers are provided over bus 1006 .
- multiple compare circuits may be provided to compare multiple priority numbers at the same time to generate the most significant priority number.
- the most significant priority number is determined, it is provided over bus 1008 to CAM 1002 and searched against the stored priority numbers to determine where it is located in CAM array 1002 .
- the indicated location is provided on internal address lines IAD 0 -IAD N-1 , which may correspond to the match lines of CAM array 1002 . Steps 708 and 710 may then be performed as previously described above.
- FIG. 11 shows DSP 1100 that includes priority index table 1101 that is another embodiment of priority index table 606 .
- priority memory 608 and priority logic 610 are merged together on a bit-by-bit basis to form priority index table 1101 .
- the priority memory includes memory elements 1102 that each store one bit of a priority number for a given row. Each memory element may be any type of storage mechanism including volatile or non-volatile memory cells.
- the priority logic includes priority logic elements 1104 . Each priority logic element 1104 is associated with, or corresponds to, one of the memory elements 1102 such that columns 1106 0 - 1106 n-1 of priority index table 1101 have a memory element/priority logic element pair for each of its rows.
- Each priority logic element 1104 effectively compares the priority number bit stored in its associated memory element 1102 with the priority number bits stored in every other memory element of its column to determine one of bits PNUM 0 -PNUM n-1 for the most significant priority number.
- Bits PNUM 0 -PNUM n-1 comprise the most significant priority number from among the policy statements that match a policy search key.
- FIG. 12A shows priority index table 1200 that is one embodiment of two rows of priority index table 1101 .
- each priority logic element 1104 includes a compare circuit 1206 and an isolation circuit 1204 .
- Each compare circuit 1206 is connected in a wired-OR configuration with the other compare circuits in its respective column by one of priority signal lines 1208 0 - 1208 n-1 .
- Each priority signal line may be pre-charged towards a power supply voltage (or any other predetermined voltage) by a pre-charge circuit 1202 .
- Each compare circuit 1206 may be any digital or analog compare circuit that, when executing step 706 of FIG. 7 , effectively compares the priority number bit stored in its respective storage element 1102 with the priority number bits stored in every other storage element 1102 of the same column.
- each compare circuit monitors the comparison result of the more significant priority number bits through the logical states of match line segments 1210 .
- Match line segments 1210 are coupled between match lines ML 0 -ML N-1 and internal address lines IAD 0 -IAD N-1 by isolation circuits 1204 .
- the isolation circuits isolate the comparison results generated for less significant priority bit locations from affecting the comparison results generated for more significant priority bit locations.
- the isolation circuits may also work together with the comparison circuits to control the state of the match line segments.
- priority index table 1200 can be illustrated with an example shown in FIG. 13 and with the aid of FIG. 7 .
- priority index table 1200 comprises a 2 ⁇ 4 matrix of rows and columns.
- Row zero stores priority number 0110 having the decimal equivalent of the number 6
- row one stores priority number 0101 having the decimal equivalent of the number 5.
- each of row zero and row one of CAM array 604 have policy statements that match the policy search key such that match lines ML 0 and ML 1 are asserted (step 702 ).
- the priority numbers are stored in ascending priority order such that 0101 is the more significant priority number between 0101 and 0110.
- compare circuits 1206 0,0 - 1206 3,1 determine that 0101 is the more significant priority number PNUM, and cause IAD 1 to be asserted to indicate that 0101 is stored in row one of the priority index table.
- Compare circuits 1206 0,0 - 1206 3,1 determine that PNUM is 0101 as follows. The most significant bit PNUM 3 is resolved first. When any memory element 1102 stores a logic zero and the corresponding match line segment 1210 is asserted, the corresponding priority signal line 1208 is discharged. Thus, each of compare circuits 1206 3,1 and 1206 3,0 discharge signal line 1208 3 such that PNUM 3 is a logic zero.
- compare circuit 1206 3,1 compares the state of priority signal line 1208 3 with the priority number bit stored in 1102 3,1 , and determines that both have the same logic state. This causes compare circuit 1206 3,1 not to affect the logical state of match line segment 1210 2,1 such that match line segment 1210 2,1 has the same logic state as match line segment 1210 3,1 (ML 1 ).
- compare circuit 1206 3,0 compares the state of priority signal line 1208 3 with the priority number bit stored in 1102 3,0 and determines that both have the same state. This causes compare circuit 1206 3,0 not to affect the logical state of match line segment 1210 2,0 such that match line segment 1210 2,0 has the same logic state as match line segment 1210 3,0 (ML 0 ).
- PNUM 2 The next most significant bit PNUM 2 is then resolved.
- Memory elements 1102 that store a logic one do not discharge their corresponding priority signal lines 1208 . Since memory elements 1102 2,1 and 1102 2,0 both store logic one states, signal line 1208 2 remains pre-charged such that PNUM 2 is a logic one.
- compare circuit 1206 2,1 compares the state of priority signal line 1208 2 with the priority number bit stored in 1102 2,1 , and determines that both have the same logic state. This causes compare circuit 1206 2,1 not to affect the logical state of match line segment 1210 1,1 such that match line segment 1210 1,1 has the same logic state as match line segment 1210 2,1 .
- compare circuit 1206 2,0 compares the state of priority signal line 1208 2 with the priority number bit stored in 1102 2,0 and determines that both have the same logic state. This causes compare circuit 1206 2,0 to not affect the logical state of match line segment 1210 1,0 such that match line segment 1210 1,0 has the same logic state as match line segment 1210 2,0 .
- PNUM 1 is resolved next. Since memory element 1102 1,1 stores a logic zero and match line segment 1210 1,1 is asserted, compare circuit 1206 1,1 discharges priority signal line 1208 1 . This causes PNUM 1 to be a logic zero. Additionally, compare circuit 1206 1,1 compares the logic zero state of priority signal line 1208 1 with the logic zero stored in 1102 1,1 and allows match line segment 1210 0,1 to have the same state as match line segment 1210 1,1 . Compare circuit 1206 1,0 , however, compares the logic zero on priority signal line 1208 1 with the logic one stored in memory element 1102 1,0 , and de-asserts match line segment 1210 0,0 .
- isolation circuits 1204 3,0 , 1204 2,0 , and 1204 1,0 isolate the de-asserted match line segment 1210 0,0 from match line segment 1210 3,0 , 1210 2,0 , and 1210 1,0 such that PNUM 3 , PNUM 2 , and PNUM 1 are not affected by the de-assertion of match line segment 1210 0,0 .
- Compare circuit 1206 0,1 alone determines PNUM 0 since compare circuit 1206 0,0 can not discharge priority signal line 1208 0 . Since memory element 1102 0,1 stores a logic one and match line segment 1210 0,1 is asserted, compare circuit 1206 0,1 leaves priority signal line 1208 0 pre-charged, and PNUM 0 is a logic one. Additionally, compare circuit 1206 0,1 allows IAD 1 to have the same state as match line segment 1210 0,1 . Since match line segment 1210 0,1 is asserted, IAD 1 will be asserted indicating that the most significant priority number is stored in that row.
- bits PNUM 3 -PNUM 0 indicate that the most significant priority number stored in the priority index table is 0101, and IAD 1 is asserted identifying that 0101 is stored in row one.
- Table 1 shows one example of a truth table for implementing each compare circuit 1206 , where X (column) and Y (row) are any integers. Other truth tables may be used (and corresponding logic generated accordingly) including those that logically complement one of more or the signals indicated in Table 1.
- state 5 may not occur since priority signal line 1208 will not be a logic one when a memory element 1102 stores a logic zero and the corresponding match line segment 1210 is asserted to a logic one state. For other truth tables, state 5 may occur.
- FIG. 14 shows one embodiment of compare circuit 1206 n-1,0 and isolation circuit 1204 n-1,0 for implementing the truth table of Table 1.
- Compare circuit 1403 is one embodiment of compare circuit 1206 n-1,0
- isolation circuit 1401 is one embodiment of isolation circuit 1204 n-1,0 .
- the embodiment of FIG. 14 may be used to implement all of the priority logic elements 1104 in the priority index table.
- Compare circuit 1403 includes inverter 1414 , transistors 1406 and 1408 connected in series between priority signal line 1208 n-1 and ground, and transistors 1410 and 1412 connected in series between match line segment 1210 n-2,0 and ground.
- N-channel transistor 1406 has its drain coupled to signal line 1208 n-1 , it gate coupled to match line segment 1210 n-1,0 , and its source coupled to the drain of n-channel transistor 1408 .
- Transistor 1408 has its gate coupled to receive the logical complement of the priority number bit (/D) stored in memory element 1102 n-1,0 , and its source coupled to ground.
- N-channel transistor 1410 has its drain coupled to match line segment 1210 n-2,0 , its gate coupled to signal line 1208 n-1 via inverter 1414 , and its source coupled to the drain of n-channel transistor 1412 .
- Transistor 1412 has its gate coupled to receive the priority number bit (D) stored in memory element 1102 n-1,0 , and its source coupled to ground. Any of transistors 1406 , 1408 , 1410 , and 1412 can be replaced with other types of transistors and the logic adjusted accordingly.
- Isolation circuit 1401 includes inverters 1402 and 1404 .
- inverters 1402 and 1404 For alternative embodiments, only one inverter may be used and the logic of the next compare circuit adjusted accordingly.
- other isolation circuits such as one or more AND, OR, or XOR logic gates or pass gates may be used.
- FIG. 15 shows another embodiment of compare circuit 1206 n-1,0 and isolation circuit 1204 n-1,0 for implementing the truth table of Table 1.
- the compare circuit and isolation circuit are merged into the same logic that includes transistors 1406 and 1408 configured as in FIG. 14 , inverter 1501 , NAND gate 1502 , and AND gate 1503 .
- NAND gate 1502 has one input coupled to signal line 1208 n-1 via inverter 1501 , and another input coupled to receive the priority number bit (D) stored in memory element 1102 n-1,0 .
- AND gate 1503 has one input coupled to match line segment 1210 n-1,0 , another input coupled to the output of NAND gate 1502 , and an output coupled to match line segment 1210 n-2,0 .
- the most significant priority number is the lowest number such that 0101 is the most significant number between 0101 and 0110.
- the priority numbers are stored in descending priority order such that 0110 is the most significant priority number between 0101 and 0110.
- compare circuits 1206 0,0 - 1206 3,1 determine that 0110 is the most significant priority number, and assert IAD 0 to indicate that 0110 is stored in row zero of the priority index table.
- inverters 1212 3 - 1212 0 output the logical states of priority signal lines 1208 3 - 1208 0 , respectively, as PNUM 3 -PNUM 0 .
- inverters 1212 3 - 1212 0 are not required.
- Compare circuits 1206 0,0 - 1206 3,1 and isolation circuits 1204 0,0 - 1204 3,1 determine that 0110 is the highest priority number as follows.
- the most significant bit PNUM 3 is resolved first.
- memory elements 1102 that store a logic zero do not discharge their corresponding priority signal lines 1208 . Since both of memory elements 1102 3,1 , and 1102 3,0 store logic zero states, signal line 1208 3 remains pre-charged such that PNUM 3 is a logic zero.
- compare circuit 1206 3,1 compares the state of priority signal line 1208 3 with the priority number bit stored in 1102 3,1 and determines that both have different logical states. This causes compare circuit 1102 3,1 to not affect the logical state of match line segment 1210 2,1 such that match line segment 1210 2,1 has the same logic state as match line segment 1210 3,1 .
- compare circuit 1206 3,0 compares the state of priority signal line 1208 3 with the priority number bit stored in 1102 3,0 and determines that both have the different logical states. This causes compare circuit 1102 3,0 to not affect the logical state of match line segment 1210 2,0 such that match line segment 1210 2,0 has the same logic state as match line segment 1210 3,0 .
- PNUM 2 is resolved next.
- the corresponding priority signal line 1208 is discharged.
- each of memory compare circuits 1206 2,1 and 1206 2,0 discharge signal line 1208 2 such that PNUM 2 is a logic one.
- compare circuit 1206 2,1 does not affect match line segment 1210 1,1 , because memory element 1102 2,1 and priority signal line 1208 2 have different logic states.
- match line segment 1210 1,1 will have the same logic state as match line segment 1210 2,1 .
- compare circuit 1206 2,0 does not affect match line segment 1210 1,0 because memory element 1102 2,0 and priority signal line 1208 2 have different logic states.
- match line segment 1210 1,0 has the same state as match line segment 1210 2,0 .
- PNUM 1 is resolved next. Since memory element 1102 1,0 stores a logic one and match line segment 1210 1,0 is asserted, compare circuit 1206 1,0 discharges priority signal line 1208 1 . This causes PNUM 1 to be a logic one. Additionally, compare circuit 1206 1,0 compares the logic zero state of priority signal line 1208 1 with the logic one stored in 1102 1,0 and allows match line segment 1210 0,0 to have the same logic state as match line segment 1210 1,0 . Compare circuit 1206 1,1 , however, compares the logic zero on priority signal line 1208 1 with the logic zero stored in memory element 1102 1,1 , and de-asserts match line segment 1210 0,1 . As in the example of FIG.
- isolation circuits 1204 3,1 , 1204 2,1 , and 1204 1,1 isolate the de-asserted match line segment 1210 0,1 from match line segments 1210 3,1 , 1210 2,1 , and 1210 1,1 such that the PNUM 3 , PNUM 2 , PNUM 1 and are not affected by de-asserted match line segment 1210 0,1 .
- compare circuit 1206 0,0 alone determines PNUM 0 since compare circuit 1206 0,1 can not discharge priority signal line 1208 0 . Since memory element 1102 0,0 stores a logic zero and match line segment 1210 0,0 is asserted, compare circuit 1206 0,0 leaves priority signal line 1208 0 pre-charged, and PNUM 0 is a logic zero. Additionally, compare circuit 1206 0,0 allows IAD 0 to have the same logic state as match line segment 1210 0,0 . Since match line segment 1210 0,0 is asserted, IAD 0 will be asserted indicating that the most significant priority number is stored in that row.
- PNUM 3 -PNUM 0 indicate that the most significant priority number stored in the priority index table is 0110, and IAD 0 is asserted identifying that 0110 is stored in row zero.
- Table 2 shows one example of a truth table for implementing each compare circuit 1206 , where X (column) and Y (row) are any integers. Other truth tables may be used (and corresponding logic generated accordingly) including those that logically complement one of more of the signals indicated in Table 2.
- TABLE 2 STATE PNUM 1102 1210 X, Y 1210 X ⁇ 1, Y 0 0 0 0 1 0 0 1 1 2 0 1 0 0 3 0 1 1 1 4 1 0 0 0 5 1 0 1 0 6 1 1 0 0 7 1 1 1 1 1
- state 3 may not occur since PNUM will not be a logic zero when a memory element 1102 stores a logic one and the corresponding match line segment 1210 is asserted to a logic one state. For other truth tables, state 3 may occur.
- any logic or circuitry may be used to implement the truth table of Table 2.
- the compare circuit 1403 and isolation circuit 1401 of FIG. 14 may be used.
- transistor 1408 has its gate coupled to receive the priority number bit (D) stored in memory element 1102 n-1,0
- transistor 1412 has its gate coupled to receive the logical complement of the priority number bit (/D) stored in memory element 1102 n-1,0
- signal line 1208 n-1 provides the logical complement of PNUM n-1 .
- the logic and circuitry of FIG. 15 may be used to implement the truth table of Table 2.
- transistor 1408 has its gate coupled to receive the priority number bit (D) stored in memory element 1102 n-1,0
- NAND gate 1502 has an input coupled to receive the logical complement of the priority number bit (/D) stored in memory element 1102 n-1,0
- signal line 1208 n-1 provides the logical complement of PNUM n-1 .
- FIG. 12B shows another embodiment of 1201 0 of the priority index table.
- IAD 0 is determined on signal line 1216 that is separate from the match line segments.
- each match line signal line is coupled to a discharge circuit that includes an inverter-transistor pair (i.e., inverters 1212 n-1,0 - 1212 0,0 and corresponding transistors 1214 n-1 - 1214 0,0 ) such that if any of the match line segments is discharged, then signal line 1216 is discharged to a low logic state.
- inverter-transistor pair i.e., inverters 1212 n-1,0 - 1212 0,0 and corresponding transistors 1214 n-1 - 1214 0,0
- Signal line 1216 is pre-charged to a high logic state by pre-charge circuit 1202 .
- signal line 1216 may be driven directly by the match line segments.
- DSP 602 can identify the address in CAM array 604 of the highest priority policy statement that matches the policy information of the incoming packet. The identified address can then be used to access routing information stored in route memory 614 . DSP 602 can perform this function without the user having to preorder the policy statements for entry into the CAM array. This can reduce the hardware and/or software needed for table management of the CAM array, and can increase the performance of a router incorporating the DSP.
- DSP 602 can operate asynchronously or synchronously. When DSP 602 operates synchronously, it receives a clock signal that may be used to clock in the policy search key and an instruction that causes the process of FIG. 7 to be performed by DSP 602 . DSP 602 may implement the classification function of FIG. 7 in one or more clock cycles.
- priority numbers for policy statements may be assigned in ascending or descending priority order such that there are gaps left between the numbers to accommodate the new priority numbers associated with new policy statements to be stored in CAM array 604 .
- the priority numbers may be assigned in consecutive ascending or descending priority order.
- New policy statements and their associated priority numbers can be added to the tables 604 and 608 in conformance with either assignment method without having to reload or physically reorder the policy statements or the priority numbers in the respective tables.
- Each new policy statement can be loaded into any location (e.g., the next free location) in CAM array 604 , and can be assigned a new priority number without having to reload or reorder CAM array 604 and priority memory 608 .
- its priority number can be compared with the existing priority numbers already stored in priority memory 608 to determine if a policy statement already exists that has been assigned that priority. It is generally desirable that no two policy statements have the same priority number.
- the network administrator or the policy-based router itself can assign the new policy statement a new priority number, or the priority number of the existing policy statement can be updated (i.e., incremented or decremented). Since the existing priority numbers are stored in ascending or descending order, updating one priority number may also result in the need to update other priority numbers such that no two priority numbers are the same.
- DSP 602 may include additional circuitry to determine if at least one of the existing priority numbers stored in memory 608 is greater than or equal to (e.g., for ascending priority order), or, alternatively, less than or equal to (e.g., for descending priority order), the new priority number. If so, the existing priority numbers that are identified by the comparison may be updated such that a new policy statement does not have the same priority number as an existing policy statement. For other embodiments described below, priority logic 610 itself in priority index table 606 may perform the additional compare function(s) and control the updating function(s).
- FIG. 17 shows one embodiment of priority index table 606 of FIG. 6 including inequality circuits INEQ 1704 0 - 1704 N-1 that perform the additional comparison functions and control the updating functions.
- the inequality circuits determine if a new priority number PNEW 1710 of a new policy statement 1708 is the same, higher, or, alternatively, lower than any of the priority numbers already stored in memory 608 .
- the inequality circuits also update the priorities of the existing policy statements in order to insert the new policy statement.
- the new policy statement and its associated priority number do not need to be physically inserted between the existing entries; rather, they can be loaded into any desirable address (e.g., the next free address) in the CAM array and priority memory, and the priority numbers of the existing policy statements updated accordingly.
- the inequality circuits may update the priority numbers of the previously stored policy statements. These updating functions can be performed by the inequality circuits without the need to physically reorder the policy statements in the CAM array, or to physically reorder the priority numbers in the priority index table. This can reduce the hardware and/or software needed for table management of the CAM array, and can increase the performance of a router incorporating the DSP.
- the inequality circuits do not update the existing, stored priority numbers during or after an insert or delete operation.
- gaps may be left in the priority numbers after an insert or delete function. These gaps may or may not be filled with subsequent priority numbers.
- each location 608 0 - 608 N-1 in the priority memory has a corresponding inequality circuit 1704 0 - 1704 N-1 that compares the new priority number PNEW 1710 with the priority number P 0 -P Z stored at that location. If PNEW is the same or a higher priority (e.g., is greater than or equal to, or, alternatively less than or equal to) than an existing priority number, an inequality circuit will update the existing priority number.
- the new policy statement 1708 , PNEW, and the associated route information RI NEW can then be written into CAM array 604 , priority memory 608 , and route memory 614 , respectively, using conventional write circuits, counters, address decoders, etc. (not shown).
- Inequality circuits 1704 0 - 1704 N-1 may be part of priority index table 606 as shown in FIG. 17 .
- inequality circuits 1704 0 - 1704 N-1 may be separate from priority index table 606 .
- the inequality circuits may be any inequality circuit that determines whether one number is greater than or equal to, greater than, less than or equal to, or less than another number.
- inequality circuits 1704 0 - 1704 N-1 may be one circuit that consecutively, or in parallel, compares PNEW with each of the priority numbers P 0 -P Z .
- FIG. 18 shows one example of adding new policy statement 1708 to CAM array 604 at the next free address of CAM array 604 (location 6043 ) and updating the existing priority numbers stored in memory 608 .
- New policy statement 1708 is assigned priority number 1, and has route information RI 3 .
- CAM array 604 already includes policy statements 201 , 202 , and 203 stored at locations 604 0 , 604 1 , and 604 2 , respectively.
- the corresponding priority numbers 2, 1, and 0 are stored at locations 608 0 , 608 1 , and 608 2 , respectively, in priority memory 608 .
- the corresponding route information RI 0 , RI 1 , and RI 2 are stored at locations 614 0 , 614 1 , and 614 2 , respectively, of route memory 614 .
- the policy statements are assigned priority numbers in ascending priority order such that policy statement 203 is identified as the highest priority policy statement by being assigned priority number 0; policy statement 201 is identified as the lowest priority policy statement by being assigned priority number 2; and, policy statement 202 is identified as having a priority greater than that of policy statement 201 , but less than that of policy statement 203 , by being assigned priority number 1.
- Inequality circuits 1704 0 - 1704 2 compare PNEW with the exiting priority numbers as follows. The comparisons are made in parallel by each inequality circuit, however, they may also be made consecutively or in groups. Inequality circuit 1704 2 compares its corresponding priority number 0 with the new priority number 1 and determines that 0 is a higher priority than 1 (i.e., that 0 is numerically less than 1). The priority number for policy statement 203 is not updated. Inequality circuit 1704 1 compares its corresponding priority number 1 with the new priority number 1 and determines that they have the same priority. The priority number for policy statement 202 is incremented (updated) by one to the number 2.
- Inequality circuit 17040 compares its corresponding priority number 2 with the new priority number 1 and determines that 1 is a higher priority than 2 (i.e., that 1 is numerically less than 2). The priority number for policy statement 201 is then incremented to 3. The modified table entries after the insertion of the new policy statement and its associated priority number are shown in FIG. 19 .
- FIG. 20 shows another example of adding new policy statement 1708 to CAM array 604 at the next free address of CAM array 604 (location 604 3 ).
- policy statements 201 , 202 , and 203 have priority numbers 14, 15, and 16, respectively, and the new policy statement has priority number 15.
- the policy statements are assigned in descending priority order such that policy statement 203 is identified as the highest priority policy statement by being assigned priority number 16; policy statement 201 is identified as the lowest priority policy statement by being assigned priority number 14; and, policy statement 202 is identified as having a priority greater than that of policy statement 201 , but less than that of policy statement 203 , by being assigned priority number 15.
- Inequality circuit 1704 2 compares its corresponding priority number 16 with the new priority number 15 and determines that 16 is a higher priority than 15 (i.e., that 16 is numerically greater than 15).
- the priority number for policy statement 203 is not updated.
- Inequality circuit 1704 0 compares its corresponding priority number 15 with the new priority number 15 and determines that they have the same priority.
- the priority number for policy statement 202 is decremented (updated) by one to the number 14.
- Inequality circuit 1704 0 compares its corresponding priority number 14 with the new priority number 15 and determines that 15 is a higher priority than 14 (i.e., that 14 is numerically less than 15).
- the priority number for policy statement 201 is then decremented to 13.
- the modified table entries after the insertion of the new policy statement and its associated priority number is shown in FIG. 21 .
- FIG. 22 shows priority memory 2208 that is one embodiment of priority memory 608 .
- Priority memory 2208 includes counters 2208 0 - 2208 N-1 . Each counter corresponds to priority memory location 608 0 - 608 N-1 , respectively, and stores the priority number for that location. The counters can be loaded (i.e., written to) using write circuits, read from using read circuits, or reset as generally known in the art. Additionally, each counter 2208 0 - 2208 N-1 provides its stored priority number P 0 -P Z to a corresponding inequality circuit 1704 0 - 1704 N-1 for comparison with PNEW.
- the inequality circuit When PNEW has a higher priority than the stored priority number for a given location, then the inequality circuit will assert its count signal COUNT 0 -COUNT N-1 such that the corresponding counter is incremented (or, alternatively, decremented). The counter will be incremented if the UP signal is asserted, or will be decremented if the DOWN signal is asserted.
- the UP and DOWN signals may be generated by an instruction decoder (not shown) on DSP 602 that receives and decodes an insert (write) instruction or a delete (invalidate) instruction to the DSP.
- the inequality circuits may each output UP and DOWN signals to their corresponding counters directly in response to one or more signals from the instruction decoder.
- the counters may be updated synchronously or asynchronously.
- Inequality circuits 1704 0 - 1704 N-1 may be disposed external to priority memory 2208 , or they may be integrated into the priority memory itself.
- FIG. 23 shows one embodiment of integrating an inequality circuit with a counter in the priority memory.
- inequality circuit 2304 0 is one embodiment of inequality circuit 1704 0 , and may also be used for any of inequality circuits 1704 1 - 1704 N-1 .
- Counter 2306 0 is one embodiment of counter 2208 0 , and may also be used for any of counters 2208 1 - 2208 N-1 .
- memory storage elements 2314 n-1,0 - 2314 0,0 have associated counter logic circuits 2316 n-1,0 - 2316 0,0 to form a conventional counter.
- Each memory storage element can be a volatile or non-volatile RAM or ROM storage element such as memory storage elements 1102 of FIG. 11 .
- the memory storage elements each store one bit of a priority number such that memory storage element 2314 n-1,0 stores the most significant priority number bit, 2314 n-2,0 stores the next most significant priority number bit, and 2314 0,0 stores the least significant priority number bit.
- Logic circuits 2316 n-1,0 - 2316 0,0 receive the UP, DOWN, and COUNT 0 signals that indicate when the counter should increment, decrement, or leave unchanged the priority number stored in the counter.
- the counter can be configured to form any type of counter including the up/down counter shown. Additionally, load and reset signals can be provided to the logic circuits to write or reset the counter. Bit lines, word lines, and read/write circuits generally known in the art may also be included to access the priority number bits stored in the memory storage elements.
- Inequality circuit 2304 0 includes compare circuits 2310 n-1,0 - 2310 0,0 .
- Each compare circuit can be any digital or analog compare circuits.
- Each compare circuit 2310 n-1,0 - 2310 0,0 compares a priority number bit, stored in a corresponding memory storage element 2314 n-1,0 - 2314 0,0 , with one of the corresponding bits PNEW n-1 -PNEW 0 provided on signal lines 2308 n-1 - 2308 0 . Additionally, each compare circuit monitors the comparison result of the more significant priority number bits with PNEW via the logical states of control line segments 2312 n-2,0 - 2312 0,0 .
- the first control line segment 2312 n-1,0 may be pre-charged by pre-charge circuit 2302 to indicate that it is the first control line segment.
- control line segment 2312 n-1,0 may be initially discharged to indicate that it is the first control line segment.
- the logical state of COUNT 0 on signal line 2318 indicates the comparison result between PNEW and the priority number stored in counter 2306 0 .
- COUNT 0 is initially pre-charged by a pre-charge circuit 2302 to a high logic state. If COUNT 0 remains pre-charged after each compare circuit performs its comparison function, then the priority number in counter 2306 0 is updated. Alternatively, if COUNT 0 is discharged by one of the compare circuits, then the priority number in counter 2306 0 is updated.
- COUNT 0 may be initially discharged to a logic low state, and counter 2306 updated if COUNT 0 remains discharged, or, alternatively, if COUNT 0 is charged to a logic one state by one of the compare circuits.
- signal line 2318 may be segmented like control line segments 2312 between compare circuits, and the segmented signal line 2318 output from compare circuit 2310 0,0 provided to counter logic circuit 2316 0,0 as COUNT 0 .
- FIG. 24 The operation of the embodiment of FIG. 23 can be illustrated with the example shown in FIG. 24 .
- a 3 ⁇ 3 matrix of rows and columns stores three priority numbers.
- any numbers of rows and columns can be used.
- Row 0 stores priority number 010 having the decimal equivalent of the number 2
- row 1 stores priority number 001 having the decimal equivalent of the number 1
- row 2 stores priority number 000 having the decimal equivalent of the number 0.
- the priority numbers are assigned in ascending priority order such that 000 is the most significant priority number, 001 is the next most significant priority number, and 010 is the least significant priority number.
- inequality circuits 2304 2 - 2304 0 compare PNEW with their corresponding priority numbers stored in counters 2306 2 - 2306 0 , respectively. Inequality circuit 2304 2 determines that 000 is a higher priority than PNEW; inequality circuit 2304 1 determines that 001 is equal to PNEW; and inequality circuit 2304 0 determines that 010 is a lower priority than PNEW. Inequality circuit 2304 2 leaves its priority number unchanged. Inequality circuits 2304 1 and 2304 0 cause counters 2306 1 and 2306 0 to increment by one their existing priority numbers such that they do not have the same priority number as PNEW. The new policy statement and PNEW can then be loaded into CAM array 604 and priority memory 608 , respectively.
- Inequality circuits 2304 1 and 2304 0 cause their corresponding counters 2306 1 and 2306 0 to be incremented as follows.
- the comparisons of the most significant bit PNEW 2 with the most significant bits stored in counters 2306 2 - 2306 0 are resolved first.
- the compare circuit leaves the COUNT signal line unaffected, and drives the next control line segment to the same logical state of the previous control line segment. Since each of memory storage elements 2314 2,2 , 2314 2,1 , and 2314 2,0 store a logic zero and PNEW 2 is a logic zero, COUNT 2 , COUNT 1 , and COUNT 0 remain unaffected.
- control signals on control line segments 2312 1,2 , 2312 1,1 , and 2312 1,0 are asserted to high logic states.
- the inequality circuits cannot yet determine, based on the most significant bits alone, whether PNEW is greater than or equal to any of the stored priority numbers.
- compare circuit 2310 1,0 When compare circuit 2310 1,0 makes this determination, it leaves COUNT 0 unaffected so that it will remain pre-charged to update the priority number stored in counter 2306 0 . Additionally, compare circuit 2310 1,0 asserts the control signal on control line segment 2312 0,0 to a low logic state. When the control signal on control line segment 2310 0,0 is low, the control signals on all subsequent control line segments are also asserted to low logic states. When the control signals are low, all lower priority compare circuits are rendered incapable of discharging COUNT 0 regardless of what is stored in their corresponding memory elements, and regardless of the logical states of the subsequent PNEW bits. For example, 2310 0,0 will be rendered incapable of discharging COUNT 0 even though PNEW 0 , a logic one, is greater than the logic zero stored in memory element 2314 0,0 .
- inequality circuit 2304 0 has already determined that 010 is greater than 001 and thus COUNT 0 remains pre-charged to update the priority number stored in counter 2306 0 . Since the control signal on control line segment 2312 0,1 is high, compare circuit 2310 0,1 compares the logic one stored in memory element 2314 0,1 with the logic one of PNEW 0 and determines that PNEW is the same number as that stored in counter 2306 1 . In response, COUNT 1 remains in a pre-charged state to update the priority number stored in counter 2306 1 .
- Compare circuit 2310 0,2 determines that PNEW 0 is greater than the logic zero stored in memory element 2314 0,2 . Since the control signal is high on control line segment 2312 0,2 , compare circuit 2310 0,2 discharges COUNT 2 such that the priority number 000 stored in counter 2306 2 will not be updated. After a sufficient amount of time has passed such that all of the bit comparisons have had time to resolve, the UP signal can be asserted such that the priority numbers in counters 2306 1 and 2306 0 are incremented. For one embodiment, the COUNT signals can be latched prior to providing the UP signal.
- Table 3 shows one example of a truth table for implementing each compare circuit 2310 , where X (column) and Y (row) are any integers. Other truth tables may be used (and corresponding logic generated accordingly) including those that logically complement one or more of the signals indicated in Table 3.
- Table 3 PNEW X 2314 X, Y 2312 X, Y 2312 X ⁇ 1, Y COUNT 0 0 0 0 COUNT 0 0 1 1 COUNT 0 1 0 0 COUNT 0 1 1 0 COUNT 1 0 0 0 COUNT 1 0 1 1 0 COUNT 1 0 0 0 COUNT 1 0 1 1 1 0 0 COUNT 1 1 1 1 COUNT
- FIG. 25A shows compare circuit 2502 that is one embodiment of compare circuit 2310 n-1,0 for implementing the truth table of Table 3.
- the embodiment of FIG. 25A may be used to implement all of the compare circuits 2310 .
- Compare circuit 2502 includes inverter 2510 , NAND gate 2508 , AND gate 2506 , and n-channel transistors 2504 , 2503 , and 2507 .
- NAND gate 2508 has one input coupled to signal line 2308 n-1 via inverter 2510 , and another input coupled to receive the data (D) stored in memory element 2314 n-1,0 .
- AND gate 2506 has one input coupled to the output of NAND gate 2508 , another input coupled to control line segment 2312 n-1,0 , and an output coupled to control line segment 2312 n-2,0 .
- Transistor 2504 has its drain coupled to the signal line 2318 , its gate coupled to the output of AND gate 2506 , and its source coupled to the drain of transistor 2503 .
- Transistor 2503 also has its gate controlled by signal line 2308 n-1 , and its source coupled to the drain of transistor 2507 .
- Transistor 2507 also has its gate coupled to receive /D from memory 2314 n-1,0 , and its source coupled to ground.
- FIG. 25B shows compare circuit 2522 that is another embodiment of compare circuit 2310 n-1,0 for implementing the truth table of Table 3.
- signal line 2318 is segmented into separate control line segments.
- the embodiment of FIG. 25B may be used to implement all of the compare circuits 2310 .
- Compare circuit 2522 includes inverter 2512 , NAND gates 2514 and 2516 , and AND gates 2518 and 2520 .
- NAND gate 2514 has one input coupled to signal line 2308 n-1 via inverter 2512 , and another input coupled to D of memory 2314 n-1,0 .
- AND gate 2520 has one input coupled to control line segment 2312 n-1,0 , another input coupled to the output of NAND gate 2514 , and an output coupled to control line segment 2312 n-2,0 .
- NAND gate 2516 has a first input coupled to control line segment 2312 n-1,0 , a second input coupled to /D of memory 2314 n-1,0 , and a third input coupled to signal line 2308 n-1 .
- AND gate 2518 has a first input coupled to count line segment 2318 n-1,0 , a second input coupled to the output of NAND gate 2516 , and an output coupled to count line segment 2318 n-2,0 .
- the most significant number is the lowest number such that 000 is the most significant number between 000, 001, and 010.
- the priority numbers are assigned in descending priority order such that the most significant priority number is the priority number with the highest numerical value, and the least significant priority number is the priority number with the lowest numerical value.
- FIG. 26 shows such an example in which the 3 ⁇ 3 matrix of FIG. 24 includes the binary priority number 111 (decimal equivalent of 7) stored in counter 23062 , priority number 110 (decimal equivalent of 6) stored in counter 2306 1 , and priority number 101 (decimal equivalent of 5) stored in counter 23060 . Numbers 7, 6, and 5 are stored in rows 2, 1, and 0, respectively.
- 111 is the most significant priority number
- 110 is the next most significant priority number
- 101 is the least significant priority number.
- Inequality circuits 2304 2 - 2304 0 compare PNEW with their corresponding priority numbers stored in counters 2306 2 - 2306 0 , respectively. Inequality circuit 2304 2 determines that 111 is a higher priority than PNEW; inequality circuit 2304 1 determines that 110 is equal to PNEW; and inequality circuit 2304 0 determines that 101 is a lower priority than PNEW. Inequality circuit 2304 2 leaves its priority number unchanged.
- Inequality circuits 2304 0,1 and 2304 0,0 each cause their corresponding counters 2306 1 and 2306 0 to decrement by one their existing priority numbers such that they do not have the same priority number as PNEW.
- the new policy statement and PNEW can then be loaded into CAM array 604 and priority memory 604 , respectively.
- Inequality circuits 2304 1 and 2304 0 cause their corresponding counters 2306 1 and 2306 0 to be decremented as follows.
- the comparisons of the most significant bit PNEW 2 with the most significant bits stored in counters 2306 2 - 2306 0 are resolved first.
- the compare circuit leaves the COUNT signal line unaffected, and drives the next control line segment to the same logical state of the previous control line segment. Since each of memory storage elements 2314 2,2 , 2314 2,1 , and 2314 2,0 store a logic one and PNEW 2 is a logic one, COUNT 2 , COUNT 1 , and COUNT 0 remain unaffected.
- control signals on control line segments 2312 1,2 , 2312 1,1 , and 2312 1,0 are asserted to high logic states.
- the inequality circuits cannot yet determine, based on the most significant bits alone, whether PNEW is less than or equal to any of the stored priority numbers.
- compare circuit 2310 1,0 When compare circuit 2310 1,0 makes this determination, it leaves COUNT 0 unaffected so that it will remain pre-charged to update the priority number stored in counter 2306 0 . Additionally, compare circuit 2310 1,0 asserts the control signal on control line segment 2312 0,0 to a low logic state. When the control signal on control line segment 2310 0,0 is low, the control signals on all subsequent control line segments are also asserted to low logic states. When the control signals are low, all lower priority compare circuits are rendered incapable of discharging COUNT 0 regardless of what is stored in their corresponding memory elements, and regardless of the logical states of the subsequent PNEW bits. For example, 2310 0,0 will be rendered incapable of discharging COUNT 0 even though PNEW 0 , a logic zero, is less than the logic one stored in memory element 2314 0,0 .
- inequality circuit 2304 0 has already determined that 101 is less than 110 and thus COUNT 0 remains pre-charged to update the priority number stored in counter 2306 0 . Since the control signal on control line segment 2312 0,1 , is high, compare circuit 2310 0,1 compares the logic zero stored in memory element 2314 0,1 with the logic one on line 2308 0 and determines that PNEW is the same number as that stored in counter 2306 1 . In response, COUNT 1 remains in a pre-charged state to update the priority number stored in counter 2306 1 .
- Compare circuit 2310 0,2 determines that PNEW 0 is less than the logical one stored in memory element 2314 0,2 . Since the control signal is high on control line segment 2312 0,2 , compare circuit 2310 0,2 discharges COUNT 2 such that the priority number 111 stored in counter 2306 2 will not be updated. After a sufficient amount of time has passed such that all of the bit comparisons have had time to resolve, the DOWN signal can be asserted such that the priority numbers in counters 2306 1 and 2306 0 are decremented. For one embodiment, the COUNT signals can be latched prior to providing the DOWN signal.
- Table 4 shows one example of a truth table for implementing each compare circuit 2310 , where X (column) and Y (row) are any integers. Other truth tables may be used (and corresponding logic generated accordingly) including those that logically complement one or more of the signals indicated in Table 4.
- Table 4 PNEW X 2314 X, Y 2312 X, Y 2312 X ⁇ 1, Y COUNT 0 0 0 0 COUNT 0 0 1 1 COUNT 0 1 0 0 0 COUNT 0 1 1 1 0 1 0 0 0 COUNT 1 0 1 0 COUNT 1 1 0 0 COUNT 1 1 1 1 COUNT 1 1 1 COUNT
- FIG. 27A shows compare circuit 2702 that is one embodiment of compare circuit 2310 n-1 ,0 the truth table of Table 4. The embodiment of FIG. 27A may be used to implement all of the compare circuits 2310 .
- Compare circuit 2702 includes inverter 2510 , NAND gate 2508 , AND gate 2506 , and n-channel transistors 2504 , 2503 , and 2507 as in FIG. 25A .
- NAND gate 2508 has one input coupled to signal line 2308 n-1 , and another input coupled to receive the /D stored in memory element 2314 n-1,0 .
- AND gate has one input coupled to the output of NAND gate 2508 , another input coupled to control line segment 2312 n-1,0 , and an output coupled to control line segment 2312 n-2,0 .
- Transistor 2504 has its drain coupled to the signal line 2318 , its gate coupled to the output of AND gate 2506 , and its source coupled to the drain of transistor 2503 .
- Transistor 2503 also has its gate controlled by signal line 2308 n-1 via inverter 2510 , and its source coupled to the drain of transistor 2507 .
- Transistor 2507 also has its gate coupled to receive D from memory 2314 n-1,0 , and its source coupled to ground.
- FIG. 27B shows compare circuit 2704 that is another embodiment of compare circuit 2310 n-1,0 for implementing the truth table of Table 4.
- signal line 2318 is segmented into separate control line segments.
- the embodiment of FIG. 27B may be used to implement all of the compare circuits 2310 .
- Compare circuit 2704 includes inverter 2708 , NAND gate 2710 , and AND gates 2706 and 2712 .
- NAND gate 2710 has a first input coupled to signal line 2308 n-1 , a second input coupled to D of memory 2314 n-1,0 , and a third input coupled to control line segment 2312 n-1,0 via inverter 2708 .
- AND gate 2706 has one input coupled to count line segment 2318 n-1,0 , another input coupled to the output of NAND gate 2710 , and an output coupled to count line segment 2318 n-2,0 .
- AND gate 2712 has a first input coupled to control line segment 2312 n-1,0 , a second input coupled to /D of memory 2314 n-1,0 , a third input coupled to signal line 2308 n-1 , and an output coupled to control line segment 2312 n-2,0 .
- FIGS. 17-27 show that a new policy statement having a priority greater than or equal to an existing policy statement, can be added to the DSP without having to physically reorder or reload the CAM array storing the policy statements, or the memory storing the priority numbers. Instead, inequality circuits update the appropriate existing priority numbers. Since the updating function can be performed without the need to physically reorder or reload the policy statements in the CAM array, or to physically reorder or reload the priority numbers in the priority index table, this can reduce the hardware and/or software needed for table management of the CAM array, and can increase the performance of a router incorporating the DSP.
- Policy statements may also be deleted from CAM array 604 using priority index table 606 of FIG. 17 .
- priority index table 606 of FIG. 17 When a policy statement is deleted, the corresponding priority number in priority memory 608 is also deleted.
- the policy statements and priority numbers can be deleted by setting one or more valid bits to an appropriate state for the row of CAM array 604 that stores the policy statement to be deleted.
- the valid bit(s) may be stored in CAM array 604 , priority memory 608 , or in each of the memory arrays.
- priority memory 608 When a priority number is deleted, the remaining priority numbers in priority memory 608 can be left unchanged. This may leave gaps in the priority numbers stored in the priority memory. These gaps may be filled in by new priority numbers associated with new policy statements to be added to the DSP, or they may remain as unused gaps. For another embodiment, remaining priority numbers in the priority memory can be updated by the inequality and counter circuits. If the priority numbers are assigned in ascending priority order, and one of the priority numbers is deleted or invalidated, then any other priority numbers that are greater than the deleted number can be decremented by the inequality circuits. For example, if priority memory 608 stores priority numbers 0, 1, and 2, and priority number 1 is invalidated, then priority number 2 can be decremented to 1.
- priority numbers are assigned in descending priority order, and one of the priority numbers is deleted or invalidated, then any other priority numbers that are less than the deleted number can be incremented by the inequality and counter circuits. For example, if priority memory 608 stores priority numbers 0, 1, and 2, and priority number 1 is invalidated, then priority number 0 can be incremented to 1. Updating the priority numbers can help to avoid unused gaps in priority memory 608 .
- FIG. 28 shows one embodiment of a process 2800 for deleting or invalidating policy statements and their associated priority numbers.
- the priority number to be deleted is compared with the entries in the priority memory.
- priority memory 608 is a CAM, or has one or more compare circuits external to the memory to perform the comparison.
- step 2802 may compare the policy statement with the existing entries in CAM array 604 .
- priority memory 608 may not be a CAM.
- step 2804 the process stops at step 2804 .
- the process could also stop if there was more than one match, and an error flag such as a multiple match flag may be asserted.
- the process may continue to delete all identified matches. If there is a match, both the policy statement and the priority number are invalidated or deleted at step 2806 . As described above, this may be accomplished by setting a valid bit(s) to an appropriate state for that row in priority memory 608 . If the valid bit(s) is located in CAM array 604 , and is inaccessible by priority memory 608 , then priority logic 610 can then identify the location of the matching entry (as described above), and provide the indication to IAD 0 -IAD N-1 .
- signal lines can then be provided through encoder 612 back to CAM array 604 (e.g., through a decoder) to address the appropriate location and invalidate the policy statement.
- signal lines IAD 0 -IAD N-1 can be provided directly to CAM array 604 to access the appropriate location for invalidation.
- the process may stop after step 2806 and potentially leave gaps in the priority numbers. Alternatively, the process may proceed to step 2808 to update the remaining entries in the priority memory.
- the deleted priority number is provided to the inequality circuits 1704 0 - 1704 N-1 , at step 2808 , to determine which entries in the priority memory have a lower priority than the deleted priority number.
- the inequality and counter circuits may then update the identified numbers, at step 2810 , as previously described.
- the delete functions illustrate one example of the DSP 602 utilizing both the inequality circuits and the priority logic to operate on the priority numbers stored in priority memory 608 .
- FIG. 29 shows one example of combining an inequality circuit with a priority logic circuit into a single array such that each circuit operates on the priority numbers on a bit-by-bit basis.
- the embodiment of FIG. 29 combines row 1201 0 of priority logic elements from FIG. 12A with inequality circuit 2304 0 and counter 2306 0 of FIG. 23 to form row zero in priority index table 606 of FIG. 17 .
- row 1201 0 from FIG. 12B may also be used.
- the priority logic elements and inequality circuits share memory elements 2314 0,0 - 2314 n-1,0 .
- Memory elements 2314 0,0 - 2314 n-1,0 also correspond to memory elements 1102 0,0 - 1102 n-1,0 .
- Conventional bit lines, word lines, read and write circuits, and optional comparand data lines and match lines i.e., when memory elements 2314 0,0 - 2314 n-1,0 are CAM cells) are not shown so as not to obscure the teachings of the figure.
- FIG. 30 shows another embodiment of combining the row 1201 0 of priority logic elements from FIG. 12A with the inequality circuit 2304 0 of FIG. 23 to form row zero in priority index table 1706 of FIG. 17 .
- signal lines 1208 n-1 - 1208 0 and 2308 n-1 - 2308 0 are replaced with a single set of signal lines 2906 n-1 - 2906 0 that can be used to determine the most significant priority number PNUM, or used to carry PNEW.
- match line segments 1210 n-1,0 - 1210 0,0 serve a dual function to operate as match line segments when determining the most significant priority number, and as control line segments (i.e., 2312 n-1,0 - 2312 0,0 of FIG. 23 ) to generate COUNT 0 when determining when a new priority number has a priority that is greater than or equal to the priority number stored in counter 2306 .
- compare circuit 2310 n-1,0 is coupled to match line segment 1210 n-2,0
- compare circuit 2310 n-2,0 is coupled to match line segment 1210 n-3,0
- COUNT 0 is provided on signal line IAD 0 .
- compare circuits 2310 may provide control signals to isolation circuits 1204 instead of directly controlling the match line segments.
- a control signal BCT indicates whether the priority index table uses the row of priority logic elements 1201 0 to determine the most significant priority number in priority memory 608 , or whether the priority index table uses the inequality circuit 2304 0 to determine if a new priority number has a priority that is greater than or equal to the priority number stored in counter 2306 .
- BCT may be provided by an instruction decoder in DSP 602 that interprets and decodes instructions provided to the DSP. When BCT is a low logic state, the priority logic elements perform their priority function; and when BCT is a high logic state, the inequality circuits perform their inequality function.
- compare circuit 1206 n-1,0 when BCT is low, compare circuit 1206 n-1,0 is enabled, compare circuit 2310 n-1,0 is disabled, and logic 2902 provides the state of ML 0 to match line segment 1210 n-1,0 .
- compare circuit 1206 n-1,0 When BCT is high, compare circuit 1206 n-1,0 is disabled, compare circuit 2310 n-1,0 is enabled, and logic 2902 drives match line segment 1210 n-1,0 to a high logic state such that match line segments 1210 n-1,0 - 1210 0,0 function as control line segments 2312 n-1,0 - 2312 0,0 of FIG. 23 .
- Logic 2902 may be any logic that performs this function.
- row 1201 0 of priority logic elements from FIG. 12B may be combined with the inequality circuit 2304 0 of FIG. 23 to form row zero in priority index table 1706 of FIG. 17 .
- the logic and circuitry for implementing the priority logic elements and the inequality circuits illustrated in FIGS. 29 and 30 may be those that implement the truth tables shown in Tables 1, 2, 3, and 4 above.
- the logic and circuits illustrated in FIGS. 14, 15 , 25 A, 25 B, 27 A, and 27 B may be used.
- the priority numbers are assigned in ascending priority order in priority memory 608 .
- logic and circuitry that implements the truth table shown in Table 1 may be used for the priority logic elements to determine the lowest priority number as the most significant priority number.
- the logic and circuitry shown in the truth table shown in Table 3 may be used for the inequality circuits to determine if a new priority number has a priority that is greater than or equal to the priority number stored in counter 2306 .
- FIG. 31 shows one example of combining the logic and circuitry of FIGS. 14 and 25 to implement the embodiment of FIG. 30 when the priority numbers are assigned in ascending priority order in priority memory 608 .
- OR gate 3102 is one embodiment of logic 2902 .
- OR gate 3102 has one input coupled to ML 0 , and another input coupled to BCT.
- BCT is also coupled to the gate of n-channel transistor 3106 and one input of OR gate 3104 via inverter 3108 .
- Transistor 3106 has its drain coupled to signal line 2906 n-1 , and its source coupled to the drain of transistor 1406 .
- OR gate 3104 has another input coupled to the output of AND gate 2506 , and has its output coupled to the gate of transistor 2504 .
- Transistor 2504 is coupled between line segment 1210 n-2,0 and node 3112 .
- Transistor pairs 1410 / 1412 and 2503 / 2507 are each coupled in series between node 3112 and ground.
- the priority numbers are assigned in descending priority order in priority memory 608 .
- logic and circuitry that implements the truth table shown in Table 2 may be used for the priority logic elements to determine the lowest priority number as the most significant priority number.
- the logic and circuitry shown in the truth table shown in Table 4 e.g., the logic and circuitry of FIG. 27A or 27 B
- FIG. 32 shows one example of combining the logic and circuitry of FIGS. 14 and 27 to implement the embodiment of FIG. 30 when the priority numbers are assigned in descending priority order in priority memory 608 .
- FIGS. 17-32 include additional inequality circuits to compare PNEW with the priority numbers already stored in the priority index table.
- priority logic 610 itself can be used to determine whether to update existing priority numbers stored in the priority index table in response to an insert or delete instruction.
- the new policy statement and PNEW do not need to be physically inserted between the existing entries; rather, they can be loaded into any desirable address (e.g., the next free address) in the CAM array and priority memory, and the priority numbers of the existing policy statements updated accordingly.
- the priority logic may update the priority numbers of the previously stored policy statements.
- These updating functions can be performed by the priority logic without the need to physically reorder the policy statements in the CAM array, or to physically reorder the priority numbers in the priority index table. This can reduce the hardware and/or software needed for table management of the CAM array, and can increase the performance of a router incorporating the DSP.
- priority logic 610 can determine whether the priority numbers already stored in memory 608 should be incremented as follows. PNEW is decremented by one (or another value) and then compared with the existing priority numbers stored in memory 608 by priority logic 610 . For this embodiment, the priority numbers in memory 608 are not compared with each other, but with the decremented PNEW. Since priority logic 610 determines which compared number is the lowest numerical value, it can also determine the logical converse; that is, it can determine which of the stored priority numbers are greater than the decremented PNEW. Each priority number that is greater than the decremented PNEW is then incremented by the priority logic. PNEW is then added to memory 608 .
- FIG. 33 shows one example of inserting policy statement 1708 and PNEW into memory 608 having priority numbers 0, 1, and 2 assigned in ascending priority order. This is the same example shown in FIGS. 18 and 19 , except that priority logic 610 is used to control updating of the priority numbers stored in memory 608 .
- PNEW having a value of 1, is decremented by 1 by decrement circuit 3304 to form PSUB having a value of zero. PSUB is then provided to priority table 606 via multiplexer 3302 .
- Decrement circuit 3304 may be any subtraction circuit that decrements PNEW by one or another value.
- Multiplexer 3302 provides PSUB to priority index table 606 in response to control signal INST, which indicates that an insert function is to take place.
- INST may be generated by an instruction decoder (not shown) that receives an insert (write) instruction.
- PSUB is compared with the stored priority numbers by priority logic 610 .
- logic 3306 asserts all match lines ML 0 -ML N-1 to a particular logic state to indicate that all stored policy statements match policy statement 1708 (even though no actual comparison takes place) such that the priority logic will compare all of the priority numbers in memory 608 with PSUB.
- Logic 3306 may be any logic circuit including, for example, AND, OR, XOR, or inverter circuits.
- BCT is a control signal that indicates an insert or delete function is being performed by DSP 602 .
- BCT may be generated by an instruction decoder (not shown).
- BCT may also be provided to priority logic 610 to indicate that priority logic 610 should compare PSUB with each entry in memory 608 rather than comparing each entry in memory 608 with each other as in the classification function of FIG. 7 .
- Priority logic 610 determines that stored priority number 0 is equal to PSUB, and leaves this stored number unchanged. Priority logic 610 also determines that stored priority numbers 1 and 2 are larger than PSUB, and increments these numbers to 2 and 3, respectively. Policy statement 1708 and PNEW are then loaded into a free address in array 604 and memory 608 , respectively. The modified table entries after the insertion are shown in FIG. 34 .
- decrement circuit 3304 would generate a negative number. For one embodiment, only priority numbers greater than 0 may be used.
- offset circuitry may be included in DSP 602 to add an offset value to PNEW before it is provided to decrement circuit 3304 and/or multiplexer 3302 . The offset circuitry may also be used to subtract the offset value when reading priority numbers from memory 608 such that the offset function is transparent to the user of DSP 602 .
- DSP 602 can detect that PNEW is equal to zero when an insert instruction is received by DSP 602 , or when an underflow condition occurs in decrement circuit 3304 .
- PNEW can then be incremented by one and be provided to decrement circuit 3304 , or PNEW can be provided directly to priority index table 606 by multiplexer 3302 . After all the priority numbers greater than zero are detected and updated by priority logic 610 , then memory 608 can be searched for any existing entry that already has priority number 0. If so, this number can be incremented (i.e., by using the COUNT and UP signals, or by overwriting priority number 0 with priority number 1), and then PNEW added to memory 608 .
- a policy statement and its priority number may also be deleted or removed using priority logic 610 .
- policy statement 1708 may be removed from array 604 , and its priority number 1 may be removed from memory 608 as follows.
- the priority number ( 3308 ) to be deleted is provided to policy index table 606 via multiplexer 3302 (INST logic low), and is compared with the entries in the priority memory.
- priority memory 608 is a CAM, or has one or more compare circuits external to the memory to perform the comparison.
- the policy statement associated with priority number 3308 may be compared with the existing entries in CAM array 604 . Since location 6083 matches priority number 3308 , both the policy statement and the priority number are invalidated or deleted.
- this may be accomplished by setting a valid bit(s) to an appropriate state for that row in priority memory 608 . If the valid bit(s) is located in CAM array 604 , and is inaccessible by priority memory 608 , then priority logic 610 can then identify the location of the matching entry (as described above), and provide the indication to IAD 0 -IAD N-1 . These signal lines can then be provided through encoder 612 back to CAM array 604 (e.g., through a decoder) to address the appropriate location and invalidate the policy statement. Alternatively, the signal lines IAD 0 -IAD N-1 can be provided directly to CAM array 604 to access the appropriate location for invalidation.
- the delete process may stop at this point and potentially leave gaps in the priority numbers. Alternatively, the process may update the remaining entries in the priority memory.
- the deleted priority number 3308 is provided to policy index table 606 via multiplexer 3302 , to determine which entries in the priority memory have a lower priority than the deleted priority number.
- Priority logic 610 identifies priority numbers 2 and 3 as having lower priorities (i.e., larger numerical values), and these numbers are decremented.
- the resultant table entries are as shown in FIG. 33 .
- decrement circuit 3304 may be replaced with an increment circuit that increments PNEW prior to supplying it to priority index table 606 for comparison as part of an insert function.
- the increment circuit would generate a number larger than can be handled by priority index table 606 .
- priority index table 606 For one embodiment, only priority numbers less than 2 n ⁇ 1 may be used.
- offset circuitry may be included in DSP 602 to subtract an offset value to PNEW before it is provided to the increment circuit and/or multiplexer 3302 . The offset circuitry may also be used to add the offset value when reading priority numbers from memory 608 such that the offset function is transparent to the user of DSP 602 .
- DSP 602 can detect that PNEW is equal to 2 n ⁇ 1 when an insert instruction is received by DSP 602 , or when an overflow condition occurs in the increment circuit. PNEW can then be decremented by one and be provided to the increment circuit, or PNEW can be provided directly to priority index table 606 by multiplexer 3302 . After all the priority numbers greater than 2 n ⁇ 1 are detected and updated by priority logic 610 , then memory 608 can be searched for any existing entry that already has priority number 2 n ⁇ 1. If so, this number can be decremented (i.e., by using the COUNT and DOWN signals, or by overwriting priority number 2 n ⁇ 1 with priority number 2 n ⁇ 2), and then PNEW added to memory 608 .
- Priority memory 608 may be formed as counters that store the priority numbers.
- FIG. 35 shows one embodiment of a row of priority memory 608 in which the priority logic 610 includes a separate priority logic element for each bit of a priority number.
- the priority logic includes row 1201 0 of priority logic elements from FIG. 12A , coupled to receive priority number bits stored in counter 2306 0 from FIG. 23 .
- PSUB bits PSUB n-1 -PSUB 0 are provided on signal lines 1208 n-1 - 1208 0 , respectively, for comparison by compare circuits 1206 n-1,0 - 1206 0,0 with the priority bits stored in memory elements 2314 n-1,0 - 2314 0,0 .
- compare circuits 1206 n-1,0 - 1206 0,0 With the priority bits stored in memory elements 2314 n-1,0 - 2314 0,0 .
- logic 3306 0 drives match line segment 1210 n-1,0 to a high logic state, and PSUB is provided on signal lines 1208 .
- BCT is also provided to each of compare circuits 1206 such that the circuits do not provide their compare results back to signal lines 1208 . Thus, only PSUB will be supplied on signal lines 1208 .
- a separate set of signal lines may be used for the updating functions and the classification function.
- BCT is asserted to a low logic state
- ML 0 is provided to match lines segment 1210 n-1,0 and the classification function of FIG. 7 may be performed.
- the logical state of IAD 0 indicates the comparison result between PSUB and the priority number stored in counter 23060 . If IAD 0 indicates that PSUB has a more significant priority than the priority number stored in counter 2306 0 , then counter 2306 0 is incremented. If, however, IAD 0 indicates that PSUB is equal to or has a less significant priority number than the priority number in counter 2306 0 , then counter 2306 0 is not incremented.
- FIG. 36 The operation of the embodiment of FIG. 35 to update priority numbers in response to an insert instruction can be illustrated with the example shown in FIG. 36 .
- a 3 ⁇ 3 matrix of rows and columns stores three priority numbers.
- any numbers of rows and columns can be used.
- Row 0 stores priority number 010 having the decimal equivalent of the number 2
- row 1 stores priority number 001 having the decimal equivalent of the number 1
- row 2 stores priority number 000 having the decimal equivalent of the number 0.
- the priority numbers are assigned in ascending priority order such that 000 is the most significant priority number, 001 is the next most significant priority number, and 010 is the least significant priority number.
- priority logic elements 1201 2 - 1201 0 compare PSUB 000 (PNEW decremented by 1) with their corresponding priority numbers stored in counters 2306 2 - 2306 0 , respectively.
- Priority logic 1201 2 determines that 000 is equal to PSUB; priority logic 1201 1 determines that 001 is greater than PSUB; and priority logic 1201 0 determines that 010 is greater than PSUB.
- Priority logic 1201 2 leaves unchanged the priority number stored in counter 2306 2 . Circuits 1201 1 and 1201 0 cause counters 2306 1 and 2306 0 to increment by one their existing priority numbers. The new policy statement and PNEW can then be loaded into CAM array 604 and priority memory 608 , respectively.
- Compare circuit 1206 0,1 determines that PSUB 0 is less than the logic one stored in memory element 2314 0,1 . Since the control signal is high on match line segment 1210 0,1 , compare circuit 1206 0,1 discharges IAD 1 such that the priority number 001 stored in counter 2306 1 will be incremented. After a sufficient amount of time has passed such that all of the bit comparisons have had time to resolve, the UP signal can be asserted such that the priority numbers in counters 2306 1 and 2306 0 are incremented. For one embodiment, the IAD signals can be latched prior to providing the UP signal. Subsequently, PNEW can be added to memory 608 .
- any compare circuits may be used for compare circuits 1206 to implement the process illustrated above.
- the compare circuits illustrated above that implement Table 1 can be used with additional control for the BCT control signal.
- One embodiment of the modified circuit of FIG. 14 is shown in FIG. 37 .
- transistor 1416 is included in series with transistors 1406 and 1408 between signal line 1208 n-1 and ground.
- the gate of transistor 1416 is controlled by BCT via inverter 1420 .
- OR gate 1418 is one embodiment of logic 3302 .
- FIG. 36 shows an embodiment of inserting a new priority number into memory 608
- the priority numbers can be decremented for a delete function by asserting the DOWN signal after all of the IAD signals are resolved.
- FIG. 35 can also insert and delete priority numbers assigned in descending priority order.
- FIG. 38 shows such an example with of a 3 ⁇ 3 matrix in which row 0 stores priority number 111 having the decimal equivalent of the number 7, row 1 stores priority number 110 having the decimal equivalent of the number 6, and row 2 stores priority number 101 having the decimal equivalent of the number 5.
- 111 is the most significant priority number
- 110 is the next most significant priority number
- 101 is the least significant priority number.
- priority logic elements 1201 2 - 1201 0 compare PSUB 111 (PNEW incremented by 1) with their corresponding priority numbers stored in counters 2306 2 - 2306 0 , respectively.
- Priority logic 1201 2 determines that 111 is equal to PSUB; priority logic 1201 1 determines that 110 is less than PSUB; and priority logic 1201 0 determines that 101 is less than PSUB.
- Priority logic 1201 2 leaves unchanged the priority number stored in counter 2306 2 . Circuits 1201 1 and 1201 0 cause counters 2306 1 and 2306 0 to decrement by one their existing priority numbers. The new policy statement and PNEW can then be loaded into CAM array 604 and priority memory 608 , respectively.
- priority logic 1201 0 has already determined that 101 is less than 111 and thus IAD 0 remains discharged to a low logic state. Since the control signal on match line segment 1210 0,2 is in a high logic state, compare circuit 1206 0,1 compares the logic one stored in memory element 2314 0,2 with the logic one of PSUB 0 , and determines that PSUB is the same number as that stored in counter 2306 2 . In response, IAD 2 remains in a high logic state such that counter 2306 2 is not updated.
- Compare circuit 1206 0,1 determines that PSUB 0 is greater than the logic zero stored in memory element 2314 0,1 . Since the control signal is asserted on match line segment 1210 0,1 , compare circuit 1206 0,1 discharges IAD 1 such that the priority number 110 stored in counter 2306 1 will be decremented. After a sufficient amount of time has passed such that all of the bit comparisons have had time to resolve, the DOWN signal can be asserted such that the priority numbers in counters 2306 1 and 2306 0 are decremented. For one embodiment, the IAD signals can be latched prior to providing the DOWN signal. Subsequently, PNEW can be added to memory 608 .
- compare circuits 1206 may be used for compare circuits 1206 to implement the process illustrated in FIG. 38 .
- the compare circuits illustrated above that implement Table 3 can be used with additional control for the BCT control signal as illustrated in FIG. 37 .
- FIG. 38 shows an embodiment of inserting a new priority number into memory 608 , the priority numbers can be incremented for a delete function by asserting the UP signal after all of the IAD signals are resolved.
- DSP 602 can perform the updating functions described above with respect to FIGS. 17-38 asynchronously or synchronously.
- DSP 602 When DSP 602 operates synchronously, it receives a clock signal that may be used to clock in the policy search key and an instruction that causes the updating functions to be performed by DSP 602 .
- DSP 602 may implement the updating functions in one or more clock cycles.
- DSP 602 stores policy statements in CAM array 604 and identifies the highest priority matching policy statement without having to presort or prearrange the policy statements in the CAM array.
- DSP 602 may also be included in a system that has multiple DSPs connected in a depth cascade configuration that expands the number of memory locations in CAM array 604 and priority memory 608 to accommodate more policy statements and their associated priority numbers.
- FIG. 39 shows a CAM system 3900 that includes three DSPs 3902 , 3904 , and 3906 interconnected in a depth cascade configuration.
- Policy statements may be arbitrarily stored in DSPs 3902 - 3906 without initially presorting or prearranging the statements.
- a classification or filtering operation such as that described in FIG. 7
- the DSP that stores the highest priority matching statement for system 3900 may be determined as will be described in more detail below.
- the DSP that includes the highest priority matching statement for system 3900 may be determined regardless of the number of CAM locations filled or occupied in each CAM array of each DSP (i.e., regardless of whether one or more CAM arrays of DSPs 3902 - 3906 is full, empty, or partially full).
- any number of DSPs may be depth cascaded as shown in FIG. 39 .
- the total memory size of system 3900 is larger than the individual memory sizes of each of the CAM arrays in each DSP 3902 - 3906 .
- each CAM array is a 16 k ⁇ 144 bit CAM array
- system 3900 may operate as a DSP that includes a 48 k ⁇ 144 bit CAM array. Any other size of CAM arrays may be used.
- CAM arrays of different widths may be depth cascaded together.
- system 3900 may be formed from stand-alone DSPs, or may be formed from integrated circuits on a common substrate.
- Each DSP receives in parallel a clock signal CLK, policy statements and priority numbers on data bus DBUS, and instructions on instruction bus IBUS.
- the policy statements and priority numbers may be provided on different buses.
- Other input signals may also be simultaneously provided to each of DSPs 3902 - 3906 including word enable signals, reset signals, chip enable signals, and the like.
- DSPs 3902 - 3906 also output their external address signal to a common HPM bus. When performing the classification function of FIG. 7 , the DSP that has the highest priority matching policy statement will take control of the HPM bus.
- Each DSP 3902 - 3906 may include any of the embodiments of DSP 602 described above.
- DSPs 3902 - 3906 may each include a cascade down input /CDI, a cascade down output /CDO, a cascade up input /CUI, a cascade up output /CUO, priority number down inputs PNDI, priority number down outputs PNDO, priority number up inputs PNUI, and priority number up outputs PNUO.
- the PNDO and PNUO outputs provide the most significant priority number PNUM generated by the priority logic in each priority index table in each of DSP 3902 - 3906 .
- the cascade up and down outputs provide signals that indicate when PNUM is valid on the PNDO and PNUO outputs.
- the cascade inputs and outputs may be omitted and CLK or match flag up and down signals may be used to validate the PNUM outputs.
- match flag up and down signals generated by flag logic coupled to CAM array 604 may be used to validate the PNUM outputs. Additional outputs or inputs such as full flag up and down pins may also be included.
- Each CAM DSP generally has its /CDO output coupled to the /CDI input of the next DSP, its /CUO output coupled to the /CUI of the previous DSP, its /CDI input coupled to the /CDO of the previous DSP, its /CUI input coupled to the /CUO of the next DSP, its PNDO outputs coupled to the PNDI inputs of the next DSP, its PNDI inputs coupled to the PNDO outputs of the previous DSP, its PNUI inputs coupled to the PNUO outputs of the next DSP, and its PNUO outputs coupled to the PNUI inputs of the previous DSP.
- DSP 3902 may be designated as the DSP that has the lowest logical system addresses by connecting its /CDI input to ground, and its PNDI inputs to VDD.
- the PNDI inputs may be connected to ground and/or the /CDI inputs of DSP 3902 may be connected to VDD.
- DSP 3902 has the lowest logical or numerical addresses of, for example, zero to N ⁇ 1.
- DSP 3904 has logical addresses N to M ⁇ 1, and DSP 3906 has logical addresses M to W ⁇ 1, where system 3900 has a total of W ternary CAM words available for storing policy statements.
- DSP 3906 may be designated as the last DSP in the cascade by connecting its PNUI inputs to VDD, and its /CUI input to ground.
- the PNUI inputs of DSP 3906 may be connected to ground, and/or the /CUI input may be connect to VDD.
- system 3900 for implementing the classification or filtering function of FIG. 7 is as follows. Initially, one or more of DSPs 3902 - 3906 is loaded with policy statements and corresponding priority numbers. In response to an instruction provided on the IBUS and the policy search key provided on the DBUS, each DSP 3902 - 3906 compares the policy search key with the policy statements stored in its CAM array 604 (step 702 ). The priority logic in each priority index table identifies its most significant priority number PNUM associated with one of its matching policy statements. Each DSP also determines the address of its PNUM in its priority memory 608 (steps 706 and 708 ).
- Each DSP then compares the priority numbers provided on its PNDI and PNUI pins with its own PNUM to determine whether it has a more significant priority number. If a DSP has a more significant or equal priority number than that provided on its PNUI pins, then the DSP outputs its own PNUM to its PNUO pins. If, however, a DSP has a lower priority PNUM than those provided on the PNUI pins, then the DSP outputs the priority number from its PNUI pins to its PNUO pins. Similarly, if a DSP has a more significant priority number than that provided on its PNDI pins, then the DSP outputs its own PNUM to its PNDO pins. If, however, a DSP has an equal or lower priority PNUM than those provided on the PNDI pins, then the DSP outputs the priority number from its PNDI pins to its PNDO pins.
- each DSP By simultaneously providing and comparing priority numbers both down and up through system 3900 , each DSP will ultimately determine whether it has the most significant priority number in the entire system.
- a device When a device has the most significant priority number for the system, it may take control of the HPM bus and output address information to access a location in route memory 614 .
- route memory 614 is as deep as the number of CAM memory location in system 3900 .
- Each of DSPs 3902 - 3906 can also update their own policy statement tables and priority memories as needed in response to an insertion or deletion instruction provided to system 3900 .
- a new policy statement and priority number may be loaded into the DSP that has the next free address of system 3900 as determined by full flag signals (not shown).
- FIG. 40 shows DSP 4000 that is one embodiment of DSPs 3902 - 3906 .
- DSP 4000 is any of the embodiments of DSP 602 that further includes cascade logic 4004 , output buffer 4002 , instruction decoder 4006 , read and write (R/W) circuits 4008 and 4010 , and flag logic 4016 .
- DSP 4000 may also include registers to store the policy statements and priority numbers prior to use by CAM array 604 and priority index table 606 .
- Cascade logic 4004 is coupled to the /CDI, /CUI, PNDI, and PNUI inputs, and to the /CDO, /CUO, PNDO, and PNUO outputs.
- cascade inputs and outputs may be omitted and thus not coupled to cascade logic 4004 .
- Cascade logic 4004 may also receive and output match flag and full flag cascade signals.
- Cascade logic 4004 receives a search signal SCH on line 4012 from instruction decoder 4006 indicating that the classification or filtering operation will be performed by DSP 4000 .
- Cascade logic 4004 may also receive a match flag signal /MF from flag logic 4016 indicating whether CAM array 604 has an address that matches a policy search key.
- Cascade logic 4004 compares the priority number on its PNDI inputs with its own most significant priority number PNUM from priority index table 606 , and outputs the more significant number from among these two priority numbers to its PNDO outputs.
- cascade logic 4004 compares the priority number on its PNUI inputs with its own PNUM, and outputs the more significant number from among these two priority numbers to its PNUO outputs. If DSP 4000 has the most significant priority number for the entire system 3900 , then it asserts its output buffer control signal OBCNTL on signal line 4014 to enable output buffer 4002 to take control of the HPM bus.
- FIG. 41 is a block diagram of cascade logic 4100 that is one embodiment of cascade logic 4004 of FIG. 40 .
- Cascade logic 4100 includes cascade down logic 4104 and cascade up logic 4102 .
- Cascade down logic 4104 generates signals for the /CDO and PNDO outputs in response to /CDI, PNDI inputs, and SCH.
- Cascade up logic 4102 generates signals on the /CUO and PNUO outputs in response to /CUI, PNUI inputs, and SCH.
- Other embodiments of cascade logic 4004 may be used. For example, match flag signals may be used.
- Cascade down logic 4104 includes signal generator 4106 and compare circuit 4110 .
- Signal generator 4106 receives CLK and SCH from instruction decoder 4006 .
- SCH indicates that the classification of filtering function is to be performed by DSP 4000 .
- Signal generator 4106 asserts /CDO to a low logic state after /CDI is asserted to a low logic state and when the priority number on the PNDO outputs is valid.
- Signal generator 4106 may be any logic circuit that performs this function.
- SCH may be omitted and signal generator 4106 may generate /CDO in response to CLK only.
- Compare circuit 4110 compares the internal most significant priority number PNUM with the priority number from the PNDI inputs.
- compare circuit 4110 If the priority number on the PNDI inputs is equal to or more significant than PNUM, then compare circuit 4110 outputs the priority number from its PNDI inputs to its PNDO outputs and asserts signal line 4116 to a low logic state. If, however, PNUM is more significant than the priority number on the PNDI inputs, then compare circuit 4110 outputs PNUM to its PNDO outputs and asserts line 4116 to a high logic state.
- Cascade up logic 4102 includes signal generator 4108 and compare circuit 4112 .
- Signal generator 4108 receives CLK and SCH from instruction decoder 4006 .
- Signal generator 4108 asserts /CUO to a low logic state after /CUI is asserted to a low logic state and when the priority number on the PNUO outputs is valid.
- Signal generator 4108 may be any logic circuit that performs this function.
- SCH may be omitted and signal generator 4108 may generate /CUO in response to CLK only.
- Compare circuit 4112 compares PNUM with the priority number from the PNUI inputs.
- compare circuit 4112 If the priority number on its PNUI inputs is more significant than PNUM, then compare circuit 4112 outputs the priority number from its PNUI inputs to its PNUO outputs and asserts signal line 4118 to a low logic state. If, however, PNUM is equal to or more significant than the priority number on its PNUI inputs, then compare circuit 4112 outputs PNUM to the PNUO outputs and asserts line 4118 to a high logic state. When signal lines 4116 and 4118 are both at high logic states, AND gate 4114 asserts OBCNTL on line 4014 to a high state to enable output buffer 4002 to take control of the HPM bus.
- signal generators 4106 and 4108 may also be omitted.
- multiple /CDI-/CDO and /CUI-/CUO pins are included, multiple pairs of signal generators may also be included, and/or match flag signals may be included.
- DSP 602 can also be used to process Internet Protocol (IP) packets that use the Classless Inter Domain Routing (CIDR) scheme.
- IP Internet Protocol
- CIDR Classless Inter Domain Routing
- an IP address has a generalized network prefix of a particular number bits of 32-bit IPv4 or a 128-bit IPv6 address.
- the network prefix or mask indicates the number of left-most contiguous bits in the IP address that are used to filter an IP address in a routing table. That is, the network prefix indicates the number of higher-order or left-most contiguous bits in the IP address that participate in an address comparison with the routing table.
- IP addresses are generally presorted or prearranged prior to entry into a CAM device such that the IP address with the longest network prefix is located in the lowest logical address of the CAM array, and the IP address with the shortest network prefix is located in the highest logical address of the CAM array.
- a search on the CAM array for a particular IP address will identify the IP address that has the longest corresponding prefix, that is, will identify the best match.
- a considerable amount of time is generally required to prearrange all of the CIDR address entries prior to loading the entries into a CAM device. Additionally, a considerable amount of time and overhead is also generally required to maintain the order of the routing table when entries are deleted or overwritten, or when new entries are to be added.
- Other architectures have been proposed that increase the size of the CAM array by adding additional logic in the CAM array itself and another match coupled to the rows of mask cells.
- DSP 602 of FIG. 6 can be used to process IP addresses without adding additional logic or signal lines to ternary CAM array 604 .
- IP addresses can be loaded into CAM cell rows 605
- the corresponding decoded prefix data can be loaded into mask rows 607 .
- Decoding logic can be provided in DSP 602 to decode the prefix number.
- the prefix data is encoded into a binary number and stored in corresponding locations in priority memory 608 .
- Encoding logic can be provided in DSP 602 to encode the prefix number into a binary (or other code) number.
- Priority logic 610 When a search is performed for the IP address with the longest prefix, all matching locations in CAM array 604 will assert their corresponding match lines ML 0 -ML N-1 .
- Priority logic 610 then compares, with each other, the encoded prefix numbers associated with the matching IP address. Priority logic 610 identifies the most significant encoded prefix number (i.e., the highest prefix number), and identifies its location in priority memory 608 to IAD 0 -IAD N-1 . The encoded most significant prefix number may also be output from DSP 600 . Encoder 612 then encodes the identified location into an address for output to HPM bus 616 . The corresponding route information can then be accessed in route memory 614 . As in the previous embodiments described above, route memory 614 may also be included within DSP 602 .
- encoder 612 may be a conventional priority encoder that determines which address to output based on a predetermined priority (i.e., based on logical address locations).
- DSP 602 can process the CIDR based IP addresses without preloading the IP addresses in the CAM array in a predetermined order. Additionally, new IP address may be added at the next free address or any other designated address in CAM array 604 without reordering or reloading the CAM array. This can reduce the hardware and/or software needed for table management of the CAM array, and can increase the performance of a router incorporating the DSP.
- ternary CAM array 604 has IP address 168.0.0.0/8 stored at location 604 0 , 168.69.0.0/16 stored at location 604 1 , and 168.69.62.0/24 stored at location 604 2 .
- each IP address is stored in array 604 as four eight-bit binary numbers. Also for this embodiment, when the decoded prefix data is a logic zero it does not mask the corresponding bits of the IP address.
- Priority memory 608 stores the prefixes 8, 16, and 24 at locations 608 0 , 608 1 , and 608 2 , and the corresponding routing information RI 0 , RI 1 , and RI 2 are stored at locations 614 0 , 614 1 , and 614 2 of route memory 614 .
- a search key of 168.69.43.100 is provided to CAM array 604 , and the IP address with the best match (i.e., the longest prefix data) is determined as follows.
- the search key is compare with the IP addresses, 168.69.0.0/16 and 168.0.0.0/8 are both identified as matches and ML 0 and ML 1 asserted. Between these two IP addresses, 168.69.0.0/16 is the best match as it has a longer prefix.
- Priority logic 610 compares the prefixes 16 and 8 stored at locations 608 0 and 608 1 and determines that 16 is greater than 8. The priority logic outputs 16 as the longest matching prefix to PNUM, and also identifies location 608 1 by asserting IAD 1 .
- Encoder 612 then encodes IAD 0 -IAD N-1 and generates an address of 1 on HPM bus 616 to access route information RI 1 in route memory 614 .
- DSP 602 can be used to implement priority logic 610 to process CIDR based IP addresses and their prefix numbers. Additionally, multiple DSPs can be depth cascaded as described in FIGS. 39-41 to implement a system that provides a sufficient number of CAM array locations for storing IP addresses.
- a digital signal processor comprising:
- a policy statement table for storing a plurality of policy statements
- a priority index table for storing a plurality of priority numbers, each priority number associated with a corresponding policy statement and indicating the priority of the corresponding policy statement relative to the other policy statements.
- the digital signal processor of clause 5 further comprising a memory array coupled to the plurality of signal lines and for storing routing information for the policy statements, the indication to select routing information from the memory array for one of the policy statements.
- priority logic further comprises means for determining whether to compare at least two priority numbers already stored in the priority index table, or to compare the priority numbers stored in the priority index table with a priority number not stored in the priority index table.
- a digital signal processor comprising:
- the digital signal processor of clause 18, further comprising priority logic coupled to the first may of storage elements, the priority logic to provide to a plurality of signal lines an indication of a location of a particular number in the first array of storage elements.
- each row of the second may coupled to one of the plurality of signal lines to receive the indication, the indication to select one of the rows of the second array.
- each row of storage elements in the first array is configured to form a counter.
- the digital signal processor of clause 30, further comprising a second plurality of compare circuits each coupled to one of the counters, and each having a first plurality of inputs to receive a first number stored in the respective counter, a second plurality of inputs each coupled to one of the plurality of priority signal lines to receive a second number provided thereon, and an output coupled to the counter, each of the second compare circuits providing a control signal on its output to indicate whether to update the first number stored in the respective counter.
- each of the second plurality of compare circuits determines when the first number is greater than or equal to the second number.
- each of the second plurality of compare circuits determines when the first number is less than or equal to the second number.
- each row of storage elements in the first array is configured to form a counter.
- each of the plurality of compare circuits determines when the first number is greater than or equal to the second number.
- each of the second plurality of compare circuits determines when the first number is less than or equal to the second number.
- a digital signal processor comprising:
- the digital signal processor of clause 41 further comprising means for determining that at least one of the numbers stored in the first storage means is less than or equal to an external number not stored in the first storage means.
- the CAM array comprises a ternary CAM array having data storage cells for storing the plurality of data words, and further having mask storage cells for storing mask data on a bit-for-bit basis for the data words.
- a method of operating a digital signal processor comprising:
- a method of adding a new policy statement to a plurality of policy statements stored in a first memory comprising:
- a method of adding a new policy statement to a plurality of policy statements stored in a first memory comprising:
- altering the new priority number comprises decrementing the new priority number.
- altering the new priority number comprises incrementing the new priority number.
- a method of deleting a policy statement from a plurality of policy statements stored in a first memory comprising:
- a system comprising:
Landscapes
- Engineering & Computer Science (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
Description
- This application is a Divisional Application of U.S. patent application Ser. No. 09/406,170 filed Sep. 23, 1999 and entitled, “Method and Apparatus for Performing Packet Classification for Policy-Based Packet Routing,” which is herein incorporated by reference in its entirety.
- The present invention relates generally to performing packet classification for policy-based packet routing.
- Routers are devices that direct traffic in a network. Each router in the network has a route table that typically includes routing information to enable incoming packets to be forwarded or routed to their destination in the network. Some routers forward packets based only on the destination address indicated in the packet. Other, more complex, routers forward or route packets based on policies defined, for example, by a network administrator. The latter routing schemes are commonly referred to as policy-based routing.
- Policy-based routing can enable packets to be forwarded or routed in a network based on any number of criteria, including the source of the packet, the destination of the packet, the cost of forwarding the packet through different routes or paths in the network, or the bandwidth available on different routes in the network. Policy-based routing can also be used to provide a certain Quality of Service (QOS) or Type of Service (TOS) to differentiated traffic in the network. For example, one or more of the various fields (e.g., the TOS bits) in the header of an Internet Protocol (IP) packet can be used by policy-based routers to forward IP packets in a network.
- Each policy-based router implements a policy through the use of route maps that define how to forward the packet in the network. Each route map statement or policy statement contains one or more match clauses and a set clause. The match clauses are a series of conditions that are used to determine if an incoming packet satisfies a particular policy. If all of the match clauses of a policy statement are satisfied, the set clause specifies how the router should forward the packet in the network. If one of the match clauses of a particular policy statement is not satisfied, then the policy-based router investigates subsequent policy statements.
-
FIG. 1 shows exemplary processedpolicy information 100 of an incoming packet to a policy-based router.Policy information 100 includesseveral policy fields 102 including a destination address (DA) for the packet, a source address (SA) of the packet, protocol type (PTCL) such as those defined by for an IP packet header, TOS, and COST.Policy information 100 may be received by a policy-based router that implements a policy such aspolicy 200 shown inFIG. 2 .Policy 200 includes threeseparate policy statements 201 through 203. Ifpolicy information 100 satisfies the match clause (i.e., the “if” clause) of one of the policy statements, the set clause (i.e., the “then” clause) of that policy statement determines routing information for the packet in the network. For example, if the destination address of the incoming packet is DA1, the source address is SA1, and the TOS field of the packet is TOS1, then routing information RI2 should be selected. - A policy-based router can use a content addressable memory (CAM)-based system to implement a filtering or classification function to determine whether an incoming packet matches a policy statement.
FIG. 3 shows one example of asystem 300 that implements policy-based routing using aternary CAM 302. The policy statements or policy words 201-203 are stored in separate rows internary CAM array 304. A ternary CAM array is one that is able to mask entries in a CAM array on a bit-by-bit basis.Ternary CAM array 304 has rows ofCAM cells 305 for storing policy field information, and corresponding rows ofmask cells 310 for storing mask data. Routing information RI0-RI2 is typically stored in anexternal memory 308 at addresses corresponding to those at which the respective policy words 201-203 are stored internary CAM array 304. Each policy field that corresponds to a match clause for a given policy statement is unmasked by having its corresponding mask bits set, for example, to a logic zero. Conversely, each policy field that does not have a match clause for a given policy statement is masked by having its corresponding mask bits set, for example, to a logic one. - When an incoming packet is received by a policy-based router, it is processed to determine the policy field information. The processed policy field information is provided to
system 300 aspolicy search key 307. For each policy statement inCAM array 304 that matches the policy search key, the corresponding match line ML0-MLN will be asserted and provided topriority encoder 306. In response to the match lines,priority encoder 306 outputs the address of the highest priority matching entry inCAM array 304 to HPMbus 312. If there is more than one matching policy statement inCAM array 304,priority encoder 306 determines that the highest priority matching policy statement is the one stored at the lowest logical address ofCAM array 304. For example, as shown inFIG. 3 , ifCAM array 304 is loaded withpolicy statement 203 at address zero (i.e., the lowest logical address),statement 202 at address one, andstatement 201 ataddress 2, and a policy search key of DA1, SA1, PTCL1, TOS1, COST1 is provided toCAM array 304, then each of policy statements 201-203 is identified as a match on match lines ML0-ML2, respectively.Priority encoder 306 outputs address zero on the HPM bus to select route information RI2 from address zero inmemory 308. - Because
priority encoder 306 determines the highest priority matching location based on predetermined logical address assignments, policy statements 201-203 are preordered or prioritized such that higher priority policy statements are stored in lower logical addresses ofCAM array 304 than lower priority policy statements. A policy statement has a higher priority than another policy statement when the route information for the first policy statement is to be selected over the second policy statement even though both policy statements may match the policy search key (e.g., with masking). The prioritizing of the policy statements is typically performed by table management hardware and/or software, which adds overhead to the router. - If the table is not prioritized correctly, then an incorrect route may be selected for the incoming packet.
FIG. 4 shows an example in whichpolicy statement 201 is incorrectly prioritized such that it, rather thanpolicy statement 203, is loaded into the lowest logical address. In this example, whenpolicy search key 307 is provided toCAM array 304,priority encoder 306 still provides address zero onHPM bus 312. This causes, however, route information RI0 to be selected inmemory 308 rather than the desired route information RI2 associated with higherpriority policy statement 201. It would be desirable to load policy statements into a CAM array without having to preorder the statements according to their respective priorities. - When a policy is changed by adding a new policy statement that has a higher (or equal) priority than at least one of the policy statements already stored in
CAM array 304, the table management hardware and software needs to reprioritize or reorder all or part ofCAM array 304. This is typically accomplished by reloading the CAM array with a new prioritized group of policy statements. This can add significant overhead to the router (e.g., delay and additional hardware and software) to change even just one policy statement in a given policy. If the table is not reloaded, CAM 302 may cause an incorrect route to be selected frommemory 308. For example,FIG. 5 shows that anew policy statement 204 is loaded into the next free address (i.e., address three) ofCAM array 304 without reprioritizing and reloadingCAM array 304. Route information RI3, corresponding topolicy statement 204, is loaded into a corresponding address inmemory 308. Ifpolicy statement 204 has a priority greater than that of 201, 202, or 203, then a search withpolicy search key 307 will not correctly identifypolicy statement 204 during a compare operation onCAM array 304.Policy statement 203, rather, will be identified as the higher priority statement, and route information RI2 will be selected. It would be desirable to be able to add a new policy statement to a policy without reloading the CAM array storing the policy statements. - A method and apparatus for performing a classification or filtering function for policy-based packet routing in a digital signal processor (DSP) is disclosed. For one embodiment, the digital signal processor includes a policy statement table for storing policy statements. The policy statement table may be stored in a ternary CAM array that stores the policy statements and associated mask data. Each policy statement has associated with it a priority number that indicates the priority of the policy statement relative to other policy statements. The priority numbers are separately stored in a priority index table. The priority index table includes priority logic that determines the most significant priority number from among the policy statements that match an incoming packet during a classification or filter operation. The priority logic also identifies the location in the priority index table of the most significant priority number. The most significant priority number may be the priority number with the lowest or highest numerical value. The identified location in the priority index table can be used to access associated route information or other information stored in a route memory array located external to, or on the DSP. When the route memory array is external to the DSP, the DSP may include an encoder to encode the identified location in the priority index table into an address for the route memory.
- The DSP configuration obviates preloading the policy statements in the policy statement table in a predetermined order. Instead, the priority logic determines the most significant priority number from among matching policy statements regardless of the order in which the policy statements are stored in the table. This can reduce the hardware and/or software needed for table management of the table, and can increase the performance of a router incorporating the DSP.
- In addition, new policy statements can be added at any location in the policy statement table, and associated priority numbers loaded into corresponding locations in the priority index table. If a new policy statement has a priority that is greater than or equal to a priority of a policy statement already stored in the policy statement table, the priority number of the previously stored policy statement may be updated to accommodate the new policy statement. Similarly, when a policy statement is removed (i.e., invalidated or overwritten) from the policy statement table, the priority numbers of the previously stored lower priority policy statements may be updated. The updating functions can be performed by the priority logic in the priority index table, or by inequality circuits in the priority index table. The updating functions can be performed without the need to physically reorder the policy statements in the policy statement table, or to physically reorder the priority numbers in the priority index table. This also can reduce the hardware and/or software needed for table management of the policy statement table, and can increase the performance of a router incorporating the DSP.
- The DSP can also be used in other non-networking applications. For example, the DSP can be used to process if-then-else functions in other applications.
- Other objects, features, and advantages of the present invention will be apparent from the accompanying drawings and from the detailed description that follows below.
- The features and advantages of the present invention are illustrated by way of example and are by no means intended to limit the scope of the present invention to the particular embodiments shown, and in which:
-
FIG. 1 is an example of policy information; -
FIG. 2 is an example of a policy having policy statements; -
FIG. 3 is an example of storing a policy in a conventional ternary CAM; -
FIG. 4 is an example of storing a particular policy in a conventional ternary CAM and comparing a search key with the policy; -
FIG. 5 is another example of storing a particular policy in a conventional ternary CAM and comparing a search key with the policy; -
FIG. 6 is one embodiment of a classification system for a policy-based router; -
FIG. 7 is one embodiment of performing the classification operation for the system ofFIG. 6 ; -
FIG. 8 is one example of performing the classification operation on a particular policy stored in the CAM array ofFIG. 6 ; -
FIG. 9 is one embodiment of reading or writing a policy statement based on a priority number stored in the priority memory ofFIG. 6 ; -
FIG. 10 is one embodiment of the priority index table ofFIG. 6 ; -
FIG. 11 is another embodiment of the priority index table ofFIG. 6 ; -
FIG. 12A is one embodiment of two rows of the priority index table ofFIG. 11 ; -
FIG. 12B is another embodiment of two rows of the priority index table ofFIG. 11 ; -
FIG. 13 is one example of determining the most significant priority number stored in the rows of the priority index table; -
FIG. 14 is one embodiment of a compare circuit and an isolation circuit for the priority logic element ofFIG. 11 ; -
FIG. 15 is another embodiment of a compare circuit and an isolation circuit for the priority logic element ofFIG. 11 ; -
FIG. 16 is another example of determining the most significant priority number stored in the rows of the priority index table; -
FIG. 17 is another embodiment of the priority index table ofFIG. 6 including inequality circuits; -
FIG. 18 is one example of a policy stored in the digital signal processor ofFIG. 17 ; -
FIG. 19 is one example of loading a new policy statement into the digital signal processor ofFIG. 18 ; -
FIG. 20 is another example of a policy stored in the digital signal processor ofFIG. 17 ; -
FIG. 21 is one example of loading a new policy statement into the digital signal processor ofFIG. 20 ; -
FIG. 22 is one embodiment of the rows of the priority index table ofFIG. 17 , in which the priority memory locations are configured as counters; -
FIG. 23 is one embodiment of an inequality circuit and counter ofFIG. 22 ; -
FIG. 24 is one example of determining whether a new number is greater than or equal to priority numbers stored in the priority memory ofFIG. 17 ; -
FIG. 25A is one embodiment of a compare circuit ofFIG. 24 ; -
FIG. 25B is another embodiment of a compare circuit ofFIG. 24 ; -
FIG. 26 is one example of determining whether a new number is less than or equal to priority numbers stored in the priority memory ofFIG. 17 ; -
FIG. 27A is one embodiment of a compare circuit ofFIG. 26 ; -
FIG. 27B is another embodiment of a compare circuit ofFIG. 26 ; -
FIG. 28 is one embodiment of a process of deleting or invalidating policy statements and their associated priority numbers from the digital signal processor ofFIG. 17 ; -
FIG. 29 is one embodiment of priority index table ofFIG. 17 ; -
FIG. 30 is another embodiment of priority index table ofFIG. 17 ; -
FIG. 31 is one embodiment of an inequality circuit element, a priority logic element and a memory element ofFIG. 17 ; -
FIG. 32 is another embodiment of an inequality circuit element, a priority logic element and a memory element ofFIG. 17 ; -
FIG. 33 is another embodiment of the priority index table ofFIG. 6 including a decrement circuit and storing a policy; -
FIG. 34 is one example of loading a new policy statement into the digital signal processor ofFIG. 33 ; -
FIG. 35 is one embodiment of a row of the priority index table inFIG. 33 ; -
FIG. 36 is one example, of determining whether a new number is greater than or equal to priority numbers stored in the priority memory ofFIG. 33 ; -
FIG. 37 is one embodiment of compare circuit and an isolation circuit for a priority logic element ofFIG. 35 ; -
FIG. 38 is one example, of determining whether a new number is less than or equal to priority numbers stored in the priority memory ofFIG. 33 ; -
FIG. 39 is a block diagram of one embodiment of depth cascading digital signal processors; -
FIG. 40 is a block diagram of one embodiment of a digital signal processor including cascade logic; -
FIG. 41 is a block diagram of one embodiment of the cascade logic ofFIG. 39 ; and -
FIG. 42 is one example of processing Internet protocol addresses based on a classless inter domain routing scheme in the digital signal processor ofFIG. 6 . - In the following description, for purposes of explanation, specific nomenclature is set forth to provide a thorough understanding of the present invention. However, it will be apparent to one skilled in the art that these specific details may not be required to practice the present invention. In other instances, well known circuits and devices are shown in block diagram form to avoid obscuring the present invention unnecessarily. Additionally, the interconnection between circuit elements or blocks may be shown as buses or as single signal lines. Each of the buses may alternatively be single signal lines, and each of the single signal lines may alternatively be buses. Additionally, the prefix symbol “/” or the suffix “B” attached to signal names indicates that the signal is an active low signal. Each of the active low signals may be changed to active high signals as generally known in the art.
- A method and apparatus for performing a classification or filtering function for policy-based packet routing in a digital signal processor (DSP) is disclosed. For one embodiment, the digital signal processor includes a policy statement table for storing policy statements. The policy statement table may be stored in a ternary CAM array that stores the policy statements and associated mask data. Each policy statement has associated with it a priority number that indicates the priority of the policy statement relative to other policy statements. The priority numbers are separately stored in a priority index table. The priority index table includes priority logic that determines the most significant priority number from among the policy statements that match an incoming packet during a classification or filter operation. The priority logic also identifies the location in the priority index table of the most significant priority number. The most significant priority number may be the priority number with the lowest or highest numerical value. The identified location in the priority index table can be used to access associated route information or other information stored in a route memory array located external to, or on the DSP. When the route memory array is external to the DSP, the DSP may include an encoder to encode the identified location in the priority index table into an address for the route memory.
- The DSP configuration obviates preloading the policy statements in the policy statement table in a predetermined order. Instead, the priority logic determines the most significant priority number from among matching policy statements regardless of the order in which the policy statements are stored in the table. This can reduce the hardware and/or software needed for table management of the table, and can increase the performance of a router incorporating the DSP.
- In addition, new policy statements can be added at any location in the policy statement table, and associated priority numbers loaded into corresponding locations in the priority index table. If a new policy statement has a priority that is greater than or equal to a priority of a policy statement already stored in the policy statement table, the priority number of the previously stored policy statement may be updated to accommodate the new policy statement. Similarly, when a policy statement is removed (i.e., invalidated or overwritten) from the policy statement table, the priority numbers of the previously stored lower priority policy statements may be updated. The updating functions can be performed by the priority logic in the priority index table, or by inequality circuits in the priority index table. The updating functions can be performed without the need to physically reorder the policy statements in the policy statement table, or to physically reorder the priority numbers in the priority index table. This also can reduce the hardware and/or software needed for table management of the policy statement table, and can increase the performance of a router incorporating the DSP.
- The DSP can also be used in other non-networking applications. For example, the DSP can be used to process if-then-else functions in other applications.
- Classifying or Filtering Policy Statements
-
FIG. 6 shows one embodiment of a classification orfiltering system 600 for a policy-based router.System 600 includesDSP 602 androute memory 614.DSP 602 includes policy statement table 604, priority index table 606 andencoder 612, and may be an integrated circuit component formed on a single semiconductor substrate. - For this embodiment, policy statement table 604 is stored in a ternary CAM array that stores policy statements or policy words.
Ternary CAM array 604 has rows ofCAM cells 605 for storing policy field information PF1-PFX, where X is any number. Each policy field PF1-PFX can include any policy information including DA, SA, PTCL, TOS, and COST, or any other type of policy field to assist in the classification or filtering of the policy statement to provide a certain Quality of Service (QoS), Class of Service (CoS), and the like. Each policy field may include any number of bits. Additional information associated with each policy field may be stored in one or more additional binary or ternary CAM cells or other types of memory cells disposed in each row ofCAM 604.Ternary CAM 604 also has rows ofmask cells 607 for storing mask data M1-MX corresponding to each row of policy fields 605. Global masks (not shown) may be used to mask entire columns inCAM array 604 as generally known in the art. For alternative embodiments,CAM array 604 may be any other type of CAM including a binary CAM, or any other type of memory to store policy statements to be compared with processed policy information of an incoming packet. -
Ternary CAM array 604 can be any ternary CAM array that is capable of locally masking each entry on a bit-by-bit basis. Each policy field that corresponds to a match clause for a given policy statement will be unmasked by having its corresponding mask bits set, for example, to a logic zero (or, alternatively, a logic one). Conversely, each policy field that does not have a match clause for a given policy statement will be masked by having its corresponding mask bits set, for example, to a logic one (or, alternatively, a logic zero). As each policy field, and/or individual bits within a policy field, can be masked,DSP 602 also supports rule and route aggregation. That is,DSP 602 supports ranges of addresses or policy field information. - Each policy statement loaded into
ternary CAM array 604 has associated with it a priority number P0-PZ and route information RI0-RIN-1. The priority number indicates the priority of the policy statement relative to other policy statements in a given policy. The policy numbers may be assigned by a user ofDSP 602 including, for example, a network administrator or the router itself. The priority numbers P0-PZ are separately stored at locations 608 0-608 N-1, respectively, ofpriority memory 608 of priority index table 606. Route information RI0-RIN-1 for the particular policy statements are stored at locations 614 0-614 N-1, respectively, inroute memory 614. The route information may include, for example, forwarding or next hop information, authentication information, QOS, TOS, time to live information or other packet filtering and classification information for an incoming packet to therouter incorporating system 600. A policy statement, its priority number, and its route information are each stored at the corresponding addresses in each of their respective memory arrays. -
Priority memory 608 androute memory 614 may each be any type of memory array including volatile, non-volatile, random access memory (RAM), and/or read only access memory (ROM). For one embodiment,priority memory 608 comprises a CAM array. -
Priority memory 608 may be n bits wide to accommodate Z=2n priority numbers, where n is any number. The number of addressable locations N inpriority memory 608 may be greater than, less than, or equal to Z. For one example,priority memory 608 may be 20 bits wide to accommodate up to 220 or 1 Meg (i.e., 1,048,576) priority numbers. - The priority numbers may be assigned in ascending priority order such that zero is the highest priority number and 2n−1 is the lowest priority number. Alternatively, the priority numbers may be assigned in descending priority order such that 2n−1 is the highest priority number and zero is the lowest priority number. Each priority number may be assigned so as to identify the priority of each policy statement relative to other policy statements. For one embodiment, the priority numbers may be assigned consecutively. For example, the highest priority policy statement can be assigned the highest priority number (e.g., zero or 2n−1), the next lower priority policy statement can be assigned the next lower priority number (e.g., one or 2n−2), and so forth. For another embodiment, gaps may be left in the priority number assignments to allow for the addition of future priority numbers associated with new policy statements.
- Priority index table 606 also includes
priority logic 610 that compares the priority numbers with each other for all corresponding policy statements that match an incoming packet.Priority logic 610 identifies the most significant priority number PNUM inmemory 608 from among the compared priority numbers, and further identifies the location of PNUM inpriority memory 608. PNUM has the lowest numerical value when the priority numbers are assigned is ascending priority order, and PNUM has the highest numerical value when the priority numbers are assigned in descending priority order.Priority logic 610 may also output PNUM fromDSP 602. The identified location of PNUM in the priority memory is provided on internal address lines IAD0-IADN-1 to encoder 612. For one example, one of IAD0-IADN-1 is asserted to indicate the location inpriority memory 608 of PNUM from among the compared priority numbers. This location also corresponds to the location of the highest priority matching policy statement internary CAM array 604. - The address of the identified location of the highest priority matching policy statement in
ternary CAM array 604 is determined byencoder 612 and output toHPM bus 616. The encoded address can then be used to access the corresponding route information frommemory 614.Encoder 612 may be any encoding logic that takes the information on address lines IAD0-IADN-1 and generates an encoded address. For one embodiment,encoder 612 is a ROM. - For another embodiment,
route memory 614 may also be included withinDSP 602. For this embodiment,encoder 612 may be omitted androute memory 614 may be accessed directly by internal address lines IAD0-IADN-1. -
FIG. 7 summarizes the classification or filtering function 700 (i.e., search or compare operation) performed byDSP 602 for an incoming packet according to a policy stored internary CAM 604. An incoming packet received by a policy-basedrouter incorporating system 600 is initially processed to determine the policy field information. The policy field information is provided tosystem 600 aspolicy search key 609. Atstep 702, the policy fields ofpolicy search key 609 are compared with the policy statements stored internary CAM array 604. For each policy statement that matches the policy search key, the corresponding match line ML0-MLN-1 is asserted. If no match is found, then the process stops atstep 704. - At step 706,
priority logic 610 determines PNUM and identifies its location inpriority memory 608. The identified location is provided on internal address lines IAD0-IADN-1 to encoder 612. Atstep 708,encoder 612 determines the address of the identified location in priority index table 606. This encoded address is also the logical address of the highest priority matching policy statement internary CAM array 604.Encoder 612 outputs the encoded address toHPM bus 616. The encoded address can then be used atstep 710 to access the corresponding route information inmemory 614.Steps 708 and/or 710 may be omitted whenencoder 612 is removed fromDSP 602, andpriority logic 610 may directly access the route information inmemory 614. - For another embodiment, IAD0-IADN-1 are provided to
CAM array 604 to access the highest priority matching policy statement, which may then be read fromDSP 602. Alternatively,HPM bus 616 may be provided to CAM array 604 (e.g., through a decoder) to access the highest priority matching policy statement. -
FIG. 8 shows one example in whichpolicy statements FIG. 2 are stored atlocations ternary CAM array 604. Thecorresponding priority numbers locations memory 608. Additionally, the corresponding route information RI0, RI1, and RI2, are stored atlocations route memory 614. The policy statements and priority numbers are written into their respective memories using conventional write circuits, counters, and/or address decoders, etc. (not shown). - For this embodiment, the priority numbers have been assigned in ascending priority order such that
policy statement 203 is identified as the highest priority policy statement by being assignedpriority number 0, the lowest numerical value;policy statement 201 is identified as the having the lowest priority policy statement by being assignedpriority number 2, the highest numerical value; and,policy statement 202 is identified as having a priority greater than that ofpolicy statement 201, but less than that ofpolicy statement 203, by being assignedpriority number 1. - For an alternative embodiment, the priority numbers may be assigned in descending priority order such that
policy statement 201 is identified as the highest priority policy statement by being assignedpriority number 2, the highest numerical value;policy statement 203 is identified as having the lowest priority policy statement by being assignedpriority number 0, the lowest numerical value; and,policy statement 202 is identified as having a priority greater than that ofpolicy statement 201, but less than that ofpolicy statement 203, by being assignedpriority number 1. - The process of determining the route information for
policy search key 609 is illustrated with the aid ofFIG. 7 . Atstep 702, the policy fields ofpolicy search key 609 are compared with the policy statements stored internary CAM array 604. In this example, the policy search key has policy fields of DA=DA1, SA=SA1, PTCL=PTCL1, TOS=TOS1, and COST=COST1.CAM array 604 determines that each of the policy statements 201-203, as masked by their respective mask data, matchespolicy search key 609. In response, each of match lines ML0-ML2 is asserted. - At step 706,
priority logic 610 compares, with each other,priority numbers policy statements Priority logic 610 determines thatpriority number 0 is the most significant priority number, asserts IAD2, and de-asserts the other internal address lines.Encoder 612 encodes the internal address information, atstep 708, and generates an external address of 2 onHPM bus 616. The external address can be used to access route information RI2 stored at address two inroute memory 614. - In contrast to the conventional system shown in
FIG. 4 ,DSP 602 is able to identify the highest priority matching policy statement stored internary CAM array 604 regardless of where the policy statements are stored inCAM array 604. - The process illustrated in
FIG. 7 identifies the location inpriority memory 608 of the most significant priority number from among the compared priority numbers. Once this location is identified, the priority number stored at the identified location can be read out fromDSP 602 by a read circuit (not shown), or a new priority number can be written into that location by a write circuit (not shown). In the former case, the user of the DSP (e.g., a network administrator or the policy-based router itself) can determine what priorities have been assigned to policy statements already stored inCAM array 604. In the latter case, priority numbers can be updated by the user for already stored policy statements. This provides the user with flexibility in the control and management of the policy statements stored inDSP 602. - For added flexibility, the user can read a policy statement (e.g., one or more of the policy fields and/or one or more of the corresponding mask data) based on priority numbers already stored in the priority memory, or write a new policy statement for a priority number already stored in the priority memory. For these embodiments,
priority memory 608 may be a CAM. For an alternative embodiment, each priority number may be separately compared by a comparison circuit with an externally applied priority number. -
FIG. 9 shows one embodiment of aprocess 900 for reading or writing a policy statement based on a priority number already stored in the priority memory. Atstep 902, an external priority number is compared with the priority numbers stored inpriority memory 608. The valid priority numbers stored inmemory 608 may be indicated by one or more validity bits (not shown) stored inCAM memory 604 ormemory 608. If there is no match, or more than one match (i.e., a multiple match), then the process stops atstep 904. Alternatively, a priority encoder circuit can be coupled to priority index table 606 to select one of the multiple matches forstep 906. - If there is a single match, then an indication of the matching location is provided on lines IAD0-IADN-1 to encoder 612 at
step 906. Whenpriority memory 608 is implemented as a CAM, IAD0-IADN-1 may be the match lines associated with each priority number storage location. Atstep 908,encoder 612 determines the address of the matching location inpriority memory 608 and provides this address onHPM bus 616. This address is also the address of the corresponding policy statement inCAM array 604. Atstep 910, the encoded address can then be provided to CAM array 604 (e.g., through a decoder), and the policy statement stored at that address either read fromDSP 602, or overwritten. For an alternative embodiment, IAD0-IADN-1 may be provided directly toCAM array 604 to access the desired policy statement and the process may omitstep 908. Additionally, a new priority number can be written topriority memory 608 at the address determined instep 910. - With reference again to
FIGS. 6 and 7 , priority index table 606 stores priority numbers inpriority memory 608, andpriority logic 610 determines the most significant priority number among those associated with policy statements that matchpolicy search key 609.FIG. 10 shows DSP 1000 that includes priority index table 1001 that is one embodiment of priority index table 606. In this embodiment,CAM array 1002 is one embodiment ofpriority memory 608, andpriority logic 1004 is one embodiment ofpriority logic 610. -
DSP 1000 performs the filtering function ofFIG. 7 as follows. Atstep 702, a policy search key is compared with the policy fields stored inCAM array 604 and match lines ML0-MLN-1 are asserted for the matching locations. The asserted match lines access the associated priority numbers stored inCAM array 1002. For one embodiment, ML0-MLN-1 are coupled to corresponding word lines of the rows of CAM cells inCAM array 1002. In response to the asserted match lines,CAM array 1002 provides the selected priority numbers topriority logic 1004 overbus 1006 to determine which priority number is the most significant priority number. The priority numbers may be provided one at a time overbus 1006, or in parallel. -
Priority logic 1004 includes one or more compare circuits to compare the priority numbers and determine the most significant priority number. For one embodiment,priority logic 1004 includes a compare circuit and a register. The first priority number provided onbus 1006 is stored in the register and compared with the second priority number. Each subsequent priority number provided overbus 1006 is then compared and the result (either greater than or less than) is then stored in the register. This process continues until no more priority numbers are provided overbus 1006. For another embodiment, multiple compare circuits may be provided to compare multiple priority numbers at the same time to generate the most significant priority number. - Once the most significant priority number is determined, it is provided over
bus 1008 toCAM 1002 and searched against the stored priority numbers to determine where it is located inCAM array 1002. The indicated location is provided on internal address lines IAD0-IADN-1, which may correspond to the match lines ofCAM array 1002.Steps -
FIG. 11 shows DSP 1100 that includes priority index table 1101 that is another embodiment of priority index table 606. In this embodiment,priority memory 608 andpriority logic 610 are merged together on a bit-by-bit basis to form priority index table 1101. The priority memory includesmemory elements 1102 that each store one bit of a priority number for a given row. Each memory element may be any type of storage mechanism including volatile or non-volatile memory cells. The priority logic includespriority logic elements 1104. Eachpriority logic element 1104 is associated with, or corresponds to, one of thememory elements 1102 such that columns 1106 0-1106 n-1 of priority index table 1101 have a memory element/priority logic element pair for each of its rows. Eachpriority logic element 1104 effectively compares the priority number bit stored in its associatedmemory element 1102 with the priority number bits stored in every other memory element of its column to determine one of bits PNUM0-PNUMn-1 for the most significant priority number. Bits PNUM0-PNUMn-1 comprise the most significant priority number from among the policy statements that match a policy search key. -
FIG. 12A shows priority index table 1200 that is one embodiment of two rows of priority index table 1101. For this embodiment, eachpriority logic element 1104 includes a comparecircuit 1206 and anisolation circuit 1204. Each comparecircuit 1206 is connected in a wired-OR configuration with the other compare circuits in its respective column by one of priority signal lines 1208 0-1208 n-1. Each priority signal line may be pre-charged towards a power supply voltage (or any other predetermined voltage) by apre-charge circuit 1202. Each comparecircuit 1206 may be any digital or analog compare circuit that, when executing step 706 ofFIG. 7 , effectively compares the priority number bit stored in itsrespective storage element 1102 with the priority number bits stored in everyother storage element 1102 of the same column. Additionally, each compare circuit monitors the comparison result of the more significant priority number bits through the logical states ofmatch line segments 1210.Match line segments 1210 are coupled between match lines ML0-MLN-1 and internal address lines IAD0-IADN-1 byisolation circuits 1204. The isolation circuits isolate the comparison results generated for less significant priority bit locations from affecting the comparison results generated for more significant priority bit locations. The isolation circuits may also work together with the comparison circuits to control the state of the match line segments. - The operation of priority index table 1200 can be illustrated with an example shown in
FIG. 13 and with the aid ofFIG. 7 . In this example, priority index table 1200 comprises a 2×4 matrix of rows and columns. For other embodiments, any numbers of rows and columns can be used. Row zero stores priority number 0110 having the decimal equivalent of the number 6, and row one stores priority number 0101 having the decimal equivalent of thenumber 5. For this example, each of row zero and row one ofCAM array 604 have policy statements that match the policy search key such that match lines ML0 and ML1 are asserted (step 702). Also, for this example, the priority numbers are stored in ascending priority order such that 0101 is the more significant priority number between 0101 and 0110. - At step 706, compare circuits 1206 0,0-1206 3,1 determine that 0101 is the more significant priority number PNUM, and cause IAD1 to be asserted to indicate that 0101 is stored in row one of the priority index table. Compare circuits 1206 0,0-1206 3,1 determine that PNUM is 0101 as follows. The most significant bit PNUM3 is resolved first. When any
memory element 1102 stores a logic zero and the correspondingmatch line segment 1210 is asserted, the correspondingpriority signal line 1208 is discharged. Thus, each of comparecircuits discharge signal line 1208 3 such that PNUM3 is a logic zero. Additionally, comparecircuit 1206 3,1 compares the state ofpriority signal line 1208 3 with the priority number bit stored in 1102 3,1, and determines that both have the same logic state. This causes comparecircuit 1206 3,1 not to affect the logical state ofmatch line segment 1210 2,1 such thatmatch line segment 1210 2,1 has the same logic state as match line segment 1210 3,1 (ML1). Similarly, comparecircuit 1206 3,0 compares the state ofpriority signal line 1208 3 with the priority number bit stored in 1102 3,0 and determines that both have the same state. This causes comparecircuit 1206 3,0 not to affect the logical state ofmatch line segment 1210 2,0 such thatmatch line segment 1210 2,0 has the same logic state as match line segment 1210 3,0 (ML0). - The next most significant bit PNUM2 is then resolved.
Memory elements 1102 that store a logic one do not discharge their correspondingpriority signal lines 1208. Sincememory elements signal line 1208 2 remains pre-charged such that PNUM2 is a logic one. Additionally, comparecircuit 1206 2,1 compares the state ofpriority signal line 1208 2 with the priority number bit stored in 1102 2,1, and determines that both have the same logic state. This causes comparecircuit 1206 2,1 not to affect the logical state ofmatch line segment 1210 1,1 such thatmatch line segment 1210 1,1 has the same logic state asmatch line segment 1210 2,1. Similarly, comparecircuit 1206 2,0 compares the state ofpriority signal line 1208 2 with the priority number bit stored in 1102 2,0 and determines that both have the same logic state. This causes comparecircuit 1206 2,0 to not affect the logical state ofmatch line segment 1210 1,0 such thatmatch line segment 1210 1,0 has the same logic state asmatch line segment 1210 2,0. - PNUM1 is resolved next. Since
memory element 1102 1,1 stores a logic zero andmatch line segment 1210 1,1 is asserted, comparecircuit 1206 1,1 dischargespriority signal line 1208 1. This causes PNUM1 to be a logic zero. Additionally, comparecircuit 1206 1,1 compares the logic zero state ofpriority signal line 1208 1 with the logic zero stored in 1102 1,1 and allowsmatch line segment 1210 0,1 to have the same state asmatch line segment 1210 1,1. Comparecircuit 1206 1,0, however, compares the logic zero onpriority signal line 1208 1 with the logic one stored inmemory element 1102 1,0, and de-assertsmatch line segment 1210 0,0. When a match line segment is de-asserted, all subsequent compare circuits for that row will de-assert the remaining match line segments of the row such that the corresponding internal address line IAD will be de-asserted. When IAD is de-asserted for a particular row, this indicates that the most significant priority number is not stored in that row. Additionally, when the remaining match line segments are de-asserted for a row, the compare circuits for that row do not discharge the remaining priority signal lines regardless of the logic states stored in the corresponding memory elements of that row. For example, comparecircuit 1206 0,0 does not dischargepriority signal line 1208 0 even thoughmemory element 1102 0,0 stores a logic zero. Additionally,isolation circuits match line segment 1210 0,0 frommatch line segment match line segment 1210 0,0. - Lastly, the least significant bit PNUM0 is resolved. Compare
circuit 1206 0,1 alone determines PNUM0 since comparecircuit 1206 0,0 can not dischargepriority signal line 1208 0. Sincememory element 1102 0,1 stores a logic one andmatch line segment 1210 0,1 is asserted, comparecircuit 1206 0,1 leavespriority signal line 1208 0 pre-charged, and PNUM0 is a logic one. Additionally, comparecircuit 1206 0,1 allows IAD1 to have the same state asmatch line segment 1210 0,1. Sincematch line segment 1210 0,1 is asserted, IAD1 will be asserted indicating that the most significant priority number is stored in that row. - Thus, when the processing of step 706 is completed, bits PNUM3-PNUM0 indicate that the most significant priority number stored in the priority index table is 0101, and IAD1 is asserted identifying that 0101 is stored in row one.
- Any circuits may be used for compare
circuits 1206 and/orisolation circuits 1204 to implement the process illustrated above. Table 1 shows one example of a truth table for implementing each comparecircuit 1206, where X (column) and Y (row) are any integers. Other truth tables may be used (and corresponding logic generated accordingly) including those that logically complement one of more or the signals indicated in Table 1.TABLE 1 STATE 1208 1102 1210X, Y 1210X−1, Y 0 0 0 0 0 1 0 0 1 1 2 0 1 0 0 3 0 1 1 0 4 1 0 0 0 5 1 0 1 1 6 1 1 0 0 7 1 1 1 1 - Note that
state 5 may not occur sincepriority signal line 1208 will not be a logic one when amemory element 1102 stores a logic zero and the correspondingmatch line segment 1210 is asserted to a logic one state. For other truth tables,state 5 may occur. - Any logic or circuitry may be used to implement the truth table of Table 1.
FIG. 14 shows one embodiment of comparecircuit 1206 n-1,0 andisolation circuit 1204 n-1,0 for implementing the truth table of Table 1. Comparecircuit 1403 is one embodiment of comparecircuit 1206 n-1,0, andisolation circuit 1401 is one embodiment ofisolation circuit 1204 n-1,0. The embodiment ofFIG. 14 may be used to implement all of thepriority logic elements 1104 in the priority index table. - Compare
circuit 1403 includesinverter 1414,transistors priority signal line 1208 n-1 and ground, andtransistors match line segment 1210 n-2,0 and ground. N-channel transistor 1406 has its drain coupled to signalline 1208 n-1, it gate coupled to matchline segment 1210 n-1,0, and its source coupled to the drain of n-channel transistor 1408.Transistor 1408 has its gate coupled to receive the logical complement of the priority number bit (/D) stored inmemory element 1102 n-1,0, and its source coupled to ground. N-channel transistor 1410 has its drain coupled to matchline segment 1210 n-2,0, its gate coupled to signalline 1208 n-1 viainverter 1414, and its source coupled to the drain of n-channel transistor 1412.Transistor 1412 has its gate coupled to receive the priority number bit (D) stored inmemory element 1102 n-1,0, and its source coupled to ground. Any oftransistors -
Isolation circuit 1401 includesinverters -
FIG. 15 shows another embodiment of comparecircuit 1206 n-1,0 andisolation circuit 1204 n-1,0 for implementing the truth table of Table 1. In this embodiment, the compare circuit and isolation circuit are merged into the same logic that includestransistors FIG. 14 ,inverter 1501,NAND gate 1502, and ANDgate 1503.NAND gate 1502 has one input coupled to signalline 1208 n-1 viainverter 1501, and another input coupled to receive the priority number bit (D) stored inmemory element 1102 n-1,0. ANDgate 1503 has one input coupled to matchline segment 1210 n-1,0, another input coupled to the output ofNAND gate 1502, and an output coupled to matchline segment 1210 n-2,0. - For the example described above with respect to
FIG. 13 , the most significant priority number is the lowest number such that 0101 is the most significant number between 0101 and 0110. For another embodiment, the priority numbers are stored in descending priority order such that 0110 is the most significant priority number between 0101 and 0110. For this embodiment, compare circuits 1206 0,0-1206 3,1 determine that 0110 is the most significant priority number, and assert IAD0 to indicate that 0110 is stored in row zero of the priority index table. For this embodiment, as shown inFIG. 16 , inverters 1212 3-1212 0 output the logical states of priority signal lines 1208 3-1208 0, respectively, as PNUM3-PNUM0. For other embodiments, inverters 1212 3-1212 0 are not required. Compare circuits 1206 0,0-1206 3,1 and isolation circuits 1204 0,0-1204 3,1 determine that 0110 is the highest priority number as follows. - As in the previous embodiment, the most significant bit PNUM3 is resolved first. For this embodiment,
memory elements 1102 that store a logic zero do not discharge their correspondingpriority signal lines 1208. Since both ofmemory elements signal line 1208 3 remains pre-charged such that PNUM3 is a logic zero. Additionally, comparecircuit 1206 3,1 compares the state ofpriority signal line 1208 3 with the priority number bit stored in 1102 3,1 and determines that both have different logical states. This causes comparecircuit 1102 3,1 to not affect the logical state ofmatch line segment 1210 2,1 such thatmatch line segment 1210 2,1 has the same logic state asmatch line segment 1210 3,1. Similarly, comparecircuit 1206 3,0 compares the state ofpriority signal line 1208 3 with the priority number bit stored in 1102 3,0 and determines that both have the different logical states. This causes comparecircuit 1102 3,0 to not affect the logical state ofmatch line segment 1210 2,0 such thatmatch line segment 1210 2,0 has the same logic state asmatch line segment 1210 3,0. - PNUM2 is resolved next. When any
memory element 1102 stores a logic one and the correspondingmatch line segment 1210 is asserted, the correspondingpriority signal line 1208 is discharged. Thus, each of memory comparecircuits discharge signal line 1208 2 such that PNUM2 is a logic one. Additionally, comparecircuit 1206 2,1 does not affectmatch line segment 1210 1,1, becausememory element 1102 2,1 andpriority signal line 1208 2 have different logic states. Thus,match line segment 1210 1,1 will have the same logic state asmatch line segment 1210 2,1. Similarly, comparecircuit 1206 2,0 does not affectmatch line segment 1210 1,0 becausememory element 1102 2,0 andpriority signal line 1208 2 have different logic states. Thus,match line segment 1210 1,0 has the same state asmatch line segment 1210 2,0. - PNUM1 is resolved next. Since
memory element 1102 1,0 stores a logic one andmatch line segment 1210 1,0 is asserted, comparecircuit 1206 1,0 dischargespriority signal line 1208 1. This causes PNUM1 to be a logic one. Additionally, comparecircuit 1206 1,0 compares the logic zero state ofpriority signal line 1208 1 with the logic one stored in 1102 1,0 and allowsmatch line segment 1210 0,0 to have the same logic state asmatch line segment 1210 1,0. Comparecircuit 1206 1,1, however, compares the logic zero onpriority signal line 1208 1 with the logic zero stored inmemory element 1102 1,1, and de-assertsmatch line segment 1210 0,1. As in the example ofFIG. 13 , when a match line segment is de-asserted, all subsequent compare circuits for that row will de-assert the remaining match line segments for the row such that the corresponding internal address line IAD will be de-asserted. When IAD is de-asserted for a particular row, this indicates that the most significant priority number in not stored in that row. Additionally, when the remaining match line segments are de-asserted, the compare circuits for that row do not discharge the remaining priority signal lines regardless of the logic states stored in their corresponding memory elements of that row. For example, comparecircuit 1206 0,1 does not dischargepriority signal line 1208 0 even thoughmemory element 1102 0,1 stores a logic one. Additionally,isolation circuits match line segment 1210 0,1 frommatch line segments match line segment 1210 0,1. - Lastly, the least significant priority number bit PNUM0 is resolved. Compare
circuit 1206 0,0 alone determines PNUM0 since comparecircuit 1206 0,1 can not dischargepriority signal line 1208 0. Sincememory element 1102 0,0 stores a logic zero andmatch line segment 1210 0,0 is asserted, comparecircuit 1206 0,0 leavespriority signal line 1208 0 pre-charged, and PNUM0 is a logic zero. Additionally, comparecircuit 1206 0,0 allows IAD0 to have the same logic state asmatch line segment 1210 0,0. Sincematch line segment 1210 0,0 is asserted, IAD0 will be asserted indicating that the most significant priority number is stored in that row. - Thus, when the processing of step 706 is completed, PNUM3-PNUM0, for this embodiment, indicate that the most significant priority number stored in the priority index table is 0110, and IAD0 is asserted identifying that 0110 is stored in row zero.
- Any circuits may be used for compare
circuits 1206 and/orisolation circuits 1204 to implement the process illustrated above forFIG. 16 . Table 2 shows one example of a truth table for implementing each comparecircuit 1206, where X (column) and Y (row) are any integers. Other truth tables may be used (and corresponding logic generated accordingly) including those that logically complement one of more of the signals indicated in Table 2.TABLE 2 STATE PNUM 1102 1210X, Y 1210X−1, Y 0 0 0 0 0 1 0 0 1 1 2 0 1 0 0 3 0 1 1 1 4 1 0 0 0 5 1 0 1 0 6 1 1 0 0 7 1 1 1 1 - Note that
state 3 may not occur since PNUM will not be a logic zero when amemory element 1102 stores a logic one and the correspondingmatch line segment 1210 is asserted to a logic one state. For other truth tables,state 3 may occur. - Any logic or circuitry may be used to implement the truth table of Table 2. For one embodiment, the compare
circuit 1403 andisolation circuit 1401 ofFIG. 14 may be used. For this embodiment,transistor 1408 has its gate coupled to receive the priority number bit (D) stored inmemory element 1102 n-1,0, andtransistor 1412 has its gate coupled to receive the logical complement of the priority number bit (/D) stored inmemory element 1102 n-1,0. Additionally,signal line 1208 n-1 provides the logical complement of PNUMn-1. - For another embodiment, the logic and circuitry of
FIG. 15 may be used to implement the truth table of Table 2. For this embodiment,transistor 1408 has its gate coupled to receive the priority number bit (D) stored inmemory element 1102 n-1,0, andNAND gate 1502 has an input coupled to receive the logical complement of the priority number bit (/D) stored inmemory element 1102 n-1,0. Additionally,signal line 1208 n-1 provides the logical complement of PNUMn-1. -
FIG. 12B shows another embodiment of 1201 0 of the priority index table. For this embodiment, IAD0 is determined onsignal line 1216 that is separate from the match line segments. As shown inFIG. 12B , each match line signal line is coupled to a discharge circuit that includes an inverter-transistor pair (i.e., inverters 1212 n-1,0-1212 0,0 and corresponding transistors 1214 n-1-1214 0,0) such that if any of the match line segments is discharged, then signalline 1216 is discharged to a low logic state. For other embodiments, other discharge circuitry may be used.Signal line 1216 is pre-charged to a high logic state bypre-charge circuit 1202. For an alternative embodiment,signal line 1216 may be driven directly by the match line segments. - The previously described embodiments of
system 600 ofFIG. 6 show that policy statements can be loaded intoCAM array 604 in any order. When an incoming packet is received,DSP 602 can identify the address inCAM array 604 of the highest priority policy statement that matches the policy information of the incoming packet. The identified address can then be used to access routing information stored inroute memory 614.DSP 602 can perform this function without the user having to preorder the policy statements for entry into the CAM array. This can reduce the hardware and/or software needed for table management of the CAM array, and can increase the performance of a router incorporating the DSP. -
DSP 602 can operate asynchronously or synchronously. WhenDSP 602 operates synchronously, it receives a clock signal that may be used to clock in the policy search key and an instruction that causes the process ofFIG. 7 to be performed byDSP 602.DSP 602 may implement the classification function ofFIG. 7 in one or more clock cycles. - Inserting and Deleting Policy Statements
- As previously described, priority numbers for policy statements may be assigned in ascending or descending priority order such that there are gaps left between the numbers to accommodate the new priority numbers associated with new policy statements to be stored in
CAM array 604. Alternatively, the priority numbers may be assigned in consecutive ascending or descending priority order. New policy statements and their associated priority numbers can be added to the tables 604 and 608 in conformance with either assignment method without having to reload or physically reorder the policy statements or the priority numbers in the respective tables. - Each new policy statement can be loaded into any location (e.g., the next free location) in
CAM array 604, and can be assigned a new priority number without having to reload or reorderCAM array 604 andpriority memory 608. When a new policy statement is received, its priority number can be compared with the existing priority numbers already stored inpriority memory 608 to determine if a policy statement already exists that has been assigned that priority. It is generally desirable that no two policy statements have the same priority number. Thus, if the priority number already exists, the network administrator or the policy-based router itself can assign the new policy statement a new priority number, or the priority number of the existing policy statement can be updated (i.e., incremented or decremented). Since the existing priority numbers are stored in ascending or descending order, updating one priority number may also result in the need to update other priority numbers such that no two priority numbers are the same. - For one embodiment,
DSP 602 may include additional circuitry to determine if at least one of the existing priority numbers stored inmemory 608 is greater than or equal to (e.g., for ascending priority order), or, alternatively, less than or equal to (e.g., for descending priority order), the new priority number. If so, the existing priority numbers that are identified by the comparison may be updated such that a new policy statement does not have the same priority number as an existing policy statement. For other embodiments described below,priority logic 610 itself in priority index table 606 may perform the additional compare function(s) and control the updating function(s). -
FIG. 17 shows one embodiment of priority index table 606 ofFIG. 6 including inequality circuits INEQ 1704 0-1704 N-1 that perform the additional comparison functions and control the updating functions. The inequality circuits determine if a newpriority number PNEW 1710 of anew policy statement 1708 is the same, higher, or, alternatively, lower than any of the priority numbers already stored inmemory 608. The inequality circuits also update the priorities of the existing policy statements in order to insert the new policy statement. The new policy statement and its associated priority number do not need to be physically inserted between the existing entries; rather, they can be loaded into any desirable address (e.g., the next free address) in the CAM array and priority memory, and the priority numbers of the existing policy statements updated accordingly. Similarly, when a policy statement is removed (i.e., invalidated or overwritten) from the CAM array, the inequality circuits may update the priority numbers of the previously stored policy statements. These updating functions can be performed by the inequality circuits without the need to physically reorder the policy statements in the CAM array, or to physically reorder the priority numbers in the priority index table. This can reduce the hardware and/or software needed for table management of the CAM array, and can increase the performance of a router incorporating the DSP. - For an alternative embodiment, the inequality circuits do not update the existing, stored priority numbers during or after an insert or delete operation. For this embodiment, gaps may be left in the priority numbers after an insert or delete function. These gaps may or may not be filled with subsequent priority numbers.
- As shown in
FIG. 17 , each location 608 0-608 N-1 in the priority memory has a corresponding inequality circuit 1704 0-1704 N-1 that compares the newpriority number PNEW 1710 with the priority number P0-PZ stored at that location. If PNEW is the same or a higher priority (e.g., is greater than or equal to, or, alternatively less than or equal to) than an existing priority number, an inequality circuit will update the existing priority number. Thenew policy statement 1708, PNEW, and the associated route information RINEW, can then be written intoCAM array 604,priority memory 608, androute memory 614, respectively, using conventional write circuits, counters, address decoders, etc. (not shown). - Inequality circuits 1704 0-1704 N-1 may be part of priority index table 606 as shown in
FIG. 17 . Alternatively, inequality circuits 1704 0-1704 N-1 may be separate from priority index table 606. The inequality circuits may be any inequality circuit that determines whether one number is greater than or equal to, greater than, less than or equal to, or less than another number. For an alternative embodiment, inequality circuits 1704 0-1704 N-1 may be one circuit that consecutively, or in parallel, compares PNEW with each of the priority numbers P0-PZ. -
FIG. 18 shows one example of addingnew policy statement 1708 toCAM array 604 at the next free address of CAM array 604 (location 6043) and updating the existing priority numbers stored inmemory 608.New policy statement 1708 is assignedpriority number 1, and has route information RI3.CAM array 604 already includespolicy statements locations corresponding priority numbers locations priority memory 608. Additionally, the corresponding route information RI0, RI1, and RI2, are stored atlocations route memory 614. For this embodiment, the policy statements are assigned priority numbers in ascending priority order such thatpolicy statement 203 is identified as the highest priority policy statement by being assignedpriority number 0;policy statement 201 is identified as the lowest priority policy statement by being assignedpriority number 2; and,policy statement 202 is identified as having a priority greater than that ofpolicy statement 201, but less than that ofpolicy statement 203, by being assignedpriority number 1. - Inequality circuits 1704 0-1704 2 compare PNEW with the exiting priority numbers as follows. The comparisons are made in parallel by each inequality circuit, however, they may also be made consecutively or in groups.
Inequality circuit 1704 2 compares itscorresponding priority number 0 with thenew priority number 1 and determines that 0 is a higher priority than 1 (i.e., that 0 is numerically less than 1). The priority number forpolicy statement 203 is not updated.Inequality circuit 1704 1 compares itscorresponding priority number 1 with thenew priority number 1 and determines that they have the same priority. The priority number forpolicy statement 202 is incremented (updated) by one to thenumber 2.Inequality circuit 17040 compares itscorresponding priority number 2 with thenew priority number 1 and determines that 1 is a higher priority than 2 (i.e., that 1 is numerically less than 2). The priority number forpolicy statement 201 is then incremented to 3. The modified table entries after the insertion of the new policy statement and its associated priority number are shown inFIG. 19 . -
FIG. 20 shows another example of addingnew policy statement 1708 toCAM array 604 at the next free address of CAM array 604 (location 604 3). For this embodiment,policy statements priority numbers priority number 15. The policy statements are assigned in descending priority order such thatpolicy statement 203 is identified as the highest priority policy statement by being assignedpriority number 16;policy statement 201 is identified as the lowest priority policy statement by being assignedpriority number 14; and,policy statement 202 is identified as having a priority greater than that ofpolicy statement 201, but less than that ofpolicy statement 203, by being assignedpriority number 15. -
Inequality circuit 1704 2 compares itscorresponding priority number 16 with thenew priority number 15 and determines that 16 is a higher priority than 15 (i.e., that 16 is numerically greater than 15). The priority number forpolicy statement 203 is not updated.Inequality circuit 1704 0 compares itscorresponding priority number 15 with thenew priority number 15 and determines that they have the same priority. The priority number forpolicy statement 202 is decremented (updated) by one to thenumber 14.Inequality circuit 1704 0 compares itscorresponding priority number 14 with thenew priority number 15 and determines that 15 is a higher priority than 14 (i.e., that 14 is numerically less than 15). The priority number forpolicy statement 201 is then decremented to 13. The modified table entries after the insertion of the new policy statement and its associated priority number is shown inFIG. 21 . - As indicated above, the priority numbers can be incremented, decremented, or left unchanged after PNEW is compared with the priority number.
FIG. 22 showspriority memory 2208 that is one embodiment ofpriority memory 608.Priority memory 2208 includes counters 2208 0-2208 N-1. Each counter corresponds to priority memory location 608 0-608 N-1, respectively, and stores the priority number for that location. The counters can be loaded (i.e., written to) using write circuits, read from using read circuits, or reset as generally known in the art. Additionally, each counter 2208 0-2208 N-1 provides its stored priority number P0-PZ to a corresponding inequality circuit 1704 0-1704 N-1 for comparison with PNEW. When PNEW has a higher priority than the stored priority number for a given location, then the inequality circuit will assert its count signal COUNT0-COUNTN-1 such that the corresponding counter is incremented (or, alternatively, decremented). The counter will be incremented if the UP signal is asserted, or will be decremented if the DOWN signal is asserted. The UP and DOWN signals may be generated by an instruction decoder (not shown) onDSP 602 that receives and decodes an insert (write) instruction or a delete (invalidate) instruction to the DSP. For an alternative embodiment, the inequality circuits may each output UP and DOWN signals to their corresponding counters directly in response to one or more signals from the instruction decoder. The counters may be updated synchronously or asynchronously. - Inequality circuits 1704 0-1704 N-1 may be disposed external to
priority memory 2208, or they may be integrated into the priority memory itself.FIG. 23 shows one embodiment of integrating an inequality circuit with a counter in the priority memory. For this embodiment,inequality circuit 2304 0 is one embodiment ofinequality circuit 1704 0, and may also be used for any of inequality circuits 1704 1-1704 N-1.Counter 2306 0 is one embodiment ofcounter 2208 0, and may also be used for any of counters 2208 1-2208 N-1. - As shown in
FIG. 23 , memory storage elements 2314 n-1,0-2314 0,0 have associated counter logic circuits 2316 n-1,0-2316 0,0 to form a conventional counter. Each memory storage element can be a volatile or non-volatile RAM or ROM storage element such asmemory storage elements 1102 ofFIG. 11 . The memory storage elements each store one bit of a priority number such thatmemory storage element 2314 n-1,0 stores the most significant priority number bit, 2314 n-2,0 stores the next most significant priority number bit, and 2314 0,0 stores the least significant priority number bit. Logic circuits 2316 n-1,0-2316 0,0 receive the UP, DOWN, and COUNT0 signals that indicate when the counter should increment, decrement, or leave unchanged the priority number stored in the counter. The counter can be configured to form any type of counter including the up/down counter shown. Additionally, load and reset signals can be provided to the logic circuits to write or reset the counter. Bit lines, word lines, and read/write circuits generally known in the art may also be included to access the priority number bits stored in the memory storage elements. -
Inequality circuit 2304 0 includes compare circuits 2310 n-1,0-2310 0,0. Each compare circuit can be any digital or analog compare circuits. Each compare circuit 2310 n-1,0-2310 0,0 compares a priority number bit, stored in a corresponding memory storage element 2314 n-1,0-2314 0,0, with one of the corresponding bits PNEWn-1-PNEW0 provided on signal lines 2308 n-1-2308 0. Additionally, each compare circuit monitors the comparison result of the more significant priority number bits with PNEW via the logical states of control line segments 2312 n-2,0-2312 0,0. The first control line segment 2312 n-1,0 may be pre-charged bypre-charge circuit 2302 to indicate that it is the first control line segment. Alternatively, control line segment 2312 n-1,0 may be initially discharged to indicate that it is the first control line segment. - For this embodiment, the logical state of COUNT0 on
signal line 2318 indicates the comparison result between PNEW and the priority number stored incounter 2306 0. COUNT0 is initially pre-charged by apre-charge circuit 2302 to a high logic state. If COUNT0 remains pre-charged after each compare circuit performs its comparison function, then the priority number incounter 2306 0 is updated. Alternatively, if COUNT0 is discharged by one of the compare circuits, then the priority number incounter 2306 0 is updated. For other embodiments, COUNT0 may be initially discharged to a logic low state, and counter 2306 updated if COUNT0 remains discharged, or, alternatively, if COUNT0 is charged to a logic one state by one of the compare circuits. For yet another embodiment,signal line 2318 may be segmented like control line segments 2312 between compare circuits, and thesegmented signal line 2318 output from comparecircuit 2310 0,0 provided to counterlogic circuit 2316 0,0 as COUNT0. - The operation of the embodiment of
FIG. 23 can be illustrated with the example shown inFIG. 24 . In this example, a 3×3 matrix of rows and columns stores three priority numbers. For other embodiments, any numbers of rows and columns can be used.Row 0 stores priority number 010 having the decimal equivalent of thenumber 2,row 1 stores priority number 001 having the decimal equivalent of thenumber 1, androw 2 stores priority number 000 having the decimal equivalent of thenumber 0. For this example, the priority numbers are assigned in ascending priority order such that 000 is the most significant priority number, 001 is the next most significant priority number, and 010 is the least significant priority number. - When a new policy statement having PNEW 001 is to be loaded into
CAM array 604, inequality circuits 2304 2-2304 0 compare PNEW with their corresponding priority numbers stored in counters 2306 2-2306 0, respectively.Inequality circuit 2304 2 determines that 000 is a higher priority than PNEW;inequality circuit 2304 1 determines that 001 is equal to PNEW; andinequality circuit 2304 0 determines that 010 is a lower priority than PNEW.Inequality circuit 2304 2 leaves its priority number unchanged.Inequality circuits CAM array 604 andpriority memory 608, respectively. -
Inequality circuits corresponding counters memory storage elements - The comparison of the next most significant bit PNEW1 with priority number bits stored in
memory elements memory elements control line segments 2312 0,2 and 2310 0,1 are asserted to high logic states. Comparecircuit 2310 1,0 however, determines that the priority number stored incounter 2306 0 is greater than PNEW because: (1)memory element 2314 1 stores a logic one; (2) the control signal is high on control line segment 2312 1,0; and (3) PNEW1 is a logic zero onsignal line 2308 1. When comparecircuit 2310 1,0 makes this determination, it leaves COUNT0 unaffected so that it will remain pre-charged to update the priority number stored incounter 2306 0. Additionally, comparecircuit 2310 1,0 asserts the control signal on control line segment 2312 0,0 to a low logic state. When the control signal oncontrol line segment 2310 0,0 is low, the control signals on all subsequent control line segments are also asserted to low logic states. When the control signals are low, all lower priority compare circuits are rendered incapable of discharging COUNT0 regardless of what is stored in their corresponding memory elements, and regardless of the logical states of the subsequent PNEW bits. For example, 2310 0,0 will be rendered incapable of discharging COUNT0 even though PNEW0, a logic one, is greater than the logic zero stored inmemory element 2314 0,0. - Lastly, the comparison of the least significant bit PNEW0 with priority number bits stored in
memory elements inequality circuit 2304 0 has already determined that 010 is greater than 001 and thus COUNT0 remains pre-charged to update the priority number stored incounter 2306 0. Since the control signal on control line segment 2312 0,1 is high, comparecircuit 2310 0,1 compares the logic one stored inmemory element 2314 0,1 with the logic one of PNEW0 and determines that PNEW is the same number as that stored incounter 2306 1. In response, COUNT1 remains in a pre-charged state to update the priority number stored incounter 2306 1. Comparecircuit 2310 0,2, however, determines that PNEW0 is greater than the logic zero stored inmemory element 2314 0,2. Since the control signal is high on control line segment 2312 0,2, comparecircuit 2310 0,2 discharges COUNT2 such that the priority number 000 stored incounter 2306 2 will not be updated. After a sufficient amount of time has passed such that all of the bit comparisons have had time to resolve, the UP signal can be asserted such that the priority numbers incounters - Any compare circuits may be used for compare
circuits 2310 to implement the process illustrated above. Table 3 shows one example of a truth table for implementing each comparecircuit 2310, where X (column) and Y (row) are any integers. Other truth tables may be used (and corresponding logic generated accordingly) including those that logically complement one or more of the signals indicated in Table 3.TABLE 3 PNEW X2314X, Y 2312X, Y 2312X−1, Y COUNT 0 0 0 0 COUNT 0 0 1 1 COUNT 0 1 0 0 COUNT 0 1 1 0 COUNT 1 0 0 0 COUNT 1 0 1 1 0 1 1 0 0 COUNT 1 1 1 1 COUNT - Any logic or circuitry may be used to implement the truth table of Table 3.
FIG. 25A shows comparecircuit 2502 that is one embodiment of comparecircuit 2310 n-1,0 for implementing the truth table of Table 3. The embodiment ofFIG. 25A may be used to implement all of the comparecircuits 2310. - Compare
circuit 2502 includesinverter 2510,NAND gate 2508, ANDgate 2506, and n-channel transistors NAND gate 2508 has one input coupled to signalline 2308 n-1 viainverter 2510, and another input coupled to receive the data (D) stored inmemory element 2314 n-1,0. ANDgate 2506 has one input coupled to the output ofNAND gate 2508, another input coupled to control line segment 2312 n-1,0, and an output coupled to control line segment 2312 n-2,0.Transistor 2504 has its drain coupled to thesignal line 2318, its gate coupled to the output of ANDgate 2506, and its source coupled to the drain oftransistor 2503.Transistor 2503 also has its gate controlled bysignal line 2308 n-1, and its source coupled to the drain oftransistor 2507.Transistor 2507 also has its gate coupled to receive /D frommemory 2314 n-1,0, and its source coupled to ground. -
FIG. 25B shows comparecircuit 2522 that is another embodiment of comparecircuit 2310 n-1,0 for implementing the truth table of Table 3. In this embodiment,signal line 2318 is segmented into separate control line segments. The embodiment ofFIG. 25B may be used to implement all of the comparecircuits 2310. - Compare
circuit 2522 includesinverter 2512,NAND gates gates NAND gate 2514 has one input coupled to signalline 2308 n-1 viainverter 2512, and another input coupled to D ofmemory 2314 n-1,0. ANDgate 2520 has one input coupled to control line segment 2312 n-1,0, another input coupled to the output ofNAND gate 2514, and an output coupled to control line segment 2312 n-2,0.NAND gate 2516 has a first input coupled to control line segment 2312 n-1,0, a second input coupled to /D ofmemory 2314 n-1,0, and a third input coupled to signalline 2308 n-1. ANDgate 2518 has a first input coupled to countline segment 2318 n-1,0, a second input coupled to the output ofNAND gate 2516, and an output coupled to countline segment 2318 n-2,0. - For the example described above with respect to
FIG. 24 , the most significant number is the lowest number such that 000 is the most significant number between 000, 001, and 010. For another embodiment, the priority numbers are assigned in descending priority order such that the most significant priority number is the priority number with the highest numerical value, and the least significant priority number is the priority number with the lowest numerical value.FIG. 26 shows such an example in which the 3×3 matrix ofFIG. 24 includes the binary priority number 111 (decimal equivalent of 7) stored in counter 23062, priority number 110 (decimal equivalent of 6) stored incounter 2306 1, and priority number 101 (decimal equivalent of 5) stored incounter 23060.Numbers 7, 6, and 5 are stored inrows - When a new policy statement having PNEW 110 is to be loaded into
CAM array 604, the logical complement of PNEW (i.e., 001) is provided to signal lines 2308 2-2308 0. Inequality circuits 2304 2-2304 0 compare PNEW with their corresponding priority numbers stored in counters 2306 2-2306 0, respectively.Inequality circuit 2304 2 determines that 111 is a higher priority than PNEW;inequality circuit 2304 1 determines that 110 is equal to PNEW; andinequality circuit 2304 0 determines that 101 is a lower priority than PNEW.Inequality circuit 2304 2 leaves its priority number unchanged.Inequality circuits corresponding counters CAM array 604 andpriority memory 604, respectively. -
Inequality circuits corresponding counters memory storage elements - The comparison of the next most significant bit PNEW1 with priority number bits stored in
memory elements memory elements control line segments 2312 0,2 and 2310 0,1 are asserted to high logic states. Comparecircuit 2310 1,0, however, determines that the priority number stored incounter 23060 is less than PNEW because: (1)memory element 2314 1,0 stores a logic zero; (2) the control signal is high on control line segment 2312 1,0; and (3) PNEW1 is a logic one. When comparecircuit 2310 1,0 makes this determination, it leaves COUNT0 unaffected so that it will remain pre-charged to update the priority number stored incounter 2306 0. Additionally, comparecircuit 2310 1,0 asserts the control signal on control line segment 2312 0,0 to a low logic state. When the control signal oncontrol line segment 2310 0,0 is low, the control signals on all subsequent control line segments are also asserted to low logic states. When the control signals are low, all lower priority compare circuits are rendered incapable of discharging COUNT0 regardless of what is stored in their corresponding memory elements, and regardless of the logical states of the subsequent PNEW bits. For example, 2310 0,0 will be rendered incapable of discharging COUNT0 even though PNEW0, a logic zero, is less than the logic one stored inmemory element 2314 0,0. - Lastly, the comparison of the least significant bit PNEW0 with priority number bits stored in
memory elements inequality circuit 2304 0 has already determined that 101 is less than 110 and thus COUNT0 remains pre-charged to update the priority number stored incounter 2306 0. Since the control signal on control line segment 2312 0,1, is high, comparecircuit 2310 0,1 compares the logic zero stored inmemory element 2314 0,1 with the logic one online 2308 0 and determines that PNEW is the same number as that stored incounter 2306 1. In response, COUNT1 remains in a pre-charged state to update the priority number stored incounter 2306 1. Comparecircuit 2310 0,2, however, determines that PNEW0 is less than the logical one stored inmemory element 2314 0,2. Since the control signal is high on control line segment 2312 0,2, comparecircuit 2310 0,2 discharges COUNT2 such that the priority number 111 stored incounter 2306 2 will not be updated. After a sufficient amount of time has passed such that all of the bit comparisons have had time to resolve, the DOWN signal can be asserted such that the priority numbers incounters - Any compare circuits may be used for compare
circuits 2310 to implement the process illustrated above. Table 4 shows one example of a truth table for implementing each comparecircuit 2310, where X (column) and Y (row) are any integers. Other truth tables may be used (and corresponding logic generated accordingly) including those that logically complement one or more of the signals indicated in Table 4.TABLE 4 PNEW X2314X, Y 2312X, Y 2312X−1, Y COUNT 0 0 0 0 COUNT 0 0 1 1 COUNT 0 1 0 0 COUNT 0 1 1 1 0 1 0 0 0 COUNT 1 0 1 0 COUNT 1 1 0 0 COUNT 1 1 1 1 COUNT - Any logic or circuitry may be used to implement the truth table of Table 4.
FIG. 27A shows comparecircuit 2702 that is one embodiment of comparecircuit FIG. 27A may be used to implement all of the comparecircuits 2310. - Compare
circuit 2702 includesinverter 2510,NAND gate 2508, ANDgate 2506, and n-channel transistors FIG. 25A .NAND gate 2508 has one input coupled to signalline 2308 n-1, and another input coupled to receive the /D stored inmemory element 2314 n-1,0. AND gate has one input coupled to the output ofNAND gate 2508, another input coupled to control line segment 2312 n-1,0, and an output coupled to control line segment 2312 n-2,0.Transistor 2504 has its drain coupled to thesignal line 2318, its gate coupled to the output of ANDgate 2506, and its source coupled to the drain oftransistor 2503.Transistor 2503 also has its gate controlled bysignal line 2308 n-1 viainverter 2510, and its source coupled to the drain oftransistor 2507.Transistor 2507 also has its gate coupled to receive D frommemory 2314 n-1,0, and its source coupled to ground. -
FIG. 27B shows comparecircuit 2704 that is another embodiment of comparecircuit 2310 n-1,0 for implementing the truth table of Table 4. In this embodiment,signal line 2318 is segmented into separate control line segments. The embodiment ofFIG. 27B may be used to implement all of the comparecircuits 2310. - Compare
circuit 2704 includesinverter 2708,NAND gate 2710, and ANDgates NAND gate 2710 has a first input coupled to signalline 2308 n-1, a second input coupled to D ofmemory 2314 n-1,0, and a third input coupled to control line segment 2312 n-1,0 viainverter 2708. ANDgate 2706 has one input coupled to countline segment 2318 n-1,0, another input coupled to the output ofNAND gate 2710, and an output coupled to countline segment 2318 n-2,0. ANDgate 2712 has a first input coupled to control line segment 2312 n-1,0, a second input coupled to /D ofmemory 2314 n-1,0, a third input coupled to signalline 2308 n-1, and an output coupled to control line segment 2312 n-2,0. - The embodiments of
FIGS. 17-27 show that a new policy statement having a priority greater than or equal to an existing policy statement, can be added to the DSP without having to physically reorder or reload the CAM array storing the policy statements, or the memory storing the priority numbers. Instead, inequality circuits update the appropriate existing priority numbers. Since the updating function can be performed without the need to physically reorder or reload the policy statements in the CAM array, or to physically reorder or reload the priority numbers in the priority index table, this can reduce the hardware and/or software needed for table management of the CAM array, and can increase the performance of a router incorporating the DSP. - Policy statements may also be deleted from
CAM array 604 using priority index table 606 ofFIG. 17 . When a policy statement is deleted, the corresponding priority number inpriority memory 608 is also deleted. The policy statements and priority numbers can be deleted by setting one or more valid bits to an appropriate state for the row ofCAM array 604 that stores the policy statement to be deleted. The valid bit(s) may be stored inCAM array 604,priority memory 608, or in each of the memory arrays. - When a priority number is deleted, the remaining priority numbers in
priority memory 608 can be left unchanged. This may leave gaps in the priority numbers stored in the priority memory. These gaps may be filled in by new priority numbers associated with new policy statements to be added to the DSP, or they may remain as unused gaps. For another embodiment, remaining priority numbers in the priority memory can be updated by the inequality and counter circuits. If the priority numbers are assigned in ascending priority order, and one of the priority numbers is deleted or invalidated, then any other priority numbers that are greater than the deleted number can be decremented by the inequality circuits. For example, ifpriority memory 608stores priority numbers priority number 1 is invalidated, thenpriority number 2 can be decremented to 1. Similarly, if the priority numbers are assigned in descending priority order, and one of the priority numbers is deleted or invalidated, then any other priority numbers that are less than the deleted number can be incremented by the inequality and counter circuits. For example, ifpriority memory 608stores priority numbers priority number 1 is invalidated, thenpriority number 0 can be incremented to 1. Updating the priority numbers can help to avoid unused gaps inpriority memory 608. -
FIG. 28 shows one embodiment of aprocess 2800 for deleting or invalidating policy statements and their associated priority numbers. Atstep 2802, the priority number to be deleted is compared with the entries in the priority memory. For this embodiment,priority memory 608 is a CAM, or has one or more compare circuits external to the memory to perform the comparison. For an alternative embodiment,step 2802 may compare the policy statement with the existing entries inCAM array 604. For this embodiment,priority memory 608 may not be a CAM. - If there is no match at
step 2802, the process stops atstep 2804. The process could also stop if there was more than one match, and an error flag such as a multiple match flag may be asserted. Alternatively, the process may continue to delete all identified matches. If there is a match, both the policy statement and the priority number are invalidated or deleted atstep 2806. As described above, this may be accomplished by setting a valid bit(s) to an appropriate state for that row inpriority memory 608. If the valid bit(s) is located inCAM array 604, and is inaccessible bypriority memory 608, thenpriority logic 610 can then identify the location of the matching entry (as described above), and provide the indication to IAD0-IADN-1. These signal lines can then be provided throughencoder 612 back to CAM array 604 (e.g., through a decoder) to address the appropriate location and invalidate the policy statement. Alternatively, the signal lines IAD0-IADN-1 can be provided directly toCAM array 604 to access the appropriate location for invalidation. - The process may stop after
step 2806 and potentially leave gaps in the priority numbers. Alternatively, the process may proceed to step 2808 to update the remaining entries in the priority memory. When the process continues, the deleted priority number is provided to the inequality circuits 1704 0-1704 N-1, atstep 2808, to determine which entries in the priority memory have a lower priority than the deleted priority number. The inequality and counter circuits may then update the identified numbers, atstep 2810, as previously described. - The delete functions illustrate one example of the
DSP 602 utilizing both the inequality circuits and the priority logic to operate on the priority numbers stored inpriority memory 608.FIG. 29 shows one example of combining an inequality circuit with a priority logic circuit into a single array such that each circuit operates on the priority numbers on a bit-by-bit basis. The embodiment ofFIG. 29 combines row 1201 0 of priority logic elements fromFIG. 12A withinequality circuit 2304 0 and counter 2306 0 ofFIG. 23 to form row zero in priority index table 606 ofFIG. 17 . For an alternative embodiment,row 1201 0 fromFIG. 12B may also be used. The priority logic elements and inequality circuits share memory elements 2314 0,0-2314 n-1,0. Memory elements 2314 0,0-2314 n-1,0 also correspond to memory elements 1102 0,0-1102 n-1,0. Conventional bit lines, word lines, read and write circuits, and optional comparand data lines and match lines (i.e., when memory elements 2314 0,0-2314 n-1,0 are CAM cells) are not shown so as not to obscure the teachings of the figure. -
FIG. 30 shows another embodiment of combining therow 1201 0 of priority logic elements fromFIG. 12A with theinequality circuit 2304 0 ofFIG. 23 to form row zero in priority index table 1706 ofFIG. 17 . In this embodiment, signal lines 1208 n-1-1208 0 and 2308 n-1-2308 0 are replaced with a single set of signal lines 2906 n-1-2906 0 that can be used to determine the most significant priority number PNUM, or used to carry PNEW. Also, for this embodiment, match line segments 1210 n-1,0-1210 0,0 serve a dual function to operate as match line segments when determining the most significant priority number, and as control line segments (i.e., 2312 n-1,0-2312 0,0 ofFIG. 23 ) to generate COUNT0 when determining when a new priority number has a priority that is greater than or equal to the priority number stored incounter 2306. Thus, comparecircuit 2310 n-1,0 is coupled to matchline segment 1210 n-2,0, comparecircuit 2310 n-2,0 is coupled to matchline segment 1210 n-3,0, and so forth, and COUNT0 is provided on signal line IAD0. For an alternative embodiment, comparecircuits 2310 may provide control signals toisolation circuits 1204 instead of directly controlling the match line segments. - A control signal BCT indicates whether the priority index table uses the row of
priority logic elements 1201 0 to determine the most significant priority number inpriority memory 608, or whether the priority index table uses theinequality circuit 2304 0 to determine if a new priority number has a priority that is greater than or equal to the priority number stored incounter 2306. BCT may be provided by an instruction decoder inDSP 602 that interprets and decodes instructions provided to the DSP. When BCT is a low logic state, the priority logic elements perform their priority function; and when BCT is a high logic state, the inequality circuits perform their inequality function. Thus, when BCT is low, comparecircuit 1206 n-1,0 is enabled, comparecircuit 2310 n-1,0 is disabled, andlogic 2902 provides the state of ML0 to matchline segment 1210 n-1,0. When BCT is high, comparecircuit 1206 n-1,0 is disabled, comparecircuit 2310 n-1,0 is enabled, andlogic 2902 drives matchline segment 1210 n-1,0 to a high logic state such that match line segments 1210 n-1,0-1210 0,0 function as control line segments 2312 n-1,0-2312 0,0 ofFIG. 23 .Logic 2902 may be any logic that performs this function. - For another embodiment of
FIG. 30 ,row 1201 0 of priority logic elements fromFIG. 12B may be combined with theinequality circuit 2304 0 ofFIG. 23 to form row zero in priority index table 1706 ofFIG. 17 . - The logic and circuitry for implementing the priority logic elements and the inequality circuits illustrated in
FIGS. 29 and 30 may be those that implement the truth tables shown in Tables 1, 2, 3, and 4 above. For example, the logic and circuits illustrated inFIGS. 14, 15 , 25A, 25B, 27A, and 27B may be used. - For one embodiment, the priority numbers are assigned in ascending priority order in
priority memory 608. For this embodiment, logic and circuitry that implements the truth table shown in Table 1 (e.g., the logic and circuitry ofFIG. 14 or 15) may be used for the priority logic elements to determine the lowest priority number as the most significant priority number. Also for this embodiment, the logic and circuitry shown in the truth table shown in Table 3 (e.g., the logic and circuitry ofFIG. 25A or 25B) may be used for the inequality circuits to determine if a new priority number has a priority that is greater than or equal to the priority number stored incounter 2306. -
FIG. 31 shows one example of combining the logic and circuitry ofFIGS. 14 and 25 to implement the embodiment ofFIG. 30 when the priority numbers are assigned in ascending priority order inpriority memory 608. For this example, ORgate 3102 is one embodiment oflogic 2902. ORgate 3102 has one input coupled to ML0, and another input coupled to BCT. BCT is also coupled to the gate of n-channel transistor 3106 and one input of ORgate 3104 viainverter 3108.Transistor 3106 has its drain coupled to signalline 2906 n-1, and its source coupled to the drain oftransistor 1406. ORgate 3104 has another input coupled to the output of ANDgate 2506, and has its output coupled to the gate oftransistor 2504.Transistor 2504 is coupled betweenline segment 1210 n-2,0 andnode 3112. Transistor pairs 1410/1412 and 2503/2507 are each coupled in series betweennode 3112 and ground. - For another embodiment, the priority numbers are assigned in descending priority order in
priority memory 608. For this embodiment, logic and circuitry that implements the truth table shown in Table 2 (e.g., the logic and circuitry ofFIG. 14 or 15 modified as indicated above) may be used for the priority logic elements to determine the lowest priority number as the most significant priority number. Also for this embodiment, the logic and circuitry shown in the truth table shown in Table 4 (e.g., the logic and circuitry ofFIG. 27A or 27B) may be used for the inequality circuits to determine if a new priority number has a priority that is less than or equal to the priority number stored incounter 2306.FIG. 32 shows one example of combining the logic and circuitry ofFIGS. 14 and 27 to implement the embodiment ofFIG. 30 when the priority numbers are assigned in descending priority order inpriority memory 608. - The embodiments of
FIGS. 17-32 include additional inequality circuits to compare PNEW with the priority numbers already stored in the priority index table. For another embodiment,priority logic 610 itself can be used to determine whether to update existing priority numbers stored in the priority index table in response to an insert or delete instruction. As with the previously described embodiments above, the new policy statement and PNEW do not need to be physically inserted between the existing entries; rather, they can be loaded into any desirable address (e.g., the next free address) in the CAM array and priority memory, and the priority numbers of the existing policy statements updated accordingly. Similarly, when a policy statement is removed (i.e., invalidated or overwritten) from the CAM array, the priority logic may update the priority numbers of the previously stored policy statements. These updating functions can be performed by the priority logic without the need to physically reorder the policy statements in the CAM array, or to physically reorder the priority numbers in the priority index table. This can reduce the hardware and/or software needed for table management of the CAM array, and can increase the performance of a router incorporating the DSP. - When the priority numbers are assigned in ascending priority order and PNEW is to be added to any free location in
memory 608,priority logic 610 can determine whether the priority numbers already stored inmemory 608 should be incremented as follows. PNEW is decremented by one (or another value) and then compared with the existing priority numbers stored inmemory 608 bypriority logic 610. For this embodiment, the priority numbers inmemory 608 are not compared with each other, but with the decremented PNEW. Sincepriority logic 610 determines which compared number is the lowest numerical value, it can also determine the logical converse; that is, it can determine which of the stored priority numbers are greater than the decremented PNEW. Each priority number that is greater than the decremented PNEW is then incremented by the priority logic. PNEW is then added tomemory 608. -
FIG. 33 shows one example of insertingpolicy statement 1708 and PNEW intomemory 608 havingpriority numbers FIGS. 18 and 19 , except thatpriority logic 610 is used to control updating of the priority numbers stored inmemory 608. PNEW, having a value of 1, is decremented by 1 bydecrement circuit 3304 to form PSUB having a value of zero. PSUB is then provided to priority table 606 viamultiplexer 3302.Decrement circuit 3304 may be any subtraction circuit that decrements PNEW by one or another value. Multiplexer 3302 provides PSUB to priority index table 606 in response to control signal INST, which indicates that an insert function is to take place. INST may be generated by an instruction decoder (not shown) that receives an insert (write) instruction. - PSUB is compared with the stored priority numbers by
priority logic 610. In response to control signal BCT,logic 3306 asserts all match lines ML0-MLN-1 to a particular logic state to indicate that all stored policy statements match policy statement 1708 (even though no actual comparison takes place) such that the priority logic will compare all of the priority numbers inmemory 608 with PSUB.Logic 3306 may be any logic circuit including, for example, AND, OR, XOR, or inverter circuits. BCT is a control signal that indicates an insert or delete function is being performed byDSP 602. BCT may be generated by an instruction decoder (not shown). BCT may also be provided topriority logic 610 to indicate thatpriority logic 610 should compare PSUB with each entry inmemory 608 rather than comparing each entry inmemory 608 with each other as in the classification function ofFIG. 7 . -
Priority logic 610 determines that storedpriority number 0 is equal to PSUB, and leaves this stored number unchanged.Priority logic 610 also determines that storedpriority numbers Policy statement 1708 and PNEW are then loaded into a free address inarray 604 andmemory 608, respectively. The modified table entries after the insertion are shown inFIG. 34 . - Note that if the
priority number 0 is to be inserted, then decrementcircuit 3304 would generate a negative number. For one embodiment, only priority numbers greater than 0 may be used. For another embodiment, offset circuitry may be included inDSP 602 to add an offset value to PNEW before it is provided to decrementcircuit 3304 and/ormultiplexer 3302. The offset circuitry may also be used to subtract the offset value when reading priority numbers frommemory 608 such that the offset function is transparent to the user ofDSP 602. For yet another embodiment,DSP 602 can detect that PNEW is equal to zero when an insert instruction is received byDSP 602, or when an underflow condition occurs indecrement circuit 3304. PNEW can then be incremented by one and be provided todecrement circuit 3304, or PNEW can be provided directly to priority index table 606 bymultiplexer 3302. After all the priority numbers greater than zero are detected and updated bypriority logic 610, thenmemory 608 can be searched for any existing entry that already haspriority number 0. If so, this number can be incremented (i.e., by using the COUNT and UP signals, or by overwritingpriority number 0 with priority number 1), and then PNEW added tomemory 608. - A policy statement and its priority number may also be deleted or removed using
priority logic 610. For example,policy statement 1708 may be removed fromarray 604, and itspriority number 1 may be removed frommemory 608 as follows. In response to a delete instruction, the priority number (3308) to be deleted is provided to policy index table 606 via multiplexer 3302 (INST logic low), and is compared with the entries in the priority memory. For this embodiment,priority memory 608 is a CAM, or has one or more compare circuits external to the memory to perform the comparison. For an alternative embodiment, the policy statement associated with priority number 3308 may be compared with the existing entries inCAM array 604. Sincelocation 6083 matches priority number 3308, both the policy statement and the priority number are invalidated or deleted. As described above, this may be accomplished by setting a valid bit(s) to an appropriate state for that row inpriority memory 608. If the valid bit(s) is located inCAM array 604, and is inaccessible bypriority memory 608, thenpriority logic 610 can then identify the location of the matching entry (as described above), and provide the indication to IAD0-IADN-1. These signal lines can then be provided throughencoder 612 back to CAM array 604 (e.g., through a decoder) to address the appropriate location and invalidate the policy statement. Alternatively, the signal lines IAD0-IADN-1 can be provided directly toCAM array 604 to access the appropriate location for invalidation. The delete process may stop at this point and potentially leave gaps in the priority numbers. Alternatively, the process may update the remaining entries in the priority memory. When the process continues, the deleted priority number 3308 is provided to policy index table 606 viamultiplexer 3302, to determine which entries in the priority memory have a lower priority than the deleted priority number.Priority logic 610 identifiespriority numbers FIG. 33 . - The comparison and updating functions described with respect to
FIGS. 33 and 34 may also be performed bypriority logic 610 on priority numbers assigned in descending priority order. For this embodiment,decrement circuit 3304 may be replaced with an increment circuit that increments PNEW prior to supplying it to priority index table 606 for comparison as part of an insert function. - Note that if the
largest priority number 2n−1 (where n is the number of bits in a row of memory 608) is to be inserted, then the increment circuit would generate a number larger than can be handled by priority index table 606. For one embodiment, only priority numbers less than 2n−1 may be used. For another embodiment, offset circuitry may be included inDSP 602 to subtract an offset value to PNEW before it is provided to the increment circuit and/ormultiplexer 3302. The offset circuitry may also be used to add the offset value when reading priority numbers frommemory 608 such that the offset function is transparent to the user ofDSP 602. For yet another embodiment,DSP 602 can detect that PNEW is equal to 2n−1 when an insert instruction is received byDSP 602, or when an overflow condition occurs in the increment circuit. PNEW can then be decremented by one and be provided to the increment circuit, or PNEW can be provided directly to priority index table 606 bymultiplexer 3302. After all the priority numbers greater than 2n−1 are detected and updated bypriority logic 610, thenmemory 608 can be searched for any existing entry that already haspriority number 2n−1. If so, this number can be decremented (i.e., by using the COUNT and DOWN signals, or by overwritingpriority number 2n−1 withpriority number 2n−2), and then PNEW added tomemory 608. - The embodiments described above in
FIGS. 33 and 34 include the ability ofpriority logic 610 to increment and/or decrement the priority numbers stored inpriority memory 608.Priority memory 608 may be formed as counters that store the priority numbers.FIG. 35 shows one embodiment of a row ofpriority memory 608 in which thepriority logic 610 includes a separate priority logic element for each bit of a priority number. For this embodiment, the priority logic includesrow 1201 0 of priority logic elements fromFIG. 12A , coupled to receive priority number bits stored incounter 2306 0 fromFIG. 23 . PSUB bits PSUBn-1-PSUB0 are provided on signal lines 1208 n-1-1208 0, respectively, for comparison by compare circuits 1206 n-1,0-1206 0,0 with the priority bits stored in memory elements 2314 n-1,0-2314 0,0. When BCT is asserted to a high logic state,logic 3306 0 drives matchline segment 1210 n-1,0 to a high logic state, and PSUB is provided onsignal lines 1208. BCT is also provided to each of comparecircuits 1206 such that the circuits do not provide their compare results back tosignal lines 1208. Thus, only PSUB will be supplied onsignal lines 1208. For an alternative embodiment, a separate set of signal lines may be used for the updating functions and the classification function. When BCT is asserted to a low logic state, ML0 is provided to matchlines segment 1210 n-1,0 and the classification function ofFIG. 7 may be performed. - For this embodiment, the logical state of IAD0 indicates the comparison result between PSUB and the priority number stored in
counter 23060. If IAD0 indicates that PSUB has a more significant priority than the priority number stored incounter 2306 0, then counter 2306 0 is incremented. If, however, IAD0 indicates that PSUB is equal to or has a less significant priority number than the priority number incounter 2306 0, then counter 2306 0 is not incremented. - The operation of the embodiment of
FIG. 35 to update priority numbers in response to an insert instruction can be illustrated with the example shown inFIG. 36 . In this example, a 3×3 matrix of rows and columns stores three priority numbers. For other embodiments, any numbers of rows and columns can be used.Row 0 stores priority number 010 having the decimal equivalent of thenumber 2,row 1 stores priority number 001 having the decimal equivalent of thenumber 1, androw 2 stores priority number 000 having the decimal equivalent of thenumber 0. For this example, the priority numbers are assigned in ascending priority order such that 000 is the most significant priority number, 001 is the next most significant priority number, and 010 is the least significant priority number. - When a new policy statement having new a priority number PNEW 001 is to be loaded into
CAM array 604, BCT is asserted to a high logic state, the control signals onmatch line segments Priority logic 1201 2 determines that 000 is equal to PSUB;priority logic 1201 1 determines that 001 is greater than PSUB; andpriority logic 1201 0 determines that 010 is greater than PSUB.Priority logic 1201 2 leaves unchanged the priority number stored incounter 2306 2.Circuits CAM array 604 andpriority memory 608, respectively. - The comparisons of the most significant bit PSUB2 with the most significant bits stored in counters 2306 2-2306 0 are resolved first. When an individual compare circuit determines that its corresponding memory storage element stores the same logic state as the PSUB bit, the compare circuit leaves the control signal of the next match line segment unaffected such that it has the same logical state of the previous match line segment. Since each of
memory storage elements match line segments - The comparison of the next most significant bit PSUB1 with priority number bits stored in
memory elements memory elements match line segments circuit 1206 1,0, however, determines that the priority number stored incounter 2306 0 is greater than PSUB because: (1)memory element 2314 1,0 stores a logic one; (2)match line segment 1210 1,0 is a high logic state; and (3) PSUB1 is a logic zero onsignal line 1208 1. When comparecircuit 1206 1,0 makes this determination, it dischargesmatch line segment 1210 0,0 to a low logic state. When a match line segment is discharged, all subsequent match line segments in that row are discharged to a low logic state such that IAD for that row is discharged to a low logic state. For this embodiment, when IAD is a low logic state, the counter associated with that row will be incremented when UP is asserted. Thus, IAD0 will cause counter 2306 0 to be incremented when UP is asserted. - Lastly, the comparison of the least significant bit PSUB0 with priority number bits stored in
memory elements priority logic 1201 0 has already determined that 010 is greater than 000 and thus IAD0 remains asserted to a low logic state. Since the control signal onmatch line segment 1210 0,2 is in a high logic state, comparecircuit 1206 0,2 compares the logic zero stored inmemory element 2314 0,2 with the logic zero of PSUB0, and determines that PSUB is the same number as that stored incounter 2306 2. In response, IAD2 remains in a high logic state such thatcounter 2306 2 is not updated. Comparecircuit 1206 0,1, however, determines that PSUB0 is less than the logic one stored inmemory element 2314 0,1. Since the control signal is high onmatch line segment 1210 0,1, comparecircuit 1206 0,1 discharges IAD1 such that the priority number 001 stored incounter 2306 1 will be incremented. After a sufficient amount of time has passed such that all of the bit comparisons have had time to resolve, the UP signal can be asserted such that the priority numbers incounters memory 608. - Any compare circuits may be used for compare
circuits 1206 to implement the process illustrated above. For example, the compare circuits illustrated above that implement Table 1 can be used with additional control for the BCT control signal. One embodiment of the modified circuit ofFIG. 14 is shown inFIG. 37 . InFIG. 37 ,transistor 1416 is included in series withtransistors signal line 1208 n-1 and ground. The gate oftransistor 1416 is controlled by BCT viainverter 1420. A similar modification can be made to any of the other embodiments of the comparecircuits 2314. ORgate 1418 is one embodiment oflogic 3302. - While
FIG. 36 shows an embodiment of inserting a new priority number intomemory 608, the priority numbers can be decremented for a delete function by asserting the DOWN signal after all of the IAD signals are resolved. - The embodiment of
FIG. 35 can also insert and delete priority numbers assigned in descending priority order.FIG. 38 shows such an example with of a 3×3 matrix in which row 0 stores priority number 111 having the decimal equivalent of the number 7,row 1 stores priority number 110 having the decimal equivalent of the number 6, androw 2 stores priority number 101 having the decimal equivalent of thenumber 5. Thus, 111 is the most significant priority number, 110 is the next most significant priority number, and 101 is the least significant priority number. - When a new policy statement having a new priority number PNEW 110 is to be loaded into
CAM array 604, BCT is asserted to a high logic state, control signals onmatch line segments Priority logic 1201 2 determines that 111 is equal to PSUB;priority logic 1201 1 determines that 110 is less than PSUB; andpriority logic 1201 0 determines that 101 is less than PSUB.Priority logic 1201 2 leaves unchanged the priority number stored incounter 2306 2.Circuits CAM array 604 andpriority memory 608, respectively. - The comparisons of the most significant bit PSUB2 with the most significant bits stored in counters 2306 2-2306 0 are resolved first. When an individual compare circuit determines that its corresponding memory storage element stores the same logic state as the PSUB bit, the compare circuit leaves the control signal of the next match line segment unaffected such that it has the same logical state of the previous match line segment. Since each of
memory storage elements match line segments - The comparison of the next most significant bit PSUB1 with priority number bits stored in
memory elements memory elements match line segments circuit 1206 1,0, however, determines that the priority number stored incounter 2306 0 is less than PSUB because: (1)memory element 2314 1,0 stores a logic zero; (2)match line segment 1210 1,0 is a high logic state; and (3) PSUB1 is a logic one onsignal line 1208 1. When comparecircuit 1206 1,0 makes this determination, it dischargesmatch line segment 1210 0,0 to a low logic state. When a match line segment is discharged to a low logic state, all subsequent match line segments in that row are discharged such that IAD for that row is discharged to a low logic state. When IAD is a low logic state, the counter associated with that row will be decremented when DOWN is asserted. Thus, IAD0 will cause counter 2306 0 to be decremented when DOWN is asserted. - Lastly, the comparison of the least significant bit PSUB0 with priority number bits stored in
memory elements priority logic 1201 0 has already determined that 101 is less than 111 and thus IAD0 remains discharged to a low logic state. Since the control signal onmatch line segment 1210 0,2 is in a high logic state, comparecircuit 1206 0,1 compares the logic one stored inmemory element 2314 0,2 with the logic one of PSUB0, and determines that PSUB is the same number as that stored incounter 2306 2. In response, IAD2 remains in a high logic state such thatcounter 2306 2 is not updated. Comparecircuit 1206 0,1, however, determines that PSUB0 is greater than the logic zero stored inmemory element 2314 0,1. Since the control signal is asserted onmatch line segment 1210 0,1, comparecircuit 1206 0,1 discharges IAD1 such that the priority number 110 stored incounter 2306 1 will be decremented. After a sufficient amount of time has passed such that all of the bit comparisons have had time to resolve, the DOWN signal can be asserted such that the priority numbers incounters memory 608. - Any compare circuits may be used for compare
circuits 1206 to implement the process illustrated inFIG. 38 . For example, the compare circuits illustrated above that implement Table 3 can be used with additional control for the BCT control signal as illustrated inFIG. 37 . WhileFIG. 38 shows an embodiment of inserting a new priority number intomemory 608, the priority numbers can be incremented for a delete function by asserting the UP signal after all of the IAD signals are resolved. -
DSP 602 can perform the updating functions described above with respect toFIGS. 17-38 asynchronously or synchronously. WhenDSP 602 operates synchronously, it receives a clock signal that may be used to clock in the policy search key and an instruction that causes the updating functions to be performed byDSP 602.DSP 602 may implement the updating functions in one or more clock cycles. - Depth Cascading DSP Devices
- As described above,
DSP 602 stores policy statements inCAM array 604 and identifies the highest priority matching policy statement without having to presort or prearrange the policy statements in the CAM array.DSP 602 may also be included in a system that has multiple DSPs connected in a depth cascade configuration that expands the number of memory locations inCAM array 604 andpriority memory 608 to accommodate more policy statements and their associated priority numbers. -
FIG. 39 shows aCAM system 3900 that includes threeDSPs FIG. 7 , is performed by each of the DSPs insystem 3900, the DSP that stores the highest priority matching statement forsystem 3900, as a whole, may be determined as will be described in more detail below. The DSP that includes the highest priority matching statement forsystem 3900 may be determined regardless of the number of CAM locations filled or occupied in each CAM array of each DSP (i.e., regardless of whether one or more CAM arrays of DSPs 3902-3906 is full, empty, or partially full). - Any number of DSPs may be depth cascaded as shown in
FIG. 39 . The total memory size ofsystem 3900 is larger than the individual memory sizes of each of the CAM arrays in each DSP 3902-3906. For example, if each CAM array is a 16 k×144 bit CAM array, thensystem 3900 may operate as a DSP that includes a 48 k×144 bit CAM array. Any other size of CAM arrays may be used. Also, CAM arrays of different widths may be depth cascaded together. Additionally,system 3900 may be formed from stand-alone DSPs, or may be formed from integrated circuits on a common substrate. - Each DSP receives in parallel a clock signal CLK, policy statements and priority numbers on data bus DBUS, and instructions on instruction bus IBUS. For alternative embodiments, the policy statements and priority numbers may be provided on different buses. Other input signals may also be simultaneously provided to each of DSPs 3902-3906 including word enable signals, reset signals, chip enable signals, and the like. DSPs 3902-3906 also output their external address signal to a common HPM bus. When performing the classification function of
FIG. 7 , the DSP that has the highest priority matching policy statement will take control of the HPM bus. Each DSP 3902-3906 may include any of the embodiments ofDSP 602 described above. - DSPs 3902-3906 may each include a cascade down input /CDI, a cascade down output /CDO, a cascade up input /CUI, a cascade up output /CUO, priority number down inputs PNDI, priority number down outputs PNDO, priority number up inputs PNUI, and priority number up outputs PNUO. The PNDO and PNUO outputs provide the most significant priority number PNUM generated by the priority logic in each priority index table in each of DSP 3902-3906. The cascade up and down outputs provide signals that indicate when PNUM is valid on the PNDO and PNUO outputs. For other embodiments, the cascade inputs and outputs may be omitted and CLK or match flag up and down signals may be used to validate the PNUM outputs. For an alternative embodiment, match flag up and down signals generated by flag logic coupled to
CAM array 604 may be used to validate the PNUM outputs. Additional outputs or inputs such as full flag up and down pins may also be included. - Each CAM DSP generally has its /CDO output coupled to the /CDI input of the next DSP, its /CUO output coupled to the /CUI of the previous DSP, its /CDI input coupled to the /CDO of the previous DSP, its /CUI input coupled to the /CUO of the next DSP, its PNDO outputs coupled to the PNDI inputs of the next DSP, its PNDI inputs coupled to the PNDO outputs of the previous DSP, its PNUI inputs coupled to the PNUO outputs of the next DSP, and its PNUO outputs coupled to the PNUI inputs of the previous DSP.
-
DSP 3902 may be designated as the DSP that has the lowest logical system addresses by connecting its /CDI input to ground, and its PNDI inputs to VDD. For another embodiment, the PNDI inputs may be connected to ground and/or the /CDI inputs ofDSP 3902 may be connected to VDD.DSP 3902 has the lowest logical or numerical addresses of, for example, zero toN− 1.DSP 3904 has logical addresses N to M−1, andDSP 3906 has logical addresses M to W−1, wheresystem 3900 has a total of W ternary CAM words available for storing policy statements.DSP 3906 may be designated as the last DSP in the cascade by connecting its PNUI inputs to VDD, and its /CUI input to ground. For another embodiment, the PNUI inputs ofDSP 3906 may be connected to ground, and/or the /CUI input may be connect to VDD. - The operation of
system 3900 for implementing the classification or filtering function ofFIG. 7 is as follows. Initially, one or more of DSPs 3902-3906 is loaded with policy statements and corresponding priority numbers. In response to an instruction provided on the IBUS and the policy search key provided on the DBUS, each DSP 3902-3906 compares the policy search key with the policy statements stored in its CAM array 604 (step 702). The priority logic in each priority index table identifies its most significant priority number PNUM associated with one of its matching policy statements. Each DSP also determines the address of its PNUM in its priority memory 608 (steps 706 and 708). Each DSP then compares the priority numbers provided on its PNDI and PNUI pins with its own PNUM to determine whether it has a more significant priority number. If a DSP has a more significant or equal priority number than that provided on its PNUI pins, then the DSP outputs its own PNUM to its PNUO pins. If, however, a DSP has a lower priority PNUM than those provided on the PNUI pins, then the DSP outputs the priority number from its PNUI pins to its PNUO pins. Similarly, if a DSP has a more significant priority number than that provided on its PNDI pins, then the DSP outputs its own PNUM to its PNDO pins. If, however, a DSP has an equal or lower priority PNUM than those provided on the PNDI pins, then the DSP outputs the priority number from its PNDI pins to its PNDO pins. - By simultaneously providing and comparing priority numbers both down and up through
system 3900, each DSP will ultimately determine whether it has the most significant priority number in the entire system. When a device has the most significant priority number for the system, it may take control of the HPM bus and output address information to access a location inroute memory 614. For this embodiment,route memory 614 is as deep as the number of CAM memory location insystem 3900. - Each of DSPs 3902-3906 can also update their own policy statement tables and priority memories as needed in response to an insertion or deletion instruction provided to
system 3900. A new policy statement and priority number, for example, may be loaded into the DSP that has the next free address ofsystem 3900 as determined by full flag signals (not shown). -
FIG. 40 shows DSP 4000 that is one embodiment of DSPs 3902-3906.DSP 4000 is any of the embodiments ofDSP 602 that further includescascade logic 4004,output buffer 4002,instruction decoder 4006, read and write (R/W)circuits flag logic 4016.DSP 4000 may also include registers to store the policy statements and priority numbers prior to use byCAM array 604 and priority index table 606.Cascade logic 4004 is coupled to the /CDI, /CUI, PNDI, and PNUI inputs, and to the /CDO, /CUO, PNDO, and PNUO outputs. For another embodiment, the cascade inputs and outputs (i.e., /CDI, /CDO, /CUI, and /CUO) may be omitted and thus not coupled to cascadelogic 4004.Cascade logic 4004 may also receive and output match flag and full flag cascade signals. -
Cascade logic 4004 receives a search signal SCH online 4012 frominstruction decoder 4006 indicating that the classification or filtering operation will be performed byDSP 4000.Cascade logic 4004 may also receive a match flag signal /MF fromflag logic 4016 indicating whetherCAM array 604 has an address that matches a policy search key.Cascade logic 4004 compares the priority number on its PNDI inputs with its own most significant priority number PNUM from priority index table 606, and outputs the more significant number from among these two priority numbers to its PNDO outputs. Similarly,cascade logic 4004 compares the priority number on its PNUI inputs with its own PNUM, and outputs the more significant number from among these two priority numbers to its PNUO outputs. IfDSP 4000 has the most significant priority number for theentire system 3900, then it asserts its output buffer control signal OBCNTL onsignal line 4014 to enableoutput buffer 4002 to take control of the HPM bus. -
FIG. 41 is a block diagram ofcascade logic 4100 that is one embodiment ofcascade logic 4004 ofFIG. 40 .Cascade logic 4100 includes cascade downlogic 4104 and cascade uplogic 4102. Cascade downlogic 4104 generates signals for the /CDO and PNDO outputs in response to /CDI, PNDI inputs, and SCH. Cascade uplogic 4102 generates signals on the /CUO and PNUO outputs in response to /CUI, PNUI inputs, and SCH. Other embodiments ofcascade logic 4004 may be used. For example, match flag signals may be used. - Cascade down
logic 4104 includessignal generator 4106 and comparecircuit 4110.Signal generator 4106 receives CLK and SCH frominstruction decoder 4006. SCH indicates that the classification of filtering function is to be performed byDSP 4000.Signal generator 4106 asserts /CDO to a low logic state after /CDI is asserted to a low logic state and when the priority number on the PNDO outputs is valid.Signal generator 4106 may be any logic circuit that performs this function. For an alternative embodiment, SCH may be omitted andsignal generator 4106 may generate /CDO in response to CLK only. Comparecircuit 4110 compares the internal most significant priority number PNUM with the priority number from the PNDI inputs. If the priority number on the PNDI inputs is equal to or more significant than PNUM, then comparecircuit 4110 outputs the priority number from its PNDI inputs to its PNDO outputs and assertssignal line 4116 to a low logic state. If, however, PNUM is more significant than the priority number on the PNDI inputs, then comparecircuit 4110 outputs PNUM to its PNDO outputs and assertsline 4116 to a high logic state. - Cascade up
logic 4102 includessignal generator 4108 and comparecircuit 4112.Signal generator 4108 receives CLK and SCH frominstruction decoder 4006.Signal generator 4108 asserts /CUO to a low logic state after /CUI is asserted to a low logic state and when the priority number on the PNUO outputs is valid.Signal generator 4108 may be any logic circuit that performs this function. For an alternative embodiment, SCH may be omitted andsignal generator 4108 may generate /CUO in response to CLK only. Comparecircuit 4112 compares PNUM with the priority number from the PNUI inputs. If the priority number on its PNUI inputs is more significant than PNUM, then comparecircuit 4112 outputs the priority number from its PNUI inputs to its PNUO outputs and assertssignal line 4118 to a low logic state. If, however, PNUM is equal to or more significant than the priority number on its PNUI inputs, then comparecircuit 4112 outputs PNUM to the PNUO outputs and assertsline 4118 to a high logic state. Whensignal lines gate 4114 asserts OBCNTL online 4014 to a high state to enableoutput buffer 4002 to take control of the HPM bus. - For an alternative embodiment in which the cascade inputs (/CDI and /CUI) and cascade outputs (/CDO and /CUO) are omitted,
signal generators - Classless Inter Domain Routing (CIDR)
-
DSP 602 can also be used to process Internet Protocol (IP) packets that use the Classless Inter Domain Routing (CIDR) scheme. With CIDR, an IP address has a generalized network prefix of a particular number bits of 32-bit IPv4 or a 128-bit IPv6 address. The network prefix or mask indicates the number of left-most contiguous bits in the IP address that are used to filter an IP address in a routing table. That is, the network prefix indicates the number of higher-order or left-most contiguous bits in the IP address that participate in an address comparison with the routing table. - Conventional ternary CAM devices such as
CAM 300 ofFIG. 3 can store the IP addresses inrows 305, and their corresponding prefixes inrows 310. Routing information associated with a particular IP address is loaded into a corresponding address location inroute memory 308. Due to the operation ofpriority encoder 306, IP addresses are generally presorted or prearranged prior to entry into a CAM device such that the IP address with the longest network prefix is located in the lowest logical address of the CAM array, and the IP address with the shortest network prefix is located in the highest logical address of the CAM array. When the IP addresses are presorted, a search on the CAM array for a particular IP address will identify the IP address that has the longest corresponding prefix, that is, will identify the best match. - A considerable amount of time is generally required to prearrange all of the CIDR address entries prior to loading the entries into a CAM device. Additionally, a considerable amount of time and overhead is also generally required to maintain the order of the routing table when entries are deleted or overwritten, or when new entries are to be added. Other architectures have been proposed that increase the size of the CAM array by adding additional logic in the CAM array itself and another match coupled to the rows of mask cells.
-
DSP 602 ofFIG. 6 can be used to process IP addresses without adding additional logic or signal lines toternary CAM array 604. IP addresses can be loaded intoCAM cell rows 605, and the corresponding decoded prefix data can be loaded intomask rows 607. Decoding logic can be provided inDSP 602 to decode the prefix number. Additionally, the prefix data is encoded into a binary number and stored in corresponding locations inpriority memory 608. Encoding logic can be provided inDSP 602 to encode the prefix number into a binary (or other code) number. - When a search is performed for the IP address with the longest prefix, all matching locations in
CAM array 604 will assert their corresponding match lines ML0-MLN-1.Priority logic 610 then compares, with each other, the encoded prefix numbers associated with the matching IP address.Priority logic 610 identifies the most significant encoded prefix number (i.e., the highest prefix number), and identifies its location inpriority memory 608 to IAD0-IADN-1. The encoded most significant prefix number may also be output fromDSP 600.Encoder 612 then encodes the identified location into an address for output toHPM bus 616. The corresponding route information can then be accessed inroute memory 614. As in the previous embodiments described above,route memory 614 may also be included withinDSP 602. - For another embodiment, there may more than one identical most significant priority numbers identified by
priority logic 610. For this embodiment,encoder 612 may be a conventional priority encoder that determines which address to output based on a predetermined priority (i.e., based on logical address locations). -
DSP 602 can process the CIDR based IP addresses without preloading the IP addresses in the CAM array in a predetermined order. Additionally, new IP address may be added at the next free address or any other designated address inCAM array 604 without reordering or reloading the CAM array. This can reduce the hardware and/or software needed for table management of the CAM array, and can increase the performance of a router incorporating the DSP. - The operation of
DSP 602 for processing CIDR based IP addresses can be illustrated by the example ofFIG. 42 . InFIG. 42 ,ternary CAM array 604 has IP address 168.0.0.0/8 stored atlocation 604 0, 168.69.0.0/16 stored atlocation 604 1, and 168.69.62.0/24 stored atlocation 604 2. For this embodiment, each IP address is stored inarray 604 as four eight-bit binary numbers. Also for this embodiment, when the decoded prefix data is a logic zero it does not mask the corresponding bits of the IP address.Priority memory 608 stores theprefixes locations locations route memory 614. - A search key of 168.69.43.100 is provided to
CAM array 604, and the IP address with the best match (i.e., the longest prefix data) is determined as follows. When the search key is compare with the IP addresses, 168.69.0.0/16 and 168.0.0.0/8 are both identified as matches and ML0 and ML1 asserted. Between these two IP addresses, 168.69.0.0/16 is the best match as it has a longer prefix.Priority logic 610 compares theprefixes locations location 608 1 by asserting IAD1.Encoder 612 then encodes IAD0-IADN-1 and generates an address of 1 onHPM bus 616 to access route information RI1 inroute memory 614. - Any of the embodiments of
DSP 602 described above can be used to implementpriority logic 610 to process CIDR based IP addresses and their prefix numbers. Additionally, multiple DSPs can be depth cascaded as described inFIGS. 39-41 to implement a system that provides a sufficient number of CAM array locations for storing IP addresses. - Various aspects of the subject-matter described herein are set forth for example and without limitation in the following numbered clauses:
- 1. A digital signal processor comprising:
- a policy statement table for storing a plurality of policy statements; and
- a priority index table for storing a plurality of priority numbers, each priority number associated with a corresponding policy statement and indicating the priority of the corresponding policy statement relative to the other policy statements.
- 2. The digital signal processor of
clause 1, wherein the policy statement table comprises a content addressable memory (CAM). - 3. The digital signal processor of
clause 3, wherein the CAM comprises a ternary CAM. - 4. The digital signal processor of
clause 1, wherein the priority index table comprises priority logic coupled to the policy statement table. - 5. The digital signal processor of clause 4, wherein the priority logic to provide to a plurality of signal lines an indication of a location of the most significant priority number in the priority index table.
- 6. The digital signal processor of
clause 5, further comprising a memory array coupled to the plurality of signal lines and for storing routing information for the policy statements, the indication to select routing information from the memory array for one of the policy statements. - 7. The digital signal processor of
clause 5, further comprising an encoder having inputs coupled to the plurality of signal lines to receive the indication. - 8. The digital signal processor of clause 4, wherein the most significant priority number is the priority number with the smallest numerical value.
- 9. The digital signal processor of clause 4, wherein the most significant priority number is the priority number with the largest numerical value.
- 10. The digital signal processor of clause 4, wherein the priority logic provides the most significant priority number to a plurality of priority signal lines.
- 11. The digital signal processor of
clause 10, wherein the priority logic comprises: -
- a plurality of compare circuits each coupled to one of the plurality of priority signal lines and the policy statement table; and
- a plurality of isolation circuits each coupled to one of the plurality of priority signal lines and one of the plurality of compare circuits.
- 12. The digital signal processor of
clause 11, wherein the priority logic further comprises means for determining whether to compare at least two priority numbers already stored in the priority index table, or to compare the priority numbers stored in the priority index table with a priority number not stored in the priority index table. - 13. The digital signal processor of
clause 1, further comprising an instruction decoder coupled to the policy statement table and the priority index table. - 14. The digital signal processor of
clause 1, further comprising decrement logic coupled to the policy index table. - 15. The digital signal processor of
clause 1, further comprising increment logic coupled to the policy index table. - 16. The digital signal processor of
clause 1, further comprising flag logic coupled to the policy statement table, the flag logic to generate a flag signal indicating when one policy statements matches a policy search key. - 17. The digital signal processor of
clause 1, further comprising cascade logic coupled to the priority index table and having inputs to receive a priority number from another digital signal processor, the cascade logic for outputting from the digital signal processor a priority number from the priority index table or the priority number from the another digital signal processor. - 18. A digital signal processor comprising:
-
- a content addressable memory (CAM) array having a plurality of rows of CAM cells; and
- a first array of storage elements coupled to the CAM array, each row of storage elements for storing a number corresponding to a data word stored in one of the rows of the CAM cells.
- 19. The digital signal processor of clause 18, wherein the rows of CAM cells are for storing policy statements, and the numbers comprise priority numbers of the policy statements, a priority number indicating the priority of a corresponding policy statement relative to the other policy statements.
- 20. The digital signal processor of clause 18, wherein the rows of CAM cells are for storing Internet Protocol (IP) addresses, and the numbers comprise prefix mask data for the IP addresses.
- 21. The digital signal processor of clause 18, further comprising priority logic coupled to the first may of storage elements, the priority logic to provide to a plurality of signal lines an indication of a location of a particular number in the first array of storage elements.
- 22. The digital signal processor of clause 21, wherein the plurality of signal lines is coupled to the CAM array.
- 23. The digital signal processor of clause 21, further comprising a second array of storage elements, each row of the second may coupled to one of the plurality of signal lines to receive the indication, the indication to select one of the rows of the second array.
- 24. The digital signal processor of clause 21, further comprising an encoder having inputs coupled to the plurality of signal lines to receive the indication, and having a plurality of outputs to provide an encoded address of the location of the number in the first array of storage elements.
- 25. The digital signal processor of
clause 24, wherein the encoded address corresponds to an address in the CAM may of the data word that corresponds with the number stored at the indicated location in the first array of storage elements. - 26. The digital signal processor of
clause 24, further comprising a decoder coupled to the CAM array, and wherein the plurality of outputs of the encoder are coupled to the decoder. - 27. The digital signal processor of
clause 24, further comprising a second may of storage elements, each row of the second array coupled to one of the plurality of outputs of the encoder. - 28. The digital signal processor of clause 19, further comprising a plurality of priority signal lines each coupled to one of the storage elements in each row of the first array.
- 29. The digital signal processor of clause 28, wherein the priority logic comprises:
-
- a first plurality of compare circuits, each compare circuit coupled to one of the storage elements in the first array of storage elements, and each compare circuit having a first input coupled to a storage element, a second input coupled to a match line, and an input/output line coupled to one of the plurality of priority signal lines; and
- an isolation circuit, the isolation circuit associated with the most significant bit of the each number having an input coupled to the match line, and an output, each additional isolation circuit associated with the other bits of each number having an output, and an input coupled to the output of the previous higher order bit.
- 30. The digital signal processor of clause 29, wherein each row of storage elements in the first array is configured to form a counter.
- 31. The digital signal processor of clause 30, further comprising a second plurality of compare circuits each coupled to one of the counters, and each having a first plurality of inputs to receive a first number stored in the respective counter, a second plurality of inputs each coupled to one of the plurality of priority signal lines to receive a second number provided thereon, and an output coupled to the counter, each of the second compare circuits providing a control signal on its output to indicate whether to update the first number stored in the respective counter.
- 32. The digital signal processor of clause 31, wherein each of the second plurality of compare circuits determines when the first number is greater than or equal to the second number.
- 33. The digital signal processor of clause 31, wherein each of the second plurality of compare circuits determines when the first number is less than or equal to the second number.
- 34. The digital signal processor of clause 28, wherein each row of storage elements in the first array is configured to form a counter.
- 35. The digital signal processor of clause 34, further comprising a plurality of compare circuits each coupled to one of the counters, and each having a first plurality of inputs to receive a first number stored in the respective counter, a second plurality of inputs each coupled to receive a second number, and an output coupled to the counter, each of the second compare circuits providing a control signal on its output to indicate whether to update the first number stored in the respective counter.
- 36. The digital signal processor of
clause 35, wherein each of the plurality of compare circuits determines when the first number is greater than or equal to the second number. - 37. The digital signal processor of
clause 35, wherein each of the second plurality of compare circuits determines when the first number is less than or equal to the second number. - 38. The digital signal processor of clause 18, further comprising priority logic having inputs coupled to the first array of storage elements, and outputs to provide the most significant number from a selection of numbers in the first array.
- 39. The digital signal processor of clause 38, wherein the first array comprises a second CAM array having a plurality of match lines, and wherein the outputs of the priority logic are coupled to the second CAM array.
- 40. The digital signal processor of clause 39, wherein the second CAM array compares the most significant number with the numbers stored in its array and asserts its plurality of match lines in response to the comparison.
- 41. A digital signal processor comprising:
-
- a content addressable memory (CAM) array for storing a plurality of data words and determining that a search key matches more than one of the data words;
- first storage means for storing a plurality of numbers, each number corresponding to a data word in the CAM array; and
- means for determining a location in the second storage means of a most significant number associated with one of the data words that matches the search key.
- 42. The digital signal processor of clause 41, further comprising means for generating an address of the location in the first storage means.
- 43. The digital signal processor of clause 42, further comprising second storage means for storing another plurality of data words, one of which is accessed in response to the address of the location in the first storage means.
- 44. The digital signal processor of clause 41, further comprising means for accessing the data word in the CAM array that corresponds to the most significant number.
- 45. The digital signal processor of clause 41, further comprising means for determining the most significant number.
- 46. The digital signal processor of clause 45, wherein the most significant number is the largest numerical number.
- 47. The digital signal processor of clause 45, wherein the most significant number is the lowest numerical number.
- 48. The digital signal processor of clause 41, further comprising means for determining that at least one of the numbers stored in the first storage means is greater than or equal to an external number not stored in the first storage means.
- 49. The digital signal processor of clause 48, further comprising means for updating the at least one number that is greater than or equal to the external number.
- 50. The digital signal processor of clause 49, further comprising means for writing the external number to the first storage means.
- 51. The digital signal processor of clause 41, further comprising means for determining that at least one of the numbers stored in the first storage means is less than or equal to an external number not stored in the first storage means.
- 52. The digital signal processor of clause 51, further comprising means for updating the at least one number that is less than or equal to the external number.
- 53. The digital signal processor of clause 52, further comprising means for writing the external number to the first storage means.
- 54. The digital signal processor of clause 41, wherein the CAM array comprises a ternary CAM array having data storage cells for storing the plurality of data words, and further having mask storage cells for storing mask data on a bit-for-bit basis for the data words.
- 55. The digital signal processor of clause 54, wherein the data words comprise policy statements, and the numbers comprise priority numbers for the policy statements.
- 56. The digital signal processor of clause 54, wherein the data words comprise Internet Protocol (IP) addresses, and the numbers comprise prefix mask data.
- 57. A method of operating a digital signal processor, comprising:
-
- determining that a search key matches more than one of a plurality of data words stored in a first memory; and
- determining a location in a second memory of a most significant number associated with the data words that match the search key.
- 58. The method of clause 57, further comprising generating an address of the location in the second memory.
- 59. The method of clause 58, further comprising accessing a location in a third memory in response to the address of the location in the second memory.
- 60. The method of clause 57, further comprising accessing a location in a third memory in response to determining the location in the second memory.
- 61. The method of clause 60, wherein the data processor is incorporated into a router, the data words comprise policy statements, and the most significant number is the most significant priority number, wherein each policy statement has a priority number that indicates the priority of the corresponding policy statement relative to other policy statements in a given policy.
- 62. The method of clause 57, further comprising determining the most significant number.
- 63. The method of 62, wherein the most significant number is the number with the largest numerical value.
- 64. The method of
clause 62, wherein the most significant number is the number with the highest numerical value. - 65. A method of adding a new policy statement to a plurality of policy statements stored in a first memory, comprising:
-
- comparing a new priority number associated with the new policy statement to a plurality of priority numbers stored in a second memory, the plurality of priority numbers each associated with a respective one of the plurality of policy statements stored in the first memory;
- determining that the new priority number is more significant than one of the plurality of priority numbers stored in the second memory;
- updating the determined priority number in the second memory without changing its physical location in the second memory;
- writing the new priority number to an available location in the second memory; and
- writing the new policy statement to an available location in the first memory.
- 66. The method of clause 65, wherein the new priority number is more significant than the determined priority number when the new priority number is greater than or equal to the determined priority number.
- 67. The method of clause 65, wherein the new priority number is more significant than the determined priority number when the new priority number is less than or equal to the determined priority number.
- 68. A method of adding a new policy statement to a plurality of policy statements stored in a first memory, comprising:
-
- altering a new priority number associated with the new policy statement;
- comparing the altered new priority number to a plurality of priority numbers stored in a second memory, the plurality of priority numbers each associated with a respective one of the plurality of policy statements stored in the first memory;
- determining that the altered new priority number is more significant than one of the plurality of priority numbers stored in the second memory;
- updating the determined priority number in the second memory without changing its physical location in the second memory;
- writing the unaltered new priority number to an available location in the second memory; and
- writing the new policy statement to an available location in the first memory.
- altering a new priority number associated with the new policy statement;
- 69. The method of clause 68, wherein the altered new priority number is more significant than the determined priority number when the altered new priority number is greater than the determined priority number.
- 70. The method of clause 68, wherein the altered new priority number is more significant than the determined priority number when the altered new priority number is less than the determined priority number.
- 71. The method of clause 68, wherein altering the new priority number comprises decrementing the new priority number.
- 72. The method of clause 68, wherein altering the new priority number comprises incrementing the new priority number.
- 73. A method of deleting a policy statement from a plurality of policy statements stored in a first memory, comprising:
-
- comparing a priority number associated with the policy statement to a plurality of priority numbers stored in a second memory, the plurality of priority numbers each associated with a respective one of the plurality of policy statements stored in the first memory;
- determining that the priority number is equal to one of the plurality of priority numbers;
- providing an indication of the location of the matched priority number in the second memory to the first memory to access the policy statement; and deleting the policy statement from the first memory.
- 74. The method of clause 73, further comprising:
-
- comparing the priority number with the plurality of priority numbers in the second memory;
- determining that the priority number is less than one of the plurality of priority numbers;
- updating the determined priority number without changing its physical location in the second memory, and without changing the physical location in the first memory of the policy statement associated with the determined priority number.
- 75. The method of clause 73, further comprising:
-
- comparing the priority number with the plurality of priority numbers in the second memory;
- determining that the priority number is greater than one of the plurality of priority numbers;
- updating the determined priority number without changing its physical location in the second memory, and without changing the physical location in the first memory of the policy statement associated with the determined priority number.
- 76. A system comprising:
-
- a first digital signal processor comprising:
- a first content addressable memory (CAM) array for storing a first plurality of data word and determining that a search key matches more than one of the first data words;
- first storage means for storing a first plurality of numbers, each number corresponding to a data word in the first CAM array;
- means for determining a first most significant number associated with one of the first data words that matches the search key; and
- outputs for providing the first most significant number; and
- a second digital signal processor comprising:
- a second CAM array for storing a second plurality of data words and determining that the search key matches more than one of the second data words;
- second storage means for storing a second plurality of numbers, each number corresponding to a data word in the second CAM array;
- means for determining a second most significant number associated with one of the second data words that matches the search key;
- first inputs coupled to the outputs of the first digital signal processor; and
- cascade logic coupled to the first inputs and the second storage means, the cascade logic to compare the first most significant number with the second most significant number.
- 77. The system of clause 76, wherein the cascade logic outputs the more significant number between the first and second most significant numbers.
- 78. The system of clause 77, wherein the more significant number is the number that has the largest numerical value.
- 79. The digital signal processor of clause 77, wherein the more significant number is the number that has the smallest numerical value.
- 80. The system of clause 76, further comprising a third digital signal processor comprising:
-
- a third content addressable memory (CAM) array for storing a third plurality of data word and determining that the search key matches more than one of the third data words;
- third storage means for storing a third plurality of numbers, each number corresponding to a data word in the third CAM array;
- means for determining a third most significant number associated with one of the third data words that matches the search key; and
- outputs for providing the third most significant number; and wherein the second digital signal processor further comprises second inputs coupled to the outputs of the third digital signal processor, and wherein the cascade logic is coupled to the second inputs to compare the second most significant number with the third most significant number.
- 81. The system of clause 80, wherein the cascade logic outputs the more significant number between the first and second most significant numbers.
- 82. The system of clause 81, wherein the more significant number is the number that has the largest numerical value.
- 83. The digital signal processor of clause 82, wherein the more significant number is the number that has the smallest numerical value.
- 84. The digital signal processor of clause 76, wherein the data words comprise policy statements, and the numbers comprise priority numbers for the policy statements.
- 85. The digital signal processor of clause 76, wherein the data words comprise Internet Protocol (IP) addresses, and the numbers comprise prefix mask data.
- 86. A priority circuit for determining the most significant number from binary numbers stored in separate rows of memory cells in a memory array, and for identifying the row in which the most significant number is stored, comprising:
-
- a plurality of priority signal lines each coupled to one of the memory cells of each of row of the array; and
- a plurality of priority logic circuits each coupled to one of the memory cells, and each having a first input, a second input coupled to one of the memory cells, an input/output (I/O) coupled to one of the plurality of priority signal lines, and an output, wherein within one row of the memory array, each priority logic circuit has its output connected to the input of the next successive priority logic circuit.
- a plurality of priority signal lines each coupled to one of the memory cells of each of row of the array; and
- 87. The priority circuit of clause 86, wherein the plurality of priority logic circuits each comprise:
-
- a compare circuit coupled to the first input, second input, and I/O line of the priority logic circuit; and
- an isolation circuit coupled to the first input and the output of the priority logic circuit.
- In the foregoing specification the invention has been described with reference to specific exemplary embodiments thereof. It will, however, be evident that various modifications and changes may be made thereto without departing from the broader spirit and scope of the invention. The specification and drawings are, accordingly, to be regarded in an illustrative rather than restrictive sense.
Claims (1)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US11/466,395 US20060280193A1 (en) | 1999-09-23 | 2006-08-22 | Method and apparatus for performing packet classification for policy-based packet routing |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US09/406,170 US7143231B1 (en) | 1999-09-23 | 1999-09-23 | Method and apparatus for performing packet classification for policy-based packet routing |
US11/466,395 US20060280193A1 (en) | 1999-09-23 | 2006-08-22 | Method and apparatus for performing packet classification for policy-based packet routing |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US09/406,170 Division US7143231B1 (en) | 1999-09-23 | 1999-09-23 | Method and apparatus for performing packet classification for policy-based packet routing |
Publications (1)
Publication Number | Publication Date |
---|---|
US20060280193A1 true US20060280193A1 (en) | 2006-12-14 |
Family
ID=23606821
Family Applications (3)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US09/406,170 Expired - Fee Related US7143231B1 (en) | 1999-09-23 | 1999-09-23 | Method and apparatus for performing packet classification for policy-based packet routing |
US09/815,921 Expired - Lifetime US7110408B1 (en) | 1999-09-23 | 2001-03-24 | Method and apparatus for selecting a most signficant priority number for a device using a partitioned priority index table |
US11/466,395 Abandoned US20060280193A1 (en) | 1999-09-23 | 2006-08-22 | Method and apparatus for performing packet classification for policy-based packet routing |
Family Applications Before (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US09/406,170 Expired - Fee Related US7143231B1 (en) | 1999-09-23 | 1999-09-23 | Method and apparatus for performing packet classification for policy-based packet routing |
US09/815,921 Expired - Lifetime US7110408B1 (en) | 1999-09-23 | 2001-03-24 | Method and apparatus for selecting a most signficant priority number for a device using a partitioned priority index table |
Country Status (1)
Country | Link |
---|---|
US (3) | US7143231B1 (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040223486A1 (en) * | 2003-05-07 | 2004-11-11 | Jan Pachl | Communication path analysis |
US20050201402A1 (en) * | 2004-03-01 | 2005-09-15 | Avici Systems, Inc. | Priority sorting |
US20070126105A1 (en) * | 2005-12-06 | 2007-06-07 | Elpida Memory Inc. | Stacked type semiconductor memory device and chip selection circuit |
US20080320129A1 (en) * | 2006-10-09 | 2008-12-25 | Hewlett-Packard Development Company, L.P. | Method of specifying a monitoring criterion for a parameter of an object in an operational management system |
US20130163595A1 (en) * | 2011-12-23 | 2013-06-27 | Electronics And Telecommunications Research Institute | Packet classification apparatus and method for classifying packet thereof |
Families Citing this family (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6985964B1 (en) * | 1999-12-22 | 2006-01-10 | Cisco Technology, Inc. | Network processor system including a central processor and at least one peripheral processor |
US7543077B1 (en) * | 2000-09-19 | 2009-06-02 | Stragent, Llc | Ternary content addressable memory embedded in a central processing unit |
US7669005B1 (en) | 2001-06-18 | 2010-02-23 | Netlogic Microsystems, Inc. | Content addressable memory (CAM) devices having soft priority resolution circuits therein and methods of operating same |
JP3947521B2 (en) * | 2002-03-05 | 2007-07-25 | 富士通株式会社 | Communication device |
US7512078B2 (en) * | 2003-10-15 | 2009-03-31 | Texas Instruments Incorporated | Flexible ethernet bridge |
JP4551668B2 (en) * | 2004-02-25 | 2010-09-29 | パイオニア株式会社 | Minute file generation method, minutes file management method, conference server, and network conference system |
US7444464B2 (en) * | 2004-11-08 | 2008-10-28 | Emc Corporation | Content addressed storage device configured to maintain content address mapping |
US8161013B2 (en) * | 2004-11-08 | 2012-04-17 | Emc Corporation | Implementing application specific management policies on a content addressed storage device |
WO2007002466A2 (en) * | 2005-06-22 | 2007-01-04 | Netlogic Microsystems, Inc. | Access control list processor |
US20080252649A1 (en) * | 2007-04-13 | 2008-10-16 | Barinder Singh Rai | Self-Automating Bandwidth Priority Memory Controller |
US20090077133A1 (en) * | 2007-09-17 | 2009-03-19 | Windsor Hsu | System and method for efficient rule updates in policy based data management |
US8095769B2 (en) * | 2008-08-19 | 2012-01-10 | Freescale Semiconductor, Inc. | Method for address comparison and a device having address comparison capabilities |
US8601205B1 (en) * | 2008-12-31 | 2013-12-03 | Synopsys, Inc. | Dynamic random access memory controller |
US8887026B2 (en) * | 2009-04-14 | 2014-11-11 | Ben Gurion University Of The Negev | Efficient detection of errors in associative memory |
US8023301B1 (en) | 2009-06-19 | 2011-09-20 | Netlogic Microsystems, Inc. | Content addressable memory device having state information processing circuitry |
US9305115B1 (en) | 2010-10-04 | 2016-04-05 | Broadcom Corporation | Method and apparatus for reducing power consumption during rule searches in a content search system |
CN102122270B (en) * | 2011-02-23 | 2013-08-07 | 华为技术有限公司 | Method and device for searching data in memory and memory |
CN103984636B (en) * | 2013-02-08 | 2017-09-29 | 上海芯豪微电子有限公司 | Storage organization and information storage, reading, addressing method |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5257416A (en) * | 1991-02-04 | 1993-10-26 | Motorola, Inc. | Method for dynamically changing channel priority for scan in a receiver |
US5574505A (en) * | 1995-05-16 | 1996-11-12 | Thomson Multimedia S.A. | Method and apparatus for operating a transport stream encoder to produce a stream of packets carrying data representing a plurality of component signals |
US5920886A (en) * | 1997-03-14 | 1999-07-06 | Music Semiconductor Corporation | Accelerated hierarchical address filtering and translation using binary and ternary CAMs |
Family Cites Families (122)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US3257646A (en) | 1963-01-24 | 1966-06-21 | Ibm | Variable word length associative memory |
US3353159A (en) | 1964-12-11 | 1967-11-14 | Burroughs Corp | Associative memory with variable word length capacity |
US3602899A (en) | 1969-06-20 | 1971-08-31 | Ibm | Associative memory system with match,no match and multiple match resolution |
US3685020A (en) | 1970-05-25 | 1972-08-15 | Cogar Corp | Compound and multilevel memories |
US3675211A (en) | 1970-09-08 | 1972-07-04 | Ibm | Data compaction using modified variable-length coding |
BE788028A (en) | 1971-08-25 | 1973-02-26 | Siemens Ag | ASSOCIATIVE MEMORY |
US4030077A (en) * | 1975-10-16 | 1977-06-14 | The Singer Company | Multistage sorter having pushdown stacks for arranging an input list into numerical order |
US4058851A (en) | 1976-10-18 | 1977-11-15 | Sperry Rand Corporation | Conditional bypass of error correction for dual memory access time selection |
JPS5489444A (en) | 1977-12-27 | 1979-07-16 | Fujitsu Ltd | Associative memory processing system |
US4377855A (en) * | 1980-11-06 | 1983-03-22 | National Semiconductor Corporation | Content-addressable memory |
US4464732A (en) * | 1982-03-19 | 1984-08-07 | Honeywell Inc. | Prioritized sorting system |
US4472805A (en) | 1982-03-26 | 1984-09-18 | Rca Corporation | Memory system with error storage |
JPS58212698A (en) | 1982-06-04 | 1983-12-10 | Matsushita Electric Ind Co Ltd | Storage device |
US4670858A (en) | 1983-06-07 | 1987-06-02 | Tektronix, Inc. | High storage capacity associative memory |
JPS60136097A (en) | 1983-12-23 | 1985-07-19 | Hitachi Ltd | Associative memory device |
US4611280A (en) * | 1984-03-12 | 1986-09-09 | At&T Bell Laboratories | Sorting method |
JPS619748A (en) | 1984-06-25 | 1986-01-17 | Nec Corp | Input and output controller |
US4656626A (en) | 1984-12-14 | 1987-04-07 | Itt Corporation | Apparatus and method for providing dynamically assigned switch paths |
US4747080A (en) | 1985-11-12 | 1988-05-24 | Nippon Telegraph & Telephone Corporation | Semiconductor memory having self correction function |
US4785398A (en) | 1985-12-19 | 1988-11-15 | Honeywell Bull Inc. | Virtual cache system using page level number generating CAM to access other memories for processing requests relating to a page |
US4758982A (en) | 1986-01-08 | 1988-07-19 | Advanced Micro Devices, Inc. | Quasi content addressable memory |
US4813002A (en) | 1986-07-21 | 1989-03-14 | Honeywell Bull Inc. | High speed high density dynamic address translator |
US4780845A (en) | 1986-07-23 | 1988-10-25 | Advanced Micro Devices, Inc. | High density, dynamic, content-addressable memory cell |
US4959811A (en) | 1986-11-03 | 1990-09-25 | Texas Instruments Incorporated | Content addressable memory including comparison inhibit and shift register circuits |
JPS63123218A (en) | 1986-11-12 | 1988-05-27 | Nec Corp | Timer/counter circuit |
US4896261A (en) * | 1986-11-24 | 1990-01-23 | Motorola Inc. | System for scheduling serial message transmission on a bus which is adoptable for rescheduling prioritized messages using a doubly-linked list |
DE3801380A1 (en) | 1987-01-20 | 1988-07-28 | Nec Corp | STRING RECOGNITION DEVICE WITH A MEMORY WHOSE STORAGE AREAS ARE SELECTIVELY ACCESSIBLE |
JP2960415B2 (en) | 1987-05-22 | 1999-10-06 | 株式会社日立製作所 | Memory protection method and device |
US4791606A (en) | 1987-09-01 | 1988-12-13 | Triad Semiconductors International Bv | High density CMOS dynamic CAM cell |
US5113339A (en) * | 1987-10-20 | 1992-05-12 | Sharp Kabushiki Kaisha | Data processor for detecting identical data simultaneously coexisting in a plurality of data sections of data transmission paths |
US4845668A (en) | 1987-12-10 | 1989-07-04 | Raytheon Company | Variable field content addressable memory |
US5010516A (en) | 1988-01-11 | 1991-04-23 | Texas Instruments Incorporated | Content addressable memory |
US5051948A (en) | 1988-02-23 | 1991-09-24 | Mitsubishi Denki Kabushiki Kaisha | Content addressable memory device |
JPH01223697A (en) | 1988-03-01 | 1989-09-06 | Mitsubishi Electric Corp | Contents addressing storage device |
US5068822A (en) * | 1988-03-16 | 1991-11-26 | Digital Equipment Corporation | Single-stage extensible sorter for sorting data and efficiently reading out sorted data, incorporating single-bit devices |
US4991134A (en) * | 1988-03-30 | 1991-02-05 | International Business Machines Corporation | Concurrent sorting apparatus and method using FIFO stacks |
US4928260A (en) | 1988-05-11 | 1990-05-22 | Advanced Micro Devices, Inc. | Content addressable memory array with priority encoder |
JPH01307095A (en) | 1988-06-01 | 1989-12-12 | Mitsubishi Electric Corp | Nonvolatile cam |
US4996666A (en) * | 1988-08-12 | 1991-02-26 | Duluk Jr Jerome F | Content-addressable memory system capable of fully parallel magnitude comparisons |
US5247688A (en) * | 1988-10-14 | 1993-09-21 | Ricoh Company, Ltd. | Character recognition sorting apparatus having comparators for simultaneous comparison of data and corresponding key against respective multistage shift arrays |
US5423015A (en) * | 1988-10-20 | 1995-06-06 | Chung; David S. F. | Memory structure and method for shuffling a stack of data utilizing buffer memory locations |
US5072422A (en) | 1989-05-15 | 1991-12-10 | E-Systems, Inc. | Content-addressed memory system with word cells having select and match bits |
US5036486A (en) | 1989-07-31 | 1991-07-30 | Hitachi, Ltd | Associative memory device |
US5053991A (en) | 1989-10-06 | 1991-10-01 | Sanders Associates, Inc. | Content-addressable memory with soft-match capability |
US5159685A (en) * | 1989-12-06 | 1992-10-27 | Racal Data Communications Inc. | Expert system for communications network |
US5485418A (en) | 1990-01-16 | 1996-01-16 | Mitsubishi Denki Kabushiki Kaisha | Associative memory |
US5107501A (en) | 1990-04-02 | 1992-04-21 | At&T Bell Laboratories | Built-in self-test technique for content-addressable memories |
US5014195A (en) | 1990-05-10 | 1991-05-07 | Digital Equipment Corporation, Inc. | Configurable set associative cache with decoded data element enable lines |
US5440715A (en) | 1990-06-27 | 1995-08-08 | Advanced Micro Devices, Inc. | Method and apparatus for expanding the width of a content addressable memory using a continuation bit |
US5317708A (en) | 1990-06-29 | 1994-05-31 | Digital Equipment Corporation | Apparatus and method for an improved content addressable memory |
US5265100A (en) | 1990-07-13 | 1993-11-23 | Sgs-Thomson Microelectronics, Inc. | Semiconductor memory with improved test mode |
US5226005A (en) | 1990-11-19 | 1993-07-06 | Unisys Corporation | Dual ported content addressable memory cell and array |
US5319763A (en) | 1991-04-02 | 1994-06-07 | Motorola, Inc. | Data processor with concurrent static and dynamic masking of operand information and method therefor |
US5239642A (en) | 1991-04-02 | 1993-08-24 | Motorola, Inc. | Data processor with shared control and drive circuitry for both breakpoint and content addressable storage devices |
JP3065736B2 (en) | 1991-10-01 | 2000-07-17 | 松下電器産業株式会社 | Semiconductor storage device |
EP0537721B1 (en) | 1991-10-15 | 1998-11-25 | Hewlett-Packard Company | Hardware-configured operating system kernel for a multitasking processor |
JP2775549B2 (en) | 1992-05-08 | 1998-07-16 | 三菱電機株式会社 | Associative memory cell and associative memory circuit |
US5383146A (en) | 1992-06-08 | 1995-01-17 | Music Semiconductors, Inc. | Memory with CAM and RAM partitions |
GB9213821D0 (en) | 1992-06-30 | 1992-08-12 | Inmos Ltd | Content addressable memory |
US5818873A (en) | 1992-08-03 | 1998-10-06 | Advanced Hardware Architectures, Inc. | Single clock cycle data compressor/decompressor with a string reversal mechanism |
US5469161A (en) | 1992-08-13 | 1995-11-21 | International Business Machines Corporation | Algorithm for the implementation of Ziv-Lempel data compression using content addressable memory |
JP2695102B2 (en) | 1992-09-30 | 1997-12-24 | 川崎製鉄株式会社 | Content addressable memory |
DE69324204T2 (en) | 1992-10-22 | 1999-12-23 | Cabletron Systems, Inc. | Searching for addresses during packet transmission using hashing and a content-addressed memory |
US5319590A (en) | 1992-12-04 | 1994-06-07 | Hal Computer Systems, Inc. | Apparatus for storing "Don't Care" in a content addressable memory cell |
US5455576A (en) | 1992-12-23 | 1995-10-03 | Hewlett Packard Corporation | Apparatus and methods for Lempel Ziv data compression with improved management of multiple dictionaries in content addressable memory |
US5446685A (en) | 1993-02-23 | 1995-08-29 | Intergraph Corporation | Pulsed ground circuit for CAM and PAL memories |
JPH06314241A (en) | 1993-03-04 | 1994-11-08 | Sharp Corp | High-speed semiconductor memory and high-speed associative storage device |
US6000008A (en) | 1993-03-11 | 1999-12-07 | Cabletron Systems, Inc. | Method and apparatus for matching data items of variable length in a content addressable memory |
DE69428408T2 (en) | 1993-04-19 | 2002-05-29 | Oki Electric Industry Co., Ltd. | DECODING CIRCUIT FOR VARIABLE LENGTH CODES, AND DECODING SYSTEM FOR VARIABLE LENGTH CODES USING THIS CIRCUIT. |
US5454094A (en) | 1993-06-24 | 1995-09-26 | Hal Computer Systems, Inc. | Method and apparatus for detecting multiple matches in a content addressable memory |
DE69425209T2 (en) | 1993-09-20 | 2001-03-15 | Codex Corp., Mansfield | Connection method and arrangement for content addressable storage |
US5422838A (en) | 1993-10-25 | 1995-06-06 | At&T Corp. | Content-addressable memory with programmable field masking |
US5396449A (en) | 1993-12-21 | 1995-03-07 | International Business Machines Corporation | Fast content addressable memory with reduced power consumption |
KR950034265A (en) | 1994-02-10 | 1995-12-28 | 도자끼 시노부 | Associative memory |
US5621677A (en) | 1994-04-29 | 1997-04-15 | Cypress Semiconductor Corp. | Method and apparatus for precharging match output in a cascaded content addressable memory system |
US5490102A (en) | 1994-06-15 | 1996-02-06 | Intel Corporation | Low capacitance content-addressable memory cell |
US5860085A (en) | 1994-08-01 | 1999-01-12 | Cypress Semiconductor Corporation | Instruction set for a content addressable memory array with read/write circuits and an interface register logic block |
US5649149A (en) | 1994-08-01 | 1997-07-15 | Cypress Semiconductor Corporation | Integrated content addressable memory array with processing logical and a host computer interface |
US5517441A (en) | 1994-12-14 | 1996-05-14 | International Business Machines Corporation | Content addressable memory circuitry and method of operation |
JP3037089B2 (en) | 1994-12-14 | 2000-04-24 | 川崎製鉄株式会社 | Associative memory |
US5513134A (en) | 1995-02-21 | 1996-04-30 | Gte Laboratories Incorporated | ATM shared memory switch with content addressing |
US5563833A (en) | 1995-03-03 | 1996-10-08 | International Business Machines Corporation | Using one memory to supply addresses to an associated memory during testing |
US5680566A (en) | 1995-03-03 | 1997-10-21 | Hal Computer Systems, Inc. | Lookaside buffer for inputting multiple address translations in a computer system |
JP3703518B2 (en) | 1995-03-30 | 2005-10-05 | 川崎マイクロエレクトロニクス株式会社 | Associative memory system |
JP3125884B2 (en) | 1995-04-18 | 2001-01-22 | インターナシヨナル・ビジネス・マシーンズ・コーポレーシヨン | Content address storage |
US5642322A (en) | 1995-05-24 | 1997-06-24 | Kawasaki Steel Corporation | Layout of semiconductor memory and content-addressable memory |
US5615135A (en) | 1995-06-01 | 1997-03-25 | International Business Machines Corporation | Event driven interface having a dynamically reconfigurable counter for monitoring a high speed data network according to changing traffic events |
US5646878A (en) | 1995-06-02 | 1997-07-08 | Motorola, Inc. | Content addressable memory system |
US5752069A (en) | 1995-08-31 | 1998-05-12 | Advanced Micro Devices, Inc. | Superscalar microprocessor employing away prediction structure |
US5592432A (en) * | 1995-09-05 | 1997-01-07 | Emc Corp | Cache management system using time stamping for replacement queue |
US6065064A (en) | 1995-10-04 | 2000-05-16 | Kawasaki Steel Corporation | Inter-network connecting device |
US6041389A (en) | 1995-11-16 | 2000-03-21 | E Cirrus Logic, Inc. | Memory architecture using content addressable memory, and systems and methods using the same |
US5745488A (en) | 1996-04-29 | 1998-04-28 | Motorola, Inc. | Method and apparatus for detection of packet types in a communications network |
US5828324A (en) | 1996-06-17 | 1998-10-27 | Hewlett-Packard Company | Match and match address signal generation in a content addressable memory encoder |
US6069573A (en) | 1996-06-17 | 2000-05-30 | Hewlett-Packard Company | Match and match address signal prioritization in a content addressable memory encoder |
US5978885A (en) | 1996-06-17 | 1999-11-02 | Hewlett Packard Co. | Method and apparatus for self-timing associative data memory |
US5699288A (en) | 1996-07-18 | 1997-12-16 | International Business Machines Corporation | Compare circuit for content-addressable memories |
JP2836596B2 (en) | 1996-08-02 | 1998-12-14 | 日本電気株式会社 | Associative memory |
US5841874A (en) | 1996-08-13 | 1998-11-24 | Motorola, Inc. | Ternary CAM memory architecture and methodology |
US5930359A (en) | 1996-09-23 | 1999-07-27 | Motorola, Inc. | Cascadable content addressable memory and system |
US5706224A (en) * | 1996-10-10 | 1998-01-06 | Quality Semiconductor, Inc. | Content addressable memory and random access memory partition circuit |
US5914956A (en) * | 1997-02-03 | 1999-06-22 | Williams; Joel R. | Cache for improving the connection capacity of a communications switch |
US6011795A (en) | 1997-03-20 | 2000-01-04 | Washington University | Method and apparatus for fast hierarchical address lookup using controlled expansion of prefixes |
JP3038649B2 (en) | 1997-04-16 | 2000-05-08 | 日本電気株式会社 | Associative search device and method |
US5940852A (en) | 1997-05-01 | 1999-08-17 | Altera Corporation | Memory cells configurable as CAM or RAM in programmable logic devices |
JP3094956B2 (en) | 1997-06-26 | 2000-10-03 | 日本電気株式会社 | Semiconductor storage device |
US5949696A (en) | 1997-06-30 | 1999-09-07 | Cypress Semiconductor Corporation | Differential dynamic content addressable memory and high speed network address filtering |
US6006306A (en) | 1997-07-02 | 1999-12-21 | Xylan Corporation | Integrated circuit with stage-implemented content-addressable memory cell |
US6181698B1 (en) | 1997-07-09 | 2001-01-30 | Yoichi Hariguchi | Network routing table using content addressable memory |
US6014732A (en) | 1997-10-22 | 2000-01-11 | Hewlett-Packard Company | Cache memory with reduced access time |
US6226710B1 (en) * | 1997-11-14 | 2001-05-01 | Utmc Microelectronic Systems Inc. | Content addressable memory (CAM) engine |
ATE224078T1 (en) | 1997-12-16 | 2002-09-15 | Ip Semiconductors As | METHOD AND APPARATUS FOR ADDRESS ANALYSIS FUNCTION IN A NETWORK USING BOOLEAN LOGIC AND PROGRAMMABLE STRUCTURES FOR FULL DESTINATION ADDRESS ANALYSIS |
US6115365A (en) * | 1998-07-30 | 2000-09-05 | Motorola, Inc. | Method and apparatus for queuing and transmitting messages |
US6081440A (en) | 1998-11-05 | 2000-06-27 | Lara Technology, Inc. | Ternary content addressable memory (CAM) having fast insertion and deletion of data values |
US5999435A (en) | 1999-01-15 | 1999-12-07 | Fast-Chip, Inc. | Content addressable memory device |
US6044005A (en) | 1999-02-03 | 2000-03-28 | Sibercore Technologies Incorporated | Content addressable memory storage device |
JP3593920B2 (en) | 1999-05-10 | 2004-11-24 | 日本電気株式会社 | Associative memory with shortest mask output function capable of continuous search operation |
US6108227A (en) | 1999-07-23 | 2000-08-22 | Lara Technology, Inc. | Content addressable memory having binary and ternary modes of operation |
US6748484B1 (en) * | 1999-08-11 | 2004-06-08 | Intel Corporation | Match resolution circuit for an associative memory |
US6467019B1 (en) * | 1999-11-08 | 2002-10-15 | Juniper Networks, Inc. | Method for memory management in ternary content addressable memories (CAMs) |
US6502163B1 (en) * | 1999-12-17 | 2002-12-31 | Lara Technology, Inc. | Method and apparatus for ordering entries in a ternary content addressable memory |
US6493791B1 (en) * | 1999-12-20 | 2002-12-10 | Intel Corporation | Prioritized content addressable memory |
-
1999
- 1999-09-23 US US09/406,170 patent/US7143231B1/en not_active Expired - Fee Related
-
2001
- 2001-03-24 US US09/815,921 patent/US7110408B1/en not_active Expired - Lifetime
-
2006
- 2006-08-22 US US11/466,395 patent/US20060280193A1/en not_active Abandoned
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5257416A (en) * | 1991-02-04 | 1993-10-26 | Motorola, Inc. | Method for dynamically changing channel priority for scan in a receiver |
US5574505A (en) * | 1995-05-16 | 1996-11-12 | Thomson Multimedia S.A. | Method and apparatus for operating a transport stream encoder to produce a stream of packets carrying data representing a plurality of component signals |
US5920886A (en) * | 1997-03-14 | 1999-07-06 | Music Semiconductor Corporation | Accelerated hierarchical address filtering and translation using binary and ternary CAMs |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040223486A1 (en) * | 2003-05-07 | 2004-11-11 | Jan Pachl | Communication path analysis |
US20040223495A1 (en) * | 2003-05-07 | 2004-11-11 | Jan Pachl | Communication path analysis |
US20050201402A1 (en) * | 2004-03-01 | 2005-09-15 | Avici Systems, Inc. | Priority sorting |
US8243744B2 (en) * | 2004-03-01 | 2012-08-14 | Futurewei Technologies, Inc. | Priority sorting |
US20070126105A1 (en) * | 2005-12-06 | 2007-06-07 | Elpida Memory Inc. | Stacked type semiconductor memory device and chip selection circuit |
US8076764B2 (en) * | 2005-12-06 | 2011-12-13 | Elpida Memory Inc. | Stacked type semiconductor memory device and chip selection circuit |
US8709871B2 (en) | 2005-12-06 | 2014-04-29 | Junji Yamada | Stacked type semiconductor memory device and chip selection circuit |
US20080320129A1 (en) * | 2006-10-09 | 2008-12-25 | Hewlett-Packard Development Company, L.P. | Method of specifying a monitoring criterion for a parameter of an object in an operational management system |
US20130163595A1 (en) * | 2011-12-23 | 2013-06-27 | Electronics And Telecommunications Research Institute | Packet classification apparatus and method for classifying packet thereof |
Also Published As
Publication number | Publication date |
---|---|
US7110408B1 (en) | 2006-09-19 |
US7143231B1 (en) | 2006-11-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20060280193A1 (en) | Method and apparatus for performing packet classification for policy-based packet routing | |
US7110407B1 (en) | Method and apparatus for performing priority encoding in a segmented classification system using enable signals | |
US7272027B2 (en) | Priority circuit for content addressable memory | |
US6499081B1 (en) | Method and apparatus for determining a longest prefix match in a segmented content addressable memory device | |
US6460112B1 (en) | Method and apparatus for determining a longest prefix match in a content addressable memory device | |
US6934795B2 (en) | Content addressable memory with programmable word width and programmable priority | |
US6574702B2 (en) | Method and apparatus for determining an exact match in a content addressable memory device | |
US6757779B1 (en) | Content addressable memory with selectable mask write mode | |
US6944709B2 (en) | Content addressable memory with block-programmable mask write mode, word width and priority | |
US6539455B1 (en) | Method and apparatus for determining an exact match in a ternary content addressable memory device | |
US7054993B1 (en) | Ternary content addressable memory device | |
US6154384A (en) | Ternary content addressable memory cell | |
US6418042B1 (en) | Ternary content addressable memory with compare operand selected according to mask value | |
US6564289B2 (en) | Method and apparatus for performing a read next highest priority match instruction in a content addressable memory device | |
US7230840B2 (en) | Content addressable memory with configurable class-based storage partition | |
US6324087B1 (en) | Method and apparatus for partitioning a content addressable memory device | |
US6876559B1 (en) | Block-writable content addressable memory device | |
US7193874B1 (en) | Content addressable memory device | |
US7382637B1 (en) | Block-writable content addressable memory device | |
US20070064461A1 (en) | Low power content addressable memory | |
US7325091B2 (en) | Disabling defective blocks in a partitioned CAM device | |
US6687785B1 (en) | Method and apparatus for re-assigning priority in a partitioned content addressable memory device | |
US7487200B1 (en) | Method and apparatus for performing priority encoding in a segmented classification system | |
WO1999023663A1 (en) | Synchronous content addressable memory with single cycle operation | |
US6892272B1 (en) | Method and apparatus for determining a longest prefix match in a content addressable memory device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STCB | Information on status: application discontinuation |
Free format text: ABANDONED -- FAILURE TO RESPOND TO AN OFFICE ACTION |
|
AS | Assignment |
Owner name: BROADCOM CORPORATION, CALIFORNIA Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:NETLOGIC I LLC;REEL/FRAME:035443/0763 Effective date: 20150327 Owner name: NETLOGIC I LLC, DELAWARE Free format text: CHANGE OF NAME;ASSIGNOR:NETLOGIC MICROSYSTEMS, INC.;REEL/FRAME:035443/0824 Effective date: 20130123 |
|
AS | Assignment |
Owner name: BANK OF AMERICA, N.A., AS COLLATERAL AGENT, NORTH CAROLINA Free format text: PATENT SECURITY AGREEMENT;ASSIGNOR:BROADCOM CORPORATION;REEL/FRAME:037806/0001 Effective date: 20160201 Owner name: BANK OF AMERICA, N.A., AS COLLATERAL AGENT, NORTH Free format text: PATENT SECURITY AGREEMENT;ASSIGNOR:BROADCOM CORPORATION;REEL/FRAME:037806/0001 Effective date: 20160201 |
|
AS | Assignment |
Owner name: AVAGO TECHNOLOGIES GENERAL IP (SINGAPORE) PTE. LTD., SINGAPORE Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:BROADCOM CORPORATION;REEL/FRAME:041706/0001 Effective date: 20170120 Owner name: AVAGO TECHNOLOGIES GENERAL IP (SINGAPORE) PTE. LTD Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:BROADCOM CORPORATION;REEL/FRAME:041706/0001 Effective date: 20170120 |
|
AS | Assignment |
Owner name: BROADCOM CORPORATION, CALIFORNIA Free format text: TERMINATION AND RELEASE OF SECURITY INTEREST IN PATENTS;ASSIGNOR:BANK OF AMERICA, N.A., AS COLLATERAL AGENT;REEL/FRAME:041712/0001 Effective date: 20170119 |