标签：des style blog http color io os ar strong
A processor of a plurality of processors includes a processor core and a message?manager. The message?manager?is in communication with the processor core. The message?manager?to receive a message from a second processor of the plurality of processors, to identify a classification rule for the message based on bits in a header of the message, and to create a queue identifier for the message using bits of a payload of the message, wherein the queue identifier is associated with a queue of the processor core.
This disclosure generally relates to data processing, and more particularly to a system and method for assigning messages within a data processing system.
A system includes a switching fabric to communicate information between multiple data processor devices that can communicate with each other via multiple data streams (streams). The switching fabric can route data for a particular stream from a source device to a destination device using a packetized message (a packet) having a header that includes header fields, and a payload that includes data. Such header fields can include a source identifier field, a destination identifier field, a class-of-service identifier field, a stream identifier field, a flow identifier field, and the like. At the destination device, a message manager?can determining a queue where a received packet is to be routed based on the header fields of the message. The message?manager?can then send the packet and an identifier indicating the target queue to a queue manager, which in turn can store the received packet in the queue indicated by the queue identifier.
The present disclosure may be better understood, and its numerous features and advantages made apparent to those skilled in the art by referencing the accompanying drawings.
FIG. 1?illustrates a system including multiple processors in accordance with at least one embodiment of the present disclosure.
FIG. 2?illustrates a message sent between two processors in the system of?FIG. 1?in accordance with at least one embodiment of the present disclosure.
FIG. 3?illustrates an input/output controller of a processor of?FIG. 1?in accordance with at least one embodiment of the present disclosure.
FIG. 4?illustrates a queue identifier for a queue in the processor of?FIG. 1?in accordance with at least one embodiment of the present disclosure.
FIG. 5?illustrates a method for assigning a message to a queue in the processor of?FIG. 1?in accordance with at least one embodiment of the present disclosure.
The following description in combination with the Figures is provided to assist in understanding the disclosure herein. The following discussion will focus on specific implementations and embodiments of one or more embodiments. This focus is provided to assist in describing the embodiments and should not be interpreted as a limitation on the scope or applicability of the disclosure, as other embodiments can be utilized.
FIGS. 1-5?illustrate embodiments of a messaging system employing multiple processors that can utilize different data streams to communicate information amongst each other via a switching fabric. During initialization, restart, or any type of configuration process, a processor that can receive a data stream from a source processor, referred to as a destination processor, can be initialized with a set of classification rules. In another embodiment, the set of classification rules can be determined and configured during a design process of the processor. According to an embodiment, when a packet of a packetized message is received by the destination device during normal operation, the message manager?selects one of the classification rules, which were provided during start-up, based upon particular bits of the received packet. The particular bits used to select the classification rule can be from one or more header fields of the packet. In determining the selected classification rule, the message?manager?can use multiple classification rule wildcard registers, each of which can indicate a portion of the one or more header fields of the message that is to be masked when selecting the classification rule. Thus, the message?manager?can determine the base queue identifier for a packet based only upon header bits of the data packet.
Once the base queue identifier is determined, the message?manager?then determines a specific queue relative to the base queue where a message is to be routed. For example, a queue offset relative to a base queue identifier can be determined. Thus, the queue offset can be based on various bits selected from the header fields and the payload of the packetized message. According to an embodiment, the specific queue is determined by combining the queue offset with the base queue identifier to create a destination queue identifier for the message. The message?manager?then sends the message and the queue identifier to a queue?manager?of the destination processor, which in turn routes the received message to a particular queue identified by the queue identifier. Various embodiments of the present disclosure will be better understood with reference to?FIGS. 1-5.
FIG. 1?illustrates a system?100?including processors?102,?104,?106, and?108(processors?102-108) and switching fabric?110?in accordance with at least one embodiment of the present disclosure. Each of the processors?102-108?includes an input/output (I/O) interface?112. The processor?102?is described herein as receiving a data stream and is therefore also referred to as destination processor?102, though it will be appreciated that in various embodiments, processor?102?can also be a source processor that provides a data stream to another processor.
Destination processor?102?includes a message?manager?114?within the I/O interface?112, a queue?manager?116, a?buffer?manager?118, multiple groups of queues?120, multiple processor cores?122, and packet buffers?124. Depending on the embodiment, queues in the groups of queues?120?can be physical queues, virtual queues, and the like. Each of the other processors?104,?106, and?108?can also include a message?manager, a queue?manager, a?buffer?manager, queues, multiple processor cores, and a packet?buffer, but these have not been shown in FIG. 1?for simplicity.
The switching fabric?110?includes multiple ports?126?and switching logic?128. Each of the processors?102-108?is connected to a corresponding port?126, and can operate as a destination processor or a source processor for receiving and sending messages, respectively. However, for simplicity it is presumed that processor?102?is operating as a destination processor and processor?108?is operating as a source processor.
A particular data stream?130?is illustrated to communicate packetized messages between source processor?108?and destination processor?102?via a particular communication path through the switching fabric?110. The message?132?is a packet that includes a header?134?and a payload?136. The header?134?can include multiple fields, such as an extended destination identification (EDID) field, a destination identification (DID) field, an extended source identification (ESID) field, a source identification (SID) field, a class-of-service (COS) field, a stream identification field, a flow identification field, a physical source port (PORT) field, a letter field, a mailbox field, and the like. Based upon the header information, the switching logic?128?determines that the message?132?is addressed to the destination processor?102, and routes the message?132?out of the switching fabric?110?via the port?126?to the destination processor?102.
In an embodiment, the communication path can include: an intra-die communication path, wherein packets are transmitted between processors of a common integrated circuit die; or an inter-die communication path, wherein packets are transmitted between processors on different die. For example, the processor?102?and processor?108?can reside on separate die that are mounted to a common substrate, such as to a printed circuit board, to a package substrate, and the like. According to one embodiment the packetized messages are compliant with the RapidIO communication protocol. Thus, the system?100can utilize the RapidIO communication protocol to transfer messages?132between different processors of the system?100, such as instruction based data processors. In one embodiment, the messages?132?can be a RapidIO doorbell messages and the like. In one embodiment, the message?132?can be transferred between the processors by the switching logic?128, which can be a cross-point switch implemented in hardware, software, or the like.
When the I/O interface?112?of the destination processor?102?receives the message?132, the message?manager?114?utilizes the header?134?information of the message?132?to determine a classification rule for the message?132?that is used to further determine a particular group of the groups of queues?120?at which the message will be stored. In an embodiment, the message?manager?114can access multiple classification rule wildcard registers (shown in?FIG. 3) to identify bits of the header?134?to be masked (not used) during the selection of a classification rule. The?buffer?manager?118?further allocates packet buffers?124, e.g., queues, in a memory cache of the destination processor?102?for the message?132?to be placed until it is retrieved by the queue?manager?for storage a specific queue. Depending on the embodiment, either the entire message?132including the header?134?and the payload?136, or only the payload?136?is placed into the packet buffers?124, and subsequently into a queue of the plurality of queues?120.
FIG. 2?illustrates a message?132?sent between the source processor?108?and the destination processor?102?in accordance with at least one embodiment of the present disclosure. The header?134?of the packet includes an EDID field?202, a DID field?204, an ESID field?206, and a SID field?208. The payload?136?of the packet includes a plurality of bits including first set of bits?218. The message?manager?114?can utilize bits from the header fields and from the payload?136?when determining a specific queue of the selected group of queues where the message?132?is to be stored. In particular, the message?manager?114?can determine the specific queue by retrieving particular bits?210?from the EDID field, particular bits212?from the DID field?204, particular bits?214?from the ESID field?206, particular bits?216?from the SID field?208, and particular bits?218?from the payload?136. The message?manager?114?can use these particular bits to generate a queue identifier for the message?132. The operation of selecting the particular bits and generating the queue identifier for the message?132?is described in more detail with reference to?FIGS. 3 and 4?below.
Referring back to?FIG. 1, after the message?manager?114?generates the queue identifier, the message?manager?114?sends the queue identifier for the message?132?to the queue?manager?116, which in turn assigns the message?132?from the packet?buffer?124?to a queue of the previously identified group of the group of queues?120?based on the queue identifier. The processor core associated with that particular queue?120?can retrieve the message?132?from the queue in a group of queues?120?to perform one or more operation associated with the message?132.
The queue identifier for the message?132?that identifies the specific queue where a message is to be stored can vary based on the particular logic states of the bits in the fields of the header?134?of the message?132?and based on the particular logic states of the bits in the payload?136?of the message?132. The available header bits of message?132?that are used to determine a queue identifier can be limited such that there may be more queues available in the destination processor?102than can be identified by using the available header bits specified in the packet header. Thus, the message?manager?114can generate more queue identifiers by also using the particular bits?218?of the payload?136. The particular bits?218?of the payload?136?may include a byte of data, such as a first or second received byte of information. Thus, the number of queues in a selected group of the groups of queues?120?that a message?132?can be assigned increases substantially by the message?manager?114?incorporating a byte of the payload?136?into the queue identifier.
The processor?102,?104,?106, or?108?transmitting the message?132?can, therefore, have more control and flexibility to vary bits in the payload?136?of the message?132?as compared to varying bits in the fields of the header?134?of the message?132. As a result, the processor sending the message?132?can actively identify a particular queue in a group of queues?120?by varying the bits of the payload?136?in the message?132. The manner in which a particular queue is determined by a destination processor based upon the payload information can be understood a priori by a source processor, or the destination processor?102, e.g. the message?manager?114, can transmit information to the source processor indicating the manner in which payload information is used to select a particular queue. For example, a destination processor can forward information to a source processor indicating the particular bits of the payload?136?used in classifying the message?132?and thereby assigning the message?132?to a particular queue in a group of queues?120. The source processors can use this information to set the particular bits?218?in the payload?136?of the message?132?to select a particular queue in a group of queues?120. It will be appreciated that in an alternate embodiment, that a source processor can send information to the destination processor indicating the manner in which payload information is used to select a particular queue.
FIG. 3?illustrates I/O interface of?FIG. 1?in greater detail and in accordance with an embodiment of the present disclosure, and in particular,?FIG. 3?illustrates an embodiment of the message?manager?114?in greater detail. The message?manager114?includes a base queue identification module?302, multiple selection registers?304, multiple classification rule wildcard registers?306, multiple classification rule base queue registers?308, and multiple classification rule mode registers?310. In an embodiment, the message?manager?114?can include the same number of selection registers?304, classification rule wildcard registers?306, classification rule base queue registers?308, and classification rule mode registers?310?as the number of classification rules?312?in the base queue identification module?302. In this embodiment, a different one of each of the selection registers?304, the classification rule wildcard registers?306, the classification rule base queue registers?308, and the classification rule mode registers?310?corresponds to a particular classification rule?312, such as on a one-to-one basis. For example, one classification rule?312?is associated with a particular selection register?304, a particular classification rule wildcard register?306, a particular classification rule base queue register?308, and a particular classification rule mode register?310. In another embodiment, the number of different registers may vary from the number of classification rules?312. For example, the message?manager?114?may include one selection register?304?for each type of message?132?that may be sent to the destination processor?102.
The base queue identification module?302?is initiated to include multiple classification rules?312, wherein each classification rule?312?can be used to select a base queue group from the group of queues?120, out of all of available the groups, for the incoming message?132. In an embodiment, the classification rules?312?can be information at storage locations, such as registers, accessible by the message?manager?114, and different bits of the stored the classification rules?312?can be set to make an selection/identification of different classification rules?312?possible for the message?manager?114. The different bits of the classification rules?312?registers can be set during a start-up or initialization of the processor?102. For example, each of the classification rules?312?can specify different bits and combinations of bits that are used for selecting the queue group (the base queue) from the groups of queues?120. In another embodiment, the classification rules?312?can be defined in software or an algorithm that is executed to determine a classification rule for an incoming message. The each selection register?304?identifies a number of bits to use in each of the fields of the header?134?and the payload?136?of the message132?to identify a specific queue in the base queue for the message?132?when the message?132?is received in the I/O interface?112.
Each classification rule wildcard register?306?identifies a number of consecutive bits in each of the fields of the header?134to be masked during the selection of a queue for the message?132. Masked bits used herein are bits that are disregarded and not used during the identification of a queue for the message?132. Each classification rule base queue register?308?can store a base queue identifier associated with the message?132?based on the classification rule?312?identified for the message?132?in the base queue identification module?302. Each classification rule mode register?310?identifies which bits of the payload?136?of the message?132?are used for identifying a particular queue in the group of queues?120?for the message132. The message?manager?114?can utilize the base queue identification module?302, the classification rule selection registers?304, the classification rule wildcard registers?306, the classification rule base queue registers?308, and the classification rule mode registers?310?to generate a queue identifier associated with the message?132?that can be used to identify a queue for the message?132?as described below with respect to?FIG. 4.
FIG. 4?illustrates the message?132, the selection register?304, the classification rule wildcard register?306, and the classification rule base queue register?308?used to generate a queue identifier?402?by the message?manager?114?of?FIG. 1in accordance with at least one embodiment of the present disclosure. The selection register?304?includes an EDID field404, a DID field?406, an ESID field?408, an SID field?410, a payload field?412, and a port field?414. The classification rule wildcard register?306?includes a DID field?416, an SID field?418, and a payload field?420.
When the I/O interface?112?receives the message?132, the base queue identification module?302?selects a classification rule based upon the bits of the header fields?202,?204,?206, and?208?and the classification rule wildcard registers. The message manager?114?can utilize each of the classification rule wildcard registers?306?to mask particular bits in the header?134?of the message?132. For example, if the message?manager?114?includes thirty-two classification rule wildcard registers?306, e.g., thirty-two classification rules, the message?manager?114?can mask the bits of the header?134?using any one of thirty-two different ways to get one of thirty-two different combinations of unmasked bits from the header?134. The message?manager114?can then compare the resulting combination of unmasked bits to select a corresponding classification rule?312, e.g., by determining which combination of unmasked bits from the header?134?matches, or most closely matches, the corresponding classification rule?312. The corresponding classification rule?312?can then be used to determine a corresponding selection register?304, a particular classification rule wildcard register?306, a particular classification rule base queue register?308, and a particular classification rule mode register?310?for the message?132. The message?manager?114?can select the classification rule?312?that matches with the resulting unmasked header bits from the corresponding classification rule wildcard register?306?as the classification rule?312?for the message?132.
For example, the message?manager?114?selects a particular classification rule?312?when to a classification rule wildcard register?306?generates a combination of unmasked bits from the header?134?of ‘01101‘ and the particular classification rule312?corresponding to the classification rule wildcard register?306?identifies ‘01101‘ as the bits for that classification rule?312. In an embodiment, if message?manager?114?does not identify a match between any of the classification rule wildcard registers?306?and the corresponding classification rules?312, the message?manager?114?can apply different factors to determine a closest match between unmasked bits from the classification rule wildcard register?306?and the corresponding classification rule?312. In this situation, the classification rule?312?with the closest match will be selected as the classification rule?312?for the message?312.
The base queue identification module?302?can then identify a classification rule base queue register?308?corresponding to the selected classification rule?312?to select a base queue identifier for the message?132. The base queue identifier can be a value that identifies a base group of queues?120?for the message?132. The message?manager?114?accesses the selection register?304?corresponding to the selected classification rule?312, or to the message type, and identifies the number of bits for each of the fields of the message?132?to be retrieved. For example, the message?manager?114?accesses the selection register?304?to identify that no bits are selected from the EDID field?404, that five bits are selected from the DID field?406, that no bits are selected from the ESID field?408, that three bits are selected from the SID field?410, that nine bits are selected from the payload field?412, and that no bits are selected from the port field?414.
One of ordinary skill in the art would recognize that not all messages received at the destination processor?102?will include a header?134?with each of the fields listed above, or that a message?132?could include a header?134?with additional fields. In another embodiment, each of the selection registers?304?can include more or less fields. The message?manager?114?then accesses the classification rule wildcard register?306?associated with the selected classification rule?312?to determine a number of consecutive bits in each of the fields of the header?134?and in the payload?136?to be masked during the identification of a queue for the message?132. Masked bits used herein are bits that are disregarded and not used during the selection/identification of a particular queue in a group of queues?120?for the message?132. The masking of bits in the fields of the header?134?enables a specific classification rule to be selected for a range of streams or messages, e.g., more than one stream or message, based on a particular number of the bits identified in the selection register?304?not being used for the queue identifier. In one embodiment, unmasked bits should be consecutive and the selection of bits by the message manager?should start with the rightmost bit of the unmasked bits. The unmasked bits should be consecutive because non-consecutive unmask bits may result in an undefined behavior in the message?manager?114.
The message?manager?114?also accesses the classification rule wildcard register?306?associated with the selected classification rule?312?to identify the masked bits in the fields of the header?134?and the payload?136. For example, the message?manager?114?accesses the classification rule wildcard register?306?associated with the selected classification rule312?to identify two bits in the DID field?416?as masked bits. In an embodiment, the classification rule wildcard register?304indicates which bits in a field are masked by setting those bits in the field to specific value, such as a digital ‘0‘ or ‘1‘. The message?manager?114?also accesses the classification rule wildcard register?304?to identify four bits in the SID field?418?as masked bits, and eight bits in the payload field?420?as masked bits. The message?manager?114?then retrieves the particular bits?212?from the DID field?204, the particular bits?216?from the SID field?208, and the particular bits?218?from the payload136?identified by the combination of the selection register?304?and the classification rule wildcard register?306. For example, the message?manager?114?identifies the masked and unmasked bits within in the DID field?204?of the message?132, and then selects the particular bits?212?from the DID field?204?starting with the rightmost bit of the unmasked bits identified in the classification rule wildcard register?306. In this example, the message?manager?114?selects the five bits in the DID field204, identified as?212?in?FIG. 4, based on the selection register?304?identifying that five bits are to selected and based on the classification rule wildcard register?306?identifying five consecutive unmasked bits in the DID field?416?of the classification rule wildcard register?306.
The message?manager?114?then identifies the number of bits to retrieve from the SID field?208?based on the selection register?304, and identifies the unmasked bits within in the SID field?208?of the message?132?based on the classification rule wildcard register?306. The message?manager?114?selects the particular bits?216?from the SID field?208?starting with the rightmost bit of the unmasked bits. For example, the message?manager?114?selects the three bits in the SID field?208?of the message?132, identified as?216?in?FIG. 4, based on the SID field?410?of the selection register?304?identifying that three bits are to be selected and that the SID field?416?of the classification rule wildcard register?306?identifying four consecutive unmasked bits. When selecting the specific bits?216, the message?manager?114?selects minimum number of bits between the number of bits identified in the SID field?410?of the selection register?304?and the number of unmasked bits identified in the SID field?418?of the classification rule wildcard register?306. Thus, the message?manager?114?selects only three bits as identified in the SID field?410?of the selection register?304?even though the SID field?418?of the classification rule wildcard register?306?identifies four unmasked bits. In an embodiment, the number of unmasked bits identified in the classification rule wildcard register?306?may be less than the number of bits identified in the selection register?304?for a given field. This situation may occur in response to a large number of bits being used for another purpose other than classification, and these bits should be masked during classification of the message?132?so that varying these bits does not effect the selection of a queue in the base group of queues?120, e.g. the most significant bits or the least significant bits can be masked during the classification of the message?132.
The message?manager?114?identifies the number of bits to retrieve from the payload?136?based on the selection register304, and identifies the unmasked bits within in the payload?136?of the message?132?based on the classification rule wildcard register?306?associated with the selected classification rule?312. The message?manager?114?accesses the classification rule mode register?310?associated with the selected classification rule?312?to determine which bits of the payload?136?of the message?132?are used for generating queue offset?422?associated with the message?132. The message?manager?114selects the particular bits?218?from the payload?136?starting with the rightmost bit of the unmasked bits. In this example, the message?manager?114?selects eight bits in the payload?136?of the message?132, identified as?218?in?FIG. 4, based on the payload field?420?of the classification rule wildcard register?306?identifying eight consecutive unmasked bits and based on the payload field?412?of the selection register?304?identifying the nine bits to be selected. When selecting the specific bits136, the message?manager?114?selects a minimum number of bits between the number of bits identified in the SID field?410of the selection register?304?and the number of unmasked bits in the SID field?418?of the classification rule wildcard register306?since in this example the ninth bit is masked to the message?manager?114?during classification of the message. Thus, the message?manager?114?compresses the zero in the ninth bit to retrieve only the eight bits identified in the payload field420?of the classification rule wildcard register?306.
The message?manager?114?assembles the particular bits?212?from the DID field?204, the particular bits?216?from the SID field?208, and the particular bits?218?from the payload?136?of the message?132?into the queue offset?422. The message manager?114?then combines the queue offset?422?with the base queue identifier of the classification rule base queue register?308?associated with the selected classification rule?312?to create the queue identifier?402. Thus, the message manager?114?generates the queue identifier?402?to utilize deterministic virtual queuing of the message?132?when the I/O interface?112?receives the message?132, such that the message?manager?114?does not utilize hashing to queue the message?132. In an embodiment, the queue identifier?402?can be limited to a specific number of bits, such as twenty four bits. Thus, the message?manager?114?truncates bits of the queue identifier?402?if the combination of the queue offset?422and the base queue identifier exceeds the specific number of bits for the queue identifier?402. The message?manager?114then sends the queue identifier?402?along with the message?132?to the queue?manager?116, which in turn places the message?132?into the queue in the base group of queues?120?associated with the queue identifier?402.
FIG. 5?illustrates a method for assigning a message to a queue in the processor of?FIG. 1?in accordance with at least one embodiment of the present disclosure. At block?502, the destination processor?102?transmits an identification of the specific bits of the payload?136?of the message?132?used in the queue identifier?402?to a source device, such as processor?104?of FIG. 1, during a session protocol between the source device and the destination device. The session protocol can be utilized by the source and destination devices to configure each of the devices to transmit message between each other. At block?504, the destination processor?102?receives a message?132?at the I/O interface?112. At block?506, the message manager?114?selects a classification rule associated with particular bits in a field of a header?134?of the received message132. The classification rule is selected based on a comparison between unmasked bits in the field of a header?134?of the message?132?for a particular classification rule?312?and the classification rule?312.
At block?508, the message?manager?114?accesses the selection register?304?associated with the selected classification rule312?to identify a number of bits of the header?134?of the message?132?to be used for the queue identifier?402. The message?manager?114?accesses the classification rule mode register?306?associated with the selected classification rule312?to identify specific bits in the payload?136?of the message?132?to be used for the queue identifier?402?at block?510. In an embodiment, the specific bits are a byte of the payload?136, such as the most significant byte, the least significant byte, or the like. The message?manager?114?retrieves the number of bits from the field of the header?134?at block?512. The message?manager?114?can retrieve the number of bits starting with a rightmost bit of unmasked bits in the field. The unmasked bits can be identified by the classification rule wildcard register?306?associated with the selected classification rule?312.
At block?514, the message?manager?114?retrieves the specific bits of the payload?136?of the message?132. The retrieved bits from the header field and the retrieved bits from the payload?136?can be the minimum number of bits identified in either the selection register?304?or the classification rule wildcard register?306. The message?manager?114?assembles the retrieved bits from the header field and the retrieved bits from the payload?136?to create an offset to a base queue identifier at block?516. At block?518, the message?manager?114?adds the offset to the base queue identifier to create the queue identifier?402. In an embodiment, if the number of bits in the offset and the base queue identifier exceed the maximum number of bits for the queue identifier, the message?manager?114?can truncate one or more of the bits. The queue?manager116?assigns the message?132?to a particular queue of a base group of queues?120?based on the queue identifier?402?at block?520.
In accordance with one aspect of the present disclosure, a method of communicating information between a plurality of processors is provided. In one embodiment, the method includes receiving, from a processor of the plurality of processors, a message in a stream. The method further includes identifying a classification rule for the message based on bits in a header of the message. The method also includes generating a queue offset based bits in a payload of the message. The method further includes assigning the message to a queue of a processor core in the processor based on the classification rule and the queue offset.
In one embodiment, the method includes retrieving bits from the header of the message. The method also includes retrieving bits from the payload of the message. The method further includes creating a queue identifier for the message using the retrieved bits of the header, and the retrieved bits of the payload. In this case, the method further includes assembling the retrieved bits of the header and the retrieved bits of the payload as the queue offset for the queue identifier. The method also includes adding the queue offset to a base queue identifier of the message to create the queue identifier.
In one embodiment, the method further includes truncating one or more bits of the queue offset in response to a total number of bits from the queue offset and the base queue identifier exceeding a maximum number of bits for the queue identifier. In one embodiment, the method further includes identifying bits in the header and specific bits of the payload of the message to use for a queue identifier based on bits in a register. The method also includes transmitting information identifying the specific bits of the payload of the message used in the queue identifier to the processor of the plurality of processors that is to provide the message prior to receiving the message.
In accordance with another aspect of the present disclosure, a method of communicating information between a plurality of processors is provided. In one embodiment, the method includes receiving, at a first processor, a message in a stream from a second processor. The method further includes creating an offset based on bits in a header and bits in a payload of the message. The method also includes identifying a classification rule for the message based on the bits in the header of the message. The method further includes combining the offset to a base queue identifier of the message to create a queue identifier, wherein the base queue identifier is based upon the classification rule. The method also includes assigning the message to a queue of a processor core in the processor based on the queue identifier.
In one embodiment, the method further includes identifying the bits of in the header and specific bits of the payload of the message to use for the queue identifier based on bits in a register of the first processor. The method further includes transmitting information identifying the specific bits of the payload of the message used to create the offset to the second processor prior to receiving the message. In one embodiment, the method includes retrieving, prior to creating the queue identifier, the bits of the message to be used in the queue identifier. The method also includes retrieving, prior to creating the queue identifier, the bits of the payload of the message to be used in the queue identifier.
In one embodiment, the method includes that identifying the bits in the header further includes: identifying masked bits in the header of the message to select specific bits to retrieve the bits of the header. In this case, identifying the bits of the payload further includes: identifying masked bits of the payload of the message to select specific bits to retrieve the bits of the payload. In one embodiment, truncating one or more bits of the offset in response to a total number of bits from the offset and the base queue identifier exceeding a maximum number of bits for the queue identifier. The method also includes that the classification rule is identified only based on bits in the header of the message. The method also includes that the queue identifier is created based on the bits in the header and the bits in the payload of the message.
In accordance with another aspect of the present disclosure, a processor of a plurality of processors of an integrated circuit device is provided. The processor includes a processor core, and a message?manager. In one embodiment, the message manager?is in communication with the processor core, and the message?manager?receives a message from a second processor of the plurality of processors, identifies a classification rule for the message based on bits in a header of the message, and creates a queue identifier for the message using bits of a payload of the message. In this case, the queue identifier is associated with a queue of the processor core. In one embodiment, the message?manager?includes: a first register to store a value to identify bits to retrieve from the header of the message for the queue identifier, a second register to store a value to identify bits of the payload of the message to retrieve for the queue identifier, and a third register to store masked bits for the header and masked bits for the payload. In this case, the message?manager?utilizes unmasked bits for the header to identify bits of the header to retrieve, and utilizes unmasked bits for the payload to identify the bits of the payload to retrieve.
In one embodiment, the processor further includes a queue?manager?in communication with the message?manager. In this case, the queue?manager?receives the queue identifier associated with the message, and assigns the message to the queue based on the queue identifier. In one embodiment, the message?manager?further retrieves, prior the queue identifier being created, bits of the header of the message to be used in the queue identifier, and retrieves, prior the queue identifier being created, the bits of the payload of the message to be used in the queue identifier. In one embodiment, the processor further includes a register to provide a base queue identifier, the message?manager?to assemble the retrieved bits of the header and the retrieved bits of the payload as an offset to a base queue identifier to create a queue identifier. In one embodiment, the message?manager?creates an offset to a base queue identifier, and the offset includes the bits of the payload.
In different embodiments, the message?manager?114?and the base queue identification module?302?can includes hardware circuitry, software, state machines, executable code, and the like. Based upon the description herein, it will be appreciated that the preceding detailed description is, therefore, not intended to be limited to the specific forms set forth herein, but on the contrary, it is intended to cover such alternatives, modifications, and equivalents, as can be reasonably included within the spirit and scope of the appended claims.
SRC=https://www.google.com.hk/patents/US20140122735
System and method for assigning a message
标签：des style blog http color io os ar strong
原文地址：http://www.cnblogs.com/coryxie/p/3973659.html
