JP2004159038A - Communication device and communication method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a communication device and a communication method which can reduce a processing burden of a host and a network load. <P>SOLUTION: A packet parser 13 is provided between a network interface card 12 and the host 11 (a reception processing part). The packet parser 13 executes header analysis of the received packet and it determines the protocol of the packet. When the determined result is an important packet, the packet parser 13 notifies the host 11 of the reception (interruption) unconditionally, that is to say, so as to preferentially execute reception processing. When it is an un-important packet, interruption is executed, for example, after a data amount exceeds a setting value. Also, when it is a real time stream packet, the packet parser 13 directly transfers it to a Codec not to the host. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a communication device and a communication method for network communication such as a local area network.
[0002]
[Prior art]
Conventionally, when a host communicates with a local area network (LAN) to perform each protocol processing, a conventional method is performed in which a network interface card, a corresponding device driver, and a protocol stack mounted on the host are involved. Was taken.
In this system, the network interface card and the device driver only transmit a packet to the LAN transmission line or receive a packet from the LAN transmission line. In other words, the generation of the transmission packet, the analysis of the reception packet, and the associated protocol processing are all performed by the host.
[0003]
FIG. 7 shows an example of a host system connected to a LAN.
In FIG. 7, it is assumed that, for example, the transfer of, for example, MPEG-based compressed moving image data is performed between two host systems 30, 40 using Ethernet (registered trademark) as a data transfer medium.
The hosts 31 and 41 in FIG. 7 are main information processing units including hardware such as a CPU and a RAM, and include application software started by the CPU and programs for various protocol processes. The host systems 30 and 40 refer to computer systems including the hosts 31 and 41 and the network interface cards 32 and 42.
[0004]
A case where MPEG data is transmitted from the host system 40 to the host system 30 will be considered. In this case, first, the host system 40 performs processing of each layer for real-time stream transfer by the CPU and software of the host 41 on data encoded by the installed software codec or hardware codec by MPEG. For example, packet processing and header processing of each protocol of RTP (Real Time Transport Protocol) / UDP (User Datagram Protocol) / IP (Internet Protocol) / MAC (Media Access Control) are performed.
Thereafter, the host 41 transfers the processed packet to the network interface card 42, and the network interface card 42 transmits the packet to the host system 30 via the transmission path 100.
[0005]
The host system 30 makes the host 31 sense that a packet has been received by an interrupt (interrupt request as a reception notification) from the mounted network interface card 32. Then, the host 31 reads the received packet stored in the buffer in the network interface card 32. The interrupt to the host 31 is generated every time a received packet arrives.
The host 31 analyzes the header read of the packet read from the network interface card 32 by the protocol processing software and the host CPU, processes the MAC protocol / IP / UDP / RTP, and then transfers it to a codec implemented by software or hardware. , MPEG video data.
[0006]
The following are known technologies applicable to such LAN communication.
[0007]
[Patent Document 1] JP-A-7-95260
[0008]
In Patent Literature 1, the data processing load on the CPU is reduced by the communication control device 101 that performs the processing of FIG. 9B with the configuration of FIG. 9A.
In this case, the communication control device 101 is connected to the application program 107 and the packet processing routine 109 via the communication interface 105.
The data received by the transmission / reception processing unit 111 is stored in the shared buffer 113 (Step S1). Here, the switching control unit 115 detects the packet type (step S2), and distributes the data to the protocol processing units 117a to 117c or the packet processing routine 109 accordingly (steps S3, S4, S5).
[0009]
[Problems to be solved by the invention]
Here, the communication between the host systems 30 and 40 described with reference to FIG. 7 is first considered. In this case, since the hosts 31 and 41 are in charge of the protocol processing, the host 31 and 41 are excellent in versatility that can easily cope with various protocols, but the processing load on the host increases as the host tries to cope with various protocols. In addition, a higher performance host CPU is required, which leads to an increase in the cost of the apparatus.
[0010]
For example, in the host system 30, an interrupt from the network interface card 32 to the host 31 occurs every time a received packet arrives. Therefore, a large number of packets are transmitted from the host system 40 or another host system connected to the LAN. In the case of receiving the packet obtained by IP packetizing the above-described MPEG stream, a very large number of packets are continuously received, and the interrupt (interruption processing) to the host 31 frequently occurs in a short time. Will be done. At the same time, depending on the implementation, it is necessary for the host 31 to access the network interface card 32 and read out the packet each time the packet reception interrupt process is performed.
The host 31 also analyzes the header of the packet read from the network interface card 32, performs necessary protocol processing, and performs processing for transferring the packet to an application program or a codec. The host 31 also needs to perform application programs and codecs (when the codec is executed by software by the host CPU).
[0011]
Therefore, when a data packet having a very high bit rate of the real-time stream is received, or when a large number of packets are received even if the data stream is not a real-time stream, the host CPU is frequently interrupted. Causes a problem that other tasks cannot be executed.
The host 30 performs a wide range of protocol processing from the second layer to the seventh layer in the OSI reference model, that is, the data link layer, the network layer, the transport layer, the session layer, the presentation layer, and the application layer. This also increases the processing load on the host.
In order to solve these problems, a high-performance and expensive CPU is required, which causes a problem such as an increase in the cost of the apparatus.
[0012]
Such an increase in the processing load on the host 30 not only increases the cost of the device but also increases the network load.
It is assumed that the model of FIG. 7 is extended and three host systems 30, 40, and 50 are connected by LAN as shown in FIG. The host system 50 includes a host 51 and a network interface card 52 similarly to the host systems 30 and 40.
[0013]
Now, it is assumed that the host system 30 is receiving a relatively large amount of packets from the host system 40 as shown as the communication DF1.
At this time, if a file transfer application is executed from the host system 50 to the host system 30, the host system 50 first makes an ARP (Address Resolution Protocol) request to obtain the MAC address of the host system 30 (communication DF2). However, the host system 30 cannot respond to the ARP request from the host system 50 until the protocol processing for the packets already received from the host system 40 is completed.
[0014]
For this reason, the host system 50 that cannot obtain an ARP response may end up having to time out the application.
This means that the communication request from the host system 50 is rejected for the host system 30 communicating with the host system 40, and the host system 50 sends a retransmission request to the Redundant packets are generated, causing an increase in network load.
However, since the retransmission processing is assumed to be a normal processing on the Ethernet network, the retransmission processing itself is not a problem. The problem is that the increase in network load is proportional to the number of host systems that communicate.
In a normal LAN system, the number of communicating hosts is not limited to three as in the model of FIG. 8, but it is generally assumed that a larger number of host systems are connected to the LAN. Therefore, it can be easily imagined that the network load due to the retransmission processing is considerably increased.
[0015]
Still another problem is that control packet reception processing, such as the above-mentioned ARP or RARP (Reverse Address Resolution Protocol), which is necessary for communication, is suspended by other packet processing during reception, and processing cannot be performed. .
Although the above-described example is the RTP / UDP / IP / MAC protocol processing of the MPEG stream, the same problem arises when TCP processing such as file transfer or remote login is required.
[0016]
As described above, in the conventional communication method, in particular, in the reception processing of the host, the problem that the task and the arithmetic processing other than the reception processing cannot be executed due to an increase in the processing load of the host, or the processing other than the reception processing is delayed There is a problem that the retransmission process occurs due to the important packet not being processed preferentially, and the network load increases. Further, if a method of increasing the speed of the host CPU is adopted to solve these problems, there is a problem that the cost of the apparatus is increased.
[0017]
Further, in the technology described in Patent Document 1, reduction of the CPU processing load is realized, but in this technology, the CPU performs the processing of the application program 107 and the packet processing routine 109, and the protocol processing is performed by communication. It is assumed that the operation is performed on the control device 101 side.
In other words, this is different from a system in which the host CPU takes charge of the protocol processing in order to expand versatility, and the above problem in a system in which the host CPU takes charge of the protocol processing is not naturally solved. In particular, this is not for reducing the processing load on the CPU including the protocol processing when the packet reception is concentrated.
[0018]
[Means for Solving the Problems]
The present invention has been made in consideration of the above problems, and has as its object to reduce, for example, a processing load on a host system and a network load in a communication processing method on a LAN.
[0019]
For this reason, the communication device of the present invention is a communication device having processing means for executing data processing according to a predetermined protocol on a received data packet by receiving a reception notification. Discriminating means for discriminating whether the received data packet is an important packet or a non-significant packet according to the included information, and if the discriminating means discriminates the packet as important, the processing means is unconditionally And a control unit that issues the reception notification to the processing unit after a predetermined condition is satisfied when the determination unit determines that the packet is the insignificant packet.
In this case, the predetermined condition is that the amount of data determined as the non-important packet is equal to or larger than a set value.
Further, the discriminating means includes table information storing header information of the important packet or the non-important packet, and compares the header information of the received data packet with the table information so that the important packet and the non-important packet are compared. An important packet is determined.
The apparatus further includes a second processing unit that performs data processing on the real-time stream packet, wherein the determining unit further determines whether the received data packet is a real-time stream packet, and the control unit determines whether the received data packet is a real-time stream packet. If the data packet is determined to be the real-time stream packet, the received data packet is transferred to the second processing means without issuing the reception notification.
[0020]
Further, the communication device of the present invention, by receiving the reception notification, performs first processing means for performing data processing on the received data packet in accordance with the first protocol, and On the other hand, a communication device having second processing means for executing data processing according to a second protocol. And determining means for determining whether or not the received data packet is a real-time stream packet according to information included in the data packet; and determining that the received data packet is not a real-time stream packet. Issue a reception notification to the first processing means, and when the discrimination means determines that the packet is a real-time stream packet, without issuing a reception notification to the first processing means, Control means for transferring the packet to the second processing means.
[0021]
The communication method of the present invention is a communication method for receiving a data packet and executing data processing according to a predetermined protocol on the received data packet, and according to information included in the data packet, A first step of determining whether the received data packet is an important packet or a non-important packet; and, if the received data packet is determined to be the non-important packet in the first step, after the predetermined condition is satisfied, the A second step of executing data processing of the data packet determined to be an important packet.
In this case, the predetermined condition is that the amount of data determined as the non-important packet is equal to or larger than a set value.
[0022]
That is, in the present invention, for example, in the host system, between the network interface card and the host (processing means), the judging means and the control means having the above configuration are provided. Alternatively, the communication method is executed as reception processing to be passed to the host processing.
In the communication device (or communication method), for example, the header of a received packet is analyzed, the protocol of the packet is determined, and the determined result is an important packet (for example, a control packet or a packet requiring real-time performance). In the case of, the host is notified to perform reception processing with higher priority.
As a result, the host (processing means) does not need to perform header processing, and can perform reception processing without burying packets requiring priority processing in packets with low processing priority.
Furthermore, especially when the received packet is a real-time stream packet, the packet is transferred to the second processing means corresponding to the real-time stream packet without passing through the host (processing means / first processing means).
[0023]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, as the embodiments of the present invention, first to third embodiments of communication control when a plurality of host systems are enabled to perform LAN communication will be described.
[0024]
<First embodiment>
The first embodiment will be described with reference to FIGS.
In FIG. 1, two host systems 10 and 20 are illustrated, and communication of various data packets is performed between the host systems 10 and 20 using Ethernet (registered trademark) as a data transfer medium.
The data packets to be communicated include texts, images, control commands, and the like as various files processed by the operation of the application programs in the host systems 10 and 20, and ARP, RARP, and ICMP (Internet Control) by communication protocol processing. For example, a control packet such as Message Protocol, and a real-time stream packet such as MPEG-based compressed moving image data are assumed.
[0025]
The hosts 11 and 21 in FIG. 1 are main information processing units including hardware such as a CPU and a RAM, and include application software started by the CPU and programs for various protocol processes. The hosts 11 and 21 function as a packet data transmission processing unit and a reception processing unit as described above with respect to LAN communication.
The host systems 10 and 20 refer to computer systems including the hosts 11 and 21 and the network interface cards 12 and 22.
[0026]
In the host system 10, a packet parser 13 is provided so as to be interposed between the host 11 and the network interface card 12. The packet parser 13 corresponds to a determination unit and a control unit of the communication device according to the present invention.
The same applies to the packet parser 23 in the host system 20.
[0027]
The codec 14 is a unit that performs packet processing (packet encoding / decoding) of real-time stream data such as MPEG moving image data, and is mounted on, for example, an apparatus including the host system 10 as a module separate from the host 11. However, an example in which the codec 14 is installed as software in the host 11 is also conceivable.
The same applies to the codec 24.
[0028]
Such communication between the host systems 10 and 20 is performed as follows.
Now, assume that a packet transmission DF is performed from the host system 20 to the host system 10.
On the host system 20 side, the host 21 generates a packet or generates a packet of an MPEG real-time stream by the codec 24. At this time, the host 21 also performs packetizing processing and header processing of each protocol such as UDP / IP / MAC.
Thereafter, the host 41 transfers the processed packet to the network interface card 42 by bypassing the packet parser 23, and the network interface card 42 transmits the packet to the host system 30 via the transmission line 100. That is, the transmission processing is the same as the conventional method described with reference to FIG.
[0029]
On the other hand, in the host system 10 on the receiving side, a data packet received from the transmission line 100 by the installed network interface card 12 is first passed to the packet parser 13.
Although the processing of the packet parser 13 will be described in detail later, header analysis and protocol determination are performed on the received data packet, and as a result, important packets and non-important packets are determined. Then, based on the packet determination result, an interrupt (reception notification interrupt request) for the packet received at a predetermined time is issued to the host 11.
The host 11 reads a data packet from the packet parser 13 according to the interrupt.
The host 11 processes the MAC protocol / IP / UDP / RTP for the packet read from the packet parser 13 by the protocol processing software and the host CPU, and then executes a predetermined application (if the packet is a real-time stream packet, the codec 14 ).
[0030]
Here, the packet parser 13 discriminates the important packet and the non-important packet as described above, and notifies the host 11 of the interrupt at different timings. This prevents the processing load on the CPU of the host 11 from becoming excessive even when packet reception is concentrated.
FIG. 2 shows a configuration example of the packet parser 13 for this purpose.
[0031]
The packet parser 13 includes a control unit 1, a setting value register 2, a header table 3, a header analysis unit 4, and a RAM 5, as shown in FIG.
The control unit 1 and the header analysis unit 4 can be formed as dedicated hardware, or can be realized by software in a general-purpose processor. The setting value register 2, the header table 3, and the RAM 5 are formed by memory elements and registers.
[0032]
The control unit 1 performs interrupt issuance to the host 11, management / access control of the RAM 5, control of transfer of packet data to the host 11, and the like. That is, at the time of packet reception, write control for buffering in the RAM 5 (RAM pointer control), interrupt issuing processing based on information from the header analysis unit 5 and information in the setting value register 2, and a transfer request from the host 11 are received. If there is, read and transfer of packet data from the RAM A5 are performed.
[0033]
The set value register 2 stores a certain set value th by processing from the host 11, for example. The set value th is used as a condition for issuing an interrupt relating to an unimportant packet described later.
The header table 3 stores, for example, header information of a packet to be regarded as an important packet by processing from the host 11 or the like. For example, a header for identifying a protocol to be received and processed with higher priority, such as ARP, RARP, ICMP, and VLAN, is stored.
[0034]
The RAM 5 is used as a reception buffer, that is, the reception packet data transferred from the network interface card 12 is buffered.
In this case, in the RAM 5, addresses 0 to N are set as the first buffer area 5 a and addresses N + 1 to M are set as the second buffer area 5 b by, for example, area division setting, and are managed by the control unit 1.
Then, the packet data transferred from the network interface card 12 is written in the first buffer area 5a.
The second buffer area 5b is an area for storing a non-important packet that waits for an interrupt to be issued until a predetermined condition is satisfied.
[0035]
The header analysis unit 4 analyzes the header of the data packet received and stored in the first buffer of the RAM 5 to determine whether the data packet is an important packet or a non-important packet. Specifically, the header of the received data packet is compared with the header of an important protocol such as ARP stored in the header table 3 to determine whether or not the header is applicable. If the packet corresponds to the header of the important packet stored in the header table 3, the packet is regarded as an important packet. Otherwise, it is a non-important packet. Information on the determined important packet / insignificant packet is supplied to the control unit 1.
[0036]
The process at the time of packet reception by the packet parser 13 will be described with reference to FIG.
When the packet data transmitted via the transmission path 100 is received by the network interface card 12 and transferred to the packet parser 13, the control unit 1 stores the packet data in the first buffer area 5a of the RAM 5. Control. For example, the control unit 1 performs write control while circulating the write pointer for the first buffer area 5a in the range of addresses 0 to N. As a result, the addresses 0 to N as the first buffer area 5a are used in the ring buffer format, and the data of the received packet is buffered to each address.
Although not shown in FIG. 3, this process is executed in response to the packet reception.
[0037]
When the packet data is buffered with the reception of the packet, first, in step F101 in FIG. 3, the header analysis unit 4 analyzes the header of the captured packet data, and refers to the information in the header table 3 as described above. It is determined whether the packet data is an important packet.
This determination result information is passed to the control unit 1, and the control unit 1 branches the process in step F102 according to the determination result information.
That is, if the packet is an important packet, the process proceeds to step F103, and an interrupt notification to the host 11 accompanying the packet reception is executed unconditionally.
Then, the process proceeds to step F107.
[0038]
On the other hand, if the packet is an insignificant packet, the process proceeds to step F104, where the received packet data, that is, the data stored in the first buffer area 5a, is transferred to the second buffer area 5b to wait for an interrupt.
The control unit 1 also performs write control on the second buffer area 5b while circulating the write pointer in the range of addresses N + 1 to M, and the addresses N + 1 to M as the second buffer area 5b are in a ring buffer format. To be used.
[0039]
Then, in step F105, it is confirmed whether or not the amount of packet data stored in the second buffer area 5b has exceeded the set value th stored in the set value register 2. If not, the process proceeds directly to step F107. That is, no interrupt is issued at this time.
[0040]
For the determination in step F105, the control unit 1 only needs to provide an internal counter and count the amount of data stored in the second buffer area 5b. For example, when data is moved to the second buffer area 5b in step F104, a value corresponding to the data amount is counted up.
Then, in step F105, a process of comparing the data amount counted at that time with the set value th may be performed.
[0041]
At some point, when the process proceeds to step F105, if it is determined that the amount of packet data stored in the second buffer area 5b has exceeded the set value th stored in the set value register 2, the process proceeds to step F106. Then, an interrupt notification is sent to the host 11. Then, the process proceeds to step F107.
In this case, since the interrupt notification is performed for the packet data accumulated in the second buffer area 5b, that is, the data to be counted, the counter is reset.
[0042]
In step F107, the existence of the next received packet is confirmed, and if there is, the flow returns to step F101 to perform the same processing. If the packet does not exist, the processing of FIG. 3 is terminated, and the next packet reception is awaited.
[0043]
If an interrupt notification is made in step F103 or F106, a transfer request for packet data is sent from the host 11 at a later point in time. In that case, the control unit 1 performs control to read the requested packet from the RAM 5 and transfer it to the host 11.
[0044]
The following operation is realized in this example by the processing of FIG.
For example, it is assumed that control packets important for communication processing, such as ARP, RARP, ICMP, and VLAN, and real-time stream packets such as MPEG data are set as important packets in the header table 3. Then, when a control packet or a real-time stream packet is received, the packet parser 13 unconditionally notifies the host 11 of an interrupt.
On the other hand, when packet data that is not regarded as important, such as a text data file or web access data, is received, they are temporarily stored in the second buffer area 5b until the total data amount exceeds the set value th. Waits for an interrupt notification. When the total data amount exceeds the set value th, it is determined that the interrupt notification condition is satisfied, and an interrupt notification in step F106 is performed.
[0045]
As a result, the concentration of interrupts on the host 11 is reduced and the number of accesses to read the received packet from the host 11 to the RAM 5 is reduced as compared with the conventional method of notifying the host 11 every time a received packet arrives. The processing load is reduced.
Also, by separating the interrupt notification of the important packet reception and the interrupt notification of the non-important packet reception, the processing of the important packet sent during the reception of a certain packet does not have to be waited, and is caused by the delay caused by the wait. Also, an improvement effect against the increase in the network load, which has been performed, can be obtained.
Of course, an important control packet or a packet requiring real-time processing is notified to the host so as to preferentially perform reception processing, so that the reliability of the reception processing of the important packet is increased, which is suitable for a communication system.
Eliminating the load of the header analysis processing by the host 11 also contributes to reducing the processing load on the host 11.
For these reasons, it is not necessary to employ an expensive CPU having a high processing capability as the CPU of the host 11, which is effective in reducing the cost of the apparatus.
[0046]
<Second embodiment>
A second embodiment will be described with reference to FIGS.
FIG. 4 shows a system configuration in the case of the second embodiment. Since the basic configuration is the same as that of FIG. 1, the same reference numerals are given to the respective portions, and the duplicate description will be avoided.
In the case of FIG. 4, on the host system 10 side, it is assumed that the codec 14 is configured as a separate module from the host 11, and the packet parser 13 transfers the received data to the codec 14 without the host 11 intervening. (Data transfer system circuit).
[0047]
The internal configuration of the packet parser 13 is basically the same as that of FIG. 2 and is not shown again, but the control unit 1 has a data transfer control function for the codec 14.
The header analysis unit 4 also determines whether or not the packet is a real-time stream packet in addition to the determination of an important packet or an unimportant packet.
The same applies to the codec 24 and the packet parser 23 on the host system 20 side.
[0048]
FIG. 5 shows a process performed by the packet parser 13 when receiving a packet.
Note that the same processing steps as those in FIG. 3 are given the same step numbers, and detailed descriptions thereof will be omitted.
In this case, at the time of packet reception, the header analysis of the packet is performed in step F101, and the process is branched in step F102 based on the result of discriminating the important packet / insignificant packet. If it is determined that the packet is an insignificant packet, the control unit 1 performs the same processing as that described with reference to FIG. 3 as steps F104, F105, and F106. That is, until the amount of data accumulated in the second buffer area 5b becomes equal to or larger than the set value th, the interrupt notification is waited, and when the data amount becomes equal to or larger than the set value th, the interrupt notification is collectively performed.
[0049]
If it is determined in step F102 that the packet is an important packet, the control unit 1 further branches the process in step F108 depending on whether or not the data packet is a real-time stream packet. Information on whether or not the packet is a real-time stream packet is supplied to the control unit 1 as discrimination information from the header analysis unit 4. If the packet is not a real-time stream packet, that is, if it is a control packet such as ARP, RARP, ICMP, or VLAN, the process proceeds to step F103, and the control unit 1 unconditionally notifies the host 11 of the interrupt notification as in the example of FIG. I do.
[0050]
On the other hand, if the received packet is a real-time stream packet such as MPEG data, the process of the control unit 1 proceeds to step F109, where the data of the received packet is transmitted to the codec 14 (the data buffered in the first buffer area 5a). To transfer.
[0051]
In other words, in the second embodiment, when a real-time stream packet is received, the received data is transferred from the packet parser 13 directly (without the intervention of the host 11) to the codec 14 without the interrupt notification to the host 11. Will be done.
As a result, it is possible to further reduce an interruption to the host 11 every time a received packet arrives, and to eliminate the access from the host 11 to the packet parser 13 associated therewith. Therefore, the processing load of the host 11 is reduced according to the first embodiment. It is further reduced compared to the form.
In particular, receiving a packet in which real-time stream data such as an MPEG moving image data stream is IP-packetized means that a large amount of packets are continuously received, and the processing load on the host 11 for this is large. Therefore, the fact that the host 11 does not need to perform the processing for the real-time stream is very effective in reducing the processing load on the host 11.
Conversely, even if there is another task being executed in the host 11 or other packet reception processing, the reception processing can be reliably performed on the real-time stream packet, and the reliability of communication can be improved. Will also help reduce network load.
[0052]
In the processing example of FIG. 5, when a real-time stream packet is received, the header analysis unit 4 sends to the control unit 1 the determination result information indicating that the packet is an important packet and a real-time stream packet. As a result, the control unit 1 branches the process depending on whether or not the packet is a real-time stream packet among the important packets in steps F102 and F108 in FIG. However, the discrimination process in the header analysis unit 4 may discriminate the real-time stream packet from the important packet. In that case, the control unit 1 may perform the processing branch of step F108 and then perform the processing branch of step F102.
[0053]
<Third embodiment>
A third embodiment will be described.
The system configuration in the case of the third embodiment is the same as that of FIG. 4, that is, the codec 14 is constituted by a module different from the host 11, and the packet parser 13 converts the received data to the codec without the host 11. 14 (data transfer circuit).
[0054]
The internal configuration of the packet parser 13 is basically the same as that of FIG. 2 described above, and the control unit 1 has a data transfer control function for the codec 14 as described in the second embodiment.
However, the header analysis unit 4 does not discriminate between important packets and non-significant packets, but simply judges whether or not a received packet is a real-time stream packet.
The RAM 5 does not need to be divided into the first buffer area 5a and the second buffer area 5b, and the entire area of the RAM 5 is simply used as the buffer area.
The same applies to the codec 24 and the packet parser 23 on the host system 20 side.
[0055]
FIG. 6 shows a process performed by the packet parser 13 when receiving a packet.
Note that the same processing steps as those in FIG. 5 are denoted by the same step numbers, and detailed description will be omitted.
In this case, at the time of packet reception, the header of the packet is analyzed in step F101. At this time, depending on the analysis by the header analysis unit 4, it is determined only whether the received packet is a real-time stream packet. Therefore, the header table 3 only needs to set the header information only for determining the real-time stream packet.
[0056]
Then, in step F108, the control unit 1 branches the processing based on the determination information from the header analysis unit 4 based on whether or not the packet is a real-time stream packet.
If the packet is not a real-time stream packet, the process proceeds to step F110, and the control unit 1 performs an interrupt notification to the host 11.
On the other hand, if the received packet is a real-time stream packet such as MPEG data, the process of the control unit 1 proceeds to step F109, and transfers the data of the received packet (data buffered in the RAM 5) to the codec 14. .
[0057]
That is, in the third embodiment, when a real-time stream packet is received, the received data is directly transferred from the packet parser 13 to the codec 14 (without the intervention of the host 11) without performing an interrupt notification to the host 11. Will be done.
If the packet is not a real-time stream packet, an interrupt notification is sent to the host 11 without distinguishing the important packet / insignificant packet described above.
[0058]
Therefore, in the case of the third embodiment, the effect obtained in the first and second embodiments by the interrupt control based on the discrimination between the important packet and the non-important packet cannot be obtained, but the processing load is the largest. The host 11 does not need to process the real-time stream packet, and in that respect, the processing load on the host 11 is reduced as compared with the conventional method. It also facilitates a reduction in network load.
[0059]
<Modification>
Although the embodiment of the present invention has been described above, the present invention is not limited to this, and various modifications can be considered. Hereinafter, an example will be described.
[0060]
The present invention can be applied not only to a wired LAN, for example, but also to a wireless LAN using a wireless transmission path (communication medium). Of course, the present invention can be applied not only to a communication system called a LAN, but also to various communication systems.
In the description of the embodiment, the two host systems 10 and 20 are used as models. However, it goes without saying that the number of host systems to be connected can actually be larger. The host system is not limited to a personal computer or the like, but may be a larger information processing device, a PDA or other small information processing device, an audio / video device having a communication function, a portable or stationary telephone, a home appliance. Various devices such as devices can be considered.
[0061]
Further, the header information stored in the header table 3 in the configuration example of FIG. 2 can be changed according to the application, and can be applied to various applications.
That is, the important packet is not limited to the control packet or the real-time stream packet, and can be appropriately changed according to the application installed in the host system. This makes it possible to provide the same expandability from the viewpoint of versatility that enables the host 11 to handle various protocols, which is an advantage when protocol processing is performed by the host 11.
[0062]
In particular, in the case of the first embodiment, the real-time stream packet may be a non-important packet. When the host system (or the application) is a device that attaches importance to the real-time processing of the real-time stream packet, it is appropriate to make the real-time stream packet an important packet. In the case of (application), by making the real-time stream packet an unimportant packet, it is suitable for reducing the processing load on the host 11.
[0063]
Further, in the above example, the header information of the important packet is set in the header table 3, but the header information of the packet to be regarded as the non-important packet may be set. Alternatively, both the header of the important packet and the header of the non-important packet may be set separately.
In any case, in the first and second embodiments, information for enabling discrimination between an important packet and a non-important packet is set. In the second and third embodiments, a real-time stream is set. What is necessary is just to set the header information which enables to determine whether the packet is a packet or not.
[0064]
In the first and second embodiments, the condition of the interrupt notification regarding the unimportant packet is such that the accumulated data amount exceeds the set value th. However, the condition setting may be variously considered.
For example, for each packet data stored in the second buffer area 5b, the time lapse from the time of reception is managed, and every time a predetermined standby time elapses, it is determined whether or not another important packet is received. If not, an interrupt notification is performed, and if an important packet is received, the apparatus waits for a further fixed time.
Also, for a packet whose elapsed time from reception exceeds a certain time value, an interrupt notification may be made at that time.
[0065]
The set value th may be the number of packets instead of the data amount. In other words, in this example, an interrupt notification is issued for a non-important packet when the number of packets as the set value th is accumulated.
In addition, the process of issuing an interrupt of an insignificant packet may be performed in consideration of both the number of packets or the set value as the data amount and the elapsed time from reception.
[0066]
In the first and second embodiments, the RAM 5 is divided into the first buffer area 5a and the second buffer area 5b. However, the first buffer and the second buffer are not divided by such area but by two memory chips. May be set.
In addition, non-important packets may be managed by a method such as address management without performing such buffer division. That is, if the address where the important packet is stored in the RAM 5 and the address where the non-important packet waiting for the interrupt is stored are managed by the control unit 1, the data transfer in the RAM 5 can be made unnecessary.
[0067]
Since the packet parser 13 performs header analysis, the analysis information may be transmitted to the host 11. For example, when notifying an interrupt of an important packet, by notifying the host 11 in detail whether the received packet is ARP or RARP, or even a header analysis result such as ICMP or VLAN, the processing load on the host 11 is further increased. Can also be reduced.
[0068]
【The invention's effect】
As can be understood from the above description, according to the present invention, for example, the function as the packet parser (determination means and control means, first step) between the network interface card and the host (processing means) in the host system And a second step). The communication device (or communication method) of the present invention analyzes the header of the received packet, determines the protocol of the packet, and if the determined result is an important packet, unconditionally, ie, preferentially performs the reception process. (Interrupt) to the host (processing means) so that the operation is performed.
As a result, the number of interrupts to the host (processing means) is reduced even when reception is concentrated, and the host (processing means) does not need to perform header processing. Thus, there is an effect that the receiving process can be surely performed without burying a packet that requires the processing in a packet with a low processing priority.
This effect also brings about effects such as a reduction in the processing load on the host (processing means) and a reduction in the network load.
Therefore, it is not necessary to use an expensive CPU having a high processing speed as a host (processing means), and the cost of the apparatus does not increase.
[0069]
In addition, a packet determined to be an unimportant packet can be appropriately handled by notifying the host of reception when the data amount of the data regarded as the unimportant packet becomes equal to or larger than a set value.
Also, the packet information is provided with table information storing header information of an important packet or an unimportant packet, and the header information of a received data packet is checked against the table information to determine an important packet and an unimportant packet. The discrimination process is easy, and important packets / insignificant packets can be set flexibly by setting the table information. Therefore, an important communication packet can be set according to the use, operating environment, performance, and other circumstances of the host system, and an optimum communication environment can be constructed.
[0070]
If the received packet is a real-time stream packet, the packet is transferred to the second processing unit corresponding to the real-time stream packet without passing through the host (processing unit / first processing unit). The processing related to the real-time stream packet becomes unnecessary, and the processing load on the host can be further reduced. Of course, this also promotes the effect of reducing the network load.
[Brief description of the drawings]
FIG. 1 is a block diagram of a communication system according to a first embodiment of this invention.
FIG. 2 is a block diagram illustrating a configuration of a packet parser according to the embodiment;
FIG. 3 is a flowchart of a packet reception process according to the first embodiment.
FIG. 4 is a block diagram of a communication system according to a second embodiment.
FIG. 5 is a flowchart of a packet reception process according to the second embodiment.
FIG. 6 is a flowchart of a packet reception process according to the third embodiment.
FIG. 7 is a block diagram of a conventional communication system.
FIG. 8 is an explanatory diagram of an increase in network load in a conventional communication system.
FIG. 9 is an explanatory diagram of a conventional communication device.
[Explanation of symbols]
1 control unit, 2 setting value register, 3 header table, 4, header analysis unit, 5 RAM, 5a first buffer, 5b second buffer, 10, 20 host system, 11, 21 host, 12, 22 network interface card, 13,23 packet parser, 14,24 codec

Claims (7)

A communication device having processing means for performing data processing according to a predetermined protocol on a received data packet by receiving a reception notification,
Determining means for determining whether the received data packet is an important packet or an unimportant packet, according to information included in the data packet;
If the discrimination means determines that the packet is the important packet, it issues the reception notification unconditionally to the processing means, and if the discrimination means determines that the packet is the non-important packet, a predetermined condition is satisfied. Control means for issuing the reception notification to the processing means after being executed;
A communication device comprising: The communication device according to claim 1, wherein the predetermined condition is that an amount of data determined as the non-important packet is equal to or larger than a set value. The discriminating means includes table information that stores header information of the important packet or the non-important packet, and compares the header information of the received data packet with the table information, so that the important packet and the non-important packet are compared. The communication device according to claim 1, wherein a packet is determined. A second processing unit that performs data processing on the real-time stream packet,
The determining means further determines whether the received data packet is a real-time stream packet,
The control means, when the received data packet is determined to be the real-time stream packet, transfers the received data packet to the second processing means without issuing the reception notification. The communication device according to claim 1, wherein: A first processing unit that executes data processing according to a first protocol on a received data packet by receiving the reception notification;
A communication device having second processing means for performing data processing according to a second protocol on the received real-time stream packet,
Determining means for determining whether or not the received data packet is a real-time stream packet according to information included in the data packet;
If the discrimination means determines that the packet is not a real-time stream packet, it issues a reception notification to the first processing means. If the discrimination means determines that the packet is a real-time stream packet, the first processing means Control means for transferring the received data packet to the second processing means without issuing a reception notification to the means,
A communication device comprising: A communication method for receiving a data packet and performing data processing on the received data packet according to a predetermined protocol,
A first step of determining whether the received data packet is an important packet or a non-important packet, according to information included in the data packet,
A second step of performing data processing of the data packet determined as the non-important packet after a predetermined condition is satisfied when the packet is determined to be the non-important packet in the first step. A communication method characterized by the above-mentioned. 7. The communication method according to claim 6, wherein the predetermined condition is that an amount of data determined as the non-important packet is equal to or larger than a set value.

Images