JP3459165B2 - Packet processing method and network architecture - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a network
The present invention relates to a packet processing method and a network architecture for establishing a connection between packets and sequentially processing a protocol hierarchically added to a packet received in a reception buffer from a lower layer to an upper layer.

[0002]

2. Description of the Related Art Conventionally, a standard network architecture of an OSI (Open System Interconnection) environment, which is built into a node of a network, has a structure in which protocols are hierarchically arranged, and ISO (International Organization) is adopted.
(zation for Standardization) etc., the positioning of each layer, functions, interfaces between layers, etc. are specified in detail.

The OSI reference model defined by ISO includes
It has a seven-layer structure including a physical layer, a data link layer, a network layer, a transport layer, a session layer, a presentation layer, and an application layer from the bottom.

The packet format data sent to the network is added with a protocol as control information for each layer as a header, and the network architecture of each node is from the network. The protocol included in the packet header is hierarchically processed in each layer, and the data included in the packet is processed.
The data is stored in the data buffer and the response packet to the received packet is transmitted to the network.

For example, the internet protocol layer (hereinafter referred to as "IP") is located in the network layer and mainly relays via one or a plurality of communication networks, and the node from the start point to the end point. Data communication up to.

[0006] The upper layer of IP (transport layer)
The transfer control protocol layer (hereinafter referred to as TCP) provides a transparent simultaneous bidirectional transfer function between the end point and the end point of the node that does not participate in the route selection and the relay function.

In order to incorporate TCP into the network architecture, processing for one connection (logical communication path for transferring data) can be performed by one TCP task, and N connections are made. For (N +
1) Start up the TCP tasks. The other one of the TCP tasks is waiting as the master for establishing the next connection.

In the IP, one task sequentially processes the header processing of the received packet received, identifies the corresponding TCP task, and deposits it in the waiting TCP task. Therefore, it always exists as one task regardless of the number of established connections.

Each TCP task processes a received packet received from a client task which has established a connection, and sends a response packet to the client task of the other party. Since the response packet to be transmitted at this time needs to pass through only one existing IP task, each TCP task acquires the right to use the IP task.

The IP task waits for a call (hereinafter referred to as POST) from the TCP task or the data link layer task. POST is a system call of an operation system (hereinafter referred to as OS) used when processing is transferred between tasks in a multitask environment.

Here, the operations of the IP task and the TCP task will be briefly described. When the data link layer task receives the first packet from the client, the data link layer task identifies the IP task from the header, and POSTs the waiting IP task.

The IP task performs a protocol process related to the network layer from the header of the received packet, identifies the TCP task from the header, and POSTs to the waiting TCP master task.

When the TCP master task determines that the content of the received packet is a connection opening request, it processes the packet and sends a response packet to the client.
Normally, the first packet from the client is a packet requesting the opening of a connection, so after transmitting the response packet to the client, the timer is activated to wait for the last packet of the three-way handshake.

When the TCP master task receives the handshake packet from the client before the timer is up, it recognizes the establishment of the connection. If the received packet after the connection is established requires a service of the session layer or higher, a new master task is generated and POST is performed. In this example, FTP (File TransferPro)
tocol) Start the master task.

When the FTP master task establishes a control connection with the client, it uses the fork () function to create a copy of itself (FTP task). FTP
The task assigns a port to the TCP master task.

The TCP master task creates a copy of itself (TCP task) using the fork () function. TCP
The task establishes a data connection with the FTP task and manages the connection until it is disconnected from the client thereafter.

The TCP master task enters a waiting state in preparation for a connection opening request from the next client.

On the other hand, the TCP task processes the transport layer each time a packet is received, and POSTs the FTP task. The FTP task receives the data received from the TCP task and executes the command from the client included in the data.

The embedding processing described above is premised on the embedding of a round-robin type OS which operates completely preemptively.

[0020]

In the conventional network architecture, since a plurality of connections are processed by the same number of TCP tasks, a function for dynamically copying a plurality of processes for processing the connections (described above). did
fork () function is required. Since this function shares and executes one program by a plurality of tasks, there is a problem in that a highly functional OS that operates in a completely preemptive manner of a round robin system that synchronizes tasks is required.

Further, the protocol processing of each layer is operated as a separate task, and the tasks of the transport layer and above are copied according to the establishment of the connection, so that the number of tasks operating on the OS increases and There was also a problem in that the overhead amount of was increased.

Further, there is a problem that the work area on the memory increases because each task in the protocol processing is performed by a different task.

An object of the present invention is to provide a packet processing method and a network architecture that reduce the load on the network layer / transport layer incorporated in the network architecture and operate more efficiently. To do.

[0024]

In order to achieve the above object, in the network architecture of the present invention, a timer interrupt that enters every fixed time is used, and the network layer of the packet received in the receiving buffer is used. For the last received packet belonging to each connection, and the received packet batch processing means for collectively processing and accumulating the protocols related to the transport layer and the transport layer during the interrupt, and designating and notifying the upper layer protocol processing means from the contents of the protocol. And a transmission packet batch processing means for collectively transmitting the response packets to the network during the interruption.

Further, in order to achieve the same object, in the packet processing method of the present invention, at least the protocol processing relating to the network layer, the transport layer, the session layer, the presentation layer and the application layer is performed. One task, network layer, transport layer as one function, session layer, presentation layer, application layer as one function,
As a packet processing method that operates by giving its function a processing priority, as a concrete means, a message processing means by an operating system function that activates the processing means called based on the message, A first packet processing means called by the message processing means, extracting a received packet from the reception buffer, processing at least a protocol relating to the network layer and the transport layer, and accumulating until a predetermined amount is reached, and a session layer. ,
Second packet processing means for processing a protocol relating to the presentation layer and application layer and outputting it to an upper layer, and a second packet processing means designated by the first packet processing means.
The message for calling the packet processing means of the above is output to the message processing means and registered in the upper queue,
Socket processing means for calling the second packet processing means by referring to the upper queue in response to a call from the message processing means.

[0026]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described with reference to the drawings. Elements common to the drawings are given the same reference numerals. First Embodiment FIG. 1 is a block diagram showing the network architecture of the printer according to the first embodiment, which includes an interrupt processing operation on a single central processing unit 1 (hereinafter referred to as CPU 1) and Indicates a task.

Note that, as a node, as shown in FIG. 2, the description will be limited to the network printer 2 (hereinafter referred to as printer 2) connected to the line.

The network architecture is a network interface (NIC) as hardware for network connection.
rface controller) 3, TCP / IP unit 4 as a protocol processing means for the transport layer and network layer, and transmission / reception of data to / from a client, and manages a reception buffer 5a allocated on the memory 5. Do I
It consists of FC task 6.

The NIC 3 is activated when receiving a packet from the network, interrupts the CPU 1 to receive the received packet in the receive buffer 5, and is used when transmitting a response packet to the received packet.

The TCP / IP unit 4 is activated at the time of a timer interrupt, and collectively processes the protocol processing relating to the network layer and the transport layer of the received packet received by the receiving buffer 5 within the interrupt time to execute the memory 5 operation. Received packet batch processing means 4a for storing the received packet in the temporary buffer 5b allocated above and POSTing to the IFC (Interface Control) task 6, and the response stored in the transmission waiting buffer 5c allocated in the memory 5. There is provided a transmission packet batch processing means 4b for transmitting packets and the like to the network within the interruption time. TCP
The / IP unit 4 operates as one task instead of separate tasks for TCP and IP.

The IFC task 6 is a communication application identifying section 8 for identifying a communication application designated by the TCP / IP section 4, and a communication for processing a protocol relating to a session layer, a presentation layer and an application layer of a received packet. Application group 9,
The input / output management unit 11 inputs the reception data included in the reception packet from the communication application group 9 and outputs it to the printing unit 10.

FIG. 3 is a time chart when processing the network architecture shown in FIG. TCP / IP
The unit 4 is set to enter a timer interrupt every time time t has elapsed, and the protocol processing of the TCP / IP unit 4 is performed within the interrupt time. The time t is set to be larger than the interrupt processing time, and when the TCP / IP unit 4 finishes the interrupt processing, POST is performed, and thereafter, the packet transmission / reception processing by the NIC 3 and the packet by the IFC task 6 are performed until the next timer interrupt occurs. Processing, printing processing by the printing unit 10, and the like are performed.

When activated by a timer interrupt, the TCP / IP unit 4 processes all the packets that have been received by the reception buffer 5 within the interruption time, creates a response packet for the last packet, and stores it in the transmission waiting buffer 7b.
For example, if three packets A1 to A3 have been received from the client A in the reception buffer 5, the three packets A1 to A3 are processed by one timer interrupt process, and a response packet for the packet A3 is created and transmitted. Store in the waiting buffer 7b.

Similarly, when packets are received from a plurality of clients, a response packet is similarly created for the last packet received from each client and stored in the transmission waiting buffer 7b. That is, three packets A1 to A3 from the client A and one packet from the client B
If one packet B1 is received from the client C and two packets C1 and C2 are received in the receiving buffer 5, T
The CP / IP unit 4 sends a response packet for the packet A3 to the client A and a packet B to the client B.
A response packet for 1 and a response packet for packet C2 are created in the client C and stored in the transmission waiting buffer 7b.

The transmission process includes a pre-process for setting information necessary for transmission to the NIC 3, a transmission process performed by the NIC 3, and a post-process performed after the transmission is completed.

The first transmission pattern is when a response is made from the TPC / IP unit 4, a response at the time of a connection establishment sequence with the TPC / IP unit 4, a response to a data packet, and an ARP (Address Resolution Protocol). ) Is a transmission process that occurs immediately after receiving a received packet, such as a response to a physical address inquiry in (1), and is performed during the timer interrupt process described above.

That is, in the case of a simple response to the stream of data packets, the transmission is not immediately performed as described above, but the response packet is created after the processing of all the received packets is completed. For other responses,
Immediately create a packet for transmission and wait for transmission 7
It is stored in b and the transmission requests are collectively issued to the NIC 3 at the end of the timer interrupt processing.

By doing so, the timeout time (turnaround time) until the response to the transmitted packet is received is collectively managed (hereinafter, Max Time).
This is referred to as out processing), and the load on the timeout processing is reduced.

The timer processing in the architecture incorporation according to the present embodiment manages the timer information as a global table together with the information for managing each connection. The timeout value of the transmission packet, the timeout counter, etc. are saved in the table, the counter is decremented during the timer interrupt, and the timeout time is managed by the number of times the timer interrupt is entered. The timeout value is TCP /
It is defined by a multiple of time t at which the timer interrupt of the IP unit 4 enters, and is increased or decreased by a multiple of t.

As the second transmission pattern, the printing unit 10
Printer firmware and communication application group 9
There is a transmission process that occurs at any time from the communication application or the like.

In the conventional transmission by incorporating the architecture, when the TCP task and the IP task receive the transmission request from the IFC task 6, as shown in FIG. 4A, the transmission processing is immediately performed and the transmission processing is performed to the NIC 3. Requested.

However, in the transmission of the application, there is no timeout, or even if there is a timeout, it is very long, so it is not necessary to transmit it urgently. Therefore, in the present invention, IFC
A method of accumulating transmission requests of task 6 for a certain period of time and transmitting them all at once is adopted.

When a transmission request is generated in the communication application function and the printer firmware, the communication application function writes the transmission request in a specific table. In actual transmission, as shown in FIG.
The CP / IP unit 4 refers to the table information set by the communication application during the next timer interrupt.

When the received packet exists in the reception buffer 5 at the time of the timer interruption, the received packet is preferentially processed and a transmission request is made to the NIC 3 together with the response packet.

Next, the operation will be described. Packets randomly sent from the client are the NIC 3 of the first and second layer parts as shown by arrows (a) and (b) in FIG.
Is received in real time by. The NIC 3 interrupts the CPU 1 each time it receives a packet, and stores the packet in the reception buffer 5 by interrupt processing.

The TCP / IP unit 4 stores the protocol processing of the network layer and the transport layer in the receiving buffer 5 at the time, as shown by the arrow (c) in FIG. The processing is performed for all received packets, and the processed packets are stored in the packet temporary buffer 7a.
To store.

When the last packet in the series of received data streams is received, the TCP / IP unit 4 identifies the upper layer protocol (communication application) from the header and sets a flag for notifying it, and the arrow ( As shown in d), POST to IFC task 6
I do.

Flags are global variables, and TCP / I
It is used by the P section 4 to specify the upper layer (communication application) to which the received data stream should be passed.
It is passed from the P section 4 to the IFC task 6. The message passing function of the OS is used as the delivery method.

All communication applications are installed as a function of the IFC task 6, and the designated upper layer is
For example, assuming FTP9a, the identification unit 8 calls the FTP function, as indicated by the arrow (e) in FIG.

The FTP 9a receives the data stream received by the TCP / IP unit 4 from the received packet temporary buffer 7a and processes the contents. If necessary, the data is passed to the input / output management unit 11, as indicated by the arrow (f) in FIG. Since the data passed to the input / output management unit 11 is connection-specific, the connection-specific queue (hereinafter, I
Create an FC-EDIT receive queue).

The IFC-EDIT receive queue is IFC.
This is a queue for transferring print data and the like from the communication application FTP 9a of the task 6 to the EDIT (edit) task 10a of the printing unit 10 via the input / output management unit 11.
The EDIT task 10a of the printing unit 10 receives data from the IFC-EDIT reception queue as shown by an arrow (g) in FIG.

Here, the difference Rec between the reception processing time rA according to the conventional method (denoted as A) and the reception processing time rB according to the method of the present embodiment (denoted as B) is shown below.

First, the processing is limited to the TCP / IP unit 4, the time is limited to t, and the packet reception time is divided into the packet processing time and the OS processing time to be obtained. If the number of packets received in a unit time is n and the time for processing one packet is E, the packet processing time E (A,
In B), both A and B are E (A, B) = E × t × n (1) Further, the processing time of the OS is the task generation time and the task switch time. Is F, the time for creating and executing a child task by the fork () function or the like is F, and the average number of connections to be accessed at the same time is N, the task creation processing (generation for each connection establishment) time F (A) at a fixed time t. Is
Only A is applicable, and F (A) = F × N ... (2) In the time required for task switching (time from POST to task activation), In the conventional installation, if the target OS is PA and the multitask OS due to interrupt is PB, the IP task for 1 packet reception,
Since the TCP tasks are activated respectively, the task change processing times P (A) and P (B) at a constant time t are as follows: In A, the IP task and the TCP task are called every time one packet is received, so P (A) = 2 × PA × t × n (3) In B, the TCP / IP unit 4 is called once within the time t, so P (B) = PB. (4) Time r required for reception processing at a fixed time t in the conventional method
A is calculated from the equations (1), (2) and (3): rA = E (A, B) + F (A) + P (A) = Etn + FN + 2 (PA) tn (5) Timer interrupt is activated, and Assuming that the time until the processing of (1) is activated is I, the time rB required for the receiving processing at the constant time t of this method is as follows from (1) and (4): rB = E (A, B) + P (B) + I = Etn + PB + I ... (6) Therefore, the difference Rec between them can be expressed by the following equation from equations (5) and (6).

Rec = rA-rB = FN + 2 (PA) tn-PB-I ... (7) Therefore, the packet receiving time at the fixed time t is shortened by FN + 2 (PA) tn-PB-I by this embedding method. be able to.

Further, in the conventional embedding method, when the receiving side node receives one packet, it passes the IP task and immediately transfers the control to the TCP task. And the TCP task is 1
It is difficult to wait for the reception of subsequent packets and respond to them in bulk, because control must be transferred to another task when the processing of the packets is finished. Therefore, a response packet is returned to the client every time one packet is received.

The difference Res between the response processing time SA of the conventional method and the response processing time SB of the method according to the present embodiment can be obtained as follows.

SA = t × n × R + t × n × PA + t × n × S (8) SB = N × R + S (9) Formula ( 8) From (9), Res = SA-SB = tnR + tn (PA) -NR + (tn-1) S ... (10) n is the number of packets received per unit time, and t is adopted by the incorporation according to the present embodiment. Timer interrupt interval, PA is the task switch time of the target OS in the conventional installation, PB is the task switch time of the multi-task OS due to interrupts, N is the average number of connections to be accessed simultaneously,
R is the time required for response processing of one packet, S is NIC3
Is the sum of the transmission request processing time and the transmission completion interrupt processing time. Therefore, the response processing time is tnR + tn (PA) −NR + (tn by the built-in processing according to the present embodiment.
-1) It can be shortened by S.

Further, as shown in FIG. 7, in actual transmission, preprocessing for setting information necessary for transmission to the NIC 3, transmission processing performed by the NIC 3, and postprocessing performed after completion of transmission are required. Particularly, since the post-processing is started as an interrupt when the transmission is completed, it takes several hundred milliseconds for saving the registers, etc. In the conventional method, this interrupt is generated every transmission. In the method according to the present embodiment, the transmission request generated within the time t and the transmission of the response packet of the received data are simultaneously activated to the NIC 3, so that the pre-processing and the post-processing in the transmission are always once.

When a difference Sen between the processing time YA of the conventional embedding method and the processing time YB of the embedding method according to the present embodiment at a constant time t of the transmission pattern 2 is calculated, YA = t × Y × S. ···················································································································· ((12) -YB = (tY-1) S ... (13) Y is the number of transmission requests generated within a unit time, S is the transmission request processing time to the NIC 3 and transmission completion interrupt processing It is the sum of time.

When the packet is received within the fixed time t, the transmission process is performed together with the response packet.
YB becomes 0, and Sen = tYS. Therefore, the transmission time is (tY-
1) It can be shortened by S or tYS.

The time-out process for the packet to be transmitted in this transmission process is also performed during the timer interrupt as in the transmission pattern 1. In the conventional turn-around time-out process for transmitted packets, it is necessary to count the timer for each of all transmitted packets.
By transmitting a fixed number of packets at the same time as in the method according to the present embodiment, it is possible to group the packet timer management and reduce the processing time.

Conventionally, the TCP task remains activated while the connection is connected, and other tasks are executed in the time slice processing, and a constant amount of CPU resource is always used for the time slice processing. Therefore, the operating efficiency of protocol processing and printer firmware processing becomes very poor.

In the embedding method according to the present embodiment, interrupts using priority are used for job scheduling, and the round robin method by time slice is not adopted.

According to the first embodiment, TCP / IP
Since the unit processes all the packets received in the reception buffer within the interrupt processing time t, it is possible to reduce the overhead of the OS and allocate the CPU to other effective processing.

Second Embodiment In general, if there are many operations in the interrupt, the influence on other processing becomes large. Therefore, in the second embodiment, the processing in the interrupt is reduced as much as possible, and most of the processing is omitted. It is designed to be incorporated in the IFC task.

FIG. 5 is a block diagram showing the network architecture of the printer according to the second embodiment.
It comprises a reception interruption means 20, an interval timer 21, and an IFC task 22 as an interface control means.

The reception interrupt means 20 is activated when a packet is received, and the reception packet is received by the reception buffer 5a of the memory 5.
Each time it is stored in the IFC task 22, a message (message
-X) is output. Interval timer 21 is counter 2
Each time 1b counts up, it outputs a message (message-S) to the IFC task 22.

The IFC task 22 receives a message and performs a calling process on the message, and a message processing means 24.
The first packet stored in the packet temporary buffer 5b of the memory 5 by processing each protocol of the link layer (2 layers), the network layer (3 layers), and the transport layer (4 layers) until a predetermined amount is reached. A packet is taken out from the processing means 23 and the temporary packet buffer 5b and the session layer (5
Layer), presentation layer (6 layers), and application (7 layers) protocols, and the data is stored in the memory 5
Of the second packet processing means 30 for outputting to the data buffer 5d and the first packet processing means 23 for outputting a message (message-A) to the message processing means 24, In response to a call from the processing means 24, the socket processing means 28 that activates the second packet processing means 30 and the control module 32 in the IFC task that controls the cables such as Centronics and RS-232C.
It has and.

The first packet processing means 23 sends a message (m) from the reception interruption means 20 to the message processing means 24.
essage-X) is output, the received packet is taken out from the reception buffer 5a and the network driver 25 that performs the protocol processing of the data link layer (layer 2) and the network driver 25 processed it. A protocol stack (TCP / I) that processes each protocol of the network layer (3 layers) and the transport layer (4 layers) of the received packet and accumulates it in the packet temporary buffer 5b of the memory 5.
P, SPX / IPX, etc.) 26 and interval timer 2
From 1 to the message processing means 24, a message (message-
S) is activated, and timer processing means 31 for managing the connection is provided.

The second packet processing means 30 is an application (FTP, RPRINTER, etc.) that actually communicates with another node by using the protocol stack 26.
27 and a job control module 29 for managing the queue of the data processed by the application 27.

The socket processing means 28 provides a common interface for communication between the application 27 and the protocol stack 26, and is called by the protocol stack 26 to specify the application 27.
A socket (Low) 28a for outputting the message (message-A) to the message processing means 24, and an application based on the contents included in the message called by the message processing means 24.
Socket (High) 28b for selecting the option 27.

FIG. 6 is a block diagram showing the structure of the application FTP. Application FTP27a
Is composed of an initial processing unit 35 and a main processing unit 36.

In response to the initial request from the printer 1, the initial processing section 35 initializes the working in the FTP, etc. and registers the application with the socket (High) 28.
Perform on b.

The main processing 36 is a message (message-
A) analysis unit 37 and connection management unit 38 that is called when the status of the lower layer is notified by message-A
From a data analysis unit 39 referred to when processing the received packet, a command branching unit 40 called when the data analysis unit 39 determines that the command is a command, and a processing group 41 for each command. Become.

FIG. 7 is an explanatory diagram of the connection management table. "Connection state" for each connection,
"Processing command", "connection No", "transmission packet number", and "destination port No" are arranged. The connection management table is the connection management unit 3
8 and is referred to by the message analysis unit 37 when processing the received packet.

FIG. 8 is an explanatory view showing the relationship between the connection state and the set value, and FIG. 9 is a load calculation table showing the relationship between the number of processing steps of each processing means of the protocol stack and the load.
FIG. 10 is a flowchart showing the processing when the packet is retransmitted.

Next, the operation will be described. The packet randomly sent from the client receives a reception interrupt signal from the NIC 3 which is a network connection H / W, is input to the CPU 1, and is received in real time by the reception interrupt means 20 including a reception interrupt program.
It is stored in the reception buffer 5a. Each time one packet is stored, the reception interrupt means 20 outputs a message (message-X) to the IFC task 22 as indicated by an arrow (a) in FIG.

Message processing means 2 of IFC task 22
No. 4 senses a message addressed to the IFC task when idle. IF before message (message-X)
If there is a message sent to the C task, the IFC task 22 executes the processing first.

The message processing means 24 uses the message (mes)
sage-X), the network driver 25 is called as indicated by the arrow (b) in FIG. Network
The driver 25 performs a normal data link layer (2 layer) process, but can also handle a plurality of NICs 3 and a plurality of 2 layer protocols (eg, Ethernet, IEEE802.3, IEEE802.5, etc.).

When the network driver 25 completes the protocol processing of the data link layer, the arrow (c) in FIG.
As shown by, the upper layer is identified and the corresponding protocol stack 26, for example, TCP / IP 26a is called. T
Since the CP / IP 26a is created as one task, not TCP and IP as separate tasks, the overhead between TCP and IP is reduced.

The TCP / IP 26a is used for the network layer (3) of the packet received from the network driver 25.
Layer) and transport layer (4 layers) protocol processing, and the packet is stored in the packet temporary buffer 5 of the memory 5.
It is stored in b and the control is returned to the network driver 25.

The network driver 25 controls the IFC.
The task 22 returns to another task and waits for the reception of the next packet.

When the above process is repeated several times, a fixed amount of packets are accumulated in the packet temporary buffer 5b. TCP
/ IP 26a, when the number of packets reaches a certain amount, as shown by the arrow (d) in FIG. 5, for socket (Low) 28a,
Information for designating the corresponding application 27 is output to request receipt of the packet.

The socket (Low) 28a is a direct socket
(High) 28b is not called, and the message processing means 24 is sent to the message (messa) as shown by an arrow (e) in FIG.
ge-A) is output. By this processing, the message processing means 24 is activated, and Centronics, RS-232
The control module 32 in the IFC task that controls a cable such as C can be actuated.

The message processing means 24 uses a message (mess).
When sage-A) is received, the socket (High) 28b is called as indicated by an arrow (f) in FIG. Socket (High)
28b is a message as indicated by an arrow (h) in FIG.
An application, for example, FTP 27a is called based on the content of (message-A).

The FTP 27a is a packet temporary buffer 5
All the packets stored in b are processed, and the job control module 29 is called as indicated by an arrow (i) in FIG.

The job control module 29 stores the data of the packet in the data buffer 5d of the memory 5, as shown by the arrow (j) in FIG.

Here, the operation of the FTP will be described in detail with reference to FIGS. FTP27a is socket (High)
It is activated by a function called by the message (message-A) received from 28b. The message analysis unit 37 analyzes the content of the message (message-A), and calls the connection management unit 38 if it notifies the state of the connection. The connection management unit 38 changes the setting value of the "connection state" of the connection management table shown in FIGS. 7 and 8 according to the contents of the message (message-A).

The message analysis section 37 displays a message (mess).
If the content of sage-A) is the reception of a packet, the data analysis unit 39 is called. The data analysis unit 39 uses a socket (Hig
h) To request the packet acquisition to 28b and to know the "connection state" of the received packet, refer to the "connection state" of the connection management table shown in FIG.

If the "connection state" is the set value 4 as shown in FIG. 8, the acquired packet is judged as print data and the packet is passed to the job control module 29. If the "connection status" is the set value 0 or 3, it means that the packet is not received.
Discard the acquired packet.

When the "connection state" is the set value 1 or 2, the acquired packet is judged to be a command and the command branching unit 40 is called. In the command branching unit 40, the "connection state" is set value 1
If the packet received in step 5 is a login request (USER command), it is determined to be login processing, and the connection management unit 38 is called.

If the "connection state" is the set value 2, the command has been logged in, and a command other than the login request can be accepted. Therefore, the corresponding command processing unit 41 is called.

When the command processing unit 41 receives a command that needs to be transmitted and received, it processes the client and requests the lower layer to establish a connection in order to establish a data connection. put out.

Until the connection is established, control is passed to other processing, so the processing after the establishment of the data connection is written and held in the "processing command" of the connection management table.

As described above, since each processing independently operates according to the set value in the "connection status" area of the connection management table, it is possible to manage a plurality of connections.

Timer processing during packet transmission is performed as shown by arrows (l) to (n) in FIG. First, the interval timer 21 outputs a message (message-S) to the message processing means 24 of the IFC task 22 as shown by an arrow (1) at regular time intervals.

The message processing means 24 uses a message (mess).
When sage-S) is received, the timer processing means 31 is called as indicated by the arrow (m). The timer processing means 31 calls the protocol stack 26 as needed, as shown by the arrow (n).

Timer processing means 31 and timer processing means 3
The operation of the protocol stack called by 1 will be described with reference to FIG. The protocol stack 26
For example, assume that TCP / IP 26a is being called. First, steps S1 and S2 are processes that must be executed every time for all connections, and are always executed by activating the timer processing means 31.

That is, the "connection state" check, the timer count process of the response waiting packet, the packet retransmission process, the Max Timeout process, etc. are executed.

The other steps are processes that do not need to be executed every time according to the protocol and often take a long processing time, resulting in large variations in the processing amount. These processes use the load coefficient G corresponding to the number of processing steps as shown in FIG.

First, in step S3, the load variable is initialized. In step S4, the processing required for each connection is performed until the processing for one connection is completed.

In step S5, the sum of the load coefficients G of the processing performed at the end of the processing for one connection is added to the load variable. If the load variable is equal to or greater than the constant value (MAX) in step S6, the process is interrupted in step S8 and the timer function is exited. At this time, the position of the connection (connection No.) at which the processing is started when the timer function is called next time is stored in a variable in the global area of the memory 5.

If the load variable is equal to or less than the fixed value (MAX), steps S3 to S7 are repeated until the processing of all connections is completed.

When the processing of all the connections is completed while the load variable remains below the fixed value (MAX), the timer processing is terminated.

According to the second embodiment, the network protocol (2nd to 7th layers) processing is divided into a lower layer and an upper layer by a socket interface on a single task.
Since the message, which is the basic function of the OS, can be used to activate the processing means of each layer, it is possible to prioritize two processes on a single task and to interrupt during network protocol processing. Can accept other interface processing, so other host interfaces (Centronics,
Enables coexistence with RS-232C).

Further, by providing the timer processing means, it is possible to maintain a stable system state by dispersing and averaging the processing loads that tend to concentrate in a temporary period. As a result, the printer can improve the performance such as overrun and overflow, and can maintain stable performance regardless of the concentration of network processing.

In addition, it is possible to establish and manage multi-connection by an application operating on a single task.

Third Embodiment In the third embodiment, the usage amount of the reception buffer 5a is periodically detected to change the allowable number of connections to speed up recovery from a processing saturation state and to reduce the number of connections. It is designed to operate efficiently even when a request connection is established.

FIG. 11 shows an interface according to the third embodiment.
FIG. 12 is a block diagram showing the main part of the val timer processing, and FIG. 12 is a table used when increasing or decreasing the allowable connections.

First, the timer processing means 31 shown in FIG. 5 judges the processing load by referring to the buffer number sensing unit 50 for sensing the usage amount of the reception buffer 5a and the table 51a shown in FIG. A unit 51 and a connection number setting unit 52 that sets the allowable number of connections are provided.

Next, the operation will be described with reference to FIG. FIG. 13 is a flow chart showing the operation of the third embodiment. In step S1, the buffer number sensing unit 50 senses the usage amount of the reception buffer 5a in the cycle of the interval timer 21, and stores the usage amount of the reception buffer 5a in the storage area X of the globally reserved memory 5 in step S2. At the time of the first sensing, since there is no "reception buffer usage amount (Y) at the last sensing" data, the storage area Y of the memory 5 is set to 0xFFFF.

In step S3, the contents of the storage area Y is 0xFF.
Whether it is FF or not is checked. If it is 0xFFFF, the process branches to step S5. If not, the process branches to step S4.
In step S4, since "reception buffer use amount (Y) at the last sense" data exists, the "reception buffer use amount (X) at the present sense" is used to compare the data with the table 51a shown in FIG. By comparison, the amount of change in the allowable number of connections is determined and written in the storage area Z of the memory 5.

In step S5, the "reception buffer usage amount (X) at this time sense" is copied to the "reception buffer usage amount (Y) at the previous sense" in preparation for the next sensing.

In step S6, it is checked whether or not the content of the storage area Z is 0. If it is 0, the procedure returns to step S1 and waits for the activation by the next interval timer 21, otherwise, in step S7 the storage area Z is returned. Increase or decrease the number of connections according to the contents of. The connection number setting unit 52 is activated when the content of the storage area Z is other than 0 and increases or decreases the number of connections according to the content of the storage area Z. In this process, when the content of the storage area Z is a positive number, an area required for each connection is secured and initialized. When the content of the storage area Z is a negative number, the area required for each connection is disabled. If the connections are connected up to the allowable number, the area required for each connection is made unavailable as soon as the connection is disconnected.

The operation of the connection number setting unit will be described with reference to FIG. The connection number setting unit 52 is activated when the content of the storage area Z is other than 0 and increases or decreases the number of connections according to the content of the storage area Z. When the content of the storage area Z is a positive number, the area required for each connection is secured and initialized. When the content of the storage area Z is a negative number, the area required for each connection is disabled. If the connection is connected up to the allowable number,
As soon as the connection is disconnected, the area required for each connection becomes unavailable.

When the used reception buffer amount at the present time and the previous time is 20% or less of the total reception buffer amount, it is highly possible that the connection with the low request load occupies the majority of all the connection buffers. Increase by 1.

If the currently used receive buffer amount is 80% or more of the total receive buffer amount, the processing load of the subsequent process is high and overflow of the receive buffer is expected. Decrease one.

If the currently used receive buffer amount is between 21% and 50% of the total receive buffer amount, the last used receive buffer amount is
If the value is 51% or more, it indicates that the processing load is decreasing, and the allowable number of connections is increased by 1 in preparation for a sudden decrease in processing load.

If the currently used receive buffer amount is between 51% and 80% of the total receive buffer amount, the last used receive buffer amount is
If it is less than 50%, it indicates that the processing load is increasing, so the allowable number of connections is decreased by 1 in preparation for a sudden increase in processing load.

According to the third embodiment, by changing the number of connections to be established depending on the weight of the processing currently being performed, recovery from the processing saturation state is speeded up, and efficiency is improved even when a low request connection is established. It can work well.

Fourth Embodiment In the fourth embodiment, the timer interval of the interval timer 21 is made variable in response to a high load notification from the expansion task 10a of the printing unit 10 shown in FIG. Thus, when a low-priority task such as the expansion task 10a has a high processing load, a high-priority task such as the IFC task 22 can be controlled.

FIG. 14 is a block diagram of a load monitoring process according to the fourth embodiment. The load monitoring unit 60 is provided in a low priority task in order to monitor the load of processing such as the expansion task 10a shown in FIG. The load notification area 61 is provided in the global area of the memory 5, receives a notification from the load monitoring unit 60, and notifies it to other processing when the load is high. The timer interval value changing means 21c is provided in the interval timer 21, and changes the timer interval value of the counter 21b of the interval timer 21 when the load is high.

FIG. 15 is a time chart of the expansion processing and engine processing in the printer.
Indicates the expansion processing, engine processing, timer processing, and other processing.

The time W indicates the time during which the expansion process can operate during the engine processing period T1. If the expansion of a certain amount of data is not completed within this time, an overrun occurs. When the timer process and other processes (both of which have higher priority than the expansion process) occur at times t1, t2, t4, and t5, the actual expansion process operation is in the range indicated by the arrow, and overrun is likely to occur.

FIG. 16 is a flow chart showing the operation of the fourth embodiment.

Next, the operation will be described according to the steps of FIG. In step S1, the timer processing 21a of the interval timer 21 senses whether or not the counter 21b counting for each timer interrupt becomes equal to or more than the timer interval value T. In this case, in step S3, a message (message-S) is output to the IFC task 22 to call the timer processing means 31 in the protocol stack. The counter 21b is reset to 0 in step S4.

In step S5, the timer interval value changing means 21
The c refers to the load notification area 61, and when the high load notification is received from the load monitoring unit 60, the timer interval value T of the counter 21b is doubled in step S8. If the high load notification has not been received, the timer interval value T of the counter 21b is checked in step S6, and if it is the default value before the change, this processing ends. If there is a change, the timer interval value T is shortened to half in step S7. When the high load notification is continuously output from the load monitoring unit 60 to the load notification area 61, the interval of the timer processing in the protocol stack 26 that is posted and activated by the timer processing unit 31 is extended by a unit of two, and the load is increased. If the high load notification is not output to the notification area 61 for a period of time,
The interval is reduced by one-half unit up to the default timer interval value.

By applying this embodiment to the printer and changing the interval of the timer processing in response to the high load notification from the developing task or the like, the timer processing time within the interval shown by the time W in FIG. It becomes possible to do. Further, by setting the load notification area shown in FIG. 14 in the global area, it is possible to notify the high load processing other than the timer processing at the same time, and it is possible to take countermeasures.

According to the fourth embodiment, by changing the timer processing interval in response to a high load notification from a deployment task or the like, a task with a low priority such as a deployment task is placed in a high processing load state. At a certain time, it is possible to control a high priority task such as IFC. As a result, the printer can be improved in performance such as overrun and overflow, and stable performance can be maintained regardless of the concentration of network processing.

Fifth Embodiment Although the first to fourth embodiments correspond to printers having a single printer description language, the fifth embodiment has a plurality of printers. Description language, printer control language (PRINTER CONTROL LANGUA
GE: hereafter referred to as PCL), Postscript (POST SCR
IPT: hereafter referred to as PS), printer job runge (PR
INTER JOB LANGUAGE: hereinafter referred to as PJL), etc., so that it can be supported.

FIG. 17 is a block diagram showing the network architecture of the printer according to the fifth embodiment, and is different from the printer according to the second embodiment in printer description languages such as PCL, PS, PJL. The network driver 25 can be re-entered so that it can be used in common by the IFC task 40 corresponding to
And that it can be called from another function before being returned), Centronics, RS
The point is that the interface such as 232C is a common function like the network driver 25.

The printer language discriminating section 41 is provided for each IFC task 40, Centronics, R corresponding to the printer description language.
An interface such as S-232C and an editing unit corresponding to each printer description language, for example, a PCL editing unit 42, P
The printer description language of the data output from each IFC task 40 is connected to the S editing unit 43 and is output to the corresponding editing unit.

FIG. 18 is a block diagram showing the structure of the IFC task. The difference from the structure of the IFC task according to the second embodiment is that the network driver 25 uses the IF
It is outside the C task 40, and the common driver function 44 called from the upper layer is provided inside the IFC task 40.

FIG. 19 is a block diagram showing the structure of the common function part, which is a network in which drivers (Ethernet system, IEEE802.2 system, IEEE802.3 system, SNAP system, etc.) for controlling access to the IFC task 40 are put together. Kudryer 25
And a NIC driver 45 which is a collection of drivers for controlling the network connection H / W.

Next, the operation will be described with reference to FIG. The difference from the second embodiment is that the reception interrupt means 2 in which the message processing means 24 comprises a NIC driver 45.
When the message indicated by the arrow (a) is received from 0, the IF
The point is that the network driver 25 outside the C task 40 is function-called, and other operations are the same as those in the second embodiment.

According to the fifth embodiment, a network driver having a large memory capacity can be commonly used by each IFC task without providing the IFC task corresponding to the printer description language such as PCL, PS, PJL. As a result, memory can be used efficiently.

Sixth Embodiment The network architecture of the printer according to the sixth embodiment is the same as that of the fifth embodiment. Messe
The message processing means 24 is an event control block for receiving messages from the respective processing means.
k: hereafter referred to as ECB), a message is taken out and processed, but there are a plurality of ECBs and the priority of processing is ECB.
If it is provided between them, it is possible to provide a priority to the processing within the task, and it is possible to operate the CPU 1 efficiently. However, if there is one ECB, the priority can be provided to the processing within the task. As a result, the CPU 1 cannot operate efficiently.

The sixth embodiment is such that an efficient operation can be performed even when there is one ECB.

FIG. 20 shows a message according to the sixth embodiment.
It is a block diagram which shows operation | movement of the image processing means. The message processing means 24 refers to the ECB 46 and the upper side queue WAPF Queue 47 of the socket processing means 28, and ECB4
Take the message from 6 and get the protocol stack 26
Request to process the protocol or call the processing of the upper side queue WAPF Queue 47.

FIG. 21 is a flow chart showing the operation of the message processing means shown in FIG. The message processing means 24 checks in step S1 whether the upper layer queue WAPFQueue 47 has been created in the socket processing means 28, and if so, branches to step S2; otherwise, the step S4. Branch to.

In step S2, it is checked whether or not the number of upper layer queues is a predetermined amount, for example, less than 10. If less than 10, the process branches to step S3, and if not, the process branches to step S5. In step S3, it is checked whether or not a message is received by the ECB 46. If it is received, the process branches to step S4, and if not, the process branches to step S5. In step S4, the protocol stack 26 is requested to process the protocol, and in step S5, the socket (High) 28b is called.

The present embodiment can be applied to the second embodiment.

According to the sixth embodiment, even if a plurality of ECBs for storing messages posted by the OS cannot be provided in one task, it is possible to give priority to the processing within the task. Therefore, the CPU can be operated efficiently.

Seventh Embodiment The network architecture of the printer according to the seventh embodiment is the same as that of the fifth embodiment. The job control module 29 has a common table for storing the client identifier of the data, stores the client identifier when outputting the data, and receives the data transmission request from the upper layer. Is a system which analyzes a client identifier based on a common table and transmits / receives a packet through the same connection.

FIG. 22 is a block diagram showing the seventh embodiment. The second packet processing means 30 has a common table for registering a data client identifier (hereinafter referred to as a client ID) in the job control module 29.

The DMD 48 used when transmitting data to the editing task of the printing section and the register when receiving the data transmission request from the printer language discrimination section 41 are registered in the common table.
There is SendDescriptor 49. For example, assume that three packets are received via the network driver 25. It is assumed that each packet is processed by the protocol and transmitted to the job control module 29 through the applications AP1, AP2, AP4.

The client IDs C1, C2 and C3 are DMD4 when the reception interrupt means 20 receives a packet.
Registered in 8. The job control module 29 receives print data received from the applications AP1, AP2 and AP4.
The client IDs C1, C2, and C3 are transmitted to the printer language discrimination unit 41 together with the data. Client ID is 32
Composed of bits. The first 8 bits are IFC task 4
0, Centronics, RS-232C, and other interface types. The next 24 bits are each interface type.
Will be unique to each. For example, in the case of the IFC task 40, an 8-bit ID (AP1
, AP2, ...) And the remaining 16 bits are the serial number in each application. With this format, it is possible to deal with the same application even if client connections are concentrated.

When the printer language discrimination unit 41 transmits the data to the received client, the printer language discrimination unit 41 discriminates the interface type from the first 8 bits of the client ID, and determines the interface type. Subscribe to Send Descriptor 49.

The job control module 29 is the Send Descrip
The bit identifying the application is read by referring to tor 49, and the corresponding application is called.

The present embodiment can be applied to the second embodiment.

According to the seventh embodiment, the upper layer, which transmits and receives packets by using the interface such as IFC task, determines the destination of the response packet without paying attention to the contents of the client ID. Since it is specified, the response packet can be accurately transmitted to the requesting client.

Eighth Embodiment The network architecture of the printer according to the eighth embodiment is the same as that of the fifth embodiment. In the eighth embodiment, when the size of the receiving buffer incorporated in the printer or the like is small, the effective receiving buffer free space is calculated and the receiving speed is improved.

FIG. 23 is an explanatory diagram showing the data transfer processing between the printer language discrimination section 41 and the job control module 29, and FIG. 24 is the data transfer between the transmission side TCP and the reception side TCP. FIG. 25 is an explanatory diagram showing processing, and FIG. 25 is a block diagram showing an eighth embodiment. As described in the seventh embodiment, the printer language discriminating section 41 requests the job control module 29 for the reception data according to the processing status. However, as shown in FIG. The polling interval is not always constant.

Further, in the reception process between TCPs, as shown in FIG. 24, a block transfer method is used in which a plurality of received packets are received and one response packet is returned. Furthermore, the distance limitation between the transmitting and receiving terminals is practically nonexistent, such as using the function of the transport layer. Therefore, the packet transfer time T may be very short or very long.

Therefore, the polling interval is regularly monitored by the job control module 29, and the job control module 2
A transfer data counting means 50 for counting the number of data R transferred from the printer 9 to the printer language discrimination unit 41 within a unit time is provided.

Further, the packet transfer time updating means 5 for calculating the packet transfer time T when processing the TCP received packet.
1 is provided in the protocol stack 26.

In addition, the job control module 29 uses the packet transfer time T, the number R of transfer data for transferring data to the upper layer in a unit time, and the current free space W of the receiving buffer for the next packet. Receivable data capacity calculation means 52 for calculating the receivable data capacity X in block transfer is provided.

The receivable data capacity X is X = (R × T + W) ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ (14) It is calculated by.

The memory 5 has a storage area 53 for storing the packet transfer time T, a storage area 54 for storing the free space W of the reception buffer updated every time the NIC driver 45 receives a reception packet, and a receivable data capacity. Calculation means 52
A storage area 55 for storing the receivable data capacity X calculated by the above is provided.

The response packet creating means 56 of the protocol stack 26 stores the response packet in the storage area 5 when creating the response packet.
The contents of 5 are included in the response packet and transmitted.

The present embodiment can also be applied to the second embodiment.

According to the eighth embodiment, high-speed and efficient reception can be realized by making the processing speed of the subsequent process and the receiving speed of the previous process close to each other regardless of the actual reception buffer size.

Ninth Embodiment The network architecture of the printer according to the ninth embodiment is the same as that of the fifth embodiment. In the ninth embodiment, the reception buffer becomes full of reception data,
A polling packet for confirming the connection from the transmitting side cannot be received and a response packet for the packet cannot be transmitted. Even if that state continues for a long time, the connection is not cut off.

FIG. 26 is a block diagram showing the ninth embodiment, and shows the receiving structure of the NIC driver for processing the data link layer and below shown in FIG. NIC driver 4
As described in the seventh embodiment, the number 5 is linked to a common table for commonly managing the received packet in its own layer and its upper layer. The common table is provided in the memory 5 as an area for storing information processed by each layer and information notified to a layer which is a post process of the own layer.

The reception buffer 5a is provided with an "always empty area" which is always free and an "always used area" which is always used. The NIC driver 45 recognizes the "always available area" and the "always used area", preferentially receives the received packet in the "always used area", and at the same time provides information (reception data size, reception data) in the common table. The start address of the data storage area, etc.) is stored. "Always used area"
When is filled with received packets, it is controlled so that it can be received in the "always free area".

The NIC driver 45 is "always empty area"
Even if it is being received in the
When an empty area is created in the "always-used area", control is performed so that the "always-used area" will be received thereafter. The "always empty area" consists of an area for 1 to 2 packets, and when a packet is received in this area, the information is stored in a common table (special) linked to the "always empty area".

FIG. 27 is a flow chart showing the operation of the ninth embodiment, where (A) shows the operation of the NIC driver and (B) shows the operation of the TCP bypass processing.

First, the operation of the NIC driver 45 will be described. In step S1, it is checked whether or not a packet is received in the "always free area", and if it is received, step S2
If not, the process branches to step S5. In step S2, it is checked whether or not there is a free space in the common table (special). If there is a free space, the process branches to step S3. If not, the process branches to step S6. In step S3, the information of the packet received in the "always empty area" is stored in the common table (special).

In step S4, the TCP bypass process is called, and after the TCP bypass process, the process returns to step S1. In step S5, normal processing is performed and the process returns to step S1. In step S6, the packet received in the "always empty area" is discarded and the process returns to step S1.

In step S7, it is checked whether or not the received packet is a polling packet from the packet source. If it is a polling packet, the process branches to step S8. If not, the process branches to step S9. When the TCP on the receiving side wants to temporarily stop the reception because the receiving buffer is full or the like, it stores 0 in a predetermined area of the response packet and transmits it to the TCP on the transmitting side. When the TCP on the transmitting side receives this response packet, it stops transmitting the packet. Then, in order to acquire the timing to start the packet transmission, the polling packet is periodically transmitted to the TCP on the receiving side.

In step S8, a response packet is created and transmitted. In step S9, the packet received in the "always free area" is discarded.

The present embodiment can also be applied to the second embodiment.

According to the ninth embodiment, even if the reception buffer is full of reception data and the reception buffer is not empty for a long time, a polling packet for confirming the connection from the transmission side can be received. Since the response packet to the packet can be transmitted, the connection can be maintained and the connection can be maintained during the low load process.

[0174]

Since the present invention is configured as described above, it has the following effects. Network
Since the packet layer and the transport layer are processed by one function of the received packet batch processing means, all the packets received in the reception buffer can be processed by one activation of the protocol processing, which results in the overhead of the OS. The amount can be reduced.

Further, by collectively processing the network layer and the transport layer, the work area on the memory can be reduced as compared with the case where the protocol is processed for each of the network layer and the transport layer.

Further, the transmission packet batch processing means for collectively transmitting the response packet to the last received packet belonging to each connection to the network during the interruption can greatly reduce the transmission processing time, As a result, the CPU can be operated for other processing, so that it is possible to provide a method of implementing a network protocol that operates efficiently.

Similarly, at least the network layer and the transport layer are processed by one function of the first packet processing means, and the session layer, the presentation layer, and the application layer are processed by the second packet. By processing with one function of processing means, and by placing a socket processing means and a message processing means between the two functions to give priority to processing,
The amount of OS overhead can be reduced.

Further, by collectively processing the network layer and the transport layer, the work area on the memory can be reduced as compared with the case where the protocol is processed for each of the network layer and the transport layer.

Since the low-function OS can be used to accept other interface processing as well, coexistence with other host interfaces (Centronics, RS-232C) on the same task is possible. You can

[Brief description of drawings]

FIG. 1 is a network of a printer according to a first embodiment.
It is a block diagram which shows a quadrature.

FIG. 2 is a configuration diagram of a network according to the first embodiment.

FIG. 3 is a time chart of the network architecture shown in FIG.

FIG. 4 is a time chart showing a comparison of transmission processing between the related art and the present invention.

FIG. 5 is a network of a printer according to a second embodiment.
It is a block diagram which shows a quadrature.

FIG. 6 is a block diagram showing a configuration of an application FTP.

FIG. 7 is an explanatory diagram of a connection management table.

FIG. 8 is an explanatory diagram showing a relationship between a connection state and a set value.

FIG. 9 is a load calculation table showing the relationship between the number of processing steps and the load.

FIG. 10 is a flowchart showing a process when a packet is retransmitted.

FIG. 11 is a block diagram showing a main part of a third embodiment.

FIG. 12 is a table used when increasing or decreasing the allowable connections.

FIG. 13 is a flowchart showing the operation of the third embodiment.
It is

FIG. 14 is a block diagram of load monitoring processing according to a fourth embodiment.

FIG. 15 is a time chart of development processing and engine processing in the printer.

FIG. 16 is a flowchart showing the operation of the fourth embodiment.
It is

FIG. 17 is a block diagram showing the network architecture of the printer according to the fifth embodiment.

FIG. 18 is a block diagram showing the structure of an IFC task.

FIG. 19 is a block diagram showing a structure of a common function part.

FIG. 20 is a block diagram showing a sixth embodiment.

FIG. 21 is a flow chart showing the operation of the message processing means.

FIG. 22 is a block diagram showing a seventh embodiment.

FIG. 23 is an explanatory diagram of data transfer processing between the printer language discrimination unit and the job control module.

FIG. 24 is an explanatory diagram of a data transfer process between TCPs.

FIG. 25 is a block diagram showing an eighth embodiment.

FIG. 26 is a block diagram showing a ninth embodiment.

FIG. 27 is a flowchart showing the operation of the ninth embodiment.
It is

[Explanation of symbols]

1 CPU 3 NIC 4 TCP / IP section 5 memory 6, 22, 40 IFC task 23 First Packet Processing Means 30 Second Packet Processing Means

Claims (14)

(57) [Claims] 1. A packet processing method for establishing a connection between nodes of a network and processing sequentially a protocol hierarchically added to a packet received in a reception buffer from a lower layer to an upper layer, at least the network layer, The processing of protocols related to the transport layer, session layer, presentation layer, and application layer is made into one task,
Network layer, a first function of a single transport layer, session layer, and presentation layer, a second function of a single application layer, both of the function is a single
It is processed by one central processing unit, and when multiple packets are received within a predetermined time, the above
A packet processing method, wherein a central processing unit is made to operate by giving priority to the first function . 2. A protocol of a received packet in a network architecture in which a connection is established between nodes of a network and a protocol hierarchically added to a packet received in a reception buffer is processed in order from a lower layer to an upper layer. At least out of the net
Work layer, transport layer, session layer, presentation
Application layer, a single processing layer
A central processing unit, said during a predetermined time using a timer interrupt to enter for each, interrupt protocols for the network layer and the transport layer of a plurality of packets received in the receiving buffer
Batch and accumulated by the central processing unit, the central processing instrumentation protocol processing means of the upper layer from the contents of the protocol
Upper and received packet batch processing means for designating notified by location, a response packet to the last received packet belonging to each connection to the network during the interruption
A network architecture comprising: a transmission packet batch processing unit that collectively transmits by a central processing unit . 3. Connection between nodes of a network
Hierarchically to the packets received in the receive buffer
The added protocols are ordered from the lower layer to the upper layer.
In a network architecture that processes in parallel, a timer interrupt that enters at regular intervals is used to
Network layer and trans
Batch processing of protocol related to port layer during interrupt
Stored in the upper layer and processed from the contents of the protocol.
Received packet batch processing means for specifying and notifying the processing means, and the last received packet belonging to each connection
Response packet to the network during the above interrupt.
Transmission packet batch processing means for collectively transmitting, the transmission packet batch processing means , the reception packet batch processing means, the response packet transmission request generated from the processing means of the upper layer during batch processing of the received packets, A network architecture characterized by comprising timer processing means for performing transmission at the time of the next timer interrupt and for counting the response timeout from the receiving side in response to the timer interrupt count. 4. A network architecture that establishes a connection between the nodes of the network, and sequentially processes the protocols hierarchically added to the packets received in the reception buffer from the lower layer to the upper layer. In the above, in the data link layer, a message processing means by an operating system function for activating the processing means called based on the message, and a received packet fetched from the receiving buffer called by the message processing means are provided. ,
A first packet processing means for processing a protocol relating to a network layer and a transport layer and accumulating until a predetermined amount is reached, and a higher layer processing a protocol relating to a session layer, a presentation layer and an application layer. A second packet processing means for outputting to the second packet processing means and a message for calling the second packet processing means designated by the first packet processing means are outputted to the message processing means, and the message is called from the message processing means. Interface provided with socket processing means for selecting the second packet processing means according to
A network comprising: a face control means; and a reception interrupt means which is activated when a packet is received and outputs a message to the interface control means each time a received packet is stored in a reception buffer. -Qua-texture. 5. The first packet processing means is called on the basis of the message of the reception interrupting means, extracts a received packet from the reception buffer, and processes a data link layer protocol. The network architecture according to claim 4. 6. A network architecture that establishes a connection between the nodes of the network, and sequentially processes the protocols hierarchically added to the packets received in the reception buffer from the lower layer to the upper layer. In the above, the message processing means by the operating system function that activates the processing means called based on the message and the protocol related to the network layer and the transport layer are processed and accumulated until a predetermined amount is reached. Designated by the first packet processing means and the second packet processing means for processing the protocol relating to the session layer, the presentation layer, and the application layer and outputting the upper layer. A message for calling the second packet processing means is output to the message processing means, and the message processing means is output. Interface control means having a socket processing means for selecting the second packet processing means in response to a call from the same, and a message activated each time the packet is received and stored in the reception buffer. Reception interrupt means for outputting to the interface control means, and a message processing means of the interface control means based on a message of the reception interrupt means,
A network driver, comprising a network driver for processing a protocol relating to a data link layer and calling the first packet processing means. 7. An interval timer which outputs a message to the message processing means of the interface control means each time the counter counts up, and is called based on the message of the interval timer. 7. The network network according to claim 4, wherein the first packet processing means is provided with timer processing means for monitoring the amount of packets accumulated by the first packet processing means.
Texture. 8. The timer processing means executes timer processing during protocol processing of the first packet processing means, sets a load coefficient to low priority processing such as response packet transmission processing, and sets a constant load. 8. The network architecture according to claim 7, wherein the timer processing is interrupted at the time when the timer processing reaches, and the continuation is executed in the next timer processing to average the load processed in one timer processing. 9. The timer processing section stores a table storage means for storing a table to be referred to when the number of connections is increased or decreased, and a first reception for storing a usage amount of a reception buffer at the previous activation. Buffer usage storage means and the above message
A buffer amount sensing unit that senses the usage amount of the reception buffer and stores it in the second reception buffer usage amount storage unit each time the processing unit is activated; the contents of the first reception buffer usage amount storage unit; Of the reception buffer usage amount storage means and the table of the table storage means, and a number of connections for setting the allowable number of connections based on the output of the determination part. The network architecture according to claim 7, further comprising a setting unit. 10. The data stored in the data buffer
Load monitoring unit provided in a low-priority processing unit for processing data, load notification storage unit for storing load notification from the load monitoring unit, and the interval timer if the content of the load notification storage unit is a high load notification. 8. The network architecture according to claim 7, further comprising timer interval value changing means for increasing the timer interval value of the counter provided therein and for decreasing the timer interval value of the counter. 11. The message processing means has one ECB (Event Control Block) for storing a message, refers to the ECB and the upper queue of the socket processing means, and stores the upper queue. The network architecture according to claim 4 or 6, wherein the upper side queue is processed when the number of registered messages is a predetermined number or more. 12. The second packet processing means has a common table for storing a client ID of data, stores the client ID when outputting the data, and stores the data from the upper layer. 5. The client ID is analyzed based on the common table when receiving the transmission request, and the packet is transmitted and received through the same connection.
The described network architecture. 13. A packet transfer time calculating means for calculating a packet transfer time from when a first packet processing means transmits a response packet to a previous received packet to when the current received packet is received, and second packet processing. Next, from the transfer data number counting means for counting the number of transfer data for transferring the data from the means to the upper layer in a unit time, the packet transfer time, the number of transfer data, and the present receiving buffer free capacity. 7. The network architecture according to claim 4, further comprising a receivable data capacity calculating means for calculating a receivable data capacity in the packet block transfer. 14. The reception interrupt means has a common table for storing information of a received packet in an “always used area” and an “always free area” of the reception buffer, and each time the packet is received, a common table is stored. If the packet received in the "always empty area" with reference to the common table is a polling packet for reception confirmation that is transmitted from the transmission side to the reception side, a response packet is created and transmitted, and reception is possible. The network architecture according to claim 4 or 6, wherein the packet is discarded if it is other than a confirmation polling packet.