OS i/o devices and device controllers

I/O The equipment is generally executed by I/O The mechanical part of operation and execution control I/O The electronic components of . These two parts are usually separated , perform I/O The mechanical part of the operation is general I/O equipment , And executive control I/O The electronic components of the are called device controllers or adapters （ adapter ). In microcomputers and minicomputers, controllers are often made into printed circuit cards, so they are often called control cards 、 Interface card or network card , It can be inserted into the expansion slot of the computer . In some big 、 In medium-sized computer systems , Also configured I/O Channel or I/O processor .

I/O equipment

I/O Type of equipment

IO There are many types of equipment , In addition to being able to divide them into block devices and character devices 、 Exclusive devices and shared devices , It can also be divided into storage devices and I/O equipment ; From the transmission rate of equipment, it can be divided into high-speed equipment 、 Medium speed sharing devices and high-speed sharing devices . These two categories are introduced below .

1） Classified by service characteristics

The first category is The storage device , Also known as external storage 、 Auxiliary deposit , It is the main device used to store information . The access speed of this kind of device is slower than that of memory , But the capacity is much larger , It's also cheap . The second is I / O equipment , It can also be divided into input devices 、 Output devices and interactive devices . Input devices are used to receive external information , Like a keyboard 、 mouse 、 Scanner 、 Video camera, etc . The output device is used to send the information processed by the computer to the device outside the processor , Such as a printer 、 Plotters, etc . Interactive devices refer to the above two types of integrated devices , Mainly display , It is used to synchronously display user commands and the results of command execution .

2） By transmission rate

According to the transmission speed , Can be I/O The equipment is divided into three categories . The first category is Low speed equipment , Its transmission rate is only a few bytes to hundreds of bytes per second . A typical low-speed device has a keyboard 、 Mouse device . The second type is Medium speed equipment , Its transmission rate ranges from thousands of bytes to hundreds of thousands of bytes per second . Typical medium speed devices are line printers 、 Laser printers, etc . The third kind is High speed equipment , Its transmission rate ranges from hundreds of thousands of bytes to gigabytes . Typical high-speed devices are tape drives 、 Disk drive 、 CD player, etc .

Interface between equipment and controller

Usually , The device is not directly related to CPU communicate , Instead, it communicates with the device controller , therefore , stay I/O The equipment shall include the interface with the equipment controller , There are three types of signals in this interface （ See the picture 6-3 Shown ）, Each corresponds to a signal line .

(1） Data signal line

This kind of signal line is used to transmit data signals between the device and the device controller . For input devices , The signal input from the outside is converted by the converter , The formed data is usually sent to the buffer first , When the amount of data reaches a certain bit （ character ） After counting , Then it is transmitted from the buffer to the device controller through a group of data signal lines , Pictured 6-3 Shown . For output devices , A batch of data transmitted from the device controller through the data signal line is temporarily stored in the dry buffer , After proper conversion by the converter , Then output character by character .

(2） Control signal line

This is done by the device controller to I / O The path through which the device sends control signals . This signal specifies the operation to be performed by the equipment , Such as reading operation （ It refers to the transmission of data from the equipment to the controller ） Or write operations （ Receive data from the controller ）, Or perform operations such as head movement .

(3） Status signal line

The signal line is used to transmit a signal indicating the current state of the device . The current state of the device is reading （ Or write ）; Device read （ Write ） complete , And prepare for new data transmission .

---

Device controller

The main functions of the device controller are , Control one or more I / O equipment , In order to realize the I / O Data exchange between devices and computers . It is CPU And I / O Interfaces between devices , Receive from CPU Orders sent , To control the I / O The equipment works , Enable the processor to free itself from complicated equipment control transactions . The device controller is an addressable device , When it controls only one device , It only has a unique device address ; If the controller can be connected to multiple devices , It should contain multiple device addresses , Each device address corresponds to a device . Device controllers can be divided into two categories ： One is the controller used to control character devices , The other is a controller for controlling block devices .

1． Basic functions of equipment controller

(1） Receive and recognize commands

The device controller can receive and recognize various commands sent by the processor . There are corresponding control registers in the controller , Used to store received commands and parameters , And decode the received command . for example , The disk controller can receive CPU It's from read 、 write 、 format etc. 15 Different commands , And some commands also have parameters . Accordingly , There are multiple registers and command decoders in the disk controller .

(2） Data exchange

The device controller can realize CPU And the controller 、 Between controller and equipment Data exchange of . For the former , Through the data bus , from CPU Write data to the controller in parallel , Or read data from the controller in parallel . For the latter , It is the device that inputs data to the controller , Or from the controller to the device . So , The data register must be set in the controller .

(3） Identify and report the status of the equipment

The controller shall record the status of the equipment for CPU understand . for example , Only when the device is in send ready state , CPU To start the controller to read data from the device . So , Set a status register in the controller , Use each of them to reflect a certain state of the equipment . When CPU After reading the contents of this register , You can know the status of the equipment .

(4） Address recognition

Just like every cell in memory has an address , Every device in the system also has an address . The device controller must be able to identify the address of each device it controls . Besides , To make CPU Be able to turn to （ Or from ） Write in the register （ Or read out ） data , These registers should have unique addresses . The controller shall be able to correctly identify these addresses . So , The address decoder should be configured in the controller .

(5） Data buffer

because I / O The speed of the device is low , and CPU And memory speed is very high , Therefore, a buffer must be set in the controller . At output time , Use this buffer to temporarily store the data transmitted by the host at a high speed , Then with I / O The device transmits the data in the buffer to at a rate matched by the device I / O equipment . On input , The buffer is used for temporary storage from I / O Data sent by the equipment , After receiving a batch of data , Then the data in the buffer is transmitted to the host at a high speed .

(6） Error control

For the I / O Data transmitted by the device , The device controller is also responsible for error detection . If an error is found in the transmission , Usually, the error detection code is set , And to CPU The report , therefore CPU Invalidate the data transmitted this time , And re transmit . This will ensure the correctness of data input .

---

The composition of the device controller

Because the device controller is located in CPU Between and equipment , It should be related to CPU signal communication , And communicate with the device , It shall also have the following functions: CPU The command sent to control the function of the equipment , therefore , Most of the existing controllers are composed of the following three parts ：

(1） Interface between equipment controller and processor

This interface is used to implement CPU Communication with the device controller , There are three types of signal lines in this interface ： cable 、 Address lines and control lines .
Data lines are usually connected to two types of registers ：

• ① The first category is Data register , There can be one or more data registers in the controller , It is used to store the data sent from the equipment （ Input ）, Or from CPU The data that came in （ Output ）.
• ② The second type is control ／ Status register , There can be one or more of these registers in the controller , Used to store from CPU Control information or equipment status information sent .

(2） The interface between the device controller and the device

On a device controller , One or more devices can be connected . Corresponding , There are one or more device interfaces in the controller . There is data in every interface 、 Three types of signals, control and status . In the controller I / O The logic selects a device port according to the address signal sent by the processor .

(3) I / O Logic

I / O Logic is used to control the equipment . It interacts with the processor through a set of control lines , The processor uses this logic to send a message to the controller I / O command . whenever CPU To start a device , On the one hand, send the start command to the controller , On the other hand, it sends the address to the controller through the address line , By the controller I / O Logic decodes the received address , Then control the selected equipment according to the translated command .
The composition of the device controller is shown in Figure 6-4 in .

---

Memory image I / O

The driver will be abstract I / O A series of specific commands converted from commands 、 Parameters and other data are loaded into the corresponding registers of the device controller , The controller executes these commands , The specific implementation is right I / O Control of equipment . This work can be done in the following two ways ：

1． Use specific I / O Instructions

In early computers , Including mainframe computers , In order to achieve CPU Communication with device controller , Assign one to each control register I / O port , This is a 8 Bit or 16 An integer , Pictured 6-5( a ） Shown . In addition, some specific I / O Instructions . for example , In order to CPU The contents of the register are copied into the controller register , Specific to be used I/O Instructions can be expressed as follows ：

io-store cpu-reg,dev-no,dev-reg 

among , cpu-reg yes CPU Some register of ;dev-no Is the designated device , Controller address ; dev-reg Specify the registers in the controller .
If it's going to be CPU The contents of the register are stored in a unit of memory （k） in , The following instructions will be used ：

Store cpu-reg,k 

The main disadvantage of this method is , Access to memory and access to devices require two different instructions .

2． Memory image I / O

In this way , In addressing, the memory unit address and the register address in the device controller are no longer distinguished , All use k . When k Value in 0~ n -1 Range time , Considered a memory address , if k Greater than or equal to n when , It is considered to be the register address of a controller . From the figure 6-5( b ） It can be seen that , When k = n when , Indicates the device controller 0 Of the 1 A register opcode The address of . therefore , If you want to CPU The contents of the register are transferred to the controller 0 Of the 1 A register opcode , Just use the following general storage instructions ：

Store cpu-reg,n 

Memory image I / O Method unifies the access methods to memory and controller , This will undoubtedly simplify I / O Programming for .

---

I / O passageway

1. l / O Introduction of channel equipment

Although in CPU And I / O After adding a device controller between devices , Can greatly reduce CPU Yes I / O Intervention , But when the host is configured with many peripherals , CPU The burden is still heavy . So , stay CPU And equipment controller I / O passageway （ I / O Channel ). Its main purpose is to establish an independent I / O operation , Not only can the transmission of data be independent of CPU , And also hope that the relevant to I / O Organization of operations 、 Manage and end processing as independently as possible , In order to make sure CPU More time for data processing ; Or say , Its purpose is to make some original CPU To deal with the I / O The task is transferred to the channel , So that CPU From complicated I / O Free yourself from the task . After setting the channel , CPU Just send one... To the channel I / O Instructions . After receiving the command, the channel , Then take out the channel program to be executed this time from the memory , Then execute the channel program , Only when the channel has completed the specified I / O After the task , Just to CPU Send an interrupt signal .
actually , I/O A channel is a special processor . It has the ability to execute I / O The ability to command , And control by executing the channel program I / O operation . but I / O The channel is different from the general processor , Mainly in the following two aspects ： One is Its instruction type is single , This is because the channel hardware is relatively simple , The command it can execute , Mainly limited to I/O Operation related instructions ; Two is Channels don't have their own memory , The channel program executed by the channel is placed in the memory of the host , In other words , It's channels and CPU Shared memory .

2． Channel type

As mentioned before , Channels are used to control peripheral devices （ Including character devices and block devices ） Of . Because there are many types of peripheral devices , And its transmission rate varies greatly , Thus, there are many types of channels . here , According to the different ways of information exchange , Channels can be divided into the following three types .

1） Byte multichannel （ Byte Multiplexor Channel )

This is a kind of Work in byte crossing The passage of . It usually contains many non distributive subchannels , The number can range from dozens to hundreds , Each subchannel is connected to one I / O equipment , And control the operation of the equipment I / O operation . These subchannels press Time slice rotation mode Share the main channel . When the first subchannel controls its I / O After the device completes the exchange of one byte , Immediately vacate the main channel , Let the second subchannel use ; When the second subchannel also completes the exchange of a byte , Also give up the main channel to the third sub channel ; And so on . When all subchannels rotate for a week , Return to the first subchannel to use byte multiplex main channel . such , As long as the byte multichannel scans each subchannel fast enough , The speed of devices connected to subchannels is not too high , Will not lose information .
chart 6-6 It shows the working principle of byte multichannel . It contains several sub channels A , B , C , D , E ,…, N …, They are respectively connected with one device through the controller . Assume that the speed of these devices is similar , And all transmit data to the host at the same time . equipment A The data stream transmitted is A₁, A₂, A₃,…; equipment B The data stream transmitted is B₁, B₂, B₃,…;…… After synthesizing these data streams （ Through the main channel ） The data flow sent to the host is A₁, B₁, C₁, D₁ ,… A₂ , B₂ , C₂ , D₂ ,…, A₃ , B₃ , C₃ , D₃ ,…

2） Array selection channel （ Block Selector Channel )

Byte multichannel is not suitable for connecting high-speed devices , This promotes the formation of array selection channels for data transmission in array mode . Although this channel can connect multiple high-speed devices , But because it only contains one distributive subchannel , Only one channel program can be executed in a period of time , Control a device for data transmission , As a result, when a device occupies the channel , It has been monopolized by it , Even if it has no data transmission , The channel is idle , Other devices are not allowed to use this channel , Until the transmission of the device is completed, release the channel . so , The utilization rate of this channel is very low .

3） Array multichannel （ Block Multiplexor Channel )

Although the array selection channel has a high transmission rate , But it allows only one device to transmit data at a time . Array multichannel is to select the channel of array. The high transmission rate and byte multichannel can make each subchannel （ equipment ） A new channel formed by the combination of the advantages of time-sharing parallel operation . It contains multiple unallocated subchannels , Therefore, this channel has high data transmission rate , It can also obtain satisfactory channel utilization . That's why , So that the channel can be widely used to connect multiple high 、 Medium speed peripherals , Its data transmission is carried out in the form of array .

3.“ bottleneck ” problem

Because the channel is expensive , As a result, the number of channels set in the machine is bound to be less , This often makes it I/O Bottleneck , Thus, the throughput of the whole system decreases . for example , In the figure 6-7 in , Suppose the device 1 To equipment 4 There are four disks , To boot the disk 4, You must use channels 1 And controller 2; But if these two have been occupied by other equipment , The disk must not start 4. Similarly , To start the disk 1 And plates 2, Because they all use channels 1, Therefore, it is impossible to start . These are caused by insufficient channels “ bottleneck ” The phenomenon .

solve “ bottleneck ” The most effective way to solve the problem , That is Increase the channel from the equipment to the host without increasing the channel , Pictured 6-8 Shown . In other words , Namely Connect a device to multiple controllers , And one controller is connected to multiple channels . The equipment in the figure 1、2、3 and 4, There are 4 A path to memory . for example , Through the controller 1 And channel 1 To memory ; You can also use the controller 2 And channel 1 To memory . The multi-channel method not only solves “ bottleneck ” problem , And it improves the reliability of the system , Because the failure of individual channels or controllers will not make there is no path between the device and the memory .

---