You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
pke-doc/chapter6_device.md

740 lines
28 KiB

This file contains ambiguous Unicode characters!

This file contains ambiguous Unicode characters that may be confused with others in your current locale. If your use case is intentional and legitimate, you can safely ignore this warning. Use the Escape button to highlight these characters.

# 第六章实验4设备管理基于[RISCV-on-PYNQ](https://gitee.com/hustos/fpga-pynq)
### 目录
- [6.1 实验4的基础知识](#fundamental)
- [6.1.1 内存映射I/O(MMIO)](#subsec_MMIO)
- [6.1.2 轮询I/O控制方式](#subsec_polling)
- [6.1.3 中断驱动I/O控制方式](#subsec_plic)
- [6.1.4 设备树](#subsec_device_tree)
- [6.2 lab4_1 POLL](#polling)
- [给定应用](#lab4_1_app)
- [实验内容](#lab4_1_content)
- [实验指导](#lab4_1_guide)
- [6.3 lab4_2_PLIC](#PLIC)
- [给定应用](#lab4_2_app)
- [实验内容](#lab4_2_content)
- [实验指导](#lab4_2_guide)
- [6.4 lab4_3_hostdevice](#hostdevice)
- [给定应用](#lab4_3_app)
- [实验内容](#lab4_3_content)
- [实验指导](#lab4_3_guide)
<a name="fundamental"></a>
## 6.1 实验4的基础知识
完成前面所有实验后PKE内核的整体功能已经得到完善。在实验四的设备实验中我们将结合fpga-pynq板在rocket chip上增加uart模块和蓝牙模块并搭载PKE内核实现蓝牙通信控制智能小车设计设备管理的相关实验。
<a name="subsec_MMIO"></a>
### 6.1.1 内存映射I/O(MMIO)
内存映射(Memory-Mapping I/O)是一种用于设备驱动程序和设备通信的方式它区别于基于I/O端口控制的Port I/O方式。RICSV指令系统的CPU通常只实现一个物理地址空间这种情况下外设I/O端口的物理地址就被映射到CPU中单一的物理地址空间成为内存的一部分CPU可以像访问一个内存单元那样访问外设I/O端口而不需要设立专门的外设I/O指令。
在MMIO中内存和I/O设备共享同一个地址空间。MMIO是应用得最为广泛的一种IO方法它使用相同的地址总线来处理内存和I/O设备I/O设备的内存和寄存器被映射到与之相关联的地址。当CPU访问某个内存地址时它可能是物理内存也可以是某个I/O设备的内存。此时用于访问内存的CPU指令就可以用来访问I/O设备。每个I/O设备监视CPU的地址总线一旦CPU访问分配给它的地址它就做出响应将数据总线连接到需要访问的设备硬件寄存器。为了容纳I/O设备CPU必须预留给I/O一个地址映射区域。
用户空间程序使用mmap系统调用将IO设备的物理内存地址映射到用户空间的虚拟内存地址上一旦映射完成用户空间的一段内存就与IO设备的内存关联起来当用户访问用户空间的这段内存地址范围时实际上会转化为对IO设备的访问。
<a name="subsec_polling"></a>
### 6.1.2 轮询I/O控制方式
在实验四中,我们设备管理的主要任务是控制设备与内存的数据传递,具体为从蓝牙设备读取到用户输入的指令字符(或传递数据给蓝牙在手机端进行打印),解析为小车前、后、左、右、停止等动作来传输数据给电机实现对小车的控制。在前两个实验中,我们分别需要对轮询控制方式和中断控制方式进行实现。
首先程序直接控制方式又称循环测试方式每次从外部设备读取一个字的数据到存储器对于读入的每个字CPU需要对外设状态进行循环检查直到确定该数据已经传入I/O数据寄存器中。在轮询的控制方式下由于CPU的高速性和I/O设备的低速性导致CPU浪费绝大多数时间处于等待I/O设备完成数据传输的循环测试中会造成大量资源浪费。
轮询I/O控制方式流程如图
<img src="pictures/fig6_1_polling.png" alt="fig6_1" style="zoom:100%;" />
<a name="subsec_plic"></a>
### 6.1.3 中断驱动I/O控制方式
在前一种轮询的控制方式中由于没有采用中断机制CPU需要不断测试I/O设备的状态造成CPU资源的极大浪费。中断驱动的方式是允许I/O设备主动打断CPU的运行并请求相应的服务请求I/O的进程首先会进入阻塞状态PLIC将字符读取操作转化为s态中断进行处理向进程传递读取的数据后唤醒进程继续运行。
采用中断驱动的控制方式在I/O操作过程中CPU可以执行其他的进程CPU与设备之间达到了部分并行的工作状态从而提升了资源利用率。
中断驱动I/O方式流程如图
<img src="pictures/fig6_2_plic.png" alt="fig6_2" style="zoom:100%;" />
<a name="subsec_device_tree"></a>
### 6.1.4 设备树
设备树Device Tree是描述计算机的特定硬件设备信息的数据结构以便于操作系统的内核可以管理和使用这些硬件包括CPU或CPU内存总线和其他一些外设。
硬件的相应信息都会写在`.dts`为后缀的文件中,`dtc`是编译`dts`的工具,`dtb(Device Tree Blob)``dts`经过`dtc`编译之后会得到`dtb`文件,`dtb`通过`Bootloader`引导程序加载到内核。所以`Bootloader`需要支持设备树才行Kernel也需要加入设备树的支持。
<img src="pictures/fig6_3.png" alt="fig6_3" style="zoom:130%;" />
在rocketchip中设备即通过设备树的方式提供给pke使用。
<a name="polling"></a>
## 6.2 lab4_1 POLL
<a name="lab4_1_app"></a>
#### **给定应用**
- user/app_poll.c
```
1 /*
2 * Below is the given application for lab4_1.
3 * The goal of this app is to control the car via Bluetooth.
4 */
5
6 #include "user_lib.h"
7 #include "util/types.h"
8
9 int main(void) {
10 printu("please input the instruction through bluetooth!\n");
11 while(1)
12 {
13 char temp = (char)uartgetchar();
14 uartputchar(temp);
15 switch (temp)
16 {
17 case '1' : gpio_reg_write(0x2e); break; //前进
18 case '2' : gpio_reg_write(0xd1); break; //后退
19 case '3' : gpio_reg_write(0x63); break; //左转
20 case '4' : gpio_reg_write(0x9c); break; //右转
21 case 'q' : exit(0); break;
22 default : gpio_reg_write(0x00); break; //停止
23 }
24 }
25 exit(0);
26 return 0;
27 }
```
应用通过轮询的方式从蓝牙端获取指令,实现对小车的控制功能。
- 切换到lab4_1继承lab3_3及之前实验所做的修改并make后的直接运行结果
```
//切换到lab4_1
$ git checkout lab4_1_poll
//继承lab3_3以及之前的答案
$ git merge lab3_3_rrsched -m "continue to work on lab4_1"
//重新构造
$ make clean; make
//运行构造结果
In m_start, hartid:0
HTIF is available!
(Emulated) memory size: 512 MB
Enter supervisor mode...
PKE kernel start 0x0000000080000000, PKE kernel end: 0x0000000080010000, PKE kernel size: 0x0000000000010000 .
free physical memory address: [0x0000000080010000, 0x000000008003ffff]
kernel memory manager is initializing ...
kernel pagetable addr is 0x000000008003e000
KERN_BASE 0x0000000080000000
physical address of _etext is: 0x0000000080005000
kernel page table is on
Switching to user mode...
in alloc_proc. user frame 0x0000000080039000, user stack 0x000000007ffff000, user kstack 0x0000000080038000
User application is loading.
Application: app_poll
CODE_SEGMENT added at mapped info offset:3
Application program entry point (virtual address): 0x00000000810000de
going to insert process 0 to ready queue.
going to schedule process 0 to run.
please input the instruction through bluetooth!
You need to implement the uart_getchar function in lab4_1 here!
System is shutting down with exit code -1.
```
从结果上来看蓝牙端端口获取用户输入指令的uartgetchar系统调用未完善所以无法进行控制小车的后续操作。按照提示我们需要实现蓝牙uart端口的获取和打印字符系统调用以及传送驱动数据给小车电机的系统调用实现对小车的控制。
<a name="lab4_1_content"></a>
#### **实验内容**
如输出提示所表示的那样需要找到并完成对uartgetcharuartputchargpio_reg_write的调用并获得以下预期结果
```
In m_start, hartid:0
HTIF is available!
(Emulated) memory size: 512 MB
Enter supervisor mode...
PKE kernel start 0x0000000080000000, PKE kernel end: 0x0000000080010000, PKE kernel size: 0x0000000000010000 .
free physical memory address: [0x0000000080010000, 0x000000008003ffff]
kernel memory manager is initializing ...
kernel pagetable addr is 0x000000008003e000
KERN_BASE 0x0000000080000000
physical address of _etext is: 0x0000000080005000
kernel page table is on
Switching to user mode...
in alloc_proc. user frame 0x0000000080039000, user stack 0x000000007ffff000, user kstack 0x0000000080038000
User application is loading.
Application: app_poll
CODE_SEGMENT added at mapped info offset:3
Application program entry point (virtual address): 0x00000000810000de
going to insert process 0 to ready queue.
going to schedule process 0 to run.
Ticks 0
please input the instruction through bluetooth!
Ticks 1
going to insert process 0 to ready queue.
going to schedule process 0 to run.
User exit with code:0.
no more ready processes, system shutdown now.
System is shutting down with exit code 0.
```
<a name="lab4_1_guide"></a>
#### **实验指导**
基于实验lab1_1你已经了解和掌握操作系统中系统调用机制的实现原理。对于本实验的应用我们发现user/app_poll.c文件中有三个函数调用uartgetcharuartputchar和gpio_reg_write。对代码进行跟踪我们发现这三个函数都在user/user_lib.c中进行了实现对应于lab1_1的流程我们可以在kernel/syscall.h中查看新增的系统调用以及编号
```
16 #define SYS_user_uart_putchar (SYS_user_base + 6)
17 #define SYS_user_uart_getchar (SYS_user_base + 7)
18 #define SYS_user_gpio_reg_write (SYS_user_base + 8)
```
继续追踪我们发现在kernel/syscall.c的do_syscall函数中新增了对应系统调用编号的实现函数对于新增系统调用分别有如下函数进行处理
```
133 case SYS_user_uart_putchar:
134 sys_user_uart_putchar(a1);return 1;
135 case SYS_user_uart_getchar:
136 return sys_user_uart_getchar();
137 case SYS_user_gpio_reg_write:
138 return sys_user_gpio_reg_write(a1);
```
读者的任务即为在kernel/syscall.c中追踪并完善对应的函数。对于uart的函数我们给出uart端口的地址映射如图
<img src="pictures/fig6_3_address.png" alt="fig6_3" style="zoom:80%;" />
我们可以看到配置uart端口的偏移地址为0x60000000对应写地址为0x60000000读地址为0x60000004同时对0x60000008的状态位进行轮询检测到信号时进行读写操作。
在kernel/syscall.c中找到函数实现空缺并根据注释完成uart系统调用
```
84 //add uart putchar getchar syscall
85 //
86 // implement the SYS_user_uart_putchar syscall
87 //
88 void sys_user_uart_putchar(uint8 ch) {
89 volatile uint32 *status = (void*)(uintptr_t)0x60000008;
90 volatile uint32 *tx = (void*)(uintptr_t)0x60000004;
91 while (*status & 0x00000008);
92 *tx = ch;
93 }
94
95 ssize_t sys_user_uart_getchar() {
96 // TODO (lab4_1): implment the syscall of sys_user_uart_getchar.
97 // hint: the functionality of sys_user_uart_getchar is to get data from UART address. therefore,
98 // we should let a pointer point, insert it in
99 // the rear of ready queue, and finally, schedule a READY process to run.
100 panic( "You have to implement sys_user_uart_getchar to get data from UART using uartgetchar in lab4_1.\n" );
101
102 }
103
104
105
106 //car control
107 ssize_t sys_user_gpio_reg_write(uint8 val) {
108 volatile uint32_t *control_reg = (void*)(uintptr_t)0x60001004;
109 volatile uint32_t *data_reg = (void*)(uintptr_t)0x60001000;
110 //*control_reg = 0;
111 *data_reg = (uint32_t)val;
112 return 1;
113 }
114
```
和uart端口读写过程类似其中电机连接端口gpio数据地址为0x60001000根据用户程序app_poll中流程我们需要将uart端口读到的驱动数据传递给电机。
安卓手机端验证首先将HC-05蓝牙模块接入pynq板接口对应关系为
| pynq接口 | HC-05接口 |
| -------- | --------- |
| VCC | VCC |
| GND | GND |
| JA4 | RXD |
| JA3 | TXD |
接入时注意对应接口错位正确插入然后在手机端下载BluetoothSerial连接hc-05蓝牙模块使用网线连接pynq板和电脑打开开发板电源。
成功连接蓝牙模块后,启动连接:
```
$ ssh xilinx@192.168.2.99
```
随后使用scp指令将编译后的pke内核和用户app文件导入
```
$ scp 文件名 xilinx@192.168.2.99:~
```
此时便成功进入pynq板环境可对结果进行验证。
**实验完毕后,记得提交修改(命令行中-m后的字符串可自行确定以便在后续实验中继承lab4_1中所做的工作**
```
$ git commit -a -m "my work on lab4_1 is done."
```
<a name="PLIC"></a>
## 6.3 lab4_2_PLIC
<a name="lab4_2_app"></a>
#### **给定应用**
- user/app_PLIC.c
```
1 /*
2 * Below is the given application for lab4_2.
3 * The goal of this app is to control the car via Bluetooth.
4 */
5
6 #include "user_lib.h"
7 #include "util/types.h"
8 void delay(unsigned int time){
9 unsigned int a = 0xfffff ,b = time;
10 volatile unsigned int i,j;
11 for(i = 0; i < a; ++i){
12 for(j = 0; j < b; ++j){
13 ;
14 }
15 }
16 }
17 int main(void) {
18 printu("Hello world!\n");
19 int i;
20 int pid = fork();
21 if(pid == 0)
22 {
23 while (1)
24 {
25 delay(3);
26 printu("waiting for you!\n");
27 }
28
29 }
30 else
31 {
32 for (;;) {
33 char temp = (char)uartgetchar();
34 printu("%c\n", temp);
35 switch (temp)
36 {
37 case '1' : gpio_reg_write(0x2e); break; //前进
38 case '2' : gpio_reg_write(0xd1); break; //后退
39 case '3' : gpio_reg_write(0x63); break; //左转
40 case '4' : gpio_reg_write(0x9c); break; //右转
41 case 'q' : exit(0); break;
42 default : gpio_reg_write(0x00); break; //停止
43 }
44 }
45 }
46
47
48 exit(0);
49
50 return 0;
51 }
```
应用通过中断的方式从蓝牙端获取指令,实现对小车的控制功能。
- 切换到lab4_2继承lab4_1及之前实验所做的修改并make后的直接运行结果
```
//切换到lab4_2
$ git checkout lab4_2_PLIC
//继承lab4_1以及之前的答案
$ git merge lab4_1_poll -m "continue to work on lab4_2"
//重新构造
$ make clean; make
//运行构造结果
In m_start, hartid:0
HTIF is available!
(Emulated) memory size: 512 MB
Enter supervisor mode...
PKE kernel start 0x0000000080000000, PKE kernel end: 0x0000000080010000, PKE kernel size: 0x0000000000010000 .
free physical memory address: [0x0000000080010000, 0x000000008003ffff]
kernel memory manager is initializing ...
kernel pagetable addr is 0x000000008003e000
KERN_BASE 0x0000000080000000
physical address of _etext is: 0x0000000080005000
kernel page table is on
Switching to user mode...
in alloc_proc. user frame 0x0000000080039000, user stack 0x000000007ffff000, user kstack 0x0000000080038000
User application is loading.
Application: app_poll
CODE_SEGMENT added at mapped info offset:3
Application program entry point (virtual address): 0x00000000810000de
going to insert process 0 to ready queue.
going to schedule process 0 to run.
please input the instruction through bluetooth!
You need to implement the uart_getchar function in lab4_2 here!
System is shutting down with exit code -1.
```
<a name="lab4_2_content"></a>
#### **实验内容**
如输出提示所表示的那样需要找到并完成对uartgetchardo_sleepgetuartvalue的调用并获得以下预期结果
```
In m_start, hartid:0
HTIF is available!
(Emulated) memory size: 512 MB
Enter supervisor mode...
PKE kernel start 0x0000000080000000, PKE kernel end: 0x0000000080010000, PKE kernel size: 0x0000000000010000 .
free physical memory address: [0x0000000080010000, 0x000000008003ffff]
kernel memory manager is initializing ...
kernel pagetable addr is 0x000000008003e000
KERN_BASE 0x0000000080000000
physical address of _etext is: 0x0000000080005000
kernel page table is on
Switching to user mode...
in alloc_proc. user frame 0x0000000080039000, user stack 0x000000007ffff000, user kstack 0x0000000080038000
User application is loading.
Application: app_polling
CODE_SEGMENT added at mapped info offset:3
Application program entry point (virtual address): 0x00000000810000de
going to insert process 0 to ready queue.
going to schedule process 0 to run.
Ticks 0
please input the instruction through bluetooth!
Ticks 1
User exit with code:0.
no more ready processes, system shutdown now.
System is shutting down with exit code 0.
```
<a name="lab4_2_guide"></a>
#### **实验指导**
对于本实验的应用我们需要在lab4_1基础上实现基于中断的uartgetchar。对代码进行跟踪我们可以在kernel/syscall.h中查看新增的系统调用以及编号
```
16 #define SYS_user_uart_putchar (SYS_user_base + 6)
17 #define SYS_user_uart_getchar (SYS_user_base + 7)
18 #define SYS_user_gpio_reg_write (SYS_user_base + 8)
```
继续追踪我们发现在kernel/syscall.c的do_syscall函数中新增了对应系统调用编号的实现函数对于新增系统调用分别有如下函数进行处理
```
139 case SYS_user_uart_getchar:
140 return sys_user_uart_getchar();
```
你的任务即为在kernel/syscall.c中追踪并完善对应的函数。
在kernel/syscall.c中找到函数实现空缺并根据注释完成uart系统调用
```
88 //
89 // implement the uart syscall
90 //
103 ssize_t sys_user_uart_getchar() {
104 panic( "You need to implement the uart_getchar function in lab4_2 here.\n" );
105 //sleep
106
107 //Wait for wake
108
109 //get the value
110
111 //Return the result character
112
113
114 }
```
当蓝牙有数据发送时pke会收到外部中断你需要完成接收到外部中断后的处理。
在kernel/strap.c中找到函数空缺并根据注释完成中断处理函数
```
100 case CAUSE_MEXTERNEL_S_TRAP:
101 {
102 panic( "You need to complete case CAUSE_MEXTERNEL_S_TRAP function in lab4_2 here.\n"
103 int irq = *(uint32 *)0xc201004L;
104 *(uint32 *)0xc201004L = irq;
105 volatile int *ctrl_reg = (void *)(uintptr_t)0x6000000c;
106 *ctrl_reg = *ctrl_reg | (1 << 4);
107
108 // get the data from MMIO.
109 // send it to the process.
110 // call function to awake process[0]
111
112
113
114 break;
115 }
```
在kernel/process.c中找到函数实现空缺并根据注释完成do_sleep函数
```
221 void do_sleep(){
222 panic( "You need to implement do_sleep function in lab4_2 here.\n"
223 // set the process BLOCKED.
224 }
```
在kernel/process.c中找到函数实现空缺并根据注释完成do_wake函数
```
226 void do_wake(){
227 panic( "You need to implement do_sleep function in lab4_2 here.\n"
228 //set the process READY.
229 //insert_to_ready_queue
230
231 //schedule
232 }
```
**实验完毕后,记得提交修改(命令行中-m后的字符串可自行确定以便在后续实验中继承lab4_2中所做的工作**
```
$ git commit -a -m "my work on lab4_2 is done."
```
<a name="hostdevice"></a>
## 6.4 lab4_3
<a name="lab4_3_app"></a>
#### **给定应用**
- user/app_host_device.c
```
1 #pragma pack(4)
2 #define _SYS__TIMEVAL_H_
3 struct timeval {
4 unsigned int tv_sec;
5 unsigned int tv_usec;
6 };
7
8 #include "user_lib.h"
9 #include "videodev2.h"
10 #define DARK 64
11 #define RATIO 7 / 10
12
13 int main() {
14 char *info = allocate_share_page();
15 int pid = do_fork();
16 if (pid == 0) {
17 int f = do_open("/dev/video0", O_RDWR), r;
18
19 struct v4l2_format fmt;
20 fmt.type = V4L2_BUF_TYPE_VIDEO_CAPTURE;
21 fmt.fmt.pix.pixelformat = V4L2_PIX_FMT_YUYV;
22 fmt.fmt.pix.width = 320;
23 fmt.fmt.pix.height = 180;
24 fmt.fmt.pix.field = V4L2_FIELD_NONE;
25 r = do_ioctl(f, VIDIOC_S_FMT, &fmt);
26 printu("Pass format: %d\n", r);
27
28 struct v4l2_requestbuffers req;
29 req.type = V4L2_BUF_TYPE_VIDEO_CAPTURE;
30 req.count = 1; req.memory = V4L2_MEMORY_MMAP;
31 r = do_ioctl(f, VIDIOC_REQBUFS, &req);
32 printu("Pass request: %d\n", r);
33
34 struct v4l2_buffer buf;
35 buf.type = V4L2_BUF_TYPE_VIDEO_CAPTURE;
36 buf.memory = V4L2_MEMORY_MMAP; buf.index = 0;
37 r = do_ioctl(f, VIDIOC_QUERYBUF, &buf);
38 printu("Pass buffer: %d\n", r);
39
40 int length = buf.length;
41 char *img = do_mmap(NULL, length, PROT_READ | PROT_WRITE, MAP_SHARED, f, buf.m.offset);
42 unsigned int type = V4L2_BUF_TYPE_VIDEO_CAPTURE;
43 r = do_ioctl(f, VIDIOC_STREAMON, &type);
44 printu("Open stream: %d\n", r);
45
46 char *img_data = allocate_page();
47 for (int i = 0; i < (length + 4095) / 4096 - 1; i++)
48 allocate_page();
49 yield();
50
51 for (;;) {
52 if (*info == '1') {
53 r = do_ioctl(f, VIDIOC_QBUF, &buf);
54 printu("Buffer enqueue: %d\n", r);
55 r = do_ioctl(f, VIDIOC_DQBUF, &buf);
56 printu("Buffer dequeue: %d\n", r);
57 r = read_mmap(img_data, img, length);
58 int num = 0;
59 for (int i = 0; i < length; i += 2)
60 if (img_data[i] < DARK) num++;
61 printu("Dark num: %d > %d\n", num, length / 2 * RATIO);
62 if (num > length / 2 * RATIO) {
63 *info = '0'; gpio_reg_write(0x00);
64 }
65 } else if (*info == 'q') break;
66 }
67
68 for (char *i = img_data; i - img_data < length; i += 4096)
69 free_page(i);
70 r = do_ioctl(f, VIDIOC_STREAMOFF, &type);
71 printu("Close stream: %d\n", r);
72 do_munmap(img, length); do_close(f); exit(0);
73 } else {
74 yield();
75 for (;;) {
76 char temp = (char)uartgetchar();
77 printu("From bluetooth: %c\n", temp);
78 *info = temp;
79 switch (temp) {
80 case '1': gpio_reg_write(0x2e); break; //前进
81 case '2': gpio_reg_write(0xd1); break; //后退
82 case '3': gpio_reg_write(0x63); break; //左转
83 case '4': gpio_reg_write(0x9c); break; //右转
84 case 'q': exit(0); break;
85 default: gpio_reg_write(0x00); break; //停止
86 }
87 }
88 }
89 return 0;
90 }
```
该用户程序包含两个进程其中主进程和实验4_2类似负责接收蓝牙发送过来的数据根据数据控制小车行动前进、后退、左转、右转、停止子进程则负责拍摄和分析首先初始化摄像头设备然后是个死循环判断摄像头拍摄的图像数据如果当前小车处于前进状态则拍摄然后检查数据如果判断前面有障碍物则控制车轮停转刹车否则如果主进程退出了则自己进行释放文件、内存、关闭设备等操作再退出。在用户程序操控摄像头的过程中使用了ioctl、mmap、munmap等系统调用需对其进行完善从而实现小车的障碍识别和停止功能。
- 切换到lab4_3、继承lab4_2中所做修改并make后的直接运行结果
```
//切换到lab4_2
$ git checkout lab4_2_PLIC
//继承lab3_3以及之前的答案
$ git merge lab4_2_PLIC -m "continue to work on lab4_2"
//重新构造
$ make clean; make
//运行构造结果
In m_start, hartid:0
HTIF is available!
(Emulated) memory size: 512 MB
Enter supervisor mode...
PKE kernel start 0x0000000080000000, PKE kernel end: 0x0000000080010000, PKE kernel size: 0x0000000000010000 .
free physical memory address: [0x0000000080010000, 0x000000008003ffff]
kernel memory manager is initializing ...
kernel pagetable addr is 0x000000008003e000
KERN_BASE 0x0000000080000000
physical address of _etext is: 0x0000000080005000
kernel page table is on
Switching to user mode...
in alloc_proc. user frame 0x0000000080039000, user stack 0x000000007ffff000, user kstack 0x0000000080038000
User application is loading.
Application: app_PLIC
CODE_SEGMENT added at mapped info offset:3
Application program entry point (virtual address): 0x00000000810000de
going to insert process 0 to ready queue.
going to schedule process 0 to run.
please input the instruction through bluetooth!
You need to implement the uart_getchar function in lab4_3 here!
System is shutting down with exit code -1.
```
<a name="lab4_3_content"></a>
#### 实验内容
如应用提示所表示的那样读者需要找到并完成对ioctl的调用使得用户能够设置设备参数从而控制摄像头实现拍照等功能获取图片后检查数据从而判断前方是否出现障碍物。
跟踪相关系统调用在kernel/file.c里可以看到需要补充的函数
```
25 int do_open(char *pathname, int flags) {
26 // TODO (lab4_3): call host open through spike_file_open and then bind fd to spike_file
27 // hint: spike_file_dup function can bind spike_file_t to an int fd.
28 panic( "You need to finish open function in lab4_3.\n" );
29 }
```
```
39 int do_ioctl(int fd, uint64 request, char *data) {
40 // TODO (lab4_3): call host ioctl through frontend_sycall
41 // hint: fronted_syscall ioctl argument:
42 // 1.call number
43 // 2.fd
44 // 3.the order to device
45 // 4.data address
46 panic( "You need to call host's ioctl by frontend_syscall in lab4_3.\n" );
47 return frontend_syscall(HTIFSYS_ioctl, spike_file_get(fd)->kfd,
48 request, (uint64)data, 0, 0, 0, 0);
49 }
```
实验预期结果:小车在前进过程中能够正常识别障碍物后并自动停车。
<a name="lab4_3_guide"></a>
#### 实验指导
##### 摄像头控制
USB摄像头最基础的控制方法是使用读写设备文件的方式。拍摄一张照片包含以下过程
* 打开设备文件使用open函数
* 设置设备参数使用ioctl函数
* 映射内存由于USB摄像头对应的设备文件不支持直接用read函数进行读写所以需要用mmap函数将文件映射到一段虚拟地址通过虚拟地址进行读写
* 拍摄使用ioctl函数控制
* 结束和清理包含使用ioctl函数关闭设备使用munmap函数解映射使用close函数关闭设备文件
其中ioctl、mmap、munmap三个函数是PKE和riscv-fesvr不支持的需要在本设计中添加因此本设计的重点包含以下三个内容
* 对riscv-fesvr的修改riscv-fesvr是PS端Arm的一个程序用于控制PL端Riscv程序的启动以及和通信。通过riscv-fesvrPL端上的程序也可以访问PS端的文件调用一些PS端系统的函数。原版的riscv-fesvr不支持ioctl和mmap等函数而操控USB摄像头的用户程序必须使用这些函数所以需要对riscv-fesvr进行修改使得PL端上运行的PKE和用户程序能够通过riscv-fesvr这个中间层调用宿主机的系统函数从而控制摄像头
* 对PKE内核代码的修改需要为riscv-fesvr新增的函数调用提供用户层接口。
* 对用户代码的修改有了对fesvr和内核的修改用户程序就可以调用各类系统调用函数操控摄像机了。为了实现避障的功能程序还需要对获得的图片信息进行解码和分析根据前方是否为障碍物选择是否刹车。
##### 图片解析
应用第21行可以看到我们从摄像头获取的数据是YUYV格式读者可进行查阅它用灰度、蓝色色度、红色色度三个属性表示颜色每个像素点都有灰度属性。由于我们分析障碍物只需要灰度图所以取每个像素点的灰度属性即可。
因此对于获取过来的数据删去奇数索引的数据就可以得到灰度图。对于障碍物的判断我们使用了一个比较简单的算法计算灰度小于64的像素点个数如果个数大于像素点总数的7/10即认为前方是障碍物。
**注意:对于灰度阈值的设定可根据环境亮度进行一定的调整,可以先根据摄像头返回的图像进行分析,计算出对应障碍物的灰度值;灰度阈值越精确,小车对于障碍物的识别将越灵敏,并能在合理的距离内识别到障碍物并停车。**
**实验完毕后,记得提交修改(命令行中-m后的字符串可自行确定以便在后续实验中继承lab4_3中所做的工作**
```
$ git commit -a -m "my work on lab4_3 is done."
```