一直想花时间来整理一下Linux内核LCD驱动，却一直都忙着做其他事情去了，这些天特意抽出时间来整理之前落下的笔记，故事就这样开始了。LCD驱动也是字符设备驱动的一种，框架上相对于字符设备驱动稍微复杂一点点，真的就是一点点，难点在对LCD硬件的配置上。
开发平台：TQ210，S5PV210处理器
内核版本：linux-3.10.46
LCD型号：AT070TN92，7英寸，TFT屏，分辨率800x480x3(RGB)，24位真彩色
一、框架分析
上图说明：①内核装载LCD驱动模块：设置并注册fb_info结构，初始化LCD硬件。②APP打开LCD设备，获取设备文件，根据设备文件进行读写显存。③在内核中，根据主设备号和次设备号定位一个fb_info结构，如果应用层的系统调用是读操作则调用fb_ops中对应的操作函数，写操作也是一样。
读写等操作代码分析：
1 fb_open 2 { 3 int fbidx = iminor(inode); //获取次设备号 4 struct fb_info *info; 5 info = get_fb_info(fbidx); 6 struct fb_info *fb_info; 7 fb_info = registered_fb[fbidx];//根据次设备号从已注册的fb_info数组中获取响应的结构 8 return fb_info; 9 ...... 10 /* 11 * 从registered_fb[]数组项里找到fb_info结构体后，将其保存到 12 * struct file结构中的私有信息成员，难道这是为了以后在某些情况方便找到并调用？？先放着... 13 * 回过来发现：这样做是为了验证在read、write、ioctl等系统调用中获得的fb_info结构和open获得的是否一样 14 */ 15 file->private_data = info; 16 //info->fbops->fb_open无定义，这是值得思考的问题！ 17 if (info->fbops->fb_open) { 18 res = info->fbops->fb_open(info,1); 19 if (res) 20 module_put(info->fbops->owner); 21 } 22 ...... 23 }
1 fb_read(struct file *file, char __user *buf, size_t count, loff_t *ppos) 2 { 3 struct fb_info *info = file_fb_info(file); 4 struct inode *inode = file_inode(file); 5 int fbidx = iminor(inode); 6 //也是根据次设备号来获取fb_info结构 7 struct fb_info *info = registered_fb[fbidx]; 8 9 if (info != file->private_data) 10 info = NULL; 11 return info; 12 //无定义 13 if (info->fbops->fb_read) 14 return info->fbops->fb_read(info, buf, count, ppos); 15 //获得显存的大小 16 total_size = info->screen_size; 17 //如果应用层要读的数据count比实际最大的显存还要大，修改count值为最大显存值 18 if (count >= total_size) 19 count = total_size; 20 //分配显存，最大只能是一页PAGE_SIZE=4KB 21 buffer = kmalloc((count > PAGE_SIZE) ? PAGE_SIZE : count,GFP_KERNEL); 22 //要读的源地址：显存虚拟基地址+偏移 23 src = (u8 __iomem *) (info->screen_base + p); 24 while (count) { 25 c = (count > PAGE_SIZE) ? PAGE_SIZE : count; 26 //读的目的地址 27 dst = buffer; 28 //读操作：拷贝数据 29 fb_memcpy_fromfb(dst, src, c); 30 dst += c; 31 src += c; 32 33 if (copy_to_user(buf, buffer, c)) { 34 err = -EFAULT; 35 break; 36 } 37 *ppos += c; 38 buf += c; 39 cnt += c; 40 count -= c; 41 } 42 kfree(buffer); //释放buffer，只起到临时中转站的作用 43 }
1 /* 2 * 这里分配的显存是在内核空间分配的，用户空间并不能直接访问， 3 * 所以需要用到这里的mmap函数，直接将这段内存空间映射到 4 * 用户空间去，用户空间就能访问这段内存空间了。 5 */ 6 static int fb_mmap(struct file *file, struct vm_area_struct * vma) 7 { 8 struct fb_info *info = file_fb_info(file); 9 struct fb_ops *fb; 10 unsigned long mmio_pgoff; 11 unsigned long start; 12 u32 len; 13 14 if (!info) 15 return -ENODEV; 16 fb = info->fbops; 17 if (!fb) 18 return -ENODEV; 19 mutex_lock(&info->mm_lock); 20 //如果fb_info->fbops->fb_mmap存在就调用该函数，实际中没有！ 21 if (fb->fb_mmap) { 22 int res; 23 res = fb->fb_mmap(info, vma); 24 mutex_unlock(&info->mm_lock); 25 return res; 26 } 27 28 /* 29 * Ugh. This can be either the frame buffer mapping, or 30 * if pgoff points past it, the mmio mapping. 31 */ 32 start = info->fix.smem_start; //fb缓冲内存的开始位置(物理地址) 33 len = info->fix.smem_len; 34 mmio_pgoff = PAGE_ALIGN((start & ~PAGE_MASK) + len) >> PAGE_SHIFT; 35 if (vma->vm_pgoff >= mmio_pgoff) { 36 if (info->var.accel_flags) { 37 mutex_unlock(&info->mm_lock); 38 return -EINVAL; 39 } 40 41 vma->vm_pgoff -= mmio_pgoff; 42 start = info->fix.mmio_start; 43 len = info->fix.mmio_len; 44 } 45 mutex_unlock(&info->mm_lock); 46 47 vma->vm_page_prot = vm_get_page_prot(vma->vm_flags); 48 fb_pgprotect(file, vma, start); 49 //映射物理内存到用户空间虚拟地址 50 return vm_iomap_memory(vma, start, len); 51 }
问题思考：
问1.什么叫帧缓冲区，他有哪些特性指标？
答1.对于应用层来说，显示图像到LCD设备就相当于往“一块内存”中写入数据，获取LCD设备上的图像就相当于拷贝“这块内存”中的数据。因此，LCD就和“一块内存”一样，专业一点术语叫帧缓冲区，它和普通的内存不太一样，除了可以“读写”操作之外还可以进行其他操作和功能设置，特性指标就是LCD的特性指标。在内核中，一个LCD显示器就相当于一个帧缓冲设备，对应一个fb_info结构。
问2.为什么要通过 registered_fb[] 数组来找到对应的 fb_info 结构体？
答2.通过对上边这几个函数的剖析发现，不管是fb_read、fb_write、fb_ioctl、fb_mmap系统调用，都是通过次设备号在已注册的fb_info结构数组中找到匹配的那一个结构之后，判断其中的fbops结构中的操作函数是否有定义，有的话就优先调用该函数，没有就使用往下的方案策略。这样的好处就是多个相同的LCD设备可以使用同一套代码，减少代码的重复性，同时对于需要特殊定义的函数又可以方便实现重定义。
问3.这个数组在哪里被注册？
答3.在register_framebuffer()函数中被注册 register_framebuffer(struct fb_info *fb_info) ret = do_register_framebuffer(fb_info); ...... registered_fb[i] = fb_info; ......
问4.fb_mmap()函数在什么场合使用？
答4.在用户空间中通过mmap()函数来进行系统调用，该函数执行成功返回的是指向被映射的帧缓冲区的指针，这样用户直接可以通过该指针来读写缓冲区。
二、驱动代码编写
#include <linux/module.h> #include <linux/kernel.h> #include <linux/errno.h> #include <linux/string.h> #include <linux/mm.h> #include <linux/slab.h> #include <linux/delay.h> #include <linux/fb.h> #include <linux/init.h> #include <linux/dma-mapping.h> #include <linux/interrupt.h> #include <linux/platform_device.h> #include <linux/clk.h> #include <linux/workqueue.h> #include <asm/io.h> #include <asm/div64.h> #include <asm/uaccess.h> #include <asm/mach/map.h> #include <mach/regs-gpio.h> #include <linux/fb.h> #define VSPW 9 //4 #define VBPD 13 //17 #define LINEVAL 479 #define VFPD 21 //26 #define HSPW 19 //4 #define HBPD 25 //40 #define HOZVAL 799 #define HFPD 209 //214 #define LeftTopX 0 #define LeftTopY 0 #define RightBotX 799 #define RightBotY 479 static struct fb_info *clb_fbinfo; /* LCD GPIO Pins */ static long unsigned long *gpf0con; static long unsigned long *gpf1con; static long unsigned long *gpf2con; static long unsigned long *gpf3con; static long unsigned long *gpd0con; static long unsigned long *gpd0dat; static long unsigned long *display_control; /* LCD Controler Pins */ struct s5pv210_lcd_regs{ volatile unsigned long vidcon0; volatile unsigned long vidcon1; volatile unsigned long vidcon2; volatile unsigned long vidcon3; volatile unsigned long vidtcon0; volatile unsigned long vidtcon1; volatile unsigned long vidtcon2; volatile unsigned long vidtcon3; volatile unsigned long wincon0; volatile unsigned long wincon1; volatile unsigned long wincon2; volatile unsigned long wincon3; volatile unsigned long wincon4; volatile unsigned long shadowcon; volatile unsigned long reserve1[2]; volatile unsigned long vidosd0a; volatile unsigned long vidosd0b; volatile unsigned long vidosd0c; }; struct clk *lcd_clk; static struct s5pv210_lcd_regs *lcd_regs; static long unsigned long *vidw00add0b0; static long unsigned long *vidw00add1b0; static u32 pseudo_palette[16]; /* from pxafb.c */ static unsigned int chan_to_field(unsigned int chan, struct fb_bitfield *bf) { chan &= 0xffff; chan >>= 16 - bf->length; return chan << bf->offset; } static int clb210_lcdfb_setcolreg(unsigned regno, unsigned red, unsigned green, unsigned blue, unsigned transp, struct fb_info *info) { unsigned int val; if (regno > 16) return 1; /* 用red,green,blue三原色构造出val */ val = chan_to_field(red, &info->var.red); val |= chan_to_field(green, &info->var.green); val |= chan_to_field(blue, &info->var.blue); pseudo_palette[regno] = val; return 0; } //帧缓冲操作函数 static struct fb_ops clb210_lcdfb_ops = { .owner = THIS_MODULE, .fb_setcolreg = clb210_lcdfb_setcolreg, //下面这3个函数是通用的 .fb_fillrect = cfb_fillrect, .fb_copyarea = cfb_copyarea, .fb_imageblit = cfb_imageblit, }; static int __init clb210_lcd_init(void) { /* 1.分配一个fb_info */ clb_fbinfo = framebuffer_alloc(0 , NULL); /* 2. 设置 */ /* 2.1 设置固定的参数 */ strcpy(clb_fbinfo->fix.id, "clb210_lcd"); clb_fbinfo->fix.smem_len = 800 * 480 * 32/8; clb_fbinfo->fix.type = FB_TYPE_PACKED_PIXELS; clb_fbinfo->fix.visual = FB_VISUAL_TRUECOLOR; clb_fbinfo->fix.line_length = 800 * 32/8; /* 2.2 设置可变的参数 */ clb_fbinfo->var.xres = 800; clb_fbinfo->var.yres = 480; clb_fbinfo->var.xres_virtual = 800; clb_fbinfo->var.yres_virtual = 480; clb_fbinfo->var.bits_per_pixel = 32; /*RGB:888*/ clb_fbinfo->var.red.offset = 16; clb_fbinfo->var.red.length = 8; clb_fbinfo->var.green.offset = 8; clb_fbinfo->var.green.length = 8; clb_fbinfo->var.blue.offset = 0; clb_fbinfo->var.blue.length = 8; clb_fbinfo->var.activate = FB_ACTIVATE_NOW ; /* 2.3 设置操作函数 */ clb_fbinfo->fbops = &clb210_lcdfb_ops; /* 2.4 其他的设置 */ /* 2.4.1 设置显存的大小 */ clb_fbinfo->screen_size = 800 * 480 * 32/8; /* 2.4.2 设置调色板 */ clb_fbinfo->pseudo_palette = pseudo_palette; /* 2.4.3 设置显存的虚拟起始地址 */ clb_fbinfo->screen_base = dma_alloc_writecombine(NULL, clb_fbinfo->fix.smem_len, (u32*)&(clb_fbinfo->fix.smem_start), GFP_KERNEL); /* 3. 硬件相关的操作 */ /* 3.1 获取lcd时钟，使能时钟 */ lcd_clk = clk_get(NULL, "lcd"); if (!lcd_clk || IS_ERR(lcd_clk)) { printk(KERN_INFO "failed to get lcd clock source\n"); } clk_enable(lcd_clk); /* 3.2 配置GPIO用于LCD */ gpf0con = ioremap(0xE0200120, 4); gpf1con = ioremap(0xE0200140, 4); gpf2con = ioremap(0xE0200160, 4); gpf3con = ioremap(0xE0200180, 4); gpd0con = ioremap(0xE02000A0, 4); gpd0dat = ioremap(0xE02000A4, 4); display_control = ioremap(0xe0107008, 4); /* 设置相关GPIO引脚用于LCD */ *gpf0con = 0x22222222; *gpf1con = 0x22222222; *gpf2con = 0x22222222; *gpf3con = 0x22222222; /* 使能LCD本身 */ *gpd0con |= 1<<4; *gpd0dat |= 1<<1; /* 显示路径的选择, 0b10: RGB=FIMD I80=FIMD ITU=FIMD */ *display_control = 2<<0; /* 3.3 映射LCD控制器对应寄存器 */ lcd_regs = ioremap(0xF8000000, sizeof(struct s5pv210_lcd_regs)); vidw00add0b0 = ioremap(0xF80000A0, 4); vidw00add1b0 = ioremap(0xF80000D0, 4); lcd_regs->vidcon0 &= ~((3<<26) | (1<<18) | (0xff<<6) | (1<<2)); lcd_regs->vidcon0 |= ((5<<6) | (1<<4) ); lcd_regs->vidcon1 &= ~(1<<7); /* 在vclk的下降沿获取数据 */ lcd_regs->vidcon1 |= ((1<<6) | (1<<5)); /* HSYNC极性反转, VSYNC极性反转 */ lcd_regs->vidtcon0 = (VBPD << 16) | (VFPD << 8) | (VSPW << 0); lcd_regs->vidtcon1 = (HBPD << 16) | (HFPD << 8) | (HSPW << 0); lcd_regs->vidtcon2 = (LINEVAL << 11) | (HOZVAL << 0); lcd_regs->wincon0 &= ~(0xf << 2); lcd_regs->wincon0 |= (0xB<<2)|(1<<15); lcd_regs->vidosd0a = (LeftTopX<<11) | (LeftTopY << 0); lcd_regs->vidosd0b = (RightBotX<<11) | (RightBotY << 0); lcd_regs->vidosd0c = (LINEVAL + 1) * (HOZVAL + 1); *vidw00add0b0 = clb_fbinfo->fix.smem_start; *vidw00add1b0 = clb_fbinfo->fix.smem_start + clb_fbinfo->fix.smem_len; lcd_regs->shadowcon = 0x1; /* 使能通道0 */ lcd_regs->vidcon0 |= 0x3; /* 开启总控制器 */ lcd_regs->wincon0 |= 1; /* 开启窗口0 */ /*4.注册*/ register_framebuffer(clb_fbinfo); return 0; } static void __exit clb210_lcd_exit(void) { unregister_framebuffer(clb_fbinfo); dma_free_writecombine(NULL, clb_fbinfo->fix.smem_len, clb_fbinfo->screen_base, clb_fbinfo->fix.smem_start); iounmap(gpf0con); iounmap(gpf1con); iounmap(gpf2con); iounmap(gpf3con); iounmap(gpd0con); iounmap(gpd0dat); iounmap(display_control); iounmap(lcd_regs); iounmap(vidw00add0b0); iounmap(vidw00add1b0); framebuffer_release(clb_fbinfo); } module_init(clb210_lcd_init); module_exit(clb210_lcd_exit); MODULE_LICENSE("GPL"); MODULE_AUTHOR("clb"); MODULE_DESCRIPTION("Lcd driver for clb210 board");
这份代码没有基于platform设备驱动来编写，在内核源码中的demo就是基于platform驱动模型来搭建的，主要内容其实一样。
三、编译测试
(1) 进入内核源码目录中，make menuconfig -> Device Drivers -> Graphics support -> [M]Support for frame buffer devices
重新编译内核 make uImage ，然后make modules，将driver/video/下的 fb.ko、cfbfillrect.ko、cfbimgblt.ko、cfbcopyarea.ko拷贝到210的根文件中，分别 insmod 安装到内核中。
(2) 安装lcd.ko驱动模块
(3) 应用层读写 frame buffer 测试程序
原文：http://www.cnblogs.com/lubiao/p/4850258.html
