回顾
上一篇说了rboot的加载流程,主要的是通过makefile将两个程序文件串了起来。这篇文章会对整个加载流程做详细讲解。
数据结构
typedef struct {
/* magic是常用的名称,用来标识这是个结构体,通常存在flash上,并且已经被初始化了 */
uint8_t magic; ///< Our magic, identifies rBoot configuration - should be BOOT_CONFIG_MAGIC
/* 用来说明当前的数据结构适用于哪个版本,不同版本常需要考虑兼容性问题 */
uint8_t version; ///< Version of configuration structure - should be BOOT_CONFIG_VERSION
/* 当前rboot的启动模式 */
uint8_t mode; ///< Boot loader mode (MODE_STANDARD | MODE_GPIO_ROM | MODE_GPIO_SKIP)
/* 当前选择的ROM区,但表示的是接下来要启动的ROM区 */
uint8_t current_rom; ///< Currently selected ROM (will be used for next standard boot)
/* MODE_GPIO_ROM模式下选择的ROM区 */
uint8_t gpio_rom; ///< ROM to use for GPIO boot (hardware switch) with mode set to MODE_GPIO_ROM
/* 可用的ROM总数 */
uint8_t count; ///< Quantity of ROMs available to boot
/* 占位,使前面长度够32位整数 */
uint8_t unused[2]; ///< Padding (not used)
/* 每个ROM区的地址 */
uint32_t roms[MAX_ROMS]; ///< Flash addresses of each ROM
#ifdef BOOT_CONFIG_CHKSUM
/* 本结构体的校验值 */
uint8_t chksum; ///< Checksum of this configuration structure (if BOOT_CONFIG_CHKSUM defined)
#endif
} rboot_config;
其实自身的注释就很详细了。
通常做Bootloader的话,都会维护自己的一个数据结构,这个结构体中放着boot配置和信息,并且存放在flash中(掉电保存)。这样做有几个目的:
知道使用哪个ROM区进行启动
可以配置ROM区地址(更灵活)
保存版本信息,使更易兼容
可以知道当前数据是否正确,有无损坏
检查固件
如果把uint32_t find_image(void)函数简化,是这样的:
uint32_t NOINLINE find_image(void) { uint8_t flag; uint32_t loadAddr; uint32_t flashsize; int32_t romToBoot; uint8_t updateConfig = 0; uint8_t buffer[SECTOR_SIZE]; rboot_config *romconf = (rboot_config*)buffer; rom_header *header = (rom_header*)buffer; ets_printf("\r\nrBoot v1.4.2 - richardaburton@gmail.com\r\n"); // read boot config SPIRead(BOOT_CONFIG_SECTOR * SECTOR_SIZE, buffer, SECTOR_SIZE); // fresh install or old version? if (romconf->magic != BOOT_CONFIG_MAGIC || romconf->version != BOOT_CONFIG_VERSION ) { // create a default config for a standard 2 rom setup ets_printf("Writing default boot config.\r\n"); // write new config sector } // try rom selected in the config, unless overriden by gpio/temp boot romToBoot = romconf->current_rom; // check valid rom number // gpio/temp boots will have already validated this if (romconf->current_rom >= romconf->count) { // if invalid rom selected try rom 0 ets_printf("Invalid rom selected, defaulting to 0.\r\n"); } // check rom is valid loadAddr = check_image(romconf->roms[romToBoot]); // check we have a good rom while (loadAddr == 0) { ets_printf("Rom %d at %x is bad.\r\n", romToBoot, romconf->roms[romToBoot]); // for normal mode try each previous rom // until we find a good one or run out updateConfig = 1; romToBoot--; if (romToBoot < 0) romToBoot = romconf->count - 1; if (romToBoot == romconf->current_rom) { // tried them all and all are bad! ets_printf("No good rom available.\r\n"); return 0; } loadAddr = check_image(romconf->roms[romToBoot]); } // re-write config, if required if (updateConfig) { romconf->current_rom = romToBoot; SPIEraseSector(BOOT_CONFIG_SECTOR); SPIWrite(BOOT_CONFIG_SECTOR * SECTOR_SIZE, buffer, SECTOR_SIZE); } ets_printf("Booting rom %d at %x, load addr %x.\r\n", romToBoot, romconf->roms[romToBoot], loadAddr); // copy the loader to top of iram ets_memcpy((void*)_text_addr, _text_data, _text_len); // return address to load from return loadAddr; }
很大一部分在做有效性判断,简化流程如下:
if(romconf->magic != BOOT_CONFIG_MAGIC) { return; } if(romconf->version != BOOT_CONFIG_VERSION) { return; } if(romconf->current_rom >= romconf->count) { return; } for(int i=0; i<romconf->count; i++) { loadAddr = check_image(romconf->roms[i]); if(loadAddr != 0) { break; } } if(loadAddr == 0) { return; } loader(loadAddr);
其中check_image主要是检测了ESP8266固件本身的有效性。
ESP8266生成的bin文件中,其实是包含了一些外部信息的,如flash模式,flash速度等,还用使用boot的版本,还有最主要的是生成文件的各段的信息。ESP8266有一些历史版本的boot,所以格式也稍有区别。加载固件
rboot-stage2a.c中:
usercode* NOINLINE load_rom(uint32_t readpos) { uint8_t sectcount; uint8_t *writepos; uint32_t remaining; usercode* usercode; rom_header header; section_header section; // read rom header SPIRead(readpos, &header, sizeof(rom_header)); readpos += sizeof(rom_header); // create function pointer for entry point usercode = header.entry; // copy all the sections for (sectcount = header.count; sectcount > 0; sectcount--) { // read section header SPIRead(readpos, §ion, sizeof(section_header)); readpos += sizeof(section_header); // get section address and length writepos = section.address; remaining = section.length; while (remaining > 0) { // work out how much to read, up to 16 bytes at a time uint32_t readlen = (remaining < READ_SIZE) ? remaining : READ_SIZE; // read the block SPIRead(readpos, writepos, readlen); readpos += readlen; // increment next write position writepos += readlen; // decrement remaining count remaining -= readlen; } } return usercode; }
前面说到,ESP8266生成的bin文件中包含了各段的信息,包括段地址、大小和内容。load_rom便是遍历bin文件中所有的段,从flash中读取,并加载到对应地址中。而如果没有Bootloader的话,这个工作是在ROM代码中执行的。
End
到这里,rboot的加载固件流程就讲完了。rboot还有不少特性,如GPIO选择ROM,加载多个ROM等,这些功能不太常用,这里就不打算说了。有需要的可以在评论区留言,我也会视情况继续写下去的。
那么下一篇,就是写zboot啦,作者说是基于rboot上做了改进,有了rboot的基础后,解读起来应该不会太难,大家一起加油啊。
本博客所有文章除特别声明外,均采用 CC BY-SA 3.0协议 。转载请注明出处!