一 基本概述
esp8266的SPI代码流程非常的清晰,主要有三部分构成: spi_init 配置 spi_trans 配置 data_transfer 配置这三块组成。
在这里,笔者就针对spi的这些流程,做一个简单的源码分析。
一 初始化源码分析
spi 源码初始化函数中,主要是完成软硬件的接口配置和参数配置,我们看一下这里面都做了一些什么呢?
虽然代码不少,但是一个函数的核心代码也就那么多:
spi_set_event_callback(host, &config->event_cb);
spi_set_mode(host, &config->mode);
spi_set_interface(host, &config->interface);
spi_set_clk_div(host, &config->clk_div);
spi_set_dummy(host, &dummy_bitlen);
spi_set_intr_enable(host, &config->intr_enable);
spi_intr_register(spi_intr, NULL);
spi_intr_enable();
这代码逻辑很清楚了:
spi_set_mode是设置是master模式还是slave模式。这里面主要涉两种模式的参数配置,记得改参数一定要用这两个啊:
核心代码如下所示:
if (SPI_MASTER_MODE == *mode) {
// Set to Master mode
SPI[host]->pin.slave_mode = false;
SPI[host]->slave.slave_mode = false;
// Master uses the entire hardware buffer to improve transmission speed
SPI[host]->user.usr_mosi_highpart = false;
SPI[host]->user.usr_miso_highpart = false;
SPI[host]->user.usr_mosi = true;
// Create hardware cs in advance
SPI[host]->user.cs_setup = true;
// Hysteresis to keep hardware cs
SPI[host]->user.cs_hold = true;
SPI[host]->user.duplex = true;
SPI[host]->user.ck_i_edge = true;
SPI[host]->ctrl2.mosi_delay_num = 0;
SPI[host]->ctrl2.miso_delay_num = 1;
} else {
// Set to Slave mode
SPI[host]->pin.slave_mode = true;
SPI[host]->slave.slave_mode = true;
SPI[host]->user.usr_miso_highpart = true;
// MOSI signals are delayed by APB_CLK(80MHz) mosi_delay_num cycles
SPI[host]->ctrl2.mosi_delay_num = 2;
SPI[host]->ctrl2.miso_delay_num = 0;
SPI[host]->slave.wr_rd_sta_en = 1;
SPI[host]->slave1.status_bitlen = 31;
SPI[host]->slave1.status_readback = 0;
// Put the slave's miso on the highpart, so you can only send 256bits
// In Slave mode miso, mosi length is the same
SPI[host]->slave1.buf_bitlen = 255;
SPI[host]->cmd.usr = 1;
}
接下来是软硬件接口的配置,不同模式下接口参数不同,GPIO是不变的:
switch (host) {
case CSPI_HOST: {
// Initialize SPI IO
PIN_PULLUP_EN(PERIPHS_IO_MUX_SD_CLK_U);
PIN_FUNC_SELECT(PERIPHS_IO_MUX_SD_CLK_U, FUNC_SPICLK);
if (interface->mosi_en) {
PIN_PULLUP_EN(PERIPHS_IO_MUX_SD_DATA1_U);
PIN_FUNC_SELECT(PERIPHS_IO_MUX_SD_DATA1_U, FUNC_SPID_MOSI);
}
if (interface->miso_en) {
PIN_PULLUP_EN(PERIPHS_IO_MUX_SD_DATA0_U);
PIN_FUNC_SELECT(PERIPHS_IO_MUX_SD_DATA0_U, FUNC_SPIQ_MISO);
}
if (interface->cs_en) {
PIN_PULLUP_EN(PERIPHS_IO_MUX_SD_CMD_U);
PIN_FUNC_SELECT(PERIPHS_IO_MUX_SD_CMD_U, FUNC_SPICS0);
}
}
break;
case HSPI_HOST: {
// Initialize HSPI IO
PIN_PULLUP_EN(PERIPHS_IO_MUX_MTMS_U);
PIN_FUNC_SELECT(PERIPHS_IO_MUX_MTMS_U, FUNC_HSPI_CLK); //GPIO14 is SPI CLK pin (Clock)
if (interface->mosi_en) {
PIN_PULLUP_EN(PERIPHS_IO_MUX_MTCK_U);
PIN_FUNC_SELECT(PERIPHS_IO_MUX_MTCK_U, FUNC_HSPID_MOSI); //GPIO13 is SPI MOSI pin (Master Data Out)
}
if (interface->miso_en) {
PIN_PULLUP_EN(PERIPHS_IO_MUX_MTDI_U);
PIN_FUNC_SELECT(PERIPHS_IO_MUX_MTDI_U, FUNC_HSPIQ_MISO); //GPIO12 is SPI MISO pin (Master Data In)
}
if (interface->cs_en) {
PIN_PULLUP_EN(PERIPHS_IO_MUX_MTDO_U);
PIN_FUNC_SELECT(PERIPHS_IO_MUX_MTDO_U, FUNC_HSPI_CS0);
}
}
break;
}
接下来是分频系数的设置,这个岁spi的速影响特别大,记住,spi的速率就靠它了:
esp_err_t spi_set_clk_div(spi_host_t host, spi_clk_div_t *clk_div)
switch (host) {
case CSPI_HOST: {
SET_PERI_REG_MASK(PERIPHS_IO_MUX_CONF_U, SPI0_CLK_EQU_SYS_CLK);
}
break;
case HSPI_HOST: {
SET_PERI_REG_MASK(PERIPHS_IO_MUX_CONF_U, SPI1_CLK_EQU_SYS_CLK);
}
break;
}
SPI[host]->clock.clk_equ_sysclk = true;
接下来就是一些简单的流程了,就是配置中断并使能:
spi_set_intr_enable(host, &config->intr_enable);
spi_intr_register(spi_intr, NULL);
spi_intr_enable();
二 配置源码分析
在配置源码中,最核心的函数就是:spi_trans(HSPI_HOST, trans);
在这个函数中,都做了一些什么呢?
if (SPI_MASTER_MODE == spi_object[host]->mode) {
ret = spi_master_trans(host, trans);
} else {
ret = spi_slave_trans(host, trans);
}
看上面源码,就知道了,所有的核心处理都在函数:spi_xxx_trans里面了。
接下来,我们看这个函数都干了啥:
// Set the cmd length and transfer cmd
if (trans.bits.cmd && trans.cmd) {
SPI[host]->user.usr_command = 1;
SPI[host]->user2.usr_command_bitlen = trans.bits.cmd - 1;
SPI[host]->user2.usr_command_value = *trans.cmd;
} else {
SPI[host]->user.usr_command = 0;
}
// Set addr length and transfer addr
if (trans.bits.addr && trans.addr) {
SPI[host]->user.usr_addr = 1;
SPI[host]->user1.usr_addr_bitlen = trans.bits.addr - 1;
SPI[host]->addr = *trans.addr;
} else {
SPI[host]->user.usr_addr = 0;
}
// Set mosi length and transmit mosi
if (trans.bits.mosi && trans.mosi) {
SPI[host]->user.usr_mosi = 1;
SPI[host]->user1.usr_mosi_bitlen = trans.bits.mosi - 1;
for (x = 0; x < trans.bits.mosi; x += 32) {
y = x / 32;
SPI[host]->data_buf[y] = trans.mosi[y];
}
} else {
SPI[host]->user.usr_mosi = 0;
}
// Set the length of the miso
if (trans.bits.miso && trans.miso) {
SPI[host]->user.usr_miso = 1;
SPI[host]->user1.usr_miso_bitlen = trans.bits.miso - 1;
} else {
SPI[host]->user.usr_miso = 0;
}
// Call the event callback function to send a transfer start event
if (spi_object[host]->event_cb) {
spi_object[host]->event_cb(SPI_TRANS_START_EVENT, NULL);
}
// Start transmission
SPI[host]->cmd.usr = 1;
大致看一下,就知道了,这个函数就是做数据传输配置的。
三 数据传输源码分析
在数据接收和发送里面,这有两个东西特别关键:
为了节省buffer空间和提升速率,这里使用了ring_buffer,这个在嵌入式中是一个非常实用的技巧,笔者见过几个系统都用过,非常的不错,有兴趣的同学可以学着写一个。
// Write data to the ESP8266 Slave use "SPI_MASTER_WRITE_DATA_TO_SLAVE_CMD" cmd
trans.cmd = &cmd;
trans.addr = &addr;
trans.mosi = write_data;
cmd = SPI_MASTER_WRITE_DATA_TO_SLAVE_CMD;
addr = 0;
write_data[0] = 1;
write_data[1] = 0x11111111;
write_data[2] = 0x22222222;
write_data[3] = 0x33333333;
write_data[4] = 0x44444444;
write_data[5] = 0x55555555;
write_data[6] = 0x66666666;
write_data[7] = 0x77777777;
while (1) {
gettimeofday(&now, NULL);
time_start = now.tv_usec;
for (x = 0;x < 100;x++) {
spi_trans(HSPI_HOST, trans);
write_data[0]++;
}
gettimeofday(&now, NULL);
time_end = now.tv_usec;
ESP_LOGI(TAG, "Master wrote 3200 bytes in %d us", (int)(time_end - time_start));
vTaskDelay(100 / portTICK_RATE_MS);
}
其实,这段代码不是很复杂,估计很多人都能看懂的,真是代码写的特别好。
四 总结感想
SPI接口应用实在是太广泛了,后来笔者使用两个系统对通,验证基本没啥问题,后面就是上一下图了,接下来,就可以上我们的牛逼系统了,wifi图传,就看同的速率了。
硬件连线:
master log:
slave log:
来源:oschina
链接:https://my.oschina.net/u/4365283/blog/4278195