Problems in SPI driver not setting DC correctly about lv_port_esp32

[kisvegabor]

Can you send a pull request to fix it?

[Boldie]

Ohh, I rewrote big parts of the driver part to support xpt2046 calibration, storing calibration to NVS, using a single spi bus (especially this is incompatible with yours, but can be changed to support both variants easily), KConfig, make gpio-pin configuration also working for JTAG pins, removed stack usage for spi_device_queue_trans (may create a crash), ...
But maybe it is of interest for others, I uploaded it as separate repo: https://github.com/Boldie/lv_drv

[kisvegabor]

I've examined this issue. As I see it can cause a problem only in ili9341_flush. However, LittlevGL won't call a new ili9341_flush while lv_disp_flush_ready() is not called in spi_ready. So theoretically it really would be wrong (as you mentioned), but int the practice it should work fine.

Do you think it still should be still fixed?

[Boldie]

No, this is not correct, because for an update you have to trigger more transfers for one flush:

	/*Column addresses*/
	ili9341_send_cmd(0x2A); <- GPIO DC needs to be 0
...
	ili9341_send_data(data, 4);  <- GPIO DC needs to be 1

	/*Page addresses*/
	ili9341_send_cmd(0x2B);   <- GPIO DC needs to be 1
...
	ili9341_send_data(data, 4);

	/*Memory write*/
	ili9341_send_cmd(0x2C);

And in my case, the DC was reset to early and the next command was not understood as it should be.

	gpio_set_level(CONFIG_LVGL_DRV_ILI9341_DC, 0);	 /*Command mode*/
	disp_spi_send_data(&cmd, 1); <<< This line returns immediately, but the transfer is not done. The DC can now be reset by the next ili9341_send_data and thus the command is not being recognized as command, it is recognized as data.

I have added a LA to the ESP and I saw the above described case. DC was set to 1 while the command is being sent out by the SPI interface.

[kisvegabor]

You are right. I'll update it.

[kisvegabor]

I've fixed it! Thank you for pointing out this issue.