Feature/esp32c3 support

[C47D]

@dastarling Could you take a look at this and test it when you have time?

Fixes lvgl/lv_port_esp32#269

[C47D]

@kisvegabor touch_driver got an update for LVGLv8, do you think we need any other update for that for this particular PR?

[kisvegabor]

Looks good in general, only a few comments above.

On a general note, for v8 instead of adding #if LVGL_VERSION_MAJOR >= 8 I started to keep only v8 stuff because people who use v7 can use the release/v7 branch.

I'm not sure how well could it work here as the goal is to have as many drivers for as many releases as possible. So if someone adds a driver to v8 maybe the preprocessor instructions are the best way to backport them to v7. Anyway, I just wanted to mention my new approach in the hope it can help here in some form 🙂

[C47D]

On a general note, for v8 instead of adding #if LVGL_VERSION_MAJOR >= 8 I started to keep only v8 stuff because people who use v7 can use the release/v7 branch.
I'm not sure how well could it work here as the goal is to have as many drivers for as many releases as possible. So if someone adds a driver to v8 maybe the preprocessor instructions are the best way to backport them to v7. Anyway, I just wanted to mention my new approach in the hope it can help here in some form 🙂

I see, I also wanted to keep a branch for drivers v7 and another for v8, but I don't know how much effort would it take to keep both in sync, as far as I've seen the differences are the indev return value of the read function is now void and the rotation support.

[embeddedt]

Rotation is also available in 7.11.

[C47D]

Rotation is also available in 7.11.

So the return type of indev read is the only change? Drivers wise. See my comment in the review, this might help.

[embeddedt]

I believe that and the requirement for drivers to be static are the only breaking changes; yes.

[kisvegabor]

I believe that and the requirement for drivers to be static are the only breaking changes; yes.

And in display driver there are some renames: disp_buf -> draw_buf.

[C47D]

Ok, will do the initial changes for v8 on this driver, some of the changes are necessary in lv_port_esp32 instead.

[C47D]

I guess this driver is complete, waiting for @dastarling for feedback, will work on v8 support for display drivers on another PR

[dastarling]

Thanks for all the work. I'll give it a check soon. I think I am still on v7, was thinking of updating to v8, but there were some differences I wasn't prepared to go thru at the time. I'll try to test it at v7 and see if I can then update to v8 and test again.

[C47D]

Thanks for all the work. I'll give it a check soon. I think I am still on v7, was thinking of updating to v8, but there were some differences I wasn't prepared to go thru at the time. I'll try to test it at v7 and see if I can then update to v8 and test again.

v7 is fine, support for v8 will take some work. Thanks for the help :), there's also a new PR to handle better I2C (#70 ), but it's not merged just yet.

[C47D]

@kisvegabor @embeddedt Think we need to merge this, some new PRs are replicating some fixes available here, what do you think?

[kisvegabor]

Sure!

[dastarling]

I tested Master and it builds and runs with lvgl v7.11.0. But the ili9488.c file does not include some changes I made to the flush function due to the limited memory of the C3. So the graphics output is banded on the screen with Resolution 480x320.
With lvgl v8.01 I get these compile errors:
./components/lvgl_esp32_drivers/lvgl_helpers.h:54:25: error: 'LV_HOR_RES_MAX' undeclared (first use in this function); did you mean 'LV_HOR_RES'? #define DISP_BUF_SIZE (LV_HOR_RES_MAX * 40) ^~~~~~~~~~~~~~ ../main/main.c:95:41: note: in expansion of macro 'DISP_BUF_SIZE' lv_color_t* buf1 = heap_caps_malloc(DISP_BUF_SIZE * sizeof(lv_color_t), MALLOC_CAP_DMA); ^~~~~~~~~~~~~ ../components/lvgl_esp32_drivers/lvgl_helpers.h:54:25: note: each undeclared identifier is reported only once for each function it appears in #define DISP_BUF_SIZE (LV_HOR_RES_MAX * 40) ^~~~~~~~~~~~~~ ../main/main.c:95:41: note: in expansion of macro 'DISP_BUF_SIZE' lv_color_t* buf1 = heap_caps_malloc(DISP_BUF_SIZE * sizeof(lv_color_t), MALLOC_CAP_DMA); ^~~~~~~~~~~~~ ../main/main.c:107:12: error: unknown type name 'lv_disp_buf_t' static lv_disp_buf_t disp_buf; ^~~~~~~~~~~~~ ../main/main.c:122:5: error: implicit declaration of function 'lv_disp_buf_init'; did you mean 'lv_disp_drv_init'? [-Werror=implicit-function-declaration] lv_disp_buf_init(&disp_buf, buf1, buf2, size_in_px); ^~~~~~~~~~~~~~~~ lv_disp_drv_init ../main/main.c:128:13: error: 'lv_disp_drv_t' {aka 'struct _lv_disp_drv_t'} has no member named 'buffer' disp_drv.buffer = &disp_buf; ^ cc1: some warnings being treated as errors [1073/1227] Building C object esp-idf/main/CMakeFiles/__idf_main.dir/gui_sm.c.obj ../main/gui_sm.c: In function 'main_menu': ../main/gui_sm.c:196:85: warning: passing argument 2 of 'lv_obj_get_child' makes integer from pointer without a cast [-Wint-conversion] current_sensor_name = lv_label_get_text(lv_obj_get_child((lv_obj_t *)param, NULL)); ^~~~ In file included from ../components/lvgl/src/core/lv_obj.h:121, from ../components/lvgl/lvgl.h:32, from ../main/gui_sm.c:22: ../components/lvgl/src/core/lv_obj_tree.h:136:20: note: expected 'int32_t' {aka 'long int'} but argument is of type 'void *' struct _lv_obj_t * lv_obj_get_child(const struct _lv_obj_t * obj, int32_t id); ^~~~~~~~~~~~~~~~

[dastarling]

The ESP32C3 displays this message in the terminal indicating the lack of memory for a full screen DMA transfer at the 480x320 resolution

W (157073) ILI9488: Could not allocate enough DMA memory!

[dastarling]

Should I open a separate case for the lack of memory?

[dastarling]

I also found a few other minor issues;

lvgl_esp32_drivers\lvgl_tft\Kconfig: 828
range 0 21 if IDF_TARGET_ESP32C3 -> range -1 21 if IDF_TARGET_ESP32C3

lvgl_esp32_drivers\lvgl_tft\Kconfig: 840
range 0 21 if IDF_TARGET_ESP32C3 -> range -1 21 if IDF_TARGET_ESP32C3

lvgl_esp32_drivers\lvgl_touch\Kconfig:139
-> range 0 21 if IDF_TARGET_ESP32C3