It’s really less information you provide here.
Problems within your app can be anything.
The code you provided should only be called once, and in this case it shouldn’t be any problem (as you wouldn’t see anything)…
If you do not know what memory allocation (and freeing!) is good for, and how it is used, just read the docs 
[robekras], which docs?
Of course I know about functions malloc/free in Arduino/C++, but does not when saying about LVGL library functions. Unfortunatelly, ESP does not print any errors/warnings.
I tried to increase or decrease size of memory which heap_caps_malloc() allocate, but nothing happen.
As I understood, heap_caps_malloc is the buffer of the frame and we can allocate memory for the full display size(sizeof(lv_color_t) * screenWidth * screenHeight); but it will not solve the problem.
Again, lv_disp_draw_buf_init is (should be) only called once, when setting up (initializing) the lvgl environment.
The disp_draw_buf is not (and doesn’t need to be) the full display frame size.
It’s in general only a small part of the full display’s frame size. This buffer is only the working buffer for lvgl.
If lvgl needs the update the full display, it does it in parts.
In the flushing function this working buffer is copied to the real display (frame buffer) which is connected with SPI to the ESP32.
The call of lv_disp_draw_buf_init (and the entire lvgl initialization) seems to be ok, else you would not see anything on display.
The problem seems to be located within your application code which manages the switching to the different screens and handling of the ‘messages’.
And this can be anything. E.g. wrong pointer handling, buffer overflow when using string functions, infinite loops etc.
[robekras], you are absolutelly right: the problem is not with allocating the memory because it allocates only 1 time in setup section. There can be an error only when some transistors don’t work, but, I don’t believe in it.
The problem is not in my sketch too, because problem appears when I just type letters on keyboard (LVGL’s component) and looks like infinitive loop - no errors, just hanging…
Maybe I need to do something with settings in lvgl configuration file…
I’m not sure whether I correctly unterstand you:
Seeing a problem when you type a character on keyboard doesn’t necessarily mean, that the problem is related to the typing function.
When your code does something wrong, e.g. writing to a wrong memory location, than the problem (exception, infinite loop or anything else) might occur at a very different time.
I do not know your code, but I don’t think it’s a problem of configuring lvgl nor is it a problem of lvgl.
As I understand it, the app is working for a while, and after some processing, it fails (what ever that means).
With a problem in configuring lvgl, you wouldn’t see anything. It wouldn’t work from the very beginning.
As I don’t think that your app does need too much resources, my guess is again, your code does something wrong.
When I click the keyboard’s button, nothing interesting appears:
// --------------------------------------------- обработка нажатия клавиши на клавиатуре ---------------------------------------
void cb_keyboard(lv_event_t * e){
lv_event_code_t code = lv_event_get_code(e);
if(code == LV_EVENT_CLICKED){ // надо попробовать LV_EVENT_VALUE_CHANGED
uint16_t bt = lv_keyboard_get_selected_btn(keyb);
//Serial.println(bt);
if (bt == 12){
//Serial.println("backspace");
lv_textarea_del_char(message_text_input);
}
else if (bt == 24){
//Serial.println("Enter");
lv_textarea_add_char(message_text_input, '\r'); // если не переносит - заменить на \n
}
else if (bt == 25){
//Serial.println("Shift");
if (lv_keyboard_get_mode(keyb) == LV_KEYBOARD_MODE_USER_1){
lv_keyboard_set_map(keyb, LV_KEYBOARD_MODE_USER_2, kb_ru_big, matrix_ru_big);
lv_keyboard_set_mode(keyb, LV_KEYBOARD_MODE_USER_2);
} else if (lv_keyboard_get_mode(keyb) == LV_KEYBOARD_MODE_USER_2){
lv_keyboard_set_map(keyb, LV_KEYBOARD_MODE_USER_1, kb_ru_small, matrix_ru_small);
lv_keyboard_set_mode(keyb, LV_KEYBOARD_MODE_USER_1);
}
}
else if (bt == 38){
//Serial.println("rus/eng");
}
else if (bt == 46){
//Serial.println("123");
lv_keyboard_set_mode(keyb, LV_KEYBOARD_MODE_SPECIAL); //LV_KEYBOARD_MODE_NUMBER
}
else{
const char * txt = lv_keyboard_get_btn_text(keyb, bt);
lv_textarea_add_text(message_text_input, txt);
}
}
}
but application hangs…
I found the reason. In configuration file I changed memory settings lika that
#if LV_MEM_CUSTOM == 0
/*Size of the memory available for `lv_mem_alloc()` in bytes (>= 2kB)*/
#define LV_MEM_SIZE (96U * 1024U) /*[bytes]*/And now everything works great! No hangs!
But, finding the way to correct is only the 1st part. The main part is find the reason why it did not work with LV_MEM_SIZE (48U * 1024U) //48 KBytes
After increasing LV_MEM_SIZE program does not hang when loading/displaying/animating messages, but sometimes does when typing long message. Too strange…
Do you use malloc and free in your program?
Do you use string functions like sprintf, strcpy, strcat in your program?
Do you do pointer arithmetic?
As far as I can see, your program isn’t doing that much, as your heap memory is running out.
A different (changed) memory layout can also lead to a different error appearance.
Sorry could not find your post where you use malloc for a string. Do you mean the malloc for disp_draw_buf? That was not for a string buffer (aka char*). That’s the drawing buffer.
I do not mean String but string functions. Don’t use strcpy, strcat. Use always the better functions with ‘n’ like strncpy, strncat.
Shouldn’t be a problem as long as you only read.
Maybe a hint: When you use utf8, have in mind, that the number of utf8 characters (in an utf8 string) is not the same as you need for storing it in char buffer. This means if you have five letters in Cyrillic you need at least (5 * 2) + 1 for your char buffer.
Not for the string. I do not use String.
Earlier you asked about malloc() function, not malloc of the string
Yes
disp_draw_buf = (lv_color_t *)malloc(sizeof(lv_color_t) * screenWidth * screenHeight/10);Thanks. But… I use them when get something from UART. Here I do not receive UART, just type on the screen
Do you process the input that you typed in, e.g. after you hit enter?
When I type, I process like this Strange error (not enought memory) - #8 by cafetera
And after clicking “send” button I process the whole message - generate ID, add sender’s name and so on…
Yes, I know your cb_keyboard.
It doesn’t look any special there.
I’m more interested in the function
I process the whole message - generate ID, add sender’s name and so on…
When the function cb_keyboard ‘hangs’, it doesn’t necessarily mean that the problem is in the cb_keyboard.
It can be on a very different place!
This function works when I press “send” button, so, don’t think of it. The problem occurs when I just type on the keyboard.
Probably I found 1 reason. I think the problem occurs when I click on the little space between the buttons.
A short story:
The boss of The Company (we name him P. Charles) instructs two employees (Tom and Jerry) to give him an overview about the current financial situation about the company.
Both walk into a meeting room and they comb through company records to compile the numbers which they write on the white board.
In the meantime P. Charles instruct another employee (named Oliver) to explain the story about the flowers and the bees to a further employee (named Stan).
Working for while Tom and Jerry decides to make a lunch break, and leave the meeting room.
Oliver and Stan are entering the now empty meeting room. Oliver cleans the white board and draws some nice sketches from flowers and bees. After a while Stan understand how the game works between the flowers and the bees, and both leave the meeting room again.
Again after a while Tom and Jerry go back to the meeting room, finding the strange drawing on the white board.
They could not understand why the white board now shows different things.
They go to P Charles and telling him that the white board seems to be broken, because the numbers on the white board has changed to some strange drawings.
P. Charles now asks all the other employees whether they observed such a strange thing with the white board.
Stan and Oliver respond to the boss that they swear everything was alright with the white board some minutes ago.
They could write to the white board and nothing unforeseen happend to the drawings.
I don’t now whether this happened in such a way.
Believe it or not.
Very interesting story!