With the Lwip TCP stack, what happens if the received message is larger than the buffer?

Question

I have been doing a small tcp project, in C, on a Raspberry Pi Pico, experimenting with the lwIP api. One of the sources of help has been the pico-sdk-examples. When the lwip stack receives a message, it summons a callback to a method of the programmer's choice. In the pico examples, that method contains the following code for handling the message, which is passed into the callback in a structure known as a pbuf. You can chain pbufs together for longer messages.

   //state is a protocol control 
   //p is a pbuf

   // Receive the buffer
        const uint16_t buffer_left = BUF_SIZE - state->recv_len; 
        state->recv_len += pbuf_copy_partial(p, state->buffer_recv + state->recv_len,
                                             p->tot_len > buffer_left ? buffer_left : p->tot_len, 0);
        tcp_recved(tpcb, p->tot_len);

My question is about the tcp_recved component. In the lwip documentation it says that

This function should be called by the application when it has processed the data. The purpose is to advertise a larger window when the data has been processed.

If the total_len is greater than the buffer size, and the buffer is filled, why does their code advertise that the tot_len was recved. Will that buffer overflow be lost? Also, why don't we use p->len?

Answer 1

In answer to why we don't use p->len, we use p->tot_len because we may want to read the entire pbuf chain. p->len might just be the first x characters, stored in the first pbuf; p->tot_len is the length of the entire string. However, I still do not know what happens if a message is too large.

Edit: If the message is larger than the buffer then the excess will be lost.