Is there a difference in format between a packet received via SLIP and one from an Ethernet?

Question

When pinging a uip device via SLIP, I get a packet in the uart that looks something like:

   uip_buf[0]   0x45    
   uip_buf[1]   0x0 
   uip_buf[2]   0x0 
   uip_buf[3]   0x3c    
   uip_buf[4]   0x1 
   uip_buf[5]   0x20    
   uip_buf[6]   0x0 
   uip_buf[7]   0x0 
   uip_buf[8]   0x80    
   uip_buf[9]   1   
   uip_buf[10]  0xb5    
   uip_buf[11]  0x51    
   uip_buf[12]  0xc0    
   uip_buf[13]  0xa8    
   uip_buf[14]  0x1 
   uip_buf[15]  0x1 
   uip_buf[16]  0xc0    
   uip_buf[17]  0xa8    

From the code, this ethernet header is supposed to be composed of a dest addr, src addr, and a type. However, in order for the packet to be processed, the type must be 0x0008, but apparently it's 0xa8c0 (uip_buf [13:12] or uip_buf [17:16]. Thus the comparison always fails:

 if(BUF->type == HTONS(UIP_ETHTYPE_IP))  // 0xa8c0 != 0x0008

The line that does the comparison came from an ethernet implementation... Is it possible the type field is different when transmitted via SLIP?

Answer 1

Yes, there are differences. SLIP does not have the ethernet header, since serial lines don't have a bus architecture: instead, there are exactly two end points, so each end already knows who it's talking to.

IP packets still need to be distinguished, so there are some magic byte values and substitutions that allow this to be done. Details at the wikipedia page:

https://en.wikipedia.org/wiki/Serial_Line_Internet_Protocol