arm: working bootloader, example app

'make' will produce two outputs:
- bootloader.elf to be loaded into the bitstream
- app.bin to be loaded through the programmer

Loading app.bin is as simple as:
- reset the board
- `python3 prog.py app.bin`

arm/bootloader.cc

@@ -0,0 +1,63 @@
+#include <cstdint>
+
+#include "gpio.h"
+#include "uart.h"
+
+namespace {
+
+uint8_t UartRead() {
+    uint8_t c;
+    while (XUartLite_Recv(uart0, &c, 1) < 1) {}
+    return c;
+}
+
+void UartWrite(uint8_t c) {
+    XUartLite_Send(uart0, &c, 1);
+}
+
+uint32_t UartRead32() {
+    uint32_t val = 0;
+
+    // little endian
+    val |= (UartRead() <<  0);
+    val |= (UartRead() <<  8);
+    val |= (UartRead() << 16);
+    val |= (UartRead() << 24);
+
+    return val;
+}
+
+}  // namespace
+
+int main() {
+    gpio0->data = 1;
+
+    InitUarts();
+
+    while(1) {
+        uint8_t c = UartRead();
+        if (c == 'c') {
+            uint32_t addr = UartRead32();
+            uint32_t bytes = UartRead32();
+
+            uint8_t* start = reinterpret_cast<uint8_t*>(addr);
+            uint8_t* end = reinterpret_cast<uint8_t*>(addr + bytes);
+            for (uint8_t* ptr = start; ptr < end; ptr++) {
+                *ptr = UartRead();
+            }
+
+            UartWrite('a');
+            UartWrite(bytes);
+            gpio0->data = 0xf0;
+        } else if (c == 'j') {
+            uint32_t addr = UartRead32();
+
+            gpio0->data = 0x55;
+
+            addr |= 0x0001;
+
+            auto jump = reinterpret_cast<void(*)()>(addr);
+            jump();
+        }
+    }
+}