Development Environment

This page describes my development environment in terms of computers, operating systems, software versions, and so on.  Why is this necessary?  A couple of reasons:

• Version-itis - Xilinx's EDK is still, to some extent, a work in progress.  It changes (for the better) with each release.
• Multiplatform - You need a Linux box to build kernels, but a Windows box to build and configure hardware.  Painful but true.

My development platform consists two machines - a Dell C400 Notebook with 30GB HDD and 512 MB RAM with running Windows XP, and an Intel Pentium 4 running RedHat Linux 8.0 (Phyche).  The Linux box is big and fast (80GB HDD, 2.4MHz CPU, 1GB physical RAM).

These two machines are configured such that my working directory (which is actually a subdir of my Linux home directory) is exported via Samba and mapped as a drive on my Windows machine.  So, there's no to-ing and fro-ing shuffling files between the two machines.  This is essential for efficient development.

So, on the Windows box I do the following:

• Run the EDK to synthesise MicroBlaze hardware platforms.  This includes using iMPACT to download bitstream and PROM files, and XMD as a debugger interface.  Software versions are
  • EDK Version 3.2 + Service Pack 1
  • ISE 5.2 + Service Pack 3

I recommend that you always upgrade to the latest version (and service pack) of EDK if possible. 

On the Linux side I have

• the MicroBlaze GCC toolchain
• The uClinux-dist and kernel source distributions
• ISE5.2 installed and running under Wine (experimental).

My home directory under Linux is exported as a Samba share and mapped as a drive on my Windows box.  This allows me to build a kernel under Linux, then switch over to Windows (and specifically, Xygwin) to build a hardware target, configure the FPGA and upload the kernel image. 

Connections to the Development Board

I use the Insight Virtex2 System Board V2MB1000 as a basic development platform.  With the P160 communications module attached, this board has 2 serial ports, and I use them both.  A serial cable connects the "top" P160 serial port to the Windows machine, for use with xmd, while the "main board" serial port is connected to the Linux box, where I run minicom as a terminal program to the uClinux console.

I'll put some photos up soon to illustrate this configuration more clearly.