[Oberon] FPGA - Boot over serial line

[Skulski, Wojciech]

I wonder if there is a way to use ChipScope to write directly to the BRAM while the Oberon System is running. The CPU would need to be stopped before doing this (stalled, perhaps), and then restarted from the known BRAM location after the updating was done.

Any thoughts?

Wojtek
________________________________________
From: Oberon [oberon-bounces at lists.inf.ethz.ch] on behalf of Magnus Karlsson [magnus at saanlima.com]
Sent: Sunday, October 15, 2017 12:33 PM
To: oberon at lists.inf.ethz.ch
Subject: Re: [Oberon] FPGA - Boot over serial line

It will also take .mem files (-bd FILENAME [<.elf>|<.mem>]).  The
example has .elf since that's the most common format directly generated
by the compiler.

You will need to find the location of the block RAM used to hold the
boot code.  The location can be found out using a tool inside ISE called
"FPGA editor" that will allow you to inspect the final logic placement
inside the FPGA.  In case of the RGB/mono bit file I sent you, the
location is RAMB16_X0Y20

Magnus


On 10/15/2017 12:16 AM, Tomas Kral wrote:
> On Fri, 13 Oct 2017 11:05:31 -0700
> Magnus Karlsson <magnus at saanlima.com> wrote:
>
> Hi
>
>> In our case the prom.mem
>> data is stored in one of the 32 block rams available in the Spartan6
>> LX9 so you would need to find out which one the synthesis toll
>> decided to use for this purpose.  The Data2Mem tool needs this
>> information in a special file called .bmm that will also specify how
>> the memory is organized (in our case it's organized as 32 bits x 512
>> words).
> The syntax is:
> $ data2mem -bm my.bmm -bd code.elf -bt my.bit -o b new.bit
>
> The difficult part is to assemble .bmm. There is an option INIT_f1
> creating BMM map for us.
>
> Verilog Coding INIT_fl Example -
> RAMB16 #(.INIT_fl("file.mem")
>            ...)
> ramb16_0 ( <port mapping>);
>
> But what is .elf, as I need to somehow supply prom.mem instead?