On Thu, 25 Jul 2013 13:49:22 +0200, Martin Decky wrote:
FPGA boards capable of running LEON - I'm looking for board
which will met LEON2 with at least PIC, Timers nad UART requirements
and will not be too expensive.
The "not too expensive" part is tricky :-)
For me, "not too expensive" typically means under $300 :-)
We have an older LatticeECP2-50-based board [1] (without the TFT
module) readily available and it can host LEON2, albeit with a
non-trivial configuration.
[1]
http://www.msc-toolguide.com/lattice/tool-family/low-cost-fpga/hpe-mini-lec2-tft.html
Was that configuration tested on this board? Will it be time-consuming
to
set working environment on that board?
QEMU lacks firmware for LEON, as it's not supported in OpenBIOS.
And I couldn't find any firmware for LEON3 (open or closed) to test
QEMU.
I'm not sure how does TSIM work - does it have a firmware?
AFAIK there is no standard platform firmware for LEON, the OS itself
acts as the firmware (*). Even on the actual hardware the OS image is
uploaded to the SRAM over a serial line or over SpaceWire before the
CPU gets initialized (at least it worked in such a way on the
evaluation boards with LEON3 from Gaisler Research back in 2008 when
I
had access to them).
I've found that LEON (in particular, LEON/TSIM) is supported by at
least
two bootloaders:
* RedBoot - http://ecos.sourceware.org/ecos/boards/sparcleon.html
* u-boot - http://gaisler.com/anonftp/u-boot/u-boot-sparc-1.0.0.pdf
Therefore I suggest to use the evaluation version of TSIM and QEMU at
least for the initial porting phase. I'll try to pull some of my
contacts at ESA and ask them whether they might give us remote access
to a machine with the commercial version of TSIM or even to a machine
connected to an evaluation board.
That would be great! Anyways, even the evaluation version should be
sufficent in initial porting stage.
The second option is, of course, to use an FPGA board with a
synthesized LEON design. This should work, but finding a proper
configuration of the design suitable for a given FPGA board can be
sometimes quite time-consuming.
Maybe there are some pre-synthesized configurations working on certain
FPGA chips out-of-the-box?
(*) Well, as an OS designer I would certainly prefer the boot loader
(even a custom one) to act as the firmware, without polluting the
kernel with low-level hardware initialization stuff.
That approach is absolutely right, and as we have existing bootloaders
for LEON, there's no point in reinventing the wheel.
_______________________________________________
HelenOS-devel mailing list
[email protected]
http://lists.modry.cz/listinfo/helenos-devel
