RISC OS news and information from Drobe


Beta! \| About us \| Contact \| RSS	Webspace \| Tech docs \| Downloads \| BBC Micro \| Gallery \| Wallpaper

Omega update fixes speed, bugs

By Chris Williams. Published: 16th Jul 2004, 19:08:54.

All news is good news

Earlier this week, while the rest of the RISC OS world looked anxiously towards the RISCOS Ltd. shareholder meeting, MicroDigital released more news on their StrongARM powered Omega computer.

The hardware developers have issued another FPGA upgrade to their Omega users which, according to MicroDigital, "includes changes to accommodate [RISC OS] Adjust, the strange of window handling seen on a small number of machines, together with general improvements and housekeeping changes required by ARMTwister plus an extra 8-10 percent speed improvement."

Presumably, the "strange ... window handling" is the graphics oddity that Archive magaziner editor Paul Beverley casually investigated on an Omega at the 2004 Wakefield show.

We gather MD have issued at least 14 FPGA updates by now, and for the curious, one such version is available for download from here. As you may recall, MD employ FPGAs (re-programmable logic chips) to implement the Omega chipset, which can be updated by end users using suitable software.

Links
MicroDigital website

Discussion

Viewing threaded comments | View comments unthreaded, listed by date

The read me files inside the upgrade zipfile and the program to do the flashing do make interesting reading for the terminally curious ;o)

blahsnr on 16/7/04 7:39PM [ Reply | Report ]

I wonder if MD really are not aware of the existence of checsums (e.g. md5)

scl4c0rn on 16/7/04 8:30PM [ Reply | Report ]

Castle don't use checksums either... Why?

scl4c0rn on 16/7/04 8:40PM [ Reply | Report ]

They still don't proof read their announcements :-(

markee174 on 16/7/04 8:40PM [ Reply | Report ]

IIRC the flash rom programmer software does several checksum test.

JGZimmerle on 16/7/04 11:21PM [ Reply | Report ]

foo

Zappo on 17/7/04 12:13AM [ Reply | Report ]

In reply to JGZimmerle: Why no md5 checksum of the zip file?

scl4c0rn on 17/7/04 3:58AM [ Reply | Report ]

So, it would be intersting to know what security measures MD have implemented around the FPGA. If you misconfigure the FPGA then you can physically damage it. You're in an intersting situation where a virus could potentially fry your chipset; forget data damage, now a virus can do hardware damage. There's a description of the types of attacks possible in this (IMHO quite intersting) paper from 1999: [link]~boosters/fpgavirus.ps

Now, off the top of my head I'm not sure what FPGA the Omega has in it, other than it's a Xilinx part IIRC. If it's a Virtex or Spartan 2 and above device, then as far as I know these things are fairly robust, and the only way to fry them is to misconfigure the I/O pads, which requires the attacker to know what the wires connected to the FPGA do, but that's just a litle time with a scope.

Even if you don't try it, simply trashing your chipset will probably require your Omega to be sent back to the mothership for reprogramming - if a virus scrambles your chipset then I'd imagine there's not much you can do with your nice new paperweight.

And then there's just the opportunity to corrupt data subtly as it it passes through the chipset. Yum.

I'm all for FPGAs in commodity hardware (my PhD topic revolved around this concept), but I just wonder what security MicroDigital have taken to prevent these things. I had a quick san over the MD web site but couldn't turn anything up. If anyone has any information on this I'd be interested to know.

-- Dougal

Dougal on 17/7/04 10:15AM [ Reply | Report ]

Dougal:

The exact same thing can be said about the Flash ROM in the Iyonix - if you reflash it with a load of rubbish, you end up with an unbootable system and your only choice is to return to base for a mobo swap.

Now, wonder if anyone will come up with a non-RISC OS ROM image for the Iyonix? ARMLinux in ROM anyone?

Cheers,

/Neil/

spellinn on 17/7/04 10:46AM [ Reply | Report ]

ARMLinux in ROM? Nah, MOS is the way :P

David

riscdomain on 17/7/04 10:57AM [ Reply | Report ]

spellinn: True, but you can reflash and fix in ROM generally, but you can physically destroy the FPGA by misconfiguring it, which is the real pain.

I'm not trying to slag the Omega in comparison to the Iyonix (I have no intention of buying either), I'm just a person interested in FPGAs and their applications.

Dougal on 17/7/04 11:00AM [ Reply | Report ]

Dougal: fair enough, but given the numerous otherways you could compromise a RISC OS system (mostly temporarily), it's probably no more concern than anything else.

Neil: Unless you genuinely had no other storage, you probably wouldn't put ARMLinux in directly ROM. You'd put a bootloader like u-boot, which will let you have a lot more flexibility if things go pear shaped. The boot loader could pull the the kernel from disk, network, or elsewhere on the flash.

mrchocky on 17/7/04 11:23AM [ Reply | Report ]

The perseverance Microdigital has shown with the Omega is defiantly paying off. With this latest FBGA upgrade, for a lot of things performance seems similar to an Iyonix now. And, no I don't work for Microdigital.

IanK on 18/7/04 8:34PM [ Reply | Report ]

So, all they have to do now is stop being silly and actually let the wider public know what they're up to through independent news outlets.

moss on 18/7/04 9:25PM [ Reply | Report ]

Well, they are updating their own news page every now and then. That's something.

The Doctor on 18/7/04 10:41PM [ Reply | Report ]

The FPGAs used in the Omega are SRAM based. This means, that when the computer is switched off, their content is lost (just like the content of the normal main memory/RAM ist lost after a power cycle. There is also a CPLD (reprogrammable chip wich does not lose its content during a power cycle) on the board, wich reads the FPGA images from the flash ROMs and writes them into the FPGAs every time the computer is switched on.

Security measures to protect against unwanted reprogramming:

1. The CPLD can only be reprogrammed using pretty expensive equipment from Xilinx, wich connects to the JTAG-connector on the mainboard. So they can not be reprogrammed by executing some code on a running Omega. 2. The FlashROMs containing the OS can be write-protected separately from the one containing the FPGA images by setting the relevant jumpers on the mainboard. 3. For the OS flash ROMs there is a backup system in the form of two ROM sockets, wich can be eqipped with the standard ROMs used in the RiscPC.

JGZimmerle on 19/7/04 10:43AM [ Reply | Report ]

Most FPGAs (Field Programmable Gate Arrays) forget their settings when they're powered down, normally reading their "program" from a ROM of some description. CPLD (Complex Programmable Logic Devices) programmers aren't that expensive - you can use them in hobby projects, for example. (FPGAs are basically CPLDs on steriods.) Lots of development boards these days are littered with FPGAs and CPLDs because they let you develop on the board with your own custom bits. I'm surprised that Microdigital havn't started selling the Omega's motherboard as a generic development board yet, considering it's likely to be much more flexible than the Iyonix's. (MIPS for example have many models of dev boards with FPGAs on so you can develop your own weirdo hardware.)

nunfetishist on 19/7/04 3:01PM [ Reply | Report ]

I consider a few thousand dollars to be expensive enough to keep most teenaged script kiddies away. Anyway, it does not matter, since from a security POV it is irrelevant, for when an attacker gains unrestricted physical access to your machine, you have lost anyway. ;-)

Oh yeah, and FPGAs are actually quite different from FPGAs in many ways. Of course they are similar in the sense, that they are both electronic devices. ;-)

JGZimmerle on 19/7/04 3:36PM [ Reply | Report ]

The programmers I've seen certainly aren't a few thousand dollars! Shop around. I also assume that "Oh yeah, and FPGAs are actually quite different from FPGAs in many ways. Of course they are similar in the sense, that they are both electronic devices." is not only typo-ridden, but also fact-devoid?

nunfetishist on 19/7/04 4:34PM [ Reply | Report ]

Xilinx's programmer costs around 405 dolars, so probably cost around £400 after tax and the markup by UK resellers, but if youre willing to spend some time, you'd probably be able to build one yourself fairly cheeply if your not after a high speed solution, I know my university programmed it's CPLD's via a simple parrell port adapter and some software to drive the JTAG protocal.

Though for the FPGA's I think they opted for the programmer as it supported the lower voltages used.

NoMercy on 19/7/04 5:31PM [ Reply | Report ]

No, I certainly meant what I wrote above. The programmer needed for altering the CPLD on the Omega mainboard costs that much and FPGAs and CPLDs are indeed very different from an elektro-technical POV. Concerning typos and spelling mistakes: I'd like to see you write as many comments/postings in German as I do in English, and then we'll count again, okay?

JGZimmerle on 19/7/04 5:32PM [ Reply | Report ]

Instead of using a CPLD programmer you can also just throw a screwdriver in there.

Why does it seem most RISC-OS fans are always ready to pick a fight by deliberately misunderstanding a post or even over the format or punctuation? I think this is the biggest turn down of RISC-OS!

Jaco on 19/7/04 5:33PM [ Reply | Report ]

Yeah, that sounds about right for the hardware alone. Add to that the cost of the software and you end up where I did.

JGZimmerle on 19/7/04 5:34PM [ Reply | Report ]

For moss: Qercus have asked MicroDigital not to send us a review machine until they consider they have fulfilled their initial promises to customers and all outstanding orders are filled. We expect to get our hands on an Omega very soon and will make extensive tests and report the result of those together with comments from customers who have had a machine for some time.

Q on 19/7/04 7:20PM [ Reply | Report ]

For Q: "We expect to get our hands on an Omega very soon..."

How soon?

M on 19/7/04 8:25PM [ Reply | Report ]

Why all the sudden change to single letter handles? This aint Spies R US!!!!!!!

Over and out 004

sa110 on 19/7/04 9:10PM [ Reply | Report ]

For M: We haven't had chance to check on how much the dispute has changed the timetable ˜ we've had other things taking up our time. I'd provisionally pencilled in the issue after next for a review but all that planning has gone back into the pot!

For sa110: The Q is fairly obvious. I'll let Mike Morris answer for his failure to repeat his alliterative initials. ;-)

Q on 19/7/04 11:06PM [ Reply | Report ]

To JGZimmerle: re comments/postings in German: having recently sung EIN DEUTSCHES REQUIEM (Brahms) in German with my choir, I'm not going to take up your challenge: it took us Brits 2 months to learn how to /say/ it properly, let alone sing it...Fantastic piece of music though, and much better in the language it was written in IMHO.

George

bucksboy on 20/7/04 11:53AM [ Reply | Report ]

@bucksboy: Yes, I know what you mean. The same applies to most music/movie really, I almost always prefer the original language version. Concerning the challenge, I was probably not entirely fair, because I have an English stepfather and have spoken English most of my life. Still, it is not my native language, and no matter how often you overly polite British tell me otherwise, I'm sure it is noticeable.

JGZimmerle on 20/7/04 12:42PM [ Reply | Report ]

Q> "Qercus have asked MicroDigital not to send us a review machine until they consider they have fulfilled their initial promises to customers and all outstanding orders are filled."

Surely if the machine is *good* enough to *sell* to people it's good enough to review ? Or have I missed something ?

AMS on 20/7/04 1:38PM [ Reply | Report ]

To AMS: Enough (too much) was written about the Omega pre-launch. We decided that we wouldn't review until: 1. All pre-orders were satisfied; 2. the critical aspects of the Omega were all running satisfactorily. Doing this means that a review will either silence the critics of the ARMTwister or show that it doesn't work. Either way we get a proper answer and an end to damaging comments.

Q on 20/07/04 5:13PM [ Reply | Report ]

Please login before posting a comment. Use the form on the right to do so or create a free account.

Create a new account
Forgot your password?

Search this website