The security world is all abuzz with the news (as reported on C/NET) that Chinese researchers have found flaws in the standard security algorithms used in a number of Internet protocols. The concern expressed in the story is the sheer scale of the problem, given the size of the Internet today and the difficulty in convincing the majority of people to upgrade their software on a regular basis.

As much of a problem as it is for that crowd, it is a much more serious concern for many embedded systems. Any device that depends on IPSEC or SSH could suddenly find itself vulnerable unless a new executable can be distributed to devices already in the field. The necessary conditions for this are a device that is designed with this type of upgradeability and the awareness that upgrades are necessary. I suspect that the first condition is more common today than the second one, given that most embedded developers are users of security protocols rather than security specialists.

The typical pattern of upgrades like these is that not much is done about them until a headline-quality event takes place and raises public awareness. The trick is to be among those reading the headline rather than being the subject of them.

Larry Mittag

The Hot Chips conference is in full swing, and as expected the emphasis is on multiple-core processors. A report in EE Times details how, as I expected, the puny multiprocessing currently available from Intel and AMD is just the tip of the iceberg. The dam that has held back hardware design is finally starting to burst and it will have serious ramifications regarding the way we create software.

Just today I got an email from a software guy wanting a characterization of the impact of not having L2 cache on a system on the CPU speed. Given that the answer depended on software that had not yet been written and was a delicate balance of cache, pipelining, and the nature of the software itself I had to explain that the answer was indeterminate. How am I going to explain it when they ask me about the impact of parallel hardware architectures that are completely outside of his (and for that matter, my) experience base?

Nick Tredennick was quoted in the article as saying that the microprocessor has held back innovation in hardware design. He said something very similar to me in a panel half a dozen years ago. One way or another, it appears that the box is being opened and that we are about to be treated to a number of novel architectures.

Larry Mittag

A recent story in EE Times generated a lot of interest on discussion boards regarding the open sourcing of IBM�s new Cell processor, but it seems that the excitement may be a bit premature. The article (IBM will unlock door to Cell) appears at first read to detail how IBM is throwing open the Cell architecture itself into the open-source community, a la the hardware specifications being developed at OpenCores.org. But what the community is actually getting are the specifications for the CPU and the system libraries necessary to make it do interesting things.

This is certainly less than open sourcing the core itself, but still no small announcement. These are the power tools necessary for writing deep OS code, the kind of code that will be necessary if the Cell is to reach wide acceptance. The question is whether systems programmers will take the architecture to heart and make it do interesting things. It shows real promise for applications that can take advantage of heavy multithreading, but this is a very different architecture. The question is whether programmers and systems designers will take it to heart.

Sun Microsystems has just jumped into the open-source pool with both feet, as reported by C/Net's news.com. This appears to be basically a surrender to the power of Linux and a high-stakes gamble for the company.

The biggest risk is that no one will care. Linux is a familiar set of code for a huge number of programmers around the world. Will Solaris offer enough value for these programmers to spend the time bringing it up?

I suspect they will. If nothing else, portions of the code may very well find its way into future Linux releases. If this does happen then someone should look over the licensing agreements very closely. I'm sure that SCO-trained lawyers would like nothing better than to find new grist for their mills.

Larry Mittag

IBM is offering a free open source implementation of the IEEE-1275 Open Firmware standard. Slimline Open Firmware (SLOF) can be used by software  engineers developing boot firmware, operating systems, or applications for PowerPC, or by hardware engineers interested in fast bring-up of PowerPC-based systems, according to IBM's DeveloperWorks website. More information about Open Firmware can be found via Google, or at the IEEE 1275 website. More details about SLOF can be found here .

 The nice thing about Open Firmware is that it allows one plug-in PCI card to be used in any chassis that understands Open Firmware, with very little concern for the host machine architecture. The combination of Open Firmware and IBM's SLOF announcement is a potential windfall for embedded systems with pluggable PCI busses, because it opens the door to the use of commodity PC hardware (and their aggressive price points) for embedded applications. The downside to Open Firmware is that PCI cards designed for Wintel boxes don't use Open Firmware; they use a ghastly technique where actual x86 machine instructions are stored in a dedicated memory area in the card itself (whose idea was *that*?!). There are a lot of nifty chipsets on PCI cards that I wish were available in affordable OEM-friendly packages, i.e. just the chips. Graphics controllers come to mind, as do gigabit ethernet controllers and GSM adapters.

 Considering the PCI cards available for Sun and Apple machines, what sorts of embedded applications could YOU now realize?

[Posted for Bill Gatliff]