To: R. Bond who wrote (19859 ) 7/1/1998 7:44:00 AM From: VoodooFL Read Replies (1) | Respond to of 31646
YK2000 Problem In MCUs Could Hurt OEMs SOURCE: Electronic Buyer's News This Week In EBN (5:10 p.m. EDT, 6/25/98) By Jack Robertson OEMs beware: You may have a Year 2000 date problem buried inside the embedded microprocessor in your product. By now, almost everyone knows that computer network and mainframe software could have Year 2000 time bombs ticking away, ready to blow up on the first day of the new century. But the same danger lurks in the embedded code of ubiquitous microcontrollers that might also have only two-digit abbreviated year designations. At midnight of the new millennium, this MCU code will revert to the year 1900, possibly with devastating consequences. MCUs are pervasive-embedded in everything from automobiles to telecommunications to home appliances to industrial equipment of all kinds. And until now, few people have given any thought to the buried software running on the MCU. But the microcontroller Y2K problem is scary, analysts and industry executives say. "It's almost impossible to get your arms around the MCU Y2K problem," said Thomas Starnes, a Dataquest Inc. analyst and former program manager for microcontrollers at Motorola Inc. "Microcontrollers are ubiquitous. They're embedded in hundreds of thousands of products, and are largely invisible to users. It's nearly impossible to get to the embedded code on MCUs. And even if you could, it's not clear how you could analyze the code or correct it." First, it is nearly impossible to extract the MCU code. In most cases, it is an inherent part of the MCU chip as an embedded mask ROM. Even code on a programmable MCU using embedded flash memory or EPROM may not be easily accessible or decoded-if you can get your hands on it. Second, MCU code is largely written in low-level assembly language, which poses extreme difficulties in dissecting and finding two-digit year designators. Many mainframe-computer Y2K fixers have thrown up their hands trying to decode assembly-language programs on the big machines. The crisis is worse for assembly codes embedded in countless millions of MCU chips. The good news is that oodles of MCUs are in products that have no need for a yearly date in their operation. The bad news is that virtually no one knows how many MCUs out there are vulnerable. One thing going for MCUs is their rapid obsolescence and replacement in a quickly changing product life cycle. And some recent MCU code writers, aware of the potential Y2K catastrophe, are embedding Year 2000-compliant software in microcontrollers. If Y2K snafus erupt in embedded MCU code, one OEM nightmare is the very real threat of big liability lawsuits. Computer network and mainframe doomsday Casandras foresee legal claims of hundreds of billions of dollars stemming from Y2K failures. This liability estimate could be dwarfed by MCU litigation if the far more pervasive microchips turn out to be as Y2K error-prone as Big Iron. The MCU chip makers have also suddenly awakened to the potential Y2K legal risk they could face. In most cases, the chip companies had no more role in the code writing than simply embedding it in their MCUs. But they have their brand names emblazoned on the microprocessors, a handy target for Y2K victims to go after. In addition, the Semiconductor Industry Association has started looking into the issue, but apparently no solution has yet to be developed. For now, no one knows the extent of the microcontroller Year 2000 danger. We will, at midnight Jan. 1, 2000.