Re:Could EU force hardware manufacturers to make working dri...
by dumpsterdiver on Sunday October 01, 2023 @08:14PM
(#37732405)
Can you imagine the nightmare alternate universe of waking up to hear that Rowhammer is raging uncontrolled in the world, and the only way to fix it is to either return a data center full of affected products to the manufacturer or pay a specialist (who would probably be booked out for a long time to come) to come onsite?
Re:Could EU force hardware manufacturers to make working dri...
by loup-vaillant on Monday October 02, 2023 @11:25AM
(#37739355)
Last I checked Rowhammer is raging uncontrolled in the world, and the only way to fix it is to swap the RAM with something that actually works. Iām not aware of any configurable firmware residing on RAM chips.
Re:Could EU force hardware manufacturers to make working dri...
by dumpsterdiver on Wednesday October 04, 2023 @02:54AM
(#37761727)
Yikes, lol. Way to put my foot in my mouth. For some reason I was thinking it was fixed with software, but that was just a taxing mitigation that only lowered the chance of success. I stand corrected, we do in fact live in that nightmare world.
Yes you can make the distinction, and no it doesn't require getting rid of the microcode: https://news.ycombinator.com/item?id=37731829
Simply put, if the microcode can be updated by the user or remotely by Intel, then it count as "software", and Intel should stop shipping it. But if those update are either impossible or require a factory recall, then it's "hardware", and Intel can happily continue to ship it.
The same logic applies to the USB cable, and even FPGA: if the bitstream is locked into the NVCM, that's hardware. If it's merely written in the rewritable SPI flash, that's software. It's not a property of the bitstream itself, it's a property of how it is stored.
Obviously the cost of micro-code bugs will sharply increase. That's kind of a feature though: now hardware vendors must be really careful.