ModernFW Was An Exciting Announcement This Week That Went Largely Unnoticed

Of Intel's keynote announcements this week kicking off their first public Open-Source Technology Summit, surprisingly not attracting too much attention this week was news of their ModernFW initiative to create a new modular and open-source firmware solution to replace aging legacy code on motherboards.

From a quick look around a few days after the announcement, ModernFW was only even mentioned on a few websites for this ambitious open-source project that has a similar trajectory to Coreboot/LinuxBoot. ModernFW is to be a new firmware stack that does away with legacy code and provides a lightweight implementation derived from TianoCore that is just enough to boot an OS kernel. Hopefully this will ultimately lead to more collaboration with LinuxBoot in particular given their similar intentions.

ModernFW code in its preliminary form can be found on GitHub.

ModernFW is an experimental approach to building a minimum viable platform firmware for machines such as cloud server platforms. Traditional PC firmware packages have evolved over time and have emphasized backwards compatibility and generality of purpose. We are seeking to reduce the overall footprint, increase efficiency, and to improve the security posture of the system by eliminating capabilities that are not needed to meet requirements for platforms that serve more vertically integrated purposes. For example, one avenue for exploration is to move any functionality that can be accomplished in the context of the operating system out of the firmware. ModernFW is based on TianoCore, and any learnings will be shared upstream to that community and others.

As for when ModernFW might be viable enough for usage, after talking with the Intel folks following the announcement, it sounds like hopefully around end of year the first modular components may be ready for testing. ModernFW is doing away with all of the legacy bits, will try to have the operating system do as much of the device initialization as possible, and enhancing security will be among their highest priorities. There is also much hope that using ModernFW will lead to quicker boot times.

Intel previously expressed plans that they want to end legacy BIOS support by 2020 so this ModernFW project appears to fit nicely with this aim.

Overall, ModernFW should be quite interesting if all their goals are reached in a timely manner. When inquiring whether Intel is still hoping to open-source the FSP, it still is being worked on but hopefully there will be news to report in the months ahead.

Another open-source firmware effort that saw a fair amount of attention this week was the Intel/Google project on Sound Open Firmware as an open-source DSP firmware and SDK solution. As covered recently, Linux 5.2 also has initial support for Sound Open Firmware. It turns out Sound Open Firmware is already in use by all Google Geminilake Chromebooks and future hardware.