Short term an OVMF-centric solution is good. But long term, I think Linux needs a Linux-friendly IBV to build native UEFI -- as well as OVMF-flavored UEFI -- with non-BSD licensed community code. If you restrict this to just OVMF, any GPL innovations will only happen at virtual level. Right now, boot loaders (eg, rEFInd) and VM projects (VirtualBox) are the only ones to benefit from UEFI non-BSD-licensed code. An IBV could offer these enhancements to Linux OEMs, not just server VMs. Linaro is already an IBV, to a degree, they produce UEFI binaries for the supported dev boards, as part of BSP build process.
Look at the coreboot blog, the other week the Purism BIOS developer was talking about a "Free Software port of FSP". That should be a shared effort by all Linux OEMs/vendors, not shouldered by a single OEM. A Linux-centric IBV could be getting help from multiple companies -- like Linaro does with ARM -- to help with this effort. It could be part of Linux Foundation's Core Infrastructure Initiative, maybe. They should also tailor Linux-friendly ACPI, like recent thread about Linux standardization of _DSD. As well as not include a WBPT table in Linux OEM systems, and look at the other tables to see what's should be removed. Alternately, the UEFI Forum should create a non-BSD tree to contain this, not just focus on BSD for the fully-closed-source end of the spectrum, and work with some FOSS-centric OSVs to better support UEFI using their community's models. I hope an IBV is listening. Create a separate project from your current closed-source code, and fully-embrace the open source community. Today, Sage Engineering is the only Open Source-friendly IBV that I'm aware of. They mostly focus on coreboot, not UEFI, though. (Like Phoenix, their blob recently went dark, I hope they're doing ok.) _______________________________________________ edk2-devel mailing list edk2-devel@lists.01.org https://lists.01.org/mailman/listinfo/edk2-devel