I wanted to bring to attention Intel's Platform Innovation Framework for Extensible Firmware Interface since its goals overlap with those LinuxBIOS a little bit. After reading the article at _http://www.linuxdevices.com/articles/AT3790951165.html_ (http://www.linuxdevices.com/articles/AT3790951165.html) , I had some questions about the ability of LinuxBIOS to support next-generation hardware. It looks like Intel wants to change the whole firmware landscape with EFI. If such a framework will be truly open and support legacy systems, it'll make LinuxBIOS redundant, wouldn't it? What about newer architectures like those built around the Itanium? Will LinuxBIOS be able to scale up as well as EFI? Those are questions that I wanted to throw out there.
Trellix78@aol.com writes:
I wanted to bring to attention Intel's Platform Innovation Framework for Extensible Firmware Interface since its goals overlap with those LinuxBIOS a little bit. After reading the article at _http://www.linuxdevices.com/articles/AT3790951165.html_ (http://www.linuxdevices.com/articles/AT3790951165.html) ,
I had some questions about the ability of LinuxBIOS to support next-generation hardware. It looks like Intel wants to change the whole firmware landscape with EFI. If such a framework will be truly open and support legacy systems, it'll make LinuxBIOS redundant, wouldn't it?
Huge IF.
But even then open source can usually find a niche.
What about newer architectures like those built around the Itanium?
EFI has a place there. So far Itanium is a niche market. It is not the x86 successor.
Will LinuxBIOS be able to scale up as well as EFI?
Has EFI scaled up to x86-64 yet? 32bit interfaces on 64bit hardware look weird.
With LinuxBIOS we support a much narrower interface so it is much easier to write. There is more work for the OS todo. But the OS is the hardware abstraction layer not the BIOS.
As for running on large hardware that really is not too hard. An OS needs to be able to do things efficiently, which makes scaling hard. After you go SMP I don't see the difficulty.
Those are questions that I wanted to throw out there.
So far all I have seen with EFI is a time sink.
Commercial things either need everyone to accept them or they tend to die.
Open source can get by with a small but growing niche, and be accepted based on the technical merits of the project at hand.
If EFI is a huge success we might have an EFI compatibility layer.
Eric
-
ebiederman＠lnxi.com -
Trellix78＠aol.com