On 20:36 Wed 22 Dec , Krzysztof Halasa wrote: > Sascha Hauer <s.hauer@xxxxxxxxxxxxxx> writes: > > >> why not put the NPE firmware in the env? > > > > That's a good idea. This way we can tell the users to put the firmware > > to arch/arm/boards/.../env/firmware before compilation and avoid > > redistribution problems if there are any. Another plus is that the > > barebox binary is not larger than it has to be. > > The only downside is that the environment gets bigger as it duplicates > > the firmware, but even this could be avoided if we add a second (non > > environment) filesystem image. > > Actually, there is more. There is plenty of free space in Barebox' > region (2 pages 128 KB each), and the microcode isn't likely to ever > change (unless Intel releases docs/sources, of course). > > I don't want it in user-writable area since it could be incidentally > erased. Without the microcode Ethernet interfaces are not functional, > and I'm not sure all users would appreciate it if they had to download > it back with [XY]-modem (or another C-kermit) :-) Even TFTP is sometimes > black magic for some of them. > > Obviously, the microcode is optional, and especially NPE-A one isn't at > all used by Barebox (Linux uses it for driving sync serial interfaces). > I don't consider the microcode a part of Barebox, like the > /lib/firmware/* blobs aren't part of Linux. > > Also it's simpler to export the microcode to Linux when it's in fixed > well-aligned locations. I'm working on a fs support in linux of the barebox env so linux could get it easly also but I prefer to be able to change my micro code as some of the people who work on IXP have acess to it to modify it. So I think its a good option so as linux it will be better to put it in a fs like /lib/firmware Best Regards, J. _______________________________________________ barebox mailing list barebox@xxxxxxxxxxxxxxxxxxx http://lists.infradead.org/mailman/listinfo/barebox
