On Wednesday 15 July 2009 19:03:36 Scott Wood wrote:
Mike Frysinger wrote:
On Wednesday 15 July 2009 18:18:20 Scott Wood wrote:
It seems pretty reasonable for U-Boot to provide functions like raise()/abort() that take the place of a hardware exception, and display an error message.
i disagree here. how much of the C library are you proposing we implement ? if libgcc keeps calling more and more functions,
Has it been?
you suggest we keep adding stubs for it ? seems like a never ending losing battle where we get screwed.
I don't see any slippery slope here, just a handful of functions that any reasonable freestanding implementation is going to want (memcpy, etc) and some way of getting an error out (raise/abort).
If it starts wanting libc functions that aren't reasonable, then of course we should complain (possibly with patches, for those willing to deal with the copyright assignment process).
i think calling raise/abort is already unreasonable for bare metal applications. -mike