[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bootstrapping Debian GNU/Linux distribution on SuperH

GOTO Masanori wrote:
 > IMHO, this situation is easy to reproduct on other archs...
 > If you use MMX extension on i386, then obsolete processor
 > cannot run any such instructions :-)

No.  You miss important point.  This is "opposite" side.  SH4 (newer
processor) cannot run older binaries.  It is Hitachi's decision, so 
blame Hitachi.

 > It needs ABI interface definition change.
 > It's problem that only this trivial things (the order of
 > argument) cause all binaries/libraries incompatibility...

It's semiconductor vendor and compiler vendor who defines ABI.  You
could say it's trivial change, but a _fundamental_ change.

 > Though I don't know it's difficult or not, but 
 > can we *break the Hitachi's ABI compatibility* ?

It's not difficult thing, in theory, but no, that's not an option.

If we break it, it means, only Linux uses strange ABI for SH4.  This
means, we need to maintain our own version of change for SH4
toolchain, and it deserves losing feedback from SH4 usage of other
operating system (or non-operating system).  SH4 is widely used in
other environment, sharing toolchain is good thing.

 > The right thing for the future is to unite these ABI 
 > interface into one... Producing many incompatible
 > binaries/libraries seems uselessness.

No, I don't think so.  Basically, SH3 and SH4 is different processor,
in this concern.  It is quite unfortunate for programmers that the
difference of CPU infrastructure (ALU(s), pipeline...) affects into
ABI level, but this is Hitachi's decision.

Well, code morphing would be the Right Thing, then.

Reply to: