Re: [APPROVE?] Fix MIPS and PowerPC with GCC 3.x

[Jonathan Larmour <jifl at eCosCentric dot com>]

Bart Veer wrote:
> > > > > > "Gary" == Gary Thomas <gary at mlbassoc dot com> writes:
>
>
>     Jifl> This would be similar to the various hal_mk_defs.c's
>     Jifl> floating around, although they requires knowledge of asm
>     Jifl> syntax. Instead the C++ file can just generate the header
>     Jifl> contents directly with #define's and the custom build step
>     Jifl> can move it into the install tree. Something similar happens
>     Jifl> for the heap magic in the memalloc package (a TCL file doing
>     Jifl> the generating that time).
>     >> 
>     >> I don't see how your approach is going to work. We need to know how
>     >> the target g++ messes about with its classes, not the host g++. Since
>     >> we cannot execute a feature test program, the way to do this is to
>     >> generate some assembler with the target g++ and analyse that, e.g.
>     >> with a Tcl script.
>
>     Gary> Which is exactly the problem solved by hal_mk_defs.c (albeit C
>     Gary> structures instead of C++ members...)
>
> Ah, I think I understand now. The proposal was to generate what
> amounts to C code using inline asm's in C++, rather than generate real
> assembler and analyse it. As opposed to hal_mk_defs.c which does
> generate assembler. I guess that would work, although I am a bit
> uncomfortable with mixing up languages to this extent.

I'm not sure that's quite what I meant either, as you describe it there. 
What I meant was something very similar to hal_mk_defs.c but the actual 
inline directive doesn't matter. Right now hal_mk_defs.c uses .equ because 
it wants the result to be available in asm, but we don't have that 
problem. I was sorta getting at using #define as the output, i.e.

#define DEFINE(sym, val) \
        asm volatile("\n\t#define \t" #sym ",%0" : : "i" (val))

The asm is never compiled so who cares :-). But on reflection this might 
be a little risky to do directly so perhaps using some other marker and 
then using sed to transform it into #define may be simpler after all :-).

It would be in C++ so there would be no mixed up languages. We would 
simply be using sizeof() from a C++ file.

The only possible problem in fact is that it isn't just the size we might 
need - it's the alignment requirements; but I'm sure we can do something 
with GCC's __alignof__ for that. I think we're so far down the road of 
relying on GCC features that we may as well give up trying to hold back!

Jifl
--
eCosCentric    http://www.eCosCentric.com/    The eCos and RedBoot experts
--[ "You can complain because roses have thorns, or you ]--
--[  can rejoice because thorns have roses." -Lincoln   ]-- Opinions==mine