Re: [Libjit] struct jit_gencode has no member named limit

[Guillaume Yziquel]

Le Monday 20 May 2013 à 14:43:33 (+0400), Aleksey Demakov a écrit :
> Hi,

Hi.

> It looks like the ARM build is broken. Besides the ARM backend is not
> complete anyway, and needs quite a bit of work before being ready for
> general use.

Indeed. I'll be switching back to the older release to start dabbling
with libjit.

However, is there a way I could specify libjit to fall back to the
interpreted backend instead of the arm backend. Some configure option?

I'd be quite interested in knowing more about what should be done to get
the ARM backend back on tracks. I have little time available, but
still... I noticed that the jit/jit-rules-arm.ins seems to be out of
date and seems to be producing an inadequate jit/jit-rules-arm.inc file.
I'm having a hard time to grasp how this 'ins to inc' code generation is
done: I've looked into gen-apply.c to see how things were going under
the hood, then saw all these gcc __builtin_apply primitives, but I kind
of lack the broader view and a high-level documentation of the
internals. Is there any such high-level doc, or am I stuck with plain
old source code?

Best regards,

-- 
     Guillaume Yziquel