Function Prolog / Epilog

[Bill Currie]

Jeffrey A Law wrote:
> 
>   In message < 374DFCD1.11A62BF1@tssc.co.nz >you write:
>   > Okay.  I'll see how far I can get this weekend.
> Cool.

Actually got to the point of figuring out how to get the i386 port to
use __attribute__((naked)) appropriately.  ie I made what I felt to be
the needed general mods, now I need to figure out how to get the back
end to use the new attribute.

>   > Phew :)  BTW, should I send in my patches anyway?
> Yes.  Presumably you've got the copyright legalities dealt with.

Not yet, but since you want my patches, even though buggy, I've now got
motivation to get around to it.  BTW, is it possible to set it up so
that *ALL* future work I do on FSF code (any package) is assigned
automaticly?

>    > They do involve some
>   > major hacking of i860.md.  I rewrote all the movXX insns to avoid
>   > multiple asm insns per rtl insn.
> Good.



>   > There's also a couple of patches
>   > scattered through the core code to allow for bogus endian machines (see
>   > the endian-ness comment in i860.h).
> Not sure about these.

Pretty small, actually.  Mostly 1-4 liners, except for combine.c (but
still in one chunk).


> It would be best if you could break the patches into managable independent
> hunks.  We do not like mega-patches with lots of intertwined stuff since
> they are difficult to evaluate.

I'll do what I can.  Separating the patches on a file by file basis is
easy and for the most part will wind up with nice managable patches, but
i860.{md,h,c} worry me a little as they're non-trivial.

Bill
-- 
Leave others their otherness