Re: Volatile MEMs in statement expressions and functions inlined as trees

[Alexandre Oliva <aoliva at redhat dot com>]

On Dec  4, 2001, Richard Henderson <rth@redhat.com> wrote:

> It isn't as diffucult as you imagine -- you only have to recognize
> the "mov" instruction -- if you get anything else you abort.

I didn't think it was too hard, indeed, but it's probably not worth
reinventing, or having to look up the insn opcodes and hope I didn't
miss any case :-)

> So, something like

Thanks!

FWIW, I've written a new patch that marks the last EXPR_STMT of a
STMT_EXPR whose value is wanted with TREE_ADDRESSABLE, and then
arranged for the expr of the EXPR_STMT to be evaluated such that its
value is wanted.  The expr containing the return value of an inlined
function is marked similarly.

I'm testing my patch now (but I'm having some trouble with Ada, whose
bootstrap compiler I've installed yesterday; I don't know whether it's
the bootstrap compiler or my patch that are faulty :-(

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me