gcc-bugs@gcc.gnu.org

[Jan Hubicka]

> This patch:
Hi,
thank you for tracing this down,

> It appears wrong to me that emit_single_push_insn is modifying
> stack_pointer_delta, because this is supposed to be done by emit_move_insn_1.

It is much cleaner to keep track of delta at lowest level - emit_single_push_insn,
that was also intention of my patch.  On i386 this obviously works, since
we are not getting crahses, but probably I've forgot to remove adustment
from emit_move_insn_1 path not used on i386.

More highlevel you do it, more places you need to modify.  emit_single_push_insn
was designed to catch all cases.

Can you please send me testcase I can drop into crosscompiler? I am still
puzzled about how the emit_single_push_insn can be called before
emit_move_insn_1 and dupplicate the work.

Honza