Fix problem with late insns in sibcall

Richard Kenner kenner@vlsi1.ultra.nyu.edu
Sun May 4 18:55:00 GMT 2003


    What do you think ?

I'd like to understand the issue more.  It seems clear to me that if such
a "late" insn did occur, things would go wrong.   What may be the case is
just that the way things are, it can't occur.  But if that's just an
"accident", then we should either leave the patch the way it is or put
in some code that will abort if a problematic case occurs.  I don't
understand the issue well enough to choose, so I think at least some
investigation is called for.



More information about the Gcc-patches mailing list