This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] Dominator opts fixes and enhancements
On Mon, 2003-08-25 at 09:27, Andrew MacLeod wrote:
> So if a variable had a real
> definition, and that got deleted or changed, the var would have no
> definition, and the annotation would *not* have a default defintion
> listed right? So we can tell that it is in error that it doesn't have a
> definition...
>
Sorry, I can't parse the above :)
The default definition is *only* created if, while renaming the code
into SSA, the very first reference we find of a variable is a USE or a
VUSE. In that case, if the variable had a default definition assigned
to it already, we just use it, otherwise we create a new one. So,
default definitions are always artificial and cannot be removed by any
optimization.
> Also, this new symbol expose of partial ranges isnt going to result in
> potential virtual/real use mixes is it?
>
Nope. I keep meaning to add a test for this in tree_verify_flow_info.
Diego.