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
- From: law at redhat dot com
- To: Diego Novillo <dnovillo at redhat dot com>
- Cc: Andrew Macleod <amacleod at redhat dot com>, gcc-patches <gcc-patches at gcc dot gnu dot org>
- Date: Mon, 25 Aug 2003 08:39:20 -0600
- Subject: Re: [tree-ssa] Dominator opts fixes and enhancements
- Reply-to: law at redhat dot com
In message <1061821474.27406.69.camel@frodo.toronto.redhat.com>, Diego Novillo
writes:
>We do not build the tree CFG with multiple entry points in mind. So,
>ENTRY_BLOCK_PTR always has exactly one successor: basic block 0.
Well, it's not terribly hard to change any code which makes that
assumption to do something like
for (e = ENTRY_BLOCK_PTR->succ; e; e = e->succ_next)
do something (e->dest)
>I'm not even sure that multiple entry points are supported by the RTL
>optimizers. I seem to remember discussions about that in the lists or
>IRC, but my memory is hazy.
The RTL optimizers try to use code like the above rather than
make assumptions about the number of entry points or the indices of
their blocks. Now reality is without input code which exercises
multiple entry points odds are some places still need to be converted.