This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] Block merging (updated)
- From: Richard Henderson <rth at redhat dot com>
- To: law at redhat dot com
- Cc: Jan Hubicka <hubicka at ucw dot cz>, Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>, Steven Bosscher <s dot bosscher at student dot tudelft dot nl>, gcc-patches at gcc dot gnu dot org
- Date: Tue, 13 Jan 2004 13:57:48 -0800
- Subject: Re: [tree-ssa] Block merging (updated)
- References: <20040113174719.GA19340@atrey.karlin.mff.cuni.cz> <200401131757.i0DHvliI024499@speedy.slc.redhat.com>
On Tue, Jan 13, 2004 at 10:57:47AM -0700, law@redhat.com wrote:
> No. We do not delete user labels. If something is deleting a user label,
> then that code is buggy.
I'm less and less interested in preserving user labels these days.
> Err, no. We have turned them into DELETED_LABELs, but that is not the
> same as removing them since they will end up in the debug information.
As DELETED_LABELs, they float around a *lot*. Quite often to
the beginning of the function. We have bug reports about this.
My personal thought is that it would actually be more useful
for the user if the label was completely removed so that we
could not emit the address in the debug info so that the
debugger could say "label has been optimized away".
That's not to say we shouldn't make efforts to preserve the user's
label when that's an alternative. E.g. don't accidentally let an
artificial label at the same block get used instead.
r~