This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: SSA_NAMES: should there be an unused, un-free limbo?
- From: Jeffrey Law <law at redhat dot com>
- To: Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>
- Cc: Daniel Berlin <dberlin at dberlin dot org>, Diego Novillo <dnovillo at redhat dot com>, Robert Kennedy <jimbob at google dot com>, gcc at gcc dot gnu dot org
- Date: Wed, 03 Jan 2007 17:12:29 -0700
- Subject: Re: SSA_NAMES: should there be an unused, un-free limbo?
- References: <1166723338.28081.168.camel@sweet.slc.redhat.com> <20061221191848.GA8856@atrey.karlin.mff.cuni.cz> <1166768298.28081.225.camel@sweet.slc.redhat.com> <20061222125603.GA14841@atrey.karlin.mff.cuni.cz> <20061224080850.GA1839@atrey.karlin.mff.cuni.cz>
- Reply-to: law at redhat dot com
On Sun, 2006-12-24 at 09:08 +0100, Zdenek Dvorak wrote:
>
> As expected, more complications than I believed appeared. The changes
> to bsi_remove and release_defs would be basically sufficient for ssa
> names for real operands, however we are losing ssa names for virtual
> operands everywhere, and on several places relying on that they are not
> released.
>
> One problem with the patch seems to be that it appears to increase
> compile times significantly (by some 4% on compiling preprocessed gcc
> sources), I did not investigate where does this slowdown come from.
If the slowdown was with an ENABLE_CHECKING compiler, then the extra
walk over the BBs and statements and the operands within the statements
could be a significant hunk.
Finding a way to mark the names as part of an existing walk in the
verification step might help siginificantly.
I already approved a hunk of this patch that Robert submitted
separately.
jeff