This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: 4 GCC regressions, 2 new, with your patch on 2002-07-18T09:56:17Z.
- From: Jan Hubicka <jh at suse dot cz>
- To: Geoff Keating <geoffk at geoffk dot org>
- Cc: Jan Hubicka <jh at suse dot cz>, gcc-patches at gcc dot gnu dot org
- Date: Fri, 9 Aug 2002 11:11:37 +0200
- Subject: Re: 4 GCC regressions, 2 new, with your patch on 2002-07-18T09:56:17Z.
- References: <200207181329.g6IDT2F03232@maat.sfbay.redhat.com> <20020718200257.GD10794@atrey.karlin.mff.cuni.cz> <20020722010558.GB2484@atrey.karlin.mff.cuni.cz> <200207221204.g6MC4N814351@desire.geoffk.org> <20020722133957.GH2426@atrey.karlin.mff.cuni.cz> <200207221458.g6MEw2l14690@desire.geoffk.org> <20020725131411.GD30520@atrey.karlin.mff.cuni.cz> <jmd6t6doy2.fsf@desire.geoffk.org>
> > c) I hope to move the code motion to hoist* functions I've
> > contributed that uses liveness information to validate motion to
> > given point. In case we will be systematically choosing points
> > where flags are alive, we will disable all motion of instructions
> > having clobbers in parallel
>
> This will be better. I hope that in the long run all of GCSE will go
> away and be replaced with a SSA-based pass.
Really? I thought that the SSA on lowlevel RTL is no-no becuase of
subregs and strict low parts etc. We need some CSE-like pass to cleanup
after last code lowering stage in any stage and it should be global.
>
> > 3) The clobbers of the instruction I move over are not causing problems
> > here as the instruction is known to have only one effect of storing
> > newly created pseudo.
>
> Then single_set is unnecessary, correct? You can use PATTERN instead.
I am reffring to instruction I am moving, not the instruction I am
examining.
Honza