This is the mail archive of the
mailing list for the GCC project.
Re: Fix PR34916, 35519 Combine (Dataflow merge regression.)
On Wed, Mar 19, 2008 at 8:30 AM, Paolo Bonzini <firstname.lastname@example.org> wrote:
> > The log_links list being walked is very short - only one element per
> > immediately linked instruction. So typical length is perhaps 2 (without
> > duplicates). The largest would be for an instruction which has many
> > input operands where each is the output of an instructions in same
> > block, with no intervening usage. Maybe a libcall call of some type.
> > But still short list!
> I agree, create_log_links doesn't even add LOG_LINKS to asms, which
> potentially have a lot of uses. I think this is entirely comparable to
> using insertion sort because you know the size of the array is small.
> I cannot approve this though (and would not like to override Richi anyway).
So I'll take your word for it and approve the patch for trunk. If there are
no problems within a few days it is also ok for the 4.3 branch as this is
obviously a regression, right?