This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Simple commit_edge_insertions fix
- To: Jeffrey A Law <law at cygnus dot com>
- Subject: Re: Simple commit_edge_insertions fix
- From: Richard Henderson <rth at cygnus dot com>
- Date: Mon, 15 May 2000 11:42:37 -0700
- Cc: Jan Hubicka <jh at suse dot cz>, egcs-patches at egcs dot cygnus dot com
- References: <20000515111629.B3407@cygnus.com> <5084.958414463@upchuck>
On Mon, May 15, 2000 at 12:14:23PM -0600, Jeffrey A Law wrote:
> That's a surprise too, though we do have a lot which are similar, but
> they should [not?] be identical except for the one mode character. That
> would be silly in most circumstances.
>
> Can you please point out an example or two so that I can look at them?
Line 1538: 16 consecutive branch on bit patterns.
[ Which could be matched with one SI and one DImode pattern,
if I read things right. In fact, most of the branch patterns
could be compressed it looks like, with judicious use of
the c_test. ]
Line 6400: SI and DImode interspace jump.
[ Which could probably completely ignore the mode of r2. ]
Line 6815: SF, SI, HI, QImode output_parallel_movb.
Those were the ones that caught my eye looking for the pattern
that you'd mentioned on this thread.
r~