This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: schedule-ebbs and trapping insns
- From: Jeff Law <law at porcupine dot slc dot redhat dot com>
- To: "David S. Miller" <davem at redhat dot com>
- Cc: gcc at gcc dot gnu dot org, bernds at redhat dot com, rth at redhat dot com, jakub at redhat dot com
- Date: Mon, 01 Jul 2002 11:30:25 -0600
- Subject: Re: schedule-ebbs and trapping insns
- Reply-to: law at redhat dot com
In message <20020627.005217.26466268.davem@redhat.com>, "David S. Miller" write
s:
>
>How can the init_ready_list used by sched-ebb.c legally not
>take into consideration trapping instructions like the
>sched-rgn.c version does? You can't move trapping instructions
>between basic blocks.
Hmmm, it seems to me that a trapping insn should act as a scheduling
barrier for any other potentially trapping insn. Right? I believe
sched-rgn already implements this and conceptually it should be
independent of whether or not we're scheduling regions or basic blocks,
so it seems to me it should probably move into sched-deps.c since these
schedulign barriers are most likely implemented by adding dependencies
into the graph.