[PATCH, rfc] combine: Make code after a new trap unreachable (PR78432)

[Segher Boessenkool]

On Mon, Nov 28, 2016 at 02:44:44PM -0700, Jeff Law wrote:
> On 11/23/2016 04:55 PM, Segher Boessenkool wrote:
> >Combine can turn a conditional trap into an unconditional trap.  If it
> >does that it should make the code after it unreachable (an unconditional
> >trap should be the last insn in its bb, and that bb has no successors).
> >
> >This patch seems to work.  It is hard to be sure, this is very hard to
> >trigger.  Quite a few other passes look like they need something similar
> >as well, but I don't see anything else handling it yet either.
> >
> >How does this look?  Any better ideas?
> It's probably a reasonable place to start as there's lots of 
> similarities to converting a conditional branch to an unconditional. 
> Like you I worry there may be other instances of this issue may be lurking.
> 
> The possibility of having a conditional trap in the IL turn into an 
> unconditional trap has always been there.  As you note, it's somewhat 
> difficult to trigger since the SSA optimizers ought to expose/optimize 
> this long before we get to RTL.

I committed this with this changelog (and the PR # fixed):


2016-11-28  Segher Boessenkool  <segher@kernel.crashing.org>

	PR rtl-optimization/78342
	* combine.c: Include "cfghooks.h".
	(try_combine): If we create an unconditional trap, break the basic
	block in two just after it, and remove the edge between; also, set
	the *new_direct_jump_p flag so that cleanup_cfg is run.