Re: Fix PR rtl-optimization/28243

[Richard Sandiford <richard at codesourcery dot com>]

Eric Botcazou <ebotcazou@libertysurf.fr> writes:
>> So the combination used the REG_EQUAL note attached to insn 83?  If so,
>> this looks very similar to PR rtl-optimization/25514 (itself fixed by
>> PR rtl-optimization/27736), which is only on mainline at the moment.
>
> Yes, it does.
>
>> You only mention 4.1 in your message above so I just wanted to check
>> that whether or not this is a separate problem, or whether a backport
>> of 25514/27736 would have worked.
>
> I think it's a sub-problem of yours and that my patch is relatively safe.
> I must admit that I haven't fully understood your solution but, if the 
> consensus is that it can eventually be backported to the 4.1 branch, my patch 
> should probably be backed out.

Sorry for dropping the ball here.  I would like to push that idea though.
The problem I have with your patch is that (IMO) it's just hacking around
the problem that 4.1 is not analysing the right live range.  The code in
the loop simply shouldn't be looking at insns that follow the death of the
register.

Richard