[0/3] Fix PR78120, in ifcvt/rtlanal/i386.

[Bernd Schmidt]

On 11/24/2016 03:36 PM, Segher Boessenkool wrote:
> On Thu, Nov 24, 2016 at 03:26:45PM +0100, Bernd Schmidt wrote:
>> On 11/24/2016 03:21 PM, Segher Boessenkool wrote:
>>
>>> Combine uses insn_rtx_cost extensively.  I have tried to change it to use
>>> the full rtx cost, not just the source cost, a few times before, and it
>>> always only regressed.  Part of it is that most ports' cost calculations
>>> are, erm, not so great -- every target we care about needs fixes.
>>>
>>> Let's please not try this in stage 3.
>>
>> It got approved and committed. Do you want me to revert it now or wait
>> for obvious signs of fallout?
>
> In my opinion it is an early stage 1 thing, not something suitable for
> stage 3.  I can do some simple tests on various targets if you want.

Sure.

I'll make the argument that stage 3 is when we fix stuff, including 
performance regressions, and this patch is very clearly a fix. When we 
have very obvious distortions like a case where costs from insn_rtx_cost 
and seq_cost aren't comparable, it's impossible to arrive at sane solutions.


Bernd