This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Combine four insns
On 08/20/2010 12:26 AM, Andrew Pinski wrote:
> On Thu, Aug 19, 2010 at 3:14 PM, Eric Botcazou <ebotcazou@adacore.com> wrote:
>>>> for the second one you checked in something without approval
>>>
>>> I don't believe this is the case. Where, specifically?
>>
>> Your message:
>> http://gcc.gnu.org/ml/gcc-patches/2010-07/msg02214.html
>>
>> Paolo's reply:
>> http://gcc.gnu.org/ml/gcc-patches/2010-07/msg02226.html
>
> That reply is a bit weird because it does seem like an approval for
> almost all of the patch. Though Paolo did mention he could not
> approve those two lines but he seems like he was saying to go ahead
> and apply it anyways. Maybe I would have waited a few more days
> before applying it or asking for a clarification to make sure people
> would not have disagreed with those two lines.
I did not expect that reasonable people would disagree with these two
lines. I still think they count as obvious if the rest is approved.
Who disagrees?
> It is tough call in my
> mind about this patch and those two lines. Though those two lines
> increased compile time because it enabled a whole new pass which was
> not there before.
Not really: it's only run if there are DImode regs, and I showed in the
thread that bootstrap times are unaffected. And the pass was there
before, it had been enabled previously and was only disabled due to
bugs. I fixed the bugs, made it faster, and reenabled it in the
specific case where it can be beneficial. I find it hard to believe
that if the first two parts are approved, the last part counts as
"checking something in without approval".
Bernd