Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <75d07691-1e4f-741f-9852-38c0b4f520bc@synopsys.com>
 <CAHk-=wjJEgjCYzHZFPxTs01p7FMEHKKqXyqwRVBk6KnvHB1qVA@mail.gmail.com> <9538bb7e-a600-2211-6b4d-561b99f1deca@synopsys.com>
In-Reply-To: <9538bb7e-a600-2211-6b4d-561b99f1deca@synopsys.com>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Fri, 30 Apr 2021 15:52:52 -0700
Message-ID: <CAHk-=wizZ-HoBJ24TPFawS5zHqT+HcreqfmFRf3Cw4SS0NxV+Q@mail.gmail.com>
Subject: Re: Heads up: gcc miscompiling initramfs zlib decompression code at -O3
To:     Vineet Gupta <Vineet.Gupta1@synopsys.com>
Cc:     Arnd Bergmann <arnd@kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Masahiro Yamada <masahiroy@kernel.org>,
        Jann Horn <jannh@google.com>,
        lkml <linux-kernel@vger.kernel.org>,
        arcml <linux-snps-arc@lists.infradead.org>,
        "linux-arch@vger.kernel.org" <linux-arch@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, Apr 30, 2021 at 3:44 PM Vineet Gupta <Vineet.Gupta1@synopsys.com> wrote:
>
> I agree that -O2 is default, but we've had -O3 default for ARC kernel
> forever, since last decade seriously. The reason I turned it on back
> then was upside of 10% performance improvement on select LMBench numbers
> on hardware at the time which for a rookie kernel hacker was yay momemt.
> I can revisit this and see if that is still true.

It would be interesting if you can actually show 10% improvement, and
also pinpoint things in the profile.

I (long long long ago) actually used -O6 for kernel builds as a "give
me everything you have" (I don't think gcc has actually ever done
anything more than O3, but whatever).

In fact, you can find

   Q: What Does gcc -O6 Do?

in some kernel FAQ's from those historical times.

We eventually gave up on it, because it just generated bigger and
slower code, and people got very tired of all the compiler bugs.

               Linus