linux-kernel - Re: [PATCH] riscv: Fix build with CONFIG_CC_OPTIMIZE_FOR

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 4 Oct 2022 18:15:11 +0100
From:   Conor Dooley <conor@...nel.org>
To:     Atish Patra <atishp@...shpatra.org>
Cc:     Heiko Stuebner <heiko@...ech.de>,
        Conor Dooley <conor.dooley@...rochip.com>,
        Jessica Clarke <jrtc27@...c27.com>,
        Palmer Dabbelt <palmer@...belt.com>,
        linux-riscv <linux-riscv@...ts.infradead.org>,
        Samuel Holland <samuel@...lland.org>,
        Albert Ou <aou@...s.berkeley.edu>,
        Anup Patel <anup@...infault.org>,
        Atish Patra <atishp@...osinc.com>, Dao Lu <daolu@...osinc.com>,
        Guo Ren <guoren@...nel.org>,
        Jisheng Zhang <jszhang@...nel.org>,
        Paul Walmsley <paul.walmsley@...ive.com>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] riscv: Fix build with CONFIG_CC_OPTIMIZE_FOR_SIZE=y

On Tue, Oct 04, 2022 at 09:52:41AM -0700, Atish Patra wrote:
> On Sat, Oct 1, 2022 at 1:13 PM Conor Dooley <conor@...nel.org> wrote:

> > > Thanks. It would be good to understand what happens when "pause" is
> > > executed on these boards ?
> >
> > The actual pause instruction? uhh, so with the usual "I don't know what
> > I am doing" disclaimer, I ran each of the .insn and pause instruction 48
> > times in a row and checked the time elapsed via rdcycle & then ran that
> > c program 1000 times in a bash loop. Got the below, the insns were run
> > first and then the pauses.
> >         insn    pause
> > min     2.3     3.2
> > max     9.5     10.6
> > avg     27.0    29.1
> > 5%      2.9     4.2
> > 95%     18.1    19.1
> >
> > Swapping the pause & insn order around made a minor difference, but not
> > enough to report on. I'd be very wary of drawing any real conclusions
> > from this data, but at least both are roughly similar (and certainly not
> > even close to doing the div w/ zero args.
> >
> 
> Yeah. That's what I was expecting. So we can't drop the div for now. Otherwise,
> the existing hardware(don't support Zhintpause) suffers by spinning faster.
> 
> Thanks for running the experiments.

I've lost track, does that mean the patch is okay as, is or needs to be
changed? The former, right?

Thanks,
Conor.