Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Thomas Gleixner <tglx@linutronix.de>
To:     Pete Swain <swine@google.com>
Cc:     Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        Dave Hansen <dave.hansen@linux.intel.com>, x86@kernel.org,
        "H. Peter Anvin" <hpa@zytor.com>,
        Paolo Bonzini <pbonzini@redhat.com>,
        Sean Christopherson <seanjc@google.com>,
        Vitaly Kuznetsov <vkuznets@redhat.com>,
        Wanpeng Li <wanpengli@tencent.com>,
        Jim Mattson <jmattson@google.com>,
        Joerg Roedel <joro@8bytes.org>,
        John Stultz <john.stultz@linaro.org>,
        Stephen Boyd <sboyd@kernel.org>,
        "Maciej W. Rozycki" <macro@orcam.me.uk>,
        Johan Hovold <johan@kernel.org>,
        Feng Tang <feng.tang@intel.com>,
        "Paul E. McKenney" <paulmck@kernel.org>,
        Juergen Gross <jgross@suse.com>, linux-kernel@vger.kernel.org,
        kvm@vger.kernel.org, Pete Swain <swine@google.com>
Subject: Re: [PATCH 2/2] timers: retpoline mitigation for time funcs
In-Reply-To: <87r165gmoz.ffs@tglx>
References: <20220218221820.950118-1-swine@google.com>
 <20220218221820.950118-2-swine@google.com> <87r165gmoz.ffs@tglx>
Date:   Mon, 11 Apr 2022 13:07:08 +0200
Message-ID: <87mtgrg9pf.ffs@tglx>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk

On Sun, Apr 10 2022 at 14:14, Thomas Gleixner wrote:
> On Fri, Feb 18 2022 at 14:18, Pete Swain wrote:
>> @@ -245,7 +245,8 @@ static int clockevents_program_min_delta(struct clock_event_device *dev)
>>  
>>  		dev->retries++;
>>  		clc = ((unsigned long long) delta * dev->mult) >> dev->shift;
>> -		if (dev->set_next_event((unsigned long) clc, dev) == 0)
>> +		if (INDIRECT_CALL_1(dev->set_next_event, lapic_next_deadline,
>> +				  (unsigned long) clc, dev) == 0)
>
> No. We are not sprinkling x86'isms into generic code.

Even if we would do that, then this is completely useless.

The hotpath function is clockevents_program_event() and not
clockevents_program_min_delta().

The latter is invoked from the former if the to be programmed event is
in the past _and_ the force argument is set, i.e. in 0.1% of the
invocations of clockevents_program_event(), which itself has an indirect
call to dev->set_next_event().

If your profiles show clockevents_program_min_delta() as dominant, then
there is something massively wrong with your kernel.

Thanks,

        tglx