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;
From:   Thomas Gleixner <tglx@linutronix.de>
To:     =?utf-8?Q?C=C3=A9dric?= Le Goater <clg@kaod.org>,
        Nicholas Piggin <npiggin@gmail.com>
Cc:     linux-kernel@vger.kernel.org
Subject: Re: [PATCH] genirq: reduce irqdebug bouncing cachelines
In-Reply-To: <b299bb0c-2997-6f6f-8646-4a7ecb2c79c0@kaod.org>
References: <20210402132037.574661-1-npiggin@gmail.com> <87im4u2vxx.ffs@nanos.tec.linutronix.de> <2dae4501-6f01-1b32-4b69-1dfc94c93425@kaod.org> <87wnt71xmm.ffs@nanos.tec.linutronix.de> <b299bb0c-2997-6f6f-8646-4a7ecb2c79c0@kaod.org>
Date:   Tue, 13 Apr 2021 22:24:28 +0200
Message-ID: <87tuoax79f.ffs@nanos.tec.linutronix.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Tue, Apr 13 2021 at 14:16, C=C3=A9dric Le Goater wrote:
>>> We could test irq_settings_no_debug() directly under handle_nested_irq(=
)=20
>>> and handle_irq_event_percpu() to avoid calling note_interrupt(), just=20
>>> like we do for noirqdebug.
>>=20
>> We can do that, but then we should not just make it:
>>=20
>>    if (!irqnodebug && !irq_settings_no_debug(desc))
>>    	note_interrupt(...);
>>=20
>> Instead have only one condition:
>>=20
>>    if (!irq_settings_no_debug(desc))
>>    	note_interrupt(...);
>>=20
>> See the uncompiled delta patch below.
>
> I merged this second part with the first and gave IRQF_NO_DEBUG a try=20
> on P8 and P9 systems and all looked fine. I should send both patches=20
> after IRQF_NO_AUTOEN is merged in mainline.=20=20=20=20=20

Does having that NODEBUG flag set on the IPI irqs make a measurable
difference or is it just too small to matter?

Thanks,

        tglx