Received-SPF: pass (google.com: domain of linux-kernel+bounces-40620-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
From: Thomas Gleixner <tglx@linutronix.de>
To: "Zhang, Rui" <rui.zhang@intel.com>, "linux-kernel@vger.kernel.org"
 <linux-kernel@vger.kernel.org>
Cc: "Raj, Ashok" <ashok.raj@intel.com>, "mhklinux@outlook.com"
 <mhklinux@outlook.com>, "arjan@linux.intel.com" <arjan@linux.intel.com>,
 "ray.huang@amd.com" <ray.huang@amd.com>, "thomas.lendacky@amd.com"
 <thomas.lendacky@amd.com>, "andrew.cooper3@citrix.com"
 <andrew.cooper3@citrix.com>, "Sivanich, Dimitri"
 <dimitri.sivanich@hpe.com>, "Tang, Feng" <feng.tang@intel.com>,
 "kan.liang@linux.intel.com" <kan.liang@linux.intel.com>, "Mehta, Sohil"
 <sohil.mehta@intel.com>, "peterz@infradead.org" <peterz@infradead.org>,
 "paulmck@kernel.org" <paulmck@kernel.org>, "kprateek.nayak@amd.com"
 <kprateek.nayak@amd.com>, "jgross@suse.com" <jgross@suse.com>,
 "andy@infradead.org" <andy@infradead.org>, "x86@kernel.org"
 <x86@kernel.org>
Subject: Re: [patch v2 21/30] x86/cpu/topology: Use topology bitmaps for sizing
In-Reply-To: <7c8874d971c69fb4c22c1b771983f8d5109a9387.camel@intel.com>
References: <20240118123127.055361964@linutronix.de>
 <20240118123649.739960607@linutronix.de>
 <7c8874d971c69fb4c22c1b771983f8d5109a9387.camel@intel.com>
Date: Fri, 26 Jan 2024 21:22:47 +0100
Message-ID: <87fryjhmjs.ffs@tglx>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

On Fri, Jan 26 2024 at 07:07, Zhang, Rui wrote:
>> > =C2=A0
>> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0cnta =3D domain_weight(TOPO=
_PKG_DOMAIN);
>> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0cntb =3D domain_weight(TOPO=
_DIE_DOMAIN);
>> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0__max_logical_packages =3D =
cnta;
>> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0__max_dies_per_package =3D =
1U << (get_count_order(cntb) - >
>> > get_count_order(cnta));
>> > +
>> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0pr_info("Max. logical packa=
ges: %3u\n", cnta);
>> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0pr_info("Max. logical dies:=
=C2=A0=C2=A0=C2=A0=C2=A0 %3u\n", cntb);
>> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0pr_info("Max. dies per pack=
age: %3u\n", >
>> > __max_dies_per_package);
>> > +
>> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0cnta =3D domain_weight(TOPO=
_CORE_DOMAIN);
>> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0cntb =3D domain_weight(TOPO=
_SMT_DOMAIN);
>> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0smp_num_siblings =3D 1U << =
(get_count_order(cntb) - >
>> > get_count_order(cnta));
>> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0pr_info("Max. threads per c=
ore: %3u\n", smp_num_siblings);
>> > +
>
> I missed this but Ashok catches it.
>
> Say, on my Adlerlake platform, which has 4 Pcores with HT + 8 Ecores,
> cnta is 12, cntb is 16, and smp_num_siblings is set to 1 erroneously.
>
> I think we should use
> 	smp_num_siblings =3D DIV_ROUND_UP(cntb, cnta);
> here.

Indeed. That's more than obvious.

> Or even check each core to get the maximum value (in case there are
> more than 2 siblings in a core some day).

We want to get rid of HT not make it worse.