From: Masayoshi Mizuma <msys.mizuma@gmail.com>
To: util-linux@vger.kernel.org
Cc: Masayoshi Mizuma <msys.mizuma@gmail.com>,
Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
Subject: [PATCH v3 1/5] lscpu: use cluster on aarch64 machine which doesn't have ACPI PPTT
Date: Fri, 20 Nov 2020 00:06:05 -0500 [thread overview]
Message-ID: <20201120050609.17409-2-msys.mizuma@gmail.com> (raw)
In-Reply-To: <20201120050609.17409-1-msys.mizuma@gmail.com>
From: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
lscpu may show the wrong number of sockets if the machine is aarch64 and
doesn't have ACPI PPTT.
That's because lscpu shows the number of sockets by using a sysfs entry
(cpu/cpuX/topology/core_siblings). The sysfs entry is set by MPIDR_EL1
register if the machine doesn't have ACPI PPTT. MPIDR_EL1 doesn't show
the physical socket information directly. It shows the affinity level.
According to linux/arch/arm64/kernel/topology.c:store_cpu_topology(),
the top level of affinity is called as 'Cluster'.
Use Cluster instead of Socket on the machine which doesn't have ACPI PPTT.
This patch is useful for aarch64 machine which is based on ARM
SBBR v1.0 and v1.1, the specs don't require ACPI PPTT. ARM SBBR v1.2
requires ACPI PPTT.
Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
---
sys-utils/lscpu-arm.c | 13 +++++++++++++
sys-utils/lscpu.1 | 3 +++
sys-utils/lscpu.c | 34 ++++++++++++++++++++++++++++------
sys-utils/lscpu.h | 2 ++
4 files changed, 46 insertions(+), 6 deletions(-)
diff --git a/sys-utils/lscpu-arm.c b/sys-utils/lscpu-arm.c
index 1cbc9775f..9391cca54 100644
--- a/sys-utils/lscpu-arm.c
+++ b/sys-utils/lscpu-arm.c
@@ -362,10 +362,23 @@ static void arm_decode(struct lscpu_cxt *cxt, struct lscpu_cputype *ct)
arm_rXpY_decode(ct);
}
+static int lscpu_is_cluster_arm(struct lscpu_cxt *cxt)
+{
+ struct stat st;
+
+ if (!(strcmp(cxt->arch->name, "aarch64")) &&
+ (stat(_PATH_ACPI_PPTT, &st) < 0) && (cxt->ncputypes == 1))
+ return 1;
+ else
+ return 0;
+}
+
void lscpu_decode_arm(struct lscpu_cxt *cxt)
{
size_t i;
+ cxt->is_cluster = lscpu_is_cluster_arm(cxt);
+
for (i = 0; i < cxt->ncputypes; i++)
arm_decode(cxt, cxt->cputypes[i]);
}
diff --git a/sys-utils/lscpu.1 b/sys-utils/lscpu.1
index ed14dc663..a41125395 100644
--- a/sys-utils/lscpu.1
+++ b/sys-utils/lscpu.1
@@ -55,6 +55,9 @@ The logical core number. A core can contain several CPUs.
.B SOCKET
The logical socket number. A socket can contain several cores.
.TP
+.B CLUSTER
+The logical cluster number. A cluster can contain several cores.
+.TP
.B BOOK
The logical book number. A book can contain several sockets.
.TP
diff --git a/sys-utils/lscpu.c b/sys-utils/lscpu.c
index 6e501b044..b9b8940a8 100644
--- a/sys-utils/lscpu.c
+++ b/sys-utils/lscpu.c
@@ -95,6 +95,7 @@ enum {
COL_CPU_CPU,
COL_CPU_CORE,
COL_CPU_SOCKET,
+ COL_CPU_CLUSTER,
COL_CPU_NODE,
COL_CPU_BOOK,
COL_CPU_DRAWER,
@@ -138,6 +139,7 @@ static struct lscpu_coldesc coldescs_cpu[] =
[COL_CPU_BOGOMIPS] = { "BOGOMIPS", N_("crude measurement of CPU speed"), SCOLS_FL_RIGHT, 1 },
[COL_CPU_CPU] = { "CPU", N_("logical CPU number"), SCOLS_FL_RIGHT, 1 },
[COL_CPU_CORE] = { "CORE", N_("logical core number"), SCOLS_FL_RIGHT },
+ [COL_CPU_CLUSTER] = { "CLUSTER", N_("logical cluster number"), SCOLS_FL_RIGHT },
[COL_CPU_SOCKET] = { "SOCKET", N_("logical socket number"), SCOLS_FL_RIGHT },
[COL_CPU_NODE] = { "NODE", N_("logical NUMA node number"), SCOLS_FL_RIGHT },
[COL_CPU_BOOK] = { "BOOK", N_("logical book number"), SCOLS_FL_RIGHT },
@@ -339,6 +341,10 @@ static char *get_cell_data(
case COL_CPU_SOCKET:
fill_id(cxt, cpu, socket, buf, bufsz);
break;
+ case COL_CPU_CLUSTER:
+ if (cxt->is_cluster)
+ fill_id(cxt, cpu, socket, buf, bufsz);
+ break;
case COL_CPU_DRAWER:
fill_id(cxt, cpu, drawer, buf, bufsz);
break;
@@ -843,7 +849,10 @@ print_summary_cputype(struct lscpu_cxt *cxt,
add_summary_s(tb, sec, _("Model:"), ct->revision ? ct->revision : ct->model);
add_summary_n(tb, sec, _("Thread(s) per core:"), ct->nthreads_per_core);
- add_summary_n(tb, sec, _("Core(s) per socket:"), ct->ncores_per_socket);
+ if (cxt->is_cluster)
+ add_summary_n(tb, sec, _("Core(s) per cluster:"), ct->ncores_per_socket);
+ else
+ add_summary_n(tb, sec, _("Core(s) per socket:"), ct->ncores_per_socket);
if (ct->nbooks) {
add_summary_n(tb, sec, _("Socket(s) per book:"), ct->nsockets_per_book);
@@ -852,8 +861,14 @@ print_summary_cputype(struct lscpu_cxt *cxt,
add_summary_n(tb, sec, _("Drawer(s):"), ct->ndrawers_per_system ?: ct->ndrawers);
} else
add_summary_n(tb, sec, _("Book(s):"), ct->nbooks_per_drawer ?: ct->nbooks);
- } else
- add_summary_n(tb, sec, _("Socket(s):"), ct->nsockets_per_book ?: ct->nsockets);
+ } else {
+ if (cxt->is_cluster)
+ add_summary_n(tb, sec, _("Cluster(s):"),
+ ct->nsockets_per_book ?: ct->nsockets);
+ else
+ add_summary_n(tb, sec, _("Socket(s):"),
+ ct->nsockets_per_book ?: ct->nsockets);
+ }
if (ct->stepping)
add_summary_s(tb, sec, _("Stepping:"), ct->stepping);
@@ -1325,8 +1340,12 @@ int main(int argc, char *argv[])
columns[ncolumns++] = COL_CPU_DRAWER;
if (ct && ct->nbooks)
columns[ncolumns++] = COL_CPU_BOOK;
- if (ct && ct->nsockets)
- columns[ncolumns++] = COL_CPU_SOCKET;
+ if (ct && ct->nsockets) {
+ if (cxt->is_cluster)
+ columns[ncolumns++] = COL_CPU_CLUSTER;
+ else
+ columns[ncolumns++] = COL_CPU_SOCKET;
+ }
if (ct && ct->ncores)
columns[ncolumns++] = COL_CPU_CORE;
if (cxt->ncaches)
@@ -1350,7 +1369,10 @@ int main(int argc, char *argv[])
if (!ncolumns) {
columns[ncolumns++] = COL_CPU_CPU;
columns[ncolumns++] = COL_CPU_CORE;
- columns[ncolumns++] = COL_CPU_SOCKET;
+ if (cxt->is_cluster)
+ columns[ncolumns++] = COL_CPU_CLUSTER;
+ else
+ columns[ncolumns++] = COL_CPU_SOCKET;
columns[ncolumns++] = COL_CPU_NODE;
columns[ncolumns++] = COL_CPU_CACHE;
cxt->show_compatible = 1;
diff --git a/sys-utils/lscpu.h b/sys-utils/lscpu.h
index 465bd3fe2..7d42816c3 100644
--- a/sys-utils/lscpu.h
+++ b/sys-utils/lscpu.h
@@ -33,6 +33,7 @@ UL_DEBUG_DECLARE_MASK(lscpu);
#define _PATH_SYS_NODE _PATH_SYS_SYSTEM "/node"
#define _PATH_SYS_DMI "/sys/firmware/dmi/tables/DMI"
#define _PATH_SYS_DMI_TYPE4 "/sys/firmware/dmi/entries/4-0/raw"
+#define _PATH_ACPI_PPTT "/sys/firmware/acpi/tables/PPTT"
struct lscpu_cache {
int id; /* unique identifier */
@@ -242,6 +243,7 @@ struct lscpu_cxt {
json : 1,
bytes : 1;
+ int is_cluster; /* For aarch64 if the machine doesn't have ACPI PPTT */
};
#define is_cpu_online(_cxt, _cpu) \
--
2.27.0
next prev parent reply other threads:[~2020-11-20 5:06 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-11-20 5:06 [PATCH v3 0/5] lscpu: Fix socket information on aarch64 machine Masayoshi Mizuma
2020-11-20 5:06 ` Masayoshi Mizuma [this message]
2020-11-20 5:06 ` [PATCH v3 2/5] lscpu-virt: split hypervisor_from_dmi_table() Masayoshi Mizuma
2020-11-20 5:06 ` [PATCH v3 3/5] lscpu-dmi: Move some functions related to DMI to lscpu-dmi Masayoshi Mizuma
2020-11-20 5:06 ` [PATCH v3 4/5] lscpu: add helper to get physical sockets Masayoshi Mizuma
2020-11-20 5:06 ` [PATCH v3 5/5] lscpu: show the number of physical socket on aarch64 machine without ACPI PPTT Masayoshi Mizuma
2020-11-20 9:07 ` [PATCH v3 0/5] lscpu: Fix socket information on aarch64 machine Karel Zak
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20201120050609.17409-2-msys.mizuma@gmail.com \
--to=msys.mizuma@gmail.com \
--cc=m.mizuma@jp.fujitsu.com \
--cc=util-linux@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).