From: Wei Chen <Wei.Chen@arm.com>
To: Julien Grall <julien@xen.org>,
"xen-devel@lists.xenproject.org" <xen-devel@lists.xenproject.org>,
"sstabellini@kernel.org" <sstabellini@kernel.org>
Cc: Bertrand Marquis <Bertrand.Marquis@arm.com>
Subject: RE: [XEN RFC PATCH 13/40] xen/arm: introduce numa_set_node for Arm
Date: Wed, 25 Aug 2021 12:07:02 +0000 [thread overview]
Message-ID: <DB9PR08MB6857E228F4C0847D4DF014ED9EC69@DB9PR08MB6857.eurprd08.prod.outlook.com> (raw)
In-Reply-To: <a07963e4-84c8-32de-4e8b-013a5419deea@xen.org>
Hi Julien,
> -----Original Message-----
> From: Julien Grall <julien@xen.org>
> Sent: 2021年8月25日 18:37
> To: Wei Chen <Wei.Chen@arm.com>; xen-devel@lists.xenproject.org;
> sstabellini@kernel.org; jbeulich@suse.com
> Cc: Bertrand Marquis <Bertrand.Marquis@arm.com>
> Subject: Re: [XEN RFC PATCH 13/40] xen/arm: introduce numa_set_node for
> Arm
>
> Hi Wei,
>
> On 11/08/2021 11:23, Wei Chen wrote:
> > This API is used to set one CPU to a NUMA node. If the system
> > configure NUMA off or system initialize NUMA failed, the
> > online NUMA node would set to only node#0. This will be done
> > in following patches. When NUMA turn off or init failed,
> > node_online_map will be cleared and set node#0 online. So we
> > use node_online_map to prevent to set a CPU to an offline node.
>
> IHMO numa_set_node() should behave exactly the same way on x86 and Arm
> because this is going to be used by the common code.
>
> From the commit message, I don't quite understand why the check is
> necessary on Arm but not on x86. Can you clarify it?
>
Yes, in patch#27, in smpboot.c, dt_smp_init_cpus function.
We will parse CPU numa-node-id from dtb CPU node. If we get
a valid node ID for one CPU, we will invoke numa_set_node to
create CPU-NODE map. But in our testing, we found when NUMA
init failed, numa_set_node still can set CPU to a offline
or invalid NODE. So we're using node_online_map to prevent
this behavior. Otherwise we have to check node_online_map
everywhere before we call numa_set_node.
x86 actually is doing the same way, but it handles node_online_map
check out of numa_set_node:
57 void __init init_cpu_to_node(void)
58 {
59 unsigned int i;
60 nodeid_t node;
61
62 for ( i = 0; i < nr_cpu_ids; i++ )
63 {
64 u32 apicid = x86_cpu_to_apicid[i];
65 if ( apicid == BAD_APICID )
66 continue;
67 node = apicid < MAX_LOCAL_APIC ? apicid_to_node[apicid] : NUMA_NO_NODE;
68 if ( node == NUMA_NO_NODE || !node_online(node) )
69 node = 0;
70 numa_set_node(i, node);
71 }
72 }
> >
> > Signed-off-by: Wei Chen <wei.chen@arm.com>
> > ---
> > xen/arch/arm/Makefile | 1 +
> > xen/arch/arm/numa.c | 31 +++++++++++++++++++++++++++++++
> > xen/include/asm-arm/numa.h | 2 ++
> > xen/include/asm-x86/numa.h | 1 -
> > xen/include/xen/numa.h | 1 +
> > 5 files changed, 35 insertions(+), 1 deletion(-)
> > create mode 100644 xen/arch/arm/numa.c
> >
> > diff --git a/xen/arch/arm/Makefile b/xen/arch/arm/Makefile
> > index 3d3b97b5b4..6e3fb8033e 100644
> > --- a/xen/arch/arm/Makefile
> > +++ b/xen/arch/arm/Makefile
> > @@ -35,6 +35,7 @@ obj-$(CONFIG_LIVEPATCH) += livepatch.o
> > obj-y += mem_access.o
> > obj-y += mm.o
> > obj-y += monitor.o
> > +obj-$(CONFIG_NUMA) += numa.o
> > obj-y += p2m.o
> > obj-y += percpu.o
> > obj-y += platform.o
> > diff --git a/xen/arch/arm/numa.c b/xen/arch/arm/numa.c
> > new file mode 100644
> > index 0000000000..1e30c5bb13
> > --- /dev/null
> > +++ b/xen/arch/arm/numa.c
> > @@ -0,0 +1,31 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Arm Architecture support layer for NUMA.
> > + *
> > + * Copyright (C) 2021 Arm Ltd
> > + *
> > + * This program is free software; you can redistribute it and/or modify
> > + * it under the terms of the GNU General Public License version 2 as
> > + * published by the Free Software Foundation.
> > + *
> > + * This program is distributed in the hope that it will be useful,
> > + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> > + * GNU General Public License for more details.
> > + *
> > + * You should have received a copy of the GNU General Public License
> > + * along with this program. If not, see <http://www.gnu.org/licenses/>.
> > + *
> > + */
> > +#include <xen/init.h>
> > +#include <xen/nodemask.h>
> > +#include <xen/numa.h>
> > +
> > +void numa_set_node(int cpu, nodeid_t nid)
> > +{
> > + if ( nid >= MAX_NUMNODES ||
> > + !nodemask_test(nid, &node_online_map) )
> > + nid = 0;
> > +
> > + cpu_to_node[cpu] = nid;
> > +}
> I think numa_set_node() will want to be implemented in common code.
>
See my above comment. If x86 is ok, I think yes, we can do it
in common code.
> > diff --git a/xen/include/asm-arm/numa.h b/xen/include/asm-arm/numa.h
> > index ab9c4a2448..1162c702df 100644
> > --- a/xen/include/asm-arm/numa.h
> > +++ b/xen/include/asm-arm/numa.h
> > @@ -27,6 +27,8 @@ extern mfn_t first_valid_mfn;
> > #define node_start_pfn(nid) (mfn_x(first_valid_mfn))
> > #define __node_distance(a, b) (20)
> >
> > +#define numa_set_node(x, y) do { } while (0)
>
> I would define it in xen/numa.h so other arch can take advantage ot it.
> Also, please use a static inline helper so the arguments are evaluated.
>
Ok
> > +
> > #endif
> >
> > #endif /* __ARCH_ARM_NUMA_H */
> > diff --git a/xen/include/asm-x86/numa.h b/xen/include/asm-x86/numa.h
> > index f8e4e15586..69859b0a57 100644
> > --- a/xen/include/asm-x86/numa.h
> > +++ b/xen/include/asm-x86/numa.h
> > @@ -22,7 +22,6 @@ extern nodeid_t pxm_to_node(unsigned int pxm);
> > #define ZONE_ALIGN (1UL << (MAX_ORDER+PAGE_SHIFT))
> >
> > extern int srat_disabled(void);
> > -extern void numa_set_node(int cpu, nodeid_t node);
> > extern nodeid_t setup_node(unsigned int pxm);
> > extern void srat_detect_node(int cpu);
> >
> > diff --git a/xen/include/xen/numa.h b/xen/include/xen/numa.h
> > index 258a5cb3db..3972aa6b93 100644
> > --- a/xen/include/xen/numa.h
> > +++ b/xen/include/xen/numa.h
> > @@ -70,6 +70,7 @@ extern int valid_numa_range(u64 start, u64 end,
> nodeid_t node);
> >
> > extern void numa_init_array(void);
> > extern void numa_initmem_init(unsigned long start_pfn, unsigned long
> end_pfn);
> > +extern void numa_set_node(int cpu, nodeid_t node);
> > extern bool numa_off;
> > extern int numa_fake;
> >
> >
>
> Cheers,
>
> --
> Julien Grall
next prev parent reply other threads:[~2021-08-25 12:07 UTC|newest]
Thread overview: 196+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-08-11 10:23 [XEN RFC PATCH 00/40] Add device tree based NUMA support to Arm64 Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 01/40] tools: Fix -Werror=maybe-uninitialized for xlu_pci_parse_bdf Wei Chen
2021-08-11 10:49 ` Jan Beulich
2021-08-13 6:28 ` Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 02/40] xen/arm: Print a 64-bit number in hex from early uart Wei Chen
2021-08-19 13:05 ` Julien Grall
2021-08-20 1:13 ` Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 03/40] xen/x86: Initialize memnodemapsize while faking NUMA node Wei Chen
2021-08-12 15:32 ` Jan Beulich
2021-08-13 7:26 ` Wei Chen
2021-08-13 8:29 ` Jan Beulich
2021-08-11 10:23 ` [XEN RFC PATCH 04/40] xen/arm: return default DMA bit width when platform is not set Wei Chen
2021-08-11 10:54 ` Jan Beulich
2021-08-13 6:54 ` Wei Chen
2021-08-13 6:56 ` Jan Beulich
2021-08-19 13:28 ` Julien Grall
2021-08-20 2:04 ` Wei Chen
2021-08-20 8:20 ` Julien Grall
2021-08-20 9:37 ` Wei Chen
2021-08-20 11:18 ` Julien Grall
2021-08-20 11:58 ` Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 05/40] xen/arm: Fix lowmem_bitsize when arch_get_dma_bitsize return 0 Wei Chen
2021-08-19 13:32 ` Julien Grall
2021-08-20 2:05 ` Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 06/40] xen: decouple NUMA from ACPI in Kconfig Wei Chen
2021-08-12 15:36 ` Jan Beulich
2021-08-13 7:27 ` Wei Chen
2021-08-12 16:54 ` Julien Grall
2021-08-13 7:28 ` Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 07/40] xen/arm: use !CONFIG_NUMA to keep fake NUMA API Wei Chen
2021-08-19 13:34 ` Julien Grall
2021-08-20 2:08 ` Wei Chen
2021-08-20 8:23 ` Julien Grall
2021-08-20 10:24 ` Wei Chen
2021-08-20 11:24 ` Julien Grall
2021-08-20 12:23 ` Wei Chen
2021-08-20 14:41 ` Julien Grall
2021-08-11 10:23 ` [XEN RFC PATCH 08/40] xen/x86: Move NUMA memory node map functions to common Wei Chen
2021-08-23 17:47 ` Julien Grall
2021-08-24 4:07 ` Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 09/40] xen/x86: Move numa_add_cpu_node " Wei Chen
2021-08-23 17:54 ` Julien Grall
2021-08-24 4:18 ` Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 10/40] xen/x86: Move NR_NODE_MEMBLKS macro " Wei Chen
2021-08-23 17:58 ` Julien Grall
2021-08-11 10:23 ` [XEN RFC PATCH 11/40] xen/x86: Move NUMA nodes and memory block ranges " Wei Chen
2021-08-24 17:40 ` Julien Grall
2021-08-25 0:57 ` Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 12/40] xen/x86: Move numa_initmem_init " Wei Chen
2021-08-25 10:21 ` Julien Grall
2021-08-25 11:15 ` Wei Chen
2021-08-25 13:26 ` Julien Grall
2021-08-11 10:23 ` [XEN RFC PATCH 13/40] xen/arm: introduce numa_set_node for Arm Wei Chen
2021-08-25 10:36 ` Julien Grall
2021-08-25 12:07 ` Wei Chen [this message]
2021-08-25 13:24 ` Julien Grall
2021-08-26 5:13 ` Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 14/40] xen/arm: set NUMA nodes max number to 64 by default Wei Chen
2021-08-25 13:28 ` Julien Grall
2021-08-25 13:36 ` Jan Beulich
2021-08-26 2:26 ` Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 15/40] xen/x86: move NUMA API from x86 header to common header Wei Chen
2021-08-11 10:23 ` [XEN RFC PATCH 16/40] xen/arm: Create a fake NUMA node to use common code Wei Chen
2021-08-26 23:10 ` Stefano Stabellini
2021-08-27 1:15 ` Wei Chen
2021-08-27 6:18 ` Jan Beulich
2021-08-27 9:32 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 17/40] xen/arm: Introduce DEVICE_TREE_NUMA Kconfig for arm64 Wei Chen
2021-08-19 13:38 ` Julien Grall
2021-08-20 2:30 ` Wei Chen
2021-08-20 8:41 ` Julien Grall
2021-08-20 10:49 ` Wei Chen
2021-08-20 11:28 ` Julien Grall
2021-08-20 12:25 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 18/40] xen/arm: Keep memory nodes in dtb for NUMA when boot from EFI Wei Chen
2021-08-19 17:35 ` Julien Grall
2021-08-20 2:18 ` Wei Chen
2021-08-26 23:24 ` Stefano Stabellini
2021-08-27 7:41 ` Julien Grall
2021-08-27 23:10 ` Stefano Stabellini
2021-08-27 9:23 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 19/40] xen: fdt: Introduce a helper to check fdt node type Wei Chen
2021-08-25 13:39 ` Julien Grall
2021-08-26 6:00 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 20/40] xen/arm: implement node distance helpers for Arm64 Wei Chen
2021-08-26 23:52 ` Stefano Stabellini
2021-08-27 9:30 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 21/40] xen/arm: introduce device_tree_numa as a switch for device tree NUMA Wei Chen
2021-08-19 17:45 ` Julien Grall
2021-08-20 2:21 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 22/40] xen/arm: introduce a helper to parse device tree processor node Wei Chen
2021-08-19 18:09 ` Julien Grall
2021-08-23 8:42 ` Wei Chen
2021-08-19 18:10 ` Julien Grall
2021-08-23 8:47 ` Wei Chen
2021-08-23 10:59 ` Julien Grall
2021-08-24 4:09 ` Wei Chen
2021-08-19 18:13 ` Julien Grall
2021-08-20 2:23 ` Wei Chen
2021-08-20 8:44 ` Julien Grall
2021-08-20 11:53 ` Wei Chen
2021-08-27 0:06 ` Stefano Stabellini
2021-08-27 9:31 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 23/40] xen/arm: introduce a helper to parse device tree memory node Wei Chen
2021-08-25 13:48 ` Julien Grall
2021-08-26 6:35 ` Wei Chen
2021-08-26 8:21 ` Julien Grall
2021-08-26 11:54 ` Wei Chen
2021-08-28 1:06 ` Stefano Stabellini
2021-08-28 3:56 ` Wei Chen
2021-08-28 10:33 ` Julien Grall
2021-08-28 13:58 ` Wei Chen
2021-09-08 7:34 ` Wei Chen
2021-09-08 22:31 ` Stefano Stabellini
2021-09-09 3:54 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 24/40] xen/arm: introduce a helper to parse device tree NUMA distance map Wei Chen
2021-08-25 13:56 ` Julien Grall
2021-08-26 7:01 ` Wei Chen
2021-08-31 0:48 ` Stefano Stabellini
2021-08-31 10:17 ` Wei Chen
2021-08-31 21:36 ` Stefano Stabellini
2021-09-01 11:04 ` Wei Chen
2021-09-01 16:21 ` Stefano Stabellini
2021-09-02 2:30 ` Wei Chen
2021-09-02 15:19 ` Stefano Stabellini
2021-09-02 6:00 ` Jan Beulich
2021-09-02 14:14 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 25/40] xen/arm: unified entry to parse all NUMA data from device tree Wei Chen
2021-08-31 0:54 ` Stefano Stabellini
2021-08-31 17:47 ` Julien Grall
2021-09-01 18:30 ` Stefano Stabellini
2021-09-02 2:48 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 26/40] xen/arm: Add boot and secondary CPU to NUMA system Wei Chen
2021-08-25 16:58 ` Julien Grall
2021-08-26 7:24 ` Wei Chen
2021-08-26 8:49 ` Julien Grall
2021-08-26 9:39 ` Jan Beulich
2021-08-26 12:08 ` Wei Chen
2021-08-26 12:26 ` Jan Beulich
2021-08-11 10:24 ` [XEN RFC PATCH 27/40] xen/arm: build CPU NUMA node map while creating cpu_logical_map Wei Chen
2021-08-25 17:06 ` Julien Grall
2021-08-26 7:26 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 28/40] xen/x86: decouple nodes_cover_memory with E820 map Wei Chen
2021-08-31 1:07 ` Stefano Stabellini
2021-08-31 10:19 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 29/40] xen/arm: implement Arm arch helpers Arm to get memory map info Wei Chen
2021-08-25 17:09 ` Julien Grall
2021-08-26 7:27 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 30/40] xen: move NUMA memory and CPU parsed nodemasks to common Wei Chen
2021-08-25 17:16 ` Julien Grall
2021-08-26 7:29 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 31/40] xen/x86: move nodes_cover_memory " Wei Chen
2021-08-31 1:16 ` Stefano Stabellini
2021-08-31 13:43 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 32/40] xen/x86: make acpi_scan_nodes to be neutral Wei Chen
2021-08-27 14:08 ` Julien Grall
2021-08-28 2:11 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 33/40] xen: export bad_srat and srat_disabled to extern Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 34/40] xen: move numa_scan_nodes from x86 to common Wei Chen
2021-08-27 14:14 ` Julien Grall
2021-08-28 2:12 ` Wei Chen
2021-08-31 1:26 ` Stefano Stabellini
2021-08-31 13:43 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 35/40] xen: enable numa_scan_nodes for device tree based NUMA Wei Chen
2021-08-27 14:19 ` Julien Grall
2021-08-28 2:13 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 36/40] xen/arm: keep guest still be NUMA unware Wei Chen
2021-08-27 14:28 ` Julien Grall
2021-08-28 2:19 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 37/40] xen: introduce an arch helper to do NUMA init failed fallback Wei Chen
2021-08-27 14:30 ` Julien Grall
2021-08-28 3:09 ` Wei Chen
2021-08-28 3:45 ` Wei Chen
2021-08-30 9:52 ` Jan Beulich
2021-08-30 10:38 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 38/40] xen/arm: enable device tree based NUMA in system init Wei Chen
2021-08-27 14:32 ` Julien Grall
2021-08-28 3:17 ` Wei Chen
2021-08-28 10:45 ` Julien Grall
2021-08-28 14:02 ` Wei Chen
2021-08-31 1:50 ` Stefano Stabellini
2021-08-31 13:43 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 39/40] xen/x86: move numa_setup to common to support NUMA switch in command line Wei Chen
2021-08-27 14:37 ` Julien Grall
2021-08-28 3:22 ` Wei Chen
2021-08-31 1:53 ` Stefano Stabellini
2021-08-31 13:44 ` Wei Chen
2021-08-11 10:24 ` [XEN RFC PATCH 40/40] xen/x86: move dump_numa info hotkey to common Wei Chen
2021-08-11 10:41 ` [XEN RFC PATCH 00/40] Add device tree based NUMA support to Arm64 Jan Beulich
2021-08-13 2:33 ` Wei Chen
2021-08-13 6:53 ` Jan Beulich
2021-08-19 13:42 ` Julien Grall
2021-08-19 14:05 ` Bertrand Marquis
2021-08-19 17:11 ` Julien Grall
2021-08-26 0:09 ` Stefano Stabellini
2021-08-26 7:31 ` Wei Chen
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=DB9PR08MB6857E228F4C0847D4DF014ED9EC69@DB9PR08MB6857.eurprd08.prod.outlook.com \
--to=wei.chen@arm.com \
--cc=Bertrand.Marquis@arm.com \
--cc=julien@xen.org \
--cc=sstabellini@kernel.org \
--cc=xen-devel@lists.xenproject.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).