* + mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug.patch added to -mm tree
@ 2015-08-17 21:36 akpm
0 siblings, 0 replies; only message in thread
From: akpm @ 2015-08-17 21:36 UTC (permalink / raw)
To: jiang.liu, hpa, mgorman, mingo, peterz, rafael.j.wysocki,
rientjes, tangchen, tglx, tj, tony.luck, umgwanakikbuti,
mm-commits
The patch titled
Subject: mm, x86: enable memoryless node support to better support CPU/memory hotplug
has been added to the -mm tree. Its filename is
mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug.patch
This patch should soon appear at
http://ozlabs.org/~akpm/mmots/broken-out/mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug.patch
and later at
http://ozlabs.org/~akpm/mmotm/broken-out/mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug.patch
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/SubmitChecklist when testing your code ***
The -mm tree is included into linux-next and is updated
there every 3-4 working days
------------------------------------------------------
From: Jiang Liu <jiang.liu@linux.intel.com>
Subject: mm, x86: enable memoryless node support to better support CPU/memory hotplug
With current implementation, all CPUs within a NUMA node will be
assocaited with another NUMA node if the node has no memory installed.
For example, on a four-node system, CPUs on node 2 and 3 are associated
with node 0 when are no memory install on node 2 and 3, which may
confuse users.
root@bkd01sdp:~# numactl --hardware
available: 2 nodes (0-1)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
node 0 size: 15602 MB
node 0 free: 15014 MB
node 1 cpus: 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89
node 1 size: 15985 MB
node 1 free: 15686 MB
node distances:
node 0 1
0: 10 21
1: 21 10
To be worse, the CPU affinity relationship won't get fixed even after
memory has been added to those nodes. After memory hot-addition to
node 2, CPUs on node 2 are still associated with node 0. This may cause
sub-optimal performance.
root@bkd01sdp:/sys/devices/system/node/node2# numactl --hardware
available: 3 nodes (0-2)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
node 0 size: 15602 MB
node 0 free: 14743 MB
node 1 cpus: 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89
node 1 size: 15985 MB
node 1 free: 15715 MB
node 2 cpus:
node 2 size: 128 MB
node 2 free: 128 MB
node distances:
node 0 1 2
0: 10 21 21
1: 21 10 21
2: 21 21 10
With support of memoryless node enabled, it will correctly report system
hardware topology for nodes without memory installed.
root@bkd01sdp:~# numactl --hardware
available: 4 nodes (0-3)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74
node 0 size: 15725 MB
node 0 free: 15129 MB
node 1 cpus: 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89
node 1 size: 15862 MB
node 1 free: 15627 MB
node 2 cpus: 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104
node 2 size: 0 MB
node 2 free: 0 MB
node 3 cpus: 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
node 3 size: 0 MB
node 3 free: 0 MB
node distances:
node 0 1 2 3
0: 10 21 21 21
1: 21 10 21 21
2: 21 21 10 21
3: 21 21 21 10
With memoryless node enabled, CPUs are correctly associated with node 2
after memory hot-addition to node 2.
root@bkd01sdp:/sys/devices/system/node/node2# numactl --hardware
available: 4 nodes (0-3)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74
node 0 size: 15725 MB
node 0 free: 14872 MB
node 1 cpus: 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89
node 1 size: 15862 MB
node 1 free: 15641 MB
node 2 cpus: 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104
node 2 size: 128 MB
node 2 free: 127 MB
node 3 cpus: 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
node 3 size: 0 MB
node 3 free: 0 MB
node distances:
node 0 1 2 3
0: 10 21 21 21
1: 21 10 21 21
2: 21 21 10 21
3: 21 21 21 10
Signed-off-by: Jiang Liu <jiang.liu@linux.intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Mike Galbraith <umgwanakikbuti@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: "Rafael J . Wysocki" <rafael.j.wysocki@intel.com>
Cc: Tang Chen <tangchen@cn.fujitsu.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---
arch/x86/Kconfig | 3 ++
arch/x86/kernel/acpi/boot.c | 4 ++
arch/x86/kernel/smpboot.c | 2 +
arch/x86/mm/numa.c | 49 ++++++++++++++++++++++++----------
4 files changed, 44 insertions(+), 14 deletions(-)
diff -puN arch/x86/Kconfig~mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug arch/x86/Kconfig
--- a/arch/x86/Kconfig~mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug
+++ a/arch/x86/Kconfig
@@ -2110,6 +2110,9 @@ config USE_PERCPU_NUMA_NODE_ID
def_bool y
depends on NUMA
+config HAVE_MEMORYLESS_NODES
+ def_bool NUMA
+
config ARCH_ENABLE_SPLIT_PMD_PTLOCK
def_bool y
depends on X86_64 || X86_PAE
diff -puN arch/x86/kernel/acpi/boot.c~mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug arch/x86/kernel/acpi/boot.c
--- a/arch/x86/kernel/acpi/boot.c~mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug
+++ a/arch/x86/kernel/acpi/boot.c
@@ -711,6 +711,7 @@ static void acpi_map_cpu2node(acpi_handl
}
set_apicid_to_node(physid, nid);
numa_set_node(cpu, nid);
+ set_cpu_numa_mem(cpu, local_memory_node(nid));
}
#endif
}
@@ -737,9 +738,10 @@ int acpi_unmap_cpu(int cpu)
{
#ifdef CONFIG_ACPI_NUMA
set_apicid_to_node(per_cpu(x86_cpu_to_apicid, cpu), NUMA_NO_NODE);
+ set_cpu_numa_mem(cpu, NUMA_NO_NODE);
#endif
- per_cpu(x86_cpu_to_apicid, cpu) = -1;
+ per_cpu(x86_cpu_to_apicid, cpu) = BAD_APICID;
set_cpu_present(cpu, false);
num_processors--;
diff -puN arch/x86/kernel/smpboot.c~mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug arch/x86/kernel/smpboot.c
--- a/arch/x86/kernel/smpboot.c~mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug
+++ a/arch/x86/kernel/smpboot.c
@@ -162,6 +162,8 @@ static void smp_callin(void)
*/
phys_id = read_apic_id();
+ set_numa_mem(local_memory_node(cpu_to_node(cpuid)));
+
/*
* the boot CPU has finished the init stage and is spinning
* on callin_map until we finish. We are free to set up this
diff -puN arch/x86/mm/numa.c~mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug arch/x86/mm/numa.c
--- a/arch/x86/mm/numa.c~mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug
+++ a/arch/x86/mm/numa.c
@@ -22,6 +22,7 @@
int __initdata numa_off;
nodemask_t numa_nodes_parsed __initdata;
+static nodemask_t numa_nodes_empty __initdata;
struct pglist_data *node_data[MAX_NUMNODES] __read_mostly;
EXPORT_SYMBOL(node_data);
@@ -562,17 +563,16 @@ static int __init numa_register_memblks(
end = max(mi->blk[i].end, end);
}
- if (start >= end)
- continue;
-
/*
* Don't confuse VM with a node that doesn't have the
* minimum amount of memory:
*/
- if (end && (end - start) < NODE_MIN_SIZE)
- continue;
-
- alloc_node_data(nid);
+ if (start < end && (end - start) >= NODE_MIN_SIZE) {
+ alloc_node_data(nid);
+ } else if (IS_ENABLED(CONFIG_HAVE_MEMORYLESS_NODES)) {
+ alloc_node_data(nid);
+ node_set(nid, numa_nodes_empty);
+ }
}
/* Dump memblock with node info and return. */
@@ -589,14 +589,18 @@ static int __init numa_register_memblks(
*/
static void __init numa_init_array(void)
{
- int rr, i;
+ int i, rr = MAX_NUMNODES;
- rr = first_node(node_online_map);
for (i = 0; i < nr_cpu_ids; i++) {
+ /* Search for an onlined node with memory */
+ do {
+ if (rr != MAX_NUMNODES)
+ rr = next_node(rr, node_online_map);
+ if (rr == MAX_NUMNODES)
+ rr = first_node(node_online_map);
+ } while (node_isset(rr, numa_nodes_empty));
+
numa_set_node(i, rr);
- rr = next_node(rr, node_online_map);
- if (rr == MAX_NUMNODES)
- rr = first_node(node_online_map);
}
}
@@ -698,9 +702,12 @@ static __init int find_near_online_node(
{
int n, val;
int min_val = INT_MAX;
- int best_node = -1;
+ int best_node = NUMA_NO_NODE;
for_each_online_node(n) {
+ if (node_isset(n, numa_nodes_empty))
+ continue;
+
val = node_distance(node, n);
if (val < min_val) {
@@ -741,6 +748,22 @@ void __init init_cpu_to_node(void)
if (!node_online(node))
node = find_near_online_node(node);
numa_set_node(cpu, node);
+ if (node_spanned_pages(node))
+ set_cpu_numa_mem(cpu, node);
+ if (IS_ENABLED(CONFIG_HAVE_MEMORYLESS_NODES))
+ node_clear(node, numa_nodes_empty);
+ }
+
+ /* Destroy empty nodes */
+ if (IS_ENABLED(CONFIG_HAVE_MEMORYLESS_NODES)) {
+ int nid;
+ const size_t nd_size = roundup(sizeof(pg_data_t), PAGE_SIZE);
+
+ for_each_node_mask(nid, numa_nodes_empty) {
+ node_set_offline(nid);
+ memblock_free(__pa(node_data[nid]), nd_size);
+ node_data[nid] = NULL;
+ }
}
}
_
Patches currently in -mm which might be from jiang.liu@linux.intel.com are
linux-next.patch
x86-numa-acpi-online-node-earlier-when-doing-cpu-hot-addition.patch
kernel-profilec-replace-cpu_to_mem-with-cpu_to_node.patch
sgi-xp-replace-cpu_to_node-with-cpu_to_mem-to-support-memoryless-node.patch
openvswitch-replace-cpu_to_node-with-cpu_to_mem-to-support-memoryless-node.patch
i40e-use-numa_mem_id-to-better-support-memoryless-node.patch
i40evf-use-numa_mem_id-to-better-support-memoryless-node.patch
x86-numa-kill-useless-code-to-improve-code-readability.patch
mm-update-_mem_id_-for-every-possible-cpu-when-memory-configuration-changes.patch
mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug.patch
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2015-08-17 21:36 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-08-17 21:36 + mm-x86-enable-memoryless-node-support-to-better-support-cpu-memory-hotplug.patch added to -mm tree akpm
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.