Re: [RFC][PATCH 2/6] x86/topo: Add TOPO_NUMA_DOMAIN

From: Peter Zijlstra

Date: Fri Feb 27 2026 - 09:14:51 EST


On Fri, Feb 27, 2026 at 06:49:36PM +0530, K Prateek Nayak wrote:
> Hello Peter,
>
> On 2/26/2026 4:19 PM, Peter Zijlstra wrote:
> > @@ -88,6 +89,14 @@ static inline u32 topo_apicid(u32 apicid
> > {
> > if (dom == TOPO_SMT_DOMAIN)
> > return apicid;
> > +#ifdef CONFIG_NUMA
> > + if (dom == TOPO_NUMA_DOMAIN) {
> > + int nid = __apicid_to_phys_node[apicid];
> > + if (nid == NUMA_NO_NODE)
> > + nid = 0;
> > + return nid;
> > + }
> > +#endif
>
> I'm not digging this override - simply because topo_apicid() was not
> meant to handle these kinds of cases where we cannot derive a topology
> ID by simply shifting and masking the APICID.
>
> Looking at the series, all we need is an equivalent of:
>
> domain_weight(TOPO_NUMA_DOMAIN)

Fair enough; but then lets replace patch 1 and 2 with something like
that.

But I must note that the nodemask API is crap; it has both node_set() and
__node_set() be the atomic version :-(

Let me go rework the other patches to fit on this.

---
diff --git a/arch/x86/include/asm/numa.h b/arch/x86/include/asm/numa.h
index 53ba39ce010c..a9063f332fa6 100644
--- a/arch/x86/include/asm/numa.h
+++ b/arch/x86/include/asm/numa.h
@@ -22,6 +22,7 @@ extern int numa_off;
*/
extern s16 __apicid_to_node[MAX_LOCAL_APIC];
extern nodemask_t numa_nodes_parsed __initdata;
+extern nodemask_t numa_phys_nodes_parsed __initdata;

static inline void set_apicid_to_node(int apicid, s16 node)
{
@@ -48,6 +49,7 @@ extern void __init init_cpu_to_node(void);
extern void numa_add_cpu(unsigned int cpu);
extern void numa_remove_cpu(unsigned int cpu);
extern void init_gi_nodes(void);
+extern int num_phys_nodes(void);
#else /* CONFIG_NUMA */
static inline void numa_set_node(int cpu, int node) { }
static inline void numa_clear_node(int cpu) { }
@@ -55,6 +57,10 @@ static inline void init_cpu_to_node(void) { }
static inline void numa_add_cpu(unsigned int cpu) { }
static inline void numa_remove_cpu(unsigned int cpu) { }
static inline void init_gi_nodes(void) { }
+static inline int num_phys_nodes(void)
+{
+ return 1;
+}
#endif /* CONFIG_NUMA */

#ifdef CONFIG_DEBUG_PER_CPU_MAPS
diff --git a/arch/x86/kernel/cpu/topology.c b/arch/x86/kernel/cpu/topology.c
index 23190a786d31..bfcd33127789 100644
--- a/arch/x86/kernel/cpu/topology.c
+++ b/arch/x86/kernel/cpu/topology.c
@@ -31,6 +31,7 @@
#include <asm/mpspec.h>
#include <asm/msr.h>
#include <asm/smp.h>
+#include <asm/numa.h>

#include "cpu.h"

diff --git a/arch/x86/mm/numa.c b/arch/x86/mm/numa.c
index 7a97327140df..99d0a9332c14 100644
--- a/arch/x86/mm/numa.c
+++ b/arch/x86/mm/numa.c
@@ -48,6 +48,8 @@ s16 __apicid_to_node[MAX_LOCAL_APIC] = {
[0 ... MAX_LOCAL_APIC-1] = NUMA_NO_NODE
};

+nodemask_t numa_phys_nodes_parsed __initdata;
+
int numa_cpu_node(int cpu)
{
u32 apicid = early_per_cpu(x86_cpu_to_apicid, cpu);
@@ -57,6 +59,11 @@ int numa_cpu_node(int cpu)
return NUMA_NO_NODE;
}

+int __init num_phys_nodes(void)
+{
+ return bitmap_weight(numa_phys_nodes_parsed.bits, MAX_NUMNODES);
+}
+
cpumask_var_t node_to_cpumask_map[MAX_NUMNODES];
EXPORT_SYMBOL(node_to_cpumask_map);

@@ -210,6 +217,7 @@ static int __init dummy_numa_init(void)
0LLU, PFN_PHYS(max_pfn) - 1);

node_set(0, numa_nodes_parsed);
+ node_set(0, numa_phys_nodes_parsed);
numa_add_memblk(0, 0, PFN_PHYS(max_pfn));

return 0;
diff --git a/arch/x86/mm/srat.c b/arch/x86/mm/srat.c
index 6f8e0f21c710..44ca66651756 100644
--- a/arch/x86/mm/srat.c
+++ b/arch/x86/mm/srat.c
@@ -57,6 +57,7 @@ acpi_numa_x2apic_affinity_init(struct acpi_srat_x2apic_cpu_affinity *pa)
}
set_apicid_to_node(apic_id, node);
node_set(node, numa_nodes_parsed);
+ node_set(node, numa_phys_nodes_parsed);
pr_debug("SRAT: PXM %u -> APIC 0x%04x -> Node %u\n", pxm, apic_id, node);
}

@@ -97,6 +98,7 @@ acpi_numa_processor_affinity_init(struct acpi_srat_cpu_affinity *pa)

set_apicid_to_node(apic_id, node);
node_set(node, numa_nodes_parsed);
+ node_set(node, numa_phys_nodes_parsed);
pr_debug("SRAT: PXM %u -> APIC 0x%02x -> Node %u\n", pxm, apic_id, node);
}