From: Thomas Gleixner <[email protected]>
XEN/PV has a completely broken vCPU enumeration scheme, which just works by
chance and provides zero topology information. Each vCPU ends up being a
single core package.
Dom0 provides MADT which can be used for topology information, but that
table is the unmodified host table, which means that there can be more CPUs
registered than the number of vCPUs XEN provides for the dom0 guest.
DomU does not have ACPI and both rely on counting the possible vCPUs via an
hypercall.
To prepare for using CPUID topology information either via MADT or via fake
APIC IDs count the number of possible CPUs during early boot and adjust
nr_cpu_ids() accordingly.
Signed-off-by: Thomas Gleixner <[email protected]>
---
arch/x86/xen/enlighten_pv.c | 3 +++
arch/x86/xen/smp.h | 2 ++
arch/x86/xen/smp_pv.c | 14 ++++++++++++++
3 files changed, 19 insertions(+)
---
--- a/arch/x86/xen/enlighten_pv.c
+++ b/arch/x86/xen/enlighten_pv.c
@@ -200,6 +200,9 @@ static void __init xen_pv_init_platform(
xen_set_mtrr_data();
else
mtrr_overwrite_state(NULL, 0, MTRR_TYPE_WRBACK);
+
+ /* Adjust nr_cpu_ids before "enumeration" happens */
+ xen_smp_count_cpus();
}
static void __init xen_pv_guest_late_init(void)
--- a/arch/x86/xen/smp.h
+++ b/arch/x86/xen/smp.h
@@ -19,6 +19,7 @@ extern void xen_smp_intr_free(unsigned i
int xen_smp_intr_init_pv(unsigned int cpu);
void xen_smp_intr_free_pv(unsigned int cpu);
+void xen_smp_count_cpus(void);
void xen_smp_cpus_done(unsigned int max_cpus);
void xen_smp_send_reschedule(int cpu);
@@ -44,6 +45,7 @@ static inline int xen_smp_intr_init_pv(u
return 0;
}
static inline void xen_smp_intr_free_pv(unsigned int cpu) {}
+static inline void xen_smp_count_cpus(void) { }
#endif /* CONFIG_SMP */
#endif
--- a/arch/x86/xen/smp_pv.c
+++ b/arch/x86/xen/smp_pv.c
@@ -411,6 +411,20 @@ static irqreturn_t xen_irq_work_interrup
return IRQ_HANDLED;
}
+void __init xen_smp_count_cpus(void)
+{
+ unsigned int cpus;
+
+ for (cpus = 0; cpus < nr_cpu_ids; cpus++) {
+ if (HYPERVISOR_vcpu_op(VCPUOP_is_up, cpus, NULL) < 0)
+ break;
+ }
+
+ pr_info("Xen PV: Detected %u vCPUS\n", cpus);
+ if (cpus < nr_cpu_ids)
+ set_nr_cpu_ids(cpus);
+}
+
static const struct smp_ops xen_smp_ops __initconst = {
.smp_prepare_boot_cpu = xen_pv_smp_prepare_boot_cpu,
.smp_prepare_cpus = xen_pv_smp_prepare_cpus,
The following commit has been merged into the x86/apic branch of tip:
Commit-ID: c8f808231f1fb63553f90d4b3796cb6804d1e693
Gitweb: https://git.kernel.org/tip/c8f808231f1fb63553f90d4b3796cb6804d1e693
Author: Thomas Gleixner <[email protected]>
AuthorDate: Tue, 13 Feb 2024 22:06:00 +01:00
Committer: Thomas Gleixner <[email protected]>
CommitterDate: Thu, 15 Feb 2024 22:07:44 +01:00
x86/xen/smp_pv: Count number of vCPUs early
XEN/PV has a completely broken vCPU enumeration scheme, which just works by
chance and provides zero topology information. Each vCPU ends up being a
single core package.
Dom0 provides MADT which can be used for topology information, but that
table is the unmodified host table, which means that there can be more CPUs
registered than the number of vCPUs XEN provides for the dom0 guest.
DomU does not have ACPI and both rely on counting the possible vCPUs via an
hypercall.
To prepare for using CPUID topology information either via MADT or via fake
APIC IDs count the number of possible CPUs during early boot and adjust
nr_cpu_ids() accordingly.
Signed-off-by: Thomas Gleixner <[email protected]>
Tested-by: Michael Kelley <[email protected]>
Tested-by: Sohil Mehta <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
---
arch/x86/xen/enlighten_pv.c | 3 +++
arch/x86/xen/smp.h | 2 ++
arch/x86/xen/smp_pv.c | 14 ++++++++++++++
3 files changed, 19 insertions(+)
diff --git a/arch/x86/xen/enlighten_pv.c b/arch/x86/xen/enlighten_pv.c
index aeb33e0..ace2eb0 100644
--- a/arch/x86/xen/enlighten_pv.c
+++ b/arch/x86/xen/enlighten_pv.c
@@ -200,6 +200,9 @@ static void __init xen_pv_init_platform(void)
xen_set_mtrr_data();
else
mtrr_overwrite_state(NULL, 0, MTRR_TYPE_WRBACK);
+
+ /* Adjust nr_cpu_ids before "enumeration" happens */
+ xen_smp_count_cpus();
}
static void __init xen_pv_guest_late_init(void)
diff --git a/arch/x86/xen/smp.h b/arch/x86/xen/smp.h
index c20cbb1..b8efdbc 100644
--- a/arch/x86/xen/smp.h
+++ b/arch/x86/xen/smp.h
@@ -19,6 +19,7 @@ extern void xen_smp_intr_free(unsigned int cpu);
int xen_smp_intr_init_pv(unsigned int cpu);
void xen_smp_intr_free_pv(unsigned int cpu);
+void xen_smp_count_cpus(void);
void xen_smp_cpus_done(unsigned int max_cpus);
void xen_smp_send_reschedule(int cpu);
@@ -44,6 +45,7 @@ static inline int xen_smp_intr_init_pv(unsigned int cpu)
return 0;
}
static inline void xen_smp_intr_free_pv(unsigned int cpu) {}
+static inline void xen_smp_count_cpus(void) { }
#endif /* CONFIG_SMP */
#endif
diff --git a/arch/x86/xen/smp_pv.c b/arch/x86/xen/smp_pv.c
index 98849b3..44706f0 100644
--- a/arch/x86/xen/smp_pv.c
+++ b/arch/x86/xen/smp_pv.c
@@ -411,6 +411,20 @@ static irqreturn_t xen_irq_work_interrupt(int irq, void *dev_id)
return IRQ_HANDLED;
}
+void __init xen_smp_count_cpus(void)
+{
+ unsigned int cpus;
+
+ for (cpus = 0; cpus < nr_cpu_ids; cpus++) {
+ if (HYPERVISOR_vcpu_op(VCPUOP_is_up, cpus, NULL) < 0)
+ break;
+ }
+
+ pr_info("Xen PV: Detected %u vCPUS\n", cpus);
+ if (cpus < nr_cpu_ids)
+ set_nr_cpu_ids(cpus);
+}
+
static const struct smp_ops xen_smp_ops __initconst = {
.smp_prepare_boot_cpu = xen_pv_smp_prepare_boot_cpu,
.smp_prepare_cpus = xen_pv_smp_prepare_cpus,