This changes FW_CFG_MAX_CPUS and FW_CFG_NUMA to use apic_id_for_cpu(),
so the NUMA table can be based on the APIC IDs, instead of CPU index
(SeaBIOS knows nothing about CPU indexes, just APIC IDs).
Signed-off-by: Eduardo Habkost <ehabk...@redhat.com>
---
Changes v1 -> v2:
- Get PC object as argument
- Add more detailed comments explaining the reason for FW_CFG_MAX_CPUS
not being simply 'max_cpus'
Changes v2 -> v3:
- Use PCInitArgs instead of PC object
---
hw/pc.c | 41 ++++++++++++++++++++++++++++++++++-------
1 file changed, 34 insertions(+), 7 deletions(-)
diff --git a/hw/pc.c b/hw/pc.c
index eb68851..8895087 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -585,6 +585,15 @@ int e820_add_entry(uint64_t address, uint64_t length,
uint32_t type)
return index;
}
+/* Returns the limit to APIC ID values
+ *
+ * This is used for FW_CFG_MAX_CPUS. See comments on bochs_bios_init().
+ */
+static unsigned int apic_id_limit(PCInitArgs *args)
+{
+ return apic_id_for_cpu(args, max_cpus - 1) + 1;
+}
+
static FWCfgState *pc_bios_init(PCInitArgs *args)
{
FWCfgState *fw_cfg;
@@ -592,6 +601,7 @@ static FWCfgState *pc_bios_init(PCInitArgs *args)
size_t smbios_len;
uint64_t *numa_fw_cfg;
int i, j;
+ unsigned int max_apic_id = apic_id_limit(args);
register_ioport_write(0x8900, 1, 1, bochs_bios_write, NULL);
@@ -600,7 +610,21 @@ static FWCfgState *pc_bios_init(PCInitArgs *args)
register_ioport_write(0x502, 1, 2, bochs_bios_write, NULL);
fw_cfg = fw_cfg_init(BIOS_CFG_IOPORT, BIOS_CFG_IOPORT + 1, 0, 0);
- fw_cfg_add_i16(fw_cfg, FW_CFG_MAX_CPUS, (uint16_t)max_cpus);
+ /* FW_CFG_MAX_CPUS is a bit confusing/problematic on x86:
+ *
+ * SeaBIOS needs FW_CFG_MAX_CPUS for CPU hotplug, but the CPU hotplug
+ * QEMU<->SeaBIOS interface is not based on the "CPU index", but on the
APIC
+ * ID of hotplugged CPUs[1]. This means that FW_CFG_MAX_CPUS is not the
+ * "maximum number of CPUs", but the "limit to the APIC ID values SeaBIOS
+ * may see".
+ *
+ * So, this means we must not use max_cpus, here, but the maximum possible
+ * APIC ID value, plus one.
+ *
+ * [1] The only kind of "CPU identifier" used between SeaBIOS and QEMU is
+ * the APIC ID, not the "CPU index"
+ */
+ fw_cfg_add_i16(fw_cfg, FW_CFG_MAX_CPUS, (uint16_t)max_apic_id);
fw_cfg_add_i32(fw_cfg, FW_CFG_ID, 1);
fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, (uint64_t)ram_size);
fw_cfg_add_bytes(fw_cfg, FW_CFG_ACPI_TABLES, (uint8_t *)acpi_tables,
@@ -620,21 +644,24 @@ static FWCfgState *pc_bios_init(PCInitArgs *args)
* of nodes, one word for each VCPU->node and one word for each node to
* hold the amount of memory.
*/
- numa_fw_cfg = g_malloc0((1 + max_cpus + nb_numa_nodes) * 8);
+ numa_fw_cfg = g_malloc0((1 + max_apic_id + nb_numa_nodes) * 8);
numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
- for (i = 0; i < max_cpus; i++) {
+ unsigned int cpu_idx;
+ for (cpu_idx = 0; cpu_idx < max_cpus; cpu_idx++) {
+ unsigned int apic_id = apic_id_for_cpu(args, cpu_idx);
+ assert(apic_id < max_apic_id);
for (j = 0; j < nb_numa_nodes; j++) {
- if (test_bit(i, node_cpumask[j])) {
- numa_fw_cfg[i + 1] = cpu_to_le64(j);
+ if (test_bit(cpu_idx, node_cpumask[j])) {
+ numa_fw_cfg[apic_id + 1] = cpu_to_le64(j);
break;
}
}
}
for (i = 0; i < nb_numa_nodes; i++) {
- numa_fw_cfg[max_cpus + 1 + i] = cpu_to_le64(node_mem[i]);
+ numa_fw_cfg[max_apic_id + 1 + i] = cpu_to_le64(node_mem[i]);
}
fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, (uint8_t *)numa_fw_cfg,
- (1 + max_cpus + nb_numa_nodes) * 8);
+ (1 + max_apic_id + nb_numa_nodes) * 8);
return fw_cfg;
}
--
1.7.11.7
