Patchwork [RFC,06/15] pc: set FW_CFG data based on APIC ID calculation

login
register
mail settings
Submitter Eduardo Habkost
Date Aug. 7, 2012, 7:56 p.m.
Message ID <1344369413-9053-7-git-send-email-ehabkost@redhat.com>
Download mbox | patch
Permalink /patch/175793/
State New
Headers show

Comments

Eduardo Habkost - Aug. 7, 2012, 7:56 p.m.
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 <ehabkost@redhat.com>
---
 hw/pc.c           | 23 ++++++++++++++++-------
 target-i386/cpu.h |  7 +++++++
 2 files changed, 23 insertions(+), 7 deletions(-)
Igor Mammedov - Aug. 13, 2012, 7:52 p.m.
On 08/07/2012 09:56 PM, Eduardo Habkost wrote:
> 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 <ehabkost@redhat.com>
> ---
>   hw/pc.c           | 23 ++++++++++++++++-------
>   target-i386/cpu.h |  7 +++++++
>   2 files changed, 23 insertions(+), 7 deletions(-)
>
> diff --git a/hw/pc.c b/hw/pc.c
> index 10449bd..9afb838 100644
> --- a/hw/pc.c
> +++ b/hw/pc.c
> @@ -581,6 +581,11 @@ int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
>       return index;
>   }
>
> +unsigned int apic_id_limit(void)
> +{
> +    return apic_id_for_cpu(max_cpus - 1) + 1;
> +}
> +
>   static void *bochs_bios_init(void)
>   {
>       void *fw_cfg;
> @@ -588,6 +593,7 @@ static void *bochs_bios_init(void)
>       size_t smbios_len;
>       uint64_t *numa_fw_cfg;
>       int i, j;
> +    unsigned int max_apic_id = apic_id_limit();
>
>       register_ioport_write(0x400, 1, 2, bochs_bios_write, NULL);
>       register_ioport_write(0x401, 1, 2, bochs_bios_write, NULL);
> @@ -602,7 +608,7 @@ static void *bochs_bios_init(void)
>       register_ioport_write(0x503, 1, 1, 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_add_i16(fw_cfg, FW_CFG_MAX_CPUS, (uint16_t)max_apic_id);
FW_CFG_MAX_CPUS becoming not MAX_CPUS sounds a bit confusing, perhaps 
short comment should be here to document this and why it's not? So code 
reader won't make false assumptions?

>       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,
> @@ -622,21 +628,24 @@ static void *bochs_bios_init(void)
>        * 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(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;
>   }
> diff --git a/target-i386/cpu.h b/target-i386/cpu.h
> index 39ea005..257d6c7 100644
> --- a/target-i386/cpu.h
> +++ b/target-i386/cpu.h
> @@ -919,6 +919,13 @@ void host_cpuid(uint32_t function, uint32_t count,
>    */
>   unsigned int apic_id_for_cpu(int cpu_index);
>
> +/* Calculate limit for the APIC ID value, based on max_cpus
> + *
> + * On PC, FW_CFG_MAX_CPUS is not max_cpus, but the limit for the APIC IDs
> + * of all CPUs (so that of all CPUs APIC ID < MAX_CPUS).
> + */
> +unsigned int apic_id_limit(void);
> +
>
>   /* helper.c */
>   int cpu_x86_handle_mmu_fault(CPUX86State *env, target_ulong addr,
>
Eduardo Habkost - Aug. 13, 2012, 8:01 p.m.
On Mon, Aug 13, 2012 at 09:52:43PM +0200, Igor Mammedov wrote:
> On 08/07/2012 09:56 PM, Eduardo Habkost wrote:
> >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 <ehabkost@redhat.com>
> >---
> >  hw/pc.c           | 23 ++++++++++++++++-------
> >  target-i386/cpu.h |  7 +++++++
> >  2 files changed, 23 insertions(+), 7 deletions(-)
> >
> >diff --git a/hw/pc.c b/hw/pc.c
> >index 10449bd..9afb838 100644
> >--- a/hw/pc.c
> >+++ b/hw/pc.c
> >@@ -581,6 +581,11 @@ int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
> >      return index;
> >  }
> >
> >+unsigned int apic_id_limit(void)
> >+{
> >+    return apic_id_for_cpu(max_cpus - 1) + 1;
> >+}
> >+
> >  static void *bochs_bios_init(void)
> >  {
> >      void *fw_cfg;
> >@@ -588,6 +593,7 @@ static void *bochs_bios_init(void)
> >      size_t smbios_len;
> >      uint64_t *numa_fw_cfg;
> >      int i, j;
> >+    unsigned int max_apic_id = apic_id_limit();
> >
> >      register_ioport_write(0x400, 1, 2, bochs_bios_write, NULL);
> >      register_ioport_write(0x401, 1, 2, bochs_bios_write, NULL);
> >@@ -602,7 +608,7 @@ static void *bochs_bios_init(void)
> >      register_ioport_write(0x503, 1, 1, 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_add_i16(fw_cfg, FW_CFG_MAX_CPUS, (uint16_t)max_apic_id);
> FW_CFG_MAX_CPUS becoming not MAX_CPUS sounds a bit confusing, perhaps
> short comment should be here to document this and why it's not? So
> code reader won't make false assumptions?

True. I tried to quickly explain the problem in the commit message, but
it sounds confusing enough to deserve a comment in the code.

(It would be even better if we had an official place to document _all_
the fw_cfg fields).

> 
> >      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,
> >@@ -622,21 +628,24 @@ static void *bochs_bios_init(void)
> >       * 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(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;
> >  }
> >diff --git a/target-i386/cpu.h b/target-i386/cpu.h
> >index 39ea005..257d6c7 100644
> >--- a/target-i386/cpu.h
> >+++ b/target-i386/cpu.h
> >@@ -919,6 +919,13 @@ void host_cpuid(uint32_t function, uint32_t count,
> >   */
> >  unsigned int apic_id_for_cpu(int cpu_index);
> >
> >+/* Calculate limit for the APIC ID value, based on max_cpus
> >+ *
> >+ * On PC, FW_CFG_MAX_CPUS is not max_cpus, but the limit for the APIC IDs
> >+ * of all CPUs (so that of all CPUs APIC ID < MAX_CPUS).
> >+ */
> >+unsigned int apic_id_limit(void);
> >+
> >
> >  /* helper.c */
> >  int cpu_x86_handle_mmu_fault(CPUX86State *env, target_ulong addr,
> >
> 
> 
> -- 
> Regards,
>   Igor

Patch

diff --git a/hw/pc.c b/hw/pc.c
index 10449bd..9afb838 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -581,6 +581,11 @@  int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
     return index;
 }
 
+unsigned int apic_id_limit(void)
+{
+    return apic_id_for_cpu(max_cpus - 1) + 1;
+}
+
 static void *bochs_bios_init(void)
 {
     void *fw_cfg;
@@ -588,6 +593,7 @@  static void *bochs_bios_init(void)
     size_t smbios_len;
     uint64_t *numa_fw_cfg;
     int i, j;
+    unsigned int max_apic_id = apic_id_limit();
 
     register_ioport_write(0x400, 1, 2, bochs_bios_write, NULL);
     register_ioport_write(0x401, 1, 2, bochs_bios_write, NULL);
@@ -602,7 +608,7 @@  static void *bochs_bios_init(void)
     register_ioport_write(0x503, 1, 1, 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_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,
@@ -622,21 +628,24 @@  static void *bochs_bios_init(void)
      * 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(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;
 }
diff --git a/target-i386/cpu.h b/target-i386/cpu.h
index 39ea005..257d6c7 100644
--- a/target-i386/cpu.h
+++ b/target-i386/cpu.h
@@ -919,6 +919,13 @@  void host_cpuid(uint32_t function, uint32_t count,
  */
 unsigned int apic_id_for_cpu(int cpu_index);
 
+/* Calculate limit for the APIC ID value, based on max_cpus
+ *
+ * On PC, FW_CFG_MAX_CPUS is not max_cpus, but the limit for the APIC IDs
+ * of all CPUs (so that of all CPUs APIC ID < MAX_CPUS).
+ */
+unsigned int apic_id_limit(void);
+
 
 /* helper.c */
 int cpu_x86_handle_mmu_fault(CPUX86State *env, target_ulong addr,