Skip to contentSkip to conversation

hw/arm/virt: Fix CPU's default NUMA node ID

PATCH Superseded Gavin Shan 4 participants 13 messages Apr 20, 2022 – Apr 22, 2022

v7 of 9 Superseded Latest is v9 →

lore
Share & tools
lore

Gavin Shan

author
#0 lore raw

v7 cover hw/arm/virt: Fix CPU's default NUMA node ID

When the CPU-to-NUMA association isn't provided by user, the default NUMA node ID for the specific CPU is returned from virt_get_default_cpu_node_id(). Unfortunately, the default NUMA node ID breaks socket boundary and leads to the broken CPU topology warning message in Linux guest. This series intends to fix the issue.

PATCH[1/4] Add cluster-id to CPU instance property
PATCH[2/4] Uses SMP configuration to populate CPU topology PATCH[3/4] Fixes the broken CPU topology by considering the socket boundary
when the default NUMA node ID is given
PATCH[4/4] Uses the populated CPU topology to build PPTT table, instead of
calculate it again

Changelog
=========
v6/v7:

   * Fixed description for 'cluster-id' and 'core-id'           (Yanan)
   * Remove '% ms->smp.sockets' in socket ID calculation        (Yanan)
   * Fixed tests/qtest/numa-test/aarch64_numa_cpu()             (Yanan)
   * Initialized offset variables in build_pptt()               (Jonathan)

* Added comments about the expected and sorted layout of
cpus[n].props.*_id and assert() on the exceptional cases (Igor)
v4/v5:
* Split PATCH[v3 1/3] to PATCH[v5 1/4] and PATCH[v5 2/4].
Verify or dump 'clsuter-id' in various spots (Yanan)
* s/within cluster/within cluster\/die/ for 'core-id' in
qapi/machine.json (Igor)
* Apply '% ms->smp.{sockets, clusters, cores, threads} in
virt_possible_cpu_arch_ids() as x86 does (Igor)
* Use [0 - possible_cpus->len] as ACPI processor UID to
build PPTT table and PATCH[v3 4/4] is dropped (Igor)
* Simplified build_pptt() to add all entries in one loop

     on ms->possible_cpus                                       (Igor)
v3:
   * Split PATCH[v2 1/3] to PATCH[v3 1/4] and PATCH[v3 2/4]     (Yanan)
   * Don't take account of die ID in CPU topology population
     and added assert(!mc->smp_props.dies_supported)            (Yanan/Igor)

* Assign cluster_id and use it when building PPTT table (Yanan/Igor)
v2:
* Populate the CPU topology in virt_possible_cpu_arch_ids() so that it can be reused in virt_get_default_cpu_node_id() (Igor)
* Added PATCH[2/3] to use the existing CPU topology when the
PPTT table is built (Igor)
* Added PATCH[3/3] to take thread ID as ACPI processor ID
in MADT and SRAT table (Gavin)

Gavin Shan (4):
qapi/machine.json: Add cluster-id
hw/arm/virt: Consider SMP configuration in CPU topology hw/arm/virt: Fix CPU's default NUMA node ID hw/acpi/aml-build: Use existing CPU topology to build PPTT table

 hw/acpi/aml-build.c        | 109 ++++++++++++++++---------------------
 hw/arm/virt.c              |  19 ++++++-
 hw/core/machine-hmp-cmds.c |   4 ++
 hw/core/machine.c          |  16 ++++++
 qapi/machine.json          |   6 +-
 tests/qtest/numa-test.c    |   3 +-
 6 files changed, 90 insertions(+), 67 deletions(-)
Signature · 1 line
2.23.0
author
#1 lore raw

PATCH v7 1/4 qapi/machine.json: Add cluster-id

This adds cluster-id in CPU instance properties, which will be used by arm/virt machine. Besides, the cluster-id is also verified or dumped in various spots:

* hw/core/machine.c::machine_set_cpu_numa_node() to associate CPU with its NUMA node.

* hw/core/machine.c::machine_numa_finish_cpu_init() to associate CPU with NUMA node when no default association isn't provided.

* hw/core/machine-hmp-cmds.c::hmp_hotpluggable_cpus() to dump cluster-id.

---
 hw/core/machine-hmp-cmds.c |  4 ++++
 hw/core/machine.c          | 16 ++++++++++++++++
 qapi/machine.json          |  6 ++++--
 3 files changed, 24 insertions(+), 2 deletions(-)
hw/core/machine-hmp-cmds.c +4 -0
index 4e2f319aeb..5cb5eecbfc 100644
@@ -77,6 +77,10 @@ void hmp_hotpluggable_cpus(Monitor *mon, const QDict *qdict)
         if (c->has_die_id) {
             monitor_printf(mon, "    die-id: \"%" PRIu64 "\"\n", c->die_id);
         }
+        if (c->has_cluster_id) {
+            monitor_printf(mon, "    cluster-id: \"%" PRIu64 "\"\n",
+                           c->cluster_id);
+        }
         if (c->has_core_id) {
             monitor_printf(mon, "    core-id: \"%" PRIu64 "\"\n", c->core_id);
         }
hw/core/machine.c +16 -0
index 1e23fdc14b..ac91dfd834 100644
@@ -679,6 +679,11 @@ void machine_set_cpu_numa_node(MachineState *machine,
             return;
         }
 
+        if (props->has_cluster_id && !slot->props.has_cluster_id) {
+            error_setg(errp, "cluster-id is not supported");
+            return;
+        }
+
         if (props->has_socket_id && !slot->props.has_socket_id) {
             error_setg(errp, "socket-id is not supported");
             return;
@@ -698,6 +703,11 @@ void machine_set_cpu_numa_node(MachineState *machine,
                 continue;
         }
 
+        if (props->has_cluster_id &&
+            props->cluster_id != slot->props.cluster_id) {
+                continue;
+        }
+
         if (props->has_die_id && props->die_id != slot->props.die_id) {
                 continue;
         }
@@ -992,6 +1002,12 @@ static char *cpu_slot_to_string(const CPUArchId *cpu)
         }
         g_string_append_printf(s, "die-id: %"PRId64, cpu->props.die_id);
     }
+    if (cpu->props.has_cluster_id) {
+        if (s->len) {
+            g_string_append_printf(s, ", ");
+        }
+        g_string_append_printf(s, "cluster-id: %"PRId64, cpu->props.cluster_id);
+    }
     if (cpu->props.has_core_id) {
         if (s->len) {
             g_string_append_printf(s, ", ");
qapi/machine.json +4 -2
index d25a481ce4..4c417e32a5 100644
@@ -868,10 +868,11 @@
 # @node-id: NUMA node ID the CPU belongs to
 # @socket-id: socket number within node/board the CPU belongs to
 # @die-id: die number within socket the CPU belongs to (since 4.1)
-# @core-id: core number within die the CPU belongs to
+# @cluster-id: cluster number within die the CPU belongs to (since 7.1)
+# @core-id: core number within cluster the CPU belongs to
 # @thread-id: thread number within core the CPU belongs to
 #
-# Note: currently there are 5 properties that could be present
+# Note: currently there are 6 properties that could be present
 #       but management should be prepared to pass through other
 #       properties with device_add command to allow for future
 #       interface extension. This also requires the filed names to be kept in
@@ -883,6 +884,7 @@
   'data': { '*node-id': 'int',
             '*socket-id': 'int',
             '*die-id': 'int',
+            '*cluster-id': 'int',
             '*core-id': 'int',
             '*thread-id': 'int'
   }
Signature · 1 line
2.23.0
#2 lore raw

Re: [PATCH v7 1/4] qapi/machine.json: Add cluster-id

Hi Gavin,

Quoted 9 lines · Gavin Shan

This adds cluster-id in CPU instance properties, which will be used by arm/virt machine. Besides, the cluster-id is also verified or dumped in various spots:

* hw/core/machine.c::machine_set_cpu_numa_node() to associate CPU with its NUMA node.

* hw/core/machine.c::machine_numa_finish_cpu_init() to associate CPU with NUMA node when no default association isn't provided.

nit: It doesn't really matter too much. But maybe clearer to read a sentence like "machine_numa_finish_cpu_init() to record CPU slots with no NUMA mapping set."

Thanks,
Yanan

Quoted 94 lines

* hw/core/machine-hmp-cmds.c::hmp_hotpluggable_cpus() to dump cluster-id.

---
  hw/core/machine-hmp-cmds.c |  4 ++++
  hw/core/machine.c          | 16 ++++++++++++++++
  qapi/machine.json          |  6 ++++--
  3 files changed, 24 insertions(+), 2 deletions(-)
hw/core/machine-hmp-cmds.c +4 -0
index 4e2f319aeb..5cb5eecbfc 100644
@@ -77,6 +77,10 @@ void hmp_hotpluggable_cpus(Monitor *mon, const QDict *qdict)
          if (c->has_die_id) {
              monitor_printf(mon, "    die-id: \"%" PRIu64 "\"\n", c->die_id);
          }
+        if (c->has_cluster_id) {
+            monitor_printf(mon, "    cluster-id: \"%" PRIu64 "\"\n",
+                           c->cluster_id);
+        }
          if (c->has_core_id) {
              monitor_printf(mon, "    core-id: \"%" PRIu64 "\"\n", c->core_id);
          }
hw/core/machine.c +16 -0
index 1e23fdc14b..ac91dfd834 100644
@@ -679,6 +679,11 @@ void machine_set_cpu_numa_node(MachineState *machine,
              return;
          }
 
+        if (props->has_cluster_id && !slot->props.has_cluster_id) {
+            error_setg(errp, "cluster-id is not supported");
+            return;
+        }
+
          if (props->has_socket_id && !slot->props.has_socket_id) {
              error_setg(errp, "socket-id is not supported");
              return;
@@ -698,6 +703,11 @@ void machine_set_cpu_numa_node(MachineState *machine,
                  continue;
          }
 
+        if (props->has_cluster_id &&
+            props->cluster_id != slot->props.cluster_id) {
+                continue;
+        }
+
          if (props->has_die_id && props->die_id != slot->props.die_id) {
                  continue;
          }
@@ -992,6 +1002,12 @@ static char *cpu_slot_to_string(const CPUArchId *cpu)
          }
          g_string_append_printf(s, "die-id: %"PRId64, cpu->props.die_id);
      }
+    if (cpu->props.has_cluster_id) {
+        if (s->len) {
+            g_string_append_printf(s, ", ");
+        }
+        g_string_append_printf(s, "cluster-id: %"PRId64, cpu->props.cluster_id);
+    }
      if (cpu->props.has_core_id) {
          if (s->len) {
              g_string_append_printf(s, ", ");
qapi/machine.json +4 -2
index d25a481ce4..4c417e32a5 100644
@@ -868,10 +868,11 @@
  # @node-id: NUMA node ID the CPU belongs to
  # @socket-id: socket number within node/board the CPU belongs to
  # @die-id: die number within socket the CPU belongs to (since 4.1)
-# @core-id: core number within die the CPU belongs to
+# @cluster-id: cluster number within die the CPU belongs to (since 7.1)
+# @core-id: core number within cluster the CPU belongs to
  # @thread-id: thread number within core the CPU belongs to
  #
-# Note: currently there are 5 properties that could be present
+# Note: currently there are 6 properties that could be present
  #       but management should be prepared to pass through other
  #       properties with device_add command to allow for future
  #       interface extension. This also requires the filed names to be kept in
@@ -883,6 +884,7 @@
    'data': { '*node-id': 'int',
              '*socket-id': 'int',
              '*die-id': 'int',
+            '*cluster-id': 'int',
              '*core-id': 'int',
              '*thread-id': 'int'
    }
author
#3 lore raw

Re: [PATCH v7 1/4] qapi/machine.json: Add cluster-id

Hi Yanan,

Quoted 5 lines · wangyanan (Y)
Quoted 4 lines · Gavin Shan

This adds cluster-id in CPU instance properties, which will be used by arm/virt machine. Besides, the cluster-id is also verified or dumped in various spots:

Quoted 7 lines
Quoted 5 lines

   * hw/core/machine.c::machine_set_cpu_numa_node() to associate CPU with its NUMA node.

   * hw/core/machine.c::machine_numa_finish_cpu_init() to associate CPU with NUMA node when no default association isn't provided.

nit: It doesn't really matter too much. But maybe clearer to read a sentence like "machine_numa_finish_cpu_init() to record CPU slots with no NUMA mapping set."

I will modify this accordingly in v8. I prefer to hold v8 posting for several days in case someone still have more comments on v7.

Quoted 84 lines
Quoted 84 lines

   * hw/core/machine-hmp-cmds.c::hmp_hotpluggable_cpus() to dump cluster-id.

---
  hw/core/machine-hmp-cmds.c |  4 ++++ hw/core/machine.c          | 16 ++++++++++++++++ qapi/machine.json          |  6 ++++--

  3 files changed, 24 insertions(+), 2 deletions(-)
hw/core/machine-hmp-cmds.c +4 -0
index 4e2f319aeb..5cb5eecbfc 100644
@@ -77,6 +77,10 @@ void hmp_hotpluggable_cpus(Monitor *mon, const QDict *qdict)
           if (c->has_die_id) {
               monitor_printf(mon, "    die-id: \"%" PRIu64 "\"\n", c->die_id);
           }
+        if (c->has_cluster_id) {
+            monitor_printf(mon, "    cluster-id: \"%" PRIu64 "\"\n",
+                           c->cluster_id);
+        }
           if (c->has_core_id) {
               monitor_printf(mon, "    core-id: \"%" PRIu64 "\"\n", c->core_id);
           }
hw/core/machine.c +16 -0
index 1e23fdc14b..ac91dfd834 100644
@@ -679,6 +679,11 @@ void machine_set_cpu_numa_node(MachineState *machine,
               return;
           }
+        if (props->has_cluster_id && !slot->props.has_cluster_id) {
+            error_setg(errp, "cluster-id is not supported");
+            return;
+        }
+
           if (props->has_socket_id && !slot->props.has_socket_id) {
               error_setg(errp, "socket-id is not supported");
               return;
@@ -698,6 +703,11 @@ void machine_set_cpu_numa_node(MachineState *machine,
                   continue;
           }
+        if (props->has_cluster_id &&
+            props->cluster_id != slot->props.cluster_id) {
+                continue;
+        }
+
           if (props->has_die_id && props->die_id != slot->props.die_id) {
                   continue;
           }
@@ -992,6 +1002,12 @@ static char *cpu_slot_to_string(const CPUArchId *cpu)
           }
           g_string_append_printf(s, "die-id: %"PRId64, cpu->props.die_id);
       }
+    if (cpu->props.has_cluster_id) {
+        if (s->len) {
+            g_string_append_printf(s, ", ");
+        }
+        g_string_append_printf(s, "cluster-id: %"PRId64, cpu->props.cluster_id);
+    }
       if (cpu->props.has_core_id) {
           if (s->len) {
               g_string_append_printf(s, ", ");
qapi/machine.json +3 -2
index d25a481ce4..4c417e32a5 100644
@@ -868,10 +868,11 @@
   # @node-id: NUMA node ID the CPU belongs to
   # @socket-id: socket number within node/board the CPU belongs to
   # @die-id: die number within socket the CPU belongs to (since 4.1)
-# @core-id: core number within die the CPU belongs to
+# @cluster-id: cluster number within die the CPU belongs to (since 7.1)
+# @core-id: core number within cluster the CPU belongs to
   # @thread-id: thread number within core the CPU belongs to
   #
-# Note: currently there are 5 properties that could be present
+# Note: currently there are 6 properties that could be present
   #       but management should be prepared to pass through other
   #       properties with device_add command to allow for future
   #       interface extension. This also requires the filed names to be kept in
Quoted 8 lines
Quoted 8 lines
qapi/machine.json +1 -0
@@ -883,6 +884,7 @@
     'data': { '*node-id': 'int',
               '*socket-id': 'int',
               '*die-id': 'int',
+            '*cluster-id': 'int',
               '*core-id': 'int',

              '*thread-id': 'int'
    }

Thanks,
Gavin

author
#4 lore raw

PATCH v7 2/4 hw/arm/virt: Consider SMP configuration in CPU topology

In reply to Gavin Shan

Currently, the SMP configuration isn't considered when the CPU topology is populated. In this case, it's impossible to provide the default CPU-to-NUMA mapping or association based on the socket ID of the given CPU.

This takes account of SMP configuration when the CPU topology is populated. The die ID for the given CPU isn't assigned since it's not supported on arm/virt machine. Besides, the used SMP configuration in qtest/numa-test/aarch64_numa_cpu() is corrcted to avoid testing failure

---
 hw/arm/virt.c           | 15 ++++++++++++++-
 tests/qtest/numa-test.c |  3 ++-
 2 files changed, 16 insertions(+), 2 deletions(-)
hw/arm/virt.c +14 -1
index d2e5ecd234..5443ecae92 100644
@@ -2505,6 +2505,7 @@ static const CPUArchIdList *virt_possible_cpu_arch_ids(MachineState *ms)
     int n;
     unsigned int max_cpus = ms->smp.max_cpus;
     VirtMachineState *vms = VIRT_MACHINE(ms);
+    MachineClass *mc = MACHINE_GET_CLASS(vms);
 
     if (ms->possible_cpus) {
         assert(ms->possible_cpus->len == max_cpus);
@@ -2518,8 +2519,20 @@ static const CPUArchIdList *virt_possible_cpu_arch_ids(MachineState *ms)
         ms->possible_cpus->cpus[n].type = ms->cpu_type;
         ms->possible_cpus->cpus[n].arch_id =
             virt_cpu_mp_affinity(vms, n);
+
+        assert(!mc->smp_props.dies_supported);
+        ms->possible_cpus->cpus[n].props.has_socket_id = true;
+        ms->possible_cpus->cpus[n].props.socket_id =
+            (n / (ms->smp.clusters * ms->smp.cores * ms->smp.threads));
+        ms->possible_cpus->cpus[n].props.has_cluster_id = true;
+        ms->possible_cpus->cpus[n].props.cluster_id =
+            (n / (ms->smp.cores * ms->smp.threads)) % ms->smp.clusters;
+        ms->possible_cpus->cpus[n].props.has_core_id = true;
+        ms->possible_cpus->cpus[n].props.core_id =
+            (n / ms->smp.threads) % ms->smp.cores;
         ms->possible_cpus->cpus[n].props.has_thread_id = true;
-        ms->possible_cpus->cpus[n].props.thread_id = n;
+        ms->possible_cpus->cpus[n].props.thread_id =
+            n % ms->smp.threads;
     }
     return ms->possible_cpus;
 }
tests/qtest/numa-test.c +2 -1
index 90bf68a5b3..aeda8c774c 100644
@@ -223,7 +223,8 @@ static void aarch64_numa_cpu(const void *data)
     QTestState *qts;
     g_autofree char *cli = NULL;
 
-    cli = make_cli(data, "-machine smp.cpus=2 "
+    cli = make_cli(data, "-machine "
+        "smp.cpus=2,smp.sockets=1,smp.clusters=1,smp.cores=1,smp.threads=2 "
         "-numa node,nodeid=0,memdev=ram -numa node,nodeid=1 "
         "-numa cpu,node-id=1,thread-id=0 "
         "-numa cpu,node-id=0,thread-id=1");
Signature · 1 line
2.23.0
#5 lore raw

Re: [PATCH v7 2/4] hw/arm/virt: Consider SMP configuration in CPU topology

Hi Gavin,
Sorry I missed the v6.

Quoted 31 lines · Gavin Shan

Currently, the SMP configuration isn't considered when the CPU topology is populated. In this case, it's impossible to provide the default CPU-to-NUMA mapping or association based on the socket ID of the given CPU.

This takes account of SMP configuration when the CPU topology is populated. The die ID for the given CPU isn't assigned since it's not supported on arm/virt machine. Besides, the used SMP configuration in qtest/numa-test/aarch64_numa_cpu() is corrcted to avoid testing failure

---
  hw/arm/virt.c           | 15 ++++++++++++++-
  tests/qtest/numa-test.c |  3 ++-
  2 files changed, 16 insertions(+), 2 deletions(-)
hw/arm/virt.c +1 -0
index d2e5ecd234..5443ecae92 100644
@@ -2505,6 +2505,7 @@ static const CPUArchIdList *virt_possible_cpu_arch_ids(MachineState *ms)
      int n;
      unsigned int max_cpus = ms->smp.max_cpus;
      VirtMachineState *vms = VIRT_MACHINE(ms);
+    MachineClass *mc = MACHINE_GET_CLASS(vms);
 
      if (ms->possible_cpus) {
          assert(ms->possible_cpus->len == max_cpus);

hw/arm/virt.c static const CPUArchIdList *virt_possible_cpu_arch_ids(MachineState *ms) · line 2526

Quoted 8 lines
hw/arm/virt.c +5 -0
@@ -2518,8 +2519,20 @@ static const CPUArchIdList *virt_possible_cpu_arch_ids(MachineState *ms)
          ms->possible_cpus->cpus[n].type = ms->cpu_type;
          ms->possible_cpus->cpus[n].arch_id =
              virt_cpu_mp_affinity(vms, n);
+
+        assert(!mc->smp_props.dies_supported);
+        ms->possible_cpus->cpus[n].props.has_socket_id = true;
+        ms->possible_cpus->cpus[n].props.socket_id =
+            (n / (ms->smp.clusters * ms->smp.cores * ms->smp.threads));

nit: so the outermost "()" is unnecessary too.

Quoted 18 lines
hw/arm/virt.c +8 -1
+        ms->possible_cpus->cpus[n].props.has_cluster_id = true;
+        ms->possible_cpus->cpus[n].props.cluster_id =
+            (n / (ms->smp.cores * ms->smp.threads)) % ms->smp.clusters;
+        ms->possible_cpus->cpus[n].props.has_core_id = true;
+        ms->possible_cpus->cpus[n].props.core_id =
+            (n / ms->smp.threads) % ms->smp.cores;
          ms->possible_cpus->cpus[n].props.has_thread_id = true;
-        ms->possible_cpus->cpus[n].props.thread_id = n;
+        ms->possible_cpus->cpus[n].props.thread_id =
+            n % ms->smp.threads;
      }
      return ms->possible_cpus;
  }
tests/qtest/numa-test.c +0 -0
index 90bf68a5b3..aeda8c774c 100644

tests/qtest/numa-test.c static void aarch64_numa_cpu(const void *data) · line 230

Quoted 9 lines
tests/qtest/numa-test.c +2 -1
@@ -223,7 +223,8 @@ static void aarch64_numa_cpu(const void *data)
      QTestState *qts;
      g_autofree char *cli = NULL;
 
-    cli = make_cli(data, "-machine smp.cpus=2 "
+    cli = make_cli(data, "-machine "
+        "smp.cpus=2,smp.sockets=1,smp.clusters=1,smp.cores=1,smp.threads=2 "
          "-numa node,nodeid=0,memdev=ram -numa node,nodeid=1 "
          "-numa cpu,node-id=1,thread-id=0 "
          "-numa cpu,node-id=0,thread-id=1");

Thanks,
Yanan

author
#6 lore raw

Re: [PATCH v7 2/4] hw/arm/virt: Consider SMP configuration in CPU topology

Hi Yanan,

Quoted 2 lines · wangyanan (Y)

Hi Gavin,
Sorry I missed the v6.

No problem at all. thanks for your review again :)

Quoted 32 lines · Gavin Shan
Quoted 31 lines

Currently, the SMP configuration isn't considered when the CPU topology is populated. In this case, it's impossible to provide the default CPU-to-NUMA mapping or association based on the socket ID of the given CPU.

This takes account of SMP configuration when the CPU topology is populated. The die ID for the given CPU isn't assigned since it's not supported on arm/virt machine. Besides, the used SMP configuration in qtest/numa-test/aarch64_numa_cpu() is corrcted to avoid testing failure

---
  hw/arm/virt.c           | 15 ++++++++++++++-
  tests/qtest/numa-test.c |  3 ++-

  2 files changed, 16 insertions(+), 2 deletions(-)
hw/arm/virt.c +1 -0
index d2e5ecd234..5443ecae92 100644
@@ -2505,6 +2505,7 @@ static const CPUArchIdList *virt_possible_cpu_arch_ids(MachineState *ms)
       int n;
       unsigned int max_cpus = ms->smp.max_cpus;
       VirtMachineState *vms = VIRT_MACHINE(ms);
+    MachineClass *mc = MACHINE_GET_CLASS(vms);
       if (ms->possible_cpus) {
           assert(ms->possible_cpus->len == max_cpus);
@@ -2518,8 +2519,20 @@ static const CPUArchIdList *virt_possible_cpu_arch_ids(MachineState *ms)
           ms->possible_cpus->cpus[n].type = ms->cpu_type;
Quoted 8 lines
Quoted 7 lines

          ms->possible_cpus->cpus[n].arch_id = virt_cpu_mp_affinity(vms, n);
+
+        assert(!mc->smp_props.dies_supported);
+        ms->possible_cpus->cpus[n].props.has_socket_id = true; +        ms->possible_cpus->cpus[n].props.socket_id =
+            (n / (ms->smp.clusters * ms->smp.cores * ms->smp.threads));

nit: so the outermost "()" is unnecessary too.

It was kept by intention so that it has same style as to other fields like cluster_id. I will remove it in v8 and it doesn't matter actually.

Quoted 18 lines
Quoted 18 lines

+        ms->possible_cpus->cpus[n].props.has_cluster_id = true; +        ms->possible_cpus->cpus[n].props.cluster_id =
+            (n / (ms->smp.cores * ms->smp.threads)) % ms->smp.clusters; +        ms->possible_cpus->cpus[n].props.has_core_id = true; +        ms->possible_cpus->cpus[n].props.core_id =
+            (n / ms->smp.threads) % ms->smp.cores; ms->possible_cpus->cpus[n].props.has_thread_id = true; -        ms->possible_cpus->cpus[n].props.thread_id = n;
+        ms->possible_cpus->cpus[n].props.thread_id =
+            n % ms->smp.threads;
      }
      return ms->possible_cpus;
  }

tests/qtest/numa-test.c +0 -0
index 90bf68a5b3..aeda8c774c 100644
@@ -223,7 +223,8 @@ static void aarch64_numa_cpu(const void *data)
Quoted 8 lines
Quoted 8 lines
tests/qtest/numa-test.c +2 -1
       QTestState *qts;
       g_autofree char *cli = NULL;
-    cli = make_cli(data, "-machine smp.cpus=2 "
+    cli = make_cli(data, "-machine "
+        "smp.cpus=2,smp.sockets=1,smp.clusters=1,smp.cores=1,smp.threads=2 "

          "-numa node,nodeid=0,memdev=ram -numa node,nodeid=1 " "-numa cpu,node-id=1,thread-id=0 "
          "-numa cpu,node-id=0,thread-id=1");

As discussed with Igor, the changes to test/qtest/numa-test.c will be split into a separate patch in v8, which goes before this one. I assume your reviewed-by tag is still valid, even for the separate patch.

Thanks,
Gavin

author
#7 lore raw

PATCH v7 3/4 hw/arm/virt: Fix CPU's default NUMA node ID

In reply to Gavin Shan

When CPU-to-NUMA association isn't explicitly provided by users, the default one is given by mc->get_default_cpu_node_id(). However, the CPU topology isn't fully considered in the default association and this causes CPU topology broken warnings on booting Linux guest.

For example, the following warning messages are observed when the Linux guest is booted with the following command lines.

  /home/gavin/sandbox/qemu.main/build/qemu-system-aarch64 \
  -accel kvm -machine virt,gic-version=host               \
  -cpu host                                               \
  -smp 6,sockets=2,cores=3,threads=1                      \
  -m 1024M,slots=16,maxmem=64G                            \
  -object memory-backend-ram,id=mem0,size=128M            \
  -object memory-backend-ram,id=mem1,size=128M            \
  -object memory-backend-ram,id=mem2,size=128M            \
  -object memory-backend-ram,id=mem3,size=128M            \
  -object memory-backend-ram,id=mem4,size=128M            \
  -object memory-backend-ram,id=mem4,size=384M            \
  -numa node,nodeid=0,memdev=mem0                         \
  -numa node,nodeid=1,memdev=mem1                         \
  -numa node,nodeid=2,memdev=mem2                         \
  -numa node,nodeid=3,memdev=mem3                         \
  -numa node,nodeid=4,memdev=mem4                         \
  -numa node,nodeid=5,memdev=mem5
         :
  alternatives: patching kernel code
  BUG: arch topology borken
  the CLS domain not a subset of the MC domain
  <the above error log repeats>
  BUG: arch topology borken
  the DIE domain not a subset of the NODE domain

With current implementation of mc->get_default_cpu_node_id(), CPU#0 to CPU#5 are associated with NODE#0 to NODE#5 separately. That's incorrect because CPU#0/1/2 should be associated with same NUMA node because they're seated in same socket.

This fixes the issue by considering the socket ID when the default CPU-to-NUMA association is provided in virt_possible_cpu_arch_ids(). With this applied, no more CPU topology broken warnings are seen from the Linux guest. The 6 CPUs are associated with NODE#0/1, but there are no CPUs associated with NODE#2/3/4/5.

---
 hw/arm/virt.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
hw/arm/virt.c +3 -1
index 5443ecae92..c3477a8a17 100644
@@ -2497,7 +2497,9 @@ virt_cpu_index_to_props(MachineState *ms, unsigned cpu_index)
 
 static int64_t virt_get_default_cpu_node_id(const MachineState *ms, int idx)
 {
-    return idx % ms->numa_state->num_nodes;
+    int64_t socket_id = ms->possible_cpus->cpus[idx].props.socket_id;
+
+    return socket_id % ms->numa_state->num_nodes;
 }
 
 static const CPUArchIdList *virt_possible_cpu_arch_ids(MachineState *ms)
Signature · 1 line
2.23.0
author
#8 lore raw

PATCH v7 4/4 hw/acpi/aml-build: Use existing CPU topology to build PPTT table

In reply to Gavin Shan

When the PPTT table is built, the CPU topology is re-calculated, but it's unecessary because the CPU topology has been populated in virt_possible_cpu_arch_ids() on arm/virt machine.

This reworks build_pptt() to avoid by reusing the existing IDs in ms->possible_cpus. Currently, the only user of build_pptt() is arm/virt machine.

---
 hw/acpi/aml-build.c | 109 +++++++++++++++++++-------------------------
 1 file changed, 47 insertions(+), 62 deletions(-)
hw/acpi/aml-build.c +47 -62
index 4086879ebf..73f4e69c29 100644
@@ -2002,86 +2002,71 @@ void build_pptt(GArray *table_data, BIOSLinker *linker, MachineState *ms,
                 const char *oem_id, const char *oem_table_id)
 {
     MachineClass *mc = MACHINE_GET_CLASS(ms);
-    GQueue *list = g_queue_new();
-    guint pptt_start = table_data->len;
-    guint parent_offset;
-    guint length, i;
-    int uid = 0;
-    int socket;
+    CPUArchIdList *cpus = ms->possible_cpus;
+    int64_t socket_id = -1, cluster_id = -1, core_id = -1;
+    uint32_t socket_offset = 0, cluster_offset = 0, core_offset = 0;
+    uint32_t pptt_start = table_data->len;
+    int n;
     AcpiTable table = { .sig = "PPTT", .rev = 2,
                         .oem_id = oem_id, .oem_table_id = oem_table_id };
 
     acpi_table_begin(&table, table_data);
 
-    for (socket = 0; socket < ms->smp.sockets; socket++) {
-        g_queue_push_tail(list,
-            GUINT_TO_POINTER(table_data->len - pptt_start));
-        build_processor_hierarchy_node(
-            table_data,
-            /*
-             * Physical package - represents the boundary
-             * of a physical package
-             */
-            (1 << 0),
-            0, socket, NULL, 0);
-    }
+    /*
+     * This works with the assumption that cpus[n].props.*_id has been
+     * sorted from top to down levels in mc->possible_cpu_arch_ids().
+     * Otherwise, the unexpected and duplicate containers will be
+     * created.
+     */
+    for (n = 0; n < cpus->len; n++) {
+        if (cpus->cpus[n].props.socket_id != socket_id) {
+            assert(cpus->cpus[n].props.socket_id > socket_id);
+            socket_id = cpus->cpus[n].props.socket_id;
+            cluster_id = -1;
+            core_id = -1;
+            socket_offset = table_data->len - pptt_start;
+            build_processor_hierarchy_node(table_data,
+                (1 << 0), /* Physical package */
+                0, socket_id, NULL, 0);
+        }
 
-    if (mc->smp_props.clusters_supported) {
-        length = g_queue_get_length(list);
-        for (i = 0; i < length; i++) {
-            int cluster;
-
-            parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-            for (cluster = 0; cluster < ms->smp.clusters; cluster++) {
-                g_queue_push_tail(list,
-                    GUINT_TO_POINTER(table_data->len - pptt_start));
-                build_processor_hierarchy_node(
-                    table_data,
-                    (0 << 0), /* not a physical package */
-                    parent_offset, cluster, NULL, 0);
+        if (mc->smp_props.clusters_supported) {
+            if (cpus->cpus[n].props.cluster_id != cluster_id) {
+                assert(cpus->cpus[n].props.cluster_id > cluster_id);
+                cluster_id = cpus->cpus[n].props.cluster_id;
+                core_id = -1;
+                cluster_offset = table_data->len - pptt_start;
+                build_processor_hierarchy_node(table_data,
+                    (0 << 0), /* Not a physical package */
+                    socket_offset, cluster_id, NULL, 0);
             }
+        } else {
+            cluster_offset = socket_offset;
         }
-    }
 
-    length = g_queue_get_length(list);
-    for (i = 0; i < length; i++) {
-        int core;
-
-        parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-        for (core = 0; core < ms->smp.cores; core++) {
-            if (ms->smp.threads > 1) {
-                g_queue_push_tail(list,
-                    GUINT_TO_POINTER(table_data->len - pptt_start));
-                build_processor_hierarchy_node(
-                    table_data,
+        if (ms->smp.threads == 1) {
+            build_processor_hierarchy_node(table_data,
+                (1 << 1) | /* ACPI Processor ID valid */
+                (1 << 3),  /* Node is a Leaf */
+                cluster_offset, n, NULL, 0);
+        } else {
+            if (cpus->cpus[n].props.core_id != core_id) {
+                assert(cpus->cpus[n].props.core_id > core_id);
+                core_id = cpus->cpus[n].props.core_id;
+                core_offset = table_data->len - pptt_start;
+                build_processor_hierarchy_node(table_data,
                     (0 << 0), /* not a physical package */
-                    parent_offset, core, NULL, 0);
-            } else {
-                build_processor_hierarchy_node(
-                    table_data,
-                    (1 << 1) | /* ACPI Processor ID valid */
-                    (1 << 3),  /* Node is a Leaf */
-                    parent_offset, uid++, NULL, 0);
+                    cluster_offset, core_id, NULL, 0);
             }
-        }
-    }
-
-    length = g_queue_get_length(list);
-    for (i = 0; i < length; i++) {
-        int thread;
 
-        parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-        for (thread = 0; thread < ms->smp.threads; thread++) {
-            build_processor_hierarchy_node(
-                table_data,
+            build_processor_hierarchy_node(table_data,
                 (1 << 1) | /* ACPI Processor ID valid */
                 (1 << 2) | /* Processor is a Thread */
                 (1 << 3),  /* Node is a Leaf */
-                parent_offset, uid++, NULL, 0);
+                core_offset, n, NULL, 0);
         }
     }
 
-    g_queue_free(list);
     acpi_table_end(linker, &table);
 }
 
Signature · 1 line
2.23.0
Acked-by on v7 4/4
#9 lore raw

Re: [PATCH v7 4/4] hw/acpi/aml-build: Use existing CPU topology to build PPTT table

Quoted 9 lines · On Wed, 20 Apr 2022 18:49:09 +0800 Gavin Shan <[email protected]> wrote:

When the PPTT table is built, the CPU topology is re-calculated, but it's unecessary because the CPU topology has been populated in virt_possible_cpu_arch_ids() on arm/virt machine.

This reworks build_pptt() to avoid by reusing the existing IDs in ms->possible_cpus. Currently, the only user of build_pptt() is arm/virt machine.

Looks fine to me, so

Also do we have a bios-tables tests that watches over/tests PPTT table? if not please add one as a patch on top.

Quoted 142 lines
---
 hw/acpi/aml-build.c | 109 +++++++++++++++++++-------------------------
 1 file changed, 47 insertions(+), 62 deletions(-)
hw/acpi/aml-build.c +47 -62
index 4086879ebf..73f4e69c29 100644
@@ -2002,86 +2002,71 @@ void build_pptt(GArray *table_data, BIOSLinker *linker, MachineState *ms,
                 const char *oem_id, const char *oem_table_id)
 {
     MachineClass *mc = MACHINE_GET_CLASS(ms);
-    GQueue *list = g_queue_new();
-    guint pptt_start = table_data->len;
-    guint parent_offset;
-    guint length, i;
-    int uid = 0;
-    int socket;
+    CPUArchIdList *cpus = ms->possible_cpus;
+    int64_t socket_id = -1, cluster_id = -1, core_id = -1;
+    uint32_t socket_offset = 0, cluster_offset = 0, core_offset = 0;
+    uint32_t pptt_start = table_data->len;
+    int n;
     AcpiTable table = { .sig = "PPTT", .rev = 2,
                         .oem_id = oem_id, .oem_table_id = oem_table_id };
 
     acpi_table_begin(&table, table_data);
 
-    for (socket = 0; socket < ms->smp.sockets; socket++) {
-        g_queue_push_tail(list,
-            GUINT_TO_POINTER(table_data->len - pptt_start));
-        build_processor_hierarchy_node(
-            table_data,
-            /*
-             * Physical package - represents the boundary
-             * of a physical package
-             */
-            (1 << 0),
-            0, socket, NULL, 0);
-    }
+    /*
+     * This works with the assumption that cpus[n].props.*_id has been
+     * sorted from top to down levels in mc->possible_cpu_arch_ids().
+     * Otherwise, the unexpected and duplicate containers will be
+     * created.
+     */
+    for (n = 0; n < cpus->len; n++) {
+        if (cpus->cpus[n].props.socket_id != socket_id) {
+            assert(cpus->cpus[n].props.socket_id > socket_id);
+            socket_id = cpus->cpus[n].props.socket_id;
+            cluster_id = -1;
+            core_id = -1;
+            socket_offset = table_data->len - pptt_start;
+            build_processor_hierarchy_node(table_data,
+                (1 << 0), /* Physical package */
+                0, socket_id, NULL, 0);
+        }
 
-    if (mc->smp_props.clusters_supported) {
-        length = g_queue_get_length(list);
-        for (i = 0; i < length; i++) {
-            int cluster;
-
-            parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-            for (cluster = 0; cluster < ms->smp.clusters; cluster++) {
-                g_queue_push_tail(list,
-                    GUINT_TO_POINTER(table_data->len - pptt_start));
-                build_processor_hierarchy_node(
-                    table_data,
-                    (0 << 0), /* not a physical package */
-                    parent_offset, cluster, NULL, 0);
+        if (mc->smp_props.clusters_supported) {
+            if (cpus->cpus[n].props.cluster_id != cluster_id) {
+                assert(cpus->cpus[n].props.cluster_id > cluster_id);
+                cluster_id = cpus->cpus[n].props.cluster_id;
+                core_id = -1;
+                cluster_offset = table_data->len - pptt_start;
+                build_processor_hierarchy_node(table_data,
+                    (0 << 0), /* Not a physical package */
+                    socket_offset, cluster_id, NULL, 0);
             }
+        } else {
+            cluster_offset = socket_offset;
         }
-    }
 
-    length = g_queue_get_length(list);
-    for (i = 0; i < length; i++) {
-        int core;
-
-        parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-        for (core = 0; core < ms->smp.cores; core++) {
-            if (ms->smp.threads > 1) {
-                g_queue_push_tail(list,
-                    GUINT_TO_POINTER(table_data->len - pptt_start));
-                build_processor_hierarchy_node(
-                    table_data,
+        if (ms->smp.threads == 1) {
+            build_processor_hierarchy_node(table_data,
+                (1 << 1) | /* ACPI Processor ID valid */
+                (1 << 3),  /* Node is a Leaf */
+                cluster_offset, n, NULL, 0);
+        } else {
+            if (cpus->cpus[n].props.core_id != core_id) {
+                assert(cpus->cpus[n].props.core_id > core_id);
+                core_id = cpus->cpus[n].props.core_id;
+                core_offset = table_data->len - pptt_start;
+                build_processor_hierarchy_node(table_data,
                     (0 << 0), /* not a physical package */
-                    parent_offset, core, NULL, 0);
-            } else {
-                build_processor_hierarchy_node(
-                    table_data,
-                    (1 << 1) | /* ACPI Processor ID valid */
-                    (1 << 3),  /* Node is a Leaf */
-                    parent_offset, uid++, NULL, 0);
+                    cluster_offset, core_id, NULL, 0);
             }
-        }
-    }
-
-    length = g_queue_get_length(list);
-    for (i = 0; i < length; i++) {
-        int thread;
 
-        parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-        for (thread = 0; thread < ms->smp.threads; thread++) {
-            build_processor_hierarchy_node(
-                table_data,
+            build_processor_hierarchy_node(table_data,
                 (1 << 1) | /* ACPI Processor ID valid */
                 (1 << 2) | /* Processor is a Thread */
                 (1 << 3),  /* Node is a Leaf */
-                parent_offset, uid++, NULL, 0);
+                core_offset, n, NULL, 0);
         }
     }
 
-    g_queue_free(list);
     acpi_table_end(linker, &table);
 }
 
author
#10 lore raw

Re: [PATCH v7 4/4] hw/acpi/aml-build: Use existing CPU topology to build PPTT table

Hi Igor,

Quoted 13 lines · Igor Mammedov
Quoted 9 lines · On Wed, 20 Apr 2022 18:49:09 +0800 Gavin Shan <[email protected]> wrote:

When the PPTT table is built, the CPU topology is re-calculated, but it's unecessary because the CPU topology has been populated in virt_possible_cpu_arch_ids() on arm/virt machine.

This reworks build_pptt() to avoid by reusing the existing IDs in ms->possible_cpus. Currently, the only user of build_pptt() is arm/virt machine.

Quoted 8 lines

Looks fine to me, so

Also do we have a bios-tables tests that watches over/tests PPTT table? if not please add one as a patch on top.

Thanks a lot for your continuous reviews. We won't have PPTT table included in bios-table tests. I will add one after this series is merged because our downstream release needs this series :)

Quoted 138 lines
Quoted 138 lines
---
  hw/acpi/aml-build.c | 109 +++++++++++++++++++-------------------------
  1 file changed, 47 insertions(+), 62 deletions(-)
hw/acpi/aml-build.c +47 -61
index 4086879ebf..73f4e69c29 100644
@@ -2002,86 +2002,71 @@ void build_pptt(GArray *table_data, BIOSLinker *linker, MachineState *ms,
                  const char *oem_id, const char *oem_table_id)
  {
      MachineClass *mc = MACHINE_GET_CLASS(ms);
-    GQueue *list = g_queue_new();
-    guint pptt_start = table_data->len;
-    guint parent_offset;
-    guint length, i;
-    int uid = 0;
-    int socket;
+    CPUArchIdList *cpus = ms->possible_cpus;
+    int64_t socket_id = -1, cluster_id = -1, core_id = -1;
+    uint32_t socket_offset = 0, cluster_offset = 0, core_offset = 0;
+    uint32_t pptt_start = table_data->len;
+    int n;
      AcpiTable table = { .sig = "PPTT", .rev = 2,
                          .oem_id = oem_id, .oem_table_id = oem_table_id };
 
      acpi_table_begin(&table, table_data);
 
-    for (socket = 0; socket < ms->smp.sockets; socket++) {
-        g_queue_push_tail(list,
-            GUINT_TO_POINTER(table_data->len - pptt_start));
-        build_processor_hierarchy_node(
-            table_data,
-            /*
-             * Physical package - represents the boundary
-             * of a physical package
-             */
-            (1 << 0),
-            0, socket, NULL, 0);
-    }
+    /*
+     * This works with the assumption that cpus[n].props.*_id has been
+     * sorted from top to down levels in mc->possible_cpu_arch_ids().
+     * Otherwise, the unexpected and duplicate containers will be
+     * created.
+     */
+    for (n = 0; n < cpus->len; n++) {
+        if (cpus->cpus[n].props.socket_id != socket_id) {
+            assert(cpus->cpus[n].props.socket_id > socket_id);
+            socket_id = cpus->cpus[n].props.socket_id;
+            cluster_id = -1;
+            core_id = -1;
+            socket_offset = table_data->len - pptt_start;
+            build_processor_hierarchy_node(table_data,
+                (1 << 0), /* Physical package */
+                0, socket_id, NULL, 0);
+        }
 
-    if (mc->smp_props.clusters_supported) {
-        length = g_queue_get_length(list);
-        for (i = 0; i < length; i++) {
-            int cluster;
-
-            parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-            for (cluster = 0; cluster < ms->smp.clusters; cluster++) {
-                g_queue_push_tail(list,
-                    GUINT_TO_POINTER(table_data->len - pptt_start));
-                build_processor_hierarchy_node(
-                    table_data,
-                    (0 << 0), /* not a physical package */
-                    parent_offset, cluster, NULL, 0);
+        if (mc->smp_props.clusters_supported) {
+            if (cpus->cpus[n].props.cluster_id != cluster_id) {
+                assert(cpus->cpus[n].props.cluster_id > cluster_id);
+                cluster_id = cpus->cpus[n].props.cluster_id;
+                core_id = -1;
+                cluster_offset = table_data->len - pptt_start;
+                build_processor_hierarchy_node(table_data,
+                    (0 << 0), /* Not a physical package */
+                    socket_offset, cluster_id, NULL, 0);
              }
+        } else {
+            cluster_offset = socket_offset;
          }
-    }
 
-    length = g_queue_get_length(list);
-    for (i = 0; i < length; i++) {
-        int core;
-
-        parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-        for (core = 0; core < ms->smp.cores; core++) {
-            if (ms->smp.threads > 1) {
-                g_queue_push_tail(list,
-                    GUINT_TO_POINTER(table_data->len - pptt_start));
-                build_processor_hierarchy_node(
-                    table_data,
+        if (ms->smp.threads == 1) {
+            build_processor_hierarchy_node(table_data,
+                (1 << 1) | /* ACPI Processor ID valid */
+                (1 << 3),  /* Node is a Leaf */
+                cluster_offset, n, NULL, 0);
+        } else {
+            if (cpus->cpus[n].props.core_id != core_id) {
+                assert(cpus->cpus[n].props.core_id > core_id);
+                core_id = cpus->cpus[n].props.core_id;
+                core_offset = table_data->len - pptt_start;
+                build_processor_hierarchy_node(table_data,
                      (0 << 0), /* not a physical package */
-                    parent_offset, core, NULL, 0);
-            } else {
-                build_processor_hierarchy_node(
-                    table_data,
-                    (1 << 1) | /* ACPI Processor ID valid */
-                    (1 << 3),  /* Node is a Leaf */
-                    parent_offset, uid++, NULL, 0);
+                    cluster_offset, core_id, NULL, 0);
              }
-        }
-    }
-
-    length = g_queue_get_length(list);
-    for (i = 0; i < length; i++) {
-        int thread;
 
-        parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-        for (thread = 0; thread < ms->smp.threads; thread++) {
-            build_processor_hierarchy_node(
-                table_data,
+            build_processor_hierarchy_node(table_data,
                  (1 << 1) | /* ACPI Processor ID valid */
                  (1 << 2) | /* Processor is a Thread */
                  (1 << 3),  /* Node is a Leaf */
-                parent_offset, uid++, NULL, 0);
+                core_offset, n, NULL, 0);
          }
      }
 
Quoted 4 lines

hw/acpi/aml-build.c void build_pptt(GArray *table_data, BIOSLinker *linker, MachineState *ms, · line 2072

Quoted 4 lines
hw/acpi/aml-build.c +0 -1
-    g_queue_free(list);
      acpi_table_end(linker, &table);
  }
 

Thanks,
Gavin

Tested-by Yanan Wang Reviewed-by Yanan Wang on v7 4/4
#11 lore raw

Re: [PATCH v7 4/4] hw/acpi/aml-build: Use existing CPU topology to build PPTT table

In reply to Gavin Shan

Hi Gavin,

Quoted 45 lines · Gavin Shan

When the PPTT table is built, the CPU topology is re-calculated, but it's unecessary because the CPU topology has been populated in virt_possible_cpu_arch_ids() on arm/virt machine.

This reworks build_pptt() to avoid by reusing the existing IDs in ms->possible_cpus. Currently, the only user of build_pptt() is arm/virt machine.

---
  hw/acpi/aml-build.c | 109 +++++++++++++++++++-------------------------
  1 file changed, 47 insertions(+), 62 deletions(-)
hw/acpi/aml-build.c +5 -14
index 4086879ebf..73f4e69c29 100644
@@ -2002,86 +2002,71 @@ void build_pptt(GArray *table_data, BIOSLinker *linker, MachineState *ms,
                  const char *oem_id, const char *oem_table_id)
  {
      MachineClass *mc = MACHINE_GET_CLASS(ms);
-    GQueue *list = g_queue_new();
-    guint pptt_start = table_data->len;
-    guint parent_offset;
-    guint length, i;
-    int uid = 0;
-    int socket;
+    CPUArchIdList *cpus = ms->possible_cpus;
+    int64_t socket_id = -1, cluster_id = -1, core_id = -1;
+    uint32_t socket_offset = 0, cluster_offset = 0, core_offset = 0;
+    uint32_t pptt_start = table_data->len;
+    int n;
      AcpiTable table = { .sig = "PPTT", .rev = 2,
                          .oem_id = oem_id, .oem_table_id = oem_table_id };
 
      acpi_table_begin(&table, table_data);
 
-    for (socket = 0; socket < ms->smp.sockets; socket++) {
-        g_queue_push_tail(list,
-            GUINT_TO_POINTER(table_data->len - pptt_start));
-        build_processor_hierarchy_node(
-            table_data,
-            /*
-             * Physical package - represents the boundary
-             * of a physical package

hw/acpi/aml-build.c void build_pptt(GArray *table_data, BIOSLinker *linker, MachineState *ms, · line 2020

Quoted 10 lines
hw/acpi/aml-build.c +6 -4
-             */
-            (1 << 0),
-            0, socket, NULL, 0);
-    }
+    /*
+     * This works with the assumption that cpus[n].props.*_id has been
+     * sorted from top to down levels in mc->possible_cpu_arch_ids().
+     * Otherwise, the unexpected and duplicate containers will be
+     * created.
+     */

s/duplicate/duplicated

Quoted 53 lines
hw/acpi/aml-build.c +24 -25
+    for (n = 0; n < cpus->len; n++) {
+        if (cpus->cpus[n].props.socket_id != socket_id) {
+            assert(cpus->cpus[n].props.socket_id > socket_id);
+            socket_id = cpus->cpus[n].props.socket_id;
+            cluster_id = -1;
+            core_id = -1;
+            socket_offset = table_data->len - pptt_start;
+            build_processor_hierarchy_node(table_data,
+                (1 << 0), /* Physical package */
+                0, socket_id, NULL, 0);
+        }
 
-    if (mc->smp_props.clusters_supported) {
-        length = g_queue_get_length(list);
-        for (i = 0; i < length; i++) {
-            int cluster;
-
-            parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-            for (cluster = 0; cluster < ms->smp.clusters; cluster++) {
-                g_queue_push_tail(list,
-                    GUINT_TO_POINTER(table_data->len - pptt_start));
-                build_processor_hierarchy_node(
-                    table_data,
-                    (0 << 0), /* not a physical package */
-                    parent_offset, cluster, NULL, 0);
+        if (mc->smp_props.clusters_supported) {
+            if (cpus->cpus[n].props.cluster_id != cluster_id) {
+                assert(cpus->cpus[n].props.cluster_id > cluster_id);
+                cluster_id = cpus->cpus[n].props.cluster_id;
+                core_id = -1;
+                cluster_offset = table_data->len - pptt_start;
+                build_processor_hierarchy_node(table_data,
+                    (0 << 0), /* Not a physical package */
+                    socket_offset, cluster_id, NULL, 0);
              }
+        } else {
+            cluster_offset = socket_offset;
          }
-    }
 
-    length = g_queue_get_length(list);
-    for (i = 0; i < length; i++) {
-        int core;
-
-        parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-        for (core = 0; core < ms->smp.cores; core++) {
-            if (ms->smp.threads > 1) {
-                g_queue_push_tail(list,
-                    GUINT_TO_POINTER(table_data->len - pptt_start));
-                build_processor_hierarchy_node(
-                    table_data,
+        if (ms->smp.threads == 1) {
+            build_processor_hierarchy_node(table_data,

hw/acpi/aml-build.c void build_pptt(GArray *table_data, BIOSLinker *linker, MachineState *ms, · line 2058

Quoted 10 lines
hw/acpi/aml-build.c +9 -0
+                (1 << 1) | /* ACPI Processor ID valid */
+                (1 << 3),  /* Node is a Leaf */
+                cluster_offset, n, NULL, 0);
+        } else {
+            if (cpus->cpus[n].props.core_id != core_id) {
+                assert(cpus->cpus[n].props.core_id > core_id);
+                core_id = cpus->cpus[n].props.core_id;
+                core_offset = table_data->len - pptt_start;
+                build_processor_hierarchy_node(table_data,
                      (0 << 0), /* not a physical package */

nit: For consistency, maybe "Not a physical package"

Quoted 23 lines
hw/acpi/aml-build.c +2 -17
-                    parent_offset, core, NULL, 0);
-            } else {
-                build_processor_hierarchy_node(
-                    table_data,
-                    (1 << 1) | /* ACPI Processor ID valid */
-                    (1 << 3),  /* Node is a Leaf */
-                    parent_offset, uid++, NULL, 0);
+                    cluster_offset, core_id, NULL, 0);
              }
-        }
-    }
-
-    length = g_queue_get_length(list);
-    for (i = 0; i < length; i++) {
-        int thread;
 
-        parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-        for (thread = 0; thread < ms->smp.threads; thread++) {
-            build_processor_hierarchy_node(
-                table_data,
+            build_processor_hierarchy_node(table_data,
                  (1 << 1) | /* ACPI Processor ID valid */
                  (1 << 2) | /* Processor is a Thread */

hw/acpi/aml-build.c void build_pptt(GArray *table_data, BIOSLinker *linker, MachineState *ms, · line 2072

Quoted 10 lines
hw/acpi/aml-build.c +1 -2
                  (1 << 3),  /* Node is a Leaf */
-                parent_offset, uid++, NULL, 0);
+                core_offset, n, NULL, 0);
          }
      }
 
-    g_queue_free(list);
      acpi_table_end(linker, &table);
  }
 

For the function of exposing vCPU topology to guest:

Thanks,
Yanan

author
#12 lore raw

Re: [PATCH v7 4/4] hw/acpi/aml-build: Use existing CPU topology to build PPTT table

Hi Yanan,

Quoted 47 lines · wangyanan (Y)
Quoted 46 lines · Gavin Shan

When the PPTT table is built, the CPU topology is re-calculated, but it's unecessary because the CPU topology has been populated in virt_possible_cpu_arch_ids() on arm/virt machine.

This reworks build_pptt() to avoid by reusing the existing IDs in ms->possible_cpus. Currently, the only user of build_pptt() is arm/virt machine.

---
  hw/acpi/aml-build.c | 109 +++++++++++++++++++-------------------------

  1 file changed, 47 insertions(+), 62 deletions(-)
hw/acpi/aml-build.c +5 -17
index 4086879ebf..73f4e69c29 100644
@@ -2002,86 +2002,71 @@ void build_pptt(GArray *table_data, BIOSLinker *linker, MachineState *ms,
                   const char *oem_id, const char *oem_table_id)
   {
       MachineClass *mc = MACHINE_GET_CLASS(ms);
-    GQueue *list = g_queue_new();
-    guint pptt_start = table_data->len;
-    guint parent_offset;
-    guint length, i;
-    int uid = 0;
-    int socket;
+    CPUArchIdList *cpus = ms->possible_cpus;
+    int64_t socket_id = -1, cluster_id = -1, core_id = -1;
+    uint32_t socket_offset = 0, cluster_offset = 0, core_offset = 0;
+    uint32_t pptt_start = table_data->len;
+    int n;
       AcpiTable table = { .sig = "PPTT", .rev = 2,
                           .oem_id = oem_id, .oem_table_id = oem_table_id };
       acpi_table_begin(&table, table_data);
-    for (socket = 0; socket < ms->smp.sockets; socket++) {
-        g_queue_push_tail(list,
-            GUINT_TO_POINTER(table_data->len - pptt_start));
-        build_processor_hierarchy_node(
-            table_data,
-            /*
-             * Physical package - represents the boundary
-             * of a physical package
-             */
-            (1 << 0),
-            0, socket, NULL, 0);
Quoted 8 lines

hw/acpi/aml-build.c void build_pptt(GArray *table_data, BIOSLinker *linker, MachineState *ms, · line 2020

Quoted 7 lines
hw/acpi/aml-build.c +6 -1
-    }
+    /*
+     * This works with the assumption that cpus[n].props.*_id has been
+     * sorted from top to down levels in mc->possible_cpu_arch_ids().
+     * Otherwise, the unexpected and duplicate containers will be
+     * created.
+     */

s/duplicate/duplicated

Both should be fine, but I will replace it with the one you prefer in v8.

Quoted 54 lines
Quoted 54 lines
hw/acpi/aml-build.c +27 -25
+    for (n = 0; n < cpus->len; n++) {
+        if (cpus->cpus[n].props.socket_id != socket_id) {
+            assert(cpus->cpus[n].props.socket_id > socket_id);
+            socket_id = cpus->cpus[n].props.socket_id;
+            cluster_id = -1;
+            core_id = -1;
+            socket_offset = table_data->len - pptt_start;
+            build_processor_hierarchy_node(table_data,
+                (1 << 0), /* Physical package */
+                0, socket_id, NULL, 0);
+        }
-    if (mc->smp_props.clusters_supported) {
-        length = g_queue_get_length(list);
-        for (i = 0; i < length; i++) {
-            int cluster;
-
-            parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-            for (cluster = 0; cluster < ms->smp.clusters; cluster++) {
-                g_queue_push_tail(list,
-                    GUINT_TO_POINTER(table_data->len - pptt_start));
-                build_processor_hierarchy_node(
-                    table_data,
-                    (0 << 0), /* not a physical package */
-                    parent_offset, cluster, NULL, 0);
+        if (mc->smp_props.clusters_supported) {
+            if (cpus->cpus[n].props.cluster_id != cluster_id) {
+                assert(cpus->cpus[n].props.cluster_id > cluster_id);
+                cluster_id = cpus->cpus[n].props.cluster_id;
+                core_id = -1;
+                cluster_offset = table_data->len - pptt_start;
+                build_processor_hierarchy_node(table_data,
+                    (0 << 0), /* Not a physical package */
+                    socket_offset, cluster_id, NULL, 0);
               }
+        } else {
+            cluster_offset = socket_offset;
           }
-    }
-    length = g_queue_get_length(list);
-    for (i = 0; i < length; i++) {
-        int core;
-
-        parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list));
-        for (core = 0; core < ms->smp.cores; core++) {
-            if (ms->smp.threads > 1) {
-                g_queue_push_tail(list,
-                    GUINT_TO_POINTER(table_data->len - pptt_start));
-                build_processor_hierarchy_node(
-                    table_data,
+        if (ms->smp.threads == 1) {
+            build_processor_hierarchy_node(table_data,
+                (1 << 1) | /* ACPI Processor ID valid */
+                (1 << 3),  /* Node is a Leaf */
+                cluster_offset, n, NULL, 0);
Quoted 8 lines
Quoted 7 lines
hw/acpi/aml-build.c +6 -0
+        } else {
+            if (cpus->cpus[n].props.core_id != core_id) {
+                assert(cpus->cpus[n].props.core_id > core_id);
+                core_id = cpus->cpus[n].props.core_id;
+                core_offset = table_data->len - pptt_start;
+                build_processor_hierarchy_node(table_data,

                      (0 << 0), /* not a physical package */

nit: For consistency, maybe "Not a physical package"

Sure. Will do in v8.

Quoted 30 lines
Quoted 22 lines

-                    parent_offset, core, NULL, 0);
-            } else {
-                build_processor_hierarchy_node(
-                    table_data,
-                    (1 << 1) | /* ACPI Processor ID valid */ -                    (1 << 3),  /* Node is a Leaf */
-                    parent_offset, uid++, NULL, 0);
+                    cluster_offset, core_id, NULL, 0); }
-        }
-    }
-
-    length = g_queue_get_length(list);
-    for (i = 0; i < length; i++) {
-        int thread;
-        parent_offset = GPOINTER_TO_UINT(g_queue_pop_head(list)); -        for (thread = 0; thread < ms->smp.threads; thread++) { -            build_processor_hierarchy_node(
-                table_data,
+            build_processor_hierarchy_node(table_data, (1 << 1) | /* ACPI Processor ID valid */ (1 << 2) | /* Processor is a Thread */

Quoted 8 lines

                  (1 << 3),  /* Node is a Leaf */
-                parent_offset, uid++, NULL, 0);
+                core_offset, n, NULL, 0);
          }
      }
-    g_queue_free(list);
      acpi_table_end(linker, &table);
  }

Quoted 4 lines

For the function of exposing vCPU topology to guest:

Thanks for testing it and your time to review this series.

Thanks,
Gavin