From: Zhao Liu <zhao1.liu@linux.intel.com>
To: "Paolo Bonzini" <pbonzini@redhat.com>,
"Alex Bennée" <alex.bennee@linaro.org>,
"Philippe Mathieu-Daudé" <philmd@linaro.org>,
"Eduardo Habkost" <eduardo@habkost.net>,
"Marcel Apfelbaum" <marcel.apfelbaum@gmail.com>,
"Yanan Wang" <wangyanan55@huawei.com>,
"Richard Henderson" <richard.henderson@linaro.org>,
"Michael S . Tsirkin" <mst@redhat.com>,
"Jason Wang" <jasowang@redhat.com>,
"Nicholas Piggin" <npiggin@gmail.com>,
"Daniel Henrique Barboza" <danielhb413@gmail.com>,
"Igor Mammedov" <imammedo@redhat.com>,
"Cédric Le Goater" <clg@kaod.org>,
"Frédéric Barrat" <fbarrat@linux.ibm.com>,
"David Gibson" <david@gibson.dropbear.id.au>,
"Harsh Prateek Bora" <harshpb@linux.ibm.com>,
"Stefano Stabellini" <sstabellini@kernel.org>,
"Anthony Perard" <anthony.perard@citrix.com>,
"Paul Durrant" <paul@xen.org>,
"Gerd Hoffmann" <kraxel@redhat.com>,
"Peter Maydell" <peter.maydell@linaro.org>,
"Alistair Francis" <alistair@alistair23.me>,
"Edgar E . Iglesias" <edgar.iglesias@gmail.com>,
"Daniel P . Berrangé" <berrange@redhat.com>,
"Bin Meng" <bin.meng@windriver.com>,
"Palmer Dabbelt" <palmer@dabbelt.com>,
"Weiwei Li" <liwei1518@gmail.com>,
"Liu Zhiwei" <zhiwei_liu@linux.alibaba.com>,
qemu-devel@nongnu.org, kvm@vger.kernel.org, qemu-ppc@nongnu.org,
xen-devel@lists.xenproject.org, qemu-arm@nongnu.org,
qemu-riscv@nongnu.org, qemu-s390x@nongnu.org
Cc: Nina Schoetterl-Glausch <nsg@linux.ibm.com>,
Thomas Huth <thuth@redhat.com>, Zhiyuan Lv <zhiyuan.lv@intel.com>,
Zhenyu Wang <zhenyu.z.wang@intel.com>,
Yongwei Ma <yongwei.ma@intel.com>, Zhao Liu <zhao1.liu@intel.com>
Subject: [RFC 39/41] hw/i386: Add the interface to search parent for QOM topology
Date: Thu, 30 Nov 2023 22:42:01 +0800 [thread overview]
Message-ID: <20231130144203.2307629-40-zhao1.liu@linux.intel.com> (raw)
In-Reply-To: <20231130144203.2307629-1-zhao1.liu@linux.intel.com>
From: Zhao Liu <zhao1.liu@intel.com>
QOM topology needs to search parent cpu-core for hotplugged CPU to
create topology child<> property in qdev_set_id().
This process is before x86_cpu_pre_plug(), thus place 2 helpers
x86_cpu_assign_apic_id() and x86_cpu_assign_topo_id() in
x86_cpu_search_parent_pre_plug() to help get the correct topology sub
indexes. Then x86_cpu_search_parent_pre_plug() searches the parent
cpu-core with these sub indexes.
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
---
hw/i386/x86.c | 128 +++++++++++++++++++++++++++++++++++++++---
include/hw/i386/x86.h | 3 +
2 files changed, 122 insertions(+), 9 deletions(-)
diff --git a/hw/i386/x86.c b/hw/i386/x86.c
index 04edd6de6aeb..595d4365fdd1 100644
--- a/hw/i386/x86.c
+++ b/hw/i386/x86.c
@@ -460,16 +460,18 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,
return;
}
- /* if 'address' properties socket-id/core-id/thread-id are not set, set them
- * so that machine_query_hotpluggable_cpus would show correct values
+ /*
+ * possible_cpus_qom_granu means the QOM topology support.
+ *
+ * TODO: Drop the "!mc->smp_props.possible_cpus_qom_granu" case when
+ * i386 completes QOM topology support.
*/
- /* TODO: move socket_id/core_id/thread_id checks into x86_cpu_realizefn()
- * once -smp refactoring is complete and there will be CPU private
- * CPUState::nr_cores and CPUState::nr_threads fields instead of globals */
- x86_topo_ids_from_apicid(cpu->apic_id, &topo_info, &topo_ids);
- x86_cpu_assign_topo_id(cpu, &topo_ids, errp);
- if (*errp) {
- return;
+ if (!mc->smp_props.possible_cpus_qom_granu) {
+ x86_topo_ids_from_apicid(cpu->apic_id, &topo_info, &topo_ids);
+ x86_cpu_assign_topo_id(cpu, &topo_ids, errp);
+ if (*errp) {
+ return;
+ }
}
if (hyperv_feat_enabled(cpu, HYPERV_FEAT_VPINDEX) &&
@@ -484,6 +486,114 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,
numa_cpu_pre_plug(cpu_slot, dev, errp);
}
+static int x86_cpu_get_topo_id_by_level(X86CPU *cpu,
+ CPUTopoLevel level)
+{
+ switch (level) {
+ case CPU_TOPO_THREAD:
+ return cpu->thread_id;
+ case CPU_TOPO_CORE:
+ return cpu->core_id;
+ case CPU_TOPO_DIE:
+ return cpu->die_id;
+ case CPU_TOPO_SOCKET:
+ return cpu->socket_id;
+ default:
+ g_assert_not_reached();
+ }
+
+ return -1;
+}
+
+typedef struct SearchCoreCb {
+ X86CPU *cpu;
+ CPUTopoState *parent;
+ int id;
+} SearchCoreCb;
+
+static int x86_cpu_search_parent_core(CPUTopoState *topo,
+ void *opaque)
+{
+ SearchCoreCb *cb = opaque;
+ CPUTopoLevel level = CPU_TOPO_LEVEL(topo);
+
+ cb->parent = topo;
+ cb->id = x86_cpu_get_topo_id_by_level(cb->cpu, level);
+
+ if (cb->id == topo->index) {
+ if (level == CPU_TOPO_CORE) {
+ return TOPO_FOREACH_END;
+ }
+ return TOPO_FOREACH_CONTINUE;
+ }
+ return TOPO_FOREACH_SIBLING;
+}
+
+Object *x86_cpu_search_parent_pre_plug(CPUTopoState *topo,
+ CPUTopoState *root,
+ Error **errp)
+{
+ int ret;
+ SearchCoreCb cb;
+ X86CPUTopoIDs topo_ids;
+ X86CPUTopoInfo topo_info;
+ X86CPU *cpu = X86_CPU(topo);
+ CPUSlot *slot = CPU_SLOT(root);
+ MachineState *ms = slot->ms;
+ DECLARE_BITMAP(foreach_bitmap, USER_AVAIL_LEVEL_NUM);
+
+ topo_info.dies_per_pkg = ms->smp.dies;
+ topo_info.cores_per_die = ms->smp.cores;
+ topo_info.threads_per_core = ms->smp.threads;
+
+ if (cpu->apic_id == UNASSIGNED_APIC_ID) {
+ x86_cpu_assign_apic_id(ms, cpu, &topo_ids, &topo_info, errp);
+ if (*errp) {
+ return NULL;
+ }
+ } else {
+ /*
+ * if 'address' properties socket-id/core-id/thread-id are not set,
+ * set them so that machine_query_hotpluggable_cpus would show
+ * correct values.
+ *
+ * TODO: move socket_id/core_id/thread_id checks into
+ * x86_cpu_realizefn() once -smp refactoring is complete and there
+ * will be CPU private CPUState::nr_cores and CPUState::nr_threads
+ * fields instead of globals.
+ */
+ x86_topo_ids_from_apicid(cpu->apic_id, &topo_info, &topo_ids);
+ }
+
+ x86_cpu_assign_topo_id(cpu, &topo_ids, errp);
+ if (*errp) {
+ return NULL;
+ }
+
+ cb.cpu = cpu;
+ cb.parent = NULL;
+ cb.id = -1;
+ bitmap_fill(foreach_bitmap, USER_AVAIL_LEVEL_NUM);
+ clear_bit(CPU_TOPO_UNKNOWN, foreach_bitmap);
+ clear_bit(CPU_TOPO_THREAD, foreach_bitmap);
+
+ ret = cpu_topo_child_foreach_recursive(root, foreach_bitmap,
+ x86_cpu_search_parent_core, &cb);
+ if (ret != TOPO_FOREACH_END) {
+ g_autofree char *search_info = NULL;
+
+ search_info = !cb.parent ? g_strdup("") :
+ g_strdup_printf(" for %s level with id: %d",
+ cpu_topo_level_to_string(CPU_TOPO_LEVEL(cb.parent)), cb.id);
+ error_setg(errp, "Can't find parent%s", search_info);
+ return NULL;
+ }
+
+ /* Keep the index of CPU topology device the same as the thread_id. */
+ topo->index = cpu->thread_id;
+ return OBJECT(cb.parent);
+}
+
CpuInstanceProperties
x86_cpu_index_to_props(MachineState *ms, unsigned cpu_index)
{
diff --git a/include/hw/i386/x86.h b/include/hw/i386/x86.h
index 19e9f93fe286..e8c9ddc36359 100644
--- a/include/hw/i386/x86.h
+++ b/include/hw/i386/x86.h
@@ -104,6 +104,9 @@ int64_t x86_get_default_cpu_node_id(const MachineState *ms, int idx);
const CPUArchIdList *x86_possible_cpu_arch_ids(MachineState *ms);
CPUArchId *x86_find_cpu_slot(MachineState *ms, uint32_t id, int *idx);
void x86_rtc_set_cpus_count(ISADevice *rtc, uint16_t cpus_count);
+Object *x86_cpu_search_parent_pre_plug(CPUTopoState *topo,
+ CPUTopoState *root,
+ Error **errp);
void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,
DeviceState *dev, Error **errp);
void x86_cpu_plug(HotplugHandler *hotplug_dev,
--
2.34.1
next prev parent reply other threads:[~2023-11-30 14:41 UTC|newest]
Thread overview: 43+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-11-30 14:41 [RFC 00/41] qom-topo: Abstract Everything about CPU Topology Zhao Liu
2023-11-30 14:41 ` [RFC 01/41] qdev: Introduce new device category to cover basic topology device Zhao Liu
2023-11-30 14:41 ` [RFC 02/41] qdev: Allow qdev_device_add() to add specific category device Zhao Liu
2023-11-30 14:41 ` [RFC 03/41] system: Create base category devices from cli before board initialization Zhao Liu
2023-11-30 14:41 ` [RFC 04/41] qom/object: Introduce helper to resolve path from non-direct parent Zhao Liu
2023-11-30 14:41 ` [RFC 05/41] qdev: Set device parent and id after setting properties Zhao Liu
2023-11-30 14:41 ` [RFC 06/41] qdev: Introduce user-child interface to collect devices from -device Zhao Liu
2023-11-30 14:41 ` [RFC 07/41] qdev: Introduce parent option in -device Zhao Liu
2023-11-30 14:41 ` [RFC 08/41] hw/core/topo: Introduce CPU topology device abstraction Zhao Liu
2023-11-30 14:41 ` [RFC 09/41] hw/core/topo: Support topology index for topology device Zhao Liu
2023-11-30 14:41 ` [RFC 10/41] hw/core/topo: Add virtual method to update topology info for parent Zhao Liu
2023-11-30 14:41 ` [RFC 11/41] hw/core/topo: Add virtual method to check topology child Zhao Liu
2023-11-30 14:41 ` [RFC 12/41] hw/core/topo: Add helpers to traverse the CPU topology tree Zhao Liu
2023-11-30 14:41 ` [RFC 13/41] hw/core/cpu: Convert CPU from general device to topology device Zhao Liu
2023-11-30 14:41 ` [RFC 14/41] PPC/ppc-core: Offload core-id to PPC specific core abstarction Zhao Liu
2023-11-30 14:41 ` [RFC 15/41] hw/cpu/core: Allow to configure plugged threads for cpu-core Zhao Liu
2023-11-30 14:41 ` [RFC 16/41] PPC/ppc-core: Limit plugged-threads and nr-threads to be equal Zhao Liu
2023-11-30 14:41 ` [RFC 17/41] hw/cpu/core: Convert cpu-core from general device to topology device Zhao Liu
2023-11-30 14:41 ` [RFC 18/41] hw/cpu/cluster: Rename CPUClusterState to CPUCluster Zhao Liu
2023-11-30 14:41 ` [RFC 19/41] hw/cpu/cluster: Wrap TCG related ops and props into CONFIG_TCG Zhao Liu
2023-11-30 14:41 ` [RFC 20/41] hw/cpu/cluster: Descript cluster is not only used for TCG in comment Zhao Liu
2023-11-30 14:41 ` [RFC 21/41] hw/cpu/cluster: Allow cpu-cluster to be created by -device Zhao Liu
2023-11-30 14:41 ` [RFC 22/41] hw/cpu/cluster: Convert cpu-cluster from general device to topology device Zhao Liu
2023-11-30 14:41 ` [RFC 23/41] hw/cpu/die: Abstract cpu-die level as " Zhao Liu
2023-11-30 14:41 ` [RFC 24/41] hw/cpu/socket: Abstract cpu-socket " Zhao Liu
2023-11-30 14:41 ` [RFC 25/41] hw/cpu/book: Abstract cpu-book " Zhao Liu
2023-11-30 14:41 ` [RFC 26/41] hw/cpu/drawer: Abstract cpu-drawer " Zhao Liu
2023-11-30 14:41 ` [RFC 27/41] hw/core/slot: Introduce CPU slot as the root of CPU topology Zhao Liu
2023-11-30 14:41 ` [RFC 28/41] hw/core/slot: Maintain the core queue in CPU slot Zhao Liu
2023-11-30 14:41 ` [RFC 29/41] hw/core/slot: Statistics topology information " Zhao Liu
2023-11-30 14:41 ` [RFC 30/41] hw/core/slot: Check topology child to be added under " Zhao Liu
2023-11-30 14:41 ` [RFC 31/41] hw/machine: Plug cpu-slot into machine to maintain topology tree Zhao Liu
2023-11-30 14:41 ` [RFC 32/41] hw/machine: Build smp topology tree from -smp Zhao Liu
2023-11-30 14:41 ` [RFC 33/41] hw/machine: Validate smp topology tree without -smp Zhao Liu
2023-11-30 14:41 ` [RFC 34/41] hw/core/topo: Implement user-child to collect topology device from cli Zhao Liu
2023-11-30 14:41 ` [RFC 35/41] hw/i386: Make x86_cpu_new() private in x86.c Zhao Liu
2023-11-30 14:41 ` [RFC 36/41] hw/i386: Allow x86_cpu_new() to specify parent for new CPU Zhao Liu
2023-11-30 14:41 ` [RFC 37/41] hw/i386: Allow i386 to create new CPUs from QOM topology Zhao Liu
2023-11-30 14:42 ` [RFC 38/41] hw/i386: Wrap apic id and topology sub ids assigning as helpers Zhao Liu
2023-11-30 14:42 ` Zhao Liu [this message]
2023-11-30 14:42 ` [RFC 40/41] hw/i386: Support QOM topology Zhao Liu
2023-11-30 14:42 ` [RFC 41/41] hw/i386: Cleanup non-QOM topology support Zhao Liu
2023-12-11 13:36 ` [RFC 00/41] qom-topo: Abstract Everything about CPU Topology Zhao Liu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20231130144203.2307629-40-zhao1.liu@linux.intel.com \
--to=zhao1.liu@linux.intel.com \
--cc=alex.bennee@linaro.org \
--cc=alistair@alistair23.me \
--cc=anthony.perard@citrix.com \
--cc=berrange@redhat.com \
--cc=bin.meng@windriver.com \
--cc=clg@kaod.org \
--cc=danielhb413@gmail.com \
--cc=david@gibson.dropbear.id.au \
--cc=edgar.iglesias@gmail.com \
--cc=eduardo@habkost.net \
--cc=fbarrat@linux.ibm.com \
--cc=harshpb@linux.ibm.com \
--cc=imammedo@redhat.com \
--cc=jasowang@redhat.com \
--cc=kraxel@redhat.com \
--cc=kvm@vger.kernel.org \
--cc=liwei1518@gmail.com \
--cc=marcel.apfelbaum@gmail.com \
--cc=mst@redhat.com \
--cc=npiggin@gmail.com \
--cc=nsg@linux.ibm.com \
--cc=palmer@dabbelt.com \
--cc=paul@xen.org \
--cc=pbonzini@redhat.com \
--cc=peter.maydell@linaro.org \
--cc=philmd@linaro.org \
--cc=qemu-arm@nongnu.org \
--cc=qemu-devel@nongnu.org \
--cc=qemu-ppc@nongnu.org \
--cc=qemu-riscv@nongnu.org \
--cc=qemu-s390x@nongnu.org \
--cc=richard.henderson@linaro.org \
--cc=sstabellini@kernel.org \
--cc=thuth@redhat.com \
--cc=wangyanan55@huawei.com \
--cc=xen-devel@lists.xenproject.org \
--cc=yongwei.ma@intel.com \
--cc=zhao1.liu@intel.com \
--cc=zhenyu.z.wang@intel.com \
--cc=zhiwei_liu@linux.alibaba.com \
--cc=zhiyuan.lv@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).