[libvirt] [PATCH v4 0/3] numa: deprecate '-numa node, mem' and default memory distribution
Changes since v3: - simplify series by dropping idea of showing property values in "qom-list-properties" and use MachineInfo in QAPI schema instead Changes since v2: - taking in account previous review, implement a way for mgmt to intospect if '-numa node,mem' is supported by machine type as suggested by Daniel at https://www.mail-archive.com/qemu-devel@nongnu.org/msg601220.html * ammend "qom-list-properties" to show property values * add "numa-mem-supported" machine property to reflect if '-numa node,mem=SZ' is supported. It culd be used with '-machine none' or at runtime with --preconfig before numa memory mapping are configured * minor fixes to deprecation documentation mentioning "numa-mem-supported" property 1) "I'm considering to deprecating -mem-path/prealloc CLI options and replacing them with a single memdev Machine property to allow interested users to pick used backend for initial RAM (fixes mixed -mem-path+hostmem backends issues) and as a transition step to modeling initial RAM as a Device instead of (ab)using MemoryRegion APIs." (for more details see: https://www.mail-archive.com/qemu-devel@nongnu.org/msg596314.html) However there is a couple of roadblocks on the way (s390x and numa memory handling). I think I finally thought out a way to hack s390x in migration compatible manner, but I don't see any way to do it for -numa node,mem and default RAM assignement to nodes. Considering both numa usecases aren't meaningfully using NUMA (aside guest side testing) and could be replaced with explicitly used memdev parameter, I'd like to propose removing these fake NUMA friends on new machine types, hence this deprecation. And once the last machie type that supported the option is removed we would be able to remove option altogether. As result of removing deprecated options and replacing initial RAM allocation with 'memdev's (1), QEMU will allocate guest RAM in consistent way, fixing mixed use-case and allowing boards to move towards modelling initial RAM as Device(s). Which in its own turn should allow to cleanup NUMA/HMP/memory accounting code more by dropping ad-hoc node_mem tracking and reusing memory device enumeration instead. Reference to previous versions: * https://www.mail-archive.com/qemu-devel@nongnu.org/msg617694.html CC: libvir-list@redhat.com CC: ehabkost@redhat.com CC: pbonzini@redhat.com CC: berrange@redhat.com CC: armbru@redhat.com Igor Mammedov (3): machine: show if CLI option '-numa node,mem' is supported in QAPI schema numa: deprecate 'mem' parameter of '-numa node' option numa: deprecate implict memory distribution between nodes include/hw/boards.h | 3 +++ hw/arm/virt.c | 1 + hw/i386/pc.c | 1 + hw/ppc/spapr.c | 1 + numa.c | 5 +++++ qapi/misc.json | 5 ++++- qemu-deprecated.texi | 24 ++++++++++++++++++++++++ vl.c | 1 + 8 files changed, 40 insertions(+), 1 deletion(-) -- 2.7.4
Legacy '-numa node,mem' option has a number of issues and mgmt often defaults to it. Unfortunately it's no possible to replace it with an alternative '-numa memdev' without breaking migration compatibility. What's possible though is to deprecate it, keeping option working with old machine types only. In order to help users to find out if being deprecated CLI option '-numa node,mem' is still supported by particular machine type, add new "numa-mem-supported" property to MachineInfo description in QAPI schema. "numa-mem-supported" is set to 'true' for machines that currently support NUMA, but it will be flipped to 'false' later on, once deprecation period expires and kept 'true' only for old machine types that used to support the legacy option so it won't break existing configuration that are using it. Signed-off-by: Igor Mammedov <imammedo@redhat.com> --- Notes: v4: * drop idea to use "qom-list-properties" and use MachineInfo instead which could be inspected with 'query-machines' include/hw/boards.h | 3 +++ hw/arm/virt.c | 1 + hw/i386/pc.c | 1 + hw/ppc/spapr.c | 1 + qapi/misc.json | 5 ++++- vl.c | 1 + 6 files changed, 11 insertions(+), 1 deletion(-) diff --git a/include/hw/boards.h b/include/hw/boards.h index 6f7916f..86894b6 100644 --- a/include/hw/boards.h +++ b/include/hw/boards.h @@ -158,6 +158,8 @@ typedef struct { * @kvm_type: * Return the type of KVM corresponding to the kvm-type string option or * computed based on other criteria such as the host kernel capabilities. + * @numa_mem_supported: + * true if '--numa node.mem' option is supported and false otherwise */ struct MachineClass { /*< private >*/ @@ -210,6 +212,7 @@ struct MachineClass { bool ignore_boot_device_suffixes; bool smbus_no_migration_support; bool nvdimm_supported; + bool numa_mem_supported; HotplugHandler *(*get_hotplug_handler)(MachineState *machine, DeviceState *dev); diff --git a/hw/arm/virt.c b/hw/arm/virt.c index bf54f10..481a603 100644 --- a/hw/arm/virt.c +++ b/hw/arm/virt.c @@ -1943,6 +1943,7 @@ static void virt_machine_class_init(ObjectClass *oc, void *data) assert(!mc->get_hotplug_handler); mc->get_hotplug_handler = virt_machine_get_hotplug_handler; hc->plug = virt_machine_device_plug_cb; + mc->numa_mem_supported = true; } static void virt_instance_init(Object *obj) diff --git a/hw/i386/pc.c b/hw/i386/pc.c index 2632b73..05b8368 100644 --- a/hw/i386/pc.c +++ b/hw/i386/pc.c @@ -2747,6 +2747,7 @@ static void pc_machine_class_init(ObjectClass *oc, void *data) nc->nmi_monitor_handler = x86_nmi; mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE; mc->nvdimm_supported = true; + mc->numa_mem_supported = true; object_class_property_add(oc, PC_MACHINE_DEVMEM_REGION_SIZE, "int", pc_machine_get_device_memory_region_size, NULL, diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c index 2ef3ce4..265ecfb 100644 --- a/hw/ppc/spapr.c +++ b/hw/ppc/spapr.c @@ -4336,6 +4336,7 @@ static void spapr_machine_class_init(ObjectClass *oc, void *data) * in which LMBs are represented and hot-added */ mc->numa_mem_align_shift = 28; + mc->numa_mem_supported = true; smc->default_caps.caps[SPAPR_CAP_HTM] = SPAPR_CAP_OFF; smc->default_caps.caps[SPAPR_CAP_VSX] = SPAPR_CAP_ON; diff --git a/qapi/misc.json b/qapi/misc.json index 8b3ca4f..d0bdccb 100644 --- a/qapi/misc.json +++ b/qapi/misc.json @@ -2018,12 +2018,15 @@ # # @hotpluggable-cpus: cpu hotplug via -device is supported (since 2.7.0) # +# @numa-mem-supported: true if '-numa node,mem' option is supported by machine +# type and false otherwise (since 4.1) +# # Since: 1.2.0 ## { 'struct': 'MachineInfo', 'data': { 'name': 'str', '*alias': 'str', '*is-default': 'bool', 'cpu-max': 'int', - 'hotpluggable-cpus': 'bool'} } + 'hotpluggable-cpus': 'bool', 'numa-mem-supported': 'bool'} } ## # @query-machines: diff --git a/vl.c b/vl.c index 5550bd7..5bf17f5 100644 --- a/vl.c +++ b/vl.c @@ -1520,6 +1520,7 @@ MachineInfoList *qmp_query_machines(Error **errp) info->name = g_strdup(mc->name); info->cpu_max = !mc->max_cpus ? 1 : mc->max_cpus; info->hotpluggable_cpus = mc->has_hotpluggable_cpus; + info->numa_mem_supported = mc->numa_mem_supported; entry = g_malloc0(sizeof(*entry)); entry->value = info; -- 2.7.4
Igor Mammedov <imammedo@redhat.com> writes:
Legacy '-numa node,mem' option has a number of issues and mgmt often defaults to it. Unfortunately it's no possible to replace it with an alternative '-numa memdev' without breaking migration compatibility. What's possible though is to deprecate it, keeping option working with old machine types only.
In order to help users to find out if being deprecated CLI option '-numa node,mem' is still supported by particular machine type, add new "numa-mem-supported" property to MachineInfo description in QAPI schema.
Suggest s/to MachineInfo description in QAPI schema/to output of query-machines/, because query-machines is the external interface people know.
"numa-mem-supported" is set to 'true' for machines that currently support NUMA, but it will be flipped to 'false' later on, once deprecation period expires and kept 'true' only for old machine types that used to support the legacy option so it won't break existing configuration that are using it.
Signed-off-by: Igor Mammedov <imammedo@redhat.com> ---
Notes: v4: * drop idea to use "qom-list-properties" and use MachineInfo instead which could be inspected with 'query-machines'
include/hw/boards.h | 3 +++ hw/arm/virt.c | 1 + hw/i386/pc.c | 1 + hw/ppc/spapr.c | 1 + qapi/misc.json | 5 ++++- vl.c | 1 + 6 files changed, 11 insertions(+), 1 deletion(-)
diff --git a/include/hw/boards.h b/include/hw/boards.h index 6f7916f..86894b6 100644 --- a/include/hw/boards.h +++ b/include/hw/boards.h @@ -158,6 +158,8 @@ typedef struct { * @kvm_type: * Return the type of KVM corresponding to the kvm-type string option or * computed based on other criteria such as the host kernel capabilities. + * @numa_mem_supported: + * true if '--numa node.mem' option is supported and false otherwise */ struct MachineClass { /*< private >*/ @@ -210,6 +212,7 @@ struct MachineClass { bool ignore_boot_device_suffixes; bool smbus_no_migration_support; bool nvdimm_supported; + bool numa_mem_supported;
HotplugHandler *(*get_hotplug_handler)(MachineState *machine, DeviceState *dev); diff --git a/hw/arm/virt.c b/hw/arm/virt.c index bf54f10..481a603 100644 --- a/hw/arm/virt.c +++ b/hw/arm/virt.c @@ -1943,6 +1943,7 @@ static void virt_machine_class_init(ObjectClass *oc, void *data) assert(!mc->get_hotplug_handler); mc->get_hotplug_handler = virt_machine_get_hotplug_handler; hc->plug = virt_machine_device_plug_cb; + mc->numa_mem_supported = true; }
static void virt_instance_init(Object *obj) diff --git a/hw/i386/pc.c b/hw/i386/pc.c index 2632b73..05b8368 100644 --- a/hw/i386/pc.c +++ b/hw/i386/pc.c @@ -2747,6 +2747,7 @@ static void pc_machine_class_init(ObjectClass *oc, void *data) nc->nmi_monitor_handler = x86_nmi; mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE; mc->nvdimm_supported = true; + mc->numa_mem_supported = true;
object_class_property_add(oc, PC_MACHINE_DEVMEM_REGION_SIZE, "int", pc_machine_get_device_memory_region_size, NULL, diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c index 2ef3ce4..265ecfb 100644 --- a/hw/ppc/spapr.c +++ b/hw/ppc/spapr.c @@ -4336,6 +4336,7 @@ static void spapr_machine_class_init(ObjectClass *oc, void *data) * in which LMBs are represented and hot-added */ mc->numa_mem_align_shift = 28; + mc->numa_mem_supported = true;
smc->default_caps.caps[SPAPR_CAP_HTM] = SPAPR_CAP_OFF; smc->default_caps.caps[SPAPR_CAP_VSX] = SPAPR_CAP_ON;
This is correct when the TYPE_VIRT_MACHINE, TYPE_PC_MACHINE and TYPE_SPAPR_MACHINE are exactly the machines supporting NUMA. How could I check that?
diff --git a/qapi/misc.json b/qapi/misc.json index 8b3ca4f..d0bdccb 100644 --- a/qapi/misc.json +++ b/qapi/misc.json @@ -2018,12 +2018,15 @@ # # @hotpluggable-cpus: cpu hotplug via -device is supported (since 2.7.0) # +# @numa-mem-supported: true if '-numa node,mem' option is supported by machine +# type and false otherwise (since 4.1)
"by the machine type", for consistency with @cpu-max. Also, rather long line.
+# # Since: 1.2.0 ## { 'struct': 'MachineInfo', 'data': { 'name': 'str', '*alias': 'str', '*is-default': 'bool', 'cpu-max': 'int', - 'hotpluggable-cpus': 'bool'} } + 'hotpluggable-cpus': 'bool', 'numa-mem-supported': 'bool'} }
## # @query-machines: diff --git a/vl.c b/vl.c index 5550bd7..5bf17f5 100644 --- a/vl.c +++ b/vl.c @@ -1520,6 +1520,7 @@ MachineInfoList *qmp_query_machines(Error **errp) info->name = g_strdup(mc->name); info->cpu_max = !mc->max_cpus ? 1 : mc->max_cpus; info->hotpluggable_cpus = mc->has_hotpluggable_cpus; + info->numa_mem_supported = mc->numa_mem_supported;
entry = g_malloc0(sizeof(*entry)); entry->value = info;
On Fri, Jun 07, 2019 at 07:39:17PM +0200, Markus Armbruster wrote:
This is correct when the TYPE_VIRT_MACHINE, TYPE_PC_MACHINE and TYPE_SPAPR_MACHINE are exactly the machines supporting NUMA. How could I check that?
parse_numa_node() rejects the -numa option if the machine doesn't implement MachineClass::get_default_cpu_node_id(). Grepping for it: $ git grep -pw get_default_cpu_node_id hw/arm/virt.c=static void virt_machine_class_init(ObjectClass *oc, void *data) hw/arm/virt.c: mc->get_default_cpu_node_id = virt_get_default_cpu_node_id; hw/core/machine.c=static void machine_numa_finish_cpu_init(MachineState *machine) hw/core/machine.c: props.node_id = mc->get_default_cpu_node_id(machine, i); hw/i386/pc.c=static void pc_machine_class_init(ObjectClass *oc, void *data) hw/i386/pc.c: mc->get_default_cpu_node_id = pc_get_default_cpu_node_id; hw/ppc/spapr.c=static void spapr_machine_class_init(ObjectClass *oc, void *data) hw/ppc/spapr.c: mc->get_default_cpu_node_id = spapr_get_default_cpu_node_id; include/hw/boards.h=typedef struct { include/hw/boards.h: * @get_default_cpu_node_id: include/hw/boards.h=struct MachineClass { include/hw/boards.h: int64_t (*get_default_cpu_node_id)(const MachineState *ms, int idx); numa.c=static void parse_numa_node(MachineState *ms, NumaNodeOptions *node, numa.c: if (!mc->cpu_index_to_instance_props || !mc->get_default_cpu_node_id) { Related: [PATCH v4 01/11] numa: move numa global variable nb_numa_nodes into MachineState which adds a MachineClass::numa_supported flag to those machines. -- Eduardo
Eduardo Habkost <ehabkost@redhat.com> writes:
On Fri, Jun 07, 2019 at 07:39:17PM +0200, Markus Armbruster wrote:
This is correct when the TYPE_VIRT_MACHINE, TYPE_PC_MACHINE and TYPE_SPAPR_MACHINE are exactly the machines supporting NUMA. How could I check that?
parse_numa_node() rejects the -numa option if the machine doesn't implement MachineClass::get_default_cpu_node_id().
Grepping for it:
$ git grep -pw get_default_cpu_node_id hw/arm/virt.c=static void virt_machine_class_init(ObjectClass *oc, void *data) hw/arm/virt.c: mc->get_default_cpu_node_id = virt_get_default_cpu_node_id; hw/core/machine.c=static void machine_numa_finish_cpu_init(MachineState *machine) hw/core/machine.c: props.node_id = mc->get_default_cpu_node_id(machine, i); hw/i386/pc.c=static void pc_machine_class_init(ObjectClass *oc, void *data) hw/i386/pc.c: mc->get_default_cpu_node_id = pc_get_default_cpu_node_id; hw/ppc/spapr.c=static void spapr_machine_class_init(ObjectClass *oc, void *data) hw/ppc/spapr.c: mc->get_default_cpu_node_id = spapr_get_default_cpu_node_id; include/hw/boards.h=typedef struct { include/hw/boards.h: * @get_default_cpu_node_id: include/hw/boards.h=struct MachineClass { include/hw/boards.h: int64_t (*get_default_cpu_node_id)(const MachineState *ms, int idx); numa.c=static void parse_numa_node(MachineState *ms, NumaNodeOptions *node, numa.c: if (!mc->cpu_index_to_instance_props || !mc->get_default_cpu_node_id) {
Related: [PATCH v4 01/11] numa: move numa global variable nb_numa_nodes into MachineState which adds a MachineClass::numa_supported flag to those machines.
Thanks, Eduardo! Preferably with commit message and doc comment tweaked: Reviewed-by: Markus Armbruster <armbru@redhat.com>
On Fri, 07 Jun 2019 19:39:17 +0200 Markus Armbruster <armbru@redhat.com> wrote:
Igor Mammedov <imammedo@redhat.com> writes:
Legacy '-numa node,mem' option has a number of issues and mgmt often defaults to it. Unfortunately it's no possible to replace it with an alternative '-numa memdev' without breaking migration compatibility. What's possible though is to deprecate it, keeping option working with old machine types only.
In order to help users to find out if being deprecated CLI option '-numa node,mem' is still supported by particular machine type, add new "numa-mem-supported" property to MachineInfo description in QAPI schema.
Suggest s/to MachineInfo description in QAPI schema/to output of query-machines/, because query-machines is the external interface people know.
fixed
"numa-mem-supported" is set to 'true' for machines that currently support NUMA, but it will be flipped to 'false' later on, once deprecation period expires and kept 'true' only for old machine types that used to support the legacy option so it won't break existing configuration that are using it.
Signed-off-by: Igor Mammedov <imammedo@redhat.com> ---
Notes: v4: * drop idea to use "qom-list-properties" and use MachineInfo instead which could be inspected with 'query-machines'
include/hw/boards.h | 3 +++ hw/arm/virt.c | 1 + hw/i386/pc.c | 1 + hw/ppc/spapr.c | 1 + qapi/misc.json | 5 ++++- vl.c | 1 + 6 files changed, 11 insertions(+), 1 deletion(-)
diff --git a/include/hw/boards.h b/include/hw/boards.h index 6f7916f..86894b6 100644 --- a/include/hw/boards.h +++ b/include/hw/boards.h @@ -158,6 +158,8 @@ typedef struct { * @kvm_type: * Return the type of KVM corresponding to the kvm-type string option or * computed based on other criteria such as the host kernel capabilities. + * @numa_mem_supported: + * true if '--numa node.mem' option is supported and false otherwise */ struct MachineClass { /*< private >*/ @@ -210,6 +212,7 @@ struct MachineClass { bool ignore_boot_device_suffixes; bool smbus_no_migration_support; bool nvdimm_supported; + bool numa_mem_supported;
HotplugHandler *(*get_hotplug_handler)(MachineState *machine, DeviceState *dev); diff --git a/hw/arm/virt.c b/hw/arm/virt.c index bf54f10..481a603 100644 --- a/hw/arm/virt.c +++ b/hw/arm/virt.c @@ -1943,6 +1943,7 @@ static void virt_machine_class_init(ObjectClass *oc, void *data) assert(!mc->get_hotplug_handler); mc->get_hotplug_handler = virt_machine_get_hotplug_handler; hc->plug = virt_machine_device_plug_cb; + mc->numa_mem_supported = true; }
static void virt_instance_init(Object *obj) diff --git a/hw/i386/pc.c b/hw/i386/pc.c index 2632b73..05b8368 100644 --- a/hw/i386/pc.c +++ b/hw/i386/pc.c @@ -2747,6 +2747,7 @@ static void pc_machine_class_init(ObjectClass *oc, void *data) nc->nmi_monitor_handler = x86_nmi; mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE; mc->nvdimm_supported = true; + mc->numa_mem_supported = true;
object_class_property_add(oc, PC_MACHINE_DEVMEM_REGION_SIZE, "int", pc_machine_get_device_memory_region_size, NULL, diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c index 2ef3ce4..265ecfb 100644 --- a/hw/ppc/spapr.c +++ b/hw/ppc/spapr.c @@ -4336,6 +4336,7 @@ static void spapr_machine_class_init(ObjectClass *oc, void *data) * in which LMBs are represented and hot-added */ mc->numa_mem_align_shift = 28; + mc->numa_mem_supported = true;
smc->default_caps.caps[SPAPR_CAP_HTM] = SPAPR_CAP_OFF; smc->default_caps.caps[SPAPR_CAP_VSX] = SPAPR_CAP_ON;
This is correct when the TYPE_VIRT_MACHINE, TYPE_PC_MACHINE and TYPE_SPAPR_MACHINE are exactly the machines supporting NUMA. How could I check that?
We don't have an interface to communicate that to users as far as I know. Eduardo pointed out to the patch that might serve as starting point for it though.
diff --git a/qapi/misc.json b/qapi/misc.json index 8b3ca4f..d0bdccb 100644 --- a/qapi/misc.json +++ b/qapi/misc.json @@ -2018,12 +2018,15 @@ # # @hotpluggable-cpus: cpu hotplug via -device is supported (since 2.7.0) # +# @numa-mem-supported: true if '-numa node,mem' option is supported by machine +# type and false otherwise (since 4.1)
"by the machine type", for consistency with @cpu-max. Also, rather long line.
fixed
+# # Since: 1.2.0 ## { 'struct': 'MachineInfo', 'data': { 'name': 'str', '*alias': 'str', '*is-default': 'bool', 'cpu-max': 'int', - 'hotpluggable-cpus': 'bool'} } + 'hotpluggable-cpus': 'bool', 'numa-mem-supported': 'bool'} }
## # @query-machines: diff --git a/vl.c b/vl.c index 5550bd7..5bf17f5 100644 --- a/vl.c +++ b/vl.c @@ -1520,6 +1520,7 @@ MachineInfoList *qmp_query_machines(Error **errp) info->name = g_strdup(mc->name); info->cpu_max = !mc->max_cpus ? 1 : mc->max_cpus; info->hotpluggable_cpus = mc->has_hotpluggable_cpus; + info->numa_mem_supported = mc->numa_mem_supported;
entry = g_malloc0(sizeof(*entry)); entry->value = info;
Legacy '-numa node,mem' option has a number of issues and mgmt often defaults to it. Unfortunately it's no possible to replace it with an alternative '-numa memdev' without breaking migration compatibility. What's possible though is to deprecate it, keeping option working with old machine types only. In order to help users to find out if being deprecated CLI option '-numa node,mem' is still supported by particular machine type, add new "numa-mem-supported" property to output of query-machines. "numa-mem-supported" is set to 'true' for machines that currently support NUMA, but it will be flipped to 'false' later on, once deprecation period expires and kept 'true' only for old machine types that used to support the legacy option so it won't break existing configuration that are using it. Signed-off-by: Igor Mammedov <imammedo@redhat.com> --- v5: (Markus Armbruster <armbru@redhat.com>) * s/by machine type/by the machine type/ * ammend commit message s/to MachineInfo description in QAPI schema/to output of query-machines/ v4: * drop idea to use "qom-list-properties" and use MachineInfo instead which could be inspected with 'query-machines' include/hw/boards.h | 3 +++ hw/arm/virt.c | 1 + hw/i386/pc.c | 1 + hw/ppc/spapr.c | 1 + qapi/misc.json | 5 ++++- vl.c | 1 + 6 files changed, 11 insertions(+), 1 deletion(-) diff --git a/include/hw/boards.h b/include/hw/boards.h index 6ff02bf..ab6badc 100644 --- a/include/hw/boards.h +++ b/include/hw/boards.h @@ -158,6 +158,8 @@ typedef struct { * @kvm_type: * Return the type of KVM corresponding to the kvm-type string option or * computed based on other criteria such as the host kernel capabilities. + * @numa_mem_supported: + * true if '--numa node.mem' option is supported and false otherwise */ struct MachineClass { /*< private >*/ @@ -210,6 +212,7 @@ struct MachineClass { bool ignore_boot_device_suffixes; bool smbus_no_migration_support; bool nvdimm_supported; + bool numa_mem_supported; HotplugHandler *(*get_hotplug_handler)(MachineState *machine, DeviceState *dev); diff --git a/hw/arm/virt.c b/hw/arm/virt.c index bf54f10..481a603 100644 --- a/hw/arm/virt.c +++ b/hw/arm/virt.c @@ -1943,6 +1943,7 @@ static void virt_machine_class_init(ObjectClass *oc, void *data) assert(!mc->get_hotplug_handler); mc->get_hotplug_handler = virt_machine_get_hotplug_handler; hc->plug = virt_machine_device_plug_cb; + mc->numa_mem_supported = true; } static void virt_instance_init(Object *obj) diff --git a/hw/i386/pc.c b/hw/i386/pc.c index edc240b..25146d7 100644 --- a/hw/i386/pc.c +++ b/hw/i386/pc.c @@ -2750,6 +2750,7 @@ static void pc_machine_class_init(ObjectClass *oc, void *data) nc->nmi_monitor_handler = x86_nmi; mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE; mc->nvdimm_supported = true; + mc->numa_mem_supported = true; object_class_property_add(oc, PC_MACHINE_DEVMEM_REGION_SIZE, "int", pc_machine_get_device_memory_region_size, NULL, diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c index e2b33e5..89d5814 100644 --- a/hw/ppc/spapr.c +++ b/hw/ppc/spapr.c @@ -4340,6 +4340,7 @@ static void spapr_machine_class_init(ObjectClass *oc, void *data) * in which LMBs are represented and hot-added */ mc->numa_mem_align_shift = 28; + mc->numa_mem_supported = true; smc->default_caps.caps[SPAPR_CAP_HTM] = SPAPR_CAP_OFF; smc->default_caps.caps[SPAPR_CAP_VSX] = SPAPR_CAP_ON; diff --git a/qapi/misc.json b/qapi/misc.json index 8b3ca4f..2dbfdf0 100644 --- a/qapi/misc.json +++ b/qapi/misc.json @@ -2018,12 +2018,15 @@ # # @hotpluggable-cpus: cpu hotplug via -device is supported (since 2.7.0) # +# @numa-mem-supported: true if '-numa node,mem' option is supported by +# the machine type and false otherwise (since 4.1) +# # Since: 1.2.0 ## { 'struct': 'MachineInfo', 'data': { 'name': 'str', '*alias': 'str', '*is-default': 'bool', 'cpu-max': 'int', - 'hotpluggable-cpus': 'bool'} } + 'hotpluggable-cpus': 'bool', 'numa-mem-supported': 'bool'} } ## # @query-machines: diff --git a/vl.c b/vl.c index cd1fbc4..f5b083f 100644 --- a/vl.c +++ b/vl.c @@ -1428,6 +1428,7 @@ MachineInfoList *qmp_query_machines(Error **errp) info->name = g_strdup(mc->name); info->cpu_max = !mc->max_cpus ? 1 : mc->max_cpus; info->hotpluggable_cpus = mc->has_hotpluggable_cpus; + info->numa_mem_supported = mc->numa_mem_supported; entry = g_malloc0(sizeof(*entry)); entry->value = info; -- 2.7.4
Igor Mammedov <imammedo@redhat.com> writes:
Legacy '-numa node,mem' option has a number of issues and mgmt often defaults to it. Unfortunately it's no possible to replace it with an alternative '-numa memdev' without breaking migration compatibility. What's possible though is to deprecate it, keeping option working with old machine types only.
In order to help users to find out if being deprecated CLI option '-numa node,mem' is still supported by particular machine type, add new "numa-mem-supported" property to output of query-machines.
"numa-mem-supported" is set to 'true' for machines that currently support NUMA, but it will be flipped to 'false' later on, once deprecation period expires and kept 'true' only for old machine types that used to support the legacy option so it won't break existing configuration that are using it.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
The parameter allows to configure fake NUMA topology where guest VM simulates NUMA topology but not actually getting performance benefits from it. The same or better results could be achieved using 'memdev' parameter. Beside of unpredictable performance, '-numa node.mem' option has other issues when it's used with combination of -mem-path + + -mem-prealloc + memdev backends (pc-dimm), breaking binding of memdev backends since mem-path/mem-prealloc are global and affect the most of RAM allocations. It's possible to make memdevs and global -mem-path/mem-prealloc to play nicely together but that will just complicate already complicated code and add unobious ways it could break on 2 different memmory allocation pathes and their combinations. Instead of it, consolidate all guest RAM allocation over memdev which still allows to create fake NUMA configurations if desired and leaves one simplifyed code path to consider when it comes to guest RAM allocation. To achieve desired simplification deprecate 'mem' parameter as its ad-hoc partitioning of initial RAM MemoryRegion can't be translated to memdev based backend transparently to users and in compatible manner (migration wise). Later down the road that will allow to consolidate means of how guest RAM is allocated and would permit us to clean up quite a bit memory allocations and numa code, leaving only 'memdev' implementation in place. Signed-off-by: Igor Mammedov <imammedo@redhat.com> --- Notes: v4: * fix up documentation to mention where users should look to check if -numa node.mem is supported numa.c | 2 ++ qemu-deprecated.texi | 16 ++++++++++++++++ 2 files changed, 18 insertions(+) diff --git a/numa.c b/numa.c index 3875e1e..2205773 100644 --- a/numa.c +++ b/numa.c @@ -121,6 +121,8 @@ static void parse_numa_node(MachineState *ms, NumaNodeOptions *node, if (node->has_mem) { numa_info[nodenr].node_mem = node->mem; + warn_report("Parameter -numa node,mem is deprecated," + " use -numa node,memdev instead"); } if (node->has_memdev) { Object *o; diff --git a/qemu-deprecated.texi b/qemu-deprecated.texi index 50292d8..eb347f5 100644 --- a/qemu-deprecated.texi +++ b/qemu-deprecated.texi @@ -82,6 +82,22 @@ The @code{-realtime mlock=on|off} argument has been replaced by the The ``-virtfs_synth'' argument is now deprecated. Please use ``-fsdev synth'' and ``-device virtio-9p-...'' instead. +@subsection -numa node,mem=@var{size} (since 4.1) + +The parameter @option{mem} of @option{-numa node} is used to assign a part of +guest RAM to a NUMA node. But when using it, it's impossible to manage specified +RAM chunk on the host side (like bind it to a host node, setting bind policy, ...), +so guest end-ups with the fake NUMA configuration with suboptiomal performance. +However since 2014 there is an alternative way to assign RAM to a NUMA node +using parameter @option{memdev}, which does the same as @option{mem} and adds +means to actualy manage node RAM on the host side. Use parameter @option{memdev} +with @var{memory-backend-ram} backend as an replacement for parameter @option{mem} +to achieve the same fake NUMA effect or a properly configured +@var{memory-backend-file} backend to actually benefit from NUMA configuration. +In future new machine versions will not accept the option but it will still +work with old machine types. User can check QAPI schema to see if the legacy +option is supported by looking at MachineInfo::numa-mem-supported property. + @section QEMU Machine Protocol (QMP) commands @subsection block-dirty-bitmap-add "autoload" parameter (since 2.12.0) -- 2.7.4
Implicit RAM distribution between nodes has exactly the same issues as: "numa: deprecate 'mem' parameter of '-numa node' option" only with QEMU being the user that's 'adding' 'mem' parameter. Deprecate it, to get it out of the way so that we could consolidate guest RAM allocation using memory backends making it consistent and possibly later on transition to using memory devices instead of adhoc memory mapping for the initial RAM. Signed-off-by: Igor Mammedov <imammedo@redhat.com> --- numa.c | 3 +++ qemu-deprecated.texi | 8 ++++++++ 2 files changed, 11 insertions(+) diff --git a/numa.c b/numa.c index 2205773..6d45a1f 100644 --- a/numa.c +++ b/numa.c @@ -409,6 +409,9 @@ void numa_complete_configuration(MachineState *ms) if (i == nb_numa_nodes) { assert(mc->numa_auto_assign_ram); mc->numa_auto_assign_ram(mc, numa_info, nb_numa_nodes, ram_size); + warn_report("Default splitting of RAM between nodes is deprecated," + " Use '-numa node,memdev' to explictly define RAM" + " allocation per node"); } numa_total = 0; diff --git a/qemu-deprecated.texi b/qemu-deprecated.texi index eb347f5..c744ba9 100644 --- a/qemu-deprecated.texi +++ b/qemu-deprecated.texi @@ -98,6 +98,14 @@ In future new machine versions will not accept the option but it will still work with old machine types. User can check QAPI schema to see if the legacy option is supported by looking at MachineInfo::numa-mem-supported property. +@subsection -numa node (without memory specified) (since 4.1) + +Splitting RAM by default between NUMA nodes has the same issues as @option{mem} +parameter described above with the difference that the role of the user plays +QEMU using implicit generic or board specific splitting rule. +Use @option{memdev} with @var{memory-backend-ram} backend or @option{mem} (if +it's supported by used machine type) to define mapping explictly instead. + @section QEMU Machine Protocol (QMP) commands @subsection block-dirty-bitmap-add "autoload" parameter (since 2.12.0) -- 2.7.4
To summarize implications on libvirt... This patch currently doesn't change the deprecation status of any machine type, so it is effectively no-op metadata addition right now. Libvirt version N can implement support for this today, and nothing would change from a functional POV with QEMU 4.1. At some later time, a QEMU release X.Y will mark a new machine type has having this deprecation enabled, at which point the dormant libvirt logic will activate, causing libvirt to use the new syntax for NUMA. IOW, libvirt version N onwards can live migrate QEMU with machine type pc-X.Y to a host with arbitrary future libvirt or QEMU version. Libvirt version N-1 can *not* live migrate QEMU with machine type pc-X.Y to a host with libvirt version N or newer. This restriction won't be a problem in practice if there's a moderate time window between libvirt implementing support for this concept, and the future QEMU X.Y which actually enables the deprecation. The main downside of this approach is the one Eduardo raised previously which is that the dormant libvirt logic will not be testable/tested until that QEMU X.Y actually arrives. The testing gap is not nice, but I don't think it is sufficient to reject this approach entirely. I wonder if there's a way to close the testing gap somehow ? Random idea would be a non-versioned "pc-no-deprecated" machine type, which blocks all use of deprecated features and does not promise any migration compat. Essentially it would exist just for testing purposem as a way todo functional tests for libvirt & mgmt apps to prove they don't use any deprecated features (any deprecated features, not merely this NUMA one). On Thu, May 30, 2019 at 10:33:16AM +0200, Igor Mammedov wrote:
Changes since v3: - simplify series by dropping idea of showing property values in "qom-list-properties" and use MachineInfo in QAPI schema instead
Changes since v2: - taking in account previous review, implement a way for mgmt to intospect if '-numa node,mem' is supported by machine type as suggested by Daniel at https://www.mail-archive.com/qemu-devel@nongnu.org/msg601220.html * ammend "qom-list-properties" to show property values * add "numa-mem-supported" machine property to reflect if '-numa node,mem=SZ' is supported. It culd be used with '-machine none' or at runtime with --preconfig before numa memory mapping are configured * minor fixes to deprecation documentation mentioning "numa-mem-supported" property
1) "I'm considering to deprecating -mem-path/prealloc CLI options and replacing them with a single memdev Machine property to allow interested users to pick used backend for initial RAM (fixes mixed -mem-path+hostmem backends issues) and as a transition step to modeling initial RAM as a Device instead of (ab)using MemoryRegion APIs." (for more details see: https://www.mail-archive.com/qemu-devel@nongnu.org/msg596314.html)
However there is a couple of roadblocks on the way (s390x and numa memory handling). I think I finally thought out a way to hack s390x in migration compatible manner, but I don't see any way to do it for -numa node,mem and default RAM assignement to nodes. Considering both numa usecases aren't meaningfully using NUMA (aside guest side testing) and could be replaced with explicitly used memdev parameter, I'd like to propose removing these fake NUMA friends on new machine types, hence this deprecation. And once the last machie type that supported the option is removed we would be able to remove option altogether.
As result of removing deprecated options and replacing initial RAM allocation with 'memdev's (1), QEMU will allocate guest RAM in consistent way, fixing mixed use-case and allowing boards to move towards modelling initial RAM as Device(s). Which in its own turn should allow to cleanup NUMA/HMP/memory accounting code more by dropping ad-hoc node_mem tracking and reusing memory device enumeration instead.
Reference to previous versions: * https://www.mail-archive.com/qemu-devel@nongnu.org/msg617694.html
CC: libvir-list@redhat.com CC: ehabkost@redhat.com CC: pbonzini@redhat.com CC: berrange@redhat.com CC: armbru@redhat.com
Igor Mammedov (3): machine: show if CLI option '-numa node,mem' is supported in QAPI schema numa: deprecate 'mem' parameter of '-numa node' option numa: deprecate implict memory distribution between nodes
include/hw/boards.h | 3 +++ hw/arm/virt.c | 1 + hw/i386/pc.c | 1 + hw/ppc/spapr.c | 1 + numa.c | 5 +++++ qapi/misc.json | 5 ++++- qemu-deprecated.texi | 24 ++++++++++++++++++++++++ vl.c | 1 + 8 files changed, 40 insertions(+), 1 deletion(-)
-- 2.7.4
Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
On Wed, Jun 05, 2019 at 06:33:11PM +0100, Daniel P. Berrangé wrote: [...]
I wonder if there's a way to close the testing gap somehow ? Random idea would be a non-versioned "pc-no-deprecated" machine type, which blocks all use of deprecated features and does not promise any migration compat. Essentially it would exist just for testing purposem as a way todo functional tests for libvirt & mgmt apps to prove they don't use any deprecated features (any deprecated features, not merely this NUMA one).
This isn't the first time I wish we had a machine type with experimental features enabled. What about calling it "pc-next"? -- Eduardo
On Wed, Jun 05, 2019 at 03:06:08PM -0300, Eduardo Habkost wrote:
On Wed, Jun 05, 2019 at 06:33:11PM +0100, Daniel P. Berrangé wrote: [...]
I wonder if there's a way to close the testing gap somehow ? Random idea would be a non-versioned "pc-no-deprecated" machine type, which blocks all use of deprecated features and does not promise any migration compat. Essentially it would exist just for testing purposem as a way todo functional tests for libvirt & mgmt apps to prove they don't use any deprecated features (any deprecated features, not merely this NUMA one).
This isn't the first time I wish we had a machine type with experimental features enabled. What about calling it "pc-next"?
No objection from me. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Igor Mammedov <imammedo@redhat.com> writes:
Changes since v3: - simplify series by dropping idea of showing property values in "qom-list-properties" and use MachineInfo in QAPI schema instead
Where did "[PATCH v3 1/6] pc: fix possible NULL pointer dereference in pc_machine_get_device_memory_region_size()" go? It fixes a crash bug...
On Fri, 07 Jun 2019 19:28:58 +0200 Markus Armbruster <armbru@redhat.com> wrote:
Igor Mammedov <imammedo@redhat.com> writes:
Changes since v3: - simplify series by dropping idea of showing property values in "qom-list-properties" and use MachineInfo in QAPI schema instead
Where did "[PATCH v3 1/6] pc: fix possible NULL pointer dereference in pc_machine_get_device_memory_region_size()" go? It fixes a crash bug...
I'll post it as separate patch as it's not more related to this series
On Thu, 30 May 2019 10:33:16 +0200 Igor Mammedov <imammedo@redhat.com> wrote:
Changes since v3: - simplify series by dropping idea of showing property values in "qom-list-properties" and use MachineInfo in QAPI schema instead
Changes since v2: - taking in account previous review, implement a way for mgmt to intospect if '-numa node,mem' is supported by machine type as suggested by Daniel at https://www.mail-archive.com/qemu-devel@nongnu.org/msg601220.html * ammend "qom-list-properties" to show property values * add "numa-mem-supported" machine property to reflect if '-numa node,mem=SZ' is supported. It culd be used with '-machine none' or at runtime with --preconfig before numa memory mapping are configured * minor fixes to deprecation documentation mentioning "numa-mem-supported" property
1) "I'm considering to deprecating -mem-path/prealloc CLI options and replacing them with a single memdev Machine property to allow interested users to pick used backend for initial RAM (fixes mixed -mem-path+hostmem backends issues) and as a transition step to modeling initial RAM as a Device instead of (ab)using MemoryRegion APIs." (for more details see: https://www.mail-archive.com/qemu-devel@nongnu.org/msg596314.html)
However there is a couple of roadblocks on the way (s390x and numa memory handling). I think I finally thought out a way to hack s390x in migration compatible manner, but I don't see any way to do it for -numa node,mem and default RAM assignement to nodes. Considering both numa usecases aren't meaningfully using NUMA (aside guest side testing) and could be replaced with explicitly used memdev parameter, I'd like to propose removing these fake NUMA friends on new machine types, hence this deprecation. And once the last machie type that supported the option is removed we would be able to remove option altogether.
As result of removing deprecated options and replacing initial RAM allocation with 'memdev's (1), QEMU will allocate guest RAM in consistent way, fixing mixed use-case and allowing boards to move towards modelling initial RAM as Device(s). Which in its own turn should allow to cleanup NUMA/HMP/memory accounting code more by dropping ad-hoc node_mem tracking and reusing memory device enumeration instead.
Eduardo, could you take and merge it via numa/machine tree?
Reference to previous versions: * https://www.mail-archive.com/qemu-devel@nongnu.org/msg617694.html
CC: libvir-list@redhat.com CC: ehabkost@redhat.com CC: pbonzini@redhat.com CC: berrange@redhat.com CC: armbru@redhat.com
Igor Mammedov (3): machine: show if CLI option '-numa node,mem' is supported in QAPI schema numa: deprecate 'mem' parameter of '-numa node' option numa: deprecate implict memory distribution between nodes
include/hw/boards.h | 3 +++ hw/arm/virt.c | 1 + hw/i386/pc.c | 1 + hw/ppc/spapr.c | 1 + numa.c | 5 +++++ qapi/misc.json | 5 ++++- qemu-deprecated.texi | 24 ++++++++++++++++++++++++ vl.c | 1 + 8 files changed, 40 insertions(+), 1 deletion(-)
participants (4)
-
Daniel P. Berrangé -
Eduardo Habkost -
Igor Mammedov -
Markus Armbruster