Build ACPI Heterogeneous Memory Attribute Table (HMAT)
v20 cover Build ACPI Heterogeneous Memory Attribute Table (HMAT)
This series of patches will build Heterogeneous Memory Attribute Table (HMAT) according to the command line. The ACPI HMAT describes the memory attributes, such as memory side cache attributes and bandwidth and latency details, related to the Memory Proximity Domain.
The software is expected to use HMAT information as hint for optimization.
In the linux kernel, the codes in drivers/acpi/hmat/hmat.c parse and report the platform's HMAT tables.
The V19 patches link:
https://patchwork.kernel.org/cover/11265525/
Changelog:
v20:
- Use g_assert_true and g_assert_false to replace g_assert (Thomas and Markus)
- Rename assoc as associativity, update the QAPI description (Markus)
- Disable cache level 0 in hmat-cache option (Igor)
- Keep base and bitmap unchanged when latency or bandwidth out of range
- Fix the broken CI case when user input latency or bandwidth less than required.
v19:
- Add description about the machine property 'hmat' in commit message (Markus)
- Update the QAPI comments
- Add a check for no memory side cache
- Add some fail cases for hmat-cache when level=0
v18:
- Defer patches 01/14~06/14 of V17, use qapi type uint64 and only nanosecond for latency (Markus)
- Rewrite the lines over 80 characters(Igor)
v17:
- Add check when user input latency or bandwidth 0, the lb_info_provided should also be 0. Because in ACPI 6.3 5.2.27.4, 0 means the corresponding latency or bandwidth information is not provided.
- Fix the infinite loop when node->latency is 0.
- Use NumaHmatCacheOptions to replace HMAT_Cache_Info (Igor)
- Add check for unordered cache level input (Igor)
- Add some fail test cases (Igor)
v16:
- Add and use qemu_strtold_finite to parse size, support full 64bit precision, modify related test cases (Eduardo and Markus)
- Simplify struct HMAT_LB_Info and related code, unify latency and bandwidth (Igor)
- Add cross check with hmat_lb data (Igor)
- Fields in Cache Attributes are promoted to uint32_t before shifting (Igor)
- Add case for QMP build HMAT (Igor)
v15:
- Add a new patch to refactor do_strtosz() (Eduardo)
- Make tests without breaking CI (Michael)
v14:
- Reuse the codes of do_strtosz to build qemu_strtotime_ns (Eduardo)
- Squash patch v13 01/12 and 02/12 together (Daniel and Eduardo)
- Drop time unit picosecond (Eric)
- Use qemu ctz64 and clz64 instead of builtin function
v13:
- Modify some text description
- Drop "initiator_valid" field in struct NodeInfo
- Reuse Garray to store the raw bandwidth and bandwidth data
- Calculate common base unit using range bitmap
- Add a patch to alculate hmat latency and bandwidth entry list
- Drop the total_levels option and use readable cache size
- Remove the unnecessary head file
- Use decimal notation with appropriate suffix for cache size
Liu Jingqi (5):
numa: Extend CLI to provide memory latency and bandwidth information
numa: Extend CLI to provide memory side cache information hmat acpi: Build Memory Proximity Domain Attributes Structure(s) hmat acpi: Build System Locality Latency and Bandwidth Information
Structure(s)
hmat acpi: Build Memory Side Cache Information Structure(s)
Tao Xu (3):
numa: Extend CLI to provide initiator information for numa nodes tests/numa: Add case for QMP build HMAT tests/bios-tables-test: add test cases for ACPI HMAT
hw/acpi/Kconfig | 7 +- hw/acpi/Makefile.objs | 1 + hw/acpi/hmat.c | 268 +++++++++++++++++++++++ hw/acpi/hmat.h | 42 ++++ hw/core/machine.c | 64 ++++++ hw/core/numa.c | 297 ++++++++++++++++++++++++++ hw/i386/acpi-build.c | 5 + include/sysemu/numa.h | 63 ++++++ qapi/machine.json | 180 +++++++++++++++- qemu-options.hx | 95 +++++++- tests/bios-tables-test-allowed-diff.h | 8 + tests/bios-tables-test.c | 44 ++++ tests/data/acpi/pc/APIC.acpihmat | 0 tests/data/acpi/pc/DSDT.acpihmat | 0 tests/data/acpi/pc/HMAT.acpihmat | 0 tests/data/acpi/pc/SRAT.acpihmat | 0 tests/data/acpi/q35/APIC.acpihmat | 0 tests/data/acpi/q35/DSDT.acpihmat | 0 tests/data/acpi/q35/HMAT.acpihmat | 0 tests/data/acpi/q35/SRAT.acpihmat | 0 tests/numa-test.c | 213 ++++++++++++++++++ 21 files changed, 1276 insertions(+), 11 deletions(-) create mode 100644 hw/acpi/hmat.c create mode 100644 hw/acpi/hmat.h create mode 100644 tests/data/acpi/pc/APIC.acpihmat create mode 100644 tests/data/acpi/pc/DSDT.acpihmat create mode 100644 tests/data/acpi/pc/HMAT.acpihmat create mode 100644 tests/data/acpi/pc/SRAT.acpihmat create mode 100644 tests/data/acpi/q35/APIC.acpihmat create mode 100644 tests/data/acpi/q35/DSDT.acpihmat create mode 100644 tests/data/acpi/q35/HMAT.acpihmat create mode 100644 tests/data/acpi/q35/SRAT.acpihmat
Signature · 1 line
2.20.1
PATCH v20 1/8 numa: Extend CLI to provide initiator information for numa nodes
In ACPI 6.3 chapter 5.2.27 Heterogeneous Memory Attribute Table (HMAT), The initiator represents processor which access to memory. And in 5.2.27.3 Memory Proximity Domain Attributes Structure, the attached initiator is defined as where the memory controller responsible for a memory proximity domain. With attached initiator information, the topology of heterogeneous memory can be described. Add new machine property 'hmat' to enable all HMAT specific options.
Extend CLI of "-numa node" option to indicate the initiator numa node-id. In the linux kernel, the codes in drivers/acpi/hmat/hmat.c parse and report the platform's HMAT tables. Before using initiator option, enable HMAT with -machine hmat=on.
---
No changes in v20.
Changes in v19:
- Add description about the machine property 'hmat' in commit message (Markus)
Changes in v15:
- Change the QAPI version tag to 5.0 (Eric)
--- hw/core/machine.c | 64 +++++++++++++++++++++++++++++++++++++++++++ hw/core/numa.c | 23 ++++++++++++++++ include/sysemu/numa.h | 5 ++++ qapi/machine.json | 10 ++++++- qemu-options.hx | 35 +++++++++++++++++++---- 5 files changed, 131 insertions(+), 6 deletions(-)
hw/core/machine.c +64 -0
@@ -518,6 +518,20 @@ static void machine_set_nvdimm(Object *obj, bool value, Error **errp) ms->nvdimms_state->is_enabled = value; } +static bool machine_get_hmat (Object *obj,Error **errp) +{ +MachineState *ms =MACHINE (obj); + +return ms->numa_state->hmat_enabled; +} + +static void machine_set_hmat (Object *obj,bool value,Error **errp) +{ +MachineState *ms =MACHINE (obj); + + ms->numa_state->hmat_enabled = value; +} +static char *machine_get_nvdimm_persistence (Object *obj,Error **errp) {MachineState *ms =MACHINE (obj);
@@ -645,6 +659,7 @@ void machine_set_cpu_numa_node(MachineState *machine, const CpuInstanceProperties *props, Error **errp) { MachineClass *mc = MACHINE_GET_CLASS(machine); +NodeInfo *numa_info = machine->numa_state->nodes;bool match = false;int i;
@@ -714,6 +729,17 @@ void machine_set_cpu_numa_node(MachineState *machine, match = true; slot->props.node_id = props->node_id; slot->props.has_node_id = props->has_node_id; + +if (machine->numa_state->hmat_enabled) { +if ((numa_info[props->node_id].initiator <MAX_NODES ) && + (props->node_id != numa_info[props->node_id].initiator)) { +error_setg (errp,"The initiator of CPU NUMA node %" PRId64 +" should be itself" , props->node_id); +return ; + } + numa_info[props->node_id].has_cpu = true; + numa_info[props->node_id].initiator = props->node_id; + } }if (!match) {
@@ -960,6 +986,13 @@ static void machine_initfn(Object *obj)if (mc->numa_mem_supported) { ms->numa_state =g_new0 (NumaState,1 ); +object_property_add_bool (obj,"hmat" , + machine_get_hmat, machine_set_hmat, + &error_abort); +object_property_set_description (obj,"hmat" , +"Set on/off to enable/disable " +"ACPI Heterogeneous Memory Attribute " +"Table (HMAT)" ,NULL ); }/* Register notifier when init is done for sysbus sanity checks */
@@ -1048,6 +1081,32 @@ static char *cpu_slot_to_string(const CPUArchId *cpu)return g_string_free (s, false); } +static void numa_validate_initiator (NumaState *numa_state) +{ +int i; +NodeInfo *numa_info = numa_state->nodes; + +for (i =0 ; i < numa_state->num_nodes; i++) { +if (numa_info[i].initiator ==MAX_NODES ) { +error_report ("The initiator of NUMA node %d is missing, use " +"'-numa node,initiator' option to declare it" , i); +exit (1 ); + } + +if (!numa_info[numa_info[i].initiator].present) { +error_report ("NUMA node %" PRIu16" is missing, use " +"'-numa node' option to declare it first" , + numa_info[i].initiator); +exit (1 ); + } + +if (!numa_info[numa_info[i].initiator].has_cpu) { +error_report ("The initiator of NUMA node %d is invalid" , i); +exit (1 ); + } + } +} +static void machine_numa_finish_cpu_init (MachineState *machine) {int i;
@@ -1088,6 +1147,11 @@ static void machine_numa_finish_cpu_init(MachineState *machine)machine_set_cpu_numa_node (machine, &props, &error_fatal); } } + +if (machine->numa_state->hmat_enabled) { +numa_validate_initiator (machine->numa_state); + } +if (s->len && !qtest_enabled ()) {warn_report ("CPU(s) not present in any NUMA nodes: %s" , s->str);
hw/core/numa.c +23 -0
@@ -133,6 +133,29 @@ static void parse_numa_node(MachineState *ms, NumaNodeOptions *node, numa_info[nodenr].node_mem =object_property_get_uint (o,"size" ,NULL ); numa_info[nodenr].node_memdev =MEMORY_BACKEND (o); } + +/* +* If not set the initiator, set it to MAX_NODES. And if +* HMAT is enabled and this node has no cpus, QEMU will raise error. +*/ + numa_info[nodenr].initiator =MAX_NODES ; +if (node->has_initiator) { +if (!ms->numa_state->hmat_enabled) { +error_setg (errp,"ACPI Heterogeneous Memory Attribute Table " +"(HMAT) is disabled, enable it with -machine hmat=on " +"before using any of hmat specific options" ); +return ; + } + +if (node->initiator >=MAX_NODES ) { +error_report ("The initiator id %" PRIu16" expects an integer " +"between 0 and %d" , node->initiator, +MAX_NODES -1 ); +return ; + } + + numa_info[nodenr].initiator = node->initiator; + } numa_info[nodenr].present = true; max_numa_nodeid =MAX (max_numa_nodeid, nodenr +1 ); ms->numa_state->num_nodes++;
include/sysemu/numa.h +5 -0
@@ -18,6 +18,8 @@ struct NodeInfo {uint64_t node_mem;struct HostMemoryBackend *node_memdev;bool present; +bool has_cpu; +uint16_t initiator;uint8_t distance[MAX_NODES ]; };
@@ -33,6 +35,9 @@ struct NumaState {/* Allow setting NUMA distance for different NUMA nodes */ bool have_numa_distance; +/* Detect if HMAT support is enabled. */ +bool hmat_enabled; +/* NUMA nodes information */ NodeInfo nodes[MAX_NODES ]; };
qapi/machine.json +9 -1
@@ -463,6 +463,13 @@ # @memdev: memory backend object. If specified for one node, # it must be specified for all nodes. # +# @initiator: defined in ACPI 6.3 Chapter 5.2.27.3 Table 5-145, +# points to the nodeid which has the memory controller +# responsible for this NUMA node. This field provides +# additional information as to the initiator node that +# is closest (as in directly attached) to this node, and +# therefore has the best performance (since 5.0) +# # Since: 2.1 ## { 'struct': 'NumaNodeOptions',
@@ -470,7 +477,8 @@ '*nodeid': 'uint16', '*cpus': ['uint16'], '*mem': 'size', - '*memdev': 'str' }} + '*memdev': 'str', + '*initiator': 'uint16' }} ## # @NumaDistOptions:
qemu-options.hx +30 -5
@@ -43,7 +43,8 @@ DEF("machine", HAS_ARG, QEMU_OPTION_machine, \ " suppress-vmdesc=on|off disables self-describing migration (default=off)\n" " nvdimm=on|off controls NVDIMM support (default=off)\n" " enforce-config-section=on|off enforce configuration section migration (default=off)\n" - " memory-encryption=@var{} memory encryption object to use (default=none)\n", + " memory-encryption=@var{} memory encryption object to use (default=none)\n" + " hmat=on|off controls ACPI HMAT support (default=off)\n", QEMU_ARCH_ALL) STEXI @item -machine [type=]@var{name}[,prop=@var{value}[,...]]
@@ -103,6 +104,9 @@ NOTE: this parameter is deprecated. Please use @option{-global} @option{migration.send-configuration}=@var{on|off} instead. @item memory-encryption=@var{} Memory encryption object to use. The default is none. +@item hmat=on|off +Enables or disables ACPI Heterogeneous Memory Attribute Table (HMAT) support. +The default is off. @end table ETEXI
@@ -161,14 +165,14 @@ If any on the three values is given, the total number of CPUs @var{n} can be omi ETEXI DEF("numa", HAS_ARG, QEMU_OPTION_numa, - "-numa node[,mem=size][,cpus=firstcpu[-lastcpu]][,nodeid=node]\n" - "-numa node[,memdev=id][,cpus=firstcpu[-lastcpu]][,nodeid=node]\n" + "-numa node[,mem=size][,cpus=firstcpu[-lastcpu]][,nodeid=node][,initiator=node]\n" + "-numa node[,memdev=id][,cpus=firstcpu[-lastcpu]][,nodeid=node][,initiator=node]\n" "-numa dist,src=source,dst=destination,val=distance\n" "-numa cpu,node-id=node[,socket-id=x][,core-id=y][,thread-id=z]\n", QEMU_ARCH_ALL) STEXI -@item -numa node[,mem=@var{size}][,cpus=@var{firstcpu}[-@var{lastcpu}]][,nodeid=@var{node}] -@itemx -numa node[,memdev=@var{id}][,cpus=@var{firstcpu}[-@var{lastcpu}]][,nodeid=@var{node}] +@item -numa node[,mem=@var{size}][,cpus=@var{firstcpu}[-@var{lastcpu}]][,nodeid=@var{node}][,initiator=@var{initiator}] +@itemx -numa node[,memdev=@var{id}][,cpus=@var{firstcpu}[-@var{lastcpu}]][,nodeid=@var{node}][,initiator=@var{initiator}] @itemx -numa dist,src=@var{source},dst=@var{destination},val=@var{distance} @itemx -numa cpu,node-id=@var{node}[,socket-id=@var{x}][,core-id=@var{y}][,thread-id=@var{z}] @findex -numa
@@ -215,6 +219,27 @@ split equally between them. @samp{mem} and @samp{memdev} are mutually exclusive. Furthermore, if one node uses @samp{memdev}, all of them have to use it. +@samp{initiator} is an additional option that points to an @var{initiator} +NUMA node that has best performance (the lowest latency or largest bandwidth) +to this NUMA @var{node}. Note that this option can be set only when +the machine property 'hmat' is set to 'on'. + +Following example creates a machine with 2 NUMA nodes, node 0 has CPU. +node 1 has only memory, and its initiator is node 0. Note that because +node 0 has CPU, by default the initiator of node 0 is itself and must be +itself. +@example +-machine hmat=on \ +-m 2G,slots=2,maxmem=4G \ +-object memory-backend-ram,size=1G,id=m0 \ +-object memory-backend-ram,size=1G,id=m1 \ +-numa node,nodeid=0,memdev=m0 \ +-numa node,nodeid=1,memdev=m1,initiator=0 \ +-smp 2,sockets=2,maxcpus=2 \ +-numa cpu,node-id=0,socket-id=0 \ +-numa cpu,node-id=0,socket-id=1 +@end example + @var{source} and @var{destination} are NUMA node IDs. @var{distance} is the NUMA distance from @var{source} to @var{destination}. The distance from a node to itself is always 10. If any pair of nodes is
Signature · 1 line
2.20.1
PATCH v20 2/8 numa: Extend CLI to provide memory latency and bandwidth information
In reply to Tao XuFrom: Liu Jingqi <[email protected]>
Add -numa hmat-lb option to provide System Locality Latency and Bandwidth Information. These memory attributes help to build System Locality Latency and Bandwidth Information Structure(s) in ACPI Heterogeneous Memory Attribute Table (HMAT). Before using hmat-lb option, enable HMAT with -machine hmat=on.
---
Changes in v20:
- Update the QAPI description (Markus)
- Keep base and bitmap unchanged when latency or bandwidth out of range
Changes in v19:
- Add description about the machine property 'hmat' in commit message (Markus)
Changes in v18:
- Use qapi type uint64 and only nanosecond for latency (Markus)
Changes in v17:
- Add check when user input latency or bandwidth 0, the lb_info_provided should also be 0. Because in ACPI 6.3 5.2.27.4, 0 means the corresponding latency or bandwidth information is not provided.
- Fix the infinite loop when node->latency is 0.
--- hw/core/numa.c | 194 ++++++++++++++++++++++++++++++++++++++++++ include/sysemu/numa.h | 53 ++++++++++++ qapi/machine.json | 93 +++++++++++++++++++- qemu-options.hx | 47 +++++++++- 4 files changed, 384 insertions(+), 3 deletions(-)
hw/core/numa.c +194 -0
@@ -23,6 +23,7 @@*/ #include "qemu/osdep.h" +#include "qemu/units.h" #include "sysemu/hostmem.h" #include "sysemu/numa.h" #include "sysemu/sysemu.h"
@@ -198,6 +199,186 @@ void parse_numa_distance(MachineState *ms, NumaDistOptions *dist, Error **errp) ms->numa_state->have_numa_distance = true; } +void parse_numa_hmat_lb (NumaState *numa_state,NumaHmatLBOptions *node, +Error **errp) +{ +int i, first_bit, last_bit; +uint64_t max_entry, temp_base, bitmap_copy; +NodeInfo *numa_info = numa_state->nodes; +HMAT_LB_Info *hmat_lb = + numa_state->hmat_lb[node->hierarchy][node->data_type]; +HMAT_LB_Data lb_data = {}; +HMAT_LB_Data *lb_temp; + +/* Error checking */ +if (node->initiator > numa_state->num_nodes) { +error_setg (errp,"Invalid initiator=%d, it should be less than %d" , + node->initiator, numa_state->num_nodes); +return ; + } +if (node->target > numa_state->num_nodes) { +error_setg (errp,"Invalid target=%d, it should be less than %d" , + node->target, numa_state->num_nodes); +return ; + } +if (!numa_info[node->initiator].has_cpu) { +error_setg (errp,"Invalid initiator=%d, it isn't an " +"initiator proximity domain" , node->initiator); +return ; + } +if (!numa_info[node->target].present) { +error_setg (errp,"The target=%d should point to an existing node" , + node->target); +return ; + } + +if (!hmat_lb) { + hmat_lb =g_malloc0 (sizeof (*hmat_lb)); + numa_state->hmat_lb[node->hierarchy][node->data_type] = hmat_lb; + hmat_lb->list =g_array_new (false, true,sizeof (HMAT_LB_Data)); + } + hmat_lb->hierarchy = node->hierarchy; + hmat_lb->data_type = node->data_type; + lb_data.initiator = node->initiator; + lb_data.target = node->target; + +if (node->data_type <=HMATLB_DATA_TYPE_WRITE_LATENCY ) { +/* Input latency data */ + +if (!node->has_latency) { +error_setg (errp,"Missing 'latency' option" ); +return ; + } +if (node->has_bandwidth) { +error_setg (errp,"Invalid option 'bandwidth' since " +"the data type is latency" ); +return ; + } + +/* Detect duplicate configuration */ +for (i =0 ; i < hmat_lb->list->len; i++) { + lb_temp = &g_array_index (hmat_lb->list, HMAT_LB_Data, i); + +if (node->initiator == lb_temp->initiator && + node->target == lb_temp->target) { +error_setg (errp,"Duplicate configuration of the latency for " +"initiator=%d and target=%d" , node->initiator, + node->target); +return ; + } + } + + hmat_lb->base = hmat_lb->base ? hmat_lb->base :UINT64_MAX ; + +if (node->latency) { +/* Calculate the temporary base and compressed latency */ + max_entry = node->latency; + temp_base =1 ; +while (QEMU_IS_ALIGNED (max_entry,10 )) { + max_entry /=10 ; + temp_base *=10 ; + } + +/* Calculate the max compressed latency */ + temp_base =MIN (hmat_lb->base, temp_base); + max_entry = node->latency / hmat_lb->base; + max_entry =MAX (hmat_lb->range_bitmap, max_entry); + +/* +* For latency hmat_lb->range_bitmap record the max compressed +* latency which should be less than 0xFFFF (UINT16_MAX) +*/ +if (max_entry >=UINT16_MAX ) { +error_setg (errp,"Latency %" PRIu64" between initiator=%d and " +"target=%d should not differ from previously entered " +"min or max values on more than %d" , node->latency, + node->initiator, node->target,UINT16_MAX -1 ); +return ; + }else { + hmat_lb->base = temp_base; + hmat_lb->range_bitmap = max_entry; + } + +/* +* Set lb_info_provided bit 0 as 1, +* latency information is provided +*/ + numa_info[node->target].lb_info_provided |=BIT (0 ); + } + lb_data.data = node->latency; + }else if (node->data_type >=HMATLB_DATA_TYPE_ACCESS_BANDWIDTH ) { +/* Input bandwidth data */ +if (!node->has_bandwidth) { +error_setg (errp,"Missing 'bandwidth' option" ); +return ; + } +if (node->has_latency) { +error_setg (errp,"Invalid option 'latency' since " +"the data type is bandwidth" ); +return ; + } +if (!QEMU_IS_ALIGNED (node->bandwidth, MiB)) { +error_setg (errp,"Bandwidth %" PRIu64" between initiator=%d and " +"target=%d should be 1MB aligned" , node->bandwidth, + node->initiator, node->target); +return ; + } + +/* Detect duplicate configuration */ +for (i =0 ; i < hmat_lb->list->len; i++) { + lb_temp = &g_array_index (hmat_lb->list, HMAT_LB_Data, i); + +if (node->initiator == lb_temp->initiator && + node->target == lb_temp->target) { +error_setg (errp,"Duplicate configuration of the bandwidth for " +"initiator=%d and target=%d" , node->initiator, + node->target); +return ; + } + } + + hmat_lb->base = hmat_lb->base ? hmat_lb->base :1 ; + +if (node->bandwidth) { +/* Keep bitmap unchanged when bandwidth out of range */ + bitmap_copy = hmat_lb->range_bitmap; + bitmap_copy |= node->bandwidth; + first_bit =ctz64 (bitmap_copy); + temp_base =UINT64_C (1 ) << first_bit; + max_entry = node->bandwidth / temp_base; + last_bit =64 -clz64 (bitmap_copy); + +/* +* For bandwidth, first_bit record the base unit of bandwidth bits, +* last_bit record the last bit of the max bandwidth. The max +* compressed bandwidth should be less than 0xFFFF (UINT16_MAX) +*/ +if ((last_bit - first_bit) >UINT16_BITS || + max_entry >=UINT16_MAX ) { +error_setg (errp,"Bandwidth %" PRIu64" between initiator=%d " +"and target=%d should not differ from previously " +"entered values on more than %d" , node->bandwidth, + node->initiator, node->target,UINT16_MAX -1 ); +return ; + }else { + hmat_lb->base = temp_base; + hmat_lb->range_bitmap = bitmap_copy; + } + +/* +* Set lb_info_provided bit 1 as 1, +* bandwidth information is provided +*/ + numa_info[node->target].lb_info_provided |=BIT (1 ); + } + lb_data.data = node->bandwidth; + }else { +assert (0 ); + } + +g_array_append_val (hmat_lb->list, lb_data); +} +void set_numa_options (MachineState *ms,NumaOptions *object,Error **errp) {Error *err =NULL ;
@@ -236,6 +417,19 @@ void set_numa_options(MachineState *ms, NumaOptions *object, Error **errp)machine_set_cpu_numa_node (ms,qapi_NumaCpuOptions_base (&object->u.cpu), &err);break ; +case NUMA_OPTIONS_TYPE_HMAT_LB : +if (!ms->numa_state->hmat_enabled) { +error_setg (errp,"ACPI Heterogeneous Memory Attribute Table " +"(HMAT) is disabled, enable it with -machine hmat=on " +"before using any of hmat specific options" ); +return ; + } + +parse_numa_hmat_lb (ms->numa_state, &object->u.hmat_lb, &err); +if (err) { + goto end; + } +break ;default :abort (); }
include/sysemu/numa.h +53 -0
81 lines not on this page · raw message
qapi/machine.json +91 -2
114 lines not on this page · raw message
qemu-options.hx +46 -1
68 lines not on this page · raw message
Signature · 1 line
2.20.1
Re: [PATCH v20 2/8] numa: Extend CLI to provide memory latency and bandwidth information
Quoted 11 lines · On Fri, 29 Nov 2019 15:56:28 +0800 Tao Xu <[email protected]> wrote:
From: Liu Jingqi <[email protected]>
Add -numa hmat-lb option to provide System Locality Latency and Bandwidth Information. These memory attributes help to build System Locality Latency and Bandwidth Information Structure(s) in ACPI Heterogeneous Memory Attribute Table (HMAT). Before using hmat-lb option, enable HMAT with -machine hmat=on.
Signed-off-by Liu Jingqi <[email protected]>Signed-off-by Tao Xu <[email protected]>
Quoted 529 lines
---
Changes in v20:
- Update the QAPI description (Markus)
- Keep base and bitmap unchanged when latency or bandwidth out of rangeChanges in v19:
- Add description about the machine property 'hmat' in commit message (Markus)Changes in v18:
- Use qapi type uint64 and only nanosecond for latency (Markus)Changes in v17:
- Add check when user input latency or bandwidth 0, thelb_info_providedshould also be 0. Because in ACPI 6.3 5.2.27.4, 0 means the corresponding latency or bandwidth information is not provided.
- Fix the infinite loop whennode->latencyis 0.--- hw/core/numa.c | 194 ++++++++++++++++++++++++++++++++++++++++++ include/sysemu/numa.h | 53 ++++++++++++ qapi/machine.json | 93 +++++++++++++++++++- qemu-options.hx | 47 +++++++++- 4 files changed, 384 insertions(+), 3 deletions(-)hw/core/numa.c +194
-0@@ -23,6 +23,7 @@*/ #include "qemu/osdep.h" +#include "qemu/units.h" #include "sysemu/hostmem.h" #include "sysemu/numa.h" #include "sysemu/sysemu.h" @@ -198,6 +199,186 @@ void parse_numa_distance(MachineState *ms, NumaDistOptions *dist, Error **errp) ms->numa_state->have_numa_distance = true; } +void parse_numa_hmat_lb (NumaState *numa_state,NumaHmatLBOptions *node, +Error **errp) +{ +int i, first_bit, last_bit; +uint64_t max_entry, temp_base, bitmap_copy; +NodeInfo *numa_info = numa_state->nodes; +HMAT_LB_Info *hmat_lb = + numa_state->hmat_lb[node->hierarchy][node->data_type]; +HMAT_LB_Data lb_data = {}; +HMAT_LB_Data *lb_temp; + +/* Error checking */ +if (node->initiator > numa_state->num_nodes) { +error_setg (errp,"Invalid initiator=%d, it should be less than %d" , + node->initiator, numa_state->num_nodes); +return ; + } +if (node->target > numa_state->num_nodes) { +error_setg (errp,"Invalid target=%d, it should be less than %d" , + node->target, numa_state->num_nodes); +return ; + } +if (!numa_info[node->initiator].has_cpu) { +error_setg (errp,"Invalid initiator=%d, it isn't an " +"initiator proximity domain" , node->initiator); +return ; + } +if (!numa_info[node->target].present) { +error_setg (errp,"The target=%d should point to an existing node" , + node->target); +return ; + } + +if (!hmat_lb) { + hmat_lb =g_malloc0 (sizeof (*hmat_lb)); + numa_state->hmat_lb[node->hierarchy][node->data_type] = hmat_lb; + hmat_lb->list =g_array_new (false, true,sizeof (HMAT_LB_Data)); + } + hmat_lb->hierarchy = node->hierarchy; + hmat_lb->data_type = node->data_type; + lb_data.initiator = node->initiator; + lb_data.target = node->target; + +if (node->data_type <=HMATLB_DATA_TYPE_WRITE_LATENCY ) { +/* Input latency data */ + +if (!node->has_latency) { +error_setg (errp,"Missing 'latency' option" ); +return ; + } +if (node->has_bandwidth) { +error_setg (errp,"Invalid option 'bandwidth' since " +"the data type is latency" ); +return ; + } + +/* Detect duplicate configuration */ +for (i =0 ; i < hmat_lb->list->len; i++) { + lb_temp = &g_array_index (hmat_lb->list, HMAT_LB_Data, i); + +if (node->initiator == lb_temp->initiator && + node->target == lb_temp->target) { +error_setg (errp,"Duplicate configuration of the latency for " +"initiator=%d and target=%d" , node->initiator, + node->target); +return ; + } + } + + hmat_lb->base = hmat_lb->base ? hmat_lb->base :UINT64_MAX ; + +if (node->latency) { +/* Calculate the temporary base and compressed latency */ + max_entry = node->latency; + temp_base =1 ; +while (QEMU_IS_ALIGNED (max_entry,10 )) { + max_entry /=10 ; + temp_base *=10 ; + } + +/* Calculate the max compressed latency */ + temp_base =MIN (hmat_lb->base, temp_base); + max_entry = node->latency / hmat_lb->base; + max_entry =MAX (hmat_lb->range_bitmap, max_entry); + +/* +* For latency hmat_lb->range_bitmap record the max compressed +* latency which should be less than 0xFFFF (UINT16_MAX) +*/ +if (max_entry >=UINT16_MAX ) { +error_setg (errp,"Latency %" PRIu64" between initiator=%d and " +"target=%d should not differ from previously entered " +"min or max values on more than %d" , node->latency, + node->initiator, node->target,UINT16_MAX -1 ); +return ; + }else { + hmat_lb->base = temp_base; + hmat_lb->range_bitmap = max_entry; + } + +/* +* Set lb_info_provided bit 0 as 1, +* latency information is provided +*/ + numa_info[node->target].lb_info_provided |=BIT (0 ); + } + lb_data.data = node->latency; + }else if (node->data_type >=HMATLB_DATA_TYPE_ACCESS_BANDWIDTH ) { +/* Input bandwidth data */ +if (!node->has_bandwidth) { +error_setg (errp,"Missing 'bandwidth' option" ); +return ; + } +if (node->has_latency) { +error_setg (errp,"Invalid option 'latency' since " +"the data type is bandwidth" ); +return ; + } +if (!QEMU_IS_ALIGNED (node->bandwidth, MiB)) { +error_setg (errp,"Bandwidth %" PRIu64" between initiator=%d and " +"target=%d should be 1MB aligned" , node->bandwidth, + node->initiator, node->target); +return ; + } + +/* Detect duplicate configuration */ +for (i =0 ; i < hmat_lb->list->len; i++) { + lb_temp = &g_array_index (hmat_lb->list, HMAT_LB_Data, i); + +if (node->initiator == lb_temp->initiator && + node->target == lb_temp->target) { +error_setg (errp,"Duplicate configuration of the bandwidth for " +"initiator=%d and target=%d" , node->initiator, + node->target); +return ; + } + } + + hmat_lb->base = hmat_lb->base ? hmat_lb->base :1 ; + +if (node->bandwidth) { +/* Keep bitmap unchanged when bandwidth out of range */ + bitmap_copy = hmat_lb->range_bitmap; + bitmap_copy |= node->bandwidth; + first_bit =ctz64 (bitmap_copy); + temp_base =UINT64_C (1 ) << first_bit; + max_entry = node->bandwidth / temp_base; + last_bit =64 -clz64 (bitmap_copy); + +/* +* For bandwidth, first_bit record the base unit of bandwidth bits, +* last_bit record the last bit of the max bandwidth. The max +* compressed bandwidth should be less than 0xFFFF (UINT16_MAX) +*/ +if ((last_bit - first_bit) >UINT16_BITS || + max_entry >=UINT16_MAX ) { +error_setg (errp,"Bandwidth %" PRIu64" between initiator=%d " +"and target=%d should not differ from previously " +"entered values on more than %d" , node->bandwidth, + node->initiator, node->target,UINT16_MAX -1 ); +return ; + }else { + hmat_lb->base = temp_base; + hmat_lb->range_bitmap = bitmap_copy; + } + +/* +* Set lb_info_provided bit 1 as 1, +* bandwidth information is provided +*/ + numa_info[node->target].lb_info_provided |=BIT (1 ); + } + lb_data.data = node->bandwidth; + }else { +assert (0 ); + } + +g_array_append_val (hmat_lb->list, lb_data); +} +void set_numa_options (MachineState *ms,NumaOptions *object,Error **errp) {Error *err =NULL ;@@ -236,6 +417,19 @@ void set_numa_options(MachineState *ms, NumaOptions *object, Error **errp)machine_set_cpu_numa_node (ms,qapi_NumaCpuOptions_base (&object->u.cpu), &err);break ; +case NUMA_OPTIONS_TYPE_HMAT_LB : +if (!ms->numa_state->hmat_enabled) { +error_setg (errp,"ACPI Heterogeneous Memory Attribute Table " +"(HMAT) is disabled, enable it with -machine hmat=on " +"before using any of hmat specific options" ); +return ; + } + +parse_numa_hmat_lb (ms->numa_state, &object->u.hmat_lb, &err); +if (err) { + goto end; + } +break ;default :abort (); }include/sysemu/numa.h +53
-0@@ -14,11 +14,34 @@ struct CPUArchId;#define NUMA_DISTANCE_MAX 254#define NUMA_DISTANCE_UNREACHABLE 255 +/* the value of AcpiHmatLBInfo flags */ +enum { +HMAT_LB_MEM_MEMORY =0 , +HMAT_LB_MEM_CACHE_1ST_LEVEL =1 , +HMAT_LB_MEM_CACHE_2ND_LEVEL =2 , +HMAT_LB_MEM_CACHE_3RD_LEVEL =3 , +HMAT_LB_LEVELS /* must be the last entry */ +}; + +/* the value of AcpiHmatLBInfo data type */ +enum { +HMAT_LB_DATA_ACCESS_LATENCY =0 , +HMAT_LB_DATA_READ_LATENCY =1 , +HMAT_LB_DATA_WRITE_LATENCY =2 , +HMAT_LB_DATA_ACCESS_BANDWIDTH =3 , +HMAT_LB_DATA_READ_BANDWIDTH =4 , +HMAT_LB_DATA_WRITE_BANDWIDTH =5 , +HMAT_LB_TYPES /* must be the last entry */ +}; + +#define UINT16_BITS 16 +struct NodeInfo {uint64_t node_mem;struct HostMemoryBackend *node_memdev;bool present;bool has_cpu; +uint8_t lb_info_provided;uint16_t initiator;uint8_t distance[MAX_NODES ]; };@@ -28,6 +51,31 @@ struct NumaNodeMem {uint64_t node_plugged_mem; }; +struct HMAT_LB_Data { +uint8_t initiator; +uint8_t target; +uint64_t data; +}; +typedef struct HMAT_LB_Data HMAT_LB_Data ; + +struct HMAT_LB_Info { +/* Indicates it's memory or the specified level memory side cache. */ +uint8_t hierarchy; + +/* Present the type of data, access/read/write latency or bandwidth. */ +uint8_t data_type; + +/* The range bitmap of bandwidth for calculating common base */ +uint64_t range_bitmap; + +/* The common base unit for latencies or bandwidths */ +uint64_t base; + +/* Array to store the latencies or bandwidths */ +GArray *list; +}; +typedef struct HMAT_LB_Info HMAT_LB_Info ; +struct NumaState {/* Number of NUMA nodes */ int num_nodes;@@ -40,11 +88,16 @@ struct NumaState {/* NUMA nodes information */ NodeInfo nodes[MAX_NODES ]; + +/* NUMA nodes HMAT Locality Latency and Bandwidth Information */ +HMAT_LB_Info *hmat_lb[HMAT_LB_LEVELS ][HMAT_LB_TYPES ]; };typedef struct NumaState NumaState ;void set_numa_options (MachineState *ms,NumaOptions *object,Error **errp);void parse_numa_opts (MachineState *ms); +void parse_numa_hmat_lb (NumaState *numa_state,NumaHmatLBOptions *node, +Error **errp);void numa_complete_configuration (MachineState *ms);void query_numa_node_mem (NumaNodeMem node_mem[],MachineState *ms);extern QemuOptsList qemu_numa_opts;qapi/machine.json +91
-2@@ -426,10 +426,12 @@ # # @cpu: property based CPU(s) to node mapping (Since: 2.10) # +# @hmat-lb: memory latency and bandwidth information (Since: 5.0) +# # Since: 2.1 ## { 'enum': 'NumaOptionsType', - 'data': [ 'node', 'dist', 'cpu' ] } + 'data': [ 'node', 'dist', 'cpu', 'hmat-lb' ] } ## # @NumaOptions:@@ -444,7 +446,8 @@ 'data': { 'node': 'NumaNodeOptions', 'dist': 'NumaDistOptions', - 'cpu': 'NumaCpuOptions' }} + 'cpu': 'NumaCpuOptions', + 'hmat-lb': 'NumaHmatLBOptions' }} ## # @NumaNodeOptions:@@ -557,6 +560,92 @@ 'base': 'CpuInstanceProperties', 'data' : {} } +## +# @HmatLBMemoryHierarchy: +# +# The memory hierarchy in the System Locality Latency and Bandwidth +# Information Structure of HMAT (Heterogeneous Memory Attribute Table) +# +# For more information about @HmatLBMemoryHierarchy, see chapter +# 5.2.27.4: Table 5-146: Field "Flags" of ACPI 6.3 spec. +# +# @memory: the structure represents the memory performance +# +# @first-level: first level of memory side cache +# +# @second-level: second level of memory side cache +# +# @third-level: third level of memory side cache +# +# Since: 5.0 +## +{ 'enum': 'HmatLBMemoryHierarchy', + 'data': [ 'memory', 'first-level', 'second-level', 'third-level' ] } + +## +# @HmatLBDataType: +# +# Data type in the System Locality Latency and Bandwidth +# Information Structure of HMAT (Heterogeneous Memory Attribute Table) +# +# For more information about @HmatLBDataType, see chapter +# 5.2.27.4: Table 5-146: Field "Data Type" of ACPI 6.3 spec. +# +# @access-latency: access latency (nanoseconds) +# +# @read-latency: read latency (nanoseconds) +# +# @write-latency: write latency (nanoseconds) +# +# @access-bandwidth: access bandwidth (Bytes per second) +# +# @read-bandwidth: read bandwidth (Bytes per second) +# +# @write-bandwidth: write bandwidth (Bytes per second) +# +# Since: 5.0 +## +{ 'enum': 'HmatLBDataType', + 'data': [ 'access-latency', 'read-latency', 'write-latency', + 'access-bandwidth', 'read-bandwidth', 'write-bandwidth' ] } + +## +# @NumaHmatLBOptions: +# +# Set the system locality latency and bandwidth information +# between Initiator and Target proximity Domains. +# +# For more information about @NumaHmatLBOptions, see chapter +# 5.2.27.4: Table 5-146 of ACPI 6.3 spec. +# +# @initiator: the Initiator Proximity Domain. +# +# @target: the Target Proximity Domain. +# +# @hierarchy: the Memory Hierarchy. Indicates the performance +# of memory or side cache. +# +# @data-type: presents the type of data, access/read/write +# latency or hit latency. +# +# @latency: the value of latency from @initiator to @target +# proximity domain, the latency unit is "ns(nanosecond)". +# +# @bandwidth: the value of bandwidth between @initiator and @target +# proximity domain, the bandwidth unit is +# "Bytes per second". +# +# Since: 5.0 +## +{ 'struct': 'NumaHmatLBOptions', + 'data': { + 'initiator': 'uint16', + 'target': 'uint16', + 'hierarchy': 'HmatLBMemoryHierarchy', + 'data-type': 'HmatLBDataType', + '*latency': 'uint64', + '*bandwidth': 'size' }} + ## # @HostMemPolicy: #qemu-options.hx +46
-1@@ -168,16 +168,19 @@ DEF("numa", HAS_ARG, QEMU_OPTION_numa, "-numa node[,mem=size][,cpus=firstcpu[-lastcpu]][,nodeid=node][,initiator=node]\n" "-numa node[,memdev=id][,cpus=firstcpu[-lastcpu]][,nodeid=node][,initiator=node]\n" "-numa dist,src=source,dst=destination,val=distance\n" - "-numa cpu,node-id=node[,socket-id=x][,core-id=y][,thread-id=z]\n", + "-numa cpu,node-id=node[,socket-id=x][,core-id=y][,thread-id=z]\n" + "-numa hmat-lb,initiator=node,target=node,hierarchy=memory|first-level|second-level|third-level,data-type=access-latency|read-latency|write-latency[,latency=lat][,bandwidth=bw]\n", QEMU_ARCH_ALL) STEXI @item -numa node[,mem=@var{size}][,cpus=@var{firstcpu}[-@var{lastcpu}]][,nodeid=@var{node}][,initiator=@var{initiator}] @itemx -numa node[,memdev=@var{id}][,cpus=@var{firstcpu}[-@var{lastcpu}]][,nodeid=@var{node}][,initiator=@var{initiator}] @itemx -numa dist,src=@var{source},dst=@var{destination},val=@var{distance} @itemx -numa cpu,node-id=@var{node}[,socket-id=@var{x}][,core-id=@var{y}][,thread-id=@var{z}] +@itemx -numa hmat-lb,initiator=@var{node},target=@var{node},hierarchy=@var{hierarchy},data-type=@var{tpye}[,latency=@var{lat}][,bandwidth=@var{bw}] @findex -numa Define a NUMA node and assign RAM and VCPUs to it. Set the NUMA distance from a source node to a destination node. +Set the ACPI Heterogeneous Memory Attributes for the given nodes. Legacy VCPU assignment uses @samp{cpus} option where @var{firstcpu} and @var{lastcpu} are CPU indexes. Each@@ -256,6 +259,48 @@ specified resources, it just assigns existing resources to NUMA nodes. This means that one still has to use the @option{-m}, @option{-smp} options to allocate RAM and VCPUs respectively. +Use @samp{hmat-lb} to set System Locality Latency and Bandwidth Information +between initiator and target NUMA nodes in ACPI Heterogeneous Attribute Memory Table (HMAT). +Initiator NUMA node can create memory requests, usually it has one or more processors. +Target NUMA node contains addressable memory. + +In @samp{hmat-lb} option, @var{node} are NUMA node IDs. @var{hierarchy} is the memory +hierarchy of the target NUMA node: if @var{hierarchy} is 'memory', the structure +represents the memory performance; if @var{hierarchy} is 'first-level|second-level|third-level', +this structure represents aggregated performance of memory side caches for each domain. +@var{type} of 'data-type' is type of data represented by this structure instance: +if 'hierarchy' is 'memory', 'data-type' is 'access|read|write' latency or 'access|read|write' +bandwidth of the target memory; if 'hierarchy' is 'first-level|second-level|third-level', +'data-type' is 'access|read|write' hit latency or 'access|read|write' hit bandwidth of the +target memory side cache. + +@var{lat} is latency value in nanoseconds. @var{bw} is bandwidth value, +the possible value and units are NUM[M|G|T], mean that the bandwidth value are +NUM byte per second (or MB/s, GB/s or TB/s depending on used suffix). +Note that if latency or bandwidth value is 0, means the corresponding latency or +bandwidth information is not provided. + +For example, the following options describe 2 NUMA nodes. Node 0 has 2 cpus and +a ram, node 1 has only a ram. The processors in node 0 access memory in node +0 with access-latency 5 nanoseconds, access-bandwidth is 200 MB/s; +The processors in NUMA node 0 access memory in NUMA node 1 with access-latency 10 +nanoseconds, access-bandwidth is 100 MB/s. +@example +-machine hmat=on \ +-m 2G \ +-object memory-backend-ram,size=1G,id=m0 \ +-object memory-backend-ram,size=1G,id=m1 \ +-smp 2 \ +-numa node,nodeid=0,memdev=m0 \ +-numa node,nodeid=1,memdev=m1,initiator=0 \ +-numa cpu,node-id=0,socket-id=0 \ +-numa cpu,node-id=0,socket-id=1 \ +-numa hmat-lb,initiator=0,target=0,hierarchy=memory,data-type=access-latency,latency=5 \ +-numa hmat-lb,initiator=0,target=0,hierarchy=memory,data-type=access-bandwidth,bandwidth=200M \ +-numa hmat-lb,initiator=0,target=1,hierarchy=memory,data-type=access-latency,latency=10 \ +-numa hmat-lb,initiator=0,target=1,hierarchy=memory,data-type=access-bandwidth,bandwidth=100M +@end example + ETEXI DEF("add-fd", HAS_ARG, QEMU_OPTION_add_fd,
PATCH v20 3/8 numa: Extend CLI to provide memory side cache information
In reply to Tao XuFrom: Liu Jingqi <[email protected]>
Add -numa hmat-cache option to provide Memory Side Cache Information. These memory attributes help to build Memory Side Cache Information Structure(s) in ACPI Heterogeneous Memory Attribute Table (HMAT). Before using hmat-cache option, enable HMAT with -machine hmat=on.
---
Changes in v20:
- Disable cache level 0 in hmat-cache option (Igor)
- Update the QAPI description (Markus)
Changes in v19:
- Add description about the machine property 'hmat' in commit message (Markus)
- Update the QAPI comments
- Add a check for no memory side cache
Changes in v18:
- Update the error message (Igor)
Changes in v17:
- Use NumaHmatCacheOptions to replace HMAT_Cache_Info (Igor)
- Add check for unordered cache level input (Igor)
Changes in v16:
- Add cross check with hmat_lb data (Igor)
- Drop total_levels in struct HMAT_Cache_Info (Igor)
- Correct the error table number (Igor)
--- hw/core/numa.c | 80 ++++++++++++++++++++++++++++++++++++++++++ include/sysemu/numa.h | 5 +++ qapi/machine.json | 81 +++++++++++++++++++++++++++++++++++++++++-- qemu-options.hx | 17 +++++++-- 4 files changed, 179 insertions(+), 4 deletions(-)
hw/core/numa.c +80 -0
@@ -379,6 +379,73 @@ void parse_numa_hmat_lb(NumaState *numa_state, NumaHmatLBOptions *node,g_array_append_val (hmat_lb->list, lb_data); } +void parse_numa_hmat_cache (MachineState *ms,NumaHmatCacheOptions *node, +Error **errp) +{ +int nb_numa_nodes = ms->numa_state->num_nodes; +NodeInfo *numa_info = ms->numa_state->nodes; +NumaHmatCacheOptions *hmat_cache =NULL ; + +if (node->node_id >= nb_numa_nodes) { +error_setg (errp,"Invalid node-id=%" PRIu32", it should be less " +"than %d" , node->node_id, nb_numa_nodes); +return ; + } + +if (numa_info[node->node_id].lb_info_provided != (BIT (0 ) |BIT (1 ))) { +error_setg (errp,"The latency and bandwidth information of " +"node-id=%" PRIu32" should be provided before memory side " +"cache attributes" , node->node_id); +return ; + } + +if (node->level <1 || node->level >=HMAT_LB_LEVELS ) { +error_setg (errp,"Invalid level=%" PRIu8", it should be larger than 0 " +"and less than or equal to %d" , node->level, +HMAT_LB_LEVELS -1 ); +return ; + } + +assert (node->associativity <HMAT_CACHE_ASSOCIATIVITY__MAX ); +assert (node->policy <HMAT_CACHE_WRITE_POLICY__MAX ); +if (ms->numa_state->hmat_cache[node->node_id][node->level]) { +error_setg (errp,"Duplicate configuration of the side cache for " +"node-id=%" PRIu32" and level=%" PRIu8, + node->node_id, node->level); +return ; + } + +if ((node->level >1 ) && + ms->numa_state->hmat_cache[node->node_id][node->level -1 ] && + (node->size >= + ms->numa_state->hmat_cache[node->node_id][node->level -1 ]->size)) { +error_setg (errp,"Invalid size=%" PRIu64", the size of level=%" PRIu8 +" should be less than the size(%" PRIu64") of " +"level=%" PRIu8, node->size, node->level, + ms->numa_state->hmat_cache[node->node_id] + [node->level -1 ]->size, + node->level -1 ); +return ; + } + +if ((node->level <HMAT_LB_LEVELS -1 ) && + ms->numa_state->hmat_cache[node->node_id][node->level +1 ] && + (node->size <= + ms->numa_state->hmat_cache[node->node_id][node->level +1 ]->size)) { +error_setg (errp,"Invalid size=%" PRIu64", the size of level=%" PRIu8 +" should be larger than the size(%" PRIu64") of " +"level=%" PRIu8, node->size, node->level, + ms->numa_state->hmat_cache[node->node_id] + [node->level +1 ]->size, + node->level +1 ); +return ; + } + + hmat_cache =g_malloc0 (sizeof (*hmat_cache)); +memcpy (hmat_cache, node,sizeof (*hmat_cache)); + ms->numa_state->hmat_cache[node->node_id][node->level] = hmat_cache; +} +void set_numa_options (MachineState *ms,NumaOptions *object,Error **errp) {Error *err =NULL ;
@@ -430,6 +497,19 @@ void set_numa_options(MachineState *ms, NumaOptions *object, Error **errp) goto end; }break ; +case NUMA_OPTIONS_TYPE_HMAT_CACHE : +if (!ms->numa_state->hmat_enabled) { +error_setg (errp,"ACPI Heterogeneous Memory Attribute Table " +"(HMAT) is disabled, enable it with -machine hmat=on " +"before using any of hmat specific options" ); +return ; + } + +parse_numa_hmat_cache (ms, &object->u.hmat_cache, &err); +if (err) { + goto end; + } +break ;default :abort (); }
include/sysemu/numa.h +5 -0
@@ -91,6 +91,9 @@ struct NumaState {/* NUMA nodes HMAT Locality Latency and Bandwidth Information */ HMAT_LB_Info *hmat_lb[HMAT_LB_LEVELS ][HMAT_LB_TYPES ]; + +/* Memory Side Cache Information Structure */ +NumaHmatCacheOptions *hmat_cache[MAX_NODES ][HMAT_LB_LEVELS ]; };typedef struct NumaState NumaState ;
@@ -98,6 +101,8 @@ void set_numa_options(MachineState *ms, NumaOptions *object, Error **errp);void parse_numa_opts (MachineState *ms);void parse_numa_hmat_lb (NumaState *numa_state,NumaHmatLBOptions *node,Error **errp); +void parse_numa_hmat_cache (MachineState *ms,NumaHmatCacheOptions *node, +Error **errp);void numa_complete_configuration (MachineState *ms);void query_numa_node_mem (NumaNodeMem node_mem[],MachineState *ms);extern QemuOptsList qemu_numa_opts;
qapi/machine.json +79 -2
102 lines not on this page · raw message
qemu-options.hx +15 -2
@@ -169,7 +169,8 @@ DEF("numa", HAS_ARG, QEMU_OPTION_numa, "-numa node[,memdev=id][,cpus=firstcpu[-lastcpu]][,nodeid=node][,initiator=node]\n" "-numa dist,src=source,dst=destination,val=distance\n" "-numa cpu,node-id=node[,socket-id=x][,core-id=y][,thread-id=z]\n" - "-numa hmat-lb,initiator=node,target=node,hierarchy=memory|first-level|second-level|third-level,data-type=access-latency|read-latency|write-latency[,latency=lat][,bandwidth=bw]\n", + "-numa hmat-lb,initiator=node,target=node,hierarchy=memory|first-level|second-level|third-level,data-type=access-latency|read-latency|write-latency[,latency=lat][,bandwidth=bw]\n" + "-numa hmat-cache,node-id=node,size=size,level=level[,associativity=none|direct|complex][,policy=none|write-back|write-through][,line=size]\n", QEMU_ARCH_ALL) STEXI @item -numa node[,mem=@var{size}][,cpus=@var{firstcpu}[-@var{lastcpu}]][,nodeid=@var{node}][,initiator=@var{initiator}]
@@ -177,6 +178,7 @@ STEXI @itemx -numa dist,src=@var{source},dst=@var{destination},val=@var{distance} @itemx -numa cpu,node-id=@var{node}[,socket-id=@var{x}][,core-id=@var{y}][,thread-id=@var{z}] @itemx -numa hmat-lb,initiator=@var{node},target=@var{node},hierarchy=@var{hierarchy},data-type=@var{tpye}[,latency=@var{lat}][,bandwidth=@var{bw}] +@itemx -numa hmat-cache,node-id=@var{node},size=@var{size},level=@var{level}[,associativity=@var{str}][,policy=@var{str}][,line=@var{size}] @findex -numa Define a NUMA node and assign RAM and VCPUs to it. Set the NUMA distance from a source node to a destination node.
@@ -280,11 +282,20 @@ NUM byte per second (or MB/s, GB/s or TB/s depending on used suffix). Note that if latency or bandwidth value is 0, means the corresponding latency or bandwidth information is not provided. +In @samp{hmat-cache} option, @var{node-id} is the NUMA-id of the memory belongs. +@var{size} is the size of memory side cache in bytes. @var{level} is the cache +level described in this structure, note that the cache level 0 should not be used +with @samp{hmat-cache} option. @var{associativity} is the cache associativity, +the possible value is 'none/direct(direct-mapped)/complex(complex cache indexing)'. +@var{policy} is the write policy. @var{line} is the cache Line size in bytes. + For example, the following options describe 2 NUMA nodes. Node 0 has 2 cpus and a ram, node 1 has only a ram. The processors in node 0 access memory in node 0 with access-latency 5 nanoseconds, access-bandwidth is 200 MB/s; The processors in NUMA node 0 access memory in NUMA node 1 with access-latency 10 nanoseconds, access-bandwidth is 100 MB/s. +And for memory side cache information, NUMA node 0 and 1 both have 1 level memory +cache, size is 10KB, policy is write-back, the cache Line size is 8 bytes: @example -machine hmat=on \ -m 2G \
@@ -298,7 +309,9 @@ nanoseconds, access-bandwidth is 100 MB/s. -numa hmat-lb,initiator=0,target=0,hierarchy=memory,data-type=access-latency,latency=5 \ -numa hmat-lb,initiator=0,target=0,hierarchy=memory,data-type=access-bandwidth,bandwidth=200M \ -numa hmat-lb,initiator=0,target=1,hierarchy=memory,data-type=access-latency,latency=10 \ --numa hmat-lb,initiator=0,target=1,hierarchy=memory,data-type=access-bandwidth,bandwidth=100M +-numa hmat-lb,initiator=0,target=1,hierarchy=memory,data-type=access-bandwidth,bandwidth=100M \ +-numa hmat-cache,node-id=0,size=10K,level=1,associativity=direct,policy=write-back,line=8 \ +-numa hmat-cache,node-id=1,size=10K,level=1,associativity=direct,policy=write-back,line=8 @end example ETEXI
Signature · 1 line
2.20.1
Re: [PATCH v20 3/8] numa: Extend CLI to provide memory side cache information
Quoted 10 lines · On Fri, 29 Nov 2019 15:56:29 +0800 Tao Xu <[email protected]> wrote:
From: Liu Jingqi <[email protected]>
Add -numa hmat-cache option to provide Memory Side Cache Information. These memory attributes help to build Memory Side Cache Information Structure(s) in ACPI Heterogeneous Memory Attribute Table (HMAT). Before using hmat-cache option, enable HMAT with -machine hmat=on.
Signed-off-by Liu Jingqi <[email protected]>Signed-off-by Tao Xu <[email protected]>
Quoted 314 lines
---
Changes in v20:
- Disable cache level 0 in hmat-cache option (Igor)
- Update the QAPI description (Markus)Changes in v19:
- Add description about the machine property 'hmat' in commit message (Markus)
- Update the QAPI comments
- Add a check for no memory side cacheChanges in v18:
- Update the error message (Igor)Changes in v17:
- Use NumaHmatCacheOptions to replaceHMAT_Cache_Info(Igor)
- Add check for unordered cache level input (Igor)Changes in v16:
- Add cross check withhmat_lbdata (Igor)
- Droptotal_levelsin structHMAT_Cache_Info(Igor)
- Correct the error table number (Igor)--- hw/core/numa.c | 80 ++++++++++++++++++++++++++++++++++++++++++ include/sysemu/numa.h | 5 +++ qapi/machine.json | 81 +++++++++++++++++++++++++++++++++++++++++-- qemu-options.hx | 17 +++++++-- 4 files changed, 179 insertions(+), 4 deletions(-)hw/core/numa.c +80
-0@@ -379,6 +379,73 @@ void parse_numa_hmat_lb(NumaState *numa_state, NumaHmatLBOptions *node,g_array_append_val (hmat_lb->list, lb_data); } +void parse_numa_hmat_cache (MachineState *ms,NumaHmatCacheOptions *node, +Error **errp) +{ +int nb_numa_nodes = ms->numa_state->num_nodes; +NodeInfo *numa_info = ms->numa_state->nodes; +NumaHmatCacheOptions *hmat_cache =NULL ; + +if (node->node_id >= nb_numa_nodes) { +error_setg (errp,"Invalid node-id=%" PRIu32", it should be less " +"than %d" , node->node_id, nb_numa_nodes); +return ; + } + +if (numa_info[node->node_id].lb_info_provided != (BIT (0 ) |BIT (1 ))) { +error_setg (errp,"The latency and bandwidth information of " +"node-id=%" PRIu32" should be provided before memory side " +"cache attributes" , node->node_id); +return ; + } + +if (node->level <1 || node->level >=HMAT_LB_LEVELS ) { +error_setg (errp,"Invalid level=%" PRIu8", it should be larger than 0 " +"and less than or equal to %d" , node->level, +HMAT_LB_LEVELS -1 ); +return ; + } + +assert (node->associativity <HMAT_CACHE_ASSOCIATIVITY__MAX ); +assert (node->policy <HMAT_CACHE_WRITE_POLICY__MAX ); +if (ms->numa_state->hmat_cache[node->node_id][node->level]) { +error_setg (errp,"Duplicate configuration of the side cache for " +"node-id=%" PRIu32" and level=%" PRIu8, + node->node_id, node->level); +return ; + } + +if ((node->level >1 ) && + ms->numa_state->hmat_cache[node->node_id][node->level -1 ] && + (node->size >= + ms->numa_state->hmat_cache[node->node_id][node->level -1 ]->size)) { +error_setg (errp,"Invalid size=%" PRIu64", the size of level=%" PRIu8 +" should be less than the size(%" PRIu64") of " +"level=%" PRIu8, node->size, node->level, + ms->numa_state->hmat_cache[node->node_id] + [node->level -1 ]->size, + node->level -1 ); +return ; + } + +if ((node->level <HMAT_LB_LEVELS -1 ) && + ms->numa_state->hmat_cache[node->node_id][node->level +1 ] && + (node->size <= + ms->numa_state->hmat_cache[node->node_id][node->level +1 ]->size)) { +error_setg (errp,"Invalid size=%" PRIu64", the size of level=%" PRIu8 +" should be larger than the size(%" PRIu64") of " +"level=%" PRIu8, node->size, node->level, + ms->numa_state->hmat_cache[node->node_id] + [node->level +1 ]->size, + node->level +1 ); +return ; + } + + hmat_cache =g_malloc0 (sizeof (*hmat_cache)); +memcpy (hmat_cache, node,sizeof (*hmat_cache)); + ms->numa_state->hmat_cache[node->node_id][node->level] = hmat_cache; +} +void set_numa_options (MachineState *ms,NumaOptions *object,Error **errp) {Error *err =NULL ;@@ -430,6 +497,19 @@ void set_numa_options(MachineState *ms, NumaOptions *object, Error **errp) goto end; }break ; +case NUMA_OPTIONS_TYPE_HMAT_CACHE : +if (!ms->numa_state->hmat_enabled) { +error_setg (errp,"ACPI Heterogeneous Memory Attribute Table " +"(HMAT) is disabled, enable it with -machine hmat=on " +"before using any of hmat specific options" ); +return ; + } + +parse_numa_hmat_cache (ms, &object->u.hmat_cache, &err); +if (err) { + goto end; + } +break ;default :abort (); }include/sysemu/numa.h +5
-0@@ -91,6 +91,9 @@ struct NumaState {/* NUMA nodes HMAT Locality Latency and Bandwidth Information */ HMAT_LB_Info *hmat_lb[HMAT_LB_LEVELS ][HMAT_LB_TYPES ]; + +/* Memory Side Cache Information Structure */ +NumaHmatCacheOptions *hmat_cache[MAX_NODES ][HMAT_LB_LEVELS ]; };typedef struct NumaState NumaState ;@@ -98,6 +101,8 @@ void set_numa_options(MachineState *ms, NumaOptions *object, Error **errp);void parse_numa_opts (MachineState *ms);void parse_numa_hmat_lb (NumaState *numa_state,NumaHmatLBOptions *node,Error **errp); +void parse_numa_hmat_cache (MachineState *ms,NumaHmatCacheOptions *node, +Error **errp);void numa_complete_configuration (MachineState *ms);void query_numa_node_mem (NumaNodeMem node_mem[],MachineState *ms);extern QemuOptsList qemu_numa_opts;qapi/machine.json +79
-2@@ -428,10 +428,12 @@ # # @hmat-lb: memory latency and bandwidth information (Since: 5.0) # +# @hmat-cache: memory side cache information (Since: 5.0) +# # Since: 2.1 ## { 'enum': 'NumaOptionsType', - 'data': [ 'node', 'dist', 'cpu', 'hmat-lb' ] } + 'data': [ 'node', 'dist', 'cpu', 'hmat-lb', 'hmat-cache' ] } ## # @NumaOptions:@@ -447,7 +449,8 @@ 'node': 'NumaNodeOptions', 'dist': 'NumaDistOptions', 'cpu': 'NumaCpuOptions', - 'hmat-lb': 'NumaHmatLBOptions' }} + 'hmat-lb': 'NumaHmatLBOptions', + 'hmat-cache': 'NumaHmatCacheOptions' }} ## # @NumaNodeOptions:@@ -646,6 +649,80 @@ '*latency': 'uint64', '*bandwidth': 'size' }} +## +# @HmatCacheAssociativity: +# +# Cache associativity in the Memory Side Cache Information Structure +# of HMAT +# +# For more information of @HmatCacheAssociativity, see chapter +# 5.2.27.5: Table 5-147 of ACPI 6.3 spec. +# +# @none: None (no memory side cache in this proximity domain, +# or cache associativity unknown) +# +# @direct: Direct Mapped +# +# @complex: Complex Cache Indexing (implementation specific) +# +# Since: 5.0 +## +{ 'enum': 'HmatCacheAssociativity', + 'data': [ 'none', 'direct', 'complex' ] } + +## +# @HmatCacheWritePolicy: +# +# Cache write policy in the Memory Side Cache Information Structure +# of HMAT +# +# For more information of @HmatCacheWritePolicy, see chapter +# 5.2.27.5: Table 5-147: Field "Cache Attributes" of ACPI 6.3 spec. +# +# @none: None (no memory side cache in this proximity domain, +# or cache write policy unknown) +# +# @write-back: Write Back (WB) +# +# @write-through: Write Through (WT) +# +# Since: 5.0 +## +{ 'enum': 'HmatCacheWritePolicy', + 'data': [ 'none', 'write-back', 'write-through' ] } + +## +# @NumaHmatCacheOptions: +# +# Set the memory side cache information for a given memory domain. +# +# For more information of @NumaHmatCacheOptions, see chapter +# 5.2.27.5: Table 5-147: Field "Cache Attributes" of ACPI 6.3 spec. +# +# @node-id: the memory proximity domain to which the memory belongs. +# +# @size: the size of memory side cache in bytes. +# +# @level: the cache level described in this structure. +# +# @associativity: the cache associativity, +# none/direct-mapped/complex(complex cache indexing). +# +# @policy: the write policy, none/write-back/write-through. +# +# @line: the cache Line size in bytes. +# +# Since: 5.0 +## +{ 'struct': 'NumaHmatCacheOptions', + 'data': { + 'node-id': 'uint32', + 'size': 'size', + 'level': 'uint8', + 'associativity': 'HmatCacheAssociativity', + 'policy': 'HmatCacheWritePolicy', + 'line': 'uint16' }} + ## # @HostMemPolicy: #qemu-options.hx +15
-2@@ -169,7 +169,8 @@ DEF("numa", HAS_ARG, QEMU_OPTION_numa, "-numa node[,memdev=id][,cpus=firstcpu[-lastcpu]][,nodeid=node][,initiator=node]\n" "-numa dist,src=source,dst=destination,val=distance\n" "-numa cpu,node-id=node[,socket-id=x][,core-id=y][,thread-id=z]\n" - "-numa hmat-lb,initiator=node,target=node,hierarchy=memory|first-level|second-level|third-level,data-type=access-latency|read-latency|write-latency[,latency=lat][,bandwidth=bw]\n", + "-numa hmat-lb,initiator=node,target=node,hierarchy=memory|first-level|second-level|third-level,data-type=access-latency|read-latency|write-latency[,latency=lat][,bandwidth=bw]\n" + "-numa hmat-cache,node-id=node,size=size,level=level[,associativity=none|direct|complex][,policy=none|write-back|write-through][,line=size]\n", QEMU_ARCH_ALL) STEXI @item -numa node[,mem=@var{size}][,cpus=@var{firstcpu}[-@var{lastcpu}]][,nodeid=@var{node}][,initiator=@var{initiator}]@@ -177,6 +178,7 @@ STEXI @itemx -numa dist,src=@var{source},dst=@var{destination},val=@var{distance} @itemx -numa cpu,node-id=@var{node}[,socket-id=@var{x}][,core-id=@var{y}][,thread-id=@var{z}] @itemx -numa hmat-lb,initiator=@var{node},target=@var{node},hierarchy=@var{hierarchy},data-type=@var{tpye}[,latency=@var{lat}][,bandwidth=@var{bw}] +@itemx -numa hmat-cache,node-id=@var{node},size=@var{size},level=@var{level}[,associativity=@var{str}][,policy=@var{str}][,line=@var{size}] @findex -numa Define a NUMA node and assign RAM and VCPUs to it. Set the NUMA distance from a source node to a destination node.@@ -280,11 +282,20 @@ NUM byte per second (or MB/s, GB/s or TB/s depending on used suffix). Note that if latency or bandwidth value is 0, means the corresponding latency or bandwidth information is not provided. +In @samp{hmat-cache} option, @var{node-id} is the NUMA-id of the memory belongs. +@var{size} is the size of memory side cache in bytes. @var{level} is the cache +level described in this structure, note that the cache level 0 should not be used +with @samp{hmat-cache} option. @var{associativity} is the cache associativity, +the possible value is 'none/direct(direct-mapped)/complex(complex cache indexing)'. +@var{policy} is the write policy. @var{line} is the cache Line size in bytes. + For example, the following options describe 2 NUMA nodes. Node 0 has 2 cpus and a ram, node 1 has only a ram. The processors in node 0 access memory in node 0 with access-latency 5 nanoseconds, access-bandwidth is 200 MB/s; The processors in NUMA node 0 access memory in NUMA node 1 with access-latency 10 nanoseconds, access-bandwidth is 100 MB/s. +And for memory side cache information, NUMA node 0 and 1 both have 1 level memory +cache, size is 10KB, policy is write-back, the cache Line size is 8 bytes: @example -machine hmat=on \ -m 2G \@@ -298,7 +309,9 @@ nanoseconds, access-bandwidth is 100 MB/s. -numa hmat-lb,initiator=0,target=0,hierarchy=memory,data-type=access-latency,latency=5 \ -numa hmat-lb,initiator=0,target=0,hierarchy=memory,data-type=access-bandwidth,bandwidth=200M \ -numa hmat-lb,initiator=0,target=1,hierarchy=memory,data-type=access-latency,latency=10 \ --numa hmat-lb,initiator=0,target=1,hierarchy=memory,data-type=access-bandwidth,bandwidth=100M +-numa hmat-lb,initiator=0,target=1,hierarchy=memory,data-type=access-bandwidth,bandwidth=100M \ +-numa hmat-cache,node-id=0,size=10K,level=1,associativity=direct,policy=write-back,line=8 \ +-numa hmat-cache,node-id=1,size=10K,level=1,associativity=direct,policy=write-back,line=8 @end example ETEXI
PATCH v20 4/8 hmat acpi: Build Memory Proximity Domain Attributes Structure(s)
In reply to Tao XuFrom: Liu Jingqi <[email protected]>
HMAT is defined in ACPI 6.3: 5.2.27 Heterogeneous Memory Attribute Table (HMAT). The specification references below link:
http://www.uefi.org/sites/default/files/resources/ACPI_6_3_final_Jan30.pdf
It describes the memory attributes, such as memory side cache attributes and bandwidth and latency details, related to the Memory Proximity Domain. The software is
expected to use this information as hint for optimization.
This structure describes Memory Proximity Domain Attributes by memory subsystem and its associativity with processor proximity domain as well as hint for memory usage.
In the linux kernel, the codes in drivers/acpi/hmat/hmat.c parse and report the platform's HMAT tables.
---
No changes in v20.
Changes in v16:
- Use uint32_t for initiator and mem_node
Changes in v13:
- Remove the unnecessary head file.
--- hw/acpi/Kconfig | 7 ++- hw/acpi/Makefile.objs | 1 + hw/acpi/hmat.c | 99 +++++++++++++++++++++++++++++++++++++++++++ hw/acpi/hmat.h | 42 ++++++++++++++++++ hw/i386/acpi-build.c | 5 +++ 5 files changed, 152 insertions(+), 2 deletions(-) create mode 100644 hw/acpi/hmat.c create mode 100644 hw/acpi/hmat.h
hw/acpi/Kconfig +5 -2
@@ -7,6 +7,7 @@ config ACPI_X86 select ACPI_NVDIMM select ACPI_CPU_HOTPLUG select ACPI_MEMORY_HOTPLUG + select ACPI_HMAT config ACPI_X86_ICH bool
@@ -23,6 +24,10 @@ config ACPI_NVDIMM bool depends on ACPI +config ACPI_HMAT + bool + depends on ACPI + config ACPI_PCI bool depends on ACPI && PCI
@@ -33,5 +38,3 @@ config ACPI_VMGENID depends on PC config ACPI_HW_REDUCED - bool - depends on ACPI
hw/acpi/Makefile.objs +1 -0
@@ -7,6 +7,7 @@ common-obj-$(CONFIG_ACPI_CPU_HOTPLUG) += cpu.o common-obj-$(CONFIG_ACPI_NVDIMM) += nvdimm.o common-obj-$(CONFIG_ACPI_VMGENID) += vmgenid.o common-obj-$(CONFIG_ACPI_HW_REDUCED) += generic_event_device.o +common-obj-$(CONFIG_ACPI_HMAT) += hmat.o common-obj-$(call lnot,$(CONFIG_ACPI_X86)) += acpi-stub.o common-obj-y += acpi_interface.o
hw/acpi/hmat.c new file +99 -0
99 lines not on this page · raw message
hw/acpi/hmat.h new file +42 -0
42 lines not on this page · raw message
hw/i386/acpi-build.c +5 -0
@@ -67,6 +67,7 @@#include "hw/i386/intel_iommu.h" #include "hw/acpi/ipmi.h" +#include "hw/acpi/hmat.h" /* These are used to size the ACPI tables for -M pc-i440fx-1.7 and * -M pc-i440fx-2.0. Even if the actual amount of AML generated grows
@@ -2834,6 +2835,10 @@ void acpi_build(AcpiBuildTables *tables, MachineState *machine)acpi_add_table (table_offsets, tables_blob);build_slit (tables_blob, tables->linker, machine); } +if (machine->numa_state->hmat_enabled) { +acpi_add_table (table_offsets, tables_blob); +build_hmat (tables_blob, tables->linker, machine->numa_state); + } }if (acpi_get_mcfg (&mcfg)) {acpi_add_table (table_offsets, tables_blob);
Signature · 1 line
2.20.1
PATCH v20 5/8 hmat acpi: Build System Locality Latency and Bandwidth Information Structure(s)
In reply to Tao XuFrom: Liu Jingqi <[email protected]>
This structure describes the memory access latency and bandwidth information from various memory access initiator proximity domains. The latency and bandwidth numbers represented in this structure correspond to rated latency and bandwidth for the platform. The software could use this information as hint for optimization.
---
Changes in v20:
- Fix the broken CI case when user input latency or bandwidth less than required
Changes in v17:
- Remove unnecessary header file (Igor)
Changes in v16:
- Add more description for lb_length (Igor)
- Drop entry_list and calculate entries in this patch (Igor)
Changes in v13:
- Calculate the entries in a new patch.
--- hw/acpi/hmat.c | 104 ++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 103 insertions(+), 1 deletion(-)
hw/acpi/hmat.c +103 -1
@@ -25,6 +25,7 @@*/ #include "qemu/osdep.h" +#include "qemu/units.h" #include "sysemu/numa.h" #include "hw/acpi/hmat.h"
@@ -67,11 +68,89 @@ static void build_hmat_mpda(GArray *table_data, uint16_t flags,build_append_int_noprefix (table_data,0 ,8 ); } +/* +* ACPI 6.3: 5.2.27.4 System Locality Latency and Bandwidth Information +* Structure: Table 5-146 +*/ +static void build_hmat_lb (GArray *table_data,HMAT_LB_Info *hmat_lb, +uint32_t num_initiator,uint32_t num_target, +uint32_t *initiator_list) +{ +int i, index; +HMAT_LB_Data *lb_data; +uint16_t *entry_list; +uint32_t base; +/* Length in bytes for entire structure */ +uint32_t lb_length + =32 /* Table length upto and including Entry Base Unit */ + +4 * num_initiator/* Initiator Proximity Domain List */ + +4 * num_target/* Target Proximity Domain List */ + +2 * num_initiator * num_target;/* Latency or Bandwidth Entries */ + +/* Type */ +build_append_int_noprefix (table_data,1 ,2 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,2 ); +/* Length */ +build_append_int_noprefix (table_data, lb_length,4 ); +/* Flags: Bits [3:0] Memory Hierarchy, Bits[7:4] Reserved */ +assert (!(hmat_lb->hierarchy >>4 )); +build_append_int_noprefix (table_data, hmat_lb->hierarchy,1 ); +/* Data Type */ +build_append_int_noprefix (table_data, hmat_lb->data_type,1 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,2 ); +/* Number of Initiator Proximity Domains (s) */ +build_append_int_noprefix (table_data, num_initiator,4 ); +/* Number of Target Proximity Domains (t) */ +build_append_int_noprefix (table_data, num_target,4 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,4 ); + +/* Entry Base Unit */ +if (hmat_lb->data_type <=HMAT_LB_DATA_WRITE_LATENCY ) { +/* Convert latency base from nanoseconds to picosecond */ + base = hmat_lb->base *1000 ; + }else { +/* Convert bandwidth base from Byte to Megabyte */ + base = hmat_lb->base / MiB; + } +build_append_int_noprefix (table_data, base,8 ); + +/* Initiator Proximity Domain List */ +for (i =0 ; i < num_initiator; i++) { +build_append_int_noprefix (table_data, initiator_list[i],4 ); + } + +/* Target Proximity Domain List */ +for (i =0 ; i < num_target; i++) { +build_append_int_noprefix (table_data, i,4 ); + } + +/* Latency or Bandwidth Entries */ + entry_list =g_malloc0 (num_initiator * num_target *sizeof (uint16_t )); +for (i =0 ; i < hmat_lb->list->len; i++) { + lb_data = &g_array_index (hmat_lb->list, HMAT_LB_Data, i); + index = lb_data->initiator * num_target + lb_data->target; + + entry_list[index] = (uint16_t )(lb_data->data / hmat_lb->base); + } + +for (i =0 ; i < num_initiator * num_target; i++) { +build_append_int_noprefix (table_data, entry_list[i],2 ); + } + +g_free (entry_list); +} +/* Build HMAT sub table structures */ static void hmat_build_table_structs (GArray *table_data,NumaState *numa_state) {uint16_t flags; -int i; +uint32_t num_initiator =0 ; +uint32_t initiator_list[MAX_NODES ]; +int i, hierarchy, type; +HMAT_LB_Info *hmat_lb;for (i =0 ; i < numa_state->num_nodes; i++) { flags =0 ;
@@ -82,6 +161,29 @@ static void hmat_build_table_structs(GArray *table_data, NumaState *numa_state)build_hmat_mpda (table_data, flags, numa_state->nodes[i].initiator, i); } + +for (i =0 ; i < numa_state->num_nodes; i++) { +if (numa_state->nodes[i].has_cpu) { + initiator_list[num_initiator++] = i; + } + } + +/* +* ACPI 6.3: 5.2.27.4 System Locality Latency and Bandwidth Information +* Structure: Table 5-146 +*/ +for (hierarchy =HMAT_LB_MEM_MEMORY ; + hierarchy <=HMAT_LB_MEM_CACHE_3RD_LEVEL ; hierarchy++) { +for (type =HMAT_LB_DATA_ACCESS_LATENCY ; + type <=HMAT_LB_DATA_WRITE_BANDWIDTH ; type++) { + hmat_lb = numa_state->hmat_lb[hierarchy][type]; + +if (hmat_lb && hmat_lb->list->len) { +build_hmat_lb (table_data, hmat_lb, num_initiator, + numa_state->num_nodes, initiator_list); + } + } + } }void build_hmat (GArray *table_data,BIOSLinker *linker,NumaState *numa_state)
Signature · 1 line
2.20.1
PATCH v20 6/8 hmat acpi: Build Memory Side Cache Information Structure(s)
In reply to Tao XuFrom: Liu Jingqi <[email protected]>
This structure describes memory side cache information for memory proximity domains if the memory side cache is present and the physical device forms the memory side cache.
The software could use this information to effectively place the data in memory to maximize the performance of the system memory that use the memory side cache.
---
No changes in v20.
Changes in v16:
- Use checks and assert to replace masks (Igor)
- Fields in Cache Attributes are promoted to uint32_t before shifting (Igor)
- Drop cpu_to_le32() (Igor)
Changes in v13:
- rename level as cache_level
--- hw/acpi/hmat.c | 69 +++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 68 insertions(+), 1 deletion(-)
hw/acpi/hmat.c +68 -1
@@ -143,14 +143,62 @@ static void build_hmat_lb(GArray *table_data, HMAT_LB_Info *hmat_lb,g_free (entry_list); } +/* ACPI 6.3: 5.2.27.5 Memory Side Cache Information Structure: Table 5-147 */ +static void build_hmat_cache (GArray *table_data,uint8_t total_levels, +NumaHmatCacheOptions *hmat_cache) +{ +/* +* Cache Attributes: Bits [3:0] – Total Cache Levels +* for this Memory Proximity Domain +*/ +uint32_t cache_attr = total_levels; + +/* Bits [7:4] : Cache Level described in this structure */ + cache_attr |= (uint32_t ) hmat_cache->level <<4 ; + +/* Bits [11:8] - Cache Associativity */ + cache_attr |= (uint32_t ) hmat_cache->associativity <<8 ; + +/* Bits [15:12] - Write Policy */ + cache_attr |= (uint32_t ) hmat_cache->policy <<12 ; + +/* Bits [31:16] - Cache Line size in bytes */ + cache_attr |= (uint32_t ) hmat_cache->line <<16 ; + +/* Type */ +build_append_int_noprefix (table_data,2 ,2 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,2 ); +/* Length */ +build_append_int_noprefix (table_data,32 ,4 ); +/* Proximity Domain for the Memory */ +build_append_int_noprefix (table_data, hmat_cache->node_id,4 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,4 ); +/* Memory Side Cache Size */ +build_append_int_noprefix (table_data, hmat_cache->size,8 ); +/* Cache Attributes */ +build_append_int_noprefix (table_data, cache_attr,4 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,2 ); +/* +* Number of SMBIOS handles (n) +* Linux kernel uses Memory Side Cache Information Structure +* without SMBIOS entries for now, so set Number of SMBIOS handles +* as 0. +*/ +build_append_int_noprefix (table_data,0 ,2 ); +} +/* Build HMAT sub table structures */ static void hmat_build_table_structs (GArray *table_data,NumaState *numa_state) {uint16_t flags;uint32_t num_initiator =0 ;uint32_t initiator_list[MAX_NODES ]; -int i, hierarchy, type; +int i, hierarchy, type, cache_level, total_levels;HMAT_LB_Info *hmat_lb; +NumaHmatCacheOptions *hmat_cache;for (i =0 ; i < numa_state->num_nodes; i++) { flags =0 ;
@@ -184,6 +232,25 @@ static void hmat_build_table_structs(GArray *table_data, NumaState *numa_state) } } } + +/* +* ACPI 6.3: 5.2.27.5 Memory Side Cache Information Structure: +* Table 5-147 +*/ +for (i =0 ; i < numa_state->num_nodes; i++) { + total_levels =0 ; +for (cache_level =1 ; cache_level <HMAT_LB_LEVELS ; cache_level++) { +if (numa_state->hmat_cache[i][cache_level]) { + total_levels++; + } + } +for (cache_level =0 ; cache_level <= total_levels; cache_level++) { + hmat_cache = numa_state->hmat_cache[i][cache_level]; +if (hmat_cache) { +build_hmat_cache (table_data, total_levels, hmat_cache); + } + } + } }void build_hmat (GArray *table_data,BIOSLinker *linker,NumaState *numa_state)
Signature · 1 line
2.20.1
PATCH v20 7/8 tests/numa: Add case for QMP build HMAT
In reply to Tao XuCheck configuring HMAT usecase
---
Changes in v20:
- Use g_assert_true and g_assert_false to replace g_assert (Thomas and Markus)
Changes in v19:
- Add some fail cases for hmat-cache when level=0
Changes in v18:
- Rewrite the lines over 80 characters
Chenges in v17:
- Add some fail test cases (Igor)
--- tests/numa-test.c | 213 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 213 insertions(+)
tests/numa-test.c +213 -0
@@ -327,6 +327,216 @@ static void pc_dynamic_cpu_cfg(const void *data)qtest_quit (qs); } +static void pc_hmat_build_cfg (const void *data) +{ +QTestState *qs =qtest_initf ("%s -nodefaults --preconfig -machine hmat=on " +"-smp 2,sockets=2 " +"-m 128M,slots=2,maxmem=1G " +"-object memory-backend-ram,size=64M,id=m0 " +"-object memory-backend-ram,size=64M,id=m1 " +"-numa node,nodeid=0,memdev=m0 " +"-numa node,nodeid=1,memdev=m1,initiator=0 " +"-numa cpu,node-id=0,socket-id=0 " +"-numa cpu,node-id=0,socket-id=1" , + data ? (char *)data :"" ); + +/* Fail: Initiator should be less than the number of nodes */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 2, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"access-latency\" } }" ))); + +/* Fail: Target should be less than the number of nodes */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 2," +" 'hierarchy': \"memory\", 'data-type': \"access-latency\" } }" ))); + +/* Fail: Initiator should contain cpu */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 1, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"access-latency\" } }" ))); + +/* Fail: Data-type mismatch */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"write-latency\"," +" 'bandwidth': 524288000 } }" ))); +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"read-bandwidth\"," +" 'latency': 5 } }" ))); + +/* Fail: Bandwidth should be 1MB (1048576) aligned */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"access-bandwidth\"," +" 'bandwidth': 1048575 } }" ))); + +/* Configuring HMAT bandwidth and latency details */ +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"access-latency\"," +" 'latency': 1 } }" )));/* 1 ns */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"access-latency\"," +" 'latency': 5 } }" )));/* Fail: Duplicate configuration */ +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"access-bandwidth\"," +" 'bandwidth': 68717379584 } }" )));/* 65534 MB/s */ +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 1," +" 'hierarchy': \"memory\", 'data-type': \"access-latency\"," +" 'latency': 65534 } }" )));/* 65534 ns */ +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 1," +" 'hierarchy': \"memory\", 'data-type': \"access-bandwidth\"," +" 'bandwidth': 34358689792 } }" )));/* 32767 MB/s */ + +/* Fail: node_id should be less than the number of nodes */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 2, 'size': 10240," +" 'level': 1, 'associativity': \"direct\", 'policy': \"write-back\"," +" 'line': 8 } }" ))); + +/* Fail: level should be less than HMAT_LB_LEVELS (4) */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 0, 'size': 10240," +" 'level': 4, 'associativity': \"direct\", 'policy': \"write-back\"," +" 'line': 8 } }" ))); + +/* Fail: associativity option should be 'none', if level is 0 */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 0, 'size': 10240," +" 'level': 0, 'associativity': \"direct\", 'policy': \"none\"," +" 'line': 0 } }" ))); +/* Fail: policy option should be 'none', if level is 0 */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 0, 'size': 10240," +" 'level': 0, 'associativity': \"none\", 'policy': \"write-back\"," +" 'line': 0 } }" ))); +/* Fail: line option should be 0, if level is 0 */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 0, 'size': 10240," +" 'level': 0, 'associativity': \"none\", 'policy': \"none\"," +" 'line': 8 } }" ))); + +/* Configuring HMAT memory side cache attributes */ +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 0, 'size': 10240," +" 'level': 1, 'associativity': \"direct\", 'policy': \"write-back\"," +" 'line': 8 } }" ))); +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 0, 'size': 10240," +" 'level': 1, 'associativity': \"direct\", 'policy': \"write-back\"," +" 'line': 8 } }" )));/* Fail: Duplicate configuration */ +/* Fail: The size of level 2 size should be small than level 1 */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 0, 'size': 10240," +" 'level': 2, 'associativity': \"direct\", 'policy': \"write-back\"," +" 'line': 8 } }" ))); +/* Fail: The size of level 0 size should be larger than level 1 */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 0, 'size': 10240," +" 'level': 0, 'associativity': \"direct\", 'policy': \"write-back\"," +" 'line': 8 } }" ))); +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 1, 'size': 10240," +" 'level': 1, 'associativity': \"direct\", 'policy': \"write-back\"," +" 'line': 8 } }" ))); + +/* let machine initialization to complete and run */ +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs, +"{ 'execute': 'x-exit-preconfig' }" ))); +qtest_qmp_eventwait (qs,"RESUME" ); + +qtest_quit (qs); +} + +static void pc_hmat_off_cfg (const void *data) +{ +QTestState *qs =qtest_initf ("%s -nodefaults --preconfig " +"-smp 2,sockets=2 " +"-m 128M,slots=2,maxmem=1G " +"-object memory-backend-ram,size=64M,id=m0 " +"-object memory-backend-ram,size=64M,id=m1 " +"-numa node,nodeid=0,memdev=m0" , + data ? (char *)data :"" ); + +/* +* Fail: Enable HMAT with -machine hmat=on +* before using any of hmat specific options +*/ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'node', 'nodeid': 1, 'memdev': \"m1\"," +" 'initiator': 0 } }" ))); +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'node', 'nodeid': 1, 'memdev': \"m1\" } }" ))); +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"access-latency\"," +" 'latency': 1 } }" ))); +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 0, 'size': 10240," +" 'level': 1, 'associativity': \"direct\", 'policy': \"write-back\"," +" 'line': 8 } }" ))); + +/* let machine initialization to complete and run */ +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs, +"{ 'execute': 'x-exit-preconfig' }" ))); +qtest_qmp_eventwait (qs,"RESUME" ); + +qtest_quit (qs); +} + +static void pc_hmat_erange_cfg (const void *data) +{ +QTestState *qs =qtest_initf ("%s -nodefaults --preconfig -machine hmat=on " +"-smp 2,sockets=2 " +"-m 128M,slots=2,maxmem=1G " +"-object memory-backend-ram,size=64M,id=m0 " +"-object memory-backend-ram,size=64M,id=m1 " +"-numa node,nodeid=0,memdev=m0 " +"-numa node,nodeid=1,memdev=m1,initiator=0 " +"-numa cpu,node-id=0,socket-id=0 " +"-numa cpu,node-id=0,socket-id=1" , + data ? (char *)data :"" ); + +/* Can't store the compressed latency */ +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"access-latency\"," +" 'latency': 1 } }" )));/* 1 ns */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"access-latency\"," +" 'latency': 65535 } }" )));/* 65535 ns */ + +/* Test the 0 input (bandwidth not provided) */ +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 0," +" 'hierarchy': \"memory\", 'data-type': \"access-bandwidth\"," +" 'bandwidth': 0 } }" )));/* 0 MB/s */ +/* Fail: bandwidth should be provided before memory side cache attributes */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-cache', 'node-id': 0, 'size': 10240," +" 'level': 1, 'associativity': \"direct\", 'policy': \"write-back\"," +" 'line': 8 } }" ))); + +/* Can't store the compressed bandwidth */ +g_assert_true (qmp_rsp_is_err (qtest_qmp (qs,"{ 'execute': 'set-numa-node'," +" 'arguments': { 'type': 'hmat-lb', 'initiator': 0, 'target': 1," +" 'hierarchy': \"memory\", 'data-type': \"access-bandwidth\"," +" 'bandwidth': 68718428160 } }" )));/* 65535 MB/s */ + +/* let machine initialization to complete and run */ +g_assert_false (qmp_rsp_is_err (qtest_qmp (qs, +"{ 'execute': 'x-exit-preconfig' }" ))); +qtest_qmp_eventwait (qs,"RESUME" ); + +qtest_quit (qs); +} +int main (int argc,char **argv) {const char *args =NULL ;
@@ -346,6 +556,9 @@ int main(int argc, char **argv)if (!strcmp (arch,"i386" ) || !strcmp (arch,"x86_64" )) {qtest_add_data_func ("/numa/pc/cpu/explicit" , args, pc_numa_cpu);qtest_add_data_func ("/numa/pc/dynamic/cpu" , args, pc_dynamic_cpu_cfg); +qtest_add_data_func ("/numa/pc/hmat/build" , args, pc_hmat_build_cfg); +qtest_add_data_func ("/numa/pc/hmat/off" , args, pc_hmat_off_cfg); +qtest_add_data_func ("/numa/pc/hmat/erange" , args, pc_hmat_erange_cfg); }if (!strcmp (arch,"ppc64" )) {
Signature · 1 line
2.20.1
PATCH v20 8/8 tests/bios-tables-test: add test cases for ACPI HMAT
In reply to Tao XuACPI table HMAT has been introduced, QEMU now builds HMAT tables for Heterogeneous Memory with boot option '-numa node'.
Add test cases on PC and Q35 machines with 2 numa nodes. Because HMAT is generated when system enable numa, the following tables need to be added for this test:
tests/data/acpi/pc/APIC.acpihmat tests/data/acpi/pc/SRAT.acpihmat tests/data/acpi/pc/HMAT.acpihmat tests/data/acpi/pc/DSDT.acpihmat tests/data/acpi/q35/APIC.acpihmat tests/data/acpi/q35/SRAT.acpihmat tests/data/acpi/q35/HMAT.acpihmat tests/data/acpi/q35/DSDT.acpihmat
---
No changes in v20.
Changes in v18:
- Remove unit "ns".
Changes in v17:
- Update the latency and bandwidth
Changes in v15:
- Make tests without breaking CI (Michael)
Changes in v13:
- Use decimal notation with appropriate suffix for cache size
--- tests/bios-tables-test-allowed-diff.h | 8 +++++ tests/bios-tables-test.c | 44 +++++++++++++++++++++++++++ tests/data/acpi/pc/APIC.acpihmat | 0 tests/data/acpi/pc/DSDT.acpihmat | 0 tests/data/acpi/pc/HMAT.acpihmat | 0 tests/data/acpi/pc/SRAT.acpihmat | 0 tests/data/acpi/q35/APIC.acpihmat | 0 tests/data/acpi/q35/DSDT.acpihmat | 0 tests/data/acpi/q35/HMAT.acpihmat | 0 tests/data/acpi/q35/SRAT.acpihmat | 0 10 files changed, 52 insertions(+) create mode 100644 tests/data/acpi/pc/APIC.acpihmat create mode 100644 tests/data/acpi/pc/DSDT.acpihmat create mode 100644 tests/data/acpi/pc/HMAT.acpihmat create mode 100644 tests/data/acpi/pc/SRAT.acpihmat create mode 100644 tests/data/acpi/q35/APIC.acpihmat create mode 100644 tests/data/acpi/q35/DSDT.acpihmat create mode 100644 tests/data/acpi/q35/HMAT.acpihmat create mode 100644 tests/data/acpi/q35/SRAT.acpihmat
tests/bios-tables-test-allowed-diff.h +8 -0
@@ -1 +1,9 @@/* List of comma-separated changed AML files to ignore */ +"tests/data/acpi/pc/APIC.acpihmat" , +"tests/data/acpi/pc/SRAT.acpihmat" , +"tests/data/acpi/pc/HMAT.acpihmat" , +"tests/data/acpi/pc/DSDT.acpihmat" , +"tests/data/acpi/q35/APIC.acpihmat" , +"tests/data/acpi/q35/SRAT.acpihmat" , +"tests/data/acpi/q35/HMAT.acpihmat" , +"tests/data/acpi/q35/DSDT.acpihmat",
tests/bios-tables-test.c +44 -0
56 lines not on this page · raw message
tests/data/acpi/pc/APIC.acpihmat new file +0 -0
tests/data/acpi/pc/DSDT.acpihmat new file +0 -0
tests/data/acpi/pc/HMAT.acpihmat new file +0 -0
tests/data/acpi/pc/SRAT.acpihmat new file +0 -0
tests/data/acpi/q35/APIC.acpihmat new file +0 -0
tests/data/acpi/q35/DSDT.acpihmat new file +0 -0
tests/data/acpi/q35/HMAT.acpihmat new file +0 -0
tests/data/acpi/q35/SRAT.acpihmat new file +0 -0
Signature · 1 line
2.20.1
Hi Michael,
Could this patch series be queued?
Thank you very much!
Tao
Quoted 115 lines · On 11/29/2019 3:56 PM, Xu, Tao3 wrote:
This series of patches will build Heterogeneous Memory Attribute Table (HMAT) according to the command line. The ACPI HMAT describes the memory attributes, such as memory side cache attributes and bandwidth and latency details, related to the Memory Proximity Domain.
The software is expected to use HMAT information as hint for optimization.In the linux kernel, the codes in
drivers/acpi/hmat/hmat.cparse and report the platform's HMAT tables.The V19 patches link:
https://patchwork.kernel.org/cover/11265525/Changelog:
v20:
- Useg_assert_trueandg_assert_falseto replaceg_assert(Thomas and Markus)
- Rename assoc as associativity, update the QAPI description (Markus)
- Disable cache level 0 in hmat-cache option (Igor)
- Keep base and bitmap unchanged when latency or bandwidth out of range
- Fix the broken CI case when user input latency or bandwidth less than required.
v19:
- Add description about the machine property 'hmat' in commit message (Markus)
- Update the QAPI comments
- Add a check for no memory side cache
- Add some fail cases for hmat-cache when level=0
v18:
- Defer patches 01/14~06/14 of V17, use qapi type uint64 and only nanosecond for latency (Markus)
- Rewrite the lines over 80 characters(Igor)
v17:
- Add check when user input latency or bandwidth 0, thelb_info_providedshould also be 0. Because in ACPI 6.3 5.2.27.4, 0 means the corresponding latency or bandwidth information is not provided.
- Fix the infinite loop whennode->latencyis 0.
- Use NumaHmatCacheOptions to replaceHMAT_Cache_Info(Igor)
- Add check for unordered cache level input (Igor)
- Add some fail test cases (Igor)
v16:
- Add and useqemu_strtold_finiteto parse size, support full 64bit precision, modify related test cases (Eduardo and Markus)
- Simplify structHMAT_LB_Infoand related code, unify latency and bandwidth (Igor)
- Add cross check withhmat_lbdata (Igor)
- Fields in Cache Attributes are promoted touint32_tbefore shifting (Igor)
- Add case for QMP build HMAT (Igor)
v15:
- Add a new patch to refactordo_strtosz()(Eduardo)
- Make tests without breaking CI (Michael)
v14:
- Reuse the codes ofdo_strtoszto buildqemu_strtotime_ns(Eduardo)
- Squash patch v13 01/12 and 02/12 together (Daniel and Eduardo)
- Drop time unit picosecond (Eric)
- Use qemu ctz64 and clz64 instead of builtin function
v13:
- Modify some text description
- Drop "initiator_valid" field in struct NodeInfo
- Reuse Garray to store the raw bandwidth and bandwidth data
- Calculate common base unit using range bitmap
- Add a patch to alculate hmat latency and bandwidth entry list
- Drop thetotal_levelsoption and use readable cache size
- Remove the unnecessary head file
- Use decimal notation with appropriate suffix for cache sizeLiu Jingqi (5):
numa: Extend CLI to provide memory latency and bandwidth information
numa: Extend CLI to provide memory side cache information hmat acpi: Build Memory Proximity Domain Attributes Structure(s) hmat acpi: Build System Locality Latency and Bandwidth Information
Structure(s)
hmat acpi: Build Memory Side Cache Information Structure(s)Tao Xu (3):
numa: Extend CLI to provide initiator information for numa nodes tests/numa: Add case for QMP build HMAT tests/bios-tables-test: add test cases for ACPI HMAThw/acpi/Kconfig | 7 +- hw/acpi/Makefile.objs | 1 + hw/acpi/hmat.c | 268 +++++++++++++++++++++++ hw/acpi/hmat.h | 42 ++++ hw/core/machine.c | 64 ++++++ hw/core/numa.c | 297 ++++++++++++++++++++++++++ hw/i386/acpi-build.c | 5 + include/sysemu/numa.h | 63 ++++++ qapi/machine.json | 180 +++++++++++++++- qemu-options.hx | 95 +++++++- tests/bios-tables-test-allowed-diff.h | 8 + tests/bios-tables-test.c | 44 ++++ tests/data/acpi/pc/APIC.acpihmat | 0 tests/data/acpi/pc/DSDT.acpihmat | 0 tests/data/acpi/pc/HMAT.acpihmat | 0 tests/data/acpi/pc/SRAT.acpihmat | 0 tests/data/acpi/q35/APIC.acpihmat | 0 tests/data/acpi/q35/DSDT.acpihmat | 0 tests/data/acpi/q35/HMAT.acpihmat | 0 tests/data/acpi/q35/SRAT.acpihmat | 0 tests/numa-test.c | 213 ++++++++++++++++++ 21 files changed, 1276 insertions(+), 11 deletions(-)create mode 100644
hw/acpi/hmat.c
create mode 100644hw/acpi/hmat.h
create mode 100644 tests/data/acpi/pc/APIC.acpihmat create mode 100644 tests/data/acpi/pc/DSDT.acpihmat create mode 100644 tests/data/acpi/pc/HMAT.acpihmat create mode 100644 tests/data/acpi/pc/SRAT.acpihmat create mode 100644 tests/data/acpi/q35/APIC.acpihmat create mode 100644 tests/data/acpi/q35/DSDT.acpihmat create mode 100644 tests/data/acpi/q35/HMAT.acpihmat create mode 100644 tests/data/acpi/q35/SRAT.acpihmat
Quoted 6 lines · Tao Xu
Hi Michael,
Could this patch series be queued?
Thank you very much!Tao
QEMU is in freeze, so not yet. Please ping after the release.
Signature · 1 line
MST
Quoted 10 lines · Michael S. Tsirkin
Quoted 6 lines · Tao Xu
Hi Michael,
Could this patch series be queued?
Thank you very much!Tao
QEMU is in freeze, so not yet. Please ping after the release.
OK, Thank you!
Quoted 9 lines · Michael S. Tsirkin
Quoted 6 lines · Tao Xu
Hi Michael,
Could this patch series be queued?
Thank you very much!Tao
QEMU is in freeze, so not yet. Please ping after the release.
Just to avoid confusion: it's Michael's personal preference not to process patches for the next version during freeze. Other maintainers do, and that's actually the project's policy:
Subject: QEMU Summit 2017: minutes
Message-ID: <CAFEAcA-b9oDkPfZbntWfhWSv1HOnbUf75p_xB_tF74h_NBGPmw@mail.gmail.com>
https://lists.nongnu.org/archive/html/qemu-devel/2017-11/msg04453.html
qemu-next:
* Problem 1: Contributors cannot get patches merged during freeze (bad experience)
[...]
* Markus Armbruster: Problem 1 is solved if maintainers keep their own -next trees
* Paolo Bonzini: Maintaining -next could slow down or create work for -freeze (e.g. who does backports)
* Action: Maintainers mustn't tell submitters to go away just because we're in a release freeze (it's up to them whether they prefer to maintain a "-next" tree for their subsystem with patches queued for the following release, or track which patches they've accepted some other way)
* We're not going to have an official project-wide "-next" tree, though
Michael, would queuing up patches in a -next branch really be too much trouble for you?
Quoted 35 lines · Markus Armbruster
Quoted 9 lines · Michael S. Tsirkin
Quoted 6 lines · Tao Xu
Hi Michael,
Could this patch series be queued?
Thank you very much!Tao
QEMU is in freeze, so not yet. Please ping after the release.
Just to avoid confusion: it's Michael's personal preference not to process patches for the next version during freeze. Other maintainers do, and that's actually the project's policy:
Subject: QEMU Summit 2017: minutes
Message-ID: <CAFEAcA-b9oDkPfZbntWfhWSv1HOnbUf75p_xB_tF74h_NBGPmw@mail.gmail.com>
https://lists.nongnu.org/archive/html/qemu-devel/2017-11/msg04453.htmlqemu-next:
* Problem 1: Contributors cannot get patches merged during freeze (bad experience)
[...]
* Markus Armbruster: Problem 1 is solved if maintainers keep their own -next trees
* Paolo Bonzini: Maintaining -next could slow down or create work for -freeze (e.g. who does backports)
* Action: Maintainers mustn't tell submitters to go away just because we're in a release freeze (it's up to them whether they prefer to maintain a "-next" tree for their subsystem with patches queued for the following release, or track which patches they've accepted some other way)
* We're not going to have an official project-wide "-next" tree, though
Quoted 2 lines
Michael, would queuing up patches in a -next branch really be too much trouble for you?
Thanks for pointing this out!
I stopped asking for re-post since awhile ago. I don't queue patches in a public tree but I do review and do keep track of pending patches.
I tend to ask contributors to also ping because sometimes there's a problem with rebase, I drop the patch but forget to tell the contributor, and it tends to happen more with big patchsets posted during freeze as there's a rush to merge changes right after that.
I usually don't bother people with this for small patches though.
I'll try to be clearer in my communication so contributors don't feel stressed.
Would something like:
"I'll queue it for merge after the release. If possible please ping me after the release to help make sure it didn't get dropped."
be clearer?
Hopefully windows CI efforts will soon bear fruit to the point where they stress PCI enough to make maintaining next worth the effort.
Signature · 1 line
MST
Quoted 55 lines · Michael S. Tsirkin
Quoted 35 lines · Markus Armbruster
Quoted 9 lines · Michael S. Tsirkin
Quoted 6 lines · Tao Xu
Hi Michael,
Could this patch series be queued?
Thank you very much!Tao
QEMU is in freeze, so not yet. Please ping after the release.
Just to avoid confusion: it's Michael's personal preference not to process patches for the next version during freeze. Other maintainers do, and that's actually the project's policy:
Subject: QEMU Summit 2017: minutes
Message-ID: <CAFEAcA-b9oDkPfZbntWfhWSv1HOnbUf75p_xB_tF74h_NBGPmw@mail.gmail.com>
https://lists.nongnu.org/archive/html/qemu-devel/2017-11/msg04453.htmlqemu-next:
* Problem 1: Contributors cannot get patches merged during freeze (bad experience)
[...]
* Markus Armbruster: Problem 1 is solved if maintainers keep their own -next trees
* Paolo Bonzini: Maintaining -next could slow down or create work for -freeze (e.g. who does backports)
* Action: Maintainers mustn't tell submitters to go away just because we're in a release freeze (it's up to them whether they prefer to maintain a "-next" tree for their subsystem with patches queued for the following release, or track which patches they've accepted some other way)
* We're not going to have an official project-wide "-next" tree, thoughQuoted 2 lines
Michael, would queuing up patches in a -next branch really be too much trouble for you?
Thanks for pointing this out!
I stopped asking for re-post since awhile ago. I don't queue patches in a public tree but I do review and do keep track of pending patches.
I tend to ask contributors to also ping because sometimes there's a problem with rebase, I drop the patch but forget to tell the contributor, and it tends to happen more with big patchsets posted during freeze as there's a rush to merge changes right after that.
I usually don't bother people with this for small patches though.I'll try to be clearer in my communication so contributors don't feel stressed.
Would something like:
Quoted 8 lines
"I'll queue it for merge after the release. If possible please ping me after the release to help make sure it didn't get dropped."
be clearer?
Hopefully windows CI efforts will soon bear fruit to the point where they stress PCI enough to make maintaining next worth the effort.
I see. Thanks for Markus and Michael's kindly response. I feel happy rather than stressed in QEMU community :)
Quoted 53 lines · Michael S. Tsirkin
Quoted 35 lines · Markus Armbruster
Quoted 9 lines · Michael S. Tsirkin
Quoted 6 lines · Tao Xu
Hi Michael,
Could this patch series be queued?
Thank you very much!Tao
QEMU is in freeze, so not yet. Please ping after the release.
Just to avoid confusion: it's Michael's personal preference not to process patches for the next version during freeze. Other maintainers do, and that's actually the project's policy:
Subject: QEMU Summit 2017: minutes
Message-ID: <CAFEAcA-b9oDkPfZbntWfhWSv1HOnbUf75p_xB_tF74h_NBGPmw@mail.gmail.com>
https://lists.nongnu.org/archive/html/qemu-devel/2017-11/msg04453.htmlqemu-next:
* Problem 1: Contributors cannot get patches merged during freeze (bad experience)
[...]
* Markus Armbruster: Problem 1 is solved if maintainers keep their own -next trees
* Paolo Bonzini: Maintaining -next could slow down or create work for -freeze (e.g. who does backports)
* Action: Maintainers mustn't tell submitters to go away just because we're in a release freeze (it's up to them whether they prefer to maintain a "-next" tree for their subsystem with patches queued for the following release, or track which patches they've accepted some other way)
* We're not going to have an official project-wide "-next" tree, thoughQuoted 2 lines
Michael, would queuing up patches in a -next branch really be too much trouble for you?
Thanks for pointing this out!
I stopped asking for re-post since awhile ago. I don't queue patches in a public tree but I do review and do keep track of pending patches.
I tend to ask contributors to also ping because sometimes there's a problem with rebase, I drop the patch but forget to tell the contributor, and it tends to happen more with big patchsets posted during freeze as there's a rush to merge changes right after that.
I usually don't bother people with this for small patches though.I'll try to be clearer in my communication so contributors don't feel stressed.
Quoted 6 lines
Would something like:
"I'll queue it for merge after the release. If possible please ping me after the release to help make sure it didn't get dropped."
be clearer?
Yes, that's both clearer and friendlier. Thank you!
Quoted 2 lines
Hopefully windows CI efforts will soon bear fruit to the point where they stress PCI enough to make maintaining next worth the effort.
CI++ :)