Support smp.clusters for x86
cover Support smp.clusters for x86
From: Zhao Liu <[email protected]>
Hi list,
This is the RESEND version of this patch series, rebased on the latest commit of the master branch (3b33ae4: Merge tag 'block- pull-request' of https://gitlab.com/stefanha/qemu into staging), so that we can send subsequent RFCs of the hybrid topology series.
This series add the cluster support for x86 PC machine, which allows x86 can use smp.clusters to configure the modlue level CPU topology of x86.
And since the compatibility issue (see section: ## Why not share L2 cache in cluster directly), this series also introduce a new command to adjust the topology of the x86 L2 cache.
Welcome your comments!
# Backgroud
The "clusters" parameter in "smp" is introduced by ARM [1], but x86 hasn't supported it.
At present, x86 defaults L2 cache is shared in one core, but this is not enough. There're some platforms that multiple cores share the same L2 cache, e.g., Alder Lake-P shares L2 cache for one module of Atom cores [2], that is, every four Atom cores shares one L2 cache. Therefore, we need the new CPU topology level (cluster/module).
Another reason is for hybrid architecture. cluster support not only provides another level of topology definition in x86, but would aslo provide required code change for future our hybrid topology support.
# Overview
## Introduction of module level for x86
"cluster" in smp is the CPU topology level which is between "core" and die.
For x86, the "cluster" in smp is corresponding to the module level [3], which is above the core level. So use the "module" other than "cluster" in x86 code.
And please note that x86 already has a cpu topology level also named "cluster" [3], this level is at the upper level of the package. Here, the cluster in x86 cpu topology is completely different from the "clusters" as the smp parameter. After the module level is introduced, the cluster as the smp parameter will actually refer to the module level of x86.
## Why not share L2 cache in cluster directly
Though "clusters" was introduced to help define L2 cache topology [1], using cluster to define x86's L2 cache topology will cause the compatibility problem:
Currently, x86 defaults that the L2 cache is shared in one core, which actually implies a default setting "cores per L2 cache is 1" and therefore implicitly defaults to having as many L2 caches as cores.
For example (i386 PC machine):
-smp 16,sockets=2,dies=2,cores=2,threads=2,maxcpus=16 (*)
Considering the topology of the L2 cache, this (*) implicitly means "1 core per L2 cache" and "2 L2 caches per die".
If we use cluster to configure L2 cache topology with the new default setting "clusters per L2 cache is 1", the above semantics will change to "2 cores per cluster" and "1 cluster per L2 cache", that is, "2 cores per L2 cache".
So the same command (*) will cause changes in the L2 cache topology, further affecting the performance of the virtual machine.
Therefore, x86 should only treat cluster as a cpu topology level and avoid using it to change L2 cache by default for compatibility.
## module level in CPUID
Currently, we don't expose module level in CPUID.1FH because currently linux (v6.2-rc6) doesn't support module level. And exposing module and die levels at the same time in CPUID.1FH will cause linux to calculate wrong die_id. The module level should be exposed until the real machine has the module level in CPUID.1FH.
We can configure CPUID.04H.02H (L2 cache topology) with module level by a new command:
"-cpu,x-l2-cache-topo=cluster"
More information about this command, please see the section: "## New
property: x-l2-cache-topo".
## New cache topology info in CPUCacheInfo
Currently, by default, the cache topology is encoded as:
1. i/d cache is shared in one core.
2. L2 cache is shared in one core.
3. L3 cache is shared in one die.
This default general setting has caused a misunderstanding, that is, the cache topology is completely equated with a specific cpu topology, such as the connection between L2 cache and core level, and the connection between L3 cache and die level.
In fact, the settings of these topologies depend on the specific platform and are not static. For example, on Alder Lake-P, every four Atom cores share the same L2 cache [2].
Thus, in this patch set, we explicitly define the corresponding cache topology for different cpu models and this has two benefits:
1. Easy to expand to new CPU models in the future, which has different cache topology.
2. It can easily support custom cache topology by some command (e.g., x-l2-cache-topo).
## New property: x-l2-cache-topo
The property l2-cache-topo will be used to change the L2 cache topology in CPUID.04H.
Now it allows user to set the L2 cache is shared in core level or cluster level.
If user passes "-cpu x-l2-cache-topo=[core|cluster]" then older L2 cache topology will be overrided by the new topology setting.
Since CPUID.04H is used by intel cpus, this property is available on intel cpus as for now.
When necessary, it can be extended to CPUID[0x8000001D] for amd cpus.
# Patch description
patch 1-5 Cleanups and fixes related to subsequent changes.
patch 6-13 Add the module as the new CPU topology level in x86, and it
is corresponding to the cluster level in generic code.
patch 14-17 Add cache topology infomation in cache models.
patch 18 Introduce a new command to configure L2 cache topology.
[1]: https://patchew.org/QEMU/[email protected]/
[2]: https://www.intel.com/content/www/us/en/products/platforms/details/alder-lake-p.html
[3]: SDM, vol.3, ch.9, 9.9.1 Hierarchical Mapping of Shared Resources.
---
Zhao Liu (10):
machine: Fix comment of machine_parse_smp_config()
tests: Rename test-x86-cpuid.c to test-x86-apicid.c i386/cpu: Fix number of addressable IDs in CPUID.04H i386/cpu: Consolidate the use of topo_info in cpu_x86_cpuid()
i386: Fix comment style in topology.h
i386: Add cache topology info in CPUCacheInfo
i386: Use CPUCacheInfo.share_level to encode CPUID[4].EAX[bits 25:14]
i386: Fix NumSharingCache for CPUID[0x8000001D].EAX[bits 25:14]
i386: Use CPUCacheInfo.share_level to encode
CPUID[0x8000001D].EAX[bits 25:14]
i386: Add new property to control L2 cache topo in CPUID.04H
Zhuocheng Ding (8):
softmmu: Fix CPUSTATE.nr_cores' calculation
i386: Introduce module-level cpu topology to CPUX86State
i386: Support modules_per_die in X86CPUTopoInfo
i386: Support module_id in X86CPUTopoIDs
i386: Update APIC ID parsing rule to support module level i386/cpu: Introduce cluster-id to X86CPU
tests: Add test case of APIC ID for module level parsing hw/i386/pc: Support smp.clusters for x86 PC machine
MAINTAINERS | 2 +-
hw/core/machine-smp.c | 7 +-
hw/i386/pc.c | 1 +
hw/i386/x86.c | 49 +++++-
include/hw/core/cpu.h | 2 +-
include/hw/i386/topology.h | 68 +++++---
qemu-options.hx | 10 +-
softmmu/cpus.c | 2 +-
target/i386/cpu.c | 146 ++++++++++++++----
target/i386/cpu.h | 25 +++
tests/unit/meson.build | 4 +-
.../{test-x86-cpuid.c => test-x86-apicid.c} | 58 ++++---
12 files changed, 275 insertions(+), 99 deletions(-)
rename tests/unit/{test-x86-cpuid.c => test-x86-apicid.c} (73%)Signature · 1 line
2.34.1
PATCH RESEND 1/18 machine: Fix comment of machine_parse_smp_config()
From: Zhao Liu <[email protected]>
Now smp supports dies and clusters, so add description about these 2 levels in the comment of machine_parse_smp_config().
--- hw/core/machine-smp.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-)
hw/core/machine-smp.c +4 -3
@@ -51,8 +51,8 @@ static char *cpu_hierarchy_to_string(MachineState *ms)* machine_parse_smp_config: Generic function used to parse the given * SMP configuration * -* Any missing parameter in "cpus/maxcpus/sockets/cores/threads" will be -* automatically computed based on the provided ones. +* Any missing parameter in "cpus/maxcpus/sockets/dies/clusters/cores/threads" +* will be automatically computed based on the provided ones. * * In the calculation of omitted sockets/cores/threads: we prefer sockets * over cores over threads before 6.2, while preferring cores over sockets
@@ -66,7 +66,8 @@ static char *cpu_hierarchy_to_string(MachineState *ms)* * For compatibility, apart from the parameters that will be computed, newly * introduced topology members which are likely to be target specific should -* be directly set as 1 if they are omitted (e.g. dies for PC since 4.1). +* be directly set as 1 if they are omitted (e.g. dies for PC since v4.1 and +* clusters for arm since v7.0). */ void machine_parse_smp_config (MachineState *ms,const SMPConfiguration *config,Error **errp)
Signature · 1 line
2.34.1
Re: [PATCH RESEND 01/18] machine: Fix comment of machine_parse_smp_config()
Quoted 17 lines · Zhao Liu
From: Zhao Liu <[email protected]>
Now smp supports dies and clusters, so add description about these 2 levels in the comment of
machine_parse_smp_config().Fixes 864c3b5 (hw/core/machine: Introduce CPU cluster topology support)Suggested-by Robert Hoo <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- hw/core/machine-smp.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-)hw/core/machine-smp.c +0
-0
hw/core/machine-smp.c static char *cpu_hierarchy_to_string(MachineState *ms) · line 55
Quoted 7 lines
hw/core/machine-smp.c +2
-2@@ -51,8 +51,8 @@ static char *cpu_hierarchy_to_string(MachineState *ms)* machine_parse_smp_config: Generic function used to parse the given * SMP configuration * -* Any missing parameter in "cpus/maxcpus/sockets/cores/threads" will be -* automatically computed based on the provided ones. +* Any missing parameter in "cpus/maxcpus/sockets/dies/clusters/cores/threads" +* will be automatically computed based on the provided ones.
This is intential. Newly added topo params (apart from maxcpus/ socket/cores/threads) wiil be assigned to 1 and not computed based the provided ones. There is no problem about this part.
hw/core/machine-smp.c static char *cpu_hierarchy_to_string(MachineState *ms) · line 71
Quoted 11 lines
hw/core/machine-smp.c +2
-1* * In the calculation of omitted sockets/cores/threads: we prefer sockets * over cores over threads before 6.2, while preferring cores over sockets @@ -66,7 +66,8 @@ static char *cpu_hierarchy_to_string(MachineState *ms)* * For compatibility, apart from the parameters that will be computed, newly * introduced topology members which are likely to be target specific should -* be directly set as 1 if they are omitted (e.g. dies for PC since 4.1). +* be directly set as 1 if they are omitted (e.g. dies for PC since v4.1 and +* clusters for arm since v7.0). */
Given that we are going to support cluster for PC machine. Maybe simple "(i.e. dies for PC since 4.1)" here is good enough?
Thanks,
Yanan
Quoted 2 lines
hw/core/machine-smp.c +0
-0void machine_parse_smp_config (MachineState *ms,const SMPConfiguration *config,Error **errp)
Re: [PATCH RESEND 01/18] machine: Fix comment of machine_parse_smp_config()
Quoted 19 lines · wangyanan (Y)
Date: Mon, 13 Feb 2023 21:31:53 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 01/18] machine: Fix comment of
machine_parse_smp_config()Quoted 12 lines · Zhao Liu
From: Zhao Liu <[email protected]>
Now smp supports dies and clusters, so add description about these 2 levels in the comment of
machine_parse_smp_config().Fixes 864c3b5 (hw/core/machine: Introduce CPU cluster topology support)Suggested-by Robert Hoo <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- hw/core/machine-smp.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-)
Quoted 15 lines
hw/core/machine-smp.c static char *cpu_hierarchy_to_string(MachineState *ms) · line 55
Quoted 12 lines
hw/core/machine-smp.c +2
-2@@ -51,8 +51,8 @@ static char *cpu_hierarchy_to_string(MachineState *ms)* machine_parse_smp_config: Generic function used to parse the given * SMP configuration * -* Any missing parameter in "cpus/maxcpus/sockets/cores/threads" will be -* automatically computed based on the provided ones. +* Any missing parameter in "cpus/maxcpus/sockets/dies/clusters/cores/threads" +* will be automatically computed based on the provided ones. This is intential. Newly added topo params (apart from maxcpus/ socket/cores/threads) wiil be assigned to 1 and not computed based the provided ones. There is no problem about this part.
Sorry, I see. Will fix.
Quoted 5 lines
Quoted 5 lines
hw/core/machine-smp.c +0
-0* * In the calculation of omitted sockets/cores/threads: we prefer sockets * over cores over threads before 6.2, while preferring cores over sockets @@ -66,7 +66,8 @@ static char *cpu_hierarchy_to_string(MachineState *ms)*
Quoted 8 lines
hw/core/machine-smp.c static char *cpu_hierarchy_to_string(MachineState *ms) · line 71
Quoted 6 lines
hw/core/machine-smp.c +2
-1* For compatibility, apart from the parameters that will be computed, newly * introduced topology members which are likely to be target specific should -* be directly set as 1 if they are omitted (e.g. dies for PC since 4.1). +* be directly set as 1 if they are omitted (e.g. dies for PC since v4.1 and +* clusters for arm since v7.0). */ Given that we are going to support cluster for PC machine. Maybe simple "(i.e. dies for PC since 4.1)" here is good enough?
Makes sense. Now I understand this logic, and I will drop this commit.
Thanks,
Zhao
Quoted 6 lines
Thanks,
YananQuoted 2 lines
hw/core/machine-smp.c +0
-0void machine_parse_smp_config (MachineState *ms,const SMPConfiguration *config,Error **errp)
PATCH RESEND 2/18 tests: Rename test-x86-cpuid.c to test-x86-apicid.c
In reply to Zhao LiuFrom: Zhao Liu <[email protected]>
In fact, this unit tests APIC ID other than CPUID. Rename to test-x86-apicid.c to make its name more in line with its actual content.
---
MAINTAINERS | 2 +-
tests/unit/meson.build | 4 ++--
tests/unit/{test-x86-cpuid.c => test-x86-apicid.c} | 2 +-
3 files changed, 4 insertions(+), 4 deletions(-)
rename tests/unit/{test-x86-cpuid.c => test-x86-apicid.c} (99%)MAINTAINERS +1 -1
@@ -1679,7 +1679,7 @@ F: include/hw/southbridge/piix.h F: hw/misc/sga.c F: hw/isa/apm.c F: include/hw/isa/apm.h -F: tests/unit/test-x86-cpuid.c +F: tests/unit/test-x86-apicid.c F: tests/qtest/test-x86-cpuid-compat.c PC Chipset
tests/unit/meson.build +2 -2
@@ -20,8 +20,8 @@ tests = { 'test-opts-visitor': [testqapi], 'test-visitor-serialization': [testqapi], 'test-bitmap': [], - # all code tested by test-x86-cpuid is inside topology.h - 'test-x86-cpuid': [], + # all code tested by test-x86-apicid is inside topology.h + 'test-x86-apicid': [], 'test-cutils': [], 'test-div128': [], 'test-shift128': [],
tests/unit/test-x86-apicid.c renamed from tests/unit/test-x86-cpuid.c +1 -1
@@ -1,5 +1,5 @@/* -* Test code for x86 CPUID and Topology functions +* Test code for x86 APIC ID and Topology functions * * Copyright (c) 2012 Red Hat Inc. *
Signature · 1 line
2.34.1
Re: [PATCH RESEND 02/18] tests: Rename test-x86-cpuid.c to test-x86-apicid.c
Quoted 50 lines · Zhao Liu
From: Zhao Liu <[email protected]>
In fact, this unit tests APIC ID other than CPUID. Rename to
test-x86-apicid.cto make its name more in line with its actual content.Signed-off-by Zhao Liu <[email protected]>--- MAINTAINERS | 2 +- tests/unit/meson.build | 4 ++-- tests/unit/{test-x86-cpuid.c => test-x86-apicid.c} | 2 +- 3 files changed, 4 insertions(+), 4 deletions(-)rename
tests/unit/{test-x86-cpuid.c=>test-x86-apicid.c} (99%)MAINTAINERS +1
-1@@ -1679,7 +1679,7 @@ F: include/hw/southbridge/piix.h F: hw/misc/sga.c F: hw/isa/apm.c F: include/hw/isa/apm.h -F: tests/unit/test-x86-cpuid.c +F: tests/unit/test-x86-apicid.c F: tests/qtest/test-x86-cpuid-compat.c PC Chipsettests/unit/meson.build +2
-2@@ -20,8 +20,8 @@ tests = { 'test-opts-visitor': [testqapi], 'test-visitor-serialization': [testqapi], 'test-bitmap': [], - # all code tested by test-x86-cpuid is inside topology.h - 'test-x86-cpuid': [], + # all code tested by test-x86-apicid is inside topology.h + 'test-x86-apicid': [], 'test-cutils': [], 'test-div128': [], 'test-shift128': [],tests/unit/test-x86-apicid.c renamed from tests/unit/test-x86-cpuid.c +0
-0
tests/unit/test-x86-apicid.c line 3
Quoted 4 lines
tests/unit/test-x86-apicid.c renamed from tests/unit/test-x86-cpuid.c +1
-1@@ -1,5 +1,5 @@/* -* Test code for x86 CPUID and Topology functions +* Test code for x86 APIC ID and Topology functions *
I'm not very sure. The "CPUID" sounds like a general test for kinds of CPU IDs. Besides APIC IDs computed from x86_apicid_from_cpu_idx(), there are also topo IDs computed from x86_topo_ids_from_idx() although this kind of IDs are not tested in test-x86-cpuid.c so far.
Thanks,
Yanan
Quoted 2 lines
tests/unit/test-x86-apicid.c renamed from tests/unit/test-x86-cpuid.c +0
-0* Copyright (c) 2012 Red Hat Inc. *
Re: [PATCH RESEND 02/18] tests: Rename test-x86-cpuid.c to test-x86-apicid.c
Quoted 53 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 10:36:34 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 02/18] tests: Renametest-x86-cpuid.cto
test-x86-apicid.cQuoted 47 lines · Zhao Liu
From: Zhao Liu <[email protected]>
In fact, this unit tests APIC ID other than CPUID. Rename to
test-x86-apicid.cto make its name more in line with its actual content.Signed-off-by Zhao Liu <[email protected]>--- MAINTAINERS | 2 +- tests/unit/meson.build | 4 ++-- tests/unit/{test-x86-cpuid.c => test-x86-apicid.c} | 2 +- 3 files changed, 4 insertions(+), 4 deletions(-)rename
tests/unit/{test-x86-cpuid.c=>test-x86-apicid.c} (99%)MAINTAINERS +1
-1@@ -1679,7 +1679,7 @@ F: include/hw/southbridge/piix.h F: hw/misc/sga.c F: hw/isa/apm.c F: include/hw/isa/apm.h -F: tests/unit/test-x86-cpuid.c +F: tests/unit/test-x86-apicid.c F: tests/qtest/test-x86-cpuid-compat.c PC Chipsettests/unit/meson.build +2
-2@@ -20,8 +20,8 @@ tests = { 'test-opts-visitor': [testqapi], 'test-visitor-serialization': [testqapi], 'test-bitmap': [], - # all code tested by test-x86-cpuid is inside topology.h - 'test-x86-cpuid': [], + # all code tested by test-x86-apicid is inside topology.h + 'test-x86-apicid': [], 'test-cutils': [], 'test-div128': [], 'test-shift128': [],tests/unit/test-x86-apicid.c renamed from tests/unit/test-x86-cpuid.c +0
-1--- a/tests/unit/test-x86-cpuid.c
Quoted 8 lines
tests/unit/test-x86-apicid.c line 3
Quoted 6 lines
+++
b/tests/unit/test-x86-apicid.ctests/unit/test-x86-apicid.c renamed from tests/unit/test-x86-cpuid.c +1
-1@@ -1,5 +1,5 @@/* -* Test code for x86 CPUID and Topology functions +* Test code for x86 APIC ID and Topology functions * I'm not very sure. The "CPUID" sounds like a general test for kinds of CPU IDs.
CPUID usually refers to that basic instruction in the x86 to obtain basic cpu information. So such naming is prone to ambiguity.
The cpu topology info of x86 is parsed based on the apic id, including the sub-ids of each topology levels (such as thread id/core id...etc.). These sub-ids are all part of the apic id.
Quoted 3 lines
Besides APIC IDs computed from
x86_apicid_from_cpu_idx(), there are also topo IDs computed fromx86_topo_ids_from_idx()although this kind of IDs are not tested intest-x86-cpuid.cso far.
What about "test-x86-topo.c" or "test-x86-topo-ids.c"?
Quoted 6 lines
Thanks,
YananQuoted 2 lines
tests/unit/test-x86-apicid.c renamed from tests/unit/test-x86-cpuid.c +0
-0* Copyright (c) 2012 Red Hat Inc. *
Re: [PATCH RESEND 02/18] tests: Rename test-x86-cpuid.c to test-x86-apicid.c
Quoted 65 lines · Zhao Liu
Quoted 53 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 10:36:34 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 02/18] tests: Renametest-x86-cpuid.cto
test-x86-apicid.cQuoted 47 lines · Zhao Liu
From: Zhao Liu <[email protected]>
In fact, this unit tests APIC ID other than CPUID. Rename to
test-x86-apicid.cto make its name more in line with its actual content.Signed-off-by Zhao Liu <[email protected]>--- MAINTAINERS | 2 +- tests/unit/meson.build | 4 ++-- tests/unit/{test-x86-cpuid.c => test-x86-apicid.c} | 2 +- 3 files changed, 4 insertions(+), 4 deletions(-)rename
tests/unit/{test-x86-cpuid.c=>test-x86-apicid.c} (99%)MAINTAINERS +1
-1@@ -1679,7 +1679,7 @@ F: include/hw/southbridge/piix.h F: hw/misc/sga.c F: hw/isa/apm.c F: include/hw/isa/apm.h -F: tests/unit/test-x86-cpuid.c +F: tests/unit/test-x86-apicid.c F: tests/qtest/test-x86-cpuid-compat.c PC Chipsettests/unit/meson.build +2
-2@@ -20,8 +20,8 @@ tests = { 'test-opts-visitor': [testqapi], 'test-visitor-serialization': [testqapi], 'test-bitmap': [], - # all code tested by test-x86-cpuid is inside topology.h - 'test-x86-cpuid': [], + # all code tested by test-x86-apicid is inside topology.h + 'test-x86-apicid': [], 'test-cutils': [], 'test-div128': [], 'test-shift128': [],tests/unit/test-x86-apicid.c renamed from tests/unit/test-x86-cpuid.c +0
-1--- a/tests/unit/test-x86-cpuid.cQuoted 8 lines
tests/unit/test-x86-apicid.c line 3
Quoted 6 lines
+++
b/tests/unit/test-x86-apicid.ctests/unit/test-x86-apicid.c renamed from tests/unit/test-x86-cpuid.c +1
-1@@ -1,5 +1,5 @@/* -* Test code for x86 CPUID and Topology functions +* Test code for x86 APIC ID and Topology functions * I'm not very sure. The "CPUID" sounds like a general test for kinds of CPU IDs.
CPUID usually refers to that basic instruction in the x86 to obtain basic cpu information. So such naming is prone to ambiguity.
Quoted 8 lines
The cpu topology info of x86 is parsed based on the apic id, including the sub-ids of each topology levels (such as thread id/core id...etc.). These sub-ids are all part of the apic id.
Quoted 3 lines
Besides APIC IDs computed from
x86_apicid_from_cpu_idx(), there are also topo IDs computed fromx86_topo_ids_from_idx()although this kind of IDs are not tested intest-x86-cpuid.cso far.What about "
test-x86-topo.c" or "test-x86-topo-ids.c"?
The first one is general enough, I think.
Quoted 4 lines
Quoted 4 lines
Thanks,
YananQuoted 2 lines
tests/unit/test-x86-apicid.c renamed from tests/unit/test-x86-cpuid.c +0
-0* Copyright (c) 2012 Red Hat Inc. *
PATCH RESEND 3/18 softmmu: Fix CPUSTATE.nr_cores' calculation
In reply to Zhao LiuFrom: Zhuocheng Ding <[email protected]>
From CPUState.nr_cores' comment, it represents "number of cores within this CPU package".
After 003f230 (machine: Tweak the order of topology members in struct CpuTopology), the meaning of smp.cores changed to "the number of cores in one die", but this commit missed to change CPUState.nr_cores' caculation, so that CPUState.nr_cores became wrong and now it misses to consider numbers of clusters and dies.
At present, only i386 is using CPUState.nr_cores.
But as for i386, which supports die level, the uses of CPUState.nr_cores are very confusing:
Early uses are based on the meaning of "cores per package" (before die is introduced into i386), and later uses are based on "cores per die" (after die's introduction).
This difference is due to that commit a94e142 (target/i386: Add CPUID.1F generation support for multi-dies PCMachine) misunderstood that CPUState.nr_cores means "cores per die" when caculated CPUID.1FH.01H:EBX. After that, the changes in i386 all followed this wrong understanding.
With the influence of 003f230 and a94e142, for i386 currently the result of CPUState.nr_cores is "cores per die", thus the original uses of CPUState.cores based on the meaning of "cores per package" are wrong when mutiple dies exist:
1. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.01H:EBX[bits 23:16] is incorrect because it expects "cpus per package" but now the result is "cpus per die".
2. In cpu_x86_cpuid() of target/i386/cpu.c, for all leaves of CPUID.04H:
EAX[bits 31:26] is incorrect because they expect "cpus per package" but now the result is "cpus per die". The error not only impacts the EAX caculation in cache_info_passthrough case, but also impacts other cases of setting cache topology for Intel CPU according to cpu topology (specifically, the incoming parameter "num_cores" expects "cores per package" in encode_cache_cpuid4()).
3. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.0BH.01H:EBX[bits 15:00] is incorrect because the EBX of 0BH.01H (core level) expects "cpus per package", which may be different with 1FH.01H (The reason is 1FH can support more levels. For QEMU, 1FH also supports die, 1FH.01H:EBX[bits 15:00] expects "cpus per die").
4. In cpu_x86_cpuid() of target/i386/cpu.c, when CPUID.80000001H is caculated, here "cpus per package" is expected to be checked, but in fact, now it checks "cpus per die". Though "cpus per die" also works for this code logic, this isn't consistent with AMD's APM.
5. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.80000008H:ECX expects "cpus per package" but it obtains "cpus per die".
6. In simulate_rdmsr() of target/i386/hvf/x86_emu.c, in kvm_rdmsr_core_thread_count() of target/i386/kvm/kvm.c, and in helper_rdmsr() of target/i386/tcg/sysemu/misc_helper.c, MSR_CORE_THREAD_COUNT expects "cpus per package" and "cores per package", but in these functions, it obtains "cpus per die" and "cores per die".
On the other hand, these uses are correct now (they are added in/after a94e142):
1. In cpu_x86_cpuid() of target/i386/cpu.c, topo_info.cores_per_die meets the actual meaning of CPUState.nr_cores ("cores per die").
2. In cpu_x86_cpuid() of target/i386/cpu.c, vcpus_per_socket (in CPUID. 04H's caculation) considers number of dies, so it's correct.
3. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.1FH.01H:EBX[bits 15:00] needs "cpus per die" and it gets the correct result, and CPUID.1FH.02H:EBX[bits 15:00] gets correct "cpus per package".
When CPUState.nr_cores is correctly changed to "cores per package" again , the above errors will be fixed without extra work, but the "currently" correct cases will go wrong and need special handling to pass correct "cpus/cores per die" they want.
Thus in this patch, we fix CPUState.nr_cores' caculation to fit the original meaning "cores per package", as well as changing caculation of topo_info.cores_per_die, vcpus_per_socket and CPUID.1FH.
In addition, in the nr_threads' comment, specify it represents the number of threads in the "core" to avoid confusion.
--- include/hw/core/cpu.h | 2 +- softmmu/cpus.c | 2 +- target/i386/cpu.c | 9 ++++----- 3 files changed, 6 insertions(+), 7 deletions(-)
include/hw/core/cpu.h +1 -1
@@ -274,7 +274,7 @@ struct qemu_work_item;* QOM parent. * @tcg_cflags: Pre-computed cflags for this cpu. * @nr_cores: Number of cores within this CPU package. -* @nr_threads: Number of threads within this CPU. +* @nr_threads: Number of threads within this CPU core. * @running: #true if CPU is currently running (lockless). * @has_waiter: #true if a CPU is currently waiting for the cpu_exec_end; * valid under cpu_list_lock.
softmmu/cpus.c +1 -1
@@ -630,7 +630,7 @@ void qemu_init_vcpu(CPUState *cpu) {MachineState *ms =MACHINE (qdev_get_machine ()); - cpu->nr_cores = ms->smp.cores; + cpu->nr_cores = ms->smp.dies * ms->smp.clusters * ms->smp.cores; cpu->nr_threads = ms->smp.threads; cpu->stopped = true; cpu->random_seed =qemu_guest_random_seed_thread_part1 ();
target/i386/cpu.c +4 -5
@@ -5218,7 +5218,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,X86CPUTopoInfo topo_info; topo_info.dies_per_pkg = env->nr_dies; - topo_info.cores_per_die = cs->nr_cores; + topo_info.cores_per_die = cs->nr_cores / env->nr_dies; topo_info.threads_per_core = cs->nr_threads;/* Calculate & apply limits for different index ranges */
@@ -5294,8 +5294,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,*/ if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 ); -int vcpus_per_socket = env->nr_dies * cs->nr_cores * - cs->nr_threads; +int vcpus_per_socket = cs->nr_cores * cs->nr_threads;if (cs->nr_cores >1 ) { *eax &= ~0xFC000000 ; *eax |= (pow2ceil (cs->nr_cores) -1 ) <<26 ;
@@ -5468,12 +5467,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 1 : *eax =apicid_die_offset (&topo_info); - *ebx = cs->nr_cores * cs->nr_threads; + *ebx = topo_info.cores_per_die * topo_info.threads_per_core; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;case 2 : *eax =apicid_pkg_offset (&topo_info); - *ebx = env->nr_dies * cs->nr_cores * cs->nr_threads; + *ebx = cs->nr_cores * cs->nr_threads; *ecx |=CPUID_TOPOLOGY_LEVEL_DIE ;break ;default :
Signature · 1 line
2.34.1
Re: [PATCH RESEND 03/18] softmmu: Fix CPUSTATE.nr_cores' calculation
Hi Zhao,
Quoted 122 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
Quoted 1 line
From CPUState.nr_cores' comment, it represents "number of cores within
this CPU package".
After 003f230 (machine: Tweak the order of topology members in struct CpuTopology), the meaning of smp.cores changed to "the number of cores in one die", but this commit missed to change CPUState.nr_cores' caculation, so that CPUState.nr_cores became wrong and now it misses to consider numbers of clusters and dies.
At present, only i386 is using CPUState.nr_cores.
But as for i386, which supports die level, the uses of CPUState.nr_cores are very confusing:
Early uses are based on the meaning of "cores per package" (before die is introduced into i386), and later uses are based on "cores per die" (after die's introduction).
This difference is due to that commit a94e142 (target/i386: Add CPUID.1F generation support for multi-dies PCMachine) misunderstood that CPUState.nr_cores means "cores per die" when caculated CPUID.1FH.01H:EBX. After that, the changes in i386 all followed this wrong understanding.
With the influence of 003f230 and a94e142, for i386 currently the result of CPUState.nr_cores is "cores per die", thus the original uses of CPUState.cores based on the meaning of "cores per package" are wrong when mutiple dies exist:
1. Incpu_x86_cpuid()oftarget/i386/cpu.c, CPUID.01H:EBX[bits 23:16] is incorrect because it expects "cpus per package" but now the result is "cpus per die".
2. Incpu_x86_cpuid()oftarget/i386/cpu.c, for all leaves of CPUID.04H:
EAX[bits 31:26] is incorrect because they expect "cpus per package" but now the result is "cpus per die". The error not only impacts the EAX caculation incache_info_passthroughcase, but also impacts other cases of setting cache topology for Intel CPU according to cpu topology (specifically, the incoming parameter "num_cores" expects "cores per package" inencode_cache_cpuid4()).
3. Incpu_x86_cpuid()oftarget/i386/cpu.c, CPUID.0BH.01H:EBX[bits 15:00] is incorrect because the EBX of 0BH.01H (core level) expects "cpus per package", which may be different with 1FH.01H (The reason is 1FH can support more levels. For QEMU, 1FH also supports die, 1FH.01H:EBX[bits 15:00] expects "cpus per die").
4. Incpu_x86_cpuid()oftarget/i386/cpu.c, when CPUID.80000001H is caculated, here "cpus per package" is expected to be checked, but in fact, now it checks "cpus per die". Though "cpus per die" also works for this code logic, this isn't consistent with AMD's APM.
5. Incpu_x86_cpuid()oftarget/i386/cpu.c, CPUID.80000008H:ECX expects "cpus per package" but it obtains "cpus per die".
6. Insimulate_rdmsr()oftarget/i386/hvf/x86_emu.c, inkvm_rdmsr_core_thread_count()oftarget/i386/kvm/kvm.c, and inhelper_rdmsr()oftarget/i386/tcg/sysemu/misc_helper.c,MSR_CORE_THREAD_COUNTexpects "cpus per package" and "cores per package", but in these functions, it obtains "cpus per die" and "cores per die".On the other hand, these uses are correct now (they are added in/after a94e142):
1. Incpu_x86_cpuid()oftarget/i386/cpu.c, topo_info.cores_per_die meets the actual meaning of CPUState.nr_cores ("cores per die").
2. Incpu_x86_cpuid()oftarget/i386/cpu.c,vcpus_per_socket(in CPUID. 04H's caculation) considers number of dies, so it's correct.
3. Incpu_x86_cpuid()oftarget/i386/cpu.c, CPUID.1FH.01H:EBX[bits 15:00] needs "cpus per die" and it gets the correct result, and CPUID.1FH.02H:EBX[bits 15:00] gets correct "cpus per package".When CPUState.nr_cores is correctly changed to "cores per package" again , the above errors will be fixed without extra work, but the "currently" correct cases will go wrong and need special handling to pass correct "cpus/cores per die" they want.
Thus in this patch, we fix CPUState.nr_cores' caculation to fit the original meaning "cores per package", as well as changing caculation of topo_info.cores_per_die,
vcpus_per_socketand CPUID.1FH.In addition, in the
nr_threads' comment, specify it represents the number of threads in the "core" to avoid confusion.Fixes a94e142 (target/i386: Add CPUID.1F generation support for multi-dies PCMachine)Fixes 003f230 (machine: Tweak the order of topology members in struct CpuTopology)Signed-off-by Zhuocheng Ding <[email protected]>Co-developed-by Zhao Liu <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- include/hw/core/cpu.h | 2 +- softmmu/cpus.c | 2 +- target/i386/cpu.c | 9 ++++----- 3 files changed, 6 insertions(+), 7 deletions(-)include/hw/core/cpu.h +1
-1@@ -274,7 +274,7 @@ struct qemu_work_item;* QOM parent. * @tcg_cflags: Pre-computed cflags for this cpu. * @nr_cores: Number of cores within this CPU package. -* @nr_threads: Number of threads within this CPU. +* @nr_threads: Number of threads within this CPU core. * @running: #true if CPU is currently running (lockless). * @has_waiter: #true if a CPU is currently waiting for the cpu_exec_end; * valid under cpu_list_lock. softmmu/cpus.c +1
-1@@ -630,7 +630,7 @@ void qemu_init_vcpu(CPUState *cpu) {MachineState *ms =MACHINE (qdev_get_machine ()); - cpu->nr_cores = ms->smp.cores; + cpu->nr_cores = ms->smp.dies * ms->smp.clusters * ms->smp.cores; cpu->nr_threads = ms->smp.threads; cpu->stopped = true; cpu->random_seed =qemu_guest_random_seed_thread_part1 ();target/i386/cpu.c +0
-0
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5221
Quoted 5 lines
target/i386/cpu.c +1
-1@@ -5218,7 +5218,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,X86CPUTopoInfo topo_info; topo_info.dies_per_pkg = env->nr_dies; - topo_info.cores_per_die = cs->nr_cores; + topo_info.cores_per_die = cs->nr_cores / env->nr_dies;
Is it better to also add a description for env->nr_dies in X86CPUState, like "/* Number of dies within this CPU package */", for less confusion?
Quoted 18 lines
target/i386/cpu.c +1
-3topo_info.threads_per_core = cs->nr_threads;/* Calculate & apply limits for different index ranges */ @@ -5294,8 +5294,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,*/ if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 ); -int vcpus_per_socket = env->nr_dies * cs->nr_cores * - cs->nr_threads; +int vcpus_per_socket = cs->nr_cores * cs->nr_threads;if (cs->nr_cores >1 ) { *eax &= ~0xFC000000 ; *eax |= (pow2ceil (cs->nr_cores) -1 ) <<26 ;@@ -5468,12 +5467,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 1 : *eax =apicid_die_offset (&topo_info); - *ebx = cs->nr_cores * cs->nr_threads;
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5478
Quoted 10 lines
target/i386/cpu.c +2
-1+ *ebx = topo_info.cores_per_die * topo_info.threads_per_core; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;case 2 : *eax =apicid_pkg_offset (&topo_info); - *ebx = env->nr_dies * cs->nr_cores * cs->nr_threads; + *ebx = cs->nr_cores * cs->nr_threads; *ecx |=CPUID_TOPOLOGY_LEVEL_DIE ;break ;default :
Otherwise:
Thanks,
Yanan
Re: [PATCH RESEND 03/18] softmmu: Fix CPUSTATE.nr_cores' calculation
Quoted 127 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 10:58:07 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 03/18] softmmu: Fix CPUSTATE.nr_cores'
calculationHi Zhao,
Quoted 119 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
Quoted 1 line
From CPUState.nr_cores' comment, it represents "number of cores within
this CPU package".
After 003f230 (machine: Tweak the order of topology members in struct CpuTopology), the meaning of smp.cores changed to "the number of cores in one die", but this commit missed to change CPUState.nr_cores' caculation, so that CPUState.nr_cores became wrong and now it misses to consider numbers of clusters and dies.
At present, only i386 is using CPUState.nr_cores.
But as for i386, which supports die level, the uses of CPUState.nr_cores are very confusing:
Early uses are based on the meaning of "cores per package" (before die is introduced into i386), and later uses are based on "cores per die" (after die's introduction).
This difference is due to that commit a94e142 (target/i386: Add CPUID.1F generation support for multi-dies PCMachine) misunderstood that CPUState.nr_cores means "cores per die" when caculated CPUID.1FH.01H:EBX. After that, the changes in i386 all followed this wrong understanding.
With the influence of 003f230 and a94e142, for i386 currently the result of CPUState.nr_cores is "cores per die", thus the original uses of CPUState.cores based on the meaning of "cores per package" are wrong when mutiple dies exist:
1. Incpu_x86_cpuid()oftarget/i386/cpu.c, CPUID.01H:EBX[bits 23:16] is incorrect because it expects "cpus per package" but now the result is "cpus per die".
2. Incpu_x86_cpuid()oftarget/i386/cpu.c, for all leaves of CPUID.04H:
EAX[bits 31:26] is incorrect because they expect "cpus per package" but now the result is "cpus per die". The error not only impacts the EAX caculation incache_info_passthroughcase, but also impacts other cases of setting cache topology for Intel CPU according to cpu topology (specifically, the incoming parameter "num_cores" expects "cores per package" inencode_cache_cpuid4()).
3. Incpu_x86_cpuid()oftarget/i386/cpu.c, CPUID.0BH.01H:EBX[bits 15:00] is incorrect because the EBX of 0BH.01H (core level) expects "cpus per package", which may be different with 1FH.01H (The reason is 1FH can support more levels. For QEMU, 1FH also supports die, 1FH.01H:EBX[bits 15:00] expects "cpus per die").
4. Incpu_x86_cpuid()oftarget/i386/cpu.c, when CPUID.80000001H is caculated, here "cpus per package" is expected to be checked, but in fact, now it checks "cpus per die". Though "cpus per die" also works for this code logic, this isn't consistent with AMD's APM.
5. Incpu_x86_cpuid()oftarget/i386/cpu.c, CPUID.80000008H:ECX expects "cpus per package" but it obtains "cpus per die".
6. Insimulate_rdmsr()oftarget/i386/hvf/x86_emu.c, inkvm_rdmsr_core_thread_count()oftarget/i386/kvm/kvm.c, and inhelper_rdmsr()oftarget/i386/tcg/sysemu/misc_helper.c,MSR_CORE_THREAD_COUNTexpects "cpus per package" and "cores per package", but in these functions, it obtains "cpus per die" and "cores per die".On the other hand, these uses are correct now (they are added in/after a94e142):
1. Incpu_x86_cpuid()oftarget/i386/cpu.c, topo_info.cores_per_die meets the actual meaning of CPUState.nr_cores ("cores per die").
2. Incpu_x86_cpuid()oftarget/i386/cpu.c,vcpus_per_socket(in CPUID. 04H's caculation) considers number of dies, so it's correct.
3. Incpu_x86_cpuid()oftarget/i386/cpu.c, CPUID.1FH.01H:EBX[bits 15:00] needs "cpus per die" and it gets the correct result, and CPUID.1FH.02H:EBX[bits 15:00] gets correct "cpus per package".When CPUState.nr_cores is correctly changed to "cores per package" again , the above errors will be fixed without extra work, but the "currently" correct cases will go wrong and need special handling to pass correct "cpus/cores per die" they want.
Thus in this patch, we fix CPUState.nr_cores' caculation to fit the original meaning "cores per package", as well as changing caculation of topo_info.cores_per_die,
vcpus_per_socketand CPUID.1FH.In addition, in the
nr_threads' comment, specify it represents the number of threads in the "core" to avoid confusion.Fixes a94e142 (target/i386: Add CPUID.1F generation support for multi-dies PCMachine)Fixes 003f230 (machine: Tweak the order of topology members in struct CpuTopology)Signed-off-by Zhuocheng Ding <[email protected]>Co-developed-by Zhao Liu <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- include/hw/core/cpu.h | 2 +- softmmu/cpus.c | 2 +- target/i386/cpu.c | 9 ++++----- 3 files changed, 6 insertions(+), 7 deletions(-)include/hw/core/cpu.h +1
-1@@ -274,7 +274,7 @@ struct qemu_work_item;* QOM parent. * @tcg_cflags: Pre-computed cflags for this cpu. * @nr_cores: Number of cores within this CPU package. -* @nr_threads: Number of threads within this CPU. +* @nr_threads: Number of threads within this CPU core. * @running: #true if CPU is currently running (lockless). * @has_waiter: #true if a CPU is currently waiting for the cpu_exec_end; * valid under cpu_list_lock. softmmu/cpus.c +1
-1@@ -630,7 +630,7 @@ void qemu_init_vcpu(CPUState *cpu) {MachineState *ms =MACHINE (qdev_get_machine ()); - cpu->nr_cores = ms->smp.cores; + cpu->nr_cores = ms->smp.dies * ms->smp.clusters * ms->smp.cores; cpu->nr_threads = ms->smp.threads; cpu->stopped = true; cpu->random_seed =qemu_guest_random_seed_thread_part1 ();target/i386/cpu.c +0
-1--- a/target/i386/cpu.c
Quoted 8 lines
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5220
Quoted 6 lines
+++
b/target/i386/cpu.ctarget/i386/cpu.c +1
-1@@ -5218,7 +5218,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,X86CPUTopoInfo topo_info; topo_info.dies_per_pkg = env->nr_dies; - topo_info.cores_per_die = cs->nr_cores; + topo_info.cores_per_die = cs->nr_cores / env->nr_dies;Is it better to also add a description for
env->nr_diesin X86CPUState, like "/* Number of dies within this CPU package */", for less confusion?
Yeah, thanks. I'll add this comment.
Quoted 33 lines
Quoted 17 lines
target/i386/cpu.c +1
-3topo_info.threads_per_core = cs->nr_threads;/* Calculate & apply limits for different index ranges */ @@ -5294,8 +5294,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,*/ if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 ); -int vcpus_per_socket = env->nr_dies * cs->nr_cores * - cs->nr_threads; +int vcpus_per_socket = cs->nr_cores * cs->nr_threads;if (cs->nr_cores >1 ) { *eax &= ~0xFC000000 ; *eax |= (pow2ceil (cs->nr_cores) -1 ) <<26 ;@@ -5468,12 +5467,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 1 : *eax =apicid_die_offset (&topo_info); - *ebx = cs->nr_cores * cs->nr_threads;target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5478
Quoted 10 lines
target/i386/cpu.c +2
-1+ *ebx = topo_info.cores_per_die * topo_info.threads_per_core; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;case 2 : *eax =apicid_pkg_offset (&topo_info); - *ebx = env->nr_dies * cs->nr_cores * cs->nr_threads; + *ebx = cs->nr_cores * cs->nr_threads; *ecx |=CPUID_TOPOLOGY_LEVEL_DIE ;break ;default :Otherwise:
Reviewed-by Yanan Wang <[email protected]>Thanks,
Yanan
PATCH RESEND 4/18 i386/cpu: Fix number of addressable IDs in CPUID.04H
In reply to Zhao LiuFrom: Zhao Liu <[email protected]>
For i-cache and d-cache, the maximum IDs for CPUs sharing cache ( CPUID.04H.00H:EAX[bits 25:14] and CPUID.04H.01H:EAX[bits 25:14]) are both 0, and this means i-cache and d-cache are shared in the SMT level. This is correct if there's single thread per core, but is wrong for the hyper threading case (one core contains multiple threads) since the i-cache and d-cache are shared in the core level other than SMT level.
Therefore, in order to be compatible with both multi-threaded and single-threaded situations, we should set i-cache and d-cache be shared at the core level by default.
Refer to the fixes of cache_info_passthrough ([1], [2]) and SDM, the CPUID.04H:EAX[bits 25:14] and CPUID.04H:EAX[bits 31:26] should use the nearest power-of-2 integer.
The nearest power-of-2 integer can be caculated by pow2ceil() or by using APIC ID offset (like L3 topology using 1 << die_offset [3]).
But in fact, CPUID.04H:EAX[bits 25:14] and CPUID.04H:EAX[bits 31:26] are associated with APIC ID. For example, in linux kernel, the field "num_threads_sharing" (Bits 25 - 14) is parsed with APIC ID. And for another example, on Alder Lake P, the CPUID.04H:EAX[bits 31:26] is not matched with actual core numbers and it's caculated by:
"(1 << (pkg_offset - core_offset)) - 1".
Therefore the offset of APIC ID should be preferred to caculate nearest power-of-2 integer for CPUID.04H:EAX[bits 25:14] and CPUID.04H:EAX[bits 31:26]:
1. d/i cache is shared in a core, 1 << core_offset should be used instand of "1" in encode_cache_cpuid4() for CPUID.04H.00H:EAX[bits 25:14] and CPUID.04H.01H:EAX[bits 25:14].
2. L2 cache is supposed to be shared in a core as for now, thereby 1 << core_offset should also be used instand of "cs->nr_threads" in encode_cache_cpuid4() for CPUID.04H.02H:EAX[bits 25:14].
3. Similarly, the value for CPUID.04H:EAX[bits 31:26] should also be replaced by the offsets upper SMT level in APIC ID.
And since [1] and [2] are good enough to make cache_into_passthrough work well, its "pow2ceil()" uses are enough so that they're no need to be replaced by APIC ID offset way.
[1]: efb3934 (x86: cpu: make sure number of addressable IDs for processor cores meets the spec) [2]: d7caf13 (x86: cpu: fixup number of addressable IDs for logical processors sharing cache) [3]: d65af28 (i386: Update new x86_apicid parsing rules with die_offset support)
--- target/i386/cpu.c | 20 +++++++++++++++----- 1 file changed, 15 insertions(+), 5 deletions(-)
target/i386/cpu.c +15 -5
@@ -5212,7 +5212,6 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, {X86CPU *cpu =env_archcpu (env);CPUState *cs =env_cpu (env); -uint32_t die_offset;uint32_t limit;uint32_t signature[3 ];X86CPUTopoInfo topo_info;
@@ -5308,27 +5307,38 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *eax = *ebx = *ecx = *edx =0 ; }else { *eax =0 ; +int addressable_cores_offset =apicid_pkg_offset (&topo_info) - +apicid_core_offset (&topo_info); +int core_offset, die_offset; +switch (count) {case 0 :/* L1 dcache info */ + core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l1d_cache, -1 , cs->nr_cores, + (1 << core_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ;case 1 :/* L1 icache info */ + core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l1i_cache, -1 , cs->nr_cores, + (1 << core_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ;case 2 :/* L2 cache info */ + core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l2_cache, - cs->nr_threads, cs->nr_cores, + (1 << core_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ;case 3 :/* L3 cache info */ die_offset =apicid_die_offset (&topo_info);if (cpu->enable_l3_cache) {encode_cache_cpuid4 (env->cache_info_cpuid4.l3_cache, - (1 << die_offset), cs->nr_cores, + (1 << die_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ; }
Signature · 1 line
2.34.1
Re: [PATCH RESEND 04/18] i386/cpu: Fix number of addressable IDs in CPUID.04H
Hi Zhao,
Quoted 34 lines · Zhao Liu
From: Zhao Liu <[email protected]>
For i-cache and d-cache, the maximum IDs for CPUs sharing cache ( CPUID.04H.00H:EAX[bits 25:14] and CPUID.04H.01H:EAX[bits 25:14]) are both 0, and this means i-cache and d-cache are shared in the SMT level. This is correct if there's single thread per core, but is wrong for the hyper threading case (one core contains multiple threads) since the i-cache and d-cache are shared in the core level other than SMT level.
Therefore, in order to be compatible with both multi-threaded and single-threaded situations, we should set i-cache and d-cache be shared at the core level by default.
Refer to the fixes of
cache_info_passthrough([1], [2]) and SDM, the CPUID.04H:EAX[bits 25:14] and CPUID.04H:EAX[bits 31:26] should use the nearest power-of-2 integer.The nearest power-of-2 integer can be caculated by
pow2ceil()or by using APIC ID offset (like L3 topology using 1 <<die_offset[3]).But in fact, CPUID.04H:EAX[bits 25:14] and CPUID.04H:EAX[bits 31:26] are associated with APIC ID. For example, in linux kernel, the field "
num_threads_sharing" (Bits 25 - 14) is parsed with APIC ID. And for another example, on Alder Lake P, the CPUID.04H:EAX[bits 31:26] is not matched with actual core numbers and it's caculated by:
"(1 << (pkg_offset-core_offset)) - 1".Therefore the offset of APIC ID should be preferred to caculate nearest power-of-2 integer for CPUID.04H:EAX[bits 25:14] and CPUID.04H:EAX[bits 31:26]:
1. d/i cache is shared in a core, 1 <<core_offsetshould be used instand of "1" inencode_cache_cpuid4()for CPUID.04H.00H:EAX[bits 25:14] and CPUID.04H.01H:EAX[bits 25:14].
2. L2 cache is supposed to be shared in a core as for now, thereby
Quoted 8 lines
1 << core_offset should also be used instand of "cs->nr_threads" in encode_cache_cpuid4() for CPUID.04H.02H:EAX[bits 25:14].3. Similarly, the value for CPUID.04H:EAX[bits 31:26] should also be replaced by the offsets upper SMT level in APIC ID.
And since [1] and [2] are good enough to make
cache_into_passthroughwork well, its "pow2ceil()" uses are enough so that they're no need to be replaced by APIC ID offset way.
If you uniformly tweak these two places with APIC ID offset too, then you can also use the more spec-compliant helpers
(e.g max_processor_ids_for_cache and max_core_ids_in_pkg) here in future patch #18. Would be it best to uniform the code?
Thanks,
Yanan
Quoted 66 lines
[1]: efb3934 (x86: cpu: make sure number of addressable IDs for processor cores meets the spec) [2]: d7caf13 (x86: cpu: fixup number of addressable IDs for logical processors sharing cache) [3]: d65af28 (i386: Update new
x86_apicidparsing rules withdie_offsetsupport)Fixes 7e3482f (i386: Helpers to encode cache information consistently)Suggested-by Robert Hoo <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 20 +++++++++++++++----- 1 file changed, 15 insertions(+), 5 deletions(-)target/i386/cpu.c +15
-5@@ -5212,7 +5212,6 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, {X86CPU *cpu =env_archcpu (env);CPUState *cs =env_cpu (env); -uint32_t die_offset;uint32_t limit;uint32_t signature[3 ];X86CPUTopoInfo topo_info;@@ -5308,27 +5307,38 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *eax = *ebx = *ecx = *edx =0 ; }else { *eax =0 ; +int addressable_cores_offset =apicid_pkg_offset (&topo_info) - +apicid_core_offset (&topo_info); +int core_offset, die_offset; +switch (count) {case 0 :/* L1 dcache info */ + core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l1d_cache, -1 , cs->nr_cores, + (1 << core_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ;case 1 :/* L1 icache info */ + core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l1i_cache, -1 , cs->nr_cores, + (1 << core_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ;case 2 :/* L2 cache info */ + core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l2_cache, - cs->nr_threads, cs->nr_cores, + (1 << core_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ;case 3 :/* L3 cache info */ die_offset =apicid_die_offset (&topo_info);if (cpu->enable_l3_cache) {encode_cache_cpuid4 (env->cache_info_cpuid4.l3_cache, - (1 << die_offset), cs->nr_cores, + (1 << die_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ; }
Re: [PATCH RESEND 04/18] i386/cpu: Fix number of addressable IDs in CPUID.04H
Quoted 47 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 18:11:23 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 04/18] i386/cpu: Fix number of addressable IDs
in CPUID.04HHi Zhao,
Quoted 39 lines · Zhao Liu
From: Zhao Liu <[email protected]>
For i-cache and d-cache, the maximum IDs for CPUs sharing cache ( CPUID.04H.00H:EAX[bits 25:14] and CPUID.04H.01H:EAX[bits 25:14]) are both 0, and this means i-cache and d-cache are shared in the SMT level. This is correct if there's single thread per core, but is wrong for the hyper threading case (one core contains multiple threads) since the i-cache and d-cache are shared in the core level other than SMT level.
Therefore, in order to be compatible with both multi-threaded and single-threaded situations, we should set i-cache and d-cache be shared at the core level by default.
Refer to the fixes of
cache_info_passthrough([1], [2]) and SDM, the CPUID.04H:EAX[bits 25:14] and CPUID.04H:EAX[bits 31:26] should use the nearest power-of-2 integer.The nearest power-of-2 integer can be caculated by
pow2ceil()or by using APIC ID offset (like L3 topology using 1 <<die_offset[3]).But in fact, CPUID.04H:EAX[bits 25:14] and CPUID.04H:EAX[bits 31:26] are associated with APIC ID. For example, in linux kernel, the field "
num_threads_sharing" (Bits 25 - 14) is parsed with APIC ID. And for another example, on Alder Lake P, the CPUID.04H:EAX[bits 31:26] is not matched with actual core numbers and it's caculated by:
"(1 << (pkg_offset-core_offset)) - 1".Therefore the offset of APIC ID should be preferred to caculate nearest power-of-2 integer for CPUID.04H:EAX[bits 25:14] and CPUID.04H:EAX[bits 31:26]:
1. d/i cache is shared in a core, 1 <<core_offsetshould be used instand of "1" inencode_cache_cpuid4()for CPUID.04H.00H:EAX[bits 25:14] and CPUID.04H.01H:EAX[bits 25:14].
2. L2 cache is supposed to be shared in a core as for now, thereby1 << core_offset should also be used instand of "cs->nr_threads" in encode_cache_cpuid4() for CPUID.04H.02H:EAX[bits 25:14].3. Similarly, the value for CPUID.04H:EAX[bits 31:26] should also be replaced by the offsets upper SMT level in APIC ID.
Quoted 7 lines
Quoted 3 lines
And since [1] and [2] are good enough to make
cache_into_passthroughwork well, its "pow2ceil()" uses are enough so that they're no need to be replaced by APIC ID offset way.If you uniformly tweak these two places with APIC ID offset too, then you can also use the more spec-compliant helpers
(e.gmax_processor_ids_for_cacheandmax_core_ids_in_pkg) here in future patch #18. Would be it best to uniform the code?
Yes, it makes sense! Will also do that. Thanks!
Quoted 70 lines
Thanks,
YananQuoted 66 lines
[1]: efb3934 (x86: cpu: make sure number of addressable IDs for processor cores meets the spec) [2]: d7caf13 (x86: cpu: fixup number of addressable IDs for logical processors sharing cache) [3]: d65af28 (i386: Update new
x86_apicidparsing rules withdie_offsetsupport)Fixes 7e3482f (i386: Helpers to encode cache information consistently)Suggested-by Robert Hoo <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 20 +++++++++++++++----- 1 file changed, 15 insertions(+), 5 deletions(-)target/i386/cpu.c +15
-5@@ -5212,7 +5212,6 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, {X86CPU *cpu =env_archcpu (env);CPUState *cs =env_cpu (env); -uint32_t die_offset;uint32_t limit;uint32_t signature[3 ];X86CPUTopoInfo topo_info;@@ -5308,27 +5307,38 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *eax = *ebx = *ecx = *edx =0 ; }else { *eax =0 ; +int addressable_cores_offset =apicid_pkg_offset (&topo_info) - +apicid_core_offset (&topo_info); +int core_offset, die_offset; +switch (count) {case 0 :/* L1 dcache info */ + core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l1d_cache, -1 , cs->nr_cores, + (1 << core_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ;case 1 :/* L1 icache info */ + core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l1i_cache, -1 , cs->nr_cores, + (1 << core_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ;case 2 :/* L2 cache info */ + core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l2_cache, - cs->nr_threads, cs->nr_cores, + (1 << core_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ;case 3 :/* L3 cache info */ die_offset =apicid_die_offset (&topo_info);if (cpu->enable_l3_cache) {encode_cache_cpuid4 (env->cache_info_cpuid4.l3_cache, - (1 << die_offset), cs->nr_cores, + (1 << die_offset), + (1 << addressable_cores_offset), eax, ebx, ecx, edx);break ; }
Re: [PATCH RESEND 04/18] i386/cpu: Fix number of addressable IDs in CPUID.04H
In reply to Zhao LiuQuoted 10 lines · Zhao Liu
For i-cache and d-cache, the maximum IDs for CPUs sharing cache ( CPUID.04H.00H:EAX[bits 25:14] and CPUID.04H.01H:EAX[bits 25:14]) are both 0, and this means i-cache and d-cache are shared in the SMT level. This is correct if there's single thread per core, but is wrong for the hyper threading case (one core contains multiple threads) since the i-cache and d-cache are shared in the core level other than SMT level.
Therefore, in order to be compatible with both multi-threaded and single-threaded situations, we should set i-cache and d-cache be shared at the core level by default.
It's true for VM only when the exactly HW topology is configured to VM. i.e., two virtual LPs of one virtual CORE are pinned to two physical LPs that of one physical CORE. Otherwise it's incorrect for VM.
for example. given a VM of 4 threads and 2 cores. If not pinning the 4 threads to physical 4 LPs of 2 CORES. It's likely each vcpu running on a LP of different physical cores. So no vcpu shares L1i/L1d cache at core level.
Re: [PATCH RESEND 04/18] i386/cpu: Fix number of addressable IDs in CPUID.04H
Hi Xiaoyao,
Thanks, I've spent some time thinking about it here.
Quoted 13 lines · Xiaoyao Li
Date: Mon, 20 Feb 2023 14:59:20 +0800
From: Xiaoyao Li <[email protected]>
Subject: Re: [PATCH RESEND 04/18] i386/cpu: Fix number of addressable IDs
in CPUID.04HQuoted 7 lines · Zhao Liu
For i-cache and d-cache, the maximum IDs for CPUs sharing cache ( CPUID.04H.00H:EAX[bits 25:14] and CPUID.04H.01H:EAX[bits 25:14]) are both 0, and this means i-cache and d-cache are shared in the SMT level. This is correct if there's single thread per core, but is wrong for the hyper threading case (one core contains multiple threads) since the i-cache and d-cache are shared in the core level other than SMT level.
Quoted 7 lines
Quoted 3 lines
Therefore, in order to be compatible with both multi-threaded and single-threaded situations, we should set i-cache and d-cache be shared at the core level by default.
It's true for VM only when the exactly HW topology is configured to VM. i.e., two virtual LPs of one virtual CORE are pinned to two physical LPs that of one physical CORE.
Yes, in this case, host and guest has the same topology, and their topology can match.
Quoted 1 line
Otherwise it's incorrect for VM.
My understanding here is that what we do in QEMU is to create self-consistent CPU topology and cache topology for the guest.
If the VM topology is self-consistent and emulated to be almost identical to the real machine, then the emulation in QEMU is correct, right? ;-)
Quoted 4 lines
for example. given a VM of 4 threads and 2 cores. If not pinning the 4 threads to physical 4 LPs of 2 CORES. It's likely each vcpu running on a LP of different physical cores.
Thanks for bringing this up, this is worth discussing.
I looked into it and found that the specific scheduling policy for the vCPU actually depends on the host setting. For example, (IIUC) if host
enables core scheduling, then host will schedule the vCPU on the SMT threads of same core.
Also, to explore the original purpose of the "per thread" i/d cache topology, I have retraced its history.
The related commit should be in '09, which is 400281a (set CPUID bits to present cores and threads topology). In this commit, the multithreading cache topology is added in CPUID.04H. In particular, here it set the L2 cache level to per core, but it did not change the level of L1 (i/d cache), that is, L1 still remains per thread.
I think that here is the problem, L1 should also be per core in multithreading case. (the fix for this patch is worth it?)
Another thing we can refer to is that AMD's i/d cache topology is per core rather than per thread (different CPUID leaf than intel): In encode_cache_cpuid8000001d() (target/i386/cpu.c), i/d cache and L2 are encoded as core level in EAX. They set up the per core supposedly to emulate the L1 topology of the real machine as well.
So, I guess this example is "unintentionally" benefiting from the "per thread" level of i/d cache.
What do you think?
Quoted 1 line
So no vcpu shares L1i/L1d cache at core level.
Yes. The scheduling of host is not guaranteed, and workload balance policies in various scenarios and some security mitigation ways may break the delicate balance we have carefully set up.
Perhaps another way is to also add a new command "x-l1-cache-topo" (like
[1] did for L2) that can adjust the i/d cache level from core to smt to benefit cases like this.
[1]: https://lists.gnu.org/archive/html/qemu-devel/2023-02/msg03201.html
Thanks,
Zhao
Re: [PATCH RESEND 04/18] i386/cpu: Fix number of addressable IDs in CPUID.04H
Quoted 29 lines · Zhao Liu
Hi Xiaoyao,
Thanks, I've spent some time thinking about it here.
Quoted 13 lines · Xiaoyao Li
Date: Mon, 20 Feb 2023 14:59:20 +0800
From: Xiaoyao Li <[email protected]>
Subject: Re: [PATCH RESEND 04/18] i386/cpu: Fix number of addressable IDs
in CPUID.04HQuoted 7 lines · Zhao Liu
For i-cache and d-cache, the maximum IDs for CPUs sharing cache ( CPUID.04H.00H:EAX[bits 25:14] and CPUID.04H.01H:EAX[bits 25:14]) are both 0, and this means i-cache and d-cache are shared in the SMT level. This is correct if there's single thread per core, but is wrong for the hyper threading case (one core contains multiple threads) since the i-cache and d-cache are shared in the core level other than SMT level.
Quoted 7 lines
Quoted 3 lines
Therefore, in order to be compatible with both multi-threaded and single-threaded situations, we should set i-cache and d-cache be shared at the core level by default.
It's true for VM only when the exactly HW topology is configured to VM. i.e., two virtual LPs of one virtual CORE are pinned to two physical LPs that of one physical CORE.
Yes, in this case, host and guest has the same topology, and their topology can match.
Quoted 8 lines
Quoted 1 line
Otherwise it's incorrect for VM.
My understanding here is that what we do in QEMU is to create self-consistent CPU topology and cache topology for the guest.
If the VM topology is self-consistent and emulated to be almost identical to the real machine, then the emulation in QEMU is correct, right? ;-)
Real machine tells two threads in the same CORE share the L1 cahche via CUPID because it's the fact and it is how exactly hardware resource lay out. However, for VM, when you tell the same thing (two threads share the L1 cache), is it true for vcpus?
The target is to emulate things correctly, not emulate it identical to real machine. In fact, for these shared resources, it's mostly infeasible to emulate correctly if not pinning vcpus to physical LPs.
Quoted 51 lines
Quoted 4 lines
for example. given a VM of 4 threads and 2 cores. If not pinning the 4 threads to physical 4 LPs of 2 CORES. It's likely each vcpu running on a LP of different physical cores.
Thanks for bringing this up, this is worth discussing.
I looked into it and found that the specific scheduling policy for the vCPU actually depends on the host setting. For example, (IIUC) if host
enables core scheduling, then host will schedule the vCPU on the SMT threads of same core.
Also, to explore the original purpose of the "per thread" i/d cache topology, I have retraced its history.
The related commit should be in '09, which is 400281a (set CPUID bits to present cores and threads topology). In this commit, the multithreading cache topology is added in CPUID.04H. In particular, here it set the L2 cache level to per core, but it did not change the level of L1 (i/d cache), that is, L1 still remains per thread.
I think that here is the problem, L1 should also be per core in multithreading case. (the fix for this patch is worth it?)
Another thing we can refer to is that AMD's i/d cache topology is per core rather than per thread (different CPUID leaf than intel): In
encode_cache_cpuid8000001d()(target/i386/cpu.c), i/d cache and L2 are encoded as core level in EAX. They set up the per core supposedly to emulate the L1 topology of the real machine as well.So, I guess this example is "unintentionally" benefiting from the "per thread" level of i/d cache.
What do you think?
Quoted 1 line
So no vcpu shares L1i/L1d cache at core level.
Yes. The scheduling of host is not guaranteed, and workload balance policies in various scenarios and some security mitigation ways may break the delicate balance we have carefully set up.
Perhaps another way is to also add a new command "x-l1-cache-topo" (like
[1] did for L2) that can adjust the i/d cache level from core to smt to benefit cases like this.[1]: https://lists.gnu.org/archive/html/qemu-devel/2023-02/msg03201.html
Thanks,
Zhao
PATCH RESEND 5/18 i386/cpu: Consolidate the use of topo_info in cpu_x86_cpuid()
In reply to Zhao LiuFrom: Zhao Liu <[email protected]>
In cpu_x86_cpuid(), there are many variables in representing the cpu topology, e.g., topo_info, cs->nr_cores/cs->nr_threads.
Since the names of cs->nr_cores/cs->nr_threads does not accurately represent its meaning, the use of cs->nr_cores/cs->nr_threads is prone to confusion and mistakes.
And the structure X86CPUTopoInfo names its memebers clearly, thus the variable "topo_info" should be preferred.
--- target/i386/cpu.c | 30 ++++++++++++++++++------------ 1 file changed, 18 insertions(+), 12 deletions(-)
target/i386/cpu.c +18 -12
@@ -5215,11 +5215,15 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,uint32_t limit;uint32_t signature[3 ];X86CPUTopoInfo topo_info; +uint32_t cpus_per_pkg; topo_info.dies_per_pkg = env->nr_dies; topo_info.cores_per_die = cs->nr_cores / env->nr_dies; topo_info.threads_per_core = cs->nr_threads; + cpus_per_pkg = topo_info.dies_per_pkg * topo_info.cores_per_die * + topo_info.threads_per_core; +/* Calculate & apply limits for different index ranges */ if (index >=0xC0000000 ) { limit = env->cpuid_xlevel2;
@@ -5255,8 +5259,8 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *ecx |=CPUID_EXT_OSXSAVE ; } *edx = env->features[FEAT_1_EDX ]; -if (cs->nr_cores * cs->nr_threads >1 ) { - *ebx |= (cs->nr_cores * cs->nr_threads) <<16 ; +if (cpus_per_pkg >1 ) { + *ebx |= cpus_per_pkg <<16 ; *edx |=CPUID_HT ; }if (!cpu->enable_pmu) {
@@ -5293,10 +5297,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,*/ if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 ); -int vcpus_per_socket = cs->nr_cores * cs->nr_threads; -if (cs->nr_cores >1 ) { +int vcpus_per_socket = cpus_per_pkg; +int cores_per_socket = topo_info.cores_per_die * + topo_info.dies_per_pkg; +if (cores_per_socket >1 ) { *eax &= ~0xFC000000 ; - *eax |= (pow2ceil (cs->nr_cores) -1 ) <<26 ; + *eax |= (pow2ceil (cores_per_socket) -1 ) <<26 ; }if (host_vcpus_per_cache > vcpus_per_socket) { *eax &= ~0x3FFC000 ;
@@ -5436,12 +5442,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,switch (count) {case 0 : *eax =apicid_core_offset (&topo_info); - *ebx = cs->nr_threads; + *ebx = topo_info.threads_per_core; *ecx |=CPUID_TOPOLOGY_LEVEL_SMT ;break ;case 1 : *eax =apicid_pkg_offset (&topo_info); - *ebx = cs->nr_cores * cs->nr_threads; + *ebx = cpus_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;default :
@@ -5472,7 +5478,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,switch (count) {case 0 : *eax =apicid_core_offset (&topo_info); - *ebx = cs->nr_threads; + *ebx = topo_info.threads_per_core; *ecx |=CPUID_TOPOLOGY_LEVEL_SMT ;break ;case 1 :
@@ -5482,7 +5488,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 2 : *eax =apicid_pkg_offset (&topo_info); - *ebx = cs->nr_cores * cs->nr_threads; + *ebx = cpus_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_DIE ;break ;default :
@@ -5707,7 +5713,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,* discards multiple thread information if it is set. * So don't set it here for Intel to make Linux guests happy. */ -if (cs->nr_cores * cs->nr_threads >1 ) { +if (cpus_per_pkg >1 ) {if (env->cpuid_vendor1 !=CPUID_VENDOR_INTEL_1 || env->cpuid_vendor2 !=CPUID_VENDOR_INTEL_2 || env->cpuid_vendor3 !=CPUID_VENDOR_INTEL_3 ) {
@@ -5769,7 +5775,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *eax |= (cpu_x86_virtual_addr_width (env) <<8 ); } *ebx = env->features[FEAT_8000_0008_EBX ]; -if (cs->nr_cores * cs->nr_threads >1 ) { +if (cpus_per_pkg >1 ) {/* * Bits 15:12 is "The number of bits in the initial * Core::X86::Apic::ApicId[ApicId] value that indicate
@@ -5777,7 +5783,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,* Bits 7:0 is "The number of threads in the package is NC+1" */ *ecx = (apicid_pkg_offset (&topo_info) <<12 ) | - ((cs->nr_cores * cs->nr_threads) -1 ); + (cpus_per_pkg -1 ); }else { *ecx =0 ; }
Signature · 1 line
2.34.1
Re: [PATCH RESEND 05/18] i386/cpu: Consolidate the use of topo_info in cpu_x86_cpuid()
Quoted 50 lines · Zhao Liu
From: Zhao Liu <[email protected]>
In
cpu_x86_cpuid(), there are many variables in representing the cpu topology, e.g.,topo_info, cs->nr_cores/cs->nr_threads.Since the names of cs->nr_cores/cs->nr_threads does not accurately represent its meaning, the use of cs->nr_cores/cs->nr_threads is prone to confusion and mistakes.
And the structure X86CPUTopoInfo names its memebers clearly, thus the variable "
topo_info" should be preferred.Suggested-by Robert Hoo <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 30 ++++++++++++++++++------------ 1 file changed, 18 insertions(+), 12 deletions(-)target/i386/cpu.c +6
-2@@ -5215,11 +5215,15 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,uint32_t limit;uint32_t signature[3 ];X86CPUTopoInfo topo_info; +uint32_t cpus_per_pkg; topo_info.dies_per_pkg = env->nr_dies; topo_info.cores_per_die = cs->nr_cores / env->nr_dies; topo_info.threads_per_core = cs->nr_threads; + cpus_per_pkg = topo_info.dies_per_pkg * topo_info.cores_per_die * + topo_info.threads_per_core; +/* Calculate & apply limits for different index ranges */ if (index >=0xC0000000 ) { limit = env->cpuid_xlevel2;@@ -5255,8 +5259,8 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *ecx |=CPUID_EXT_OSXSAVE ; } *edx = env->features[FEAT_1_EDX ]; -if (cs->nr_cores * cs->nr_threads >1 ) { - *ebx |= (cs->nr_cores * cs->nr_threads) <<16 ; +if (cpus_per_pkg >1 ) { + *ebx |= cpus_per_pkg <<16 ; *edx |=CPUID_HT ; }if (!cpu->enable_pmu) {
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5300
Quoted 6 lines
target/i386/cpu.c +1
-2@@ -5293,10 +5297,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,*/ if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 ); -int vcpus_per_socket = cs->nr_cores * cs->nr_threads; -if (cs->nr_cores >1 ) { +int vcpus_per_socket = cpus_per_pkg;
Would it make sense to directly use cpus_per_pkg here
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5302
Quoted 2 lines
target/i386/cpu.c +2
-0+int cores_per_socket = topo_info.cores_per_die * + topo_info.dies_per_pkg;
There are other places in cpu_x86_cpuid where cs->nr_cores is used separately, why not make a global "cores_per_pkg" like cpus_per_pkg and also tweak the other places?
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5445
Quoted 13 lines
target/i386/cpu.c +3
-2+if (cores_per_socket >1 ) { *eax &= ~0xFC000000 ; - *eax |= (pow2ceil (cs->nr_cores) -1 ) <<26 ; + *eax |= (pow2ceil (cores_per_socket) -1 ) <<26 ; }if (host_vcpus_per_cache > vcpus_per_socket) { *eax &= ~0x3FFC000 ;@@ -5436,12 +5442,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,switch (count) {case 0 : *eax =apicid_core_offset (&topo_info); - *ebx = cs->nr_threads; + *ebx = topo_info.threads_per_core;
There are many other places in cpu_x86_cpuid where cs->nr_threads is used separately, such as encode_cache_cpuid4(***), should we replace them all?
Quoted 46 lines
target/i386/cpu.c +5
-5*ecx |=CPUID_TOPOLOGY_LEVEL_SMT ;break ;case 1 : *eax =apicid_pkg_offset (&topo_info); - *ebx = cs->nr_cores * cs->nr_threads; + *ebx = cpus_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;default :@@ -5472,7 +5478,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,switch (count) {case 0 : *eax =apicid_core_offset (&topo_info); - *ebx = cs->nr_threads; + *ebx = topo_info.threads_per_core; *ecx |=CPUID_TOPOLOGY_LEVEL_SMT ;break ;case 1 :@@ -5482,7 +5488,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 2 : *eax =apicid_pkg_offset (&topo_info); - *ebx = cs->nr_cores * cs->nr_threads; + *ebx = cpus_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_DIE ;break ;default :@@ -5707,7 +5713,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,* discards multiple thread information if it is set. * So don't set it here for Intel to make Linux guests happy. */ -if (cs->nr_cores * cs->nr_threads >1 ) { +if (cpus_per_pkg >1 ) {if (env->cpuid_vendor1 !=CPUID_VENDOR_INTEL_1 || env->cpuid_vendor2 !=CPUID_VENDOR_INTEL_2 || env->cpuid_vendor3 !=CPUID_VENDOR_INTEL_3 ) {@@ -5769,7 +5775,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *eax |= (cpu_x86_virtual_addr_width (env) <<8 ); } *ebx = env->features[FEAT_8000_0008_EBX ]; -if (cs->nr_cores * cs->nr_threads >1 ) { +if (cpus_per_pkg >1 ) {/* * Bits 15:12 is "The number of bits in the initial * Core::X86::Apic::ApicId[ApicId] value that indicate
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5789
Quoted 8 lines
target/i386/cpu.c +1
-1@@ -5777,7 +5783,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,* Bits 7:0 is "The number of threads in the package is NC+1" */ *ecx = (apicid_pkg_offset (&topo_info) <<12 ) | - ((cs->nr_cores * cs->nr_threads) -1 ); + (cpus_per_pkg -1 ); }else { *ecx =0 ; }
Thanks,
Yanan
Re: [PATCH RESEND 05/18] i386/cpu: Consolidate the use of topo_info in cpu_x86_cpuid()
Quoted 61 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 11:28:25 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 05/18] i386/cpu: Consolidate the use of
topo_infoincpu_x86_cpuid()Quoted 48 lines · Zhao Liu
From: Zhao Liu <[email protected]>
In
cpu_x86_cpuid(), there are many variables in representing the cpu topology, e.g.,topo_info, cs->nr_cores/cs->nr_threads.Since the names of cs->nr_cores/cs->nr_threads does not accurately represent its meaning, the use of cs->nr_cores/cs->nr_threads is prone to confusion and mistakes.
And the structure X86CPUTopoInfo names its memebers clearly, thus the variable "
topo_info" should be preferred.Suggested-by Robert Hoo <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 30 ++++++++++++++++++------------ 1 file changed, 18 insertions(+), 12 deletions(-)target/i386/cpu.c +6
-2@@ -5215,11 +5215,15 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,uint32_t limit;uint32_t signature[3 ];X86CPUTopoInfo topo_info; +uint32_t cpus_per_pkg; topo_info.dies_per_pkg = env->nr_dies; topo_info.cores_per_die = cs->nr_cores / env->nr_dies; topo_info.threads_per_core = cs->nr_threads; + cpus_per_pkg = topo_info.dies_per_pkg * topo_info.cores_per_die * + topo_info.threads_per_core; +/* Calculate & apply limits for different index ranges */ if (index >=0xC0000000 ) { limit = env->cpuid_xlevel2;@@ -5255,8 +5259,8 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *ecx |=CPUID_EXT_OSXSAVE ; } *edx = env->features[FEAT_1_EDX ]; -if (cs->nr_cores * cs->nr_threads >1 ) { - *ebx |= (cs->nr_cores * cs->nr_threads) <<16 ; +if (cpus_per_pkg >1 ) { + *ebx |= cpus_per_pkg <<16 ; *edx |=CPUID_HT ; }if (!cpu->enable_pmu) {target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5300
Quoted 6 lines
target/i386/cpu.c +1
-2@@ -5293,10 +5297,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,*/ if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 ); -int vcpus_per_socket = cs->nr_cores * cs->nr_threads; -if (cs->nr_cores >1 ) { +int vcpus_per_socket = cpus_per_pkg;Would it make sense to directly use
cpus_per_pkghere
Quoted 5 lines
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5302
Quoted 2 lines
target/i386/cpu.c +2
-0+int cores_per_socket = topo_info.cores_per_die * + topo_info.dies_per_pkg;There are other places in
cpu_x86_cpuidwherecs->nr_coresis used separately, why not make a global "cores_per_pkg" likecpus_per_pkgand also tweak the other places?
Yeah, good idea.
Quoted 8 lines
Quoted 8 lines
target/i386/cpu.c +2
-1+if (cores_per_socket >1 ) { *eax &= ~0xFC000000 ; - *eax |= (pow2ceil (cs->nr_cores) -1 ) <<26 ; + *eax |= (pow2ceil (cores_per_socket) -1 ) <<26 ; }if (host_vcpus_per_cache > vcpus_per_socket) { *eax &= ~0x3FFC000 ;@@ -5436,12 +5442,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
Quoted 8 lines
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5445
Quoted 5 lines
target/i386/cpu.c +1
-1switch (count) {case 0 : *eax =apicid_core_offset (&topo_info); - *ebx = cs->nr_threads; + *ebx = topo_info.threads_per_core;There are many other places in
cpu_x86_cpuidwherecs->nr_threadsis used separately, such asencode_cache_cpuid4(***), should we replace them all?
In a previous patch [1], I replaced the use of cs->nr_threads/nr_cores in the call of encode_cache_cpuid4().
The cleanest way is to pass topo_info to encode_cache_cpuid4(), but this involves the modification of the interface format and the use of the cache topo level, so I included it in a follow-up patch [2].
[1]: [PATCH RESEND 04/18] i386/cpu: Fix number of addressable IDs in
CPUID.04,
https://lists.gnu.org/archive/html/qemu-devel/2023-02/msg03188.html
[2]: [PATCH RESEND 15/18] i386: Use CPUCacheInfo.share_level to encode
CPUID[4].EAX[bits 25:14],
https://lists.gnu.org/archive/html/qemu-devel/2023-02/msg03199.htmlQuoted 56 lines
Quoted 46 lines
target/i386/cpu.c +5
-5*ecx |=CPUID_TOPOLOGY_LEVEL_SMT ;break ;case 1 : *eax =apicid_pkg_offset (&topo_info); - *ebx = cs->nr_cores * cs->nr_threads; + *ebx = cpus_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;default :@@ -5472,7 +5478,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,switch (count) {case 0 : *eax =apicid_core_offset (&topo_info); - *ebx = cs->nr_threads; + *ebx = topo_info.threads_per_core; *ecx |=CPUID_TOPOLOGY_LEVEL_SMT ;break ;case 1 :@@ -5482,7 +5488,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 2 : *eax =apicid_pkg_offset (&topo_info); - *ebx = cs->nr_cores * cs->nr_threads; + *ebx = cpus_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_DIE ;break ;default :@@ -5707,7 +5713,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,* discards multiple thread information if it is set. * So don't set it here for Intel to make Linux guests happy. */ -if (cs->nr_cores * cs->nr_threads >1 ) { +if (cpus_per_pkg >1 ) {if (env->cpuid_vendor1 !=CPUID_VENDOR_INTEL_1 || env->cpuid_vendor2 !=CPUID_VENDOR_INTEL_2 || env->cpuid_vendor3 !=CPUID_VENDOR_INTEL_3 ) {@@ -5769,7 +5775,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *eax |= (cpu_x86_virtual_addr_width (env) <<8 ); } *ebx = env->features[FEAT_8000_0008_EBX ]; -if (cs->nr_cores * cs->nr_threads >1 ) { +if (cpus_per_pkg >1 ) {/* * Bits 15:12 is "The number of bits in the initial * Core::X86::Apic::ApicId[ApicId] value that indicate target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5789
Quoted 8 lines
target/i386/cpu.c +1
-1@@ -5777,7 +5783,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,* Bits 7:0 is "The number of threads in the package is NC+1" */ *ecx = (apicid_pkg_offset (&topo_info) <<12 ) | - ((cs->nr_cores * cs->nr_threads) -1 ); + (cpus_per_pkg -1 ); }else { *ecx =0 ; }Thanks,
Yanan
Re: [PATCH RESEND 05/18] i386/cpu: Consolidate the use of topo_info in cpu_x86_cpuid()
Quoted 85 lines · Zhao Liu
Quoted 61 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 11:28:25 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 05/18] i386/cpu: Consolidate the use of
topo_infoincpu_x86_cpuid()Quoted 48 lines · Zhao Liu
From: Zhao Liu <[email protected]>
In
cpu_x86_cpuid(), there are many variables in representing the cpu topology, e.g.,topo_info, cs->nr_cores/cs->nr_threads.Since the names of cs->nr_cores/cs->nr_threads does not accurately represent its meaning, the use of cs->nr_cores/cs->nr_threads is prone to confusion and mistakes.
And the structure X86CPUTopoInfo names its memebers clearly, thus the variable "
topo_info" should be preferred.Suggested-by Robert Hoo <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 30 ++++++++++++++++++------------ 1 file changed, 18 insertions(+), 12 deletions(-)target/i386/cpu.c +6
-2@@ -5215,11 +5215,15 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,uint32_t limit;uint32_t signature[3 ];X86CPUTopoInfo topo_info; +uint32_t cpus_per_pkg; topo_info.dies_per_pkg = env->nr_dies; topo_info.cores_per_die = cs->nr_cores / env->nr_dies; topo_info.threads_per_core = cs->nr_threads; + cpus_per_pkg = topo_info.dies_per_pkg * topo_info.cores_per_die * + topo_info.threads_per_core; +/* Calculate & apply limits for different index ranges */ if (index >=0xC0000000 ) { limit = env->cpuid_xlevel2;@@ -5255,8 +5259,8 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *ecx |=CPUID_EXT_OSXSAVE ; } *edx = env->features[FEAT_1_EDX ]; -if (cs->nr_cores * cs->nr_threads >1 ) { - *ebx |= (cs->nr_cores * cs->nr_threads) <<16 ; +if (cpus_per_pkg >1 ) { + *ebx |= cpus_per_pkg <<16 ; *edx |=CPUID_HT ; }if (!cpu->enable_pmu) {target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5300
Quoted 6 lines
target/i386/cpu.c +1
-2@@ -5293,10 +5297,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,*/ if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 ); -int vcpus_per_socket = cs->nr_cores * cs->nr_threads; -if (cs->nr_cores >1 ) { +int vcpus_per_socket = cpus_per_pkg;Would it make sense to directly use
cpus_per_pkghereQuoted 5 lines
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5302
Quoted 2 lines
target/i386/cpu.c +2
-0+int cores_per_socket = topo_info.cores_per_die * + topo_info.dies_per_pkg;There are other places in
cpu_x86_cpuidwherecs->nr_coresis used separately, why not make a global "cores_per_pkg" likecpus_per_pkgand also tweak the other places?Yeah, good idea.
Quoted 8 lines
Quoted 8 lines
target/i386/cpu.c +2
-1+if (cores_per_socket >1 ) { *eax &= ~0xFC000000 ; - *eax |= (pow2ceil (cs->nr_cores) -1 ) <<26 ; + *eax |= (pow2ceil (cores_per_socket) -1 ) <<26 ; }if (host_vcpus_per_cache > vcpus_per_socket) { *eax &= ~0x3FFC000 ;@@ -5436,12 +5442,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,Quoted 8 lines
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5445
Quoted 5 lines
target/i386/cpu.c +1
-1switch (count) {case 0 : *eax =apicid_core_offset (&topo_info); - *ebx = cs->nr_threads; + *ebx = topo_info.threads_per_core;There are many other places in
cpu_x86_cpuidwherecs->nr_threadsis used separately, such asencode_cache_cpuid4(***), should we replace them all?
Quoted 6 lines
In a previous patch [1], I replaced the use of cs->nr_threads/nr_cores in the call of
encode_cache_cpuid4().The cleanest way is to pass
topo_infotoencode_cache_cpuid4(), but this involves the modification of the interface format and the use of the cache topo level, so I included it in a follow-up patch [2].
Ok, I see. I have not reached there.😉
Quoted 64 lines
[1]: [PATCH RESEND 04/18] i386/cpu: Fix number of addressable IDs in
CPUID.04, https://lists.gnu.org/archive/html/qemu-devel/2023-02/msg03188.html [2]: [PATCH RESEND 15/18] i386: Use CPUCacheInfo.share_level to encode CPUID[4].EAX[bits 25:14], https://lists.gnu.org/archive/html/qemu-devel/2023-02/msg03199.htmlQuoted 56 lines
Quoted 46 lines
target/i386/cpu.c +5
-5*ecx |=CPUID_TOPOLOGY_LEVEL_SMT ;break ;case 1 : *eax =apicid_pkg_offset (&topo_info); - *ebx = cs->nr_cores * cs->nr_threads; + *ebx = cpus_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;default :@@ -5472,7 +5478,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,switch (count) {case 0 : *eax =apicid_core_offset (&topo_info); - *ebx = cs->nr_threads; + *ebx = topo_info.threads_per_core; *ecx |=CPUID_TOPOLOGY_LEVEL_SMT ;break ;case 1 :@@ -5482,7 +5488,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 2 : *eax =apicid_pkg_offset (&topo_info); - *ebx = cs->nr_cores * cs->nr_threads; + *ebx = cpus_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_DIE ;break ;default :@@ -5707,7 +5713,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,* discards multiple thread information if it is set. * So don't set it here for Intel to make Linux guests happy. */ -if (cs->nr_cores * cs->nr_threads >1 ) { +if (cpus_per_pkg >1 ) {if (env->cpuid_vendor1 !=CPUID_VENDOR_INTEL_1 || env->cpuid_vendor2 !=CPUID_VENDOR_INTEL_2 || env->cpuid_vendor3 !=CPUID_VENDOR_INTEL_3 ) {@@ -5769,7 +5775,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *eax |= (cpu_x86_virtual_addr_width (env) <<8 ); } *ebx = env->features[FEAT_8000_0008_EBX ]; -if (cs->nr_cores * cs->nr_threads >1 ) { +if (cpus_per_pkg >1 ) {/* * Bits 15:12 is "The number of bits in the initial * Core::X86::Apic::ApicId[ApicId] value that indicate target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5789
Quoted 8 lines
target/i386/cpu.c +1
-1@@ -5777,7 +5783,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,* Bits 7:0 is "The number of threads in the package is NC+1" */ *ecx = (apicid_pkg_offset (&topo_info) <<12 ) | - ((cs->nr_cores * cs->nr_threads) -1 ); + (cpus_per_pkg -1 ); }else { *ecx =0 ; }Thanks,
Yanan
PATCH RESEND 6/18 i386: Introduce module-level cpu topology to CPUX86State
In reply to Zhao LiuFrom: Zhuocheng Ding <[email protected]>
smp command has the "clusters" parameter but x86 hasn't supported that level. Though "clusters" was introduced to help define L2 cache topology [1], using cluster to define x86's L2 cache topology will cause the compatibility problem:
Currently, x86 defaults that the L2 cache is shared in one core, which actually implies a default setting "cores per L2 cache is 1" and therefore implicitly defaults to having as many L2 caches as cores.
For example (i386 PC machine):
-smp 16,sockets=2,dies=2,cores=2,threads=2,maxcpus=16 (*)
Considering the topology of the L2 cache, this (*) implicitly means "1 core per L2 cache" and "2 L2 caches per die".
If we use cluster to configure L2 cache topology with the new default setting "clusters per L2 cache is 1", the above semantics will change to "2 cores per cluster" and "1 cluster per L2 cache", that is, "2 cores per L2 cache".
So the same command (*) will cause changes in the L2 cache topology, further affecting the performance of the virtual machine.
Therefore, x86 should only treat cluster as a cpu topology level and avoid using it to change L2 cache by default for compatibility.
"cluster" in smp is the CPU topology level which is between "core" and die.
For x86, the "cluster" in smp is corresponding to the module level [2], which is above the core level. So use the "module" other than "cluster" in i386 code.
And please note that x86 already has a cpu topology level also named "cluster" [2], this level is at the upper level of the package. Here, the cluster in x86 cpu topology is completely different from the "clusters" as the smp parameter. After the module level is introduced, the cluster as the smp parameter will actually refer to the module level of x86.
[1]: 0d87178 (hw/core/machine: Introduce CPU cluster topology support) [2]: SDM, vol.3, ch.9, 9.9.1 Hierarchical Mapping of Shared Resources.
--- hw/i386/x86.c | 1 + target/i386/cpu.c | 1 + target/i386/cpu.h | 6 ++++++ 3 files changed, 8 insertions(+)
hw/i386/x86.c +1 -0
@@ -306,6 +306,7 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,init_topo_info (&topo_info, x86ms); env->nr_dies = ms->smp.dies; + env->nr_modules = ms->smp.clusters;/* * If APIC ID is not set,
target/i386/cpu.c +1 -0
@@ -6781,6 +6781,7 @@ static void x86_cpu_initfn(Object *obj)CPUX86State *env = &cpu->env; env->nr_dies =1 ; + env->nr_modules =1 ;cpu_set_cpustate_pointers (cpu); object_property_add(obj, "feature-words", "X86CPUFeatureWordInfo",
target/i386/cpu.h +6 -0
@@ -1810,7 +1810,13 @@ typedef struct CPUArchState {TPRAccess tpr_access_type; +/* Number of dies per package. */ unsigned nr_dies; +/* +* Number of modules per die. Module level in x86 cpu topology is +* corresponding to smp.clusters. +*/ +unsigned nr_modules; } CPUX86State;struct kvm_msrs ;
Signature · 1 line
2.34.1
Re: [PATCH RESEND 06/18] i386: Introduce module-level cpu topology to CPUX86State
Quoted 6 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
smp command has the "clusters" parameter but x86 hasn't supported that level. Though "clusters" was introduced to help define L2 cache topology [1], using cluster to define x86's L2 cache topology will cause the compatibility problem:
Well, the smp parameter "clusters" isn't destined to define L2 cache topology. It's actually a CPU topology level concept above cores, in which the cores may share some resources (the resources can be L2 cache or some others like L3 cache tags, depending on the Archs).
On some ARM64 chips, cores in the same cluster share a L2 and hold their own L1D/I separately. There are also chips, where cores in the same cluster have their own L2 & L1D/I cache separately, and share a L3 cache tag.
Quoted 15 lines
Currently, x86 defaults that the L2 cache is shared in one core, which actually implies a default setting "cores per L2 cache is 1" and therefore implicitly defaults to having as many L2 caches as cores.
For example (i386 PC machine):
-smp 16,sockets=2,dies=2,cores=2,threads=2,maxcpus=16 (*)Considering the topology of the L2 cache, this (*) implicitly means "1 core per L2 cache" and "2 L2 caches per die".
If we use cluster to configure L2 cache topology with the new default setting "clusters per L2 cache is 1", the above semantics will change to "2 cores per cluster" and "1 cluster per L2 cache", that is, "2 cores per L2 cache".
Quoted 5 lines
So the same command (*) will cause changes in the L2 cache topology, further affecting the performance of the virtual machine.
Therefore, x86 should only treat cluster as a cpu topology level and avoid using it to change L2 cache by default for compatibility.
Agree. I think all the smp parameters only indicates the CPU hierarchy, while the cache layout is much more flexible.
Quoted 8 lines
"cluster" in smp is the CPU topology level which is between "core" and die.
For x86, the "cluster" in smp is corresponding to the module level [2], which is above the core level. So use the "module" other than "cluster" in i386 code.
Quoted 6 lines
And please note that x86 already has a cpu topology level also named "cluster" [2], this level is at the upper level of the package. Here, the cluster in x86 cpu topology is completely different from the "clusters" as the smp parameter. After the module level is introduced, the cluster as the smp parameter will actually refer to the module level of x86.
I see. So the reason for use of "module" instead of "cluster" is that there is already cluster concept above Package in the x86 reference.
Thanks,
Yanan
Quoted 54 lines
[1]: 0d87178 (hw/core/machine: Introduce CPU cluster topology support) [2]: SDM, vol.3, ch.9, 9.9.1 Hierarchical Mapping of Shared Resources.
Signed-off-by Zhuocheng Ding <[email protected]>Co-developed-by Zhao Liu <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- hw/i386/x86.c | 1 + target/i386/cpu.c | 1 + target/i386/cpu.h | 6 ++++++ 3 files changed, 8 insertions(+)hw/i386/x86.c +1
-0@@ -306,6 +306,7 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,init_topo_info (&topo_info, x86ms); env->nr_dies = ms->smp.dies; + env->nr_modules = ms->smp.clusters;/* * If APIC ID is not set, target/i386/cpu.c +1
-0@@ -6781,6 +6781,7 @@ static void x86_cpu_initfn(Object *obj)CPUX86State *env = &cpu->env; env->nr_dies =1 ; + env->nr_modules =1 ;cpu_set_cpustate_pointers (cpu); object_property_add(obj, "feature-words", "X86CPUFeatureWordInfo",target/i386/cpu.h +6
-0@@ -1810,7 +1810,13 @@ typedef struct CPUArchState {TPRAccess tpr_access_type; +/* Number of dies per package. */ unsigned nr_dies; +/* +* Number of modules per die. Module level in x86 cpu topology is +* corresponding to smp.clusters. +*/ +unsigned nr_modules; } CPUX86State;struct kvm_msrs ;
PATCH RESEND 7/18 i386: Support modules_per_die in X86CPUTopoInfo
In reply to Zhao LiuFrom: Zhuocheng Ding <[email protected]>
Support module level in i386 cpu topology structure "X86CPUTopoInfo".
Before updating APIC ID parsing rule with module level, the apicid_core_width() temporarily combines the core and module levels together.
At present, we don't expose module level in CPUID.1FH because currently linux (v6.2-rc6) doesn't support module level. And exposing module and die levels at the same time in CPUID.1FH will cause linux to calculate the wrong die_id. The module level should be exposed until the real machine has the module level in CPUID.1FH.
In addition, update topology structure in test-x86-apicid.c.
--- hw/i386/x86.c | 3 ++- include/hw/i386/topology.h | 13 ++++++++--- target/i386/cpu.c | 17 ++++++++------ tests/unit/test-x86-apicid.c | 45 +++++++++++++++++++----------------- 4 files changed, 46 insertions(+), 32 deletions(-)
hw/i386/x86.c +2 -1
@@ -71,7 +71,8 @@ inline void init_topo_info(X86CPUTopoInfo *topo_info,MachineState *ms =MACHINE (x86ms); topo_info->dies_per_pkg = ms->smp.dies; - topo_info->cores_per_die = ms->smp.cores; + topo_info->modules_per_die = ms->smp.clusters; + topo_info->cores_per_module = ms->smp.cores; topo_info->threads_per_core = ms->smp.threads; }
include/hw/i386/topology.h +10 -3
@@ -54,7 +54,8 @@ typedef struct X86CPUTopoIDs {typedef struct X86CPUTopoInfo {unsigned dies_per_pkg; -unsigned cores_per_die; +unsigned modules_per_die; +unsigned cores_per_module;unsigned threads_per_core; }X86CPUTopoInfo ;
@@ -78,7 +79,12 @@ static inline unsigned apicid_smt_width(X86CPUTopoInfo *topo_info)*/ static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) { -return apicid_bitwidth_for_count (topo_info->cores_per_die); +/* +* TODO: Will separate module info from core_width when update +* APIC ID with module level. +*/ +return apicid_bitwidth_for_count (topo_info->cores_per_module * + topo_info->modules_per_die); }/* Bit width of the Die_ID field */
@@ -128,7 +134,8 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info, X86CPUTopoIDs *topo_ids) {unsigned nr_dies = topo_info->dies_per_pkg; -unsigned nr_cores = topo_info->cores_per_die; +unsigned nr_cores = topo_info->cores_per_module * + topo_info->modules_per_die;unsigned nr_threads = topo_info->threads_per_core; topo_ids->pkg_id = cpu_index / (nr_dies * nr_cores * nr_threads);
target/i386/cpu.c +10 -7
@@ -336,7 +336,9 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache,/* L3 is shared among multiple cores */ if (cache->level ==3 ) { - l3_threads = topo_info->cores_per_die * topo_info->threads_per_core; + l3_threads = topo_info->modules_per_die * + topo_info->cores_per_module * + topo_info->threads_per_core; *eax |= (l3_threads -1 ) <<14 ; }else { *eax |= ((topo_info->threads_per_core -1 ) <<14 );
@@ -5218,11 +5220,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,uint32_t cpus_per_pkg; topo_info.dies_per_pkg = env->nr_dies; - topo_info.cores_per_die = cs->nr_cores / env->nr_dies; + topo_info.modules_per_die = env->nr_modules; + topo_info.cores_per_module = cs->nr_cores / env->nr_dies / env->nr_modules; topo_info.threads_per_core = cs->nr_threads; - cpus_per_pkg = topo_info.dies_per_pkg * topo_info.cores_per_die * - topo_info.threads_per_core; + cpus_per_pkg = topo_info.dies_per_pkg * topo_info.modules_per_die * + topo_info.cores_per_module * topo_info.threads_per_core;/* Calculate & apply limits for different index ranges */ if (index >=0xC0000000 ) {
@@ -5298,8 +5301,8 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 );int vcpus_per_socket = cpus_per_pkg; -int cores_per_socket = topo_info.cores_per_die * - topo_info.dies_per_pkg; +int cores_per_socket = cpus_per_pkg / + topo_info.threads_per_core;if (cores_per_socket >1 ) { *eax &= ~0xFC000000 ; *eax |= (pow2ceil (cores_per_socket) -1 ) <<26 ;
@@ -5483,7 +5486,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 1 : *eax =apicid_die_offset (&topo_info); - *ebx = topo_info.cores_per_die * topo_info.threads_per_core; + *ebx = cpus_per_pkg / topo_info.dies_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;case 2 :
tests/unit/test-x86-apicid.c +24 -21
94 lines not on this page · raw message
Signature · 1 line
2.34.1
Re: [PATCH RESEND 07/18] i386: Support modules_per_die in X86CPUTopoInfo
Quoted 110 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
Support module level in i386 cpu topology structure "X86CPUTopoInfo".
Before updating APIC ID parsing rule with module level, the
apicid_core_width()temporarily combines the core and module levels together.At present, we don't expose module level in CPUID.1FH because currently linux (v6.2-rc6) doesn't support module level. And exposing module and die levels at the same time in CPUID.1FH will cause linux to calculate the wrong
die_id. The module level should be exposed until the real machine has the module level in CPUID.1FH.In addition, update topology structure in
test-x86-apicid.c.Signed-off-by Zhuocheng Ding <[email protected]>Co-developed-by Zhao Liu <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- hw/i386/x86.c | 3 ++- include/hw/i386/topology.h | 13 ++++++++--- target/i386/cpu.c | 17 ++++++++------ tests/unit/test-x86-apicid.c | 45 +++++++++++++++++++----------------- 4 files changed, 46 insertions(+), 32 deletions(-)hw/i386/x86.c +2
-1@@ -71,7 +71,8 @@ inline void init_topo_info(X86CPUTopoInfo *topo_info,MachineState *ms =MACHINE (x86ms); topo_info->dies_per_pkg = ms->smp.dies; - topo_info->cores_per_die = ms->smp.cores; + topo_info->modules_per_die = ms->smp.clusters; + topo_info->cores_per_module = ms->smp.cores; topo_info->threads_per_core = ms->smp.threads; }include/hw/i386/topology.h +10
-3@@ -54,7 +54,8 @@ typedef struct X86CPUTopoIDs {typedef struct X86CPUTopoInfo {unsigned dies_per_pkg; -unsigned cores_per_die; +unsigned modules_per_die; +unsigned cores_per_module;unsigned threads_per_core; }X86CPUTopoInfo ;@@ -78,7 +79,12 @@ static inline unsigned apicid_smt_width(X86CPUTopoInfo *topo_info)*/ static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) { -return apicid_bitwidth_for_count (topo_info->cores_per_die); +/* +* TODO: Will separate module info from core_width when update +* APIC ID with module level. +*/ +return apicid_bitwidth_for_count (topo_info->cores_per_module * + topo_info->modules_per_die); }/* Bit width of the Die_ID field */ @@ -128,7 +134,8 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info, X86CPUTopoIDs *topo_ids) {unsigned nr_dies = topo_info->dies_per_pkg; -unsigned nr_cores = topo_info->cores_per_die; +unsigned nr_cores = topo_info->cores_per_module * + topo_info->modules_per_die;unsigned nr_threads = topo_info->threads_per_core; topo_ids->pkg_id = cpu_index / (nr_dies * nr_cores * nr_threads);target/i386/cpu.c +7
-4@@ -336,7 +336,9 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache,/* L3 is shared among multiple cores */ if (cache->level ==3 ) { - l3_threads = topo_info->cores_per_die * topo_info->threads_per_core; + l3_threads = topo_info->modules_per_die * + topo_info->cores_per_module * + topo_info->threads_per_core; *eax |= (l3_threads -1 ) <<14 ; }else { *eax |= ((topo_info->threads_per_core -1 ) <<14 );@@ -5218,11 +5220,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,uint32_t cpus_per_pkg; topo_info.dies_per_pkg = env->nr_dies; - topo_info.cores_per_die = cs->nr_cores / env->nr_dies; + topo_info.modules_per_die = env->nr_modules; + topo_info.cores_per_module = cs->nr_cores / env->nr_dies / env->nr_modules; topo_info.threads_per_core = cs->nr_threads; - cpus_per_pkg = topo_info.dies_per_pkg * topo_info.cores_per_die * - topo_info.threads_per_core; + cpus_per_pkg = topo_info.dies_per_pkg * topo_info.modules_per_die * + topo_info.cores_per_module * topo_info.threads_per_core;/* Calculate & apply limits for different index ranges */ if (index >=0xC0000000 ) {
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5305
Quoted 7 lines
target/i386/cpu.c +2
-2@@ -5298,8 +5301,8 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 );int vcpus_per_socket = cpus_per_pkg; -int cores_per_socket = topo_info.cores_per_die * - topo_info.dies_per_pkg; +int cores_per_socket = cpus_per_pkg / + topo_info.threads_per_core;
As mentioned in patch 5, cores_per_socket can be function-global.
Quoted 113 lines
target/i386/cpu.c +1
-1if (cores_per_socket >1 ) { *eax &= ~0xFC000000 ; *eax |= (pow2ceil (cores_per_socket) -1 ) <<26 ;@@ -5483,7 +5486,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 1 : *eax =apicid_die_offset (&topo_info); - *ebx = topo_info.cores_per_die * topo_info.threads_per_core; + *ebx = cpus_per_pkg / topo_info.dies_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;case 2 :tests/unit/test-x86-apicid.c +24
-21@@ -30,13 +30,16 @@ static void test_topo_bits(void) {X86CPUTopoInfo topo_info = {0 }; -/* simple tests for 1 thread per core, 1 core per die, 1 die per package */ - topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 }; +/* +* simple tests for 1 thread per core, 1 core per module, +* 1 module per die, 1 die per package +*/ + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_core_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 0), ==, 0); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1), ==, 1); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 2), ==, 2);@@ -45,39 +48,39 @@ static void test_topo_bits(void)/* Test field width calculation for multiple values */ - topo_info = (X86CPUTopoInfo ) {1 ,1 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,2 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 1); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,3 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,4 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,4 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,14 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,14 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,15 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,15 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,16 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,16 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,17 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,17 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,31 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,31 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,32 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,32 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,33 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,33 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 6); - topo_info = (X86CPUTopoInfo ) {1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); - topo_info = (X86CPUTopoInfo ) {2 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {2 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 1); - topo_info = (X86CPUTopoInfo ) {3 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {3 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {4 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {4 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2);/* build a weird topology and see if IDs are calculated correctly @@ -85,18 +88,18 @@ static void test_topo_bits(void)/* This will use 2 bits for thread ID and 3 bits for core ID */ - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); g_assert_cmpuint(apicid_core_offset(&topo_info), ==, 2); g_assert_cmpuint(apicid_die_offset(&topo_info), ==, 5); g_assert_cmpuint(apicid_pkg_offset(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 0), ==, 0); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1), ==, 1); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 2), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1 * 3 + 0), ==, (1 <<2 ) |0 ); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1 * 3 + 1), ==,
Re: [PATCH RESEND 07/18] i386: Support modules_per_die in X86CPUTopoInfo
Quoted 106 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 18:38:31 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 07/18] i386: Supportmodules_per_diein
X86CPUTopoInfoQuoted 100 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
Support module level in i386 cpu topology structure "X86CPUTopoInfo".
Before updating APIC ID parsing rule with module level, the
apicid_core_width()temporarily combines the core and module levels together.At present, we don't expose module level in CPUID.1FH because currently linux (v6.2-rc6) doesn't support module level. And exposing module and die levels at the same time in CPUID.1FH will cause linux to calculate the wrong
die_id. The module level should be exposed until the real machine has the module level in CPUID.1FH.In addition, update topology structure in
test-x86-apicid.c.Signed-off-by Zhuocheng Ding <[email protected]>Co-developed-by Zhao Liu <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- hw/i386/x86.c | 3 ++- include/hw/i386/topology.h | 13 ++++++++--- target/i386/cpu.c | 17 ++++++++------ tests/unit/test-x86-apicid.c | 45 +++++++++++++++++++----------------- 4 files changed, 46 insertions(+), 32 deletions(-)hw/i386/x86.c +2
-1@@ -71,7 +71,8 @@ inline void init_topo_info(X86CPUTopoInfo *topo_info,MachineState *ms =MACHINE (x86ms); topo_info->dies_per_pkg = ms->smp.dies; - topo_info->cores_per_die = ms->smp.cores; + topo_info->modules_per_die = ms->smp.clusters; + topo_info->cores_per_module = ms->smp.cores; topo_info->threads_per_core = ms->smp.threads; }include/hw/i386/topology.h +10
-3@@ -54,7 +54,8 @@ typedef struct X86CPUTopoIDs {typedef struct X86CPUTopoInfo {unsigned dies_per_pkg; -unsigned cores_per_die; +unsigned modules_per_die; +unsigned cores_per_module;unsigned threads_per_core; }X86CPUTopoInfo ;@@ -78,7 +79,12 @@ static inline unsigned apicid_smt_width(X86CPUTopoInfo *topo_info)*/ static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) { -return apicid_bitwidth_for_count (topo_info->cores_per_die); +/* +* TODO: Will separate module info from core_width when update +* APIC ID with module level. +*/ +return apicid_bitwidth_for_count (topo_info->cores_per_module * + topo_info->modules_per_die); }/* Bit width of the Die_ID field */ @@ -128,7 +134,8 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info, X86CPUTopoIDs *topo_ids) {unsigned nr_dies = topo_info->dies_per_pkg; -unsigned nr_cores = topo_info->cores_per_die; +unsigned nr_cores = topo_info->cores_per_module * + topo_info->modules_per_die;unsigned nr_threads = topo_info->threads_per_core; topo_ids->pkg_id = cpu_index / (nr_dies * nr_cores * nr_threads);target/i386/cpu.c +7
-4@@ -336,7 +336,9 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache,/* L3 is shared among multiple cores */ if (cache->level ==3 ) { - l3_threads = topo_info->cores_per_die * topo_info->threads_per_core; + l3_threads = topo_info->modules_per_die * + topo_info->cores_per_module * + topo_info->threads_per_core; *eax |= (l3_threads -1 ) <<14 ; }else { *eax |= ((topo_info->threads_per_core -1 ) <<14 );@@ -5218,11 +5220,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,uint32_t cpus_per_pkg; topo_info.dies_per_pkg = env->nr_dies; - topo_info.cores_per_die = cs->nr_cores / env->nr_dies; + topo_info.modules_per_die = env->nr_modules; + topo_info.cores_per_module = cs->nr_cores / env->nr_dies / env->nr_modules; topo_info.threads_per_core = cs->nr_threads; - cpus_per_pkg = topo_info.dies_per_pkg * topo_info.cores_per_die * - topo_info.threads_per_core; + cpus_per_pkg = topo_info.dies_per_pkg * topo_info.modules_per_die * + topo_info.cores_per_module * topo_info.threads_per_core;/* Calculate & apply limits for different index ranges */ if (index >=0xC0000000 ) {@@ -5298,8 +5301,8 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
Quoted 8 lines
target/i386/cpu.c void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, · line 5305
Quoted 7 lines
target/i386/cpu.c +2
-2if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 );int vcpus_per_socket = cpus_per_pkg; -int cores_per_socket = topo_info.cores_per_die * - topo_info.dies_per_pkg; +int cores_per_socket = cpus_per_pkg / + topo_info.threads_per_core;As mentioned in patch 5,
cores_per_socketcan be function-global.
Yeah, got it.
Quoted 103 lines
Quoted 102 lines
target/i386/cpu.c +1
-1if (cores_per_socket >1 ) { *eax &= ~0xFC000000 ; *eax |= (pow2ceil (cores_per_socket) -1 ) <<26 ;@@ -5483,7 +5486,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 1 : *eax =apicid_die_offset (&topo_info); - *ebx = topo_info.cores_per_die * topo_info.threads_per_core; + *ebx = cpus_per_pkg / topo_info.dies_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;case 2 :tests/unit/test-x86-apicid.c +24
-21@@ -30,13 +30,16 @@ static void test_topo_bits(void) {X86CPUTopoInfo topo_info = {0 }; -/* simple tests for 1 thread per core, 1 core per die, 1 die per package */ - topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 }; +/* +* simple tests for 1 thread per core, 1 core per module, +* 1 module per die, 1 die per package +*/ + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_core_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 0), ==, 0); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1), ==, 1); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 2), ==, 2);@@ -45,39 +48,39 @@ static void test_topo_bits(void)/* Test field width calculation for multiple values */ - topo_info = (X86CPUTopoInfo ) {1 ,1 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,2 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 1); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,3 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,4 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,4 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,14 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,14 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,15 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,15 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,16 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,16 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,17 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,17 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,31 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,31 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,32 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,32 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,33 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,33 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 6); - topo_info = (X86CPUTopoInfo ) {1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); - topo_info = (X86CPUTopoInfo ) {2 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {2 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 1); - topo_info = (X86CPUTopoInfo ) {3 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {3 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {4 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {4 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2);/* build a weird topology and see if IDs are calculated correctly @@ -85,18 +88,18 @@ static void test_topo_bits(void)/* This will use 2 bits for thread ID and 3 bits for core ID */ - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); g_assert_cmpuint(apicid_core_offset(&topo_info), ==, 2); g_assert_cmpuint(apicid_die_offset(&topo_info), ==, 5); g_assert_cmpuint(apicid_pkg_offset(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 0), ==, 0); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1), ==, 1); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 2), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1 * 3 + 0), ==, (1 <<2 ) |0 ); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1 * 3 + 1), ==,
Re: [PATCH RESEND 07/18] i386: Support modules_per_die in X86CPUTopoInfo
In reply to Zhao LiuHi Zhao,
Quoted 7 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
Support module level in i386 cpu topology structure "X86CPUTopoInfo".
Before updating APIC ID parsing rule with module level, the
apicid_core_width()temporarily combines the core and module levels together.
If we dont merge this one with the followed patches, then nits below may be meaningful.
Quoted 23 lines
At present, we don't expose module level in CPUID.1FH because currently linux (v6.2-rc6) doesn't support module level. And exposing module and die levels at the same time in CPUID.1FH will cause linux to calculate the wrong
die_id. The module level should be exposed until the real machine has the module level in CPUID.1FH.In addition, update topology structure in
test-x86-apicid.c.Signed-off-by Zhuocheng Ding <[email protected]>Co-developed-by Zhao Liu <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- hw/i386/x86.c | 3 ++- include/hw/i386/topology.h | 13 ++++++++--- target/i386/cpu.c | 17 ++++++++------ tests/unit/test-x86-apicid.c | 45 +++++++++++++++++++----------------- 4 files changed, 46 insertions(+), 32 deletions(-)hw/i386/x86.c +0
-0
hw/i386/x86.c inline void init_topo_info(X86CPUTopoInfo *topo_info, · line 75
Quoted 6 lines
hw/i386/x86.c +2
-1@@ -71,7 +71,8 @@ inline void init_topo_info(X86CPUTopoInfo *topo_info,MachineState *ms =MACHINE (x86ms); topo_info->dies_per_pkg = ms->smp.dies; - topo_info->cores_per_die = ms->smp.cores; + topo_info->modules_per_die = ms->smp.clusters; + topo_info->cores_per_module = ms->smp.cores;
Here we can ensure that topo_info->modules_per_die is always 1, so...
Quoted 19 lines
hw/i386/x86.c +0
-0topo_info->threads_per_core = ms->smp.threads; }include/hw/i386/topology.h +2
-1@@ -54,7 +54,8 @@ typedef struct X86CPUTopoIDs {typedef struct X86CPUTopoInfo {unsigned dies_per_pkg; -unsigned cores_per_die; +unsigned modules_per_die; +unsigned cores_per_module;unsigned threads_per_core; }X86CPUTopoInfo ;@@ -78,7 +79,12 @@ static inline unsigned apicid_smt_width(X86CPUTopoInfo *topo_info)*/
include/hw/i386/topology.h static inline unsigned apicid_smt_width(X86CPUTopoInfo *topo_info) · line 88
Quoted 10 lines
include/hw/i386/topology.h +6
-1static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) { -return apicid_bitwidth_for_count (topo_info->cores_per_die); +/* +* TODO: Will separate module info from core_width when update +* APIC ID with module level. +*/ +return apicid_bitwidth_for_count (topo_info->cores_per_module * + topo_info->modules_per_die); }
...We can directly add apicid_module_width (which returns 0 so far) and apicid_module_offset here which don't rely on the APIC ID rule change, and avoid the "TODO..".
Then patch 8 and 10 are about module_id, so can be merged. Is this good?
Thanks,
Yanan
Quoted 163 lines
include/hw/i386/topology.h +2
-1/* Bit width of the Die_ID field */ @@ -128,7 +134,8 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info, X86CPUTopoIDs *topo_ids) {unsigned nr_dies = topo_info->dies_per_pkg; -unsigned nr_cores = topo_info->cores_per_die; +unsigned nr_cores = topo_info->cores_per_module * + topo_info->modules_per_die;unsigned nr_threads = topo_info->threads_per_core; topo_ids->pkg_id = cpu_index / (nr_dies * nr_cores * nr_threads);target/i386/cpu.c +10
-7@@ -336,7 +336,9 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache,/* L3 is shared among multiple cores */ if (cache->level ==3 ) { - l3_threads = topo_info->cores_per_die * topo_info->threads_per_core; + l3_threads = topo_info->modules_per_die * + topo_info->cores_per_module * + topo_info->threads_per_core; *eax |= (l3_threads -1 ) <<14 ; }else { *eax |= ((topo_info->threads_per_core -1 ) <<14 );@@ -5218,11 +5220,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,uint32_t cpus_per_pkg; topo_info.dies_per_pkg = env->nr_dies; - topo_info.cores_per_die = cs->nr_cores / env->nr_dies; + topo_info.modules_per_die = env->nr_modules; + topo_info.cores_per_module = cs->nr_cores / env->nr_dies / env->nr_modules; topo_info.threads_per_core = cs->nr_threads; - cpus_per_pkg = topo_info.dies_per_pkg * topo_info.cores_per_die * - topo_info.threads_per_core; + cpus_per_pkg = topo_info.dies_per_pkg * topo_info.modules_per_die * + topo_info.cores_per_module * topo_info.threads_per_core;/* Calculate & apply limits for different index ranges */ if (index >=0xC0000000 ) {@@ -5298,8 +5301,8 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 );int vcpus_per_socket = cpus_per_pkg; -int cores_per_socket = topo_info.cores_per_die * - topo_info.dies_per_pkg; +int cores_per_socket = cpus_per_pkg / + topo_info.threads_per_core;if (cores_per_socket >1 ) { *eax &= ~0xFC000000 ; *eax |= (pow2ceil (cores_per_socket) -1 ) <<26 ;@@ -5483,7 +5486,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 1 : *eax =apicid_die_offset (&topo_info); - *ebx = topo_info.cores_per_die * topo_info.threads_per_core; + *ebx = cpus_per_pkg / topo_info.dies_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;case 2 :tests/unit/test-x86-apicid.c +24
-21@@ -30,13 +30,16 @@ static void test_topo_bits(void) {X86CPUTopoInfo topo_info = {0 }; -/* simple tests for 1 thread per core, 1 core per die, 1 die per package */ - topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 }; +/* +* simple tests for 1 thread per core, 1 core per module, +* 1 module per die, 1 die per package +*/ + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_core_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 0), ==, 0); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1), ==, 1); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 2), ==, 2);@@ -45,39 +48,39 @@ static void test_topo_bits(void)/* Test field width calculation for multiple values */ - topo_info = (X86CPUTopoInfo ) {1 ,1 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,2 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 1); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,3 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,4 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,4 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,14 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,14 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,15 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,15 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,16 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,16 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,17 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,17 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,31 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,31 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,32 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,32 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,33 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,33 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 6); - topo_info = (X86CPUTopoInfo ) {1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); - topo_info = (X86CPUTopoInfo ) {2 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {2 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 1); - topo_info = (X86CPUTopoInfo ) {3 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {3 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {4 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {4 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2);/* build a weird topology and see if IDs are calculated correctly @@ -85,18 +88,18 @@ static void test_topo_bits(void)/* This will use 2 bits for thread ID and 3 bits for core ID */ - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); g_assert_cmpuint(apicid_core_offset(&topo_info), ==, 2); g_assert_cmpuint(apicid_die_offset(&topo_info), ==, 5); g_assert_cmpuint(apicid_pkg_offset(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 0), ==, 0); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1), ==, 1); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 2), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1 * 3 + 0), ==, (1 <<2 ) |0 ); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1 * 3 + 1), ==,
Re: [PATCH RESEND 07/18] i386: Support modules_per_die in X86CPUTopoInfo
Quoted 76 lines · wangyanan (Y)
Date: Thu, 16 Feb 2023 10:34:24 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 07/18] i386: Supportmodules_per_diein
X86CPUTopoInfoHi Zhao,
Quoted 7 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
Support module level in i386 cpu topology structure "X86CPUTopoInfo".
Before updating APIC ID parsing rule with module level, the
apicid_core_width()temporarily combines the core and module levels together.If we dont merge this one with the followed patches, then nits below may be meaningful.
Quoted 23 lines
At present, we don't expose module level in CPUID.1FH because currently linux (v6.2-rc6) doesn't support module level. And exposing module and die levels at the same time in CPUID.1FH will cause linux to calculate the wrong
die_id. The module level should be exposed until the real machine has the module level in CPUID.1FH.In addition, update topology structure in
test-x86-apicid.c.Signed-off-by Zhuocheng Ding <[email protected]>Co-developed-by Zhao Liu <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- hw/i386/x86.c | 3 ++- include/hw/i386/topology.h | 13 ++++++++--- target/i386/cpu.c | 17 ++++++++------ tests/unit/test-x86-apicid.c | 45 +++++++++++++++++++----------------- 4 files changed, 46 insertions(+), 32 deletions(-)hw/i386/x86.c +0
-0hw/i386/x86.c inline void init_topo_info(X86CPUTopoInfo *topo_info, · line 74
Quoted 5 lines
hw/i386/x86.c +2
-1@@ -71,7 +71,8 @@ inline void init_topo_info(X86CPUTopoInfo *topo_info,MachineState *ms =MACHINE (x86ms); topo_info->dies_per_pkg = ms->smp.dies; - topo_info->cores_per_die = ms->smp.cores; + topo_info->modules_per_die = ms->smp.clusters; + topo_info->cores_per_module = ms->smp.cores;Here we can ensure that
topo_info->modules_per_dieis always 1, so...Quoted 16 lines
hw/i386/x86.c +0
-0topo_info->threads_per_core = ms->smp.threads; }include/hw/i386/topology.h +2
-1@@ -54,7 +54,8 @@ typedef struct X86CPUTopoIDs {typedef struct X86CPUTopoInfo {unsigned dies_per_pkg; -unsigned cores_per_die; +unsigned modules_per_die; +unsigned cores_per_module;unsigned threads_per_core; }X86CPUTopoInfo ;@@ -78,7 +79,12 @@ static inline unsigned apicid_smt_width(X86CPUTopoInfo *topo_info)*/ include/hw/i386/topology.h static inline unsigned apicid_smt_width(X86CPUTopoInfo *topo_info) · line 88
Quoted 10 lines
include/hw/i386/topology.h +6
-1static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) { -return apicid_bitwidth_for_count (topo_info->cores_per_die); +/* +* TODO: Will separate module info from core_width when update +* APIC ID with module level. +*/ +return apicid_bitwidth_for_count (topo_info->cores_per_module * + topo_info->modules_per_die); }...We can directly add
apicid_module_width(which returns 0 so far) andapicid_module_offsethere which don't rely on the APIC ID rule change, and avoid the "TODO..".
Quoted 2 lines
Then patch 8 and 10 are about
module_id, so can be merged. Is this good?
Thanks Yanan, this makes sense. I'll do it.
Zhao
Quoted 151 lines
Thanks,
YananQuoted 147 lines
include/hw/i386/topology.h +2
-1/* Bit width of the Die_ID field */ @@ -128,7 +134,8 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info, X86CPUTopoIDs *topo_ids) {unsigned nr_dies = topo_info->dies_per_pkg; -unsigned nr_cores = topo_info->cores_per_die; +unsigned nr_cores = topo_info->cores_per_module * + topo_info->modules_per_die;unsigned nr_threads = topo_info->threads_per_core; topo_ids->pkg_id = cpu_index / (nr_dies * nr_cores * nr_threads);target/i386/cpu.c +10
-7@@ -336,7 +336,9 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache,/* L3 is shared among multiple cores */ if (cache->level ==3 ) { - l3_threads = topo_info->cores_per_die * topo_info->threads_per_core; + l3_threads = topo_info->modules_per_die * + topo_info->cores_per_module * + topo_info->threads_per_core; *eax |= (l3_threads -1 ) <<14 ; }else { *eax |= ((topo_info->threads_per_core -1 ) <<14 );@@ -5218,11 +5220,12 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,uint32_t cpus_per_pkg; topo_info.dies_per_pkg = env->nr_dies; - topo_info.cores_per_die = cs->nr_cores / env->nr_dies; + topo_info.modules_per_die = env->nr_modules; + topo_info.cores_per_module = cs->nr_cores / env->nr_dies / env->nr_modules; topo_info.threads_per_core = cs->nr_threads; - cpus_per_pkg = topo_info.dies_per_pkg * topo_info.cores_per_die * - topo_info.threads_per_core; + cpus_per_pkg = topo_info.dies_per_pkg * topo_info.modules_per_die * + topo_info.cores_per_module * topo_info.threads_per_core;/* Calculate & apply limits for different index ranges */ if (index >=0xC0000000 ) {@@ -5298,8 +5301,8 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,if (*eax &31 ) {int host_vcpus_per_cache =1 + ((*eax &0x3FFC000 ) >>14 );int vcpus_per_socket = cpus_per_pkg; -int cores_per_socket = topo_info.cores_per_die * - topo_info.dies_per_pkg; +int cores_per_socket = cpus_per_pkg / + topo_info.threads_per_core;if (cores_per_socket >1 ) { *eax &= ~0xFC000000 ; *eax |= (pow2ceil (cores_per_socket) -1 ) <<26 ;@@ -5483,7 +5486,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,break ;case 1 : *eax =apicid_die_offset (&topo_info); - *ebx = topo_info.cores_per_die * topo_info.threads_per_core; + *ebx = cpus_per_pkg / topo_info.dies_per_pkg; *ecx |=CPUID_TOPOLOGY_LEVEL_CORE ;break ;case 2 :tests/unit/test-x86-apicid.c +24
-21@@ -30,13 +30,16 @@ static void test_topo_bits(void) {X86CPUTopoInfo topo_info = {0 }; -/* simple tests for 1 thread per core, 1 core per die, 1 die per package */ - topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 }; +/* +* simple tests for 1 thread per core, 1 core per module, +* 1 module per die, 1 die per package +*/ + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_core_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 0), ==, 0); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1), ==, 1); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 2), ==, 2);@@ -45,39 +48,39 @@ static void test_topo_bits(void)/* Test field width calculation for multiple values */ - topo_info = (X86CPUTopoInfo ) {1 ,1 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,2 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 1); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,3 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,4 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,4 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,14 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,14 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,15 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,15 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,16 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,16 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 4); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,17 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,17 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,31 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,31 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,32 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,32 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,33 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,33 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 6); - topo_info = (X86CPUTopoInfo ) {1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); - topo_info = (X86CPUTopoInfo ) {2 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {2 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 1); - topo_info = (X86CPUTopoInfo ) {3 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {3 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {4 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {4 ,1 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2);/* build a weird topology and see if IDs are calculated correctly @@ -85,18 +88,18 @@ static void test_topo_bits(void)/* This will use 2 bits for thread ID and 3 bits for core ID */ - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); g_assert_cmpuint(apicid_core_offset(&topo_info), ==, 2); g_assert_cmpuint(apicid_die_offset(&topo_info), ==, 5); g_assert_cmpuint(apicid_pkg_offset(&topo_info), ==, 5); - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 0), ==, 0); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1), ==, 1); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 2), ==, 2); - topo_info = (X86CPUTopoInfo ) {1 ,6 ,3 }; + topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1 * 3 + 0), ==, (1 <<2 ) |0 ); g_assert_cmpuint(x86_apicid_from_cpu_idx(&topo_info, 1 * 3 + 1), ==,
PATCH RESEND 8/18 i386: Support module_id in X86CPUTopoIDs
In reply to Zhao LiuFrom: Zhuocheng Ding <[email protected]>
Add module_id member in X86CPUTopoIDs.
module_id can be parsed from APIC ID, so before updating the parsing rule of APIC ID, temporarily set the module_id generated in this way to 0.
module_id can be also generated from cpu topology, and before i386 supports "clusters" in smp, the default "clusters per die" is only 1, thus the module_id generated in this way is 0, so that it will not conflict with the module_id generated by APIC ID.
--- hw/i386/x86.c | 17 +++++++++++++++++ include/hw/i386/topology.h | 24 ++++++++++++++++++++---- 2 files changed, 37 insertions(+), 4 deletions(-)
hw/i386/x86.c +17 -0
@@ -362,6 +362,14 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev, topo_ids.die_id = cpu->die_id; topo_ids.core_id = cpu->core_id; topo_ids.smt_id = cpu->thread_id; + +/* +* TODO: This is the temporary initialization for topo_ids.module_id to +* avoid "maybe-uninitialized" compilation errors. Will remove when +* X86CPU supports cluster_id. +*/ + topo_ids.module_id =0 ; + cpu->apic_id =x86_apicid_from_topo_ids (&topo_info, &topo_ids); }
@@ -370,6 +378,11 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,MachineState *ms =MACHINE (x86ms);x86_topo_ids_from_apicid (cpu->apic_id, &topo_info, &topo_ids); + +/* +* TODO: Before APIC ID supports module level parsing, there's no need +* to expose module_id info. +*/ error_setg(errp, "Invalid CPU [socket: %u, die: %u, core: %u, thread: %u] with" " APIC ID %" PRIu32 ", valid index range 0:%d",
@@ -495,6 +508,10 @@ const CPUArchIdList *x86_possible_cpu_arch_ids(MachineState *ms) ms->possible_cpus->cpus[i].props.has_die_id = true; ms->possible_cpus->cpus[i].props.die_id = topo_ids.die_id; } +if (ms->smp.clusters >1 ) { + ms->possible_cpus->cpus[i].props.has_cluster_id = true; + ms->possible_cpus->cpus[i].props.cluster_id = topo_ids.module_id; + } ms->possible_cpus->cpus[i].props.has_core_id = true; ms->possible_cpus->cpus[i].props.core_id = topo_ids.core_id; ms->possible_cpus->cpus[i].props.has_thread_id = true;
include/hw/i386/topology.h +20 -4
@@ -48,6 +48,7 @@ typedef uint32_t apic_id_t;typedef struct X86CPUTopoIDs {unsigned pkg_id;unsigned die_id; +unsigned module_id;unsigned core_id;unsigned smt_id; }X86CPUTopoIDs ;
@@ -134,12 +135,21 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info, X86CPUTopoIDs *topo_ids) {unsigned nr_dies = topo_info->dies_per_pkg; -unsigned nr_cores = topo_info->cores_per_module * - topo_info->modules_per_die; +unsigned nr_modules = topo_info->modules_per_die; +unsigned nr_cores = topo_info->cores_per_module;unsigned nr_threads = topo_info->threads_per_core; - topo_ids->pkg_id = cpu_index / (nr_dies * nr_cores * nr_threads); - topo_ids->die_id = cpu_index / (nr_cores * nr_threads) % nr_dies; +/* +* Currently smp for i386 doesn't support "clusters", modules_per_die is +* only 1. Therefore, the module_id generated from the module topology will +* not conflict with the module_id generated according to the apicid. +*/ + topo_ids->pkg_id = cpu_index / (nr_dies * nr_modules * + nr_cores * nr_threads); + topo_ids->die_id = cpu_index / (nr_modules * nr_cores * + nr_threads) % nr_dies; + topo_ids->module_id = cpu_index / (nr_cores * nr_threads) % + nr_modules; topo_ids->core_id = cpu_index / nr_threads % nr_cores; topo_ids->smt_id = cpu_index % nr_threads; }
@@ -156,6 +166,12 @@ static inline void x86_topo_ids_from_apicid(apic_id_t apicid, topo_ids->core_id = (apicid >>apicid_core_offset (topo_info)) & ~(0xFFFFFFFFUL <<apicid_core_width (topo_info)); +/* +* TODO: This is the temporary initialization for topo_ids.module_id to +* avoid "maybe-uninitialized" compilation errors. Will remove when APIC +* ID supports module level parsing. +*/ + topo_ids->module_id =0 ; topo_ids->die_id = (apicid >>apicid_die_offset (topo_info)) & ~(0xFFFFFFFFUL <<apicid_die_width (topo_info));
Signature · 1 line
2.34.1
PATCH RESEND 9/18 i386: Fix comment style in topology.h
In reply to Zhao LiuFrom: Zhao Liu <[email protected]>
For function comments in this file, keep the comment style consistent with other places.
--- include/hw/i386/topology.h | 33 +++++++++++++++++---------------- 1 file changed, 17 insertions(+), 16 deletions(-)
include/hw/i386/topology.h +17 -16
@@ -24,7 +24,8 @@#ifndef HW_I386_TOPOLOGY_H #define HW_I386_TOPOLOGY_H -/* This file implements the APIC-ID-based CPU topology enumeration logic, +/* +* This file implements the APIC-ID-based CPU topology enumeration logic, * documented at the following document: * Intel® 64 Architecture Processor Topology Enumeration * http://software.intel.com/en-us/articles/intel-64-architecture-processor-topology-enumeration/
@@ -41,7 +42,8 @@#include "qemu/bitops.h" -/* APIC IDs can be 32-bit, but beware: APIC IDs > 255 require x2APIC support +/* +* APIC IDs can be 32-bit, but beware: APIC IDs > 255 require x2APIC support */ typedef uint32_t apic_id_t ;
@@ -60,8 +62,7 @@ typedef struct X86CPUTopoInfo {unsigned threads_per_core; } X86CPUTopoInfo; -/* Return the bit width needed for 'count' IDs -*/ +/* Return the bit width needed for 'count' IDs */ static unsigned apicid_bitwidth_for_count (unsigned count) {g_assert (count >=1 );
@@ -69,15 +70,13 @@ static unsigned apicid_bitwidth_for_count(unsigned count)return count ?32 -clz32 (count) :0 ; } -/* Bit width of the SMT_ID (thread ID) field on the APIC ID -*/ +/* Bit width of the SMT_ID (thread ID) field on the APIC ID */ static inline unsigned apicid_smt_width (X86CPUTopoInfo *topo_info) {return apicid_bitwidth_for_count (topo_info->threads_per_core); } -/* Bit width of the Core_ID field -*/ +/* Bit width of the Core_ID field */ static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) {/*
@@ -94,8 +93,7 @@ static inline unsigned apicid_die_width(X86CPUTopoInfo *topo_info)return apicid_bitwidth_for_count (topo_info->dies_per_pkg); } -/* Bit offset of the Core_ID field -*/ +/* Bit offset of the Core_ID field */ static inline unsigned apicid_core_offset (X86CPUTopoInfo *topo_info) {return apicid_smt_width (topo_info);
@@ -107,14 +105,14 @@ static inline unsigned apicid_die_offset(X86CPUTopoInfo *topo_info)return apicid_core_offset (topo_info) +apicid_core_width (topo_info); } -/* Bit offset of the Pkg_ID (socket ID) field -*/ +/* Bit offset of the Pkg_ID (socket ID) field */ static inline unsigned apicid_pkg_offset (X86CPUTopoInfo *topo_info) {return apicid_die_offset (topo_info) +apicid_die_width (topo_info); } -/* Make APIC ID for the CPU based on Pkg_ID, Core_ID, SMT_ID +/* +* Make APIC ID for the CPU based on Pkg_ID, Core_ID, SMT_ID * * The caller must make sure core_id < nr_cores and smt_id < nr_threads. */
@@ -127,7 +125,8 @@ static inline apic_id_t x86_apicid_from_topo_ids(X86CPUTopoInfo *topo_info, topo_ids->smt_id; } -/* Calculate thread/core/package IDs for a specific topology, +/* +* Calculate thread/core/package IDs for a specific topology, * based on (contiguous) CPU index */ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info,
@@ -154,7 +153,8 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info, topo_ids->smt_id = cpu_index % nr_threads; } -/* Calculate thread/core/package IDs for a specific topology, +/* +* Calculate thread/core/package IDs for a specific topology, * based on APIC ID */ static inline void x86_topo_ids_from_apicid(apic_id_t apicid,
@@ -178,7 +178,8 @@ static inline void x86_topo_ids_from_apicid(apic_id_t apicid, topo_ids->pkg_id = apicid >>apicid_pkg_offset (topo_info); } -/* Make APIC ID for the CPU 'cpu_index' +/* +* Make APIC ID for the CPU 'cpu_index' * * 'cpu_index' is a sequential, contiguous ID for the CPU. */
Signature · 1 line
2.34.1
Re: [PATCH RESEND 09/18] i386: Fix comment style in topology.h
Quoted 9 lines · Zhao Liu
From: Zhao Liu <[email protected]>
For function comments in this file, keep the comment style consistent with other places.
Signed-off-by Zhao Liu <[email protected]>--- include/hw/i386/topology.h | 33 +++++++++++++++++---------------- 1 file changed, 17 insertions(+), 16 deletions(-)
Re: [PATCH RESEND 09/18] i386: Fix comment style in topology.h
In reply to Zhao LiuQuoted 111 lines · Zhao Liu
From: Zhao Liu <[email protected]>
For function comments in this file, keep the comment style consistent with other places.
Signed-off-by Zhao Liu <[email protected]>--- include/hw/i386/topology.h | 33 +++++++++++++++++---------------- 1 file changed, 17 insertions(+), 16 deletions(-)include/hw/i386/topology.h +15
-15@@ -24,7 +24,8 @@#ifndef HW_I386_TOPOLOGY_H #define HW_I386_TOPOLOGY_H -/* This file implements the APIC-ID-based CPU topology enumeration logic, +/* +* This file implements the APIC-ID-based CPU topology enumeration logic, * documented at the following document: * Intel® 64 Architecture Processor Topology Enumeration * http://software.intel.com/en-us/articles/intel-64-architecture-processor-topology-enumeration/ @@ -41,7 +42,8 @@#include "qemu/bitops.h" -/* APIC IDs can be 32-bit, but beware: APIC IDs > 255 require x2APIC support +/* +* APIC IDs can be 32-bit, but beware: APIC IDs > 255 require x2APIC support */ typedef uint32_t apic_id_t ;@@ -60,8 +62,7 @@ typedef struct X86CPUTopoInfo {unsigned threads_per_core; } X86CPUTopoInfo; -/* Return the bit width needed for 'count' IDs -*/ +/* Return the bit width needed for 'count' IDs */ static unsigned apicid_bitwidth_for_count (unsigned count) {g_assert (count >=1 );@@ -69,15 +70,13 @@ static unsigned apicid_bitwidth_for_count(unsigned count)return count ?32 -clz32 (count) :0 ; } -/* Bit width of the SMT_ID (thread ID) field on the APIC ID -*/ +/* Bit width of the SMT_ID (thread ID) field on the APIC ID */ static inline unsigned apicid_smt_width (X86CPUTopoInfo *topo_info) {return apicid_bitwidth_for_count (topo_info->threads_per_core); } -/* Bit width of the Core_ID field -*/ +/* Bit width of the Core_ID field */ static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) {/* @@ -94,8 +93,7 @@ static inline unsigned apicid_die_width(X86CPUTopoInfo *topo_info)return apicid_bitwidth_for_count (topo_info->dies_per_pkg); } -/* Bit offset of the Core_ID field -*/ +/* Bit offset of the Core_ID field */ static inline unsigned apicid_core_offset (X86CPUTopoInfo *topo_info) {return apicid_smt_width (topo_info);@@ -107,14 +105,14 @@ static inline unsigned apicid_die_offset(X86CPUTopoInfo *topo_info)return apicid_core_offset (topo_info) +apicid_core_width (topo_info); } -/* Bit offset of the Pkg_ID (socket ID) field -*/ +/* Bit offset of the Pkg_ID (socket ID) field */ static inline unsigned apicid_pkg_offset (X86CPUTopoInfo *topo_info) {return apicid_die_offset (topo_info) +apicid_die_width (topo_info); } -/* Make APIC ID for the CPU based on Pkg_ID, Core_ID, SMT_ID +/* +* Make APIC ID for the CPU based on Pkg_ID, Core_ID, SMT_ID * * The caller must make sure core_id < nr_cores and smt_id < nr_threads. */ @@ -127,7 +125,8 @@ static inline apic_id_t x86_apicid_from_topo_ids(X86CPUTopoInfo *topo_info, topo_ids->smt_id; } -/* Calculate thread/core/package IDs for a specific topology, +/* +* Calculate thread/core/package IDs for a specific topology, * based on (contiguous) CPU index */ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info,@@ -154,7 +153,8 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info, topo_ids->smt_id = cpu_index % nr_threads; } -/* Calculate thread/core/package IDs for a specific topology, +/* +* Calculate thread/core/package IDs for a specific topology, * based on APIC ID */ static inline void x86_topo_ids_from_apicid(apic_id_t apicid,
include/hw/i386/topology.h static inline void x86_topo_ids_from_apicid(apic_id_t apicid, · line 185
Quoted 9 lines
include/hw/i386/topology.h +2
-1@@ -178,7 +178,8 @@ static inline void x86_topo_ids_from_apicid(apic_id_t apicid, topo_ids->pkg_id = apicid >>apicid_pkg_offset (topo_info); } -/* Make APIC ID for the CPU 'cpu_index' +/* +* Make APIC ID for the CPU 'cpu_index' * * 'cpu_index' is a sequential, contiguous ID for the CPU. */
Thanks,
Yanan
Re: [PATCH RESEND 09/18] i386: Fix comment style in topology.h
In reply to Zhao LiuQuoted 6 lines · Zhao Liu
From: Zhao Liu <[email protected]>
For function comments in this file, keep the comment style consistent with other places.
Signed-off-by Zhao Liu <[email protected]>
nit:Better to move this cleanup patch to top of the series.
Quoted 114 lines
--- include/hw/i386/topology.h | 33 +++++++++++++++++---------------- 1 file changed, 17 insertions(+), 16 deletions(-)include/hw/i386/topology.h +17
-16@@ -24,7 +24,8 @@#ifndef HW_I386_TOPOLOGY_H #define HW_I386_TOPOLOGY_H -/* This file implements the APIC-ID-based CPU topology enumeration logic, +/* +* This file implements the APIC-ID-based CPU topology enumeration logic, * documented at the following document: * Intel® 64 Architecture Processor Topology Enumeration * http://software.intel.com/en-us/articles/intel-64-architecture-processor-topology-enumeration/ @@ -41,7 +42,8 @@#include "qemu/bitops.h" -/* APIC IDs can be 32-bit, but beware: APIC IDs > 255 require x2APIC support +/* +* APIC IDs can be 32-bit, but beware: APIC IDs > 255 require x2APIC support */ typedef uint32_t apic_id_t ;@@ -60,8 +62,7 @@ typedef struct X86CPUTopoInfo {unsigned threads_per_core; } X86CPUTopoInfo; -/* Return the bit width needed for 'count' IDs -*/ +/* Return the bit width needed for 'count' IDs */ static unsigned apicid_bitwidth_for_count (unsigned count) {g_assert (count >=1 );@@ -69,15 +70,13 @@ static unsigned apicid_bitwidth_for_count(unsigned count)return count ?32 -clz32 (count) :0 ; } -/* Bit width of the SMT_ID (thread ID) field on the APIC ID -*/ +/* Bit width of the SMT_ID (thread ID) field on the APIC ID */ static inline unsigned apicid_smt_width (X86CPUTopoInfo *topo_info) {return apicid_bitwidth_for_count (topo_info->threads_per_core); } -/* Bit width of the Core_ID field -*/ +/* Bit width of the Core_ID field */ static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) {/* @@ -94,8 +93,7 @@ static inline unsigned apicid_die_width(X86CPUTopoInfo *topo_info)return apicid_bitwidth_for_count (topo_info->dies_per_pkg); } -/* Bit offset of the Core_ID field -*/ +/* Bit offset of the Core_ID field */ static inline unsigned apicid_core_offset (X86CPUTopoInfo *topo_info) {return apicid_smt_width (topo_info);@@ -107,14 +105,14 @@ static inline unsigned apicid_die_offset(X86CPUTopoInfo *topo_info)return apicid_core_offset (topo_info) +apicid_core_width (topo_info); } -/* Bit offset of the Pkg_ID (socket ID) field -*/ +/* Bit offset of the Pkg_ID (socket ID) field */ static inline unsigned apicid_pkg_offset (X86CPUTopoInfo *topo_info) {return apicid_die_offset (topo_info) +apicid_die_width (topo_info); } -/* Make APIC ID for the CPU based on Pkg_ID, Core_ID, SMT_ID +/* +* Make APIC ID for the CPU based on Pkg_ID, Core_ID, SMT_ID * * The caller must make sure core_id < nr_cores and smt_id < nr_threads. */ @@ -127,7 +125,8 @@ static inline apic_id_t x86_apicid_from_topo_ids(X86CPUTopoInfo *topo_info, topo_ids->smt_id; } -/* Calculate thread/core/package IDs for a specific topology, +/* +* Calculate thread/core/package IDs for a specific topology, * based on (contiguous) CPU index */ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info,@@ -154,7 +153,8 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info, topo_ids->smt_id = cpu_index % nr_threads; } -/* Calculate thread/core/package IDs for a specific topology, +/* +* Calculate thread/core/package IDs for a specific topology, * based on APIC ID */ static inline void x86_topo_ids_from_apicid(apic_id_t apicid,@@ -178,7 +178,8 @@ static inline void x86_topo_ids_from_apicid(apic_id_t apicid, topo_ids->pkg_id = apicid >>apicid_pkg_offset (topo_info); } -/* Make APIC ID for the CPU 'cpu_index' +/* +* Make APIC ID for the CPU 'cpu_index' * * 'cpu_index' is a sequential, contiguous ID for the CPU. */
Re: [PATCH RESEND 09/18] i386: Fix comment style in topology.h
Quoted 4 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 18:54:15 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 09/18] i386: Fix comment style intopology.h
Quoted 8 lines
Quoted 6 lines · Zhao Liu
From: Zhao Liu <[email protected]>
For function comments in this file, keep the comment style consistent with other places.
Signed-off-by Zhao Liu <[email protected]>nit:Better to move this cleanup patch to top of the series.
Ok, will do that. Thanks!
Quoted 102 lines
Quoted 101 lines
--- include/hw/i386/topology.h | 33 +++++++++++++++++---------------- 1 file changed, 17 insertions(+), 16 deletions(-)include/hw/i386/topology.h +17
-16@@ -24,7 +24,8 @@#ifndef HW_I386_TOPOLOGY_H #define HW_I386_TOPOLOGY_H -/* This file implements the APIC-ID-based CPU topology enumeration logic, +/* +* This file implements the APIC-ID-based CPU topology enumeration logic, * documented at the following document: * Intel® 64 Architecture Processor Topology Enumeration * http://software.intel.com/en-us/articles/intel-64-architecture-processor-topology-enumeration/ @@ -41,7 +42,8 @@#include "qemu/bitops.h" -/* APIC IDs can be 32-bit, but beware: APIC IDs > 255 require x2APIC support +/* +* APIC IDs can be 32-bit, but beware: APIC IDs > 255 require x2APIC support */ typedef uint32_t apic_id_t ;@@ -60,8 +62,7 @@ typedef struct X86CPUTopoInfo {unsigned threads_per_core; } X86CPUTopoInfo; -/* Return the bit width needed for 'count' IDs -*/ +/* Return the bit width needed for 'count' IDs */ static unsigned apicid_bitwidth_for_count (unsigned count) {g_assert (count >=1 );@@ -69,15 +70,13 @@ static unsigned apicid_bitwidth_for_count(unsigned count)return count ?32 -clz32 (count) :0 ; } -/* Bit width of the SMT_ID (thread ID) field on the APIC ID -*/ +/* Bit width of the SMT_ID (thread ID) field on the APIC ID */ static inline unsigned apicid_smt_width (X86CPUTopoInfo *topo_info) {return apicid_bitwidth_for_count (topo_info->threads_per_core); } -/* Bit width of the Core_ID field -*/ +/* Bit width of the Core_ID field */ static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) {/* @@ -94,8 +93,7 @@ static inline unsigned apicid_die_width(X86CPUTopoInfo *topo_info)return apicid_bitwidth_for_count (topo_info->dies_per_pkg); } -/* Bit offset of the Core_ID field -*/ +/* Bit offset of the Core_ID field */ static inline unsigned apicid_core_offset (X86CPUTopoInfo *topo_info) {return apicid_smt_width (topo_info);@@ -107,14 +105,14 @@ static inline unsigned apicid_die_offset(X86CPUTopoInfo *topo_info)return apicid_core_offset (topo_info) +apicid_core_width (topo_info); } -/* Bit offset of the Pkg_ID (socket ID) field -*/ +/* Bit offset of the Pkg_ID (socket ID) field */ static inline unsigned apicid_pkg_offset (X86CPUTopoInfo *topo_info) {return apicid_die_offset (topo_info) +apicid_die_width (topo_info); } -/* Make APIC ID for the CPU based on Pkg_ID, Core_ID, SMT_ID +/* +* Make APIC ID for the CPU based on Pkg_ID, Core_ID, SMT_ID * * The caller must make sure core_id < nr_cores and smt_id < nr_threads. */ @@ -127,7 +125,8 @@ static inline apic_id_t x86_apicid_from_topo_ids(X86CPUTopoInfo *topo_info, topo_ids->smt_id; } -/* Calculate thread/core/package IDs for a specific topology, +/* +* Calculate thread/core/package IDs for a specific topology, * based on (contiguous) CPU index */ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info,@@ -154,7 +153,8 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info, topo_ids->smt_id = cpu_index % nr_threads; } -/* Calculate thread/core/package IDs for a specific topology, +/* +* Calculate thread/core/package IDs for a specific topology, * based on APIC ID */ static inline void x86_topo_ids_from_apicid(apic_id_t apicid,@@ -178,7 +178,8 @@ static inline void x86_topo_ids_from_apicid(apic_id_t apicid, topo_ids->pkg_id = apicid >>apicid_pkg_offset (topo_info); } -/* Make APIC ID for the CPU 'cpu_index' +/* +* Make APIC ID for the CPU 'cpu_index' * * 'cpu_index' is a sequential, contiguous ID for the CPU. */
PATCH RESEND 10/18 i386: Update APIC ID parsing rule to support module level
In reply to Zhao LiuFrom: Zhuocheng Ding <[email protected]>
Add the module level parsing support for APIC ID.
With this support, now the conversion between X86CPUTopoIDs, X86CPUTopoInfo and APIC ID is completed.
--- hw/i386/x86.c | 19 ++++++++----------- include/hw/i386/topology.h | 36 ++++++++++++++++++------------------ 2 files changed, 26 insertions(+), 29 deletions(-)
hw/i386/x86.c +8 -11
@@ -311,11 +311,11 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,/* * If APIC ID is not set, -* set it based on socket/die/core/thread properties. +* set it based on socket/die/cluster/core/thread properties. */ if (cpu->apic_id ==UNASSIGNED_APIC_ID ) { -int max_socket = (ms->smp.max_cpus -1 ) / - smp_threads / smp_cores / ms->smp.dies; +int max_socket = (ms->smp.max_cpus -1 ) / smp_threads / smp_cores / + ms->smp.clusters / ms->smp.dies;/* * die-id was optional in QEMU 4.0 and older, so keep it optional
@@ -379,15 +379,12 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,x86_topo_ids_from_apicid (cpu->apic_id, &topo_info, &topo_ids); -/* -* TODO: Before APIC ID supports module level parsing, there's no need -* to expose module_id info. -*/ error_setg (errp, -"Invalid CPU [socket: %u, die: %u, core: %u, thread: %u] with" -" APIC ID %" PRIu32", valid index range 0:%d" , - topo_ids.pkg_id, topo_ids.die_id, topo_ids.core_id, topo_ids.smt_id, - cpu->apic_id, ms->possible_cpus->len -1 ); +"Invalid CPU [socket: %u, die: %u, module: %u, core: %u, thread: %u]" +" with APIC ID %" PRIu32", valid index range 0:%d" , + topo_ids.pkg_id, topo_ids.die_id, topo_ids.module_id, + topo_ids.core_id, topo_ids.smt_id, cpu->apic_id, + ms->possible_cpus->len -1 );return ; }
include/hw/i386/topology.h +18 -18
@@ -79,12 +79,13 @@ static inline unsigned apicid_smt_width(X86CPUTopoInfo *topo_info)/* Bit width of the Core_ID field */ static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) { -/* -* TODO: Will separate module info from core_width when update -* APIC ID with module level. -*/ -return apicid_bitwidth_for_count (topo_info->cores_per_module * - topo_info->modules_per_die); +return apicid_bitwidth_for_count (topo_info->cores_per_module); +} + +/* Bit width of the Module_ID (cluster ID) field */ +static inline unsigned apicid_module_width (X86CPUTopoInfo *topo_info) +{ +return apicid_bitwidth_for_count (topo_info->modules_per_die); }/* Bit width of the Die_ID field */
@@ -99,10 +100,16 @@ static inline unsigned apicid_core_offset(X86CPUTopoInfo *topo_info)return apicid_smt_width (topo_info); } +/* Bit offset of the Module_ID (cluster ID) field */ +static inline unsigned apicid_module_offset (X86CPUTopoInfo *topo_info) +{ +return apicid_core_offset (topo_info) +apicid_core_width (topo_info); +} +/* Bit offset of the Die_ID field */ static inline unsigned apicid_die_offset (X86CPUTopoInfo *topo_info) { -return apicid_core_offset (topo_info) +apicid_core_width (topo_info); +return apicid_module_offset (topo_info) +apicid_module_width (topo_info); }/* Bit offset of the Pkg_ID (socket ID) field */
@@ -121,6 +128,7 @@ static inline apic_id_t x86_apicid_from_topo_ids(X86CPUTopoInfo *topo_info, {return (topo_ids->pkg_id <<apicid_pkg_offset (topo_info)) | (topo_ids->die_id <<apicid_die_offset (topo_info)) | + (topo_ids->module_id <<apicid_module_offset (topo_info)) | (topo_ids->core_id <<apicid_core_offset (topo_info)) | topo_ids->smt_id; }
@@ -138,11 +146,6 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info,unsigned nr_cores = topo_info->cores_per_module;unsigned nr_threads = topo_info->threads_per_core; -/* -* Currently smp for i386 doesn't support "clusters", modules_per_die is -* only 1. Therefore, the module_id generated from the module topology will -* not conflict with the module_id generated according to the apicid. -*/ topo_ids->pkg_id = cpu_index / (nr_dies * nr_modules * nr_cores * nr_threads); topo_ids->die_id = cpu_index / (nr_modules * nr_cores *
@@ -166,12 +169,9 @@ static inline void x86_topo_ids_from_apicid(apic_id_t apicid, topo_ids->core_id = (apicid >>apicid_core_offset (topo_info)) & ~(0xFFFFFFFFUL <<apicid_core_width (topo_info)); -/* -* TODO: This is the temporary initialization for topo_ids.module_id to -* avoid "maybe-uninitialized" compilation errors. Will remove when APIC -* ID supports module level parsing. -*/ - topo_ids->module_id =0 ; + topo_ids->module_id = + (apicid >>apicid_module_offset (topo_info)) & + ~(0xFFFFFFFFUL <<apicid_module_width (topo_info)); topo_ids->die_id = (apicid >>apicid_die_offset (topo_info)) & ~(0xFFFFFFFFUL <<apicid_die_width (topo_info));
Signature · 1 line
2.34.1
Re: [PATCH RESEND 10/18] i386: Update APIC ID parsing rule to support module level
Hi Zhao,
Quoted 6 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
Add the module level parsing support for APIC ID.
With this support, now the conversion between X86CPUTopoIDs, X86CPUTopoInfo and APIC ID is completed.
IIUC, contents in patch 6-8 and 10 are all about "Introduce the module-level CPU topology support for x86", why do we need gradually do this with kinds of temporary things instead of warp them into one patch? Before support for smp.clusters in the CLI for x86, we can ensure that modules_per_dies is always 1 so that the code is save in one diff. Or do I miss something?
Thanks,
Yanan
Quoted 126 lines
Signed-off-by Zhuocheng Ding <[email protected]>Co-developed-by Zhao Liu <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- hw/i386/x86.c | 19 ++++++++----------- include/hw/i386/topology.h | 36 ++++++++++++++++++------------------ 2 files changed, 26 insertions(+), 29 deletions(-)hw/i386/x86.c +8
-11@@ -311,11 +311,11 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,/* * If APIC ID is not set, -* set it based on socket/die/core/thread properties. +* set it based on socket/die/cluster/core/thread properties. */ if (cpu->apic_id ==UNASSIGNED_APIC_ID ) { -int max_socket = (ms->smp.max_cpus -1 ) / - smp_threads / smp_cores / ms->smp.dies; +int max_socket = (ms->smp.max_cpus -1 ) / smp_threads / smp_cores / + ms->smp.clusters / ms->smp.dies;/* * die-id was optional in QEMU 4.0 and older, so keep it optional @@ -379,15 +379,12 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,x86_topo_ids_from_apicid (cpu->apic_id, &topo_info, &topo_ids); -/* -* TODO: Before APIC ID supports module level parsing, there's no need -* to expose module_id info. -*/ error_setg (errp, -"Invalid CPU [socket: %u, die: %u, core: %u, thread: %u] with" -" APIC ID %" PRIu32", valid index range 0:%d" , - topo_ids.pkg_id, topo_ids.die_id, topo_ids.core_id, topo_ids.smt_id, - cpu->apic_id, ms->possible_cpus->len -1 ); +"Invalid CPU [socket: %u, die: %u, module: %u, core: %u, thread: %u]" +" with APIC ID %" PRIu32", valid index range 0:%d" , + topo_ids.pkg_id, topo_ids.die_id, topo_ids.module_id, + topo_ids.core_id, topo_ids.smt_id, cpu->apic_id, + ms->possible_cpus->len -1 );return ; }include/hw/i386/topology.h +18
-18@@ -79,12 +79,13 @@ static inline unsigned apicid_smt_width(X86CPUTopoInfo *topo_info)/* Bit width of the Core_ID field */ static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) { -/* -* TODO: Will separate module info from core_width when update -* APIC ID with module level. -*/ -return apicid_bitwidth_for_count (topo_info->cores_per_module * - topo_info->modules_per_die); +return apicid_bitwidth_for_count (topo_info->cores_per_module); +} + +/* Bit width of the Module_ID (cluster ID) field */ +static inline unsigned apicid_module_width (X86CPUTopoInfo *topo_info) +{ +return apicid_bitwidth_for_count (topo_info->modules_per_die); }/* Bit width of the Die_ID field */ @@ -99,10 +100,16 @@ static inline unsigned apicid_core_offset(X86CPUTopoInfo *topo_info)return apicid_smt_width (topo_info); } +/* Bit offset of the Module_ID (cluster ID) field */ +static inline unsigned apicid_module_offset (X86CPUTopoInfo *topo_info) +{ +return apicid_core_offset (topo_info) +apicid_core_width (topo_info); +} +/* Bit offset of the Die_ID field */ static inline unsigned apicid_die_offset (X86CPUTopoInfo *topo_info) { -return apicid_core_offset (topo_info) +apicid_core_width (topo_info); +return apicid_module_offset (topo_info) +apicid_module_width (topo_info); }/* Bit offset of the Pkg_ID (socket ID) field */ @@ -121,6 +128,7 @@ static inline apic_id_t x86_apicid_from_topo_ids(X86CPUTopoInfo *topo_info, {return (topo_ids->pkg_id <<apicid_pkg_offset (topo_info)) | (topo_ids->die_id <<apicid_die_offset (topo_info)) | + (topo_ids->module_id <<apicid_module_offset (topo_info)) | (topo_ids->core_id <<apicid_core_offset (topo_info)) | topo_ids->smt_id; }@@ -138,11 +146,6 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info,unsigned nr_cores = topo_info->cores_per_module;unsigned nr_threads = topo_info->threads_per_core; -/* -* Currently smp for i386 doesn't support "clusters", modules_per_die is -* only 1. Therefore, the module_id generated from the module topology will -* not conflict with the module_id generated according to the apicid. -*/ topo_ids->pkg_id = cpu_index / (nr_dies * nr_modules * nr_cores * nr_threads); topo_ids->die_id = cpu_index / (nr_modules * nr_cores *@@ -166,12 +169,9 @@ static inline void x86_topo_ids_from_apicid(apic_id_t apicid, topo_ids->core_id = (apicid >>apicid_core_offset (topo_info)) & ~(0xFFFFFFFFUL <<apicid_core_width (topo_info)); -/* -* TODO: This is the temporary initialization for topo_ids.module_id to -* avoid "maybe-uninitialized" compilation errors. Will remove when APIC -* ID supports module level parsing. -*/ - topo_ids->module_id =0 ; + topo_ids->module_id = + (apicid >>apicid_module_offset (topo_info)) & + ~(0xFFFFFFFFUL <<apicid_module_width (topo_info)); topo_ids->die_id = (apicid >>apicid_die_offset (topo_info)) & ~(0xFFFFFFFFUL <<apicid_die_width (topo_info));
Re: [PATCH RESEND 10/18] i386: Update APIC ID parsing rule to support module level
Quoted 10 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 19:06:32 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 10/18] i386: Update APIC ID parsing rule to
support module levelHi Zhao,
Quoted 2 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
Quoted 7 lines
Quoted 4 lines
Add the module level parsing support for APIC ID.
With this support, now the conversion between X86CPUTopoIDs, X86CPUTopoInfo and APIC ID is completed.
IIUC, contents in patch 6-8 and 10 are all about "Introduce the module-level CPU topology support for x86", why do we need gradually do this with kinds of temporary things instead of warp them into one patch?
Patch 6 is about CPUX86State.nr_dies, which is independent from patch 7, 8, 10.
Patch 7 (X86CPUTopoInfo.modules_per_die), patch 8 (X86CPUTopoIDs.module_id), and patch 10 (APIC ID parsing rule) are related but have their own relatively clear little themes, and are gradually completing full support for module level in apic id.
Patch 7, 8, 10 can be combined into one big patch. This current splitting way is actually designed to make it easier to review... But if you think this is not convenient for review, sorry for that, and I'm willing to merge them together. ;-)
Thanks,
Zhao
Quoted 124 lines
Before support
for smp.clusters in the CLI for x86, we can ensure thatmodules_per_diesis always 1 so that the code is save in one diff. Or do I miss something?Thanks,
YananQuoted 117 lines
Signed-off-by Zhuocheng Ding <[email protected]>Co-developed-by Zhao Liu <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- hw/i386/x86.c | 19 ++++++++----------- include/hw/i386/topology.h | 36 ++++++++++++++++++------------------ 2 files changed, 26 insertions(+), 29 deletions(-)hw/i386/x86.c +8
-11@@ -311,11 +311,11 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,/* * If APIC ID is not set, -* set it based on socket/die/core/thread properties. +* set it based on socket/die/cluster/core/thread properties. */ if (cpu->apic_id ==UNASSIGNED_APIC_ID ) { -int max_socket = (ms->smp.max_cpus -1 ) / - smp_threads / smp_cores / ms->smp.dies; +int max_socket = (ms->smp.max_cpus -1 ) / smp_threads / smp_cores / + ms->smp.clusters / ms->smp.dies;/* * die-id was optional in QEMU 4.0 and older, so keep it optional @@ -379,15 +379,12 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,x86_topo_ids_from_apicid (cpu->apic_id, &topo_info, &topo_ids); -/* -* TODO: Before APIC ID supports module level parsing, there's no need -* to expose module_id info. -*/ error_setg (errp, -"Invalid CPU [socket: %u, die: %u, core: %u, thread: %u] with" -" APIC ID %" PRIu32", valid index range 0:%d" , - topo_ids.pkg_id, topo_ids.die_id, topo_ids.core_id, topo_ids.smt_id, - cpu->apic_id, ms->possible_cpus->len -1 ); +"Invalid CPU [socket: %u, die: %u, module: %u, core: %u, thread: %u]" +" with APIC ID %" PRIu32", valid index range 0:%d" , + topo_ids.pkg_id, topo_ids.die_id, topo_ids.module_id, + topo_ids.core_id, topo_ids.smt_id, cpu->apic_id, + ms->possible_cpus->len -1 );return ; }include/hw/i386/topology.h +18
-18@@ -79,12 +79,13 @@ static inline unsigned apicid_smt_width(X86CPUTopoInfo *topo_info)/* Bit width of the Core_ID field */ static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) { -/* -* TODO: Will separate module info from core_width when update -* APIC ID with module level. -*/ -return apicid_bitwidth_for_count (topo_info->cores_per_module * - topo_info->modules_per_die); +return apicid_bitwidth_for_count (topo_info->cores_per_module); +} + +/* Bit width of the Module_ID (cluster ID) field */ +static inline unsigned apicid_module_width (X86CPUTopoInfo *topo_info) +{ +return apicid_bitwidth_for_count (topo_info->modules_per_die); }/* Bit width of the Die_ID field */ @@ -99,10 +100,16 @@ static inline unsigned apicid_core_offset(X86CPUTopoInfo *topo_info)return apicid_smt_width (topo_info); } +/* Bit offset of the Module_ID (cluster ID) field */ +static inline unsigned apicid_module_offset (X86CPUTopoInfo *topo_info) +{ +return apicid_core_offset (topo_info) +apicid_core_width (topo_info); +} +/* Bit offset of the Die_ID field */ static inline unsigned apicid_die_offset (X86CPUTopoInfo *topo_info) { -return apicid_core_offset (topo_info) +apicid_core_width (topo_info); +return apicid_module_offset (topo_info) +apicid_module_width (topo_info); }/* Bit offset of the Pkg_ID (socket ID) field */ @@ -121,6 +128,7 @@ static inline apic_id_t x86_apicid_from_topo_ids(X86CPUTopoInfo *topo_info, {return (topo_ids->pkg_id <<apicid_pkg_offset (topo_info)) | (topo_ids->die_id <<apicid_die_offset (topo_info)) | + (topo_ids->module_id <<apicid_module_offset (topo_info)) | (topo_ids->core_id <<apicid_core_offset (topo_info)) | topo_ids->smt_id; }@@ -138,11 +146,6 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info,unsigned nr_cores = topo_info->cores_per_module;unsigned nr_threads = topo_info->threads_per_core; -/* -* Currently smp for i386 doesn't support "clusters", modules_per_die is -* only 1. Therefore, the module_id generated from the module topology will -* not conflict with the module_id generated according to the apicid. -*/ topo_ids->pkg_id = cpu_index / (nr_dies * nr_modules * nr_cores * nr_threads); topo_ids->die_id = cpu_index / (nr_modules * nr_cores *@@ -166,12 +169,9 @@ static inline void x86_topo_ids_from_apicid(apic_id_t apicid, topo_ids->core_id = (apicid >>apicid_core_offset (topo_info)) & ~(0xFFFFFFFFUL <<apicid_core_width (topo_info)); -/* -* TODO: This is the temporary initialization for topo_ids.module_id to -* avoid "maybe-uninitialized" compilation errors. Will remove when APIC -* ID supports module level parsing. -*/ - topo_ids->module_id =0 ; + topo_ids->module_id = + (apicid >>apicid_module_offset (topo_info)) & + ~(0xFFFFFFFFUL <<apicid_module_width (topo_info)); topo_ids->die_id = (apicid >>apicid_die_offset (topo_info)) & ~(0xFFFFFFFFUL <<apicid_die_width (topo_info));
Re: [PATCH RESEND 10/18] i386: Update APIC ID parsing rule to support module level
Quoted 13 lines · Zhao Liu
Quoted 12 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 19:06:32 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 10/18] i386: Update APIC ID parsing rule to
support module levelHi Zhao,
Quoted 4 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
Add the module level parsing support for APIC ID.
Quoted 7 lines
Quoted 5 lines
Quoted 2 lines
With this support, now the conversion between X86CPUTopoIDs, X86CPUTopoInfo and APIC ID is completed.
IIUC, contents in patch 6-8 and 10 are all about "Introduce the module-level CPU topology support for x86", why do we need gradually do this with kinds of temporary things instead of warp them into one patch?
Patch 6 is about CPUX86State.nr_dies, which is independent from patch 7, 8, 10.
Ok
Quoted 10 lines
Patch 7 (X86CPUTopoInfo.modules_per_die), patch 8 (X86CPUTopoIDs.module_id), and patch 10 (APIC ID parsing rule) are related but have their own relatively clear little themes, and are gradually completing full support for module level in apic id.
Patch 7, 8, 10 can be combined into one big patch. This current splitting way is actually designed to make it easier to review... But if you think this is not convenient for review, sorry for that, and I'm willing to merge them together. ;-)
So comments in patch 7, I think merging 8 and 10 will be clean enough.
Thanks,
Yanan
Quoted 127 lines
Thanks,
ZhaoQuoted 123 lines
Before support
for smp.clusters in the CLI for x86, we can ensure thatmodules_per_diesis always 1 so that the code is save in one diff. Or do I miss something?Thanks,
YananQuoted 117 lines
Signed-off-by Zhuocheng Ding <[email protected]>Co-developed-by Zhao Liu <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- hw/i386/x86.c | 19 ++++++++----------- include/hw/i386/topology.h | 36 ++++++++++++++++++------------------ 2 files changed, 26 insertions(+), 29 deletions(-)hw/i386/x86.c +8
-11@@ -311,11 +311,11 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,/* * If APIC ID is not set, -* set it based on socket/die/core/thread properties. +* set it based on socket/die/cluster/core/thread properties. */ if (cpu->apic_id ==UNASSIGNED_APIC_ID ) { -int max_socket = (ms->smp.max_cpus -1 ) / - smp_threads / smp_cores / ms->smp.dies; +int max_socket = (ms->smp.max_cpus -1 ) / smp_threads / smp_cores / + ms->smp.clusters / ms->smp.dies;/* * die-id was optional in QEMU 4.0 and older, so keep it optional @@ -379,15 +379,12 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev,x86_topo_ids_from_apicid (cpu->apic_id, &topo_info, &topo_ids); -/* -* TODO: Before APIC ID supports module level parsing, there's no need -* to expose module_id info. -*/ error_setg (errp, -"Invalid CPU [socket: %u, die: %u, core: %u, thread: %u] with" -" APIC ID %" PRIu32", valid index range 0:%d" , - topo_ids.pkg_id, topo_ids.die_id, topo_ids.core_id, topo_ids.smt_id, - cpu->apic_id, ms->possible_cpus->len -1 ); +"Invalid CPU [socket: %u, die: %u, module: %u, core: %u, thread: %u]" +" with APIC ID %" PRIu32", valid index range 0:%d" , + topo_ids.pkg_id, topo_ids.die_id, topo_ids.module_id, + topo_ids.core_id, topo_ids.smt_id, cpu->apic_id, + ms->possible_cpus->len -1 );return ; }include/hw/i386/topology.h +18
-18@@ -79,12 +79,13 @@ static inline unsigned apicid_smt_width(X86CPUTopoInfo *topo_info)/* Bit width of the Core_ID field */ static inline unsigned apicid_core_width (X86CPUTopoInfo *topo_info) { -/* -* TODO: Will separate module info from core_width when update -* APIC ID with module level. -*/ -return apicid_bitwidth_for_count (topo_info->cores_per_module * - topo_info->modules_per_die); +return apicid_bitwidth_for_count (topo_info->cores_per_module); +} + +/* Bit width of the Module_ID (cluster ID) field */ +static inline unsigned apicid_module_width (X86CPUTopoInfo *topo_info) +{ +return apicid_bitwidth_for_count (topo_info->modules_per_die); }/* Bit width of the Die_ID field */ @@ -99,10 +100,16 @@ static inline unsigned apicid_core_offset(X86CPUTopoInfo *topo_info)return apicid_smt_width (topo_info); } +/* Bit offset of the Module_ID (cluster ID) field */ +static inline unsigned apicid_module_offset (X86CPUTopoInfo *topo_info) +{ +return apicid_core_offset (topo_info) +apicid_core_width (topo_info); +} +/* Bit offset of the Die_ID field */ static inline unsigned apicid_die_offset (X86CPUTopoInfo *topo_info) { -return apicid_core_offset (topo_info) +apicid_core_width (topo_info); +return apicid_module_offset (topo_info) +apicid_module_width (topo_info); }/* Bit offset of the Pkg_ID (socket ID) field */ @@ -121,6 +128,7 @@ static inline apic_id_t x86_apicid_from_topo_ids(X86CPUTopoInfo *topo_info, {return (topo_ids->pkg_id <<apicid_pkg_offset (topo_info)) | (topo_ids->die_id <<apicid_die_offset (topo_info)) | + (topo_ids->module_id <<apicid_module_offset (topo_info)) | (topo_ids->core_id <<apicid_core_offset (topo_info)) | topo_ids->smt_id; }@@ -138,11 +146,6 @@ static inline void x86_topo_ids_from_idx(X86CPUTopoInfo *topo_info,unsigned nr_cores = topo_info->cores_per_module;unsigned nr_threads = topo_info->threads_per_core; -/* -* Currently smp for i386 doesn't support "clusters", modules_per_die is -* only 1. Therefore, the module_id generated from the module topology will -* not conflict with the module_id generated according to the apicid. -*/ topo_ids->pkg_id = cpu_index / (nr_dies * nr_modules * nr_cores * nr_threads); topo_ids->die_id = cpu_index / (nr_modules * nr_cores *@@ -166,12 +169,9 @@ static inline void x86_topo_ids_from_apicid(apic_id_t apicid, topo_ids->core_id = (apicid >>apicid_core_offset (topo_info)) & ~(0xFFFFFFFFUL <<apicid_core_width (topo_info)); -/* -* TODO: This is the temporary initialization for topo_ids.module_id to -* avoid "maybe-uninitialized" compilation errors. Will remove when APIC -* ID supports module level parsing. -*/ - topo_ids->module_id =0 ; + topo_ids->module_id = + (apicid >>apicid_module_offset (topo_info)) & + ~(0xFFFFFFFFUL <<apicid_module_width (topo_info)); topo_ids->die_id = (apicid >>apicid_die_offset (topo_info)) & ~(0xFFFFFFFFUL <<apicid_die_width (topo_info));
PATCH RESEND 11/18 i386/cpu: Introduce cluster-id to X86CPU
In reply to Zhao LiuFrom: Zhuocheng Ding <[email protected]>
We introduce cluster-id other than module-id to be consistent with CpuInstanceProperties.cluster-id, and this avoids the confusion of parameter names when hotplugging.
Following the legacy smp check rules, also add the cluster_id validity into x86_cpu_pre_plug().
--- hw/i386/x86.c | 33 +++++++++++++++++++++++++-------- target/i386/cpu.c | 2 ++ target/i386/cpu.h | 1 + 3 files changed, 28 insertions(+), 8 deletions(-)
hw/i386/x86.c +25 -8
@@ -325,6 +325,14 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev, cpu->die_id =0 ; } +/* +* cluster-id was optional in QEMU 8.0 and older, so keep it optional +* if there's only one cluster per die. +*/ +if (cpu->cluster_id <0 && ms->smp.clusters ==1 ) { + cpu->cluster_id =0 ; + } +if (cpu->socket_id <0 ) {error_setg (errp,"CPU socket-id is not set" );return ;
@@ -341,6 +349,14 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev, cpu->die_id, ms->smp.dies - 1);return ; } +if (cpu->cluster_id <0 ) { +error_setg (errp,"CPU cluster-id is not set" ); +return ; + }else if (cpu->cluster_id > ms->smp.clusters -1 ) { +error_setg (errp,"Invalid CPU cluster-id: %u must be in range 0:%u" , + cpu->cluster_id, ms->smp.clusters -1 ); +return ; + }if (cpu->core_id <0 ) {error_setg (errp,"CPU core-id is not set" );return ;
@@ -360,16 +376,9 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev, topo_ids.pkg_id = cpu->socket_id; topo_ids.die_id = cpu->die_id; + topo_ids.module_id = cpu->cluster_id; topo_ids.core_id = cpu->core_id; topo_ids.smt_id = cpu->thread_id; - -/* -* TODO: This is the temporary initialization for topo_ids.module_id to -* avoid "maybe-uninitialized" compilation errors. Will remove when -* X86CPU supports cluster_id. -*/ - topo_ids.module_id =0 ; - cpu->apic_id =x86_apicid_from_topo_ids (&topo_info, &topo_ids); }
@@ -416,6 +425,14 @@ void x86_cpu_pre_plug(HotplugHandler *hotplug_dev, } cpu->die_id = topo_ids.die_id; +if (cpu->cluster_id !=-1 && cpu->cluster_id != topo_ids.module_id) { +error_setg (errp,"property cluster-id: %u doesn't match set apic-id:" +" 0x%x (cluster-id: %u)" , cpu->cluster_id, cpu->apic_id, + topo_ids.module_id); +return ; + } + cpu->cluster_id = topo_ids.module_id; +if (cpu->core_id !=-1 && cpu->core_id != topo_ids.core_id) { error_setg(errp, "property core-id: %u doesn't match set apic-id:" " 0x%x (core-id: %u)", cpu->core_id, cpu->apic_id,
target/i386/cpu.c +2 -0
@@ -6980,12 +6980,14 @@ static Property x86_cpu_properties[] = {DEFINE_PROP_UINT32 ("apic-id" ,X86CPU , apic_id,0 ),DEFINE_PROP_INT32 ("thread-id" ,X86CPU , thread_id,0 ),DEFINE_PROP_INT32 ("core-id" ,X86CPU , core_id,0 ), +DEFINE_PROP_INT32 ("cluster-id" ,X86CPU , cluster_id,0 ),DEFINE_PROP_INT32 ("die-id" ,X86CPU , die_id,0 ),DEFINE_PROP_INT32 ("socket-id" ,X86CPU , socket_id,0 ), #elseDEFINE_PROP_UINT32 ("apic-id" ,X86CPU , apic_id,UNASSIGNED_APIC_ID ),DEFINE_PROP_INT32 ("thread-id" ,X86CPU , thread_id,-1 ),DEFINE_PROP_INT32 ("core-id" ,X86CPU , core_id,-1 ), +DEFINE_PROP_INT32 ("cluster-id" ,X86CPU , cluster_id,-1 ),DEFINE_PROP_INT32 ("die-id" ,X86CPU , die_id,-1 ), DEFINE_PROP_INT32("socket-id", X86CPU, socket_id, -1), #endif
target/i386/cpu.h +1 -0
@@ -1966,6 +1966,7 @@ struct ArchCPU {int32_t node_id;/* NUMA node this CPU belongs to */ int32_t socket_id;int32_t die_id; +int32_t cluster_id;int32_t core_id;int32_t thread_id;
Signature · 1 line
2.34.1
PATCH RESEND 12/18 tests: Add test case of APIC ID for module level parsing
In reply to Zhao LiuFrom: Zhuocheng Ding <[email protected]>
After i386 supports module level, it's time to add the test for module level's parsing.
--- tests/unit/test-x86-apicid.c | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-)
tests/unit/test-x86-apicid.c +15 -4
@@ -37,6 +37,7 @@ static void test_topo_bits(void) topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_core_width(&topo_info), ==, 0); + g_assert_cmpuint(apicid_module_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 };
@@ -74,13 +75,22 @@ static void test_topo_bits(void) topo_info = (X86CPUTopoInfo ) {1 ,1 ,33 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 6); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,6 ,30 ,2 }; + g_assert_cmpuint(apicid_module_width(&topo_info), ==, 3); + topo_info = (X86CPUTopoInfo ) {1 ,7 ,30 ,2 }; + g_assert_cmpuint(apicid_module_width(&topo_info), ==, 3); + topo_info = (X86CPUTopoInfo ) {1 ,8 ,30 ,2 }; + g_assert_cmpuint(apicid_module_width(&topo_info), ==, 3); + topo_info = (X86CPUTopoInfo ) {1 ,9 ,30 ,2 }; + g_assert_cmpuint(apicid_module_width(&topo_info), ==, 4); + + topo_info = (X86CPUTopoInfo ) {1 ,6 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); - topo_info = (X86CPUTopoInfo ) {2 ,1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {2 ,6 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 1); - topo_info = (X86CPUTopoInfo ) {3 ,1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {3 ,6 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {4 ,1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {4 ,6 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2);/* build a weird topology and see if IDs are calculated correctly
@@ -91,6 +101,7 @@ static void test_topo_bits(void) topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); g_assert_cmpuint(apicid_core_offset(&topo_info), ==, 2); + g_assert_cmpuint(apicid_module_offset(&topo_info), ==, 5); g_assert_cmpuint(apicid_die_offset(&topo_info), ==, 5); g_assert_cmpuint(apicid_pkg_offset(&topo_info), ==, 5);
Signature · 1 line
2.34.1
Re: [PATCH RESEND 12/18] tests: Add test case of APIC ID for module level parsing
Quoted 52 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
After i386 supports module level, it's time to add the test for module level's parsing.
Signed-off-by Zhuocheng Ding <[email protected]>Co-developed-by Zhao Liu <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- tests/unit/test-x86-apicid.c | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-)tests/unit/test-x86-apicid.c +14
-4@@ -37,6 +37,7 @@ static void test_topo_bits(void) topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_core_width(&topo_info), ==, 0); + g_assert_cmpuint(apicid_module_width(&topo_info), ==, 0); g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); topo_info = (X86CPUTopoInfo ) {1 ,1 ,1 ,1 };@@ -74,13 +75,22 @@ static void test_topo_bits(void) topo_info = (X86CPUTopoInfo ) {1 ,1 ,33 ,2 }; g_assert_cmpuint(apicid_core_width(&topo_info), ==, 6); - topo_info = (X86CPUTopoInfo ) {1 ,1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {1 ,6 ,30 ,2 }; + g_assert_cmpuint(apicid_module_width(&topo_info), ==, 3); + topo_info = (X86CPUTopoInfo ) {1 ,7 ,30 ,2 }; + g_assert_cmpuint(apicid_module_width(&topo_info), ==, 3); + topo_info = (X86CPUTopoInfo ) {1 ,8 ,30 ,2 }; + g_assert_cmpuint(apicid_module_width(&topo_info), ==, 3); + topo_info = (X86CPUTopoInfo ) {1 ,9 ,30 ,2 }; + g_assert_cmpuint(apicid_module_width(&topo_info), ==, 4); + + topo_info = (X86CPUTopoInfo ) {1 ,6 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 0); - topo_info = (X86CPUTopoInfo ) {2 ,1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {2 ,6 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 1); - topo_info = (X86CPUTopoInfo ) {3 ,1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {3 ,6 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2); - topo_info = (X86CPUTopoInfo ) {4 ,1 ,30 ,2 }; + topo_info = (X86CPUTopoInfo ) {4 ,6 ,30 ,2 }; g_assert_cmpuint(apicid_die_width(&topo_info), ==, 2);/* build a weird topology and see if IDs are calculated correctly
tests/unit/test-x86-apicid.c static void test_topo_bits(void) · line 107
Quoted 7 lines
tests/unit/test-x86-apicid.c +1
-0@@ -91,6 +101,7 @@ static void test_topo_bits(void) topo_info = (X86CPUTopoInfo ) {1 ,1 ,6 ,3 }; g_assert_cmpuint(apicid_smt_width(&topo_info), ==, 2); g_assert_cmpuint(apicid_core_offset(&topo_info), ==, 2); + g_assert_cmpuint(apicid_module_offset(&topo_info), ==, 5); g_assert_cmpuint(apicid_die_offset(&topo_info), ==, 5); g_assert_cmpuint(apicid_pkg_offset(&topo_info), ==, 5);
Thanks,
Yanan
PATCH RESEND 13/18 hw/i386/pc: Support smp.clusters for x86 PC machine
In reply to Zhao LiuFrom: Zhuocheng Ding <[email protected]>
As module-level topology support is added to X86CPU, now we can enable the support for the cluster parameter on PC machines. With this support, we can define a 5-level x86 CPU topology with "-smp":
-smp cpus=*,maxcpus=*,sockets=*,dies=*,clusters=*,cores=*,threads=*.
Additionally, add the 5-level topology example in description of "-smp".
--- hw/i386/pc.c | 1 + qemu-options.hx | 10 +++++----- 2 files changed, 6 insertions(+), 5 deletions(-)
hw/i386/pc.c +1 -0
@@ -1929,6 +1929,7 @@ static void pc_machine_class_init(ObjectClass *oc, void *data) mc->default_cpu_type =TARGET_DEFAULT_CPU_TYPE ; mc->nvdimm_supported = true; mc->smp_props.dies_supported = true; + mc->smp_props.clusters_supported = true; mc->default_ram_id ="pc.ram" ; object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
qemu-options.hx +5 -5
@@ -312,14 +312,14 @@ SRST -smp 8,sockets=2,cores=2,threads=2,maxcpus=8 The following sub-option defines a CPU topology hierarchy (2 sockets - totally on the machine, 2 dies per socket, 2 cores per die, 2 threads - per core) for PC machines which support sockets/dies/cores/threads. - Some members of the option can be omitted but their values will be - automatically computed: + totally on the machine, 2 dies per socket, 2 clusters per die, 2 cores per + cluster, 2 threads per core) for PC machines which support sockets/dies + /clusters/cores/threads. Some members of the option can be omitted but + their values will be automatically computed: :: - -smp 16,sockets=2,dies=2,cores=2,threads=2,maxcpus=16 + -smp 32,sockets=2,dies=2,clusters=2,cores=2,threads=2,maxcpus=32 The following sub-option defines a CPU topology hierarchy (2 sockets totally on the machine, 2 clusters per socket, 2 cores per cluster,
Signature · 1 line
2.34.1
Re: [PATCH RESEND 13/18] hw/i386/pc: Support smp.clusters for x86 PC machine
Quoted 8 lines · Zhao Liu
From: Zhuocheng Ding <[email protected]>
As module-level topology support is added to X86CPU, now we can enable the support for the cluster parameter on PC machines. With this support, we can define a 5-level x86 CPU topology with "-smp":
-smp cpus=*,maxcpus=*,sockets=*,dies=*,clusters=*,cores=*,threads=*.
Quoted 8 lines
Additionally, add the 5-level topology example in description of "-smp".
Signed-off-by Zhuocheng Ding <[email protected]>Signed-off-by Zhao Liu <[email protected]>--- hw/i386/pc.c | 1 + qemu-options.hx | 10 +++++----- 2 files changed, 6 insertions(+), 5 deletions(-)
Thanks,
Yanan
Quoted 37 lines
hw/i386/pc.c +1
-0@@ -1929,6 +1929,7 @@ static void pc_machine_class_init(ObjectClass *oc, void *data) mc->default_cpu_type =TARGET_DEFAULT_CPU_TYPE ; mc->nvdimm_supported = true; mc->smp_props.dies_supported = true; + mc->smp_props.clusters_supported = true; mc->default_ram_id ="pc.ram" ; object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",qemu-options.hx +5
-5@@ -312,14 +312,14 @@ SRST -smp 8,sockets=2,cores=2,threads=2,maxcpus=8 The following sub-option defines a CPU topology hierarchy (2 sockets - totally on the machine, 2 dies per socket, 2 cores per die, 2 threads - per core) for PC machines which support sockets/dies/cores/threads. - Some members of the option can be omitted but their values will be - automatically computed: + totally on the machine, 2 dies per socket, 2 clusters per die, 2 cores per + cluster, 2 threads per core) for PC machines which support sockets/dies + /clusters/cores/threads. Some members of the option can be omitted but + their values will be automatically computed: :: - -smp 16,sockets=2,dies=2,cores=2,threads=2,maxcpus=16 + -smp 32,sockets=2,dies=2,clusters=2,cores=2,threads=2,maxcpus=32 The following sub-option defines a CPU topology hierarchy (2 sockets totally on the machine, 2 clusters per socket, 2 cores per cluster,
PATCH RESEND 14/18 i386: Add cache topology info in CPUCacheInfo
In reply to Zhao LiuFrom: Zhao Liu <[email protected]>
Currently, by default, the cache topology is encoded as:
1. i/d cache is shared in one core.
2. L2 cache is shared in one core.
3. L3 cache is shared in one die.
This default general setting has caused a misunderstanding, that is, the cache topology is completely equated with a specific cpu topology, such as the connection between L2 cache and core level, and the connection between L3 cache and die level.
In fact, the settings of these topologies depend on the specific platform and are not static. For example, on Alder Lake-P, every four Atom cores share the same L2 cache.
Thus, we should explicitly define the corresponding cache topology for different cache models to increase scalability.
Except legacy_l2_cache_cpuid2 (its default topo level is INVALID), explicitly set the corresponding topology level for all other cache models. In order to be compatible with the existing cache topology, set the CORE level for the i/d cache, set the CORE level for L2 cache, and set the DIE level for L3 cache.
The field for CPUID[4].EAX[bits 25:14] or CPUID[0x8000001D].EAX[bits 25:14] will be set based on CPUCacheInfo.share_level.
--- target/i386/cpu.c | 19 +++++++++++++++++++ target/i386/cpu.h | 16 ++++++++++++++++ 2 files changed, 35 insertions(+)
target/i386/cpu.c +19 -0
@@ -433,6 +433,7 @@ static CPUCacheInfo legacy_l1d_cache = { .sets = 64, .partitions = 1, .no_invd_sharing = true, + .share_level = CORE, };/*FIXME: CPUID leaf 0x80000005 is inconsistent with leaves 2 & 4 */
@@ -447,6 +448,7 @@ static CPUCacheInfo legacy_l1d_cache_amd = { .partitions = 1, .lines_per_tag = 1, .no_invd_sharing = true, + .share_level = CORE, };/* L1 instruction cache: */
@@ -460,6 +462,7 @@ static CPUCacheInfo legacy_l1i_cache = { .sets = 64, .partitions = 1, .no_invd_sharing = true, + .share_level = CORE, };/*FIXME: CPUID leaf 0x80000005 is inconsistent with leaves 2 & 4 */
@@ -474,6 +477,7 @@ static CPUCacheInfo legacy_l1i_cache_amd = { .partitions = 1, .lines_per_tag = 1, .no_invd_sharing = true, + .share_level = CORE, };/* Level 2 unified cache: */
@@ -487,6 +491,7 @@ static CPUCacheInfo legacy_l2_cache = { .sets = 4096, .partitions = 1, .no_invd_sharing = true, + .share_level = CORE, };/*FIXME: CPUID leaf 2 descriptor is inconsistent with CPUID leaf 4 */
@@ -509,6 +514,7 @@ static CPUCacheInfo legacy_l2_cache_amd = { .associativity = 16, .sets = 512, .partitions = 1, + .share_level = CORE, };/* Level 3 unified cache: */
@@ -524,6 +530,7 @@ static CPUCacheInfo legacy_l3_cache = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, };/* TLB definitions: */
@@ -1668,6 +1675,7 @@ static const CPUCaches epyc_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l1i_cache = &(CPUCacheInfo ) { .type =INSTRUCTION_CACHE ,
@@ -1680,6 +1688,7 @@ static const CPUCaches epyc_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l2_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,
@@ -1690,6 +1699,7 @@ static const CPUCaches epyc_cache_info = { .partitions = 1, .sets = 1024, .lines_per_tag = 1, + .share_level = CORE, }, .l3_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,
@@ -1703,6 +1713,7 @@ static const CPUCaches epyc_cache_info = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, }, };
@@ -1718,6 +1729,7 @@ static const CPUCaches epyc_rome_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l1i_cache = &(CPUCacheInfo ) { .type =INSTRUCTION_CACHE ,
@@ -1730,6 +1742,7 @@ static const CPUCaches epyc_rome_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l2_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,
@@ -1740,6 +1753,7 @@ static const CPUCaches epyc_rome_cache_info = { .partitions = 1, .sets = 1024, .lines_per_tag = 1, + .share_level = CORE, }, .l3_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,
@@ -1753,6 +1767,7 @@ static const CPUCaches epyc_rome_cache_info = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, }, };
@@ -1768,6 +1783,7 @@ static const CPUCaches epyc_milan_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l1i_cache = &(CPUCacheInfo ) { .type =INSTRUCTION_CACHE ,
@@ -1780,6 +1796,7 @@ static const CPUCaches epyc_milan_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l2_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,
@@ -1790,6 +1807,7 @@ static const CPUCaches epyc_milan_cache_info = { .partitions = 1, .sets = 1024, .lines_per_tag = 1, + .share_level = CORE, }, .l3_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,
@@ -1803,6 +1821,7 @@ static const CPUCaches epyc_milan_cache_info = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, }, };
target/i386/cpu.h +16 -0
@@ -1476,6 +1476,15 @@ enum CacheType {UNIFIED_CACHE }; +enum CPUTopoLevel { +INVALID =0 , +SMT , +CORE , +MODULE , +DIE , +PACKAGE , +}; + typedef struct CPUCacheInfo { enum CacheType type; uint8_t level;
@@ -1517,6 +1526,13 @@ typedef struct CPUCacheInfo {* address bits. CPUID[4].EDX[bit 2]. */ bool complex_indexing; + +/* +* Cache Topology. The level that cache is shared in. +* Used to encode CPUID[4].EAX[bits 25:14] or +* CPUID[0x8000001D].EAX[bits 25:14]. +*/ +enum CPUTopoLevel share_level; } CPUCacheInfo;
Signature · 1 line
2.34.1
Re: [PATCH RESEND 14/18] i386: Add cache topology info in CPUCacheInfo
Hi Zhao,
Quoted 12 lines · Zhao Liu
From: Zhao Liu <[email protected]>
Currently, by default, the cache topology is encoded as:
1. i/d cache is shared in one core.
2. L2 cache is shared in one core.
3. L3 cache is shared in one die.This default general setting has caused a misunderstanding, that is, the cache topology is completely equated with a specific cpu topology, such as the connection between L2 cache and core level, and the connection between L3 cache and die level.
Quoted 8 lines
In fact, the settings of these topologies depend on the specific platform and are not static. For example, on Alder Lake-P, every four Atom cores share the same L2 cache.
Thus, we should explicitly define the corresponding cache topology for different cache models to increase scalability.
Except
legacy_l2_cache_cpuid2(its default topo level is INVALID),
It seems like its default topo level is UNKNOWN in this case.
Quoted 175 lines
explicitly set the corresponding topology level for all other cache models. In order to be compatible with the existing cache topology, set the CORE level for the i/d cache, set the CORE level for L2 cache, and set the DIE level for L3 cache.
The field for CPUID[4].EAX[bits 25:14] or CPUID[0x8000001D].EAX[bits 25:14] will be set based on CPUCacheInfo.share_level.
Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 19 +++++++++++++++++++ target/i386/cpu.h | 16 ++++++++++++++++ 2 files changed, 35 insertions(+)target/i386/cpu.c +19
-0@@ -433,6 +433,7 @@ static CPUCacheInfo legacy_l1d_cache = { .sets = 64, .partitions = 1, .no_invd_sharing = true, + .share_level = CORE, };/*FIXME: CPUID leaf 0x80000005 is inconsistent with leaves 2 & 4 */ @@ -447,6 +448,7 @@ static CPUCacheInfo legacy_l1d_cache_amd = { .partitions = 1, .lines_per_tag = 1, .no_invd_sharing = true, + .share_level = CORE, };/* L1 instruction cache: */ @@ -460,6 +462,7 @@ static CPUCacheInfo legacy_l1i_cache = { .sets = 64, .partitions = 1, .no_invd_sharing = true, + .share_level = CORE, };/*FIXME: CPUID leaf 0x80000005 is inconsistent with leaves 2 & 4 */ @@ -474,6 +477,7 @@ static CPUCacheInfo legacy_l1i_cache_amd = { .partitions = 1, .lines_per_tag = 1, .no_invd_sharing = true, + .share_level = CORE, };/* Level 2 unified cache: */ @@ -487,6 +491,7 @@ static CPUCacheInfo legacy_l2_cache = { .sets = 4096, .partitions = 1, .no_invd_sharing = true, + .share_level = CORE, };/*FIXME: CPUID leaf 2 descriptor is inconsistent with CPUID leaf 4 */ @@ -509,6 +514,7 @@ static CPUCacheInfo legacy_l2_cache_amd = { .associativity = 16, .sets = 512, .partitions = 1, + .share_level = CORE, };/* Level 3 unified cache: */ @@ -524,6 +530,7 @@ static CPUCacheInfo legacy_l3_cache = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, };/* TLB definitions: */ @@ -1668,6 +1675,7 @@ static const CPUCaches epyc_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l1i_cache = &(CPUCacheInfo ) { .type =INSTRUCTION_CACHE ,@@ -1680,6 +1688,7 @@ static const CPUCaches epyc_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l2_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1690,6 +1699,7 @@ static const CPUCaches epyc_cache_info = { .partitions = 1, .sets = 1024, .lines_per_tag = 1, + .share_level = CORE, }, .l3_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1703,6 +1713,7 @@ static const CPUCaches epyc_cache_info = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, }, };@@ -1718,6 +1729,7 @@ static const CPUCaches epyc_rome_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l1i_cache = &(CPUCacheInfo ) { .type =INSTRUCTION_CACHE ,@@ -1730,6 +1742,7 @@ static const CPUCaches epyc_rome_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l2_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1740,6 +1753,7 @@ static const CPUCaches epyc_rome_cache_info = { .partitions = 1, .sets = 1024, .lines_per_tag = 1, + .share_level = CORE, }, .l3_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1753,6 +1767,7 @@ static const CPUCaches epyc_rome_cache_info = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, }, };@@ -1768,6 +1783,7 @@ static const CPUCaches epyc_milan_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l1i_cache = &(CPUCacheInfo ) { .type =INSTRUCTION_CACHE ,@@ -1780,6 +1796,7 @@ static const CPUCaches epyc_milan_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l2_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1790,6 +1807,7 @@ static const CPUCaches epyc_milan_cache_info = { .partitions = 1, .sets = 1024, .lines_per_tag = 1, + .share_level = CORE, }, .l3_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1803,6 +1821,7 @@ static const CPUCaches epyc_milan_cache_info = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, }, };target/i386/cpu.h +0
-0
target/i386/cpu.h enum CacheType { · line 1480
Quoted 5 lines
target/i386/cpu.h +2
-0@@ -1476,6 +1476,15 @@ enum CacheType {UNIFIED_CACHE }; +enum CPUTopoLevel { +INVALID =0 ,
Maybe UNKNOWN?
target/i386/cpu.h enum CacheType { · line 1485
Quoted 5 lines
target/i386/cpu.h +5
-0+SMT , +CORE , +MODULE , +DIE , + PACKAGE,
Do we need a prefix like CPU_TOPO_LEVEL_* or shorter CPU_TL_* to avoid possible naming conflicts with other micros or enums? Not sure..
Thanks,
Yanan
Quoted 19 lines
target/i386/cpu.h +9
-0+}; + typedef struct CPUCacheInfo { enum CacheType type; uint8_t level;@@ -1517,6 +1526,13 @@ typedef struct CPUCacheInfo {* address bits. CPUID[4].EDX[bit 2]. */ bool complex_indexing; + +/* +* Cache Topology. The level that cache is shared in. +* Used to encode CPUID[4].EAX[bits 25:14] or +* CPUID[0x8000001D].EAX[bits 25:14]. +*/ +enum CPUTopoLevel share_level; } CPUCacheInfo;
Re: [PATCH RESEND 14/18] i386: Add cache topology info in CPUCacheInfo
Quoted 191 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 20:17:22 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 14/18] i386: Add cache topology info in
CPUCacheInfoHi Zhao,
Quoted 12 lines · Zhao Liu
From: Zhao Liu <[email protected]>
Currently, by default, the cache topology is encoded as:
1. i/d cache is shared in one core.
2. L2 cache is shared in one core.
3. L3 cache is shared in one die.This default general setting has caused a misunderstanding, that is, the cache topology is completely equated with a specific cpu topology, such as the connection between L2 cache and core level, and the connection between L3 cache and die level.
Quoted 8 lines
In fact, the settings of these topologies depend on the specific platform and are not static. For example, on Alder Lake-P, every four Atom cores share the same L2 cache.
Thus, we should explicitly define the corresponding cache topology for different cache models to increase scalability.
Except
legacy_l2_cache_cpuid2(its default topo level is INVALID),It seems like its default topo level is UNKNOWN in this case.
Quoted 162 lines
explicitly set the corresponding topology level for all other cache models. In order to be compatible with the existing cache topology, set the CORE level for the i/d cache, set the CORE level for L2 cache, and set the DIE level for L3 cache.
The field for CPUID[4].EAX[bits 25:14] or CPUID[0x8000001D].EAX[bits 25:14] will be set based on CPUCacheInfo.share_level.
Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 19 +++++++++++++++++++ target/i386/cpu.h | 16 ++++++++++++++++ 2 files changed, 35 insertions(+)target/i386/cpu.c +19
-0@@ -433,6 +433,7 @@ static CPUCacheInfo legacy_l1d_cache = { .sets = 64, .partitions = 1, .no_invd_sharing = true, + .share_level = CORE, };/*FIXME: CPUID leaf 0x80000005 is inconsistent with leaves 2 & 4 */ @@ -447,6 +448,7 @@ static CPUCacheInfo legacy_l1d_cache_amd = { .partitions = 1, .lines_per_tag = 1, .no_invd_sharing = true, + .share_level = CORE, };/* L1 instruction cache: */ @@ -460,6 +462,7 @@ static CPUCacheInfo legacy_l1i_cache = { .sets = 64, .partitions = 1, .no_invd_sharing = true, + .share_level = CORE, };/*FIXME: CPUID leaf 0x80000005 is inconsistent with leaves 2 & 4 */ @@ -474,6 +477,7 @@ static CPUCacheInfo legacy_l1i_cache_amd = { .partitions = 1, .lines_per_tag = 1, .no_invd_sharing = true, + .share_level = CORE, };/* Level 2 unified cache: */ @@ -487,6 +491,7 @@ static CPUCacheInfo legacy_l2_cache = { .sets = 4096, .partitions = 1, .no_invd_sharing = true, + .share_level = CORE, };/*FIXME: CPUID leaf 2 descriptor is inconsistent with CPUID leaf 4 */ @@ -509,6 +514,7 @@ static CPUCacheInfo legacy_l2_cache_amd = { .associativity = 16, .sets = 512, .partitions = 1, + .share_level = CORE, };/* Level 3 unified cache: */ @@ -524,6 +530,7 @@ static CPUCacheInfo legacy_l3_cache = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, };/* TLB definitions: */ @@ -1668,6 +1675,7 @@ static const CPUCaches epyc_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l1i_cache = &(CPUCacheInfo ) { .type =INSTRUCTION_CACHE ,@@ -1680,6 +1688,7 @@ static const CPUCaches epyc_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l2_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1690,6 +1699,7 @@ static const CPUCaches epyc_cache_info = { .partitions = 1, .sets = 1024, .lines_per_tag = 1, + .share_level = CORE, }, .l3_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1703,6 +1713,7 @@ static const CPUCaches epyc_cache_info = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, }, };@@ -1718,6 +1729,7 @@ static const CPUCaches epyc_rome_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l1i_cache = &(CPUCacheInfo ) { .type =INSTRUCTION_CACHE ,@@ -1730,6 +1742,7 @@ static const CPUCaches epyc_rome_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l2_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1740,6 +1753,7 @@ static const CPUCaches epyc_rome_cache_info = { .partitions = 1, .sets = 1024, .lines_per_tag = 1, + .share_level = CORE, }, .l3_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1753,6 +1767,7 @@ static const CPUCaches epyc_rome_cache_info = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, }, };@@ -1768,6 +1783,7 @@ static const CPUCaches epyc_milan_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l1i_cache = &(CPUCacheInfo ) { .type =INSTRUCTION_CACHE ,@@ -1780,6 +1796,7 @@ static const CPUCaches epyc_milan_cache_info = { .lines_per_tag = 1, .self_init = 1, .no_invd_sharing = true, + .share_level = CORE, }, .l2_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1790,6 +1807,7 @@ static const CPUCaches epyc_milan_cache_info = { .partitions = 1, .sets = 1024, .lines_per_tag = 1, + .share_level = CORE, }, .l3_cache = &(CPUCacheInfo ) { .type =UNIFIED_CACHE ,@@ -1803,6 +1821,7 @@ static const CPUCaches epyc_milan_cache_info = { .self_init = true, .inclusive = true, .complex_indexing = true, + .share_level = DIE, }, };target/i386/cpu.h +0
-0
Quoted 8 lines
target/i386/cpu.h enum CacheType { · line 1479
Quoted 7 lines
target/i386/cpu.h +2
-0@@ -1476,6 +1476,15 @@ enum CacheType {UNIFIED_CACHE }; +enum CPUTopoLevel { +INVALID =0 ,Maybe UNKNOWN?
I agree, it's a better name than INVALID.
Quoted 8 lines
target/i386/cpu.h enum CacheType { · line 1485
Quoted 5 lines
target/i386/cpu.h +5
-0+SMT , +CORE , +MODULE , +DIE , + PACKAGE,Do we need a prefix like CPU_TOPO_LEVEL_* or shorter CPU_TL_* to avoid possible naming conflicts with other micros or enums? Not sure..
Yes, adding a prefix for enum is more in line with the QEMU naming style, I will add it. I support the first name, it looks more clear.
Thanks,
Zhao
Quoted 21 lines
Thanks,
YananQuoted 17 lines
target/i386/cpu.h +9
-0+}; + typedef struct CPUCacheInfo { enum CacheType type; uint8_t level;@@ -1517,6 +1526,13 @@ typedef struct CPUCacheInfo {* address bits. CPUID[4].EDX[bit 2]. */ bool complex_indexing; + +/* +* Cache Topology. The level that cache is shared in. +* Used to encode CPUID[4].EAX[bits 25:14] or +* CPUID[0x8000001D].EAX[bits 25:14]. +*/ +enum CPUTopoLevel share_level; } CPUCacheInfo;
PATCH RESEND 15/18 i386: Use CPUCacheInfo.share_level to encode CPUID[4].EAX[bits 25:14]
In reply to Zhao LiuFrom: Zhao Liu <[email protected]>
CPUID[4].EAX[bits 25:14] is used to represent the cache topology for intel CPUs.
After cache models have topology information, we can use CPUCacheInfo.share_level to decide which topology level to be encoded into CPUID[4].EAX[bits 25:14].
Additionally, since maximum_processor_id (original "num_apic_ids") is parsed based on cpu topology levels, which are verified when parsing smp, it's no need to check this value by "assert(num_apic_ids > 0)" again, so remove this assert.
--- target/i386/cpu.c | 55 +++++++++++++++++++++++++++++++---------------- 1 file changed, 36 insertions(+), 19 deletions(-)
target/i386/cpu.c +36 -19
@@ -231,22 +231,50 @@ static uint8_t cpuid2_cache_descriptor(CPUCacheInfo *cache) ((t) == UNIFIED_CACHE) ? CACHE_TYPE_UNIFIED : \ 0 /* Invalid value */) +static uint32_t max_processor_ids_for_cache (CPUCacheInfo *cache, +X86CPUTopoInfo *topo_info) +{ +uint32_t num_ids =0 ; + +switch (cache->share_level) { +case CORE : + num_ids =1 <<apicid_core_offset (topo_info); +break ; +case DIE : + num_ids =1 <<apicid_die_offset (topo_info); +break ; +default : +/* +* Currently there is no use case for SMT, MODULE and PACKAGE, so use +* assert directly to facilitate debugging. +*/ +g_assert_not_reached (); + } + +return num_ids -1 ; +} + +static uint32_t max_core_ids_in_package (X86CPUTopoInfo *topo_info) +{ +uint32_t num_cores =1 << (apicid_pkg_offset (topo_info) - +apicid_core_offset (topo_info)); +return num_cores -1 ; +}/* Encode cache info for CPUID[4] */ static void encode_cache_cpuid4 (CPUCacheInfo *cache, -int num_apic_ids,int num_cores, +X86CPUTopoInfo *topo_info,uint32_t *eax,uint32_t *ebx,uint32_t *ecx,uint32_t *edx) {assert (cache->size == cache->line_size * cache->associativity * cache->partitions * cache->sets); -assert (num_apic_ids >0 ); *eax =CACHE_TYPE (cache->type) |CACHE_LEVEL (cache->level) | (cache->self_init ?CACHE_SELF_INIT_LEVEL :0 ) | - ((num_cores -1 ) <<26 ) | - ((num_apic_ids -1 ) <<14 ); + (max_core_ids_in_package (topo_info) <<26 ) | + (max_processor_ids_for_cache (cache, topo_info) <<14 );assert (cache->line_size >0 );assert (cache->partitions >0 );
@@ -5335,38 +5363,27 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count, *eax = *ebx = *ecx = *edx =0 ; }else { *eax =0 ; -int addressable_cores_offset =apicid_pkg_offset (&topo_info) - -apicid_core_offset (&topo_info); -int core_offset, die_offset;switch (count) {case 0 :/* L1 dcache info */ - core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l1d_cache, - (1 << core_offset), - (1 << addressable_cores_offset), + &topo_info, eax, ebx, ecx, edx);break ;case 1 :/* L1 icache info */ - core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l1i_cache, - (1 << core_offset), - (1 << addressable_cores_offset), + &topo_info, eax, ebx, ecx, edx);break ;case 2 :/* L2 cache info */ - core_offset =apicid_core_offset (&topo_info);encode_cache_cpuid4 (env->cache_info_cpuid4.l2_cache, - (1 << core_offset), - (1 << addressable_cores_offset), + &topo_info, eax, ebx, ecx, edx);break ;case 3 :/* L3 cache info */ - die_offset =apicid_die_offset (&topo_info);if (cpu->enable_l3_cache) {encode_cache_cpuid4 (env->cache_info_cpuid4.l3_cache, - (1 << die_offset), - (1 << addressable_cores_offset), + &topo_info, eax, ebx, ecx, edx);break ; }
Signature · 1 line
2.34.1
PATCH RESEND 16/18 i386: Fix NumSharingCache for CPUID[0x8000001D].EAX[bits 25:14]
In reply to Zhao LiuFrom: Zhao Liu <[email protected]>
From AMD's APM, NumSharingCache (CPUID[0x8000001D].EAX[bits 25:14]) means [1]:
The number of logical processors sharing this cache is the value of this field incremented by 1. To determine which logical processors are sharing a cache, determine a Share Id for each processor as follows:
ShareId = LocalApicId >> log2(NumSharingCache+1)
Logical processors with the same ShareId then share a cache. If NumSharingCache+1 is not a power of two, round it up to the next power of two.
From the description above, the caculation of this feild should be same as CPUID[4].EAX[bits 25:14] for intel cpus. So also use the offsets of APIC ID to caculate this field.
Note: I don't have the hardware available, hope someone can help me to confirm whether this calculation is correct, thanks!
[1]: APM, vol.3, appendix.E.4.15 Function 8000_001Dh--Cache Topology
Information
--- target/i386/cpu.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)
target/i386/cpu.c +4 -6
@@ -355,7 +355,7 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache, uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx) { - uint32_t l3_threads; + uint32_t sharing_apic_ids;assert (cache->size == cache->line_size * cache->associativity * cache->partitions * cache->sets);
@@ -364,13 +364,11 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache,/* L3 is shared among multiple cores */ if (cache->level ==3 ) { - l3_threads = topo_info->modules_per_die * - topo_info->cores_per_module * - topo_info->threads_per_core; - *eax |= (l3_threads -1 ) <<14 ; + sharing_apic_ids =1 <<apicid_die_offset (topo_info); }else { - *eax |= ((topo_info->threads_per_core -1 ) <<14 ); + sharing_apic_ids =1 <<apicid_core_offset (topo_info); } + *eax |= (sharing_apic_ids -1 ) <<14 ;assert (cache->line_size >0 );assert (cache->partitions >0 );
Signature · 1 line
2.34.1
Re: [PATCH RESEND 16/18] i386: Fix NumSharingCache for CPUID[0x8000001D].EAX[bits 25:14]
Quoted 35 lines · Zhao Liu
From: Zhao Liu <[email protected]>
Quoted 1 line
From AMD's APM, NumSharingCache (CPUID[0x8000001D].EAX[bits 25:14])
means [1]:
The number of logical processors sharing this cache is the value of this field incremented by 1. To determine which logical processors are sharing a cache, determine a Share Id for each processor as follows:
ShareId = LocalApicId >> log2(NumSharingCache+1)
Logical processors with the same ShareId then share a cache. If NumSharingCache+1 is not a power of two, round it up to the next power of two.
Quoted 1 line
From the description above, the caculation of this feild should be same
as CPUID[4].EAX[bits 25:14] for intel cpus. So also use the offsets of APIC ID to caculate this field.
Note: I don't have the hardware available, hope someone can help me to confirm whether this calculation is correct, thanks!
[1]: APM, vol.3, appendix.E.4.15 Function 8000_001Dh--Cache Topology
InformationSigned-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)target/i386/cpu.c +0
-0
target/i386/cpu.c static void encode_cache_cpuid8000001d(CPUCacheInfo *cache, · line 358
Quoted 5 lines
target/i386/cpu.c +1
-1@@ -355,7 +355,7 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache, uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx) { - uint32_t l3_threads; + uint32_t sharing_apic_ids;
maybe num_apic_ids or num_ids?
Quoted 20 lines
target/i386/cpu.c +3
-5assert (cache->size == cache->line_size * cache->associativity * cache->partitions * cache->sets);@@ -364,13 +364,11 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache,/* L3 is shared among multiple cores */ if (cache->level ==3 ) { - l3_threads = topo_info->modules_per_die * - topo_info->cores_per_module * - topo_info->threads_per_core; - *eax |= (l3_threads -1 ) <<14 ; + sharing_apic_ids =1 <<apicid_die_offset (topo_info); }else { - *eax |= ((topo_info->threads_per_core -1 ) <<14 ); + sharing_apic_ids =1 <<apicid_core_offset (topo_info); } + *eax |= (sharing_apic_ids -1 ) <<14 ;assert (cache->line_size >0 );assert (cache->partitions >0 );
Re: [PATCH RESEND 16/18] i386: Fix NumSharingCache for CPUID[0x8000001D].EAX[bits 25:14]
Quoted 36 lines · wangyanan (Y)
Date: Wed, 15 Feb 2023 20:32:39 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 16/18] i386: Fix NumSharingCache for
CPUID[0x8000001D].EAX[bits 25:14]Quoted 30 lines · Zhao Liu
From: Zhao Liu <[email protected]>
Quoted 1 line
From AMD's APM, NumSharingCache (CPUID[0x8000001D].EAX[bits 25:14])
means [1]:
The number of logical processors sharing this cache is the value of this field incremented by 1. To determine which logical processors are sharing a cache, determine a Share Id for each processor as follows:
ShareId = LocalApicId >> log2(NumSharingCache+1)
Logical processors with the same ShareId then share a cache. If NumSharingCache+1 is not a power of two, round it up to the next power of two.
Quoted 1 line
From the description above, the caculation of this feild should be same
as CPUID[4].EAX[bits 25:14] for intel cpus. So also use the offsets of APIC ID to caculate this field.
Note: I don't have the hardware available, hope someone can help me to confirm whether this calculation is correct, thanks!
[1]: APM, vol.3, appendix.E.4.15 Function 8000_001Dh--Cache Topology
InformationSigned-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)
Quoted 11 lines
target/i386/cpu.c static void encode_cache_cpuid8000001d(CPUCacheInfo *cache, · line 358
Quoted 10 lines
target/i386/cpu.c +1
-1@@ -355,7 +355,7 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache, uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx) { - uint32_t l3_threads; + uint32_t sharing_apic_ids;maybe
num_apic_idsor num_ids?
Thanks, num is better as a prefix. I would use num_apic_ids.
Zhao
Quoted 18 lines
Quoted 17 lines
target/i386/cpu.c +3
-5assert (cache->size == cache->line_size * cache->associativity * cache->partitions * cache->sets);@@ -364,13 +364,11 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache,/* L3 is shared among multiple cores */ if (cache->level ==3 ) { - l3_threads = topo_info->modules_per_die * - topo_info->cores_per_module * - topo_info->threads_per_core; - *eax |= (l3_threads -1 ) <<14 ; + sharing_apic_ids =1 <<apicid_die_offset (topo_info); }else { - *eax |= ((topo_info->threads_per_core -1 ) <<14 ); + sharing_apic_ids =1 <<apicid_core_offset (topo_info); } + *eax |= (sharing_apic_ids -1 ) <<14 ;assert (cache->line_size >0 );assert (cache->partitions >0 );
PATCH RESEND 17/18 i386: Use CPUCacheInfo.share_level to encode CPUID[0x8000001D].EAX[bits 25:14]
In reply to Zhao LiuFrom: Zhao Liu <[email protected]>
CPUID[0x8000001D].EAX[bits 25:14] is used to represent the cache topology for amd CPUs.
After cache models have topology information, we can use CPUCacheInfo.share_level to decide which topology level to be encoded into CPUID[0x8000001D].EAX[bits 25:14].
--- target/i386/cpu.c | 10 +--------- 1 file changed, 1 insertion(+), 9 deletions(-)
target/i386/cpu.c +1 -9
@@ -355,20 +355,12 @@ static void encode_cache_cpuid8000001d(CPUCacheInfo *cache, uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx) { - uint32_t sharing_apic_ids; assert(cache->size == cache->line_size * cache->associativity * cache->partitions * cache->sets); *eax =CACHE_TYPE (cache->type) |CACHE_LEVEL (cache->level) | (cache->self_init ?CACHE_SELF_INIT_LEVEL :0 ); - -/* L3 is shared among multiple cores */ -if (cache->level ==3 ) { - sharing_apic_ids =1 <<apicid_die_offset (topo_info); - }else { - sharing_apic_ids =1 <<apicid_core_offset (topo_info); - } - *eax |= (sharing_apic_ids -1 ) <<14 ; + *eax |=max_processor_ids_for_cache (cache, topo_info) <<14 ;assert (cache->line_size >0 );assert (cache->partitions >0 );
Signature · 1 line
2.34.1
PATCH RESEND 18/18 i386: Add new property to control L2 cache topo in CPUID.04H
In reply to Zhao LiuFrom: Zhao Liu <[email protected]>
The property x-l2-cache-topo will be used to change the L2 cache topology in CPUID.04H.
Now it allows user to set the L2 cache is shared in core level or cluster level.
If user passes "-cpu x-l2-cache-topo=[core|cluster]" then older L2 cache topology will be overrided by the new topology setting.
Here we expose to user "cluster" instead of "module", to be consistent with "cluster-id" naming.
Since CPUID.04H is used by intel CPUs, this property is available on intel CPUs as for now.
When necessary, it can be extended to CPUID.8000001DH for amd CPUs.
--- target/i386/cpu.c | 33 ++++++++++++++++++++++++++++++++- target/i386/cpu.h | 2 ++ 2 files changed, 34 insertions(+), 1 deletion(-)
target/i386/cpu.c +32 -1
@@ -240,12 +240,15 @@ static uint32_t max_processor_ids_for_cache(CPUCacheInfo *cache,case CORE : num_ids =1 <<apicid_core_offset (topo_info);break ; +case MODULE : + num_ids =1 <<apicid_module_offset (topo_info); +break ;case DIE : num_ids =1 <<apicid_die_offset (topo_info);break ;default :/* -* Currently there is no use case for SMT, MODULE and PACKAGE, so use +* Currently there is no use case for SMT and PACKAGE, so use * assert directly to facilitate debugging. */ g_assert_not_reached ();
@@ -6633,6 +6636,33 @@ static void x86_cpu_realizefn(DeviceState *dev, Error **errp) env->cache_info_amd.l3_cache = &legacy_l3_cache; } +if (cpu->l2_cache_topo_level) { +/* +* FIXME: Currently only supports changing CPUID[4] (for intel), and +* will support changing CPUID[0x8000001D] when necessary. +*/ +if (!IS_INTEL_CPU (env)) { +error_setg (errp,"only intel cpus supports x-l2-cache-topo" ); +return ; + } + +if (!strcmp (cpu->l2_cache_topo_level,"core" )) { + env->cache_info_cpuid4.l2_cache->share_level =CORE ; + }else if (!strcmp (cpu->l2_cache_topo_level,"cluster" )) { +/* +* We expose to users "cluster" instead of "module", to be +* consistent with "cluster-id" naming. +*/ + env->cache_info_cpuid4.l2_cache->share_level =MODULE ; + }else { +error_setg (errp, +"x-l2-cache-topo doesn't support '%s', " +"and it only supports 'core' or 'cluster'" , + cpu->l2_cache_topo_level); +return ; + } + } +#ifndef CONFIG_USER_ONLY MachineState *ms =MACHINE (qdev_get_machine ());qemu_register_reset (x86_cpu_machine_reset_cb, cpu);
@@ -7135,6 +7165,7 @@ static Property x86_cpu_properties[] = { false),DEFINE_PROP_BOOL ("x-intel-pt-auto-level" ,X86CPU , intel_pt_auto_level, true), +DEFINE_PROP_STRING ("x-l2-cache-topo" ,X86CPU , l2_cache_topo_level), DEFINE_PROP_END_OF_LIST() };
target/i386/cpu.h +2 -0
@@ -1987,6 +1987,8 @@ struct ArchCPU {int32_t thread_id;int32_t hv_max_vps; + +char *l2_cache_topo_level; };
Signature · 1 line
2.34.1
Re: [PATCH RESEND 18/18] i386: Add new property to control L2 cache topo in CPUID.04H
Quoted 10 lines · Zhao Liu
From: Zhao Liu <[email protected]>
The property x-l2-cache-topo will be used to change the L2 cache topology in CPUID.04H.
Now it allows user to set the L2 cache is shared in core level or cluster level.
If user passes "-cpu x-l2-cache-topo=[core|cluster]" then older L2 cache topology will be overrided by the new topology setting.
Currently x-l2-cache-topo only defines the share level *globally*. I'm thinking how we can make the property more powerful so that it can specify which CPUs share l2 on core level and which CPUs share l2 on cluster level.
What would Intel's Hybrid CPUs do? Determine the l2 share level is core or cluster according to the CPU core type (Atom or Core)? While ARM does not have the core type concept but have CPUs that l2 is shared on different levels in the same system.
Thanks,
Yanan
Quoted 90 lines
Here we expose to user "cluster" instead of "module", to be consistent with "cluster-id" naming.
Since CPUID.04H is used by intel CPUs, this property is available on intel CPUs as for now.
When necessary, it can be extended to CPUID.8000001DH for amd CPUs.
Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 33 ++++++++++++++++++++++++++++++++- target/i386/cpu.h | 2 ++ 2 files changed, 34 insertions(+), 1 deletion(-)target/i386/cpu.c +32
-1@@ -240,12 +240,15 @@ static uint32_t max_processor_ids_for_cache(CPUCacheInfo *cache,case CORE : num_ids =1 <<apicid_core_offset (topo_info);break ; +case MODULE : + num_ids =1 <<apicid_module_offset (topo_info); +break ;case DIE : num_ids =1 <<apicid_die_offset (topo_info);break ;default :/* -* Currently there is no use case for SMT, MODULE and PACKAGE, so use +* Currently there is no use case for SMT and PACKAGE, so use * assert directly to facilitate debugging. */ g_assert_not_reached ();@@ -6633,6 +6636,33 @@ static void x86_cpu_realizefn(DeviceState *dev, Error **errp) env->cache_info_amd.l3_cache = &legacy_l3_cache; } +if (cpu->l2_cache_topo_level) { +/* +* FIXME: Currently only supports changing CPUID[4] (for intel), and +* will support changing CPUID[0x8000001D] when necessary. +*/ +if (!IS_INTEL_CPU (env)) { +error_setg (errp,"only intel cpus supports x-l2-cache-topo" ); +return ; + } + +if (!strcmp (cpu->l2_cache_topo_level,"core" )) { + env->cache_info_cpuid4.l2_cache->share_level =CORE ; + }else if (!strcmp (cpu->l2_cache_topo_level,"cluster" )) { +/* +* We expose to users "cluster" instead of "module", to be +* consistent with "cluster-id" naming. +*/ + env->cache_info_cpuid4.l2_cache->share_level =MODULE ; + }else { +error_setg (errp, +"x-l2-cache-topo doesn't support '%s', " +"and it only supports 'core' or 'cluster'" , + cpu->l2_cache_topo_level); +return ; + } + } +#ifndef CONFIG_USER_ONLY MachineState *ms =MACHINE (qdev_get_machine ());qemu_register_reset (x86_cpu_machine_reset_cb, cpu);@@ -7135,6 +7165,7 @@ static Property x86_cpu_properties[] = { false),DEFINE_PROP_BOOL ("x-intel-pt-auto-level" ,X86CPU , intel_pt_auto_level, true), +DEFINE_PROP_STRING ("x-l2-cache-topo" ,X86CPU , l2_cache_topo_level), DEFINE_PROP_END_OF_LIST() };target/i386/cpu.h +2
-0@@ -1987,6 +1987,8 @@ struct ArchCPU {int32_t thread_id;int32_t hv_max_vps; + +char *l2_cache_topo_level; };
Re: [PATCH RESEND 18/18] i386: Add new property to control L2 cache topo in CPUID.04H
Quoted 11 lines · wangyanan (Y)
Date: Thu, 16 Feb 2023 21:14:54 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 5 lines · Zhao Liu
From: Zhao Liu <[email protected]>
The property x-l2-cache-topo will be used to change the L2 cache topology in CPUID.04H.
Quoted 6 lines
Quoted 5 lines
Now it allows user to set the L2 cache is shared in core level or cluster level.
If user passes "-cpu x-l2-cache-topo=[core|cluster]" then older L2 cache topology will be overrided by the new topology setting.
Currently x-l2-cache-topo only defines the share level *globally*.
Yes, will set for all CPUs.
Quoted 8 lines
I'm thinking how we can make the property more powerful so that it can specify which CPUs share l2 on core level and which CPUs share l2 on cluster level.
What would Intel's Hybrid CPUs do? Determine the l2 share level is core or cluster according to the CPU core type (Atom or Core)? While ARM does not have the core type concept but have CPUs that l2 is shared on different levels in the same system.
For example, Alderlake's "core" shares 1 L2 per core and every 4 "atom"s share 1 L2. For this case, we can set the topology as:
cluster0 has 1 "core" and cluster1 has 4 "atom". Then set L2 shared on cluster level.
Since cluster0 has only 1 "core" type core, then L2 per "core" works.
Not sure if this idea can be applied to arm?
Quoted 89 lines
Thanks,
YananQuoted 85 lines
Here we expose to user "cluster" instead of "module", to be consistent with "cluster-id" naming.
Since CPUID.04H is used by intel CPUs, this property is available on intel CPUs as for now.
When necessary, it can be extended to CPUID.8000001DH for amd CPUs.
Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 33 ++++++++++++++++++++++++++++++++- target/i386/cpu.h | 2 ++ 2 files changed, 34 insertions(+), 1 deletion(-)target/i386/cpu.c +32
-1@@ -240,12 +240,15 @@ static uint32_t max_processor_ids_for_cache(CPUCacheInfo *cache,case CORE : num_ids =1 <<apicid_core_offset (topo_info);break ; +case MODULE : + num_ids =1 <<apicid_module_offset (topo_info); +break ;case DIE : num_ids =1 <<apicid_die_offset (topo_info);break ;default :/* -* Currently there is no use case for SMT, MODULE and PACKAGE, so use +* Currently there is no use case for SMT and PACKAGE, so use * assert directly to facilitate debugging. */ g_assert_not_reached ();@@ -6633,6 +6636,33 @@ static void x86_cpu_realizefn(DeviceState *dev, Error **errp) env->cache_info_amd.l3_cache = &legacy_l3_cache; } +if (cpu->l2_cache_topo_level) { +/* +* FIXME: Currently only supports changing CPUID[4] (for intel), and +* will support changing CPUID[0x8000001D] when necessary. +*/ +if (!IS_INTEL_CPU (env)) { +error_setg (errp,"only intel cpus supports x-l2-cache-topo" ); +return ; + } + +if (!strcmp (cpu->l2_cache_topo_level,"core" )) { + env->cache_info_cpuid4.l2_cache->share_level =CORE ; + }else if (!strcmp (cpu->l2_cache_topo_level,"cluster" )) { +/* +* We expose to users "cluster" instead of "module", to be +* consistent with "cluster-id" naming. +*/ + env->cache_info_cpuid4.l2_cache->share_level =MODULE ; + }else { +error_setg (errp, +"x-l2-cache-topo doesn't support '%s', " +"and it only supports 'core' or 'cluster'" , + cpu->l2_cache_topo_level); +return ; + } + } +#ifndef CONFIG_USER_ONLY MachineState *ms =MACHINE (qdev_get_machine ());qemu_register_reset (x86_cpu_machine_reset_cb, cpu);@@ -7135,6 +7165,7 @@ static Property x86_cpu_properties[] = { false),DEFINE_PROP_BOOL ("x-intel-pt-auto-level" ,X86CPU , intel_pt_auto_level, true), +DEFINE_PROP_STRING ("x-l2-cache-topo" ,X86CPU , l2_cache_topo_level), DEFINE_PROP_END_OF_LIST() };target/i386/cpu.h +2
-0@@ -1987,6 +1987,8 @@ struct ArchCPU {int32_t thread_id;int32_t hv_max_vps; + +char *l2_cache_topo_level; };
Re: [PATCH RESEND 18/18] i386: Add new property to control L2 cache topo in CPUID.04H
Quoted 28 lines · Zhao Liu
Quoted 11 lines · wangyanan (Y)
Date: Thu, 16 Feb 2023 21:14:54 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 5 lines · Zhao Liu
From: Zhao Liu <[email protected]>
The property x-l2-cache-topo will be used to change the L2 cache topology in CPUID.04H.
Quoted 6 lines
Quoted 5 lines
Now it allows user to set the L2 cache is shared in core level or cluster level.
If user passes "-cpu x-l2-cache-topo=[core|cluster]" then older L2 cache topology will be overrided by the new topology setting.
Currently x-l2-cache-topo only defines the share level *globally*.
Yes, will set for all CPUs.
Quoted 8 lines
I'm thinking how we can make the property more powerful so that it can specify which CPUs share l2 on core level and which CPUs share l2 on cluster level.
What would Intel's Hybrid CPUs do? Determine the l2 share level is core or cluster according to the CPU core type (Atom or Core)? While ARM does not have the core type concept but have CPUs that l2 is shared on different levels in the same system.
Quoted 7 lines
For example, Alderlake's "core" shares 1 L2 per core and every 4 "atom"s share 1 L2. For this case, we can set the topology as:
cluster0 has 1 "core" and cluster1 has 4 "atom". Then set L2 shared on cluster level.
Since cluster0 has only 1 "core" type core, then L2 per "core" works.
This brings restriction to the users that cluster0 must have *1* core-type core. What if we set 2 vCores in cluster0 and 4 vCores in cluster1, and bind cores in cluster0 to 2 core-type pCores and bind cores in cluster1 to 4 atom-type pCores?I think this is a necessary use case too.
Quoted 89 lines
Not sure if this idea can be applied to arm?
Quoted 87 lines
Thanks,
YananQuoted 85 lines
Here we expose to user "cluster" instead of "module", to be consistent with "cluster-id" naming.
Since CPUID.04H is used by intel CPUs, this property is available on intel CPUs as for now.
When necessary, it can be extended to CPUID.8000001DH for amd CPUs.
Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 33 ++++++++++++++++++++++++++++++++- target/i386/cpu.h | 2 ++ 2 files changed, 34 insertions(+), 1 deletion(-)target/i386/cpu.c +32
-1@@ -240,12 +240,15 @@ static uint32_t max_processor_ids_for_cache(CPUCacheInfo *cache,case CORE : num_ids =1 <<apicid_core_offset (topo_info);break ; +case MODULE : + num_ids =1 <<apicid_module_offset (topo_info); +break ;case DIE : num_ids =1 <<apicid_die_offset (topo_info);break ;default :/* -* Currently there is no use case for SMT, MODULE and PACKAGE, so use +* Currently there is no use case for SMT and PACKAGE, so use * assert directly to facilitate debugging. */ g_assert_not_reached ();@@ -6633,6 +6636,33 @@ static void x86_cpu_realizefn(DeviceState *dev, Error **errp) env->cache_info_amd.l3_cache = &legacy_l3_cache; } +if (cpu->l2_cache_topo_level) { +/* +* FIXME: Currently only supports changing CPUID[4] (for intel), and +* will support changing CPUID[0x8000001D] when necessary. +*/ +if (!IS_INTEL_CPU (env)) { +error_setg (errp,"only intel cpus supports x-l2-cache-topo" ); +return ; + } + +if (!strcmp (cpu->l2_cache_topo_level,"core" )) { + env->cache_info_cpuid4.l2_cache->share_level =CORE ; + }else if (!strcmp (cpu->l2_cache_topo_level,"cluster" )) { +/* +* We expose to users "cluster" instead of "module", to be +* consistent with "cluster-id" naming. +*/ + env->cache_info_cpuid4.l2_cache->share_level =MODULE ; + }else { +error_setg (errp, +"x-l2-cache-topo doesn't support '%s', " +"and it only supports 'core' or 'cluster'" , + cpu->l2_cache_topo_level); +return ; + } + } +#ifndef CONFIG_USER_ONLY MachineState *ms =MACHINE (qdev_get_machine ());qemu_register_reset (x86_cpu_machine_reset_cb, cpu);@@ -7135,6 +7165,7 @@ static Property x86_cpu_properties[] = { false),DEFINE_PROP_BOOL ("x-intel-pt-auto-level" ,X86CPU , intel_pt_auto_level, true), +DEFINE_PROP_STRING ("x-l2-cache-topo" ,X86CPU , l2_cache_topo_level), DEFINE_PROP_END_OF_LIST() };target/i386/cpu.h +2
-0@@ -1987,6 +1987,8 @@ struct ArchCPU {int32_t thread_id;int32_t hv_max_vps; + +char *l2_cache_topo_level; };
Re: [PATCH RESEND 18/18] i386: Add new property to control L2 cache topo in CPUID.04H
Quoted 40 lines · wangyanan (Y)
Date: Fri, 17 Feb 2023 11:45:58 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 34 lines · Zhao Liu
Quoted 11 lines · wangyanan (Y)
Date: Thu, 16 Feb 2023 21:14:54 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 5 lines · Zhao Liu
From: Zhao Liu <[email protected]>
The property x-l2-cache-topo will be used to change the L2 cache topology in CPUID.04H.
Quoted 6 lines
Quoted 5 lines
Now it allows user to set the L2 cache is shared in core level or cluster level.
If user passes "-cpu x-l2-cache-topo=[core|cluster]" then older L2 cache topology will be overrided by the new topology setting.
Currently x-l2-cache-topo only defines the share level *globally*.
Yes, will set for all CPUs.
Quoted 8 lines
I'm thinking how we can make the property more powerful so that it can specify which CPUs share l2 on core level and which CPUs share l2 on cluster level.
What would Intel's Hybrid CPUs do? Determine the l2 share level is core or cluster according to the CPU core type (Atom or Core)? While ARM does not have the core type concept but have CPUs that l2 is shared on different levels in the same system.
For example, Alderlake's "core" shares 1 L2 per core and every 4 "atom"s share 1 L2. For this case, we can set the topology as:
cluster0 has 1 "core" and cluster1 has 4 "atom". Then set L2 shared on cluster level.
Quoted 7 lines
Quoted 1 line
Since cluster0 has only 1 "core" type core, then L2 per "core" works.
This brings restriction to the users that cluster0 must have *1* core-type core.
What if we set 2 vCores in cluster0 and 4 vCores in cluster1, and bind cores in
cluster0 to 2 core-type pCores and bind cores in cluster1 to 4 atom-type pCores?I think this is a necessary use case too.
At present, the cache topology level and core type are not bound, so the cache topology level can also be adjusted for any vCores.
Quoted 90 lines
Quoted 89 lines
Not sure if this idea can be applied to arm?
Quoted 87 lines
Thanks,
YananQuoted 85 lines
Here we expose to user "cluster" instead of "module", to be consistent with "cluster-id" naming.
Since CPUID.04H is used by intel CPUs, this property is available on intel CPUs as for now.
When necessary, it can be extended to CPUID.8000001DH for amd CPUs.
Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 33 ++++++++++++++++++++++++++++++++- target/i386/cpu.h | 2 ++ 2 files changed, 34 insertions(+), 1 deletion(-)target/i386/cpu.c +32
-1@@ -240,12 +240,15 @@ static uint32_t max_processor_ids_for_cache(CPUCacheInfo *cache,case CORE : num_ids =1 <<apicid_core_offset (topo_info);break ; +case MODULE : + num_ids =1 <<apicid_module_offset (topo_info); +break ;case DIE : num_ids =1 <<apicid_die_offset (topo_info);break ;default :/* -* Currently there is no use case for SMT, MODULE and PACKAGE, so use +* Currently there is no use case for SMT and PACKAGE, so use * assert directly to facilitate debugging. */ g_assert_not_reached ();@@ -6633,6 +6636,33 @@ static void x86_cpu_realizefn(DeviceState *dev, Error **errp) env->cache_info_amd.l3_cache = &legacy_l3_cache; } +if (cpu->l2_cache_topo_level) { +/* +* FIXME: Currently only supports changing CPUID[4] (for intel), and +* will support changing CPUID[0x8000001D] when necessary. +*/ +if (!IS_INTEL_CPU (env)) { +error_setg (errp,"only intel cpus supports x-l2-cache-topo" ); +return ; + } + +if (!strcmp (cpu->l2_cache_topo_level,"core" )) { + env->cache_info_cpuid4.l2_cache->share_level =CORE ; + }else if (!strcmp (cpu->l2_cache_topo_level,"cluster" )) { +/* +* We expose to users "cluster" instead of "module", to be +* consistent with "cluster-id" naming. +*/ + env->cache_info_cpuid4.l2_cache->share_level =MODULE ; + }else { +error_setg (errp, +"x-l2-cache-topo doesn't support '%s', " +"and it only supports 'core' or 'cluster'" , + cpu->l2_cache_topo_level); +return ; + } + } +#ifndef CONFIG_USER_ONLY MachineState *ms =MACHINE (qdev_get_machine ());qemu_register_reset (x86_cpu_machine_reset_cb, cpu);@@ -7135,6 +7165,7 @@ static Property x86_cpu_properties[] = { false),DEFINE_PROP_BOOL ("x-intel-pt-auto-level" ,X86CPU , intel_pt_auto_level, true), +DEFINE_PROP_STRING ("x-l2-cache-topo" ,X86CPU , l2_cache_topo_level), DEFINE_PROP_END_OF_LIST() };target/i386/cpu.h +2
-0@@ -1987,6 +1987,8 @@ struct ArchCPU {int32_t thread_id;int32_t hv_max_vps; + +char *l2_cache_topo_level; };
Re: [PATCH RESEND 18/18] i386: Add new property to control L2 cache topo in CPUID.04H
In reply to Zhao LiuQuoted 31 lines · Zhao Liu
Quoted 11 lines · wangyanan (Y)
Date: Thu, 16 Feb 2023 21:14:54 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 5 lines · Zhao Liu
From: Zhao Liu <[email protected]>
The property x-l2-cache-topo will be used to change the L2 cache topology in CPUID.04H.
Quoted 6 lines
Quoted 5 lines
Now it allows user to set the L2 cache is shared in core level or cluster level.
If user passes "-cpu x-l2-cache-topo=[core|cluster]" then older L2 cache topology will be overrided by the new topology setting.
Currently x-l2-cache-topo only defines the share level *globally*.
Yes, will set for all CPUs.
Quoted 8 lines
I'm thinking how we can make the property more powerful so that it can specify which CPUs share l2 on core level and which CPUs share l2 on cluster level.
What would Intel's Hybrid CPUs do? Determine the l2 share level is core or cluster according to the CPU core type (Atom or Core)? While ARM does not have the core type concept but have CPUs that l2 is shared on different levels in the same system.
For example, Alderlake's "core" shares 1 L2 per core and every 4 "atom"s share 1 L2. For this case, we can set the topology as:
Quoted 6 lines
cluster0 has 1 "core" and cluster1 has 4 "atom". Then set L2 shared on cluster level.
Since cluster0 has only 1 "core" type core, then L2 per "core" works.
Not sure if this idea can be applied to arm?
For a CPU topopoly where we have 2 clusters totally, 2 cores in cluster0 have their own L1/L2 cache and 2 threads in each core, 4 cores in cluster1 share one L2 cache and 1 thread in each core. The global way does not work well.
What about defining something general, which looks like -numa config:
-cache-topo cache=l2, share_level="core", cpus='0-3' -cache-topo cache=l2, share_level="cluster", cpus='4-7'
If we ever want to support custom share-level for L3/L1, no extra work is needed. We can also extend the CLI to support custom cache size, etc..
If you thinks this a good idea to explore, I can work on it, since I'm planing to add support cache topology for ARM.
Thanks,
Yanan
Quoted 87 lines
Quoted 87 lines
Thanks,
YananQuoted 85 lines
Here we expose to user "cluster" instead of "module", to be consistent with "cluster-id" naming.
Since CPUID.04H is used by intel CPUs, this property is available on intel CPUs as for now.
When necessary, it can be extended to CPUID.8000001DH for amd CPUs.
Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 33 ++++++++++++++++++++++++++++++++- target/i386/cpu.h | 2 ++ 2 files changed, 34 insertions(+), 1 deletion(-)target/i386/cpu.c +32
-1@@ -240,12 +240,15 @@ static uint32_t max_processor_ids_for_cache(CPUCacheInfo *cache,case CORE : num_ids =1 <<apicid_core_offset (topo_info);break ; +case MODULE : + num_ids =1 <<apicid_module_offset (topo_info); +break ;case DIE : num_ids =1 <<apicid_die_offset (topo_info);break ;default :/* -* Currently there is no use case for SMT, MODULE and PACKAGE, so use +* Currently there is no use case for SMT and PACKAGE, so use * assert directly to facilitate debugging. */ g_assert_not_reached ();@@ -6633,6 +6636,33 @@ static void x86_cpu_realizefn(DeviceState *dev, Error **errp) env->cache_info_amd.l3_cache = &legacy_l3_cache; } +if (cpu->l2_cache_topo_level) { +/* +* FIXME: Currently only supports changing CPUID[4] (for intel), and +* will support changing CPUID[0x8000001D] when necessary. +*/ +if (!IS_INTEL_CPU (env)) { +error_setg (errp,"only intel cpus supports x-l2-cache-topo" ); +return ; + } + +if (!strcmp (cpu->l2_cache_topo_level,"core" )) { + env->cache_info_cpuid4.l2_cache->share_level =CORE ; + }else if (!strcmp (cpu->l2_cache_topo_level,"cluster" )) { +/* +* We expose to users "cluster" instead of "module", to be +* consistent with "cluster-id" naming. +*/ + env->cache_info_cpuid4.l2_cache->share_level =MODULE ; + }else { +error_setg (errp, +"x-l2-cache-topo doesn't support '%s', " +"and it only supports 'core' or 'cluster'" , + cpu->l2_cache_topo_level); +return ; + } + } +#ifndef CONFIG_USER_ONLY MachineState *ms =MACHINE (qdev_get_machine ());qemu_register_reset (x86_cpu_machine_reset_cb, cpu);@@ -7135,6 +7165,7 @@ static Property x86_cpu_properties[] = { false),DEFINE_PROP_BOOL ("x-intel-pt-auto-level" ,X86CPU , intel_pt_auto_level, true), +DEFINE_PROP_STRING ("x-l2-cache-topo" ,X86CPU , l2_cache_topo_level), DEFINE_PROP_END_OF_LIST() };target/i386/cpu.h +2
-0@@ -1987,6 +1987,8 @@ struct ArchCPU {int32_t thread_id;int32_t hv_max_vps; + +char *l2_cache_topo_level; };
Re: [PATCH RESEND 18/18] i386: Add new property to control L2 cache topo in CPUID.04H
Quoted 49 lines · wangyanan (Y)
Date: Fri, 17 Feb 2023 12:07:01 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 31 lines · Zhao Liu
Quoted 11 lines · wangyanan (Y)
Date: Thu, 16 Feb 2023 21:14:54 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 5 lines · Zhao Liu
From: Zhao Liu <[email protected]>
The property x-l2-cache-topo will be used to change the L2 cache topology in CPUID.04H.
Quoted 6 lines
Quoted 5 lines
Now it allows user to set the L2 cache is shared in core level or cluster level.
If user passes "-cpu x-l2-cache-topo=[core|cluster]" then older L2 cache topology will be overrided by the new topology setting.
Currently x-l2-cache-topo only defines the share level *globally*.
Yes, will set for all CPUs.
Quoted 8 lines
I'm thinking how we can make the property more powerful so that it can specify which CPUs share l2 on core level and which CPUs share l2 on cluster level.
What would Intel's Hybrid CPUs do? Determine the l2 share level is core or cluster according to the CPU core type (Atom or Core)? While ARM does not have the core type concept but have CPUs that l2 is shared on different levels in the same system.
For example, Alderlake's "core" shares 1 L2 per core and every 4 "atom"s share 1 L2. For this case, we can set the topology as:
Quoted 6 lines
cluster0 has 1 "core" and cluster1 has 4 "atom". Then set L2 shared on cluster level.
Since cluster0 has only 1 "core" type core, then L2 per "core" works.
Not sure if this idea can be applied to arm?
For a CPU topopoly where we have 2 clusters totally, 2 cores in cluster0 have their own L1/L2 cache and 2 threads in each core, 4 cores in cluster1 share one L2 cache and 1 thread in each core. The global way does not work well.
What about defining something general, which looks like -numa config:
Review
Quoted 2 lines
(quoted diff) +0
-2-cache-topo cache=l2, share_level="core", cpus='0-3' -cache-topo cache=l2, share_level="cluster", cpus='4-7'
Hi Yanan, here it may be necessary to check whether the cpu index set in "cpus" is reasonable through the specific cpu topology.
For example, core0 has 2 CPUs: cpu0 and cpu1, and core1 has 2 CPUs: cpu2 and cpu3, then set l2 as:
-cache-topo cache=l2, share_level="core", cpus='0-2' -cache-topo cache=l2, share_level="core", cpus='3'
Whether this command is legal depends on the meaning we give to the parameter "cpu":
1. If "cpu" means all cpus share the cache set in this command, then this command should fail since cpu2 and cpu3 are in a core.
2. If "cpu" means the affected cpus, then this command should find the cores they belong to according to the cpu topology, and set L2 for those cores. This command may return success.
What about removing share_level and ask "cpu" to mean all the sharing cpus to avoid checking the cpu topology?
Then the above example should be:
-cache-topo cache=l2, cpus='0-1'
-cache-topo cache=l2, cpus='2-3'
This decouples cpu topology and cache topology completely and very simple. In this way, determining the cache by specifying the shared cpu is similar to that in x86 CPUID.04H.
But the price of simplicity is we may build a cache topology that doesn't match the reality.
But if the cache topology must be set based on the cpu topology, another way is consider specifying the cache when setting the topology structure, which can be based on @Daniel's format [1]:
-object cpu-socket,id=sock0,cache=l3
-object cpu-die,id=die0,parent=sock0
-object cpu-cluster,id=cluster0,parent=die0
-object cpu-cluster,id=cluster1,parent=die0,cache=l2
-object x86-cpu-model-core,id=cpu0,parent=cluster0,threads=2,cache=l1i,lid,l2 -object x86-cpu-model-atom,id=cpu1,parent=cluster1,cache=l1i,lid -object x86-cpu-model-atom,id=cpu2,parent=cluster1,cache=l1i,l1d
Then from this command, cpu0 has a l2, and cpu1 and cpu2 shares a l2 (the l2 is inserted in cluster1).
This whole process is like when designing or building a CPU, the user decides where to insert the caches. The advantage is that it is easier to verify the rationality and is intuitive. But complicated.
(Also CC @Daniel for comments).
[1]: https://lists.gnu.org/archive/html/qemu-devel/2023-02/msg03320.html
Thanks,
Zhao
Quoted 96 lines
If we ever want to support custom share-level for L3/L1, no extra work is needed. We can also extend the CLI to support custom cache size, etc..
If you thinks this a good idea to explore, I can work on it, since I'm planing to add support cache topology for ARM.
Thanks,
YananQuoted 87 lines
Quoted 87 lines
Thanks,
YananQuoted 85 lines
Here we expose to user "cluster" instead of "module", to be consistent with "cluster-id" naming.
Since CPUID.04H is used by intel CPUs, this property is available on intel CPUs as for now.
When necessary, it can be extended to CPUID.8000001DH for amd CPUs.
Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 33 ++++++++++++++++++++++++++++++++- target/i386/cpu.h | 2 ++ 2 files changed, 34 insertions(+), 1 deletion(-)target/i386/cpu.c +32
-1@@ -240,12 +240,15 @@ static uint32_t max_processor_ids_for_cache(CPUCacheInfo *cache,case CORE : num_ids =1 <<apicid_core_offset (topo_info);break ; +case MODULE : + num_ids =1 <<apicid_module_offset (topo_info); +break ;case DIE : num_ids =1 <<apicid_die_offset (topo_info);break ;default :/* -* Currently there is no use case for SMT, MODULE and PACKAGE, so use +* Currently there is no use case for SMT and PACKAGE, so use * assert directly to facilitate debugging. */ g_assert_not_reached ();@@ -6633,6 +6636,33 @@ static void x86_cpu_realizefn(DeviceState *dev, Error **errp) env->cache_info_amd.l3_cache = &legacy_l3_cache; } +if (cpu->l2_cache_topo_level) { +/* +* FIXME: Currently only supports changing CPUID[4] (for intel), and +* will support changing CPUID[0x8000001D] when necessary. +*/ +if (!IS_INTEL_CPU (env)) { +error_setg (errp,"only intel cpus supports x-l2-cache-topo" ); +return ; + } + +if (!strcmp (cpu->l2_cache_topo_level,"core" )) { + env->cache_info_cpuid4.l2_cache->share_level =CORE ; + }else if (!strcmp (cpu->l2_cache_topo_level,"cluster" )) { +/* +* We expose to users "cluster" instead of "module", to be +* consistent with "cluster-id" naming. +*/ + env->cache_info_cpuid4.l2_cache->share_level =MODULE ; + }else { +error_setg (errp, +"x-l2-cache-topo doesn't support '%s', " +"and it only supports 'core' or 'cluster'" , + cpu->l2_cache_topo_level); +return ; + } + } +#ifndef CONFIG_USER_ONLY MachineState *ms =MACHINE (qdev_get_machine ());qemu_register_reset (x86_cpu_machine_reset_cb, cpu);@@ -7135,6 +7165,7 @@ static Property x86_cpu_properties[] = { false),DEFINE_PROP_BOOL ("x-intel-pt-auto-level" ,X86CPU , intel_pt_auto_level, true), +DEFINE_PROP_STRING ("x-l2-cache-topo" ,X86CPU , l2_cache_topo_level), DEFINE_PROP_END_OF_LIST() };target/i386/cpu.h +2
-0@@ -1987,6 +1987,8 @@ struct ArchCPU {int32_t thread_id;int32_t hv_max_vps; + +char *l2_cache_topo_level; };
Re: [PATCH RESEND 18/18] i386: Add new property to control L2 cache topo in CPUID.04H
Quoted 49 lines · Zhao Liu
Quoted 48 lines · wangyanan (Y)
Date: Fri, 17 Feb 2023 12:07:01 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 31 lines · Zhao Liu
Quoted 11 lines · wangyanan (Y)
Date: Thu, 16 Feb 2023 21:14:54 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 5 lines · Zhao Liu
From: Zhao Liu <[email protected]>
The property x-l2-cache-topo will be used to change the L2 cache topology in CPUID.04H.
Quoted 6 lines
Quoted 5 lines
Now it allows user to set the L2 cache is shared in core level or cluster level.
If user passes "-cpu x-l2-cache-topo=[core|cluster]" then older L2 cache topology will be overrided by the new topology setting.
Currently x-l2-cache-topo only defines the share level *globally*.
Yes, will set for all CPUs.
Quoted 8 lines
I'm thinking how we can make the property more powerful so that it can specify which CPUs share l2 on core level and which CPUs share l2 on cluster level.
What would Intel's Hybrid CPUs do? Determine the l2 share level is core or cluster according to the CPU core type (Atom or Core)? While ARM does not have the core type concept but have CPUs that l2 is shared on different levels in the same system.
For example, Alderlake's "core" shares 1 L2 per core and every 4 "atom"s share 1 L2. For this case, we can set the topology as:
Quoted 6 lines
cluster0 has 1 "core" and cluster1 has 4 "atom". Then set L2 shared on cluster level.
Since cluster0 has only 1 "core" type core, then L2 per "core" works.
Not sure if this idea can be applied to arm?
For a CPU topopoly where we have 2 clusters totally, 2 cores in cluster0 have their own L1/L2 cache and 2 threads in each core, 4 cores in cluster1 share one L2 cache and 1 thread in each core. The global way does not work well.
Quoted 5 lines
Review
Quoted 3 lines
What about defining something general, which looks like -numa config:
(quoted diff) +0
-2-cache-topo cache=l2, share_level="core", cpus='0-3' -cache-topo cache=l2, share_level="cluster", cpus='4-7'Hi Yanan, here it may be necessary to check whether the cpu index set in "cpus" is reasonable through the specific cpu topology.
Yes, the validity of the cache topo configs from the users should be check in machine_parse_cache_topo ( if we will have this func). It's not a big deal, we always need the validity checks.
In summary:
1、There can not be the same cpus in different "cpus" list. 2、A combination of all the "cpus" list should *just* cover all the CPUs in the machine
3、Most importantly, cpus in the same cluster must be set with the same cache "share_level" (core or cluster) and cpus in the same core must also be set with the same cache "share_level".
Quoted 8 lines
For example, core0 has 2 CPUs: cpu0 and cpu1, and core1 has 2 CPUs: cpu2 and cpu3, then set l2 as:
(quoted diff) +0
-2-cache-topo cache=l2, share_level="core", cpus='0-2' -cache-topo cache=l2, share_level="core", cpus='3'Whether this command is legal depends on the meaning we give to the parameter "cpu":
s/cpus/cpu.
It means all the afftected CPUs, e.g, the second case.
Quoted 12 lines
1. If "cpu" means all cpus share the cache set in this command, then this command should fail since cpu2 and cpu3 are in a core.
2. If "cpu" means the affected cpus, then this command should find the cores they belong to according to the cpu topology, and set L2 for those cores. This command may return success.
What about removing
share_leveland ask "cpu" to mean all the sharing cpus to avoid checking the cpu topology?Then the above example should be:
Review
Quoted 2 lines
(quoted diff) +0
-2-cache-topo cache=l2, cpus='0-1' -cache-topo cache=l2, cpus='2-3'
Sorry, I dont understand how we will know the cache share_level of cpus '0-1' or '2-3'. What will the CLIs will be like if we change the belows CLIs by removing the "share_level" params.
-cache-topo cache=l2, share_level="core", cpus='0-3' -cache-topo cache=l2, share_level="cluster", cpus='4-7'
Quoted 19 lines
This decouples cpu topology and cache topology completely and very simple. In this way, determining the cache by specifying the shared cpu is similar to that in x86 CPUID.04H.
But the price of simplicity is we may build a cache topology that doesn't match the reality.
But if the cache topology must be set based on the cpu topology, another way is consider specifying the cache when setting the topology structure, which can be based on @Daniel's format [1]:
-object cpu-socket,id=sock0,cache=l3 -object cpu-die,id=die0,parent=sock0 -object cpu-cluster,id=cluster0,parent=die0 -object cpu-cluster,id=cluster1,parent=die0,cache=l2 -object x86-cpu-model-core,id=cpu0,parent=cluster0,threads=2,cache=l1i,lid,l2 -object x86-cpu-model-atom,id=cpu1,parent=cluster1,cache=l1i,lid -object x86-cpu-model-atom,id=cpu2,parent=cluster1,cache=l1i,l1d
Quoted 6 lines
Then from this command, cpu0 has a l2, and cpu1 and cpu2 shares a l2 (the l2 is inserted in cluster1).
This whole process is like when designing or building a CPU, the user decides where to insert the caches. The advantage is that it is easier to verify the rationality and is intuitive. But complicated.
Yeah, this is also a way.
Most of the concern is that it will not be easy/readable to extand the cache configs, for example if we ever want to support custom cache size, cache type or other cache properties in the future. And yes, will also complex the -objects by making them huge.
I think keeping cache and cpu configs decouped will leave simplicity to the users, just like we keep numa resources config from cpu topology currently.
On the other hand, -cache-topo is not just for hybrid, it's also for current smp, which make it inappropriate to bind -cache-topo
with hybrid case. For example, "-cache-topo name=l2, share_level=cluster" will indicates that l2 cache is shared on cluster level globally. And this is "x-l2-cache-topo" in this patch does.
Thanks,
Yanan
Quoted 102 lines
(Also CC @Daniel for comments).
[1]: https://lists.gnu.org/archive/html/qemu-devel/2023-02/msg03320.html
Thanks,
ZhaoQuoted 95 lines
If we ever want to support custom share-level for L3/L1, no extra work is needed. We can also extend the CLI to support custom cache size, etc..
If you thinks this a good idea to explore, I can work on it, since I'm planing to add support cache topology for ARM.
Thanks,
YananQuoted 87 lines
Quoted 87 lines
Thanks,
YananQuoted 85 lines
Here we expose to user "cluster" instead of "module", to be consistent with "cluster-id" naming.
Since CPUID.04H is used by intel CPUs, this property is available on intel CPUs as for now.
When necessary, it can be extended to CPUID.8000001DH for amd CPUs.
Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 33 ++++++++++++++++++++++++++++++++- target/i386/cpu.h | 2 ++ 2 files changed, 34 insertions(+), 1 deletion(-)target/i386/cpu.c +32
-1@@ -240,12 +240,15 @@ static uint32_t max_processor_ids_for_cache(CPUCacheInfo *cache,case CORE : num_ids =1 <<apicid_core_offset (topo_info);break ; +case MODULE : + num_ids =1 <<apicid_module_offset (topo_info); +break ;case DIE : num_ids =1 <<apicid_die_offset (topo_info);break ;default :/* -* Currently there is no use case for SMT, MODULE and PACKAGE, so use +* Currently there is no use case for SMT and PACKAGE, so use * assert directly to facilitate debugging. */ g_assert_not_reached ();@@ -6633,6 +6636,33 @@ static void x86_cpu_realizefn(DeviceState *dev, Error **errp) env->cache_info_amd.l3_cache = &legacy_l3_cache; } +if (cpu->l2_cache_topo_level) { +/* +* FIXME: Currently only supports changing CPUID[4] (for intel), and +* will support changing CPUID[0x8000001D] when necessary. +*/ +if (!IS_INTEL_CPU (env)) { +error_setg (errp,"only intel cpus supports x-l2-cache-topo" ); +return ; + } + +if (!strcmp (cpu->l2_cache_topo_level,"core" )) { + env->cache_info_cpuid4.l2_cache->share_level =CORE ; + }else if (!strcmp (cpu->l2_cache_topo_level,"cluster" )) { +/* +* We expose to users "cluster" instead of "module", to be +* consistent with "cluster-id" naming. +*/ + env->cache_info_cpuid4.l2_cache->share_level =MODULE ; + }else { +error_setg (errp, +"x-l2-cache-topo doesn't support '%s', " +"and it only supports 'core' or 'cluster'" , + cpu->l2_cache_topo_level); +return ; + } + } +#ifndef CONFIG_USER_ONLY MachineState *ms =MACHINE (qdev_get_machine ());qemu_register_reset (x86_cpu_machine_reset_cb, cpu);@@ -7135,6 +7165,7 @@ static Property x86_cpu_properties[] = { false),DEFINE_PROP_BOOL ("x-intel-pt-auto-level" ,X86CPU , intel_pt_auto_level, true), +DEFINE_PROP_STRING ("x-l2-cache-topo" ,X86CPU , l2_cache_topo_level), DEFINE_PROP_END_OF_LIST() };target/i386/cpu.h +2
-0@@ -1987,6 +1987,8 @@ struct ArchCPU {int32_t thread_id;int32_t hv_max_vps; + +char *l2_cache_topo_level; };
Re: [PATCH RESEND 18/18] i386: Add new property to control L2 cache topo in CPUID.04H
Quoted 55 lines · wangyanan (Y)
Date: Fri, 17 Feb 2023 17:08:31 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 49 lines · Zhao Liu
Quoted 48 lines · wangyanan (Y)
Date: Fri, 17 Feb 2023 12:07:01 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 31 lines · Zhao Liu
Quoted 11 lines · wangyanan (Y)
Date: Thu, 16 Feb 2023 21:14:54 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 5 lines · Zhao Liu
From: Zhao Liu <[email protected]>
The property x-l2-cache-topo will be used to change the L2 cache topology in CPUID.04H.
Quoted 6 lines
Quoted 5 lines
Now it allows user to set the L2 cache is shared in core level or cluster level.
If user passes "-cpu x-l2-cache-topo=[core|cluster]" then older L2 cache topology will be overrided by the new topology setting.
Currently x-l2-cache-topo only defines the share level *globally*.
Yes, will set for all CPUs.
Quoted 8 lines
I'm thinking how we can make the property more powerful so that it can specify which CPUs share l2 on core level and which CPUs share l2 on cluster level.
What would Intel's Hybrid CPUs do? Determine the l2 share level is core or cluster according to the CPU core type (Atom or Core)? While ARM does not have the core type concept but have CPUs that l2 is shared on different levels in the same system.
For example, Alderlake's "core" shares 1 L2 per core and every 4 "atom"s share 1 L2. For this case, we can set the topology as:
Quoted 6 lines
cluster0 has 1 "core" and cluster1 has 4 "atom". Then set L2 shared on cluster level.
Since cluster0 has only 1 "core" type core, then L2 per "core" works.
Not sure if this idea can be applied to arm?
For a CPU topopoly where we have 2 clusters totally, 2 cores in cluster0 have their own L1/L2 cache and 2 threads in each core, 4 cores in cluster1 share one L2 cache and 1 thread in each core. The global way does not work well.
Quoted 8 lines
Quoted 5 lines
Review
Quoted 3 lines
What about defining something general, which looks like -numa config:
(quoted diff) +0
-2-cache-topo cache=l2, share_level="core", cpus='0-3' -cache-topo cache=l2, share_level="cluster", cpus='4-7'Hi Yanan, here it may be necessary to check whether the cpu index set in "cpus" is reasonable through the specific cpu topology.
Yes, the validity of the cache topo configs from the users should be check in
machine_parse_cache_topo( if we will have this func). It's not a big deal, we always need the validity checks.
I guess that verification needs to build up the full cpu topology, as done in another hybrid RFC. So, should the cpu-topology.h related patches in that RFC be split out and sent first?
Quoted 8 lines
In summary:
1、There can not be the same cpus in different "cpus" list. 2、A combination of all the "cpus" list should *just* cover all the CPUs in the machine
3、Most importantly, cpus in the same cluster must be set with the same cache "share_level" (core or cluster) and cpus in the same core must also be set with the same cache "share_level".
Got it, thx.
Quoted 177 lines
Quoted 8 lines
For example, core0 has 2 CPUs: cpu0 and cpu1, and core1 has 2 CPUs: cpu2 and cpu3, then set l2 as:
(quoted diff) +0
-2-cache-topo cache=l2, share_level="core", cpus='0-2' -cache-topo cache=l2, share_level="core", cpus='3'Whether this command is legal depends on the meaning we give to the parameter "cpu":
s/cpus/cpu.
It means all the afftected CPUs, e.g, the second case.Quoted 12 lines
1. If "cpu" means all cpus share the cache set in this command, then this command should fail since cpu2 and cpu3 are in a core.
2. If "cpu" means the affected cpus, then this command should find the cores they belong to according to the cpu topology, and set L2 for those cores. This command may return success.
What about removing
share_leveland ask "cpu" to mean all the sharing cpus to avoid checking the cpu topology?Then the above example should be:
Review
Quoted 2 lines
(quoted diff) +0
-2-cache-topo cache=l2, cpus='0-1' -cache-topo cache=l2, cpus='2-3'Sorry, I dont understand how we will know the cache
share_levelof cpus '0-1' or '2-3'. What will the CLIs will be like if we change the belows CLIs by removing the "share_level" params.(quoted diff) +0
-2-cache-topo cache=l2, share_level="core", cpus='0-3' -cache-topo cache=l2, share_level="cluster", cpus='4-7'Quoted 19 lines
This decouples cpu topology and cache topology completely and very simple. In this way, determining the cache by specifying the shared cpu is similar to that in x86 CPUID.04H.
But the price of simplicity is we may build a cache topology that doesn't match the reality.
But if the cache topology must be set based on the cpu topology, another way is consider specifying the cache when setting the topology structure, which can be based on @Daniel's format [1]:
-object cpu-socket,id=sock0,cache=l3 -object cpu-die,id=die0,parent=sock0 -object cpu-cluster,id=cluster0,parent=die0 -object cpu-cluster,id=cluster1,parent=die0,cache=l2 -object x86-cpu-model-core,id=cpu0,parent=cluster0,threads=2,cache=l1i,lid,l2 -object x86-cpu-model-atom,id=cpu1,parent=cluster1,cache=l1i,lid -object x86-cpu-model-atom,id=cpu2,parent=cluster1,cache=l1i,l1dQuoted 6 lines
Then from this command, cpu0 has a l2, and cpu1 and cpu2 shares a l2 (the l2 is inserted in cluster1).
This whole process is like when designing or building a CPU, the user decides where to insert the caches. The advantage is that it is easier to verify the rationality and is intuitive. But complicated.
Yeah, this is also a way.
Most of the concern is that it will not be easy/readable to extand the cache configs, for example if we ever want to support custom cache size, cache type or other cache properties in the future. And yes, will also complex the -objects by making them huge.I think keeping cache and cpu configs decouped will leave simplicity to the users, just like we keep numa resources config from cpu topology currently.
On the other hand, -cache-topo is not just for hybrid, it's also for current smp, which make it inappropriate to bind -cache-topo with hybrid case. For example, "-cache-topo name=l2, share_level=cluster" will indicates that l2 cache is shared on cluster level globally. And this is
"x-l2-cache-topo" in this patch does.Thanks,
YananQuoted 102 lines
(Also CC @Daniel for comments).
[1]: https://lists.gnu.org/archive/html/qemu-devel/2023-02/msg03320.html
Thanks,
ZhaoQuoted 95 lines
If we ever want to support custom share-level for L3/L1, no extra work is needed. We can also extend the CLI to support custom cache size, etc..
If you thinks this a good idea to explore, I can work on it, since I'm planing to add support cache topology for ARM.
Thanks,
YananQuoted 87 lines
Quoted 87 lines
Thanks,
YananQuoted 85 lines
Here we expose to user "cluster" instead of "module", to be consistent with "cluster-id" naming.
Since CPUID.04H is used by intel CPUs, this property is available on intel CPUs as for now.
When necessary, it can be extended to CPUID.8000001DH for amd CPUs.
Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 33 ++++++++++++++++++++++++++++++++- target/i386/cpu.h | 2 ++ 2 files changed, 34 insertions(+), 1 deletion(-)target/i386/cpu.c +32
-1@@ -240,12 +240,15 @@ static uint32_t max_processor_ids_for_cache(CPUCacheInfo *cache,case CORE : num_ids =1 <<apicid_core_offset (topo_info);break ; +case MODULE : + num_ids =1 <<apicid_module_offset (topo_info); +break ;case DIE : num_ids =1 <<apicid_die_offset (topo_info);break ;default :/* -* Currently there is no use case for SMT, MODULE and PACKAGE, so use +* Currently there is no use case for SMT and PACKAGE, so use * assert directly to facilitate debugging. */ g_assert_not_reached ();@@ -6633,6 +6636,33 @@ static void x86_cpu_realizefn(DeviceState *dev, Error **errp) env->cache_info_amd.l3_cache = &legacy_l3_cache; } +if (cpu->l2_cache_topo_level) { +/* +* FIXME: Currently only supports changing CPUID[4] (for intel), and +* will support changing CPUID[0x8000001D] when necessary. +*/ +if (!IS_INTEL_CPU (env)) { +error_setg (errp,"only intel cpus supports x-l2-cache-topo" ); +return ; + } + +if (!strcmp (cpu->l2_cache_topo_level,"core" )) { + env->cache_info_cpuid4.l2_cache->share_level =CORE ; + }else if (!strcmp (cpu->l2_cache_topo_level,"cluster" )) { +/* +* We expose to users "cluster" instead of "module", to be +* consistent with "cluster-id" naming. +*/ + env->cache_info_cpuid4.l2_cache->share_level =MODULE ; + }else { +error_setg (errp, +"x-l2-cache-topo doesn't support '%s', " +"and it only supports 'core' or 'cluster'" , + cpu->l2_cache_topo_level); +return ; + } + } +#ifndef CONFIG_USER_ONLY MachineState *ms =MACHINE (qdev_get_machine ());qemu_register_reset (x86_cpu_machine_reset_cb, cpu);@@ -7135,6 +7165,7 @@ static Property x86_cpu_properties[] = { false),DEFINE_PROP_BOOL ("x-intel-pt-auto-level" ,X86CPU , intel_pt_auto_level, true), +DEFINE_PROP_STRING ("x-l2-cache-topo" ,X86CPU , l2_cache_topo_level), DEFINE_PROP_END_OF_LIST() };target/i386/cpu.h +2
-0@@ -1987,6 +1987,8 @@ struct ArchCPU {int32_t thread_id;int32_t hv_max_vps; + +char *l2_cache_topo_level; };
Re: [PATCH RESEND 18/18] i386: Add new property to control L2 cache topo in CPUID.04H
Quoted 61 lines · Zhao Liu
Quoted 60 lines · wangyanan (Y)
Date: Fri, 17 Feb 2023 17:08:31 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 54 lines · Zhao Liu
Quoted 49 lines · wangyanan (Y)
Date: Fri, 17 Feb 2023 12:07:01 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 31 lines · Zhao Liu
Quoted 11 lines · wangyanan (Y)
Date: Thu, 16 Feb 2023 21:14:54 +0800
From: "wangyanan (Y)" <[email protected]>
Subject: Re: [PATCH RESEND 18/18] i386: Add new property to control L2
cache topo in CPUID.04HQuoted 5 lines · Zhao Liu
From: Zhao Liu <[email protected]>
The property x-l2-cache-topo will be used to change the L2 cache topology in CPUID.04H.
Quoted 6 lines
Quoted 5 lines
Now it allows user to set the L2 cache is shared in core level or cluster level.
If user passes "-cpu x-l2-cache-topo=[core|cluster]" then older L2 cache topology will be overrided by the new topology setting.
Currently x-l2-cache-topo only defines the share level *globally*.
Yes, will set for all CPUs.
Quoted 8 lines
I'm thinking how we can make the property more powerful so that it can specify which CPUs share l2 on core level and which CPUs share l2 on cluster level.
What would Intel's Hybrid CPUs do? Determine the l2 share level is core or cluster according to the CPU core type (Atom or Core)? While ARM does not have the core type concept but have CPUs that l2 is shared on different levels in the same system.
For example, Alderlake's "core" shares 1 L2 per core and every 4 "atom"s share 1 L2. For this case, we can set the topology as:
Quoted 6 lines
cluster0 has 1 "core" and cluster1 has 4 "atom". Then set L2 shared on cluster level.
Since cluster0 has only 1 "core" type core, then L2 per "core" works.
Not sure if this idea can be applied to arm?
For a CPU topopoly where we have 2 clusters totally, 2 cores in cluster0 have their own L1/L2 cache and 2 threads in each core, 4 cores in cluster1 share one L2 cache and 1 thread in each core. The global way does not work well.
What about defining something general, which looks like -numa config:
Review
Quoted 2 lines
(quoted diff) +0
-2-cache-topo cache=l2, share_level="core", cpus='0-3' -cache-topo cache=l2, share_level="cluster", cpus='4-7'Hi Yanan, here it may be necessary to check whether the cpu index set in "cpus" is reasonable through the specific cpu topology.
Quoted 6 lines
Quoted 3 lines
Yes, the validity of the cache topo configs from the users should be check in
machine_parse_cache_topo( if we will have this func). It's not a big deal, we always need the validity checks.I guess that verification needs to build up the full cpu topology, as done in another hybrid RFC. So, should the
cpu-topology.hrelated patches in that RFC be split out and sent first?
Which patches, do you mean the re-work/generalization of smp? Will it be reasonable if they are sent separately without the hybrid introduction? I'm not yet sure if -cache must need the cpu-topology.h related things, maybe current upstream QEMU code is enough.
Quoted 185 lines
Quoted 7 lines
In summary:
1、There can not be the same cpus in different "cpus" list. 2、A combination of all the "cpus" list should *just* cover all the CPUs in the machine
3、Most importantly, cpus in the same cluster must be set with the same cache "share_level" (core or cluster) and cpus in the same core must also be set with the same cache "share_level".Got it, thx.
Quoted 176 lines
Quoted 8 lines
For example, core0 has 2 CPUs: cpu0 and cpu1, and core1 has 2 CPUs: cpu2 and cpu3, then set l2 as:
(quoted diff) +0
-2-cache-topo cache=l2, share_level="core", cpus='0-2' -cache-topo cache=l2, share_level="core", cpus='3'Whether this command is legal depends on the meaning we give to the parameter "cpu":
s/cpus/cpu.
It means all the afftected CPUs, e.g, the second case.Quoted 12 lines
1. If "cpu" means all cpus share the cache set in this command, then this command should fail since cpu2 and cpu3 are in a core.
2. If "cpu" means the affected cpus, then this command should find the cores they belong to according to the cpu topology, and set L2 for those cores. This command may return success.
What about removing
share_leveland ask "cpu" to mean all the sharing cpus to avoid checking the cpu topology?Then the above example should be:
Review
Quoted 2 lines
(quoted diff) +0
-2-cache-topo cache=l2, cpus='0-1' -cache-topo cache=l2, cpus='2-3'Sorry, I dont understand how we will know the cache
share_levelof cpus '0-1' or '2-3'. What will the CLIs will be like if we change the belows CLIs by removing the "share_level" params.(quoted diff) +0
-2-cache-topo cache=l2, share_level="core", cpus='0-3' -cache-topo cache=l2, share_level="cluster", cpus='4-7'Quoted 19 lines
This decouples cpu topology and cache topology completely and very simple. In this way, determining the cache by specifying the shared cpu is similar to that in x86 CPUID.04H.
But the price of simplicity is we may build a cache topology that doesn't match the reality.
But if the cache topology must be set based on the cpu topology, another way is consider specifying the cache when setting the topology structure, which can be based on @Daniel's format [1]:
-object cpu-socket,id=sock0,cache=l3 -object cpu-die,id=die0,parent=sock0 -object cpu-cluster,id=cluster0,parent=die0 -object cpu-cluster,id=cluster1,parent=die0,cache=l2 -object x86-cpu-model-core,id=cpu0,parent=cluster0,threads=2,cache=l1i,lid,l2 -object x86-cpu-model-atom,id=cpu1,parent=cluster1,cache=l1i,lid -object x86-cpu-model-atom,id=cpu2,parent=cluster1,cache=l1i,l1dQuoted 6 lines
Then from this command, cpu0 has a l2, and cpu1 and cpu2 shares a l2 (the l2 is inserted in cluster1).
This whole process is like when designing or building a CPU, the user decides where to insert the caches. The advantage is that it is easier to verify the rationality and is intuitive. But complicated.
Yeah, this is also a way.
Most of the concern is that it will not be easy/readable to extand the cache configs, for example if we ever want to support custom cache size, cache type or other cache properties in the future. And yes, will also complex the -objects by making them huge.I think keeping cache and cpu configs decouped will leave simplicity to the users, just like we keep numa resources config from cpu topology currently.
On the other hand, -cache-topo is not just for hybrid, it's also for current smp, which make it inappropriate to bind -cache-topo with hybrid case. For example, "-cache-topo name=l2, share_level=cluster" will indicates that l2 cache is shared on cluster level globally. And this is
"x-l2-cache-topo" in this patch does.Thanks,
YananQuoted 102 lines
(Also CC @Daniel for comments).
[1]: https://lists.gnu.org/archive/html/qemu-devel/2023-02/msg03320.html
Thanks,
ZhaoQuoted 95 lines
If we ever want to support custom share-level for L3/L1, no extra work is needed. We can also extend the CLI to support custom cache size, etc..
If you thinks this a good idea to explore, I can work on it, since I'm planing to add support cache topology for ARM.
Thanks,
YananQuoted 87 lines
Quoted 87 lines
Thanks,
YananQuoted 85 lines
Here we expose to user "cluster" instead of "module", to be consistent with "cluster-id" naming.
Since CPUID.04H is used by intel CPUs, this property is available on intel CPUs as for now.
When necessary, it can be extended to CPUID.8000001DH for amd CPUs.
Signed-off-by Zhao Liu <[email protected]>--- target/i386/cpu.c | 33 ++++++++++++++++++++++++++++++++- target/i386/cpu.h | 2 ++ 2 files changed, 34 insertions(+), 1 deletion(-)target/i386/cpu.c +32
-1@@ -240,12 +240,15 @@ static uint32_t max_processor_ids_for_cache(CPUCacheInfo *cache,case CORE : num_ids =1 <<apicid_core_offset (topo_info);break ; +case MODULE : + num_ids =1 <<apicid_module_offset (topo_info); +break ;case DIE : num_ids =1 <<apicid_die_offset (topo_info);break ;default :/* -* Currently there is no use case for SMT, MODULE and PACKAGE, so use +* Currently there is no use case for SMT and PACKAGE, so use * assert directly to facilitate debugging. */ g_assert_not_reached ();@@ -6633,6 +6636,33 @@ static void x86_cpu_realizefn(DeviceState *dev, Error **errp) env->cache_info_amd.l3_cache = &legacy_l3_cache; } +if (cpu->l2_cache_topo_level) { +/* +* FIXME: Currently only supports changing CPUID[4] (for intel), and +* will support changing CPUID[0x8000001D] when necessary. +*/ +if (!IS_INTEL_CPU (env)) { +error_setg (errp,"only intel cpus supports x-l2-cache-topo" ); +return ; + } + +if (!strcmp (cpu->l2_cache_topo_level,"core" )) { + env->cache_info_cpuid4.l2_cache->share_level =CORE ; + }else if (!strcmp (cpu->l2_cache_topo_level,"cluster" )) { +/* +* We expose to users "cluster" instead of "module", to be +* consistent with "cluster-id" naming. +*/ + env->cache_info_cpuid4.l2_cache->share_level =MODULE ; + }else { +error_setg (errp, +"x-l2-cache-topo doesn't support '%s', " +"and it only supports 'core' or 'cluster'" , + cpu->l2_cache_topo_level); +return ; + } + } +#ifndef CONFIG_USER_ONLY MachineState *ms =MACHINE (qdev_get_machine ());qemu_register_reset (x86_cpu_machine_reset_cb, cpu);@@ -7135,6 +7165,7 @@ static Property x86_cpu_properties[] = { false),DEFINE_PROP_BOOL ("x-intel-pt-auto-level" ,X86CPU , intel_pt_auto_level, true), +DEFINE_PROP_STRING ("x-l2-cache-topo" ,X86CPU , l2_cache_topo_level), DEFINE_PROP_END_OF_LIST() };target/i386/cpu.h +2
-0@@ -1987,6 +1987,8 @@ struct ArchCPU {int32_t thread_id;int32_t hv_max_vps; + +char *l2_cache_topo_level; };