| Message ID | 20240704031603.1744546-1-zhao1.liu@intel.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Introduce SMP Cache Topology | expand |
Hi Daniel and Markus, A gentle ping. Would you kindly have a look at this version of the API design? If it could meet your satisfaction, I’ll continue iterating. Thanks, Zhao On Thu, Jul 04, 2024 at 11:15:55AM +0800, Zhao Liu wrote: > Date: Thu, 4 Jul 2024 11:15:55 +0800 > From: Zhao Liu <zhao1.liu@intel.com> > Subject: [PATCH 0/8] Introduce SMP Cache Topology > X-Mailer: git-send-email 2.34.1 > > Hi all, > > Since the previous RFC v2, I've reimplemented smp-cache object based on > Daniel's comment (thanks Daniel!), which is both flexible to support > current cache topology requirements and extensible. > > So, I officially convert the RFC to PATCH. > > Background on smp cache topology can be found in the previous RFC v2 > cover letter: > > https://lore.kernel.org/qemu-devel/20240530101539.768484-1-zhao1.liu@intel.com/ > > The following content focuses on this series implementation of the > smp-cache object. > > > About smp-cache > =============== > > In fact, the smp-cache object introduced in this series is a bit > different from Daniel's original suggestion. Instead of being integrated > into -smp, it is created explicitly via -object, and the smp-cache > property is added to -machine to link to this object. > > An example is as follows: > > -object '{"qom-type":"smp-cache","id":"cache0","caches":[{"name":"l1d","topo":"core"},{"name":"l1i","topo":"core"},{"name":"l2","topo":"module"},{"name":"l3","topo":"socket"}]}' > -machine smp-cache=cache0 > > > Such the design change is based on the following 2 reasons: > * Defining the cache with a JSON list is very extensible and can > support more cache properties. > > * But -smp, as the sugar of machine property "smp", is based on keyval > format, and doesn't support options with JSON format. Thus it's not > possible to add a JSON format based option to -smp/-machine for now. > > So, I decoupled the creation of the smp-cache object from the -machine > to make both -machine and -object happy! > > > Welcome your feedback! > > > Thanks and Best Regards, > Zhao > --- > Changelog: > > Main changes since RFC v2: > * Dropped cpu-topology.h and cpu-topology.c since QAPI has the helper > (CpuTopologyLevel_str) to convert enum to string. (Markus) > * Fixed text format in machine.json (CpuTopologyLevel naming, 2 spaces > between sentences). (Markus) > * Added a new level "default" to de-compatibilize some arch-specific > topo settings. (Daniel) > * Moved CpuTopologyLevel to qapi/machine-common.json, at where the > cache enumeration and smp-cache object would be added. > - If smp-cache object is defined in qapi/machine.json, storage-daemon > will complain about the qmp cmds in qapi/machine.json during > compiling. > * Referred to Daniel's suggestion to introduce cache JSON list, though > as a standalone object since -smp/-machine can't support JSON. > * Linked machine's smp_cache to smp-cache object instead of a builtin > structure. This is to get around the fact that the keyval format of > -machine can't support JSON. > * Wrapped the cache topology level access into a helper. > * Split as a separate commit to just include compatibility checking and > topology checking. > * Allow setting "default" topology level even though the cache > isn't supported by machine. (Daniel) > * Rewrote the document of smp-cache object. > > Main changes since RFC v1: > * Split CpuTopology renaimg out of this RFC. > * Use QAPI to enumerate CPU topology levels. > * Drop string_to_cpu_topo() since QAPI will help to parse the topo > levels. > * Set has_*_cache field in machine_get_smp(). (JeeHeng) > * Use "*_cache=topo_level" as -smp example as the original "level" > term for a cache has a totally different meaning. (Jonathan) > --- > Zhao Liu (8): > hw/core: Make CPU topology enumeration arch-agnostic > qapi/qom: Introduce smp-cache object > hw/core: Add smp cache topology for machine > hw/core: Check smp cache topology support for machine > i386/cpu: Support thread and module level cache topology > i386/cpu: Update cache topology with machine's configuration > i386/pc: Support cache topology in -machine for PC machine > qemu-options: Add the description of smp-cache object > > MAINTAINERS | 2 + > hw/core/machine-smp.c | 86 ++++++++++++++++++++++++++++++ > hw/core/machine.c | 22 ++++++++ > hw/core/meson.build | 1 + > hw/core/smp-cache.c | 103 ++++++++++++++++++++++++++++++++++++ > hw/i386/pc.c | 4 ++ > include/hw/boards.h | 11 +++- > include/hw/core/smp-cache.h | 27 ++++++++++ > include/hw/i386/topology.h | 18 +------ > qapi/machine-common.json | 97 ++++++++++++++++++++++++++++++++- > qapi/qapi-schema.json | 4 +- > qapi/qom.json | 3 ++ > qemu-options.hx | 58 ++++++++++++++++++++ > target/i386/cpu.c | 83 +++++++++++++++++++++-------- > target/i386/cpu.h | 4 +- > 15 files changed, 478 insertions(+), 45 deletions(-) > create mode 100644 hw/core/smp-cache.c > create mode 100644 include/hw/core/smp-cache.h > > -- > 2.34.1 >
On Mon, Jul 22, 2024 at 03:33:37PM +0800, Zhao Liu wrote: > Hi Daniel and Markus, > > A gentle ping. Would you kindly have a look at this version of the API > design? If it could meet your satisfaction, I’ll continue iterating. > > Thanks, > Zhao I'm not really a QMP guy, you want Markus.
Hi all, Since the previous RFC v2, I've reimplemented smp-cache object based on Daniel's comment (thanks Daniel!), which is both flexible to support current cache topology requirements and extensible. So, I officially convert the RFC to PATCH. Background on smp cache topology can be found in the previous RFC v2 cover letter: https://lore.kernel.org/qemu-devel/20240530101539.768484-1-zhao1.liu@intel.com/ The following content focuses on this series implementation of the smp-cache object. About smp-cache =============== In fact, the smp-cache object introduced in this series is a bit different from Daniel's original suggestion. Instead of being integrated into -smp, it is created explicitly via -object, and the smp-cache property is added to -machine to link to this object. An example is as follows: -object '{"qom-type":"smp-cache","id":"cache0","caches":[{"name":"l1d","topo":"core"},{"name":"l1i","topo":"core"},{"name":"l2","topo":"module"},{"name":"l3","topo":"socket"}]}' -machine smp-cache=cache0 Such the design change is based on the following 2 reasons: * Defining the cache with a JSON list is very extensible and can support more cache properties. * But -smp, as the sugar of machine property "smp", is based on keyval format, and doesn't support options with JSON format. Thus it's not possible to add a JSON format based option to -smp/-machine for now. So, I decoupled the creation of the smp-cache object from the -machine to make both -machine and -object happy! Welcome your feedback! Thanks and Best Regards, Zhao --- Changelog: Main changes since RFC v2: * Dropped cpu-topology.h and cpu-topology.c since QAPI has the helper (CpuTopologyLevel_str) to convert enum to string. (Markus) * Fixed text format in machine.json (CpuTopologyLevel naming, 2 spaces between sentences). (Markus) * Added a new level "default" to de-compatibilize some arch-specific topo settings. (Daniel) * Moved CpuTopologyLevel to qapi/machine-common.json, at where the cache enumeration and smp-cache object would be added. - If smp-cache object is defined in qapi/machine.json, storage-daemon will complain about the qmp cmds in qapi/machine.json during compiling. * Referred to Daniel's suggestion to introduce cache JSON list, though as a standalone object since -smp/-machine can't support JSON. * Linked machine's smp_cache to smp-cache object instead of a builtin structure. This is to get around the fact that the keyval format of -machine can't support JSON. * Wrapped the cache topology level access into a helper. * Split as a separate commit to just include compatibility checking and topology checking. * Allow setting "default" topology level even though the cache isn't supported by machine. (Daniel) * Rewrote the document of smp-cache object. Main changes since RFC v1: * Split CpuTopology renaimg out of this RFC. * Use QAPI to enumerate CPU topology levels. * Drop string_to_cpu_topo() since QAPI will help to parse the topo levels. * Set has_*_cache field in machine_get_smp(). (JeeHeng) * Use "*_cache=topo_level" as -smp example as the original "level" term for a cache has a totally different meaning. (Jonathan) --- Zhao Liu (8): hw/core: Make CPU topology enumeration arch-agnostic qapi/qom: Introduce smp-cache object hw/core: Add smp cache topology for machine hw/core: Check smp cache topology support for machine i386/cpu: Support thread and module level cache topology i386/cpu: Update cache topology with machine's configuration i386/pc: Support cache topology in -machine for PC machine qemu-options: Add the description of smp-cache object MAINTAINERS | 2 + hw/core/machine-smp.c | 86 ++++++++++++++++++++++++++++++ hw/core/machine.c | 22 ++++++++ hw/core/meson.build | 1 + hw/core/smp-cache.c | 103 ++++++++++++++++++++++++++++++++++++ hw/i386/pc.c | 4 ++ include/hw/boards.h | 11 +++- include/hw/core/smp-cache.h | 27 ++++++++++ include/hw/i386/topology.h | 18 +------ qapi/machine-common.json | 97 ++++++++++++++++++++++++++++++++- qapi/qapi-schema.json | 4 +- qapi/qom.json | 3 ++ qemu-options.hx | 58 ++++++++++++++++++++ target/i386/cpu.c | 83 +++++++++++++++++++++-------- target/i386/cpu.h | 4 +- 15 files changed, 478 insertions(+), 45 deletions(-) create mode 100644 hw/core/smp-cache.c create mode 100644 include/hw/core/smp-cache.h