| Message ID | 20240820204558.1296319-1-srinivas.pandruvada@linux.intel.com (mailing list archive) |
|---|---|
| State | Accepted, archived |
| Headers | show |
| Series | [v2] platform/x86/intel-uncore-freq: Do not present separate package-die domain | expand |
On Tue, 20 Aug 2024, Srinivas Pandruvada wrote: > The scope of uncore control is per power domain with TPMI. > > There are two types of processor topologies can be presented by CPUID > extended topology leaf irrespective of the hardware architecture: > > 1. A die is not enumerated in CPUID. In this case there is only one die > in a package is visible. In this case there can be multiple power domains > in a single die. > 2. A power domain in a package is enumerated as a die in CPUID. So > there is one power domain per die. > > To allow die level controls, the current implementation creates a root > domain and aggregates all information from power domains in it. This > is well suited for configuration 1 above. > > But for configuration 2 above, the root domain will present the same > information as present by power domain. So, no use of aggregating. To > check the configuration, call topology_max_dies_per_package(). If it is > more than one, avoid creating root domain. > > Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Hi, On 8/20/24 10:45 PM, Srinivas Pandruvada wrote: > The scope of uncore control is per power domain with TPMI. > > There are two types of processor topologies can be presented by CPUID > extended topology leaf irrespective of the hardware architecture: > > 1. A die is not enumerated in CPUID. In this case there is only one die > in a package is visible. In this case there can be multiple power domains > in a single die. > 2. A power domain in a package is enumerated as a die in CPUID. So > there is one power domain per die. > > To allow die level controls, the current implementation creates a root > domain and aggregates all information from power domains in it. This > is well suited for configuration 1 above. > > But for configuration 2 above, the root domain will present the same > information as present by power domain. So, no use of aggregating. To > check the configuration, call topology_max_dies_per_package(). If it is > more than one, avoid creating root domain. > > Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> > --- > This is a forward looking change as TPMI is architectural and should > support all topologies. Thank you for your patch, I've applied this patch to my review-hans branch: https://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86.git/log/?h=review-hans Note it will show up in my review-hans branch once I've pushed my local branch there, which might take a while. Once I've run some tests on this branch the patches there will be added to the platform-drivers-x86/for-next branch and eventually will be included in the pdx86 pull-request to Linus for the next merge-window. Regards, Hans > > v2 > Updated commit description as suggested. > > .../x86/intel/uncore-frequency/uncore-frequency-tpmi.c | 7 ++++++- > 1 file changed, 6 insertions(+), 1 deletion(-) > > diff --git a/drivers/platform/x86/intel/uncore-frequency/uncore-frequency-tpmi.c b/drivers/platform/x86/intel/uncore-frequency/uncore-frequency-tpmi.c > index 2016acf44f6a..e6047fbbea76 100644 > --- a/drivers/platform/x86/intel/uncore-frequency/uncore-frequency-tpmi.c > +++ b/drivers/platform/x86/intel/uncore-frequency/uncore-frequency-tpmi.c > @@ -557,6 +557,9 @@ static int uncore_probe(struct auxiliary_device *auxdev, const struct auxiliary_ > > auxiliary_set_drvdata(auxdev, tpmi_uncore); > > + if (topology_max_dies_per_package() > 1) > + return 0; > + > tpmi_uncore->root_cluster.root_domain = true; > tpmi_uncore->root_cluster.uncore_root = tpmi_uncore; > > @@ -580,7 +583,9 @@ static void uncore_remove(struct auxiliary_device *auxdev) > { > struct tpmi_uncore_struct *tpmi_uncore = auxiliary_get_drvdata(auxdev); > > - uncore_freq_remove_die_entry(&tpmi_uncore->root_cluster.uncore_data); > + if (tpmi_uncore->root_cluster.root_domain) > + uncore_freq_remove_die_entry(&tpmi_uncore->root_cluster.uncore_data); > + > remove_cluster_entries(tpmi_uncore); > > uncore_freq_common_exit();
diff --git a/drivers/platform/x86/intel/uncore-frequency/uncore-frequency-tpmi.c b/drivers/platform/x86/intel/uncore-frequency/uncore-frequency-tpmi.c index 2016acf44f6a..e6047fbbea76 100644 --- a/drivers/platform/x86/intel/uncore-frequency/uncore-frequency-tpmi.c +++ b/drivers/platform/x86/intel/uncore-frequency/uncore-frequency-tpmi.c @@ -557,6 +557,9 @@ static int uncore_probe(struct auxiliary_device *auxdev, const struct auxiliary_ auxiliary_set_drvdata(auxdev, tpmi_uncore); + if (topology_max_dies_per_package() > 1) + return 0; + tpmi_uncore->root_cluster.root_domain = true; tpmi_uncore->root_cluster.uncore_root = tpmi_uncore; @@ -580,7 +583,9 @@ static void uncore_remove(struct auxiliary_device *auxdev) { struct tpmi_uncore_struct *tpmi_uncore = auxiliary_get_drvdata(auxdev); - uncore_freq_remove_die_entry(&tpmi_uncore->root_cluster.uncore_data); + if (tpmi_uncore->root_cluster.root_domain) + uncore_freq_remove_die_entry(&tpmi_uncore->root_cluster.uncore_data); + remove_cluster_entries(tpmi_uncore); uncore_freq_common_exit();
The scope of uncore control is per power domain with TPMI. There are two types of processor topologies can be presented by CPUID extended topology leaf irrespective of the hardware architecture: 1. A die is not enumerated in CPUID. In this case there is only one die in a package is visible. In this case there can be multiple power domains in a single die. 2. A power domain in a package is enumerated as a die in CPUID. So there is one power domain per die. To allow die level controls, the current implementation creates a root domain and aggregates all information from power domains in it. This is well suited for configuration 1 above. But for configuration 2 above, the root domain will present the same information as present by power domain. So, no use of aggregating. To check the configuration, call topology_max_dies_per_package(). If it is more than one, avoid creating root domain. Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> --- This is a forward looking change as TPMI is architectural and should support all topologies. v2 Updated commit description as suggested. .../x86/intel/uncore-frequency/uncore-frequency-tpmi.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)