| Message ID | 20220621192034.3332546-17-sudeep.holla@arm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | arch_topology: Updates to add socket support and fix cluster ids | expand |
On Tuesday 21 Jun 2022 at 20:20:30 (+0100), Sudeep Holla wrote: > With the support of socket node parsing from the device tree and > assigning 0 as package_id in absence of socket nodes, there is no need > to check for invalid package_id. It is always initialised to 0 or values > from the device tree socket nodes. > > Just drop that now redundant check for uninitialised package_id. > > Signed-off-by: Sudeep Holla <sudeep.holla@arm.com> > --- > drivers/base/arch_topology.c | 11 ----------- > 1 file changed, 11 deletions(-) > > diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c > index 46fa1b70b02b..42448a5a9412 100644 > --- a/drivers/base/arch_topology.c > +++ b/drivers/base/arch_topology.c > @@ -611,7 +611,6 @@ static int __init parse_dt_topology(void) > { > struct device_node *cn, *map; > int ret = 0; > - int cpu; > > cn = of_find_node_by_path("/cpus"); > if (!cn) { > @@ -633,16 +632,6 @@ static int __init parse_dt_topology(void) > > topology_normalize_cpu_scale(); > > - /* > - * Check that all cores are in the topology; the SMP code will > - * only mark cores described in the DT as possible. > - */ > - for_each_possible_cpu(cpu) > - if (cpu_topology[cpu].package_id < 0) { > - ret = -EINVAL; > - break; > - } > - Maybe it would still be good to keep this for systems with potential errors in DT, where one forgets to add a core in cpu-map. For example, if I modify juno.dts as follows: --- a/arch/arm64/boot/dts/arm/juno.dts +++ b/arch/arm64/boot/dts/arm/juno.dts @@ -51,12 +51,9 @@ core0 { cpu = <&A53_0>; }; core1 { - cpu = <&A53_1>; - }; - core2 { cpu = <&A53_2>; }; - core3 { + core2 { cpu = <&A53_3>; }; }; and miss a little core in cluster1, I would end up with an incomplete topology: core3 would have cluster_id as -1, while all other CPUs would have a valid value; also, core3 would have package_id = -1. Thanks, Ionela. > out_map: > of_node_put(map); > out: > -- > 2.36.1 >
On Mon, Jun 27, 2022 at 02:12:52PM +0100, Ionela Voinescu wrote: > On Tuesday 21 Jun 2022 at 20:20:30 (+0100), Sudeep Holla wrote: > > With the support of socket node parsing from the device tree and > > assigning 0 as package_id in absence of socket nodes, there is no need > > to check for invalid package_id. It is always initialised to 0 or values > > from the device tree socket nodes. > > > > Just drop that now redundant check for uninitialised package_id. > > > > Signed-off-by: Sudeep Holla <sudeep.holla@arm.com> > > --- > > drivers/base/arch_topology.c | 11 ----------- > > 1 file changed, 11 deletions(-) > > > > diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c > > index 46fa1b70b02b..42448a5a9412 100644 > > --- a/drivers/base/arch_topology.c > > +++ b/drivers/base/arch_topology.c > > @@ -611,7 +611,6 @@ static int __init parse_dt_topology(void) > > { > > struct device_node *cn, *map; > > int ret = 0; > > - int cpu; > > > > cn = of_find_node_by_path("/cpus"); > > if (!cn) { > > @@ -633,16 +632,6 @@ static int __init parse_dt_topology(void) > > > > topology_normalize_cpu_scale(); > > > > - /* > > - * Check that all cores are in the topology; the SMP code will > > - * only mark cores described in the DT as possible. > > - */ > > - for_each_possible_cpu(cpu) > > - if (cpu_topology[cpu].package_id < 0) { > > - ret = -EINVAL; > > - break; > > - } > > - > > Maybe it would still be good to keep this for systems with potential > errors in DT, where one forgets to add a core in cpu-map. > Though I would ideally prefer to catch such DT issues with schema, I know we are no there yet. So I agree to retain this. > For example, if I modify juno.dts as follows: > > --- a/arch/arm64/boot/dts/arm/juno.dts > +++ b/arch/arm64/boot/dts/arm/juno.dts > @@ -51,12 +51,9 @@ core0 { > cpu = <&A53_0>; > }; > core1 { > - cpu = <&A53_1>; > - }; > - core2 { > cpu = <&A53_2>; > }; > - core3 { > + core2 { > cpu = <&A53_3>; > }; > }; > > and miss a little core in cluster1, I would end up with an incomplete > topology: core3 would have cluster_id as -1, while all other CPUs would > have a valid value; also, core3 would have package_id = -1. > Indeed. I didn't consider the case where DT would have issues when I dropped this check. I will drop this patch. Thanks again.
diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c index 46fa1b70b02b..42448a5a9412 100644 --- a/drivers/base/arch_topology.c +++ b/drivers/base/arch_topology.c @@ -611,7 +611,6 @@ static int __init parse_dt_topology(void) { struct device_node *cn, *map; int ret = 0; - int cpu; cn = of_find_node_by_path("/cpus"); if (!cn) { @@ -633,16 +632,6 @@ static int __init parse_dt_topology(void) topology_normalize_cpu_scale(); - /* - * Check that all cores are in the topology; the SMP code will - * only mark cores described in the DT as possible. - */ - for_each_possible_cpu(cpu) - if (cpu_topology[cpu].package_id < 0) { - ret = -EINVAL; - break; - } - out_map: of_node_put(map); out:
With the support of socket node parsing from the device tree and assigning 0 as package_id in absence of socket nodes, there is no need to check for invalid package_id. It is always initialised to 0 or values from the device tree socket nodes. Just drop that now redundant check for uninitialised package_id. Signed-off-by: Sudeep Holla <sudeep.holla@arm.com> --- drivers/base/arch_topology.c | 11 ----------- 1 file changed, 11 deletions(-)