| Message ID | 20220225031255.3647599-3-mario.limonciello@amd.com |
|---|---|
| State | New |
| Headers | show |
| Series | Avoid enabling LPI on non-ARM | expand |
On Thu, Feb 24, 2022 at 09:12:55PM -0600, Mario Limonciello wrote: > When `osc_pc_lpi_support_confirmed` is set through `_OSC` and `_LPI` is > populated then the cpuidle driver assumes that LPI is fully functional. > > However currently the kernel only provides architectural support for LPI > on ARM. This leads to high power consumption on X86 platforms that > otherwise try to enable LPI. > > So probe whether or not LPI support is implemented before enabling LPI in > the kernel. This is done by overloading `acpi_processor_ffh_lpi_probe` to > check whether it returns `-EOPNOTSUPP`. It also means that all future > implementations of `acpi_processor_ffh_lpi_probe` will need to follow > these semantics as well. > > Signed-off-by: Mario Limonciello <mario.limonciello@amd.com> > --- > drivers/acpi/processor_idle.c | 15 ++++++++++----- > 1 file changed, 10 insertions(+), 5 deletions(-) > > diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c > index f8e9fa82cb9b..0092fd479527 100644 > --- a/drivers/acpi/processor_idle.c > +++ b/drivers/acpi/processor_idle.c > @@ -1080,6 +1080,11 @@ static int flatten_lpi_states(struct acpi_processor *pr, > return 0; > } > > +int __weak acpi_processor_ffh_lpi_probe(unsigned int cpu) > +{ > + return -EOPNOTSUPP; > +} > + > static int acpi_processor_get_lpi_info(struct acpi_processor *pr) > { > int ret, i; > @@ -1094,6 +1099,11 @@ static int acpi_processor_get_lpi_info(struct acpi_processor *pr) > if (!acpi_has_method(handle, "_LPI")) > return -EINVAL; > > + /* make sure our architecture has support */ > + ret = acpi_processor_ffh_lpi_probe(pr->id); > + if (ret == -EOPNOTSUPP) > + return ret; > + You can do this check first, before even checking osc_pc_lpi_support_confirmed and looking for _LPI method IMO as they are not useful if arch doesn't support handling LPIs. Either way, Reviewed-by: Sudeep Holla <sudeep.holla@arm.com>
diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c index f8e9fa82cb9b..0092fd479527 100644 --- a/drivers/acpi/processor_idle.c +++ b/drivers/acpi/processor_idle.c @@ -1080,6 +1080,11 @@ static int flatten_lpi_states(struct acpi_processor *pr, return 0; } +int __weak acpi_processor_ffh_lpi_probe(unsigned int cpu) +{ + return -EOPNOTSUPP; +} + static int acpi_processor_get_lpi_info(struct acpi_processor *pr) { int ret, i; @@ -1094,6 +1099,11 @@ static int acpi_processor_get_lpi_info(struct acpi_processor *pr) if (!acpi_has_method(handle, "_LPI")) return -EINVAL; + /* make sure our architecture has support */ + ret = acpi_processor_ffh_lpi_probe(pr->id); + if (ret == -EOPNOTSUPP) + return ret; + flat_state_cnt = 0; prev = &info[0]; curr = &info[1]; @@ -1139,11 +1149,6 @@ static int acpi_processor_get_lpi_info(struct acpi_processor *pr) return 0; } -int __weak acpi_processor_ffh_lpi_probe(unsigned int cpu) -{ - return -ENODEV; -} - int __weak acpi_processor_ffh_lpi_enter(struct acpi_lpi_state *lpi) { return -ENODEV;
When `osc_pc_lpi_support_confirmed` is set through `_OSC` and `_LPI` is populated then the cpuidle driver assumes that LPI is fully functional. However currently the kernel only provides architectural support for LPI on ARM. This leads to high power consumption on X86 platforms that otherwise try to enable LPI. So probe whether or not LPI support is implemented before enabling LPI in the kernel. This is done by overloading `acpi_processor_ffh_lpi_probe` to check whether it returns `-EOPNOTSUPP`. It also means that all future implementations of `acpi_processor_ffh_lpi_probe` will need to follow these semantics as well. Signed-off-by: Mario Limonciello <mario.limonciello@amd.com> --- drivers/acpi/processor_idle.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-)