additional sysfs entries for CPPC
Subsystems (3)
v2 cover additional sysfs entries for CPPC
This patch-set adds few additional sysfs entries to expose the performance capabilities of each CPU. The performance capabilities include highest perf, lowest perf, nominal perf and lowest non-linear perf. See 8.4.7.1 for ACPI 6.1 spec for details on these capabilities.
cppc_cpufreq driver operates in KHz scale whereas the delivered performance computed in userspace will be in abstract CPPC scale, so exposing perf capabilities should allow userspace to figure out the conversion factor from CPPC scale to KHz.
Changes in V2:
- Removed caching of perf caps based on Rafael's inputs
Prashanth Prakash (2):
ACPI / CPPC: Read lowest non linear perf in cppc_get_perf_caps ACPI / CPPC: add sysfs entries for CPPC perf capabilities
drivers/acpi/cppc_acpi.c | 80 ++++++++++++++++++++++++++---------------------- include/acpi/cppc_acpi.h | 3 +- 2 files changed, 46 insertions(+), 37 deletions(-)
Signature · 3 lines
Qualcomm Datacenter Technologies on behalf of Qualcomm Technologies, Inc. Qualcomm Technologies, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project.
PATCH v2 1/2 ACPI / CPPC: Read lowest non linear perf in cppc_get_perf_caps
Read lowest non linear perf in cppc_get_perf_caps so that it can be exposed via sysfs to the usespace. Lowest non linear perf is the lowest performance level at which nonlinear power savings are achieved.
--- drivers/acpi/cppc_acpi.c | 19 +++++++++++-------- include/acpi/cppc_acpi.h | 1 + 2 files changed, 12 insertions(+), 8 deletions(-)
drivers/acpi/cppc_acpi.c +11 -8
@@ -972,9 +972,9 @@ static int cpc_write(int cpu, struct cpc_register_resource *reg_res, u64 val)int cppc_get_perf_caps (int cpunum,struct cppc_perf_caps *perf_caps) {struct cpc_desc *cpc_desc =per_cpu (cpc_desc_ptr, cpunum); -struct cpc_register_resource *highest_reg, *lowest_reg, *ref_perf, - *nom_perf; -u64 high, low, nom; +struct cpc_register_resource *highest_reg, *lowest_reg, + *lowest_non_linear_reg, *nominal_reg; +u64 high, low, nom, low_non_linear;int ret =0 , regs_in_pcc =0 ;if (!cpc_desc) {
@@ -984,12 +984,12 @@ int cppc_get_perf_caps(int cpunum, struct cppc_perf_caps *perf_caps) highest_reg = &cpc_desc->cpc_regs[HIGHEST_PERF ]; lowest_reg = &cpc_desc->cpc_regs[LOWEST_PERF ]; - ref_perf = &cpc_desc->cpc_regs[REFERENCE_PERF ]; - nom_perf = &cpc_desc->cpc_regs[NOMINAL_PERF ]; + lowest_non_linear_reg = &cpc_desc->cpc_regs[LOW_NON_LINEAR_PERF ]; + nominal_reg = &cpc_desc->cpc_regs[NOMINAL_PERF ];/* Are any of the regs PCC ?*/ if (CPC_IN_PCC (highest_reg) ||CPC_IN_PCC (lowest_reg) || -CPC_IN_PCC (ref_perf) ||CPC_IN_PCC (nom_perf)) { +CPC_IN_PCC (lowest_non_linear_reg) ||CPC_IN_PCC (nominal_reg)) { regs_in_pcc =1 ;down_write (&pcc_data.pcc_lock);/* Ring doorbell once to update PCC subspace */
@@ -1005,10 +1005,13 @@ int cppc_get_perf_caps(int cpunum, struct cppc_perf_caps *perf_caps)cpc_read (cpunum, lowest_reg, &low); perf_caps->lowest_perf = low; -cpc_read (cpunum, nom_perf, &nom); +cpc_read (cpunum, nominal_reg, &nom); perf_caps->nominal_perf = nom; -if (!high || !low || !nom) +cpc_read (cpunum, lowest_non_linear_reg, &low_non_linear); + perf_caps->lowest_non_linear_perf = low_non_linear; + +if (!high || !low || !nom || !low_non_linear) ret = -EFAULT ; out_err:
include/acpi/cppc_acpi.h +1 -0
@@ -103,6 +103,7 @@ struct cppc_perf_caps {u32 highest_perf;u32 nominal_perf;u32 lowest_perf; +u32 lowest_non_linear_perf; };struct cppc_perf_ctrls {
Signature · 3 lines
Qualcomm Datacenter Technologies on behalf of Qualcomm Technologies, Inc. Qualcomm Technologies, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project.
Re: [PATCH V2 1/2] ACPI / CPPC: Read lowest non linear perf in cppc_get_perf_caps
Quoted 5 lines · Prashanth Prakash
Read lowest non linear perf in
cppc_get_perf_capsso that it can be exposed via sysfs to the usespace. Lowest non linear perf is the lowest performance level at which nonlinear power savings are achieved.Signed-off-by Prashanth Prakash <[email protected]>
Except that I would spell "nonlinear" as one word which appears to be a common practice, it looks OK to me.
Quoted 10 lines
--- drivers/acpi/cppc_acpi.c | 19 +++++++++++-------- include/acpi/cppc_acpi.h | 1 + 2 files changed, 12 insertions(+), 8 deletions(-)drivers/acpi/cppc_acpi.c +0
-0
drivers/acpi/cppc_acpi.c static int cpc_write(int cpu, struct cpc_register_resource *reg_res, u64 val) · line 977
Quoted 9 lines
drivers/acpi/cppc_acpi.c +3
-3@@ -972,9 +972,9 @@ static int cpc_write(int cpu, struct cpc_register_resource *reg_res, u64 val)int cppc_get_perf_caps (int cpunum,struct cppc_perf_caps *perf_caps) {struct cpc_desc *cpc_desc =per_cpu (cpc_desc_ptr, cpunum); -struct cpc_register_resource *highest_reg, *lowest_reg, *ref_perf, - *nom_perf; -u64 high, low, nom; +struct cpc_register_resource *highest_reg, *lowest_reg, + *lowest_non_linear_reg, *nominal_reg; +u64 high, low, nom, low_non_linear;
min_nonlinear here?
Quoted 41 lines
drivers/acpi/cppc_acpi.c +8
-5int ret =0 , regs_in_pcc =0 ;if (!cpc_desc) {@@ -984,12 +984,12 @@ int cppc_get_perf_caps(int cpunum, struct cppc_perf_caps *perf_caps) highest_reg = &cpc_desc->cpc_regs[HIGHEST_PERF ]; lowest_reg = &cpc_desc->cpc_regs[LOWEST_PERF ]; - ref_perf = &cpc_desc->cpc_regs[REFERENCE_PERF ]; - nom_perf = &cpc_desc->cpc_regs[NOMINAL_PERF ]; + lowest_non_linear_reg = &cpc_desc->cpc_regs[LOW_NON_LINEAR_PERF ]; + nominal_reg = &cpc_desc->cpc_regs[NOMINAL_PERF ];/* Are any of the regs PCC ?*/ if (CPC_IN_PCC (highest_reg) ||CPC_IN_PCC (lowest_reg) || -CPC_IN_PCC (ref_perf) ||CPC_IN_PCC (nom_perf)) { +CPC_IN_PCC (lowest_non_linear_reg) ||CPC_IN_PCC (nominal_reg)) { regs_in_pcc =1 ;down_write (&pcc_data.pcc_lock);/* Ring doorbell once to update PCC subspace */ @@ -1005,10 +1005,13 @@ int cppc_get_perf_caps(int cpunum, struct cppc_perf_caps *perf_caps)cpc_read (cpunum, lowest_reg, &low); perf_caps->lowest_perf = low; -cpc_read (cpunum, nom_perf, &nom); +cpc_read (cpunum, nominal_reg, &nom); perf_caps->nominal_perf = nom; -if (!high || !low || !nom) +cpc_read (cpunum, lowest_non_linear_reg, &low_non_linear); + perf_caps->lowest_non_linear_perf = low_non_linear; + +if (!high || !low || !nom || !low_non_linear) ret = -EFAULT ; out_err:include/acpi/cppc_acpi.h +0
-0
include/acpi/cppc_acpi.h struct cppc_perf_caps { · line 109
Quoted 7 lines
include/acpi/cppc_acpi.h +1
-0@@ -103,6 +103,7 @@ struct cppc_perf_caps {u32 highest_perf;u32 nominal_perf;u32 lowest_perf; +u32 lowest_non_linear_perf; };struct cppc_perf_ctrls {
I can make the above changes for you if you agree. :-)
Thanks,
Rafael
Re: [PATCH V2 1/2] ACPI / CPPC: Read lowest non linear perf in cppc_get_perf_caps
Hi Rafael,
Quoted 76 lines · Rafael J. Wysocki
Quoted 5 lines · Prashanth Prakash
Read lowest non linear perf in
cppc_get_perf_capsso that it can be exposed via sysfs to the usespace. Lowest non linear perf is the lowest performance level at which nonlinear power savings are achieved.Signed-off-by Prashanth Prakash <[email protected]>Except that I would spell "nonlinear" as one word which appears to be a common practice, it looks OK to me.
Quoted 10 lines
--- drivers/acpi/cppc_acpi.c | 19 +++++++++++-------- include/acpi/cppc_acpi.h | 1 + 2 files changed, 12 insertions(+), 8 deletions(-)drivers/acpi/cppc_acpi.c +0
-0drivers/acpi/cppc_acpi.c static int cpc_write(int cpu, struct cpc_register_resource *reg_res, u64 val) · line 977
Quoted 9 lines
drivers/acpi/cppc_acpi.c +3
-3@@ -972,9 +972,9 @@ static int cpc_write(int cpu, struct cpc_register_resource *reg_res, u64 val)int cppc_get_perf_caps (int cpunum,struct cppc_perf_caps *perf_caps) {struct cpc_desc *cpc_desc =per_cpu (cpc_desc_ptr, cpunum); -struct cpc_register_resource *highest_reg, *lowest_reg, *ref_perf, - *nom_perf; -u64 high, low, nom; +struct cpc_register_resource *highest_reg, *lowest_reg, + *lowest_non_linear_reg, *nominal_reg; +u64 high, low, nom, low_non_linear;
min_nonlinearhere?Quoted 46 lines
drivers/acpi/cppc_acpi.c +8
-5int ret =0 , regs_in_pcc =0 ;if (!cpc_desc) {@@ -984,12 +984,12 @@ int cppc_get_perf_caps(int cpunum, struct cppc_perf_caps *perf_caps) highest_reg = &cpc_desc->cpc_regs[HIGHEST_PERF ]; lowest_reg = &cpc_desc->cpc_regs[LOWEST_PERF ]; - ref_perf = &cpc_desc->cpc_regs[REFERENCE_PERF ]; - nom_perf = &cpc_desc->cpc_regs[NOMINAL_PERF ]; + lowest_non_linear_reg = &cpc_desc->cpc_regs[LOW_NON_LINEAR_PERF ]; + nominal_reg = &cpc_desc->cpc_regs[NOMINAL_PERF ];/* Are any of the regs PCC ?*/ if (CPC_IN_PCC (highest_reg) ||CPC_IN_PCC (lowest_reg) || -CPC_IN_PCC (ref_perf) ||CPC_IN_PCC (nom_perf)) { +CPC_IN_PCC (lowest_non_linear_reg) ||CPC_IN_PCC (nominal_reg)) { regs_in_pcc =1 ;down_write (&pcc_data.pcc_lock);/* Ring doorbell once to update PCC subspace */ @@ -1005,10 +1005,13 @@ int cppc_get_perf_caps(int cpunum, struct cppc_perf_caps *perf_caps)cpc_read (cpunum, lowest_reg, &low); perf_caps->lowest_perf = low; -cpc_read (cpunum, nom_perf, &nom); +cpc_read (cpunum, nominal_reg, &nom); perf_caps->nominal_perf = nom; -if (!high || !low || !nom) +cpc_read (cpunum, lowest_non_linear_reg, &low_non_linear); + perf_caps->lowest_non_linear_perf = low_non_linear; + +if (!high || !low || !nom || !low_non_linear) ret = -EFAULT ; out_err:include/acpi/cppc_acpi.h +1
-0@@ -103,6 +103,7 @@ struct cppc_perf_caps {u32 highest_perf;u32 nominal_perf;u32 lowest_perf; +u32 lowest_non_linear_perf; };
Quoted 3 lines
include/acpi/cppc_acpi.h struct cppc_perf_caps { · line 109
Quoted 2 lines
include/acpi/cppc_acpi.h +0
-0struct cppc_perf_ctrls {I can make the above changes for you if you agree. :-)
Thanks! Yes, you can make those correction :-)
--
Thanks,
Prashanth
PATCH v2 2/2 ACPI / CPPC: add sysfs entries for CPPC perf capabilities
In reply to Prashanth PrakashComputed delivered performance using CPPC feedback counters are in the CPPC abstract scale, whereas cppc_cpufreq driver operates in KHz scale. Exposing the CPPC performance capabilities (highest,lowest, nominal, lowest non-linear) will allow userspace to figure out the conversion factor from CPPC abstract scale to KHz.
Also rename ctr_wrap_time to wraparound_time so that show_cppc_data() macro will work with it.
--- drivers/acpi/cppc_acpi.c | 61 ++++++++++++++++++++++++++---------------------- include/acpi/cppc_acpi.h | 2 +- 2 files changed, 34 insertions(+), 29 deletions(-)
drivers/acpi/cppc_acpi.c +33 -28
@@ -132,49 +132,54 @@ struct cppc_attr {#define to_cpc_desc (a) container_of(a, struct cpc_desc, kobj) +#define show_cppc_data (access_fn, struct_name, member_name) \ + static ssize_t show_##member_name(struct kobject *kobj, \ + struct attribute *attr, char *buf) \ + { \ + struct cpc_desc *cpc_ptr = to_cpc_desc(kobj); \ + struct struct_name st_name = {0}; \ + int ret; \ + \ + ret = access_fn(cpc_ptr->cpu_id, &st_name); \ + if (ret) \ + return ret; \ + \ + return scnprintf(buf, PAGE_SIZE, "%llu\n", \ + (u64)st_name.member_name); \ + } \ + define_one_cppc_ro(member_name) + +show_cppc_data (cppc_get_perf_caps, cppc_perf_caps, highest_perf); +show_cppc_data (cppc_get_perf_caps, cppc_perf_caps, lowest_perf); +show_cppc_data (cppc_get_perf_caps, cppc_perf_caps, nominal_perf); +show_cppc_data (cppc_get_perf_caps, cppc_perf_caps, lowest_non_linear_perf); +show_cppc_data (cppc_get_perf_ctrs, cppc_perf_fb_ctrs, reference_perf); +show_cppc_data (cppc_get_perf_ctrs, cppc_perf_fb_ctrs, wraparound_time); +static ssize_t show_feedback_ctrs (struct kobject *kobj,struct attribute *attr,char *buf) {struct cpc_desc *cpc_ptr =to_cpc_desc (kobj);struct cppc_perf_fb_ctrs fb_ctrs = {0 }; +int ret; -cppc_get_perf_ctrs (cpc_ptr->cpu_id, &fb_ctrs); + ret =cppc_get_perf_ctrs (cpc_ptr->cpu_id, &fb_ctrs); +if (ret) +return ret;return scnprintf (buf,PAGE_SIZE ,"ref:%llu del:%llu\n" , fb_ctrs.reference, fb_ctrs.delivered); }define_one_cppc_ro (feedback_ctrs); -static ssize_t show_reference_perf (struct kobject *kobj, -struct attribute *attr,char *buf) -{ -struct cpc_desc *cpc_ptr =to_cpc_desc (kobj); -struct cppc_perf_fb_ctrs fb_ctrs = {0 }; - -cppc_get_perf_ctrs (cpc_ptr->cpu_id, &fb_ctrs); - -return scnprintf (buf,PAGE_SIZE ,"%llu\n" , - fb_ctrs.reference_perf); -} -define_one_cppc_ro (reference_perf); - -static ssize_t show_wraparound_time (struct kobject *kobj, -struct attribute *attr,char *buf) -{ -struct cpc_desc *cpc_ptr =to_cpc_desc (kobj); -struct cppc_perf_fb_ctrs fb_ctrs = {0 }; - -cppc_get_perf_ctrs (cpc_ptr->cpu_id, &fb_ctrs); - -return scnprintf (buf,PAGE_SIZE ,"%llu\n" , fb_ctrs.ctr_wrap_time); - -} -define_one_cppc_ro (wraparound_time); -static struct attribute *cppc_attrs[] = { &feedback_ctrs.attr, &reference_perf.attr, &wraparound_time.attr, + &highest_perf.attr, + &lowest_perf.attr, + &lowest_non_linear_perf.attr, + &nominal_perf.attr,NULL };
@@ -1086,7 +1091,7 @@ int cppc_get_perf_ctrs(int cpunum, struct cppc_perf_fb_ctrs *perf_fb_ctrs) perf_fb_ctrs->delivered = delivered; perf_fb_ctrs->reference = reference; perf_fb_ctrs->reference_perf = ref_perf; - perf_fb_ctrs->ctr_wrap_time = ctr_wrap_time; + perf_fb_ctrs->wraparound_time = ctr_wrap_time; out_err:if (regs_in_pcc)up_write (&pcc_data.pcc_lock);
include/acpi/cppc_acpi.h +1 -1
@@ -116,7 +116,7 @@ struct cppc_perf_fb_ctrs {u64 reference;u64 delivered;u64 reference_perf; -u64 ctr_wrap_time; +u64 wraparound_time; };/* Per CPU container for runtime CPPC management. */
Signature · 3 lines
Qualcomm Datacenter Technologies on behalf of Qualcomm Technologies, Inc. Qualcomm Technologies, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project.
Cc++
Any feedback on this?
Quoted 14 lines · Prashanth Prakash
This patch-set adds few additional sysfs entries to expose the performance capabilities of each CPU. The performance capabilities include highest perf, lowest perf, nominal perf and lowest non-linear perf. See 8.4.7.1 for ACPI 6.1 spec for details on these capabilities.
cppc_cpufreqdriver operates in KHz scale whereas the delivered performance computed in userspace will be in abstract CPPC scale, so exposing perf capabilities should allow userspace to figure out the conversion factor from CPPC scale to KHz.Changes in V2:
- Removed caching of perf caps based on Rafael's inputs
Quoted 8 lines
Prashanth Prakash (2):
ACPI / CPPC: Read lowest non linear perf incppc_get_perf_capsACPI / CPPC: add sysfs entries for CPPC perf capabilitiesdrivers/acpi/cppc_acpi.c | 80 ++++++++++++++++++++++++++---------------------- include/acpi/cppc_acpi.h | 3 +- 2 files changed, 46 insertions(+), 37 deletions(-)
--
Thanks,
Prashanth
Quoted 26 lines · Prashanth
Cc++
Any feedback on this?
Quoted 21 lines · Prashanth Prakash
This patch-set adds few additional sysfs entries to expose the performance capabilities of each CPU. The performance capabilities include highest perf, lowest perf, nominal perf and lowest non-linear perf. See 8.4.7.1 for ACPI 6.1 spec for details on these capabilities.
cppc_cpufreqdriver operates in KHz scale whereas the delivered performance computed in userspace will be in abstract CPPC scale, so exposing perf capabilities should allow userspace to figure out the conversion factor from CPPC scale to KHz.Changes in V2:
- Removed caching of perf caps based on Rafael's inputsPrashanth Prakash (2):
ACPI / CPPC: Read lowest non linear perf incppc_get_perf_capsACPI / CPPC: add sysfs entries for CPPC perf capabilitiesdrivers/acpi/cppc_acpi.c | 80 ++++++++++++++++++++++++++---------------------- include/acpi/cppc_acpi.h | 3 +- 2 files changed, 46 insertions(+), 37 deletions(-)
I am about to get to this.
Thanks,
Rafael