cpumask: convert shared_cpu_map in acpi_processor* structs to cpumask_var_t
Impact: Reduce memory usage, use new API.
This is part of an effort to reduce structure sizes for machines
configured with large NR_CPUS. cpumask_t gets replaced by
cpumask_var_t, which is either struct cpumask[1] (small NR_CPUS) or
struct cpumask * (large NR_CPUS).
(Changes to powernow-k* by <travis>.)
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Mike Travis <travis@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
index 88ea02d..d0a0010 100644
--- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
@@ -517,6 +517,17 @@
}
}
+static void free_acpi_perf_data(void)
+{
+ unsigned int i;
+
+ /* Freeing a NULL pointer is OK, and alloc_percpu zeroes. */
+ for_each_possible_cpu(i)
+ free_cpumask_var(per_cpu_ptr(acpi_perf_data, i)
+ ->shared_cpu_map);
+ free_percpu(acpi_perf_data);
+}
+
/*
* acpi_cpufreq_early_init - initialize ACPI P-States library
*
@@ -527,6 +538,7 @@
*/
static int __init acpi_cpufreq_early_init(void)
{
+ unsigned int i;
dprintk("acpi_cpufreq_early_init\n");
acpi_perf_data = alloc_percpu(struct acpi_processor_performance);
@@ -534,6 +546,15 @@
dprintk("Memory allocation error for acpi_perf_data.\n");
return -ENOMEM;
}
+ for_each_possible_cpu(i) {
+ if (!alloc_cpumask_var(&per_cpu_ptr(acpi_perf_data, i)
+ ->shared_cpu_map, GFP_KERNEL)) {
+
+ /* Freeing a NULL pointer is OK: alloc_percpu zeroes. */
+ free_acpi_perf_data();
+ return -ENOMEM;
+ }
+ }
/* Do initialization in ACPI core */
acpi_processor_preregister_performance(acpi_perf_data);
@@ -604,9 +625,9 @@
*/
if (policy->shared_type == CPUFREQ_SHARED_TYPE_ALL ||
policy->shared_type == CPUFREQ_SHARED_TYPE_ANY) {
- policy->cpus = perf->shared_cpu_map;
+ cpumask_copy(&policy->cpus, perf->shared_cpu_map);
}
- policy->related_cpus = perf->shared_cpu_map;
+ cpumask_copy(&policy->related_cpus, perf->shared_cpu_map);
#ifdef CONFIG_SMP
dmi_check_system(sw_any_bug_dmi_table);
@@ -795,7 +816,7 @@
ret = cpufreq_register_driver(&acpi_cpufreq_driver);
if (ret)
- free_percpu(acpi_perf_data);
+ free_acpi_perf_data();
return ret;
}
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k7.c b/arch/x86/kernel/cpu/cpufreq/powernow-k7.c
index 7c7d56b..1b446d7 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k7.c
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k7.c
@@ -310,6 +310,12 @@
goto err0;
}
+ if (!alloc_cpumask_var(&acpi_processor_perf->shared_cpu_map,
+ GFP_KERNEL)) {
+ retval = -ENOMEM;
+ goto err05;
+ }
+
if (acpi_processor_register_performance(acpi_processor_perf, 0)) {
retval = -EIO;
goto err1;
@@ -412,6 +418,8 @@
err2:
acpi_processor_unregister_performance(acpi_processor_perf, 0);
err1:
+ free_cpumask_var(acpi_processor_perf->shared_cpu_map);
+err05:
kfree(acpi_processor_perf);
err0:
printk(KERN_WARNING PFX "ACPI perflib can not be used in this platform\n");
@@ -652,6 +660,7 @@
#ifdef CONFIG_X86_POWERNOW_K7_ACPI
if (acpi_processor_perf) {
acpi_processor_unregister_performance(acpi_processor_perf, 0);
+ free_cpumask_var(acpi_processor_perf->shared_cpu_map);
kfree(acpi_processor_perf);
}
#endif
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
index 7f05f44..c3c9adb 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
@@ -766,7 +766,7 @@
static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data)
{
struct cpufreq_frequency_table *powernow_table;
- int ret_val;
+ int ret_val = -ENODEV;
if (acpi_processor_register_performance(&data->acpi_data, data->cpu)) {
dprintk("register performance failed: bad ACPI data\n");
@@ -815,6 +815,13 @@
/* notify BIOS that we exist */
acpi_processor_notify_smm(THIS_MODULE);
+ if (!alloc_cpumask_var(&data->acpi_data.shared_cpu_map, GFP_KERNEL)) {
+ printk(KERN_ERR PFX
+ "unable to alloc powernow_k8_data cpumask\n");
+ ret_val = -ENOMEM;
+ goto err_out_mem;
+ }
+
return 0;
err_out_mem:
@@ -826,7 +833,7 @@
/* data->acpi_data.state_count informs us at ->exit() whether ACPI was used */
data->acpi_data.state_count = 0;
- return -ENODEV;
+ return ret_val;
}
static int fill_powernow_table_pstate(struct powernow_k8_data *data, struct cpufreq_frequency_table *powernow_table)
@@ -929,6 +936,7 @@
{
if (data->acpi_data.state_count)
acpi_processor_unregister_performance(&data->acpi_data, data->cpu);
+ free_cpumask_var(data->acpi_data.shared_cpu_map);
}
#else
@@ -1134,7 +1142,8 @@
data->cpu = pol->cpu;
data->currpstate = HW_PSTATE_INVALID;
- if (powernow_k8_cpu_init_acpi(data)) {
+ rc = powernow_k8_cpu_init_acpi(data);
+ if (rc) {
/*
* Use the PSB BIOS structure. This is only availabe on
* an UP version, and is deprecated by AMD.
@@ -1152,20 +1161,17 @@
"ACPI maintainers and complain to your BIOS "
"vendor.\n");
#endif
- kfree(data);
- return -ENODEV;
+ goto err_out;
}
if (pol->cpu != 0) {
printk(KERN_ERR FW_BUG PFX "No ACPI _PSS objects for "
"CPU other than CPU0. Complain to your BIOS "
"vendor.\n");
- kfree(data);
- return -ENODEV;
+ goto err_out;
}
rc = find_psb_table(data);
if (rc) {
- kfree(data);
- return -ENODEV;
+ goto err_out;
}
}
diff --git a/drivers/acpi/processor_core.c b/drivers/acpi/processor_core.c
index 3494836..0cc2fd3 100644
--- a/drivers/acpi/processor_core.c
+++ b/drivers/acpi/processor_core.c
@@ -826,6 +826,11 @@
if (!pr)
return -ENOMEM;
+ if (!alloc_cpumask_var(&pr->throttling.shared_cpu_map, GFP_KERNEL)) {
+ kfree(pr);
+ return -ENOMEM;
+ }
+
pr->handle = device->handle;
strcpy(acpi_device_name(device), ACPI_PROCESSOR_DEVICE_NAME);
strcpy(acpi_device_class(device), ACPI_PROCESSOR_CLASS);
@@ -845,10 +850,8 @@
pr = acpi_driver_data(device);
- if (pr->id >= nr_cpu_ids) {
- kfree(pr);
- return 0;
- }
+ if (pr->id >= nr_cpu_ids)
+ goto free;
if (type == ACPI_BUS_REMOVAL_EJECT) {
if (acpi_processor_handle_eject(pr))
@@ -873,6 +876,9 @@
per_cpu(processors, pr->id) = NULL;
per_cpu(processor_device_array, pr->id) = NULL;
+
+free:
+ free_cpumask_var(pr->throttling.shared_cpu_map);
kfree(pr);
return 0;
diff --git a/drivers/acpi/processor_perflib.c b/drivers/acpi/processor_perflib.c
index 0d7b772..846e227 100644
--- a/drivers/acpi/processor_perflib.c
+++ b/drivers/acpi/processor_perflib.c
@@ -588,12 +588,15 @@
int count, count_target;
int retval = 0;
unsigned int i, j;
- cpumask_t covered_cpus;
+ cpumask_var_t covered_cpus;
struct acpi_processor *pr;
struct acpi_psd_package *pdomain;
struct acpi_processor *match_pr;
struct acpi_psd_package *match_pdomain;
+ if (!alloc_cpumask_var(&covered_cpus, GFP_KERNEL))
+ return -ENOMEM;
+
mutex_lock(&performance_mutex);
retval = 0;
@@ -617,7 +620,7 @@
}
pr->performance = percpu_ptr(performance, i);
- cpu_set(i, pr->performance->shared_cpu_map);
+ cpumask_set_cpu(i, pr->performance->shared_cpu_map);
if (acpi_processor_get_psd(pr)) {
retval = -EINVAL;
continue;
@@ -650,18 +653,18 @@
}
}
- cpus_clear(covered_cpus);
+ cpumask_clear(covered_cpus);
for_each_possible_cpu(i) {
pr = per_cpu(processors, i);
if (!pr)
continue;
- if (cpu_isset(i, covered_cpus))
+ if (cpumask_test_cpu(i, covered_cpus))
continue;
pdomain = &(pr->performance->domain_info);
- cpu_set(i, pr->performance->shared_cpu_map);
- cpu_set(i, covered_cpus);
+ cpumask_set_cpu(i, pr->performance->shared_cpu_map);
+ cpumask_set_cpu(i, covered_cpus);
if (pdomain->num_processors <= 1)
continue;
@@ -699,8 +702,8 @@
goto err_ret;
}
- cpu_set(j, covered_cpus);
- cpu_set(j, pr->performance->shared_cpu_map);
+ cpumask_set_cpu(j, covered_cpus);
+ cpumask_set_cpu(j, pr->performance->shared_cpu_map);
count++;
}
@@ -718,8 +721,8 @@
match_pr->performance->shared_type =
pr->performance->shared_type;
- match_pr->performance->shared_cpu_map =
- pr->performance->shared_cpu_map;
+ cpumask_copy(match_pr->performance->shared_cpu_map,
+ pr->performance->shared_cpu_map);
}
}
@@ -731,14 +734,15 @@
/* Assume no coordination on any error parsing domain info */
if (retval) {
- cpus_clear(pr->performance->shared_cpu_map);
- cpu_set(i, pr->performance->shared_cpu_map);
+ cpumask_clear(pr->performance->shared_cpu_map);
+ cpumask_set_cpu(i, pr->performance->shared_cpu_map);
pr->performance->shared_type = CPUFREQ_SHARED_TYPE_ALL;
}
pr->performance = NULL; /* Will be set for real in register */
}
mutex_unlock(&performance_mutex);
+ free_cpumask_var(covered_cpus);
return retval;
}
EXPORT_SYMBOL(acpi_processor_preregister_performance);
diff --git a/drivers/acpi/processor_throttling.c b/drivers/acpi/processor_throttling.c
index a0c38c9..d278381 100644
--- a/drivers/acpi/processor_throttling.c
+++ b/drivers/acpi/processor_throttling.c
@@ -61,11 +61,14 @@
int count, count_target;
int retval = 0;
unsigned int i, j;
- cpumask_t covered_cpus;
+ cpumask_var_t covered_cpus;
struct acpi_processor *pr, *match_pr;
struct acpi_tsd_package *pdomain, *match_pdomain;
struct acpi_processor_throttling *pthrottling, *match_pthrottling;
+ if (!alloc_cpumask_var(&covered_cpus, GFP_KERNEL))
+ return -ENOMEM;
+
/*
* Now that we have _TSD data from all CPUs, lets setup T-state
* coordination between all CPUs.
@@ -91,19 +94,19 @@
if (retval)
goto err_ret;
- cpus_clear(covered_cpus);
+ cpumask_clear(covered_cpus);
for_each_possible_cpu(i) {
pr = per_cpu(processors, i);
if (!pr)
continue;
- if (cpu_isset(i, covered_cpus))
+ if (cpumask_test_cpu(i, covered_cpus))
continue;
pthrottling = &pr->throttling;
pdomain = &(pthrottling->domain_info);
- cpu_set(i, pthrottling->shared_cpu_map);
- cpu_set(i, covered_cpus);
+ cpumask_set_cpu(i, pthrottling->shared_cpu_map);
+ cpumask_set_cpu(i, covered_cpus);
/*
* If the number of processor in the TSD domain is 1, it is
* unnecessary to parse the coordination for this CPU.
@@ -144,8 +147,8 @@
goto err_ret;
}
- cpu_set(j, covered_cpus);
- cpu_set(j, pthrottling->shared_cpu_map);
+ cpumask_set_cpu(j, covered_cpus);
+ cpumask_set_cpu(j, pthrottling->shared_cpu_map);
count++;
}
for_each_possible_cpu(j) {
@@ -165,12 +168,14 @@
* If some CPUS have the same domain, they
* will have the same shared_cpu_map.
*/
- match_pthrottling->shared_cpu_map =
- pthrottling->shared_cpu_map;
+ cpumask_copy(match_pthrottling->shared_cpu_map,
+ pthrottling->shared_cpu_map);
}
}
err_ret:
+ free_cpumask_var(covered_cpus);
+
for_each_possible_cpu(i) {
pr = per_cpu(processors, i);
if (!pr)
@@ -182,8 +187,8 @@
*/
if (retval) {
pthrottling = &(pr->throttling);
- cpus_clear(pthrottling->shared_cpu_map);
- cpu_set(i, pthrottling->shared_cpu_map);
+ cpumask_clear(pthrottling->shared_cpu_map);
+ cpumask_set_cpu(i, pthrottling->shared_cpu_map);
pthrottling->shared_type = DOMAIN_COORD_TYPE_SW_ALL;
}
}
@@ -567,7 +572,7 @@
pthrottling = &pr->throttling;
pthrottling->tsd_valid_flag = 1;
pthrottling->shared_type = pdomain->coord_type;
- cpu_set(pr->id, pthrottling->shared_cpu_map);
+ cpumask_set_cpu(pr->id, pthrottling->shared_cpu_map);
/*
* If the coordination type is not defined in ACPI spec,
* the tsd_valid_flag will be clear and coordination type
@@ -826,7 +831,7 @@
static int acpi_processor_get_throttling(struct acpi_processor *pr)
{
- cpumask_t saved_mask;
+ cpumask_var_t saved_mask;
int ret;
if (!pr)
@@ -834,14 +839,20 @@
if (!pr->flags.throttling)
return -ENODEV;
+
+ if (!alloc_cpumask_var(&saved_mask, GFP_KERNEL))
+ return -ENOMEM;
+
/*
* Migrate task to the cpu pointed by pr.
*/
- saved_mask = current->cpus_allowed;
- set_cpus_allowed_ptr(current, &cpumask_of_cpu(pr->id));
+ cpumask_copy(saved_mask, ¤t->cpus_allowed);
+ /* FIXME: use work_on_cpu() */
+ set_cpus_allowed_ptr(current, cpumask_of(pr->id));
ret = pr->throttling.acpi_processor_get_throttling(pr);
/* restore the previous state */
- set_cpus_allowed_ptr(current, &saved_mask);
+ set_cpus_allowed_ptr(current, saved_mask);
+ free_cpumask_var(saved_mask);
return ret;
}
@@ -986,13 +997,13 @@
int acpi_processor_set_throttling(struct acpi_processor *pr, int state)
{
- cpumask_t saved_mask;
+ cpumask_var_t saved_mask;
int ret = 0;
unsigned int i;
struct acpi_processor *match_pr;
struct acpi_processor_throttling *p_throttling;
struct throttling_tstate t_state;
- cpumask_t online_throttling_cpus;
+ cpumask_var_t online_throttling_cpus;
if (!pr)
return -EINVAL;
@@ -1003,17 +1014,25 @@
if ((state < 0) || (state > (pr->throttling.state_count - 1)))
return -EINVAL;
- saved_mask = current->cpus_allowed;
+ if (!alloc_cpumask_var(&saved_mask, GFP_KERNEL))
+ return -ENOMEM;
+
+ if (!alloc_cpumask_var(&online_throttling_cpus, GFP_KERNEL)) {
+ free_cpumask_var(saved_mask);
+ return -ENOMEM;
+ }
+
+ cpumask_copy(saved_mask, ¤t->cpus_allowed);
t_state.target_state = state;
p_throttling = &(pr->throttling);
- cpus_and(online_throttling_cpus, cpu_online_map,
- p_throttling->shared_cpu_map);
+ cpumask_and(online_throttling_cpus, cpu_online_mask,
+ p_throttling->shared_cpu_map);
/*
* The throttling notifier will be called for every
* affected cpu in order to get one proper T-state.
* The notifier event is THROTTLING_PRECHANGE.
*/
- for_each_cpu_mask_nr(i, online_throttling_cpus) {
+ for_each_cpu(i, online_throttling_cpus) {
t_state.cpu = i;
acpi_processor_throttling_notifier(THROTTLING_PRECHANGE,
&t_state);
@@ -1025,7 +1044,8 @@
* it can be called only for the cpu pointed by pr.
*/
if (p_throttling->shared_type == DOMAIN_COORD_TYPE_SW_ANY) {
- set_cpus_allowed_ptr(current, &cpumask_of_cpu(pr->id));
+ /* FIXME: use work_on_cpu() */
+ set_cpus_allowed_ptr(current, cpumask_of(pr->id));
ret = p_throttling->acpi_processor_set_throttling(pr,
t_state.target_state);
} else {
@@ -1034,7 +1054,7 @@
* it is necessary to set T-state for every affected
* cpus.
*/
- for_each_cpu_mask_nr(i, online_throttling_cpus) {
+ for_each_cpu(i, online_throttling_cpus) {
match_pr = per_cpu(processors, i);
/*
* If the pointer is invalid, we will report the
@@ -1056,7 +1076,8 @@
continue;
}
t_state.cpu = i;
- set_cpus_allowed_ptr(current, &cpumask_of_cpu(i));
+ /* FIXME: use work_on_cpu() */
+ set_cpus_allowed_ptr(current, cpumask_of(i));
ret = match_pr->throttling.
acpi_processor_set_throttling(
match_pr, t_state.target_state);
@@ -1068,13 +1089,16 @@
* affected cpu to update the T-states.
* The notifier event is THROTTLING_POSTCHANGE
*/
- for_each_cpu_mask_nr(i, online_throttling_cpus) {
+ for_each_cpu(i, online_throttling_cpus) {
t_state.cpu = i;
acpi_processor_throttling_notifier(THROTTLING_POSTCHANGE,
&t_state);
}
/* restore the previous state */
- set_cpus_allowed_ptr(current, &saved_mask);
+ /* FIXME: use work_on_cpu() */
+ set_cpus_allowed_ptr(current, saved_mask);
+ free_cpumask_var(online_throttling_cpus);
+ free_cpumask_var(saved_mask);
return ret;
}
@@ -1120,7 +1144,7 @@
if (acpi_processor_get_tsd(pr)) {
pthrottling = &pr->throttling;
pthrottling->tsd_valid_flag = 0;
- cpu_set(pr->id, pthrottling->shared_cpu_map);
+ cpumask_set_cpu(pr->id, pthrottling->shared_cpu_map);
pthrottling->shared_type = DOMAIN_COORD_TYPE_SW_ALL;
}
diff --git a/include/acpi/processor.h b/include/acpi/processor.h
index 3795590..0574add 100644
--- a/include/acpi/processor.h
+++ b/include/acpi/processor.h
@@ -127,7 +127,7 @@
unsigned int state_count;
struct acpi_processor_px *states;
struct acpi_psd_package domain_info;
- cpumask_t shared_cpu_map;
+ cpumask_var_t shared_cpu_map;
unsigned int shared_type;
};
@@ -172,7 +172,7 @@
unsigned int state_count;
struct acpi_processor_tx_tss *states_tss;
struct acpi_tsd_package domain_info;
- cpumask_t shared_cpu_map;
+ cpumask_var_t shared_cpu_map;
int (*acpi_processor_get_throttling) (struct acpi_processor * pr);
int (*acpi_processor_set_throttling) (struct acpi_processor * pr,
int state);