Merge branches 'pm-opp', 'pm-misc', 'pm-avs' and 'pm-tools'

* pm-opp:
  opp: Don't use IS_ERR on invalid supplies
  opp: Make dev_pm_opp_set_rate() handle freq = 0 to drop performance votes
  opp: Don't overwrite rounded clk rate
  opp: Allocate genpd_virt_devs from dev_pm_opp_attach_genpd()
  opp: Attach genpds to devices from within OPP core

* pm-misc:
  PM / clk: Remove error message on out-of-memory condition
  drivers: base: power: clock_ops: Use of_clk_get_parent_count()

* pm-avs:
  power: avs: smartreflex: no need to check return value of debugfs_create functions

* pm-tools:
  cpupower : frequency-set -r option misses the last cpu in related cpu list
  cpupower: correct spelling of interval
  Add README and update pm-graph and sleepgraph docs
  Update to pm-graph 5.4
  Update to pm-graph 5.3
This commit is contained in:
Rafael J. Wysocki 2019-07-08 10:59:38 +02:00
17 changed files with 1399 additions and 341 deletions

View File

@ -12,6 +12,7 @@
#include <linux/pm_clock.h>
#include <linux/clk.h>
#include <linux/clkdev.h>
#include <linux/of_clk.h>
#include <linux/slab.h>
#include <linux/err.h>
#include <linux/pm_domain.h>
@ -92,8 +93,6 @@ static int __pm_clk_add(struct device *dev, const char *con_id,
if (con_id) {
ce->con_id = kstrdup(con_id, GFP_KERNEL);
if (!ce->con_id) {
dev_err(dev,
"Not enough memory for clock connection ID.\n");
kfree(ce);
return -ENOMEM;
}
@ -195,8 +194,7 @@ int of_pm_clk_add_clks(struct device *dev)
if (!dev || !dev->of_node)
return -EINVAL;
count = of_count_phandle_with_args(dev->of_node, "clocks",
"#clock-cells");
count = of_clk_get_parent_count(dev->of_node);
if (count <= 0)
return -ENODEV;

View File

@ -682,7 +682,7 @@ static int _set_opp_custom(const struct opp_table *opp_table,
data->old_opp.rate = old_freq;
size = sizeof(*old_supply) * opp_table->regulator_count;
if (IS_ERR(old_supply))
if (!old_supply)
memset(data->old_opp.supplies, 0, size);
else
memcpy(data->old_opp.supplies, old_supply, size);
@ -708,7 +708,7 @@ static int _set_required_opps(struct device *dev,
/* Single genpd case */
if (!genpd_virt_devs) {
pstate = opp->required_opps[0]->pstate;
pstate = likely(opp) ? opp->required_opps[0]->pstate : 0;
ret = dev_pm_genpd_set_performance_state(dev, pstate);
if (ret) {
dev_err(dev, "Failed to set performance state of %s: %d (%d)\n",
@ -726,7 +726,7 @@ static int _set_required_opps(struct device *dev,
mutex_lock(&opp_table->genpd_virt_dev_lock);
for (i = 0; i < opp_table->required_opp_count; i++) {
pstate = opp->required_opps[i]->pstate;
pstate = likely(opp) ? opp->required_opps[i]->pstate : 0;
if (!genpd_virt_devs[i])
continue;
@ -748,29 +748,37 @@ static int _set_required_opps(struct device *dev,
* @dev: device for which we do this operation
* @target_freq: frequency to achieve
*
* This configures the power-supplies and clock source to the levels specified
* by the OPP corresponding to the target_freq.
* This configures the power-supplies to the levels specified by the OPP
* corresponding to the target_freq, and programs the clock to a value <=
* target_freq, as rounded by clk_round_rate(). Device wanting to run at fmax
* provided by the opp, should have already rounded to the target OPP's
* frequency.
*/
int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq)
{
struct opp_table *opp_table;
unsigned long freq, old_freq;
unsigned long freq, old_freq, temp_freq;
struct dev_pm_opp *old_opp, *opp;
struct clk *clk;
int ret;
if (unlikely(!target_freq)) {
dev_err(dev, "%s: Invalid target frequency %lu\n", __func__,
target_freq);
return -EINVAL;
}
opp_table = _find_opp_table(dev);
if (IS_ERR(opp_table)) {
dev_err(dev, "%s: device opp doesn't exist\n", __func__);
return PTR_ERR(opp_table);
}
if (unlikely(!target_freq)) {
if (opp_table->required_opp_tables) {
ret = _set_required_opps(dev, opp_table, NULL);
} else {
dev_err(dev, "target frequency can't be 0\n");
ret = -EINVAL;
}
goto put_opp_table;
}
clk = opp_table->clk;
if (IS_ERR(clk)) {
dev_err(dev, "%s: No clock available for the device\n",
@ -793,13 +801,15 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq)
goto put_opp_table;
}
old_opp = _find_freq_ceil(opp_table, &old_freq);
temp_freq = old_freq;
old_opp = _find_freq_ceil(opp_table, &temp_freq);
if (IS_ERR(old_opp)) {
dev_err(dev, "%s: failed to find current OPP for freq %lu (%ld)\n",
__func__, old_freq, PTR_ERR(old_opp));
}
opp = _find_freq_ceil(opp_table, &freq);
temp_freq = freq;
opp = _find_freq_ceil(opp_table, &temp_freq);
if (IS_ERR(opp)) {
ret = PTR_ERR(opp);
dev_err(dev, "%s: failed to find OPP for freq %lu (%d)\n",
@ -1741,91 +1751,137 @@ void dev_pm_opp_unregister_set_opp_helper(struct opp_table *opp_table)
}
EXPORT_SYMBOL_GPL(dev_pm_opp_unregister_set_opp_helper);
static void _opp_detach_genpd(struct opp_table *opp_table)
{
int index;
for (index = 0; index < opp_table->required_opp_count; index++) {
if (!opp_table->genpd_virt_devs[index])
continue;
dev_pm_domain_detach(opp_table->genpd_virt_devs[index], false);
opp_table->genpd_virt_devs[index] = NULL;
}
kfree(opp_table->genpd_virt_devs);
opp_table->genpd_virt_devs = NULL;
}
/**
* dev_pm_opp_set_genpd_virt_dev - Set virtual genpd device for an index
* @dev: Consumer device for which the genpd device is getting set.
* @virt_dev: virtual genpd device.
* @index: index.
* dev_pm_opp_attach_genpd - Attach genpd(s) for the device and save virtual device pointer
* @dev: Consumer device for which the genpd is getting attached.
* @names: Null terminated array of pointers containing names of genpd to attach.
*
* Multiple generic power domains for a device are supported with the help of
* virtual genpd devices, which are created for each consumer device - genpd
* pair. These are the device structures which are attached to the power domain
* and are required by the OPP core to set the performance state of the genpd.
* The same API also works for the case where single genpd is available and so
* we don't need to support that separately.
*
* This helper will normally be called by the consumer driver of the device
* "dev", as only that has details of the genpd devices.
* "dev", as only that has details of the genpd names.
*
* This helper needs to be called once for each of those virtual devices, but
* only if multiple domains are available for a device. Otherwise the original
* device structure will be used instead by the OPP core.
* This helper needs to be called once with a list of all genpd to attach.
* Otherwise the original device structure will be used instead by the OPP core.
*/
struct opp_table *dev_pm_opp_set_genpd_virt_dev(struct device *dev,
struct device *virt_dev,
int index)
struct opp_table *dev_pm_opp_attach_genpd(struct device *dev, const char **names)
{
struct opp_table *opp_table;
struct device *virt_dev;
int index, ret = -EINVAL;
const char **name = names;
opp_table = dev_pm_opp_get_opp_table(dev);
if (!opp_table)
return ERR_PTR(-ENOMEM);
mutex_lock(&opp_table->genpd_virt_dev_lock);
if (unlikely(!opp_table->genpd_virt_devs ||
index >= opp_table->required_opp_count ||
opp_table->genpd_virt_devs[index])) {
dev_err(dev, "Invalid request to set required device\n");
dev_pm_opp_put_opp_table(opp_table);
mutex_unlock(&opp_table->genpd_virt_dev_lock);
return ERR_PTR(-EINVAL);
/*
* If the genpd's OPP table isn't already initialized, parsing of the
* required-opps fail for dev. We should retry this after genpd's OPP
* table is added.
*/
if (!opp_table->required_opp_count) {
ret = -EPROBE_DEFER;
goto put_table;
}
mutex_lock(&opp_table->genpd_virt_dev_lock);
opp_table->genpd_virt_devs = kcalloc(opp_table->required_opp_count,
sizeof(*opp_table->genpd_virt_devs),
GFP_KERNEL);
if (!opp_table->genpd_virt_devs)
goto unlock;
while (*name) {
index = of_property_match_string(dev->of_node,
"power-domain-names", *name);
if (index < 0) {
dev_err(dev, "Failed to find power domain: %s (%d)\n",
*name, index);
goto err;
}
if (index >= opp_table->required_opp_count) {
dev_err(dev, "Index can't be greater than required-opp-count - 1, %s (%d : %d)\n",
*name, opp_table->required_opp_count, index);
goto err;
}
if (opp_table->genpd_virt_devs[index]) {
dev_err(dev, "Genpd virtual device already set %s\n",
*name);
goto err;
}
virt_dev = dev_pm_domain_attach_by_name(dev, *name);
if (IS_ERR(virt_dev)) {
ret = PTR_ERR(virt_dev);
dev_err(dev, "Couldn't attach to pm_domain: %d\n", ret);
goto err;
}
opp_table->genpd_virt_devs[index] = virt_dev;
name++;
}
opp_table->genpd_virt_devs[index] = virt_dev;
mutex_unlock(&opp_table->genpd_virt_dev_lock);
return opp_table;
err:
_opp_detach_genpd(opp_table);
unlock:
mutex_unlock(&opp_table->genpd_virt_dev_lock);
put_table:
dev_pm_opp_put_opp_table(opp_table);
return ERR_PTR(ret);
}
EXPORT_SYMBOL_GPL(dev_pm_opp_attach_genpd);
/**
* dev_pm_opp_put_genpd_virt_dev() - Releases resources blocked for genpd device.
* @opp_table: OPP table returned by dev_pm_opp_set_genpd_virt_dev().
* @virt_dev: virtual genpd device.
* dev_pm_opp_detach_genpd() - Detach genpd(s) from the device.
* @opp_table: OPP table returned by dev_pm_opp_attach_genpd().
*
* This releases the resource previously acquired with a call to
* dev_pm_opp_set_genpd_virt_dev(). The consumer driver shall call this helper
* if it doesn't want OPP core to update performance state of a power domain
* anymore.
* This detaches the genpd(s), resets the virtual device pointers, and puts the
* OPP table.
*/
void dev_pm_opp_put_genpd_virt_dev(struct opp_table *opp_table,
struct device *virt_dev)
void dev_pm_opp_detach_genpd(struct opp_table *opp_table)
{
int i;
/*
* Acquire genpd_virt_dev_lock to make sure virt_dev isn't getting
* used in parallel.
*/
mutex_lock(&opp_table->genpd_virt_dev_lock);
for (i = 0; i < opp_table->required_opp_count; i++) {
if (opp_table->genpd_virt_devs[i] != virt_dev)
continue;
opp_table->genpd_virt_devs[i] = NULL;
dev_pm_opp_put_opp_table(opp_table);
/* Drop the vote */
dev_pm_genpd_set_performance_state(virt_dev, 0);
break;
}
_opp_detach_genpd(opp_table);
mutex_unlock(&opp_table->genpd_virt_dev_lock);
if (unlikely(i == opp_table->required_opp_count))
dev_err(virt_dev, "Failed to find required device entry\n");
dev_pm_opp_put_opp_table(opp_table);
}
EXPORT_SYMBOL_GPL(dev_pm_opp_detach_genpd);
/**
* dev_pm_opp_xlate_performance_state() - Find required OPP's pstate for src_table.

View File

@ -138,7 +138,6 @@ err:
static void _opp_table_free_required_tables(struct opp_table *opp_table)
{
struct opp_table **required_opp_tables = opp_table->required_opp_tables;
struct device **genpd_virt_devs = opp_table->genpd_virt_devs;
int i;
if (!required_opp_tables)
@ -152,10 +151,8 @@ static void _opp_table_free_required_tables(struct opp_table *opp_table)
}
kfree(required_opp_tables);
kfree(genpd_virt_devs);
opp_table->required_opp_count = 0;
opp_table->genpd_virt_devs = NULL;
opp_table->required_opp_tables = NULL;
}
@ -168,9 +165,8 @@ static void _opp_table_alloc_required_tables(struct opp_table *opp_table,
struct device_node *opp_np)
{
struct opp_table **required_opp_tables;
struct device **genpd_virt_devs = NULL;
struct device_node *required_np, *np;
int count, count_pd, i;
int count, i;
/* Traversing the first OPP node is all we need */
np = of_get_next_available_child(opp_np, NULL);
@ -183,33 +179,11 @@ static void _opp_table_alloc_required_tables(struct opp_table *opp_table,
if (!count)
goto put_np;
/*
* Check the number of power-domains to know if we need to deal
* with virtual devices. In some cases we have devices with multiple
* power domains but with only one of them being scalable, hence
* 'count' could be 1, but we still have to deal with multiple genpds
* and virtual devices.
*/
count_pd = of_count_phandle_with_args(dev->of_node, "power-domains",
"#power-domain-cells");
if (!count_pd)
goto put_np;
if (count_pd > 1) {
genpd_virt_devs = kcalloc(count, sizeof(*genpd_virt_devs),
GFP_KERNEL);
if (!genpd_virt_devs)
goto put_np;
}
required_opp_tables = kcalloc(count, sizeof(*required_opp_tables),
GFP_KERNEL);
if (!required_opp_tables) {
kfree(genpd_virt_devs);
if (!required_opp_tables)
goto put_np;
}
opp_table->genpd_virt_devs = genpd_virt_devs;
opp_table->required_opp_tables = required_opp_tables;
opp_table->required_opp_count = count;

View File

@ -899,38 +899,19 @@ static int omap_sr_probe(struct platform_device *pdev)
}
dev_info(&pdev->dev, "%s: SmartReflex driver initialized\n", __func__);
if (!sr_dbg_dir) {
if (!sr_dbg_dir)
sr_dbg_dir = debugfs_create_dir("smartreflex", NULL);
if (IS_ERR_OR_NULL(sr_dbg_dir)) {
ret = PTR_ERR(sr_dbg_dir);
pr_err("%s:sr debugfs dir creation failed(%d)\n",
__func__, ret);
goto err_list_del;
}
}
sr_info->dbg_dir = debugfs_create_dir(sr_info->name, sr_dbg_dir);
if (IS_ERR_OR_NULL(sr_info->dbg_dir)) {
dev_err(&pdev->dev, "%s: Unable to create debugfs directory\n",
__func__);
ret = PTR_ERR(sr_info->dbg_dir);
goto err_debugfs;
}
(void) debugfs_create_file("autocomp", S_IRUGO | S_IWUSR,
sr_info->dbg_dir, (void *)sr_info, &pm_sr_fops);
(void) debugfs_create_x32("errweight", S_IRUGO, sr_info->dbg_dir,
&sr_info->err_weight);
(void) debugfs_create_x32("errmaxlimit", S_IRUGO, sr_info->dbg_dir,
&sr_info->err_maxlimit);
debugfs_create_file("autocomp", S_IRUGO | S_IWUSR, sr_info->dbg_dir,
(void *)sr_info, &pm_sr_fops);
debugfs_create_x32("errweight", S_IRUGO, sr_info->dbg_dir,
&sr_info->err_weight);
debugfs_create_x32("errmaxlimit", S_IRUGO, sr_info->dbg_dir,
&sr_info->err_maxlimit);
nvalue_dir = debugfs_create_dir("nvalue", sr_info->dbg_dir);
if (IS_ERR_OR_NULL(nvalue_dir)) {
dev_err(&pdev->dev, "%s: Unable to create debugfs directory for n-values\n",
__func__);
ret = PTR_ERR(nvalue_dir);
goto err_debugfs;
}
if (sr_info->nvalue_count == 0 || !sr_info->nvalue_table) {
dev_warn(&pdev->dev, "%s: %s: No Voltage table for the corresponding vdd. Cannot create debugfs entries for n-values\n",
@ -945,12 +926,12 @@ static int omap_sr_probe(struct platform_device *pdev)
snprintf(name, sizeof(name), "volt_%lu",
sr_info->nvalue_table[i].volt_nominal);
(void) debugfs_create_x32(name, S_IRUGO | S_IWUSR, nvalue_dir,
&(sr_info->nvalue_table[i].nvalue));
debugfs_create_x32(name, S_IRUGO | S_IWUSR, nvalue_dir,
&(sr_info->nvalue_table[i].nvalue));
snprintf(name, sizeof(name), "errminlimit_%lu",
sr_info->nvalue_table[i].volt_nominal);
(void) debugfs_create_x32(name, S_IRUGO | S_IWUSR, nvalue_dir,
&(sr_info->nvalue_table[i].errminlimit));
debugfs_create_x32(name, S_IRUGO | S_IWUSR, nvalue_dir,
&(sr_info->nvalue_table[i].errminlimit));
}

View File

@ -128,8 +128,8 @@ struct opp_table *dev_pm_opp_set_clkname(struct device *dev, const char * name);
void dev_pm_opp_put_clkname(struct opp_table *opp_table);
struct opp_table *dev_pm_opp_register_set_opp_helper(struct device *dev, int (*set_opp)(struct dev_pm_set_opp_data *data));
void dev_pm_opp_unregister_set_opp_helper(struct opp_table *opp_table);
struct opp_table *dev_pm_opp_set_genpd_virt_dev(struct device *dev, struct device *virt_dev, int index);
void dev_pm_opp_put_genpd_virt_dev(struct opp_table *opp_table, struct device *virt_dev);
struct opp_table *dev_pm_opp_attach_genpd(struct device *dev, const char **names);
void dev_pm_opp_detach_genpd(struct opp_table *opp_table);
int dev_pm_opp_xlate_performance_state(struct opp_table *src_table, struct opp_table *dst_table, unsigned int pstate);
int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq);
int dev_pm_opp_set_sharing_cpus(struct device *cpu_dev, const struct cpumask *cpumask);
@ -292,12 +292,12 @@ static inline struct opp_table *dev_pm_opp_set_clkname(struct device *dev, const
static inline void dev_pm_opp_put_clkname(struct opp_table *opp_table) {}
static inline struct opp_table *dev_pm_opp_set_genpd_virt_dev(struct device *dev, struct device *virt_dev, int index)
static inline struct opp_table *dev_pm_opp_attach_genpd(struct device *dev, const char **names)
{
return ERR_PTR(-ENOTSUPP);
}
static inline void dev_pm_opp_put_genpd_virt_dev(struct opp_table *opp_table, struct device *virt_dev) {}
static inline void dev_pm_opp_detach_genpd(struct opp_table *opp_table) {}
static inline int dev_pm_opp_xlate_performance_state(struct opp_table *src_table, struct opp_table *dst_table, unsigned int pstate)
{

View File

@ -61,7 +61,7 @@ Only display specific monitors. Use the monitor string(s) provided by \-l option
.PP
\-i seconds
.RS 4
Measure intervall.
Measure interval.
.RE
.PP
\-c

View File

@ -98,7 +98,7 @@ msgstr ""
#: utils/idle_monitor/cpupower-monitor.c:74
#, c-format
msgid "\t -i: time intervall to measure for in seconds (default 1)\n"
msgid "\t -i: time interval to measure for in seconds (default 1)\n"
msgstr ""
#: utils/idle_monitor/cpupower-monitor.c:75

View File

@ -95,7 +95,7 @@ msgstr ""
#: utils/idle_monitor/cpupower-monitor.c:74
#, c-format
msgid "\t -i: time intervall to measure for in seconds (default 1)\n"
msgid "\t -i: time interval to measure for in seconds (default 1)\n"
msgstr ""
#: utils/idle_monitor/cpupower-monitor.c:75

View File

@ -95,7 +95,7 @@ msgstr ""
#: utils/idle_monitor/cpupower-monitor.c:74
#, c-format
msgid "\t -i: time intervall to measure for in seconds (default 1)\n"
msgid "\t -i: time interval to measure for in seconds (default 1)\n"
msgstr ""
#: utils/idle_monitor/cpupower-monitor.c:75

View File

@ -95,7 +95,7 @@ msgstr ""
#: utils/idle_monitor/cpupower-monitor.c:74
#, c-format
msgid "\t -i: time intervall to measure for in seconds (default 1)\n"
msgid "\t -i: time interval to measure for in seconds (default 1)\n"
msgstr ""
#: utils/idle_monitor/cpupower-monitor.c:75

View File

@ -93,7 +93,7 @@ msgstr ""
#: utils/idle_monitor/cpupower-monitor.c:74
#, c-format
msgid "\t -i: time intervall to measure for in seconds (default 1)\n"
msgid "\t -i: time interval to measure for in seconds (default 1)\n"
msgstr ""
#: utils/idle_monitor/cpupower-monitor.c:75

View File

@ -305,6 +305,8 @@ int cmd_freq_set(int argc, char **argv)
bitmask_setbit(cpus_chosen, cpus->cpu);
cpus = cpus->next;
}
/* Set the last cpu in related cpus list */
bitmask_setbit(cpus_chosen, cpus->cpu);
cpufreq_put_related_cpus(cpus);
}
}

552
tools/power/pm-graph/README Normal file
View File

@ -0,0 +1,552 @@
p m - g r a p h
pm-graph: suspend/resume/boot timing analysis tools
Version: 5.4
Author: Todd Brandt <todd.e.brandt@intel.com>
Home Page: https://01.org/pm-graph
Report bugs/issues at bugzilla.kernel.org Tools/pm-graph
- https://bugzilla.kernel.org/buglist.cgi?component=pm-graph&product=Tools
Full documentation available online & in man pages
- Getting Started:
https://01.org/pm-graph/documentation/getting-started
- Config File Format:
https://01.org/pm-graph/documentation/3-config-file-format
- upstream version in git:
https://github.com/intel/pm-graph/
Table of Contents
- Overview
- Setup
- Usage
- Basic Usage
- Dev Mode Usage
- Proc Mode Usage
- Configuration Files
- Usage Examples
- Config File Options
- Custom Timeline Entries
- Adding/Editing Timeline Functions
- Adding/Editing Dev Timeline Source Functions
- Verifying your Custom Functions
- Testing on consumer linux Operating Systems
- Android
------------------------------------------------------------------
| OVERVIEW |
------------------------------------------------------------------
This tool suite is designed to assist kernel and OS developers in optimizing
their linux stack's suspend/resume & boot time. Using a kernel image built
with a few extra options enabled, the tools will execute a suspend or boot,
and will capture dmesg and ftrace data. This data is transformed into a set of
timelines and a callgraph to give a quick and detailed view of which devices
and kernel processes are taking the most time in suspend/resume & boot.
------------------------------------------------------------------
| SETUP |
------------------------------------------------------------------
These packages are required to execute the scripts
- python
- python-requests
Ubuntu:
sudo apt-get install python python-requests
Fedora:
sudo dnf install python python-requests
The tools can most easily be installed via git clone and make install
$> git clone http://github.com/intel/pm-graph.git
$> cd pm-graph
$> sudo make install
$> man sleepgraph ; man bootgraph
Setup involves some minor kernel configuration
The following kernel build options are required for all kernels:
CONFIG_DEVMEM=y
CONFIG_PM_DEBUG=y
CONFIG_PM_SLEEP_DEBUG=y
CONFIG_FTRACE=y
CONFIG_FUNCTION_TRACER=y
CONFIG_FUNCTION_GRAPH_TRACER=y
CONFIG_KPROBES=y
CONFIG_KPROBES_ON_FTRACE=y
In kernel 3.15.0, two patches were upstreamed which enable the
v3.0 behavior. These patches allow the tool to read all the
data from trace events instead of from dmesg. You can enable
this behavior on earlier kernels with these patches:
(kernel/pre-3.15/enable_trace_events_suspend_resume.patch)
(kernel/pre-3.15/enable_trace_events_device_pm_callback.patch)
If you're using a kernel older than 3.15.0, the following
additional kernel parameters are required:
(e.g. in file /etc/default/grub)
GRUB_CMDLINE_LINUX_DEFAULT="... initcall_debug log_buf_len=32M ..."
If you're using a kernel older than 3.11-rc2, the following simple
patch must be applied to enable ftrace data:
in file: kernel/power/suspend.c
in function: int suspend_devices_and_enter(suspend_state_t state)
remove call to "ftrace_stop();"
remove call to "ftrace_start();"
There is a patch which does this for kernel v3.8.0:
(kernel/pre-3.11-rc2/enable_ftrace_in_suspendresume.patch)
------------------------------------------------------------------
| USAGE |
------------------------------------------------------------------
Basic Usage
___________
1) First configure a kernel using the instructions from the previous sections.
Then build, install, and boot with it.
2) Open up a terminal window and execute the mode list command:
%> sudo ./sleepgraph.py -modes
['freeze', 'mem', 'disk']
Execute a test using one of the available power modes, e.g. mem (S3):
%> sudo ./sleepgraph.py -m mem -rtcwake 15
or with a config file
%> sudo ./sleepgraph.py -config config/suspend.cfg
When the system comes back you'll see the script finishing up and
creating the output files in the test subdir. It generates output
files in subdirectory: suspend-mmddyy-HHMMSS. The ftrace file can
be used to regenerate the html timeline with different options
HTML output: <hostname>_<mode>.html
raw dmesg output: <hostname>_<mode>_dmesg.txt
raw ftrace output: <hostname>_<mode>_ftrace.txt
View the html in firefox or chrome.
Dev Mode Usage
______________
Developer mode adds information on low level source calls to the timeline.
The tool sets kprobes on all delay and mutex calls to see which devices
are waiting for something and when. It also sets a suite of kprobes on
subsystem dependent calls to better fill out the timeline.
The tool will also expose kernel threads that don't normally show up in the
timeline. This is useful in discovering dependent threads to get a better
idea of what each device is waiting for. For instance, the scsi_eh thread,
a.k.a. scsi resume error handler, is what each SATA disk device waits for
before it can continue resume.
The timeline will be much larger if run with dev mode, so it can be useful
to set the -mindev option to clip out any device blocks that are too small
to see easily. The following command will give a nice dev mode run:
%> sudo ./sleepgraph.py -m mem -rtcwake 15 -mindev 1 -dev
or with a config file
%> sudo ./sleepgraph.py -config config/suspend-dev.cfg
Proc Mode Usage
_______________
Proc mode adds user process info to the timeline. This is done in a manner
similar to the bootchart utility, which graphs init processes and their
execution as the system boots. This tool option does the same thing but for
the period before and after suspend/resume.
In order to see any process info, there needs to be some delay before or
after resume since processes are frozen in suspend_prepare and thawed in
resume_complete. The predelay and postdelay args allow you to do this. It
can also be useful to run in x2 mode with an x2 delay, this way you can
see process activity before and after resume, and in between two
successive suspend/resumes.
The command can be run like this:
%> sudo ./sleepgraph.py -m mem -rtcwake 15 -x2 -x2delay 1000 -predelay 1000 -postdelay 1000 -proc
or with a config file
%> sudo ./sleepgraph.py -config config/suspend-proc.cfg
------------------------------------------------------------------
| CONFIGURATION FILES |
------------------------------------------------------------------
Since 4.0 we've moved to using config files in lieu of command line options.
The config folder contains a collection of typical use cases.
There are corresponding configs for other power modes:
Simple suspend/resume with basic timeline (mem/freeze/standby)
config/suspend.cfg
config/freeze.cfg
config/standby.cfg
Dev mode suspend/resume with dev timeline (mem/freeze/standby)
config/suspend-dev.cfg
config/freeze-dev.cfg
config/standby-dev.cfg
Simple suspend/resume with timeline and callgraph (mem/freeze/standby)
config/suspend-callgraph.cfg
config/freeze-callgraph.cfg
config/standby-callgraph.cfg
Sample proc mode x2 run using mem suspend
config/suspend-x2-proc.cfg
Sample for editing timeline funcs (moves internal functions into config)
config/custom-timeline-functions.cfg
Sample debug config for serio subsystem
config/debug-serio-suspend.cfg
Usage Examples
______________
Run a simple mem suspend:
%> sudo ./sleepgraph.py -config config/suspend.cfg
Run a mem suspend with callgraph data:
%> sudo ./sleepgraph.py -config config/suspend-callgraph.cfg
Run a mem suspend with dev mode detail:
%> sudo ./sleepgraph.py -config config/suspend-dev.cfg
Config File Options
___________________
[Settings]
# Verbosity: print verbose messages (def: false)
verbose: false
# Suspend Mode: e.g. standby, mem, freeze, disk (def: mem)
mode: mem
# Output Directory Format: {hostname}, {date}, {time} give current values
output-dir: suspend-{hostname}-{date}-{time}
# Automatic Wakeup: use rtcwake to wakeup after X seconds (def: infinity)
rtcwake: 15
# Add Logs: add the dmesg and ftrace log to the html output (def: false)
addlogs: false
# Sus/Res Gap: insert a gap between sus & res in the timeline (def: false)
srgap: false
# Custom Command: Command to execute in lieu of suspend (def: "")
command: echo mem > /sys/power/state
# Proc mode: graph user processes and cpu usage in the timeline (def: false)
proc: false
# Dev mode: graph source functions in the timeline (def: false)
dev: false
# Suspend/Resume x2: run 2 suspend/resumes back to back (def: false)
x2: false
# x2 Suspend Delay: time delay between the two test runs in ms (def: 0 ms)
x2delay: 0
# Pre Suspend Delay: nclude an N ms delay before (1st) suspend (def: 0 ms)
predelay: 0
# Post Resume Delay: include an N ms delay after (last) resume (def: 0 ms)
postdelay: 0
# Min Device Length: graph only dev callbacks longer than min (def: 0.001 ms)
mindev: 0.001
# Callgraph: gather ftrace callgraph data on all timeline events (def: false)
callgraph: false
# Expand Callgraph: pre-expand the callgraph treeviews in html (def: false)
expandcg: false
# Min Callgraph Length: show callgraphs only if longer than min (def: 1 ms)
mincg: 1
# Timestamp Precision: number of sig digits in timestamps (0:S, [3:ms], 6:us)
timeprec: 3
# Device Filter: show only devs whose name/driver includes one of these strings
devicefilter: _cpu_up,_cpu_down,i915,usb
# Override default timeline entries:
# Do not use the internal default functions for timeline entries (def: false)
# Set this to true if you intend to only use the ones defined in the config
override-timeline-functions: true
# Override default dev timeline entries:
# Do not use the internal default functions for dev timeline entries (def: false)
# Set this to true if you intend to only use the ones defined in the config
override-dev-timeline-functions: true
# Call Loop Max Gap (dev mode only)
# merge loops of the same call if each is less than maxgap apart (def: 100us)
callloop-maxgap: 0.0001
# Call Loop Max Length (dev mode only)
# merge loops of the same call if each is less than maxlen in length (def: 5ms)
callloop-maxlen: 0.005
------------------------------------------------------------------
| CUSTOM TIMELINE ENTRIES |
------------------------------------------------------------------
Adding or Editing Timeline Functions
____________________________________
The tool uses an array of function names to fill out empty spaces in the
timeline where device callbacks don't appear. For instance, in suspend_prepare
the tool adds the sys_sync and freeze_processes calls as virtual device blocks
in the timeline to show you where the time is going. These calls should fill
the timeline with contiguous data so that most kernel execution is covered.
It is possible to add new function calls to the timeline by adding them to
the config. It's also possible to copy the internal timeline functions into
the config so that you can override and edit them. Place them in the
timeline_functions_ARCH section with the name of your architecture appended.
i.e. for x86_64: [timeline_functions_x86_64]
Use the override-timeline-functions option if you only want to use your
custom calls, or leave it false to append them to the internal ones.
This section includes a list of functions (set using kprobes) which use both
symbol data and function arg data. The args are pulled directly from the
stack using this architecture's registers and stack formatting. Each entry
can include up to four pieces of info: The function name, a format string,
an argument list, and a color. But only a function name is required.
For a full example config, see config/custom-timeline-functions.cfg. It pulls
all the internal timeline functions into the config and allows you to edit
them.
Entry format:
function: format{fn_arg1}_{fn_arg2} fn_arg1 fn_arg2 ... [color=purple]
Required Arguments:
function: The symbol name for the function you want probed, this is the
minimum required for an entry, it will show up as the function
name with no arguments.
example: _cpu_up:
Optional Arguments:
format: The format to display the data on the timeline in. Use braces to
enclose the arg names.
example: CPU_ON[{cpu}]
color: The color of the entry block in the timeline. The default color is
transparent, so the entry shares the phase color. The color is an
html color string, either a word, or an RGB.
example: [color=#CC00CC]
arglist: A list of arguments from registers/stack addresses. See URL:
https://www.kernel.org/doc/Documentation/trace/kprobetrace.txt
example: cpu=%di:s32
Here is a full example entry. It displays cpu resume calls in the timeline
in orange. They will appear as CPU_ON[0], CPU_ON[1], etc.
[timeline_functions_x86_64]
_cpu_up: CPU_ON[{cpu}] cpu=%di:s32 [color=orange]
Adding or Editing Dev Mode Timeline Source Functions
____________________________________________________
In dev mode, the tool uses an array of function names to monitor source
execution within the timeline entries.
The function calls are displayed inside the main device/call blocks in the
timeline. However, if a function call is not within a main timeline event,
it will spawn an entirely new event named after the caller's kernel thread.
These asynchronous kernel threads will populate in a separate section
beneath the main device/call section.
The tool has a set of hard coded calls which focus on the most common use
cases: msleep, udelay, schedule_timeout, mutex_lock_slowpath, etc. These are
the functions that add a hardcoded time delay to the suspend/resume path.
The tool also includes some common functions native to important
subsystems: ata, i915, and ACPI, etc.
It is possible to add new function calls to the dev timeline by adding them
to the config. It's also possible to copy the internal dev timeline
functions into the config so that you can override and edit them. Place them
in the dev_timeline_functions_ARCH section with the name of your architecture
appended. i.e. for x86_64: [dev_timeline_functions_x86_64]
Use the override-dev-timeline-functions option if you only want to use your
custom calls, or leave it false to append them to the internal ones.
The format is the same as the timeline_functions_x86_64 section. It's a
list of functions (set using kprobes) which use both symbol data and function
arg data. The args are pulled directly from the stack using this
architecture's registers and stack formatting. Each entry can include up
to four pieces of info: The function name, a format string, an argument list,
and a color. But only the function name is required.
For a full example config, see config/custom-timeline-functions.cfg. It pulls
all the internal dev timeline functions into the config and allows you to edit
them.
Here is a full example entry. It displays the ATA port reset calls as
ataN_port_reset in the timeline. This is where most of the SATA disk resume
time goes, so it can be helpful to see the low level call.
[dev_timeline_functions_x86_64]
ata_eh_recover: ata{port}_port_reset port=+36(%di):s32 [color=#CC00CC]
Verifying your custom functions
_______________________________
Once you have a set of functions (kprobes) defined, it can be useful to
perform a quick check to see if you formatted them correctly and if the system
actually supports them. To do this, run the tool with your config file
and the -status option. The tool will go through all the kprobes (both
custom and internal if you haven't overridden them) and actually attempts
to set them in ftrace. It will then print out success or fail for you.
Note that kprobes which don't actually exist in the kernel won't stop the
tool, they just wont show up.
For example:
sudo ./sleepgraph.py -config config/custom-timeline-functions.cfg -status
Checking this system (myhostname)...
have root access: YES
is sysfs mounted: YES
is "mem" a valid power mode: YES
is ftrace supported: YES
are kprobes supported: YES
timeline data source: FTRACE (all trace events found)
is rtcwake supported: YES
verifying timeline kprobes work:
_cpu_down: YES
_cpu_up: YES
acpi_pm_finish: YES
acpi_pm_prepare: YES
freeze_kernel_threads: YES
freeze_processes: YES
sys_sync: YES
thaw_processes: YES
verifying dev kprobes work:
__const_udelay: YES
__mutex_lock_slowpath: YES
acpi_os_stall: YES
acpi_ps_parse_aml: YES
intel_opregion_init: NO
intel_opregion_register: NO
intel_opregion_setup: NO
msleep: YES
schedule_timeout: YES
schedule_timeout_uninterruptible: YES
usleep_range: YES
------------------------------------------------------------------
| TESTING ON CONSUMER LINUX OPERATING SYSTEMS |
------------------------------------------------------------------
Android
_______
The easiest way to execute on an android device is to run the android.sh
script on the device, then pull the ftrace log back to the host and run
sleepgraph.py on it.
Here are the steps:
[download and install the tool on the device]
host%> wget https://raw.githubusercontent.com/intel/pm-graph/master/tools/android.sh
host%> adb connect 192.168.1.6
host%> adb root
# push the script to a writeable location
host%> adb push android.sh /sdcard/
[check whether the tool will run on your device]
host%> adb shell
dev%> cd /sdcard
dev%> sh android.sh status
host : asus_t100
kernel : 3.14.0-i386-dirty
modes : freeze mem
rtcwake : supported
ftrace : supported
trace events {
suspend_resume: found
device_pm_callback_end: found
device_pm_callback_start: found
}
# the above is what you see on a system that's properly patched
[execute the suspend]
# NOTE: The suspend will only work if the screen isn't timed out,
# so you have to press some keys first to wake it up b4 suspend)
dev%> sh android.sh suspend mem
------------------------------------
Suspend/Resume timing test initiated
------------------------------------
hostname : asus_t100
kernel : 3.14.0-i386-dirty
mode : mem
ftrace out : /mnt/shell/emulated/0/ftrace.txt
dmesg out : /mnt/shell/emulated/0/dmesg.txt
log file : /mnt/shell/emulated/0/log.txt
------------------------------------
INITIALIZING FTRACE........DONE
STARTING FTRACE
SUSPEND START @ 21:24:02 (rtcwake in 10 seconds)
<adb connection will now terminate>
[retrieve the data from the device]
# I find that you have to actually kill the adb process and
# reconnect sometimes in order for the connection to work post-suspend
host%> adb connect 192.168.1.6
# (required) get the ftrace data, this is the most important piece
host%> adb pull /sdcard/ftrace.txt
# (optional) get the dmesg data, this is for debugging
host%> adb pull /sdcard/dmesg.txt
# (optional) get the log, which just lists some test times for comparison
host%> adb pull /sdcard/log.txt
[create an output html file using sleepgraph.py]
host%> sleepgraph.py -ftrace ftrace.txt
You should now have an output.html with the android data, enjoy!

View File

@ -325,9 +325,9 @@ def parseKernelLog():
if(not sysvals.stamp['kernel']):
sysvals.stamp['kernel'] = sysvals.kernelVersion(msg)
continue
m = re.match('.* setting system clock to (?P<t>.*) UTC.*', msg)
m = re.match('.* setting system clock to (?P<d>[0-9\-]*)[ A-Z](?P<t>[0-9:]*) UTC.*', msg)
if(m):
bt = datetime.strptime(m.group('t'), '%Y-%m-%d %H:%M:%S')
bt = datetime.strptime(m.group('d')+' '+m.group('t'), '%Y-%m-%d %H:%M:%S')
bt = bt - timedelta(seconds=int(ktime))
data.boottime = bt.strftime('%Y-%m-%d_%H:%M:%S')
sysvals.stamp['time'] = bt.strftime('%B %d %Y, %I:%M:%S %p')
@ -348,7 +348,7 @@ def parseKernelLog():
data.newAction(phase, f, pid, start, ktime, int(r), int(t))
del devtemp[f]
continue
if(re.match('^Freeing unused kernel memory.*', msg)):
if(re.match('^Freeing unused kernel .*', msg)):
data.tUserMode = ktime
data.dmesg['kernel']['end'] = ktime
data.dmesg['user']['start'] = ktime
@ -1008,7 +1008,7 @@ if __name__ == '__main__':
updateKernelParams()
elif cmd == 'flistall':
for f in sysvals.getBootFtraceFilterFunctions():
print f
print(f)
elif cmd == 'checkbl':
sysvals.getBootLoader()
pprint('Boot Loader: %s\n%s' % (sysvals.bootloader, sysvals.blexec))

View File

@ -98,12 +98,34 @@ postdelay: 0
# graph only devices longer than min in the timeline (default: 0.001 ms)
mindev: 0.001
# Call Loop Max Gap (dev mode only)
# merge loops of the same call if each is less than maxgap apart (def: 100us)
callloop-maxgap: 0.0001
# Call Loop Max Length (dev mode only)
# merge loops of the same call if each is less than maxlen in length (def: 5ms)
callloop-maxlen: 0.005
# Override default timeline entries:
# Do not use the internal default functions for timeline entries (def: false)
# Set this to true if you intend to only use the ones defined in the config
override-timeline-functions: true
# Override default dev timeline entries:
# Do not use the internal default functions for dev timeline entries (def: false)
# Set this to true if you intend to only use the ones defined in the config
override-dev-timeline-functions: true
# ---- Debug Options ----
# Callgraph
# gather detailed ftrace callgraph data on all timeline events (default: false)
callgraph: false
# Max graph depth
# limit the callgraph trace to this depth (default: 0 = all)
maxdepth: 2
# Callgraph phase filter
# Only enable callgraphs for one phase, i.e. resume_noirq (default: all)
cgphase: suspend
@ -131,3 +153,7 @@ timeprec: 6
# Add kprobe functions to the timeline
# Add functions to the timeline from a text file (default: no-action)
# fadd: file.txt
# Ftrace buffer size
# Set trace buffer size to N kilo-bytes (default: all of free memory up to 3GB)
# bufsize: 1000

View File

@ -53,6 +53,11 @@ disable rtcwake and require a user keypress to resume.
Add the dmesg and ftrace logs to the html output. They will be viewable by
clicking buttons in the timeline.
.TP
\fB-turbostat\fR
Use turbostat to execute the command in freeze mode (default: disabled). This
will provide turbostat output in the log which will tell you which actual
power modes were entered.
.TP
\fB-result \fIfile\fR
Export a results table to a text file for parsing.
.TP
@ -121,6 +126,10 @@ be created in a new subdirectory with a summary page: suspend-xN-{date}-{time}.
Use ftrace to create device callgraphs (default: disabled). This can produce
very large outputs, i.e. 10MB - 100MB.
.TP
\fB-ftop\fR
Use ftrace on the top level call: "suspend_devices_and_enter" only (default: disabled).
This option implies -f and creates a single callgraph covering all of suspend/resume.
.TP
\fB-maxdepth \fIlevel\fR
limit the callgraph trace depth to \fIlevel\fR (default: 0=all). This is
the best way to limit the output size when using callgraphs via -f.
@ -138,8 +147,8 @@ which are barely visible in the timeline.
The value is a float: e.g. 0.001 represents 1 us.
.TP
\fB-cgfilter \fI"func1,func2,..."\fR
Reduce callgraph output in the timeline by limiting it to a list of calls. The
argument can be a single function name or a comma delimited list.
Reduce callgraph output in the timeline by limiting it certain devices. The
argument can be a single device name or a comma delimited list.
(default: none)
.TP
\fB-cgskip \fIfile\fR
@ -183,6 +192,9 @@ Print out the contents of the ACPI Firmware Performance Data Table.
\fB-battery\fR
Print out battery status and current charge.
.TP
\fB-wifi\fR
Print out wifi status and connection details.
.TP
\fB-xon/-xoff/-xstandby/-xsuspend\fR
Test xset by attempting to switch the display to the given mode. This
is the same command which will be issued by \fB-display \fImode\fR.

File diff suppressed because it is too large Load Diff