kernel/linux.git/drivers/cpufreq/cpufreq_stats.c, branch v6.3

cpufreq: stats: Convert to use sysfs_emit_at() API

2022-12-06T11:20:07+00:00

Follow the advice of the Documentation/filesystems/sysfs.rst and show() should only use sysfs_emit() or sysfs_emit_at() when formatting the value to be returned to user space. Signed-off-by: ye xingchen Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki

cpufreq: stats: Clean up local variable in cpufreq_stats_create_table()

2021-06-07T11:47:57+00:00

Local variable 'count' will be initialized and 'ret' is also not required, so remove the redundant initialization and get rid of 'ret'. Signed-off-by: Shaokun Zhang Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki

cpufreq: stats: Use local_clock() instead of jiffies

2020-12-11T18:53:58+00:00

local_clock() has better precision and accuracy as compared to jiffies, lets use it for time management in cpufreq stats. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki

cpufreq: stats: Fix string format specifier mismatch

2020-10-12T10:32:10+00:00

Fix following warning: drivers/cpufreq/cpufreq_stats.c:63:10: warning: %d in format string (no. 1) requires 'int' but the argument type is 'unsigned int' Fixes: 40c3bd4cfa6f ("cpufreq: stats: Defer stats update to cpufreq_stats_record_transition()") Reported-by: kernel test robot Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki

cpufreq: stats: Add memory barrier to store_reset()

2020-10-07T15:13:15+00:00

There is nothing to prevent the CPU or the compiler from reordering the writes to stats->reset_time and stats->reset_pending in store_reset(), in which case the readers of stats->reset_time may see a stale value. Moreover, on 32-bit arches the write to reset_time cannot be completed in one go, so the readers of it may see a partially updated value in that case. To prevent that from happening, add a write memory barrier between the writes to stats->reset_time and stats->reset_pending in store_reset() and corresponding read memory barrier in the readers of stats->reset_time. Fixes: 40c3bd4cfa6f ("cpufreq: stats: Defer stats update to cpufreq_stats_record_transition()") Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar

cpufreq: stats: Enable stats for fast-switch as well

2020-10-05T13:13:43+00:00

Now that all the blockers are gone for enabling stats in fast-switching case, enable it. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki

cpufreq: stats: Mark few conditionals with unlikely()

2020-10-05T13:13:43+00:00

Since this will be part of the scheduler's hotpath in some cases, use unlikely() for few of the obvious conditionals. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki

cpufreq: stats: Remove locking

2020-10-05T13:13:43+00:00

The locking isn't required anymore as stats can get updated only from one place at a time. Remove it. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki

cpufreq: stats: Defer stats update to cpufreq_stats_record_transition()

2020-10-05T13:13:42+00:00

In order to prepare for lock-less stats update, add support to defer any updates to it until cpufreq_stats_record_transition() is called. The stats were updated from two places earlier: - show_time_in_state(): This can be easily deferred, all we need is to calculate the delta duration again in this routine to show the current state's time-in-state. - store_reset(): This is a bit tricky as we need to clear the stats here and avoid races with simultaneous call to cpufreq_stats_record_transition(). Fix that by deferring the reset of the stats (within the code) to the next call to cpufreq_stats_record_transition(), but since we need to keep showing the right stats until that time, we capture the reset time and account for the time since last time reset was called until the time cpufreq_stats_record_transition() update the stats. User space will continue seeing the stats correctly, everything will be 0 after the stats are reset, apart from the time-in-state of the current state, until the time a frequency switch happens. Signed-off-by: Viresh Kumar [ rjw: Minor changelog edits ] Signed-off-by: Rafael J. Wysocki

cpufreq: Use scnprintf() for avoiding potential buffer overflow

2020-03-14T11:07:38+00:00

Since snprintf() returns the would-be-output size instead of the actual output size, the succeeding calls may go beyond the given buffer limit. Fix it by replacing with scnprintf(). Signed-off-by: Takashi Iwai Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki