In the current code, if we fail during a frequency transition we simply send the POSTCHANGE notification with old frequency. This isn't enough.

One of the core user of these notifications is the code responsible for keeping loops_per_jiffy aligned with frequency change. And mostly it is written as:

if ((val == CPUFREQ_PRECHANGE && freq->old < freq->new) || (val == CPUFREQ_POSTCHANGE && freq->old > freq->new)) { update-loops-per-jiffy... }

So, suppose we are changing to a higher frequency and failed during transition, then following will happen: - CPUFREQ_PRECHANGE notification with freq-new > freq-old - CPUFREQ_POSTCHANGE notification with freq-new == freq-old

The first one will update loops_per_jiffy and second one will do nothing. Even if we send the 2nd notification by exchanging values of freq-new and old, some users of these notifications might get unstable.

This can be fixed by simply calling cpufreq_notify_post_transition() with error code and this routine will take care of sending notifications in correct order.

Signed-off-by: Viresh Kumar viresh.kumar@linaro.org --- drivers/cpufreq/cpufreq.c | 13 ++----------- 1 file changed, 2 insertions(+), 11 deletions(-)

diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index a862aa9..557bb49 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -1764,17 +1764,8 @@ int __cpufreq_driver_target(struct cpufreq_policy *policy, pr_err("%s: Failed to change cpu frequency: %d\n", __func__, retval);

- if (notify) { - /* - * Notify with old freq in case we failed to change - * frequency - */ - if (retval) - freqs.new = freqs.old; - - cpufreq_notify_transition(policy, &freqs, - CPUFREQ_POSTCHANGE); - } + if (notify) + cpufreq_notify_post_transition(policy, &freqs, retval); }

out:

In the current code, if we fail during a frequency transition we simply send the POSTCHANGE notification with old frequency. This isn't enough.

One of the core user of these notifications is the code responsible for keeping loops_per_jiffy aligned with frequency change. And mostly it is written as:

if ((val == CPUFREQ_PRECHANGE && freq->old < freq->new) || (val == CPUFREQ_POSTCHANGE && freq->old > freq->new)) { update-loops-per-jiffy... }

So, suppose we are changing to a higher frequency and failed during transition, then following will happen: - CPUFREQ_PRECHANGE notification with freq-new > freq-old - CPUFREQ_POSTCHANGE notification with freq-new == freq-old

The first one will update loops_per_jiffy and second one will do nothing. Even if we send the 2nd notification by exchanging values of freq-new and old, some users of these notifications might get unstable.

This can be fixed by simply calling cpufreq_notify_post_transition() with error code and this routine will take care of sending notifications in correct order.

Signed-off-by: Viresh Kumar viresh.kumar@linaro.org --- drivers/cpufreq/powernow-k8.c | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-)

diff --git a/drivers/cpufreq/powernow-k8.c b/drivers/cpufreq/powernow-k8.c index 0023c7d..e10b646 100644 --- a/drivers/cpufreq/powernow-k8.c +++ b/drivers/cpufreq/powernow-k8.c @@ -964,14 +964,9 @@ static int transition_frequency_fidvid(struct powernow_k8_data *data, cpufreq_cpu_put(policy);

cpufreq_notify_transition(policy, &freqs, CPUFREQ_PRECHANGE); - res = transition_fid_vid(data, fid, vid); - if (res) - freqs.new = freqs.old; - else - freqs.new = find_khz_freq_from_fid(data->currfid); + cpufreq_notify_post_transition(policy, &freqs, res);

- cpufreq_notify_transition(policy, &freqs, CPUFREQ_POSTCHANGE); return res; }

On Saturday, November 30, 2013 09:26:19 PM Viresh Kumar wrote:

This introduces another routine cpufreq_notify_post_transition() which can be used to send POSTCHANGE notification for new freq with or without both {PRE|POST}CHANGE notifications for last freq in case of failures. This is useful at multiple places, specially for sending transition failure notifications.

Signed-off-by: Viresh Kumar viresh.kumar@linaro.org

Hi Rafael,

Please see if you want to take it for 3.13 or 14, as this fixes bugs which are partly introduced in 3.13..

Which ones?

Rafael

drivers/cpufreq/cpufreq.c | 14 ++++++++++++++ include/linux/cpufreq.h | 2 ++ 2 files changed, 16 insertions(+)

diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index 606224a..a862aa9 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -324,6 +324,20 @@ void cpufreq_notify_transition(struct cpufreq_policy *policy, } EXPORT_SYMBOL_GPL(cpufreq_notify_transition); +/* Do post notifications when there are chances that transition has failed */ +void cpufreq_notify_post_transition(struct cpufreq_policy *policy,

• struct cpufreq_freqs *freqs, int transition_failed)
  

+{

• cpufreq_notify_transition(policy, freqs, CPUFREQ_POSTCHANGE);
• if (!transition_failed)

• return;
  

• swap(freqs->old, freqs->new);
• cpufreq_notify_transition(policy, freqs, CPUFREQ_PRECHANGE);
• cpufreq_notify_transition(policy, freqs, CPUFREQ_POSTCHANGE);

+} +EXPORT_SYMBOL_GPL(cpufreq_notify_post_transition);

/*********************************************************************

•                      SYSFS INTERFACE                          *
  

diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h index ee5fe9d..57e48db 100644 --- a/include/linux/cpufreq.h +++ b/include/linux/cpufreq.h @@ -314,6 +314,8 @@ int cpufreq_unregister_notifier(struct notifier_block *nb, unsigned int list); void cpufreq_notify_transition(struct cpufreq_policy *policy, struct cpufreq_freqs *freqs, unsigned int state); +void cpufreq_notify_post_transition(struct cpufreq_policy *policy,

• struct cpufreq_freqs *freqs, int transition_failed);
  

#else /* CONFIG_CPU_FREQ */ static inline int cpufreq_register_notifier(struct notifier_block *nb,