Hi,

On Friday, August 07, 2015 09:11:28 AM Viresh Kumar wrote:

Hi,

I would suggest you sending such patches as reply to the earlier
threads only, instead of a new chain. This will save your time.

Please explain it more.  This patch needs to be first for cpufreq-dt
switch to be complete.  scaling_boost_freqs is available currently
for exynos-cpufreq users when boost is enabled and ideally we would
like it to be available immediately after the switch to cpufreq-dt.

For example, you will need to resend other patches unnecessarily if I
NAK this patch :)

On 06-08-15, 15:41, Bartlomiej Zolnierkiewicz wrote:

quoted

Add cpufreq_boost_enabled_generic_attr table and use it in
cpufreq-dt driver instead of cpufreq_generic_attr one when
boost support is enabled.  As a result scaling_boost_freqs
sysfs attribute is available when cpufreq-dt driver is
used and boost support is enabled.

Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Thomas Abraham <redacted>
Cc: Javier Martinez Canillas <redacted>
Cc: Krzysztof Kozlowski <redacted>
Signed-off-by: Bartlomiej Zolnierkiewicz <redacted>
---
 drivers/cpufreq/cpufreq-dt.c | 3 +++
 drivers/cpufreq/freq_table.c | 7 +++++++
 include/linux/cpufreq.h      | 1 +
 3 files changed, 11 insertions(+)

diff --git a/drivers/cpufreq/cpufreq-dt.c b/drivers/cpufreq/cpufreq-dt.c
index b9259ab..7b091c0 100644
--- a/drivers/cpufreq/cpufreq-dt.c
+++ b/drivers/cpufreq/cpufreq-dt.c

@@ -182,6 +182,8 @@ try_again:
 	return ret;
 }
 
+static struct cpufreq_driver dt_cpufreq_driver;
+
 static int cpufreq_init(struct cpufreq_policy *policy)
 {
 	struct cpufreq_frequency_table *freq_table;

@@ -336,6 +338,7 @@ static int cpufreq_init(struct cpufreq_policy *policy)
 		ret = cpufreq_enable_boost_support();
 		if (ret)
 			goto out_free_cpufreq_table;
+		dt_cpufreq_driver.attr = cpufreq_boost_enabled_generic_attr;
 	}
 
 	policy->cpuinfo.transition_latency = transition_latency;

diff --git a/drivers/cpufreq/freq_table.c b/drivers/cpufreq/freq_table.c
index a8f1daf..5a00a09 100644
--- a/drivers/cpufreq/freq_table.c
+++ b/drivers/cpufreq/freq_table.c

@@ -300,6 +300,13 @@ struct freq_attr *cpufreq_generic_attr[] = {
 };
 EXPORT_SYMBOL_GPL(cpufreq_generic_attr);
 
+struct freq_attr *cpufreq_boost_enabled_generic_attr[] = {
+	&cpufreq_freq_attr_scaling_available_freqs,
+	&cpufreq_freq_attr_scaling_boost_freqs,
+	NULL,
+};
+EXPORT_SYMBOL_GPL(cpufreq_boost_enabled_generic_attr);

This isn't scalable. We can't create a new generic structure every time
a entry comes in. Rather we should leave the generic attr for

boost entry is rather generic one so I imagine that other cpufreq
drivers that add support for boost would also like to use it.

Anyway since there are no such drivers currently I agree that
making a new generic structure may be premature and I will re-do
this patch according to your request.

Best regards,
--
Bartlomiej Zolnierkiewicz
Samsung R&D Institute Poland
Samsung Electronics