Thanks Raj.
Let me post my comments inline.
On Wed, Sep 20, 2017 at 12:52 PM, Mani, Rajmohan
<rajmohan.m...@intel.com> wrote:
> Adding Tomasz...
>
>> -----Original Message-----
>> From: Mohandass, Divagar
>> Sent: Monday, September 04, 2017 3:29 AM
>> To: robh...@kernel.org; mark.rutl...@arm.com; w...@the-dreams.de;
>> sakari.ai...@iki.fi
>> Cc: devicet...@vger.kernel.org; linux-...@vger.kernel.org; linux-
>> ker...@vger.kernel.org; Mani, Rajmohan <rajmohan.m...@intel.com>;
>> Mohandass, Divagar <divagar.mohand...@intel.com>
>> Subject: [PATCH v6 3/3] eeprom: at24: enable runtime pm support
>>
>> Currently the device is kept in D0, there is an opportunity to save power by
>> enabling runtime pm.
>>
>> Device can be daisy chained from PMIC and we can't rely on I2C core for auto
>> resume/suspend. Driver will decide when to resume/suspend.
>>
>> Signed-off-by: Divagar Mohandass <divagar.mohand...@intel.com>
>> ---
>> drivers/misc/eeprom/at24.c | 38
>> ++++++++++++++++++++++++++++++++++++++
>> 1 file changed, 38 insertions(+)
>>
>> diff --git a/drivers/misc/eeprom/at24.c b/drivers/misc/eeprom/at24.c index
>> 2199c42..d718a7a 100644
>> --- a/drivers/misc/eeprom/at24.c
>> +++ b/drivers/misc/eeprom/at24.c
>> @@ -24,6 +24,7 @@
>> #include <linux/i2c.h>
>> #include <linux/nvmem-provider.h>
>> #include <linux/platform_data/at24.h>
>> +#include <linux/pm_runtime.h>
>>
>> /*
>> * I2C EEPROMs from most vendors are inexpensive and mostly
>> interchangeable.
>> @@ -501,11 +502,21 @@ static ssize_t at24_eeprom_write_i2c(struct
>> at24_data *at24, const char *buf, static int at24_read(void *priv, unsigned
>> int
>> off, void *val, size_t count) {
>> struct at24_data *at24 = priv;
>> + struct i2c_client *client;
>> char *buf = val;
>> + int ret;
>>
>> if (unlikely(!count))
>> return count;
>>
>> + client = at24_translate_offset(at24, &off);
>> +
>> + ret = pm_runtime_get_sync(&client->dev);
>> + if (ret < 0) {
>> + pm_runtime_put_noidle(&client->dev);
>> + return ret;
>> + }
>> +
>> /*
>> * Read data from chip, protecting against concurrent updates
>> * from this host, but not from other I2C masters.
>> @@ -518,6 +529,7 @@ static int at24_read(void *priv, unsigned int off, void
>> *val, size_t count)
>> status = at24->read_func(at24, buf, off, count);
>> if (status < 0) {
>> mutex_unlock(&at24->lock);
>> + pm_runtime_put(&client->dev);
>> return status;
>> }
>> buf += status;
>> @@ -527,17 +539,29 @@ static int at24_read(void *priv, unsigned int off, void
>> *val, size_t count)
>>
>> mutex_unlock(&at24->lock);
>>
>> + pm_runtime_put(&client->dev);
>> +
>> return 0;
>> }
>>
>> static int at24_write(void *priv, unsigned int off, void *val, size_t
>> count) {
>> struct at24_data *at24 = priv;
>> + struct i2c_client *client;
>> char *buf = val;
>> + int ret;
>>
>> if (unlikely(!count))
>> return -EINVAL;
>>
>> + client = at24_translate_offset(at24, &off);
>> +
>> + ret = pm_runtime_get_sync(&client->dev);
>> + if (ret < 0) {
>> + pm_runtime_put_noidle(&client->dev);
>> + return ret;
>> + }
>> +
>> /*
>> * Write data to chip, protecting against concurrent updates
>> * from this host, but not from other I2C masters.
>> @@ -550,6 +574,7 @@ static int at24_write(void *priv, unsigned int off, void
>> *val, size_t count)
>> status = at24->write_func(at24, buf, off, count);
>> if (status < 0) {
>> mutex_unlock(&at24->lock);
>> + pm_runtime_put(&client->dev);
>> return status;
>> }
>> buf += status;
>> @@ -559,6 +584,8 @@ static int at24_write(void *priv, unsigned int off, void
>> *val, size_t count)
>>
>> mutex_unlock(&at24->lock);
>>
>> + pm_runtime_put(&client->dev);
>> +
>> return 0;
>> }
>>
>> @@ -743,11 +770,17 @@ static int at24_probe(struct i2c_client *client, const
>> struct i2c_device_id *id)
>>
>> i2c_set_clientdata(client, at24);
>>
>> + /* enable runtime pm */
>> + pm_runtime_get_noresume(&client->dev);
>> + pm_runtime_set_active(&client->dev);
>> + pm_runtime_enable(&client->dev);
Do we need this get_noresume/set_active dance? I remember it was for
some reason needed for PCI devices, but I don't see why for I2C
anything else than just pm_runtime_enable() would be necessary.
Also, we enable runtime PM, but we don't provide any callbacks. If
there is no callback in any level of the hierarchy, NULL would be
returned in [3], making [2] return -ENOSYS and [1] fail. The behavior
depends on subsystem and whether the device is attached to a
pm_domain. In our particular case I'd guess the device would be in an
ACPI pm_domain and that would work, but the driver is generic and must
work in any cases.
[1]
http://elixir.free-electrons.com/linux/v4.4.88/source/drivers/base/power/runtime.c#L738
[2]
http://elixir.free-electrons.com/linux/v4.4.88/source/drivers/base/power/runtime.c#L364
[3]
http://elixir.free-electrons.com/linux/v4.4.88/source/drivers/base/power/runtime.c#L19
Best regards,
Tomasz
