On 12.02.2016 21:54, Wolfram Sang wrote: > On Mon, Feb 01, 2016 at 04:44:05PM +0200, Andy Shevchenko wrote: >> On Mon, 2016-02-01 at 16:21 +0200, Jarkko Nikula wrote: >>> On 01/30/2016 12:31 AM, Alexey Khoroshilov wrote: >>>> It seems clk_disable_unprepare() is missed in dw_i2c_plat_remove(), >>>> so the patch adds it. >>>> >>>> Found by Linux Driver Verification project (linuxtesting.org). >>>> >>>> Signed-off-by: Alexey Khoroshilov <[email protected]> >>>> --- >>>> drivers/i2c/busses/i2c-designware-platdrv.c | 1 + >>>> 1 file changed, 1 insertion(+) >>>> >>>> diff --git a/drivers/i2c/busses/i2c-designware-platdrv.c >>>> b/drivers/i2c/busses/i2c-designware-platdrv.c >>>> index 438f1b4964c0..8f19b7b81fe0 100644 >>>> --- a/drivers/i2c/busses/i2c-designware-platdrv.c >>>> +++ b/drivers/i2c/busses/i2c-designware-platdrv.c >>>> @@ -267,6 +267,7 @@ static int dw_i2c_plat_remove(struct >>>> platform_device *pdev) >>>> i2c_del_adapter(&dev->adapter); >>>> >>>> i2c_dw_disable(dev); >>>> + i2c_dw_plat_prepare_clk(dev, false); >>>> >>> I tried this quickly and it appears more work is needed. When >>> CONFIG_PM_SLEEP is set then autosuspending will do the unprepare and >>> this patch causes double unprepare at remove. But when >>> CONFIG_PM_SLEEP >>> is not set then indeed those clk calls are out of sync. >> >> Besides that I would suggest to check carefully error patch in the >> probe(), i.e. handling error from i2c_dw_probe(). There maybe similar >> issue is hidden. > > So, waiting for V2 on this one. >
I have a fix for error handling of i2c_dw_probe(), but I am not sure what is the right approach to handle CONFIG_PM_SLEEP case. What is a safe way to distinguish a need for the unprepare in dw_i2c_plat_remove()? Should we try to avoid double i2c_dw_disable(dev) in the same case? -- Alexey
