On Thursday, September 21, 2017 3:13:56 AM CEST Rajat Jain wrote:

On Wed, Sep 20, 2017 at 5:24 PM, Rafael J. Wysocki [off-list ref] wrote:

quoted

On Thu, Sep 21, 2017 at 12:31 AM, Rajat Jain [off-list ref] wrote:

quoted

Ref: https://lkml.org/lkml/2017/9/19/649

The bus controllers should suspend the bus operations only after
all of the devices on the bus have suspended their device
completely. Since the i2c_client drivers could be talking to
their devices in their suspend_late() calls, lets ensure that the
bus is alive by that time. Thus moving the controller suspend logic to
suspend_late().

Signed-off-by: Rajat Jain <redacted>
---
 drivers/i2c/busses/i2c-designware-platdrv.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/i2c/busses/i2c-designware-platdrv.c b/drivers/i2c/busses/i2c-designware-platdrv.c
index 0e65b97842b4..66dd7f844c40 100644
--- a/drivers/i2c/busses/i2c-designware-platdrv.c
+++ b/drivers/i2c/busses/i2c-designware-platdrv.c

@@ -468,7 +468,7 @@ static int dw_i2c_plat_suspend(struct device *dev)
 static const struct dev_pm_ops dw_i2c_dev_pm_ops = {
        .prepare = dw_i2c_plat_prepare,
        .complete = dw_i2c_plat_complete,
-       SET_SYSTEM_SLEEP_PM_OPS(dw_i2c_plat_suspend, dw_i2c_plat_resume)
+       SET_LATE_SYSTEM_SLEEP_PM_OPS(dw_i2c_plat_suspend, dw_i2c_plat_resume)
        SET_RUNTIME_PM_OPS(dw_i2c_plat_runtime_suspend,
                           dw_i2c_plat_resume,
                           NULL)

No, you can't just do that.

I sent patches to do it properly before my trip to LA last week, it
shouldn't be overly difficult to find them in the mailing list
archives.  I can look them up tomorrow if need be.

Thanks, I am guessing you mean this?

https://patchwork.kernel.org/patch/9939807/

Yes, that's what I mean.

Thanks,
Rafael