On Tue, 28 May 2019 at 04:34, Wolfram Sang <wsa@xxxxxxxxxxxxx> wrote:
>
> On Tue, May 21, 2019 at 10:33:50AM +0900, Masahisa Kojima wrote:
> > master_xfer should return the number of messages successfully
> > processed.
> >
> > Fixes: 0d676a6c4390 ("i2c: add support for Socionext SynQuacer I2C controller")
> > Cc: <stable@xxxxxxxxxxxxxxx> # v4.19+
> > Signed-off-by: Okamoto Satoru <okamoto.satoru@xxxxxxxxxxxxx>
> > Signed-off-by: Masahisa Kojima <masahisa.kojima@xxxxxxxxxx>
>
> Applied to for-current, thanks!
>
Thank you for merging and your feedback.
> I just noticed you have an open coded loop in synquacer_i2c_xfer() which
> should not be needed because the I2C core does that?
I'm not sure I correctly understand the meaning of "open coded loop",
but I2C controller status such as lost arbitration and bus busy is checked
in the retry of synquacer_i2c_doxfer().
> Your code does a HW
> reset, though, but is it really needed for a lost arbitration?
Other than handling lost arbitration, this loop also handles following errors.
- transfer fails(expected size and actual transferred size is not matched)
- transfer timeout
I think it is reasonable to reset the I2C controller before xfer() retries.
Regards,
Masahisa
