Re: [PATCH v2 2/4] spi: spi-fsl-dspi: Fix delete the processing of undefined bitmask for rxdata

[Esben Haabendal <esben.haabendal@xxxxxxxxx>]

Boris Brezillon <boris.brezillon@xxxxxxxxxxx> writes:

> On Sun, 30 Sep 2018 10:10:14 +0000
> Chuanhua Han <chuanhua.han@xxxxxxx> wrote:
>
>> > -----Original Message-----
>> > From: Boris Brezillon <boris.brezillon@xxxxxxxxxxx>
>> > Sent: 2018年9月30日 18:07
>> > To: Chuanhua Han <chuanhua.han@xxxxxxx>
>> > Cc: broonie@xxxxxxxxxx; linux-spi@xxxxxxxxxxxxxxx;
>> > linux-kernel@xxxxxxxxxxxxxxx; eha@xxxxxxxx
>> > Subject: Re: [PATCH v2 2/4] spi: spi-fsl-dspi: Fix delete the processing of
>> > undefined bitmask for rxdata
>> > 
>> > On Sun, 30 Sep 2018 17:25:33 +0800
>> > Chuanhua Han <chuanhua.han@xxxxxxx> wrote:
>> >   
>> > > This patch fixes the problem of rxdata being equal to 0 during the
>> > > XSPI mode transfer of the dspi controller.
>> > > In XSPI mode, If it is not deleted, the value of rxdata will be equal
>> > > to 0, and the data received will not be received correctly, causing
>> > > the receiving transfer of the spi to fail.
>> > >
>> > > Signed-off-by: Chuanhua Han <chuanhua.han@xxxxxxx>
>> > > ---
>> > > Changes in v2:
>> > >  -The original patch is divided into multiple patches(the original
>> > > patch theme is "spi: spi-fsl-dspi: Fix support for XSPI transport
>> > > mode"),one of which is segmented.
>> > >
>> > >  drivers/spi/spi-fsl-dspi.c | 3 ---
>> > >  1 file changed, 3 deletions(-)
>> > >
>> > > diff --git a/drivers/spi/spi-fsl-dspi.c b/drivers/spi/spi-fsl-dspi.c
>> > > index 3082e72e4f6c..4dc1064bf408 100644
>> > > --- a/drivers/spi/spi-fsl-dspi.c
>> > > +++ b/drivers/spi/spi-fsl-dspi.c
>> > > @@ -243,9 +243,6 @@ static void dspi_push_rx(struct fsl_dspi *dspi, u32  
>> > rxdata)  
>> > >  	if (!dspi->rx)
>> > >  		return;
>> > >
>> > > -	/* Mask of undefined bits */
>> > > -	rxdata &= (1 << dspi->bits_per_word) - 1;
>> > > -  
>> > 
>> > Why not  
>> In xspi mode, the value of rxdata after the statement is processed is equal
>> to 0 no matter what data is received.
>
> Only if dspi->bits_per_word is 0.
>
> Actually, I just had a look, and xfer->bits_per_word should never be 0
> because spi_validate() makes sure it's initialized [1]. Don't know
> where dpsi->bits_per_word comes from, but maybe you have a problem
> there (dpsi->bits_per_word and xfer->bits_per_word not in sync).
>
> [1]https://elixir.bootlin.com/linux/v4.19-rc5/source/drivers/spi/spi.c#L2869

dspi->bits_per_word = xfer->bits_per_word

https://elixir.bootlin.com/linux/v4.19-rc5/source/drivers/spi/spi-fsl-dspi.c#L697

So it should never be out of sync, and it should never be 0.

As I mentioned in another mail, I suspect what Han is observing is
caused by byte ordering, so that the mask masks the wrong data.
Maybe related to the byte-ordering fix patch.

/Esben