[PATCH] SPI: drain MXC SPI transfer buffer when probing device

[Daniel Mack]

Hi Uwe,

On Thu, Nov 19, 2009 at 07:49:51PM +0100, Uwe Kleine-König wrote:
> On Thu, Nov 19, 2009 at 11:44:47AM +0100, Daniel Mack wrote:
> > On the MX31litekit, the bootloader seems to communicate with the MC13783
> > PMIC chip before booting Linux. However, it does not flush all the
> > buffers properly after that, which makes the imx-spi driver read
> > bogus data when probing the MC13783.
> > 
> > Fix that by draining the SPI buffer on startup.
> > 
> > Signed-off-by: Daniel Mack <daniel at caiaq.de>
> > Cc: David Brownell <dbrownell at users.sourceforge.net>
> > Cc: Andrew Morton <akpm at linux-foundation.org>
> > Cc: Sascha Hauer <s.hauer at pengutronix.de>
> > Cc: Uwe Kleine-König <u.kleine-koenig at pengutronix.de>
> > Cc: spi-devel-general at lists.sourceforge.net
> > ---
> >  drivers/spi/spi_imx.c |    7 +++++++
> >  1 files changed, 7 insertions(+), 0 deletions(-)
> > 
> > diff --git a/drivers/spi/spi_imx.c b/drivers/spi/spi_imx.c
> > index 89c22ef..a3894fd 100644
> > --- a/drivers/spi/spi_imx.c
> > +++ b/drivers/spi/spi_imx.c
> > @@ -42,8 +42,11 @@
> >  #define MXC_CSPITXDATA		0x04
> >  #define MXC_CSPICTRL		0x08
> >  #define MXC_CSPIINT		0x0c
> > +#define MXC_CSPISTAT		0x14
> On imx27 the register at offset 0x14 is called PERIODREG ...
> 
> >  #define MXC_RESET		0x1c
> >  
> > +#define MXC_CSPISTAT_RR		(1 << 3)
> ... and bits 0..14 are called WAIT.
> 
> 
> > +
> >  /* generic defines to abstract from the different register layouts */
> >  #define MXC_INT_RR	(1 << 0) /* Receive data ready interrupt */
> >  #define MXC_INT_TE	(1 << 1) /* Transmit FIFO empty interrupt */
> > @@ -593,6 +596,10 @@ static int __init spi_imx_probe(struct platform_device *pdev)
> >  	if (!cpu_is_mx31() || !cpu_is_mx35())
> >  		writel(1, spi_imx->base + MXC_RESET);
> >  
> > +	/* drain the buffer */
> > +	while (readl(spi_imx->base + MXC_CSPISTAT) & MXC_CSPISTAT_RR)
> > +		readl(spi_imx->base + MXC_CSPIRXDATA);
> > +
> So this needs protection by
> 
> 	if (cpu_is_mx31() || cpu_is_mx35())

Oh, I wasn't aware of such differences, thanks for checking!

> (note, I didn't check the mx35 reference!).

But I did now, and mx35 seems to be compatible. I also changed the
register define names to avoid confusion.

New patch below.

Thanks,
Daniel