Re: [PATCH RFC 1/2] tpm: tpm_tis_spi: Ensure SPI mode 0

[Jarkko Sakkinen]

On Fri Sep 6, 2024 at 5:46 PM EEST, Stefan Wahren wrote:
> Am 06.09.24 um 11:47 schrieb Jarkko Sakkinen:
> > On Fri Sep 6, 2024 at 9:09 AM EEST, Stefan Wahren wrote:
> >> According to TCG PC Client Platform TPM Profile (PTP) Specification
> >> only SPI mode 0 is supported. In order to ensure the SPI controller
> >> supports the necessary settings, call spi_setup() and bail out
> >> as soon as possible in error case.
> >>
> >> This change should affect all supported TPM SPI devices, because
> >> tpm_tis_spi_probe is either called direct or indirectly.
> >>
> >> Signed-off-by: Stefan Wahren <wahrenst@xxxxxxx>
> >> ---
> >>   drivers/char/tpm/tpm_tis_spi_main.c | 11 +++++++++++
> >>   1 file changed, 11 insertions(+)
> >>
> >> diff --git a/drivers/char/tpm/tpm_tis_spi_main.c b/drivers/char/tpm/tpm_tis_spi_main.c
> >> index 61b42c83ced8..e62d297b040e 100644
> >> --- a/drivers/char/tpm/tpm_tis_spi_main.c
> >> +++ b/drivers/char/tpm/tpm_tis_spi_main.c
> >> @@ -248,6 +248,17 @@ static int tpm_tis_spi_write_bytes(struct tpm_tis_data *data, u32 addr,
> >>   int tpm_tis_spi_init(struct spi_device *spi, struct tpm_tis_spi_phy *phy,
> >>   		     int irq, const struct tpm_tis_phy_ops *phy_ops)
> >>   {
> >> +	int ret;
> >> +
> >> +	spi->mode &= ~SPI_MODE_X_MASK;
> >> +	spi->mode |= SPI_MODE_0;
> >> +
> >> +	ret = spi_setup(spi);
> >> +	if (ret < 0) {
> >> +		dev_err(&spi->dev, "SPI setup failed: %d\n", ret);
> >> +		return ret;
> >> +	}
> >> +
> >>   	phy->iobuf = devm_kmalloc(&spi->dev, SPI_HDRSIZE + MAX_SPI_FRAMESIZE, GFP_KERNEL);
> >>   	if (!phy->iobuf)
> >>   		return -ENOMEM;
> >> --
> >> 2.34.1
> >
> > Why?
> SPI protocol driver usually call spi_setup to verify that the SPI
> controller accept the settings (SPI mode, bit, clock rate ...). Bailing
> out early is more helpful for debugging issues.

What problem has this patch solved for you? I think it makes the code
only less robust and more error prone.

BR, Jarkko