spi: coldfire-qspi: Fix bits per word setting when t->bits_per_word is not set
diff mbox

Message ID 1393463102.12945.3.camel@phoenix
State New, archived
Delegated to: Mark Brown
Headers show

Commit Message

Axel Lin Feb. 27, 2014, 1:05 a.m. UTC
This ensures to use spi_device's bits_per_word setting when t->bits_per_word is
not set. This looks like a bug introduced by commit 24778be20f8
"spi: convert drivers to use bits_per_word_mask".

Signed-off-by: Axel Lin <axel.lin@ingics.com>
---
 drivers/spi/spi-coldfire-qspi.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

Comments

Stephen Warren Feb. 27, 2014, 5:09 p.m. UTC | #1
On 02/26/2014 06:05 PM, Axel Lin wrote:
> This ensures to use spi_device's bits_per_word setting when t->bits_per_word is
> not set. This looks like a bug introduced by commit 24778be20f8
> "spi: convert drivers to use bits_per_word_mask".

I thought the SPI core already contained this kind of logic, so that
drivers could rely on t->bits_per_word always being set up? Yes, this:

> static int __spi_validate(struct spi_device *spi, struct spi_message *message)
...
>         /**
>          * Set transfer bits_per_word and max speed as spi device default if
>          * it is not set for this transfer.
>          * Set transfer tx_nbits and rx_nbits as single transfer default
>          * (SPI_NBITS_SINGLE) if it is not set for this transfer.
>          */
>         list_for_each_entry(xfer, &message->transfers, transfer_list) {
>                 message->frame_length += xfer->len;
>                 if (!xfer->bits_per_word)
>                         xfer->bits_per_word = spi->bits_per_word;

--
To unsubscribe from this list: send the line "unsubscribe linux-spi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Mark Brown Feb. 28, 2014, 5:23 a.m. UTC | #2
On Thu, Feb 27, 2014 at 10:09:59AM -0700, Stephen Warren wrote:
> On 02/26/2014 06:05 PM, Axel Lin wrote:
> > This ensures to use spi_device's bits_per_word setting when t->bits_per_word is
> > not set. This looks like a bug introduced by commit 24778be20f8
> > "spi: convert drivers to use bits_per_word_mask".

> I thought the SPI core already contained this kind of logic, so that
> drivers could rely on t->bits_per_word always being set up? Yes, this:

Indeed - if there weren't such logic the better fix would be to add it.
Axel Lin March 2, 2014, 2:20 p.m. UTC | #3
2014-02-28 1:09 GMT+08:00 Stephen Warren <swarren@wwwdotorg.org>:
> On 02/26/2014 06:05 PM, Axel Lin wrote:
>> This ensures to use spi_device's bits_per_word setting when t->bits_per_word is
>> not set. This looks like a bug introduced by commit 24778be20f8
>> "spi: convert drivers to use bits_per_word_mask".
>
> I thought the SPI core already contained this kind of logic, so that
> drivers could rely on t->bits_per_word always being set up? Yes, this:
Hi Stephen,
Yes, you are right and thanks for the review.

Regards,
Axel
--
To unsubscribe from this list: send the line "unsubscribe linux-spi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch
diff mbox

diff --git a/drivers/spi/spi-coldfire-qspi.c b/drivers/spi/spi-coldfire-qspi.c
index c29bbb0..8c85bea 100644
--- a/drivers/spi/spi-coldfire-qspi.c
+++ b/drivers/spi/spi-coldfire-qspi.c
@@ -315,12 +315,16 @@  static int mcfqspi_transfer_one(struct spi_master *master,
 {
 	struct mcfqspi *mcfqspi = spi_master_get_devdata(master);
 	u16 qmr = MCFQSPI_QMR_MSTR;
+	u8 bpw;
 
-	qmr |= t->bits_per_word << 10;
+	bpw = t->bits_per_word ? : spi->bits_per_word;
+
+	qmr |= bpw << 10;
 	if (spi->mode & SPI_CPHA)
 		qmr |= MCFQSPI_QMR_CPHA;
 	if (spi->mode & SPI_CPOL)
 		qmr |= MCFQSPI_QMR_CPOL;
+
 	if (t->speed_hz)
 		qmr |= mcfqspi_qmr_baud(t->speed_hz);
 	else
@@ -328,7 +332,7 @@  static int mcfqspi_transfer_one(struct spi_master *master,
 	mcfqspi_wr_qmr(mcfqspi, qmr);
 
 	mcfqspi_wr_qir(mcfqspi, MCFQSPI_QIR_SPIFE);
-	if (t->bits_per_word == 8)
+	if (bpw == 8)
 		mcfqspi_transfer_msg8(mcfqspi, t->len, t->tx_buf, t->rx_buf);
 	else
 		mcfqspi_transfer_msg16(mcfqspi, t->len / 2, t->tx_buf,