[PATCH v2 0/2] spi: new feature and fix for Marvell Orion driver

[PATCH v2 0/2] spi: new feature and fix for Marvell Orion driver

[kostap]

From: Konstantin Porotchkin <kostap@marvell.com>

This patch set contains the following changes:
- support for switching CS for every transferred byte
- fix for the clock by enabling it for every call to spi_setup()

v2:
- remove DTS entry for single byte CS, use existing SPI_CS_WORD flag
- drop changes to core SPI driver
- fix the wrong first patch signature

Marcin Wojtas (2):
  spi: orion: enable clocks before spi_setup
  spi: orion: enable support for switching CS every transferred byte

 drivers/spi/spi-orion.c | 34 +++++++++++++++++++-
 1 file changed, 33 insertions(+), 1 deletion(-)

-- 
2.17.1

[PATCH v2 1/2] spi: orion: enable clocks before spi_setup

[kostap]

From: Marcin Wojtas <mw@semihalf.com>

The spi-orion driver disables its clocks whenever it is not used.
In usual case during boot (i.e. using SPI flash) it is not a problem,
as the child device driver is present and probed along with
spi_register_master() execution.

However in case the child device driver is not ready
(e.g. when its type is module_spi_driver) the spi_setup() callback
can be called after the spi-orion probe. It may happen,
that as a result there will be an attempt to access controller's
registers with the clocks disabled.

Prevent such situations and make sure the clocks are on,
each time the spi_setup() is called.

Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: Konstantin Porotchkin <kostap@marvell.com>
---
 drivers/spi/spi-orion.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/drivers/spi/spi-orion.c b/drivers/spi/spi-orion.c
index b57b8b3cc26e..3bfda4225d45 100644
--- a/drivers/spi/spi-orion.c
+++ b/drivers/spi/spi-orion.c
@@ -507,6 +507,16 @@ static int orion_spi_transfer_one(struct spi_master *master,
 
 static int orion_spi_setup(struct spi_device *spi)
 {
+	struct orion_spi *orion_spi = spi_master_get_devdata(spi->master);
+
+	/*
+	 * Make sure the clocks are enabled before
+	 * configuring the SPI controller.
+	 */
+	clk_prepare_enable(orion_spi->clk);
+	if (!IS_ERR(orion_spi->axi_clk))
+		clk_prepare_enable(orion_spi->axi_clk);
+
 	return orion_spi_setup_transfer(spi, NULL);
 }
 
-- 
2.17.1

[PATCH v2 2/2] spi: orion: enable support for switching CS every transferred byte

[kostap]

From: Marcin Wojtas <mw@semihalf.com>

Some SPI devices, require toggling the CS every transferred byte.
Enable such possibility in the spi-orion driver.

Note that in order to use it, in the driver of a secondary device
attached to this controller, the SPI bus 'mode' field must be
updated with SPI_CS_WORD flag before calling spi_setup() routine.

In addition to that include a work-around - some devices, such as
certain models of SLIC (Subscriber Line Interface Card),
may require extra delay after CS toggling, so add a minimal
timing relaxation in relevant places.

Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: Konstantin Porotchkin <kostap@marvell.com>
---
 drivers/spi/spi-orion.c | 24 +++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

diff --git a/drivers/spi/spi-orion.c b/drivers/spi/spi-orion.c
index 3bfda4225d45..0f92dd026bee 100644
--- a/drivers/spi/spi-orion.c
+++ b/drivers/spi/spi-orion.c
@@ -369,8 +369,15 @@ orion_spi_write_read_8bit(struct spi_device *spi,
 {
 	void __iomem *tx_reg, *rx_reg, *int_reg;
 	struct orion_spi *orion_spi;
+	bool cs_single_byte;
+
+	cs_single_byte = spi->mode & SPI_CS_WORD;
 
 	orion_spi = spi_master_get_devdata(spi->master);
+
+	if (cs_single_byte)
+		orion_spi_set_cs(spi, 0);
+
 	tx_reg = spi_reg(orion_spi, ORION_SPI_DATA_OUT_REG);
 	rx_reg = spi_reg(orion_spi, ORION_SPI_DATA_IN_REG);
 	int_reg = spi_reg(orion_spi, ORION_SPI_INT_CAUSE_REG);
@@ -384,6 +391,11 @@ orion_spi_write_read_8bit(struct spi_device *spi,
 		writel(0, tx_reg);
 
 	if (orion_spi_wait_till_ready(orion_spi) < 0) {
+		if (cs_single_byte) {
+			orion_spi_set_cs(spi, 1);
+			/* Satisfy some SLIC devices requirements */
+			udelay(4);
+		}
 		dev_err(&spi->dev, "TXS timed out\n");
 		return -1;
 	}
@@ -391,6 +403,16 @@ orion_spi_write_read_8bit(struct spi_device *spi,
 	if (rx_buf && *rx_buf)
 		*(*rx_buf)++ = readl(rx_reg);
 
+	if (cs_single_byte) {
+		orion_spi_set_cs(spi, 1);
+		/*
+		 * Some devices, such as certain models of SLIC
+		 * (Subscriber Line Interface Card)
+		 * may require extra delay after CS toggling.
+		 */
+		udelay(4);
+	}
+
 	return 1;
 }
 
@@ -626,7 +648,7 @@ static int orion_spi_probe(struct platform_device *pdev)
 	}
 
 	/* we support all 4 SPI modes and LSB first option */
-	master->mode_bits = SPI_CPHA | SPI_CPOL | SPI_LSB_FIRST;
+	master->mode_bits = SPI_CPHA | SPI_CPOL | SPI_LSB_FIRST | SPI_CS_WORD;
 	master->set_cs = orion_spi_set_cs;
 	master->transfer_one = orion_spi_transfer_one;
 	master->num_chipselect = ORION_NUM_CHIPSELECTS;
-- 
2.17.1

[PATCH v2 0/2] spi: new feature and fix for Marvell Orion driver

[kostap]

From: Konstantin Porotchkin <kostap@marvell.com>

This patch set contains the following:
- support for switching CS for every transferred byte
- fix for the clock by enabling it for every call to spi_setup()

v2:
- remove DTS entry for single byte CS, use existing SPI_CS_WORD flag
- drop changes to core SPI driver
- fix the wrong first patch signature

Marcin Wojtas (2):
  spi: orion: enable clocks before spi_setup
  spi: orion: enable support for switching CS every transferred byte

 drivers/spi/spi-orion.c | 30 +++++++++++++++++++-
 1 file changed, 29 insertions(+), 1 deletion(-)

-- 
2.17.1

Re: [PATCH v2 1/2] spi: orion: enable clocks before spi_setup

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 687 bytes --]

On Thu, Dec 17, 2020 at 07:09:31PM +0200, kostap@marvell.com wrote:

> +	/*
> +	 * Make sure the clocks are enabled before
> +	 * configuring the SPI controller.
> +	 */
> +	clk_prepare_enable(orion_spi->clk);
> +	if (!IS_ERR(orion_spi->axi_clk))
> +		clk_prepare_enable(orion_spi->axi_clk);
> +
>  	return orion_spi_setup_transfer(spi, NULL);
>  }

There's no matching disable here so we'll leak the enables every time
setup() is called - we should unwind the enables after calling
_setup_transfer().  It may be more sensible to just take a runtime PM
reference rather than do the raw clock API stuff, one less call and
means if anything else gets added to runtime PM it'll be handled.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH v2 2/2] spi: orion: enable support for switching CS every transferred byte

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 1050 bytes --]

On Thu, Dec 17, 2020 at 07:09:32PM +0200, kostap@marvell.com wrote:

> +++ b/drivers/spi/spi-orion.c
> @@ -369,8 +369,15 @@ orion_spi_write_read_8bit(struct spi_device *spi,
>  {

This is only supporting SPI_CS_WORD for 8 bit operations but the driver
also supports 16 bit words, it should at least report an error if
there's an attempt to use SPI_CS_WORD for 16 bit transfers.  It also
looks like this won't work on systems where direct access is supported
since those use a separate I/O path, that can be fixed by just adding an
additional check when deciding to go down that path.

The driver should also pay attention to SPI_CS_HIGH if it's going to try
to control chip select by hand as it does, which is generally frowned
upon.  TBH I'm wondering if it might not be better to just rely on the
core support for implementing SPI_CS_WORD on controllers that can't do
it in hardware - it *is* much higher overhead since it needs to split
the transfers up but it depends how performance critical and frequent
access to such devices is likely to be.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

RE: [EXT] Re: [PATCH v2 1/2] spi: orion: enable clocks before spi_setup

[Kostya Porotchkin]

Hi, Mark,

> -----Original Message-----
> From: Mark Brown <broonie@kernel.org>

> > +	/*
> > +	 * Make sure the clocks are enabled before
> > +	 * configuring the SPI controller.
> > +	 */
> > +	clk_prepare_enable(orion_spi->clk);
> > +	if (!IS_ERR(orion_spi->axi_clk))
> > +		clk_prepare_enable(orion_spi->axi_clk);
> > +
> >  	return orion_spi_setup_transfer(spi, NULL);  }
> 
> There's no matching disable here so we'll leak the enables every time
> setup() is called - we should unwind the enables after calling
> _setup_transfer().  It may be more sensible to just take a runtime PM
> reference rather than do the raw clock API stuff, one less call and means if
> anything else gets added to runtime PM it'll be handled.

[KP] You mean we should call here to orion_spi_runtime_resume() instead of clk_prepare_enable() and make this PM function available regardless the CONFIG_PM option?

Thanks
Kosta

Re: [EXT] Re: [PATCH v2 1/2] spi: orion: enable clocks before spi_setup

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 890 bytes --]

On Tue, Dec 22, 2020 at 12:46:01PM +0000, Kostya Porotchkin wrote:

Please fix your mail client to word wrap within paragraphs at something
substantially less than 80 columns.  Doing this makes your messages much
easier to read and reply to.

> > There's no matching disable here so we'll leak the enables every time
> > setup() is called - we should unwind the enables after calling
> > _setup_transfer().  It may be more sensible to just take a runtime PM
> > reference rather than do the raw clock API stuff, one less call and means if
> > anything else gets added to runtime PM it'll be handled.

> [KP] You mean we should call here to orion_spi_runtime_resume()
> instead of clk_prepare_enable() and make this PM function available
> regardless the CONFIG_PM option?

That's one part of the suggestion, yes - the other part is that you need
to have a disable matching the enable here.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]