[PATCH V2 0/6] spi: bcm2835: improvements

[PATCH V2 0/6] spi: bcm2835: improvements

[kernel]

From: Martin Sperl <kernel@martin.sperl.org>

Patchset that implements changes for the spi-bcm2835aux now
also for the spi-bcm2835

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions

Martin Sperl (6):
  spi: bcm2835: bcm2835_spi_transfer_one_poll remove unnecessary
    argument
  spi: bcm2835: Avoid 64-bit arithmetic in xfer len calc
  spi: bcm2835: added comment about different bus behaviour of DMA mode
  spi: bcm2835: make the polling duration limits configurable
  spi: bcm2835: make the lower limit for dma mode configurable
  spi: bcm2835: add driver stats to debugfs

 drivers/spi/spi-bcm2835.c | 128 ++++++++++++++++++++++++++++++++++++++--------
 1 file changed, 107 insertions(+), 21 deletions(-)

--
2.11.0

[PATCH V2 0/6] spi: bcm2835: improvements

[kernel]

From: Martin Sperl <kernel@martin.sperl.org>

Patchset that implements changes for the spi-bcm2835aux now
also for the spi-bcm2835

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions

Martin Sperl (6):
  spi: bcm2835: bcm2835_spi_transfer_one_poll remove unnecessary
    argument
  spi: bcm2835: Avoid 64-bit arithmetic in xfer len calc
  spi: bcm2835: added comment about different bus behaviour of DMA mode
  spi: bcm2835: make the polling duration limits configurable
  spi: bcm2835: make the lower limit for dma mode configurable
  spi: bcm2835: add driver stats to debugfs

 drivers/spi/spi-bcm2835.c | 128 ++++++++++++++++++++++++++++++++++++++--------
 1 file changed, 107 insertions(+), 21 deletions(-)

--
2.11.0

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH V2 1/6] spi: bcm2835: bcm2835_spi_transfer_one_poll remove unnecessary argument

[kernel-TqfNSX0MhmxHKSADF0wUEw]

From: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>

Remove the unnecessary argument of xfer_time_us when calling
bcm2835_spi_transfer_one_poll.

Signed-off-by: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions

---
 drivers/spi/spi-bcm2835.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 35aebdfd3b4e..ef7f8dc9f18e 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -725,8 +725,7 @@ static void bcm2835_dma_init(struct spi_master *master, struct device *dev)
 static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 					 struct spi_device *spi,
 					 struct spi_transfer *tfr,
-					 u32 cs,
-					 unsigned long long xfer_time_us)
+					 u32 cs)
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long timeout;
@@ -827,8 +826,7 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,

 	/* for short requests run polling*/
 	if (xfer_time_us <= BCM2835_SPI_POLLING_LIMIT_US)
-		return bcm2835_spi_transfer_one_poll(master, spi, tfr,
-						     cs, xfer_time_us);
+		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);

 	/* run in dma mode if conditions are right */
 	if (master->can_dma && bcm2835_spi_can_dma(master, spi, tfr))
--
2.11.0

[PATCH V2 1/6] spi: bcm2835: bcm2835_spi_transfer_one_poll remove unnecessary argument

[kernel]

From: Martin Sperl <kernel@martin.sperl.org>

Remove the unnecessary argument of xfer_time_us when calling
bcm2835_spi_transfer_one_poll.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions

---
 drivers/spi/spi-bcm2835.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 35aebdfd3b4e..ef7f8dc9f18e 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -725,8 +725,7 @@ static void bcm2835_dma_init(struct spi_master *master, struct device *dev)
 static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 					 struct spi_device *spi,
 					 struct spi_transfer *tfr,
-					 u32 cs,
-					 unsigned long long xfer_time_us)
+					 u32 cs)
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long timeout;
@@ -827,8 +826,7 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,

 	/* for short requests run polling*/
 	if (xfer_time_us <= BCM2835_SPI_POLLING_LIMIT_US)
-		return bcm2835_spi_transfer_one_poll(master, spi, tfr,
-						     cs, xfer_time_us);
+		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);

 	/* run in dma mode if conditions are right */
 	if (master->can_dma && bcm2835_spi_can_dma(master, spi, tfr))
--
2.11.0

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH V2 2/6] spi: bcm2835: Avoid 64-bit arithmetic in xfer len calc

[kernel]

From: Martin Sperl <kernel@martin.sperl.org>

Avoid 64 bit aritmetics when deciding if we need to use polling or not
This replicates: commit d704afffe65c
("spi: bcm2835aux: Avoid 64-bit arithmetic in xfer len calc")
from spi-bcm2835aux

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions

---
 drivers/spi/spi-bcm2835.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index ef7f8dc9f18e..a64bfa8e0710 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -777,7 +777,6 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long spi_hz, clk_hz, cdiv;
 	unsigned long spi_used_hz;
-	unsigned long long xfer_time_us;
 	u32 cs = bcm2835_rd(bs, BCM2835_SPI_CS);

 	/* set clock */
@@ -818,14 +817,15 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	bs->tx_len = tfr->len;
 	bs->rx_len = tfr->len;

-	/* calculate the estimated time in us the transfer runs */
-	xfer_time_us = (unsigned long long)tfr->len
-		* 9 /* clocks/byte - SPI-HW waits 1 clock after each byte */
-		* 1000000;
-	do_div(xfer_time_us, spi_used_hz);
-
-	/* for short requests run polling*/
-	if (xfer_time_us <= BCM2835_SPI_POLLING_LIMIT_US)
+	/* Calculate the estimated time in us the transfer runs.  Note that
+	 * there is 1 idle clocks cycles after each byte getting transferred
+	 * so we have 9 cycles/byte.  This is used to find the number of Hz
+	 * per byte per polling limit.  E.g., we can transfer 1 byte in 30 us
+	 * per 300,000 Hz of bus clock.
+	 */
+#define HZ_PER_BYTE ((9 * 1000000) / BCM2835_SPI_POLLING_LIMIT_US)
+	/* run in polling mode for short transfers */
+	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);

 	/* run in dma mode if conditions are right */
--
2.11.0

[PATCH V2 2/6] spi: bcm2835: Avoid 64-bit arithmetic in xfer len calc

[kernel]

From: Martin Sperl <kernel@martin.sperl.org>

Avoid 64 bit aritmetics when deciding if we need to use polling or not
This replicates: commit d704afffe65c
("spi: bcm2835aux: Avoid 64-bit arithmetic in xfer len calc")
from spi-bcm2835aux

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions

---
 drivers/spi/spi-bcm2835.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index ef7f8dc9f18e..a64bfa8e0710 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -777,7 +777,6 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long spi_hz, clk_hz, cdiv;
 	unsigned long spi_used_hz;
-	unsigned long long xfer_time_us;
 	u32 cs = bcm2835_rd(bs, BCM2835_SPI_CS);

 	/* set clock */
@@ -818,14 +817,15 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	bs->tx_len = tfr->len;
 	bs->rx_len = tfr->len;

-	/* calculate the estimated time in us the transfer runs */
-	xfer_time_us = (unsigned long long)tfr->len
-		* 9 /* clocks/byte - SPI-HW waits 1 clock after each byte */
-		* 1000000;
-	do_div(xfer_time_us, spi_used_hz);
-
-	/* for short requests run polling*/
-	if (xfer_time_us <= BCM2835_SPI_POLLING_LIMIT_US)
+	/* Calculate the estimated time in us the transfer runs.  Note that
+	 * there is 1 idle clocks cycles after each byte getting transferred
+	 * so we have 9 cycles/byte.  This is used to find the number of Hz
+	 * per byte per polling limit.  E.g., we can transfer 1 byte in 30 us
+	 * per 300,000 Hz of bus clock.
+	 */
+#define HZ_PER_BYTE ((9 * 1000000) / BCM2835_SPI_POLLING_LIMIT_US)
+	/* run in polling mode for short transfers */
+	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);

 	/* run in dma mode if conditions are right */
--
2.11.0

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH V2 3/6] spi: bcm2835: added comment about different bus behaviour of DMA mode

[kernel-TqfNSX0MhmxHKSADF0wUEw]

From: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>

The DMA mode behaves slightly different than polling or interrupt driven
mode, so just document the fact

Signed-off-by: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            new in V2

---
 drivers/spi/spi-bcm2835.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index a64bfa8e0710..3140d952c03a 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -828,7 +828,10 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);

-	/* run in dma mode if conditions are right */
+	/* run in dma mode if conditions are right
+	 * Note that unlike poll or interrupt mode DMA mode does not have
+	 * this 1 idle clock cycle pattern but runs the spi clock without gaps
+	 */
 	if (master->can_dma && bcm2835_spi_can_dma(master, spi, tfr))
 		return bcm2835_spi_transfer_one_dma(master, spi, tfr, cs);

--
2.11.0

[PATCH V2 3/6] spi: bcm2835: added comment about different bus behaviour of DMA mode

[kernel]

From: Martin Sperl <kernel@martin.sperl.org>

The DMA mode behaves slightly different than polling or interrupt driven
mode, so just document the fact

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            new in V2

---
 drivers/spi/spi-bcm2835.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index a64bfa8e0710..3140d952c03a 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -828,7 +828,10 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);

-	/* run in dma mode if conditions are right */
+	/* run in dma mode if conditions are right
+	 * Note that unlike poll or interrupt mode DMA mode does not have
+	 * this 1 idle clock cycle pattern but runs the spi clock without gaps
+	 */
 	if (master->can_dma && bcm2835_spi_can_dma(master, spi, tfr))
 		return bcm2835_spi_transfer_one_dma(master, spi, tfr, cs);

--
2.11.0

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH V2 4/6] spi: bcm2835: make the polling duration limits configurable

[kernel-TqfNSX0MhmxHKSADF0wUEw]

From: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>

Under some circumstances the default 30 us polling limit is not optimal
and may lead to long delays because we are waiting on an interrupt.
with this patch we have the possibility to influence this policy.

So make this limit (in us) configurable via a module parameters
(but also modifyable via /sys/modules/...)

This replicates similar code found in spi-bcm2835aux.

Signed-off-by: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions

---
 drivers/spi/spi-bcm2835.c | 22 ++++++++++++++--------
 1 file changed, 14 insertions(+), 8 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 3140d952c03a..2d702db8689c 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -73,14 +73,18 @@

 #define BCM2835_SPI_FIFO_SIZE		64
 #define BCM2835_SPI_FIFO_SIZE_3_4	48
-#define BCM2835_SPI_POLLING_LIMIT_US	30
-#define BCM2835_SPI_POLLING_JIFFIES	2
 #define BCM2835_SPI_DMA_MIN_LENGTH	96
 #define BCM2835_SPI_MODE_BITS	(SPI_CPOL | SPI_CPHA | SPI_CS_HIGH \
 				| SPI_NO_CS | SPI_3WIRE)

 #define DRV_NAME	"spi-bcm2835"

+/* define polling limits */
+unsigned int polling_limit_us = 30;
+module_param(polling_limit_us, uint, 0664);
+MODULE_PARM_DESC(polling_limit_us,
+		 "time in us to run a transfer in polling mode\n");
+
 /**
  * struct bcm2835_spi - BCM2835 SPI controller
  * @regs: base address of register map
@@ -739,8 +743,8 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 	 */
 	bcm2835_wr_fifo_blind(bs, BCM2835_SPI_FIFO_SIZE);

-	/* set the timeout */
-	timeout = jiffies + BCM2835_SPI_POLLING_JIFFIES;
+	/* set the timeout to at least 2 jiffies */
+	timeout = jiffies + 2 + HZ * polling_limit_us / 1000000;

 	/* loop until finished the transfer */
 	while (bs->rx_len) {
@@ -775,8 +779,8 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 				    struct spi_transfer *tfr)
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
-	unsigned long spi_hz, clk_hz, cdiv;
-	unsigned long spi_used_hz;
+	unsigned long spi_hz, clk_hz, cdiv, spi_used_hz;
+	unsigned long hz_per_byte, byte_limit;
 	u32 cs = bcm2835_rd(bs, BCM2835_SPI_CS);

 	/* set clock */
@@ -823,9 +827,11 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	 * per byte per polling limit.  E.g., we can transfer 1 byte in 30 us
 	 * per 300,000 Hz of bus clock.
 	 */
-#define HZ_PER_BYTE ((9 * 1000000) / BCM2835_SPI_POLLING_LIMIT_US)
+	hz_per_byte = polling_limit_us ? (9 * 1000000) / polling_limit_us : 0;
+	byte_limit = hz_per_byte ? spi_used_hz / hz_per_byte : 1;
+
 	/* run in polling mode for short transfers */
-	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
+	if (tfr->len < byte_limit)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);

 	/* run in dma mode if conditions are right
--
2.11.0

[PATCH V2 4/6] spi: bcm2835: make the polling duration limits configurable

[kernel]

From: Martin Sperl <kernel@martin.sperl.org>

Under some circumstances the default 30 us polling limit is not optimal
and may lead to long delays because we are waiting on an interrupt.
with this patch we have the possibility to influence this policy.

So make this limit (in us) configurable via a module parameters
(but also modifyable via /sys/modules/...)

This replicates similar code found in spi-bcm2835aux.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions

---
 drivers/spi/spi-bcm2835.c | 22 ++++++++++++++--------
 1 file changed, 14 insertions(+), 8 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 3140d952c03a..2d702db8689c 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -73,14 +73,18 @@

 #define BCM2835_SPI_FIFO_SIZE		64
 #define BCM2835_SPI_FIFO_SIZE_3_4	48
-#define BCM2835_SPI_POLLING_LIMIT_US	30
-#define BCM2835_SPI_POLLING_JIFFIES	2
 #define BCM2835_SPI_DMA_MIN_LENGTH	96
 #define BCM2835_SPI_MODE_BITS	(SPI_CPOL | SPI_CPHA | SPI_CS_HIGH \
 				| SPI_NO_CS | SPI_3WIRE)

 #define DRV_NAME	"spi-bcm2835"

+/* define polling limits */
+unsigned int polling_limit_us = 30;
+module_param(polling_limit_us, uint, 0664);
+MODULE_PARM_DESC(polling_limit_us,
+		 "time in us to run a transfer in polling mode\n");
+
 /**
  * struct bcm2835_spi - BCM2835 SPI controller
  * @regs: base address of register map
@@ -739,8 +743,8 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 	 */
 	bcm2835_wr_fifo_blind(bs, BCM2835_SPI_FIFO_SIZE);

-	/* set the timeout */
-	timeout = jiffies + BCM2835_SPI_POLLING_JIFFIES;
+	/* set the timeout to at least 2 jiffies */
+	timeout = jiffies + 2 + HZ * polling_limit_us / 1000000;

 	/* loop until finished the transfer */
 	while (bs->rx_len) {
@@ -775,8 +779,8 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 				    struct spi_transfer *tfr)
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
-	unsigned long spi_hz, clk_hz, cdiv;
-	unsigned long spi_used_hz;
+	unsigned long spi_hz, clk_hz, cdiv, spi_used_hz;
+	unsigned long hz_per_byte, byte_limit;
 	u32 cs = bcm2835_rd(bs, BCM2835_SPI_CS);

 	/* set clock */
@@ -823,9 +827,11 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	 * per byte per polling limit.  E.g., we can transfer 1 byte in 30 us
 	 * per 300,000 Hz of bus clock.
 	 */
-#define HZ_PER_BYTE ((9 * 1000000) / BCM2835_SPI_POLLING_LIMIT_US)
+	hz_per_byte = polling_limit_us ? (9 * 1000000) / polling_limit_us : 0;
+	byte_limit = hz_per_byte ? spi_used_hz / hz_per_byte : 1;
+
 	/* run in polling mode for short transfers */
-	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
+	if (tfr->len < byte_limit)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);

 	/* run in dma mode if conditions are right
--
2.11.0

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH V2 5/6] spi: bcm2835: make the lower limit for dma mode configurable

[kernel-TqfNSX0MhmxHKSADF0wUEw]

From: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>

Under some circumstances the default 96 byte limit is not optimal.

The polling and interrupt mode of the spi-bcm2835 controller exhibit a
1 idle clock cycle which is missing when using DMA.

So when transferring a longer spi_transfer to some low spec MCU devices
without SPI FIFOs like a ATMega it means that the SPI has to get reduced
by a factor of 2 or more when using DMA mode compared to using PIO modes
to give the MCU enough time to handle the incoming spi byte and prepare
the next byte to get delivered - these extra 4+ CPU cycles that the idle
spi clock is providing allows bigger transfers to work at faster speeds
(assuming highly optimized MCU code).
So forcing the transfer to use PIO mode to geth this behaviour for
longer transfers is of advantage.

The oposite is true when (ab)using the spi MOSI line alone to control a
WS2812B RGB LED stripe, where the DMA mode makes it possible to use the
MOSI lines alone (with correct encoding) to control the LED.

So both extremes have there use cases and with this patch we allow to
control the policy on a controller wide basis.

Right now we only allow to control the limit on a controller wide basis,
but in the future it may be possible to configure this on a per spi_device
basis.

Signed-off-by: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>
---
 drivers/spi/spi-bcm2835.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 2d702db8689c..755e1e6d2252 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -73,7 +73,6 @@
 
 #define BCM2835_SPI_FIFO_SIZE		64
 #define BCM2835_SPI_FIFO_SIZE_3_4	48
-#define BCM2835_SPI_DMA_MIN_LENGTH	96
 #define BCM2835_SPI_MODE_BITS	(SPI_CPOL | SPI_CPHA | SPI_CS_HIGH \
 				| SPI_NO_CS | SPI_3WIRE)
 
@@ -85,6 +84,12 @@ module_param(polling_limit_us, uint, 0664);
 MODULE_PARM_DESC(polling_limit_us,
 		 "time in us to run a transfer in polling mode\n");
 
+/* define minimum transfer length for which we may use DMA */
+unsigned int dma_min_bytes_limit = 96;
+module_param(dma_min_bytes_limit, uint, 0664);
+MODULE_PARM_DESC(dma_min_bytes_limit,
+		 "minimum number of bytes to run a transfer in dma mode - 0 disables dma-mode completely\n");
+
 /**
  * struct bcm2835_spi - BCM2835 SPI controller
  * @regs: base address of register map
@@ -631,7 +636,7 @@ static bool bcm2835_spi_can_dma(struct spi_master *master,
 				struct spi_transfer *tfr)
 {
 	/* we start DMA efforts only on bigger transfers */
-	if (tfr->len < BCM2835_SPI_DMA_MIN_LENGTH)
+	if (!dma_min_bytes_limit || tfr->len < dma_min_bytes_limit)
 		return false;
 
 	/* BCM2835_SPI_DLEN has defined a max transfer size as
-- 
2.11.0

[PATCH V2 5/6] spi: bcm2835: make the lower limit for dma mode configurable

[kernel]

From: Martin Sperl <kernel@martin.sperl.org>

Under some circumstances the default 96 byte limit is not optimal.

The polling and interrupt mode of the spi-bcm2835 controller exhibit a
1 idle clock cycle which is missing when using DMA.

So when transferring a longer spi_transfer to some low spec MCU devices
without SPI FIFOs like a ATMega it means that the SPI has to get reduced
by a factor of 2 or more when using DMA mode compared to using PIO modes
to give the MCU enough time to handle the incoming spi byte and prepare
the next byte to get delivered - these extra 4+ CPU cycles that the idle
spi clock is providing allows bigger transfers to work at faster speeds
(assuming highly optimized MCU code).
So forcing the transfer to use PIO mode to geth this behaviour for
longer transfers is of advantage.

The oposite is true when (ab)using the spi MOSI line alone to control a
WS2812B RGB LED stripe, where the DMA mode makes it possible to use the
MOSI lines alone (with correct encoding) to control the LED.

So both extremes have there use cases and with this patch we allow to
control the policy on a controller wide basis.

Right now we only allow to control the limit on a controller wide basis,
but in the future it may be possible to configure this on a per spi_device
basis.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
---
 drivers/spi/spi-bcm2835.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 2d702db8689c..755e1e6d2252 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -73,7 +73,6 @@
 
 #define BCM2835_SPI_FIFO_SIZE		64
 #define BCM2835_SPI_FIFO_SIZE_3_4	48
-#define BCM2835_SPI_DMA_MIN_LENGTH	96
 #define BCM2835_SPI_MODE_BITS	(SPI_CPOL | SPI_CPHA | SPI_CS_HIGH \
 				| SPI_NO_CS | SPI_3WIRE)
 
@@ -85,6 +84,12 @@ module_param(polling_limit_us, uint, 0664);
 MODULE_PARM_DESC(polling_limit_us,
 		 "time in us to run a transfer in polling mode\n");
 
+/* define minimum transfer length for which we may use DMA */
+unsigned int dma_min_bytes_limit = 96;
+module_param(dma_min_bytes_limit, uint, 0664);
+MODULE_PARM_DESC(dma_min_bytes_limit,
+		 "minimum number of bytes to run a transfer in dma mode - 0 disables dma-mode completely\n");
+
 /**
  * struct bcm2835_spi - BCM2835 SPI controller
  * @regs: base address of register map
@@ -631,7 +636,7 @@ static bool bcm2835_spi_can_dma(struct spi_master *master,
 				struct spi_transfer *tfr)
 {
 	/* we start DMA efforts only on bigger transfers */
-	if (tfr->len < BCM2835_SPI_DMA_MIN_LENGTH)
+	if (!dma_min_bytes_limit || tfr->len < dma_min_bytes_limit)
 		return false;
 
 	/* BCM2835_SPI_DLEN has defined a max transfer size as
-- 
2.11.0


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH V2 6/6] spi: bcm2835: add driver stats to debugfs

[kernel-TqfNSX0MhmxHKSADF0wUEw]

From: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>

To estimate efficiency add statistics on transfer types
(polling, interrupt and dma) used to debugfs.

Signed-off-by: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions
	    fixed compile issue when CONFIG_DEBUG_FS is unset

---
 drivers/spi/spi-bcm2835.c | 74 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 74 insertions(+)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 755e1e6d2252..83a67eb76325 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -22,6 +22,7 @@

 #include <linux/clk.h>
 #include <linux/completion.h>
+#include <linux/debugfs.h>
 #include <linux/delay.h>
 #include <linux/dma-mapping.h>
 #include <linux/dmaengine.h>
@@ -106,6 +107,15 @@ MODULE_PARM_DESC(dma_min_bytes_limit,
  *	length is not a multiple of 4 (to overcome hardware limitation)
  * @tx_spillover: whether @tx_prologue spills over to second TX sglist entry
  * @dma_pending: whether a DMA transfer is in progress
+ * @debugfs_dir: the debugfs directory - neede to remove debugfs when
+ *      unloading the module
+ * @count_transfer_polling: count of how often polling mode is used
+ * @count_transfer_irq: count of how often interrupt mode is used
+ * @count_transfer_irq_after_polling: count of how often we fall back to
+ *      interrupt mode after starting in polling mode.
+ *      These are counted as well in @count_transfer_polling and
+ *      @count_transfer_irq
+ * @count_transfer_dma: count how often dma mode is used
  */
 struct bcm2835_spi {
 	void __iomem *regs;
@@ -120,8 +130,55 @@ struct bcm2835_spi {
 	int rx_prologue;
 	unsigned int tx_spillover;
 	unsigned int dma_pending;
+
+	struct dentry *debugfs_dir;
+	u64 count_transfer_polling;
+	u64 count_transfer_irq;
+	u64 count_transfer_irq_after_polling;
+	u64 count_transfer_dma;
 };

+#if defined(CONFIG_DEBUG_FS)
+static void bcm2835_debugfs_create(struct bcm2835_spi *bs,
+				   const char *dname)
+{
+	char name[64];
+	struct dentry *dir;
+
+	/* get full name */
+	snprintf(name, sizeof(name), "spi-bcm2835-%s", dname);
+
+	/* the base directory */
+	dir = debugfs_create_dir(name, NULL);
+	bs->debugfs_dir = dir;
+
+	/* the counters */
+	debugfs_create_u64("count_transfer_polling", 0444, dir,
+			   &bs->count_transfer_polling);
+	debugfs_create_u64("count_transfer_irq", 0444, dir,
+			   &bs->count_transfer_irq);
+	debugfs_create_u64("count_transfer_irq_after_polling", 0444, dir,
+			   &bs->count_transfer_irq_after_polling);
+	debugfs_create_u64("count_transfer_dma", 0444, dir,
+			   &bs->count_transfer_dma);
+}
+
+static void bcm2835_debugfs_remove(struct bcm2835_spi *bs)
+{
+	debugfs_remove_recursive(bs->debugfs_dir);
+	bs->debugfs_dir = NULL;
+}
+#else
+static void bcm2835_debugfs_create(struct bcm2835_spi *bs,
+				   const char *dname)
+{
+}
+
+static void bcm2835_debugfs_remove(struct bcm2835_spi *bs)
+{
+}
+#endif /* CONFIG_DEBUG_FS */
+
 static inline u32 bcm2835_rd(struct bcm2835_spi *bs, unsigned reg)
 {
 	return readl(bs->regs + reg);
@@ -325,6 +382,9 @@ static int bcm2835_spi_transfer_one_irq(struct spi_master *master,
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);

+	/* update usage statistics */
+	bs->count_transfer_irq++;
+
 	/*
 	 * Enable HW block, but with interrupts still disabled.
 	 * Otherwise the empty TX FIFO would immediately trigger an interrupt.
@@ -583,6 +643,9 @@ static int bcm2835_spi_transfer_one_dma(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	int ret;

+	/* update usage statistics */
+	bs->count_transfer_dma++;
+
 	/*
 	 * Transfer first few bytes without DMA if length of first TX or RX
 	 * sglist entry is not a multiple of 4 bytes (hardware limitation).
@@ -739,6 +802,9 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long timeout;

+	/* update usage statistics */
+	bs->count_transfer_polling++;
+
 	/* enable HW block without interrupts */
 	bcm2835_wr(bs, BCM2835_SPI_CS, cs | BCM2835_SPI_CS_TA);

@@ -768,6 +834,10 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 					    jiffies - timeout,
 					    bs->tx_len, bs->rx_len);
 			/* fall back to interrupt mode */
+
+			/* update usage statistics */
+			bs->count_transfer_irq_after_polling++;
+
 			return bcm2835_spi_transfer_one_irq(master, spi,
 							    tfr, cs, false);
 		}
@@ -1004,6 +1074,8 @@ static int bcm2835_spi_probe(struct platform_device *pdev)
 		goto out_clk_disable;
 	}

+	bcm2835_debugfs_create(bs, dev_name(&pdev->dev));
+
 	return 0;

 out_clk_disable:
@@ -1018,6 +1090,8 @@ static int bcm2835_spi_remove(struct platform_device *pdev)
 	struct spi_master *master = platform_get_drvdata(pdev);
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);

+	bcm2835_debugfs_remove(bs);
+
 	/* Clear FIFOs, and disable the HW block */
 	bcm2835_wr(bs, BCM2835_SPI_CS,
 		   BCM2835_SPI_CS_CLEAR_RX | BCM2835_SPI_CS_CLEAR_TX);
--
2.11.0

[PATCH V2 6/6] spi: bcm2835: add driver stats to debugfs

[kernel]

From: Martin Sperl <kernel@martin.sperl.org>

To estimate efficiency add statistics on transfer types
(polling, interrupt and dma) used to debugfs.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions
	    fixed compile issue when CONFIG_DEBUG_FS is unset

---
 drivers/spi/spi-bcm2835.c | 74 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 74 insertions(+)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 755e1e6d2252..83a67eb76325 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -22,6 +22,7 @@

 #include <linux/clk.h>
 #include <linux/completion.h>
+#include <linux/debugfs.h>
 #include <linux/delay.h>
 #include <linux/dma-mapping.h>
 #include <linux/dmaengine.h>
@@ -106,6 +107,15 @@ MODULE_PARM_DESC(dma_min_bytes_limit,
  *	length is not a multiple of 4 (to overcome hardware limitation)
  * @tx_spillover: whether @tx_prologue spills over to second TX sglist entry
  * @dma_pending: whether a DMA transfer is in progress
+ * @debugfs_dir: the debugfs directory - neede to remove debugfs when
+ *      unloading the module
+ * @count_transfer_polling: count of how often polling mode is used
+ * @count_transfer_irq: count of how often interrupt mode is used
+ * @count_transfer_irq_after_polling: count of how often we fall back to
+ *      interrupt mode after starting in polling mode.
+ *      These are counted as well in @count_transfer_polling and
+ *      @count_transfer_irq
+ * @count_transfer_dma: count how often dma mode is used
  */
 struct bcm2835_spi {
 	void __iomem *regs;
@@ -120,8 +130,55 @@ struct bcm2835_spi {
 	int rx_prologue;
 	unsigned int tx_spillover;
 	unsigned int dma_pending;
+
+	struct dentry *debugfs_dir;
+	u64 count_transfer_polling;
+	u64 count_transfer_irq;
+	u64 count_transfer_irq_after_polling;
+	u64 count_transfer_dma;
 };

+#if defined(CONFIG_DEBUG_FS)
+static void bcm2835_debugfs_create(struct bcm2835_spi *bs,
+				   const char *dname)
+{
+	char name[64];
+	struct dentry *dir;
+
+	/* get full name */
+	snprintf(name, sizeof(name), "spi-bcm2835-%s", dname);
+
+	/* the base directory */
+	dir = debugfs_create_dir(name, NULL);
+	bs->debugfs_dir = dir;
+
+	/* the counters */
+	debugfs_create_u64("count_transfer_polling", 0444, dir,
+			   &bs->count_transfer_polling);
+	debugfs_create_u64("count_transfer_irq", 0444, dir,
+			   &bs->count_transfer_irq);
+	debugfs_create_u64("count_transfer_irq_after_polling", 0444, dir,
+			   &bs->count_transfer_irq_after_polling);
+	debugfs_create_u64("count_transfer_dma", 0444, dir,
+			   &bs->count_transfer_dma);
+}
+
+static void bcm2835_debugfs_remove(struct bcm2835_spi *bs)
+{
+	debugfs_remove_recursive(bs->debugfs_dir);
+	bs->debugfs_dir = NULL;
+}
+#else
+static void bcm2835_debugfs_create(struct bcm2835_spi *bs,
+				   const char *dname)
+{
+}
+
+static void bcm2835_debugfs_remove(struct bcm2835_spi *bs)
+{
+}
+#endif /* CONFIG_DEBUG_FS */
+
 static inline u32 bcm2835_rd(struct bcm2835_spi *bs, unsigned reg)
 {
 	return readl(bs->regs + reg);
@@ -325,6 +382,9 @@ static int bcm2835_spi_transfer_one_irq(struct spi_master *master,
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);

+	/* update usage statistics */
+	bs->count_transfer_irq++;
+
 	/*
 	 * Enable HW block, but with interrupts still disabled.
 	 * Otherwise the empty TX FIFO would immediately trigger an interrupt.
@@ -583,6 +643,9 @@ static int bcm2835_spi_transfer_one_dma(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	int ret;

+	/* update usage statistics */
+	bs->count_transfer_dma++;
+
 	/*
 	 * Transfer first few bytes without DMA if length of first TX or RX
 	 * sglist entry is not a multiple of 4 bytes (hardware limitation).
@@ -739,6 +802,9 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long timeout;

+	/* update usage statistics */
+	bs->count_transfer_polling++;
+
 	/* enable HW block without interrupts */
 	bcm2835_wr(bs, BCM2835_SPI_CS, cs | BCM2835_SPI_CS_TA);

@@ -768,6 +834,10 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 					    jiffies - timeout,
 					    bs->tx_len, bs->rx_len);
 			/* fall back to interrupt mode */
+
+			/* update usage statistics */
+			bs->count_transfer_irq_after_polling++;
+
 			return bcm2835_spi_transfer_one_irq(master, spi,
 							    tfr, cs, false);
 		}
@@ -1004,6 +1074,8 @@ static int bcm2835_spi_probe(struct platform_device *pdev)
 		goto out_clk_disable;
 	}

+	bcm2835_debugfs_create(bs, dev_name(&pdev->dev));
+
 	return 0;

 out_clk_disable:
@@ -1018,6 +1090,8 @@ static int bcm2835_spi_remove(struct platform_device *pdev)
 	struct spi_master *master = platform_get_drvdata(pdev);
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);

+	bcm2835_debugfs_remove(bs);
+
 	/* Clear FIFOs, and disable the HW block */
 	bcm2835_wr(bs, BCM2835_SPI_CS,
 		   BCM2835_SPI_CS_CLEAR_RX | BCM2835_SPI_CS_CLEAR_TX);
--
2.11.0

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH V2 1/6] spi: bcm2835: bcm2835_spi_transfer_one_poll remove unnecessary argument

[Mark Brown]

[-- Attachment #1.1: Type: text/plain, Size: 509 bytes --]

On Tue, Apr 23, 2019 at 08:15:08PM +0000, kernel@martin.sperl.org wrote:
> From: Martin Sperl <kernel@martin.sperl.org>
> 
> Remove the unnecessary argument of xfer_time_us when calling
> bcm2835_spi_transfer_one_poll.
> 
> Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
> 
> Changelog:
>   V1 -> V2: applied feedback by Stefan Wahren
>             reorganized patchset
> 	    added extra rational, descriptions
> 
> ---

As per SubmittingPatches the changelog should go after the ---.

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

[-- Attachment #2: Type: text/plain, Size: 176 bytes --]

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH V2 5/6] spi: bcm2835: make the lower limit for dma mode configurable

[Mark Brown]

[-- Attachment #1.1: Type: text/plain, Size: 866 bytes --]

On Wed, Apr 24, 2019 at 09:07:12AM +0200, Lukas Wunner wrote:

> Indeed I'd prefer if another bit is added to "mode" in struct device
> to represent the need for another clock cycle in-between bytes.
> The SPI core could then reduce the clock speed based on this flag
> and the problem would be solved for everyone.

> Influencing this behavior with a module parameter feels a bit like a
> kludge and I fear may stay indefinitely even if a better solution
> is implemented later.

This does feel like we know enough to have a more advanced function in
the driver given a bit of information about the client device
requireemnts.  Though it's going to be complex to express them,
especially with the ATMega case where we want fast individual clocks but
lots of dead space in between bytes (is the controller capable of adding
that dead space itself in DMA mode BTW?).

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

[-- Attachment #2: Type: text/plain, Size: 176 bytes --]

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Applied "spi: bcm2835: add driver stats to debugfs" to the spi tree

[Mark Brown]

The patch

   spi: bcm2835: add driver stats to debugfs

has been applied to the spi tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

>From 154f7da56f1ecba42021d550c9e8432ac8d32c26 Mon Sep 17 00:00:00 2001
From: Martin Sperl <kernel@martin.sperl.org>
Date: Tue, 23 Apr 2019 20:15:13 +0000
Subject: [PATCH] spi: bcm2835: add driver stats to debugfs

To estimate efficiency add statistics on transfer types
(polling, interrupt and dma) used to debugfs.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions
	    fixed compile issue when CONFIG_DEBUG_FS is unset
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 drivers/spi/spi-bcm2835.c | 74 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 74 insertions(+)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 3230d37fa89a..eb67da697ef5 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -22,6 +22,7 @@
 
 #include <linux/clk.h>
 #include <linux/completion.h>
+#include <linux/debugfs.h>
 #include <linux/delay.h>
 #include <linux/dma-mapping.h>
 #include <linux/dmaengine.h>
@@ -101,6 +102,15 @@ MODULE_PARM_DESC(polling_limit_us,
  *	length is not a multiple of 4 (to overcome hardware limitation)
  * @tx_spillover: whether @tx_prologue spills over to second TX sglist entry
  * @dma_pending: whether a DMA transfer is in progress
+ * @debugfs_dir: the debugfs directory - neede to remove debugfs when
+ *      unloading the module
+ * @count_transfer_polling: count of how often polling mode is used
+ * @count_transfer_irq: count of how often interrupt mode is used
+ * @count_transfer_irq_after_polling: count of how often we fall back to
+ *      interrupt mode after starting in polling mode.
+ *      These are counted as well in @count_transfer_polling and
+ *      @count_transfer_irq
+ * @count_transfer_dma: count how often dma mode is used
  */
 struct bcm2835_spi {
 	void __iomem *regs;
@@ -115,8 +125,55 @@ struct bcm2835_spi {
 	int rx_prologue;
 	unsigned int tx_spillover;
 	unsigned int dma_pending;
+
+	struct dentry *debugfs_dir;
+	u64 count_transfer_polling;
+	u64 count_transfer_irq;
+	u64 count_transfer_irq_after_polling;
+	u64 count_transfer_dma;
 };
 
+#if defined(CONFIG_DEBUG_FS)
+static void bcm2835_debugfs_create(struct bcm2835_spi *bs,
+				   const char *dname)
+{
+	char name[64];
+	struct dentry *dir;
+
+	/* get full name */
+	snprintf(name, sizeof(name), "spi-bcm2835-%s", dname);
+
+	/* the base directory */
+	dir = debugfs_create_dir(name, NULL);
+	bs->debugfs_dir = dir;
+
+	/* the counters */
+	debugfs_create_u64("count_transfer_polling", 0444, dir,
+			   &bs->count_transfer_polling);
+	debugfs_create_u64("count_transfer_irq", 0444, dir,
+			   &bs->count_transfer_irq);
+	debugfs_create_u64("count_transfer_irq_after_polling", 0444, dir,
+			   &bs->count_transfer_irq_after_polling);
+	debugfs_create_u64("count_transfer_dma", 0444, dir,
+			   &bs->count_transfer_dma);
+}
+
+static void bcm2835_debugfs_remove(struct bcm2835_spi *bs)
+{
+	debugfs_remove_recursive(bs->debugfs_dir);
+	bs->debugfs_dir = NULL;
+}
+#else
+static void bcm2835_debugfs_create(struct bcm2835_spi *bs,
+				   const char *dname)
+{
+}
+
+static void bcm2835_debugfs_remove(struct bcm2835_spi *bs)
+{
+}
+#endif /* CONFIG_DEBUG_FS */
+
 static inline u32 bcm2835_rd(struct bcm2835_spi *bs, unsigned reg)
 {
 	return readl(bs->regs + reg);
@@ -320,6 +377,9 @@ static int bcm2835_spi_transfer_one_irq(struct spi_master *master,
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 
+	/* update usage statistics */
+	bs->count_transfer_irq++;
+
 	/*
 	 * Enable HW block, but with interrupts still disabled.
 	 * Otherwise the empty TX FIFO would immediately trigger an interrupt.
@@ -564,6 +624,9 @@ static int bcm2835_spi_transfer_one_dma(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	int ret;
 
+	/* update usage statistics */
+	bs->count_transfer_dma++;
+
 	/*
 	 * Transfer first few bytes without DMA if length of first TX or RX
 	 * sglist entry is not a multiple of 4 bytes (hardware limitation).
@@ -706,6 +769,9 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long timeout;
 
+	/* update usage statistics */
+	bs->count_transfer_polling++;
+
 	/* enable HW block without interrupts */
 	bcm2835_wr(bs, BCM2835_SPI_CS, cs | BCM2835_SPI_CS_TA);
 
@@ -735,6 +801,10 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 					    jiffies - timeout,
 					    bs->tx_len, bs->rx_len);
 			/* fall back to interrupt mode */
+
+			/* update usage statistics */
+			bs->count_transfer_irq_after_polling++;
+
 			return bcm2835_spi_transfer_one_irq(master, spi,
 							    tfr, cs, false);
 		}
@@ -982,6 +1052,8 @@ static int bcm2835_spi_probe(struct platform_device *pdev)
 		goto out_clk_disable;
 	}
 
+	bcm2835_debugfs_create(bs, dev_name(&pdev->dev));
+
 	return 0;
 
 out_clk_disable:
@@ -996,6 +1068,8 @@ static int bcm2835_spi_remove(struct platform_device *pdev)
 	struct spi_master *master = platform_get_drvdata(pdev);
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 
+	bcm2835_debugfs_remove(bs);
+
 	/* Clear FIFOs, and disable the HW block */
 	bcm2835_wr(bs, BCM2835_SPI_CS,
 		   BCM2835_SPI_CS_CLEAR_RX | BCM2835_SPI_CS_CLEAR_TX);
-- 
2.20.1

Applied "spi: bcm2835: add driver stats to debugfs" to the spi tree

[Mark Brown]

The patch

   spi: bcm2835: add driver stats to debugfs

has been applied to the spi tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

From 154f7da56f1ecba42021d550c9e8432ac8d32c26 Mon Sep 17 00:00:00 2001
From: Martin Sperl <kernel@martin.sperl.org>
Date: Tue, 23 Apr 2019 20:15:13 +0000
Subject: [PATCH] spi: bcm2835: add driver stats to debugfs

To estimate efficiency add statistics on transfer types
(polling, interrupt and dma) used to debugfs.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions
	    fixed compile issue when CONFIG_DEBUG_FS is unset
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 drivers/spi/spi-bcm2835.c | 74 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 74 insertions(+)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 3230d37fa89a..eb67da697ef5 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -22,6 +22,7 @@
 
 #include <linux/clk.h>
 #include <linux/completion.h>
+#include <linux/debugfs.h>
 #include <linux/delay.h>
 #include <linux/dma-mapping.h>
 #include <linux/dmaengine.h>
@@ -101,6 +102,15 @@ MODULE_PARM_DESC(polling_limit_us,
  *	length is not a multiple of 4 (to overcome hardware limitation)
  * @tx_spillover: whether @tx_prologue spills over to second TX sglist entry
  * @dma_pending: whether a DMA transfer is in progress
+ * @debugfs_dir: the debugfs directory - neede to remove debugfs when
+ *      unloading the module
+ * @count_transfer_polling: count of how often polling mode is used
+ * @count_transfer_irq: count of how often interrupt mode is used
+ * @count_transfer_irq_after_polling: count of how often we fall back to
+ *      interrupt mode after starting in polling mode.
+ *      These are counted as well in @count_transfer_polling and
+ *      @count_transfer_irq
+ * @count_transfer_dma: count how often dma mode is used
  */
 struct bcm2835_spi {
 	void __iomem *regs;
@@ -115,8 +125,55 @@ struct bcm2835_spi {
 	int rx_prologue;
 	unsigned int tx_spillover;
 	unsigned int dma_pending;
+
+	struct dentry *debugfs_dir;
+	u64 count_transfer_polling;
+	u64 count_transfer_irq;
+	u64 count_transfer_irq_after_polling;
+	u64 count_transfer_dma;
 };
 
+#if defined(CONFIG_DEBUG_FS)
+static void bcm2835_debugfs_create(struct bcm2835_spi *bs,
+				   const char *dname)
+{
+	char name[64];
+	struct dentry *dir;
+
+	/* get full name */
+	snprintf(name, sizeof(name), "spi-bcm2835-%s", dname);
+
+	/* the base directory */
+	dir = debugfs_create_dir(name, NULL);
+	bs->debugfs_dir = dir;
+
+	/* the counters */
+	debugfs_create_u64("count_transfer_polling", 0444, dir,
+			   &bs->count_transfer_polling);
+	debugfs_create_u64("count_transfer_irq", 0444, dir,
+			   &bs->count_transfer_irq);
+	debugfs_create_u64("count_transfer_irq_after_polling", 0444, dir,
+			   &bs->count_transfer_irq_after_polling);
+	debugfs_create_u64("count_transfer_dma", 0444, dir,
+			   &bs->count_transfer_dma);
+}
+
+static void bcm2835_debugfs_remove(struct bcm2835_spi *bs)
+{
+	debugfs_remove_recursive(bs->debugfs_dir);
+	bs->debugfs_dir = NULL;
+}
+#else
+static void bcm2835_debugfs_create(struct bcm2835_spi *bs,
+				   const char *dname)
+{
+}
+
+static void bcm2835_debugfs_remove(struct bcm2835_spi *bs)
+{
+}
+#endif /* CONFIG_DEBUG_FS */
+
 static inline u32 bcm2835_rd(struct bcm2835_spi *bs, unsigned reg)
 {
 	return readl(bs->regs + reg);
@@ -320,6 +377,9 @@ static int bcm2835_spi_transfer_one_irq(struct spi_master *master,
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 
+	/* update usage statistics */
+	bs->count_transfer_irq++;
+
 	/*
 	 * Enable HW block, but with interrupts still disabled.
 	 * Otherwise the empty TX FIFO would immediately trigger an interrupt.
@@ -564,6 +624,9 @@ static int bcm2835_spi_transfer_one_dma(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	int ret;
 
+	/* update usage statistics */
+	bs->count_transfer_dma++;
+
 	/*
 	 * Transfer first few bytes without DMA if length of first TX or RX
 	 * sglist entry is not a multiple of 4 bytes (hardware limitation).
@@ -706,6 +769,9 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long timeout;
 
+	/* update usage statistics */
+	bs->count_transfer_polling++;
+
 	/* enable HW block without interrupts */
 	bcm2835_wr(bs, BCM2835_SPI_CS, cs | BCM2835_SPI_CS_TA);
 
@@ -735,6 +801,10 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 					    jiffies - timeout,
 					    bs->tx_len, bs->rx_len);
 			/* fall back to interrupt mode */
+
+			/* update usage statistics */
+			bs->count_transfer_irq_after_polling++;
+
 			return bcm2835_spi_transfer_one_irq(master, spi,
 							    tfr, cs, false);
 		}
@@ -982,6 +1052,8 @@ static int bcm2835_spi_probe(struct platform_device *pdev)
 		goto out_clk_disable;
 	}
 
+	bcm2835_debugfs_create(bs, dev_name(&pdev->dev));
+
 	return 0;
 
 out_clk_disable:
@@ -996,6 +1068,8 @@ static int bcm2835_spi_remove(struct platform_device *pdev)
 	struct spi_master *master = platform_get_drvdata(pdev);
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 
+	bcm2835_debugfs_remove(bs);
+
 	/* Clear FIFOs, and disable the HW block */
 	bcm2835_wr(bs, BCM2835_SPI_CS,
 		   BCM2835_SPI_CS_CLEAR_RX | BCM2835_SPI_CS_CLEAR_TX);
-- 
2.20.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Applied "spi: bcm2835: make the polling duration limits configurable" to the spi tree

[Mark Brown]

The patch

   spi: bcm2835: make the polling duration limits configurable

has been applied to the spi tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

>From ff245d90ebed8d4da6751dfee1bc76e4a5e94257 Mon Sep 17 00:00:00 2001
From: Martin Sperl <kernel@martin.sperl.org>
Date: Tue, 23 Apr 2019 20:15:11 +0000
Subject: [PATCH] spi: bcm2835: make the polling duration limits configurable

Under some circumstances the default 30 us polling limit is not optimal
and may lead to long delays because we are waiting on an interrupt.
with this patch we have the possibility to influence this policy.

So make this limit (in us) configurable via a module parameters
(but also modifyable via /sys/modules/...)

This replicates similar code found in spi-bcm2835aux.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 drivers/spi/spi-bcm2835.c | 22 ++++++++++++++--------
 1 file changed, 14 insertions(+), 8 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 0d8c97502f14..3230d37fa89a 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -73,14 +73,18 @@
 
 #define BCM2835_SPI_FIFO_SIZE		64
 #define BCM2835_SPI_FIFO_SIZE_3_4	48
-#define BCM2835_SPI_POLLING_LIMIT_US	30
-#define BCM2835_SPI_POLLING_JIFFIES	2
 #define BCM2835_SPI_DMA_MIN_LENGTH	96
 #define BCM2835_SPI_MODE_BITS	(SPI_CPOL | SPI_CPHA | SPI_CS_HIGH \
 				| SPI_NO_CS | SPI_3WIRE)
 
 #define DRV_NAME	"spi-bcm2835"
 
+/* define polling limits */
+unsigned int polling_limit_us = 30;
+module_param(polling_limit_us, uint, 0664);
+MODULE_PARM_DESC(polling_limit_us,
+		 "time in us to run a transfer in polling mode\n");
+
 /**
  * struct bcm2835_spi - BCM2835 SPI controller
  * @regs: base address of register map
@@ -711,8 +715,8 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 	 */
 	bcm2835_wr_fifo_blind(bs, BCM2835_SPI_FIFO_SIZE);
 
-	/* set the timeout */
-	timeout = jiffies + BCM2835_SPI_POLLING_JIFFIES;
+	/* set the timeout to at least 2 jiffies */
+	timeout = jiffies + 2 + HZ * polling_limit_us / 1000000;
 
 	/* loop until finished the transfer */
 	while (bs->rx_len) {
@@ -747,8 +751,8 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 				    struct spi_transfer *tfr)
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
-	unsigned long spi_hz, clk_hz, cdiv;
-	unsigned long spi_used_hz;
+	unsigned long spi_hz, clk_hz, cdiv, spi_used_hz;
+	unsigned long hz_per_byte, byte_limit;
 	u32 cs = bcm2835_rd(bs, BCM2835_SPI_CS);
 
 	/* set clock */
@@ -795,9 +799,11 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	 * per byte per polling limit.  E.g., we can transfer 1 byte in 30 us
 	 * per 300,000 Hz of bus clock.
 	 */
-#define HZ_PER_BYTE ((9 * 1000000) / BCM2835_SPI_POLLING_LIMIT_US)
+	hz_per_byte = polling_limit_us ? (9 * 1000000) / polling_limit_us : 0;
+	byte_limit = hz_per_byte ? spi_used_hz / hz_per_byte : 1;
+
 	/* run in polling mode for short transfers */
-	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
+	if (tfr->len < byte_limit)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);
 
 	/* run in dma mode if conditions are right
-- 
2.20.1

Applied "spi: bcm2835: make the polling duration limits configurable" to the spi tree

[Mark Brown]

The patch

   spi: bcm2835: make the polling duration limits configurable

has been applied to the spi tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

From ff245d90ebed8d4da6751dfee1bc76e4a5e94257 Mon Sep 17 00:00:00 2001
From: Martin Sperl <kernel@martin.sperl.org>
Date: Tue, 23 Apr 2019 20:15:11 +0000
Subject: [PATCH] spi: bcm2835: make the polling duration limits configurable

Under some circumstances the default 30 us polling limit is not optimal
and may lead to long delays because we are waiting on an interrupt.
with this patch we have the possibility to influence this policy.

So make this limit (in us) configurable via a module parameters
(but also modifyable via /sys/modules/...)

This replicates similar code found in spi-bcm2835aux.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 drivers/spi/spi-bcm2835.c | 22 ++++++++++++++--------
 1 file changed, 14 insertions(+), 8 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 0d8c97502f14..3230d37fa89a 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -73,14 +73,18 @@
 
 #define BCM2835_SPI_FIFO_SIZE		64
 #define BCM2835_SPI_FIFO_SIZE_3_4	48
-#define BCM2835_SPI_POLLING_LIMIT_US	30
-#define BCM2835_SPI_POLLING_JIFFIES	2
 #define BCM2835_SPI_DMA_MIN_LENGTH	96
 #define BCM2835_SPI_MODE_BITS	(SPI_CPOL | SPI_CPHA | SPI_CS_HIGH \
 				| SPI_NO_CS | SPI_3WIRE)
 
 #define DRV_NAME	"spi-bcm2835"
 
+/* define polling limits */
+unsigned int polling_limit_us = 30;
+module_param(polling_limit_us, uint, 0664);
+MODULE_PARM_DESC(polling_limit_us,
+		 "time in us to run a transfer in polling mode\n");
+
 /**
  * struct bcm2835_spi - BCM2835 SPI controller
  * @regs: base address of register map
@@ -711,8 +715,8 @@ static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 	 */
 	bcm2835_wr_fifo_blind(bs, BCM2835_SPI_FIFO_SIZE);
 
-	/* set the timeout */
-	timeout = jiffies + BCM2835_SPI_POLLING_JIFFIES;
+	/* set the timeout to at least 2 jiffies */
+	timeout = jiffies + 2 + HZ * polling_limit_us / 1000000;
 
 	/* loop until finished the transfer */
 	while (bs->rx_len) {
@@ -747,8 +751,8 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 				    struct spi_transfer *tfr)
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
-	unsigned long spi_hz, clk_hz, cdiv;
-	unsigned long spi_used_hz;
+	unsigned long spi_hz, clk_hz, cdiv, spi_used_hz;
+	unsigned long hz_per_byte, byte_limit;
 	u32 cs = bcm2835_rd(bs, BCM2835_SPI_CS);
 
 	/* set clock */
@@ -795,9 +799,11 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	 * per byte per polling limit.  E.g., we can transfer 1 byte in 30 us
 	 * per 300,000 Hz of bus clock.
 	 */
-#define HZ_PER_BYTE ((9 * 1000000) / BCM2835_SPI_POLLING_LIMIT_US)
+	hz_per_byte = polling_limit_us ? (9 * 1000000) / polling_limit_us : 0;
+	byte_limit = hz_per_byte ? spi_used_hz / hz_per_byte : 1;
+
 	/* run in polling mode for short transfers */
-	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
+	if (tfr->len < byte_limit)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);
 
 	/* run in dma mode if conditions are right
-- 
2.20.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Applied "spi: bcm2835: Avoid 64-bit arithmetic in xfer len calc" to the spi tree

[Mark Brown]

The patch

   spi: bcm2835: Avoid 64-bit arithmetic in xfer len calc

has been applied to the spi tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

>From 7f1922eb3adba8c372fdf0d04fe9e6268f98291a Mon Sep 17 00:00:00 2001
From: Martin Sperl <kernel@martin.sperl.org>
Date: Tue, 23 Apr 2019 20:15:09 +0000
Subject: [PATCH] spi: bcm2835: Avoid 64-bit arithmetic in xfer len calc

Avoid 64 bit aritmetics when deciding if we need to use polling or not
This replicates: commit d704afffe65c
("spi: bcm2835aux: Avoid 64-bit arithmetic in xfer len calc")
from spi-bcm2835aux

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 drivers/spi/spi-bcm2835.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 6c73e694794a..b7bb07c9507d 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -749,7 +749,6 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long spi_hz, clk_hz, cdiv;
 	unsigned long spi_used_hz;
-	unsigned long long xfer_time_us;
 	u32 cs = bcm2835_rd(bs, BCM2835_SPI_CS);
 
 	/* set clock */
@@ -790,14 +789,15 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	bs->tx_len = tfr->len;
 	bs->rx_len = tfr->len;
 
-	/* calculate the estimated time in us the transfer runs */
-	xfer_time_us = (unsigned long long)tfr->len
-		* 9 /* clocks/byte - SPI-HW waits 1 clock after each byte */
-		* 1000000;
-	do_div(xfer_time_us, spi_used_hz);
-
-	/* for short requests run polling*/
-	if (xfer_time_us <= BCM2835_SPI_POLLING_LIMIT_US)
+	/* Calculate the estimated time in us the transfer runs.  Note that
+	 * there is 1 idle clocks cycles after each byte getting transferred
+	 * so we have 9 cycles/byte.  This is used to find the number of Hz
+	 * per byte per polling limit.  E.g., we can transfer 1 byte in 30 us
+	 * per 300,000 Hz of bus clock.
+	 */
+#define HZ_PER_BYTE ((9 * 1000000) / BCM2835_SPI_POLLING_LIMIT_US)
+	/* run in polling mode for short transfers */
+	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);
 
 	/* run in dma mode if conditions are right */
-- 
2.20.1

Applied "spi: bcm2835: added comment about different bus behaviour of DMA mode" to the spi tree

[Mark Brown]

The patch

   spi: bcm2835: added comment about different bus behaviour of DMA mode

has been applied to the spi tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

>From c41d62b06cb92289ab5db9d37a0f27fe6271fa34 Mon Sep 17 00:00:00 2001
From: Martin Sperl <kernel@martin.sperl.org>
Date: Tue, 23 Apr 2019 20:15:10 +0000
Subject: [PATCH] spi: bcm2835: added comment about different bus behaviour of
 DMA mode

The DMA mode behaves slightly different than polling or interrupt driven
mode, so just document the fact

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            new in V2
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 drivers/spi/spi-bcm2835.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index b7bb07c9507d..0d8c97502f14 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -800,7 +800,10 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);
 
-	/* run in dma mode if conditions are right */
+	/* run in dma mode if conditions are right
+	 * Note that unlike poll or interrupt mode DMA mode does not have
+	 * this 1 idle clock cycle pattern but runs the spi clock without gaps
+	 */
 	if (master->can_dma && bcm2835_spi_can_dma(master, spi, tfr))
 		return bcm2835_spi_transfer_one_dma(master, spi, tfr, cs);
 
-- 
2.20.1

Applied "spi: bcm2835: Avoid 64-bit arithmetic in xfer len calc" to the spi tree

[Mark Brown]

The patch

   spi: bcm2835: Avoid 64-bit arithmetic in xfer len calc

has been applied to the spi tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

From 7f1922eb3adba8c372fdf0d04fe9e6268f98291a Mon Sep 17 00:00:00 2001
From: Martin Sperl <kernel@martin.sperl.org>
Date: Tue, 23 Apr 2019 20:15:09 +0000
Subject: [PATCH] spi: bcm2835: Avoid 64-bit arithmetic in xfer len calc

Avoid 64 bit aritmetics when deciding if we need to use polling or not
This replicates: commit d704afffe65c
("spi: bcm2835aux: Avoid 64-bit arithmetic in xfer len calc")
from spi-bcm2835aux

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 drivers/spi/spi-bcm2835.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 6c73e694794a..b7bb07c9507d 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -749,7 +749,6 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long spi_hz, clk_hz, cdiv;
 	unsigned long spi_used_hz;
-	unsigned long long xfer_time_us;
 	u32 cs = bcm2835_rd(bs, BCM2835_SPI_CS);
 
 	/* set clock */
@@ -790,14 +789,15 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	bs->tx_len = tfr->len;
 	bs->rx_len = tfr->len;
 
-	/* calculate the estimated time in us the transfer runs */
-	xfer_time_us = (unsigned long long)tfr->len
-		* 9 /* clocks/byte - SPI-HW waits 1 clock after each byte */
-		* 1000000;
-	do_div(xfer_time_us, spi_used_hz);
-
-	/* for short requests run polling*/
-	if (xfer_time_us <= BCM2835_SPI_POLLING_LIMIT_US)
+	/* Calculate the estimated time in us the transfer runs.  Note that
+	 * there is 1 idle clocks cycles after each byte getting transferred
+	 * so we have 9 cycles/byte.  This is used to find the number of Hz
+	 * per byte per polling limit.  E.g., we can transfer 1 byte in 30 us
+	 * per 300,000 Hz of bus clock.
+	 */
+#define HZ_PER_BYTE ((9 * 1000000) / BCM2835_SPI_POLLING_LIMIT_US)
+	/* run in polling mode for short transfers */
+	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);
 
 	/* run in dma mode if conditions are right */
-- 
2.20.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Applied "spi: bcm2835: added comment about different bus behaviour of DMA mode" to the spi tree

[Mark Brown]

The patch

   spi: bcm2835: added comment about different bus behaviour of DMA mode

has been applied to the spi tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

From c41d62b06cb92289ab5db9d37a0f27fe6271fa34 Mon Sep 17 00:00:00 2001
From: Martin Sperl <kernel@martin.sperl.org>
Date: Tue, 23 Apr 2019 20:15:10 +0000
Subject: [PATCH] spi: bcm2835: added comment about different bus behaviour of
 DMA mode

The DMA mode behaves slightly different than polling or interrupt driven
mode, so just document the fact

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            new in V2
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 drivers/spi/spi-bcm2835.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index b7bb07c9507d..0d8c97502f14 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -800,7 +800,10 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 	if (tfr->len < spi_used_hz / HZ_PER_BYTE)
 		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);
 
-	/* run in dma mode if conditions are right */
+	/* run in dma mode if conditions are right
+	 * Note that unlike poll or interrupt mode DMA mode does not have
+	 * this 1 idle clock cycle pattern but runs the spi clock without gaps
+	 */
 	if (master->can_dma && bcm2835_spi_can_dma(master, spi, tfr))
 		return bcm2835_spi_transfer_one_dma(master, spi, tfr, cs);
 
-- 
2.20.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Applied "spi: bcm2835: bcm2835_spi_transfer_one_poll remove unnecessary argument" to the spi tree

[Mark Brown]

The patch

   spi: bcm2835: bcm2835_spi_transfer_one_poll remove unnecessary argument

has been applied to the spi tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

>From 9ac3f90db46b7805a37df2bc419faf369025d64a Mon Sep 17 00:00:00 2001
From: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>
Date: Tue, 23 Apr 2019 20:15:08 +0000
Subject: [PATCH] spi: bcm2835: bcm2835_spi_transfer_one_poll remove
 unnecessary argument

Remove the unnecessary argument of xfer_time_us when calling
bcm2835_spi_transfer_one_poll.

Signed-off-by: Martin Sperl <kernel-TqfNSX0MhmxHKSADF0wUEw@public.gmane.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions
Signed-off-by: Mark Brown <broonie-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 drivers/spi/spi-bcm2835.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 8aa22713c483..6c73e694794a 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -697,8 +697,7 @@ static void bcm2835_dma_init(struct spi_master *master, struct device *dev)
 static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 					 struct spi_device *spi,
 					 struct spi_transfer *tfr,
-					 u32 cs,
-					 unsigned long long xfer_time_us)
+					 u32 cs)
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long timeout;
@@ -799,8 +798,7 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 
 	/* for short requests run polling*/
 	if (xfer_time_us <= BCM2835_SPI_POLLING_LIMIT_US)
-		return bcm2835_spi_transfer_one_poll(master, spi, tfr,
-						     cs, xfer_time_us);
+		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);
 
 	/* run in dma mode if conditions are right */
 	if (master->can_dma && bcm2835_spi_can_dma(master, spi, tfr))
-- 
2.20.1

Applied "spi: bcm2835: bcm2835_spi_transfer_one_poll remove unnecessary argument" to the spi tree

[Mark Brown]

The patch

   spi: bcm2835: bcm2835_spi_transfer_one_poll remove unnecessary argument

has been applied to the spi tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

From 9ac3f90db46b7805a37df2bc419faf369025d64a Mon Sep 17 00:00:00 2001
From: Martin Sperl <kernel@martin.sperl.org>
Date: Tue, 23 Apr 2019 20:15:08 +0000
Subject: [PATCH] spi: bcm2835: bcm2835_spi_transfer_one_poll remove
 unnecessary argument

Remove the unnecessary argument of xfer_time_us when calling
bcm2835_spi_transfer_one_poll.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Changelog:
  V1 -> V2: applied feedback by Stefan Wahren
            reorganized patchset
	    added extra rational, descriptions
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 drivers/spi/spi-bcm2835.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index 8aa22713c483..6c73e694794a 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -697,8 +697,7 @@ static void bcm2835_dma_init(struct spi_master *master, struct device *dev)
 static int bcm2835_spi_transfer_one_poll(struct spi_master *master,
 					 struct spi_device *spi,
 					 struct spi_transfer *tfr,
-					 u32 cs,
-					 unsigned long long xfer_time_us)
+					 u32 cs)
 {
 	struct bcm2835_spi *bs = spi_master_get_devdata(master);
 	unsigned long timeout;
@@ -799,8 +798,7 @@ static int bcm2835_spi_transfer_one(struct spi_master *master,
 
 	/* for short requests run polling*/
 	if (xfer_time_us <= BCM2835_SPI_POLLING_LIMIT_US)
-		return bcm2835_spi_transfer_one_poll(master, spi, tfr,
-						     cs, xfer_time_us);
+		return bcm2835_spi_transfer_one_poll(master, spi, tfr, cs);
 
 	/* run in dma mode if conditions are right */
 	if (master->can_dma && bcm2835_spi_can_dma(master, spi, tfr))
-- 
2.20.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH V2 5/6] spi: bcm2835: make the lower limit for dma mode configurable

[kernel-TqfNSX0MhmxHKSADF0wUEw]

> On 08.05.2019, at 10:42, Mark Brown <broonie-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org> wrote:
> 
> especially with the ATMega case where we want fast individual clocks but
> lots of dead space in between bytes (is the controller capable of adding
> that dead space itself in DMA mode BTW?).

No it is not - at least not so far as I know of (and I just went over the
available documentation again).

Martin

Re: [PATCH V2 5/6] spi: bcm2835: make the lower limit for dma mode configurable

[kernel]

> On 08.05.2019, at 10:42, Mark Brown <broonie@kernel.org> wrote:
> 
> especially with the ATMega case where we want fast individual clocks but
> lots of dead space in between bytes (is the controller capable of adding
> that dead space itself in DMA mode BTW?).

No it is not - at least not so far as I know of (and I just went over the
available documentation again).

Martin

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel