diff mbox series

i2c: designware: Introduce cooldown timer to AMDPSP driver

Message ID 20220725080240.106619-1-jsd@semihalf.com
State Superseded
Headers show
Series i2c: designware: Introduce cooldown timer to AMDPSP driver | expand

Commit Message

Jan Dąbroś July 25, 2022, 8:02 a.m. UTC
In order to optimize performance, limit amount of back and forth
transactions between x86 and PSP. This is done by introduction of
cooldown period - that is window in which x86 isn't releasing the bus
immediately after each I2C transaction.

In order to protect PSP from being starved while waiting for
arbitration, after a programmed time bus is automatically released by a
deferred function.

Signed-off-by: Jan Dabros <jsd@semihalf.com>
---
 drivers/i2c/busses/i2c-designware-amdpsp.c | 68 +++++++++++++++++-----
 1 file changed, 53 insertions(+), 15 deletions(-)

Comments

Jan Dąbroś Aug. 9, 2022, 7:41 a.m. UTC | #1
Hi Jarko, Andriy,

Could you please take a look?

Best Regards,
Jan


pon., 25 lip 2022 o 10:03 Jan Dabros <jsd@semihalf.com> napisał(a):
>
> In order to optimize performance, limit amount of back and forth
> transactions between x86 and PSP. This is done by introduction of
> cooldown period - that is window in which x86 isn't releasing the bus
> immediately after each I2C transaction.
>
> In order to protect PSP from being starved while waiting for
> arbitration, after a programmed time bus is automatically released by a
> deferred function.
>
> Signed-off-by: Jan Dabros <jsd@semihalf.com>
> ---
>  drivers/i2c/busses/i2c-designware-amdpsp.c | 68 +++++++++++++++++-----
>  1 file changed, 53 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/i2c/busses/i2c-designware-amdpsp.c b/drivers/i2c/busses/i2c-designware-amdpsp.c
> index b624356c945f..2e1bb5ae72c3 100644
> --- a/drivers/i2c/busses/i2c-designware-amdpsp.c
> +++ b/drivers/i2c/busses/i2c-designware-amdpsp.c
> @@ -6,6 +6,7 @@
>  #include <linux/io-64-nonatomic-lo-hi.h>
>  #include <linux/psp-sev.h>
>  #include <linux/types.h>
> +#include <linux/workqueue.h>
>
>  #include <asm/msr.h>
>
> @@ -15,6 +16,8 @@
>  #define PSP_MBOX_OFFSET                0x10570
>  #define PSP_CMD_TIMEOUT_US     (500 * USEC_PER_MSEC)
>
> +#define PSP_I2C_COOLDOWN_TIME_MS 100
> +
>  #define PSP_I2C_REQ_BUS_CMD            0x64
>  #define PSP_I2C_REQ_RETRY_CNT          400
>  #define PSP_I2C_REQ_RETRY_DELAY_US     (25 * USEC_PER_MSEC)
> @@ -240,6 +243,42 @@ static int psp_send_i2c_req(enum psp_i2c_req_type i2c_req_type)
>         return ret;
>  }
>
> +static void release_bus_now(void)
> +{
> +       int status;
> +
> +       if (!psp_i2c_sem_acquired)
> +               return;
> +
> +       status = psp_send_i2c_req(PSP_I2C_REQ_RELEASE);
> +       if (status)
> +               return;
> +
> +       dev_dbg(psp_i2c_dev, "PSP semaphore held for %ums\n",
> +               jiffies_to_msecs(jiffies - psp_i2c_sem_acquired));
> +
> +       psp_i2c_sem_acquired = 0;
> +}
> +
> +static void psp_release_i2c_bus_deferred(struct work_struct *work)
> +{
> +
> +       mutex_lock(&psp_i2c_access_mutex);
> +
> +       /*
> +        * If there is any pending transaction, cannot release the bus here.
> +        * psp_release_i2c_bus will take care of this later.
> +        */
> +       if (psp_i2c_access_count)
> +               goto cleanup;
> +
> +       release_bus_now();
> +
> +cleanup:
> +       mutex_unlock(&psp_i2c_access_mutex);
> +}
> +static DECLARE_DELAYED_WORK(release_queue, psp_release_i2c_bus_deferred);
> +
>  static int psp_acquire_i2c_bus(void)
>  {
>         int status;
> @@ -250,21 +289,23 @@ static int psp_acquire_i2c_bus(void)
>         if (psp_i2c_mbox_fail)
>                 goto cleanup;
>
> +       psp_i2c_access_count++;
> +
>         /*
> -        * Simply increment usage counter and return if PSP semaphore was
> -        * already taken by kernel.
> +        * No need to request bus arbitration once we are inside cooldown
> +        * period.
>          */
> -       if (psp_i2c_access_count) {
> -               psp_i2c_access_count++;
> +       if (psp_i2c_sem_acquired)
>                 goto cleanup;
> -       }
>
>         status = psp_send_i2c_req(PSP_I2C_REQ_ACQUIRE);
>         if (status)
>                 goto cleanup;
>
>         psp_i2c_sem_acquired = jiffies;
> -       psp_i2c_access_count++;
> +
> +       schedule_delayed_work(&release_queue,
> +                             msecs_to_jiffies(PSP_I2C_COOLDOWN_TIME_MS));
>
>         /*
>          * In case of errors with PSP arbitrator psp_i2c_mbox_fail variable is
> @@ -279,8 +320,6 @@ static int psp_acquire_i2c_bus(void)
>
>  static void psp_release_i2c_bus(void)
>  {
> -       int status;
> -
>         mutex_lock(&psp_i2c_access_mutex);
>
>         /* Return early if mailbox was malfunctional */
> @@ -295,13 +334,12 @@ static void psp_release_i2c_bus(void)
>         if (psp_i2c_access_count)
>                 goto cleanup;
>
> -       /* Send a release command to PSP */
> -       status = psp_send_i2c_req(PSP_I2C_REQ_RELEASE);
> -       if (status)
> -               goto cleanup;
> -
> -       dev_dbg(psp_i2c_dev, "PSP semaphore held for %ums\n",
> -               jiffies_to_msecs(jiffies - psp_i2c_sem_acquired));
> +       /*
> +        * Send a release command to PSP if the cooldown timeout elapsed but x86 still
> +        * owns the ctrlr.
> +        */
> +       if (!delayed_work_pending(&release_queue))
> +               release_bus_now();
>
>  cleanup:
>         mutex_unlock(&psp_i2c_access_mutex);
> --
> 2.37.1.359.gd136c6c3e2-goog
>
Jarkko Nikula Aug. 9, 2022, 12:05 p.m. UTC | #2
Hi

Sorry the delay, this slipped through my eyes during vacation. Couple 
minor comments below.

On 7/25/22 11:02, Jan Dabros wrote:
> In order to optimize performance, limit amount of back and forth
> transactions between x86 and PSP. This is done by introduction of
> cooldown period - that is window in which x86 isn't releasing the bus
> immediately after each I2C transaction.
> 
> In order to protect PSP from being starved while waiting for
> arbitration, after a programmed time bus is automatically released by a
> deferred function.
> 
> Signed-off-by: Jan Dabros <jsd@semihalf.com>
> ---
>   drivers/i2c/busses/i2c-designware-amdpsp.c | 68 +++++++++++++++++-----
>   1 file changed, 53 insertions(+), 15 deletions(-)
> 
> diff --git a/drivers/i2c/busses/i2c-designware-amdpsp.c b/drivers/i2c/busses/i2c-designware-amdpsp.c
> index b624356c945f..2e1bb5ae72c3 100644
> --- a/drivers/i2c/busses/i2c-designware-amdpsp.c
> +++ b/drivers/i2c/busses/i2c-designware-amdpsp.c
> @@ -6,6 +6,7 @@
>   #include <linux/io-64-nonatomic-lo-hi.h>
>   #include <linux/psp-sev.h>
>   #include <linux/types.h>
> +#include <linux/workqueue.h>
>   
>   #include <asm/msr.h>
>   
> @@ -15,6 +16,8 @@
>   #define PSP_MBOX_OFFSET		0x10570
>   #define PSP_CMD_TIMEOUT_US	(500 * USEC_PER_MSEC)
>   
> +#define PSP_I2C_COOLDOWN_TIME_MS 100
> +

"cooldown" distract me thinking thermal management. Would semaphore 
reservation time/timer fit better?

> +static void release_bus_now(void)
> +static void psp_release_i2c_bus_deferred(struct work_struct *work)
> +static DECLARE_DELAYED_WORK(release_queue, psp_release_i2c_bus_deferred);
> +

I'd use the same namespace here. Perhaps _now can be dropped from the 
name since the release_bus and release_bus_deferred are near to each 
other and _deferred variant implies it's called after timeout.

> +	/*
> +	 * Send a release command to PSP if the cooldown timeout elapsed but x86 still
> +	 * owns the ctrlr.
> +	 */

Replace "ctrlr" -> "control" here since then it doesn't lead to think 
is't some technical object like register etc.
Jan Dąbroś Aug. 12, 2022, 6:08 a.m. UTC | #3
wt., 9 sie 2022 o 14:05 Jarkko Nikula <jarkko.nikula@linux.intel.com>
napisał(a):
>
> Hi
>
> Sorry the delay, this slipped through my eyes during vacation. Couple
> minor comments below.
>
> On 7/25/22 11:02, Jan Dabros wrote:
> > In order to optimize performance, limit amount of back and forth
> > transactions between x86 and PSP. This is done by introduction of
> > cooldown period - that is window in which x86 isn't releasing the bus
> > immediately after each I2C transaction.
> >
> > In order to protect PSP from being starved while waiting for
> > arbitration, after a programmed time bus is automatically released by a
> > deferred function.
> >
> > Signed-off-by: Jan Dabros <jsd@semihalf.com>
> > ---
> >   drivers/i2c/busses/i2c-designware-amdpsp.c | 68 +++++++++++++++++-----
> >   1 file changed, 53 insertions(+), 15 deletions(-)
> >
> > diff --git a/drivers/i2c/busses/i2c-designware-amdpsp.c b/drivers/i2c/busses/i2c-designware-amdpsp.c
> > index b624356c945f..2e1bb5ae72c3 100644
> > --- a/drivers/i2c/busses/i2c-designware-amdpsp.c
> > +++ b/drivers/i2c/busses/i2c-designware-amdpsp.c
> > @@ -6,6 +6,7 @@
> >   #include <linux/io-64-nonatomic-lo-hi.h>
> >   #include <linux/psp-sev.h>
> >   #include <linux/types.h>
> > +#include <linux/workqueue.h>
> >
> >   #include <asm/msr.h>
> >
> > @@ -15,6 +16,8 @@
> >   #define PSP_MBOX_OFFSET             0x10570
> >   #define PSP_CMD_TIMEOUT_US  (500 * USEC_PER_MSEC)
> >
> > +#define PSP_I2C_COOLDOWN_TIME_MS 100
> > +
>
> "cooldown" distract me thinking thermal management. Would semaphore
> reservation time/timer fit better?

Yes, it makes sense. I will change this here and in the commit message
to "semaphore reservation timer".

>
> > +static void release_bus_now(void)
> > +static void psp_release_i2c_bus_deferred(struct work_struct *work)
> > +static DECLARE_DELAYED_WORK(release_queue, psp_release_i2c_bus_deferred);
> > +
>
> I'd use the same namespace here. Perhaps _now can be dropped from the
> name since the release_bus and release_bus_deferred are near to each
> other and _deferred variant implies it's called after timeout.

Right, release_bus_now -> release_bus.

>
> > +     /*
> > +      * Send a release command to PSP if the cooldown timeout elapsed but x86 still
> > +      * owns the ctrlr.
> > +      */
>
> Replace "ctrlr" -> "control" here since then it doesn't lead to think
> is't some technical object like register etc.

This is about "controller" not "control", but I think your comment is
still applicable.

Best Regards,
Jan
diff mbox series

Patch

diff --git a/drivers/i2c/busses/i2c-designware-amdpsp.c b/drivers/i2c/busses/i2c-designware-amdpsp.c
index b624356c945f..2e1bb5ae72c3 100644
--- a/drivers/i2c/busses/i2c-designware-amdpsp.c
+++ b/drivers/i2c/busses/i2c-designware-amdpsp.c
@@ -6,6 +6,7 @@ 
 #include <linux/io-64-nonatomic-lo-hi.h>
 #include <linux/psp-sev.h>
 #include <linux/types.h>
+#include <linux/workqueue.h>
 
 #include <asm/msr.h>
 
@@ -15,6 +16,8 @@ 
 #define PSP_MBOX_OFFSET		0x10570
 #define PSP_CMD_TIMEOUT_US	(500 * USEC_PER_MSEC)
 
+#define PSP_I2C_COOLDOWN_TIME_MS 100
+
 #define PSP_I2C_REQ_BUS_CMD		0x64
 #define PSP_I2C_REQ_RETRY_CNT		400
 #define PSP_I2C_REQ_RETRY_DELAY_US	(25 * USEC_PER_MSEC)
@@ -240,6 +243,42 @@  static int psp_send_i2c_req(enum psp_i2c_req_type i2c_req_type)
 	return ret;
 }
 
+static void release_bus_now(void)
+{
+	int status;
+
+	if (!psp_i2c_sem_acquired)
+		return;
+
+	status = psp_send_i2c_req(PSP_I2C_REQ_RELEASE);
+	if (status)
+		return;
+
+	dev_dbg(psp_i2c_dev, "PSP semaphore held for %ums\n",
+		jiffies_to_msecs(jiffies - psp_i2c_sem_acquired));
+
+	psp_i2c_sem_acquired = 0;
+}
+
+static void psp_release_i2c_bus_deferred(struct work_struct *work)
+{
+
+	mutex_lock(&psp_i2c_access_mutex);
+
+	/*
+	 * If there is any pending transaction, cannot release the bus here.
+	 * psp_release_i2c_bus will take care of this later.
+	 */
+	if (psp_i2c_access_count)
+		goto cleanup;
+
+	release_bus_now();
+
+cleanup:
+	mutex_unlock(&psp_i2c_access_mutex);
+}
+static DECLARE_DELAYED_WORK(release_queue, psp_release_i2c_bus_deferred);
+
 static int psp_acquire_i2c_bus(void)
 {
 	int status;
@@ -250,21 +289,23 @@  static int psp_acquire_i2c_bus(void)
 	if (psp_i2c_mbox_fail)
 		goto cleanup;
 
+	psp_i2c_access_count++;
+
 	/*
-	 * Simply increment usage counter and return if PSP semaphore was
-	 * already taken by kernel.
+	 * No need to request bus arbitration once we are inside cooldown
+	 * period.
 	 */
-	if (psp_i2c_access_count) {
-		psp_i2c_access_count++;
+	if (psp_i2c_sem_acquired)
 		goto cleanup;
-	}
 
 	status = psp_send_i2c_req(PSP_I2C_REQ_ACQUIRE);
 	if (status)
 		goto cleanup;
 
 	psp_i2c_sem_acquired = jiffies;
-	psp_i2c_access_count++;
+
+	schedule_delayed_work(&release_queue,
+			      msecs_to_jiffies(PSP_I2C_COOLDOWN_TIME_MS));
 
 	/*
 	 * In case of errors with PSP arbitrator psp_i2c_mbox_fail variable is
@@ -279,8 +320,6 @@  static int psp_acquire_i2c_bus(void)
 
 static void psp_release_i2c_bus(void)
 {
-	int status;
-
 	mutex_lock(&psp_i2c_access_mutex);
 
 	/* Return early if mailbox was malfunctional */
@@ -295,13 +334,12 @@  static void psp_release_i2c_bus(void)
 	if (psp_i2c_access_count)
 		goto cleanup;
 
-	/* Send a release command to PSP */
-	status = psp_send_i2c_req(PSP_I2C_REQ_RELEASE);
-	if (status)
-		goto cleanup;
-
-	dev_dbg(psp_i2c_dev, "PSP semaphore held for %ums\n",
-		jiffies_to_msecs(jiffies - psp_i2c_sem_acquired));
+	/*
+	 * Send a release command to PSP if the cooldown timeout elapsed but x86 still
+	 * owns the ctrlr.
+	 */
+	if (!delayed_work_pending(&release_queue))
+		release_bus_now();
 
 cleanup:
 	mutex_unlock(&psp_i2c_access_mutex);