PSC controller has a limitation that it can only power-up the second core
when the first core is in ON state. Power-state for core0 should be equal
to or higher than core1, else the kernel is seen hanging during rproc
loading.
Make the powering up of cores sequential, by waiting for the current core
to power-up before proceeding to the next core, with a timeout of 2sec.
Add a wait queue event in k3_r5_cluster_rproc_init call, that will wait
for the current core to be released from reset before proceeding with the
next core.
Also, ensure that core1 can not be powered on before core0 when starting
cores from sysfs. Similarly, ensure that core0 can not be shutdown
before core1 from sysfs.
v2: Changelog:
1) Fixed multi-line comment format
2) Included root cause of bug in comments
3) Added a patch to ensure power-up/shutdown is sequential via sysfs
Link to v1:
https://lore.kernel.org/all/[email protected]/
Apurva Nandan (1):
remoteproc: k3-r5: Wait for core0 power-up before powering up core1
Beleswar Padhi (1):
remoteproc: k3-r5: Do not allow core1 to power up before core0 via
sysfs
drivers/remoteproc/ti_k3_r5_remoteproc.c | 51 +++++++++++++++++++++++-
1 file changed, 49 insertions(+), 2 deletions(-)
--
2.34.1
From: Apurva Nandan <[email protected]>
PSC controller has a limitation that it can only power-up the second core
when the first core is in ON state. Power-state for core0 should be equal
to or higher than core1, else the kernel is seen hanging during rproc
loading.
Make the powering up of cores sequential, by waiting for the current core
to power-up before proceeding to the next core, with a timeout of 2sec.
Add a wait queue event in k3_r5_cluster_rproc_init call, that will wait
for the current core to be released from reset before proceeding with the
next core.
Fixes: 6dedbd1d5443 ("remoteproc: k3-r5: Add a remoteproc driver for R5F subsystem")
Signed-off-by: Apurva Nandan <[email protected]>
---
drivers/remoteproc/ti_k3_r5_remoteproc.c | 28 ++++++++++++++++++++++++
1 file changed, 28 insertions(+)
diff --git a/drivers/remoteproc/ti_k3_r5_remoteproc.c b/drivers/remoteproc/ti_k3_r5_remoteproc.c
index ad3415a3851b..5a9bd5d4a2ea 100644
--- a/drivers/remoteproc/ti_k3_r5_remoteproc.c
+++ b/drivers/remoteproc/ti_k3_r5_remoteproc.c
@@ -103,12 +103,14 @@ struct k3_r5_soc_data {
* @dev: cached device pointer
* @mode: Mode to configure the Cluster - Split or LockStep
* @cores: list of R5 cores within the cluster
+ * @core_transition: wait queue to sync core state changes
* @soc_data: SoC-specific feature data for a R5FSS
*/
struct k3_r5_cluster {
struct device *dev;
enum cluster_mode mode;
struct list_head cores;
+ wait_queue_head_t core_transition;
const struct k3_r5_soc_data *soc_data;
};
@@ -128,6 +130,7 @@ struct k3_r5_cluster {
* @atcm_enable: flag to control ATCM enablement
* @btcm_enable: flag to control BTCM enablement
* @loczrama: flag to dictate which TCM is at device address 0x0
+ * @released_from_reset: flag to signal when core is out of reset
*/
struct k3_r5_core {
struct list_head elem;
@@ -144,6 +147,7 @@ struct k3_r5_core {
u32 atcm_enable;
u32 btcm_enable;
u32 loczrama;
+ bool released_from_reset;
};
/**
@@ -460,6 +464,8 @@ static int k3_r5_rproc_prepare(struct rproc *rproc)
ret);
return ret;
}
+ core->released_from_reset = true;
+ wake_up_interruptible(&cluster->core_transition);
/*
* Newer IP revisions like on J7200 SoCs support h/w auto-initialization
@@ -1140,6 +1146,7 @@ static int k3_r5_rproc_configure_mode(struct k3_r5_rproc *kproc)
return ret;
}
+ core->released_from_reset = c_state;
ret = ti_sci_proc_get_status(core->tsp, &boot_vec, &cfg, &ctrl,
&stat);
if (ret < 0) {
@@ -1280,6 +1287,26 @@ static int k3_r5_cluster_rproc_init(struct platform_device *pdev)
cluster->mode == CLUSTER_MODE_SINGLECPU ||
cluster->mode == CLUSTER_MODE_SINGLECORE)
break;
+
+ /*
+ * R5 cores require to be powered on sequentially, core0
+ * should be in higher power state than core1 in a cluster
+ * So, wait for current core to power up before proceeding
+ * to next core and put timeout of 2sec for each core.
+ *
+ * This waiting mechanism is necessary because
+ * rproc_auto_boot_callback() for core1 can be called before
+ * core0 due to thread execution order.
+ */
+ ret = wait_event_interruptible_timeout(cluster->core_transition,
+ core->released_from_reset,
+ msecs_to_jiffies(2000));
+ if (ret <= 0) {
+ dev_err(dev,
+ "Timed out waiting for %s core to power up!\n",
+ rproc->name);
+ return ret;
+ }
}
return 0;
@@ -1709,6 +1736,7 @@ static int k3_r5_probe(struct platform_device *pdev)
cluster->dev = dev;
cluster->soc_data = data;
INIT_LIST_HEAD(&cluster->cores);
+ init_waitqueue_head(&cluster->core_transition);
ret = of_property_read_u32(np, "ti,cluster-mode", &cluster->mode);
if (ret < 0 && ret != -EINVAL) {
--
2.34.1
PSC controller has a limitation that it can only power-up the second
core when the first core is in ON state. Power-state for core0 should be
equal to or higher than core1.
Therefore, prevent core1 from powering up before core0 during the start
process from sysfs. Similarly, prevent core0 from shutting down before
core1 has been shut down from sysfs.
Fixes: 6dedbd1d5443 ("remoteproc: k3-r5: Add a remoteproc driver for R5F subsystem")
Signed-off-by: Beleswar Padhi <[email protected]>
---
drivers/remoteproc/ti_k3_r5_remoteproc.c | 23 +++++++++++++++++++++--
1 file changed, 21 insertions(+), 2 deletions(-)
diff --git a/drivers/remoteproc/ti_k3_r5_remoteproc.c b/drivers/remoteproc/ti_k3_r5_remoteproc.c
index 5a9bd5d4a2ea..b9d2a16af563 100644
--- a/drivers/remoteproc/ti_k3_r5_remoteproc.c
+++ b/drivers/remoteproc/ti_k3_r5_remoteproc.c
@@ -548,7 +548,7 @@ static int k3_r5_rproc_start(struct rproc *rproc)
struct k3_r5_rproc *kproc = rproc->priv;
struct k3_r5_cluster *cluster = kproc->cluster;
struct device *dev = kproc->dev;
- struct k3_r5_core *core;
+ struct k3_r5_core *core0, *core;
u32 boot_addr;
int ret;
@@ -574,6 +574,15 @@ static int k3_r5_rproc_start(struct rproc *rproc)
goto unroll_core_run;
}
} else {
+ /* do not allow core 1 to start before core 0 */
+ core0 = list_first_entry(&cluster->cores, struct k3_r5_core,
+ elem);
+ if (core != core0 && core0->rproc->state == RPROC_OFFLINE) {
+ dev_err(dev, "%s: can not start core 1 before core 0\n",
+ __func__);
+ return -EPERM;
+ }
+
ret = k3_r5_core_run(core);
if (ret)
goto put_mbox;
@@ -619,7 +628,8 @@ static int k3_r5_rproc_stop(struct rproc *rproc)
{
struct k3_r5_rproc *kproc = rproc->priv;
struct k3_r5_cluster *cluster = kproc->cluster;
- struct k3_r5_core *core = kproc->core;
+ struct device *dev = kproc->dev;
+ struct k3_r5_core *core1, *core = kproc->core;
int ret;
/* halt all applicable cores */
@@ -632,6 +642,15 @@ static int k3_r5_rproc_stop(struct rproc *rproc)
}
}
} else {
+ /* do not allow core 0 to stop before core 1 */
+ core1 = list_last_entry(&cluster->cores, struct k3_r5_core,
+ elem);
+ if (core != core1 && core1->rproc->state != RPROC_OFFLINE) {
+ dev_err(dev, "%s: can not stop core 0 before core 1\n",
+ __func__);
+ return -EPERM;
+ }
+
ret = k3_r5_core_halt(core);
if (ret)
goto out;
--
2.34.1
Good day,
On Wed, Apr 24, 2024 at 06:35:03PM +0530, Beleswar Padhi wrote:
> From: Apurva Nandan <[email protected]>
>
> PSC controller has a limitation that it can only power-up the second core
> when the first core is in ON state. Power-state for core0 should be equal
> to or higher than core1, else the kernel is seen hanging during rproc
> loading.
>
> Make the powering up of cores sequential, by waiting for the current core
> to power-up before proceeding to the next core, with a timeout of 2sec.
> Add a wait queue event in k3_r5_cluster_rproc_init call, that will wait
> for the current core to be released from reset before proceeding with the
> next core.
>
> Fixes: 6dedbd1d5443 ("remoteproc: k3-r5: Add a remoteproc driver for R5F subsystem")
>
> Signed-off-by: Apurva Nandan <[email protected]>
You need to add your own SoB as well.
> ---
> drivers/remoteproc/ti_k3_r5_remoteproc.c | 28 ++++++++++++++++++++++++
> 1 file changed, 28 insertions(+)
>
> diff --git a/drivers/remoteproc/ti_k3_r5_remoteproc.c b/drivers/remoteproc/ti_k3_r5_remoteproc.c
> index ad3415a3851b..5a9bd5d4a2ea 100644
> --- a/drivers/remoteproc/ti_k3_r5_remoteproc.c
> +++ b/drivers/remoteproc/ti_k3_r5_remoteproc.c
> @@ -103,12 +103,14 @@ struct k3_r5_soc_data {
> * @dev: cached device pointer
> * @mode: Mode to configure the Cluster - Split or LockStep
> * @cores: list of R5 cores within the cluster
> + * @core_transition: wait queue to sync core state changes
> * @soc_data: SoC-specific feature data for a R5FSS
> */
> struct k3_r5_cluster {
> struct device *dev;
> enum cluster_mode mode;
> struct list_head cores;
> + wait_queue_head_t core_transition;
> const struct k3_r5_soc_data *soc_data;
> };
>
> @@ -128,6 +130,7 @@ struct k3_r5_cluster {
> * @atcm_enable: flag to control ATCM enablement
> * @btcm_enable: flag to control BTCM enablement
> * @loczrama: flag to dictate which TCM is at device address 0x0
> + * @released_from_reset: flag to signal when core is out of reset
> */
> struct k3_r5_core {
> struct list_head elem;
> @@ -144,6 +147,7 @@ struct k3_r5_core {
> u32 atcm_enable;
> u32 btcm_enable;
> u32 loczrama;
> + bool released_from_reset;
> };
>
> /**
> @@ -460,6 +464,8 @@ static int k3_r5_rproc_prepare(struct rproc *rproc)
> ret);
> return ret;
> }
> + core->released_from_reset = true;
> + wake_up_interruptible(&cluster->core_transition);
>
> /*
> * Newer IP revisions like on J7200 SoCs support h/w auto-initialization
> @@ -1140,6 +1146,7 @@ static int k3_r5_rproc_configure_mode(struct k3_r5_rproc *kproc)
> return ret;
> }
>
> + core->released_from_reset = c_state;
I understand why this is needed but it line could be very cryptic for people
trying to understand this driver. Please add a comment to describe what is
happening here.
> ret = ti_sci_proc_get_status(core->tsp, &boot_vec, &cfg, &ctrl,
> &stat);
> if (ret < 0) {
> @@ -1280,6 +1287,26 @@ static int k3_r5_cluster_rproc_init(struct platform_device *pdev)
> cluster->mode == CLUSTER_MODE_SINGLECPU ||
> cluster->mode == CLUSTER_MODE_SINGLECORE)
> break;
> +
> + /*
> + * R5 cores require to be powered on sequentially, core0
> + * should be in higher power state than core1 in a cluster
> + * So, wait for current core to power up before proceeding
> + * to next core and put timeout of 2sec for each core.
> + *
> + * This waiting mechanism is necessary because
> + * rproc_auto_boot_callback() for core1 can be called before
> + * core0 due to thread execution order.
> + */
> + ret = wait_event_interruptible_timeout(cluster->core_transition,
> + core->released_from_reset,
> + msecs_to_jiffies(2000));
> + if (ret <= 0) {
> + dev_err(dev,
> + "Timed out waiting for %s core to power up!\n",
> + rproc->name);
> + return ret;
> + }
> }
>
> return 0;
> @@ -1709,6 +1736,7 @@ static int k3_r5_probe(struct platform_device *pdev)
> cluster->dev = dev;
> cluster->soc_data = data;
> INIT_LIST_HEAD(&cluster->cores);
> + init_waitqueue_head(&cluster->core_transition);
>
> ret = of_property_read_u32(np, "ti,cluster-mode", &cluster->mode);
> if (ret < 0 && ret != -EINVAL) {
> --
> 2.34.1
>
Hello,
On 26/04/24 22:39, Mathieu Poirier wrote:
> Good day, On Wed, Apr 24, 2024 at 06: 35: 03PM +0530, Beleswar Padhi wrote: >
> From: Apurva Nandan <a-nandan@ ti. com> > > PSC controller has a limitation that
> it can only power-up the second core > when the first core is in ON
> ZjQcmQRYFpfptBannerStart
> This message was sent from outside of Texas Instruments.
> Do not click links or open attachments unless you recognize the source of this
> email and know the content is safe. If you wish to report this message to IT
> Security, please forward the message as an attachment to [email protected]
> ZjQcmQRYFpfptBannerEnd
>
> Good day,
>
> On Wed, Apr 24, 2024 at 06:35:03PM +0530, Beleswar Padhi wrote:
> > From: Apurva Nandan <[email protected]>
> >
> > PSC controller has a limitation that it can only power-up the second core
> > when the first core is in ON state. Power-state for core0 should be equal
> > to or higher than core1, else the kernel is seen hanging during rproc
> > loading.
> >
> > Make the powering up of cores sequential, by waiting for the current core
> > to power-up before proceeding to the next core, with a timeout of 2sec.
> > Add a wait queue event in k3_r5_cluster_rproc_init call, that will wait
> > for the current core to be released from reset before proceeding with the
> > next core.
> >
> > Fixes: 6dedbd1d5443 ("remoteproc: k3-r5: Add a remoteproc driver for R5F subsystem")
> >
> > Signed-off-by: Apurva Nandan <[email protected]>
>
> You need to add your own SoB as well.
>
> > ---
> > drivers/remoteproc/ti_k3_r5_remoteproc.c | 28 ++++++++++++++++++++++++
> > 1 file changed, 28 insertions(+)
> >
> > diff --git a/drivers/remoteproc/ti_k3_r5_remoteproc.c b/drivers/remoteproc/ti_k3_r5_remoteproc.c
> > index ad3415a3851b..5a9bd5d4a2ea 100644
> > --- a/drivers/remoteproc/ti_k3_r5_remoteproc.c
> > +++ b/drivers/remoteproc/ti_k3_r5_remoteproc.c
> > @@ -103,12 +103,14 @@ struct k3_r5_soc_data {
> > * @dev: cached device pointer
> > * @mode: Mode to configure the Cluster - Split or LockStep
> > * @cores: list of R5 cores within the cluster
> > + * @core_transition: wait queue to sync core state changes
> > * @soc_data: SoC-specific feature data for a R5FSS
> > */
> > struct k3_r5_cluster {
> > struct device *dev;
> > enum cluster_mode mode;
> > struct list_head cores;
> > + wait_queue_head_t core_transition;
> > const struct k3_r5_soc_data *soc_data;
> > };
> >
> > @@ -128,6 +130,7 @@ struct k3_r5_cluster {
> > * @atcm_enable: flag to control ATCM enablement
> > * @btcm_enable: flag to control BTCM enablement
> > * @loczrama: flag to dictate which TCM is at device address 0x0
> > + * @released_from_reset: flag to signal when core is out of reset
> > */
> > struct k3_r5_core {
> > struct list_head elem;
> > @@ -144,6 +147,7 @@ struct k3_r5_core {
> > u32 atcm_enable;
> > u32 btcm_enable;
> > u32 loczrama;
> > + bool released_from_reset;
> > };
> >
> > /**
> > @@ -460,6 +464,8 @@ static int k3_r5_rproc_prepare(struct rproc *rproc)
> > ret);
> > return ret;
> > }
> > + core->released_from_reset = true;
> > + wake_up_interruptible(&cluster->core_transition);
> >
> > /*
> > * Newer IP revisions like on J7200 SoCs support h/w auto-initialization
> > @@ -1140,6 +1146,7 @@ static int k3_r5_rproc_configure_mode(struct k3_r5_rproc *kproc)
> > return ret;
> > }
> >
> > + core->released_from_reset = c_state;
>
> I understand why this is needed but it line could be very cryptic for people
> trying to understand this driver. Please add a comment to describe what is
> happening here.
Thanks for the review. I will send v3 addressing these comments shortly!
>
> > ret = ti_sci_proc_get_status(core->tsp, &boot_vec, &cfg, &ctrl,
> > &stat);
> > if (ret < 0) {
> > @@ -1280,6 +1287,26 @@ static int k3_r5_cluster_rproc_init(struct platform_device *pdev)
> > cluster->mode == CLUSTER_MODE_SINGLECPU ||
> > cluster->mode == CLUSTER_MODE_SINGLECORE)
> > break;
> > +
> > + /*
> > + * R5 cores require to be powered on sequentially, core0
> > + * should be in higher power state than core1 in a cluster
> > + * So, wait for current core to power up before proceeding
> > + * to next core and put timeout of 2sec for each core.
> > + *
> > + * This waiting mechanism is necessary because
> > + * rproc_auto_boot_callback() for core1 can be called before
> > + * core0 due to thread execution order.
> > + */
> > + ret = wait_event_interruptible_timeout(cluster->core_transition,
> > + core->released_from_reset,
> > + msecs_to_jiffies(2000));
> > + if (ret <= 0) {
> > + dev_err(dev,
> > + "Timed out waiting for %s core to power up!\n",
> > + rproc->name);
> > + return ret;
> > + }
> > }
> >
> > return 0;
> > @@ -1709,6 +1736,7 @@ static int k3_r5_probe(struct platform_device *pdev)
> > cluster->dev = dev;
> > cluster->soc_data = data;
> > INIT_LIST_HEAD(&cluster->cores);
> > + init_waitqueue_head(&cluster->core_transition);
> >
> > ret = of_property_read_u32(np, "ti,cluster-mode", &cluster->mode);
> > if (ret < 0 && ret != -EINVAL) {
> > --
> > 2.34.1
> >
>