2023-07-07 05:35:15

by Tinghan Shen

Subject: [PATCH v14 10/12] remoteproc: mediatek: Report watchdog crash to all cores

Once the SCP watchdog timeout happened, trigger the rproc recovery flow on
all cores.

Signed-off-by: Tinghan Shen <[email protected]>
---
drivers/remoteproc/mtk_scp.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/remoteproc/mtk_scp.c b/drivers/remoteproc/mtk_scp.c
index 9fed5161f094..7e786aaf04e1 100644
--- a/drivers/remoteproc/mtk_scp.c
+++ b/drivers/remoteproc/mtk_scp.c
@@ -80,8 +80,14 @@ EXPORT_SYMBOL_GPL(scp_put);

static void scp_wdt_handler(struct mtk_scp *scp, u32 scp_to_host)
{
+ struct mtk_scp_of_cluster *scp_cluster = platform_get_drvdata(scp->cluster_pdev);
+ struct mtk_scp *scp_node;
+
dev_err(scp->dev, "SCP watchdog timeout! 0x%x", scp_to_host);
- rproc_report_crash(scp->rproc, RPROC_WATCHDOG);
+
+ /* report watchdog timeout to all cores */
+ list_for_each_entry(scp_node, &scp_cluster->mtk_scp_list, elem)
+ rproc_report_crash(scp_node->rproc, RPROC_WATCHDOG);
}

static void scp_init_ipi_handler(void *data, unsigned int len, void *priv)
--
2.18.0

2023-07-12 20:01:35

by Mathieu Poirier

Subject: Re: [PATCH v14 10/12] remoteproc: mediatek: Report watchdog crash to all cores

On Fri, Jul 07, 2023 at 01:26:10PM +0800, Tinghan Shen wrote:
> Once the SCP watchdog timeout happened, trigger the rproc recovery flow on
> all cores.
>

You are missing the most important part... "Why" is this needed? Without it I
don't know what is going on and can't review your work.

> Signed-off-by: Tinghan Shen <[email protected]>
> ---
> drivers/remoteproc/mtk_scp.c | 8 +++++++-
> 1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/remoteproc/mtk_scp.c b/drivers/remoteproc/mtk_scp.c
> index 9fed5161f094..7e786aaf04e1 100644
> --- a/drivers/remoteproc/mtk_scp.c
> +++ b/drivers/remoteproc/mtk_scp.c
> @@ -80,8 +80,14 @@ EXPORT_SYMBOL_GPL(scp_put);
>
> static void scp_wdt_handler(struct mtk_scp *scp, u32 scp_to_host)
> {
> + struct mtk_scp_of_cluster *scp_cluster = platform_get_drvdata(scp->cluster_pdev);
> + struct mtk_scp *scp_node;
> +
> dev_err(scp->dev, "SCP watchdog timeout! 0x%x", scp_to_host);
> - rproc_report_crash(scp->rproc, RPROC_WATCHDOG);
> +
> + /* report watchdog timeout to all cores */
> + list_for_each_entry(scp_node, &scp_cluster->mtk_scp_list, elem)
> + rproc_report_crash(scp_node->rproc, RPROC_WATCHDOG);
> }
>
> static void scp_init_ipi_handler(void *data, unsigned int len, void *priv)
> --
> 2.18.0
>

2023-07-13 08:37:38

by Tinghan Shen

Subject: Re: [PATCH v14 10/12] remoteproc: mediatek: Report watchdog crash to all cores

Hi Mathieu,

On Wed, 2023-07-12 at 13:38 -0600, Mathieu Poirier wrote:
> On Fri, Jul 07, 2023 at 01:26:10PM +0800, Tinghan Shen wrote:
> > Once the SCP watchdog timeout happened, trigger the rproc recovery flow on
> > all cores.
> >
>
> You are missing the most important part... "Why" is this needed? Without it I
> don't know what is going on and can't review your work.

Sorry, I'll improve the commit message.

I think that the SCP should have the same behavior as an SMP system. In the event
of a watchdog timeout occuring on any of the cores, the entire SCP should be recovered.


>
> > Signed-off-by: Tinghan Shen <[email protected]>
> > ---
> > drivers/remoteproc/mtk_scp.c | 8 +++++++-
> > 1 file changed, 7 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/remoteproc/mtk_scp.c b/drivers/remoteproc/mtk_scp.c
> > index 9fed5161f094..7e786aaf04e1 100644
> > --- a/drivers/remoteproc/mtk_scp.c
> > +++ b/drivers/remoteproc/mtk_scp.c
> > @@ -80,8 +80,14 @@ EXPORT_SYMBOL_GPL(scp_put);
> >
> > static void scp_wdt_handler(struct mtk_scp *scp, u32 scp_to_host)
> > {
> > + struct mtk_scp_of_cluster *scp_cluster = platform_get_drvdata(scp->cluster_pdev);
> > + struct mtk_scp *scp_node;
> > +
> > dev_err(scp->dev, "SCP watchdog timeout! 0x%x", scp_to_host);
> > - rproc_report_crash(scp->rproc, RPROC_WATCHDOG);
> > +
> > + /* report watchdog timeout to all cores */
> > + list_for_each_entry(scp_node, &scp_cluster->mtk_scp_list, elem)
> > + rproc_report_crash(scp_node->rproc, RPROC_WATCHDOG);
> > }
> >
> > static void scp_init_ipi_handler(void *data, unsigned int len, void *priv)
> > --
> > 2.18.0

--
Best regards,
TingHan