2021-08-09 06:40:57

by Clark Wang

[permalink] [raw]
Subject: [PATCH V4 8/8] i3c: master: svc: enable the interrupt in the enable ibi function

If enable interrupt in the svc_i3c_master_bus_init() but do not call
enable ibi in the device driver, it will cause a kernel dump in the
svc_i3c_master_handle_ibi() when a slave start occurs on the i3c bus,
because the data->ibi_pool is not initialized.
So only enable the interrupt in svc_i3c_master_enable_ibi() function.

Signed-off-by: Clark Wang <[email protected]>
---
V4: New patch in this patchset
---
drivers/i3c/master/svc-i3c-master.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/i3c/master/svc-i3c-master.c b/drivers/i3c/master/svc-i3c-master.c
index 77f67d407acd..b802afd8eb7d 100644
--- a/drivers/i3c/master/svc-i3c-master.c
+++ b/drivers/i3c/master/svc-i3c-master.c
@@ -569,8 +569,6 @@ static int svc_i3c_master_bus_init(struct i3c_master_controller *m)
if (ret)
goto rpm_out;

- svc_i3c_master_enable_interrupts(master, SVC_I3C_MINT_SLVSTART);
-
rpm_out:
pm_runtime_mark_last_busy(master->dev);
pm_runtime_put_autosuspend(master->dev);
@@ -1398,6 +1396,8 @@ static int svc_i3c_master_enable_ibi(struct i3c_dev_desc *dev)
return ret;
}

+ svc_i3c_master_enable_interrupts(master, SVC_I3C_MINT_SLVSTART);
+
return i3c_master_enec_locked(m, dev->info.dyn_addr, I3C_CCC_EVENT_SIR);
}

@@ -1407,6 +1407,8 @@ static int svc_i3c_master_disable_ibi(struct i3c_dev_desc *dev)
struct svc_i3c_master *master = to_svc_i3c_master(m);
int ret;

+ svc_i3c_master_disable_interrupts(master);
+
ret = i3c_master_disec_locked(m, dev->info.dyn_addr, I3C_CCC_EVENT_SIR);

pm_runtime_mark_last_busy(master->dev);
--
2.25.1


2021-08-16 07:04:19

by Miquel Raynal

[permalink] [raw]
Subject: Re: [PATCH V4 8/8] i3c: master: svc: enable the interrupt in the enable ibi function

Hi Clark,

Clark Wang <[email protected]> wrote on Mon, 9 Aug 2021 14:36:45
+0800:

> If enable interrupt in the svc_i3c_master_bus_init() but do not call
> enable ibi in the device driver, it will cause a kernel dump in the
> svc_i3c_master_handle_ibi() when a slave start occurs on the i3c bus,
> because the data->ibi_pool is not initialized.
> So only enable the interrupt in svc_i3c_master_enable_ibi() function.

Is this situation actually happening or is this purely theoretical?

Anyway this doesn't hurt so:

Reviewed-by: Miquel Raynal <[email protected]>

>
> Signed-off-by: Clark Wang <[email protected]>
> ---
> V4: New patch in this patchset
> ---
> drivers/i3c/master/svc-i3c-master.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/i3c/master/svc-i3c-master.c b/drivers/i3c/master/svc-i3c-master.c
> index 77f67d407acd..b802afd8eb7d 100644
> --- a/drivers/i3c/master/svc-i3c-master.c
> +++ b/drivers/i3c/master/svc-i3c-master.c
> @@ -569,8 +569,6 @@ static int svc_i3c_master_bus_init(struct i3c_master_controller *m)
> if (ret)
> goto rpm_out;
>
> - svc_i3c_master_enable_interrupts(master, SVC_I3C_MINT_SLVSTART);
> -
> rpm_out:
> pm_runtime_mark_last_busy(master->dev);
> pm_runtime_put_autosuspend(master->dev);
> @@ -1398,6 +1396,8 @@ static int svc_i3c_master_enable_ibi(struct i3c_dev_desc *dev)
> return ret;
> }
>
> + svc_i3c_master_enable_interrupts(master, SVC_I3C_MINT_SLVSTART);
> +
> return i3c_master_enec_locked(m, dev->info.dyn_addr, I3C_CCC_EVENT_SIR);
> }
>
> @@ -1407,6 +1407,8 @@ static int svc_i3c_master_disable_ibi(struct i3c_dev_desc *dev)
> struct svc_i3c_master *master = to_svc_i3c_master(m);
> int ret;
>
> + svc_i3c_master_disable_interrupts(master);
> +
> ret = i3c_master_disec_locked(m, dev->info.dyn_addr, I3C_CCC_EVENT_SIR);
>
> pm_runtime_mark_last_busy(master->dev);

Thanks,
Miquèl