* Re: [PATCH] thermal/drivers/rcar: add error checking in probe()
2026-06-23 7:49 [PATCH] thermal/drivers/rcar: add error checking in probe() Dan Carpenter
@ 2026-06-23 8:03 ` Geert Uytterhoeven
2026-06-23 8:13 ` Dan Carpenter
2026-06-23 8:18 ` Niklas Söderlund
1 sibling, 1 reply; 5+ messages in thread
From: Geert Uytterhoeven @ 2026-06-23 8:03 UTC (permalink / raw)
To: Dan Carpenter
Cc: Niklas Söderlund, Niklas Söderlund, Rafael J. Wysocki,
Daniel Lezcano, Zhang Rui, Lukasz Luba, Magnus Damm,
linux-renesas-soc, linux-pm, linux-kernel, kernel-janitors,
Andrzej Pietrasiewicz
Hi Dan,
Thanks for your patch!
On Tue, 23 Jun 2026 at 09:49, Dan Carpenter <error27@gmail.com> wrote:
> The thermal_zone_device_register_with_trips() can fail for a number of
> reasons, including allocation failures. Check for error pointers to
> avoid an error pointer dereference.
>
> Fixes: 9d617949d490 ("thermal/drivers/renesas: Group all renesas thermal drivers together")
This is not the commit you are looking for...
Fixes: bbcf90c0646ac797 ("thermal: Explicitly enable non-changing
thermal zone devices")
> Signed-off-by: Dan Carpenter <error27@gmail.com>
> --- a/drivers/thermal/renesas/rcar_thermal.c
> +++ b/drivers/thermal/renesas/rcar_thermal.c
> @@ -492,6 +492,11 @@ static int rcar_thermal_probe(struct platform_device *pdev)
> "rcar_thermal", trips, ARRAY_SIZE(trips), priv,
> &rcar_thermal_zone_ops, NULL, 0,
> idle);
> + if (IS_ERR(priv->zone)) {
> + ret = PTR_ERR(priv->zone);
> + priv->zone = NULL;
> + goto error_unregister;
> + }
This check is already present below (out of context), so it would be good
to avoid duplicating it.
>
> ret = thermal_zone_device_enable(priv->zone);
> if (ret) {
The issue is that this call was added before the error checking is done.
So this should be moved below instead (to the else branch of the next
chip->use_of_thermal test?).
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [PATCH] thermal/drivers/rcar: add error checking in probe()
2026-06-23 7:49 [PATCH] thermal/drivers/rcar: add error checking in probe() Dan Carpenter
2026-06-23 8:03 ` Geert Uytterhoeven
@ 2026-06-23 8:18 ` Niklas Söderlund
2026-06-23 8:25 ` Dan Carpenter
1 sibling, 1 reply; 5+ messages in thread
From: Niklas Söderlund @ 2026-06-23 8:18 UTC (permalink / raw)
To: Dan Carpenter
Cc: Rafael J. Wysocki, Daniel Lezcano, Zhang Rui, Lukasz Luba,
Geert Uytterhoeven, Magnus Damm, linux-renesas-soc, linux-pm,
linux-kernel, kernel-janitors
Hi Dan,
Thanks for your work.
On 2026-06-23 10:49:29 +0300, Dan Carpenter wrote:
> The thermal_zone_device_register_with_trips() can fail for a number of
> reasons, including allocation failures. Check for error pointers to
> avoid an error pointer dereference.
>
> Fixes: 9d617949d490 ("thermal/drivers/renesas: Group all renesas thermal drivers together")
I don't think this is correct as this commits just moves the file.
> Signed-off-by: Dan Carpenter <error27@gmail.com>
> ---
> drivers/thermal/renesas/rcar_thermal.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/drivers/thermal/renesas/rcar_thermal.c b/drivers/thermal/renesas/rcar_thermal.c
> index 6e5dcac5d47a..71f836fbc698 100644
> --- a/drivers/thermal/renesas/rcar_thermal.c
> +++ b/drivers/thermal/renesas/rcar_thermal.c
> @@ -492,6 +492,11 @@ static int rcar_thermal_probe(struct platform_device *pdev)
> "rcar_thermal", trips, ARRAY_SIZE(trips), priv,
> &rcar_thermal_zone_ops, NULL, 0,
> idle);
> + if (IS_ERR(priv->zone)) {
> + ret = PTR_ERR(priv->zone);
> + priv->zone = NULL;
> + goto error_unregister;
> + }
While this indeed is an issue that should be fixed I don't think this is
the correct fix. Below the if .. else .. block where this is added there
already is a check for IS_ERR(priv->zone). That however does not guard
against the usage of priv->zone for thermal_zone_device_enable().
We should only call thermal_zone_device_enable() if we are on a system
that uses OF (gated by chip->use_of_thermal) which is the reason for the
if .. else .. block in the first place. As chance have it we also have a
check on chip->use_of_thermal directly after the existing
IS_ERR(priv->zone) check. I think it would be better to move the call to
thermal_zone_device_enable() there and avoid having two checks for
IS_ERR(priv->zone)? Something like this,
diff --git a/drivers/thermal/renesas/rcar_thermal.c b/drivers/thermal/renesas/rcar_thermal.c
index 6e5dcac5d47a..dd13cf971ddb 100644
--- a/drivers/thermal/renesas/rcar_thermal.c
+++ b/drivers/thermal/renesas/rcar_thermal.c
@@ -492,12 +492,6 @@ static int rcar_thermal_probe(struct platform_device *pdev)
"rcar_thermal", trips, ARRAY_SIZE(trips), priv,
&rcar_thermal_zone_ops, NULL, 0,
idle);
-
- ret = thermal_zone_device_enable(priv->zone);
- if (ret) {
- thermal_zone_device_unregister(priv->zone);
- priv->zone = ERR_PTR(ret);
- }
}
if (IS_ERR(priv->zone)) {
dev_err(dev, "can't register thermal zone\n");
@@ -507,13 +501,16 @@ static int rcar_thermal_probe(struct platform_device *pdev)
}
if (chip->use_of_thermal) {
+ ret = thermal_zone_device_enable(priv->zone);
+ if (ret)
+ goto error_of_thermal;
+
ret = thermal_add_hwmon_sysfs(priv->zone);
if (ret)
- goto error_unregister;
+ goto error_of_thermal;
}
rcar_thermal_irq_enable(priv);
-
list_move_tail(&priv->list, &common->head);
/* update ENR bits */
@@ -528,6 +525,8 @@ static int rcar_thermal_probe(struct platform_device *pdev)
return 0;
+error_of_thermal:
+ thermal_zone_device_unregister(priv->zone);
error_unregister:
rcar_thermal_remove(pdev);
>
> ret = thermal_zone_device_enable(priv->zone);
> if (ret) {
> --
> 2.53.0
>
--
Kind Regards,
Niklas Söderlund
^ permalink raw reply related [flat|nested] 5+ messages in thread