reset: don't overwrite fwnode_reset_n_cells

[PATCH] reset: don't overwrite fwnode_reset_n_cells

Posted by Bartosz Golaszewski 4 weeks ago

Fix a logic bug in reser_register() where we set fwnode_reset_n_cells to
1 if fwnode is set and fwnode_xlate is not but we do it after assigning
of_fwnode_handle(of_node) to fwnode.

Modify the logic to: assign fwnode from of_node if applicable, if fwnode
is still not set, try to get it from the device and only then check
of_xlate and fwnode_xlate and either assign fwnode_reset_n_cells from OF
or default to fwnode_reset_simple_xlate and fwnode_reset_n_cells = 1.

Fixes: ba8dbbb14b7e ("reset: convert the core API to using firmware nodes")
Reported-by: Mark Brown <broonie@kernel.org>
Closes: https://lore.kernel.org/all/0b72286b-33dd-4bc9-8c0e-161c2f4baed8@sirena.org.uk/
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>
---
 drivers/reset/core.c | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)

diff --git a/drivers/reset/core.c b/drivers/reset/core.c
index e625cf59cfb0..4d27bbea1a69 100644
--- a/drivers/reset/core.c
+++ b/drivers/reset/core.c
@@ -132,20 +132,21 @@ int reset_controller_register(struct reset_controller_dev *rcdev)
 	if ((rcdev->of_node && rcdev->fwnode) || (rcdev->of_xlate && rcdev->fwnode_xlate))
 		return -EINVAL;
 
-	if (!rcdev->of_node && !rcdev->fwnode) {
+	if (rcdev->of_node && !rcdev->fwnode)
+		rcdev->fwnode = of_fwnode_handle(rcdev->of_node);
+
+	if (!rcdev->fwnode) {
 		rcdev->fwnode = dev_fwnode(rcdev->dev);
 		if (!rcdev->fwnode)
 			return -EINVAL;
 	}
 
-	if (rcdev->of_node) {
-		rcdev->fwnode = of_fwnode_handle(rcdev->of_node);
+	if (rcdev->of_xlate)
 		rcdev->fwnode_reset_n_cells = rcdev->of_reset_n_cells;
-	}
 
-	if (rcdev->fwnode && !rcdev->fwnode_xlate) {
-		rcdev->fwnode_reset_n_cells = 1;
+	if (!rcdev->fwnode_xlate && !rcdev->of_xlate) {
 		rcdev->fwnode_xlate = fwnode_reset_simple_xlate;
+		rcdev->fwnode_reset_n_cells = 1;
 	}
 
 	INIT_LIST_HEAD(&rcdev->reset_control_head);
-- 
2.47.3

Re: [PATCH] reset: don't overwrite fwnode_reset_n_cells

Posted by Tommaso Merciai 3 weeks, 6 days ago

Hi Bartosz,
Thanks for your patch.

On Tue, Mar 10, 2026 at 04:15:15PM +0100, Bartosz Golaszewski wrote:
> Fix a logic bug in reser_register() where we set fwnode_reset_n_cells to
> 1 if fwnode is set and fwnode_xlate is not but we do it after assigning
> of_fwnode_handle(of_node) to fwnode.
> 
> Modify the logic to: assign fwnode from of_node if applicable, if fwnode
> is still not set, try to get it from the device and only then check
> of_xlate and fwnode_xlate and either assign fwnode_reset_n_cells from OF
> or default to fwnode_reset_simple_xlate and fwnode_reset_n_cells = 1.
> 
> Fixes: ba8dbbb14b7e ("reset: convert the core API to using firmware nodes")

Tested on RZ/G3E.
Tested-by: Tommaso Merciai <tommaso.merciai.xr@bp.renesas.com>

Kind Regards,
Tommaso

> Reported-by: Mark Brown <broonie@kernel.org>
> Closes: https://lore.kernel.org/all/0b72286b-33dd-4bc9-8c0e-161c2f4baed8@sirena.org.uk/
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>
> ---
>  drivers/reset/core.c | 13 +++++++------
>  1 file changed, 7 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/reset/core.c b/drivers/reset/core.c
> index e625cf59cfb0..4d27bbea1a69 100644
> --- a/drivers/reset/core.c
> +++ b/drivers/reset/core.c
> @@ -132,20 +132,21 @@ int reset_controller_register(struct reset_controller_dev *rcdev)
>  	if ((rcdev->of_node && rcdev->fwnode) || (rcdev->of_xlate && rcdev->fwnode_xlate))
>  		return -EINVAL;
>  
> -	if (!rcdev->of_node && !rcdev->fwnode) {
> +	if (rcdev->of_node && !rcdev->fwnode)
> +		rcdev->fwnode = of_fwnode_handle(rcdev->of_node);
> +
> +	if (!rcdev->fwnode) {
>  		rcdev->fwnode = dev_fwnode(rcdev->dev);
>  		if (!rcdev->fwnode)
>  			return -EINVAL;
>  	}
>  
> -	if (rcdev->of_node) {
> -		rcdev->fwnode = of_fwnode_handle(rcdev->of_node);
> +	if (rcdev->of_xlate)
>  		rcdev->fwnode_reset_n_cells = rcdev->of_reset_n_cells;
> -	}
>  
> -	if (rcdev->fwnode && !rcdev->fwnode_xlate) {
> -		rcdev->fwnode_reset_n_cells = 1;
> +	if (!rcdev->fwnode_xlate && !rcdev->of_xlate) {
>  		rcdev->fwnode_xlate = fwnode_reset_simple_xlate;
> +		rcdev->fwnode_reset_n_cells = 1;
>  	}
>  
>  	INIT_LIST_HEAD(&rcdev->reset_control_head);
> -- 
> 2.47.3
>

Re: [PATCH] reset: don't overwrite fwnode_reset_n_cells

Posted by Philipp Zabel 3 weeks, 6 days ago

Hi Bartosz, Mark,

On Di, 2026-03-10 at 16:15 +0100, Bartosz Golaszewski wrote:
> Fix a logic bug in reser_register() where we set fwnode_reset_n_cells to
> 1 if fwnode is set and fwnode_xlate is not but we do it after assigning
> of_fwnode_handle(of_node) to fwnode.
> 
> Modify the logic to: assign fwnode from of_node if applicable, if fwnode
> is still not set, try to get it from the device and only then check
> of_xlate and fwnode_xlate and either assign fwnode_reset_n_cells from OF
> or default to fwnode_reset_simple_xlate and fwnode_reset_n_cells = 1.
> 
> Fixes: ba8dbbb14b7e ("reset: convert the core API to using firmware nodes")
> Reported-by: Mark Brown <broonie@kernel.org>
> Closes: https://lore.kernel.org/all/0b72286b-33dd-4bc9-8c0e-161c2f4baed8@sirena.org.uk/
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>

Applied to reset/next, thank you!

[1/1] reset: don't overwrite fwnode_reset_n_cells
      https://git.pengutronix.de/cgit/pza/linux/commit/?id=62d11b80ea5d

regards
Philipp

Re: [PATCH] reset: don't overwrite fwnode_reset_n_cells

Posted by Mark Brown 4 weeks ago

On Tue, Mar 10, 2026 at 04:15:15PM +0100, Bartosz Golaszewski wrote:
> Fix a logic bug in reser_register() where we set fwnode_reset_n_cells to
> 1 if fwnode is set and fwnode_xlate is not but we do it after assigning
> of_fwnode_handle(of_node) to fwnode.

Tested-by: Mark Brown <broonie@kernel.org>

Re: [PATCH] reset: don't overwrite fwnode_reset_n_cells

Posted by Geert Uytterhoeven 4 weeks ago

Hi Bartosz,

Hi Bartosz,

Thanks for your patch!

On Tue, 10 Mar 2026 at 16:15, Bartosz Golaszewski
<bartosz.golaszewski@oss.qualcomm.com> wrote:
> Fix a logic bug in reser_register() where we set fwnode_reset_n_cells to

One more non-existing function ;-)

reset_controller_register()

> 1 if fwnode is set and fwnode_xlate is not but we do it after assigning
> of_fwnode_handle(of_node) to fwnode.
>
> Modify the logic to: assign fwnode from of_node if applicable, if fwnode
> is still not set, try to get it from the device and only then check
> of_xlate and fwnode_xlate and either assign fwnode_reset_n_cells from OF
> or default to fwnode_reset_simple_xlate and fwnode_reset_n_cells = 1.
>
> Fixes: ba8dbbb14b7e ("reset: convert the core API to using firmware nodes")
> Reported-by: Mark Brown <broonie@kernel.org>
> Closes: https://lore.kernel.org/all/0b72286b-33dd-4bc9-8c0e-161c2f4baed8@sirena.org.uk/
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>

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

Re: [PATCH] reset: don't overwrite fwnode_reset_n_cells

Posted by Bartosz Golaszewski 4 weeks ago

On Tue, Mar 10, 2026 at 4:17 PM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>
> Hi Bartosz,
>
> Hi Bartosz,
>
> Thanks for your patch!
>
> On Tue, 10 Mar 2026 at 16:15, Bartosz Golaszewski
> <bartosz.golaszewski@oss.qualcomm.com> wrote:
> > Fix a logic bug in reser_register() where we set fwnode_reset_n_cells to
>
> One more non-existing function ;-)
>
> reset_controller_register()
>

Gah! If Mark says it fixes the issue, I'll either resend it with
commit message fixed or maybe Philipp can fix it when applying.

Bart

Re: [PATCH] reset: don't overwrite fwnode_reset_n_cells

Posted by Philipp Zabel 3 weeks, 6 days ago

On Di, 2026-03-10 at 16:19 +0100, Bartosz Golaszewski wrote:
> On Tue, Mar 10, 2026 at 4:17 PM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> > 
> > Hi Bartosz,
> > 
> > Hi Bartosz,
> > 
> > Thanks for your patch!
> > 
> > On Tue, 10 Mar 2026 at 16:15, Bartosz Golaszewski
> > <bartosz.golaszewski@oss.qualcomm.com> wrote:
> > > Fix a logic bug in reser_register() where we set fwnode_reset_n_cells to
> > 
> > One more non-existing function ;-)
> > 
> > reset_controller_register()
> > 
> 
> Gah! If Mark says it fixes the issue, I'll either resend it with
> commit message fixed or maybe Philipp can fix it when applying.

I'll do that.

Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>

regards
Philipp