[PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Charles Keepax]

This reverts commit 25decf0469d4c91d90aa2e28d996aed276bfc622.

This software node change doesn't actually fix any current issues
with the kernel, it is an improvement to the lookup process rather
than fixing a live bug. It also causes a couple of regressions with
shipping laptops, which relied on the label based lookup.

There is a fix for the regressions in mainline, the first 5 patches
of [1]. However, those patches are fairly substantial changes and
given the patch causing the regression doesn't actually fix a bug
it seems better to just revert it in stable.

CC: stable@vger.kernel.org # 6.12, 6.17
Link: https://lore.kernel.org/linux-sound/20251120-reset-gpios-swnodes-v7-0-a100493a0f4b@linaro.org/ [1]
Closes: https://github.com/thesofproject/linux/issues/5599
Closes: https://github.com/thesofproject/linux/issues/5603
Acked-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
---

I wasn't exactly sure of the proceedure for reverting a patch that was
cherry-picked to stable, so apologies if I have made any mistakes here
but happy to update if necessary.

Thanks,
Charles

 drivers/gpio/gpiolib-swnode.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpio/gpiolib-swnode.c b/drivers/gpio/gpiolib-swnode.c
index e3806db1c0e07..f21dbc28cf2c8 100644
--- a/drivers/gpio/gpiolib-swnode.c
+++ b/drivers/gpio/gpiolib-swnode.c
@@ -41,7 +41,7 @@ static struct gpio_device *swnode_get_gpio_device(struct fwnode_handle *fwnode)
 	    !strcmp(gdev_node->name, GPIOLIB_SWNODE_UNDEFINED_NAME))
 		return ERR_PTR(-ENOENT);
 
-	gdev = gpio_device_find_by_fwnode(fwnode);
+	gdev = gpio_device_find_by_label(gdev_node->name);
 	return gdev ?: ERR_PTR(-EPROBE_DEFER);
 }
 
-- 
2.47.3

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Bartosz Golaszewski]

On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
<ckeepax@opensource.cirrus.com> wrote:
>
> This reverts commit 25decf0469d4c91d90aa2e28d996aed276bfc622.
>
> This software node change doesn't actually fix any current issues
> with the kernel, it is an improvement to the lookup process rather
> than fixing a live bug. It also causes a couple of regressions with
> shipping laptops, which relied on the label based lookup.
>
> There is a fix for the regressions in mainline, the first 5 patches
> of [1]. However, those patches are fairly substantial changes and
> given the patch causing the regression doesn't actually fix a bug
> it seems better to just revert it in stable.
>
> CC: stable@vger.kernel.org # 6.12, 6.17
> Link: https://lore.kernel.org/linux-sound/20251120-reset-gpios-swnodes-v7-0-a100493a0f4b@linaro.org/ [1]
> Closes: https://github.com/thesofproject/linux/issues/5599
> Closes: https://github.com/thesofproject/linux/issues/5603
> Acked-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
> ---
>
> I wasn't exactly sure of the proceedure for reverting a patch that was
> cherry-picked to stable, so apologies if I have made any mistakes here
> but happy to update if necessary.
>

Yes, I'd like to stress the fact that this MUST NOT be reverted in
mainline, only in v6.12 and v6.17 stable branches.

Thanks,
Bartosz

> Thanks,
> Charles
>
>  drivers/gpio/gpiolib-swnode.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpio/gpiolib-swnode.c b/drivers/gpio/gpiolib-swnode.c
> index e3806db1c0e07..f21dbc28cf2c8 100644
> --- a/drivers/gpio/gpiolib-swnode.c
> +++ b/drivers/gpio/gpiolib-swnode.c
> @@ -41,7 +41,7 @@ static struct gpio_device *swnode_get_gpio_device(struct fwnode_handle *fwnode)
>             !strcmp(gdev_node->name, GPIOLIB_SWNODE_UNDEFINED_NAME))
>                 return ERR_PTR(-ENOENT);
>
> -       gdev = gpio_device_find_by_fwnode(fwnode);
> +       gdev = gpio_device_find_by_label(gdev_node->name);
>         return gdev ?: ERR_PTR(-EPROBE_DEFER);
>  }
>
> --
> 2.47.3
>

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Charles Keepax]

On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> <ckeepax@opensource.cirrus.com> wrote:
> >
> > This reverts commit 25decf0469d4c91d90aa2e28d996aed276bfc622.
> >
> > This software node change doesn't actually fix any current issues
> > with the kernel, it is an improvement to the lookup process rather
> > than fixing a live bug. It also causes a couple of regressions with
> > shipping laptops, which relied on the label based lookup.
> >
> > There is a fix for the regressions in mainline, the first 5 patches
> > of [1]. However, those patches are fairly substantial changes and
> > given the patch causing the regression doesn't actually fix a bug
> > it seems better to just revert it in stable.
> >
> > CC: stable@vger.kernel.org # 6.12, 6.17
> > Link: https://lore.kernel.org/linux-sound/20251120-reset-gpios-swnodes-v7-0-a100493a0f4b@linaro.org/ [1]
> > Closes: https://github.com/thesofproject/linux/issues/5599
> > Closes: https://github.com/thesofproject/linux/issues/5603
> > Acked-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> > Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
> > ---
> >
> > I wasn't exactly sure of the proceedure for reverting a patch that was
> > cherry-picked to stable, so apologies if I have made any mistakes here
> > but happy to update if necessary.
> >
> 
> Yes, I'd like to stress the fact that this MUST NOT be reverted in
> mainline, only in v6.12 and v6.17 stable branches.

Indeed yes, apologies if that wasn't clear.

Thanks,
Charles

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Greg KH]

On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> <ckeepax@opensource.cirrus.com> wrote:
> >
> > This reverts commit 25decf0469d4c91d90aa2e28d996aed276bfc622.
> >
> > This software node change doesn't actually fix any current issues
> > with the kernel, it is an improvement to the lookup process rather
> > than fixing a live bug. It also causes a couple of regressions with
> > shipping laptops, which relied on the label based lookup.
> >
> > There is a fix for the regressions in mainline, the first 5 patches
> > of [1]. However, those patches are fairly substantial changes and
> > given the patch causing the regression doesn't actually fix a bug
> > it seems better to just revert it in stable.
> >
> > CC: stable@vger.kernel.org # 6.12, 6.17
> > Link: https://lore.kernel.org/linux-sound/20251120-reset-gpios-swnodes-v7-0-a100493a0f4b@linaro.org/ [1]
> > Closes: https://github.com/thesofproject/linux/issues/5599
> > Closes: https://github.com/thesofproject/linux/issues/5603
> > Acked-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> > Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
> > ---
> >
> > I wasn't exactly sure of the proceedure for reverting a patch that was
> > cherry-picked to stable, so apologies if I have made any mistakes here
> > but happy to update if necessary.
> >
> 
> Yes, I'd like to stress the fact that this MUST NOT be reverted in
> mainline, only in v6.12 and v6.17 stable branches.

But why?  Why not take the upstream changes instead?  We would much
rather do that as it reduces the divergance.  5 patches is trivial for
us to take.

thanks,

greg k-h

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Charles Keepax]

On Tue, Nov 25, 2025 at 12:43:16PM +0100, Greg KH wrote:
> On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> > On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> > <ckeepax@opensource.cirrus.com> wrote:
> > >
> > > This reverts commit 25decf0469d4c91d90aa2e28d996aed276bfc622.
> > >
> > > This software node change doesn't actually fix any current issues
> > > with the kernel, it is an improvement to the lookup process rather
> > > than fixing a live bug. It also causes a couple of regressions with
> > > shipping laptops, which relied on the label based lookup.
> > >
> > > There is a fix for the regressions in mainline, the first 5 patches
> > > of [1]. However, those patches are fairly substantial changes and
> > > given the patch causing the regression doesn't actually fix a bug
> > > it seems better to just revert it in stable.
> > >
> > > CC: stable@vger.kernel.org # 6.12, 6.17
> > > Link: https://lore.kernel.org/linux-sound/20251120-reset-gpios-swnodes-v7-0-a100493a0f4b@linaro.org/ [1]
> > > Closes: https://github.com/thesofproject/linux/issues/5599
> > > Closes: https://github.com/thesofproject/linux/issues/5603
> > > Acked-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> > > Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
> > > ---
> > >
> > > I wasn't exactly sure of the proceedure for reverting a patch that was
> > > cherry-picked to stable, so apologies if I have made any mistakes here
> > > but happy to update if necessary.
> > >
> > 
> > Yes, I'd like to stress the fact that this MUST NOT be reverted in
> > mainline, only in v6.12 and v6.17 stable branches.
> 
> But why?  Why not take the upstream changes instead?  We would much
> rather do that as it reduces the divergance.  5 patches is trivial for
> us to take.

My thinking was that they are a bit invasive for backports, as
noted in the commit message. But if that is the preferred option
I can do a series with those instead?

Thanks,
Charles

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Greg KH]

On Tue, Nov 25, 2025 at 11:48:02AM +0000, Charles Keepax wrote:
> On Tue, Nov 25, 2025 at 12:43:16PM +0100, Greg KH wrote:
> > On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> > > On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> > > <ckeepax@opensource.cirrus.com> wrote:
> > > >
> > > > This reverts commit 25decf0469d4c91d90aa2e28d996aed276bfc622.
> > > >
> > > > This software node change doesn't actually fix any current issues
> > > > with the kernel, it is an improvement to the lookup process rather
> > > > than fixing a live bug. It also causes a couple of regressions with
> > > > shipping laptops, which relied on the label based lookup.
> > > >
> > > > There is a fix for the regressions in mainline, the first 5 patches
> > > > of [1]. However, those patches are fairly substantial changes and
> > > > given the patch causing the regression doesn't actually fix a bug
> > > > it seems better to just revert it in stable.
> > > >
> > > > CC: stable@vger.kernel.org # 6.12, 6.17
> > > > Link: https://lore.kernel.org/linux-sound/20251120-reset-gpios-swnodes-v7-0-a100493a0f4b@linaro.org/ [1]
> > > > Closes: https://github.com/thesofproject/linux/issues/5599
> > > > Closes: https://github.com/thesofproject/linux/issues/5603
> > > > Acked-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> > > > Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
> > > > ---
> > > >
> > > > I wasn't exactly sure of the proceedure for reverting a patch that was
> > > > cherry-picked to stable, so apologies if I have made any mistakes here
> > > > but happy to update if necessary.
> > > >
> > > 
> > > Yes, I'd like to stress the fact that this MUST NOT be reverted in
> > > mainline, only in v6.12 and v6.17 stable branches.
> > 
> > But why?  Why not take the upstream changes instead?  We would much
> > rather do that as it reduces the divergance.  5 patches is trivial for
> > us to take.
> 
> My thinking was that they are a bit invasive for backports, as
> noted in the commit message. But if that is the preferred option
> I can do a series with those instead?

I'd prefer to take what is upstream, it's simpler over the long term to
do so.

thanks,

greg k-h

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Charles Keepax]

On Tue, Nov 25, 2025 at 12:58:30PM +0100, Greg KH wrote:
> On Tue, Nov 25, 2025 at 11:48:02AM +0000, Charles Keepax wrote:
> > On Tue, Nov 25, 2025 at 12:43:16PM +0100, Greg KH wrote:
> > > On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> > > > On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> > > > <ckeepax@opensource.cirrus.com> wrote:
> > > > >
> > > > > This reverts commit 25decf0469d4c91d90aa2e28d996aed276bfc622.
> > > > >
> > > > > This software node change doesn't actually fix any current issues
> > > > > with the kernel, it is an improvement to the lookup process rather
> > > > > than fixing a live bug. It also causes a couple of regressions with
> > > > > shipping laptops, which relied on the label based lookup.
> > > > >
> > > > > There is a fix for the regressions in mainline, the first 5 patches
> > > > > of [1]. However, those patches are fairly substantial changes and
> > > > > given the patch causing the regression doesn't actually fix a bug
> > > > > it seems better to just revert it in stable.
> > > > >
> > > > > CC: stable@vger.kernel.org # 6.12, 6.17
> > > > > Link: https://lore.kernel.org/linux-sound/20251120-reset-gpios-swnodes-v7-0-a100493a0f4b@linaro.org/ [1]
> > > > > Closes: https://github.com/thesofproject/linux/issues/5599
> > > > > Closes: https://github.com/thesofproject/linux/issues/5603
> > > > > Acked-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> > > > > Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
> > > > > ---
> > > > >
> > > > > I wasn't exactly sure of the proceedure for reverting a patch that was
> > > > > cherry-picked to stable, so apologies if I have made any mistakes here
> > > > > but happy to update if necessary.
> > > > >
> > > > 
> > > > Yes, I'd like to stress the fact that this MUST NOT be reverted in
> > > > mainline, only in v6.12 and v6.17 stable branches.
> > > 
> > > But why?  Why not take the upstream changes instead?  We would much
> > > rather do that as it reduces the divergance.  5 patches is trivial for
> > > us to take.
> > 
> > My thinking was that they are a bit invasive for backports, as
> > noted in the commit message. But if that is the preferred option
> > I can do a series with those instead?
> 
> I'd prefer to take what is upstream, it's simpler over the long term to
> do so.
> 

I really doubt this will end up simpler, as the comparison here
is a) not backporting a change that probably shouldn't have gone
to stable in the first place vs. b) backport a bunch of quite
invasive changes.

Do we have to wait for the fixes to hit Linus's tree before
pushing them to stable? As they are still in Philipp Zabel's
reset tree at the moment and I would quite like to stem the
rising tide of tickets I am getting about audio breaking on
peoples laptops as soon as possible.

Thanks,
Charles

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Greg KH]

On Tue, Nov 25, 2025 at 12:49:59PM +0000, Charles Keepax wrote:
> On Tue, Nov 25, 2025 at 12:58:30PM +0100, Greg KH wrote:
> > On Tue, Nov 25, 2025 at 11:48:02AM +0000, Charles Keepax wrote:
> > > On Tue, Nov 25, 2025 at 12:43:16PM +0100, Greg KH wrote:
> > > > On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> > > > > On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> > > > > <ckeepax@opensource.cirrus.com> wrote:
> > > > > >
> > > > > > This reverts commit 25decf0469d4c91d90aa2e28d996aed276bfc622.
> > > > > >
> > > > > > This software node change doesn't actually fix any current issues
> > > > > > with the kernel, it is an improvement to the lookup process rather
> > > > > > than fixing a live bug. It also causes a couple of regressions with
> > > > > > shipping laptops, which relied on the label based lookup.
> > > > > >
> > > > > > There is a fix for the regressions in mainline, the first 5 patches
> > > > > > of [1]. However, those patches are fairly substantial changes and
> > > > > > given the patch causing the regression doesn't actually fix a bug
> > > > > > it seems better to just revert it in stable.
> > > > > >
> > > > > > CC: stable@vger.kernel.org # 6.12, 6.17
> > > > > > Link: https://lore.kernel.org/linux-sound/20251120-reset-gpios-swnodes-v7-0-a100493a0f4b@linaro.org/ [1]
> > > > > > Closes: https://github.com/thesofproject/linux/issues/5599
> > > > > > Closes: https://github.com/thesofproject/linux/issues/5603
> > > > > > Acked-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> > > > > > Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
> > > > > > ---
> > > > > >
> > > > > > I wasn't exactly sure of the proceedure for reverting a patch that was
> > > > > > cherry-picked to stable, so apologies if I have made any mistakes here
> > > > > > but happy to update if necessary.
> > > > > >
> > > > > 
> > > > > Yes, I'd like to stress the fact that this MUST NOT be reverted in
> > > > > mainline, only in v6.12 and v6.17 stable branches.
> > > > 
> > > > But why?  Why not take the upstream changes instead?  We would much
> > > > rather do that as it reduces the divergance.  5 patches is trivial for
> > > > us to take.
> > > 
> > > My thinking was that they are a bit invasive for backports, as
> > > noted in the commit message. But if that is the preferred option
> > > I can do a series with those instead?
> > 
> > I'd prefer to take what is upstream, it's simpler over the long term to
> > do so.
> > 
> 
> I really doubt this will end up simpler, as the comparison here
> is a) not backporting a change that probably shouldn't have gone
> to stable in the first place vs. b) backport a bunch of quite
> invasive changes.

But think about future changes/fixes.  6.12 is going to be around for 5
more years, doing one-off fixes ensures that any future changes/fixes
will NOT apply to 6.12.y and require custom changes that are almost
guaranteed to break.

Again, it is almost always better to take the same changes that are in
Linus's tree as they are better tested and future fixes apply cleaner.

But I defer to the maintainer, of the maintainer says to take this
one-off change (i.e. revert), I'll gladly do so.  Just trying to explain
that taking lots of upstream changes is almost always the right thing to
do in the long run.  And we are in this for the long run.

> Do we have to wait for the fixes to hit Linus's tree before
> pushing them to stable? As they are still in Philipp Zabel's
> reset tree at the moment and I would quite like to stem the
> rising tide of tickets I am getting about audio breaking on
> peoples laptops as soon as possible.

Yes, we need the fixes there first.

thanks,

greg k-h

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Charles Keepax]

On Thu, Nov 27, 2025 at 02:51:50PM +0100, Greg KH wrote:
> On Tue, Nov 25, 2025 at 12:49:59PM +0000, Charles Keepax wrote:
> > On Tue, Nov 25, 2025 at 12:58:30PM +0100, Greg KH wrote:
> > > On Tue, Nov 25, 2025 at 11:48:02AM +0000, Charles Keepax wrote:
> > > > On Tue, Nov 25, 2025 at 12:43:16PM +0100, Greg KH wrote:
> > > > > On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> > > > > > On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> > > > > > <ckeepax@opensource.cirrus.com> wrote:
> > Do we have to wait for the fixes to hit Linus's tree before
> > pushing them to stable? As they are still in Philipp Zabel's
> > reset tree at the moment and I would quite like to stem the
> > rising tide of tickets I am getting about audio breaking on
> > peoples laptops as soon as possible.
> 
> Yes, we need the fixes there first.

Fair enough, but it is super sad that everyone has to sit around
with broken devices until after the merge window. This is not a
theoretical issue people are complaining about this now.

Thanks,
Charles

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Greg KH]

On Thu, Nov 27, 2025 at 02:04:48PM +0000, Charles Keepax wrote:
> On Thu, Nov 27, 2025 at 02:51:50PM +0100, Greg KH wrote:
> > On Tue, Nov 25, 2025 at 12:49:59PM +0000, Charles Keepax wrote:
> > > On Tue, Nov 25, 2025 at 12:58:30PM +0100, Greg KH wrote:
> > > > On Tue, Nov 25, 2025 at 11:48:02AM +0000, Charles Keepax wrote:
> > > > > On Tue, Nov 25, 2025 at 12:43:16PM +0100, Greg KH wrote:
> > > > > > On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> > > > > > > On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> > > > > > > <ckeepax@opensource.cirrus.com> wrote:
> > > Do we have to wait for the fixes to hit Linus's tree before
> > > pushing them to stable? As they are still in Philipp Zabel's
> > > reset tree at the moment and I would quite like to stem the
> > > rising tide of tickets I am getting about audio breaking on
> > > peoples laptops as soon as possible.
> > 
> > Yes, we need the fixes there first.
> 
> Fair enough, but it is super sad that everyone has to sit around
> with broken devices until after the merge window. This is not a
> theoretical issue people are complaining about this now.

Are people sitting around with this issue in 6.18-rc releases now?  Is
6.18-final going to be broken in the same way?

confused,

greg k-h

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Charles Keepax]

On Thu, Nov 27, 2025 at 03:07:48PM +0100, Greg KH wrote:
> On Thu, Nov 27, 2025 at 02:04:48PM +0000, Charles Keepax wrote:
> > On Thu, Nov 27, 2025 at 02:51:50PM +0100, Greg KH wrote:
> > > On Tue, Nov 25, 2025 at 12:49:59PM +0000, Charles Keepax wrote:
> > > > On Tue, Nov 25, 2025 at 12:58:30PM +0100, Greg KH wrote:
> > > > > On Tue, Nov 25, 2025 at 11:48:02AM +0000, Charles Keepax wrote:
> > > > > > On Tue, Nov 25, 2025 at 12:43:16PM +0100, Greg KH wrote:
> > > > > > > On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> > > > > > > > On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> > > > > > > > <ckeepax@opensource.cirrus.com> wrote:
> > > > Do we have to wait for the fixes to hit Linus's tree before
> > > > pushing them to stable? As they are still in Philipp Zabel's
> > > > reset tree at the moment and I would quite like to stem the
> > > > rising tide of tickets I am getting about audio breaking on
> > > > peoples laptops as soon as possible.
> > > 
> > > Yes, we need the fixes there first.
> > 
> > Fair enough, but it is super sad that everyone has to sit around
> > with broken devices until after the merge window. This is not a
> > theoretical issue people are complaining about this now.
> 
> Are people sitting around with this issue in 6.18-rc releases now?  Is
> 6.18-final going to be broken in the same way?

Yeah regrettably that is going to be broken too, at least until
the first stable release either does the same revert or backports
the same fixes.

Thanks,
Charles

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Greg KH]

On Thu, Nov 27, 2025 at 02:10:32PM +0000, Charles Keepax wrote:
> On Thu, Nov 27, 2025 at 03:07:48PM +0100, Greg KH wrote:
> > On Thu, Nov 27, 2025 at 02:04:48PM +0000, Charles Keepax wrote:
> > > On Thu, Nov 27, 2025 at 02:51:50PM +0100, Greg KH wrote:
> > > > On Tue, Nov 25, 2025 at 12:49:59PM +0000, Charles Keepax wrote:
> > > > > On Tue, Nov 25, 2025 at 12:58:30PM +0100, Greg KH wrote:
> > > > > > On Tue, Nov 25, 2025 at 11:48:02AM +0000, Charles Keepax wrote:
> > > > > > > On Tue, Nov 25, 2025 at 12:43:16PM +0100, Greg KH wrote:
> > > > > > > > On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> > > > > > > > > On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> > > > > > > > > <ckeepax@opensource.cirrus.com> wrote:
> > > > > Do we have to wait for the fixes to hit Linus's tree before
> > > > > pushing them to stable? As they are still in Philipp Zabel's
> > > > > reset tree at the moment and I would quite like to stem the
> > > > > rising tide of tickets I am getting about audio breaking on
> > > > > peoples laptops as soon as possible.
> > > > 
> > > > Yes, we need the fixes there first.
> > > 
> > > Fair enough, but it is super sad that everyone has to sit around
> > > with broken devices until after the merge window. This is not a
> > > theoretical issue people are complaining about this now.
> > 
> > Are people sitting around with this issue in 6.18-rc releases now?  Is
> > 6.18-final going to be broken in the same way?
> 
> Yeah regrettably that is going to be broken too, at least until
> the first stable release either does the same revert or backports
> the same fixes.

Great, we are "bug compatible!"  :)

Seriously, this happens for minor things all the time, not that big of a
deal normally.

thanks,

greg k-h

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Bartosz Golaszewski]

On Thu, Nov 27, 2025 at 3:17 PM Greg KH <gregkh@linuxfoundation.org> wrote:
>
> On Thu, Nov 27, 2025 at 02:10:32PM +0000, Charles Keepax wrote:
> > On Thu, Nov 27, 2025 at 03:07:48PM +0100, Greg KH wrote:
> > > On Thu, Nov 27, 2025 at 02:04:48PM +0000, Charles Keepax wrote:
> > > > On Thu, Nov 27, 2025 at 02:51:50PM +0100, Greg KH wrote:
> > > > > On Tue, Nov 25, 2025 at 12:49:59PM +0000, Charles Keepax wrote:
> > > > > > On Tue, Nov 25, 2025 at 12:58:30PM +0100, Greg KH wrote:
> > > > > > > On Tue, Nov 25, 2025 at 11:48:02AM +0000, Charles Keepax wrote:
> > > > > > > > On Tue, Nov 25, 2025 at 12:43:16PM +0100, Greg KH wrote:
> > > > > > > > > On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> > > > > > > > > > On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> > > > > > > > > > <ckeepax@opensource.cirrus.com> wrote:
> > > > > > Do we have to wait for the fixes to hit Linus's tree before
> > > > > > pushing them to stable? As they are still in Philipp Zabel's
> > > > > > reset tree at the moment and I would quite like to stem the
> > > > > > rising tide of tickets I am getting about audio breaking on
> > > > > > peoples laptops as soon as possible.
> > > > >
> > > > > Yes, we need the fixes there first.
> > > >
> > > > Fair enough, but it is super sad that everyone has to sit around
> > > > with broken devices until after the merge window. This is not a
> > > > theoretical issue people are complaining about this now.
> > >
> > > Are people sitting around with this issue in 6.18-rc releases now?  Is
> > > 6.18-final going to be broken in the same way?
> >
> > Yeah regrettably that is going to be broken too, at least until
> > the first stable release either does the same revert or backports
> > the same fixes.
>
> Great, we are "bug compatible!"  :)
>
> Seriously, this happens for minor things all the time, not that big of a
> deal normally.
>

Just my two cents: this feature interacts quite a lot with another new
feature: shared GPIOs in GPIOLIB. I've already either queued or have
under review ~7 other fixes. Since in stable, the code from this
series would not interact with gpiolib-shared (because no way this
should get backported), we'd still have a bit of a different
environment in mainline and stable branches.

I would very much prefer to revert the patch in question than worry
about divergences in logic.

Bartosz

Re: [PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Greg KH]

On Thu, Nov 27, 2025 at 03:45:29PM +0100, Bartosz Golaszewski wrote:
> On Thu, Nov 27, 2025 at 3:17 PM Greg KH <gregkh@linuxfoundation.org> wrote:
> >
> > On Thu, Nov 27, 2025 at 02:10:32PM +0000, Charles Keepax wrote:
> > > On Thu, Nov 27, 2025 at 03:07:48PM +0100, Greg KH wrote:
> > > > On Thu, Nov 27, 2025 at 02:04:48PM +0000, Charles Keepax wrote:
> > > > > On Thu, Nov 27, 2025 at 02:51:50PM +0100, Greg KH wrote:
> > > > > > On Tue, Nov 25, 2025 at 12:49:59PM +0000, Charles Keepax wrote:
> > > > > > > On Tue, Nov 25, 2025 at 12:58:30PM +0100, Greg KH wrote:
> > > > > > > > On Tue, Nov 25, 2025 at 11:48:02AM +0000, Charles Keepax wrote:
> > > > > > > > > On Tue, Nov 25, 2025 at 12:43:16PM +0100, Greg KH wrote:
> > > > > > > > > > On Tue, Nov 25, 2025 at 11:31:56AM +0100, Bartosz Golaszewski wrote:
> > > > > > > > > > > On Tue, Nov 25, 2025 at 11:29 AM Charles Keepax
> > > > > > > > > > > <ckeepax@opensource.cirrus.com> wrote:
> > > > > > > Do we have to wait for the fixes to hit Linus's tree before
> > > > > > > pushing them to stable? As they are still in Philipp Zabel's
> > > > > > > reset tree at the moment and I would quite like to stem the
> > > > > > > rising tide of tickets I am getting about audio breaking on
> > > > > > > peoples laptops as soon as possible.
> > > > > >
> > > > > > Yes, we need the fixes there first.
> > > > >
> > > > > Fair enough, but it is super sad that everyone has to sit around
> > > > > with broken devices until after the merge window. This is not a
> > > > > theoretical issue people are complaining about this now.
> > > >
> > > > Are people sitting around with this issue in 6.18-rc releases now?  Is
> > > > 6.18-final going to be broken in the same way?
> > >
> > > Yeah regrettably that is going to be broken too, at least until
> > > the first stable release either does the same revert or backports
> > > the same fixes.
> >
> > Great, we are "bug compatible!"  :)
> >
> > Seriously, this happens for minor things all the time, not that big of a
> > deal normally.
> >
> 
> Just my two cents: this feature interacts quite a lot with another new
> feature: shared GPIOs in GPIOLIB. I've already either queued or have
> under review ~7 other fixes. Since in stable, the code from this
> series would not interact with gpiolib-shared (because no way this
> should get backported), we'd still have a bit of a different
> environment in mainline and stable branches.
> 
> I would very much prefer to revert the patch in question than worry
> about divergences in logic.

Ok, fair enough, now reverted!

thanks,

greg k-h

[PATCH] Revert "gpio: swnode: don't use the swnode's name as the key for GPIO lookup"

[Charles Keepax]

This reverts commit 25decf0469d4c91d90aa2e28d996aed276bfc622.

This software node change doesn't actually fix any current issues
with the kernel, it is an improvement to the lookup process rather
than fixing a live bug. It also causes a couple of regressions with
shipping laptops, which relied on the label based lookup.

There is a fix for the regressions in mainline, the first 5 patches
of [1]. However, those patches are fairly substantial changes and
given the patch causing the regression doesn't actually fix a bug
it seems better to just revert it in stable.

CC: stable@vger.kernel.org # 6.18
Link: https://lore.kernel.org/linux-sound/20251120-reset-gpios-swnodes-v7-0-a100493a0f4b@linaro.org/ [1]
Link: https://lore.kernel.org/stable/20251125102924.3612459-1-ckeepax@opensource.cirrus.com/ [2]
Closes: https://github.com/thesofproject/linux/issues/5599
Closes: https://github.com/thesofproject/linux/issues/5603
Acked-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
---

This fix for the software node lookups is also required on 6.18 stable,
see the discussion for 6.12/6.17 in [2] for why we are doing a revert
rather than backporting the other fixes. The "full" fixes are merged in
6.19 so this should be the last kernel we need to push this revert onto.

Thanks,
Charles

 drivers/gpio/gpiolib-swnode.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpio/gpiolib-swnode.c b/drivers/gpio/gpiolib-swnode.c
index e3806db1c0e07..f21dbc28cf2c8 100644
--- a/drivers/gpio/gpiolib-swnode.c
+++ b/drivers/gpio/gpiolib-swnode.c
@@ -41,7 +41,7 @@ static struct gpio_device *swnode_get_gpio_device(struct fwnode_handle *fwnode)
 	    !strcmp(gdev_node->name, GPIOLIB_SWNODE_UNDEFINED_NAME))
 		return ERR_PTR(-ENOENT);
 
-	gdev = gpio_device_find_by_fwnode(fwnode);
+	gdev = gpio_device_find_by_label(gdev_node->name);
 	return gdev ?: ERR_PTR(-EPROBE_DEFER);
 }
 
-- 
2.47.3