This may be useful for debugging and develompent purposes, when there are
drivers that depend on regulators to be enabled but do not request them.
It is inspired from the clk_ignore_unused and pd_ignore_unused parameters,
that are used to keep firmware-enabled clocks and power domains on even if
these are not used by drivers.
The parameter is not expected to be used in normal cases and should not be
needed on a platform with proper driver support.
Signed-off-by: Javier Martinez Canillas <[email protected]>
---
Documentation/admin-guide/kernel-parameters.txt | 7 +++++++
drivers/regulator/core.c | 17 +++++++++++++++++
2 files changed, 24 insertions(+)
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 0a1731a0f0ef..91b58d767c2c 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -5461,6 +5461,13 @@
print every Nth verbose statement, where N is the value
specified.
+ regulator_ignore_unused
+ [REGULATOR]
+ Prevents the regulator framework to disable regulators
+ that are unused due not driver claiming them. This may
+ be useful for debug and development, but should not be
+ needed on a platform with proper driver support.
+
relax_domain_level=
[KNL, SMP] Set scheduler's default relax_domain_level.
See Documentation/admin-guide/cgroup-v1/cpusets.rst.
diff --git a/drivers/regulator/core.c b/drivers/regulator/core.c
index 3137e40fcd3e..220034ff0273 100644
--- a/drivers/regulator/core.c
+++ b/drivers/regulator/core.c
@@ -6234,6 +6234,14 @@ static int regulator_late_cleanup(struct device *dev, void *data)
return 0;
}
+static bool regulator_ignore_unused;
+static int __init regulator_ignore_unused_setup(char *__unused)
+{
+ regulator_ignore_unused = true;
+ return 1;
+}
+__setup("regulator_ignore_unused", regulator_ignore_unused_setup);
+
static void regulator_init_complete_work_function(struct work_struct *work)
{
/*
@@ -6246,6 +6254,15 @@ static void regulator_init_complete_work_function(struct work_struct *work)
class_for_each_device(®ulator_class, NULL, NULL,
regulator_register_resolve_supply);
+ /*
+ * For debugging purposes, it may be useful to prevent unused
+ * regulators to be disabled.
+ */
+ if (regulator_ignore_unused) {
+ pr_warn("regulator: Not disabling unused regulators\n");
+ return;
+ }
+
/* If we have a full configuration then disable any regulators
* we have permission to change the status for and which are
* not in use or always_on. This is effectively the default
--
2.41.0
Hi,
On 10/28/23 03:24, Javier Martinez Canillas wrote:
> This may be useful for debugging and develompent purposes, when there are
> drivers that depend on regulators to be enabled but do not request them.
>
> It is inspired from the clk_ignore_unused and pd_ignore_unused parameters,
> that are used to keep firmware-enabled clocks and power domains on even if
> these are not used by drivers.
>
> The parameter is not expected to be used in normal cases and should not be
> needed on a platform with proper driver support.
>
> Signed-off-by: Javier Martinez Canillas <[email protected]>
> ---
>
> Documentation/admin-guide/kernel-parameters.txt | 7 +++++++
> drivers/regulator/core.c | 17 +++++++++++++++++
> 2 files changed, 24 insertions(+)
>
> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> index 0a1731a0f0ef..91b58d767c2c 100644
> --- a/Documentation/admin-guide/kernel-parameters.txt
> +++ b/Documentation/admin-guide/kernel-parameters.txt
> @@ -5461,6 +5461,13 @@
> print every Nth verbose statement, where N is the value
> specified.
>
> + regulator_ignore_unused
> + [REGULATOR]
> + Prevents the regulator framework to disable regulators
from disabling regulators
> + that are unused due not driver claiming them. This may
no driver claiming them.
> + be useful for debug and development, but should not be
> + needed on a platform with proper driver support.
> +
> relax_domain_level=
> [KNL, SMP] Set scheduler's default relax_domain_level.
> See Documentation/admin-guide/cgroup-v1/cpusets.rst.
> diff --git a/drivers/regulator/core.c b/drivers/regulator/core.c
> index 3137e40fcd3e..220034ff0273 100644
> --- a/drivers/regulator/core.c
> +++ b/drivers/regulator/core.c
> @@ -6234,6 +6234,14 @@ static int regulator_late_cleanup(struct device *dev, void *data)
> return 0;
> }
>
> +static bool regulator_ignore_unused;
> +static int __init regulator_ignore_unused_setup(char *__unused)
> +{
> + regulator_ignore_unused = true;
> + return 1;
> +}
> +__setup("regulator_ignore_unused", regulator_ignore_unused_setup);
> +
> static void regulator_init_complete_work_function(struct work_struct *work)
> {
> /*
> @@ -6246,6 +6254,15 @@ static void regulator_init_complete_work_function(struct work_struct *work)
> class_for_each_device(®ulator_class, NULL, NULL,
> regulator_register_resolve_supply);
>
> + /*
> + * For debugging purposes, it may be useful to prevent unused
> + * regulators to be disabled.
from being disabled.
> + */
> + if (regulator_ignore_unused) {
> + pr_warn("regulator: Not disabling unused regulators\n");
> + return;
> + }
> +
> /* If we have a full configuration then disable any regulators
> * we have permission to change the status for and which are
> * not in use or always_on. This is effectively the default
HTH.
--
~Randy
On Sat, Oct 28, 2023 at 12:24:03PM +0200, Javier Martinez Canillas wrote:
> This may be useful for debugging and develompent purposes, when there are
> drivers that depend on regulators to be enabled but do not request them.
>
> It is inspired from the clk_ignore_unused and pd_ignore_unused parameters,
> that are used to keep firmware-enabled clocks and power domains on even if
> these are not used by drivers.
>
> The parameter is not expected to be used in normal cases and should not be
> needed on a platform with proper driver support.
>
> Signed-off-by: Javier Martinez Canillas <[email protected]>
Thanks for posting this! I've had a need for this. With Randy's feedback
addressed:
Reviewed-by: Brian Masney <[email protected]>
On Sat, Oct 28, 2023 at 12:24:03PM +0200, Javier Martinez Canillas wrote:
> This may be useful for debugging and develompent purposes, when there are
> drivers that depend on regulators to be enabled but do not request them.
>
> It is inspired from the clk_ignore_unused and pd_ignore_unused parameters,
> that are used to keep firmware-enabled clocks and power domains on even if
> these are not used by drivers.
>
> The parameter is not expected to be used in normal cases and should not be
> needed on a platform with proper driver support.
>
> Signed-off-by: Javier Martinez Canillas <[email protected]>
Acked-by: Manivannan Sadhasivam <[email protected]>
Thanks for looking into this. This patch is very handy to debug display related
issues on platforms without serial console.
On the other note, I'm wondering if we could use sync_state() for handling the
regulator_init_complete() work. This would ensure that the regulators are only
disabled when all the consumers are probed.
- Mani
> ---
>
> Documentation/admin-guide/kernel-parameters.txt | 7 +++++++
> drivers/regulator/core.c | 17 +++++++++++++++++
> 2 files changed, 24 insertions(+)
>
> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> index 0a1731a0f0ef..91b58d767c2c 100644
> --- a/Documentation/admin-guide/kernel-parameters.txt
> +++ b/Documentation/admin-guide/kernel-parameters.txt
> @@ -5461,6 +5461,13 @@
> print every Nth verbose statement, where N is the value
> specified.
>
> + regulator_ignore_unused
> + [REGULATOR]
> + Prevents the regulator framework to disable regulators
> + that are unused due not driver claiming them. This may
> + be useful for debug and development, but should not be
> + needed on a platform with proper driver support.
> +
> relax_domain_level=
> [KNL, SMP] Set scheduler's default relax_domain_level.
> See Documentation/admin-guide/cgroup-v1/cpusets.rst.
> diff --git a/drivers/regulator/core.c b/drivers/regulator/core.c
> index 3137e40fcd3e..220034ff0273 100644
> --- a/drivers/regulator/core.c
> +++ b/drivers/regulator/core.c
> @@ -6234,6 +6234,14 @@ static int regulator_late_cleanup(struct device *dev, void *data)
> return 0;
> }
>
> +static bool regulator_ignore_unused;
> +static int __init regulator_ignore_unused_setup(char *__unused)
> +{
> + regulator_ignore_unused = true;
> + return 1;
> +}
> +__setup("regulator_ignore_unused", regulator_ignore_unused_setup);
> +
> static void regulator_init_complete_work_function(struct work_struct *work)
> {
> /*
> @@ -6246,6 +6254,15 @@ static void regulator_init_complete_work_function(struct work_struct *work)
> class_for_each_device(®ulator_class, NULL, NULL,
> regulator_register_resolve_supply);
>
> + /*
> + * For debugging purposes, it may be useful to prevent unused
> + * regulators to be disabled.
> + */
> + if (regulator_ignore_unused) {
> + pr_warn("regulator: Not disabling unused regulators\n");
> + return;
> + }
> +
> /* If we have a full configuration then disable any regulators
> * we have permission to change the status for and which are
> * not in use or always_on. This is effectively the default
> --
> 2.41.0
>
--
மணிவண்ணன் சதாசிவம்
On Wed, Nov 01, 2023 at 10:26:52AM +0530, Manivannan Sadhasivam wrote:
> On the other note, I'm wondering if we could use sync_state() for handling the
> regulator_init_complete() work. This would ensure that the regulators are only
> disabled when all the consumers are probed.
That assumes that everything defined in the DT both has a driver and has
the driver available for the currently running kernel neither of which
is a good assumption.
Mark Brown <[email protected]> writes:
> On Wed, Nov 01, 2023 at 10:26:52AM +0530, Manivannan Sadhasivam wrote:
>
>> On the other note, I'm wondering if we could use sync_state() for handling the
>> regulator_init_complete() work. This would ensure that the regulators are only
>> disabled when all the consumers are probed.
>
> That assumes that everything defined in the DT both has a driver and has
> the driver available for the currently running kernel neither of which
> is a good assumption.
Agreed. I believe the current logic of disabling all regulators using a
workqueue is the correct thing to do.
The only better option I think is to make user-space notify the kernel
that it won't load kernel modules anymore. But the delayed work would
sill be needed, since the kernel can't make an assumption on whether
user-space will notify of this or not.
--
Best regards,
Javier Martinez Canillas
Core Platforms
Red Hat