1. Patchew
  2. linux
  3. [PATCH v4 0/3] PM: runtime: Auto-cleanup macros for runtime PM
  4. View patch

[PATCH v4 2/3] PCI/sysfs: Use runtime PM guard macro for auto-cleanup

Rafael J. Wysocki posted 3 patches 2 months, 3 weeks ago
[PATCH v4 2/3] PCI/sysfs: Use runtime PM guard macro for auto-cleanup
Posted by Rafael J. Wysocki 2 months, 3 weeks ago 
From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Use the newly introduced pm_runtime_active_try guard to simplify
the code and add the proper error handling for PM runtime resume
errors.

Based on an earlier patch from Takashi Iwai <tiwai@suse.de> [1].

Link: https://patch.msgid.link/20250919163147.4743-3-tiwai@suse.de [1]
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---

v3 -> v4:
   * Use ACQUIRE()/ACQUIRE_ERR() (Jonathan)
   * Adjust subject and changelog
   * Take patch ownership (it's all different now)
   * Pick up Bjorn's ACK from v3 (Bjorn, please let me know if that's not OK)

v2 -> v3: No changes

v1 -> v2:
   * Adjust the name of the class to handle the disabled runtime PM case
     transparently (like the original code).

---
 drivers/pci/pci-sysfs.c |    5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

--- a/drivers/pci/pci-sysfs.c
+++ b/drivers/pci/pci-sysfs.c
@@ -1475,8 +1475,9 @@ static ssize_t reset_method_store(struct
 		return count;
 	}
 
-	pm_runtime_get_sync(dev);
-	struct device *pmdev __free(pm_runtime_put) = dev;
+	ACQUIRE(pm_runtime_active_try, pm)(dev);
+	if (ACQUIRE_ERR(pm_runtime_active_try, &pm))
+		return -ENXIO;
 
 	if (sysfs_streq(buf, "default")) {
 		pci_init_reset_methods(pdev);
Re: [PATCH v4 2/3] PCI/sysfs: Use runtime PM guard macro for auto-cleanup
Posted by Farhan Ali 1 month, 3 weeks ago 
On 9/26/2025 9:24 AM, Rafael J. Wysocki wrote:
> From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
>
> Use the newly introduced pm_runtime_active_try guard to simplify
> the code and add the proper error handling for PM runtime resume
> errors.
>
> Based on an earlier patch from Takashi Iwai <tiwai@suse.de> [1].
>
> Link: https://patch.msgid.link/20250919163147.4743-3-tiwai@suse.de [1]
> Acked-by: Bjorn Helgaas <bhelgaas@google.com>
> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> ---
>
> v3 -> v4:
>     * Use ACQUIRE()/ACQUIRE_ERR() (Jonathan)
>     * Adjust subject and changelog
>     * Take patch ownership (it's all different now)
>     * Pick up Bjorn's ACK from v3 (Bjorn, please let me know if that's not OK)
>
> v2 -> v3: No changes
>
> v1 -> v2:
>     * Adjust the name of the class to handle the disabled runtime PM case
>       transparently (like the original code).
>
> ---
>   drivers/pci/pci-sysfs.c |    5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
>
> --- a/drivers/pci/pci-sysfs.c
> +++ b/drivers/pci/pci-sysfs.c
> @@ -1475,8 +1475,9 @@ static ssize_t reset_method_store(struct
>   		return count;
>   	}
>   
> -	pm_runtime_get_sync(dev);
> -	struct device *pmdev __free(pm_runtime_put) = dev;
> +	ACQUIRE(pm_runtime_active_try, pm)(dev);
> +	if (ACQUIRE_ERR(pm_runtime_active_try, &pm))
> +		return -ENXIO;
>   
>   	if (sysfs_streq(buf, "default")) {
>   		pci_init_reset_methods(pdev);
>
>
Hi Rafael,

This patch breaks updating the 'reset_method' sysfs file on s390. If we 
try to update the reset_method, we are hitting the ENXIO error. eg:

echo 'bus' > /sys/bus/pci/devices/0007\:00\:10.1/reset_method
-bash: echo: write error: No such device or address

I don't think s390 does anything different in this path, so this could 
also impact other platforms? Changing this to something like this fixes it


diff --git a/drivers/pci/pci-sysfs.c b/drivers/pci/pci-sysfs.c
index 9d6f74bd95f8..d7fc0dc81c30 100644
--- a/drivers/pci/pci-sysfs.c
+++ b/drivers/pci/pci-sysfs.c
@@ -1517,8 +1517,8 @@ static ssize_t reset_method_store(struct device *dev,
                 return count;
         }

-       ACQUIRE(pm_runtime_active_try, pm)(dev);
-       if (ACQUIRE_ERR(pm_runtime_active_try, &pm))
+       ACQUIRE(pm_runtime_active, pm)(dev);
+       if (ACQUIRE_ERR(pm_runtime_active, &pm))
                 return -ENXIO;

This changes the logic to what it was previously which used 
pm_runtime_get_sync and pm_runtime_put. But I am not familiar with the 
PM runtime code, so not sure what would be the right fix here.

Thanks

Farhan
Re: [PATCH v4 2/3] PCI/sysfs: Use runtime PM guard macro for auto-cleanup
Posted by Rafael J. Wysocki 1 month, 3 weeks ago 
On Tue, Oct 21, 2025 at 12:07 AM Farhan Ali <alifm@linux.ibm.com> wrote:
>
>
> On 9/26/2025 9:24 AM, Rafael J. Wysocki wrote:
> > From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> >
> > Use the newly introduced pm_runtime_active_try guard to simplify
> > the code and add the proper error handling for PM runtime resume
> > errors.
> >
> > Based on an earlier patch from Takashi Iwai <tiwai@suse.de> [1].
> >
> > Link: https://patch.msgid.link/20250919163147.4743-3-tiwai@suse.de [1]
> > Acked-by: Bjorn Helgaas <bhelgaas@google.com>
> > Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > ---
> >
> > v3 -> v4:
> >     * Use ACQUIRE()/ACQUIRE_ERR() (Jonathan)
> >     * Adjust subject and changelog
> >     * Take patch ownership (it's all different now)
> >     * Pick up Bjorn's ACK from v3 (Bjorn, please let me know if that's not OK)
> >
> > v2 -> v3: No changes
> >
> > v1 -> v2:
> >     * Adjust the name of the class to handle the disabled runtime PM case
> >       transparently (like the original code).
> >
> > ---
> >   drivers/pci/pci-sysfs.c |    5 +++--
> >   1 file changed, 3 insertions(+), 2 deletions(-)
> >
> > --- a/drivers/pci/pci-sysfs.c
> > +++ b/drivers/pci/pci-sysfs.c
> > @@ -1475,8 +1475,9 @@ static ssize_t reset_method_store(struct
> >               return count;
> >       }
> >
> > -     pm_runtime_get_sync(dev);
> > -     struct device *pmdev __free(pm_runtime_put) = dev;
> > +     ACQUIRE(pm_runtime_active_try, pm)(dev);
> > +     if (ACQUIRE_ERR(pm_runtime_active_try, &pm))
> > +             return -ENXIO;
> >
> >       if (sysfs_streq(buf, "default")) {
> >               pci_init_reset_methods(pdev);
> >
> >
> Hi Rafael,
>
> This patch breaks updating the 'reset_method' sysfs file on s390. If we
> try to update the reset_method, we are hitting the ENXIO error. eg:
>
> echo 'bus' > /sys/bus/pci/devices/0007\:00\:10.1/reset_method
> -bash: echo: write error: No such device or address
>
> I don't think s390 does anything different in this path, so this could
> also impact other platforms? Changing this to something like this fixes it
>
>
> diff --git a/drivers/pci/pci-sysfs.c b/drivers/pci/pci-sysfs.c
> index 9d6f74bd95f8..d7fc0dc81c30 100644
> --- a/drivers/pci/pci-sysfs.c
> +++ b/drivers/pci/pci-sysfs.c
> @@ -1517,8 +1517,8 @@ static ssize_t reset_method_store(struct device *dev,
>                  return count;
>          }
>
> -       ACQUIRE(pm_runtime_active_try, pm)(dev);
> -       if (ACQUIRE_ERR(pm_runtime_active_try, &pm))
> +       ACQUIRE(pm_runtime_active, pm)(dev);
> +       if (ACQUIRE_ERR(pm_runtime_active, &pm))
>                  return -ENXIO;
>
> This changes the logic to what it was previously which used
> pm_runtime_get_sync and pm_runtime_put. But I am not familiar with the
> PM runtime code, so not sure what would be the right fix here.

Can you please check if this helps:

https://lore.kernel.org/linux-pm/5943878.DvuYhMxLoT@rafael.j.wysocki/

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.