usb: dwc3: mark PM function as __maybe_unused
Subsystems (2)
PATCH 1/1 usb: dwc3: mark PM function as __maybe_unused
Starting with v4.8-rc5, we get a warning about the dwc3_pci_pm_dummy function when CONFIG_PM_SLEEP is disabled:
drivers/usb/dwc3/dwc3-pci.c:253:12: warning: 'dwc3_pci_pm_dummy' defined but not used
Using __maybe_unused instead of #ifdef lets the compiler figure out whether it should drop the definitions silently and avoids this warning.
--- drivers/usb/dwc3/dwc3-pci.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-)
After we fixed all other warnings reported by kernelci.org, this is now the only thing we get a warning for in the configurations we build, so it would be good to get this fixed before v4.8 to get a clean build for the first time in a release.
drivers/usb/dwc3/dwc3-pci.c +3 -5
@@ -234,8 +234,7 @@ static const struct pci_device_id dwc3_pci_id_table[] = { };MODULE_DEVICE_TABLE (pci, dwc3_pci_id_table); -#ifdef CONFIG_PM -static int dwc3_pci_runtime_suspend (struct device *dev) +static int __maybe_unused dwc3_pci_runtime_suspend (struct device *dev) {if (device_run_wake (dev))return 0 ;
@@ -243,14 +242,14 @@ static int dwc3_pci_runtime_suspend(struct device *dev)return -EBUSY ; } -static int dwc3_pci_runtime_resume (struct device *dev) +static int __maybe_unused dwc3_pci_runtime_resume (struct device *dev) {struct platform_device *dwc3 =dev_get_drvdata (dev);return pm_runtime_get (&dwc3->dev); } -static int dwc3_pci_pm_dummy (struct device *dev) +static int __maybe_unused dwc3_pci_pm_dummy (struct device *dev) {/* * There's nothing to do here. No, seriously. Everything is either taken
@@ -262,7 +261,6 @@ static int dwc3_pci_pm_dummy(struct device *dev)*/ return 0 ; } -#endif /* CONFIG_PM */ static struct dev_pm_ops dwc3_pci_dev_pm_ops = {SET_SYSTEM_SLEEP_PM_OPS (dwc3_pci_pm_dummy, dwc3_pci_pm_dummy)
Signature · 1 line
2.9.0
Hi,
Quoted 51 lines · Arnd Bergmann
Starting with v4.8-rc5, we get a warning about the
dwc3_pci_pm_dummyfunction whenCONFIG_PM_SLEEPis disabled:
drivers/usb/dwc3/dwc3-pci.c:253:12: warning: 'dwc3_pci_pm_dummy' defined but not usedUsing
__maybe_unusedinstead of #ifdef lets the compiler figure out whether it should drop the definitions silently and avoids this warning.Signed-off-by Arnd Bergmann <[email protected]>Fixes f6c274e11e3b ("usb: dwc3: pci: runtime_resume child device")--- drivers/usb/dwc3/dwc3-pci.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-)After we fixed all other warnings reported by kernelci.org, this is now the only thing we get a warning for in the configurations we build, so it would be good to get this fixed before v4.8 to get a clean build for the first time in a release.
drivers/usb/dwc3/dwc3-pci.c +3
-4@@ -234,8 +234,7 @@ static const struct pci_device_id dwc3_pci_id_table[] = { };MODULE_DEVICE_TABLE (pci, dwc3_pci_id_table); -#ifdef CONFIG_PM -static int dwc3_pci_runtime_suspend (struct device *dev) +static int __maybe_unused dwc3_pci_runtime_suspend (struct device *dev) {if (device_run_wake (dev))return 0 ;@@ -243,14 +242,14 @@ static int dwc3_pci_runtime_suspend(struct device *dev)return -EBUSY ; } -static int dwc3_pci_runtime_resume (struct device *dev) +static int __maybe_unused dwc3_pci_runtime_resume (struct device *dev) {struct platform_device *dwc3 =dev_get_drvdata (dev);return pm_runtime_get (&dwc3->dev); } -static int dwc3_pci_pm_dummy (struct device *dev) +static int __maybe_unused dwc3_pci_pm_dummy (struct device *dev) {/* * There's nothing to do here. No, seriously. Everything is either taken
drivers/usb/dwc3/dwc3-pci.c static int dwc3_pci_pm_dummy(struct device *dev) · line 265
Quoted 4 lines
drivers/usb/dwc3/dwc3-pci.c +0
-1@@ -262,7 +261,6 @@ static int dwc3_pci_pm_dummy(struct device *dev)*/ return 0 ; } -#endif /* CONFIG_PM */
seems like it would've been enough to define dwc3_pci_pm_dummy() if CONFIG_PM_SLEEP. Right? IOW:
drivers/usb/dwc3/dwc3-pci.c +3 -1
@@ -249,7 +249,9 @@ static int dwc3_pci_runtime_resume(struct device *dev)return pm_runtime_get (&dwc3->dev); } +#endif /* CONFIG_PM */ +#ifdef CONFIG_PM_SLEEP static int dwc3_pci_pm_dummy (struct device *dev) {/*
@@ -262,7 +264,7 @@ static int dwc3_pci_pm_dummy(struct device *dev)*/ return 0 ; } -#endif /* CONFIG_PM */ +#endif /* CONFIG_PM_SLEEP */ static struct dev_pm_ops dwc3_pci_dev_pm_ops = {SET_SYSTEM_SLEEP_PM_OPS (dwc3_pci_pm_dummy, dwc3_pci_pm_dummy)
Can you check if this is enough for you? Works just fine here.
Signature · 1 line
balbi
Arnd Bergmann
Acked-byQuoted 28 lines · CEST Felipe Balbi
seems like it would've been enough to define
dwc3_pci_pm_dummy()ifCONFIG_PM_SLEEP. Right? IOW:drivers/usb/dwc3/dwc3-pci.c +3
-1@@ -249,7 +249,9 @@ static int dwc3_pci_runtime_resume(struct device *dev)return pm_runtime_get (&dwc3->dev); } +#endif /* CONFIG_PM */ +#ifdef CONFIG_PM_SLEEP static int dwc3_pci_pm_dummy (struct device *dev) {/* @@ -262,7 +264,7 @@ static int dwc3_pci_pm_dummy(struct device *dev)*/ return 0 ; } -#endif /* CONFIG_PM */ +#endif /* CONFIG_PM_SLEEP */ static struct dev_pm_ops dwc3_pci_dev_pm_ops = {SET_SYSTEM_SLEEP_PM_OPS (dwc3_pci_pm_dummy, dwc3_pci_pm_dummy)
Quoted 2 lines
Can you check if this is enough for you? Works just fine here.
I'm sure this solves the problem too
I usually prefer the __maybe_unused annotation over the #ifdef because it's much harder to get wrong, but I'm really only interested in killing that last warning in the v4.8 kernel build at the moment ;-)
I am getting a bit annoyed with seeing at least half a dozen such bugs every week (in linux-next) and I have an idea for how to replace SET_SYSTEM_SLEEP_PM_OPS() etc with something that doesn't need any #ifdef of __maybe_unused, but that is a bigger chunk of work that will require careful planning.
Arnd