> -----Original Message-----
> From: Greg Kroah-Hartman <[email protected]>
> Sent: Wednesday, March 2, 2022 4:58 PM
> To: Youngjin Jang <[email protected]>
> Cc: Rafael J. Wysocki <[email protected]>; Pavel Machek <[email protected]>;
> Len Brown <[email protected]>; Bjorn Helgaas <[email protected]>;
> [email protected]; [email protected]; linux-
> [email protected]; [email protected]; linux-
[email protected];
> [email protected]
> Subject: Re: [PATCH] PM: Add device name to suspend_report_result()
>
> On Wed, Mar 02, 2022 at 03:49:17PM +0900, Youngjin Jang wrote:
> > From: "yj84.jang" <[email protected]>
> >
> > currently, suspend_report_result() prints only function information.
> > If any driver uses common pm function, nobody knows who called failed
> > function exactly.
> >
> > So, device information is needed to recognize specific wrong driver.
> >
> > e.g.)
> > PM: dpm_run_callback(): pm_generic_suspend+0x0/0x48 returns 0
> > PM: dpm_run_callback(): platform_pm_suspend+0x0/0x68 returns 0 after
> > patch,
> > PM: dpm_run_callback(): pm_generic_suspend+0x0/0x48 (amba) returns 0
> > PM: dpm_run_callback(): platform_pm_suspend+0x0/0x68 (armv7-pmu)
> > returns 0
> >
> > Signed-off-by: yj84.jang <[email protected]>
> > ---
> > drivers/base/power/main.c | 10 +++++-----
> > drivers/pci/pci-driver.c | 14 +++++++-------
> > drivers/pnp/driver.c | 2 +-
> > drivers/usb/core/hcd-pci.c | 4 ++--
> > include/linux/pm.h | 8 ++++----
> > 5 files changed, 19 insertions(+), 19 deletions(-)
> >
> > diff --git a/drivers/base/power/main.c b/drivers/base/power/main.c
> > index 04ea92c..a762fe8 100644
> > --- a/drivers/base/power/main.c
> > +++ b/drivers/base/power/main.c
> > @@ -485,7 +485,7 @@ static int dpm_run_callback(pm_callback_t cb, struct
> device *dev,
> > trace_device_pm_callback_start(dev, info, state.event);
> > error = cb(dev);
> > trace_device_pm_callback_end(dev, error);
> > - suspend_report_result(cb, error);
> > + suspend_report_result(dev, cb, error);
> >
> > initcall_debug_report(dev, calltime, cb, error);
> >
> > @@ -1568,7 +1568,7 @@ static int legacy_suspend(struct device *dev,
> pm_message_t state,
> > trace_device_pm_callback_start(dev, info, state.event);
> > error = cb(dev, state);
> > trace_device_pm_callback_end(dev, error);
> > - suspend_report_result(cb, error);
> > + suspend_report_result(dev, cb, error);
> >
> > initcall_debug_report(dev, calltime, cb, error);
> >
> > @@ -1855,7 +1855,7 @@ static int device_prepare(struct device *dev,
> pm_message_t state)
> > device_unlock(dev);
> >
> > if (ret < 0) {
> > - suspend_report_result(callback, ret);
> > + suspend_report_result(dev, callback, ret);
> > pm_runtime_put(dev);
> > return ret;
> > }
> > @@ -1960,10 +1960,10 @@ int dpm_suspend_start(pm_message_t state) }
> > EXPORT_SYMBOL_GPL(dpm_suspend_start);
> >
> > -void __suspend_report_result(const char *function, void *fn, int ret)
> > +void __suspend_report_result(const char *function, struct device
> > +*dev, void *fn, int ret)
> > {
> > if (ret)
> > - pr_err("%s(): %pS returns %d\n", function, fn, ret);
> > + pr_err("%s(): %pS (%s) returns %d\n", function, fn,
> > +dev_driver_string(dev), ret);
>
> If you have a struct device, please use dev_err().
>
> thanks,
>
> greg k-h
Hello,
Thanks for your review.
I think dev_err() is nice option, but we can see a minor issue.
Prefix log "PM: " would be lost, If I use dev_err() in this context.
As you know, all logs in power management include "PM :" prefix.
So, I think pr_err() with detail information would be better than dev_err().
- PM: amba 1740000.etm: dpm_run_callback(): pm_generic_resume+0x0/0x48
returns 0
On Wed, Mar 02, 2022 at 08:00:14PM +0900, �念��/TV S/W Lab(VD)/Staff Engineer/�Ｚ���� wrote:
> > -----Original Message-----
> > From: Greg Kroah-Hartman <[email protected]>
> > Sent: Wednesday, March 2, 2022 4:58 PM
> > To: Youngjin Jang <[email protected]>
> > Cc: Rafael J. Wysocki <[email protected]>; Pavel Machek <[email protected]>;
> > Len Brown <[email protected]>; Bjorn Helgaas <[email protected]>;
> > [email protected]; [email protected]; linux-
> > [email protected]; [email protected]; linux-
> [email protected];
> > [email protected]
> > Subject: Re: [PATCH] PM: Add device name to suspend_report_result()
> >
> > On Wed, Mar 02, 2022 at 03:49:17PM +0900, Youngjin Jang wrote:
> > > From: "yj84.jang" <[email protected]>
> > >
> > > currently, suspend_report_result() prints only function information.
> > > If any driver uses common pm function, nobody knows who called failed
> > > function exactly.
> > >
> > > So, device information is needed to recognize specific wrong driver.
> > >
> > > e.g.)
> > > PM: dpm_run_callback(): pm_generic_suspend+0x0/0x48 returns 0
> > > PM: dpm_run_callback(): platform_pm_suspend+0x0/0x68 returns 0 after
> > > patch,
> > > PM: dpm_run_callback(): pm_generic_suspend+0x0/0x48 (amba) returns 0
> > > PM: dpm_run_callback(): platform_pm_suspend+0x0/0x68 (armv7-pmu)
> > > returns 0
> > >
> > > Signed-off-by: yj84.jang <[email protected]>
> > > ---
> > > drivers/base/power/main.c | 10 +++++-----
> > > drivers/pci/pci-driver.c | 14 +++++++-------
> > > drivers/pnp/driver.c | 2 +-
> > > drivers/usb/core/hcd-pci.c | 4 ++--
> > > include/linux/pm.h | 8 ++++----
> > > 5 files changed, 19 insertions(+), 19 deletions(-)
> > >
> > > diff --git a/drivers/base/power/main.c b/drivers/base/power/main.c
> > > index 04ea92c..a762fe8 100644
> > > --- a/drivers/base/power/main.c
> > > +++ b/drivers/base/power/main.c
> > > @@ -485,7 +485,7 @@ static int dpm_run_callback(pm_callback_t cb, struct
> > device *dev,
> > > trace_device_pm_callback_start(dev, info, state.event);
> > > error = cb(dev);
> > > trace_device_pm_callback_end(dev, error);
> > > - suspend_report_result(cb, error);
> > > + suspend_report_result(dev, cb, error);
> > >
> > > initcall_debug_report(dev, calltime, cb, error);
> > >
> > > @@ -1568,7 +1568,7 @@ static int legacy_suspend(struct device *dev,
> > pm_message_t state,
> > > trace_device_pm_callback_start(dev, info, state.event);
> > > error = cb(dev, state);
> > > trace_device_pm_callback_end(dev, error);
> > > - suspend_report_result(cb, error);
> > > + suspend_report_result(dev, cb, error);
> > >
> > > initcall_debug_report(dev, calltime, cb, error);
> > >
> > > @@ -1855,7 +1855,7 @@ static int device_prepare(struct device *dev,
> > pm_message_t state)
> > > device_unlock(dev);
> > >
> > > if (ret < 0) {
> > > - suspend_report_result(callback, ret);
> > > + suspend_report_result(dev, callback, ret);
> > > pm_runtime_put(dev);
> > > return ret;
> > > }
> > > @@ -1960,10 +1960,10 @@ int dpm_suspend_start(pm_message_t state) }
> > > EXPORT_SYMBOL_GPL(dpm_suspend_start);
> > >
> > > -void __suspend_report_result(const char *function, void *fn, int ret)
> > > +void __suspend_report_result(const char *function, struct device
> > > +*dev, void *fn, int ret)
> > > {
> > > if (ret)
> > > - pr_err("%s(): %pS returns %d\n", function, fn, ret);
> > > + pr_err("%s(): %pS (%s) returns %d\n", function, fn,
> > > +dev_driver_string(dev), ret);
> >
> > If you have a struct device, please use dev_err().
> >
> > thanks,
> >
> > greg k-h
>
> Hello,
> Thanks for your review.
>
> I think dev_err() is nice option, but we can see a minor issue.
> Prefix log "PM: " would be lost, If I use dev_err() in this context.
> As you know, all logs in power management include "PM :" prefix.
Why does that matter? Fix them all to use the struct device pointer and
then they will be properly unified with the rest of the kernel log
infrastructure.
thanks,
greg k-h
On Wed, Mar 02, 2022 at 03:52:51PM +0100, 'Greg Kroah-Hartman' wrote:
> On Wed, Mar 02, 2022 at 08:00:14PM +0900, �念��/TV S/W Lab(VD)/Staff Engineer/�Ｚ���� wrote:
> > > -----Original Message-----
> > > From: Greg Kroah-Hartman <[email protected]>
> > > Sent: Wednesday, March 2, 2022 4:58 PM
> > > To: Youngjin Jang <[email protected]>
> > > Cc: Rafael J. Wysocki <[email protected]>; Pavel Machek <[email protected]>;
> > > Len Brown <[email protected]>; Bjorn Helgaas <[email protected]>;
> > > [email protected]; [email protected]; linux-
> > > [email protected]; [email protected]; linux-
> > [email protected];
> > > [email protected]
> > > Subject: Re: [PATCH] PM: Add device name to suspend_report_result()
> > >
> > > On Wed, Mar 02, 2022 at 03:49:17PM +0900, Youngjin Jang wrote:
> > > > From: "yj84.jang" <[email protected]>
> > > >
> > > > currently, suspend_report_result() prints only function information.
> > > > If any driver uses common pm function, nobody knows who called failed
> > > > function exactly.
> > > >
> > > > So, device information is needed to recognize specific wrong driver.
> > > >
> > > > e.g.)
> > > > PM: dpm_run_callback(): pm_generic_suspend+0x0/0x48 returns 0
> > > > PM: dpm_run_callback(): platform_pm_suspend+0x0/0x68 returns 0 after
> > > > patch,
> > > > PM: dpm_run_callback(): pm_generic_suspend+0x0/0x48 (amba) returns 0
> > > > PM: dpm_run_callback(): platform_pm_suspend+0x0/0x68 (armv7-pmu)
> > > > returns 0
> > > > - pr_err("%s(): %pS returns %d\n", function, fn, ret);
> > > > + pr_err("%s(): %pS (%s) returns %d\n", function, fn,
> > > > +dev_driver_string(dev), ret);
> > >
> > > If you have a struct device, please use dev_err().
> >
> > I think dev_err() is nice option, but we can see a minor issue.
> > Prefix log "PM: " would be lost, If I use dev_err() in this context.
> > As you know, all logs in power management include "PM :" prefix.
>
> Why does that matter? Fix them all to use the struct device pointer and
> then they will be properly unified with the rest of the kernel log
> infrastructure.
You can #define dev_fmt if you need a prefix.
> -----Original Message-----
> From: Bjorn Helgaas <[email protected]>
> Sent: Thursday, March 3, 2022 5:16 AM
> To: 'Greg Kroah-Hartman' <[email protected]>
> Cc: �念��/TV S/W Lab(VD)/Staff Engineer/�Ｚ���� <[email protected]>;
> 'Rafael J. Wysocki' <[email protected]>; 'Pavel Machek' <[email protected]>;
> 'Len Brown' <[email protected]>; 'Bjorn Helgaas' <[email protected]>;
> [email protected]; [email protected]; linux-
> [email protected]; [email protected]; [email protected];
> [email protected]
> Subject: Re: [PATCH] PM: Add device name to suspend_report_result()
>
> On Wed, Mar 02, 2022 at 03:52:51PM +0100, 'Greg Kroah-Hartman' wrote:
> > On Wed, Mar 02, 2022 at 08:00:14PM +0900, 念 /TV S/W Lab(VD)/Staff
> Engineer/ Ｚ wrote:
> > > > -----Original Message-----
> > > > From: Greg Kroah-Hartman <[email protected]>
> > > > Sent: Wednesday, March 2, 2022 4:58 PM
> > > > To: Youngjin Jang <[email protected]>
> > > > Cc: Rafael J. Wysocki <[email protected]>; Pavel Machek
> > > > <[email protected]>; Len Brown <[email protected]>; Bjorn Helgaas
> > > > <[email protected]>; [email protected];
> > > > [email protected]; linux- [email protected];
> > > > [email protected]; linux-
> > > [email protected];
> > > > [email protected]
> > > > Subject: Re: [PATCH] PM: Add device name to
> > > > suspend_report_result()
> > > >
> > > > On Wed, Mar 02, 2022 at 03:49:17PM +0900, Youngjin Jang wrote:
> > > > > From: "yj84.jang" <[email protected]>
> > > > >
> > > > > currently, suspend_report_result() prints only function
> information.
> > > > > If any driver uses common pm function, nobody knows who called
> > > > > failed function exactly.
> > > > >
> > > > > So, device information is needed to recognize specific wrong
> driver.
> > > > >
> > > > > e.g.)
> > > > > PM: dpm_run_callback(): pm_generic_suspend+0x0/0x48 returns 0
> > > > > PM: dpm_run_callback(): platform_pm_suspend+0x0/0x68 returns 0
> > > > > after patch,
> > > > > PM: dpm_run_callback(): pm_generic_suspend+0x0/0x48 (amba)
> > > > > returns 0
> > > > > PM: dpm_run_callback(): platform_pm_suspend+0x0/0x68 (armv7-pmu)
> > > > > returns 0
>
> > > > > - pr_err("%s(): %pS returns %d\n", function, fn, ret);
> > > > > + pr_err("%s(): %pS (%s) returns %d\n", function, fn,
> > > > > +dev_driver_string(dev), ret);
> > > >
> > > > If you have a struct device, please use dev_err().
> > >
> > > I think dev_err() is nice option, but we can see a minor issue.
> > > Prefix log "PM: " would be lost, If I use dev_err() in this context.
> > > As you know, all logs in power management include "PM :" prefix.
> >
> > Why does that matter? Fix them all to use the struct device pointer
> > and then they will be properly unified with the rest of the kernel log
> > infrastructure.
>
> You can #define dev_fmt if you need a prefix.
I tested dev_fmt before, but I feel that not a good solution.
Because the readability is not so great than I expected.
I didn't want to break the PM logging rules.
Anyway, I got you guys opinion.
Let me try second patch with dev_err().
On Thu, Mar 03, 2022 at 07:56:37AM +0900, 장영진/TV S/W Lab(VD)/Staff Engineer/삼성전자 wrote:
> > -----Original Message-----
> > From: Bjorn Helgaas <[email protected]>
> > Sent: Thursday, March 3, 2022 5:16 AM
> > To: 'Greg Kroah-Hartman' <[email protected]>
> > Cc: �念��/TV S/W Lab(VD)/Staff Engineer/�Ｚ���� <[email protected]>;
> > 'Rafael J. Wysocki' <[email protected]>; 'Pavel Machek' <[email protected]>;
> > 'Len Brown' <[email protected]>; 'Bjorn Helgaas' <[email protected]>;
> > [email protected]; [email protected]; linux-
> > [email protected]; [email protected]; [email protected];
> > [email protected]
> > Subject: Re: [PATCH] PM: Add device name to suspend_report_result()
> >
> > On Wed, Mar 02, 2022 at 03:52:51PM +0100, 'Greg Kroah-Hartman' wrote:
> > > On Wed, Mar 02, 2022 at 08:00:14PM +0900, 念 /TV S/W Lab(VD)/Staff
> > Engineer/ Ｚ wrote:
> > > > > -----Original Message-----
> > > > > From: Greg Kroah-Hartman <[email protected]>
> > > > > Sent: Wednesday, March 2, 2022 4:58 PM
> > > > > To: Youngjin Jang <[email protected]>
> > > > > Cc: Rafael J. Wysocki <[email protected]>; Pavel Machek
> > > > > <[email protected]>; Len Brown <[email protected]>; Bjorn Helgaas
> > > > > <[email protected]>; [email protected];
> > > > > [email protected]; linux- [email protected];
> > > > > [email protected]; linux-
> > > > [email protected];
> > > > > [email protected]
> > > > > Subject: Re: [PATCH] PM: Add device name to
> > > > > suspend_report_result()
> > > > >
> > > > > On Wed, Mar 02, 2022 at 03:49:17PM +0900, Youngjin Jang wrote:
> > > > > > From: "yj84.jang" <[email protected]>
> > > > > > - pr_err("%s(): %pS returns %d\n", function, fn, ret);
> > > > > > + pr_err("%s(): %pS (%s) returns %d\n", function, fn,
> > > > > > +dev_driver_string(dev), ret);
> > > > >
> > > > > If you have a struct device, please use dev_err().
> > > >
> > > > I think dev_err() is nice option, but we can see a minor issue.
> > > > Prefix log "PM: " would be lost, If I use dev_err() in this context.
> > > > As you know, all logs in power management include "PM :" prefix.
> > >
> > > Why does that matter? Fix them all to use the struct device pointer
> > > and then they will be properly unified with the rest of the kernel log
> > > infrastructure.
> >
> > You can #define dev_fmt if you need a prefix.
>
> I tested dev_fmt before, but I feel that not a good solution.
> Because the readability is not so great than I expected.
> I didn't want to break the PM logging rules.
I didn't catch your meaning here. Some examples would probably help.
The patch above is from __suspend_report_result() in
drivers/base/power/main.c. That file already defines both pr_fmt and
dev_fmt to be "PM: ", so I would expect dev_err() output to already
include "PM: ".