| Message ID | 3a20bf20-247d-1242-dcd0-aef1bbc6e308@infradead.org (mailing list archive) |
|---|---|
| State | Not Applicable |
| Headers | show |
| Series | [-next] cpufreq: add stub for get_cpu_idle_time() to fix scsi/lpfc driver build | expand |
On 06-07-20, 09:44, Randy Dunlap wrote: > From: Randy Dunlap <rdunlap@infradead.org> > > To fix a build error in drivers/scsi/lpfc/lpfc_sli.c when > CONFIG_CPU_FREQ is not set/enabled, add a stub function for > get_cpu_idle_time() in <linux/cpufreq.h>. > > ../drivers/scsi/lpfc/lpfc_sli.c: In function ‘lpfc_init_idle_stat_hb’: > ../drivers/scsi/lpfc/lpfc_sli.c:7330:26: error: implicit declaration of function ‘get_cpu_idle_time’; did you mean ‘set_cpu_active’? [-Werror=implicit-function-declaration] And why is lpfc_sli.c using a cpufreq (supposedly internal, i.e. for cpufreq related parts) routine ? I think if you really need this, then it should be moved to a better common place and let everyone use it. I also see that drivers/macintosh/rack-meter.c has its own implementation for this. > idle_stat->prev_idle = get_cpu_idle_time(i, &wall, 1); > > Signed-off-by: Randy Dunlap <rdunlap@infradead.org> > Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net> > Cc: Viresh Kumar <viresh.kumar@linaro.org> > Cc: linux-pm@vger.kernel.org > Cc: James Smart <james.smart@broadcom.com> > Cc: Dick Kennedy <dick.kennedy@broadcom.com> > Cc: linux-scsi@vger.kernel.org > --- > include/linux/cpufreq.h | 4 ++++ > 1 file changed, 4 insertions(+) > > --- linux-next-20200706.orig/include/linux/cpufreq.h > +++ linux-next-20200706/include/linux/cpufreq.h > @@ -237,6 +237,10 @@ static inline unsigned int cpufreq_get_h > { > return 0; > } > +static inline u64 get_cpu_idle_time(unsigned int cpu, u64 *wall, int io_busy) > +{ > + return 0; > +} > static inline void disable_cpufreq(void) { } > #endif >
On 7/6/20 8:09 PM, Viresh Kumar wrote: > On 06-07-20, 09:44, Randy Dunlap wrote: >> From: Randy Dunlap <rdunlap@infradead.org> >> >> To fix a build error in drivers/scsi/lpfc/lpfc_sli.c when >> CONFIG_CPU_FREQ is not set/enabled, add a stub function for >> get_cpu_idle_time() in <linux/cpufreq.h>. >> >> ../drivers/scsi/lpfc/lpfc_sli.c: In function ‘lpfc_init_idle_stat_hb’: >> ../drivers/scsi/lpfc/lpfc_sli.c:7330:26: error: implicit declaration of function ‘get_cpu_idle_time’; did you mean ‘set_cpu_active’? [-Werror=implicit-function-declaration] > > And why is lpfc_sli.c using a cpufreq (supposedly internal, i.e. for > cpufreq related parts) routine ? I think if you really need this, then > it should be moved to a better common place and let everyone use it. Viresh: James Smart replied in another email thread with lpfc explanation for using get_cpu_idle_time(). Please see https://lore.kernel.org/linux-scsi/7ae1c7e3-ce8d-836b-1ae7-d4d00bd8f95c@broadcom.com/T/#md083717b1ff3a428c3b419dcc6d11cd03fee44c7 for this text: ""The driver is using cpu utilization in order to choose between softirq or work queues in handling an interrupt. Less-utilized, softirq is used. higher utilized, work queue is used. The utilization is checked periodically via a heartbeat. "" > I also see that drivers/macintosh/rack-meter.c has its own > implementation for this. > >> idle_stat->prev_idle = get_cpu_idle_time(i, &wall, 1); >> >> Signed-off-by: Randy Dunlap <rdunlap@infradead.org> >> Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net> >> Cc: Viresh Kumar <viresh.kumar@linaro.org> >> Cc: linux-pm@vger.kernel.org >> Cc: James Smart <james.smart@broadcom.com> >> Cc: Dick Kennedy <dick.kennedy@broadcom.com> >> Cc: linux-scsi@vger.kernel.org >> --- >> include/linux/cpufreq.h | 4 ++++ >> 1 file changed, 4 insertions(+) >> >> --- linux-next-20200706.orig/include/linux/cpufreq.h >> +++ linux-next-20200706/include/linux/cpufreq.h >> @@ -237,6 +237,10 @@ static inline unsigned int cpufreq_get_h >> { >> return 0; >> } >> +static inline u64 get_cpu_idle_time(unsigned int cpu, u64 *wall, int io_busy) >> +{ >> + return 0; >> +} >> static inline void disable_cpufreq(void) { } >> #endif >> >
On 07-07-20, 11:18, Randy Dunlap wrote: > Viresh: > > James Smart replied in another email thread with lpfc explanation for using > get_cpu_idle_time(). Please see > https://lore.kernel.org/linux-scsi/7ae1c7e3-ce8d-836b-1ae7-d4d00bd8f95c@broadcom.com/T/#md083717b1ff3a428c3b419dcc6d11cd03fee44c7 > > for this text: > ""The driver is using cpu utilization in order to choose between softirq or work queues in handling an interrupt. Less-utilized, softirq is used. higher utilized, work queue is used. The utilization is checked periodically via a heartbeat. "" To avoid ping pong of messages, I have instead replied to that email directly now. Lets follow it there.
--- linux-next-20200706.orig/include/linux/cpufreq.h +++ linux-next-20200706/include/linux/cpufreq.h @@ -237,6 +237,10 @@ static inline unsigned int cpufreq_get_h { return 0; } +static inline u64 get_cpu_idle_time(unsigned int cpu, u64 *wall, int io_busy) +{ + return 0; +} static inline void disable_cpufreq(void) { } #endif