On 02/08/23(Wed) 18:27, Claudio Jeker wrote:
> On Wed, Aug 02, 2023 at 10:15:20AM -0500, Scott Cheloha wrote:
> > Now that the proc0 wakeup(9) is gone we can retry the other part of
> > the uvm_meter() patch.
> >
> > uvm_meter() is meant to run every 5 seconds, but for historical
> > reasons it is called from schedcpu() and it is scheduled against the
> > UTC clock. schedcpu() and uvm_meter() have different periods, so
> > uvm_meter() ought to be a separate timeout. uvm_meter() is started
> > alongside schedcpu() so the two will still run in sync.
> >
> > v1: https://marc.info/?l=openbsd-tech&m=168710929409153&w=2
> >
> > ok?
>
> I would refer if uvm_meter is killed and the load calcualtion moved to the
> scheduler.
Me too.
> > Index: sys/uvm/uvm_meter.c
> > ===================================================================
> > RCS file: /cvs/src/sys/uvm/uvm_meter.c,v
> > retrieving revision 1.46
> > diff -u -p -r1.46 uvm_meter.c
> > --- sys/uvm/uvm_meter.c 2 Aug 2023 13:54:45 -0000 1.46
> > +++ sys/uvm/uvm_meter.c 2 Aug 2023 15:13:49 -0000
> > @@ -85,10 +85,12 @@ void uvmexp_read(struct uvmexp *);
> > * uvm_meter: calculate load average
> > */
> > void
> > -uvm_meter(void)
> > +uvm_meter(void *unused)
> > {
> > - if ((gettime() % 5) == 0)
> > - uvm_loadav(&averunnable);
> > + static struct timeout to = TIMEOUT_INITIALIZER(uvm_meter, NULL);
> > +
> > + timeout_add_sec(&to, 5);
> > + uvm_loadav(&averunnable);
> > }
> >
> > /*
> > Index: sys/uvm/uvm_extern.h
> > ===================================================================
> > RCS file: /cvs/src/sys/uvm/uvm_extern.h,v
> > retrieving revision 1.170
> > diff -u -p -r1.170 uvm_extern.h
> > --- sys/uvm/uvm_extern.h 21 Jun 2023 21:16:21 -0000 1.170
> > +++ sys/uvm/uvm_extern.h 2 Aug 2023 15:13:49 -0000
> > @@ -414,7 +414,7 @@ void uvmspace_free(struct vmspace *);
> > struct vmspace *uvmspace_share(struct process *);
> > int uvm_share(vm_map_t, vaddr_t, vm_prot_t,
> > vm_map_t, vaddr_t, vsize_t);
> > -void uvm_meter(void);
> > +void uvm_meter(void *);
> > int uvm_sysctl(int *, u_int, void *, size_t *,
> > void *, size_t, struct proc *);
> > struct vm_page *uvm_pagealloc(struct uvm_object *,
> > Index: sys/kern/sched_bsd.c
> > ===================================================================
> > RCS file: /cvs/src/sys/kern/sched_bsd.c,v
> > retrieving revision 1.78
> > diff -u -p -r1.78 sched_bsd.c
> > --- sys/kern/sched_bsd.c 25 Jul 2023 18:16:19 -0000 1.78
> > +++ sys/kern/sched_bsd.c 2 Aug 2023 15:13:50 -0000
> > @@ -235,7 +235,6 @@ schedcpu(void *arg)
> > }
> > SCHED_UNLOCK(s);
> > }
> > - uvm_meter();
> > wakeup(&lbolt);
> > timeout_add_sec(to, 1);
> > }
> > @@ -688,6 +687,7 @@ scheduler_start(void)
> >
> > rrticks_init = hz / 10;
> > schedcpu(&schedcpu_to);
> > + uvm_meter(NULL);
> >
> > #ifndef SMALL_KERNEL
> > if (perfpolicy == PERFPOL_AUTO)
> > Index: share/man/man9/uvm_init.9
> > ===================================================================
> > RCS file: /cvs/src/share/man/man9/uvm_init.9,v
> > retrieving revision 1.7
> > diff -u -p -r1.7 uvm_init.9
> > --- share/man/man9/uvm_init.9 21 Jun 2023 21:16:21 -0000 1.7
> > +++ share/man/man9/uvm_init.9 2 Aug 2023 15:13:50 -0000
> > @@ -168,7 +168,7 @@ argument is ignored.
> > .Ft void
> > .Fn uvm_kernacc "caddr_t addr" "size_t len" "int rw"
> > .Ft void
> > -.Fn uvm_meter
> > +.Fn uvm_meter "void *arg"
> > .Ft int
> > .Fn uvm_sysctl "int *name" "u_int namelen" "void *oldp" "size_t *oldlenp"
> > "void *newp " "size_t newlen" "struct proc *p"
> > .Ft int
> > @@ -212,7 +212,7 @@ access, in the kernel address space.
> > .Pp
> > The
> > .Fn uvm_meter
> > -function calculates the load average and wakes up the swapper if necessary.
> > +timeout updates system load averages every five seconds.
> > .Pp
> > The
> > .Fn uvm_sysctl
>
> --
> :wq Claudio
>
