Messages in this thread | | | Date | Fri, 21 Nov 2008 17:07:43 -0800 | From | Keika Kobayashi <> | Subject | Re: [PATCH 2/3] proc: Export statistics for softirq to /proc |
| |
> > +/* > > + * /proc/softirqs ... display the number of softirqs > > + */ > > +static int show_softirqs(struct seq_file *p, void *v) > > +{ > > + int i, j; > > + > > + seq_printf(p, " "); > > + for_each_online_cpu(i) > > + seq_printf(p, "CPU%-8d", i); > > + seq_printf(p, "\n"); > > + > > + for_each_softirq_nr(i) { > > + seq_printf(p, "%-10s", desc_array[i]); > > + for_each_online_cpu(j) > > + seq_printf(p, "%10u ", kstat_softirqs_cpu(i, j)); > > + seq_printf(p, "\n"); > > + } > > + return 0; > > +} > > This uses for_each_online_cpu(), but below we use for_each_possible_cpu(). > > Shouldn't we be consistent here so that at least the numbers will add > up to the same thing? > > Probably for_each_possible_cpu() is best - people might want to see how > many softirqs happened on a CPU which was recently offlined.
I understand this point. I'll fix it later.
There is same problem regarding /proc/interrupts. Should we change from for_each_online_cpu() to for_each_possible_cpu(), or is it too late?
| |