Re: /proc/stat weirdness

[Denis Vlasenko]

On 22 April 2002 04:18, Mark Hahn wrote:
> why not figure out if theres a code-based reason?
> just add some code in the kernel to check for the case,
> and dump some extra data.

Hehe. I caught it (imho, not tested).
jiffies is saved to local variable before user,nice,system is calculated,
what if delta(user+nice+system) was == delta(jiffies)
[i.e. no idle ticks since last readout by e.g. top]
and a jiffy just ended and got accounted into one of user,nice,system?
We'll get delta(user+nice+system) > delta(jiffies),
delta(idle)=delta(jiffies)-delta(user+nice+system) - negative!

File: proc_misc.c
....
static int kstat_read_proc(char *page, char **start, off_t off,
                                 int count, int *eof, void *data)
{
        int i, len;
        extern unsigned long total_forks;
        unsigned long jif = jiffies; <*********** jiffies saved
        unsigned int sum = 0, user = 0, nice = 0, system = 0;
        int major, disk;

        for (i = 0 ; i < smp_num_cpus; i++) {
                int cpu = cpu_logical_map(i), j;

                user += kstat.per_cpu_user[cpu]; <***
                nice += kstat.per_cpu_nice[cpu]; <*** accounting
                system += kstat.per_cpu_system[cpu]; <***
#if !defined(CONFIG_ARCH_S390)
                for (j = 0 ; j < NR_IRQS ; j++)
                        sum += kstat.irqs[cpu][j];
#endif
        }

        len = sprintf(page, "cpu  %u %u %u %lu\n", user, nice, system,
                      jif * smp_num_cpus - (user + nice + system));
--
vda