Re: [PATCH] 2.6.0-test4 -- add context switch counters

[William Lee Irwin III <wli@holomorphy.com>]

On Wed, Aug 27, 2003 at 10:57:44AM +1000, Peter Chubb wrote:
> Currently, the context switch counters reported by getrusage() are
> always zero.  The appended patch adds fields to struct task_struct to
> count context switches, and adds code to do the counting.
> The patch adds 4 longs to struct task struct, and a single addition to
> the fast path in schedule().

Thanks, this will be useful. We're still missing a fair number of them:

struct  rusage {
        struct timeval ru_utime;        /* user time used */
        struct timeval ru_stime;        /* system time used */
        long    ru_maxrss;              /* maximum resident set size */
        long    ru_ixrss;               /* integral shared memory size */
        long    ru_idrss;               /* integral unshared data size */
        long    ru_isrss;               /* integral unshared stack size */
        long    ru_minflt;              /* page reclaims */
        long    ru_majflt;              /* page faults */
        long    ru_nswap;               /* swaps */
        long    ru_inblock;             /* block input operations */
        long    ru_oublock;             /* block output operations */
        long    ru_msgsnd;              /* messages sent */
        long    ru_msgrcv;              /* messages received */
        long    ru_nsignals;            /* signals received */
        long    ru_nvcsw;               /* voluntary context switches */
        long    ru_nivcsw;              /* involuntary " */
};

...

                case RUSAGE_SELF:
                        jiffies_to_timeval(p->utime, &r.ru_utime);
                        jiffies_to_timeval(p->stime, &r.ru_stime);
                        r.ru_minflt = p->min_flt;
                        r.ru_majflt = p->maj_flt;
                        r.ru_nswap = p->nswap;
                        break;

and we're worse off yet: "FIXME! Get the fault counts properly!" ...
AFAICT literally the only useful number here is utime/stime.


-- wli

P.S.:
The stuff in /proc/$PID/statm isn't a big deal; I've got full 2.4.x
semantics (modulo the VSZ correction) with fully O(1) algorithmic
overhead in some patch originally by bcrl I forward ported somewhere.