On Wed, Sep 12, 2018 at 04:28:28PM -0700, Andrew Morton wrote: > On Tue, 28 Aug 2018 13:22:53 -0400 Johannes Weiner <hannes@xxxxxxxxxxx> wrote: > > > There are several definitions of those functions/macros in places that > > mess with fixed-point load averages. Provide an official version. > > missed blk-iolatency.c for some reason? Ah, that callsite came in with this merge window. Thanks for the fixup. > --- a/block/blk-iolatency.c~sched-loadavg-consolidate-load_int-load_frac-calc_load-fix > +++ a/block/blk-iolatency.c > @@ -512,7 +512,7 @@ static void iolatency_check_latencies(st > exp_idx = min_t(int, BLKIOLATENCY_NR_EXP_FACTORS - 1, > div64_u64(iolat->cur_win_nsec, > BLKIOLATENCY_EXP_BUCKET_SIZE)); > - CALC_LOAD(iolat->lat_avg, iolatency_exp_factors[exp_idx], stat.mean); > + calc_load(iolat->lat_avg, iolatency_exp_factors[exp_idx], stat.mean); The macro used to modify the avg parameter in place, but with the function we need an explicit assignment to update the variable: Signed-off-by: Johannes Weiner <hannes@xxxxxxxxxxx> --- diff --git a/block/blk-iolatency.c b/block/blk-iolatency.c index 335c22317757..8793f1344e11 100644 --- a/block/blk-iolatency.c +++ b/block/blk-iolatency.c @@ -512,7 +512,8 @@ static void iolatency_check_latencies(struct iolatency_grp *iolat, u64 now) exp_idx = min_t(int, BLKIOLATENCY_NR_EXP_FACTORS - 1, div64_u64(iolat->cur_win_nsec, BLKIOLATENCY_EXP_BUCKET_SIZE)); - calc_load(iolat->lat_avg, iolatency_exp_factors[exp_idx], stat.mean); + iolat->lat_avg = calc_load(iolat->lat_avg, + iolatency_exp_factors[exp_idx], stat.mean); /* Everything is ok and we don't need to adjust the scale. */ if (stat.mean <= iolat->min_lat_nsec &&
