We used the full device/file size for the zipf or pareto init,
this breaks if size= is specifically given. This causes fio
to generate offsets outside the specified size= range, and fio
will then just exit.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
range_size = min(td->o.min_bs[DDIR_READ], td->o.min_bs[DDIR_WRITE]);
- nranges = (f->real_file_size + range_size - 1) / range_size;
+ nranges = (f->io_size + range_size - 1) / range_size;
seed = jhash(f->file_name, strlen(f->file_name), 0) * td->thread_number;
if (!td->o.rand_repeatable)