On 2022-09-19 02:41, Xiao Ni wrote: > When creating a raid device without specifying data offset, > dv->data_offset is always 1 (INVALID_SECTORS). > If users specify data offset, we need to use dv->data_offset. If users > don't specify it, we need to use > st->data_offset. > > For super1, in the function write_init_super1, before writing > superblock to disk, it checks data_offset. If it's > INVALID_SECTORS, it will use st->data_offset (the default value of > specific superblock) > > And for s->size, if the raid device is a raid0, s->size is 0 here. I > see you and Mariusz talked about the raid0 > zone. If the raid0 is created with disks of different size, it can > have more than one zone. > e.g. disk1(10G) disk2(20G) disk3(30G) > disk1 disk2 disk3 > 10G 10G 10G (zone0) > 10G 10G (zone1) (The zone1 is behind zone0) > 10G (zone2) > > If the user doesn't specify size, it will use all space of the disk > and create zones and we can discard > the whole disk in this situation. Thanks, this was really useful information. I've incorporated it into v3 which I'll be sending out in a bit. Logan
