Hi all
It's the patch I want to fix this problem. But I'm not sure I'm right or not. So this
is just for review. If it's right, I'll send the patch formally.
Test:
mdadm -CR /dev/md0 -l1 -n2 /dev/loop0 /dev/loop1 --size=1G (loop0 and loop1 are 8GB)
Result:
rdev->sectors(kernel space) will be 8GB, it's wrong. It should be 1GB-superblock-bitmap-badblock.
During creating super1.0/1.2 raid device, kernel space calls super_1_load, it sets rdev->sectors
with sb->data_size. If the raid device is created with specifying --size, it doesn't use total
space of disk. But rdev->sectors is still set with the total space of disk(decreasing superblock/bitmap/badblock)
So sb->data_size shoule be the raid member size rather than the total disk size.
Signed-off-by: Xiao Ni <xni@xxxxxxxxxx>
---
super1.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/super1.c b/super1.c
index e0d80be..d5d022b 100644
--- a/super1.c
+++ b/super1.c
@@ -2029,7 +2029,7 @@ static int write_init_super1(struct supertype *st)
sb->super_offset = __cpu_to_le64(sb_offset);
if (sb_offset < array_size + bm_space)
bm_space = sb_offset - array_size;
- sb->data_size = __cpu_to_le64(sb_offset - bm_space);
+ sb->data_size = __cpu_to_le64(array_size);
if (bm_space >= 8) {
sb->bblog_size = __cpu_to_le16(8);
sb->bblog_offset = __cpu_to_le32((unsigned)-8);
@@ -2043,7 +2043,7 @@ static int write_init_super1(struct supertype *st)
data_offset = sb_offset + 16;
sb->data_offset = __cpu_to_le64(data_offset);
- sb->data_size = __cpu_to_le64(dsize - data_offset);
+ sb->data_size = __cpu_to_le64(array_size);
if (data_offset >= sb_offset+bm_offset+bm_space+8) {
sb->bblog_size = __cpu_to_le16(8);
sb->bblog_offset = __cpu_to_le32(bm_offset +
--
2.7.5
