On Friday June 20, chris@xxxxxxxxxxxx wrote:
> Chris Webb <chris@xxxxxxxxxxxx> writes:
>
> > I'd like to support metadata types 0.90 and 1.0 too, relocating the
> > superblock. I think it'd be sufficient to write a new one at the end of
> > the device then update rdev->sb_offset. I'll try this next week.
>
> I've had a quick stab at this before I leave along with fixing a couple of
> other oversights. (I just return EBUSY if there's a bitmap present in 0.90
> and 1.0.) It appears to work correctly for 0.90 but generates a corrupt
> superblock for 1.0, which I'll need to fix on Monday.
Thanks for doing this!!!
- I am perfectly happy with getting -EBUSY if there is an internal
bitmap. It is quite easy to turn off the bitmap, resize the
devices, then turn it on again, and there is little cost in
doing this.
- I would much prefer using "sector" numbers rather than "K" numbers
in any new code. I'd eventually like to alway use sector numbers
internally, but that's a lower priority. So if you could change
the rdev_size_change methods to convert to sectors and then use
that I'd appreciate it.
- I think you really should call md_super_wait after md_super_write.
You really don't want the device to appear bigger until the
new metadata really is safe of disk.
- I think you need some protection to make sure that size doesn't get
set below my_mddev->size while the array is active. That would be
bad.
What I'd really like is for md to get a call-back when the device size
changes, so that the metadata can be relocated immediately. However
that is a little way off, and I think this is a useful thing to have
now.
Thanks,
NeilBrown
>
> Cheers,
>
> Chris.
> --- linux-2.6.24.4/drivers/md/md.c 2008-03-24 18:49:18.000000000 +0000
> +++ linux-2.6.24.4-cdwmd/drivers/md/md.c 2008-06-20 15:02:01.000000000 +0100
> @@ -652,11 +652,12 @@ static unsigned int calc_sb_csum(mdp_sup
> */
>
> struct super_type {
> - char *name;
> - struct module *owner;
> - int (*load_super)(mdk_rdev_t *rdev, mdk_rdev_t *refdev, int minor_version);
> - int (*validate_super)(mddev_t *mddev, mdk_rdev_t *rdev);
> - void (*sync_super)(mddev_t *mddev, mdk_rdev_t *rdev);
> + char *name;
> + struct module *owner;
> + int (*load_super)(mdk_rdev_t *rdev, mdk_rdev_t *refdev, int minor_version);
> + int (*validate_super)(mddev_t *mddev, mdk_rdev_t *rdev);
> + void (*sync_super)(mddev_t *mddev, mdk_rdev_t *rdev);
> + unsigned long long (*rdev_size_change)(mdk_rdev_t *rdev, unsigned long long size);
> };
>
> /*
> @@ -994,6 +995,23 @@ static void super_90_sync(mddev_t *mddev
> }
>
> /*
> + * rdev_size_change for 0.90.0
> + */
> +static unsigned long long
> +super_90_rdev_size_change(mdk_rdev_t *rdev, unsigned long long size)
> +{
> + if (rdev->mddev->bitmap_offset)
> + return 0; /* can't move bitmap */
> + rdev->sb_offset = calc_dev_sboffset(rdev->bdev);
> + if (!size || size > rdev->sb_offset)
> + size = rdev->sb_offset;
> + md_super_write(rdev->mddev, rdev, rdev->sb_offset << 1, rdev->sb_size,
> + rdev->sb_page);
> + return size;
> +}
> +
> +
> +/*
> * version 1 superblock
> */
>
> @@ -1310,21 +1328,49 @@ static void super_1_sync(mddev_t *mddev,
> sb->sb_csum = calc_sb_1_csum(sb);
> }
>
> +static unsigned long long
> +super_1_rdev_size_change(mdk_rdev_t *rdev, unsigned long long size)
> +{
> + struct mdp_superblock_1 *sb;
> + sb = (struct mdp_superblock_1 *) page_address(rdev->sb_page);
> + if (rdev->sb_offset < rdev->data_offset/2) {
> + /* minor versions 1 and 2; superblock before data */
> + unsigned long long max_size;
> + max_size = (rdev->bdev->bd_inode->i_size >> 10) - rdev->data_offset/2;
> + if (!size || size > max_size)
> + size = max_size;
> + } else {
> + /* minor version 0; superblock after data */
> + if (rdev->mddev->bitmap_offset)
> + return 0; /* can't move bitmap */
> + rdev->sb_offset = (rdev->bdev->bd_inode->i_size >> 10) - 8;
> + rdev->sb_offset &= ~(sector_t)(4 - 1);
> + if (!size || size > rdev->sb_offset - rdev->data_offset/2)
> + size = rdev->sb_offset - rdev->data_offset/2;
> + }
> + sb->data_size = cpu_to_le64(size*2);
> + sb->sb_csum = calc_sb_1_csum(sb);
> + md_super_write(rdev->mddev, rdev, rdev->sb_offset << 1, rdev->sb_size,
> + rdev->sb_page);
> + return size;
> +}
>
> static struct super_type super_types[] = {
> [0] = {
> .name = "0.90.0",
> .owner = THIS_MODULE,
> - .load_super = super_90_load,
> - .validate_super = super_90_validate,
> - .sync_super = super_90_sync,
> + .load_super = super_90_load,
> + .validate_super = super_90_validate,
> + .sync_super = super_90_sync,
> + .rdev_size_change = super_90_rdev_size_change,
> },
> [1] = {
> .name = "md-1",
> .owner = THIS_MODULE,
> - .load_super = super_1_load,
> - .validate_super = super_1_validate,
> - .sync_super = super_1_sync,
> + .load_super = super_1_load,
> + .validate_super = super_1_validate,
> + .sync_super = super_1_sync,
> + .rdev_size_change = super_1_rdev_size_change,
> },
> };
>
> @@ -1946,8 +1992,13 @@ rdev_size_store(mdk_rdev_t *rdev, const
> unsigned long long size = simple_strtoull(buf, &e, 10);
> if (e==buf || (*e && *e != '\n'))
> return -EINVAL;
> - if (rdev->mddev->pers)
> - return -EBUSY;
> + if (rdev->mddev->pers) {
> + mdp_super_t *sb;
> + sb = (mdp_super_t *) page_address(rdev->sb_page);
> + size = super_types[sb->major_version].rdev_size_change(rdev, size);
> + if (!size)
> + return -EBUSY;
> + }
> rdev->size = size;
> if (size < rdev->mddev->size || rdev->mddev->size == 0)
> rdev->mddev->size = size;
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
