mdadm --create supports --logical-block-size option

[wuguanghao3]

The kernel block branch has started supporting the
configuration of logical block size after mergeing commit
62ed1b582246 (md:allow configuring logical block size),
therefore a new parameter should be added to allow specifying
the logical block size when creating a RAID device.

Links: https://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux.git/commit/?h=for-6.19/block&id=62ed1b582246

[mtkaczyk]

Hi, bunch of questions :)

The kernel already supports configuring logical block size, therefore a new
parameter should be added to allow specifying the logical block size when
creating a RAID device.

Please mention when it was added. Should we check kernel version to determine if it is supported?
If it is fresh change, what will happen if kernel is not supporting it?

Please also update README.

What about recovery path? if change for one disk fail, should we restore previous the logical block size for all?
What will happen if we will try to set same block size twice?

LGTM,. but we need README do be updated!

[mtkaczyk]

I don't understand the special_options at all concept so please care to explain why it is added here.

[wuguanghao3]

My understanding is that these options are not general options but specific to certain modes. LogicalBlocksize only takes effect when used with --create to create something. Then, we need to define one value through O(Create, xxx) to reach the corresponding branch.

[wuguanghao3]

Hi, bunch of questions :)

The kernel already supports configuring logical block size, therefore a new
parameter should be added to allow specifying the logical block size when
creating a RAID device.

Please mention when it was added. Should we check kernel version to determine if it is supported? If it is fresh change, what will happen if kernel is not supporting it?

Please also update README.

What about recovery path? if change for one disk fail, should we restore previous the logical block size for all? What will happen if we will try to set same block size twice?

LGTM,. but we need README do be updated!

1. I don't think it's necessary to check the kernel version. Because if it's an old kernel version, the logical_block_size file does not exist under the /sys path, so it would directly result in an error.
2. OK, I will update the README later.
3. The user space only writes values to the /sys path of the md device and does not directly modify the metadata information of the member disks. If the setting fails, the md device will also fail to be created, and there is no need for recovery.