Re: [PATCH 1/1] mdadm/super1: Add MD_FEATURE_RAID0_LAYOUT if sb->layout is set

[Mariusz Tkaczyk <mariusz.tkaczyk@linux.intel.com>]

On Fri, 13 Oct 2023 20:12:38 +0800
Xiao Ni <xni@redhat.com> wrote:

> > > > So, it forces the calculations made by Neil back but I think that we can
> > > > simply compare dev_size and data_offset between members.  
> > >
> > > We don't need to consider the compatibility anymore in future?
> > >  
> > Not sure if I get your question correctly. This property is supported now so
> > why we should? It is already there so we are safe to set it.  
> 
> I asked because you said we can remove the check in future. So I don't
> know why we don't need the check in future. The check here should be
> the kernel version check, right?


We are not supporting old kernels forever. At some point of time, we would
decide that kernels older than 5.5 are no longer a valid case and then we will
free to remove verification. If we are not supporting something older than the
version where it was added, we can assume that MD_RAID0_LAYOUT is always
available and we don't need to care anymore, right?

Here a recent example:
https://git.kernel.org/pub/scm/utils/mdadm/mdadm.git/commit/?id=f8d2c4286a

Thanks,
Mariusz