[PATCH v3 0/3] simplify split calculation

[Liam R. Howlett]

* Andrew Morton <akpm at linux-foundation.org> [241129 23:31]:
> On Wed, 13 Nov 2024 03:16:13 +0000 Wei Yang <richard.weiyang at gmail.com> wrote:
> 
> > In version 1 [1], we found current split would result into deficient node.
> > 
> > By discussion, current implementation would lead to jitter problem. Since this
> > is a rare case in real world, we decide to simplify the split calculation.
> > 
> > Patch 1: simplify split calculation
> > Patch 2: add a test case to check deficient node
> > Patch 3: validate deficient node except for root node
> > 
> > [1]: https://lkml.kernel.org/r/20241020024628.22469-1-richard.weiyang@gmail.com
> 
> There's nothing here which is really useful as a [0/n] overview.
> 
> Why is [1/3] being proposed for -stable?  The changelog doesn't
> describe what benefit such a backport would provide to -stable users.
> 

The split calculation may cause an insufficient node in rare cases.
This could cause issues down the line for the tree during merging
operations, which would lead to wasted space at best and stability
issues at worse.  Although we haven't seen an insufficient node occur
(or at least not cause issues), it seems prudent to backport the fix to
remove the risk.

The benefit is to remove risk from tree operations by ensuring the nodes
remain in a known good (and well tested) state.

Does that seem reasonable?  I'd be happy to hear any guidance you can
provide for such issues and if you think it is worth sending to stable.

Thanks,
Liam