Re: [PATCH v2] mm/fake-numa: fix under-allocation detection in uniform split

[Mike Rapoport <rppt@kernel.org>]

On Wed, Apr 22, 2026 at 06:26:34PM +0900, Sang-Heon Jeon wrote:
> Hello,
> 
> On Wed, Apr 22, 2026 at 5:36 PM Mike Rapoport <rppt@kernel.org> wrote:
> >
> > On Tue, Apr 21, 2026 at 03:56:01PM +0900, Sang-Heon Jeon wrote:
> > > Hi
> > >
> > > On Tue, Apr 21, 2026 at 3:29 PM Mike Rapoport <rppt@kernel.org> wrote:
> > > >
> > > > On Mon, Apr 20, 2026 at 11:26:13PM +0900, Sang-Heon Jeon wrote:
> > > > > On Mon, Apr 20, 2026 at 3:31 PM Mike Rapoport <rppt@kernel.org> wrote:
> > > > > > On Fri, Apr 17, 2026 at 10:58:05PM +0900, Sang-Heon Jeon wrote:
> > > > > > > When split NUMA node uniformly, split_nodes_size_interleave_uniform()
> > > > > > > returns the next absolute node ID, not the number of nodes created.
> > > > > > >
> > > > > > > The existing under-allocation detection logic compares next absolute node
> > > > > > > ID (ret) and request count (n), which only works when nid starts at 0.
> > > > > > >
> > > > > > > For example, on a system with 2 physical NUMA nodes (node 0: 2GB, node
> > > > > > > 1: 128MB) and numa=fake=8U, 8 fake nodes are successfully created from
> > > > > > > node 0 and split_nodes_size_interleave_uniform() returns 8. For node 1,
> > > > > > > fake node nid starts at 8, but only 4 fake nodes are created due to
> > > > > > > current FAKE_NODE_MIN_SIZE being 32MB, and
> > > > > > > split_nodes_size_interleave_uniform() returns 12. By existing
> > > > > > > under-allocation detection logic, "ret < n" (12 < 8) is false, so the
> > > > > >
> > > > > > In this example it would be 11, won't it?
> > > > > > I'll update when applying.
> > > > >
> > > > > Oops, my previous reply is not the main point of this review. (Please ignore it)
> > > > > But 12 is still correct, because split_nodes_size_interleave_uniform()
> > > > > returns next available node ID.
> > > > >
> > > > > IMHO, it would be clearer if both the commit message and the function
> > > > > comment change into "next available node ID".
> > > > > If you're okay with it, I'll create v3 patch soon, including minor fix below
> > > >
> > > > No need for v3, I updated the comment and the whitespace:
> > > > https://git.kernel.org/rppt/memblock/c/a34436ac05e76bd6634a89fc0a399ab3a48f509a
> > >
> > > Thanks for your help, Mike. However, it seems like '12' was changed to
> > > '11' in the commit message. I think this change should be rolled back.
> > > Could you check?
> >
> > Node IDs are counted from 0, hence for 12 nodes the last nid would be 11.
> 
> Right, the last nid is indeed 11. However, the return value `ret` here
> refers to the next nid, so it should be 12 in this example.

Fixed now, thanks!
 
> Best Regards,
> Sang-Heon Jeon

-- 
Sincerely yours,
Mike.