On 05.08.25 15:25, Jason Gunthorpe wrote:
On Tue, Aug 05, 2025 at 04:00:53PM +0300, Linus Torvalds wrote:
On Tue, 5 Aug 2025 at 10:47, David Hildenbrand <david@xxxxxxxxxx> wrote:
The concern is rather false positives, meaning, you want consecutive
PFNs (just like within a folio), but -- because the stars aligned --
you get consecutive "struct page" that do not translate to consecutive PFNs.
So I don't think that can happen with a valid 'struct page', because
if the 'struct page's are in different sections, they will have been
allocated separately too.
This is certainly true for the CONFIG_SPARSEMEM_VMEMMAP case, but in
the other cases I thought we end up with normal allocations for struct
page? This is what David was talking about.
So then we are afraid of this:
a = kvmalloc_array(nelms_a);
b = kvmalloc_array(nelms_b);
assert(a + nelms_a != b)
I thought this was possible with our allocator, especially vmemmap?
David, there is another alternative to prevent this, simple though a
bit wasteful, just allocate a bit bigger to ensure the allocation
doesn't end on an exact PAGE_SIZE boundary?
:/ in particular doing that through the memblock in sparse_init_nid(), I
am not so sure that's a good idea.
I prefer Linus' proposal and avoids the one nth_page(), unless any other
approach can help us get rid of more nth_page() usage -- and I don't
think your proposal could, right?
--
Cheers,
David / dhildenb
