On Thu, 19 Sept 2024 at 06:36, Matthew Wilcox <willy@xxxxxxxxxxxxx> wrote:
>
> Probably xas_split_alloc() needs to just do the alloc, like the name
> says, and drop the 'entry' argument. ICBW, but I think it explains
> what you're seeing? Maybe it doesn't?
.. or we make the rule be that you have to re-check that the order and
the entry still matches when you do the actual xas_split()..
Like commit 6758c1128ceb does, in this case.
We do have another xas_split_alloc() - in the hugepage case - but
there we do have
xas_lock(&xas);
xas_reset(&xas);
if (xas_load(&xas) != folio)
goto fail;
and the folio is locked over the whole sequence, so I think that code
is probably safe and guarantees that we're splitting with the same
details we alloc'ed.
Linus
