On Sat, Feb 14, 2026 at 12:39:22PM -0800, Linus Torvalds wrote:
> On Sat, 14 Feb 2026 at 12:33, Eric Biggers <[email protected]> wrote:
> >
> > -               if (fsverity_verify_page(dic->vi, rpage))
> > +               if (fsverity_verify_blocks(dic->vi, page_folio(rpage),
> > +                                          PAGE_SIZE, 0))
> 
> This really is very wrong. It may be equivalent to the old code, but
> the old code was *also* wrong.
> 
> If you use "page_folio()", you need to do the proper offsetting of the
> page inside the folio, unless the filesystem is purely using the old
> legacy "folio is the same as page", which is simply not true in f2fs.
> 
> It might be true in this particular case, but considering that it was
> *NOT* true in another case I fixed up, I really don't want to see this
> same mistake done over and over again.
> 
> So either it's the whole folio, in which case PAGE_SIZE is wrong.
> 
> Or it really is PAGE_SIZE, in which case you need to use the proper
> offset within the folio.
> 
> Don't take the old buggy garbage that was fsverity_verify_page() and
> repeat the bug when you remove it.

The reason I went with the direct conversion is that
f2fs_verify_cluster() clearly assumes small folios already, and indeed
it's called only with small folios.  But sure, we can make that specific
line in it large-folio-aware.

- Eric


_______________________________________________
Linux-f2fs-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

Reply via email to