linux-kernel - Re: [PATCH v3 08/16] mm/rmap: drop "compound" parameter from page_add_new_anon

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 13 Apr 2022 13:48:51 +0100
From:   Matthew Wilcox <willy@...radead.org>
To:     David Hildenbrand <david@...hat.com>
Cc:     Vlastimil Babka <vbabka@...e.cz>, linux-kernel@...r.kernel.org,
        Andrew Morton <akpm@...ux-foundation.org>,
        Hugh Dickins <hughd@...gle.com>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        David Rientjes <rientjes@...gle.com>,
        Shakeel Butt <shakeelb@...gle.com>,
        John Hubbard <jhubbard@...dia.com>,
        Jason Gunthorpe <jgg@...dia.com>,
        Mike Kravetz <mike.kravetz@...cle.com>,
        Mike Rapoport <rppt@...ux.ibm.com>,
        Yang Shi <shy828301@...il.com>,
        "Kirill A . Shutemov" <kirill.shutemov@...ux.intel.com>,
        Jann Horn <jannh@...gle.com>, Michal Hocko <mhocko@...nel.org>,
        Nadav Amit <namit@...are.com>, Rik van Riel <riel@...riel.com>,
        Roman Gushchin <guro@...com>,
        Andrea Arcangeli <aarcange@...hat.com>,
        Peter Xu <peterx@...hat.com>,
        Donald Dutile <ddutile@...hat.com>,
        Christoph Hellwig <hch@....de>,
        Oleg Nesterov <oleg@...hat.com>, Jan Kara <jack@...e.cz>,
        Liang Zhang <zhangliang5@...wei.com>,
        Pedro Gomes <pedrodemargomes@...il.com>,
        Oded Gabbay <oded.gabbay@...il.com>, linux-mm@...ck.org
Subject: Re: [PATCH v3 08/16] mm/rmap: drop "compound" parameter from
 page_add_new_anon_rmap()

On Wed, Apr 13, 2022 at 02:28:38PM +0200, David Hildenbrand wrote:
> On 13.04.22 14:26, Matthew Wilcox wrote:
> > On Tue, Apr 12, 2022 at 11:37:09AM +0200, David Hildenbrand wrote:
> >> On 12.04.22 10:47, Vlastimil Babka wrote:
> >>> There's a VM_BUG_ON_PAGE(PageTransCompound(page), page); later in a
> >>> !compound branch. Since compound is now determined by the same check, could
> >>> be deleted.
> >>
> >> Yes, eventually we could get rid of both VM_BUG_ON_PAGE() on both
> >> branches and add a single VM_BUG_ON_PAGE(PageTail(page), page) check on
> >> the compound branch. (we could also make sure that we're not given a
> >> hugetlb page)
> > 
> > As a rule of thumb, if you find yourself wanting to add
> > VM_BUG_ON_PAGE(PageTail(page), page), you probably want to change the
> > interface to take a folio.
> 
> Yeah, I had the same in mind. Might be a reasonable addon on top --
> although it would stick out in the rmap code a bit because most
> functions deal with both, folios and subpages.

I have the start of a series which starts looking at the fault path
to see where it makes sense to use folios and where it makes sense to
use pages.

We're (generally) faulting on a PTE, so we need the precise page to
be returned in vmf->page.  However vmf->cow_page can/should be a
folio (because it's definitely not a tail page).  That trickles
down into copy_present_page() (new_page and prealloc both become folios)
and so page_add_new_anon_rmap() then looks like a good target to
take a folio.

The finish_fault() -> do_set_pte() -> page_add_new_anon_rmap() looks
like the only kind of strange place where we don't necessarily have a
folio (all the others we just allocated it).