linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Peter Collingbourne <pcc@google.com>
To: Catalin Marinas <catalin.marinas@arm.com>
Cc: "Peter Collingbourne" <pcc@google.com>,
	"Qun-wei Lin (林群崴)" <Qun-wei.Lin@mediatek.com>,
	linux-arm-kernel@lists.infradead.org, linux-mm@kvack.org,
	linux-kernel@vger.kernel.org,
	"surenb@google.com" <surenb@google.com>,
	"david@redhat.com" <david@redhat.com>,
	"Chinwen Chang (張錦文)" <chinwen.chang@mediatek.com>,
	"kasan-dev@googlegroups.com" <kasan-dev@googlegroups.com>,
	"Kuan-Ying Lee (李冠穎)" <Kuan-Ying.Lee@mediatek.com>,
	"Casper Li (李中榮)" <casper.li@mediatek.com>,
	"gregkh@linuxfoundation.org" <gregkh@linuxfoundation.org>,
	vincenzo.frascino@arm.com,
	"Alexandru Elisei" <alexandru.elisei@arm.com>,
	will@kernel.org, eugenis@google.com,
	"Steven Price" <steven.price@arm.com>,
	stable@vger.kernel.org
Subject: [PATCH 2/3] mm: Call arch_swap_restore() from arch_do_swap_page() and deprecate the latter
Date: Fri, 12 May 2023 16:57:51 -0700	[thread overview]
Message-ID: <20230512235755.1589034-3-pcc@google.com> (raw)
In-Reply-To: <20230512235755.1589034-1-pcc@google.com>

The previous patch made it possible for MTE to restore tags before they
are freed by hooking arch_do_swap_page().

However, the arch_do_swap_page() hook API is incompatible with swap
restoration in circumstances where we do not have an mm or a vma,
such as swapoff with swapped out shmem, and I expect that ADI will
currently fail to restore tags in these circumstances. This implies that
arch-specific metadata stores ought to be indexed by swap index, as MTE
does, rather than by mm and vma, as ADI does, and we should discourage
hooking arch_do_swap_page(), preferring to hook arch_swap_restore()
instead, as MTE already does.

Therefore, instead of directly hooking arch_do_swap_page() for
MTE, deprecate that hook, change its default implementation to call
arch_swap_restore() and rely on the existing implementation of the latter
for MTE.

Fixes: c145e0b47c77 ("mm: streamline COW logic in do_swap_page()")
Link: https://linux-review.googlesource.com/id/Id2f1ad76eaf606ae210e1d2dd0b7fe287e5f7d87
Signed-off-by: Peter Collingbourne <pcc@google.com>
Reported-by: Qun-wei Lin (林群崴) <Qun-wei.Lin@mediatek.com>
Link: https://lore.kernel.org/all/5050805753ac469e8d727c797c2218a9d780d434.camel@mediatek.com/
Cc: <stable@vger.kernel.org> # 6.1
---
 include/linux/pgtable.h | 26 +++++++++++++-------------
 1 file changed, 13 insertions(+), 13 deletions(-)

diff --git a/include/linux/pgtable.h b/include/linux/pgtable.h
index c63cd44777ec..fc0259cf60fb 100644
--- a/include/linux/pgtable.h
+++ b/include/linux/pgtable.h
@@ -740,6 +740,12 @@ static inline int pgd_same(pgd_t pgd_a, pgd_t pgd_b)
 	set_pgd(pgdp, pgd); \
 })
 
+#ifndef __HAVE_ARCH_SWAP_RESTORE
+static inline void arch_swap_restore(swp_entry_t entry, struct folio *folio)
+{
+}
+#endif
+
 #ifndef __HAVE_ARCH_DO_SWAP_PAGE
 /*
  * Some architectures support metadata associated with a page. When a
@@ -748,14 +754,14 @@ static inline int pgd_same(pgd_t pgd_a, pgd_t pgd_b)
  * processors support an ADI (Application Data Integrity) tag for the
  * page as metadata for the page. arch_do_swap_page() can restore this
  * metadata when a page is swapped back in.
+ *
+ * This hook is deprecated. Architectures should hook arch_swap_restore()
+ * instead, because this hook is not called on all code paths that can
+ * swap in a page, particularly those where mm and vma are not available
+ * (e.g. swapoff for shmem pages).
  */
-static inline void arch_do_swap_page(struct mm_struct *mm,
-				     struct vm_area_struct *vma,
-				     unsigned long addr,
-				     pte_t pte, pte_t oldpte)
-{
-
-}
+#define arch_do_swap_page(mm, vma, addr, pte, oldpte) \
+	arch_swap_restore(pte_to_swp_entry(oldpte), page_folio(pte_page(pte)))
 #endif
 
 #ifndef __HAVE_ARCH_UNMAP_ONE
@@ -798,12 +804,6 @@ static inline void arch_swap_invalidate_area(int type)
 }
 #endif
 
-#ifndef __HAVE_ARCH_SWAP_RESTORE
-static inline void arch_swap_restore(swp_entry_t entry, struct folio *folio)
-{
-}
-#endif
-
 #ifndef __HAVE_ARCH_PGD_OFFSET_GATE
 #define pgd_offset_gate(mm, addr)	pgd_offset(mm, addr)
 #endif
-- 
2.40.1.606.ga4b1b128d6-goog



  parent reply	other threads:[~2023-05-12 23:58 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-05-12 23:57 [PATCH 0/3] mm: Fix bug affecting swapping in MTE tagged pages Peter Collingbourne
2023-05-12 23:57 ` [PATCH 1/3] mm: Move arch_do_swap_page() call to before swap_free() Peter Collingbourne
2023-05-13  3:29   ` David Hildenbrand
2023-05-15 17:34     ` Catalin Marinas
2023-05-15 23:40       ` Peter Collingbourne
2023-05-16 12:35         ` David Hildenbrand
2023-05-17  1:57           ` Peter Collingbourne
2023-05-17  8:30             ` David Hildenbrand
2023-05-18 20:06               ` Peter Collingbourne
2023-05-19  9:21                 ` David Hildenbrand
2023-05-19 16:21                   ` Catalin Marinas
2023-05-16 12:30       ` David Hildenbrand
2023-05-17  1:37         ` Peter Collingbourne
2023-05-17  8:31           ` David Hildenbrand
2023-05-16  0:16     ` Peter Collingbourne
2023-05-16  2:35       ` Peter Collingbourne
2023-05-17  8:34         ` David Hildenbrand
2023-05-16 12:40       ` David Hildenbrand
2023-05-17  2:13         ` Peter Collingbourne
2023-05-12 23:57 ` Peter Collingbourne [this message]
2023-05-13  3:34   ` [PATCH 2/3] mm: Call arch_swap_restore() from arch_do_swap_page() and deprecate the latter David Hildenbrand
2023-05-12 23:57 ` [PATCH 3/3] arm64: mte: Simplify swap tag restoration logic and fix uninitialized tag issue Peter Collingbourne

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230512235755.1589034-3-pcc@google.com \
    --to=pcc@google.com \
    --cc=Kuan-Ying.Lee@mediatek.com \
    --cc=Qun-wei.Lin@mediatek.com \
    --cc=alexandru.elisei@arm.com \
    --cc=casper.li@mediatek.com \
    --cc=catalin.marinas@arm.com \
    --cc=chinwen.chang@mediatek.com \
    --cc=david@redhat.com \
    --cc=eugenis@google.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=kasan-dev@googlegroups.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=stable@vger.kernel.org \
    --cc=steven.price@arm.com \
    --cc=surenb@google.com \
    --cc=vincenzo.frascino@arm.com \
    --cc=will@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox