From: Peter Xu <peterx@redhat.com>
To: linux-kernel@vger.kernel.org, linux-mm@kvack.org
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>,
Mike Kravetz <mike.kravetz@oracle.com>,
peterx@redhat.com, Jerome Glisse <jglisse@redhat.com>,
"Kirill A . Shutemov" <kirill@shutemov.name>,
Hugh Dickins <hughd@google.com>,
Axel Rasmussen <axelrasmussen@google.com>,
Matthew Wilcox <willy@infradead.org>,
Andrew Morton <akpm@linux-foundation.org>,
Andrea Arcangeli <aarcange@redhat.com>,
Nadav Amit <nadav.amit@gmail.com>
Subject: [PATCH RFC 04/30] shmem/userfaultfd: Take care of UFFDIO_COPY_MODE_WP
Date: Fri, 15 Jan 2021 12:08:41 -0500 [thread overview]
Message-ID: <20210115170907.24498-5-peterx@redhat.com> (raw)
In-Reply-To: <20210115170907.24498-1-peterx@redhat.com>
Firstly, pass wp_copy into shmem_mfill_atomic_pte() through the stack.
Then apply the UFFD_WP bit properly when the UFFDIO_COPY on shmem is with
UFFDIO_COPY_MODE_WP.
One thing to mention is that shmem_mfill_atomic_pte() needs to set the dirty
bit in pte even if UFFDIO_COPY_MODE_WP is set. The reason is similar to
dcf7fe9d8976 ("userfaultfd: shmem: UFFDIO_COPY: set the page dirty if VM_WRITE
is not set") where we need to set page as dirty even if VM_WRITE is no there.
It's just that shmem can drop the pte any time later, and if it's not dirty the
data will be dropped. For uffd-wp, that could lead to data loss if without the
dirty bit set.
Note that shmem_mfill_zeropage_pte() will always call shmem_mfill_atomic_pte()
with wp_copy==false because UFFDIO_ZEROCOPY does not support UFFDIO_COPY_MODE_WP.
Signed-off-by: Peter Xu <peterx@redhat.com>
---
include/linux/shmem_fs.h | 5 +++--
mm/shmem.c | 26 +++++++++++++++++++-------
mm/userfaultfd.c | 2 +-
3 files changed, 23 insertions(+), 10 deletions(-)
diff --git a/include/linux/shmem_fs.h b/include/linux/shmem_fs.h
index a5a5d1d4d7b1..9d6fc68a1e57 100644
--- a/include/linux/shmem_fs.h
+++ b/include/linux/shmem_fs.h
@@ -123,14 +123,15 @@ extern int shmem_mcopy_atomic_pte(struct mm_struct *dst_mm, pmd_t *dst_pmd,
struct vm_area_struct *dst_vma,
unsigned long dst_addr,
unsigned long src_addr,
- struct page **pagep);
+ struct page **pagep,
+ bool wp_copy);
extern int shmem_mfill_zeropage_pte(struct mm_struct *dst_mm,
pmd_t *dst_pmd,
struct vm_area_struct *dst_vma,
unsigned long dst_addr);
#else
#define shmem_mcopy_atomic_pte(dst_mm, dst_pte, dst_vma, dst_addr, \
- src_addr, pagep) ({ BUG(); 0; })
+ src_addr, pagep, wp_copy) ({ BUG(); 0; })
#define shmem_mfill_zeropage_pte(dst_mm, dst_pmd, dst_vma, \
dst_addr) ({ BUG(); 0; })
#endif
diff --git a/mm/shmem.c b/mm/shmem.c
index 537c137698f8..de45333626f7 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -2363,7 +2363,8 @@ static int shmem_mfill_atomic_pte(struct mm_struct *dst_mm,
unsigned long dst_addr,
unsigned long src_addr,
bool zeropage,
- struct page **pagep)
+ struct page **pagep,
+ bool wp_copy)
{
struct inode *inode = file_inode(dst_vma->vm_file);
struct shmem_inode_info *info = SHMEM_I(inode);
@@ -2425,9 +2426,18 @@ static int shmem_mfill_atomic_pte(struct mm_struct *dst_mm,
goto out_release;
_dst_pte = mk_pte(page, dst_vma->vm_page_prot);
- if (dst_vma->vm_flags & VM_WRITE)
- _dst_pte = pte_mkwrite(pte_mkdirty(_dst_pte));
- else {
+ if (dst_vma->vm_flags & VM_WRITE) {
+ if (wp_copy)
+ _dst_pte = pte_mkuffd_wp(pte_wrprotect(_dst_pte));
+ else
+ _dst_pte = pte_mkwrite(_dst_pte);
+ /*
+ * Similar reason to set_page_dirty(), that we need to mark the
+ * pte dirty even if wp_copy==true here, otherwise the pte and
+ * its page could be dropped at anytime when e.g. swapped out.
+ */
+ _dst_pte = pte_mkdirty(_dst_pte);
+ } else {
/*
* We don't set the pte dirty if the vma has no
* VM_WRITE permission, so mark the page dirty or it
@@ -2485,10 +2495,12 @@ int shmem_mcopy_atomic_pte(struct mm_struct *dst_mm,
struct vm_area_struct *dst_vma,
unsigned long dst_addr,
unsigned long src_addr,
- struct page **pagep)
+ struct page **pagep,
+ bool wp_copy)
{
return shmem_mfill_atomic_pte(dst_mm, dst_pmd, dst_vma,
- dst_addr, src_addr, false, pagep);
+ dst_addr, src_addr, false, pagep,
+ wp_copy);
}
int shmem_mfill_zeropage_pte(struct mm_struct *dst_mm,
@@ -2499,7 +2511,7 @@ int shmem_mfill_zeropage_pte(struct mm_struct *dst_mm,
struct page *page = NULL;
return shmem_mfill_atomic_pte(dst_mm, dst_pmd, dst_vma,
- dst_addr, 0, true, &page);
+ dst_addr, 0, true, &page, false);
}
#ifdef CONFIG_TMPFS
diff --git a/mm/userfaultfd.c b/mm/userfaultfd.c
index 9a3d451402d7..6d4b3b7c7f9f 100644
--- a/mm/userfaultfd.c
+++ b/mm/userfaultfd.c
@@ -445,7 +445,7 @@ static __always_inline ssize_t mfill_atomic_pte(struct mm_struct *dst_mm,
if (!zeropage)
err = shmem_mcopy_atomic_pte(dst_mm, dst_pmd,
dst_vma, dst_addr,
- src_addr, page);
+ src_addr, page, wp_copy);
else
err = shmem_mfill_zeropage_pte(dst_mm, dst_pmd,
dst_vma, dst_addr);
--
2.26.2
next prev parent reply other threads:[~2021-01-15 17:11 UTC|newest]
Thread overview: 43+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-01-15 17:08 [PATCH RFC 00/30] userfaultfd-wp: Support shmem and hugetlbfs Peter Xu
2021-01-15 17:08 ` [PATCH RFC 01/30] mm/thp: Simplify copying of huge zero page pmd when fork Peter Xu
2021-01-15 17:08 ` [PATCH RFC 02/30] mm/userfaultfd: Fix uffd-wp special cases for fork() Peter Xu
2021-01-15 17:08 ` [PATCH RFC 03/30] mm/userfaultfd: Fix a few thp pmd missing uffd-wp bit Peter Xu
2021-01-15 17:08 ` Peter Xu [this message]
2021-01-15 17:08 ` [PATCH RFC 05/30] mm: Clear vmf->pte after pte_unmap_same() returns Peter Xu
2021-01-15 17:08 ` [PATCH RFC 06/30] mm/userfaultfd: Introduce special pte for unmapped file-backed mem Peter Xu
2021-01-15 17:08 ` [PATCH RFC 07/30] mm/swap: Introduce the idea of special swap ptes Peter Xu
2021-01-18 19:40 ` Jason Gunthorpe
2021-01-19 14:24 ` Peter Xu
2021-01-15 17:08 ` [PATCH RFC 08/30] shmem/userfaultfd: Handle uffd-wp special pte in page fault handler Peter Xu
2021-01-15 17:08 ` [PATCH RFC 09/30] mm: Drop first_index/last_index in zap_details Peter Xu
2021-01-15 17:08 ` [PATCH RFC 10/30] mm: Introduce zap_details.zap_flags Peter Xu
2021-01-15 17:08 ` [PATCH RFC 11/30] mm: Introduce ZAP_FLAG_SKIP_SWAP Peter Xu
2021-01-15 17:08 ` [PATCH RFC 12/30] mm: Pass zap_flags into unmap_mapping_pages() Peter Xu
2021-01-15 17:08 ` [PATCH RFC 13/30] shmem/userfaultfd: Persist uffd-wp bit across zapping for file-backed Peter Xu
2021-01-15 17:08 ` [PATCH RFC 14/30] shmem/userfaultfd: Allow wr-protect none pte for file-backed mem Peter Xu
2021-01-15 17:08 ` [PATCH RFC 15/30] shmem/userfaultfd: Allows file-back mem to be uffd wr-protected on thps Peter Xu
2021-01-15 17:08 ` [PATCH RFC 16/30] shmem/userfaultfd: Handle the left-overed special swap ptes Peter Xu
2021-01-15 17:08 ` [PATCH RFC 17/30] shmem/userfaultfd: Pass over uffd-wp special swap pte when fork() Peter Xu
2021-01-15 17:08 ` [PATCH RFC 18/30] hugetlb/userfaultfd: Hook page faults for uffd write protection Peter Xu
2021-01-15 17:08 ` [PATCH RFC 19/30] hugetlb/userfaultfd: Take care of UFFDIO_COPY_MODE_WP Peter Xu
2021-01-15 17:08 ` [PATCH RFC 20/30] hugetlb/userfaultfd: Handle UFFDIO_WRITEPROTECT Peter Xu
2021-01-15 17:08 ` [PATCH RFC 21/30] hugetlb: Pass vma into huge_pte_alloc() Peter Xu
2021-01-28 22:59 ` Axel Rasmussen
2021-01-29 22:31 ` Peter Xu
2021-01-30 8:08 ` Axel Rasmussen
2021-01-15 17:08 ` [PATCH RFC 22/30] hugetlb/userfaultfd: Forbid huge pmd sharing when uffd enabled Peter Xu
2021-01-15 17:09 ` [PATCH RFC 23/30] mm/hugetlb: Introduce huge version of special swap pte helpers Peter Xu
2021-01-15 17:09 ` [PATCH RFC 24/30] mm/hugetlb: Move flush_hugetlb_tlb_range() into hugetlb.h Peter Xu
2021-01-15 17:09 ` [PATCH RFC 25/30] hugetlb/userfaultfd: Unshare all pmds for hugetlbfs when register wp Peter Xu
2021-01-15 17:09 ` [PATCH RFC 26/30] hugetlb/userfaultfd: Handle uffd-wp special pte in hugetlb pf handler Peter Xu
2021-01-15 17:09 ` [PATCH RFC 27/30] hugetlb/userfaultfd: Allow wr-protect none ptes Peter Xu
2021-01-15 17:09 ` [PATCH RFC 28/30] hugetlb/userfaultfd: Only drop uffd-wp special pte if required Peter Xu
2021-01-15 17:09 ` [PATCH RFC 29/30] userfaultfd: Enable write protection for shmem & hugetlbfs Peter Xu
2021-01-15 17:12 ` [PATCH RFC 30/30] userfaultfd/selftests: Enable uffd-wp for shmem/hugetlbfs Peter Xu
2021-01-29 22:49 ` [PATCH RFC 00/30] userfaultfd-wp: Support shmem and hugetlbfs Peter Xu
2021-02-05 21:53 ` Mike Kravetz
2021-02-06 2:36 ` Peter Xu
2021-02-09 19:29 ` Mike Kravetz
2021-02-09 22:00 ` Peter Xu
2021-02-05 22:21 ` Hugh Dickins
2021-02-06 2:47 ` Peter Xu
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=20210115170907.24498-5-peterx@redhat.com \
--to=peterx@redhat.com \
--cc=aarcange@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=axelrasmussen@google.com \
--cc=hughd@google.com \
--cc=jglisse@redhat.com \
--cc=kirill@shutemov.name \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mike.kravetz@oracle.com \
--cc=nadav.amit@gmail.com \
--cc=rppt@linux.vnet.ibm.com \
--cc=willy@infradead.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;
as well as URLs for NNTP newsgroup(s).