* + userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed.patch added to -mm tree
@ 2017-03-03 22:17 akpm
0 siblings, 0 replies; only message in thread
From: akpm @ 2017-03-03 22:17 UTC (permalink / raw)
To: aarcange, dgilbert, hillf.zj, mike.kravetz, rppt, xemul, mm-commits
The patch titled
Subject: userfaultfd: non-cooperative: userfaultfd_remove revalidate vma in MADV_DONTNEED
has been added to the -mm tree. Its filename is
userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed.patch
This patch should soon appear at
http://ozlabs.org/~akpm/mmots/broken-out/userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed.patch
and later at
http://ozlabs.org/~akpm/mmotm/broken-out/userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed.patch
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/SubmitChecklist when testing your code ***
The -mm tree is included into linux-next and is updated
there every 3-4 working days
------------------------------------------------------
From: Andrea Arcangeli <aarcange@redhat.com>
Subject: userfaultfd: non-cooperative: userfaultfd_remove revalidate vma in MADV_DONTNEED
userfaultfd_remove() has to be execute before zapping the pagetables or
UFFDIO_COPY could keep filling pages after zap_page_range returned,
which would result in non zero data after a MADV_DONTNEED.
However userfaultfd_remove() may have to release the mmap_sem. This
was handled correctly in MADV_REMOVE, but MADV_DONTNEED accessed a
potentially stale vma (the very vma passed to zap_page_range(vma, ...)).
The fix consists in revalidating the vma in case userfaultfd_remove()
had to release the mmap_sem.
This also optimizes away an unnecessary down_read/up_read in the
MADV_REMOVE case if UFFD_EVENT_FORK had to be delivered.
It all remains zero runtime cost in case CONFIG_USERFAULTFD=n as
userfaultfd_remove() will be defined as "true" at build time.
Link: http://lkml.kernel.org/r/20170302173738.18994-3-aarcange@redhat.com
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Pavel Emelyanov <xemul@parallels.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---
fs/userfaultfd.c | 9 ++----
include/linux/userfaultfd_k.h | 7 ++---
mm/madvise.c | 44 +++++++++++++++++++++++++++++---
3 files changed, 47 insertions(+), 13 deletions(-)
diff -puN fs/userfaultfd.c~userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed fs/userfaultfd.c
--- a/fs/userfaultfd.c~userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed
+++ a/fs/userfaultfd.c
@@ -695,8 +695,7 @@ void mremap_userfaultfd_complete(struct
userfaultfd_event_wait_completion(ctx, &ewq);
}
-void userfaultfd_remove(struct vm_area_struct *vma,
- struct vm_area_struct **prev,
+bool userfaultfd_remove(struct vm_area_struct *vma,
unsigned long start, unsigned long end)
{
struct mm_struct *mm = vma->vm_mm;
@@ -705,13 +704,11 @@ void userfaultfd_remove(struct vm_area_s
ctx = vma->vm_userfaultfd_ctx.ctx;
if (!ctx || !(ctx->features & UFFD_FEATURE_EVENT_REMOVE))
- return;
+ return true;
userfaultfd_ctx_get(ctx);
up_read(&mm->mmap_sem);
- *prev = NULL; /* We wait for ACK w/o the mmap semaphore */
-
msg_init(&ewq.msg);
ewq.msg.event = UFFD_EVENT_REMOVE;
@@ -720,7 +717,7 @@ void userfaultfd_remove(struct vm_area_s
userfaultfd_event_wait_completion(ctx, &ewq);
- down_read(&mm->mmap_sem);
+ return false;
}
static bool has_unmap_ctx(struct userfaultfd_ctx *ctx, struct list_head *unmaps,
diff -puN include/linux/userfaultfd_k.h~userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed include/linux/userfaultfd_k.h
--- a/include/linux/userfaultfd_k.h~userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed
+++ a/include/linux/userfaultfd_k.h
@@ -61,8 +61,7 @@ extern void mremap_userfaultfd_complete(
unsigned long from, unsigned long to,
unsigned long len);
-extern void userfaultfd_remove(struct vm_area_struct *vma,
- struct vm_area_struct **prev,
+extern bool userfaultfd_remove(struct vm_area_struct *vma,
unsigned long start,
unsigned long end);
@@ -118,11 +117,11 @@ static inline void mremap_userfaultfd_co
{
}
-static inline void userfaultfd_remove(struct vm_area_struct *vma,
- struct vm_area_struct **prev,
+static inline bool userfaultfd_remove(struct vm_area_struct *vma,
unsigned long start,
unsigned long end)
{
+ return true;
}
static inline int userfaultfd_unmap_prep(struct vm_area_struct *vma,
diff -puN mm/madvise.c~userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed mm/madvise.c
--- a/mm/madvise.c~userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed
+++ a/mm/madvise.c
@@ -513,7 +513,43 @@ static long madvise_dontneed(struct vm_a
if (!can_madv_dontneed_vma(vma))
return -EINVAL;
- userfaultfd_remove(vma, prev, start, end);
+ if (!userfaultfd_remove(vma, start, end)) {
+ *prev = NULL; /* mmap_sem has been dropped, prev is stale */
+
+ down_read(¤t->mm->mmap_sem);
+ vma = find_vma(current->mm, start);
+ if (!vma)
+ return -ENOMEM;
+ if (start < vma->vm_start) {
+ /*
+ * This "vma" under revalidation is the one
+ * with the lowest vma->vm_start where start
+ * is also < vma->vm_end. If start <
+ * vma->vm_start it means an hole materialized
+ * in the user address space within the
+ * virtual range passed to MADV_DONTNEED.
+ */
+ return -ENOMEM;
+ }
+ if (!can_madv_dontneed_vma(vma))
+ return -EINVAL;
+ if (end > vma->vm_end) {
+ /*
+ * Don't fail if end > vma->vm_end. If the old
+ * vma was splitted while the mmap_sem was
+ * released the effect of the concurrent
+ * operation may not cause MADV_DONTNEED to
+ * have an undefined result. There may be an
+ * adjacent next vma that we'll walk
+ * next. userfaultfd_remove() will generate an
+ * UFFD_EVENT_REMOVE repetition on the
+ * end-vma->vm_end range, but the manager can
+ * handle a repetition fine.
+ */
+ end = vma->vm_end;
+ }
+ VM_WARN_ON(start >= end);
+ }
zap_page_range(vma, start, end - start);
return 0;
}
@@ -554,8 +590,10 @@ static long madvise_remove(struct vm_are
* mmap_sem.
*/
get_file(f);
- userfaultfd_remove(vma, prev, start, end);
- up_read(¤t->mm->mmap_sem);
+ if (userfaultfd_remove(vma, start, end)) {
+ /* mmap_sem was not released by userfaultfd_remove() */
+ up_read(¤t->mm->mmap_sem);
+ }
error = vfs_fallocate(f,
FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE,
offset, end - start);
_
Patches currently in -mm which might be from aarcange@redhat.com are
userfaultfd-shmem-__do_fault-requires-vm_fault_nopage.patch
userfaultfd-non-cooperative-rollback-userfaultfd_exit.patch
userfaultfd-non-cooperative-robustness-check.patch
userfaultfd-non-cooperative-release-all-ctx-in-dup_userfaultfd_complete.patch
userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed.patch
userfaultfd-selftest-vm-allow-to-build-in-vm-directory.patch
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2017-03-03 22:40 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-03-03 22:17 + userfaultfd-non-cooperative-userfaultfd_remove-revalidate-vma-in-madv_dontneed.patch added to -mm tree akpm
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).