[PATCH v2 0/2] make VM_SOFTDIRTY a sticky VMA flag

[PATCH v2 0/2] make VM_SOFTDIRTY a sticky VMA flag

[Lorenzo Stoakes]

Currently we set VM_SOFTDIRTY when a new mapping is set up (whether by
establishing a new VMA, or via merge) as implemented in __mmap_complete()
and do_brk_flags().

However, when performing a merge of existing mappings such as when
performing mprotect(), we may lose the VM_SOFTDIRTY flag.

Now we have the concept of making VMA flags 'sticky', that is that they
both don't prevent merge and, importantly, are propagated to merged VMAs,
this seems a sensible alternative to the existing special-casing of
VM_SOFTDIRTY.

We additionally add a self-test that demonstrates that this logic behaves
as expected.

v2:
* Propagated tags (thanks David, Pedro!)
* Corrected comments and added ASCII diagrams as per David.
* Expanded self-test to check for mprotect() case.
* Updated incomplete cover letter.
* Updated commit messages accordingly.

v1:
https://lore.kernel.org/all/cover.1763142412.git.lorenzo.stoakes@oracle.com/

Lorenzo Stoakes (2):
  mm: propagate VM_SOFTDIRTY on merge
  testing/selftests/mm: add soft-dirty merge self-test

 include/linux/mm.h                      |  23 ++---
 tools/testing/selftests/mm/soft-dirty.c | 127 +++++++++++++++++++++++-
 tools/testing/vma/vma_internal.h        |  23 ++---
 3 files changed, 148 insertions(+), 25 deletions(-)

--
2.51.0

[PATCH v2 1/2] mm: propagate VM_SOFTDIRTY on merge

[Lorenzo Stoakes]

Currently we set VM_SOFTDIRTY when a new mapping is set up (whether by
establishing a new VMA, or via merge) as implemented in __mmap_complete()
and do_brk_flags().

However, when performing a merge of existing mappings such as when
performing mprotect(), we may lose the VM_SOFTDIRTY flag.

This is because currently we simply ignore VM_SOFTDIRTY for the purposes
of merge, so one VMA may possess the flag and another not, and whichever
happens to be the target VMA will be the one upon which the merge is
performed which may or may not have VM_SOFTDIRTY set.

Now we have the concept of 'sticky' VMA flags, let's make VM_SOFTDIRTY one
which solves this issue.

Additionally update VMA userland tests to propagate changes.

Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Suggested-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: David Hildenbrand (Red Hat) <david@kernel.org>
Reviewed-by: Pedro Falcato <pfalcato@suse.de>
---
 include/linux/mm.h               | 23 +++++++++++------------
 tools/testing/vma/vma_internal.h | 23 +++++++++++------------
 2 files changed, 22 insertions(+), 24 deletions(-)

diff --git a/include/linux/mm.h b/include/linux/mm.h
index 73fc90cbad94..057649b948f2 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -532,29 +532,28 @@ extern unsigned int kobjsize(const void *objp);
  * possesses it but the other does not, the merged VMA should nonetheless have
  * applied to it:
  *
+ *   VM_SOFTDIRTY - if a VMA is marked soft-dirty, that is has not had its
+ *                  references cleared via /proc/$pid/clear_refs, any merged VMA
+ *                  should be considered soft-dirty also as it operates at a VMA
+ *                  granularity.
+ *
  * VM_MAYBE_GUARD - If a VMA may have guard regions in place it implies that
  *                  mapped page tables may contain metadata not described by the
  *                  VMA and thus any merged VMA may also contain this metadata,
  *                  and thus we must make this flag sticky.
  */
-#define VM_STICKY VM_MAYBE_GUARD
+#define VM_STICKY (VM_SOFTDIRTY | VM_MAYBE_GUARD)

 /*
  * VMA flags we ignore for the purposes of merge, i.e. one VMA possessing one
  * of these flags and the other not does not preclude a merge.
  *
- * VM_SOFTDIRTY - Should not prevent from VMA merging, if we match the flags but
- *                dirty bit -- the caller should mark merged VMA as dirty. If
- *                dirty bit won't be excluded from comparison, we increase
- *                pressure on the memory system forcing the kernel to generate
- *                new VMAs when old one could be extended instead.
- *
- *    VM_STICKY - If one VMA has flags which most be 'sticky', that is ones
- *                which should propagate to all VMAs, but the other does not,
- *                the merge should still proceed with the merge logic applying
- *                sticky flags to the final VMA.
+ * VM_STICKY - If one VMA has flags which most be 'sticky', that is ones
+ *             which should propagate to all VMAs, but the other does not,
+ *             the merge should still proceed with the merge logic applying
+ *             sticky flags to the final VMA.
  */
-#define VM_IGNORE_MERGE (VM_SOFTDIRTY | VM_STICKY)
+#define VM_IGNORE_MERGE VM_STICKY

 /*
  * Flags which should result in page tables being copied on fork. These are
diff --git a/tools/testing/vma/vma_internal.h b/tools/testing/vma/vma_internal.h
index bd6352a5f24d..10f46a95a73a 100644
--- a/tools/testing/vma/vma_internal.h
+++ b/tools/testing/vma/vma_internal.h
@@ -122,29 +122,28 @@ extern unsigned long dac_mmap_min_addr;
  * possesses it but the other does not, the merged VMA should nonetheless have
  * applied to it:
  *
+ *   VM_SOFTDIRTY - if a VMA is marked soft-dirty, that is has not had its
+ *                  references cleared via /proc/$pid/clear_refs, any merged VMA
+ *                  should be considered soft-dirty also as it operates at a VMA
+ *                  granularity.
+ *
  * VM_MAYBE_GUARD - If a VMA may have guard regions in place it implies that
  *                  mapped page tables may contain metadata not described by the
  *                  VMA and thus any merged VMA may also contain this metadata,
  *                  and thus we must make this flag sticky.
  */
-#define VM_STICKY VM_MAYBE_GUARD
+#define VM_STICKY (VM_SOFTDIRTY | VM_MAYBE_GUARD)

 /*
  * VMA flags we ignore for the purposes of merge, i.e. one VMA possessing one
  * of these flags and the other not does not preclude a merge.
  *
- * VM_SOFTDIRTY - Should not prevent from VMA merging, if we match the flags but
- *                dirty bit -- the caller should mark merged VMA as dirty. If
- *                dirty bit won't be excluded from comparison, we increase
- *                pressure on the memory system forcing the kernel to generate
- *                new VMAs when old one could be extended instead.
- *
- *    VM_STICKY - If one VMA has flags which most be 'sticky', that is ones
- *                which should propagate to all VMAs, but the other does not,
- *                the merge should still proceed with the merge logic applying
- *                sticky flags to the final VMA.
+ * VM_STICKY - If one VMA has flags which most be 'sticky', that is ones
+ *             which should propagate to all VMAs, but the other does not,
+ *             the merge should still proceed with the merge logic applying
+ *             sticky flags to the final VMA.
  */
-#define VM_IGNORE_MERGE (VM_SOFTDIRTY | VM_STICKY)
+#define VM_IGNORE_MERGE VM_STICKY

 /*
  * Flags which should result in page tables being copied on fork. These are
--
2.51.0

[PATCH v2 2/2] testing/selftests/mm: add soft-dirty merge self-test

[Lorenzo Stoakes]

Assert that we correctly merge VMAs containing VM_SOFTDIRTY flags now that
we correctly handle these as sticky.

In order to do so, we have to account for the fact the pagemap interface
checks soft dirty PTEs and additionally that newly merged VMAs are marked
VM_SOFTDIRTY.

We do this by using use unfaulted anon VMAs, establishing one and clearing
references on that one, before establishing another and merging the two
before checking that soft-dirty is propagated as expected.

We check that this functions correctly with mremap() and mprotect() as
sample cases, because VMA merge of adjacent newly mapped VMAs will
automatically be made soft-dirty due to existing logic which does so.

We are therefore exercising other means of merging VMAs.

Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
---
 tools/testing/selftests/mm/soft-dirty.c | 127 +++++++++++++++++++++++-
 1 file changed, 126 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/mm/soft-dirty.c b/tools/testing/selftests/mm/soft-dirty.c
index 4ee4db3750c1..4c73820070ba 100644
--- a/tools/testing/selftests/mm/soft-dirty.c
+++ b/tools/testing/selftests/mm/soft-dirty.c
@@ -184,6 +184,130 @@ static void test_mprotect(int pagemap_fd, int pagesize, bool anon)
 		close(test_fd);
 }

+static void test_merge(int pagemap_fd, int pagesize)
+{
+	char *reserved, *map, *map2;
+
+	/*
+	 * Reserve space for tests:
+	 *
+	 *   ---padding to ---
+	 *   |   avoid adj.  |
+	 *   v     merge     v
+	 * |---|---|---|---|---|
+	 * |   | 1 | 2 | 3 |   |
+	 * |---|---|---|---|---|
+	 */
+	reserved = mmap(NULL, 5 * pagesize, PROT_NONE,
+			MAP_ANON | MAP_PRIVATE, -1, 0);
+	if (reserved == MAP_FAILED)
+		ksft_exit_fail_msg("mmap failed\n");
+	munmap(reserved, 4 * pagesize);
+
+	/*
+	 * Establish initial VMA:
+	 *
+	 *      S/D
+	 * |---|---|---|---|---|
+	 * |   | 1 |   |   |   |
+	 * |---|---|---|---|---|
+	 */
+	map = mmap(&reserved[pagesize], pagesize, PROT_READ | PROT_WRITE,
+		   MAP_ANON | MAP_PRIVATE | MAP_FIXED, -1, 0);
+	if (map == MAP_FAILED)
+		ksft_exit_fail_msg("mmap failed\n");
+
+	/* This will clear VM_SOFTDIRTY too. */
+	clear_softdirty();
+
+	/*
+	 * Now place a new mapping which will be marked VM_SOFTDIRTY. Away from
+	 * map:
+	 *
+	 *       -      S/D
+	 * |---|---|---|---|---|
+	 * |   | 1 |   | 2 |   |
+	 * |---|---|---|---|---|
+	 */
+	map2 = mmap(&reserved[3 * pagesize], pagesize, PROT_READ | PROT_WRITE,
+		    MAP_ANON | MAP_PRIVATE | MAP_FIXED, -1, 0);
+	if (map2 == MAP_FAILED)
+		ksft_exit_fail_msg("mmap failed\n");
+
+	/*
+	 * Now remap it immediately adjacent to map, if the merge correctly
+	 * propagates VM_SOFTDIRTY, we should then observe the VMA as a whole
+	 * being marked soft-dirty:
+	 *
+	 *       merge
+	 *        S/D
+	 * |---|-------|---|---|
+	 * |   |   1   |   |   |
+	 * |---|-------|---|---|
+	 */
+	map2 = mremap(map2, pagesize, pagesize, MREMAP_FIXED | MREMAP_MAYMOVE,
+		      &reserved[2 * pagesize]);
+	if (map2 == MAP_FAILED)
+		ksft_exit_fail_msg("mremap failed\n");
+	ksft_test_result(pagemap_is_softdirty(pagemap_fd, map) == 1,
+			 "Test %s-anon soft-dirty after remap merge 1st pg\n",
+			 __func__);
+	ksft_test_result(pagemap_is_softdirty(pagemap_fd, map2) == 1,
+			 "Test %s-anon soft-dirty after remap merge 2nd pg\n",
+			 __func__);
+
+	munmap(map, 2 * pagesize);
+
+	/*
+	 * Now establish another VMA:
+	 *
+	 *      S/D
+	 * |---|---|---|---|---|
+	 * |   | 1 |   |   |   |
+	 * |---|---|---|---|---|
+	 */
+	map = mmap(&reserved[pagesize], pagesize, PROT_READ | PROT_WRITE,
+		   MAP_ANON | MAP_PRIVATE | MAP_FIXED, -1, 0);
+	if (map == MAP_FAILED)
+		ksft_exit_fail_msg("mmap failed\n");
+
+	/* Clear VM_SOFTDIRTY... */
+	clear_softdirty();
+	/* ...and establish incompatible adjacent VMA:
+	 *
+	 *       -  S/D
+	 * |---|---|---|---|---|
+	 * |   | 1 | 2 |   |   |
+	 * |---|---|---|---|---|
+	 */
+	map2 = mmap(&reserved[2 * pagesize], pagesize,
+	PROT_READ | PROT_WRITE | PROT_EXEC,
+		   MAP_ANON | MAP_PRIVATE | MAP_FIXED, -1, 0);
+	if (map2 == MAP_FAILED)
+		ksft_exit_fail_msg("mmap failed\n");
+
+	/*
+	 * Now mprotect() VMA 1 so it's compatible with 2 and therefore merges:
+	 *
+	 *       merge
+	 *        S/D
+	 * |---|-------|---|---|
+	 * |   |   1   |   |   |
+	 * |---|-------|---|---|
+	 */
+	if (mprotect(map, pagesize, PROT_READ | PROT_WRITE | PROT_EXEC))
+		ksft_exit_fail_msg("mprotect failed\n");
+
+	ksft_test_result(pagemap_is_softdirty(pagemap_fd, map) == 1,
+			 "Test %s-anon soft-dirty after mprotect merge 1st pg\n",
+			 __func__);
+	ksft_test_result(pagemap_is_softdirty(pagemap_fd, map2) == 1,
+			 "Test %s-anon soft-dirty after mprotect merge 2nd pg\n",
+			 __func__);
+
+	munmap(map, 2 * pagesize);
+}
+
 static void test_mprotect_anon(int pagemap_fd, int pagesize)
 {
 	test_mprotect(pagemap_fd, pagesize, true);
@@ -204,7 +328,7 @@ int main(int argc, char **argv)
 	if (!softdirty_supported())
 		ksft_exit_skip("soft-dirty is not support\n");

-	ksft_set_plan(15);
+	ksft_set_plan(19);
 	pagemap_fd = open(PAGEMAP_FILE_PATH, O_RDONLY);
 	if (pagemap_fd < 0)
 		ksft_exit_fail_msg("Failed to open %s\n", PAGEMAP_FILE_PATH);
@@ -216,6 +340,7 @@ int main(int argc, char **argv)
 	test_hugepage(pagemap_fd, pagesize);
 	test_mprotect_anon(pagemap_fd, pagesize);
 	test_mprotect_file(pagemap_fd, pagesize);
+	test_merge(pagemap_fd, pagesize);

 	close(pagemap_fd);

--
2.51.0

Re: [PATCH v2 1/2] mm: propagate VM_SOFTDIRTY on merge

[Andrei Vagin]

On Mon, Nov 17, 2025 at 9:40 AM Lorenzo Stoakes
<lorenzo.stoakes@oracle.com> wrote:
>
> Currently we set VM_SOFTDIRTY when a new mapping is set up (whether by
> establishing a new VMA, or via merge) as implemented in __mmap_complete()
> and do_brk_flags().
>
> However, when performing a merge of existing mappings such as when
> performing mprotect(), we may lose the VM_SOFTDIRTY flag.
>
> This is because currently we simply ignore VM_SOFTDIRTY for the purposes
> of merge, so one VMA may possess the flag and another not, and whichever
> happens to be the target VMA will be the one upon which the merge is
> performed which may or may not have VM_SOFTDIRTY set.
>
> Now we have the concept of 'sticky' VMA flags, let's make VM_SOFTDIRTY one
> which solves this issue.
>
> Additionally update VMA userland tests to propagate changes.
>
> Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
> Suggested-by: Vlastimil Babka <vbabka@suse.cz>
> Acked-by: David Hildenbrand (Red Hat) <david@kernel.org>
> Reviewed-by: Pedro Falcato <pfalcato@suse.de>

Acked-by: Andrey Vagin <avagin@gmail.com>

Thanks,
Andrei

Re: [PATCH v2 1/2] mm: propagate VM_SOFTDIRTY on merge

[Andrew Morton]

On Mon, 17 Nov 2025 17:33:38 +0000 Lorenzo Stoakes <lorenzo.stoakes@oracle.com> wrote:

> Currently we set VM_SOFTDIRTY when a new mapping is set up (whether by
> establishing a new VMA, or via merge) as implemented in __mmap_complete()
> and do_brk_flags().
> 
> However, when performing a merge of existing mappings such as when
> performing mprotect(), we may lose the VM_SOFTDIRTY flag.
> 
> This is because currently we simply ignore VM_SOFTDIRTY for the purposes
> of merge, so one VMA may possess the flag and another not, and whichever
> happens to be the target VMA will be the one upon which the merge is
> performed which may or may not have VM_SOFTDIRTY set.
> 
> Now we have the concept of 'sticky' VMA flags, let's make VM_SOFTDIRTY one
> which solves this issue.
> 
> Additionally update VMA userland tests to propagate changes.
> 

Oh.  This patch messes with the comments which
mm-implement-sticky-vma-flags-fix-2.patch just altered.  Not sure what
you intend here, so I left it as below - please advise.

(also, I fixed a typo: s/most/must/, both files)

--- a/include/linux/mm.h~mm-propagate-vm_softdirty-on-merge
+++ a/include/linux/mm.h
@@ -532,28 +532,28 @@ extern unsigned int kobjsize(const void
  * possesses it but the other does not, the merged VMA should nonetheless have
  * applied to it:
  *
+ *   VM_SOFTDIRTY - if a VMA is marked soft-dirty, that is has not had its
+ *                  references cleared via /proc/$pid/clear_refs, any merged VMA
+ *                  should be considered soft-dirty also as it operates at a VMA
+ *                  granularity.
+ *
  * VM_MAYBE_GUARD - If a VMA may have guard regions in place it implies that
  *                  mapped page tables may contain metadata not described by the
  *                  VMA and thus any merged VMA may also contain this metadata,
  *                  and thus we must make this flag sticky.
  */
-#define VM_STICKY VM_MAYBE_GUARD
+#define VM_STICKY (VM_SOFTDIRTY | VM_MAYBE_GUARD)
 
 /*
  * VMA flags we ignore for the purposes of merge, i.e. one VMA possessing one
  * of these flags and the other not does not preclude a merge.
  *
- * VM_SOFTDIRTY - Should not prevent from VMA merging, if we match the flags but
- *                dirty bit -- the caller should mark merged VMA as dirty. If
- *                dirty bit won't be excluded from comparison, we increase
- *                pressure on the memory system forcing the kernel to generate
- *                new VMAs when old one could be extended instead.
- *
- *    VM_STICKY - When merging VMAs, VMA flags must match, unless they are
- *                'sticky'. If any sticky flags exist in either VMA, we simply
- *                set all of them on the merged VMA.
+ * VM_STICKY - If one VMA has flags which must be 'sticky', that is ones
+ *             which should propagate to all VMAs, but the other does not,
+ *             the merge should still proceed with the merge logic applying
+ *             sticky flags to the final VMA.
  */
-#define VM_IGNORE_MERGE (VM_SOFTDIRTY | VM_STICKY)
+#define VM_IGNORE_MERGE VM_STICKY
 
 /*
  * Flags which should result in page tables being copied on fork. These are
--- a/tools/testing/vma/vma_internal.h~mm-propagate-vm_softdirty-on-merge
+++ a/tools/testing/vma/vma_internal.h
@@ -122,28 +122,23 @@ extern unsigned long dac_mmap_min_addr;
  * possesses it but the other does not, the merged VMA should nonetheless have
  * applied to it:
  *
- * VM_MAYBE_GUARD - If a VMA may have guard regions in place it implies that
- *                  mapped page tables may contain metadata not described by the
- *                  VMA and thus any merged VMA may also contain this metadata,
- *                  and thus we must make this flag sticky.
+ *   VM_SOFTDIRTY - if a VMA is marked soft-dirty, that is has not had its
+ *                  references cleared via /proc/$pid/clear_refs, any merged VMA
+ *                  should be considered soft-dirty also as it operates at a VMA
+ *                  granularity.
  */
-#define VM_STICKY VM_MAYBE_GUARD
+#define VM_STICKY (VM_SOFTDIRTY | VM_MAYBE_GUARD)
 
 /*
  * VMA flags we ignore for the purposes of merge, i.e. one VMA possessing one
  * of these flags and the other not does not preclude a merge.
  *
- * VM_SOFTDIRTY - Should not prevent from VMA merging, if we match the flags but
- *                dirty bit -- the caller should mark merged VMA as dirty. If
- *                dirty bit won't be excluded from comparison, we increase
- *                pressure on the memory system forcing the kernel to generate
- *                new VMAs when old one could be extended instead.
- *
- *    VM_STICKY - When merging VMAs, VMA flags must match, unless they are
- *                'sticky'. If any sticky flags exist in either VMA, we simply
- *                set all of them on the merged VMA.
+ * VM_STICKY - If one VMA has flags which most be 'sticky', that is ones
+ *             which should propagate to all VMAs, but the other does not,
+ *             the merge should still proceed with the merge logic applying
+ *             sticky flags to the final VMA.
  */
-#define VM_IGNORE_MERGE (VM_SOFTDIRTY | VM_STICKY)
+#define VM_IGNORE_MERGE VM_STICKY
 
 /*
  * Flags which should result in page tables being copied on fork. These are
_

Re: [PATCH v2 1/2] mm: propagate VM_SOFTDIRTY on merge

[Lorenzo Stoakes]

On Mon, Nov 17, 2025 at 09:52:52AM -0800, Andrei Vagin wrote:
> On Mon, Nov 17, 2025 at 9:40 AM Lorenzo Stoakes
> <lorenzo.stoakes@oracle.com> wrote:
> >
> > Currently we set VM_SOFTDIRTY when a new mapping is set up (whether by
> > establishing a new VMA, or via merge) as implemented in __mmap_complete()
> > and do_brk_flags().
> >
> > However, when performing a merge of existing mappings such as when
> > performing mprotect(), we may lose the VM_SOFTDIRTY flag.
> >
> > This is because currently we simply ignore VM_SOFTDIRTY for the purposes
> > of merge, so one VMA may possess the flag and another not, and whichever
> > happens to be the target VMA will be the one upon which the merge is
> > performed which may or may not have VM_SOFTDIRTY set.
> >
> > Now we have the concept of 'sticky' VMA flags, let's make VM_SOFTDIRTY one
> > which solves this issue.
> >
> > Additionally update VMA userland tests to propagate changes.
> >
> > Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
> > Suggested-by: Vlastimil Babka <vbabka@suse.cz>
> > Acked-by: David Hildenbrand (Red Hat) <david@kernel.org>
> > Reviewed-by: Pedro Falcato <pfalcato@suse.de>
>
> Acked-by: Andrey Vagin <avagin@gmail.com>

Thanks :)

>
> Thanks,
> Andrei

Re: [PATCH v2 1/2] mm: propagate VM_SOFTDIRTY on merge

[Vlastimil Babka]

On 11/17/25 18:33, Lorenzo Stoakes wrote:
> Currently we set VM_SOFTDIRTY when a new mapping is set up (whether by
> establishing a new VMA, or via merge) as implemented in __mmap_complete()
> and do_brk_flags().
> 
> However, when performing a merge of existing mappings such as when
> performing mprotect(), we may lose the VM_SOFTDIRTY flag.
> 
> This is because currently we simply ignore VM_SOFTDIRTY for the purposes
> of merge, so one VMA may possess the flag and another not, and whichever
> happens to be the target VMA will be the one upon which the merge is
> performed which may or may not have VM_SOFTDIRTY set.
> 
> Now we have the concept of 'sticky' VMA flags, let's make VM_SOFTDIRTY one
> which solves this issue.
> 
> Additionally update VMA userland tests to propagate changes.
> 
> Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
> Suggested-by: Vlastimil Babka <vbabka@suse.cz>
> Acked-by: David Hildenbrand (Red Hat) <david@kernel.org>
> Reviewed-by: Pedro Falcato <pfalcato@suse.de>

Seems it's been like that since 34228d473efe ("mm: ignore VM_SOFTDIRTY on
VMA merging") (unless it was in the meanwhile fixed and broken again) but as
was discussed, not urgent for stable.

Reviewed-by: Vlastimil Babka <vbabka@suse.cz>

Re: [PATCH v2 1/2] mm: propagate VM_SOFTDIRTY on merge

[Lorenzo Stoakes]

On Mon, Nov 17, 2025 at 02:54:13PM -0800, Andrew Morton wrote:
> On Mon, 17 Nov 2025 17:33:38 +0000 Lorenzo Stoakes <lorenzo.stoakes@oracle.com> wrote:
>
> > Currently we set VM_SOFTDIRTY when a new mapping is set up (whether by
> > establishing a new VMA, or via merge) as implemented in __mmap_complete()
> > and do_brk_flags().
> >
> > However, when performing a merge of existing mappings such as when
> > performing mprotect(), we may lose the VM_SOFTDIRTY flag.
> >
> > This is because currently we simply ignore VM_SOFTDIRTY for the purposes
> > of merge, so one VMA may possess the flag and another not, and whichever
> > happens to be the target VMA will be the one upon which the merge is
> > performed which may or may not have VM_SOFTDIRTY set.
> >
> > Now we have the concept of 'sticky' VMA flags, let's make VM_SOFTDIRTY one
> > which solves this issue.
> >
> > Additionally update VMA userland tests to propagate changes.
> >
>
> Oh.  This patch messes with the comments which
> mm-implement-sticky-vma-flags-fix-2.patch just altered.  Not sure what
> you intend here, so I left it as below - please advise.

Ah sorry, missed this!

Yes a couple things to fixup below, sorry to be a pain!

>
> (also, I fixed a typo: s/most/must/, both files)
>
> --- a/include/linux/mm.h~mm-propagate-vm_softdirty-on-merge
> +++ a/include/linux/mm.h
> @@ -532,28 +532,28 @@ extern unsigned int kobjsize(const void
>   * possesses it but the other does not, the merged VMA should nonetheless have
>   * applied to it:
>   *
> + *   VM_SOFTDIRTY - if a VMA is marked soft-dirty, that is has not had its
> + *                  references cleared via /proc/$pid/clear_refs, any merged VMA
> + *                  should be considered soft-dirty also as it operates at a VMA
> + *                  granularity.
> + *
>   * VM_MAYBE_GUARD - If a VMA may have guard regions in place it implies that
>   *                  mapped page tables may contain metadata not described by the
>   *                  VMA and thus any merged VMA may also contain this metadata,
>   *                  and thus we must make this flag sticky.
>   */
> -#define VM_STICKY VM_MAYBE_GUARD
> +#define VM_STICKY (VM_SOFTDIRTY | VM_MAYBE_GUARD)
>
>  /*
>   * VMA flags we ignore for the purposes of merge, i.e. one VMA possessing one
>   * of these flags and the other not does not preclude a merge.
>   *
> - * VM_SOFTDIRTY - Should not prevent from VMA merging, if we match the flags but
> - *                dirty bit -- the caller should mark merged VMA as dirty. If
> - *                dirty bit won't be excluded from comparison, we increase
> - *                pressure on the memory system forcing the kernel to generate
> - *                new VMAs when old one could be extended instead.
> - *
> - *    VM_STICKY - When merging VMAs, VMA flags must match, unless they are
> - *                'sticky'. If any sticky flags exist in either VMA, we simply
> - *                set all of them on the merged VMA.
> + * VM_STICKY - If one VMA has flags which must be 'sticky', that is ones
> + *             which should propagate to all VMAs, but the other does not,
> + *             the merge should still proceed with the merge logic applying
> + *             sticky flags to the final VMA.

Could we fix this up actually? Sorry to be a pain but better to have the new wording:

+ *    VM_STICKY - When merging VMAs, VMA flags must match, unless they are
+ *                'sticky'. If any sticky flags exist in either VMA, we simply
+ *                set all of them on the merged VMA.

>   */
> -#define VM_IGNORE_MERGE (VM_SOFTDIRTY | VM_STICKY)
> +#define VM_IGNORE_MERGE VM_STICKY
>
>  /*
>   * Flags which should result in page tables being copied on fork. These are
> --- a/tools/testing/vma/vma_internal.h~mm-propagate-vm_softdirty-on-merge
> +++ a/tools/testing/vma/vma_internal.h
> @@ -122,28 +122,23 @@ extern unsigned long dac_mmap_min_addr;
>   * possesses it but the other does not, the merged VMA should nonetheless have
>   * applied to it:
>   *
> - * VM_MAYBE_GUARD - If a VMA may have guard regions in place it implies that
> - *                  mapped page tables may contain metadata not described by the
> - *                  VMA and thus any merged VMA may also contain this metadata,
> - *                  and thus we must make this flag sticky.
> + *   VM_SOFTDIRTY - if a VMA is marked soft-dirty, that is has not had its
> + *                  references cleared via /proc/$pid/clear_refs, any merged VMA
> + *                  should be considered soft-dirty also as it operates at a VMA
> + *                  granularity.
>   */
> -#define VM_STICKY VM_MAYBE_GUARD
> +#define VM_STICKY (VM_SOFTDIRTY | VM_MAYBE_GUARD)
>
>  /*
>   * VMA flags we ignore for the purposes of merge, i.e. one VMA possessing one
>   * of these flags and the other not does not preclude a merge.
>   *
> - * VM_SOFTDIRTY - Should not prevent from VMA merging, if we match the flags but
> - *                dirty bit -- the caller should mark merged VMA as dirty. If
> - *                dirty bit won't be excluded from comparison, we increase
> - *                pressure on the memory system forcing the kernel to generate
> - *                new VMAs when old one could be extended instead.
> - *
> - *    VM_STICKY - When merging VMAs, VMA flags must match, unless they are
> - *                'sticky'. If any sticky flags exist in either VMA, we simply
> - *                set all of them on the merged VMA.
> + * VM_STICKY - If one VMA has flags which most be 'sticky', that is ones
> + *             which should propagate to all VMAs, but the other does not,
> + *             the merge should still proceed with the merge logic applying
> + *             sticky flags to the final VMA.

Also could we do the same here?

+ *    VM_STICKY - When merging VMAs, VMA flags must match, unless they are
+ *                'sticky'. If any sticky flags exist in either VMA, we simply
+ *                set all of them on the merged VMA.

>   */
> -#define VM_IGNORE_MERGE (VM_SOFTDIRTY | VM_STICKY)
> +#define VM_IGNORE_MERGE VM_STICKY
>
>  /*
>   * Flags which should result in page tables being copied on fork. These are
> _
>

Thanks, Lorenzo