Skip to contentSkip to conversation

mm/rmap: clear vma->anon_vma on error

PATCH Applied Lorenzo Stoakes (Oracle) 5 participants 8 messages Mar 18 – Mar 19

Subsystems (2)
lore
Share & tools
lore

Lorenzo Stoakes (Oracle)

author maintainer
#0 lore raw

PATCH 1/1 mm/rmap: clear vma->anon_vma on error

Commit 542eda1a8329 ("mm/rmap: improve anon_vma_clone(), unlink_anon_vmas() comments, add asserts") alters the way errors are handled, but overlooked one important aspect of clean up.

When a VMA encounters an error state in anon_vma_clone() (that is, on attempted allocation of anon_vma_chain objects), it cleans up partially established state in cleanup_partial_anon_vmas(), before returning an error.

However, this occurs prior to anon_vma->num_active_vmas being incremented, and it also fails to clear the VMA's vma->anon_vma field, which remains in place.

This is immediately an inconsistent state, because anon_vma->num_active_vmas is supposed to track the number of VMAs whose vma->anon_vma field references that anon_vma, and now that count is off-by-negative-1 for each VMA for which this error state has occurred.

When VMAs are unlinked from this anon_vma, unlink_anon_vmas() will eventually underflow anon_vma->num_active_vmas, which will trigger a warning.

This will always eventually happen, as we unlink anon_vma's at process teardown.

It could also cause maybe_reuse_anon_vma() to incorrectly permit the reuse of an anon_vma which has active VMAs attached, which will lead to a persistently invalid state.

The solution is to clear the VMA's anon_vma field when we clean up partial state, as the fact we are doing so indicates clearly that the VMA is not correctly integrated into the anon_vma tree and thus this field is invalid.

---
 mm/rmap.c | 7 +++++++
 1 file changed, 7 insertions(+)
mm/rmap.c +7 -0
index 6398d7eef393..abe4712a220c 100644
@@ -457,6 +457,13 @@ static void cleanup_partial_anon_vmas(struct vm_area_struct *vma)
 		list_del(&avc->same_vma);
 		anon_vma_chain_free(avc);
 	}
+
+	/*
+	 * The anon_vma assigned to this VMA is no longer valid, as we were not
+	 * able to correctly clone AVC state. Avoid inconsistent anon_vma tree
+	 * state by resetting.
+	 */
+	vma->anon_vma = NULL;
 }
 
 /**

--
2.53.0

maintainer Acked-by on v1 1/1
#1 lore raw
Quoted 48 lines · Lorenzo Stoakes (Oracle)

Commit 542eda1a8329 ("mm/rmap: improve anon_vma_clone(), unlink_anon_vmas() comments, add asserts") alters the way errors are handled, but overlooked one important aspect of clean up.

When a VMA encounters an error state in anon_vma_clone() (that is, on attempted allocation of anon_vma_chain objects), it cleans up partially established state in cleanup_partial_anon_vmas(), before returning an error.

However, this occurs prior to anon_vma->num_active_vmas being incremented, and it also fails to clear the VMA's vma->anon_vma field, which remains in place.

This is immediately an inconsistent state, because anon_vma->num_active_vmas is supposed to track the number of VMAs whose vma->anon_vma field references that anon_vma, and now that count is off-by-negative-1 for each VMA for which this error state has occurred.

When VMAs are unlinked from this anon_vma, unlink_anon_vmas() will eventually underflow anon_vma->num_active_vmas, which will trigger a warning.

This will always eventually happen, as we unlink anon_vma's at process teardown.

It could also cause maybe_reuse_anon_vma() to incorrectly permit the reuse of an anon_vma which has active VMAs attached, which will lead to a persistently invalid state.

The solution is to clear the VMA's anon_vma field when we clean up partial state, as the fact we are doing so indicates clearly that the VMA is not correctly integrated into the anon_vma tree and thus this field is invalid.

---
 mm/rmap.c | 7 +++++++
 1 file changed, 7 insertions(+)
mm/rmap.c +0 -0
index 6398d7eef393..abe4712a220c 100644

mm/rmap.c static void cleanup_partial_anon_vmas(struct vm_area_struct *vma) · line 466

Quoted 10 lines
mm/rmap.c +7 -0
@@ -457,6 +457,13 @@ static void cleanup_partial_anon_vmas(struct vm_area_struct *vma)
 		list_del(&avc->same_vma);
 		anon_vma_chain_free(avc);
 	}
+
+	/*
+	 * The anon_vma assigned to this VMA is no longer valid, as we were not
+	 * able to correctly clone AVC state. Avoid inconsistent anon_vma tree
+	 * state by resetting.
+	 */
+	vma->anon_vma = NULL;

LGTM. I was wondering whether anon_vma_clone() should take care of setting up dst->anon_vma. It looks a bit odd in dup_anon_vma.

Signature · 3 lines
Cheers,

David
author maintainer
#2 lore raw
Quoted 3 lines · David Hildenbrand (Arm)

Thanks!

Quoted 2 lines

LGTM. I was wondering whether anon_vma_clone() should take care of setting up dst->anon_vma. It looks a bit odd in dup_anon_vma.

Well in the fork case we don't want to set vma->anon_vma, as we use vma->anon_vma to denote whether or not we reused another anon_vma. For split + remap, we already duplicated it via vm_area_dup() -> vm_area_init_from().

But... now we have enum vma_operation threaded through here, we can just do that there like:

	/* dst is unfaulted, so inherit src's anon_vma. */
	if (operation == VMA_OP_MERGE_UNFAULTED)
		dst->anon_vma = src->anon_vma;

?

Quoted 5 lines

--
Cheers,

David

Cheers, Lorenzo

#3 lore raw
Quoted 15 lines · Lorenzo Stoakes (Oracle)
Quoted 3 lines · David Hildenbrand (Arm)

Thanks!

Quoted 2 lines

LGTM. I was wondering whether anon_vma_clone() should take care of setting up dst->anon_vma. It looks a bit odd in dup_anon_vma.

Well in the fork case we don't want to set vma->anon_vma, as we use vma->anon_vma to denote whether or not we reused another anon_vma. For split + remap, we already duplicated it via vm_area_dup() -> vm_area_init_from().

Quoted 7 lines

But... now we have enum vma_operation threaded through here, we can just do that there like:

	/* dst is unfaulted, so inherit src's anon_vma. */
	if (operation == VMA_OP_MERGE_UNFAULTED)
		dst->anon_vma = src->anon_vma;

Exactly what I had in mind.

Signature · 3 lines
Cheers,

David
author maintainer
#4 lore raw
Quoted 19 lines · David Hildenbrand (Arm)
Quoted 18 lines · Lorenzo Stoakes (Oracle)
Quoted 3 lines · David Hildenbrand (Arm)

Thanks!

Quoted 2 lines

LGTM. I was wondering whether anon_vma_clone() should take care of setting up dst->anon_vma. It looks a bit odd in dup_anon_vma.

Well in the fork case we don't want to set vma->anon_vma, as we use vma->anon_vma to denote whether or not we reused another anon_vma. For split + remap, we already duplicated it via vm_area_dup() -> vm_area_init_from().

But... now we have enum vma_operation threaded through here, we can just do that there like:

Quoted 6 lines
Quoted 4 lines
	/* dst is unfaulted, so inherit src's anon_vma. */
	if (operation == VMA_OP_MERGE_UNFAULTED)
		dst->anon_vma = src->anon_vma;

Exactly what I had in mind.

That's fair, will send a separate patch for that!

Quoted 5 lines

--
Cheers,

David

Cheers, Lorenzo

reviewer Acked-by on v1 1/1
#5 lore raw
In reply to Lorenzo Stoakes (Oracle)
Quoted 33 lines · Lorenzo Stoakes (Oracle)

Commit 542eda1a8329 ("mm/rmap: improve anon_vma_clone(), unlink_anon_vmas() comments, add asserts") alters the way errors are handled, but overlooked one important aspect of clean up.

When a VMA encounters an error state in anon_vma_clone() (that is, on attempted allocation of anon_vma_chain objects), it cleans up partially established state in cleanup_partial_anon_vmas(), before returning an error.

However, this occurs prior to anon_vma->num_active_vmas being incremented, and it also fails to clear the VMA's vma->anon_vma field, which remains in place.

This is immediately an inconsistent state, because anon_vma->num_active_vmas is supposed to track the number of VMAs whose vma->anon_vma field references that anon_vma, and now that count is off-by-negative-1 for each VMA for which this error state has occurred.

When VMAs are unlinked from this anon_vma, unlink_anon_vmas() will eventually underflow anon_vma->num_active_vmas, which will trigger a warning.

This will always eventually happen, as we unlink anon_vma's at process teardown.

It could also cause maybe_reuse_anon_vma() to incorrectly permit the reuse of an anon_vma which has active VMAs attached, which will lead to a persistently invalid state.

The solution is to clear the VMA's anon_vma field when we clean up partial state, as the fact we are doing so indicates clearly that the VMA is not correctly integrated into the anon_vma tree and thus this field is invalid.

Quoted 6 lines

Thanks!

Quoted 24 lines
---
 mm/rmap.c | 7 +++++++
 1 file changed, 7 insertions(+)
mm/rmap.c +7 -1
index 6398d7eef393..abe4712a220c 100644
@@ -457,6 +457,13 @@ static void cleanup_partial_anon_vmas(struct vm_area_struct *vma)
 		list_del(&avc->same_vma);
 		anon_vma_chain_free(avc);
 	}
+
+	/*
+	 * The anon_vma assigned to this VMA is no longer valid, as we were not
+	 * able to correctly clone AVC state. Avoid inconsistent anon_vma tree
+	 * state by resetting.
+	 */
+	vma->anon_vma = NULL;
 }
 
 /**
--

2.53.0

Tested-by on v1 1/1
#6 lore raw
In reply to Lorenzo Stoakes (Oracle)
Quoted 33 lines

Commit 542eda1a8329 ("mm/rmap: improve anon_vma_clone(), unlink_anon_vmas() comments, add asserts") alters the way errors are handled, but overlooked one important aspect of clean up.

When a VMA encounters an error state in anon_vma_clone() (that is, on attempted allocation of anon_vma_chain objects), it cleans up partially established state in cleanup_partial_anon_vmas(), before returning an error.

However, this occurs prior to anon_vma->num_active_vmas being incremented, and it also fails to clear the VMA's vma->anon_vma field, which remains in place.

This is immediately an inconsistent state, because anon_vma->num_active_vmas is supposed to track the number of VMAs whose vma->anon_vma field references that anon_vma, and now that count is off-by-negative-1 for each VMA for which this error state has occurred.

When VMAs are unlinked from this anon_vma, unlink_anon_vmas() will eventually underflow anon_vma->num_active_vmas, which will trigger a warning.

This will always eventually happen, as we unlink anon_vma's at process teardown.

It could also cause maybe_reuse_anon_vma() to incorrectly permit the reuse of an anon_vma which has active VMAs attached, which will lead to a persistently invalid state.

The solution is to clear the VMA's anon_vma field when we clean up partial state, as the fact we are doing so indicates clearly that the VMA is not correctly integrated into the anon_vma tree and thus this field is invalid.

Quoted 6 lines

Thanks!

Quoted 24 lines
---
 mm/rmap.c | 7 +++++++
 1 file changed, 7 insertions(+)
mm/rmap.c +7 -1
index 6398d7eef393..abe4712a220c 100644
@@ -457,6 +457,13 @@ static void cleanup_partial_anon_vmas(struct vm_area_struct *vma)
 		list_del(&avc->same_vma);
 		anon_vma_chain_free(avc);
 	}
+
+	/*
+	 * The anon_vma assigned to this VMA is no longer valid, as we were not
+	 * able to correctly clone AVC state. Avoid inconsistent anon_vma tree
+	 * state by resetting.
+	 */
+	vma->anon_vma = NULL;
 }
 
 /**
--

2.53.0

Acked-by on v1 1/1
#7 lore raw
In reply to Lorenzo Stoakes (Oracle)
Quoted 33 lines · Lorenzo Stoakes (Oracle)

Commit 542eda1a8329 ("mm/rmap: improve anon_vma_clone(), unlink_anon_vmas() comments, add asserts") alters the way errors are handled, but overlooked one important aspect of clean up.

When a VMA encounters an error state in anon_vma_clone() (that is, on attempted allocation of anon_vma_chain objects), it cleans up partially established state in cleanup_partial_anon_vmas(), before returning an error.

However, this occurs prior to anon_vma->num_active_vmas being incremented, and it also fails to clear the VMA's vma->anon_vma field, which remains in place.

This is immediately an inconsistent state, because anon_vma->num_active_vmas is supposed to track the number of VMAs whose vma->anon_vma field references that anon_vma, and now that count is off-by-negative-1 for each VMA for which this error state has occurred.

When VMAs are unlinked from this anon_vma, unlink_anon_vmas() will eventually underflow anon_vma->num_active_vmas, which will trigger a warning.

This will always eventually happen, as we unlink anon_vma's at process teardown.

It could also cause maybe_reuse_anon_vma() to incorrectly permit the reuse of an anon_vma which has active VMAs attached, which will lead to a persistently invalid state.

The solution is to clear the VMA's anon_vma field when we clean up partial state, as the fact we are doing so indicates clearly that the VMA is not correctly integrated into the anon_vma tree and thus this field is invalid.

Quoted 7 lines

---

Signature · 2 lines
Cheers,
Harry / Hyeonggon