Skip to contentSkip to conversation

FAILED: patch "[PATCH] apparmor: fix use-after-free in rawdata dedup loop" failed to apply to 5.10-stable tree

PATCH New [email protected] 2 participants 2 messages Jul 2

Subsystems (3)
lore
Share & tools
lore

[email protected]

author maintainer
#0 lore raw

The patch below does not apply to the 5.10-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.

To reproduce the conflict and resubmit, you may use the following commands:

git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.10.y
git checkout FETCH_HEAD
git cherry-pick -x 6f060496d03e4dc560a40f73770bd08335cb7a27
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<[email protected]>' --in-reply-to '2026070221-sulphate-cauterize-488e@gregkh' --subject-prefix 'PATCH 5.10.y' 'HEAD^..'

Possible dependencies:

thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

From 6f060496d03e4dc560a40f73770bd08335cb7a27 Mon Sep 17 00:00:00 2001
From: Ruslan Valiyev <[email protected]>
Date: Tue, 26 May 2026 00:04:46 +0200
Subject: [PATCH] apparmor: fix use-after-free in rawdata dedup loop

aa_replace_profiles() walks ns->rawdata_list to dedup the incoming policy blob against entries already attached to existing profiles. Per the kernel-doc on struct aa_loaddata, list membership does not hold a reference: profiles hold pcount, and when the last pcount drops, do_ploaddata_rmfs() is queued on a workqueue that takes ns->lock and removes the entry. Between dropping the last pcount and the workqueue running, an entry remains on the list with pcount == 0.

aa_get_profile_loaddata() is an unconditional kref_get() on pcount, so when the dedup loop hits such an entry, refcount hardening reports

refcount_t: addition on 0; use-after-free.

inside aa_replace_profiles(), and the poisoned counter then trips "saturated" and "underflow" warnings on the subsequent uses of the same loaddata.

Before commit a0b7091c4de4 ("apparmor: fix race on rawdata dereference") the dedup path used a get_unless_zero-style helper on a single counter, so the existing "if (tmp)" guard was meaningful. The split-refcount refactor introduced aa_get_profile_loaddata(), which has plain kref_get() semantics, and the guard quietly became a no-op.

Introduce aa_get_profile_loaddata_not0(), matching the existing _not0 convention used by aa_get_profile_not0(), and use it for the rawdata_list dedup lookup so dying entries are skipped.

Reproduced on x86_64 with v7.1-rc5 in QEMU+KVM running Ubuntu 24.04 + stress-ng 0.17.06:

  stress-ng --apparmor 1 --klog-check --timeout 60s

Without this patch the three refcount_t warnings fire within a few seconds. With it the same 60 s run is clean. Coverage is a smoke-test only; a longer soak with CONFIG_KASAN, CONFIG_KCSAN and CONFIG_PROVE_LOCKING would be welcome from anyone with the cycles.

security/apparmor/include/policy_unpack.h +19 -0
index e5a95dc4da1f..b9de0fdf9ee5 100644
@@ -163,6 +163,25 @@ aa_get_profile_loaddata(struct aa_loaddata *data)
 	return data;
 }
 
+/**
+ * aa_get_profile_loaddata_not0 - get a profile reference count if not zero
+ * @data: reference to get a count on
+ *
+ * Like aa_get_profile_loaddata(), but safe to call on an entry that may
+ * be on a list (e.g. ns->rawdata_list) where the last pcount has already
+ * dropped and the deferred cleanup has not yet run.
+ *
+ * Returns: pointer to reference, or %NULL if @data is NULL or its
+ *          profile refcount has already reached zero.
+ */
+static inline struct aa_loaddata *
+aa_get_profile_loaddata_not0(struct aa_loaddata *data)
+{
+	if (data && kref_get_unless_zero(&data->pcount))
+		return data;
+	return NULL;
+}
+
 void __aa_loaddata_update(struct aa_loaddata *data, long revision);
 bool aa_rawdata_eq(struct aa_loaddata *l, struct aa_loaddata *r);
 void aa_loaddata_kref(struct kref *kref);
security/apparmor/policy.c +6 -2
index 847b0ff450c5..b59e827747da 100644
@@ -1230,8 +1230,12 @@ ssize_t aa_replace_profiles(struct aa_ns *policy_ns, struct aa_label *label,
 			if (aa_rawdata_eq(rawdata_ent, udata)) {
 				struct aa_loaddata *tmp;
 
-				tmp = aa_get_profile_loaddata(rawdata_ent);
-				/* check we didn't fail the race */
+				/*
+				 * Entries remain on rawdata_list with
+				 * pcount == 0 until do_ploaddata_rmfs()
+				 * runs; only take a live profile ref.
+				 */
+				tmp = aa_get_profile_loaddata_not0(rawdata_ent);
 				if (tmp) {
 					aa_put_profile_loaddata(udata);
 					udata = tmp;
maintainer
#1 lore raw

PATCH 1/1 apparmor: fix use-after-free in rawdata dedup loop

From: Ruslan Valiyev <[email protected]>

[ Upstream commit 6f060496d03e4dc560a40f73770bd08335cb7a27 ]

aa_replace_profiles() walks ns->rawdata_list to dedup the incoming policy blob against entries already attached to existing profiles. Per the kernel-doc on struct aa_loaddata, list membership does not hold a reference: profiles hold pcount, and when the last pcount drops, do_ploaddata_rmfs() is queued on a workqueue that takes ns->lock and removes the entry. Between dropping the last pcount and the workqueue running, an entry remains on the list with pcount == 0.

aa_get_profile_loaddata() is an unconditional kref_get() on pcount, so when the dedup loop hits such an entry, refcount hardening reports

refcount_t: addition on 0; use-after-free.

inside aa_replace_profiles(), and the poisoned counter then trips "saturated" and "underflow" warnings on the subsequent uses of the same loaddata.

Before commit a0b7091c4de4 ("apparmor: fix race on rawdata dereference") the dedup path used a get_unless_zero-style helper on a single counter, so the existing "if (tmp)" guard was meaningful. The split-refcount refactor introduced aa_get_profile_loaddata(), which has plain kref_get() semantics, and the guard quietly became a no-op.

Introduce aa_get_profile_loaddata_not0(), matching the existing _not0 convention used by aa_get_profile_not0(), and use it for the rawdata_list dedup lookup so dying entries are skipped.

Reproduced on x86_64 with v7.1-rc5 in QEMU+KVM running Ubuntu 24.04 + stress-ng 0.17.06:

  stress-ng --apparmor 1 --klog-check --timeout 60s

Without this patch the three refcount_t warnings fire within a few seconds. With it the same 60 s run is clean. Coverage is a smoke-test only; a longer soak with CONFIG_KASAN, CONFIG_KCSAN and CONFIG_PROVE_LOCKING would be welcome from anyone with the cycles.

---
 security/apparmor/include/policy_unpack.h | 19 +++++++++++++++++++
 security/apparmor/policy.c                |  8 ++++++--
 2 files changed, 25 insertions(+), 2 deletions(-)
security/apparmor/include/policy_unpack.h +19 -0
index 61cc9b5b8a1f36..b6af40885860b4 100644
@@ -122,6 +122,25 @@ aa_get_profile_loaddata(struct aa_loaddata *data)
 	return data;
 }
 
+/**
+ * aa_get_profile_loaddata_not0 - get a profile reference count if not zero
+ * @data: reference to get a count on
+ *
+ * Like aa_get_profile_loaddata(), but safe to call on an entry that may
+ * be on a list (e.g. ns->rawdata_list) where the last pcount has already
+ * dropped and the deferred cleanup has not yet run.
+ *
+ * Returns: pointer to reference, or %NULL if @data is NULL or its
+ *          profile refcount has already reached zero.
+ */
+static inline struct aa_loaddata *
+aa_get_profile_loaddata_not0(struct aa_loaddata *data)
+{
+	if (data && kref_get_unless_zero(&data->pcount))
+		return data;
+	return NULL;
+}
+
 void __aa_loaddata_update(struct aa_loaddata *data, long revision);
 bool aa_rawdata_eq(struct aa_loaddata *l, struct aa_loaddata *r);
 void aa_loaddata_kref(struct kref *kref);
security/apparmor/policy.c +6 -2
index 62ac50db5f803e..46d84d3a099866 100644
@@ -976,8 +976,12 @@ ssize_t aa_replace_profiles(struct aa_ns *policy_ns, struct aa_label *label,
 		if (aa_rawdata_eq(rawdata_ent, udata)) {
 			struct aa_loaddata *tmp;
 
-			tmp = aa_get_profile_loaddata(rawdata_ent);
-			/* check we didn't fail the race */
+			/*
+			 * Entries remain on rawdata_list with
+			 * pcount == 0 until do_ploaddata_rmfs()
+			 * runs; only take a live profile ref.
+			 */
+			tmp = aa_get_profile_loaddata_not0(rawdata_ent);
 			if (tmp) {
 				aa_put_profile_loaddata(udata);
 				udata = tmp;
Signature · 1 line
2.53.0