x86/amd_node: Fix PCI device reference counting in amd_smn_init()grafted

The local "root" pointer is a temporary variable used during the device
search. Therefore, refcount related to the search iterators should be cleaned
up after the search is complete.

Use the __free() cleanup macro to ensure the refcount is decremented when the
temporary pointer goes out of scope.

Additionally, increment the refcount when caching a root pointer. This ensures
the in-use refcount is separate from the temporary search refcounting.

Finally, drop the redundant "root = NULL" before the second search loop. The
pci_get_class() iterator always decrements the refcount of its "from"
argument, so the first loop can only fall through with "root" already NULL.

Fixes: 0a4b61d9c2e4 ("x86/amd_node: Fix AMD root device caching")
Closes: https://sashiko.dev/#/patchset/20260806160159.230453-1-jason.andryuk%40amd.com
Reported-by: Sashiko <sashiko-bot@kernel.org>
Assisted-by: LLM
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>
Cc: <stable@kernel.org>
Link: https://patch.msgid.link/20260903154325.74343-1-yazen.ghannam@amd.com

1 files changed, 88 insertions, 0 deletions

diff --git a/include/net/cls_cgroup.h b/include/net/cls_cgroup.h
new file mode 100644
index 000000000..668aeee9b
--- /dev/null
+++ b/include/net/cls_cgroup.h
@@ -0,0 +1,88 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * cls_cgroup.h			Control Group Classifier
+ *
+ * Authors:	Thomas Graf <tgraf@suug.ch>
+ */
+
+#ifndef _NET_CLS_CGROUP_H
+#define _NET_CLS_CGROUP_H
+
+#include <linux/cgroup.h>
+#include <linux/hardirq.h>
+#include <linux/rcupdate.h>
+#include <net/sock.h>
+#include <net/inet_sock.h>
+
+#ifdef CONFIG_CGROUP_NET_CLASSID
+struct cgroup_cls_state {
+	struct cgroup_subsys_state css;
+	u32 classid;
+};
+
+struct cgroup_cls_state *task_cls_state(struct task_struct *p);
+
+static inline u32 task_cls_classid(struct task_struct *p)
+{
+	u32 classid;
+
+	if (in_interrupt())
+		return 0;
+
+	rcu_read_lock();
+	classid = container_of(task_css(p, net_cls_cgrp_id),
+			       struct cgroup_cls_state, css)->classid;
+	rcu_read_unlock();
+
+	return classid;
+}
+
+static inline void sock_update_classid(struct sock_cgroup_data *skcd)
+{
+	u32 classid;
+
+	classid = task_cls_classid(current);
+	sock_cgroup_set_classid(skcd, classid);
+}
+
+static inline u32 __task_get_classid(struct task_struct *task)
+{
+	return task_cls_state(task)->classid;
+}
+
+static inline u32 task_get_classid(const struct sk_buff *skb)
+{
+	u32 classid = __task_get_classid(current);
+
+	/* Due to the nature of the classifier it is required to ignore all
+	 * packets originating from softirq context as accessing `current'
+	 * would lead to false results.
+	 *
+	 * This test assumes that all callers of dev_queue_xmit() explicitly
+	 * disable bh. Knowing this, it is possible to detect softirq based
+	 * calls by looking at the number of nested bh disable calls because
+	 * softirqs always disables bh.
+	 */
+	if (softirq_count()) {
+		struct sock *sk = skb_to_full_sk(skb);
+
+		/* If there is an sock_cgroup_classid we'll use that. */
+		if (!sk || !sk_fullsock(sk))
+			return 0;
+
+		classid = sock_cgroup_classid(&sk->sk_cgrp_data);
+	}
+
+	return classid;
+}
+#else /* !CONFIG_CGROUP_NET_CLASSID */
+static inline void sock_update_classid(struct sock_cgroup_data *skcd)
+{
+}
+
+static inline u32 task_get_classid(const struct sk_buff *skb)
+{
+	return 0;
+}
+#endif /* CONFIG_CGROUP_NET_CLASSID */
+#endif  /* _NET_CLS_CGROUP_H */