[PATCH 1/1] of: property: Fix device_node cleanup

[Alexander Stein <alexander.stein@ew.tq-group.com>]

'__free(device_node)' attached to remote will cause a double release:
> OF: ERROR: of_node_release() detected bad of_node_put() on
> /soc@0/bus@32c00000/dsi@32e60000

In case remote is to be returned it must not be cleaned up by a call
to of_node_put. The caller has to do that as the documentation mentions.
Partly revert commit b94d24c08ee1a ("of: property: Remove calls to
of_node_put") to fix it.

Fixes: b94d24c08ee1 ("of: property: Remove calls to of_node_put")
Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
---
I get the following trace starting with next-20240529:

OF: ERROR: of_node_release() detected bad of_node_put() on /soc@0/bus@32c00000/dsi@32e60000
CPU: 0 PID: 11 Comm: kworker/u16:0 Not tainted 6.10.0-rc1-next-20240529+ #2511 610eecbb638fc5fab9a27e01b78203e8a379a280
Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL (DT)
Workqueue: events_unbound deferred_probe_work_func
Call trace:
 dump_backtrace+0x90/0x10c
 show_stack+0x14/0x1c
 dump_stack_lvl+0x6c/0x80
 dump_stack+0x14/0x1c
 of_node_release+0x11c/0x188
 kobject_cleanup+0x48/0x17c
 kobject_put+0x78/0xc0
 of_node_put+0x14/0x20
 drm_of_find_panel_or_bridge+0x84/0xd4
 devm_drm_of_get_bridge+0x3c/0x8c
 lcdif_attach_bridge+0x88/0x270
 lcdif_load+0x144/0x28c
 lcdif_probe+0x34/0xcc
 platform_probe+0x64/0xe8
 really_probe+0xc8/0x3ac
 __driver_probe_device+0x84/0x188
 driver_probe_device+0x38/0x150
 __device_attach_driver+0xcc/0x194
 bus_for_each_drv+0x80/0xdc
 __device_attach+0x9c/0x1d0
 device_initial_probe+0x10/0x18
 bus_probe_device+0xa4/0xa8
 deferred_probe_work_func+0x9c/0xe8
 process_one_work+0x154/0x3fc
 worker_thread+0x2f4/0x404
 kthread+0xf4/0x100
 ret_from_fork+0x10/0x20
OF: ERROR: next of_node_put() on this node will result in a kobject warning 'refcount_t: underflow; use-after-free.'
------------[ cut here ]------------
refcount_t: addition on 0; use-after-free.
WARNING: CPU: 0 PID: 11 at lib/refcount.c:25 refcount_warn_saturate+0x150/0x214
Modules linked in:
CPU: 0 PID: 11 Comm: kworker/u16:0 Not tainted 6.10.0-rc1-next-20240529+ #2511 610eecbb638fc5fab9a27e01b78203e8a379a280
Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL (DT)
[...]

 drivers/of/property.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/of/property.c b/drivers/of/property.c
index 83536216ed4fe..e5c6301643e4c 100644
--- a/drivers/of/property.c
+++ b/drivers/of/property.c
@@ -836,9 +836,7 @@ struct device_node *of_graph_get_remote_node(const struct device_node *node,
 {
 	struct device_node *endpoint_node __free(device_node) =
 			    of_graph_get_endpoint_by_regs(node, port, endpoint);
-
-	struct device_node *remote __free(device_node) =
-			    of_graph_get_remote_port_parent(endpoint_node);
+	struct device_node *remote;
 
 	if (!endpoint_node) {
 		pr_debug("no valid endpoint (%d, %d) for node %pOF\n",
@@ -846,6 +844,7 @@ struct device_node *of_graph_get_remote_node(const struct device_node *node,
 		return NULL;
 	}
 
+	remote = of_graph_get_remote_port_parent(endpoint_node);
 	if (!remote) {
 		pr_debug("no valid remote node\n");
 		return NULL;
@@ -853,6 +852,7 @@ struct device_node *of_graph_get_remote_node(const struct device_node *node,
 
 	if (!of_device_is_available(remote)) {
 		pr_debug("not available for remote node\n");
+		of_node_put(remote);
 		return NULL;
 	}
 
-- 
2.34.1