Re: [PATCH] net: ethernet: ti: cpts: Fix refclk node leak in cpts_of_mux_clk_setup()
From: Simon Horman <horms@kernel.org>
Date: 2026-09-21 11:02:30
Also in:
lkml, stable
On Thu, Sep 17, 2026 at 11:46:16AM +0000, Wentao Liang wrote:
quoted hunk ↗ jump to hunk
cpts_of_mux_clk_setup() takes a reference on the "cpts-refclk-mux" child node that is only released on the mux_fail error path. When the mux clock provider is registered successfully, the reference is leaked. of_clk_add_hw_provider() keeps its own reference on the node, so release the one taken here before returning. Fixes: a3047a81ba13 ("net: ethernet: ti: cpts: add support for ext rftclk selection") Cc: stable@vger.kernel.org Signed-off-by: Wentao Liang <redacted> --- drivers/net/ethernet/ti/cpts.c | 2 ++ 1 file changed, 2 insertions(+)diff --git a/drivers/net/ethernet/ti/cpts.c b/drivers/net/ethernet/ti/cpts.c index 2ba4c8795d60..372cef13d81e 100644 --- a/drivers/net/ethernet/ti/cpts.c +++ b/drivers/net/ethernet/ti/cpts.c@@ -714,6 +714,8 @@ static int cpts_of_mux_clk_setup(struct cpts *cpts, struct device_node *node) goto mux_fail; } + of_node_put(refclk_np); + return ret;
I believe that of_node_put() should not be added here, but rather it should be added to cpts_clk_del_provider(). Otherwise, what ensures that the node hasn't been reclaimed by the time cpts_clk_del_provider() runs? -- pw-bot: changes-requested