[PATCH] clk: xilinx: fix NULL pointer dereference in xvcu_clk_hw_unregister_leaf()

[PATCH] clk: xilinx: fix NULL pointer dereference in xvcu_clk_hw_unregister_leaf()

[Haotian Zhang]

xvcu_clk_hw_unregister_leaf() gets the mux with
clk_hw_get_parent(divider), which returns NULL when the divider has no
parent, and passes it to clk_hw_unregister_mux() without checking it.
The following check tests the wrong variable: divider has already been
verified to be non-NULL at this point, so "if (!divider) return;" is
dead code and a NULL mux is dereferenced, crashing in
clk_hw_unregister_mux().

Only unregister the mux when it is non-NULL, and always release the
divider.

Fixes: 9c789deea206 ("soc: xilinx: vcu: implement clock provider for output clocks")
Assisted-by: DeepSeek-V4.1-Flash
Signed-off-by: Haotian Zhang <vulab@iscas.ac.cn>
---
 drivers/clk/xilinx/xlnx_vcu.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/drivers/clk/xilinx/xlnx_vcu.c b/drivers/clk/xilinx/xlnx_vcu.c
index f14bda375e35..b004c16692d3 100644
--- a/drivers/clk/xilinx/xlnx_vcu.c
+++ b/drivers/clk/xilinx/xlnx_vcu.c
@@ -516,9 +516,8 @@ static void xvcu_clk_hw_unregister_leaf(struct clk_hw *hw)
 		return;
 
 	mux = clk_hw_get_parent(divider);
-	clk_hw_unregister_mux(mux);
-	if (!divider)
-		return;
+	if (mux)
+		clk_hw_unregister_mux(mux);
 
 	clk_hw_unregister_divider(divider);
 }
-- 
2.25.1

Re: [PATCH] clk: xilinx: fix NULL pointer dereference in xvcu_clk_hw_unregister_leaf()

[Krzysztof Kozlowski]

On 09/10/2026 08:18, Haotian Zhang wrote:
> xvcu_clk_hw_unregister_leaf() gets the mux with
> clk_hw_get_parent(divider), which returns NULL when the divider has no
> parent, and passes it to clk_hw_unregister_mux() without checking it.
> The following check tests the wrong variable: divider has already been
> verified to be non-NULL at this point, so "if (!divider) return;" is
> dead code and a NULL mux is dereferenced, crashing in
> clk_hw_unregister_mux().
> 
> Only unregister the mux when it is non-NULL, and always release the
> divider.

So you fired up your OpenClaw or whatever LLM tool and don't bother to
slow down, right?

Best regards,
Krzysztof