Re: [PATCH] counter: ti-eqep: fix refcount leak in ti_eqep_probe()

On Thu, Jun 04, 2026 at 09:40:36AM +0000, Wentao Liang wrote:
> After a successful pm_runtime_get_sync(), ti_eqep_probe() can fail
> if devm_clk_get_enabled() returns an error.  In that case the
> runtime PM reference is never released, causing a refcount leak.
> 
> Fix this by adding pm_runtime_put_sync() and pm_runtime_disable()
> calls before returning the error in the clock enable failure path.
> 
> The same cleanup pattern is already used when counter_add() fails,
> so this change makes the error handling consistent.
> 
> Cc: [email protected]
> Fixes: 0cf81c73e4c6 ("counter: ti-eqep: enable clock at probe")
> Signed-off-by: Wentao Liang <[email protected]>
> ---
>  drivers/counter/ti-eqep.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/counter/ti-eqep.c b/drivers/counter/ti-eqep.c
> index d21c157e531a..84fbf500b0f8 100644
> --- a/drivers/counter/ti-eqep.c
> +++ b/drivers/counter/ti-eqep.c
> @@ -548,8 +548,11 @@ static int ti_eqep_probe(struct platform_device *pdev)
>  	pm_runtime_get_sync(dev);
> 
>  	clk = devm_clk_get_enabled(dev, NULL);
> -	if (IS_ERR(clk))
> +	if (IS_ERR(clk)) {
> +		pm_runtime_put_sync(dev);
> +		pm_runtime_disable(dev);
>  		return dev_err_probe(dev, PTR_ERR(clk), "failed to enable clock\n");
> +	}
> 
>  	err = counter_add(counter);
>  	if (err < 0) {
> --
> 2.34.1
> 

Hello Wentao,

This seems similar to another patch we're working on.[^1] I recommend
reviewing and commenting on that thread so we can consolidate there.

Thanks,

William Breathitt Gray

[^1]: https://lore.kernel.org/all/[email protected]/