Re: [PATCH 2/2] clk: Fix OOPS calling hlist_del on an already poisoned hlist.

[Stephen Boyd <sboyd@codeaurora.org>]

On 02/05/15 10:24, Alban Browaeys wrote:
> Put is called more than once, thus hlist_del ends up on a poisoned
> list.

Why is clk_put() called more than once on the same clk pointer? Where
does this happe

>
> Move hlist_del to the __clk_release handler managed by kref instead
> of calling it in _clk_put.
>
> Fixes: 1c8e600440c7 ("clk: Add rate constraints to clocks")
> Signed-off-by: Alban Browaeys <prahal@yahoo.com>
> ---
>  drivers/clk/clk.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
> index 8f33722..f1d4b33 100644
> --- a/drivers/clk/clk.c
> +++ b/drivers/clk/clk.c
> @@ -2489,6 +2489,9 @@ static void __clk_release(struct kref *ref)
>  	struct clk *clk = container_of(&core, struct clk, core);
>  	int i = core->num_parents;
>  
> +	hlist_del(&clk->child_node);
> +	clk_core_set_rate_nolock(clk->core, clk->core->req_rate);
> +
>  	kfree(core->parents);
>  	while (--i >= 0)
>  		kfree_const(core->parent_names[i]);
> @@ -2666,8 +2669,6 @@ void __clk_put(struct clk *clk)
>  
>  	clk_prepare_lock();
>  
> -	hlist_del(&clk->child_node);

Ugh we should call this clks_node or something so that we don't confuse
it with child_node in clk_core.

> -	clk_core_set_rate_nolock(clk->core, clk->core->req_rate);
>  	owner = clk->core->owner;
>  	kref_put(&clk->core->ref, __clk_release);
>  


-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project