On 10/06/2022 18:15, Christophe JAILLET wrote:
Le 29/05/2022 à 10:06, Krzysztof Kozlowski a écrit :
On 27/05/2022 08:54, Yihao Han wrote:
clk_unregister() already checks the clk ptr using !clk || WARN_ON_ONCE(IS_ERR(clk)) so there is no need to check it again before calling it.
No, this explanation does not make sense. clk_unregister() warns and this code is not equivalent.
Best regards, Krzysztof
Hi,
Moreover, as pointed out by greg in [1] on some plateform the assertion in the commit description is wrong. His message is about clk_disable() but, IIUC, it makes sense for clk_unregister() as well. See [2] on the sh plateform.
Yes, this is true as well, although does not have the practical impact on this driver as it uses platforms with common clock framework.
Best regards, Krzysztof