Re: [bug report] clk: berlin: Migrate to clk_hw based registration and OF APIs

[Stephen Boyd <sboyd@codeaurora.org>]

On 11/16, Dan Carpenter wrote:
> Hello Stephen Boyd,
> 
> The patch f6475e298297: "clk: berlin: Migrate to clk_hw based
> registration and OF APIs" from Aug 16, 2016, leads to the following
> static checker warning:
> 
> 	drivers/clk/berlin/bg2.c:688 berlin2_clock_setup()
> 	warn: did you mean to pass the address of 'clk_data'
> 
> drivers/clk/berlin/bg2.c
>    674          hws[CLKID_TWD] =
>    675                  clk_hw_register_fixed_factor(NULL, "twd", "cpu", 0, 1, 3);
>    676  
>    677          /* check for errors on leaf clocks */
>    678          for (n = 0; n < MAX_CLKS; n++) {
>    679                  if (!IS_ERR(hws[n]))
>    680                          continue;
>    681  
>    682                  pr_err("%s: Unable to register leaf clock %d\n",
>    683                         np->full_name, n);
>    684                  goto bg2_fail;
>    685          }
>    686  
>    687          /* register clk-provider */
>    688          of_clk_add_hw_provider(np, of_clk_hw_onecell_get, &clk_data);
>                                                                   ^^^^^^^^^
> Pretty sure this was overlooked when we switched to allocating this
> with kmalloc().  Should just be passing clk_data.  Same for the
> following warnings as well:
> 
> drivers/clk/berlin/bg2q.c:385 berlin2q_clock_setup() warn: did you mean to pass the address of 'clk_data'
> drivers/clk/clk-efm32gg.c:85 efm32gg_cmu_init() warn: did you mean to pass the address of 'clk_data'

Thanks! I guess there are new checks in smatch that have been
introduced lately to find these problems? I'll go fix all these
up now and add a Reported-by appropriately.

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