[bug report] cpufreq: brcmstb-avs-cpufreq: Fix initial command check

[bug report] cpufreq: brcmstb-avs-cpufreq: Fix initial command check

[Dan Carpenter]

Hello Florian Fainelli,

The patch 22a26cc6a51e: "cpufreq: brcmstb-avs-cpufreq: Fix initial
command check" from May 22, 2019, leads to the following static
checker warning:

	drivers/cpufreq/brcmstb-avs-cpufreq.c:449 brcm_avs_is_firmware_loaded()
	warn: was && intended here instead of ||?

drivers/cpufreq/brcmstb-avs-cpufreq.c
   435  /*
   436   * To ensure the right firmware is running we need to
   437   *    - check the MAGIC matches what we expect
   438   *    - brcm_avs_get_pmap() doesn't return -ENOTSUPP or -EINVAL
   439   * We need to set up our interrupt handling before calling brcm_avs_get_pmap()!
   440   */
   441  static bool brcm_avs_is_firmware_loaded(struct private_data *priv)
   442  {
   443          u32 magic;
   444          int rc;
   445  
   446          rc = brcm_avs_get_pmap(priv, NULL);
   447          magic = readl(priv->base + AVS_MBOX_MAGIC);
   448  
   449          return (magic == AVS_FIRMWARE_MAGIC) && ((rc != -ENOTSUPP) ||
   450                  (rc != -EINVAL));

This is quite an interesting one because originally this was && instead
of || as the static checker suggests.  Obviously "(rc != -ENOTSUPP ||
rc != -EINVAL) is always true because "rc" can't possibly be both values
at the same time.

I see the comment at the top of the function, but I don't understand why
those particular error codes are special.  The temptation for me would
be to do:

	rc = brcm_avs_get_pmap(priv, NULL);
	if (rc)
		return false;

	magic = readl(priv->base + AVS_MBOX_MAGIC);
	return magic == AVS_FIRMWARE_MAGIC;

But I'm clearly missing some information so I'm just going to report
this bug instead.

   451  }

regards,
dan carpenter