Re: [PATCH] staging: brcm80211: brcmfmac: Add and use dhd_dbg

["Henry Ptasinski" <henryp@broadcom.com>]

On 05/18/2011 05:32 PM, Joe Perches wrote:
> On Wed, 2011-05-18 at 15:45 -0700, Henry Ptasinski wrote:
>> On 05/18/2011 11:23 AM, Joe Perches wrote:
>>> All uses of DHD_<TYPE>   macros are for debugging only.
>>> Change the multiple uses of DHD_<TYPE>((...)) to dhd_dbg(TYPE, ...)
>>> for a more consistent style.
>> I generally like this approach, but in brcmsmac we've been switching to
>> wiphy_err() and related instead.  Any strong argument for one over the
>> other?
>
> These aren't described as errors but are debugging messages.

Yes, so wiphy_debug(), wiphy_info(), wiphy_notice(), etc could be used. 
  It's not as fine grained as the current logging mechanism, so that 
would be one argument against using wiphy_*.  I don't have a strong 
opinion either way, just looking to see which is the generally preferred 
approach in the kernel (if there is any preference).


>>> -       DHD_TRACE(("%s: Enter\n", __func__));
>>> +       dhd_dbg(TRACE, "%s: Enter\n", __func__);
>> I'd propose moving __func__ into the macro definition itself, which
>> would help ensure consistency (and shorten all the debug lines a bit).
>
> I think TRACE is unnecessary and can be eliminated
> and replaced by the function tracer.
>
> Not all uses use __func__.

That's just sloppiness on our part.  They should all be consistent.

> I think __func__ unnecessary and it should be avoided.

If it's moved into the macro, then there's only one place to change if 
you want to delete "__func__"/add it back/include it conditionally.


> Other than that, I've no objections.
>
>> Also, perhaps rename to "brcm_dbg()", "bcm_dbg()" or something like that
>> and move it into include/bcmutils.h, so brcmsmac can use it as well.
>
> Your choice.

"brcm_dbg()" seems fine to me.

- Henry