WITH_CTF vs -g

[George Neville-Neil]

On 10 Sep 2014, at 15:31, John-Mark Gurney wrote:

> Pedro Giffuni wrote this message on Wed, Sep 10, 2014 at 13:45 -0500:
>> Hi Andriy;
>>
>> Il giorno 10/set/2014, alle ore 12:23, Andriy Gapon <avg at FreeBSD.org> 
>> ha scritto:
>>
>>>
>>> In my opinion WITH_CTF should imply -g in CFLAGS otherwise, as far 
>>> as I can see,
>>> there is nothing to generate CTF data from.  Forcing an end-user to 
>>> remember to
>>> additionally pass -g is not nice.
>>>
>>
>> My understanding is that CTF is meant to be a debugging format 
>> independent of DWARF,
>> so it should be especially useful for the cases where there is no 
>> debugging information.
>
> Except that the CTF data is generated from the DWARF data...  Hence
> why you need to compile w/ -g...  ctfconvert uses the DWARF data to
> make the CTF data...
>
>> Just like Illumos, we haven?t really made much (or any) use of CTF 
>> outside the kernel
>> but now that is an option:
>>
>> http://dtrace.org/blogs/rm/2013/11/14/userland-ctf-in-dtrace/
>>
>>
>>> Also, I think that we can always have -g in CTFFLAGS, because the 
>>> stripping step
>>> takes care of the original DWARF data in any case.  But I am not 
>>> 100% sure about
>>> this.
>>>
>>
>>> What do you think?
>>
>>
>> BTW, it would be nice to see what we can take from the CTF/DDB GSoC 
>> [1]. I understand
>> the BSD-licensed CTF library has advanced greatly but still needs 
>> more work.
>
> Yeh, I need to look at this more too as there are somethings I would
> like to do w/ CTF that I can't because the library we have doesn't
> export all the data..
>

And on the main topic, yes, one should imply the other.  Go for it.

Best,
George