Android Native GCC 4.9.2 Build Fails at Dynamic libgcc

[Cyd Haselton]

On Tue, Jan 6, 2015 at 3:34 AM, Andrew Haley <aph@redhat.com> wrote:
> On 05/01/15 20:12, Cyd Haselton wrote:
>>>>>
>>>>> If you run that command with the additional argument "-Wl,-Map,mapfile"
>>>>> you'll get a link map (called "mapfile") which should tell you where
>>>>> dlopen is linked from.
>>>>>
>>>>> Andrew.
>>>>>
>>>> Neither -Wl,-Map,mapfile, -Wl,-Map mapfile or -Wl,-Map=mapfile
>>>> generates said link map
>>>
>>>
>>> Update: The link map isn't generated in the 4.9 build dir because...I
>>> suspect...of the original error.  The link map IS generated in the 4.8
>>> build dir, but grep "dlopen" mapfile yields zero results.  I'm running
>>> make clean && make to see if more helpfulndata is generated
>>
>> And nada.  Zilch.  To be on the safe side I cat-ed through the map
>> file...no reference to dlopen anywhere.  And, as mentioned earlier, no
>> mapfile is generated for the 4.9 libgcc build.
>> I'm going to go back a step and add the -Wl,-Map=mapfile flag to both
>> the 4.9 and 4.8 gcc Makefiles and run make clean && make, but if you
>> or anyone else watching this email thread have a better suggestion
>> please let me know.
>
> Aha!  So 4.8 doesn't call dlopen(), but 4.9 does.
>
> Which input file in 4.9 contains the call to dlopen() ?  See if
> it is different in 4.8.
>
> Andrew.
>

Regarding the generated map file for the gcc build, is it possible
that either a) i'm looking in the wrong place or b) I need to add the
generate map file command to Makefiles in the gcc build subdirs and/or
different places in the gcc Makefile?

I'll look int0 b). Let me know about a)