RE: gcc 4.0.x arm softfloat difficulties

["Dave Korn" <dave dot korn at artimi dot com>]

Lennert Buytenhek wrote:
> On Wed, Jan 04, 2006 at 06:14:15PM -0000, Dave Korn wrote:
> 
>>> I can't really imagine that gcc 4.0.x arm softfloat doesn't work at all
>>> or that nobody is using it, so maybe someone here has an idea what to do
>>> before I proceed to plod on?
>> 
>>   So just out of curiosity (here's where I either say something very
>> very stupid or solve the problem for you), if old versions of gcc didn't
>> have soft float, and you've taken a new version of gcc, and tried to
>> add soft float, and got an error from having _two_ versions of the
>> softfloat functions ... 
>> 
>>  ... doesn't that mean that someone else has _already_ added soft float
>> to gcc, sometime between 3.3 and 4.0.1, and your patch was unneccessary?
> 
> gcc 3.4 comes with softfloat support, but needs the patch that I quoted
> to prevent it from giving an error about undefined symbols.  The end
> product works fine.
> 
> gcc 4.0 also comes with softfloat support, but doesn't work without the
> patch (gives an error about undefined symbols) nor with the patch (gives
> an error about duplicate symbols.)
> 

  Hm.  Adding those things to LIB1ASMFUNCS is the wrong thing, I think,
normally you just add the DPBIT/FPBIT macros.  I don't have the relevant files
to hand, but what happens if you compare the gcc-3.4 version of t-linux
against the gcc-4.0 version of t-linux (without the patch applied); it's
almost certain to be the case that _something_ related has been changed
'beneath your feet' between 3.4 and 4.0.x that is now causing this new
problem.  If you haven't read it yet, the relevant chapter of the gcc
internals docs is 12.1, "Target Makefile Fragments"

    cheers,
      DaveK
-- 
Can't think of a witty .sigline today....



------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com