Re: [Simulavr-devel] OSX compilation

[Bill]

Thanks.

I've recently upgraded too....I'm using GCC 4.0.0 for avr as well, and I
see you patched for the latest binutils.

One thing I found puzzling was that it looks like you disabled warnings
on undefined symbols...where are undefined symbols coming from? I would
have thought that undefined symbols would be pretty serious. Comments?

Thanks again.

Jakob Schwendner wrote:

>Bill,
>
>I have checked the config.h and indeed it has WORDS_BIGENDIAN 1 set.
>Sorry it took me a while to test all this, but I updated to OSX 10.4
>and that included a bit of work, to get all the dev environment back
>up.
>I send you a new patch as well, as I think a few things have changed
>in 10.4 (gcc 4.0 to start with).
>
>to get simulavr working I installed a copy of DarwinPorts and
>installed the following modules:
>port install -k avr-binutils (keep the build files as they are needed)
>port install avr-gcc (I changed the Portfile to not use the version
>suffix on binaries)
>
>get the simulavr cvs
>apply the darwin-patch.1.diff
>./configure address@hidden/work/build
>address@hidden/include
>
>let me know if you need to know anything about the patch. I really
>don't know very much about that autoconf and libtool stuff, so I had
>to fight my way through, and hope its not too dirty.
>
>cheers,
>
>Jakob
>
>
>On 5/5/05, Bill <address@hidden> wrote:
>  
>
>>Thanks.
>>
>>The config.h macro WORDS_BIGENDIAN should be used, I think, to handle
>>the difference in code.
>>
>>Please do:
>>cvs update configure.ac
>>
>>Then run ./bootstrap and configure again....I you should see
>>AC_C_BIGENDIAN added to configure.ac..this is what I see in src/config.h
>>on my intel(little-endian) box:
>>
>>/* Define to 1 if your processor stores words with the most significant byte
>>   first (like Motorola and SPARC, unlike Intel and VAX). */
>>/* #undef WORDS_BIGENDIAN */
>>
>>If you can confirm this works(i.e. you see #define WORDS_BIGENDIAN 1),
>>I'll see that it(i.e. including your patch in proper conditionals if
>>needed) is handled in CVS. (or Klaus will, we'll work that detail out)
>>
>>I'll take a look at your patch. This has been a drawn out process I
>>know. Thank you for your effors Jakob.
>>
>>Thanks,
>>
>>Bill
>>
>>Jakob Schwendner wrote:
>>
>>    
>>
>>>Hey guys,
>>>
>>>finally I got everything working fine.
>>>In my first try I did the endianess swap in the wrong place.
>>>I've done a diff to the current CVS that compiles fine on my 10.3.9 OSX.
>>>There seemed to be a small change in the binutils (2.15-1) as well. I
>>>haven't really tried to do the config changes very variable as I don't
>>>know that much about autoconf and the like.
>>>In the docs it says something about a printf style debug facility. How
>>>can I get that to work?
>>>Anyway, here is the patch.
>>>
>>>
>>>      
>>>
>>    
>>