Dead Hackers Society
Other BBSes »
 
 
Demoscene  Coding  CT60  Buy/sell

Atari coding BBS
 
Re: GCC 4.1.2 for MiNT....
Posted by: KeithS Apr,01.2007-16:10 

First off, the patch adds support for the Atari/TOS/MiNT target, much like the previous GCC3 patches. Some of the architecture within GCC has changed, so some juggling of how the original patch was written had to be done, mostly macros GCC has removed from use and replaced with newer ones.

Next, I dropped support for building the mshort libraries. Main reason, it reduced the build time for GCC4 greatly, since you need to build support for it for every target. I might be saying this wrong, but when you build GCC, old way, you built support libs for 68000, mshort, 68000/mshort, 68020, 68020/mshort, 68020-60, 68020-60/mshort. I dropped mshort, which I am still not 100% sure we can do (based on building the MiNT kernel, should not matter otherwise, except for the inability to link libs with PureC). I have been unable to build the kernel because cvs right now is setup for building a Coldfire target, which I do not support and I have been too lazy to debug kernal's make process to disable it (It can be easily added to GCC I think, Coldfire support that is).

Lastly, 2 problems in libiberty are commented out (not fixed, since I do not know the cause yet) and then a fix in libsdtc++ compile where, for some reason, GCC was not carrying over something it discovered in configure.

I can't think of much else, the 3.3.6 patches were applied, and changed/fixed where they did apply correctly.

The *ONLY* problems I have had so far is a problem with using -floop-optimize2 for an optimization, the compiler ICEs. Since this optimization is still being developed, I do not believe this is a big issue yet. Otherwise, the compiler did not build MiNTLib because it was unhappy with the use of the register a6, but I need to examine that further.

I have a built, mostly functioning ScummVM that I might post. SDL was built using -O3 -floop-optimize2 -funroll-loops -fvariable-expansion-in-unroller except where the loop optimizer caused the compiler to ICE, then that source was just built with -O3. ScummVM was built with the same parameters, except with g++ instead of GCC. It did have the same issue a few times. And this build of ScummVM had everything, all engines enabled (including unsupported one) OGG/MP3 support (Useless right now bevause they are CPU hungery) and the high quality scalers (which do not compile with GCC 2.95.3). Made a large, 7.5 Meg executable after being stripped....







[All messages in this thread]    [Start new thread]

Topic Posted by  Date 
GCC 4.1.2 for MiNT.... KeithS Mar,30.2007-21:02
  Re: GCC 4.1.2 for MiNT.... mikro Apr,01.2007-12:21
    Re: GCC 4.1.2 for MiNT.... KeithS Apr,01.2007-16:10
      Re: GCC 4.1.2 for MiNT.... mikro Apr,02.2007-10:33
        Re: GCC 4.1.2 for MiNT.... KeithS Apr,02.2007-18:59
  Re: GCC 4.1.2 for MiNT.... ggn Apr,02.2007-17:25
  Re: GCC 4.1.2 for MiNT.... Peter Apr,02.2007-18:55
    Re: GCC 4.1.2 for MiNT.... KeithS Apr,02.2007-21:02
      Re: GCC 4.1.2 for MiNT.... Peter Apr,03.2007-16:43


Reply to this message
Name:
Topic:
Anti-troll code:

Message:

What's the anti-troll code?
That's your personal code to be able to add comments and messages on the dhs.nu site.
Don't have a code or forgot it? Fix it here.
© 1994-2024 Dead Hackers Society Contact: Anders Eriksson