26-Apr-2024 13:51 GMT.		 UNDER CONSTRUCTION
Anonymous, there are 20 items in your selection
[Files] 68k hosted gcc for x86-AROS portedANN.lu
Posted on 19-May-2004 23:56 GMT by whoosh20 comments
View flat
View list
On a 68k AmigaOS system such as A1200 or WinUAE or Morphos or Cloanto or OS4 or Amithlon, cross compile programs to x86-AROS, via this gcc-3.3.1 cross compiler. Click here This cross compiler was built entirely on 68k WinUAE by combining the existing sources of gcc-3.3.1-aros and 68k hosted gcc-3.3.1-amigaos,

The c compiler part has been tested out on several graphics demos from the archives on the AROS site: blackhole, dawafire, firework,flamme, metaballs, newvox, parallax,

the binaries were generated via 68k WinUAE using this cross compiler and the binaries ran correctly via a i686 compile on my PC.

Hello world also ran correctly. The c++ part of the compiler is fully untested, so any feedback on it will be of interest.

People talk about preventing the AmigaOS alternatives fragmenting, well cross compilers are how you prevent fragmentation: you can generate binaries for machines you dont have.

Previously you needed Linux to develop for AROS although there is also an AROS-native gcc,

68k hosted gcc for x86-AROS ported : Comment 1 of 20ANN.lu
Posted by Anonymous on 19-May-2004 22:50 GMT
Yeah! :)

http://www.openamiga.org

The latest weeks we have seen lots of efforts on the MorphOS side that
will bring benefit to OS4/AROS ports as well. We are one community, no
matter what samface say or other people that wants to separate some
people from others, the Amiga API together with some third party API's
will make it very easy to develop for both AROS, MorphOS, AmigaOS, and
OS4 at the same time! No need to leave someone out! Keep them all
supported! Long live Amiga! :)
68k hosted gcc for x86-AROS ported : Comment 2 of 20ANN.lu
Posted by Crumb // AAT on 20-May-2004 05:04 GMT
Excellent job! I fully agree with you, cross compilers are great to avoid the community fragmentation.
Would it be difficult to have a big gcc distro uploaded to aminet with support for OS3.x,OS4.x,WarpOS,PowerUP,MorphOS,Amithlon,AROS-x86?
I ask this because I'm not sure about how can I have various cross compilers in the same "installation" of gcc.
I tried once with gcc-68k and gcc-Amithlon but it seems to overwrite some files... it would be great if you added a txt explaining how to have all the targets in the same gcc installation.

Thank you for your effort!

Once someone has a gcc distro that compiles for all the targets it may be a good idea to upload it to aminet. If it came with an installer that used fd2pragma to create all the link libs and required includes so newbies had a easier job it would be great...
68k hosted gcc for x86-AROS ported : Comment 3 of 20ANN.lu
Posted by Anonymous on 20-May-2004 06:15 GMT
In reply to Comment 2 (Crumb // AAT):
> Would it be difficult to have a big gcc distro uploaded to aminet with support for OS3.x,OS4.x,WarpOS,PowerUP,MorphOS,Amithlon,AROS-x86?
I ask this because I'm not sure about how can I have various cross compilers in the same "installation" of gcc

You can do that with GoldED as your dev environment. It has differend folders for each gcc and a compiler switcher that unloads the currently used gcc and activates another gcc (making new assigns etc). The active gcc is indiated in the menu and a project can have different makefiles and different options for each gcc, which are selected automatically.
68k hosted gcc for x86-AROS ported : Comment 4 of 20ANN.lu
Posted by Frederik on 20-May-2004 07:48 GMT
great..
I must get back to developing amiga stuff again :)
68k hosted gcc for x86-AROS ported : Comment 5 of 20ANN.lu
Posted by Amon_Re on 20-May-2004 08:17 GMT
Excellent work :)
68k hosted gcc for x86-AROS ported : Comment 6 of 20ANN.lu
Posted by Anonymous on 20-May-2004 14:19 GMT
Wow, this is cool! Could the AROS guys please host this stuff on
their server?
68k hosted gcc for x86-AROS ported : Comment 7 of 20ANN.lu
Posted by Anonymous on 20-May-2004 19:31 GMT
In reply to Comment 2 (Crumb // AAT):
@Crumb

>Excellent job! I fully agree with you, cross compilers are great to avoid
>the community fragmentation.
>Would it be difficult to have a big gcc distro uploaded to aminet with
>support for OS3.x,OS4.x,WarpOS,PowerUP,MorphOS,Amithlon,AROS-x86?
>I ask this because I'm not sure about how can I have various cross compilers
>in the same "installation" of gcc.

with the merging of gcc-3.3.1-aros and gcc-3.3.3-amigaos source that I have
done I can in theory generate cross compilers between any pair of:

68k-amigaos, ppc-amigaos, x86-aros, ppc-aros,

if someone binmails me the source for any gcc between 3.3.1 and 3.3.3
for any of the other AmigaOS variants, I can probably merge that source into this,

currently I only have access to Morphos gcc-2.95.3-5 and Amithlon gcc-2.95.3-4
sources, which are probably too far apart from 3.3.1 to be merge-able,

on the whole 68k hosted cross compilers are the most useful because most
of the above list emulates 68k,

>I tried once with gcc-68k and gcc-Amithlon but it seems to overwrite some
>files...

yes, the default build when installed into gg: will not run!

Firstly 68k "as" gets called which then fails because it doesnt understand -Qy,
also I think for linking 68k "nm" gets called which also fails.
Its a total malfunction!

With this port I managed to hack around this problem, my port will merge
seamlessly directly into gg: so you can switch between 68k-gcc and x86-aros-gcc,

ie gg:bin/gcc and gg:bin/i686-pc-aros-gcc-3.3.1

68k gcc's havent quite got things in the right place, however I have
worked around this problem,

make sure when you merge my port that you skip existing files,
the instructions I give for decompressing .tar.bz2 archives will
skip existing files. The tar -k flag makes sure nothing is overwritten,
I think the only overwrites that will try to happen
are of manual pages. The build generates lib/libiberty.a which does clash,
but I moved it to I think i686-pc-aros/lib

Backup gg: before attempting this so you can restore the unmerged correct
version if necessary.


The shell output generated if you follow the instructions in the readme
will tell you which overwrites were skipped,

I have now on the URL a script called whoosh_gg_startup,

this script allows you to have multi-assign Geekgadgets,

this takes some effort to write but some months ago I spent some hours
carefully studying the Geek startup to improve the startup to a
generalised multiassign version,

so if you have gcc-Amithlon in directory gg_amithlon: and gcc-Morphos in
directory gg_morphos: and your 68k Geekgadgets in gg_68k:

you would type:

whoosh_gg_startup gg_68k:
whoosh_gg_startup gg_morphos:
whoosh_gg_startup gg_amithlon:


if there are clashes of paths then call the script on the directory you
wish to be first in the path search,

so if you want to have a session with gcc-Morphos you would type after
booting up:

whoosh_gg_startup gg_morphos:
whoosh_gg_startup gg_68k:
whoosh_gg_startup gg_amithlon:

and if you wanted a session with gcc-Amithlon type:

whoosh_gg_startup gg_amithlon:
whoosh_gg_startup gg_68k:
whoosh_gg_startup gg_morphos:

You need to reboot then if you want to change gcc,

>it would be great if you added a txt explaining how to have all the
>targets in the same gcc installation.

Maybe I can look into this, I agree its how it should be having all the
gcc's all in the same gg:bin/ directory and all the installs in subdirectories
of gg:


Now I assume you mean that you want eg ppc-morphos-gcc, i686-amithlon-gcc
all in the same directory structure?

whether its possible to have 1 gcc which can generate Morphos + Amithlon + AROS
binaries I dont know. :I will try and ask some gcc experts on this,

>Thank you for your effort!

>Once someone has a gcc distro that compiles for all the targets it may be a
>good idea to upload it to aminet. If it came with an installer that used
>fd2pragma to create all the link libs and required includes so newbies had a
>easier job it would be great...

how do I upload to aminet? I couldnt see an upload link,

For this port you need to get the includes + libs from an archive from the
AROS site, the reason is there are a lot of licenses involved,

rather than delay the upload by some weeks as I try and figure out what
all the license requirements are I decided it was simpler to ask people
to obtain those parts directly from the aros site,
68k hosted gcc for x86-AROS ported : Comment 8 of 20ANN.lu
Posted by whoosh on 20-May-2004 19:40 GMT
In reply to Comment 4 (Frederik):
@Frederik

>great..
>I must get back to developing amiga stuff again :)

making available cross compilers makes such a difference,

the motivation for doing this port was I was so demoralised
by the idea of having to cross compile from Linux!

I thought I would rather attempt at least a port to 68k-amigaos hosted
aros-gcc than go via Linux. Linux probably is the ultimate build platform,
but if you are a true 68k-amigaos person you wont want to go through
Linux. I am really pleased that the build of this gcc itself was
entirely 68k-amigaos hosted, (emulated 68k),
68k hosted gcc for x86-AROS ported : Comment 9 of 20ANN.lu
Posted by whoosh on 20-May-2004 20:02 GMT
In reply to Comment 6 (Anonymous):
>Wow, this is cool! Could the AROS guys please host this stuff on
>their server?

I want them to! it would free up webspace on my site,

However I cannot force them to as currently I am
a 3rd party developer,

whoosh
68k hosted gcc for x86-AROS ported : Comment 10 of 20ANN.lu
Posted by top on 20-May-2004 20:32 GMT
In reply to Comment 9 (whoosh):
is there any way to unify more the developpement of AmigaOS running environnements on PC hardware ?
for example with a web site for AmigaOS on X86 (with different sections WinUAE, AROS, Amithlon, linux UAE)
Is there any chance to see AmitlonOpen and Aros find a common way of developpement especially for apps coded X86/68k ?
Hope that Cloanto will be an important actor to allow AmigaOS4 running on X86 hardware in future.
recompiling AmigaOS4 68k & PPC apps for native 68k/X86 code will be great, and will allow Amiga community get more users from PC markets.
This will bring naturaly more talents to our community and more popularity and more business and more money to improve Amiga technology (software and hardware).

Hope that Hyperion, AmigaInc/Kmos and others will make some efforts to help developpers to improve the port the new AmigaOS technologies on X86 hardware, in parallel of PowerPC hardware developpement.
68k hosted gcc for x86-AROS ported : Comment 11 of 20ANN.lu
Posted by whoosh on 20-May-2004 20:46 GMT
In reply to Comment 1 (Anonymous):
>Yeah! :)
>
> http://www.openamiga.org

>The latest weeks we have seen lots of efforts on the MorphOS side that
>will bring benefit to OS4/AROS ports as well. We are one community, no
>matter what samface say or other people that wants to separate some
>people from others, the Amiga API together with some third party API's
>will make it very easy to develop for both AROS, MorphOS, AmigaOS, and
>OS4 at the same time! No need to leave someone out! Keep them all
>supported! Long live Amiga! :)


imagine if all the alternative Amiga paths combined into one,


maybe through some sort of amnesty or treaty!


that would be such a powerful entity,


Currently I am based on WinUAE, and after one month of having
Windows XP I know for a fact now that AmigaOS is a better system,


I think even AmigaOS 1.2 totally outdoes XP in many respects,


(I concede that XP has good memory protection and rarely crashes)


When I finish a WinUAE session and return to Windows XP its like
stepping out of wonderland onto a drab pavement!


You know Windows XP doesnt even understand ascii files!


If you double click an ascii file on XP, it tells you that Windows XP
has been unable to determine what type of file it is,


do you want XP to consult some Microsoft website to figure out what this file is?


If you say yes, the Microsoft website also doesnt know what the ascii file is!


ascii files confound the MS empire!


It also doesnt understand eg .tar.bz2 files,


IMO the best way to restore your faith in AmigaOS is actually to buy
a current Windows XP machine.


The hardware is theoretically so powerful and the machine is so unresponsive,
if I switch on an external USB drive, I have to wait for literally minutes
as it does an "autoplay" scan of everything. When you switch off the machine
you have to go through a very long shutdown procedure,


XP can only cope with 4 real hard disk partitions (blue in colour), 1 of these can be an
"extended" partition which can be subdivided into any number of "logical" partitions
(green in colour), the design and thinking is so clunky,


To partition a drive you click through a maze of obscure links:


start --> control-panel --> performance&maintanence --> administrative-tools -->
computer-management --> storage --> disk-management


its as if you are discouraged from low level activities,


The CD writer s/w (RecordNow! (I think)) that came with the XP machine
cannot even cope with creating a data disk by specifying a list of
files + directories, it always malfunctions. So the only way to do it is to use

mkisofs from Geekgadgets to convert the files + directories into an
iso image. :the CD writer copes with this alright. It also duplicates CD's
correctly.
68k hosted gcc for x86-AROS ported : Comment 12 of 20ANN.lu
Posted by Anonymous on 20-May-2004 22:44 GMT
In reply to Comment 7 (Anonymous):
> I have now on the URL a script called whoosh_gg_startup, this script allows you to have multi-assign Geekgadgets,

Be careful, multi-assigns sounds like a good idea but some geek gadget tools may not handle those correctly (at least man does not, there might be more programs that traverse folders and fail with multi-assigns).
68k hosted gcc for x86-AROS ported : Comment 13 of 20ANN.lu
Posted by Crumb // AAT on 21-May-2004 05:13 GMT
In reply to Comment 7 (Anonymous):
Mmm OS4 should be merged well because there's a gcc3.3 compiler available... I heard that the diffs were in the main gcc tree but I'm not sure...
I think that there's an option in the assign command to remove assigns so this could avoid the need to reboot the machine to change the target.
The idea of using AROS includes instead of AmigaOS ones is quite good :-) so there won't be any problem uploading it...

To upload stuff to aminet you should connect via ftp and leave the stuff in the upload or incomming directory.
Remember to include a .readme file of the style of other files, something like:
short: GCC compiler for multiple targets (AROS-x86, AmigaOS3...)
Author: Whoosh your@email.com
uploader: you again.... then you put a description of the package above... the main files should be in lha format. I think that it may be a good idea to upload it to AmigaShare too. I hope you get at least some web space in the AROS website, otherwise tell me and we'll upload it to my website...
68k hosted gcc for x86-AROS ported : Comment 14 of 20ANN.lu
Posted by Crumb // AAT on 21-May-2004 05:18 GMT
In reply to Comment 11 (whoosh):
A 68k AROS version would help to get more popularity in the Amiga community because some people will not upgrade to ppc or simply will upgrade later, so having AROS-68k would be quite interesting.

I guess that Olliver (the guy of the ColdFusion accelerator) will find a 68k AROS version really useful because it would help to avoid the need of emulating exec, gaining a lot of speed.

It's not prioritary, but a 68k version would rock :-)
68k hosted gcc for x86-AROS ported : Comment 15 of 20ANN.lu
Posted by whoosh on 21-May-2004 18:57 GMT
In reply to Comment 13 (Crumb // AAT):
>Mmm OS4 should be merged well because there's a gcc3.3 compiler available...
>I heard that the diffs were in the main gcc tree but I'm not sure...


if anyone gets hold of any diffs files for any gcc's for any of
Morphos, OS4, WarpOS...
and you send them to me I will try and merge them with other gcc's,


to synchronise with this gcc-3.3.1-AROS, a version in the range 3.3.1 to 3.3.3
should be merge-able, though any version at all will be useful,


Hyperion say they have a 68k hosted gcc for OS4, however I havent been
able to get hold of it,


>I think that there's an option in the assign command to remove assigns
>so this could avoid the need to reboot the machine to change the target.
>The idea of using AROS includes instead of AmigaOS ones is quite good :-)
>so there won't be any problem uploading it...

in theory yes, but you will need to figure out the license dependencies,

ie there are many external opensource submodules eg libjpeg.a, libpng.a,
etc. Also there will be include files for these, so some of these may
require that you also upload the source code as well.

One way around all this is to upload all the AROS archives, but some are
quite large archives

If you are prepared to trawl through the license conditions and upload
my port complete with the includes and libs, then let me know and
once its uploaded I can replace the downloads with redirection links to
www.aminet on my website.

just 2 conditions: state that I did the port and that people should email me
if they run into problems, and that you are taking responsibility for verifying the licensing conditions,

If you upload to aminet then you can upload the material as .lha archives,
My WinUAE environment cannot deal with file links so all file links are done
with full copies, the advantage is that this will then be installable into any
filesystem at all as-is,

> To upload stuff to aminet you should connect via ftp and leave the stuff in the
> upload or incomming directory.
> Remember to include a .readme file of the style of other files, something like:
> short: GCC compiler for multiple targets (AROS-x86, AmigaOS3...)
> Author: Whoosh your@email.com
> uploader: you again.... then you put a description of the package above...
> the main files should be in lha format.

so thats how you do it!

ok, let me know if you want to complete the license dependency uploads,
I dont have the energy to do this myself but maybe you are motivated enough:
licenses are where peoples paranoia is focussed to the highest extent,
which I find very stressful,

> I think that it may be a good idea to
> upload it to AmigaShare too.

what is AmigaShare?

> I hope you get at least some web space in the
> AROS website, otherwise tel
> me and we'll upload it to my website...


any webspace you can lend would be very useful,


maybe also if I fix some other related cross compiler matters eg I will look into
getting gcc-amithlon to merge into gg: then if you can host this on
your site,


I will also look into getting the existing
68k-hosted gcc-morphos to merge into gg:,
that way developers will be able to churn out binaries for
68k, AROS, Amithlon, Morphos just by changing the variable CC,
(I think)


eg


"CC = -m68020 -s -O2" for 68020 version,


"CC = i686-pc-aros-gcc-3.3.1 -s -O2 -march=i586 " for x86-aros,


"CC = i686be-amithlon-gcc -s -O2 " (probably!) for amithlon,


"CC = ppc-morphos-gcc -s -O2 " for Morphos,


it will be like spinning round with a machine gun,


I dont know what the licensing conditions on the Morphos includes is,


I think I may need to recompile some parts of gcc-amithlon to be able
to merge it into gg:,


gcc-amithlon was the first cross compiler I tried out,
and it was what
made me believe a 68k-hosted gcc-aros was possible, it will be very
useful to get it to run seamlessly from within gg:


perhaps send me an email so I can contact you later on by replying to it!


Note that using your own website gives extra flexibility because you can
modify everything directly,


Currently I think I only have about 2Mb webspace left of my 30Mb allocation,


blueyonder has a facility where they tell you how much free webspace you have
left,


I will try and clear up some webspace by moving more obscure things into
.tar.bz2 format. .tar.bz2 is very space efficient eg
.tar.gz and .lha are very similar but where they might be 13Mb, .tar.bz2
will be eg 10Mb. So .tar.bz2 is a way to gain a lot of extra webspace,
68k hosted gcc for x86-AROS ported : Comment 16 of 20ANN.lu
Posted by whoosh on 21-May-2004 19:45 GMT
In reply to Comment 14 (Crumb // AAT):
>A 68k AROS version would help to get more popularity in the Amiga community
>because some people will not upgrade to ppc or simply will upgrade later,
>so having AROS-68k would be quite interesting.


the idea certainly interests me, but I dont want to make any promises on this,
also perhaps someone else or even a group of people will attempt it, I heard that someone begun on it but abandoned the project,


there is major learning curve between where I stand now and 68k-AROS,


AROS is something you have to approach gradually, for me
step 1. was to create this cross compiler,


once I have the cross compiler it becomes a sort of portal into
x86 AROS: I can transmit all sort of things through this portal to
x86-AROS,


once enough stuff is transmitted through, I can then begin using
the stuff on the other side of the portal,


this will also be true collectively, any stuff that anyone transmits
through the cross compiler then becomes available for use on x86-AROS
to everyone,


for me the 68k-AROS concept will be much further down the line,
what I would do is try and measure up the problem from a distance
as a first step, however I will have to do a lot of other stuff before I can even think about this. If you want to catch a plane the first problem
is to get to the airport, so you may have to figure out how to get to a
railway station first, but to do that you may need to get hold of a taxi
first, but to do that you have to get hold of a phone book first,


eventually though you will be sitting in the aeroplane as it soars
through the clouds up towards the stratosphere,


>I guess that Olliver (the guy of the ColdFusion accelerator) will find a
>68k AROS version really useful because it would help to avoid the need of
>emulating exec, gaining a lot of speed.


having used Windows XP I think a ColdFusion Amiga is actually a very viable
path, one of the advantages is that you would have better use of CPU caches:
I expect that ColdFusion Amiga will be much faster than people expect,
it could be the fastest way to do OS3.1,


what is the URL for the ColdFusion project?


>It's not prioritary, but a 68k version would rock :-)


it would certainly help to break the grip of Amiga Inc on this platform,


now if Amiga Inc were pushing the platform forwards with endless good
ideas I would be completely supporting them, but they arent,
so its just "survival instinct" to back the alternatives,


and as AROS is open source all the OS bugs in 68k AmigaOS could be fixed,


with 68k AmigaOS there are many bugs which remain unfixed even though
they were known to the OS developers in 1992! eg they are documented in
the autodocs


there is a saying: many hands make light work,
68k hosted gcc for x86-AROS ported : Comment 17 of 20ANN.lu
Posted by whoosh on 22-May-2004 20:11 GMT
In reply to Comment 2 (Crumb // AAT):
@Crumb

>I tried once with gcc-68k and gcc-Amithlon but it seems to overwrite some
>files... it would be great if you added a txt explaining how to have all the
>targets in the same gcc installation


I've now looked into this more carefully, what I have found out is:


1. gcc-morphos on www.aminet is incomplete,


2. binutils and gcc of ftp.geekgadgets.org have wrongly named things,


3. It looks like you can merge gcc-amithlon, gcc-morphos, gcc-amigaos,
but only if you download everything from:


http://www.lysator.liu.se/~lcs/files/gg-cross/tar


basically you will have to install over 20 things from this directory,
the amigaos things from this directory appear to be correctly done,


however the archives here are not in gg: format, so for each archive
you will have to decompress it individually, fix any links if you use
a filesystem which doesnt have links, then you would copy a subdirectory,
/opt/gg to gg: watching out for overwrites eg


sh


cd gg-????????/opt/gg


cp -r -p -v -i . /gg


is an interactive copy ie queries you on any repeats,


make a full backup of gg: before attempting this,
I have just begun on this but what I am doing is recompressing
all the archives into gg: format via .tar.bz2, once done I will
then decompress these to gg:,


I use 2 scripts: one decompresses, I then fix by hand if necessary
any links and then the second script creates 2 archives one for gg:
and the other recompresses the docs from ???????/usr to another directory
and then deletes the decompression,

my 68k-x-x86-aros gcc should fit in with either scheme,


At the centre of the problems is that 68k gcc and binutils use bin/as instead of
something like bin/m68k-amigaos-as etc:


binutils-2.9.1-bin.tgz:


bin/addr2line


bin/ar


bin/as


bin/c++filt


bin/gasp


bin/gprof


bin/ld


bin/nm


bin/objcopy


bin/objdump


bin/ranlib


bin/size


bin/strings


bin/strip



possibly just renaming bin/as etc in binutils and some things in
gcc will fix all the problems, not totally sure though,


gcc-2.95.3-4 from ftp.geekgadgets.org has the following wrong names:

bin/c++, bin/cpp, bin/g++, bin/gcc, bin/gcov

I think all these need to be prefixed probably by m68k-amigaos-
so eg bin/g++ would become bin/m68k-amigaos-g++


I will try to find out what the exact changes required are,
68k hosted gcc for x86-AROS ported : Comment 18 of 20ANN.lu
Posted by whoosh on 22-May-2004 20:46 GMT
In reply to Comment 17 (whoosh):
AAAAAAARGHHHH!the archives on http://www.lysator.liu.se/~lcs/files/gg-cross/tarappear to be correct, but they arent 68k-amigaos binaries!so dont do some of what I said!its possible they may be runnable on Amithlon but probably not on 68k,ok, I was just about to waste a lot of hours,I will try and work on this so that gcc-morphos, gcc-amithlon, gcc-amigaoswill all run from the same m68k gg: environment,I may have quite a bit of work to do but I think it will be worth it,BTW I have never managed to get gcc-morphos to function on my setup,though I havent tried very recently,and there doesnt appear to be morphos binutils on www.aminet,though I should be able to rebuild it from available material,
68k hosted gcc for x86-AROS ported : Comment 19 of 20ANN.lu
Posted by whoosh on 23-May-2004 16:26 GMT
In reply to Comment 18 (whoosh):
>BTW I have never managed to get gcc-morphos to function on my setup,>though I havent tried very recently,>and there doesnt appear to be morphos binutils on www.aminet,>though I should be able to rebuild it from available material,well I have managed to port 68k-hosted morphos-ppc binutils,quite smoothly just one small hitch, this is 2.9.1 (I think),so its not the most recent binutils, but it is quite sufficient,Also I have just completed porting 68k-hosted morphos-gcc-2.95.3-5this went totally smoothly, no hitches at all, literally "just add water",its quite time consuming though,There is some further work to do though:1. the install for gcc didnt function, so I need to look into that next,2. I will need to resolve clashes with the 68k environment,gcc-2.95.3 seems much better behaved than gcc-3.3.?Next I will look into gcc-amithlon,I have a request if any Morphos people are reading this:please can you make the Morphos include files and linker librariesdirectly available to everyone, ie outside of mdc and any other registration,ie can you make some download links for this so peoplewho dont use Morphos can cross compilework to Morphos without needing to go through a registration process,"dont look a gift horse in the mouth" so to speak,even if its just the OS3.1 subset of Morphos,none of www.aminet.net, ftp.geekgadgets.org, http://ftp.gnu.org, www.aros.org,require any registration to access their material: you just type a URL,click a link and thats it,if you want to put barbed wire around Morphos specific parts of Morphosthat is fine, but please open up the access to the reimplemented parts,(this is a request not a demand!)BTW when I typed mossdk into www.google.com it asked me if perhaps Imeant mossad!The 2 items that need to be made freely available are:mossdk_devenv_includes.lha for the includes andmossdk_devenv.lha for the linker libraries,thats all I need to make this cross compiler complete,I think the linker libraries are probably just stubs so I dont think there is any security risk,
68k hosted gcc for x86-AROS ported : Comment 20 of 20ANN.lu
Posted by whoosh on 24-May-2004 18:55 GMT
In reply to Comment 19 (whoosh):
re gcc-morphos: good news and bad!the good news is that I can compile source files eg if add.c is:int f( int x , int y ){ return( x + y ) ; }then (with AmigaOS shell prompt %)% gg:bin/ppc-morphos-gcc -O2 -S add.c -o add.s% gg:bin/ppc-morphos-gcc -O2 -c add.c -o add.oppc-morphos-gcc: installation problem, cannot exec `as': No such file or directory% assign c: ports3:binutils291/morphos/install/ppc-morphos/bin ADD% gg:bin/ppc-morphos-gcc -O2 -c add.c -o add.owith the assign it compiles correctly,the assign statement forcing the ppc-morphos/bin/as into the search path,and the contents of add.s are: .file "add.c"gcc2_compiled.: .section ".text" .align 2 .globl f .type f,@functionf: add 3,3,4 blr.Lfe1: .size f,.Lfe1-f .ident "GCC: (GNU) 2.95.3 20010315 (release/lcs-2002-04-12)"entirely generated through rebuilt "equipment",However it looks tricky to integrate this into the 68k environment,as the reference to "as" seems to be hardwired,also without the mossdk_#? files I cannot link this into a program,The install procedure seems to start building other stuff which it cannot complete without the mossdk_#? header files, however gcc gets done without problem,One other thing, I found a source file which if you activate some stuff allows you to have m68k alignment instead of the default different ppc alignment,that way 68k AmigaOS headers could be used as-is, however it wouldnt be consistent with other Morphos things,:so anything outside of classic AmigaOS would have to be rebuilt from scratch,this approach may be more useful for porting, you would have 2 parallel Morphoses,(Morphi?) one with ppc alignment the other with m68k alignment for porting purposes,(dont know what problems you would run into in trying to set this up),gcc is a very geological thing, eg ppc appears to be a customisation of RS6000,
Anonymous, there are 20 items in your selection
Back to Top