Open-source software currently supported on macOS PowerPC

[barracuda156]

wow, I turned down optimization level to O1, disabled fortify_sources in configure at the same line, disabled stack protector by configure switch and it worked!

Mouse kinda weird in gtk3 UI, but I am about co compile qemu 5.0. with SDL ui and see how it perform

View attachment 2639478

Nice, we can probably update the port accordingly then.

 

[Andrew-R]

attached two hacks I used for qemu 5.0.0 and 5.1.0

edit: attached hack for configure where I set O1 instead of O2

./qemu-5.1.0/ppc-softmmu/qemu-system-ppc -hda ~/VMs/osx-tiger_10.4.11_installed.qcow2 -cpu G4 -m 1024 -device usb-tablet -usb

this line seems to work better wrt wild mouse in guest (guest is 10.4.11 Tiger)

 

[barracuda156]

attached two hacks I used for qemu 5.0.0 and 5.1.0

I don’t understand sdl_main_hack. What it does is just to exclude whatever the chunk inside CONFIG_SDL has (assuming that is intended, there is no point to change anything inside that chunk). `#if 1` does nothing, it is always true.

What is the issue about cache_info?

 

[Andrew-R]

I don’t understand sdl_main_hack. What it does is just to exclude whatever the chunk inside CONFIG_SDL has (assuming that is intended, there is no point to change anything inside that chunk). `#if 1` does nothing, it is always true.

What is the issue about cache_info?

Without this sdl_main hack compiler was complaining about function w/o prototype. You surely can simplify it just into not doing its __APPLE__ and __sdl__ dance, not sure, may be it was needed on new osx?

cacheinfo.c was asserting at line 188 at runtime.

 

[barracuda156]

Without this sdl_main hack compiler was complaining about function w/o prototype. You surely can simplify it just into not doing its __APPLE__ and __sdl__ dance, not sure, may be it was needed on new osx?

Excluding SDL block may or may not be required, but two other changes are redundant, since they have no effect on what gets compiled.

cacheinfo.c was asserting at line 188 at runtime.

Drop the assert then?

 

[Andrew-R]

Excluding SDL block may or may not be required, but two other changes are redundant, since they have no effect on what gets compiled.



Drop the assert then?

I have no idea what else might break if I remove this assert .....

 

[Andrew-R]

Compiled qemu 0.9.0 with Rhapsody/OpenSTEP hacks by neozeed on Leopard 10.5.8 ppc32:

GitHub - neozeed/qemu-090-ns: Qemu 0.90 patched for NeXTSTEP/Rhapsody

Qemu 0.90 patched for NeXTSTEP/Rhapsody. Contribute to neozeed/qemu-090-ns development by creating an account on GitHub.

github.com

./configure   --host-cc=gcc-3.3 --cc=gcc-3.3 --target-list=i386-softmmu --enable-sdl --disable-gcc-check --disable-cocoa --extra-cflags="-I/opt/local/include/" --extra-ldflags="-Wl,-framework,CoreAudio"

but it required some hacks, to be attached next (git diff)

 

[Andrew-R]

local hacks for qemu 0.9.0 above

 

[barracuda156]

Compiled qemu 0.9.0 with Rhapsody/OpenSTEP hacks by neozeed on Leopard 10.5.8 ppc32:

GitHub - neozeed/qemu-090-ns: Qemu 0.90 patched for NeXTSTEP/Rhapsody

Qemu 0.90 patched for NeXTSTEP/Rhapsody. Contribute to neozeed/qemu-090-ns development by creating an account on GitHub.

github.com

View attachment 2639609

./configure   --host-cc=gcc-3.3 --cc=gcc-3.3 --target-list=i386-softmmu --enable-sdl --disable-gcc-check --disable-cocoa --extra-cflags="-I/opt/local/include/" --extra-ldflags="-Wl,-framework,CoreAudio"

but it required some hacks, to be attached next (git diff)

This is the first time I see gcc-3.3 being used LOL

 

[Matias_]

I have been trying to install VeraCrypt from MacPorts because I think it has a chance of actually working on OSXPPC. But despite me specifying NOASM=1 NOSSE2=1 makeflags in the portfile, it insists I have an Intel CPU with SSE2 and goes ahead invoking x86 assembly anyway....

 

[barracuda156]

I have been trying to install VeraCrypt from MacPorts because I think it has a chance of actually working on OSXPPC. But despite me specifying NOASM=1 NOSSE2=1 makeflags, it insists I have an Intel CPU with SSE2 and goes ahead invoking x86 assembly anyway....

I tried to build it ages ago and remember it was broken, but forgot why.
As long as it has some generic, non-asm implementation, should be fixable.