Code: Select all
sudo ./mupen64plus --testshots 501 [rom]
Code: Select all
# install required packages pacman -S git freetype2 freeglut glu # download source git clone http://github.com/ricrpi/mupen64plus cd mupen64plus # comment out references to apt-get to stop it failing due to incorrect gcc version sed -i 's/apt-get/#apt-get/g' ./dev_build.sh # substitute gnu stubs-soft.h ln -s /usr/include/gnu/stubs-hard.h /usr/include/gnu/stubs-soft.h # go! ./dev_build.sh
You're right, it was running out of memory. I increased the size of the swap file and it finished compiling successfully. The problem now is that it won't link libmupen64plus.so.2.0.0 as I get the following error:ric_rpi wrote:mashley
under raspbian you need around 750MB RAM + swap, to compile the core. Could you provide the system status that is output when you run ./build.sh please.
From what I have read this is a hard/soft float issue, though I have no idea how to fix it/usr/bin/ld: error: libmupen64plus.so.2.0.0 uses VFP register arguments, /tmp/ccJtECfO.ltrans0.ltrans.o does not
Thanks ric. Line 30 in that file for me reads:ric_rpi wrote:Sounds like you have just hit the issue I need to fix and push back to the official mupen64plus team.
Afraid I can't provide a software update due to the state of mupen64plus on my system however if you open mupen64plus/ricrpi/mupen64plus-core/src/ r4300/new_dynarec/linkage_arm.S and comment out .eabi_attribute 28,1 (line 30), You should then be able to link successfully.
AFAIK yes. It creates a folder called "ricrpi" in the mupen64plus folder when I run dev_build.sh, and I get messages from git like "Downloading core from ricrpi to ./ricrpi/mupen64plus-core".ric_rpi wrote:eabi_attribute 28 is the flag to the linker that states the file is using Hard floating point. Don't know why it is missing in your file as it is currently unmodified in mine. Is the core repository set to use the branch ric_dev?
So should I use build.sh instead of dev_build.sh?./build.sh will checkout the correct branch if it is wrong but ./dev_build.sh doesn't.
Hold on, don't I want it to use hard float? I've got /usr/include/gnu/stubs-hard.h, so presumably gcc is using hard float. So shouldn't I add the flags in to make the compiler use hard float?I can't remember what 24 does but compiling the linkage_arm.S with -S and (--verbose-asm???) will write comments next to the attributes.