Closed ctirad closed 2 years ago
I think I hit this problem a while back when I first built fvdi (and I didn't do anything about it - sorry!). PortAsm isn't even available on the platforms that I use (M1 Mac and ARM Linux).,
The problem that I hit is that the buildit.sh script always builds drivers/radeon, and drivers/radeon/Makefile always sets CPU = v4e
. A simple fix would be to only build radeon if the CPU is set to v4e. I'd be happy to raise a PR to do that when I get some time. How does that sound?
FYI the hack that I added to get it to build for me on my Mac:
index ad3b2bf..2ae9972 100755
--- a/buildit.sh
+++ b/buildit.sh
@@ -1,5 +1,7 @@
#!/bin/bash
+set -ex
+
CPU=${CPU:-"020"} # assign defaults if not set
DEBUG=${DEBUG:-"yes"}
@@ -30,15 +32,15 @@ rm -f fvdi_gnu.prg
(cd ${BPL_DIR}; DEBUG=${DEBUG} CPU=${CPU} M68K_ATARI_MINT_CROSS=yes make clean) || exit
(cd ${BPL_DIR}; DEBUG=${DEBUG} CPU=${CPU} M68K_ATARI_MINT_CROSS=yes make -j) || exit
-(cd ${RAD_DIR}; DEBUG=${DEBUG} CPU=${CPU} M68K_ATARI_MINT_CROSS=yes make clean) || exit
-(cd ${RAD_DIR}; DEBUG=${DEBUG} CPU=${CPU} M68K_ATARI_MINT_CROSS=yes make -j) || exit
+# (cd ${RAD_DIR}; DEBUG=${DEBUG} CPU=${CPU} M68K_ATARI_MINT_CROSS=yes make clean) || exit
+# (cd ${RAD_DIR}; DEBUG=${DEBUG} CPU=${CPU} M68K_ATARI_MINT_CROSS=yes make -j) || exit
(cd ${FBE_DIR}; DEBUG=${DEBUG} CPU=${CPU} M68K_ATARI_MINT_CROSS=yes make clean) || exit
(cd ${FBE_DIR}; DEBUG=${DEBUG} CPU=${CPU} M68K_ATARI_MINT_CROSS=yes make -j) || exit
cp ${BPL_DIR}/bitplane.sys .
cp ${ENG_DIR}/fvdi_gnu.prg .
-cp ${RAD_DIR}/radeon.sys .
+# cp ${RAD_DIR}/radeon.sys .
cp ${FBE_DIR}/firebee.sys .
#m68k-atari-mint-strip bitplane.sys
diff --git a/fvdi/drivers/radeon/Makefile b/fvdi/drivers/radeon/Makefile
index 93e8937..0167ce4 100644
--- a/fvdi/drivers/radeon/Makefile
+++ b/fvdi/drivers/radeon/Makefile
@@ -14,7 +14,7 @@ all: $(TARGET)
top_srcdir = ../..
-CPU = v4e
+# CPU = v4e
OPTS = -mshort -O2 -fomit-frame-pointer
include $(top_srcdir)/CONFIGVARS
I'm not quite sure how much of this is needed but I didn't spend much time on it... I just wanted the build to work!
The script was meant to just help myself in a very early stage when we took "reactivated" fVDI development and the Makefiles weren't doing what they are supposed to.
As this is not the case anymore, I would just recommend to get rid of buildit.sh
alltogether.
Ah nice, I was wondering why we have both buildit.sh and a top level makefile. It looks like the CI calls the top level makefile too.
So if I get the time too raise a PR later on, it'll be a PR that deletes buildit.sh (and possibly changes the docs, which I think refer to it).
I raised #42 to remove buildit.sh.
I've tested building the 000 and 020 versions on my Mac M1. (Can't test v4e because no PortAsm :-). Hopefully the CI should do a good job of confirming that it builds on Linux as well.
closed by #42
Please, remove PortAsm dependency for 68k.