03442: unidasm fails to link on Linux ppc targets

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
03442	Compiling	Minor	Always	Sep 17, 2009, 16:01	Oct 5, 2009, 04:36

Tester	belegdol	View Status	Public	Platform	SDLMAME
Assigned To		Resolution	Unable to reproduce	OS	Linux
Status [?]	Closed			Driver
Version	0.133u4	Fixed in Version		Build	PowerPC
		Fixed in Git Commit		Github Pull Request #

Summary	03442: unidasm fails to link on Linux ppc targets
Description	Compiling src/tools/unidasm.c... src/tools/unidasm.c:261: error: 'cpu_disassemble_powerpc' undeclared here (not in a function) make: *** [obj/sdl/mamed/tools/unidasm.o] Error 1 This happens both with gcc 4.3.2 (Fedora 10) and 4.4.1 (Fedora 11 and rawhide): http://buildsys.rpmfusion.org/logs/fedora-10-rpmfusion_nonfree/4908-sdlmame-0134-0.3.0133u4.fc10/ppc/build.log http://buildsys.rpmfusion.org/logs/fedora-11-rpmfusion_nonfree/4907-sdlmame-0134-0.3.0133u4.fc11/ppc/build.log http://buildsys.rpmfusion.org/logs/fedora-development-rpmfusion_nonfree/4906-sdlmame-0134-0.3.0133u4.fc12/ppc/build.log I checked and the problem is not specific to the debug build.
Steps To Reproduce	make -j4 BIGENDIAN=1 DEBUG=1 SYMBOLS=1 OPTIMIZE=2 'OPT_FLAGS=-O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -DINI_PATH="\"/etc/mame;\""'
Additional Information	Sorry if it's not the right place for ppc-specific bugs, but I wouldn't like this bug to get lost in the multitude of posts in bannister.org forums.
Github Commit

Flags
Regression Version	0.133u4
Affected Sets / Systems

Attached Files

No.04951 Tafoid Administrator Sep 17, 2009, 16:18	Closing. The best way to handle this is the forums. SDL/Linux/PowerPC support is there - not here. The time it should be here is when it can be duplicated in Windows (baseline). AFAIK, it has not been. Work it out with RB and the others there. If this needs to be reopened as a baseline bug, he can do that easily enough.
No.05003 R. Belmont Developer Oct 5, 2009, 04:36	This appears to be a compiler bug - there's no way it shouldn't be able to find that symbol like that.