emmanuel maillard e.rsz@libertysurf.fr writes:
Sorry my previous mail include a broken patch This one is the good one.
Changelog : Fix spec32 file generation for Darwin
Why are you removing the init code and the SizeOfImage thing? This doesn't look right at all.
init() fini() and end are not part of mach-o. Initialisation are already call by dlopen and previous version from pierre make recall initialisation stuff for all dlls each time you load one. So __wine_register_dll_16 failed due to assertion.
SizeOfImge can be compute like this :
if you want i'll add this for computing image size.
emmanuel
Le 19 avr. 04, à 23:35, Alexandre Julliard a écrit :
emmanuel maillard e.rsz@libertysurf.fr writes:
Sorry my previous mail include a broken patch This one is the good one.
Changelog : Fix spec32 file generation for Darwin
Why are you removing the init code and the SizeOfImage thing? This doesn't look right at all.
-- Alexandre Julliard julliard@winehq.org
This SF.Net email is sponsored by: IBM Linux Tutorials Free Linux tutorial presented by Daniel Robbins, President and CEO of GenToo technologies. Learn everything from fundamentals to system administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click _______________________________________________ Darwine-devel mailing list Darwine-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/darwine-devel
emmanuel maillard e.rsz@libertysurf.fr writes:
init() fini() and end are not part of mach-o. Initialisation are already call by dlopen and previous version from pierre make recall initialisation stuff for all dlls each time you load one. So __wine_register_dll_16 failed due to assertion.
But your way won't work right either, because we need to intercept constructors and run them from the dll entry point, which is the point of the init/fini stuff we are doing there. So you will need to adapt that for whatever constructor mechanism mach-o is using.
SizeOfImge can be compute like this :
[...]
if you want i'll add this for computing image size.
No, I think the current _end hack is much simpler, and good enough for our purposes.
Le 20 avr. 04, à 02:16, Alexandre Julliard a écrit :
emmanuel maillard e.rsz@libertysurf.fr writes:
init() fini() and end are not part of mach-o. Initialisation are already call by dlopen and previous version from pierre make recall initialisation stuff for all dlls each time you load one. So __wine_register_dll_16 failed due to assertion.
But your way won't work right either, because we need to intercept constructors and run them from the dll entry point, which is the point of the init/fini stuff we are doing there. So you will need to adapt that for whatever constructor mechanism mach-o is using.
My way work (at least enought to run WineLib apps), i test it and others people too. For now this is this the only way i find to have Wine work on Darwin. But i'm agree, we need to have a better mach-o support.
emmanuel
-
Alexandre Julliard -
emmanuel maillard