Robert Schuster <[EMAIL PROTECTED]> writes: > Clinton Ebadi schrieb: >>. The FFI problems require a bit of assembly hacking which is >> new to me (ah munging with the stack! The black abyss of horror). > > Are we talking about the same libffi here? AFAIK libffi gained ARM EABI > support by now (last missing bit was the closure API). This should for > example allow GCJ 4.3 on the Neo.
GNU clisp uses ffcall[0]. I investigated building a simple ffcall-libffi-compat, but ffcall has a few features that libffi does not support (varargs is the big one). Someone else attempted[1] to fix ffcall, but he was unable to fix the final issues with stack and register corruption owing to his lack of detailed ARM knowledge. Alas, I too lack this detailed knowledge and so I gave up temporarily after an initial stab at it. The annoying part is that ffcall works fine on OABI ARM. The new (variable rather than fixed) alignment restrictions for structs seem to be causing the breakage. [0] http://savannah.gnu.org/projects/libffcall/ [1] http://mat.exon.name/logs/clisp -- emacsen: "Like... windows are portals man... emacsen: Dude... let's yank this shit out of the kill ring" _______________________________________________ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community