> I don't want to ask OpenBSD.org-developers because they always think > they are right. I want to have a talk with developing users which have > experience with compiling to subarchitectures.
As one of the aforementioned developers, who makes mistakes and often acknowledges them (but not all the time. come on. I am a sick bastard, right?), I consider this insulting. Mind you, your problem might simply be that an embedded assembly construct does not compile correctly if -march=i386 because some 486+ specific code is not wrapped with proper ``#if CPU_486''-like defines. I don't know if idea is correct, as you did not paste any error messages. I had added this on my list of ``things to check on a rainy day''. But suddenly I don't see the need to research this problem further. Have a nice day, Miod