Jan Capek <jen@xxxxxxxx> writes: > thanks for reply. I think I have tried the -E option at first to see what > the sources looked after the preprocessing step, cuz I was suspecting > some 'void' macro breaking things, however that didn't happen.. The > --save-temps gave me the same result. Below is the relevant part of > the libgcc2.i file. That seems very strange. I have no explanation for that, unless perhaps the compiler was itself miscompiled. > I am wondering if some of the 'define_insn "jump"' or 'define_insn > "indirect_jump"' definitions in the .md file would be responsible for this > behavior. I am really new to all this, so I need to study a few papers > that I found on the web about gcc porting. Before I can make any rough > guess of what is going on. > > Still any input on what to look at in the .md file is very welcome. It can not be anything in the .md file. This error is coming from the frontend. (Unless perhaps this is a bootstrap and something in the .md file in the previous stage is breaking the next stage of the compiler). Ian
