Alan Hourihane wrote :
I do have a testcase, but I've not delved into it to find why the optimiser is causing the crash. Given that the patch cures it and malloc's memory rather than from thestack, I'm not inclined to go digging much further.
I'm very frightened by that kind of bug, because the problem could be in the MiNT GCC patch. If it is the case, we have to fix it. Or it may be in GCC, in that case we have to make a bug report.
You have a testcase: please send it to me, I'll dig it further. Anyway, your getcwd fix is really a good thing: - it makes the MiNTLib better, by using less stack - it avoids a potential GCC bug -- Vincent Rivière