https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120513
--- Comment #16 from Greg Chandler <chandleg at wizardsworks dot org> --- It's entirely possible I spent days chasing down the wrong thing, but I wanted to make sure I'm not missing something much farther up stream... I don't think the configure stuff is a distro patch problem, I'm basically having to build an OS from scratch as all of my stuff was too old to build anything new. I started with a running system with glibc 2.1.3 and gcc 2.95. Without going into too much detail I'll say that after a month's work I could do was get that to glibc 2.7 and gcc 4.2.4, so I started over with a basic linux-from-scratch build with a cross compiler. I've got a fully multiuser system built with glibc 2.41 going this route, but due to so many people switching to meson for build, I can't build roughly half of a distribution without a native compiler.... Longstory short, somewhere I think around gcc 4.9 stack protection was introduced, and when trying to compile for alpha it clearly tells you the architecture isn't supported. There is no configure flag to disable that, so no matter what you do the alpha-cross will always spit out: cc1: warning: '-fstack-protector' not supported for this target For every single iteration as it's called. This gets roped into a ton of the configure scripts' (somewhat poorly maintained) checks as errors even though they are not errors, and not on stderr. I do understand some warnings are harbingers of worse to come, but some things are not... I really wish that every time there was a patch/release for gcc full regression tests run against all platforms, all cross platforms, and all candaian cross platforms, just so things don't get kicked down the road for others to find out. I'd probably even be willing to foot the bill for a system dedicated to that exact purpose if there was someone (or group) in charge enough to make that a reality... Anyway: Since I noticed the configures weren't detecting right, for some pretty key things, and I was having to patch the code to fix c and header files in gcc's unpatched sources, I figured I should make sure I am not dealing with something that got munged from that before digging even deeper. So at least now, I can say unless I've missing something so obvious that everyone else has missed it on the configure options it's about time to move away from that and back towards the debugger. Last I checked the cross couldn't compile gdb, I can't remember if it was because of a missing dependency or a dependency that required a meson build that didn't have support for cross compiling. I'll be working on that again tonight... :( Not that most folks reading this would be interested, but the progress for reviving the Slackware Alpha port (what I've been working on) is here: http://wizardsworks.org/alpha/
