with "stack smashing detected" crash, is there a way to get a bit more information than this? pastebin.com/tDHsW2va (i.e. any hints on what exactly happened and where?)

I'd suggest trying `::whatis` on the address in the stack trace that looks plausible, and main+0x74::dis to at least work out where it went from there. It's always an issue if the stack has been corrupted.

The ssp protection is usually checking for overflows in fixed arrays on the stack, so code inspection around those is another approach, but it depends on how complicated the code is. You can usually tell if SSP is used on a particular function from the disassembly.

Also try a few bytes less than that address fffffc7feea0cd90::whatis

ok, thanks

yeah, i found the issue, but was wondering if it could say something like "you got that array size wrong" :)

Great. Heh.. that is what is usually means.

You corrupted the stack canary, but that could be a few different ways. However, SSP doesn't get applied to all functions, just ones where the compiler thinks there's something that could be overflowed.

and our ssp implementation in libc deliberately just terminates the process as quickly as possible, hence upanic().

there are some other places we could probably utilize upanic()

Definitely. It has started to appear in some places like bhyve, pcieadm and prtconf already.

andyf: there are options so that it gets applied to every function

back on mainline omnios bloody :D with my change