David Bremner <da...@tethera.net> writes: > A helpful sysadmin (hi pabs!) ran the command on the autobuilder as well, and > got > _almost_ the same output. > > It most likely is significant that the gcc version is different. I'll > try upgrading the porterbox chroot and see if it duplicates the > autobuilder failure. >
I managed this earlier than expected, and indeed I can confirm the jit buffer overflow in racketcgc is repeatable on the porterbox (even the address that triggers the error is the same) in my experiments. So in principle I could run racketcgc under gdb, if it would help. Sorry for making this more complicated than necessary, I missed the variation in gcc version. In some sense it's a blessing that the crash is more deterministic under gcc 5.3
