emerging -e @world two times led to this (in both): after sandbox (and pax-utils) got their emerge turn, all the other emerges fails with sandbox segmentation fault Between the two emerge -e @world I restored sandbox from a stage4 and re-emerged it with debug symbols (with success) https://gist.github.com/Alessandro-Barbieri/a6f5993d6f8d0ed66f08e9581d0ad7ba#file-emerge_history https://gist.github.com/Alessandro-Barbieri/a6f5993d6f8d0ed66f08e9581d0ad7ba#file-sys-apps-sandbox-2-10-r3-20170503-002558-log https://gist.github.com/Alessandro-Barbieri/a6f5993d6f8d0ed66f08e9581d0ad7ba#file-app-misc-pax-utils-1-1-7-20170503-002145-log https://gist.github.com/Alessandro-Barbieri/a6f5993d6f8d0ed66f08e9581d0ad7ba#file-strace https://gist.github.com/Alessandro-Barbieri/a6f5993d6f8d0ed66f08e9581d0ad7ba#file-gdb-txt https://gist.github.com/Alessandro-Barbieri/a6f5993d6f8d0ed66f08e9581d0ad7ba#file-info_pax-utils https://gist.github.com/Alessandro-Barbieri/a6f5993d6f8d0ed66f08e9581d0ad7ba#file-info_sandbox
now segfaults after every emerge tried removing one flag at time, found that if I remove *both* -lpthread and -lgomp does not segfaults
i've seen this before, but flaky, where an active libsandbox that gets updated starts crashing. but not clear why that's happening -- the copy in memory should be isolated from disk, and we don't pass binary info between it. for future reference, please do not link to logs hosted on external sites. only attach those logs here.
