Messages in this thread | | | From | Steven Stewart-Gallus <> | Date | Mon, 01 Sep 2014 21:19:08 +0000 (GMT) | Subject | How is pivot_root intended to be used? |
| |
Hello,
I am not confused about how I can currently use pivot_root for containers on my kernel (version 3.13). Currently a sequence like:
if (-1 == syscall(__NR_pivot_root, ".", ".")) { perror("pivot_root"); return EXIT_FAILURE; }
if (-1 == umount2(".", MNT_DETACH)) { perror("umount"); return EXIT_FAILURE; }
/* pivot_root() may or may not affect its current working * directory. It is therefore recommended to call chdir("/") * immediately after pivot_root(). * * - http://man7.org/linux/man-pages/man2/pivot_root.2.html */ if (-1 == chdir("/")) { perror("chdir"); return EXIT_FAILURE; }
works. However, I am completely and totally confused about how the pivot_root system call is intended to be used here and have absolutely no idea if this will work in the future. This concerns me because I don't want my program to potentially develop silent security bugs on future versions of the Linux kernel.
Some information about the context. I am sandboxing a program just after it has done a fork, unshared the mount namespace, setup a sandbox directory and changed into it. If you just want to look at the code, the actual code is avaliable at https://gitorious.org/linted/linted/source/b25685ba4762bfb794c8f36ae74276d32d2b0ca8:src/spawn/spawn.c.
Thank you, Steven Stewart-Gallus
| |