I just got this (from utimensat, not utimens, but they share a lot of code) while doing a massively parallel 'make check' of coreutils 8.21 on Linux 3.7.
The filesystem in question is ext4 atop loopback atop LVM atop md (raid1), a fairly typical setup for coreutils test runs, I'd think. (The source tree is ultimately mounted over NFSv3 and gnulib bootstrapped there, but that shouldn't matter because the first thing I do after that is cp -a the whole thing into the loopback filesystem and then configure, build, and test from there.) #0 0x00007f6bd07858f5 in raise () from /lib/libc.so.6 (gdb) bt #0 0x00007f6bd07858f5 in raise () from /lib/libc.so.6 #1 0x00007f6bd07889fb in __GI_abort () at abort.c:92 #2 0x00000000004021e0 in test_utimens (print=print@entry=true, func=0x401890 <do_utimensat>) at test-utimens.h:35 #3 0x0000000000401449 in main () at test-utimensat.c:86 As before, it's not reproducible on demand (e.g. under a debugger).