On Tue, 2013-07-16 at 10:42 +0100, Richard W.M. Jones wrote:
FWIW this change caused a segfault in OpenStack
This phrase is very dramatic. I'd say "triggered a double free in an untested libguestfs error path" is more accurate and less dramatic.
Really it had nothing to do with hard links at all - the same symptom could have happened on ENOSPC for write().