This came out of the investigation done for https://issues.redhat.com/browse/RHEL-120314. One of the tests we were considering upstreaming was an old "unsafe unlink" heap exploit. There is an up-to-date version of this exploit here:
https://github.com/shellphish/how2heap/blob/master/glibc_2.41/unsafe_unlink.c

We should consider working on hardening upstream against this type of exploit. One way suggested by Florian is to obfuscate the fd and bk pointers.

Acceptance Criteria:

• Investigate the practicality and performance cost of said hardening, and if it's worth doing, then:
• submit upstream along with a suitable test case (with multi-threaded variants)
• find reviewers, address any upstream comments, and work towards consensus