Few failures, and even fewer than appears (5) as at least one is already fixed.
"Failed test cases:
lib/libarchive/t_libarchive:libarchive, lib/libc/c063/t_utimensat:utimensat_fd, lib/libc/c063/t_utimensat:utimensat_fdcwd, lib/libc/c063/t_utimensat:utimensat_fdlink, usr.sbin/tcpdump/t_tcpdump:promiscuous
Summary for 931 test programs:
9417 passed test cases.
5 failed test cases.
59 expected failed test cases.
692 skipped test cases.
"
Wrong direction; 7 failures this time.
Summary for 931 test programs:
9415 passed test cases.
7 failed test cases.
59 expected failed test cases.
692 skipped test cases.
Of the last 7 failures seen, 5 are covered by open PRs, leaving 2 new tests to review.
I am now running each of them every 3 minutes to see how frequently they pass/fail; so far one "interesting" failure:
$ atf-run fs/vfs/t_vnops:nfs_rename_reg_nodir
tc-end: 1681852322.816669, nfs_rename_reg_nodir, failed, /usr/src/tests/fs/vfs/t_vnops.c:486: sb.st_nlink != 1
This result from the full run may be difficult to reproduce:
mountfuzz8: [1.488390s] Failed: Test program received signal 6 (core dumped)
(hard to hold a wriggly dump)
Found one instance of the source code on a static site:
http://web.mit.edu/freebsd/head/contrib/netbsd-tests/fs/puffs/t_fuzz.c
I postulate this needs a little refactoring:
"Our metric of success is crash / no crash."
Easier to reproduce than I expected.
tc-so:[ 1.0700050] panic: kernel diagnostic assertion "pmp->pmp_status == PUFFSTAT_MOUNTING" failed: file ".../fs/puffs/puffs_vfsops.c", line 330
tc-so:[ 1.0700050] rump kernel halting...
tc-so:halted
tc-se:Test program crashed; attempting to get stack trace
tc-se:Backtrace stopped: previous frame identical to this frame (corrupt stack?)
tc-end: 1681857121.515223, mountfuzz8, failed, Test program received signal 6 (core dumped)
Jim Spath