mtree -c { -O | -X }: Avoid silently ignoring errors from fts(3)

[bdrewery]

Summary

With an only (-O), or excludes (-X), filter it is possible for fts_read(3)
to encounter some error that sets FTS_STOP while looking at an ignored
path. No error is printed because it is not a wanted path. Further reads
return NULL indicating a successful end. Thus later encountered paths
that are expected never are recorded and no error is indicated on stderr
nor for exit status.

While here don't handle jflag logic if this path is ignored.

This seems trivial but I put it out for review as it is hard to test well.

Test Plan

/usr/tests/usr.sbin/nmtree # kyua test
nmtree_test:mtree_check  ->  passed  [0.028s]
nmtree_test:mtree_convert_C  ->  passed  [0.020s]
nmtree_test:mtree_convert_C_S  ->  passed  [0.020s]
nmtree_test:mtree_convert_D  ->  passed  [0.020s]
nmtree_test:mtree_convert_D_S  ->  passed  [0.020s]
nmtree_test:mtree_create  ->  passed  [0.028s]
nmtree_test:mtree_ignore  ->  passed  [0.050s]
nmtree_test:mtree_merge  ->  passed  [0.024s]
nmtree_test:mtree_nonemptydir  ->  passed  [0.028s]
nmtree_test:mtree_specspec_type  ->  passed  [0.035s]
nmtree_test:netbsd6_check  ->  passed  [0.027s]
nmtree_test:netbsd6_convert_C  ->  passed  [0.020s]
nmtree_test:netbsd6_convert_C_S  ->  passed  [0.020s]
nmtree_test:netbsd6_convert_D  ->  passed  [0.019s]
nmtree_test:netbsd6_convert_D_S  ->  passed  [0.019s]
nmtree_test:netbsd6_create  ->  passed  [0.028s]
nmtree_test:netbsd6_ignore  ->  passed  [0.035s]
nmtree_test:netbsd6_merge  ->  passed  [0.023s]
nmtree_test:netbsd6_nonemptydir  ->  passed  [0.027s]

Results file id is usr_tests_usr.sbin_nmtree.20201110-180914-198829
Results saved to /root/.kyua/store/results.usr_tests_usr.sbin_nmtree.20201110-180914-198829.db

19/19 passed (0 failed)

Also used custom kernel failpoints in a downstream kernel (OneFS) to inject EIO.

Expected result:

> cat ~/only
./ifs/blah
./include

> mtree -c -O ~/only -p / -R all -k uid,gid,mode; echo $?
#          user: root
#       machine: bdrewer-lcnn0c9-3
#          tree: /
#          date: Tue Nov 10 10:14:10 2020

# .
/set type=file uid=0 gid=0 mode=0755
.               type=dir

# ./ifs
/set type=file uid=0 gid=0 mode=0644
ifs             type=dir mode=0777
# ./ifs
..


# ./include
/set type=file uid=0 gid=0 mode=0755
include         type=dir
# ./include
..

0

Actual result that lead to this fix:

> mtree -c -O ~/only -p / -R all -k uid,gid,mode; echo $?
#          user: root
#       machine: bdrewer-lcnn0c9-3
#          tree: /
#          date: Tue Nov 10 10:14:11 2020

# .
/set type=file uid=0 gid=0 mode=0755
.               type=dir

# ./ifs
/set type=file uid=0 gid=0 mode=0644
ifs             type=dir mode=0777
0

Behavior now (added 'xx' to the error to ensure hitting the expected one):

> mtree -c -O ~/only -p / -R all -k uid,gid,mode; echo $?
#          user: root
#       machine: bdrewer-lcnn0c9-3
#          tree: /
#          date: Tue Nov 10 10:16:28 2020

# .
/set type=file uid=0 gid=0 mode=0755
.               type=dir

# ./ifs
/set type=file uid=0 gid=0 mode=0644
ifs             type=dir mode=0777
mtree: xx./ifs/home: Input/output error
1

[bdrewery]

In all cases in fts.c, NULL is returned when FTS_STOP is set. The downstream fts is the problem, not FreeBSD or mtree.