On 16/03/2023 20:28, Sam James wrote:
# grep -rsin "FAIL:"
/var/tmp/portage/sys-apps/coreutils-9.1_p20230313/temp/build.log
8920:# XFAIL: 0
8921:# FAIL: 2
12462:FAIL: tests/misc/cksum-raw
17659:FAIL: tests/df/df-symlink
36761:# XFAIL: 0
36762:# FAIL: 2
37856:# XFAIL: 0
37857:# FAIL: 0
The cksum-raw issue has already been reported by Bruno so I'm not
worried about that, but df-symlink is a bit odd.
I've put the whole log at tests/df/df-symlink.log at
http://sprunge.us/H2Sr6e.
There are multiple file system roots on that system which may be confusing
things.
Does the attached avoid the issue?
make TESTS=tests/df/df-symlink.sh SUBDIRS=. check
thanks for all the testing,
Pádraig
diff --git a/tests/df/df-symlink.sh b/tests/df/df-symlink.sh
index 34e8a8fed..5d33fddba 100755
--- a/tests/df/df-symlink.sh
+++ b/tests/df/df-symlink.sh
@@ -30,14 +30,17 @@ df --out=source,target symlink > out || fail=1
compare exp out || fail=1
# Ensure we output the same values for device nodes and '.'
-# This was not the case in coreutil-8.22 on systems
+# This was not the case in coreutils-8.22 on systems
# where the device in the mount list was a symlink itself.
# I.e., '.' => /dev/mapper/fedora-home -> /dev/dm-2
# Restrict this test to systems with a 1:1 mapping between
# source and target. This excludes for example BTRFS sub-volumes.
if test "$(df --output=source | grep -F "$file_system" | wc -l)" = 1; then
- df --out=source,target '.' > out || fail=1
- compare exp out || fail=1
+ # Restrict to systems with a single file system root (and have findmnt(1))
+ if test "$(findmnt -n -r -o FSROOT | uniq | wc -l)" = 1; then
+ df --out=source,target '.' > out || fail=1
+ compare exp out || fail=1
+ fi
fi
test "$fail" = 1 && dump_mount_list_
