In the `O_RDONLY` case, we were calling `openat` without a mode
argument. However, it's perfectly legal (albeit unusual) to do
`open(O_RDONLY|O_CREAT)`. One such application that makes use of this is
`flock(1)`.
This was actually caught by `_FORTIFY_SOURCE=2`, and once we run
`rofiles-fuse` with `-f`, the message is clear:
```
*** invalid openat64 call: O_CREAT or O_TMPFILE without mode ***:
rofiles-fuse terminated
======= Backtrace: =========
/lib64/libc.so.6(+0x7c8dc)[0x7f36d9f188dc]
/lib64/libc.so.6(__fortify_fail+0x37)[0x7f36d9fbfaa7]
/lib64/libc.so.6(+0x10019a)[0x7f36d9f9c19a]
rofiles-fuse[0x401768]
...
```
Without `_FORTIFY_SOURCE`, the file gets created, but its mode is
completely random.
I ran into this while investigating
https://github.com/projectatomic/rpm-ostree/pull/1003.
Closes: #1200
Approved by: cgwalters
if ((finfo->flags & O_ACCMODE) == O_RDONLY)
{
/* Read */
- fd = openat (basefd, path, finfo->flags);
+ fd = openat (basefd, path, finfo->flags, mode);
if (fd == -1)
return -errno;
}
setup_test_repository "bare"
-echo "1..7"
+echo "1..8"
cd ${test_tmpdir}
mkdir mnt
assert_file_has_content err.txt "Unable to do hardlink checkout across devices"
echo "ok checkout copy fallback"
+
+# check that O_RDONLY|O_CREAT is handled correctly; used by flock(1) at least
+flock mnt/nonexistent-file echo "ok create file in ro mode"