(1)
Create a 1 GiB file instead of 1 TiB, because
apparently, on MacOS, the file (sometimes?) is not
created sparse, and fills up users' disks:
https://github.com/rfjakob/gocryptfs/issues/625
(2)
On darwin, SEEK_DATA is not the same as on Linux
( 2f8b555de2 )
so use the value provided by the unix package.
Our git version is v2+ for some time now, but go.mod
still declared v1. Hopefully making both match makes
https://pkg.go.dev/github.com/rfjakob/gocryptfs/v2 work.
All the import paths have been fixed like this:
find . -name \*.go | xargs sed -i s%github.com/rfjakob/gocryptfs/%github.com/rfjakob/gocryptfs/v2/%
mv is unhappy when we return EPERM when it tries to set
system.posix_acl_access:
mv: preserving permissions for ‘b/x’: Operation not permitted
Now we return EOPNOTSUPP like tmpfs does and mv seems happy.
We relied on the finalizer to close a few fds.
For some reason, this did not cause problems on Linux,
but on MacOS, it causes unmount failures:
umount(/private/tmp/gocryptfs-test-parent/194654785/default-plain): Resource busy -- try 'diskutil unmount'
Due to RMW, we always need read permissions on the backing file. This is a
problem if the file permissions do not allow reading (i.e. 0200 permissions).
This patch works around that problem by chmod'ing the file, obtaining a fd,
and chmod'ing it back.
Test included.
Issue reported at: https://github.com/rfjakob/gocryptfs/issues/125
This test reproduces the problem xfstests generic/124 uncovered.
The warning itself is harmless, but we should either (1) add locking
so that this cannot happen anymore or (2) drop the warning.
Currently fails:
$ go test -v
=== RUN Test1980Tar
--- PASS: Test1980Tar (0.00s)
=== RUN TestCtlSock
--- PASS: TestCtlSock (0.10s)
=== RUN TestOpenTruncateRead
--- PASS: TestOpenTruncateRead (0.00s)
=== RUN TestWORead
--- PASS: TestWORead (0.00s)
=== RUN TestXfs124
cipherSize 18 == header size: interrupted write?
-wpanic turns this warning into a panic: cipherSize 18 == header size: interrupted write?
We do not have to track the writeOnly status because the kernel
will not forward read requests on a write-only FD to us anyway.
I have verified this behavoir manually on a 4.10.8 kernel and also
added a testcase.
Test that we get the right timestamp when extracting a tarball.
Also simplify the workaround in doTestUtimesNano() and fix the
fact that it was running no test at all.
