Age | Commit message (Collapse) | Author |
|
Directly accessing the Nodes does not work properly,
as there is no way to attach a newly LOOKUPped Node
to the tree. This means Path() does not work.
Use an actual mount instead and walk the tree.
|
|
|
|
|
|
Horribly broken, but it compiles.
.../tests/fsck$ ./run_fsck.bash
Reading password from extpass program "echo", arguments: ["test"]
Decrypting master key
OpenDir ".": invalid entry "invalid_file_name.3": illegal base64 data at input byte 17
OpenDir ".": invalid entry "invalid_file_name_2": bad message
fsck: corrupt entry in dir "": "invalid_file_name.3"
fsck: corrupt entry in dir "": "invalid_file_name_2"
OpenDir ".": invalid entry "invalid_file_name____1": bad message
fsck: corrupt entry in dir "": "invalid_file_name____1"
fsck: error stating file ".go-fuse.5577006791947779410/deleted": no such file or directory
fsck: error listing xattrs on ".go-fuse.13260572831089785859/deleted": no such file or directory
fsck: error opening dir "i10488239 (dir): ": no such file or directory
fsck: error reading symlink ".go-fuse.10667007354186551956/deleted": no such file or directory
fsck: error listing xattrs on ".go-fuse.11998794077335055257/deleted": no such file or directory
[...]
|
|
Helpful for fsck.
|
|
|
|
gocryptfs/tests/xattr passes.
|
|
|
|
Chasing a bug that seems to have nothing to do
with magic names, as it already triggers during
warmup:
--- FAIL: TestMagicNames (0.00s)
matrix_test.go:773: Testing n="warmup1"
matrix_test.go:773: Testing n="warmup2"
matrix_test.go:820: no such file or directory
|
|
Don't let go-fuse mess úp 0000 permissions.
|
|
|
|
We used to always open a *File2 and letting the *File2
handle Setattr. This does not work it we cannot open the file!
Before:
$ go test
2020/07/12 20:14:57 writer: Write/Writev failed, err: 2=no such file or directory. opcode: INTERRUPT
2020/07/12 20:14:57 writer: Write/Writev failed, err: 2=no such file or directory. opcode: INTERRUPT
--- FAIL: TestLchown (0.00s)
matrix_test.go:634: lchown /tmp/gocryptfs-test-parent-1026/014500839/default-plain/symlink: too many levels of symbolic links
touch: setting times of '/tmp/gocryptfs-test-parent-1026/014500839/default-plain/utimesnano_symlink': Too many levels of symbolic links
--- FAIL: TestUtimesNanoSymlink (0.00s)
matrix_test.go:655: exit status 1
--- FAIL: TestMkfifo (0.00s)
matrix_test.go:755: file exists
--- FAIL: TestMagicNames (0.00s)
matrix_test.go:773: Testing n="gocryptfs.longname.QhUr5d9FHerwEs--muUs6_80cy6JRp89c1otLwp92Cs"
matrix_test.go:773: Testing n="gocryptfs.diriv"
matrix_test.go:815: open /tmp/gocryptfs-test-parent-1026/014500839/default-plain/linktarget: permission denied
--- FAIL: TestChmod (0.00s)
matrix_test.go:840: chmod 444 -> 000 failed: permission denied
matrix_test.go:840: chmod 444 -> 111 failed: permission denied
matrix_test.go:840: chmod 444 -> 123 failed: permission denied
matrix_test.go:840: chmod 444 -> 321 failed: permission denied
FAIL
exit status 1
FAIL github.com/rfjakob/gocryptfs/tests/matrix 0.790s
After:
$ go test
--- FAIL: TestMkfifo (0.00s)
matrix_test.go:755: file exists
--- FAIL: TestMagicNames (0.00s)
matrix_test.go:773: Testing n="gocryptfs.longname.QhUr5d9FHerwEs--muUs6_80cy6JRp89c1otLwp92Cs"
matrix_test.go:773: Testing n="gocryptfs.diriv"
matrix_test.go:815: open /tmp/gocryptfs-test-parent-1026/501766059/default-plain/linktarget: permission denied
--- FAIL: TestChmod (0.00s)
matrix_test.go:849: modeHave 0644 != modeWant 0
FAIL
exit status 1
FAIL github.com/rfjakob/gocryptfs/tests/matrix 0.787s
|
|
|
|
|
|
If we don't, reads hang with -serialize_reads.
|
|
We used to do this
[pid 99182] renameat2(14, "Y_4DAxKvj1QnXmJx2AkrKA", 15, ".", RENAME_NOREPLACE <unfinished ...>
which was not the intention.
|
|
|
|
|
|
|
|
|
|
Fortunately, this just means fixing up the function
signatures.
|
|
|
|
|
|
|
|
Instead, use the new toFuseCtx() function introduced
in an earlier commit.
|
|
|
|
|
|
|
|
|
|
fs.Mount already called srv.Serve() for us.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This was an attempt to make the C code more
similar to Go (which also reads from multiple threads).
However, I still could not repro the ENOENT problems.
https://github.com/rfjakob/gocryptfs/issues/483
|
|
Compiles, but untested otherwise. No caching.
|
|
Also, fix the namespaceMap comment.
|
|
Compiles and mounts but does nothing useful.
|
|
Simplify the tests by using empty files. Empty
files are valid, and we don't check the content
anyway.
Also adjust comment style a little and add
a missing break statement.
|
|
At least 16 bytes due to AES --> at least 22 characters in base64
|
|
Changed invalid file decoding and decryption. Function
DecryptName now shortens the filename until the filename is
decodable and decryptable. Will work with valid **and**
invalid Base64URL delimiter (valid delimiter [0-9a-zA-z_\\-].
If the filename is not decryptable at all, it returns the
original cipher name with flag suffix Changed cli tests to
generate decryptable and undecryptable file names with correct
encrypted content. Replacing #474, extends #393
|