Commit Graph

22 Commits

Author SHA1 Message Date
Jakob Unterwurzacher
393e531afd Fix warnings reported by Go 1.6 "go tool vet -shadow=true"
Warnings were:

  main.go:234: declaration of err shadows declaration at main.go:163:
  internal/fusefrontend/file.go:401: declaration of err shadows declaration at internal/fusefrontend/file.go:379:
  internal/fusefrontend/file.go:419: declaration of err shadows declaration at internal/fusefrontend/file.go:379:
  internal/fusefrontend/fs_dir.go:140: declaration of err shadows declaration at internal/fusefrontend/fs_dir.go:97:
2016-06-14 22:46:23 +02:00
Jakob Unterwurzacher
5b1eed35ee fusefrontend: Utimens: convert ENOENT to EBADF
If /proc/self/fd/X did not exist, the actual error is that the file
descriptor was invalid.

go-fuse's pathfs prefers using an open fd even for path-based operations
but does not take any locks to prevent the fd from being closed.
Instead, it retries the operation by path if it get EBADF. So this
change allows the retry logic to work correctly.

This fixes the error

    rsync: failed to set times on "/tmp/ping.Kgw.mnt/linux-3.0/[...]/.dvb_demux.c.N7YlEM":
    No such file or directory (2)

that was triggered by pingpong-rsync.bash.
2016-06-09 22:21:00 +02:00
Jakob Unterwurzacher
5da292828c fusefrontend: fix chown on dangling symlinks
We (actually, go-fuse) used to call Chown() instead of Lchown()
which meant that the operation would fail on dangling symlinks.

Fix this by calling os.Lchown() ourself. Also add a test case
for this.
2016-06-08 00:32:44 +02:00
Jakob Unterwurzacher
a602e798b1 fusefrontend: report an error if all files in a directory were invalid
Just presenting an empty directory means that the user does not know
that things went wrong unless he checks the syslog or tries to delete
the directory.

It would be nice to report the error even if only some files were
invalid. However, go-fuse does not allow returning the valid
directory entries AND an error.
2016-06-04 16:39:27 +02:00
Jakob Unterwurzacher
5dd9576a11 fusefrontend: replace unreliable "fd < 0" check
... with the "released" boolean.

For some reason, the "f.fd.Fd() < 0" check did not work reliably,
leading to nil pointer panics on the following wlock.lock().

The problem was discovered during fsstress testing and is unlikely
to happen in normal operations.

With this change, we passed 1700+ fsstress iterations.
2016-05-30 09:36:06 +02:00
Jakob Unterwurzacher
1648c54adb fusefrontend: use sync.Once for one-time warnings
Using a simple boolean was racy (which was harmless
in this case) and non-idomatic.
2016-05-29 22:50:03 +02:00
Jakob Unterwurzacher
fd53dfd2ad fusefronted: check Fstat return value on file create
The Fstat call should never fail, but still, if it does return an error
it should be handled properly.
2016-05-29 22:43:48 +02:00
Jakob Unterwurzacher
77813bdc13 fusefrontend: simplify wlockMap
mapMutex can be anonymous and using an RWMutex is overkill
because the lock protects very short sections.
2016-05-29 22:40:05 +02:00
Jakob Unterwurzacher
4c0cb37c50 fusefrontend: remove unused "forgotten" variable
The functionality has long been replaced by the fd < 0
check.
2016-05-29 13:46:47 +02:00
David Gnedt
a93bcabe9c Encrypt path in statfs() calls
Paths in statfs() calls were not encrypted resulting in
an Function not implemented error, when the unencrypted
path didn't exist in the underlying (encrypted)
filesystem.

$ df plain/existingdir
df: ‘plain/existingdir’: Function not implemented
2016-05-12 23:24:39 +02:00
Jakob Unterwurzacher
ba7c798418 fusefrontend: fix panic due to concurrently unregistered wlock
Commit 730291feab properly freed wlock when the file descriptor is
closed. However, concurrently running Write and Truncates may
still want to lock it. Check if the fd has been closed first.
2016-05-08 23:21:20 +02:00
Jakob Unterwurzacher
730291feab fusefrontend: fix wlock memory leak
The write lock was not freed on release, causing a slow memory leak.

This was noticed by running extractloop.bash for 10 hours.
2016-05-05 13:38:39 +02:00
Jakob Unterwurzacher
4d79fba285 prelloc: warn and continue if fallocate(2) is not supported
This makes gocryptfs work at all on ZFS.
See https://github.com/rfjakob/gocryptfs/issues/22 .
2016-04-20 22:47:31 +02:00
Jakob Unterwurzacher
bd5405189e Fix "go tool vet -shadow=true" warnings
Among those one real bug.
2016-04-10 21:31:15 +02:00
Jakob Unterwurzacher
db72fcea41 longnames: fix fsstress failure, use dirfd
Using dirfd-relative operations allows safe lockless handling
of the ".name" files.
2016-04-10 21:31:08 +02:00
Jakob Unterwurzacher
63d3e51734 longnames: use symbolic constants instead of naked ints 2016-04-10 12:36:43 +02:00
Jakob Unterwurzacher
653d4a619c longnames part II: Rename, Unlink, Rmdir, Mknod, Mkdir + tests 2016-02-07 14:02:09 +01:00
Jakob Unterwurzacher
6b5d977cce Move OpenDir to fs_dir.go 2016-02-07 11:29:54 +01:00
Jakob Unterwurzacher
e111e20649 longnames part I: Create and OpenDir work with long filenames > 176 bytes
Todo: Rename, Unlink, Rmdir, Mknod, Mkdir
2016-02-06 22:54:14 +01:00
Jakob Unterwurzacher
c74772bc8d Run go fmt 2016-02-06 20:23:36 +01:00
Jakob Unterwurzacher
b0ee5258b1 Fix tests - were broken by the refactoring 2016-02-06 20:22:45 +01:00
Jakob Unterwurzacher
9078a77850 Move pathfs_frontend to internal/fusefrontend
"git status" for reference:

renamed:    pathfs_frontend/args.go -> internal/fusefrontend/args.go
renamed:    pathfs_frontend/compat_darwin.go -> internal/fusefrontend/compat_darwin.go
renamed:    pathfs_frontend/compat_linux.go -> internal/fusefrontend/compat_linux.go
renamed:    pathfs_frontend/file.go -> internal/fusefrontend/file.go
renamed:    pathfs_frontend/file_holes.go -> internal/fusefrontend/file_holes.go
renamed:    pathfs_frontend/fs.go -> internal/fusefrontend/fs.go
renamed:    pathfs_frontend/fs_dir.go -> internal/fusefrontend/fs_dir.go
renamed:    pathfs_frontend/names.go -> internal/fusefrontend/names.go
renamed:    pathfs_frontend/write_lock.go -> internal/fusefrontend/write_lock.go
modified:   main.go
2016-02-06 19:27:59 +01:00