Released 3.7.2

Calling through to request_free() from clean_req() causes deadlock since
sshfs.lock is already held by the caller of clean_req().

.appveyor.yml

@@ -0,0 +1,19 @@
+version: '{build}'
+
+install:
+
+# install WinFsp
+- appveyor DownloadFile https://github.com/billziss-gh/winfsp/releases/download/v1.4B2/winfsp-1.4.18211.msi
+- for %%f in ("winfsp-*.msi") do start /wait msiexec /i %%f /qn INSTALLLEVEL=1000
+
+# install FUSE for Cygwin (64-bit and 32-bit)
+- C:\cygwin64\bin\env.exe -i PATH=/bin bash "%ProgramFiles(x86)%\WinFsp\opt\cygfuse\install.sh"
+- C:\cygwin\bin\env.exe -i PATH=/bin bash "%ProgramFiles(x86)%\WinFsp\opt\cygfuse\install.sh"
+
+# install additional Cygwin packages (64-bit and 32-bit)
+- C:\cygwin64\setup-x86_64.exe -qnNdO -R C:\cygwin64 -s http://cygwin.mirror.constant.com -l C:\cygwin64\var\cache\setup -P libglib2.0-devel -P meson
+- C:\cygwin\setup-x86.exe -qnNdO -R C:\cygwin -s http://cygwin.mirror.constant.com -l C:\cygwin\var\cache\setup -P libglib2.0-devel -P meson
+
+build_script:
+- C:\cygwin64\bin\env.exe -i PATH=/bin bash test\appveyor-build.sh
+- C:\cygwin\bin\env.exe -i PATH=/bin bash test\appveyor-build.sh

.github/ISSUE_TEMPLATE/issue-report.md

@@ -0,0 +1,17 @@
+---
+name: Issue report
+about: Create a report to help us improve
+title: ''
+labels: ''
+assignees: ''
+
+---
+
+PLEASE READ BEFORE REPORTING AN ISSUE
+
+SSHFS does not have any active, regular contributors or developers. The current maintainer continues to apply pull requests and tries to make regular releases, but unfortunately has no capacity to do any development beyond addressing high-impact issues. When reporting bugs, please understand that unless you are including a pull request or are reporting a critical issue, you will probably not get a response.
+
+To prevent the issue tracker from being flooded with issues that no-one is intending to work on, and to give more visibilty to critical issues that users should be aware of and that most urgently need attention, I will also close most bug reports once they've been inactive for a while.
+
+Please note that this isn't meant to imply that you haven't found a bug - you most likely have and I'm grateful that you took the time to report it. Unfortunately, SSHFS is a purely volunteer driven project,
+and at the moment there simply aren't any volunteers.

.gitignore

@@ -34,3 +34,4 @@ sshfs.1
 /patches
 /m4
 .deps/
+/build

.travis.yml

@@ -1,21 +1,21 @@
-sudo: required
-dist: trusty
+dist: focal
+sudo: enabled
 
 language:
     - c
 addons:
   apt:
-    sources:
-    - ubuntu-toolchain-r-test
     packages:
     - valgrind
-    - clang
     - gcc
-    - gcc-6
+    - clang
     - python-docutils
     - python3-pip
-before_install:
-   - sudo -H python3 -m pip install pip==8.1.1
+    - python3-setuptools
+    - ninja-build
+    - meson
+    - python3-pytest
+    - libglib2.0-dev
 install: test/travis-install.sh
 script: test/travis-build.sh
 

AUTHORS

@@ -13,25 +13,56 @@ Miklos Szeredi <miklos@szeredi.hu> (until 12/2015)
 Contributors (autogenerated list)
 ---------------------------------
 
+
 Alan Jenkins <alan.christopher.jenkins@gmail.com>
 Alexander Neumann <alexander@bumpern.de>
 Anatol Pomozov <anatol.pomozov@gmail.com>
+Andrew Stone <a@stne.dev>
 Benjamin Fleischer <fleiben@gmail.com>
 Berserker <berserker.troll@yandex.com>
 Bill Zissimopoulos <billziss@navimatics.com>
+bjoe2k4 <bjoe2k4@users.noreply.github.com>
+Brandon Carter <b-carter@users.noreply.github.com>
 Chris Wolfe <cwolfe@chromium.org>
+Clayton G. Hobbs <clay@lakeserv.net>
+Daniel Lublin <daniel@lublin.se>
+Dominique Martinet <asmadeus@codewreck.org>
+DrDaveD <2129743+DrDaveD@users.noreply.github.com>
+Fabrice Fontaine <fontaine.fabrice@gmail.com>
 gala <gala132@users.noreply.github.com>
+Galen Getsov <4815620+ggetsov@users.noreply.github.com>
 George Vlahavas <vlahavas@gmail.com>
+G.raud Meyer <graud@gmx.com>
 harrim4n <git@harrim4n.com>
 Jakub Jelen <jjelen@redhat.com>
+jeg139 <54814784+jeg139@users.noreply.github.com>
+Josh Triplett <josh@joshtriplett.org>
 Julio Merino <jmmv@google.com>
 Julio Merino <jmmv@meroh.net>
+Junichi Uekawa <dancerj@gmail.com>
+Junichi Uekawa <dancer@netfort.gr.jp>
+kalvdans <github@kalvdans.no-ip.org>
+Michael Forney <mforney@mforney.org>
 Mike Kelly <mike@pair.com>
+Mike Salvatore <mike.s.salvatore@gmail.com>
 Miklos Szeredi <miklos@szeredi.hu>
 Miklos Szeredi <mszeredi@suse.cz>
+mssalvatore <mike.s.salvatore@gmail.com>
 Nikolaus Rath <Nikolaus@rath.org>
 Percy Jahn <email@percyjahn.de>
+Peter Belm <peterbelm@gmail.com>
 Qais Patankar <qaisjp@gmail.com>
+Quentin Rameau <quinq@fifth.space>
+Reid Wagner <wagnerreid@gmail.com>
 Rian Hunter <rian@alum.mit.edu>
+Rian Hunter <rianhunter@users.noreply.github.com>
+Samuel Murray <samuel.murray@outlook.com>
+S. D. Cloudt <s.d.cloudt@student.tue.nl>
+Simon Arlott <70171+nomis@users.noreply.github.com>
+smheidrich <smheidrich@weltenfunktion.de>
+sunwire <50745572+sunwire@users.noreply.github.com>
+Tim Harder <radhermit@gmail.com>
+Timo Savola <timo.savola@iki.fi>
 tpoindessous <thomas@poindessous.com>
+Viktor Szakats <vszakats@users.noreply.github.com>
 Zoltan Kuscsik <zoltan.kuscsik@linaro.org>

ChangeLog.rst

@@ -1,3 +1,84 @@
+Release 3.7.2 (2021-06-08)
+--------------------------
+
+* Added a secondary check so if a mkdir request fails with EPERM an access request will be
+  tried - returning EEXIST if the access was successful.
+	Fixes: https://github.com/libfuse/sshfs/issues/243
+
+
+Release 3.7.1 (2020-11-09)
+--------------------------
+
+* Minor bugfixes.
+
+
+Release 3.7.0 (2020-01-03)
+--------------------------
+
+* New max_conns option enables the use of multiple connections to improve responsiveness
+  during large file transfers. Thanks to Timo Savola for doing most of the implementation
+  work, and thanks to CEA.fr for sponsoring remaining bugfixes and cleanups!
+
+* The `buflimit` workaround is now disabled by default. The corresponding bug in OpenSSH
+  has been fixed in 2007
+  (cf. https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=365541#37), so this shouldn't be
+  needed anymore. If you depend on this workaround, please let the SSHFS maintainers know,
+  otherwise support for the workaround will be removed completely in a future version.
+
+  
+Release 3.6.0 (2019-11-03)
+--------------------------
+
+* Added "-o direct_io" option.
+  This option disables the use of page cache in kernel.
+  This is useful for example if the file size is not known before reading it.
+  For example if you mount /proc dir from a remote host without the direct_io
+  option, the read always will return zero bytes instead of actual data.
+* Added --verbose option.
+* Fixed a number of compiler warnings.
+* Improved performance under OS X.
+
+
+Release 3.5.2 (2019-04-13)
+--------------------------
+
+* Fixed "-o idmap=user" to map both UID and GID on all OSs.
+* Fixed improper handling of sequential spaces spaces in "ssh_command" option
+
+Release 3.5.1 (2018-12-22)
+--------------------------
+
+* Documentation updates
+* Build system updates
+* Added "BindInterface" as valid "-o" option.
+
+Release 3.5.0 (2018-08-28)
+--------------------------
+
+* Fixed error code returned by rename(), allowing proper fallback.
+* Port to Cygwin.
+
+Release 3.4.0 (2018-06-29)
+--------------------------
+
+* Make utimens(NULL) result in timestamp "now" -- no more touched files
+  dated 1970-01-01
+* New `createmode` workaround.
+* Fix `fstat` workaround regression.
+
+Release 3.3.2 (2018-04-29)
+--------------------------
+
+* New `renamexdev` workaround.
+
+Release 3.3.1 (2017-10-25)
+--------------------------
+
+* Manpage is now installed in correct directory.
+* SSHFS now supports (or rather: ignores) some options that it may
+  receive as result of being mounted from ``/etc/mtab``. This includes
+  things like ``user``, ``netdev``, or ``auto``.
+
 SSHFS 3.3.0 (2017-09-20)
 ------------------------
 

README.rst

@@ -10,6 +10,20 @@ servers support and enable this SFTP access by default, so SSHFS is
 very simple to use - there's nothing to do on the server-side.
 
 
+Development Status
+------------------
+
+SSHFS is shipped by all major Linux distributions and has been in
+production use across a wide range of systems for many years. However,
+at present SSHFS does not have any active, regular contributors, and
+there are a number of known issues (see the bugtracker).  The current
+maintainer continues to apply pull requests and makes regular
+releases, but unfortunately has no capacity to do any development
+beyond addressing high-impact issues. When reporting bugs, please
+understand that unless you are including a pull request or are
+reporting a critical issue, you will probably not get a response.
+
+
 How to use
 ----------
 
@@ -32,7 +46,7 @@ To unmount the filesystem::
 
     fusermount -u mountpoint
 
-On BSD and OS-X, to unmount the filesystem::
+On BSD and macOS, to unmount the filesystem::
 
     umount mountpoint
 
@@ -42,7 +56,7 @@ Installation
 
 First, download the latest SSHFS release from
 https://github.com/libfuse/sshfs/releases. On Linux and BSD, you will
-also need to install libfuse_ 3.1.0 or newer. On OS-X, you need
+also need to install libfuse_ 3.1.0 or newer. On macOS, you need
 OSXFUSE_ instead. Finally, you need the Glib_ library with development
 headers (which should be available from your operating system's
 package manager).
@@ -51,7 +65,7 @@ To build and install, we recommend to use Meson_ (version 0.38 or
 newer) and Ninja_.  After extracting the sshfs tarball, create a
 (temporary) build directory and run Meson::
 
-    $ md build; cd build
+    $ mkdir build; cd build
     $ meson ..
 
 Normally, the default build options will work fine. If you
@@ -84,9 +98,3 @@ https://lists.sourceforge.net/lists/listinfo/fuse-sshfs).
 
 Please report any bugs on the GitHub issue tracker at
 https://github.com/libfuse/libfuse/issues.
-
-Professional Support
---------------------
-
-Professional support is available. Please contact Nikolaus Rath
-<Nikolaus@rath.org> for details.

cache.c

@@ -258,7 +258,7 @@ static void *cache_init(struct fuse_conn_info *conn,
 	res = cache.next_oper->init(conn, cfg);
 	
 	// Cache requires a path for each request
-        cfg->nullpath_ok = 0;
+	cfg->nullpath_ok = 0;
 
 	return res;
 }

make_release_tarball.sh

@@ -19,7 +19,8 @@ git archive --format=tar "${TAG}" | tar -x "--directory=${TAG}"
 find "${TAG}" -name .gitignore -delete
 rm "${TAG}/make_release_tarball.sh" \
    "${TAG}/.travis.yml" \
-   "${TAG}/test/travis-*.sh"
+   "${TAG}/test/travis-build.sh" \
+   "${TAG}/test/travis-install.sh"
 tar -cJf "${TAG}.tar.xz" "${TAG}/"
 gpg --armor --detach-sign "${TAG}.tar.xz"
 

meson.build

@@ -1,5 +1,5 @@
-project('sshfs', 'c', version: '3.3.0',
-        meson_version: '>= 0.38',
+project('sshfs', 'c', version: '3.7.2',
+        meson_version: '>= 0.40',
         default_options: [ 'buildtype=debugoptimized' ])
 
 add_global_arguments('-D_REENTRANT', '-DHAVE_CONFIG_H',
@@ -25,7 +25,7 @@ if not cc.compiles(code, args: [ '-O0', '-Werror=unused-result' ])
 endif
 
 
-rst2man = find_program('rst2man', required: false)
+rst2man = find_program('rst2man', 'rst2man.py', required: false)
   
 cfg = configuration_data()
 
@@ -58,7 +58,7 @@ executable('sshfs', sshfs_sources,
 if rst2man.found()
     custom_target('manpages', input: [ 'sshfs.rst' ], output: [ 'sshfs.1' ],
                   command: [rst2man, '@INPUT@', '@OUTPUT@'], install: true,
-                  install_dir: join_paths(get_option('mandir'), '1'))
+                  install_dir: join_paths(get_option('mandir'), 'man1'))
 else
     message('rst2man not found, not building manual page.')
 endif

sshfs.rst

@@ -27,21 +27,24 @@ To unmount it::
 Description
 ===========
 
-SSHFS allows you to mount a remote filesystem using SSH (more
-precisely, the SFTP subsystem). Most SSH servers support and enable
-this SFTP access by default, so SSHFS is very simple to use - there's
-nothing to do on the server-side.
+SSHFS allows you to mount a remote filesystem using SSH (more precisely, the SFTP
+subsystem). Most SSH servers support and enable this SFTP access by default, so SSHFS is
+very simple to use - there's nothing to do on the server-side.
 
-SSHFS uses FUSE (Filesystem in Userspace) and should work on any
-operating system that provides a FUSE implementation. Currently,
-this includes Linux, FreeBSD and Mac OS X.
+By default, file permissions are ignored by SSHFS. Any user that can access the filesystem
+will be able to perform any operation that the remote server permits - based on the
+credentials that were used to connect to the server. If this is undesired, local
+permission checking can be enabled with ``-o default_permissions``.
 
-It is recommended to run SSHFS as regular user (not as root).  For
-this to work the mountpoint must be owned by the user.  If username is
-omitted SSHFS will use the local username. If the directory is
-omitted, SSHFS will mount the (remote) home directory.  If you need to
-enter a password sshfs will ask for it (actually it just runs ssh
-which ask for the password if needed).
+By default, only the mounting user will be able to access the filesystem. Access for other
+users can be enabled by passing ``-o allow_other``. In this case you most likely also
+want to use ``-o default_permissions``.
+
+It is recommended to run SSHFS as regular user (not as root).  For this to work the
+mountpoint must be owned by the user.  If username is omitted SSHFS will use the local
+username. If the directory is omitted, SSHFS will mount the (remote) home directory.  If
+you need to enter a password sshfs will ask for it (actually it just runs ssh which ask
+for the password if needed).
 
 
 Options
@@ -107,12 +110,16 @@ Options
 
    :rename: Emulate overwriting an existing file by deleting and
         renaming.
+   :renamexdev: Make rename fail with EXDEV instead of the default EPERM
+        to allow moving files across remote filesystems.
    :truncate: Work around servers that don't support truncate by
         coping the whole file, truncating it locally, and sending it
         back.
    :fstat: Work around broken servers that don't support *fstat()* by
            using *stat* instead.
    :buflimit: Work around OpenSSH "buffer fillup" bug.
+   :createmode: Work around broken servers that produce an error when passing a
+                non-zero mode to create, by always passing a mode of 0.
 
 -o idmap=TYPE
    How to map remote UID/GIDs to local values. Possible values are:
@@ -149,8 +156,11 @@ Options
 -o directport=PORT
    directly connect to PORT bypassing ssh
 
--o slave
-   communicate over stdin and stdout bypassing network
+-o passive
+   communicate over stdin and stdout bypassing network. Useful for
+   mounting local filesystem on the remote side.  An example using
+   dpipe command would be ``dpipe /usr/lib/openssh/sftp-server = ssh
+   RemoteHostname sshfs :/directory/to/be/shared ~/mnt/src -o passive``
 
 -o disable_hardlink
    With this option set, attempts to call `link(2)` will fail with
@@ -198,6 +208,27 @@ Options
    sets the interval for forced cleaning of the directory cache
    when full.
 
+-o direct_io
+   This option disables the use of page cache (file content cache) in 
+   the kernel for this filesystem.
+   This has several affects:
+
+   1. Each read() or write() system call will initiate one or more read or
+      write operations, data will not be cached in the kernel.
+
+   2. The return value of the read() and write() system calls will correspond 
+      to the return values of the read and write operations. This is useful 
+      for example if the file size is not known in advance (before reading it).
+      e.g. /proc filesystem 
+
+-o max_conns=N
+   sets the maximum number of simultaneous SSH connections
+   to use. Each connection is established with a separate SSH process.
+   The primary purpose of this feature is to improve the responsiveness of the
+   file system during large file transfers. When using more than once
+   connection, the *password_stdin* and *passive* options can not be
+   used, and the *buflimit* workaround is not supported.
+
 In addition, SSHFS accepts several options common to all FUSE file
 systems. These are described in the `mount.fuse` manpage (look
 for "general", "libfuse specific", and "high-level API" options).
@@ -228,8 +259,19 @@ removed it, but before SSHFS had the time to rename the old file. In
 this case, the rename will still fail.
 
 
-SSHFS hangs
-~~~~~~~~~~~
+Permission denied when moving files across remote filesystems
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Most SFTP servers return only a generic "failure" when failing to rename
+across filesystem boundaries (EXDEV).  sshfs normally converts this generic
+failure to a permission denied error (EPERM).  If the option ``-o
+workaround=renamexdev`` is given, generic failures will be considered EXDEV
+errors which will make programs like `mv(1)` attempt to actually move the
+file after the failed rename.
+
+
+SSHFS hangs for no apparent reason
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 In some cases, attempts to access the SSHFS mountpoint may freeze if
 no filesystem activity has occured for some time. This is typically
@@ -238,6 +280,34 @@ without SSHFS being informed about that. As a workaround, you can try
 to mount with ``-o ServerAliveInterval=15``. This will force the SSH
 connection to stay alive even if you have no activity.
 
+
+SSHFS hangs after the connection was interrupted
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+By default, network operations in SSHFS run without timeouts, mirroring the
+default behavior of SSH itself. As a consequence, if the connection to the
+remote host is interrupted (e.g. because a network cable was removed),
+operations on files or directories under the mountpoint will block until the
+connection is either restored or closed altogether (e.g. manually).
+Applications that try to access such files or directories will generally appear
+to "freeze" when this happens.
+
+If it is acceptable to discard data being read or written, a quick workaround
+is to kill the responsible ``sshfs`` process, which will make any blocking
+operations on the mounted filesystem error out and thereby "unfreeze" the
+relevant applications. Note that force unmounting with ``fusermount -zu``, on
+the other hand, does not help in this case and will leave read/write operations
+in the blocking state.
+
+For a more automatic solution, one can use the ``-o ServerAliveInterval=15``
+option mentioned above, which will drop the connection after not receiving a
+response for 3 * 15 = 45 seconds from the remote host. By also supplying ``-o
+reconnect``, one can ensure that the connection is re-established as soon as
+possible afterwards. As before, this will naturally lead to loss of data that
+was in the process of being read or written at the time when the connection was
+interrupted.
+
+
 Mounting from /etc/fstab
 ========================
 

test/appveyor-build.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+
+machine=$(uname -m)
+mkdir build-$machine
+cd build-$machine
+meson ..
+ninja

test/meson.build

@@ -2,7 +2,7 @@ test_scripts = [ 'conftest.py', 'pytest.ini', 'test_sshfs.py',
                  'util.py' ]
 custom_target('test_scripts', input: test_scripts,
               output: test_scripts, build_by_default: true,
-              command: ['cp', '-fP', '--preserve=mode',
+              command: ['cp', '-fPp', 
                         '@INPUT@', meson.current_build_dir() ])
 
 # Provide something helpful when running 'ninja test'

test/test_sshfs.py

@@ -33,13 +33,17 @@ def name_generator(__ctr=[0]):
 @pytest.mark.parametrize("debug", (False, True))
 @pytest.mark.parametrize("cache_timeout", (0,1))
 @pytest.mark.parametrize("sync_rd", (True, False))
-def test_sshfs(tmpdir, debug, cache_timeout, sync_rd, capfd):
+@pytest.mark.parametrize("multiconn", (True,False))
+def test_sshfs(tmpdir, debug, cache_timeout, sync_rd, multiconn, capfd):
     
     # Avoid false positives from debug messages
     #if debug:
     #    capfd.register_output(r'^   unique: [0-9]+, error: -[0-9]+ .+$',
     #                          count=0)
 
+    # Avoid false positives from storing key for localhost 
+    capfd.register_output(r"^Warning: Permanently added 'localhost' .+", count=0)
+    
     # Test if we can ssh into localhost without password
     try:
         res = subprocess.call(['ssh', '-o', 'KbdInteractiveAuthentication=no',
@@ -74,6 +78,8 @@ def test_sshfs(tmpdir, debug, cache_timeout, sync_rd, capfd):
     cmdline += [ '-o', 'entry_timeout=0',
                  '-o', 'attr_timeout=0' ]
 
+    if multiconn:
+        cmdline += [ '-o', 'max_conns=3' ]
     
     new_env = dict(os.environ) # copy, don't modify
 
@@ -102,13 +108,14 @@ def test_sshfs(tmpdir, debug, cache_timeout, sync_rd, capfd):
         # SSHFS only supports one second resolution when setting
         # file timestamps.
         tst_utimens(mnt_dir, tol=1)
+        tst_utimens_now(mnt_dir)
 
         tst_link(mnt_dir, cache_timeout)
         tst_truncate_path(mnt_dir)
         tst_truncate_fd(mnt_dir)
         tst_open_unlink(mnt_dir)
     except:
-        cleanup(mnt_dir)
+        cleanup(mount_process, mnt_dir)
         raise
     else:
         umount(mount_process, mnt_dir)
@@ -403,6 +410,18 @@ def tst_utimens(mnt_dir, tol=0):
         assert abs(fstat.st_atime_ns - atime_ns) < tol*1e9
         assert abs(fstat.st_mtime_ns - mtime_ns) < tol*1e9
 
+def tst_utimens_now(mnt_dir):
+    fullname = pjoin(mnt_dir, name_generator())
+
+    fd = os.open(fullname, os.O_CREAT | os.O_RDWR)
+    os.close(fd)
+    os.utime(fullname, None)
+
+    fstat = os.lstat(fullname)
+    # We should get now-timestamps
+    assert fstat.st_atime != 0
+    assert fstat.st_mtime != 0
+
 def tst_passthrough(src_dir, mnt_dir, cache_timeout):
     name = name_generator()
     src_name = pjoin(src_dir, name)

test/travis-build.sh

@@ -12,7 +12,7 @@ export CC
 TEST_CMD="python3 -m pytest --maxfail=99 test/"
 
 # Standard build with Valgrind
-for CC in gcc gcc-6 clang; do
+for CC in gcc clang; do
     mkdir build-${CC}; cd build-${CC}
     if [ ${CC} == 'gcc-6' ]; then
         build_opts='-D b_lundef=false'

test/travis-install.sh

@@ -2,23 +2,12 @@
 
 set -e
 
-sudo python3 -m pip install pytest meson
-wget https://github.com/ninja-build/ninja/releases/download/v1.7.2/ninja-linux.zip
-unzip ninja-linux.zip
-chmod 755 ninja
-sudo chown root:root ninja
-sudo mv -fv ninja /usr/local/bin
-valgrind --version
-ninja --version
-meson --version
-
 # Install fuse
 wget https://github.com/libfuse/libfuse/archive/master.zip
 unzip master.zip
 cd libfuse-master
 mkdir build
 cd build
-export CC=gcc-6
 meson ..
 ninja
 sudo ninja install
@@ -28,7 +17,7 @@ ls -d1 /usr/local/lib/*-linux-gnu | sudo tee /etc/ld.so.conf.d/usrlocal.conf
 sudo ldconfig
 
 # Setup ssh
-ssh-keygen -b 768 -t rsa -f ~/.ssh/id_rsa -P ''
+ssh-keygen -b 1024 -t rsa -f ~/.ssh/id_rsa -P ''
 cat ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys
 chmod 600 ~/.ssh/authorized_keys
 ssh -o "StrictHostKeyChecking=no" localhost echo "SSH connection succeeded"

test/util.py

@@ -20,13 +20,19 @@ def wait_for_mount(mount_process, mnt_dir,
         elapsed += 0.1
     pytest.fail("mountpoint failed to come up")
 
-def cleanup(mnt_dir):
+def cleanup(mount_process, mnt_dir):
     subprocess.call(['fusermount', '-z', '-u', mnt_dir],
                     stdout=subprocess.DEVNULL,
                     stderr=subprocess.STDOUT)
+    mount_process.terminate()
+    try:
+        mount_process.wait(1)
+    except subprocess.TimeoutExpired:
+        mount_process.kill()
+    
 
 def umount(mount_process, mnt_dir):
-    subprocess.check_call(['fusermount', '-z', '-u', mnt_dir ])
+    subprocess.check_call(['fusermount3', '-z', '-u', mnt_dir ])
     assert not os.path.ismount(mnt_dir)
 
     # Give mount process a little while to terminate. Popen.wait(timeout)