Compare commits

...

19 Commits

Author SHA1 Message Date
Nikolaus Rath 7f56cfc602 Released 3.3.1 2017-10-25 19:32:34 +01:00
Nikolaus Rath fd3aa6594f meson.build: don't use cp -p instead of --preserve-mode
The latter is not supported on OS X.
2017-10-25 19:30:34 +01:00
Timo Savola 2731a001e5 Fix memory leak on opendir error 2017-10-15 14:31:56 +01:00
Nikolaus Rath 608e7ab89c Ignore fstab-specific mount options
Fixes: #96.
2017-09-26 12:41:30 +01:00
Nikolaus Rath a73dda48bd Install manpage in man1/, not 1/
Fixes: #95.
2017-09-22 10:35:52 +01:00
Nikolaus Rath 7c2550a7e4 Don't attempt to remove non-existing file. 2017-09-20 18:09:59 +01:00
Nikolaus Rath 71aa3b97f6 Released 3.3.0 2017-09-20 18:08:27 +01:00
Nikolaus Rath 14636931ac Add support for printing release contributors. 2017-09-20 18:07:29 +01:00
Nikolaus Rath d193b19a8f Dropped support for writeback caching
As of kernel 4.14, the FUSE module's + writeback implementation is not
compatible with network filesystems, and there are no imminent plans
to change that.

For more details, see
https://marc.info/?l=fuse-devel&m=150592103107662&w=2 or

As a consequence, the -o unreliable_append option has become obsolete
as well.

Fixes: #93
Fixes: #88
Fixes: #81
2017-09-20 17:56:44 +01:00
Nikolaus Rath 949d76d1a2 Add support for mounting from /etc/fstab
Fixes: #92.
2017-09-20 14:43:34 +01:00
Nikolaus Rath 848ff30d90 Dropped FAQ file
Most entries were obsolete. Useful content has been moved to the man
page.
2017-09-20 14:43:34 +01:00
Nikolaus Rath 37c8c1c80b Dropped support for autotools build
It's just too much of a pain to maintain.
2017-09-20 14:43:31 +01:00
Nikolaus Rath 85b950d3c6 Updated manual page with missing options.
Fixes: #82.
2017-09-20 13:32:33 +01:00
Nikolaus Rath b32c4a4ca1 tst_link(): fix test failure 2017-09-20 12:06:31 +01:00
Nikolaus Rath bc47eeb68c travis: fix build procedure. 2017-09-20 11:58:43 +01:00
Nikolaus Rath e4de7d257f Improve documentation for symlink handling options.
Fixes: #51.
2017-09-19 22:34:50 +01:00
Nikolaus Rath 2a818faea5 Use 'buildtype' to set debug/optimization options
Until now, running `mesonconf -D buildtype=debug` didn't have any
effect - which was confusing.
2017-08-22 11:09:03 +02:00
Nikolaus Rath 48cf30d3e2 Converted manpage to Restructured Text 2017-08-18 10:58:47 +02:00
Anatol Pomozov 1f20b5dff5 Bump version number 2017-08-14 10:28:29 +02:00
18 changed files with 390 additions and 673 deletions
+3 -1
View File
@@ -12,8 +12,10 @@ addons:
- clang
- gcc
- gcc-6
- python-docutils
- python3-pip
before_install:
- pip install pip==8.1.1 && pip3 install pip==8.1.1
- sudo -H python3 -m pip install pip==8.1.1
install: test/travis-install.sh
script: test/travis-build.sh
+3
View File
@@ -15,6 +15,7 @@ Contributors (autogenerated list)
Alan Jenkins <alan.christopher.jenkins@gmail.com>
Alexander Neumann <alexander@bumpern.de>
Anatol Pomozov <anatol.pomozov@gmail.com>
Benjamin Fleischer <fleiben@gmail.com>
Berserker <berserker.troll@yandex.com>
Bill Zissimopoulos <billziss@navimatics.com>
@@ -31,6 +32,8 @@ Miklos Szeredi <mszeredi@suse.cz>
Nikolaus Rath <Nikolaus@rath.org>
Percy Jahn <email@percyjahn.de>
Qais Patankar <qaisjp@gmail.com>
Reid Wagner <wagnerreid@gmail.com>
Rian Hunter <rian@alum.mit.edu>
Timo Savola <timo.savola@iki.fi>
tpoindessous <thomas@poindessous.com>
Zoltan Kuscsik <zoltan.kuscsik@linaro.org>
+19
View File
@@ -1,3 +1,22 @@
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)
------------------------
* Dropped support for writeback caching (and, as a consequence,
"unreliable append" operation). As of kernel 4.14, the FUSE module's
writeback implementation is not compatible with network filesystems
and there are no imminent plans to change that.
* Add support for mounting from /etc/fstab
* Dropped support for building with autotools.
* Added missing options to man page.
Release 3.2.0 (2017-08-06)
--------------------------
-242
View File
@@ -1,242 +0,0 @@
SSHFS FAQ
=========
1. I've found a bug and there's no solution in this FAQ, what
should I do?
2. Advantage of sshfs over NFS and Samba
3. Create the device node
4. mv fails with "Operation not permitted"
5. Configuring the ssh connection
6. What are the no_readahead and sshfs_sync options for?
7. Why does df return strange values on partitions mounted via
sshfs?
8. How do I specify the remote mount point (since the example
defaults to the home directory)
9. sshfs hangs after a while
10. Following symlinks on the server side
11. Making absolute symlinks work
12. Mounting as root
13. Exporting via NFS
14. Automatical mounting using /etc/fstab
15. Why does SVN (etc...) fail with permission denied?
16. Why does SVN (etc...) fail to rename files?
17. Is there some neat way to do it in reverse?
18. Might it be reasonable to disallow loops?
19. How to mount through an intermediary ssh server, eg:
localhost -> A -> B; mount B from localhost
20. Alternative Solution
21. I seem to have successfully mounted a remote directory, but
performing an `ls -l` on the directory above the mount point shows
the mount point's attributes as `? ? ? ? ? ?`. Nothing shows up in
the directory either. What am I doing wrong?
--------------------------------------------------------------------------
1. I've found a bug and there's no solution in this FAQ, what should I do?
Please report the bug in the Github issue tracker:
https://github.com/libfuse/sshfs/issues
Also logs with debugging output can be useful for diagnosing the
problem. Try running sshfs with the following options:
sshfs -odebug,sshfs_debug,loglevel=debug ...
Doing strace on the application which fails may also sometimes help:
strace -f -o /tmp/strace application args ...
Note that large messages (over 40k) will be rejected from the mailing
list. So try to keep the logs as short as possible.
2. Advantage of sshfs over NFS and Samba
Users can mount remote resources they already have ssh access to,
without requiring the remote machine to export the resource.
The remote resource can be mounted when it is needed in a location
that is convenient for the user at that time, without needing to rely
on a central, root-controlled file system table.
Automatic mounting, if desired, can be added to a shell script such as
.bashrc (provided authentication is done using RSA/DSA keys).
Resources can be mounted over slow and unreliable (distant)
connections.
3. Create the device node
If you don't use udev, you may get this error message:
fusermount: failed to open /dev/fuse: No such device or address
Before loading the fuse kernel module, create the device node
manually:
mknod -m 666 /dev/fuse c 10 229
4. mv fails with "Operation not permitted".
Use -o workaround=rename (requires sshfs version >= 1.3).
5. Configuring the ssh connection
In addition to flags like -C, -p, and -o SSHOPT...=, you may find it
easier to edit your /.ssh/config file. You can add an entry with any
customization you want, test it with ssh, and finally use it with
sshfs. As a bonus, you get a short mnemonic for your configuration.
6. What are the no_readahead and sshfs_sync options for?
These disable read and write optimizations respectively. They don't
really make sense unless you're doing something special.
7. Why does df return strange values on partitions mounted via sshfs?
Because the SFTP protocol doesn't have a statfs operation this is
currently not possible to display proper usage on remote partition.
8. How do I specfy the remote mount point (since the example defaults to the
home directory)
The example shows:
sshfs hostname: mountpoint
To specify a remote mount point use:
sshfs hostname:remotemountpoint mountpoint
This might be obvious to others, but I ended up looking up the
interface to sftp to see if I could learn how to specify the remote
mount point, then thought about the way that scp specifies the remote
directory, and it worked.
9. sshfs hangs after a while
Mounting works fine, I can use the files in Mountpoint as good as any
other files on my system, but after bit of time, changing nothing on
the remote files sshfs crashes. This means, I can not cd into the
Mountpoint (xterm hangs, nautilus hangs... every program trying to
access the Mountpoint gets stuck, and won't return).
Solution: add
ServerAliveInterval 15
in your .ssh/config (or use -o ServerAliveInterval=15 on the sshfs
command line but I did not test that solution). This will force the
ssh connection to stay alive even if you have no activity.
10. Following symlinks on the server side
The -o follow_symlinks option will enable this.
11. Making absolute symlinks work
Use the -o transform_symlinks option, which will transform absolute
symlinks (ones which point somewhere inside the mount) into relative
ones.
12. Mounting as root
Generally it's not possible to use an sshfs mount as a "real"
filesystem shared between multiple users. Some of this functionality
can be enabled with the -o allow_other and -o default_permissions
options, but files will not be created with the correct ownership,
etc...
13. Exporting via NFS
Use the userspace NFS daemon http://sourceforge.net/projects/unfs
14. Automatical mounting using /etc/fstab
A line in /etc/fstab has the following format:
sshfs#USERNAME@REMOTE_HOST:REMOTE_PATH MOUNT_POINT fuse SSHFS_OPTIONS 0 0
eg.
sshfs#guest@guest.login.com:data /mnt/guest fuse \
uid=1003,gid=100,umask=0,allow_other 0 0
15. Why does SVN (etc...) fail with permission denied?
This is a bug that happens when an application creates a read-only
file opened for writing (e.g. open("foo", O_WRONLY|O_CREAT, 0444))
It has been fixed in sshfs version 1.3, but also requires FUSE version
>=2.5.X and Linux kernel version >=2.6.15.
16. Why does SVN (etc...) fail to rename files?
$ svn co svn://anonsvn.kde.org/home/kde/trunk/KDE/kdelibs
svn: Can't move 'kdelibs/.svn/tmp/entries' to 'kdelibs/.svn/entries':
Operation not permitted
The reason is that SFTP protocol version 3 (which is implemented by
OpenSSH's sftp-server) defines the rename operation slightly
differently than POSIX. The difference is that renaming to an existing
file or directory will fail instead of atomically replacing the old
file.
The -o workaround=rename option will try to emulate POSIX rename
semantics, but it cannot guarantee atomicity. In most of the cases
this doesn't matter, and things will work fine with this option.
17. Is there some neat way to do it in reverse?
You want to mount a USB thumb drive onto a file server that is rather
remote.
Assuming this is difficult because the laptop with the thumb drive is
sitting behind NAT, firewalls, etc. then you need to create a
port-forward:
client$ ssh -R 2222:localhost:22 server
server$ sshfs -p 2222 localhost:/media/usb1 myusb1
(Now, is there is there a smarter way that does not involve port
opening login permissions in an undesireable direction?)
18. Might it be reasonable to disallow loops?
sshfs localhost:/mnt /mnt
This seems to produce undesirable results. --JoshuaRodman
19. How to mount through an intermediary ssh server, eg: localhost -> A -> B;
mount B from localhost
Start by mounting the folder you need that is on "a" to a folder on
"b" then mount the new folder that is on "a" to a folder on localhost.
IE: (These are NOT real commands, but a sequence of steps.
A mounts B:/home/x on /mnt/Bx
localhost mounts A:/mnt/Bx on ~/mydir
20. Alternative Solution:
1) Create a shell script to wrap the tunneling of one ssh command over
another,
$ cat >Atunnel <<EOF
#!/bin/bash
ssh -q A ssh -q "$@"
EOF
$ chmod u+x Atunnel
N.B. make sure to put this somewhere on your path. 2) sshfs mount as
normal but using this script as the ssh command.
$ sshfs -o ssh_command='Atunnel' B: ~/mydir
21. I seem to have successfully mounted a remote directory, but performing an
`ls -l` on the directory above the mount point shows the mount point's
attributes as `? ? ? ? ? ?`. Nothing shows up in the directory either. What
am I doing wrong?
You probably specified a remote path with the tilde (~) in it. This
doesn't seem to work. Instead, specify an absolute remote path:
sshfs username@remote.host:/home/username/whatever my/mount
What options do i use to make playing media files (music) over sshfs more
efficient?
MacFUSE doesn't seem to let me move files from one directory to another. It
first asks for my local user password (i.e. the password on my Macbook Pro)
and then produces the error message "The operation cannot be completed
because one or more required items cannot be found. (Error code -120).
-30
View File
@@ -1,30 +0,0 @@
## Process this file with automake to produce Makefile.in
bin_PROGRAMS = sshfs
SUBDIRS = test
sshfs_SOURCES = sshfs.c cache.c cache.h
if FUSE_OPT_COMPAT
sshfs_SOURCES += compat/fuse_opt.c compat/fuse_opt.h
endif
if DARWIN_COMPAT
sshfs_SOURCES += compat/darwin_compat.c compat/darwin_compat.h
endif
sshfs_LDADD = $(SSHFS_LIBS)
sshfs_CFLAGS = $(SSHFS_CFLAGS)
sshfs_CPPFLAGS = -D_REENTRANT -DFUSE_USE_VERSION=31 -DLIBDIR=\"$(libdir)\" \
-DIDMAP_DEFAULT="\"$(IDMAP_DEFAULT)\""
EXTRA_DIST = sshfs.1.in meson.build ChangeLog.rst
CLEANFILES = sshfs.1 sshfs.1.tmp
dist_man_MANS = sshfs.1
sshfs.1: sshfs.1.in
$(AM_V_GEN)sed \
-e 's/[@]IDMAP_DEFAULT@/$(IDMAP_DEFAULT)/g' \
-e 's/[@]UNMOUNT_COMMAND@/$(UNMOUNT_COMMAND)/g' \
<$(srcdir)/sshfs.1.in >sshfs.1.tmp || exit 1; \
mv sshfs.1.tmp sshfs.1
-51
View File
@@ -75,57 +75,6 @@ tests requires the `py.test`_ Python module)::
.. _Ninja: https://ninja-build.org/
.. _`py.test`: http://www.pytest.org/
Alternate Installation
----------------------
If you are not able to use Meson and Ninja, please report this to the
sshfs mailing list. Until the problem is resolved, you may fall back
to an in-source build using autotools::
$ ./configure
$ make
$ sudo make install
Note that support for building with autotools may disappear at some
point, so if you depend on using autotools for some reason please let
the sshfs developers know!
Caveats
-------
Rename
~~~~~~
Some SSH servers do not support atomically overwriting the destination
when renaming a file. In this case you will get an error when you
attempt to rename a file and the destination already exists. A
workaround is to first remove the destination file, and then do the
rename. SSHFS can do this automatically if you call it with `-o
workaround=rename`. However, in this case it is still possible that
someone (or something) recreates the destination file after SSHFS has
removed it, but before SSHFS had the time to rename the old file. In
this case, the rename will still fail.
Hardlinks
~~~~~~~~~
If the SSH server supports the *hardlinks* extension, SSHFS will allow
you to create hardlinks. However, hardlinks will always appear as
individual files when seen through an SSHFS mount, i.e. they will
appear to have different inodes and an *st_nlink* value of 1.
O_APPEND
~~~~~~~~
When writeback caching is enabled, SSHFS cannot reliably support the
``O_APPEND`` open flag and thus signals an error on open. To enable
support for unreliable ``O_APPEND`` (which may overwrite data if the
file changes on the server at a bad time), mount the file system with
``-o unreliable_append``.
Getting Help
------------
-44
View File
@@ -1,44 +0,0 @@
AC_INIT(sshfs, 3.2.0)
AC_CANONICAL_TARGET
AM_INIT_AUTOMAKE([foreign subdir-objects])
AM_CONFIG_HEADER(config.h)
AC_PROG_CC
AM_PROG_CC_C_O
CFLAGS="$CFLAGS -Wall -W"
LIBS=
AC_SEARCH_LIBS(dlsym, [dl])
LIBS=
case "$target_os" in
*linux*) osname=linux;;
*darwin*) osname=darwin;;
*) osname=unknown;;
esac
export PKG_CONFIG_PATH=/usr/local/lib/pkgconfig:$PKG_CONFIG_PATH
PKG_CHECK_MODULES([SSHFS], [fuse3 >= 3.1 glib-2.0 gthread-2.0])
have_fuse_opt_parse=no
oldlibs="$LIBS"
LIBS="$LIBS $SSHFS_LIBS"
AC_CHECK_FUNC([fuse_opt_parse], [have_fuse_opt_parse=yes])
LIBS="$oldlibs"
if test "$have_fuse_opt_parse" = no -o "$osname" = darwin; then
CFLAGS="$CFLAGS -I${srcdir}/compat"
fi
AM_CONDITIONAL(FUSE_OPT_COMPAT, test "$have_fuse_opt_parse" = no)
AM_CONDITIONAL(DARWIN_COMPAT, test "$osname" = darwin)
AC_CHECK_PROG(UNMOUNT_COMMAND, fusermount, fusermount -u, umount)
# TODO: Figure out why we special-case this in Darwin. Would be nice if
# the default setting was consistent across platforms so we wouldn't need
# to care about it here.
case "$osname" in
darwin) IDMAP_DEFAULT=user ;;
*) IDMAP_DEFAULT=none ;;
esac
AC_SUBST(IDMAP_DEFAULT)
AC_CONFIG_FILES([Makefile test/Makefile])
AC_OUTPUT
+30
View File
@@ -0,0 +1,30 @@
#!/bin/sh
#
# Create tarball from Git tag, removing and adding
# some files.
#
set -e
if [ -z "$1" ]; then
TAG="$(git tag --list 'sshfs-3*' --sort=-taggerdate | head -1)"
else
TAG="$1"
fi
echo "Creating release tarball for ${TAG}..."
mkdir "${TAG}"
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-build.sh" \
"${TAG}/test/travis-install.sh"
tar -cJf "${TAG}.tar.xz" "${TAG}/"
gpg --armor --detach-sign "${TAG}.tar.xz"
PREV_TAG="$(git tag --list 'sshfs-3*' --sort=-taggerdate --merged "${TAG}^"| head -1)"
echo "Contributors from ${PREV_TAG} to ${TAG}:"
git log --pretty="format:%an <%aE>" "${PREV_TAG}..${TAG}" | sort -u
+17 -9
View File
@@ -1,8 +1,8 @@
project('sshfs', 'c', version: '3.1.0',
project('sshfs', 'c', version: '3.3.1',
meson_version: '>= 0.38',
default_options: [ 'buildtype=plain' ])
default_options: [ 'buildtype=debugoptimized' ])
add_global_arguments('-D_REENTRANT', '-DHAVE_CONFIG_H', '-O2', '-g',
add_global_arguments('-D_REENTRANT', '-DHAVE_CONFIG_H',
'-Wall', '-Wextra', '-Wno-sign-compare',
'-Wmissing-declarations', '-Wwrite-strings',
language: 'c')
@@ -25,6 +25,8 @@ if not cc.compiles(code, args: [ '-O0', '-Werror=unused-result' ])
endif
rst2man = find_program('rst2man', required: false)
cfg = configuration_data()
cfg.set_quoted('PACKAGE_VERSION', meson.project_version())
@@ -39,9 +41,6 @@ else
cfg.set_quoted('IDMAP_DEFAULT', 'none')
endif
configure_file(input: 'sshfs.1.in',
output: 'sshfs.1',
configuration : cfg)
configure_file(output: 'config.h',
configuration : cfg)
@@ -56,9 +55,18 @@ executable('sshfs', sshfs_sources,
install: true,
install_dir: get_option('bindir'))
# This is a little ugly. Is there a better way to tell Meson that the
# manpage is in the build directory?
install_man(join_paths(meson.current_build_dir(), 'sshfs.1'))
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'), 'man1'))
else
message('rst2man not found, not building manual page.')
endif
meson.add_install_script('utils/install_helper.sh',
get_option('sbindir'),
get_option('bindir'))
subdir('test')
-239
View File
@@ -1,239 +0,0 @@
.TH SSHFS "1" "April 2008" "SSHFS version 2.0" "User Commands"
.SH NAME
SSHFS \- filesystem client based on ssh
.SH SYNOPSIS
.SS mounting
.TP
\fBsshfs\fP [\fIuser\fP@]\fBhost\fP:[\fIdir\fP] \fBmountpoint\fP [\fIoptions\fP]
.SS unmounting
.TP
\fB@UNMOUNT_COMMAND@ mountpoint\fP
.SH DESCRIPTION
SSHFS (Secure SHell FileSystem) is a file system for Linux (and other
operating systems with a FUSE implementation, such as Mac OS X or FreeBSD)
capable of operating on files on a remote computer using just a secure
shell login on the remote computer. On the local computer where the SSHFS
is mounted, the implementation makes use of the FUSE (Filesystem in
Userspace) kernel module. The practical effect of this is that the end user
can seamlessly interact with remote files being securely served over SSH
just as if they were local files on his/her computer. On the remote
computer the SFTP subsystem of SSH is used.
.PP
If \fIhost\fP is a numeric IPv6 address, it needs to be enclosed in
square brackets.
.SH OPTIONS
.SS "general options:"
.TP
\fB\-o\fR opt,[opt...]
mount options
.TP
\fB\-h\fR \fB\-\-help\fR
print help
.TP
\fB\-V\fR \fB\-\-version\fR
print version
.SS "SSHFS options:"
.TP
\fB\-p\fR PORT
equivalent to '\-o port=PORT'
.TP
\fB\-C\fR
equivalent to '\-o compression=yes'
.TP
\fB\-F\fR ssh_configfile
specifies alternative ssh configuration file
.TP
\fB\-1\fR
equivalent to '\-o ssh_protocol=1'
.TP
\fB\-o\fR reconnect
reconnect to server
.TP
\fB\-o\fR delay_connect
delay connection to server
.TP
\fB\-o\fR sshfs_sync
synchronous writes
.TP
\fB\-o\fR no_readahead
synchronous reads (no speculative readahead)
.TP
\fB\-o\fR sync_readdir
synchronous readdir
.TP
\fB\-o\fR sshfs_debug
print some debugging information
.TP
\fB\-o\fR workaround=LIST
colon separated list of workarounds
.RS 8
.TP
none
no workarounds enabled
.TP
[no]rename
fix renaming to existing file (default: off)
.TP
[no]nodelaysrv
set nodelay tcp flag in ssh (default: off)
.TP
[no]truncate
fix truncate for old servers (default: off)
.TP
[no]buflimit
fix buffer fillup bug in server (default: on)
.RE
.TP
\fB\-o\fR idmap=TYPE
user/group ID mapping (default: @IDMAP_DEFAULT@)
.RS 8
.TP
none
no translation of the ID space
.TP
user
only translate UID/GID of connecting user
.TP
file
translate UIDs/GIDs based upon the contents of \fBuidfile \fR and
\fBgidfile\fR
.RE
.TP
\fB\-o\fR uidfile=FILE
file containing username:uid mappings for \fBidmap=file\fR
.RE
.TP
\fB\-o\fR gidfile=FILE
file containing groupname:gid mappings for \fBidmap=file\fR
.RE
.TP
\fB\-o\fR nomap=TYPE
with idmap=file, how to handle missing mappings
.RS 8
.TP
ignore
don't do any re-mapping
.TP
error
return an error (default)
.RE
.TP
\fB\-o\fR ssh_command=CMD
execute CMD instead of 'ssh'
.TP
\fB\-o\fR ssh_protocol=N
ssh protocol to use (default: 2)
.TP
\fB\-o\fR sftp_server=SERV
path to sftp server or subsystem (default: sftp)
.TP
\fB\-o\fR directport=PORT
directly connect to PORT bypassing ssh
.TP
\fB\-o\fR slave
communicate over stdin and stdout bypassing network
.TP
\fB\-o\fR disable_hardlink
link(2) will return with errno set to ENOSYS. Hard links don't currently work
perfectly on sshfs, and this confuses some programs. If that happens try
disabling hard links with this option.
.TP
\fB\-o\fR transform_symlinks
transform absolute symlinks to relative
.TP
\fB\-o\fR follow_symlinks
follow symlinks on the server
.TP
\fB\-o\fR no_check_root
don't check for existence of 'dir' on server
.TP
\fB\-o\fR password_stdin
read password from stdin (only for pam_mount!)
.TP
\fB\-o\fR SSHOPT=VAL
ssh options (see man ssh_config)
.SS "FUSE options:"
.TP
\fB\-d\fR \fB\-o\fR debug
enable debug output (implies \fB\-f\fR)
.TP
\fB\-f\fR
foreground operation
.TP
\fB\-s\fR
disable multi\-threaded operation
.TP
\fB\-o\fR allow_other
allow access to other users
.TP
\fB\-o\fR allow_root
allow access to root
.HP
\fB\-o\fR default_permissions
enable permission checking by kernel
.TP
\fB\-o\fR fsname=NAME
set filesystem name
.TP
\fB\-o\fR subtype=NAME
set filesystem type
.TP
\fB\-o\fR max_read=N
set maximum size of read requests
.TP
\fB\-o\fR kernel_cache
cache files in kernel
.TP
\fB\-o\fR [no]auto_cache
enable caching based on modification times
.TP
\fB\-o\fR umask=M
set file permissions (octal)
.TP
\fB\-o\fR uid=N
set file owner
.TP
\fB\-o\fR gid=N
set file group
.TP
\fB\-o\fR entry_timeout=T
cache timeout for names (1.0s)
.TP
\fB\-o\fR negative_timeout=T
cache timeout for deleted names (0.0s)
.TP
\fB\-o\fR attr_timeout=T
cache timeout for attributes (1.0s)
.TP
\fB\-o\fR ac_attr_timeout=T
auto cache timeout for attributes (attr_timeout)
.TP
\fB\-o\fR modules=M1[:M2...]
names of modules to push onto filesystem stack
.TP
\fB\-o\fR max_write=N
set maximum size of write requests
.SS "Module options:"
.TP
[subdir]
.TP
\fB\-o\fR subdir=DIR
prepend this directory to all paths (mandatory)
.TP
\fB\-o\fR [no]rellinks
transform absolute symlinks to relative
.TP
[iconv]
.TP
\fB\-o\fR from_code=CHARSET
original encoding of file names (default: UTF-8)
.TP
\fB\-o\fR to_code=CHARSET
new encoding of the file names (default: ISO-8859-2)
.PD
.SH "AUTHORS"
.LP
SSHFS has been written by Miklos Szeredi <miklos@szeredi.hu>.
.LP
This man page was written by Bartosz Fenski <fenio@debian.org> for the
Debian GNU/Linux distribution (but it may be used by others).
+13 -32
View File
@@ -226,7 +226,6 @@ struct sshfs {
int nomap;
int disable_hardlink;
int dir_cache;
int writeback_cache;
int show_version;
int show_help;
int singlethread;
@@ -408,9 +407,6 @@ static struct fuse_opt sshfs_opts[] = {
SSHFS_OPT("disable_hardlink", disable_hardlink, 1),
SSHFS_OPT("dir_cache=yes", dir_cache, 1),
SSHFS_OPT("dir_cache=no", dir_cache, 0),
SSHFS_OPT("writeback_cache=yes", writeback_cache, 1),
SSHFS_OPT("writeback_cache=no", writeback_cache, 0),
SSHFS_OPT("unreliable_append", unrel_append, 1),
SSHFS_OPT("-h", show_help, 1),
SSHFS_OPT("--help", show_help, 1),
@@ -429,8 +425,18 @@ static struct fuse_opt sshfs_opts[] = {
SSHFS_OPT("cache=yes", dir_cache, 1),
SSHFS_OPT("cache=no", dir_cache, 0),
FUSE_OPT_END
FUSE_OPT_KEY("writeback_cache=no", FUSE_OPT_KEY_DISCARD),
FUSE_OPT_KEY("unreliable_append", FUSE_OPT_KEY_DISCARD),
/* These may come in from /etc/fstab - we just ignore them */
FUSE_OPT_KEY("auto", FUSE_OPT_KEY_DISCARD),
FUSE_OPT_KEY("noauto", FUSE_OPT_KEY_DISCARD),
FUSE_OPT_KEY("user", FUSE_OPT_KEY_DISCARD),
FUSE_OPT_KEY("nouser", FUSE_OPT_KEY_DISCARD),
FUSE_OPT_KEY("users", FUSE_OPT_KEY_DISCARD),
FUSE_OPT_KEY("_netdev", FUSE_OPT_KEY_DISCARD),
FUSE_OPT_END
};
static struct fuse_opt workaround_opts[] = {
@@ -1736,10 +1742,6 @@ static void *sshfs_init(struct fuse_conn_info *conn,
// Lookup of . and .. is supported
conn->capable |= FUSE_CAP_EXPORT_SUPPORT;
// Enable writeback cache if supported
if (sshfs.writeback_cache && (conn->capable & FUSE_CAP_WRITEBACK_CACHE))
conn->want |= FUSE_CAP_WRITEBACK_CACHE;
if (!sshfs.delay_connect)
start_processing_thread();
@@ -2150,7 +2152,8 @@ static int sshfs_opendir(const char *path, struct fuse_file_info *fi)
if (!err) {
buf_finish(handle);
fi->fh = (unsigned long) handle;
}
} else
free(handle);
buf_free(&buf);
return err;
}
@@ -2502,24 +2505,6 @@ static int sshfs_open_common(const char *path, mode_t mode,
if (sshfs.dir_cache)
wrctr = cache_get_write_ctr();
/* With writeback cache, kernel may send read requests even
when userspace opened write-only */
if (sshfs.writeback_cache &&
(fi->flags & O_ACCMODE) == O_WRONLY) {
fi->flags &= ~O_ACCMODE;
fi->flags |= O_RDWR;
}
/* Having the kernel handle O_APPEND doesn't work reliably, if
the file changes on the server at the wrong time, we will
overwrite data instead of appending. */
if ((fi->flags & O_APPEND) && sshfs.writeback_cache) {
if(sshfs.unrel_append)
fi->flags &= ~O_APPEND;
else
return -EINVAL;
}
if ((fi->flags & O_ACCMODE) == O_RDONLY)
pflags = SSH_FXF_READ;
else if((fi->flags & O_ACCMODE) == O_WRONLY)
@@ -3364,9 +3349,7 @@ static void usage(const char *progname)
" -o sshfs_sync synchronous writes\n"
" -o no_readahead synchronous reads (no speculative readahead)\n"
" -o sync_readdir synchronous readdir\n"
" -o unreliable_append Enable (unreliable) O_APPEND support\n"
" -d, --debug print some debugging information (implies -f)\n"
" -o writeback_cache=BOOL enable writeback cache {yes,no} (default: yes)\n"
" -o dir_cache=BOOL enable caching of directory contents (names,\n"
" attributes, symlink targets) {yes,no} (default: yes)\n"
" -o dcache_max_size=N sets the maximum size of the directory cache (default: 10000)\n"
@@ -3876,8 +3859,6 @@ int main(int argc, char *argv[])
sshfs.wfd = -1;
sshfs.ptyfd = -1;
sshfs.dir_cache = 1;
sshfs.writeback_cache = 1;
sshfs.unrel_append = 0;
sshfs.show_help = 0;
sshfs.show_version = 0;
sshfs.singlethread = 0;
+273
View File
@@ -0,0 +1,273 @@
=======
SSHFS
=======
---------------------------------------------
filesystem client based on SSH
---------------------------------------------
:Manual section: 1
:Manual group: User Commands
Synopsis
========
To mount a filesystem::
sshfs [user@]host:[dir] mountpoint [options]
If *host* is a numeric IPv6 address, it needs to be enclosed in square
brackets.
To unmount it::
fusermount3 -u mountpoint # Linux
umount mountpoint # OS X, FreeBSD
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 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.
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
=======
-o opt,[opt...]
mount options, see below for details. A a variety of SSH options can
be given here as well, see the manual pages for *sftp(1)* and
*ssh_config(5)*.
-h, --help
print help and exit.
-V, --version
print version information and exit.
-d, --debug
print debugging information.
-p PORT
equivalent to '-o port=PORT'
-f
do not daemonize, stay in foreground.
-s
Single threaded operation.
-C
equivalent to '-o compression=yes'
-F ssh_configfile
specifies alternative ssh configuration file
-1
equivalent to '-o ssh_protocol=1'
-o reconnect
automatically reconnect to server if connection is
interrupted. Attempts to access files that were opened before the
reconnection will give errors and need to be re-opened.
-o delay_connect
Don't immediately connect to server, wait until mountpoint is first
accessed.
-o sshfs_sync
synchronous writes. This will slow things down, but may be useful
in some situations.
-o no_readahead
Only read exactly the data that was requested, instead of
speculatively reading more to anticipate the next read request.
-o sync_readdir
synchronous readdir. This will slow things down, but may be useful
in some situations.
-o workaround=LIST
Enable the specified workaround. See the `Caveats` section below
for some additional information. Possible values are:
:rename: Emulate overwriting an existing file by deleting and
renaming.
: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.
-o idmap=TYPE
How to map remote UID/GIDs to local values. Possible values are:
:none: no translation of the ID space (default).
:user: map the UID/GID of the remote user to UID/GID of the
mounting user.
:file: translate UIDs/GIDs based upon the contents of `--uidfile`
and `--gidfile`.
-o uidfile=FILE
file containing ``username:uid`` mappings for `-o idmap=file`
-o gidfile=FILE
file containing ``groupname:gid`` mappings for `-o idmap=file`
-o nomap=TYPE
with idmap=file, how to handle missing mappings:
:ignore: don't do any re-mapping
:error: return an error (default)
-o ssh_command=CMD
execute CMD instead of 'ssh'
-o ssh_protocol=N
ssh protocol to use (default: 2)
-o sftp_server=SERV
path to sftp server or subsystem (default: sftp)
-o directport=PORT
directly connect to PORT bypassing ssh
-o slave
communicate over stdin and stdout bypassing network
-o disable_hardlink
With this option set, attempts to call `link(2)` will fail with
error code ENOSYS.
-o transform_symlinks
transform absolute symlinks on remote side to relative
symlinks. This means that if e.g. on the server side
``/foo/bar/com`` is a symlink to ``/foo/blub``, SSHFS will
transform the link target to ``../blub`` on the client side.
-o follow_symlinks
follow symlinks on the server, i.e. present them as regular
files on the client. If a symlink is dangling (i.e, the target does
not exist) the behavior depends on the remote server - the entry
may appear as a symlink on the client, or it may appear as a
regular file that cannot be accessed.
-o no_check_root
don't check for existence of 'dir' on server
-o password_stdin
read password from stdin (only for pam_mount!)
-o dir_cache=BOOL
Enables (*yes*) or disables (*no*) the SSHFS directory cache. The
directory cache holds the names of directory entries. Enabling it
allows `readdir(3)` system calls to be processed without network
access.
-o dcache_max_size=N
sets the maximum size of the directory cache.
-o dcache_timeout=N
sets timeout for directory cache in seconds.
-o dcache_{stat,link,dir}_timeout=N
sets separate timeout for {attributes, symlinks, names} in the
directory cache.
-o dcache_clean_interval=N
sets the interval for automatic cleaning of the directory cache.
-o dcache_min_clean_interval=N
sets the interval for forced cleaning of the directory cache
when full.
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).
Caveats / Workarounds
=====================
Hardlinks
~~~~~~~~~
If the SSH server supports the *hardlinks* extension, SSHFS will allow
you to create hardlinks. However, hardlinks will always appear as
individual files when seen through an SSHFS mount, i.e. they will
appear to have different inodes and an *st_nlink* value of 1.
Rename
~~~~~~
Some SSH servers do not support atomically overwriting the destination
when renaming a file. In this case you will get an error when you
attempt to rename a file and the destination already exists. A
workaround is to first remove the destination file, and then do the
rename. SSHFS can do this automatically if you call it with `-o
workaround=rename`. However, in this case it is still possible that
someone (or something) recreates the destination file after SSHFS has
removed it, but before SSHFS had the time to rename the old file. In
this case, the rename will still fail.
SSHFS hangs
~~~~~~~~~~~
In some cases, attempts to access the SSHFS mountpoint may freeze if
no filesystem activity has occured for some time. This is typically
caused by the SSH connection being dropped because of inactivity
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.
Mounting from /etc/fstab
========================
To mount an SSHFS filesystem from ``/etc/fstab``, simply use ``sshfs`
as the file system type. (For backwards compatibility, you may also
use ``fuse.sshfs``).
See also
========
The `mount.fuse(8)` manpage.
Getting Help
============
If you need help, please ask on the <fuse-sshfs@lists.sourceforge.net>
mailing list (subscribe at
https://lists.sourceforge.net/lists/listinfo/fuse-sshfs).
Please report any bugs on the GitHub issue tracker at
https://github.com/libfuse/libfuse/issues.
Authors
=======
SSHFS is currently maintained by Nikolaus Rath <Nikolaus@rath.org>,
and was created by Miklos Szeredi <miklos@szeredi.hu>.
This man page was originally written by Bartosz Fenski
<fenio@debian.org> for the Debian GNU/Linux distribution (but it may
be used by others).
-4
View File
@@ -1,4 +0,0 @@
## Process this file with automake to produce Makefile.in
EXTRA_DIST = meson.build conftest.py pytest.ini test_sshfs.py \
util.py wrong_command.c
+1 -1
View File
@@ -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'
+11 -11
View File
@@ -33,9 +33,7 @@ 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))
@pytest.mark.parametrize("writeback", (False, True))
def test_sshfs(tmpdir, debug, cache_timeout, sync_rd,
writeback, capfd):
def test_sshfs(tmpdir, debug, cache_timeout, sync_rd, capfd):
# Avoid false positives from debug messages
#if debug:
@@ -65,12 +63,6 @@ def test_sshfs(tmpdir, debug, cache_timeout, sync_rd,
if sync_rd:
cmdline += [ '-o', 'sync_readdir' ]
if writeback:
cmdline += [ '-o', 'writeback_cache=yes',
'-o', 'unreliable_append' ]
else:
cmdline += [ '-o', 'writeback_cache=no' ]
# SSHFS Cache
if cache_timeout == 0:
cmdline += [ '-o', 'dir_cache=no' ]
@@ -111,7 +103,7 @@ def test_sshfs(tmpdir, debug, cache_timeout, sync_rd,
# file timestamps.
tst_utimens(mnt_dir, tol=1)
tst_link(mnt_dir)
tst_link(mnt_dir, cache_timeout)
tst_truncate_path(mnt_dir)
tst_truncate_fd(mnt_dir)
tst_open_unlink(mnt_dir)
@@ -282,7 +274,7 @@ def tst_open_unlink(mnt_dir):
def tst_statvfs(mnt_dir):
os.statvfs(mnt_dir)
def tst_link(mnt_dir):
def tst_link(mnt_dir, cache_timeout):
name1 = pjoin(mnt_dir, name_generator())
name2 = pjoin(mnt_dir, name_generator())
shutil.copyfile(TEST_FILE, name1)
@@ -293,6 +285,14 @@ def tst_link(mnt_dir):
os.link(name1, name2)
# The link operation changes st_ctime, and if we're unlucky
# the kernel will keep the old value cached for name1, and
# retrieve the new value for name2 (at least, this is the only
# way I can explain the test failure). To avoid this problem,
# we need to wait until the cached value has expired.
if cache_timeout:
safe_sleep(cache_timeout)
fstat1 = os.lstat(name1)
fstat2 = os.lstat(name2)
for attr in ('st_mode', 'st_dev', 'st_uid', 'st_gid',
+1 -8
View File
@@ -36,13 +36,6 @@ for san in undefined address; do
meson -D b_sanitize=${san} -D b_lundef=false -D werror=true ..
ninja
${TEST_CMD}
sudo ninja install
cd ..
done
# Autotools build
CC=gcc
autoreconf -i
./configure
make
${TEST_CMD}
sudo make install
-1
View File
@@ -2,7 +2,6 @@
set -e
sudo ln -svf $(which python3) /usr/bin/python3
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
+19
View File
@@ -0,0 +1,19 @@
#!/bin/sh
#
# Don't call this script. It is used internally by the Meson
# build system. Thank you for your cooperation.
#
set -e
bindir="$2"
sbindir="$1"
prefix="${MESON_INSTALL_DESTDIR_PREFIX}"
mkdir -p "${prefix}/${sbindir}"
ln -svf --relative "${prefix}/${bindir}/sshfs" \
"${prefix}/${sbindir}/mount.sshfs"
ln -svf --relative "${prefix}/${bindir}/sshfs" \
"${prefix}/${sbindir}/mount.fuse.sshfs"